Encyclopedia Of Chemical Processing volume 1 [1, 1 ed.] 0824755006, 9780824755003

Citation preview

Encyclopedia of Chemical Processing volume 1

Encyclopedias from Taylor & Francis Group Encyclopedia of Biomaterials and Biomedical Engineering (2 Volume Set) Edited by Gary E. Wnek and Gary Bowlin ISBN: 0-8247-5562-6 Encyclopedia of Biopharmaceutical Statistics, Second Edition Edited by Shein-Chung Chow ISBN: 0-8247-4261-3 Encyclopedia of Chemical Processing (5 Volume Set) Edited by Sunggyu Lee ISBN: 0-8247-5563-4 Encyclopedia of Chromatography, Second Edition (2 Volume Set) Edited by Jack Cazes ISBN: 0-8247-2785-1

Encyclopedia of Supramolecular Chemistry (2 Volume Set) Edited by Jerry L. Atwood and Jonathan Steed ISBN: 0-8247-5056-X Encyclopedia of Surface and Colloid Science (4 Volume Set) Edited by Arthur T. Hubbard ISBN: 0-8247-0633-1

Dekker Agropedia Collection (7 Volume Set) ISBN: 0-8247-2194-2 (Also available individually) Encyclopedia of Agricultural, Food, and Biological Engineering Edited by Dennis R. Heldman ISBN: 0-8247-0938-1 Encyclopedia of Animal Science Edited by Wilson G. Pond and Alan Bell ISBN: 0-8247-5496-4

Encyclopedia of Clinical Pharmacy Edited by Joseph DiPiro ISBN: 0-8247-0752-4 Encyclopedia of Corrosion Technology, Second Edition Edited by P.E. Schweitzer ISBN: 0-8247-4878-6 Encyclopedia of Dietary Supplements Edited by Paul M. Coates, Marc R. Blackman, Gordon M. Cragg, Mark Levine, Joel Moss, and Jeffrey D. White ISBN: 0-8247-5504-9 Encyclopedia of Library and Information Science, Second Edition (4 Volume Set) Edited by Miriam Drake ISBN: 0-8247-2075-X Encyclopedia of Medical Genomics and Proteomics (2 Volume Set) Edited by Jürgen Fuchs and Maurizio Podda ISBN: 0-8247-5564-2 Dekker Encyclopedia of Nanoscience and Nanotechnology (5 Volume Set) Edited by James A. Schwarz, Cristian Contescu, and Karol Putyera ISBN: 0-8247-5055-1 Encyclopedia of Optical Engineering (3 Volume Set) Edited by Ronald Driggers ISBN: 0-8247-0940-3 Encyclopedia of Pharmaceutical Technology, Second Edition (3 Volume Set) Edited by James Swarbrick ISBN: 0-8247-2825-4 Encyclopedia of Public Administration and Public Policy (2 Volume Set) Edited by Jack Rabin ISBN: 0-8247-4748-8

Encyclopedia of Pest Management Edited by David Pimentel ISBN: 0-8247-0632-3 Encyclopedia of Plant and Crop Science Edited by Robert M. Goodman ISBN: 0-8247-0944-6 Encyclopedia of Soil Science, Second Edition Edited by Rattan Lal ISBN: 0-8493-3830-1 Encyclopedia of Water Science Edited by B.A. Stewart and Terry Howell ISBN: 0-8247-0948-9

Coming Soon Encyclopedia of Biotechnology in Agriculture and Food Edited by Dennis R. Heldman ISBN: 0-8493-5027-1 Encyclopedia of Pharmaceutical Technology, Third Edition (6 Volume Set) Edited by James Swarbrick ISBN: 0-8493-9399-X Encyclopedia of Surface and Colloid Science, Second Edition (8 Volume Set) Edited by Poniss Somasundaran ISBN: 0-8493-9615-8 Encyclopedia of Energy Engineering (2 Volume Set) Edited by Barney L. Capehart ISBN: 0-8493-3653-8

These titles are available both in print and online. To order, visit: www.crcpress.com Telephone: 1-800-272-7737 • Fax: 1-800-374-3401 E-Mail: [email protected]

Encyclopedia of Chemical Processing volume 1

Edited by:

Sunggyu Lee Department of Chemical Engineering University of Missouri - Columbia Columbia, Missouri U.S.A.

New York London

Published in 2006 by Taylor & Francis Group 270 Madison Avenue New York, NY 10016 © 2006 by Taylor & Francis Group, LLC No claim to original U.S. Government works Printed in the United States of America on acid-free paper 10 9 8 7 6 5 4 3 2 1 International International International International International International International International International International International International

Standard Standard Standard Standard Standard Standard Standard Standard Standard Standard Standard Standard

Book Number-10: Book Number-10: Book Number-10: Book Number-10: Book Number-10: Book Number-10: Book Number-13: Book Number-13: Book Number-13: Book Number-13: Book Number-13: Book Number-13:

0-8247-5563-4 (Set) 0-8247-5500-6 (Vol 1) 0-8247-5557-X (Vol 2) 0-8247-5558-8 (Vol 3) 0-8247-5559-6 (Vol 4) 0-8247-5560-X (Vol 5) 978-0-8247-5563-8 (Set) 978-0-8247-5500-3 (Vol 1) 978-0-8247-5557-7 (Vol 2) 978-0-8247-5558-4 (Vol 3) 978-0-8247-5559-1 (Vol 4) 978-0-8247-5560-7 (Vol 5)

This book contains information obtained from authentic and highly regarded sources. Reprinted material is quoted with permission, and sources are indicated. A wide variety of references are listed. Reasonable efforts have been made to publish reliable data and information, but the author and the publisher cannot assume responsibility for the validity of all materials or for the consequences of their use. No part of this book may be reprinted, reproduced, transmitted, or utilized in any form by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying, microfilming, and recording, or in any information storage or retrieval system, without written permission from the publishers. For permission to photocopy or use material electronically from this work, please access www.copyright.com (http://www.copyright.com/) or contact the Copyright Clearance Center, Inc. (CCC) 222 Rosewood Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organization that provides licenses and registration for a variety of users. For organizations that have been granted a photocopy license by the CCC, a separate system of payment has been arranged. Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe. Library of Congress Cataloging-in-Publication Data Catalog record is available from the Library of Congress

Visit the Taylor & Francis Web site at http://www.taylorandfrancis.com Taylor & Francis Group is the Academic Division of Informa plc.

This Encyclopedia is dedicated to my wife Kyung Paik Lee who put up with me, lovingly supported me, and helped make this venture a success

Sunggyu Lee Editor Department of Chemical Engineering University of Missouri–Columbia Columbia, Missouri, U.S.A.

Editorial Advisory Board John C. Angus Case Western Reserve University, Cleveland, Ohio, U.S.A.

L. James Lee The Ohio State University, Columbus, Ohio, U.S.A.

Alexis T. Bell University of California, Berkeley, California, U.S.A.

Ki-Jun Lee Seoul National University, Seoul, Korea

Dibakar Bhattacharyya University of Kentucky, Lexington, Kentucky, U.S.A.

Chung-Chiun Liu Case Western Reserve University, Cleveland, Ohio, U.S.A.

Milorad P. Dudukovic Washington University, St. Louis, Missouri, U.S.A.

Badie I. Morsi University of Pittsburgh, Pittsburgh, Pennsylvania, U.S.A.

Robert Fulton Dye Dye Engineering & Technology, Sugar Land, Texas, U.S.A. James R. Fair University of Texas, Austin, Texas, U.S.A. Liang-Shih Fan The Ohio State University, Columbus, Ohio, U.S.A. Mehmet Gencer IMET Corporation, Akron, Ohio, U.S.A. Sun-Tak Hwang University of Cincinnati, Cincinnati, Ohio, U.S.A. Michael T. Klein Rutgers University, Piscataway, New Jersey, U.S.A.

Peter R. Pujado UOP LLC, Des Plaines, Illinois, U.S.A. M. M. Sharma HONY, Chembur, Mumbai, India James G. Speight CD&W Inc., Laramie, Wyoming, U.S.A. David G. Wood University of Melbourne, Victoria, Australia Jeffrey Yen Atofina Chemicals, Inc., King of Prussia, Pennsylvania, U.S.A.

Contributors

Kim Aasberg-Petersen = Haldor Topsøe A=S, Lyngby, Denmark Mohamed O. Abdalla = Department of Chemistry, Tuskegee University, Tuskegee, Alabama, U.S.A. Abdullah M. Aitani = King Fahd University of Petroleum and Minerals, Dhahran, Saudi Arabia G. Akay = Process Intensification and Miniaturization Centre, School of Chemical Engineering and Advanced Materials and Institute for Nanoscale Science and Technology, University of Newcastle, Newcastle Upon Tyne, U.K. S. Al-Malaika = Polymer Processing and Performance Research Unit, School of Engineering and Applied Science, Aston University, Birmingham, U.K. Lyle F. Albright = School of Chemical Engineering, Purdue University, West Lafayette, Indiana, U.S.A. T. L. Alford = Department of Chemical and Materials Engineering, Arizona State University, Tempe, Arizona, U.S.A. S. W. Allison = Engineering Science and Technology Division, Oak Ridge National Laboratory (ORNL), Oak Ridge, Tennessee, U.S.A. M. Cengiz Altan = School of Aerospace and Mechanical Engineering, University of Oklahoma, Norman, Oklahoma, U.S.A. Ramin Amin-Sanayei = Arkema Inc., King of Prussia, Pennsylvania, U.S.A. Paul Andersen = Coperion Corporation, Ramsey, New Jersey, U.S.A. Ee Lui Ang = Department of Chemical and Biomolecular Engineering, University of Illinois, Urbana, Illinois, U.S.A. Piero M. Armenante = Otto H. York Department of Chemical Engineering, New Jersey Institute of Technology, Newark, New Jersey, U.S.A. David S. J. Arney = 3M Company, St. Paul, Minnesota, U.S.A. David B. Asay = Department of Chemical Engineering, The Pennsylvania State University, University Park, Pennsylvania, U.S.A. Gabriel Ascanio = URPEI, Department of Chemical Engineering, Ecole Polytechnique, Montreal, Quebec, Canada W. R. Ashurst = University of California–Berkeley, Berkeley, California, U.S.A. Naveed Aslam = University of South Florida, Tampa, Florida, U.S.A. Victor Atiemo-Obeng = Engineering Science and Market Development, The Dow Chemical Company, Midland, Michigan, U.S.A. Andre´ Bakker = Fluent Inc., Lebanon, New Hampshire, U.S.A. Michael W. Balakos = R&D, Su€d-Chemie Inc., Louisville, Kentucky, U.S.A. Shankha K. Banerji = Department of Civil & Environmental Engineering, University of Missouri–Columbia, Columbia, Missouri, U.S.A. Jimmie R. Baran = 3M Company, St. Paul, Minnesota, U.S.A. Stanley M. Barnett = University of Rhode Island, Kingston, Rhode Island, U.S.A. Shubhayu Basu = Department of Chemical and Biomolecular Engineering, The Ohio State University, Columbus, Ohio, U.S.A. Roger N. Beers = The Goodyear Tire & Rubber Company, Akron, Ohio, U.S.A. Ce´line T. Bellehumeur = Department of Chemical and Petroleum Engineering, University of Calgary, Calgary, Alberta, Canada T. J. Bencic = Optical Instrumentation Technology Branch, NASA John H. Glenn Research Center at Lewis Field, Cleveland, Ohio, U.S.A. ix

x

Jonathan W. Bender = Department of Chemical Engineering, University of South Carolina, Columbia, South Carolina, U.S.A. David A. Benko = The Goodyear Tire & Rubber Company, Akron, Ohio, U.S.A. Sujata K. Bhatia = Dupont Central Research and Development, Wilmington, Delaware, U.S.A. Surita R. Bhatia = Department of Chemical Engineering, University of Massachusetts–Amherst, Amherst, Massachusetts, U.S.A. P. R. Bishnoi = Department of Chemical and Petroleum Engineering, University of Calgary, Calgary, Alberta, Canada T. Reg. Bott = School of Engineering, Chemical Engineering, University of Birmingham, Birmingham, U.K. Andrea Bozzano = UOP LLC, Des Plaines, Illinois, U.S.A. Mike Bradford = Jacobs Engineering Group Inc., Houston, Texas, U.S.A. Ian D. Brindle = Brock University, St. Catharines, Ontario, Canada Edmundo Brito-De La Fuente = Departamento de Alimentos y Biotecnologı´a, UNAM, Me´xico, Me´xico Michael C. Brooks = U.S. Environmental Protection Agency, Kerr Research Center, Ada, Oklahoma, U.S.A. Nigel D. Browning = Department of Chemical Engineering and Materials Science, University of California Davis, Lawrence Berkeley National Laboratory, Berkeley, California, U.S.A. David A. Bruce = Department of Chemical and Biomolecular Engineering, Clemson University, Clemson, South Carolina, U.S.A. Joel G. Burken = Department of Civil, Architectural and Environmental Engineering, University of Missouri–Rolla, Rolla, Missouri, U.S.A. J. R. Burns = Protensive Ltd., Bioscience Centre, Centre for Life, Newcastle Upon Tyne, U.K. Richard V. Calabrese = Department of Chemical and Biomolecular Engineering, University of Maryland, College Park, Maryland, U.S.A. Gregg Caldwell = 3M Company, St. Paul, Minnesota, U.S.A. Gerard T. Caneba = Department of Chemical Engineering, Michigan Technological University, Houghton, Michigan, U.S.A. Giovanni Maria Carlomagno = Department of Energetics, Thermofluidynamics and Environmental Control (DETEC), University of Naples Federico II, Napoli, Italy C. Carraro = University of California–Berkeley, Berkeley, California, U.S.A. Eldon D. Case = Chemical Engineering and Materials Science Department, Michigan State University, East Lansing, Michigan, U.S.A. M. R. Cates = Engineering Science and Technology Division, Oak Ridge National Laboratory (ORNL), Oak Ridge, Tennessee, U.S.A. John C. Chadwick = Dutch Polymer Institute (DPI), Laboratory of Polymer Chemistry, Eindhoven University of Technology, Eindhoven, The Netherlands Louay M. Chamra = Southeast Cooling, Heating and Power Application Center, Department of Mechanical Engineering, Mississippi State University, Mississippi State, Mississippi, U.S.A. Jamal Chaouki = Ecole Polytechnique, Montreal, Quebec, Canada Vicki Chen = UNESCO Centre for Membrane Science and Technology, School of Chemical Engineering, University of New South Wales, Sydney, New South Wales, Australia Zhilei Chen = Center for Biophysics and Computational Biology, University of Illinois, Urbana, Illinois, U.S.A. Hyoung J. Choi = Department of Polymer Science and Engineering, Inha University, Incheon, Korea Kyu Yong Choi = Department of Chemical Engineering, University of Maryland, College Park, Maryland, U.S.A. Pyoungho Choi = Albany Nanotech, The College of Nanoscale Science and Engineering (CNSE), State University of New York, Albany, New York, U.S.A.

xi

T. C. Chung = Department of Materials Science and Engineering, The Pennsylvania State University, University Park, Pennsylvania, U.S.A. Matthew A. Clarke = Department of Chemical and Petroleum Engineering, University of Calgary, Calgary, Alberta, Canada Gary Combes = School of Chemical Engineering and Industrial Chemistry, The University of New South Wales, Sydney, New South Wales, Australia Richard F. Cope = Fluid Mechanics and Mixing Group, The Dow Chemical Company, Midland, Michigan, U.S.A. Richard Corkish = ARC Centre of Excellence for Advanced Silicon Photovoltaics and Photonics, University of New South Wales, Sydney, New South Wales, Australia R. A. Cottis = School of Materials, Corrosion and Protection Centre, University of Manchester, Manchester, U.K. Sean A. Curran = Honeywell International Inc., Morristown, New Jersey, U.S.A. Wayne R. Curtis = Department of Chemical Engineering, The Pennsylvania State University, University Park, Pennsylvania, U.S.A. Teresa J. Cutright = Department of Civil Engineering, The University of Akron, Akron, Ohio, U.S.A. Qizhou Dai = Advanced Biomaterials Chemistry, University of British Columbia, Vancouver, British Columbia, Canada A. K. Dalai = Catalysis and Chemical Reactor Engineering Laboratories, Department of Chemical Engineering, University of Saskatchewan, Saskatoon, Canada Douglas A. Dale = Genencor International, Palo Alto, California, U.S.A. Rohit P. Datar = Technical Operations, CPKelco, Okmulgee, Oklahoma, U.S.A. Ravindra Datta = Fuel Cell Center, Worcester Polytechnic Institute, Worcester, Massachusetts, U.S.A. Sharad M. Dave = Bhabha Atomic Research Center, Mumbai, India Forrest M. Davidson, III = The University of Texas, Austin, Texas, U.S.A. Frank Davis = Cranfield University, Silsoe, U.K. Mark DeDecker = Firestone Polymers, Bridgestone=Firestone Research LLC, Akron, Ohio, U.S.A. Fariba Dehghani = School of Chemical Engineering and Industrial Chemistry, The University of New South Wales, Sydney, New South Wales, Australia Shuguang Deng = Chemical Engineering Department, New Mexico State University, Las Cruces, New Mexico, U.S.A. Amy S. Determan = Department of Chemical and Biological Engineering, Iowa State University, Ames, Iowa, U.S.A. Glenn B. DeWolf = URS Corporation, Austin, Texas, U.S.A. Partha Dey = P.A. Consulting, Nashville, Tennessee, U.S.A. R. Dhib = Department of Chemical Engineering, Ryerson University, Toronto, Ontario, Canada Huu D. Doan = Department of Chemical Engineering, Ryerson University, Toronto, Ontario, Canada Mildred S. Dresselhaus = Massachusetts Institute of Technology, Cambridge, Massachusetts, U.S.A. Nishith Dwivedi = Department of Chemical Engineering, Indian Institute of Technology, New Delhi, India Vahid Ebadat = Chilworth Technology, Inc., Princeton, New Jersey, U.S.A. Sina Ebnesajjad = DuPont Fluoroproducts, Chestnut Run Plaza, Wilmington, Delaware, U.S.A. Jeremy S. Edwards = Department of Chemical Engineering, University of Delaware, Newark, Delaware, U.S.A. Brian W. Eggiman = Purdue University, West Lafayette, Indiana, U.S.A. J. Richard Elliott, Jr. = Department of Chemical Engineering, University of Akron, Akron, Ohio, U.S.A.

xii

Morinobu Endo = Faculty of Engineering, Shinshu University, Wakasato, Nagano-shi, Japan Rolf Erni = Department of Chemical Engineering and Materials Science, University of California Davis, Lawrence Berkeley National Laboratory, Berkeley, California, U.S.A. Arthur W. Etchells = DuPont Fellow, Philadelphia, Pennsylvania, U.S.A. L.-S. Fan = Department of Chemical and Biomolecular Engineering, The Ohio State University, Columbus, Ohio, U.S.A. Rajeev Farwaha = Celanese Polymers, Bridgewater, New Jersey, U.S.A. James J. Feng = Department of Chemical and Biological Engineering and Department of Mathematics, University of British Columbia, Vancouver, British Columbia, Canada D. Ferdous = Catalysis and Chemical Reactor Engineering Laboratories, Department of Chemical Engineering, University of Saskatchewan, Saskatoon, Canada Jim C. Fitch = Noria Corporation, Tulsa, Oklahoma, U.S.A. J. F. Forbes = Department of Chemical and Materials Engineering, University of Alberta, Edmonton, Alberta, Canada Neil R. Foster = School of Chemical Engineering and Industrial Chemistry, The University of New South Wales, Sydney, New South Wales, Australia Jonathan Francis = University of Central Lancashire, Preston, U.K. Matthew H. Frey = 3M Company, St. Paul, Minnesota, U.S.A. Joanna D. Fromstein = Department of Chemical Engineering and Applied Chemistry, University of Toronto, Toronto, Ontario, Canada Mina Gabriel = Honeywell International Inc., Morristown, New Jersey, U.S.A. Alfred Gaertner = Genencor International, Palo Alto, California, U.S.A. Prabhu Ganesan = Department of Chemical Engineering, University of South Carolina, Columbia, South Carolina, U.S.A. Shubhra Gangopadhyay = University of Missouri–Columbia, Columbia, Missouri, U.S.A. Hugo S. Garcia = UNIDA, Instituto Tecnologico de Veracruz, Veracruz, Mexico Dinesh Gera = Fluent Incorporated, Morgantown, West Virginia, U.S.A. Richard Gilbert = Department of Wood and Paper Science, North Carolina State University, Raleigh, North Carolina, U.S.A. Giuseppe Giorleo = Department of Materials and Production Engineering (DIMP), University of Naples Federico II, Napoli, Italy S. M. Goedeke = Engineering Science and Technology Division, Oak Ridge National Laboratory (ORNL), Oak Ridge, Tennessee, U.S.A. Scott Gold = Department of Chemical and Biomolecular Engineering, University of Illinois, Urbana, Illinois, U.S.A. Vincent G. Gomes = University of Sydney, Sydney, New South Wales, Australia James G. Goodwin, Jr. = Department of Chemical Engineering, Clemson University, Clemson, South Carolina, U.S.A. John R. Grace = University of British Columbia, Vancouver, British Columbia, Canada Brian P. Grady = School of Chemical, Biological and Materials Engineering, University of Oklahoma, Norman, Oklahoma, U.S.A. Dan F. Graves = Firestone Polymers, Bridgestone=Firestone Research LLC, Akron, Ohio, U.S.A. Erich Grotewold = Department of Plant Cellular and Molecular Biology and Plant Biotechnology Center, The Ohio State University, Columbus, Ohio, U.S.A. Rajiv Grover = Jacobs Engineering Group Inc., Houston, Texas, U.S.A. Ronald W. Gumbs = Gumbs Associates Ltd., East Brunswick, New Jersey, U.S.A. L. Jay Guo = Department of Electrical Engineering and Computer Science, The University of Michigan, Ann Arbor, Michigan, U.S.A. Ram B. Gupta = Department of Chemical Engineering, Auburn University, Auburn, Alabama, U.S.A.

xiii

Byung Gwon Lee = Environment and Process Technology Division, Korea Institute of Science and Technology, Cheongryang, Seoul, South Korea Don C. Haddox = Waterbury, Vermont, U.S.A. Joel M. Haight = Pennsylvania State University, University Park, Pennsylvania, U.S.A. Pradeep Haldar = Albany Nanotech, The College of Nanoscale Science and Engineering (CNSE), State University of New York, Albany, New York, U.S.A. Ian Hamerton = University of Surrey, Surrey, U.K. Youssef K. Hamidi = School of Aerospace and Mechanical Engineering, University of Oklahoma, Norman, Oklahoma, U.S.A. Tobias Hanrath = The University of Texas, Austin, Texas, U.S.A. Douglas P. Harrison = Gordon A. and Mary Cain Department of Chemical Engineering, Louisiana State University, Baton Rouge, Louisiana, U.S.A. Gareth P. Harrison = Institute for Energy Systems, School of Engineering and Electronics, University of Edinburgh, Edinburgh, U.K. Takuya Hayashi = Faculty of Engineering, Shinshu University, Wakasato, Nagano-shi, Japan Douglas G. Hayes = Department Biosystems Engineering and Environmental Science, University of Tennessee, Knoxville, Tennessee, U.S.A. Donald T. Haynie = Biomedical Engineering and Physics, Bionanosystems Engineering Laboratory, Center for Applied Physics Studies, Louisiana Tech University, Ruston, Louisiana, U.S.A. Roland H. Heck = Princeton University, Princeton, New Jersey, U.S.A. William L. Hergenrother = Bridgestone Americas Center for Research and Technology, Bridgestone=Firestone Research LLC, Akron, Ohio, U.S.A. Andrew M. Herring = Department of Chemical Engineering, Colorado School of Mines, Golden, Colorado, U.S.A. Se´amus P. J. Higson = Cranfield University, Silsoe, U.K. Charles G. Hill, Jr. = Department of Chemical and Biological Engineering, University of Wisconsin–Madison, Madison, Wisconsin, U.S.A. John O. Hill = La Trobe University, Melbourne, Victoria, Australia Hugh W. Hillhouse = Purdue University, West Lafayette, Indiana, U.S.A. B. Keith Hodge = Southeast Cooling, Heating and Power Application Center, Department of Mechanical Engineering, Mississippi State University, Mississippi State, Mississippi, U.S.A. J. D. Holladay = Pacific Northwest National Laboratory, Richland, Washington, U.S.A. W. A. Hollerman = Department of Physics, University of Louisiana at Lafayette, Lafayette, Louisiana, U.S.A. W. C. Hsu = Sino-American Silicon Product Inc., Hsinchu, Taiwan, ROC X. D. Hu = R&D, Su€d-Chemie Inc., Louisville, Kentucky, U.S.A. Yinlun Huang = Department of Chemical Engineering and Materials Science, Wayne State University, Detroit, Michigan, U.S.A. Anton Huber = PolySaccharide Initiative, Institut fu¨r Chemie, Karl-Franzens, Universita¨t Graz, Graz, Austria Kang Moo Huh = Department of Polymer Science and Engineering, Chungnam National University, Daejeon, South Korea Raymond L. Huhnke = Biosystems and Agricultural Engineering, Oklahoma State Unviersity, Stillwater, Oklahoma, U.S.A. Steven C. Hukvari = Parr Instrument Company, Moline, Illinois, U.S.A. Scott M. Husson = Department of Chemical and Biomolecular Engineering, Clemson University, Clemson, South Carolina, U.S.A. Sangchul Hwang = Department of Civil Engineering and Surveying, University of Puerto Rico, Mayagu€ez, Puerto Rico Oleg Ilinich = Engelhard Corporation, Iselin, New Jersey, U.S.A.

xiv

R. J. J. Jachuck = Process Intensification and Clean Technology (PICT) Group, Department of Chemical Engineering, Clarkson University, Potsdam, New York, U.S.A. Manish Jain = 3M Company, St. Paul, Minnesota, U.S.A. Krishnan Jayaraman = University of Auckland, Auckland, New Zealand Nigil Satish Jeyashekar = Department of Chemical Engineering, University of Mississippi, University, Mississippi, U.S.A. Myung S. Jhon = Department of Chemical Engineering and Data Storage Systems Center, Carnegie Mellon University, Pittsburgh, Pennsylvania, U.S.A. Tyler Johannes = Department of Chemical and Biomolecular Engineering, University of Illinois, Urbana, Illinois, U.S.A. Bob Johnson = School of Physical and Chemical Sciences, Queensland University of Technology, Brisbane, Queensland, Australia Joshua Jurs = Departments of Chemistry, Mechanical Engineering and Materials Science, and Center for Nanoscale Science and Technology, Rice University, Houston, Texas, U.S.A. John F. Kadla = Faculty of Forestry, Biomaterials Chemistry, The University of British Columbia, Vancouver, British Columbia, Canada Kishore K. Kar = Fluid Mechanics and Mixing Group, The Dow Chemical Company, Midland, Michigan, U.S.A. Thomas E. Karis = Hitachi Global Storage Technologies, San Jose Research Center, San Jose, California, U.S.A. S. Komar Kawatra = Department of Chemical Engineering, Michigan Technological University, Houghton, Michigan, U.S.A. David O. Kazmer = Department of Plastics Engineering, University of Massachusetts, Lowell, Massachusetts, U.S.A. Jason M. Keith = Department of Chemical Engineering, Michigan Technological University, Houghton, Michigan, U.S.A. Sunil Kesavan = Akebono Corporation, Farmington Hills, Michigan, U.S.A. Daeik Kim = University of Southern California, Los Angeles, California, U.S.A. Ji W. Kim = Department of Chemical Engineering and Data Storage Systems Center, Carnegie Mellon University, Pittsburgh, Pennsylvania, U.S.A. Jin-Kuk Kim = Centre for Process Integration, University of Manchester, Manchester, U.K. Seong H. Kim = Department of Chemical Engineering, Materials Research Institute, The Pennsylvania State University, University Park, Pennsylvania, U.S.A. Yoong-Ahm Kim = Faculty of Engineering, Shinshu University, Wakasato, Nagano-shi, Japan Brian P. Kirkmeyer = International Flavors and Fragrances, Union Beach, New Jersey, U.S.A. Dehong Kong = Chilworth Technology, Inc., Monmouth Junction, New Jersey, U.S.A. Brian A. Korgel = The University of Texas, Austin, Texas, U.S.A. Milivoje M. Kostic = Department of Mechanical Engineering, Northern Illinois University, DeKalb, Illinois, U.S.A. J. Kouvetakis = Department of Chemistry and Biochemistry, Arizona State University, Tempe, Arizona, U.S.A. Andrzej Kraslawski = Lappeenranta University of Technology, Lappeenranta, Finland Suzanne M. Kresta = Department of Chemical and Materials Engineering, University of Alberta, Edmonton, Alberta, Canada Satoshi Kubo = Faculty of Forestry, Biomaterials Chemistry, The University of British Columbia, Vancouver, British Columbia, Canada Mark A. Kuehne = NL Chemical Technology, Inc., Mount Prospect, Illinois, U.S.A. Ashok Kumar = Department of Civil Engineering, University of Toledo, Toledo, Ohio, U.S.A. Vimal Kumar = Department of Chemical Engineering, Indian Institute of Technology, New Delhi, India

xv

Arunabha Kundu = Department of Chemical Engineering, Indian Institute of Technology, New Delhi, India Charlotte T. M. Kwok = Department of Chemical and Biomolecular Engineering, University of Illinois, Urbana, Illinois, U.S.A. K. S. Lackner = Department of Earth and Environmental Engineering, Columbia University, New York, New York, U.S.A. Harry J. Lader = Harry Lader and Associates, Inc., Cleveland, Ohio, U.S.A. Yadunandan Lal Dar = Corporate Research, National Starch and Chemical Company, Bridgewater, New Jersey, U.S.A. Joseph M. Lambert, Jr. = Parr Instrument Company, Moline, Illinois, U.S.A. C. W. Lan = Department of Chemical Engineering, National Taiwan University, Taipei, Taiwan, ROC H. Bryan Lanterman = Department of Chemical Engineering, University of Missouri–Columbia, Columbia, Missouri, U.S.A. Pierre Le-Clech = UNESCO Centre for Membrane Science and Technology, School of Chemical Engineering, University of New South Wales, Sydney, New South Wales, Australia James M. Lee = Department of Chemical Engineering, Washington State University, Pullman, Washington, U.S.A. L. James Lee = Department of Chemical Engineering, The Ohio State University, Columbus, Ohio, U.S.A. Sunggyu Lee = Department of Chemical Engineering, University of Missouri–Columbia, Columbia, Missouri, U.S.A. Yoon Seob Lee = Department of Chemical and Biomolecular Engineering, The Ohio State University, Columbus, Ohio, U.S.A. Youn-Woo Lee = Environment and Process Technology Division, Korea Institute of Science and Technology, Cheongryang, Seoul, South Korea Kwok-Wai Lem = Honeywell International Inc., Morristown, New Jersey, U.S.A. Douglas E. Leng = Leng Associates, Midland, Michigan, U.S.A. Alan G. Letki = Alfa Laval Inc., Warminster, Pennsylvania, U.S.A. Christopher Lew = Department of Chemical and Environmental Engineering, University of California at Riverside, Riverside, California, U.S.A. Randy S. Lewis = School of Chemical Engineering, Oklahoma State University, Stillwater, Oklahoma, U.S.A. Bingyun Li = Biomedical Engineering and Physics, Bionanosystems Engineering Laboratory, Center for Applied Physics Studies, Louisiana Tech University, Ruston, Louisiana, U.S.A. Jane C. Li = NL Chemical Technology, Inc., Mount Prospect, Illinois, U.S.A. Jun Li = Department of Polymer Science, University of Southern Mississippi, Hattiesburg, Mississippi, U.S.A. Norman N. Li = NL Chemical Technology, Inc., Mount Prospect, Illinois, U.S.A. Shuang Li = Department of Chemical and Environmental Engineering, University of California at Riverside, Riverside, California, U.S.A. Wenping Li = Department of Chemical Engineering, University of Houston, Houston, Texas, U.S.A. Xiaodong Li = Department of Mechanical Engineering, University of South Carolina, Columbia, South Carolina, U.S.A. Zijian Li = Department of Chemical and Environmental Engineering, University of California at Riverside, Riverside, California, U.S.A. Michael K. Lindell = Hazard Reduction and Recovery Center, Texas A&M University, College Station, Texas, U.S.A. Chung-Chiun Liu = Electronic Design Center and Chemical Engineering Department, Case Western Reserve University, Cleveland, Ohio, U.S.A. Henry Liu = Freight Pipeline Company, Columbia, Missouri, U.S.A.

xvi

Daniel G. Lo¨ffler = Quarens Technologies, Inc., Bend, Oregon, U.S.A. Ali Lohi = Department of Chemical Engineering, Ryerson University, Toronto, Ontario, Canada Stephen J. Lombardo = Department of Chemical Engineering, University of Missouri–Columbia, Columbia, Missouri, U.S.A. Helen H. Lou = Department of Chemical Engineering, Lamar University, Beaumont, Texas, U.S.A. Jorge A. Lubguban = University of Missouri–Columbia, Columbia, Missouri, U.S.A. Douglas K. Ludlow = Chemical and Biological Engineering, University of Missouri–Rolla, Rolla, Missouri, U.S.A. Adriane G. Ludwick = Department of Chemistry, Tuskegee University, Tuskegee, Alabama, U.S.A. Juergen Lueske = Fluid Mechanics and Mixing Group, The Dow Chemical Company, Niedersachsen, Germany Xiaoliang Ma = Clean Fuels and Catalysis Program, The Energy Institute, and Department of Energy and Geo-Environmental Engineering, The Pennsylvania State University, University Park, Pennsylvania, U.S.A. R. Maboudian = University of California–Berkeley, Berkeley, California, U.S.A. Sundararajan V. Madihally = School of Chemical Engineering, Oklahoma State University, Stillwater, Oklahoma, U.S.A. Surya K. Mallapragada = Department of Chemical and Biological Engineering, Iowa State University, Ames, Iowa, U.S.A. Raffaella Mammucari = School of Chemical Engineering and Industrial Chemistry, The University of New South Wales, Sydney, New South Wales, Australia Elizabeth Marden Marshall = Fluent Inc., Lebanon, New Hampshire, U.S.A. T. E. Marlin = Department of Chemical Engineering, McMaster University, Hamilton, Ontario, Canada Stanley Marple = Chemical Engineering Department, University of Houston, Houston, Texas, U.S.A. Richard I. Masel = Department of Chemical and Biomolecular Engineering, University of Illinois, Urbana, Illinois, U.S.A. Takeshi Matsuura = Department of Chemical Engineering, University of Ottawa, Ottawa, Ontario, Canada J. W. Mayer = Department of Chemical and Materials Engineering, Arizona State University, Tempe, Arizona, U.S.A. Kevin P. Menard = PerkinElmer Thermal Laboratory, Materials Science Department, University of North Texas, Denton, Texas, U.S.A. Carosena Meola = Department of Energetics, Thermofluidynamics and Environmental Control (DETEC), University of Naples Federico II, Napoli, Italy John C. Middleton = BHR Group Ltd., Cranfield, Bedfordshire, U.K. Jan D. Miller = Department of Metallurgical Engineering, University of Utah, Salt Lake City, Utah, U.S.A. Patrick L. Mills = Chemical Science and Engineering Laboratory, DuPont Company, Wilmington, Delaware, U.S.A. Kishore K. Mohanty = Department of Chemical Engineering, University of Houston, Houston, Texas, U.S.A. Sanat Mohanty = 3M Company, St. Paul, Minnesota, U.S.A. Satish C. Mohapatra = Advanced Fluid Technologies, Inc., Dynalene Heat Transfer Fluids, Whitehall, Pennsylvania, U.S.A. Alexander B. Morgan = Dow Chemical Company, Core R&D, Midland, Michigan, U.S.A. Sarah E. Morgan = Department of Polymer Science, University of Southern Mississippi, Hattiesburg, Mississippi, U.S.A. Maximiliano Luis Munford = Group of Organic Optoelectronic Devices, Departamento de Fı´sica, Universidade Federal do Parana´, Curitiba, Parana´, Brazil Balaji Narasimhan = Department of Chemical and Biological Engineering, Iowa State University, Ames, Iowa, U.S.A.

xvii

Amarnath Nareddy = Department of Chemical and Biomolecular Engineering, Clemson University, Clemson, South Carolina, U.S.A. Robert Naumann = University of Alabama, Huntsville, Alabama, U.S.A. Flora T. T. Ng = Department of Chemical Engineering, University of Waterloo, Waterloo, Ontario, Canada Anh V. Nguyen = Discipline of Chemical Engineering, The University of Newcastle, Callaghan, New South Wales, Australia Zheng Ni = Department of Chemical and Biomolecular Engineering, University of Illinois, Urbana, Illinois, U.S.A. K. D. P. Nigam = Department of Chemical Engineering, Indian Institute of Technology, New Delhi, India Takamasa Nonaka = Faculty of Engineering, Department of Applied Chemistry and Biochemistry, Kumamoto University, Kurokami, Kumamoto-shi, Japan Paul O’Connor = Akzo Nobel Catalysts, Amersfoort, The Netherlands Kimberly Ogden = Department of Chemical and Environmental Engineering, University of Arizona, Tucson, Arizona, U.S.A. Masahiro Ohshima = Department of Chemical Engineering, Kyoto University, Nishikyo-ku, Kyoto, Japan Cristina Otero = Departamento de Biocata´lisis, Instituto de Catalisis y Petroleoquimica, CSIC, Campus Universidad Autonoma, Cantoblanco, Madrid, Spain Richard P. Palluzi = ExxonMobil Research and Engineering Company, Annandale, New Jersey, U.S.A. A.-H. Park = Department of Chemical and Biomolecular Engineering, The Ohio State University, Columbus, Ohio, U.S.A. Kinam Park = Departments of Pharmaceutics and Biomedical Engineering, Purdue University, West Lafayette, Indiana, U.S.A. Jae Hyung Park = Department of Advanced Polymer and Fiber Materials, Kyung Hee University, Gyeonggi-do, South Korea Kenneth R. Parker = Ken Parker Consultant APL, West Midlands, U.K. Andre´ Avelino Pasa = Thin Films and Surfaces Group, Departamento de Fı´sica, Universidade Federal de Santa Catarina, Floriano´polis, Santa Catarina, Brazil Shwetal Patel = Department of Chemical Engineering, University of Delaware, Newark, Delaware, U.S.A. Rajam Pattabiraman = Central Electrochemical Research Institute, Karaikudi, India Edward L. Paul = Merck and Co., Inc., Sea Girt, New Jersey, U.S.A. Jing Peng = Department of Applied Chemistry, College of Chemistry, Peking University, Beijing, People’s Republic of China A. Penlidis = Institute for Polymer Research, Department of Chemical Engineering, University of Waterloo, Waterloo, Ontario, Canada W. Roy Penney = Department of Chemical Engineering, University of Arkansas, Fayetteville, Arkansas, U.S.A. Mario A. Perez = 3M Company, St. Paul, Minnesota, U.S.A. Ronald W. Perry = School of Public Affairs, Arizona State University, Tempe, Arizona, U.S.A. Ralph W. Pike = Department of Chemical Engineering, Louisiana State University, Baton Rouge, Louisiana, U.S.A. Branko N. Popov = Department of Chemical Engineering, University of South Carolina, Columbia, South Carolina, U.S.A. Mark A. Prelas = Nuclear Science and Engineering Institute, University of Missouri–Columbia, Columbia, Missouri, U.S.A. Todd Pugsley = University of Saskatchewan, Saskatoon, Saskatchewan, Canada Peter R. Pujado´ = UOP LLC, Des Plaines, Illinois, U.S.A. Yunying Qi = Shell Global Solutions (US) Inc., Westhollow Technology Center, Houston, Texas, U.S.A.

xviii

Mehrdad Rafat = Department of Chemical Engineering, University of Ottawa, Ottawa, Ontario, Canada Theodore W. Randolph = Department of Chemical Engineering, University of Colorado, Boulder, Colorado, U.S.A. Harish G. Rao = LFR Inc., Elgin, Illinois, U.S.A. Sanjeev N. Rao = University of Auckland, Auckland, New Zealand James F. Rathman = Department of Chemical and Biomolecular Engineering, The Ohio State University, Columbus, Ohio, U.S.A. Louis G. Reifschneider = Department of Technology, Illinois State University, Normal, Illinois, U.S.A. David G. Retzloff = Department of Chemical Engineering, University of Missouri–Columbia, Columbia, Missouri, U.S.A. Richard W. Rice = Department of Chemical Engineering, Clemson University, Clemson, South Carolina, U.S.A. J. A. Richardson = Anticorrosion Consulting, Durham, U.K. Peter L. Rinaldi = Department of Chemistry, The University of Akron, Akron, Ohio, U.S.A. Clayt Robinson = Department of Chemical and Biomolecular Engineering, The Ohio State University, Columbus, Ohio, U.S.A. Ken K. Robinson = Mega-Carbon Company, St. Charles, Illinois, U.S.A. Alexander F. Routh = Department of Chemical and Process Engineering, University of Sheffield, Sheffield, U.K. Wolfgang Ruettinger = Engelhard Corporation, Iselin, New Jersey, U.S.A. Gregory T. Rushton = Department of Chemistry and Biochemistry, Kennesaw State University, Kennesaw, Georgia, U.S.A. James M. Ryan = Ryan Consulting, Inc., Fort Myers, Florida, U.S.A. Ajit Sadana = Department of Chemical Engineering, University of Mississippi, Life Sciences Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee, U.S.A. Krishnendu Saha = Nuclear Science and Engineering Institute, University of Missouri–Columbia, Columbia, Missouri, U.S.A. Sangrama K. Sahoo = Department of Chemistry, The University of Akron, Akron, Ohio, U.S.A. C. Sanchez = Chimie de la Matie`re Condense´e, Universite´ Pierre et Marie Curie, Jussieu, Paris, France Abhay Sardesai = Department of Chemical Engineering, University of Missouri–Columbia, Columbia, Missouri, U.S.A. Robert J. Schmidt = UOP LLC, Des Plaines, Illinois, U.S.A. Karl B. Schnelle, Jr. = Department of Chemical Engineering, Vanderbilt University, Nashville, Tennessee, U.S.A. M. J. Scorah = Institute for Polymer Research, Department of Chemical Engineering, University of Waterloo, Waterloo, Ontario, Canada Edmund G. Seebauer = Department of Chemical and Biomolecular Engineering, University of Illinois, Urbana, Illinois, U.S.A. Arup K. Sengupta = Department of Civil and Environmental Engineering, Lehigh University, Bethlehem, Pennsylvania, U.S.A. Sukalyan Sengupta = Civil and Environmental Engineering Department, University of Massachusetts, Dartmouth, Massachusetts, U.S.A. Selim M. Senkan = Department of Chemical Engineering, University of California, Los Angeles, California, U.S.A. Xinming Shao = Akebono Corporation, Farmington Hills, Michigan, U.S.A. Zengyi Shao = Department of Chemical and Biomolecular Engineering, University of Illinois, Urbana, Illinois, U.S.A. J. M. Shaw = Department of Chemical and Materials Engineering, University of Alberta, Edmonton, Alberta, Canada

xix

H. H. Sheena = Polymer Processing and Performance Research Unit, School of Engineering and Applied Science, Aston University, Birmingham, U.K. Ken D. Shimizu = Department of Chemistry and Biochemistry, University of South Carolina, Columbia, South Carolina, U.S.A. Allen R. Siedle = 3M Company, St. Paul, Minnesota, U.S.A. Michael R. Simurdiak = Department of Chemical and Biomolecular Engineering, University of Illinois, Urbana, Illinois, U.S.A. Azad Singh = Department of Chemical Engineering, Indian Institute of Technology, New Delhi, India Shivaji Sircar = Chemical Engineering Department, Lehigh University, Bethlehem, Pennsylvania, U.S.A. Andrew W. Sloley = VECO USA, Inc., Bellingham, Washington, U.S.A. Gary S. Smith = Arkema lnc., King of Prussia, Pennsylvania, U.S.A. John M. Smith = School of Engineering, University of Surrey, Guildford, Surrey, U.K. Robin Smith = Centre for Process Integration, University of Manchester, Manchester, U.K. Ryan G. Soderquist = Department of Chemical Engineering, Washington State University, Pullman, Washington, U.S.A. Stephen Sohn = UOP LLC, Des Plaines, Illinois, U.S.A. G. J. A. A. Soler-Illia = Unidad de Actividad Quı´mica, CNEA, Centro Ato´mico Constituyentes, San Martı´n, Buenos Aires, Argentina Chunshan Song = Clean Fuels and Catalysis Program, The Energy Institute and Department of Energy and Geo-Environmental Engineering, The Pennsylvania State University, University Park, Pennsylvania, U.S.A. Richard Q. Song = NL Chemical Technology, Inc., Mount Prospect, Illinois, U.S.A. James G. Speight = CD & W Inc., Laramie, Wyoming, U.S.A. Vijay R. Srinivas = Arkema lnc., King of Prussia, Pennsylvania, U.S.A. Jason Stephenson = Departments of Chemistry, Mechanical Engineering and Materials Science, and Center for Nanoscale Science and Technology, Rice University, Houston, Texas, U.S.A. Michael R. Stoner = Department of Chemical Engineering, University of Colorado, Boulder, Colorado, U.S.A. Truman S. Storvick = Chemical Engineering Department, University of Missouri–Columbia, Columbia, Missouri, U.S.A. Kenneth Strawhecker = Department of Chemical Engineering, The Pennsylvania State University, University Park, Pennsylvania, U.S.A. Kyung W. Suh = Midland, Michigan, U.S.A. Timothy P. Sullivan = Randolph Air Force Base, Texas, U.S.A. Steve Sund = Honeywell International Inc., Morristown, New Jersey, U.S.A. Aydin K. Sunol = University of South Florida, Tampa, Florida, U.S.A. Sermin G. Sunol = University of South Florida, Tampa, Florida, U.S.A. Kenji Takeuchi = Faculty of Engineering, Shinshu University, Wakasato, Nagano-shi, Japan Philippe A. Tanguy = URPEI, Department of Chemical Engineering, Ecole Polytechnique, Montreal, Quebec, Canada Andreas Taubert = Department of Chemistry, University of Basel, Basel, Switzerland Maruicio Terrones = Advanced Materials Department, IPICyT, San Luis Potosı´, SLP, Me´xico Francis Thibault = URPEI, Department of Chemical Engineering, Ecole Polytechnique, Montreal, Quebec, Canada Louis J. Thibodeaux = Louisiana State University, Baton Rouge, Louisiana, U.S.A. Cristina U. Thomas = 3M Company, St. Paul, Minnesota, U.S.A. Karsten E. Thompson = Gordon A. and Mary Cain Department of Chemical Engineering, Louisiana State University, Baton Rouge, Louisiana, U.S.A. Frank M. Tiller = Department of Chemical Engineering, University of Houston, Houston, Texas, U.S.A.

xx

Maria P. Torres = Department of Chemical and Biological Engineering, Iowa State University, Ames, Iowa, U.S.A. James M. Tour = Departments of Chemistry, Mechanical Engineering and Materials Science, and Center for Nanoscale Science, and Technology, Rice University, Houston, Texas, U.S.A. Linh T. T. Tran = Discipline of Chemical Engineering, The University of Newcastle, Callaghan, New South Wales, Australia Drew D. Troyer = Noria Corporation, Tulsa, Oklahoma, U.S.A. Maxwell Tsai = NL Chemical Technology, Inc., Mount Prospect, Illinois, U.S.A. Tom Chunghu Tsai = The Dow Chemical Company, Houston, Texas, U.S.A. Uday T. Turaga = Clean Fuels and Catalysis Program, The Energy Institute and Department of Energy and Geo-Environmental Engineering, The Pennsylvania State University, University Park, Pennsylvania, U.S.A. Simant R. Upreti = Department of Chemical Engineering, Ryerson University, Toronto, Ontario, Canada Vikrant N. Urade = Purdue University, West Lafayette, Indiana, U.S.A. Ivo F. J. Vankelecom = Centre for Surface Chemistry and Catalysis, Department of Interphase Chemistry, Faculty of Agricultural and Applied Biological Sciences, Katholieke Universiteit Leuven, Leuven, Belgium Charanya Varadarajan = Department of Civil Engineering, University of Toledo, Toledo, Ohio, U.S.A. Angel Velez = Nuclear Science and Engineering Institute, University of Missouri–Columbia, Columbia, Missouri, U.S.A. Abhilash Vijayan = Department of Civil Engineering, University of Toledo, Toledo, Ohio, U.S.A. V. V. Viswanathan = Pacific Northwest National Laboratory, Richland, Washington, U.S.A. Dionisios G. Vlachos = Department of Chemical Engineering and Center for Catalytic Science and Technology, University of Delaware, Newark, Delaware, U.S.A. Tuan Vo-Dinh = Life Sciences Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee, U.S.A. Bipin V. Vora = UOP LLC, Des Plaines, Illinois, U.S.A. Chun Wang = Department of Chemical Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts, U.S.A. J. R. White = School of Chemical Engineering and Advanced Materials, University of Newcastle Upon Tyne, Newcastle Upon Tyne, U.K. Kimberly A. Woodhouse = Department of Chemical Engineering and Applied Chemistry, University of Toronto, Toronto, Ontario, Canada Guy B. Woodle = UOP LLC, Des Plaines, Illinois, U.S.A. Jiangning Wu = Department of Chemical Engineering, Ryerson University, Toronto, Ontario, Canada Eleanore T. Wurtzel = Department of Biological Sciences, Lehman College, The City University of New York, Bronx, New York, U.S.A. Nicholas Patrick Wynn = Sulzer Chemtech GmbH, Neunkirchen, Germany Xuekun Xing = NTK Powderdex, Inc., Wixom, Michigan, U.S.A. and Department of Chemical Engineering, Case Western Reserve University, Cleveland, Ohio, U.S.A. Yushan Yan = Department of Chemical and Environmental Engineering, University of California at Riverside, Riverside, California, U.S.A. Shang-Tian Yang = Department of Chemical and Biomolecular Engineering, The Ohio State University, Columbus, Ohio, U.S.A. Xiaobo Yang = Institute of Physical Chemistry and Electrochemistry, University of Hanover, Hanover, Germany Hirotsugu K. Yasuda = University of Missouri–Columbia, Columbia, Missouri, U.S.A. Mingli Ye = Departments of Pharmaceutics and Biomedical Engineering, Purdue University, West Lafayette, Indiana, U.S.A. Jeffrey H. Yen = Arkema Inc., King of Prussia, Pennsylvania, U.S.A.

xxi

Teh Fu Yen = University of Southern California, Los Angeles, California, U.S.A. W. S. Yip = Suncor Energy Inc., Fort McMurray, Alberta, Canada W. C. Yu = Department of Molecular Science and Engineering, National Taipei University of Technology, Taipei, Taiwan, ROC John Zabasajja = MOS 12, Technology and Manufacturing, Motorola Semiconductor Products Sector (SPS), Chandler, Arizona, U.S.A. Jacques L. Zakin = Department of Chemical Engineering, The Ohio State University, Columbus, Ohio, U.S.A. Sergei V. Zelentsov = Chemical Department, Nizhny Novgorod State University, Nizhny Novgorod, Russia Nadezda V. Zelentsova = Chemical Department, Nizhny Novgorod State University, Nizhny Novgorod, Russia Ying Zhang = Department of Chemical Engineering, The Ohio State University, Columbus, Ohio, U.S.A. Huimin Zhao = Department of Chemical and Biomolecular Engineering, University of Illinois, Urbana, Illinois, U.S.A. Haishan Zheng = Department of Chemical Engineering, Michigan Technological University, Houghton, Michigan, U.S.A. X.-Y. Zou = Oilphase-DBR, Schlumberger, Edmonton, Alberta, Canada

Contents

Contributors Topical Table of Contents Preface

Volume 1 Absorption Equipment = Karl B. Schnelle, Jr. and Partha Dey . . . . . . . . . . . . . Activated Sludge Process = Shankha K. Banerji . . . . . . . . . . . . . . . . . . . . . . Adsorption = Shivaji Sircar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Advanced Oxidation = Sangchul Hwang . . . . . . . . . . . . . . . . . . . . . . . . . . . Alkaline Zn–MnO2 Batteries = Chung-Chiun Liu and Xuekun Xing . . . . . . . . . Alkylation Processes to Produce High-Quality Gasolines = Lyle F. Albright and James M. Ryan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Animal Cell Culture = Shang-Tian Yang and Shubhayu Basu . . . . . . . . . . . . . . Antioxidants = S. Al-Malaika and H. H. Sheena . . . . . . . . . . . . . . . . . . . . . . Biocatalysis = Tyler Johannes, Michael R. Simurdiak, and Huimin Zhao . . . . . . Biofilms = T. Reg. Bott . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Biofuels and Bioenergy = Dinesh Gera . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bioinformatics and Modeling Biological Systems = Shwetal Patel and Jeremy S. Edwards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Biomass to Ethanol = Randy S. Lewis, Rohit P. Datar, and Raymond L. Huhnke . Biomaterials = Sujata K. Bhatia and Surita R. Bhatia . . . . . . . . . . . . . . . . . . BioMEMS = L. James Lee . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Biomolecular Engineering = Zengyi Shao, Ee Lui Ang, and Huimin Zhao . . . . . . Bioprocess and Chemical Process Intensification = G. Akay . . . . . . . . . . . . . . Bioprocessing = Ryan G. Soderquist and James M. Lee . . . . . . . . . . . . . . . . . Bioremediation = Teresa J. Cutright . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bioseparations = Shubhayu Basu and Shang-Tian Yang . . . . . . . . . . . . . . . . . Blowing Agent = Kyung W. Suh . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Branching Level Detection in Polymers = M.J. Scorah, R. Dhib, and A. Penlidis . Bubble Cap Tray = Stanley Marple . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bulk Molding and Sheet Molding Compounds = Sanjeev N. Rao and Krishnan Jayaraman . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Capsule Pipeline = Henry Liu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Carbon Dioxide Capture and Disposal: Carbon Sequestration = K. S. Lackner, A.-H. Park, and L.-S. Fan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Carbon Fibers from Lignin-Recyclable Plastic Blends = Satoshi Kubo and John F. Kadla . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Carbon Nanotubes = Morinobu Endo, Yoong-Ahm Kim, Takuya Hayashi, Kenji Takeuchi, Maruicio Terrones, and Mildred S. Dresselhaus . . . . . . . . . Catalyst Preparation = X. D. Hu and Michael W. Balakos . . . . . . . . . . . . . . . xxiii

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. 57 . 67 . 81 . 101 . 111 . 121

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

1 11 25 41 51

131 143 153 161 171 183 199 207 221 237 251 269

. . . . . . . . . . 283 . . . . . . . . . . 295 . . . . . . . . . . 305 . . . . . . . . . . 317 . . . . . . . . . . 333 . . . . . . . . . . 345

xxiv

Catalytic Combustion for Thermal Energy Generation = Daniel G. Lo¨ffler . . . . . Catalytic Cracking = Paul O’Connor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Catalytic Dehydrogenation = Bipin V. Vora and Peter R. Pujado´ . . . . . . . . . . . Catalytic Naphtha Reforming = Abdullah M. Aitani . . . . . . . . . . . . . . . . . . . Centrifuges = Alan G. Letki . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ceramics = Stephen J. Lombardo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chemical Mechanical Planarization in Integrated Circuit Manufacturing = John Zabasajja . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chemical Vapor Deposition = David G. Retzloff . . . . . . . . . . . . . . . . . . . . . Chiral Drug Separation = Bingyun Li and Donald T. Haynie . . . . . . . . . . . . . . Chlorofluorocarbons = Byung Gwon Lee and Youn-Woo Lee . . . . . . . . . . . . . . CHP Technology/Systems = Louay M. Chamra and B. K. Hodge . . . . . . . . . . . Chromatographic Separations = Scott M. Husson . . . . . . . . . . . . . . . . . . . . . . Coal–Water Slurries = S. Komar Kawatra . . . . . . . . . . . . . . . . . . . . . . . . . . Computational Fluid Dynamics = Andre´ Bakker and Elizabeth Marden Marshall . Computer-Aided Process Engineering = Andrzej Kraslawski . . . . . . . . . . . . . . . Conductive Polymers = Ronald W. Gumbs . . . . . . . . . . . . . . . . . . . . . . . . . . Contact Angles, Surface Tension, and Capillarity = Peter R. Pujado´ . . . . . . . . . Corrosion in the Process Industries = J. A. Richardson and R. A. Cottis . . . . . . . Critical Phase Behavior = J. Richard Elliott, Jr. . . . . . . . . . . . . . . . . . . . . . . Cross-Linked Polyethylene = Carosena Meola, Giovanni Maria Carlomagno, and Giuseppe Giorleo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Crystal Growth = C. W. Lan, W. C. Yu, and W. C. Hsu . . . . . . . . . . . . . . . . . . Cumene Production = Robert J. Schmidt . . . . . . . . . . . . . . . . . . . . . . . . . . . Dehumidification = Louay M. Chamra and B. Keith Hodge . . . . . . . . . . . . . . . Denitrogenation = Daeik Kim and Teh Fu Yen . . . . . . . . . . . . . . . . . . . . . . . Design of Extrusion Dies = Milivoje M. Kostic and Louis G. Reifschneider . . . . Desulfurization = Chunshan Song, Uday T. Turaga, and Xiaoliang Ma . . . . . . . Detergent Alkylate = Bipin Vora, Andrea Bozzano, and Stephen Sohn . . . . . . . Detergent Enzymes = Michael R. Stoner, Douglas A. Dale, Alfred Gaertner, and Theodore W. Randolph . . . . . . . . . . . . . . . . . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

361 371 379 397 407 417

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

429 441 449 459 469 481 495 505 517 527 539 549 563

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

577 589 603 617 627 633 651 663

. . . . . . . . . . 673

Volume 2 Diamond Films = Angel L. Velez and Mark A. Prelas . . . . . . . . . . . . Diamond-Like Carbon Films = Angel Velez and Mark A. Prelas . . . . . . Differential Scanning Calorimetry = John O. Hill . . . . . . . . . . . . . . . Dimethyl Ether = Abhay Sardesai . . . . . . . . . . . . . . . . . . . . . . . . . Dimethylcarbonate = Byung Gwon Lee . . . . . . . . . . . . . . . . . . . . . . Distillation Column Design: Packing = Andrew W. Sloley . . . . . . . . . . Distillation Column Design: Trays = Andrew W. Sloley . . . . . . . . . . . . Drag Reducing Agents = Jacques L. Zakin, Ying Zhang, and Yunying Qi Dust Explosion Hazard Assessment and Control = Vahid Ebadat . . . . . Dynamic Mechanical Thermal Analysis = Kevin P. Menard . . . . . . . . . Education on Plant Design = Truman S. Storvick . . . . . . . . . . . . . . . . Electrodeposition = Andre´ Avelino Pasa and Maximiliano Luis Munford Electronic Chemical Sensors = Chung-Chiun Liu . . . . . . . . . . . . . . . . Electroplating = Helen H. Lou and Yinlun Huang . . . . . . . . . . . . . . . Electrostatic Precipitation = Kenneth R. Parker . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

685 695 699 707 719 729 749 767 787 799 813 821 833 839 849

xxv

Emulsion Polymerization = Vincent G. Gomes . . . . . . . . . . . . . . . . . . . . . . . . Enhanced Oil Recovery = Kishore K. Mohanty and Gerard T. Caneba . . . . . . . . . Environmental Chemodynamics = Louis J. Thibodeaux . . . . . . . . . . . . . . . . . . . Environmental Law and Policy = Don C. Haddox and Teresa J. Cutright . . . . . . . Epoxy Resins = Ian Hamerton . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ethylbenzene = Guy B. Woodle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fermentation Processes = Kimberly Ogden . . . . . . . . . . . . . . . . . . . . . . . . . . Fermenter Design = Kishore K. Kar, Juergen Lueske, and Richard F. Cope . . . . . Fluid Flow = Theodore Reginald Bott . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fluid Transport in Porous Media = Michael C. Brooks . . . . . . . . . . . . . . . . . . . Fluidization = A.-H. Park and L.-S. Fan . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fluidized Bed Reactor = John R. Grace, Jamal Chaouki, and Todd Pugsley . . . . . Fluorescent Coatings for High Temperature Phosphor Thermometry = S. W. Allison, W. A. Hollerman, S. M. Goedeke, M. R. Cates, and T. J. Bencic . . . . . . . . . . Fluoropolymers = Sina Ebnesajjad . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fouling of Heat Exchangers = T. Reg. Bott . . . . . . . . . . . . . . . . . . . . . . . . . . Fractal Geometry: Applications = Douglas K. Ludlow . . . . . . . . . . . . . . . . . . . . Free-Radical Polymerization = Yadunandan Lal Dar, Rajeev Farwaha, and Gerard T. Caneba . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Friction Materials = Sunil Kesavan and Xinming Shao . . . . . . . . . . . . . . . . . . Fuel Cell Membranes = Andrew M. Herring . . . . . . . . . . . . . . . . . . . . . . . . . . Functional Biomaterials = Chun Wang . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Gas Explosion Hazard: Prevention and Protection = Dehong Kong . . . . . . . . . . . Gas–Liquid Contactors = Kishore K. Kar, Richard F. Cope, and Juergen Lueske . . Gas–Liquid Mixing in Agitated Reactors = John C. Middleton, John M. Smith, and Piero M. Armenante . . . . . . . . . . . . . . . . . . . . . . . . . Gas-Phase Lubrication of MEMS Devices: Using Alcohol Vapor Adsorption Isotherm for Lubrication of Silicon Oxides = Kenneth Strawhecker, David B. Asay, and Seong H. Kim . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Gas–Solid Reactions = Douglas P. Harrison . . . . . . . . . . . . . . . . . . . . . . . . . . Gas-to-Liquid Mass Transfer = Huu D. Doan, Simant R. Upreti, and Ali Lohi . . . Geothermal Energy = Sunggyu Lee and H. Bryan Lanterman . . . . . . . . . . . . . . Greenhouse Gas Management for Multiplant Complexes = Ralph W. Pike . . . . . . . Heat Exchanger Operation and Troubleshooting = T. Reg. Bott . . . . . . . . . . . . . . Heat Transfer Fluids = Satish C. Mohapatra . . . . . . . . . . . . . . . . . . . . . . . . . Heavy Water (Deuterium Oxide) = Sharad M. Dave . . . . . . . . . . . . . . . . . . . . Heterogeneous Catalysis = Richard W. Rice and James G. Goodwin, Jr. . . . . . . . High-Pressure Reactor Design = Joseph M. Lambert, Jr. and Steven C. Hukvari . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hollow Fiber Technology = Vicki Chen and Pierre Le-Clech . . . . . . . . . . . . . . . Hybrid Materials (Organic–Inorganic) = C. Sanchez and G. J. A. A. Soler-Illia . . . Hydrocracking = James G. Speight . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hydrodesulfurization = James G. Speight . . . . . . . . . . . . . . . . . . . . . . . . . . . Hydrodynamics of Trickle-Bed Reactors = K. D. P. Nigam and Arunabha Kundu . . Hydrogels = Jae Hyung Park, Kang Moo Huh, Mingli, Ye, and Kinam Park . . . . . Hydrogen Bonding = J. Richard Elliott, Jr. . . . . . . . . . . . . . . . . . . . . . . . . . . Hydrogenation = Xiaobo Yang . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hydrogenation Reactions in Dense Gas Systems = Gary Combes, Fariba Dehghani, Raffaella Mammucari, and Neil R. Foster . . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

863 881 891 899 911 929 941 951 975 987 997 1009

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

1021 1031 1043 1053

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

1057 1071 1085 1099 1109 1119

. . . . . . . . . 1131

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

1143 1151 1163 1175 1189 1203 1211 1221 1235

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

1245 1253 1267 1281 1289 1297 1307 1319 1325

. . . . . . . . . 1337

xxvi

Hydrophilic Polymers for Biomedical Applications = Frank Davis and Se´amus P. J. Higson . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1349 Hydrotreating Catalysts and Processes: Current Status and Path Forward = Arunabha Kundu, Nishith Dwivedi, Azad Singh, and K. D. P. Nigam . . . . . . . . . . . . . . . . 1357

Volume 3 Immobilized Enzyme Technology = Charles G. Hill, Jr., Cristina Otero, and Hugo S. Garcia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Incineration and Combustion = Selim M. Senkan . . . . . . . . . . . . . . . . . . . . . . . Injection Molding = David O. Kazmer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ion Exchange = Sukalyan Sengupta and Arup K. Sengupta . . . . . . . . . . . . . . . . Ion Exchange Resin = Sukalyan Sengupta and Arup K. SenGupta . . . . . . . . . . . Latex Processing = Alexander F. Routh . . . . . . . . . . . . . . . . . . . . . . . . . . . . Liquid–Liquid Mixing in Agitated Reactors = Richard V. Calabrese, Douglas E. Leng, and Piero M. Armenante . . . . . . . . . . . . . . . . . . . . . . . . Lithium–Ion Battery = Chung-Chiun Liu and Xuekun Xing . . . . . . . . . . . . . . . . Loss Prevention in Chemical Processing = Joel M. Haight . . . . . . . . . . . . . . . . . Low-Pressure Cascade Arc Torch = Hirotsugu Yasuda . . . . . . . . . . . . . . . . . . . Lubrication Performance Factors for Chemical Process Plant Machinery = Jim C. Fitch and Drew D. Troyer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mass Transfer Enhancement Because of Flow Instabilities = Vimal Kumar and K. D. P. Nigam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Materials Modeling = Sanat Mohanty, Gregg Caldwell, Manish Jain, and Cristina U. Thomas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Measuring Experimental Quantities Using Simple Fluorescence = W. A. Hollerman, S. W. Allison, S. M. Goedeke, and M. R. Cates . . . . . . . . . . . . . . . . . . . . . Membrane Reactors = Ivo F. J. Vankelecom . . . . . . . . . . . . . . . . . . . . . . . . . . Mesoporous Silica Films = Hugh W. Hillhouse and Brian W. Eggiman . . . . . . . . . Metallocene Catalysts for Olefin Polymerization = T. C. Chung . . . . . . . . . . . . . Microelectronics Fabrication = Edmund G. Seebauer and Charlotte T. M. Kwok . . Microfabrication = Chung-Chiun Liu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Microgravity Processing of Materials = Robert Naumann . . . . . . . . . . . . . . . . . Microreactors and Microreaction Engineering = Richard I. Masel, Scott Gold, and Zheng Ni . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Microscale Fuel Cells = J. D. Holladay and V. V. Viswanathan . . . . . . . . . . . . . Microscopy of Ionomers = Andreas Taubert and Brian P. Kirkmeyer . . . . . . . . . . Microwave Processing of Ceramics = Eldon D. Case . . . . . . . . . . . . . . . . . . . . Mixing and Chemical Reactions = Edward L. Paul, Suzanne M. Kresta, and Arthur W. Etchells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Molecular Bioengineering = Sundararajan V. Madihally . . . . . . . . . . . . . . . . . . Molecular Modeling for Nonequilibrium Chemical Processes = Dionisios G. Vlachos Molecular Self-Assembly = Yoon Seob Lee and James F. Rathman . . . . . . . . . . . Molecularly Imprinted Polymers = Gregory T. Rushton and Ken D. Shimizu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Molten Carbonate Fuel Cells = Prabhu Ganesan, Branko N. Popov, and Rajam Pattabiraman . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Multiphase Mixing and Solid–Liquid Mixing in Agitated Reactors = Piero M. Armenante, Victor Atiemo-Obeng, and W. Roy Penney . . . . . . . . . . Multiphase Reactors = Stanley M. Barnett . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

1367 1381 1401 1411 1427 1445

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

1457 1469 1483 1493

. . . . . . . . . 1511 . . . . . . . . . 1531 . . . . . . . . . 1551 . . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

1561 1575 1587 1599 1615 1627 1633

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

1643 1663 1673 1687

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

1699 1709 1717 1727

. . . . . . . . . 1737 . . . . . . . . . 1747 . . . . . . . . . 1767 . . . . . . . . . 1781

xxvii

Nanoimprint Technology and Its Applications = L. Jay Guo . . . . . . . . . . . . Nanomaterials = David S. J. Arney, Jimmie R. Baran, Allen R. Siedle, and Matthew H. Frey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nanoporous Dielectric Materials = Jorge A. Lubguban and Shubhra Gangopadhyay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nanostructured Materials = Vikrant N. Urade and Hugh W. Hillhouse . . . . . Nanotribology = Jonathan W. Bender and Xiaodong Li . . . . . . . . . . . . . . Natural Gas Hydrates = P. R. Bishnoi and Matthew A. Clarke . . . . . . . . . Natural Gas Utilization = Peter R. Pujado´ . . . . . . . . . . . . . . . . . . . . . . . New Flame Retardant Materials: Nonhalogenated Additives from Brominated Starting Materials and Inherently Low-Flammability Polymers = Alexander B. Morgan, Joshua Jurs, Jason Stephenson, and James M. Tour . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nickel–Cadmium Battery = Chung-Chiun Liu and Xuekun Xing . . . . . . . . . NMR in Chemical Processing = Sangrama K. Sahoo and Peter L. Rinaldi . . NMR Spectroscopy of Polymers in Solution = Sangrama K. Sahoo and Peter L. Rinaldi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . NOx Removal = Mike Bradford and Rajiv Grover . . . . . . . . . . . . . . . . . Numerical Computations for Chemical Process Analysis and Design = David G. Retzloff . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Onsite and Offsite Emergency Preparedness for Chemical Facilities and Chemical Transportation = Michael K. Lindell and Ronald W. Perry . . Oriented Morphologies: Development in Polymer Processing = Mario A. Perez Osmotic Distillation = Bob Johnson . . . . . . . . . . . . . . . . . . . . . . . . . . . Ozone Treatment = Jiangning Wu . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . 1791 . . . . . . . . . . . . . 1803 . . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

1813 1825 1837 1849 1865

. . . . . . . . . . . . . 1879 . . . . . . . . . . . . . 1897 . . . . . . . . . . . . . 1907 . . . . . . . . . . . . . 1919 . . . . . . . . . . . . . 1935 . . . . . . . . . . . . . 1949 . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

1959 1973 1985 1993

Volume 4 Packed Absorption Column Design = Karl B. Schnelle, Jr. and Partha Dey . . . . . . Particle–Particle Interaction: Improvements in the Prediction of DLVO Forces = Anh V. Nguyen, Linh T. T. Tran, and Jan D. Miller . . . . . . . . . . . . . . . . . . Pervaporation: Vapor Permeation = Nicholas Patrick Wynn . . . . . . . . . . . . . . . . Petroleum Refinery Distillation = Stanley Marple . . . . . . . . . . . . . . . . . . . . . . Phase Behavior of Hydrocarbon Mixtures = X.-Y. Zou and J. M. Shaw . . . . . . . . . Phase Equilibria = Karl B. Schnelle, Jr. . . . . . . . . . . . . . . . . . . . . . . . . . . . . Phenolic Resins = Adriane G. Ludwick and Mohamed O. Abdalla . . . . . . . . . . . Photodegradation of Polymers = J. R. White . . . . . . . . . . . . . . . . . . . . . . . . . . Photoresists = Sergei V. Zelentsov and Nadezda V. Zelentsova . . . . . . . . . . . . . Photovoltaic Materials = Richard Corkish . . . . . . . . . . . . . . . . . . . . . . . . . . . Phytoremediation = Joel G. Burken . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pilot Plant and Minipilot Units = Richard P. Palluzi . . . . . . . . . . . . . . . . . . . . Pinch Design and Analysis = Robin Smith and Jin-Kuk Kim . . . . . . . . . . . . . . . Pipeline Safety = Glenn B. DeWolf . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Plant Metabolic Engineering = Eleanore T. Wurtzel and Erich Grotewold . . . . . . . Plasma Etching = David G. Retzloff . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Plasma Polymerization Coating = Hirotsugu K. Yasuda . . . . . . . . . . . . . . . . . . . Pollution Prevention = Ashok Kumar, Harish G. Rao, Abhilash Vijayan, and Charanya Varadarajan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Polyanhydrides = Maria P. Torres, Amy S. Determan, Surya K. Mallapragada, and Balaji Narasimhan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . 2003 . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

2017 2031 2053 2067 2077 2089 2101 2111 2129 2139 2147 2165 2181 2191 2201 2215

. . . . . . . . . 2231 . . . . . . . . . 2247

xxviii

Polybutadiene = William L. Hergenrother, Mark DeDecker, and Dan F. Graves Polycarbonate (PC) = Sarah E. Morgan and Jun Li . . . . . . . . . . . . . . . . . . Polycyclic Aromatic Hydrocarbons (PAHs) = Teresa J. Cutright and Sangchul Hwang . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Polymer Clay Nanocomposites = Hyoung J. Choi, Ji W. Kim, and Myung S. Jhon Polymer Composites = Youssef K. Hamidi, M. Cengiz Altan, and Brian P. Grady . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Polymeric Membranes = Takeshi Matsuura and Mehrdad Rafat . . . . . . . . . . Polymerization Reactions: Modeling, Design, and Control = Kyu Yong Choi . . . . Polysaccharides = Anton Huber . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Polyurethanes = Joanna D. Fromstein and Kimberly A. Woodhouse . . . . . . . . Polyvinylidene Fluoride = Jeffrey H. Yen and Ramin Amin-Sanayei . . . . . . . . Porous Media = Karsten E. Thompson . . . . . . . . . . . . . . . . . . . . . . . . . . . Powder Coating Application Processes = Harry J. Lader . . . . . . . . . . . . . . . . Power Factor = Peter R. Pujado´ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pressure-Relief Valve Design = Jonathan Francis . . . . . . . . . . . . . . . . . . . . Process Optimization = Ralph W. Pike . . . . . . . . . . . . . . . . . . . . . . . . . . . Processing of Pharmaceuticals Using Dense Gas Technologies = R. Mammucari, F. Dehghani, and N. R. Foster . . . . . Propylene Production = Abdullah M. Aitani . . . . . . . . . . . . . . . . . . . . . . . . Protein Design = Zhilei Chen and Huimin Zhao . . . . . . . . . . . . . . . . . . . . . Protein Folding: Biomedical Implications = Ajit Sadana, Tuan Vo-Dinh, and Nigil Satish Jeyashekar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Protein Production in Transgenic Plants: Development and Commercialization = Wayne R. Curtis . . . . . . . . . . . . . . . . . . . . . Proton-Exchange Membrane Fuel Cells = Pyoungho Choi, Pradeep Haldar, and Ravindra Datta . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Reactive Extrusion = Gerard T. Caneba . . . . . . . . . . . . . . . . . . . . . . . . . . Reactive Separation = Vincent G. Gomes . . . . . . . . . . . . . . . . . . . . . . . . . . Reactor Engineering = Ken K. Robinson . . . . . . . . . . . . . . . . . . . . . . . . . . Real-Time Optimization: Status, Issues, and Opportunities = J. F. Forbes, T. E. Marlin, and W. S. Yip . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Recent Advances in Catalytic Distillation = Flora T. T. Ng . . . . . . . . . . . . . . . Recycling of Spent Tires = Roger N. Beers and David A. Benko . . . . . . . . . . . Reformulated Gasoline = A. K. Dalai and D. Ferdous . . . . . . . . . . . . . . . . . Renewable Energy = Gareth P. Harrison . . . . . . . . . . . . . . . . . . . . . . . . . . Reprocessing of Domestic Spent Nuclear Fuel = Truman S. Storvick . . . . . . . . . Resid Conversion = James G. Speight . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . 2259 . . . . . . . . . . . 2277 . . . . . . . . . . . 2291 . . . . . . . . . . . 2301 . . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

2313 2323 2335 2349 2369 2379 2391 2405 2417 2423 2439

. . . . . . . . . . . 2451 . . . . . . . . . . . 2461 . . . . . . . . . . . 2467 . . . . . . . . . . . 2479 . . . . . . . . . . . 2489 . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

2501 2531 2541 2557

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

2585 2599 2613 2625 2635 2647 2655

Volume 5 Rheology of Cellulose Liquid Crystalline Polymers = Qizhou Dai, Richard Gilbert, and John F. Kadla . . . . . . . . . . . . . . . . . . Rotational Molding of Polymers = Ce´line T. Bellehumeur . . . . . . . Rubber Devulcanization = David A. Benko and Roger N. Beers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Scrubbers = S. Komar Kawatra . . . . . . . . . . . . . . . . . . . . . . . . Six Sigma Design: An Overview of Design for Six Sigma (DFSS) = Sean A. Curran, Kwok-Wai Lem, Steve Sund, and Mina Gabriel

. . . . . . . . . . . . . . . . . . . 2663 . . . . . . . . . . . . . . . . . . . 2677 . . . . . . . . . . . . . . . . . . . 2691 . . . . . . . . . . . . . . . . . . . 2701 . . . . . . . . . . . . . . . . . . . 2719

xxix

Size Reduction = Sunil Kesavan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Soave’s Modified Redlich–Kwong Equation of State = J. Richard Elliott, Jr. . . . . . . . . . Solid–Liquid Mixing: Numerical Simulation and Physical Experiments = Philippe A. Tanguy, Francis Thibault, Gabriel Ascanio, and Edmundo Brito-De La Fuente . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Solid–Liquid Separation = Frank M. Tiller and Wenping Li . . . . . . . . . . . . . . . . . . . . Solvent Refining Processes = Roland H. Heck . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Solvents = Satish C. Mohapatra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sonochemical Reaction Engineering = David A. Bruce and Amarnath Nareddy . . . . . . . Sorbent Technology = Shuguang Deng . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Spinning Disk Reactor = R. J. J. Jachuck and J. R. Burns . . . . . . . . . . . . . . . . . . . . . Styrene = Guy B. Woodle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Styrene–Butadiene Rubber = Jing Peng . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Superabsorbents = Takamasa Nonaka . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Supercritical CO2-Assisted Surface Coating Injection Molding = Masahiro Ohshima . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Supercritical Fluid Extraction (SFE) = Ram B. Gupta . . . . . . . . . . . . . . . . . . . . . . . Supercritical Fluid Technology: Reactions = Aydin K. Sunol, Sermin G. Sunol, and Naveed Aslam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Supercritical Water Oxidation = Ram B. Gupta . . . . . . . . . . . . . . . . . . . . . . . . . . . Synthesis Gas = Kim Aasberg-Petersen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Tar Sand = James G. Speight . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Theoretical Aspects of Liquid Crystals and Liquid Crystalline Polymers = James J. Feng . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Thermal Analysis Techniques: Overview = John O. Hill . . . . . . . . . . . . . . . . . . . . . . . Thermal Cracking of Hydrocarbons = Tom Chunghu Tsai and Lyle F. Albright . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Thermal Desorption = Timothy P. Sullivan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Thermal Stability of Chemical Reactors = Haishan Zheng and Jason M. Keith . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Thermogravimetric Analysis = John O. Hill . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Thermomechanical Analysis = Kevin P. Menard . . . . . . . . . . . . . . . . . . . . . . . . . . . Thermosets: Materials, Processes, and Waste Minimization = Kwok-Wai Lem, Sean A. Curran, Steve Sund, and Mina Gabriel . . . . . . . . . . . . . . . . . . . . . . . . . Thin Film Processes in MEMS and NEMS Technologies = W. R. Ashurst, C. Carraro and R. Maboudian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Thin Film Science and Technology = T. L. Alford, J. Kouvetakis, and J. W. Mayer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Thin Liquid Film Deposition = Myung S. Jhon and Thomas E. Karis . . . . . . . . . . . . . . Thiochemicals: Mercaptans, Sulfides, and Polysulfides = Jeffrey Yen, Vijay R. Srinivas, and Gary S. Smith . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Thiochemicals: Mercapto Acids and Organosulfur (IV/VI) Compounds = Jeffrey H. Yen, Gary S. Smith, and Vijay R. Srinivas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Tissue Engineering = Shang-Tian Yang and Clayt Robinson . . . . . . . . . . . . . . . . . . . Trace Elements = Ian D. Brindle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Transmission Electron Microscopy for Materials Science = Rolf Erni and Nigel D. Browning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Tubular Reactors: Reactor Types and Selected Process Applications = Patrick L. Mills and Joseph M. Lambert, Jr. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Twin-Screw Extrusion = Paul Andersen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . 2735 . . . . . 2747

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

2753 2769 2791 2799 2811 2825 2847 2859 2871 2881

. . . . . 2897 . . . . . 2907 . . . .

. . . .

. . . .

. . . .

. . . .

2915 2927 2933 2947

. . . . . 2955 . . . . . 2965 . . . . . 2975 . . . . . 2987 . . . . . 2997 . . . . . 3009 . . . . . 3023 . . . . . 3031 . . . . . 3049 . . . . . 3061 . . . . . 3075 . . . . . 3089 . . . . . 3101 . . . . . 3115 . . . . . 3129 . . . . . 3139 . . . . . 3151 . . . . . 3167

xxx

Use of Lipases to Isolate Polyunsaturated and Oxygenated Fatty Acids and Form Value-Added Ester Products = Douglas G. Hayes . . . . . . . . . . . . . Vapor–Liquid–Solid Synthesis of Nanowires = Brian A. Korgel, Tobias Hanrath, and Forrest M. Davidson, III . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Water Gas Shift Reaction = Wolfgang Ruettinger and Oleg Ilinich . . . . . . . . . . . . Water Reclamation = Mark A. Kuehne, Norman N. Li, Richard Q. Song, Maxwell Tsai, and Jane C. Li . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Wide Band-Gap Electronics Materials = Mark A. Prelas and Krishnendu Saha . . . . Zeolite Membranes = Yushan Yan, Zijian Li, Shuang Li, and Christopher Lew . . . . Ziegler–Natta Catalysis = John C. Chadwick . . . . . . . . . . . . . . . . . . . . . . . . . . Index

. . . . . . . . 3179 . . . . . . . . 3191 . . . . . . . . 3205 . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

3217 3227 3237 3247

Topical Table of Contents

Advanced Materials Carbon Nanotubes = Morinobu Endo, Yoong-Ahm Kim, Takuya Hayashi, Kenji Takeuchi, Maruicio Terrones, and Mildred S. Dresselhaus . . . . . . . . . . Ceramics = Stephen J. Lombardo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conductive Polymers = Ronald W. Gumbs . . . . . . . . . . . . . . . . . . . . . . . . . . . Diamond Films = Angel Velez and Mark A. Prelas . . . . . . . . . . . . . . . . . . . . . Diamond-Like Carbon Films = Angel Velez and Mark A. Prelas . . . . . . . . . . . . . Electrodeposition = Andre´ Avelino Pasa and Maximiliano Luis Munford . . . . . . . Friction Materials = Sunil Kesavan and Xinming Shao . . . . . . . . . . . . . . . . . . Fuel Cell Membranes = Andrew M. Herring . . . . . . . . . . . . . . . . . . . . . . . . . . Functional Biomaterials = Chun Wang . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Heavy Water (Deuterium Oxide) = Sharad M. Dave . . . . . . . . . . . . . . . . . . . . Hollow Fiber Technology = Vicki Chen and Pierre Le-Clech . . . . . . . . . . . . . . . Hybrid Materials (Organic–Inorganic) = C. Sanchez and G. J. A. A. Soler-Illia . . . Hydrogels = Jae Hyung Park, Kang Moo Huh, Mingli Ye, and Kinam Park . . . . . Ion Exchange Resin = Sukalyan Sengupta and Arup K. Sengupta . . . . . . . . . . . . Materials Modeling = Sanat Mohanty, Gregg Caldwell, Manish Jain, and Cristina U. Thomas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Membrane Reactors = Ivo F. J. Vankelecom . . . . . . . . . . . . . . . . . . . . . . . . . . Mesoporous Silica Films = Brian W. Eggiman and Hugh W. Hillhouse . . . . . . . . . Molecularly Imprinted Polymers = Gregory T. Rushton and Ken D. Shimizu . . . . . Nanostructured Materials = Vikrant N. Urade and Hugh W. Hillhouse . . . . . . . . . New Flame Retardant Materials: Nonhalogenated Additives from Brominated Starting Materials and Inherently Low-Flammability Polymers = Alexander B. Morgan, Joshua Jurs, Jason Stephenson, and James M. Tour . . . Photovoltaic Materials = Richard Corkish . . . . . . . . . . . . . . . . . . . . . . . . . . . Polyanhydrides = Maria P. Torres, Amy S. Determan, Surya K. Mallapragada, and Balaji Narasimhan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Porous Media = Karsten E. Thompson . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Superabsorbents = Takamasa Nonaka . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Theoretical Aspects of Liquid Crystals and Liquid Crystalline Polymers = James J. Feng . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Thin Film Processes in MEMS and NEMS Technologies = W. R. Ashurst, C. Carraro, and R. Maboudian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Thin Film Science and Technology = T. L. Alford, J. Kouvetakis, and J. W. Mayer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Thin Liquid Film Deposition = Myung S. Jhon and Thomas E. Karis . . . . . . . . . . Wide Band-Gap Electronics Materials = Mark A. Prelas and Krishnendu Saha . . . Zeolite Membranes = Yushan Yan, Zijian Li, Shuang Li, and Christopher Lew . . .

xxxi

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

333 417 527 685 695 821 1071 1085 1099 1221 1253 1267 1307 1427

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

1551 1575 1587 1737 1825

. . . . . . . . . 1879 . . . . . . . . . 2129 . . . . . . . . . 2247 . . . . . . . . . 2391 . . . . . . . . . 2881 . . . . . . . . . 2955 . . . . . . . . . 3049 . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

3061 3075 3227 3237

xxxii

Analytical Methods and Characterization Branching Level Detection in Polymers = M. J. Scorah, R. Dhib, and A. Penlidis . . Chromatographic Separations = Scott M. Husson . . . . . . . . . . . . . . . . . . . . . . . Contact Angles, Surface Tension, and Capillarity = Peter R. Pujado´ . . . . . . . . . . Differential Scanning Calorimetry = John O. Hill . . . . . . . . . . . . . . . . . . . . . . Dynamic Mechanical Thermal Analysis = Kevin P. Menard . . . . . . . . . . . . . . . . Fluorescent Coatings for High Temperature Phosphor Thermometry = S. W. Allison, W. A. Hollerman, S. M. Goedeke, M. R. Cates, and T. J. Bencic . . . . . . . . . . Measuring Experimental Quantities Using Simple Fluorescence = W. A. Hollerman, S. W. Allison, S. M. Goedeke, and M. R. Cates . . . . . . . . . Microscopy of Ionomers = Andreas Taubert and Brian P. Kirkmeyer . . . . . . . . . . NMR in Chemical Processing = Sangrama K. Sahoo and Peter L. Rinaldi . . . . . . NMR Spectroscopy of Polymers in Solution = Sangrama K. Sahoo and Peter L. Rinaldi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Thermal Analysis Techniques: Overview = John O. Hill . . . . . . . . . . . . . . . . . . . Thermogravimetric Analysis = John O. Hill . . . . . . . . . . . . . . . . . . . . . . . . . . Thermomechanical Analysis = Kevin P. Menard . . . . . . . . . . . . . . . . . . . . . . . Transmission Electron Microscopy for Materials Science = Rolf Erni and Nigel D. Browning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

251 481 539 699 799

. . . . . . . . . 1021 . . . . . . . . . 1561 . . . . . . . . . 1673 . . . . . . . . . 1907 . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

1919 2965 3009 3023

. . . . . . . . . 3139

Batteries and Fuel Cells Alkaline Zn–MnO2 Batteries = Chung-Chiun Liu and Xuekun Xing . . . . . . . . Fuel Cell Membranes = Andrew M. Herring . . . . . . . . . . . . . . . . . . . . . . . . Lithium–Ion Battery = Chung-Chiun Liu and Xuekun Xing . . . . . . . . . . . . . . Microscale Fuel Cells = J. D. Holladay and V. V. Viswanathan . . . . . . . . . . . Molten Carbonate Fuel Cells = Prabhu Ganesan, Branko N. Popov, and Rajam Pattabiraman . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nickel–Cadmium Battery = Chung-Chiun Liu and Xuekun Xing . . . . . . . . . . . Proton-Exchange Membrane Fuel Cells = Pyoungho Choi, Pradeep Haldar, and Ravindra Datta . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. 51 . 1085 . 1469 . 1663

. . . . . . . . . . . 1747 . . . . . . . . . . . 1897 . . . . . . . . . . . 2501

Biomaterials, Materials for Biological and Biomedical Applications Biomaterials = Sujata K. Bhatia and Surita R. Bhatia . . . . . . . . . . . . . . . . . . BioMEMS = L. James Lee . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Biomolecular Engineering = Zengyi Shao, Ee Lui Ang, and Huimin Zhao . . . . . . Chiral Drug Separation = Bingyun Li and Donald T. Haynie . . . . . . . . . . . . . . Detergent Enzymes = Michael R. Stoner, Douglas A. Dale, Alfred Gaertner, and Theodore W. Randolph . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Functional Biomaterials = Chun Wang . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hydrogels = Jae Hyung Park, Kang Moo Huh, Mingli Ye, and Kinam Park . . . . Hydrophilic Polymers for Biomedical Applications = Frank Davis and Se´amus P. J. Higson . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Materials Modeling = Sanat Mohanty, Gregg Caldwell, Manish Jain, and Cristina U. Thomas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Molecular Bioengineering = Sundararajan V. Madihally . . . . . . . . . . . . . . . . . Molecularly Imprinted Polymers = Gregory T. Rushton and Ken D. Shimizu . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

153 161 171 449

. . . . . . . . . . 673 . . . . . . . . . . 1099 . . . . . . . . . . 1307 . . . . . . . . . . 1349 . . . . . . . . . . 1551 . . . . . . . . . . 1709 . . . . . . . . . . 1737

xxxiii

Polyanhydrides = Maria P. Torres, Amy S. Determan, Surya K. Mallapragada, and Balaji Narasimhan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Processing of Pharmaceuticals Using Dense Gas Technologies = Raffaella Mammucari, Fariba Dehghani, and Neil R. Foster . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Protein Folding: Biomedical Implications = Ajit Sadana, Tuan Vo-Dinh, and Nigil Satish Jeyashekar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Protein Production in Transgenic Plants: Development and Commercialization = Wayne R. Curtis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Tissue Engineering = Shang-Tian Yang and Clayt Robinson . . . . . . . . . . . . . . . . . .

. . . . . . 2247 . . . . . . 2451 . . . . . . 2479 . . . . . . 2489 . . . . . . 3115

Biotechnology and Biological Processing Animal Cell Culture = Shang-Tian Yang and Shubhayu Basu . . . . . . . . . . . . . . Biocatalysis = Tyler Johannes, Michael R. Simurdiak, and Huimin Zhao . . . . . . Biofilms = T. Reg. Bott . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Biofuels and Bioenergy = Dinesh Gera . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bioinformatics and Modeling Biological Systems = Shwetal Patel and Jeremy S. Edwards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Biomass to Ethanol = Randy S. Lewis, Rohit P. Datar, and Raymond L. Huhnke . Biomolecular Engineering = Zengyi Shao, Ee Lui Ang, and Huimin Zhao . . . . . . Bioprocess and Chemical Process Intensification = G. Akay . . . . . . . . . . . . . . Bioprocessing = Ryan G. Soderquist and James M. Lee . . . . . . . . . . . . . . . . . Bioremediation = Teresa J. Cutright . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bioseparations = Shubhayu Basu and Shang-Tian Yang . . . . . . . . . . . . . . . . . Chiral Drug Separation = Bingyun Li and Donald T. Haynie . . . . . . . . . . . . . . Detergent Enzymes = Michael R. Stoner, Douglas A. Dale, Alfred Gaertner, and Theodore W. Randolph . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fermentation Processes = Kimberly Ogden . . . . . . . . . . . . . . . . . . . . . . . . . Fermenter Design = Kishore K. Kar, Juergen Lueske, and Richard F. Cope . . . . Immobilized Enzyme Technology = Charles G. Hill, Jr., Cristina Otero, and Hugo S. Garcia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Molecular Bioengineering = Sundararajan V. Madihally . . . . . . . . . . . . . . . . . Molecular Self-Assembly = Yoon Seob Lee and James F. Rathman . . . . . . . . . . Plant Metabolic Engineering = Eleanore T. Wurtzel and Erich Grotewold . . . . . . Polysaccharides = Anton Huber . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Protein Design = Zhilei Chen and Huimin Zhao . . . . . . . . . . . . . . . . . . . . . . Protein Folding: Biomedical Implications = Ajit Sadana, Tuan Vo-Dinh, and Nigil Satish Jeyashekar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Protein Production in Transgenic Plants: Development and Commercialization = Wayne R. Curtis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Tissue Engineering = Shang-Tian Yang and Clayt Robinson . . . . . . . . . . . . . . Trace Elements = Ian D. Brindle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Use of Lipases to Isolate Polyunsaturated and Oxygenated Fatty Acids and Form Value-Added Ester Products = Douglas G. Hayes . . . . . . . . . . . . . . . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. 67 . 101 . 111 . 121

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

131 143 171 183 199 207 221 449

. . . . . . . . . . 673 . . . . . . . . . . 941 . . . . . . . . . . 951 . . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

1367 1709 1727 2191 2349 2467

. . . . . . . . . . 2479 . . . . . . . . . . 2489 . . . . . . . . . . 3115 . . . . . . . . . . 3129 . . . . . . . . . . 3179

Catalysis and Catalyst Preparation Alkylation Processes to Produce High-Quality Gasolines = Lyle F. Albright and James M. Ryan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 Biocatalysis = Tyler Johannes, Michael R. Simurdiak, and Huimin Zhao . . . . . . . . . . . . . . . . 101

xxxiv

Catalysis and Catalyst Preparation (cont’d) Catalyst Preparation = X. D. Hu and Michael W. Balakos . . . . . . . . . . . . . . . . Catalytic Combustion for Thermal Energy Generation = Daniel G. Lo¨ffler . . . . . . Catalytic Cracking = Paul O’Connor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Catalytic Dehydrogenation = Bipin V. Vora and Peter R. Pujado´ . . . . . . . . . . . . Catalytic Naphtha Reforming = Abdullah M. Aitani . . . . . . . . . . . . . . . . . . . . Dimethyl Ether = Abhay Sardesai . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Heterogeneous Catalysis = Richard W. Rice and James G. Goodwin, Jr. . . . . . . . Hydrocracking = James G. Speight . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hydrodesulfurization = James G. Speight . . . . . . . . . . . . . . . . . . . . . . . . . . . Hydrogenation = Xiaobo Yang . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hydrogenation Reactions in Dense Gas Systems = Gary Combes, Fariba Dehghani, Raffaella Mammucari, and Neil R. Foster . . . . . . . . . . . . . . . . . . . . . . . . Hydrotreating Catalysts and Processes: Current Status and Path Forward = Arunabha Kundu, Nishith Dwivedi, Azad Singh, and K. D. P. Nigam . . . . . . . Metallocene Catalysts for Olefin Polymerization = T. C. Chung . . . . . . . . . . . . . Recent Advances in Catalytic Distillation = Flora T. T. Ng . . . . . . . . . . . . . . . . . Resid Conversion = James G. Speight . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Water Gas Shift Reaction = Wolfgang Ruettinger and Oleg Ilinich . . . . . . . . . . . Ziegler–Natta Catalysis = John C. Chadwick . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

345 361 371 379 397 707 1235 1281 1289 1325

. . . . . . . . . 1337 . . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

1357 1599 2599 2655 3205 3247

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

821 833 839 849

Electrochemical Processes and Products Electrodeposition = Andre´ Avelino Pasa and Maximiliano Luis Munford Electronic Chemical Sensors = Chung-Chiun Liu . . . . . . . . . . . . . . . . Electroplating = Helen H. Lou and Yinlun Huang . . . . . . . . . . . . . . . Electrostatic Precipitation = Kenneth R. Parker . . . . . . . . . . . . . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

Energy Processing and Renewable Energy Biofuels and Bioenergy = Dinesh Gera . . . . . . . . . . . . . . . . . . . . . . . . . . . . Biomass to Ethanol = Randy S. Lewis, Rohit P. Datar, and Raymond L. Huhnke . Capsule Pipeline = Henry Liu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Carbon Dioxide Capture and Disposal: Carbon Sequestration = K. S. Lackner, A.-H. Park, and L.-S. Fan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . CHP Technology/Systems = Louay M. Chamra and B. K. Hodge . . . . . . . . . . . Coal–Water Slurries = S. Komar Kawatra . . . . . . . . . . . . . . . . . . . . . . . . . . Geothermal Energy = Sunggyu Lee and H. Bryan Lanterman . . . . . . . . . . . . . Natural Gas Hydrates = P. R. Bishnoi and Matthew A. Clarke . . . . . . . . . . . . Natural Gas Utilization = Peter R. Pujado´ . . . . . . . . . . . . . . . . . . . . . . . . . . Photovoltaic Materials = Richard Corkish . . . . . . . . . . . . . . . . . . . . . . . . . . Renewable Energy = Gareth P. Harrison . . . . . . . . . . . . . . . . . . . . . . . . . . . Reprocessing of Domestic Spent Nuclear Fuel = Truman S. Storvick . . . . . . . . . . Resid Conversion = James G. Speight . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Synthesis Gas = Kim Aasberg-Petersen . . . . . . . . . . . . . . . . . . . . . . . . . . . . Tar Sand = James G. Speight . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . 121 . . . . . . . . . . 143 . . . . . . . . . . 295 . . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

305 469 495 1175 1849 1865 2129 2635 2647 2655 2933 2947

xxxv

Environmental Technology and Regulations Activated Sludge Process = Shankha K. Banerji . . . . . . . . . . . . . . . . . . . . . . . . . Advanced Oxidation = Sangchul Hwang . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bioremediation = Teresa J. Cutright . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Carbon Dioxide Capture and Disposal: Carbon Sequestration = K. S. Lackner, A.-H. Park, and L.-S. Fan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Carbon Fibers from Lignin-Recyclable Plastic Blends = Satoshi Kubo and John F. Kadla . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Catalytic Combustion for Thermal Energy Generation = Daniel G. Lo¨ffler . . . . . . . . Chlorofluorocarbons = Byung Gwon Lee and Youn-Woo Lee . . . . . . . . . . . . . . . . . Denitrogenation = Daeik Kim and Teh Fu Yen . . . . . . . . . . . . . . . . . . . . . . . . . . Desulfurization = Chunshan Song, Uday T. Turaga, and Xiaoliang Ma . . . . . . . . . . Dust Explosion Hazard Assessment and Control = Vahid Ebadat . . . . . . . . . . . . . . Electrostatic Precipitation = Kenneth R. Parker . . . . . . . . . . . . . . . . . . . . . . . . . Environmental Chemodynamics = Louis J. Thibodeaux . . . . . . . . . . . . . . . . . . . . . Environmental Law and Policy = Don C. Haddox and Teresa J. Cutright . . . . . . . . . Gas Explosion Hazard: Prevention and Protection = Dehong Kong . . . . . . . . . . . . . Greenhouse Gas Management for Multiplant Complexes = Ralph W. Pike . . . . . . . . . Incineration and Combustion = Selim M. Senkan . . . . . . . . . . . . . . . . . . . . . . . . . NOx Removal = Mike Bradford and Rajiv Grover . . . . . . . . . . . . . . . . . . . . . . . Ozone Treatment = Jiangning Wu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Photodegradation of Polymers = J. R. White . . . . . . . . . . . . . . . . . . . . . . . . . . . . Phytoremediation = Joel G. Burken . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pollution Prevention = Ashok Kumar, Harish G. Rao, Abhilash Vijayan, and Charanya Varadarajan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Polycyclic Aromatic Hydrocarbons (PAHs) = Teresa J. Cutright and Sangchul Hwang Recycling of Spent Tires = Roger N. Beers and David A. Benko . . . . . . . . . . . . . . . Rubber Devulcanization = David A. Benko and Roger N. Beers . . . . . . . . . . . . . . . Scrubbers = S. Komar Kawatra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sorbent Technology = Shuguang Deng . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Thermosets: Materials, Processes, and Waste Minimization = Kwok-Wai Lem, Sean A. Curran, Steve Sund, and Mina Gabriel . . . . . . . . . . . . . . . . . . . . . . . Trace Elements = Ian D. Brindle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Water Reclamation = Mark A. Kuehne, Norman N. Li, Richard Q. Song, Maxwell Tsai, and Jane C. Li . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . 11 . . . . . . . 41 . . . . . . . 207 . . . . . . . 305 . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

317 361 459 627 651 787 849 891 899 1109 1189 1381 1935 1993 2101 2139

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

2231 2291 2613 2691 2701 2825

. . . . . . . 3031 . . . . . . . 3129 . . . . . . . 3217

Mass Transfer and Mixing Gas–Liquid Contactors = Kishore K. Kar, Richard F. Cope, and Juergen Lueske Gas–Liquid Mixing in Agitated Reactors = John C. Middleton, John M. Smith, and Piero M. Armenante . . . . . . . . . . . . . . . . . . . . . . . Gas-to-Liquid Mass Transfer = Huu D. Doan, Simant R. Upreti, and Ali Lohi . Liquid–Liquid Mixing in Agitated Reactors = Richard V. Calabrese, Douglas E. Leng, and Piero M. Armenante . . . . . . . . . . . . . . . . . . . . . . Mass Transfer Enhancement Because of Flow Instabilities = Vimal Kumar and K. D. P. Nigam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mixing and Chemical Reactions = Edward L. Paul, Suzanne M. Kresta, and Arthur W. Etchells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . 1119 . . . . . . . . . . . 1131 . . . . . . . . . . . 1163 . . . . . . . . . . . 1457 . . . . . . . . . . . 1531 . . . . . . . . . . . 1699

xxxvi

Mass Transfer and Mixing (cont’d) Multiphase Mixing and Solid–Liquid Mixing in Agitated Reactors = Piero M. Armenante, Victor Atiemo-Obeng, and W. Roy Penney . . . . . . . . . . . . . . . . . . . 1767 Multiphase Reactors = Stanley M. Barnett . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1781 Solid–Liquid Mixing: Numerical Simulation and Physical Experiments = Philippe A. Tanguy, Francis Thibault, Gabriel Ascanio, and Edmundo Brito-De La Fuente . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2753

Materials Application and Processing BioMEMS = L. James Lee . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ceramics = Stephen J. Lombardo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chemical Mechanical Planarization in Integrated Circuit Manufacturing = John Zabasajja . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chemical Vapor Deposition = David G. Retzloff . . . . . . . . . . . . . . . . . . . . . Crystal Growth = C. W. Lan, W. C. Yu, and W. C. Hsu . . . . . . . . . . . . . . . . . . Gas-Phase Lubrication of MEMS Devices: Using Alcohol Vapor Adsorption Isotherm for Lubrication of Silicon Oxides = Kenneth Strawhecker, David B. Asay, and Seong H. Kim . . . . . . . . . . . . . . . . . . . . . . . . . . . . Microelectronics Fabrication = Edmund G. Seebauer and Charlotte T. M. Kwok . Microfabrication = Chung-Chiun Liu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Microgravity Processing of Materials = Robert Naumann . . . . . . . . . . . . . . . . Microwave Processing of Ceramics = Eldon D. Case . . . . . . . . . . . . . . . . . . . Photoresists = Sergei V. Zelentsov and Nadezda V. Zelentsova . . . . . . . . . . . . Plasma Etching = David G. Retzloff . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . 161 . . . . . . . . . . 417 . . . . . . . . . . 429 . . . . . . . . . . 441 . . . . . . . . . . 589

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

1143 1615 1627 1633 1687 2111 2201

Nanotechnology Carbon Nanotubes = Morinobu Endo, Yoong-Ahm Kim, Takuya Hayashi, Kenji Takeuchi, Maruicio Terrones, and Mildred S. Dresselhaus . . . . . . . . . . . Molecular Self-Assembly = Yoon Seob Lee and James F. Rathman . . . . . . . . . . . . Nanoimprint Technology = L. Jay Guo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nanomaterials = David S. J. Arney, Jimmie R. Baran, Allen R. Siedle, and Matthew H. Frey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nanoporous Dielectric Materials = Jorge A. Lubguban and Shubhra Gangopadhyay . Nanostructured Materials = Vikrant N. Urade and Hugh W. Hillhouse . . . . . . . . . . Nanotribology = Jonathan W. Bender and Xiaodong Li . . . . . . . . . . . . . . . . . . . Photoresists = Sergei V. Zelentsov and Nadezda V. Zelentsova . . . . . . . . . . . . . . Polymer Clay Nanocomposites = Hyoung J. Choi, Ji W. Kim, and Myung S. Jhon . . . Vapor–Liquid–Solid Synthesis of Nanowires = Brian A. Korgel, Tobias Hanrath, and Forrest M. Davidson, III . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . 333 . . . . . . . . 1727 . . . . . . . . 1791 . . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

1803 1813 1825 1837 2111 2301

. . . . . . . . 3191

Particle Technology Coal–Water Slurries = S. Komar Kawatra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 495 Particle–Particle Interaction: Improvements in the Prediction of DLVO Forces = Anh V. Nguyen, Linh T. T. Tran, and Jan D. Miller . . . . . . . . . . . . . . . . . . . . . . . . . . . 2017

xxxvii

Petrochemicals and Petrochemical Processing Cumene Production = Robert J. Schmidt . . . . . . . . . . . . . . . . . . . . Detergent Alkylate = Bipin Vora, Andrea Bozzano, and Stephen Sohn Dimethyl Ether = Abhay Sardesai . . . . . . . . . . . . . . . . . . . . . . . . Dimethylcarbonate = Byung Gwon Lee . . . . . . . . . . . . . . . . . . . . . Ethylbenzene = Guy B. Woodle . . . . . . . . . . . . . . . . . . . . . . . . . . Propylene Production = Abdullah M. Aitani . . . . . . . . . . . . . . . . . . Solvents = Satish C. Mohapatra . . . . . . . . . . . . . . . . . . . . . . . . . Styrene = Guy B. Woodle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Thiochemicals: Mercaptans, Sulfides, and Polysulfides = Jeffrey Yen, Vijay R. Srinivas, and Gary S. Smith . . . . . . . . . . . . . . . . . . . . Thiochemicals: Mercapto Acids and Organosulfur (IV/VI) Compounds = Jeffrey H. Yen, Gary S. Smith, and Vijay R. Srinivas . . . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

603 663 707 719 929 2461 2799 2859

. . . . . . . . . . . . . . . . . 3089 . . . . . . . . . . . . . . . . . 3101

Petroleum and Fuel Processing Alkylation Processes to Produce High-Quality Gasolines = Lyle F. Albright and James M. Ryan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Catalytic Cracking = Paul O’Connor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Catalytic Dehydrogenation = Bipin V. Vora and Peter R. Pujado´ . . . . . . . . . . . . . . Catalytic Naphtha Reforming = Abdullah M. Aitani . . . . . . . . . . . . . . . . . . . . . . Denitrogenation = Daeik Kim and Teh Fu Yen . . . . . . . . . . . . . . . . . . . . . . . . . . Desulfurization = Chunshan Song, Uday T. Turaga, and Xiaoliang Ma . . . . . . . . . . Enhanced Oil Recovery = Kishore K. Mohanty and Gerard T. Caneba . . . . . . . . . . . Hydrocracking = James G. Speight . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hydrodesulfurization = James G. Speight . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hydrogenation = Xiaobo Yang . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hydrotreating Catalysts and Processes: Current Status and Path Forward = Arunabha Kundu, Nishith Dwivedi, Azad Singh, and K. D. P. Nigam . . . . . . . . . Natural Gas Utilization = Peter R. Pujado´ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Petroleum Refinery Distillation = Stanley Marple . . . . . . . . . . . . . . . . . . . . . . . . Polycyclic Aromatic Hydrocarbons (PAHs) = Teresa J. Cutright and Sangchul Hwang Propylene Production = Abdullah M. Aitani . . . . . . . . . . . . . . . . . . . . . . . . . . . . Reformulated Gasoline = A. K. Dalai and D. Ferdous . . . . . . . . . . . . . . . . . . . . . Resid Conversion = James G. Speight . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Solvent Refining Processes = Roland H. Heck . . . . . . . . . . . . . . . . . . . . . . . . . . . Synthesis Gas = Kim Aasberg-Petersen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Tar Sand = James G. Speight . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Thermal Cracking of Hydrocarbons = Tom Chunghu Tsai and Lyle F. Albright . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

57 371 379 397 627 651 881 1281 1289 1325

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

1357 1865 2053 2291 2461 2625 2655 2791 2933 2947 2975

Pipeline Technology Capsule Pipeline = Henry Liu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295 Pipeline Safety = Glenn B. DeWolf . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2181

Plasma Technology Low-Pressure Cascade Arc Torch = Hirotsugu Yasuda . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1493 Plasma Etching = David G. Retzloff . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2201 Plasma Polymerization Coating = Hirotsugu K. Yasuda . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2215

xxxviii

Polymer Processing Blowing Agent = Kyung W. Suh . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bulk Molding and Sheet Molding Compounds = Sanjeev N. Rao and Krishnan Jayaraman . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cross-Linked Polyethylene = Carosena Meola, Giovanni Maria Carlomagno, and Giuseppe Giorleo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Design of Extrusion Dies = Milivoje M. Kostic and Louis G. Reifschneider . . . . . Drag Reducing Agents = Jacques L. Zakin, Ying Zhang, and Yunying Qi . . . . . . . Injection Molding = David O. Kazmer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Latex Processing = Alexander F. Routh . . . . . . . . . . . . . . . . . . . . . . . . . . . . Oriented Morphologies: Development in Polymer Processing = Mario A. Perez . . . . Polymer Clay Nanocomposites = Hyoung J. Choi, Ji W. Kim, and Myung S. Jhon . . Polymer Composites = Youssef K. Hamidi, M. Cengiz Altan, and Brian P. Grady . Powder Coating Application Processes = Harry J. Lader . . . . . . . . . . . . . . . . . . Reactive Extrusion = Gerard T. Caneba . . . . . . . . . . . . . . . . . . . . . . . . . . . . Rotational Molding of Polymers = Ce´line T. Bellehumeur . . . . . . . . . . . . . . . . . Rubber Devulcanization = David A. Benko and Roger N. Beers . . . . . . . . . . . . . Supercritical CO2-Assisted Surface Coating Injection Molding = Masahiro Ohshima Twin-Screw Extrusion = Paul Andersen . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . 237 . . . . . . . . . 283 . . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

577 633 767 1401 1445 1973 2301 2313 2405 2531 2677 2691 2897 3167

Polymeric Materials and Polymerization Branching Level Detection in Polymers = M. J. Scorah, R. Dhib, and A. Penlidis . Bulk Molding and Sheet Molding Compounds = Sanjeev N. Rao and Krishnan Jayaraman . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Carbon Fibers from Lignin-Recyclable Plastic Blends = Satoshi Kubo and John F. Kadla . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conductive Polymers = Ronald W. Gumbs . . . . . . . . . . . . . . . . . . . . . . . . . . Cross-Linked Polyethylene = Carosena Meola, Giovanni Maria Carlomagno, and Giuseppe Giorleo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Emulsion Polymerization = Vincent G. Gomes . . . . . . . . . . . . . . . . . . . . . . . Epoxy Resins = Ian Hamerton . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fluoropolymers = Sina Ebnesajjad . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Free-Radical Polymerization = Yadunandan Lal Dar, Rajeev Farwaha, and Gerard T. Caneba . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fuel Cell Membranes = Andrew M. Herring . . . . . . . . . . . . . . . . . . . . . . . . . Hydrogels = Jae Hyung Park, Kang Moo Huh, Mingli Ye, and Kinam Park . . . . Hydrophilic Polymers for Biomedical Applications = Frank Davis and Se´amus P. J. Higson . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ion Exchange Resin = Sukalyan Sengupta and Arup K. Sengupta . . . . . . . . . . . Latex Processing = Alexander F. Routh . . . . . . . . . . . . . . . . . . . . . . . . . . . Metallocene Catalysts for Olefin Polymerization = T. C. Chung . . . . . . . . . . . . Microscopy of Ionomers = Andreas Taubert and Brian P. Kirkmeyer . . . . . . . . . Molecularly Imprinted Polymers = Gregory T. Rushton and Ken D. Shimizu . . . . New Flame Retardant Materials: Nonhalogenated Additives from Brominated Starting Materials and Inherently Low-Flammability Polymers = Alexander B. Morgan, Joshua Jurs, Jason Stephenson, and James M. Tour . . NMR Spectroscopy of Polymers in Solution = Sangrama K. Sahoo and Peter L. Rinaldi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . 251 . . . . . . . . . . 283 . . . . . . . . . . 317 . . . . . . . . . . 527 . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

577 863 911 1031

. . . . . . . . . . 1057 . . . . . . . . . . 1085 . . . . . . . . . . 1307 . . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

1349 1427 1445 1599 1673 1737

. . . . . . . . . . 1879 . . . . . . . . . . 1919

xxxix

Oriented Morphologies: Development in Polymer Processing = Mario A. Perez . . . . . . . Phenolic Resins = Adriane G. Ludwick and Mohamed O. Abdalla . . . . . . . . . . . . . . Photodegradation of Polymers = J. R. White . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Plasma Polymerization Coating = Hirotsugu K. Yasuda . . . . . . . . . . . . . . . . . . . . . . Polyanhydrides = Maria P. Torres, Amy S. Determan, Surya K. Mallapragada, and Balaji Narasimhan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Polybutadiene = William L. Hergenrother, Mark DeDecker, and Dan F. Graves . . . . . Polycarbonate (PC) = Sarah E. Morgan and Jun Li . . . . . . . . . . . . . . . . . . . . . . . Polymer Clay Nanocomposites = Hyoung J. Choi, Ji W. Kim, and Myung S. Jhon . . . . . Polymer Composites = Youssef K. Hamidi, M. Cengiz Altan, and Brian P. Grady . . . . Polymeric Membranes = Takeshi Matsuura and Mehrdad Rafat . . . . . . . . . . . . . . . Polymerization Reactions: Modeling, Design, and Control = Kyu Yong Choi . . . . . . . . . Polyurethanes = Joanna D. Fromstein and Kimberly A. Woodhouse . . . . . . . . . . . . . Polyvinylidene Fluoride = Jeffrey H. Yen and Ramin Amin-Sanayei . . . . . . . . . . . . . Proton-Exchange Membrane Fuel Cells = Pyoungho Choi, Pradeep Haldar, and Ravindra Datta . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Rheology of Cellulose Liquid Crystalline Polymers = Qizhou Dai, Richard Gilbert, and John F. Kadla . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Styrene–Butadiene Rubber = Jing Peng . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Superabsorbents = Takamasa Nonaka . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Theoretical Aspects of Liquid Crystals and Liquid Crystalline Polymers = James J. Feng . Thermosets: Materials, Processes, and Waste Minimization = Kwok-Wai Lem, Sean A. Curran, Steve Sund, and Mina Gabriel . . . . . . . . . . . . . . . . . . . . . . . . Ziegler–Natta Catalysis = John C. Chadwick . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

1973 2089 2101 2215

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

2247 2259 2277 2301 2313 2323 2335 2369 2379

. . . . . . 2501 . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

2663 2871 2881 2955

. . . . . . 3031 . . . . . . 3247

Process Design, Control, Optimization, and Simulation Bioprocess and Chemical Process Intensification = G. Akay . . . . . . . . . . . . . . . . . . CHP Technology/Systems = Louay M. Chamra and B. K. Hodge . . . . . . . . . . . . . . . Computational Fluid Dynamics = Andre´ Bakker and Elizabeth Marden Marshall . . . . . Computer-Aided Process Engineering = Andrzej Kraslawski . . . . . . . . . . . . . . . . . . . Dust Explosion Hazard Assessment and Control = Vahid Ebadat . . . . . . . . . . . . . . . Education on Plant Design = Truman S. Storvick . . . . . . . . . . . . . . . . . . . . . . . . . . Fractal Geometry: Applications = Douglas K. Ludlow . . . . . . . . . . . . . . . . . . . . . . . Greenhouse Gas Management for Multiplant Complexes = Ralph W. Pike . . . . . . . . . . Lubrication Performance Factors for Chemical Process Plant Machinery = Jim C. Fitch and Drew D. Troyer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Molecular Modeling for Nonequilibrium Chemical Processes = Dionisios G. Vlachos . . . Numerical Computations for Chemical Process Analysis and Design = David G. Retzloff Pilot Plant and Minipilot Units = Richard P. Palluzi . . . . . . . . . . . . . . . . . . . . . . . Pinch Design and Analysis = Robin Smith and Jin-Kuk Kim . . . . . . . . . . . . . . . . . . Power Factor = Peter R. Pujado´ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Process Optimization = Ralph W. Pike . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Real-Time Optimization: Status, Issues, and Opportunities = J. F. Forbes, T. E. Marlin, and W. S. Yip . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Six Sigma Design: An Overview of Design for Six Sigma (DFSS) = Sean A. Curran, Kwok-Wai Lem, Steve Sund, and Mina Gabriel . . . . . . . . . . . . . Solid–Liquid Mixing: Numerical Simulation and Physical Experiments = Philippe A. Tanguy, Francis Thibault, Gabriel Ascanio, and Edmundo Brito-De La Fuente . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

183 469 505 517 787 813 1053 1189

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

1511 1717 1949 2147 2165 2417 2439

. . . . . . 2585 . . . . . . 2719

. . . . . . 2753

xl

Process Safety and Loss Prevention Corrosion in the Process Industries = J. A. Richardson and R. A. Cottis . . . . Dust Explosion Hazard Assessment and Control = Vahid Ebadat . . . . . . . . Fouling of Heat Exchangers = T. Reg. Bott . . . . . . . . . . . . . . . . . . . . . . Gas Explosion Hazard: Prevention and Protection = Dehong Kong . . . . . . . Loss Prevention in Chemical Processing = Joel M. Haight . . . . . . . . . . . . . Lubrication Performance Factors for Chemical Process Plant Machinery = Jim C. Fitch and Drew D. Troyer . . . . . . . . . . . . . . . . . . . . . . . . . . Onsite and Offsite Emergency Preparedness for Chemical Facilities and Chemical Transportation = Michael K. Lindell and Ronald W. Perry . . Pipeline Safety = Glenn B. DeWolf . . . . . . . . . . . . . . . . . . . . . . . . . . . Pollution Prevention = Ashok Kumar, Harish G. Rao, Abhilash Vijayan, and Charanya Varadarajan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

549 787 1043 1109 1483

. . . . . . . . . . . . . 1511 . . . . . . . . . . . . . 1959 . . . . . . . . . . . . . 2181 . . . . . . . . . . . . . 2231

Reactor Engineering and Design Advanced Oxidation = Sangchul Hwang . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bioprocessing = Ryan G. Soderquist and James M. Lee . . . . . . . . . . . . . . . . . . Catalyst Preparation = X. D. Hu and Michael W. Balakos . . . . . . . . . . . . . . . . Emulsion Polymerization = Vincent G. Gomes . . . . . . . . . . . . . . . . . . . . . . . . Fermentation Processes = Kimberly Ogden . . . . . . . . . . . . . . . . . . . . . . . . . . Fermenter Design = Kishore K. Kar, Juergen Lueske, and Richard F. Cope . . . . . Fluidization = A.-H. Park and L.-S. Fan . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fluidized Bed Reactor = John R. Grace, Jamal Chaouki, and Todd Pugsley . . . . . Free-Radical Polymerization = Yadunandan Lal Dar, Rajeev Farwaha, and Gerard T. Caneba . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Gas–Liquid Mixing in Agitated Reactors = John C. Middleton, John M. Smith, and Piero M. Armenante . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Gas–Solid Reactions = Douglas P. Harrison . . . . . . . . . . . . . . . . . . . . . . . . . . Gas-to-Liquid Mass Transfer = Huu D. Doan, Simant R. Upreti, and Ali Lohi . . . High-Pressure Reactor Design = Joseph M. Lambert, Jr. and Steven C. Hukvari . . Hydrodynamics of Trickle-Bed Reactors = K. D. P. Nigam and Arunabha Kundu . . Hydrogenation Reactions in Dense Gas Systems = Gary Combes, Fariba Dehghani, Raffaella Mammucari, and Neil R. Foster . . . . . . . . . . . . . . . . . . . . . . . . Incineration and Combustion = Selim M. Senkan . . . . . . . . . . . . . . . . . . . . . . . Liquid–Liquid Mixing in Agitated Reactors = Richard V. Calabrese, Douglas E. Leng, and Piero M. Armenante . . . . . . . . . . . . . . . . . . . . . . . . Membrane Reactors = Ivo F. J. Vankelecom . . . . . . . . . . . . . . . . . . . . . . . . . . Microreactors and Microreaction Engineering = Richard I. Masel, Scott Gold, and Zheng Ni . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mixing and Chemical Reactions = Edward L. Paul, Suzanne M. Kresta, and Arthur W. Etchells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Multiphase Mixing and Solid–Liquid Mixing in Agitated Reactors = Piero M. Armenante, Victor Atiemo-Obeng, and W. Roy Penney . . . . . . . . . . Multiphase Reactors = Stanley M. Barnett . . . . . . . . . . . . . . . . . . . . . . . . . . . Polymerization Reactions: Modeling, Design, and Control = Kyu Yong Choi . . . . . . Reactive Extrusion = Gerard T. Caneba . . . . . . . . . . . . . . . . . . . . . . . . . . . . Reactive Separation = Vincent G. Gomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . Reactor Engineering = Ken K. Robinson . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. 41 199 345 863 941 951 997 1009

. . . . . . . . . 1057 . . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

1131 1151 1163 1245 1297

. . . . . . . . . 1337 . . . . . . . . . 1381 . . . . . . . . . 1457 . . . . . . . . . 1575 . . . . . . . . . 1643 . . . . . . . . . 1699 . . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

1767 1781 2335 2531 2541 2557

xli

Sonochemical Reaction Engineering = David A. Bruce and Amarnath Nareddy . . Spinning Disk Reactor = R. J. J. Jachuck and J. R. Burns . . . . . . . . . . . . . . . . Supercritical Fluid Technology: Reactions = Aydin K. Sunol, Sermin G. Sunol, and Naveed Aslam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Supercritical Water Oxidation = Ram B. Gupta . . . . . . . . . . . . . . . . . . . . . . Thermal Stability of Chemical Reactors = Haishan Zheng and Jason M. Keith . . Tubular Reactors: Reactor Types and Selected Process Applications = Patrick L. Mills and Joseph M. Lambert Jr. . . . . . . . . . . . . . . . . . . . . . . Water Gas Shift Reaction = Wolfgang Ruettinger and Oleg Ilinich . . . . . . . . . .

. . . . . . . . . . 2811 . . . . . . . . . . 2847 . . . . . . . . . . 2915 . . . . . . . . . . 2927 . . . . . . . . . . 2997 . . . . . . . . . . 3151 . . . . . . . . . . 3205

Semiconductor Processing and Microelectronics Chemical Mechanical Planarization in Integrated Circuit Manufacturing = John Zabasajja . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chemical Vapor Deposition = David G. Retzloff . . . . . . . . . . . . . . . . . . . . . . Crystal Growth = C. W. Lan, W. C. Yu, and W. C. Hsu . . . . . . . . . . . . . . . . . . . Diamond Films = Angel Velez and Mark A. Prelas . . . . . . . . . . . . . . . . . . . . . Diamond-Like Carbon Films = Angel Velez and Mark A. Prelas . . . . . . . . . . . . . Electrodeposition = Andre´ Avelino Pasa and Maximiliano Luis Munford . . . . . . . Gas-Phase Lubrication of MEMS Devices: Using Alcohol Vapor Adsorption Isotherm for Lubrication of Silicon Oxides = Kenneth Strawhecker, David B. Asay, and Seong H. Kim . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Microelectronics Fabrication = Edmund G. Seebauer and Charlotte T. M. Kwok . . Nanoporous Dielectric Materials = Jorge A. Lubguban and Shubhra Gangopadhyay Photoresists = Sergei V. Zelentsov and Nadezda V. Zelentsova . . . . . . . . . . . . . Plasma Etching = David G. Retzloff . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Thin Film Processes in MEMS and NEMS Technologies = W. R. Ashurst, C. Carraro, and R. Maboudian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Thin Film Science and Technology = T. L. Alford, J. Kouvetakis, and J. W. Mayer . Thin Liquid Film Deposition = Myung S. Jhon and Thomas E. Karis . . . . . . . . . . Wide Band-Gap Electronics Materials = Mark A. Prelas and Krishnendu Saha . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

429 441 589 685 695 821

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

1143 1615 1813 2111 2201

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

3049 3061 3075 3227

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

.. 1 . 25 221 269 407 449 481 617 729 749 1985 2003 2053 2541 2599 2701

Separation Processes Absorption Equipment = Karl B. Schnelle Jr. and Partha Dey . . . . . . . . . Adsorption = Shivaji Sircar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bioseparations = Shubhayu Basu and Shang-Tian Yang . . . . . . . . . . . . . Bubble Cap Tray = Stanley Marple . . . . . . . . . . . . . . . . . . . . . . . . . . Centrifuges = Alan G. Letki . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chiral Drug Separation = Bingyun Li and Donald T. Haynie . . . . . . . . . . Chromatographic Separations = Scott M. Husson . . . . . . . . . . . . . . . . . . Dehumidification = Louay M. Chamra and B. Keith Hodge . . . . . . . . . . . Distillation Column Design: Packing = Andrew W. Sloley . . . . . . . . . . . . Distillation Column Design: Trays = Andrew W. Sloley . . . . . . . . . . . . . . Osmotic Distillation = Bob Johnson . . . . . . . . . . . . . . . . . . . . . . . . . . Packed Absorption Column Design = Karl B. Schnelle, Jr. and Partha Dey . Petroleum Refinery Distillation = Stanley Marple . . . . . . . . . . . . . . . . . Reactive Separation = Vincent G. Gomes . . . . . . . . . . . . . . . . . . . . . . . Recent Advances in Catalytic Distillation = Flora T. T. Ng . . . . . . . . . . . . Scrubbers = S. Komar Kawatra . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

xlii

Separation Processes (cont’d) Solid–Liquid Separation = Frank M. Tiller and Wenping Li . . . . . . . . . . . . . . . . . . . . . . . . . 2769 Sorbent Technology = Shuguang Deng . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2825 Supercritical Fluid Extraction (SFE) = Ram B. Gupta . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2907

Specialty Chemicals Antioxidants = S. Al-Malaika and H. H. Sheena . . . . . . . . . . . . . . . . . . . . Blowing Agent = Kyung W. Suh . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chlorofluorocarbons = Byung Gwon Lee and Youn-Woo Lee . . . . . . . . . . . . Detergent Alkylate = Bipin Vora, Andrea Bozzano, and Stephen Sohn . . . . . Dimethylcarbonate = Byung Gwon Lee . . . . . . . . . . . . . . . . . . . . . . . . . . Drag Reducing Agents = Jacques L. Zakin, Ying Zhang, and Yunying Qi . . . . Heat Transfer Fluids = Satish C. Mohapatra . . . . . . . . . . . . . . . . . . . . . . Heavy Water (Deuterium Oxide) = Sharad M. Dave . . . . . . . . . . . . . . . . . Solvents = Satish C. Mohapatra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Thiochemicals: Mercaptans, Sulfides, and Polysulfides = Jeffrey Yen, Vijay R. Srinivas, and Gary S. Smith . . . . . . . . . . . . . . . . . . . . . . . . . Thiochemicals: Mercapto Acids and Organosulfur (IV/VI) Compounds = Jeffrey H. Yen, Gary S. Smith, and Vijay R. Srinivas . . . . . . . . . . . . . . Use of Lipases to Isolate Polyunsaturated and Oxygenated Fatty Acids and Form Value-Added Ester Products = Douglas G. Hayes . . . . . . . . . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. 81 237 459 663 719 767 1211 1221 2799

. . . . . . . . . . . . 3089 . . . . . . . . . . . . 3101 . . . . . . . . . . . . 3179

Supercritical Fluid Technology Critical Phase Behavior = J. Richard Elliott Jr. . . . . . . . . . . . . . . . . . . . . . . . Hydrogenation Reactions in Dense Gas Systems = Gary Combes, Fariba Dehghani, Raffaella Mammucari, and Neil R. Foster . . . . . . . . . . . . Processing of Pharmaceuticals Using Dense Gas Technologies = Raffaella Mammucari, Fariba Dehghani, and Neil R. Foster . . . . . . . . . . . . Supercritical CO2-Assisted Surface Coating Injection Molding = Masahiro Ohshima Supercritical Fluid Extraction (SFE) = Ram B. Gupta . . . . . . . . . . . . . . . . . . . Supercritical Fluid Technology: Reactions = Aydin K. Sunol, Sermin G. Sunol, and Naveed Aslam . . . . . . . . . . . . . . . . . . . . . . . . . . . . Supercritical Water Oxidation = Ram B. Gupta . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . 563 . . . . . . . . . 1337 . . . . . . . . . 2451 . . . . . . . . . 2897 . . . . . . . . . 2907 . . . . . . . . . 2915 . . . . . . . . . 2927

Thermodynamics and Process Applications Adsorption = Shivaji Sircar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Critical Phase Behavior = J. Richard Elliott Jr. . . . . . . . . . . . . . . . . . . . . . . . Hydrogen Bonding = J. Richard Elliott, Jr. . . . . . . . . . . . . . . . . . . . . . . . . . . Molecular Modeling for Nonequilibrium Chemical Processes = Dionisios G. Vlachos Phase Behavior of Hydrocarbon Mixtures = X.-Y. Zou and J. M. Shaw . . . . . . . . . Phase Equilibria = Karl B. Schnelle Jr. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Processing of Pharmaceuticals Using Dense Gas Technologies = Raffaella Mammucari, Fariba Dehghani, and Neil R. Foster . . . . . . . . . . . . Soave’s Modified Redlich–Kwong Equation of State = J. Richard Elliott, Jr. . . . . . Thermal Desorption = Timothy P. Sullivan . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. 25 563 1319 1717 2067 2077

. . . . . . . . . 2451 . . . . . . . . . 2747 . . . . . . . . . 2987

xliii

Transport Phenomena and Applications Computational Fluid Dynamics = Andre´ Bakker and Elizabeth Marden Marshall Fluid Flow = T. Reg. Bott . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fluid Transport in Porous Media = Michael C. Brooks . . . . . . . . . . . . . . . . . Fluidization = A.-H. Park and L.-S. Fan . . . . . . . . . . . . . . . . . . . . . . . . . . Heat Transfer Fluids = Satish C. Mohapatra . . . . . . . . . . . . . . . . . . . . . . . Hydrodynamics of Trickle-Bed Reactors = K. D. P. Nigam and Arunabha Kundu Mass Transfer Enhancement Because of Flow Instabilities = Vimal Kumar and K. D. P. Nigam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Porous Media = Karsten E. Thompson . . . . . . . . . . . . . . . . . . . . . . . . . . . Rheology of Cellulose Liquid Crystalline Polymers = Qizhou Dai, Richard Gilbert, and John F. Kadla . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

505 975 987 997 1211 1297

. . . . . . . . . . . 1531 . . . . . . . . . . . 2391 . . . . . . . . . . . 2663

Unit Operations and Design Absorption Equipment = Karl B. Schnelle Jr. and Partha Dey . . . . . . . . . . . . . . Adsorption = Shivaji Sircar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Biofilms = T. Reg. Bott . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bioseparations = Shubhayu Basu and Shang-Tian Yang . . . . . . . . . . . . . . . . . . Bubble Cap Tray = Stanley Marple . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Capsule Pipeline = Henry Liu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Centrifuges = Alan G. Letki . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Dehumidification = Louay M. Chamra and B. Keith Hodge . . . . . . . . . . . . . . . . Distillation Column Design: Packing = Andrew W. Sloley . . . . . . . . . . . . . . . . . Distillation Column Design: Trays = Andrew W. Sloley . . . . . . . . . . . . . . . . . . . Fouling of Heat Exchangers = T. Reg. Bott . . . . . . . . . . . . . . . . . . . . . . . . . . Gas–Liquid Contactors = Kishore K. Kar, Richard F. Cope, and Juergen Lueske . . Gas–Liquid Mixing in Agitated Reactors = John C. Middleton, John M. Smith, and Piero M. Armenante . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Heat Exchanger Operation and Troubleshooting = T. Reg. Bott . . . . . . . . . . . . . . Heat Transfer Fluids = Satish C. Mohapatra . . . . . . . . . . . . . . . . . . . . . . . . . Ion Exchange = Sukalyan Sengupta and Arup K. Sengupta . . . . . . . . . . . . . . . . Osmotic Distillation = Bob Johnson . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Packed Absorption Column Design = Karl B. Schnelle, Jr. and Partha Dey . . . . . . Pervaporation: Vapor Permeation = Nicholas Patrick Wynn . . . . . . . . . . . . . . . . Pressure-Relief Valve Design = Jonathan Francis . . . . . . . . . . . . . . . . . . . . . . Size Reduction = Sunil Kesavan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Solid–Liquid Separation = Frank M. Tiller and Wenping Li . . . . . . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

1 25 111 221 269 295 407 617 729 749 1043 1119

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

1131 1203 1211 1411 1985 2003 2031 2423 2735 2769

Preface

The Encyclopedia of Chemical Processing is an authoritative, dynamic, and most comprehensive multi-volume reference work on the broad subject of chemical processing, which will enable readers to have an enriching experience about general as well as targeted knowledge in this field. The Encyclopedia caters to engineers, scientists, researchers, inventors, professors, and students, as well as general readers in academia, industry, research institutions, government, and legal practices. In addition, the Encyclopedia has been designed to address the needs of practicing engineers and scientists, businessmen, lawyers, industrial executives, and professionals in the chemical processing and technology arena. The Encyclopedia encompasses the entire realm of chemical processing, offering upto-date, reliable, and comprehensive coverage of process technologies that have steadily progressed over the years, and at the same time identifying and addressing new breakthroughs and emerging technologies in chemical processing. The Encyclopedia contains a large number of entries that are devoted to life science subjects and futuristic materials and technologies, namely, biotechnology, nanomaterials and nanotechnology, and materials and technologies geared for microelectronics. Under the advice of an editorial advisory board comprised of distinguished and renowned scholars from around the world, the Encyclopedia will serve as the most respected reference work in the field of chemical processing. The Encyclopedia covers cradle-to-grave information on processing novel materials, emerging process technologies and resultant materials, and manufacturing organic and inorganic chemicals. Specific topics of interest include synthesis reactions, properties and characterization of materials, appropriate choice of catalysts, reactor design, process flowsheets, energy integration practices, pinch design, design of separation equipment and peripherals, environmental aspects of chemical plant operation such as safety and loss prevention, obedience of environmental regulations, waste reduction and management, and much more. The Encyclopedia also contains descriptions of different types of reactors and separation systems and their design, unit operations, system integration, process system peripherals such as pumps, valves, and controllers, analytical techniques and equipment, as well as pilot plant design and scale-up criteria. Fundamental aspects of industrial catalytic processes are detailed including catalyst preparation, characterization, structure-property relationships, deactivation and defouling, and catalyst regeneration methods. Examples of industrial processes that use different types of catalysts for chemical manufacture are also detailed. Identification and utilization of alternative resources for complementing our energy needs are addressed, which include renewable energy resources, oxygenated fuels, biofuels, fuel cells, and batteries. Polymers are ubiquitous in today’s life, and their utilization is limited only by chemists’ and chemical engineers’ imaginations. The Encyclopedia attempts to cover the wide spectrum of polymerization and polymer processing, including metallocene processes, description of structure, properties and end use of different polymers, copolymers, polymer blends and composites, polymer coatings, and rubber compounds. Additional topics of interest that are also covered include but are not limited to polymer characterization, molding technology, and polymer and rubber recycling. Advanced materials being used in myriad applications are also accounted for; examples of these are ceramics, nanomaterials, nanocomposites, carbon nanotubes, hydrophilic polymers, photovoltaic materials, biomaterials, and biomedical materials. xlv

xlvi

The issues potentially related to global warming cannot be understated in the present world, largely due to the increasing use of fossil fuels by automobiles and industries around the world. The degree to which this environmental issue affects society and the remedial measures needed in order to alleviate these concerns are well-addressed in the Encyclopedia, which covers emerging environmental technology, bioremediation, greenhouse gas alleviation, waste minimization, sequestration of carbon dioxide, etc. Biotechnology is a rapidly growing area of chemical, biological, and life sciences, and as such is also well-covered in the Encyclopedia. Enzymes, biomaterials, bioseparation, bioprocessing, bioreactor design, biocatalysis, BioMEMS, protein design, chiral drug separation, and hydrogels are few of the topics of merit that are included. The Encyclopedia also identifies and addresses emerging technologies in great detail including but not limited to nanotechnology, plasma technology, thin film technology, supercritical fluid technology and its applications, as well as microfabrication and micromachining for the microelectronics area. The authors of this initial printed version of the Encyclopedia are recognized experts in their fields, lending credibility and prestige to the Encyclopedia. All the authors were invited based on their records of accomplishment in the chosen topical areas. All entries were individually reviewed by peers as well as the Editor. As part of the review and revision processes, every effort was exercised to maintain the consistency, accuracy, readability, and up-to-date nature of the information presented. The Encyclopedia is published in both online and printed formats. The printed version consists of multiple traditional hardbound volumes with articles arranged alphabetically. The online version of the Encyclopedia is created by coupling the content of the printed edition with a powerful search engine, user-friendly interface, and customer-focused features. The online database is dynamic and evolving in nature, with additional articles added each quarter. The Editor feels honored to have been asked to undertake the important and challenging endeavor of developing the Encyclopedia of Chemical Processing that will cater to the needs of the rapidly changing world of the 21st century. The Editor is humbled to follow the impeccable work of the previous editor, Professor John J. McKetta, who led the development of the Encyclopedia of Chemical Processing and Design, a total of 69 volumes, which has become one of the most authoritative reference sources for scientists, engineers, and practitioners for several decades. I would like to express my most sincere thanks and appreciation to the authors for their excellent professionalism and dedicated work. Needless to say, an encyclopedia of this nature would never exist if the expert authors had not devoted their valuable time to preparing the authoritative entries on their assigned topics. I wish to thank all my colleagues and friends as well as the editorial board members for all their suggestions, comments, assistance, volunteerism, and patience. In particular, I appreciate the encouragements, guidance, and assistance provided by Mr. Russell Dekker, Dr. Chai-sung Lee, Dr. John C. Angus, Dr. C. C. Liu, Dr. James G. Speight, Dr. Robert Dye, Dr. Sunil Kesavan, Dr. John Zabasajja, Dr. J. Richard Elliott, Jr., Dr. Abhay Sardesai, Dr. Hirotsugu Yasuda, Dr. David G. Retzloff, Dr. Patricia Roberts, Dr. Kelly Clark, Dr. Jeffrey Yen, Dr. Peter Pujado, and Dr. Stephen J. Lombardo. I also would like to thank Mr. Jonathan E. Wenzel, Ms. Leah A. Leavitt, Dr. Teresa J. Cutright, Dr. H. Bryan Lanterman, Dr. Qingsong Yu, and Dr. Patricia A. Darcy for providing various assistance while editing. I am also deeply indebted to the former and current employees of the Publisher for their dedicated work toward successful completion of the project, to name a few, Ms. Alison Cohen, Ms. Oona Schmid, Ms. Marisa Hoheb, Ms. Maria Kelley, Ms. Meaghan Johnson, and Ms. Joanne Jay. The contributions of those mentioned made this Encyclopedia possible. Sunggyu Lee Editor

Absorption Equipment A Karl B. Schnelle, Jr. Chemical Engineering Department, Vanderbilt University, Nashville, Tennessee, U.S.A.

Partha Dey P. A. Consulting, Nashville, Tennessee, U.S.A.

INTRODUCTION Absorption is a mass transfer operation in which a soluble gaseous component is removed from a gas stream by dissolving in a liquid. Absorption can be used to recover valuable gaseous components such as hydrocarbons or to remove unwanted gaseous components such as hydrogen sulfide from a stream. A valuable solute can be separated from the absorbing liquid and recovered in a pure, concentrated form by distillation or stripping (desorption). The absorbing liquid is then used in a closed circuit and is continuously regenerated and recycled. Examples of regeneration alternatives to distillation or stripping are removal through precipitation and settling; chemical destruction through neutralization, oxidation, or reduction; hydrolysis; solvent extraction; and liquid adsorption. Absorption is one of the main methods of separation used in the chemical processing industry. Accompanied by chemical reaction between the absorbed component and a reagent in the absorbing fluid, absorption can become a very effective means of separation. Absorption can also be used to remove an air pollutant like an acid gas from stream. Then, the system could be a simple absorption in which the absorbing liquid is used in a single pass and then disposed of while containing the absorbed pollutant.

Operations of Absorption Towers In the past it was the custom to call absorbers operating as cleanup towers to remove undesirable gaseous effluents by the name of scrubber. At that time most of the effluent gases being removed were acid gases being scrubbed with water. The designation of scrubber to scrub the discharge gas and clean it seemed rather natural. Today the same kind of operation is carried out, but with more stringent regulations imposed by the local air pollution control agency. The name scrubber is now applied to those operations in which particulate matter is removed but the scrubbing operation may also include the simultaneous removal of gaseous pollutants. Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120007642 Copyright # 2006 by Taylor & Francis. All rights reserved.

In this chapter the term absorber will refer to the removal of gaseous contaminants.

General Considerations Filters, heat exchangers, dryers, bubble cap columns, cyclones, etc., are ordinarily designed and built by process equipment manufacturers. However, units of special design for one-of-a-kind operations such as packed or plate towers are quite often designed and built under the supervision of plant engineers. Thus, there is a large variety of this type of equipment, none of it essentially standard.

TYPES OF ABSORPTION EQUIPMENT Absorption takes place in either staged or plate towers or continuous or packed contactor. However, in both cases the flow is continuous. In the ideal equilibrium stage model, two phases are contacted, well mixed, come to equilibrium, and then are separated with no carryover. Real processes are evaluated by expressing efficiency as a percentage of the change that would occur in the ideal stages. Any liquid carryover is removed by mechanical means. In the continuous absorber the two immiscible phases are in continuous and tumultuous contact within a vessel that is usually a tall column. A large surface is made available by packing the column with ceramic or metal materials. The packing provides more surface area and a greater degree of turbulence to promote mass transfer. The penalty for using packing is in the increased pressure loss in moving the fluids through the column, which causes an increased demand for energy. In the usual countercurrent flow column, the lighter phase enters the bottom and passes upward. Transfer of material takes place by molecular and eddy diffusion processes across the interface between the immiscible phases. Contact may be also cocurrent or cross-flow. Columns for the removal of 1

2

air contaminants are usually designed for countercurrent or cross-flow operation. Absorption can take place in a countercurrent, cocurrent, or cross-flow device. Vertical countercurrent towers are either built with a metal, plastic, or ceramic packing or constructed as plate towers with various types of plates. This chapter will discuss the solvents used to carry out absorption and the various types of absorption equipment.

ABSORPTION SOLVENTS Absorption systems can be divided into those that use water as the primary absorbing liquid and those that use a low-volatility organic liquid. The gas solubility should be high in the absorbing solvent. The gas leaving an absorber is usually saturated with the solvent; therefore, the solvent should have a low vapor pressure. A lower viscosity solvent is advantageous to promote more rapid absorption rates and improve flooding characteristics. The solvent should not be corrosive to the materials of construction of the absorber. It should be nontoxic and nonflammable. Depending on the region where the absorber is to be constructed, the solvent should have a low freezing point. Nonaqueous Systems At first glance, an organic liquid appears to be the preferred solvent for absorbing hydrocarbon and organic vapors from a gas stream because of improved solubility and miscibility. The lower heat of vaporization of organic liquids results in energy conservation when solvent regeneration must occur by stripping. Many heavy oils such as No. 2 fuel oil or heavier and other solvents with low vapor pressure can do extremely well in reducing organic vapor concentrations to low levels. Care must be exercised in picking a solvent that will have sufficiently low vapor pressure so that the solvent itself will not become a source of volatile organic pollution. Obviously, the treated gas will be saturated with the absorbing solvent. An absorber–stripper system for recovery of benzene vapors has been described by Crocker.[1] Other aspects of organic solvent absorption requiring consideration are stability of the solvent in the gas solvent system, for example, its resistance to oxidation, and its possible fire and explosion hazard. Although water is the most common liquid used for absorbing acidic gases, amines (monoethanol-, diethanol-, and triethanolamine; methyldiethanolamine; and dimethylaniline) have been used for absorbing SO2 and H2S from hydrocarbon gas streams. Such absorbents are generally limited to solid particulate free systems because solids can produce difficult to handle

Absorption Equipment

sludge as well as use up valuable organic absorbents. Furthermore, because of absorbent cost, absorbent regeneration must be practiced in almost all cases.

Aqueous Systems Absorption is one of the most frequently used methods for removal of water-soluble gases. Acidic gases such as HCl, HF, and SiF4 can be absorbed in water efficiently and readily, especially if the last contact is made with water that has been made alkaline. Less soluble acidic gases such as SO2, C12, and H2S can be absorbed more readily in a dilute caustic solution. The scrubbing liquid may be made alkaline with dissolved soda ash or sodium bicarbonate, or with sodium hydroxide, usually with no higher a concentration in the scrubbing liquid than 5–10%. Lime is a cheaper and more plentiful alkali, but its use directly in the absorber may lead to plugging or coating problems if the calcium salts produced have only limited solubility. A technique often used is the two-step flue gas desulfurization process, where the absorbing solution containing NaOH is used inside the absorption tower, and then the tower effluent is treated with lime externally, precipitating the absorbed component as a slightly soluble calcium salt. The precipitate may be removed by thickening and the regenerated sodium alkali solution is recycled to the absorber. Scrubbing with an ammonium salt solution can also be employed. In such cases, the gas is often first contacted with the more alkaline solution and then with the neutral or slightly acid contact to prevent stripping losses of NH3 to the atmosphere. When flue gases containing CO2 are being scrubbed with an alkaline solution to remove other acidic components, the caustic consumption can be inordinately high if CO2 is absorbed. However, if the pH of the scrubbing liquid entering the absorber is kept below 9.0, the amount of CO2 absorbed can be kept low. Conversely, alkaline gases, such as NH3, can be removed from the main gas stream with acidic water solutions such as dilute H2SO4, H3PO4, or HNO3. Single-pass scrubbing solutions so used can often be disposed of as fertilizer ingredients. Alternatives are to remove the absorbed component by concentration and crystallization. The absorbing gas must have adequate solubility in the scrubbing liquid at the resulting temperature of the gas–liquid system. For pollutant gases with limited water solubility, such as SO2 or benzene vapors, the large quantities of water that would be required are generally impractical on a single-pass basis, but may be used in unusual circumstances. An early example from the United Kingdom is the removal of SO2 from flue gas at the Battersea and Bankside electric power stations, which is described by Rees.[2] Here, the normally alkaline

Absorption Equipment

water from the Thames tidal estuary is used in a large quantity on a one-pass basis.

PACKED TOWERS There are two major types of packing, random dumped pieces and structured modular forms. The structured packing is usually crimped or corrugated sheets. The packing provides a large interfacial area for mass transfer and should have a low-pressure drop. However, it must permit passage of large volumes of fluid without flooding. The pressure drop should be the result of skin friction and not form drag. Thus, flow should be through the packing and not around the packing. The packing should have enough mechanical strength to carry the load and allow easy handling and installation. It should be able to resist thermal shock and possible extreme temperature changes, and it must be chemically resistant to the fluids being processed. Random or Dumped Packing Random packings are dumped into the tower during construction and are allowed to fall at random. The tower might be filled with water, first to allow a gentler settling and to prevent breakage, especially with ceramictype packing. Random dumped tower packing comes in many different shapes. Two of the most popular are rings and saddles. Sizes range from 0.25 in. to 3.5 in., with 1 in. being a very common size. The choice of a packing is mostly dependent on the service in which the tower will be engaged. Packings are made of ceramic, metal, or plastic, depending on the service. Ceramic materials will withstand corrosion and are therefore used where the solutions resulting are aqueous and corrosive. Metals are used where noncorrosive organic liquids are present. Plastic packing may be used in the case of corrosive aqueous solutions and for organic liquids that are not solvents for the plastic of which the packing is made. Metal packing is more expensive, but provides lower pressure drop and higher efficiency. When using plastic materials, care must be taken that the temperature is not too high and that oxidizing agents are not present. Ring-type packings are commonly made of metal or plastic, except for Raschig rings, which are generally ceramic. Ring-type packings lend themselves to distillation because of their good turndown properties and availability in metals of all types that can be press formed. Usually, ring-type packings are used in handling organic solutions when there are no corrosive problems. However, rings do not promote redistribution of liquids, and Raschig rings may even cause maldistribution. Saddles are commonly made from ceramic or plastic and

3

give good corrosion resistance. Saddles are best for redistribution of liquid and, thus, serve as a good packing for absorption towers. Structured Packing Early on after the production of random packings had been used extensively, stacked beds of the conventional random packings such as larger-sized Raschig rings were used as ordered packings. Owing to the high cost of installation of this type of packing, it was largely discontinued. At that time multiple layers of corrugated metal lath formed into a honeycomb structure came into use. Later on, a woven wire mesh arranged in rows of vertically corrugated elements came into use. Subsequently, other wire-mesh structures have gained favor. Then, a sheet metal structured packing was developed to reduce the expense of the wire-mesh type. The use of this structured type of packing not only promotes mass transfer owing to increased surface area, but also has less pressure drop in many different services. Tower Considerations Materials of construction Random packing can be made from ceramic materials, plastic, or metal. Most structured packing and plates in staged towers are made from metal although there are simple woven types of plastic materials that can be considered as structured packing. The tower packing, plates, and tower materials must be compatible with the fluids flowing through the towers. Of particular significance would be acid gases that may have a deleterious effect on metal tower internal parts and organic solvents that may have a serious effect on plastic materials. It is also necessary to consider the case where there may be a high heat of absorption emitted. The internal tower may have to be cooled to withstand the temperature that results from the heat of absorption. Flow arrangements In diffusional operations such as absorption where mass is to be transferred from one phase to another, it is necessary to bring the two phases into contact to permit the change toward equilibrium to take place. The transfer may take place with both streams flowing in the same direction, in which case the operation is called concurrent or cocurrent flow. When the two streams flow in the opposite direction, the operation is termed countercurrent flow, an operation carried out with the gas entering at the bottom and flowing

A

4

Absorption Equipment

upward and the liquid entering at the top and flowing down. This process is illustrated in Fig. 1. A combined operation in which the contaminated gas is first cleaned in a countercurrent operation, as shown in Fig. 2, and then the gas is further treated to remove more of the contaminant as shown in the cocurrent operation that follows. Countercurrent operation is the most widely used absorption equipment arrangement. As the gas flow increases at constant liquid flow, liquid holdup must increase. The maximum gas flow is limited by the pressure drop and the liquid holdup that will build up to flooding. Contact time is controlled by the bed depth and the gas velocity. In countercurrent flow mass transfer driving force is maximum at the gas entrance and liquid exit. Cocurrent operation can be carried out at high gas velocities because there is no flooding limit. In fact, liquid holdup decreases as velocity increases. However, the mass transfer driving force is smaller than in countercurrent operation. Some processes for both absorption and the removal of particulates employ a cross-flow spray

Fig. 1 Countercurrent flow packed tower.

chamber operation. Here, the water is sprayed down on a bed of packing material. The carrier gas containing pollutant gas or the particulate flows horizontally through the packing, with the spray and packing causing the absorbed gas or particles to be forced down to the bottom of the spray chamber where they can be removed. Fig. 3 illustrates a cross-flow absorber. The design of cross-flow absorption equipment is more difficult than vertical towers because the area for mass transfer is different for the gas and liquid phases. Continuous and steady-state operation is usually most economical. However, when smaller quantities of material are processed, it is often more advantageous to charge the entire batch at once. In fact, in many cases this is the only way the process can be done. This is called batch operation and is a transient operation from start-up to shut-down. A batch operation presents a more difficult design problem.

Packed tower internals In addition to the packing, absorption towers must include internal parts to make a successful piece of operating equipment. Fig. 4 illustrates the placement of the tower internals. These internals begin with a packing support plate at the bottom of the tower. The packing support plate must physically support the weight of the packing. It must incorporate a high percentage of free area to permit relatively unrestricted flow of downcoming liquid. A flat plate has the disadvantage in that both liquids and gases must pass countercurrently through the same holes. Therefore, a substantial hydrostatic head may develop. Furthermore, the bottom layer of packing partially blocks many of the openings reducing the free space. Both of these conditions lower tower capacity. A gas injection plate provides separate passage for gas and liquid and prevents buildup of hydrostatic head. Liquid distributors are used at all locations where an external liquid stream is introduced. Absorbers and strippers generally require only one distributor, while continuous distillation towers require at least two, at the feed and reflux inlets. The distributors should be 6–12 in. above packing to allow for gas disengagement from the bed. The distributor should provide uniform liquid distribution and a large free area for gas flow. Liquid redistributors collect downcoming liquid and distribute it uniformly to the bed below. Initially, after entering the tower the liquid tends to flow out to the wall, the redistributor makes that portion of the liquid more available again to the gas flow. It also breaks up the coalescence of the downcoming liquid, and it will eliminate factors that cause a loss of efficiency in the tower and reestablish a uniform pattern of liquid

Absorption Equipment

5

A

Fig. 2 Combined countercurrent and cocurrent operation.

irrigation. A bed depth of up to 6 m (20 ft) should be alright before redistribution is needed. Retaining and hold-down plates are used only with ceramic or carbon tower packing. They prevent the upper portion of the packed bed from becoming fluidized and from breaking up during surges in pressure or at high-pressure drop. The plates rest directly on packing and restrict movement by virtue of the weight of the plate. Retainers or bed limiters prevent bed expansion or fluidization. When operating at highpressure drops, retainers are fastened to the wall. They are designed to prevent individual packing pieces from passing through the plate openings.

PLATE TOWERS Plate or tray towers are vertical cylinders in which the gas and liquid are contacted on horizontal plates in a stepwise fashion. By the nature of the operation plate towers are countercurrent flow devices. Fig. 5 shows a typical arrangement. In plate columns the gas is introduced at the bottom. Contact between gas and liquid is obtained by forcing the gas to pass upward through small orifices, bubbling through a liquid layer flowing across a plate. The liquid is introduced at the top and passes downward by gravity over the plate and through a downcomer onto the next plate. The

Fig. 3 Cross-flow absorber operation.

6

Absorption Equipment

Fig. 4 Packed column with internal hardware indicated. (View this art in color at www.dekker. com.)

bubble cap tower is the classical contacting device. Each plate of the tower is a stage in which interphase diffusion occurs and the fluids are separated. Ideally, the vapor and liquid would reach equilibrium at each stage. The number of these ideal stages that are required is determined by the difficulty of separation. The number is calculated from the mass and energy balances around the plate and the tower. The stage or tray efficiency is determined by the mechanical design. Higher contact times result in higher efficiencies. Deeper pools of liquid on the plates promote

higher contact times, and higher gas velocities promote better efficiency as well. Unfortunately, these conditions can lead to flooding of the plates and a severe drop in efficiency or foaming of the liquid on the plates. Thus, an inoperative situation might result. Turndown ratio is defined as the ratio of design rate to minimum rate. In many instances of tower design and operation, the performance when operating below the design rate becomes important. Furthermore, plate towers experience the same materials limitations as discussed earlier for packed towers. When selecting

Absorption Equipment

7

a cap that is mounted on top of the riser. A series of slots are cut into the cap through which the gas passes into the liquid that is flowing across the plate. They have the advantage of being able to handle wide ranges of liquid and gas flows. However, the new types of trays are much less expensive; therefore, bubble caps are being phased out of use in new tower designs. Sieve Trays

Fig. 5 Plate column. (View this art in color at www.dekker. com.)

plate types in addition to materials of construction, turndown ratio, pressure drop, capacity, and efficiency must be considered. Plate Types A plate type column may be operated in either a crossflow or a counterflow method. Cross-flow plates are the most common types. Perry and Green and Wankat discuss the advantages of both type plates.[3,4] Crossflow plates use a downcomer to transport the liquid from the upper plate to the lower plate. They offer greater mass transfer efficiency and operating range. The downcomer may be located to control the liquid flow pattern. Newer designs of cross-flow plates employ perforations, which may be simple round orifices or may contain movable valve-like vents that act like variable orifices. These type plates will be discussed in the following section. In counterflow plates there are no downcomers and the liquid and gas use the same openings for flow. The openings are usually round perforations or long slots. The plates may be corrugated to segregate the liquid and gas flow.

Sieve plates are simple flat plates perforated with small holes. The advantages are low cost and high plate efficiency but they have narrow gas flow operating ranges. These trays may be subject to flooding because of liquid backing up in the downspouts or excessive entrainment. Fig. 6 is a schematic illustrating bubble cap and sieve trays. Efficiency remains good at design conditions. However, turndown is relatively poor, and therefore, the trays are not flexible in operating conditions. Sieve trays are relatively resistant to clogging and they can have large holes that make them easier to clean. Entrainment is much less than that experienced in bubble cap trays; therefore, plate spacing can be smaller than in bubble cap trays. Valve Trays A variation of the bubble cap tray is the valve tray, which permits greater variations in gas flow without dumping the liquid through the gas passages. Valve trays are also cross-flow type and can be described as sieve trays with large variable openings. The openings are covered with movable caps that rise and fall as the gas rate increases and decreases. This keeps the gas velocity through the slots essentially constant. Valve

Bubble Caps Bubble cap trays, a cross-flow type of plate, were originally the most common type of tray. On these trays risers lead the gas up through the tray and underneath

Fig. 6 Bubble cap and sieve tray. (View this art in color at www.dekker.com.)

A

8

trays are designed to have better turndown ratios than sieve trays and their efficiency remains high as the gas rate drops. Valve trays are more likely to plug if solids are present and are more costly than sieve trays.

Absorption Equipment

poor, and information on design parameters is hard to find.

Spray Chambers Baffle Tower A counterflow plate-contacting device for absorption is the baffle tower, which has been employed occasionally when plugging and scaling problems are expected to be severe. Fig. 7 illustrates a baffle tower. Gases passing up the tower must pass through sheets of downwardly cascading liquid, providing some degree of contact and liquid atomization. Baffle tower design may use alternating segmental or disk and doughnut plates. Here, the gas alternately flows upward through central orifices and annuli traversing through liquid curtains with each change in direction. Mass transfer is generally

Fig. 7 Baffle tray tower. (View this art in color at www. dekker.com.)

A liquid may be introduced into a tower as fine drops through a nozzle. This device is known as a spray chamber. The flow could be cocurrent or countercurrent. A countercurrent spray chamber is shown in Fig. 8. These towers are considerably more resistant to plugging when solid particulates are present in the inlet gas. However, difficulties with plugging in spray towers and erosion can be troublesome when the spray liquid is recycled. Particle settling followed by fine strainers or even coarse filters is beneficial to eliminate

Fig. 8 Spray chamber. (View this art in color at www. dekker.com.)

Absorption Equipment

this problem. These devices have the advantage of low-pressure drop but there is a tendency for the liquid to be entrained in the gas leaving the tower. Mist eliminators can help reduce this problem. An additional disadvantage is the cost of pumping the liquid to force it through the nozzles. The efficiency of spray chambers can be improved by introducing the feed into the tower in a cyclonic manner.

ABSORPTION FOR AIR POLLUTION CONTROL Absorption plays a major role today in air pollution control. In the first part of this chapter it was noted that absorption was referred to as scrubbing especially when associated with cleaning up a stream containing an acid gas before it was emitted to the atmosphere. Today in air pollution control technology the term scrubber continues to be used with reference to cleanup of pollutant gases but usually when the equipment used also removes particulate matter. The combined action of removal of particulates and gases takes place in venturi scrubbers, spray towers, plate towers, and other types of devices. Absorption of sulfur oxides and nitrogen oxides are the most common devices where the combined action of particulate removal and absorption of gases takes place. It must be noted, however, that if an absorption process is going to be used to remove sulfur oxides it should not precede an electrostatic precipitator. Removal of the sulfur molecules before the precipitator will change the electrical properties of the gas and may result in loss of the ability to remove the particulate matter in the precipitator. There continue to be many absorbers for the removal of water-soluble gases. Acid gases and some volatile organic compounds can be absorbed readily in water by the types of equipment previously discussed. These processes are essentially absorption with chemical reaction. For a discussion of absorption in air pollution control and a description of several absorption systems for sulfur dioxide and nitrogen oxide removal, see Schnelle and Brown.[5] A more detailed discussion of many more processes for flue gas desulfurization employing absorption is given by Lunt and Cunic.[6] The United Kingdom has a long history of flu gas desulfurization. The world’s first such system was installed in a power plant at Battersea, England, in 1936.[7] At Bankside a 228 MW boiler was the first installed. It was fitted with a once-through scrubber system using water from the condensers, which was dosed with alkali. A further unit was commissioned in 1949. Both units continued to operate until the early 1970s. At Fulham power station, a 120 MW boiler system was operated with recycled lime dosed scrubbing liquor from 1936 to 1940.

9

It should also be noted that absorption has been used to remove contaminants from natural gas streams during processing. In the early 1930s di-ethanolamine was used as an absorbent for both hydrogen sulfide and carbon dioxide.[8] This process became known as the ‘‘Gerbitol Process.’’ Other alkanolamines such as mono-ethanolamine and di-isopropanolamine have also found wide application. Plate Towers Originally, bubble cap plates had been used for absorption of pollutant gases such as sulfur dioxide. However, the solids in the slurries used as absorbents can more readily plug bubble caps. Typical absorbents used in current processes include, for example, conventional lime slurry; lime-limestone slurries; mixed sodium sulfite= lime slurries; and magnesium sulfite=bisulfite mixed with lime slurries. Conventional lime slurry towers may consist of a multilevel spray tower combined with a venturi scrubber. Venturi scrubbers will be discussed briefly below. Mixed sodium sulfite=lime slurries may be contacted in a plate tower. Sieve plates might be used with larger than normal holes to help prevent plugging due to the solids in the slurries. Venturi Scrubbers Venturi scrubbers are designed on the basis of the venturi flow-metering device. The flow channel is narrowed down so that the velocity will greatly increase at the throat. Then, as in the flow-metering device the flow channel widens out. For ease of fabrication venturi scrubbers are designed with a rectangular cross section. The absorption fluid is injected into the Venturi at the throat where the velocity is the greatest. For particulate removal plain water could be used. As in the plate columns discussed above, for the simultaneous removal of particulate matter and sulfur dioxide, soda ash or caustic soda slurries could be used for absorption of the gas. Venturis are frequently used in conjunction with plate towers. They also serve as stand-alone removal devices in some cases.

CONCLUSIONS—A COMPARISON OF PACKED COLUMNS AND PLATE COLUMNS Absorption is a mass transfer operation that is commonly used to recover valuable gaseous components or to remove undesirable components of a gas stream. It is one of the main methods of separation in the chemical process industries. Absorption can take place in packed towers or in plate towers. Perry and Green (9) compare the advantages of packed and plate towers.[3]

A

10

Absorption Equipment

Table 1 Economic factors in packed tower design operating and capital cost factors Operating costs

Economic Factors Affecting Packed Tower Construction

Capital costs

Pumping power for gas and liquid

Tower and shell, packing or plates

Labor and maintenance

Packing support

Steam and cooling water

Gas and liquid distributor

Loss of unabsorbed material

Pumps, blowers, and compressors

Disposal of absorbed material

Piping and ducts

Solvent makeup

Heat exchangers

Solvent purification

Solvent recovery system

When column diameters are less than 0.6 m (2.0 ft) packed towers can be considerably cheaper. However, if alloy metals are necessary, plate towers may result in less cost. Using ceramic or other similar resistant materials for packing and materials of construction, packed towers can serve to handle corrosive materials and acids. Because the gas flow in packed towers may offer less degree of agitation, packed tower operation may be better for liquids that tend to foam. When liquids are thermally sensitive, packed columns may offer less holdup and thus prevent changes taking place in the liquids due to thermal reaction. When solids are contained in liquids or when solids have the chance of condensing out of the gas steam, plate columns offer the advantage of being able to be designed to be more readily cleaned. Plate columns can more readily absorb thermal expansion, which might result in breakage of packing as it is inserted into the column or during the operation. Cooling coils can be more readily installed in plate towers than in packed towers. Flooding may occur with high liquid rates in packed columns, whereas a plate tower may be designed to handle the higher liquid rate. Low liquid flow rates may result in poor wetting of packing, thereby resulting in poor mass transfer. Thus, a plate tower may more readily handle lower liquid flow rates.

When a tower is designed for treating a given quantity of gas per hour, the height of the tower, especially in packed towers, is determined from mass transfer considerations. The diameter or cross-sectional area is determined by fluid dynamics from the gas velocity in the empty tower cross section in packed towers and by the velocity through the bubble caps or other openings in plate towers. The smaller the diameter of the tower, the higher the gas velocity that will help the gas to overcome the tower pressure drop. This could result in a lower cost of pushing the gas through the tower. The chief economic factors to be considered in tower design are listed in Table 1.

REFERENCES 1. Crocker, B.B. Capture of hazardous emissions. In Control of Specific Toxic Pollutants, Proceedings of the Conference, Air Pollution Control Association, Gainesville, FL, Feb 1979; Air and Waste Management Association: Pittsburgh, PA, 1979; 414–433. 2. Rees, R.L. The removal of oxides of sulfur from flue gases. J. Inst. Fuel 1953, 25, 350–357. 3. Perry, R.H.; Green, D.W. Perry’s Chemical Engineer’s Handbook; 7th Ed.; McGraw-Hill: New York, 1997; 14–24, 14–25, 14–39, 14–40. 4. Wankat, P.C. Equlibrium Staged Separations; Prentice Hall: Upper Saddle River, NJ, 1988; 369–379. 5. Schnelle, K.B., Jr.; Brown, C.A. Air Pollution Control Technology Handbook; CRC Press: Boca Raton, FL, 2002. 6. Lunt, R.R.; Cunic, J.D. Profiles in Flue Gas Desulfurization; American Institute of Chemical Engineers: New York, 2000. 7. http:==www.dti.gov.uk=energy=coal=cfft=pub= cb013-a.pdf (accessed Dec 2004). 8. http:==www.r-t-o-l.com=learning=studentsguide= sru.htm (accessed Dec 2004).

Activated Sludge Process A Shankha K. Banerji Department of Civil and Environmental Engineering, University of Missouri–Columbia, Columbia, Missouri, U.S.A.

INTRODUCTION Wastewater treatment occurs in a treatment plant in several stages depending on the degree of treatment desired. In the first stage, the preliminary treatment processes prepare the influent wastewater for treatment in subsequent processes. Bar screens, grit chamber, and flow equalization tank are some of the processes included in the preliminary treatment. There is no significant removal of biodegradable organic matter expressed in terms of 5-day biochemical oxygen demand (BOD) or suspended solids by these processes. The next stage is the primary treatment process where settleable (and floatable) solids present in the wastewater are removed by gravity sedimentation. In some rare instances, the flotation process can be used instead of gravity sedimentation for the removal of settleable solids. The primary treatment process can remove up to 40% of the incoming BOD and 50–70% of the suspended solids.[1] The subsequent stage is the secondary treatment process, which is needed to remove the remaining soluble and colloidal organic matter from the wastewater that was not removed during the primary treatment processes. The secondary processes invariably use aerobic biological treatment processes to remove the soluble and colloidal organic matter from the wastewater. The biological treatment process converts the soluble and colloidal organic matter into settleable solids and micro-organisms (sludge), which are removed in the secondary settling tank leaving a clearer supernatant effluent for discharge. Thus, the settling tank following the aeration tank is an integral part of the process. In this entry, the secondary tank details are not included. These processes in combination with the primary process can remove 90þ% BOD (carbonaceous BOD) and suspended solids. Thus, the secondary wastewater treatment processes can meet the current US Environmental Protection Agency mandated effluent requirements of 30 mg=L of BOD and 30 mg=L of suspended solids for municipal wastewater treatment.[1] There are two types of secondary aerobic biological treatment processes: suspended growth processes and attached growth processes. In the suspended growth process, the micro-organisms responsible for the biochemical conversion of organic matter are kept in Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120007653 Copyright # 2005 by Taylor & Francis. All rights reserved.

suspension by aeration or agitation in a tank where the wastewater is introduced. The micro-organisms assimilate the organic compounds for synthesis of new cells (biomass) and for respiration, which provides the energy for the synthesis and other cellular processes. Activated sludge process and its modifications are suspended growth processes. In the attached growth process, the micro-organisms are present in an attached form (biofilm) on a medium, either stone, treated wood, or synthetic plastic materials. The wastewater comes in contact with these attached microorganisms, and the same biochemical processes as in the suspended growth process take place, namely, cell synthesis and respiration. Trickling filters and rotating biological contactors are the two most common attached growth secondary biological treatment processes used.[1] In this article, activated sludge process and some of its modifications are discussed at some length. Only the details of carbonaceous BOD removal from wastewater are included.

CONVENTIONAL PROCESS The conventional activated sludge process consists of an aeration tank followed by a settling tank, as shown in Fig. 1. The wastewater from the primary settling tank enters the aeration tank and mixes with the micro-organisms or biomass present. A portion of the settled sludge (biomass) in the secondary settling tank is recycled back to the head of the aeration tank. This recycled sludge is referred to as return activated sludge (RAS). The term sludge in the secondary settling tank refers to solids that have settled in the tank bottom because of gravity forces. The recycling of the sludge maintains a desired amount of biomass concentration in the aeration tank. The solids responsible for the bio-oxidation of organic matter consist of microorganisms (biomass), biodegradable and nonbiodegradable organic matter, and some inert solids in the aeration tank known as mixed liquor suspended solids (MLSS). The mixture of wastewater and these solids is called mixed liquor. The organic components of the MLSS are known as mixed liquor volatile suspended 11

12

Activated Sludge Process

Fig. 1 Conventional activated sludge process.

solids (MLVSS). Mixed liquor volatile suspended solids are often considered to represent the active biomass in the system. The remaining settled sludge from the secondary settling tank is withdrawn as waste activated sludge (WAS) for further processing before disposal. The biological oxidation process taking place in the aeration tank can be described by the following equation: Organic matter þ N þ P þ O2 þ micro-organisms ¼ CO2 þ H2 O þ new micro-organisms In the above equation, N and P are the nitrogen and phosphorus compounds (sometimes called nutrients) that are needed for micro-organism metabolism and growth. In most instances, they are already in excess amounts in domestic wastewater, but for special cases they have to be supplied if not present in adequate quantities. The nutrient requirements in an aerobic biological treatment process are based on the BOD load imposed on the system. For every 100 kg of BOD introduced to the system, 5 kg of N and 1 kg of P should be available.[2] The oxygen needed has to be in a dissolved state to be available to the cells. The organic matter must be a biodegradable type that can be utilized by the micro-organisms present. The transport of the organic molecules inside the cells occurs through a microbial cell ˚ , which membrane having a pore size of the order of 5 A means only small molecules that are soluble can be assimilated by the cell. Larger molecules are broken down to smaller sizes outside the cell through exogenous enzymes secreted by the microbes. The pH of the process should be within the range 6.5–8.5. Under some conditions, the ammonia present in the wastewater can be oxidized by the nitrifying bacteria to nitrite and nitrate molecules. This is known as nitrification process. If

necessary, nitrification can be achieved in the process by appropriate sludge recycling and the organic loading rate of the process. The nitrification process is not discussed in this entry, but more information can be obtained in Ref.[1]. The activated sludge process was developed in England by Ardern and Lockett in 1914 based on experiments conducted at the Lawrence Experiment Station in Lawrence, MA, in the early 1900s.[2] Presently, it is one of the most common secondary treatment processes used throughout the world. The conventional process has been modified to improve its performance. These modifications are described later.

MICROBIOLOGY OF THE PROCESS The micro-organisms present in the aeration tank of the activated sludge process are quite varied and depend on the type of wastewater being introduced and the environment of the aeration tank (e.g., temperature, pH, etc.). The predominant micro-organisms constantly change depending on the environmental conditions. The micro-organisms range from very small virus particles to much larger multicellular worms. But the predominant species are heterotrophic bacteria, with lesser amounts of autotrophic bacteria. Taxonomically, bacteria are prokaryotic protista, having mostly single-cell structure. The heterotrophic bacteria use organic compounds as their source of energy and cell carbon (electron donor), while autotrophic micro-organisms oxidize inorganic compounds for generating energy (electron donor) and use inorganic carbon (bicarbonates or CO2) as a source of cell carbon. One of the prominent autotrophic bacteria present is the nitrifying bacteria, which transforms ammonia to nitrite and finally to nitrate.[1]

Activated Sludge Process

Usually, there are large numbers of bacterial species present, making the system a mixed culture. Because the wastewater contains many different types of organic compounds, the presence of these different micro-organisms with varied metabolic capabilities enhances the possibility of degradation of these compounds. Most of the bacteria present in the activated sludge process are aerobic, meaning that they use oxygen as the ultimate electron acceptor, which produces energy for their growth and other uses. There are some facultative anaerobic bacteria present as well, which can survive in the presence or absence of oxygen. In the absence of oxygen, they can use organic compounds as the ultimate electron acceptor to form reduced organic compounds. Besides bacteria, other micro-organisms present are protozoa, fungi, and rotifers. Protozoa are eukaryotic protists. Most of the protozoa are unicellular organisms. Wastewater contains many different species of protozoa—flagellates, ciliates, amoebas, and rotifers. These organisms are predators for the bacteria and may help in flocculation and clarification in the secondary settling tank. Fungi are common in the activated sludge process operating at lower pH values. They include organisms such as yeasts and molds. They are basically saprophytic organisms feeding on organic matter. Their numbers are smaller than other species in activated sludge. In addition to these organisms, activated sludge may also contain nematodes (roundworms) and other worms. They play no part in the wastewater treatment process. The activated sludge occurs in the aeration tank in the form of flocs. These flocs are made up of microorganisms, inorganic and organic colloidal, and particulate matter. They are bound together in an organic matrix. Their size may vary from 50 to 1000 mm.[3] The shape depends on the materials encased, the organisms present, and the mechanical forces applied inside the tank. The floc containing mixed liquor leaves the aeration tank and normally settles out in a compact form in the secondary settling tank within 2–4 hr. Sometimes the settling of the sludge in the secondary settling tank is disrupted. The sludge is said to be a ‘‘bulking sludge.’’ One of the proposed theories on bulking suggests that the preponderance of filamentous bacteria can cause poor settling sludge. It has been suggested that in a good settling sludge floc, the floc forming bacteria and filamentous bacteria are present in balanced numbers to give a compact sludge mass. In the floc, the filamentous bacteria form a backbone that provides its structure and strength, while the floc forming bacteria grow around the filamentous types. The gelatinous matrices of the floc forming bacteria, sometimes known as Zoogloea bacteria, entrap other micro-organisms, colloidal and particulate matter to give the floc its shape. The preponderance of

13

filamentous bacteria over floc forming bacteria could cause an unbalanced situation and hence poor settling of the resulting floc. The filamentous bacteria found in activated sludge are Thiotrix spp., Nocardia spp., Sphaerotilus natans, Beggiatoa.[4]

KINETICS AND DESIGN EQUATIONS FOR CONVENTIONAL PROCESSES The design of conventional biological wastewater treatment processes depends on the reaction rates of the metabolism of organic matter by the microorganisms present. They use a part of the substrate (organic matter) for cell growth and the balance to produce energy to satisfy the cell needs. The reactor (tank) hydraulics is also an important factor in the design of the process. There are two idealized flow patterns that are considered in a suspended growth reactor—completely mixed and plug-flow conditions. In the completely mixed reactor, the influent flow is instantaneously mixed with the reactor content such that the concentration of the organic matter in terms of (BOD or COD) is the same throughout the reactor and in the effluent. In the plug-flow reactor, there is no such mixing in the longitudinal direction, but right angles to the flow there is complete mixing in the reactor section. Thus, there is a substrate concentration gradient longitudinally along the reactor as the wastewater organic matter is metabolized by the microorganisms. From reactor engineering, it can be shown that for the same influent substrate concentration, reaction rate, and set removal efficiency, plug-flow reactor will have a lower volume than a completely mixed reactor. Thus, for most municipal wastewater treatment applications, where the wastewater does not contain toxic ingredients, plug-flow reactors have been most commonly used. However, there are instances where a completely mixed reactor may be advantageous. In situations where periodically hazardous or toxic wastes are present in the influent wastewater, the entry of such a waste to the reactor causes an immediate reduction of concentration of the toxic component because of dilution with the entire tank content. This may reduce the adverse impact of the toxic component to the micro-organisms present with no significant impact on their waste treatment ability. If such a waste were introduced to a plug-flow reactor, it would cause an immediate toxic effect on the micro-organisms at the head end of the plant causing progressive process failure. In addition, the oxygen uptake rates throughout the completely mixed reactor are the same, which makes the design of the aeration process simpler. In a plug-flow reactor, the oxygen uptake rate is higher at the head end and decreases as the wastewater proceeds

A

14

Activated Sludge Process

down the reactor length. This may cause an unbalanced oxygenation system in the tank, especially if the aeration devices are equally spaced throughout the tank length. Higher influent substrate concentration at the head end of the plug-flow reactor also favors the floc forming bacteria over the filamentous type in the reactor, which helps in the settling of these micro-organisms later in the settling tank.[4]

It should be noted that substrate S is expressed as BOD or soluble COD (sCOD). The sCOD is obtained by using a wastewater sample that has been filtered through a 0.45 mm membrane filter. The inverse of the term on the left-hand side of Eq. (1) is known as solid retention time (SRT) or sludge age, yc: yc ¼

Completely Mixed Reactor with Recycle Using a mass balance on biomass X and substrate S around the reactor in Fig. 2 under steady state conditions (dX=dt ¼ 0 and dS=dt ¼ 0), one can arrive at the following equation:

VX ðQ
Qw ÞXe þ Qw Xr

ð2Þ

The specific growth rate constant m is affected by the substrate concentration S. The relationship between S and m is given by the following equation named after Monod:[1] m ¼

mm S Ks þ S

ð3Þ

where

ðQ
Qw ÞXe þ Qw Xr m ¼ VX

ð1Þ

mm is the maximum specific growth rate coefficient (time
1); S is the substrate concentration (BOD or sCOD) in the reactor (g=m3);

where m is the specific growth rate of cell mass ðtime
1 Þ ¼ dX=dt X ; 3

V is the aeration tank (reactor) volume (m ); 3

and Ks is the half-velocity constant (g=m3).

Q is the wastewater influent flow rate (m =day); Qw is the sludge waste rate from the settling tank bottom (m3=day); X is the biomass concentration in the aeration tank (g VSS=m3); Xe is the concentration of biomass in the effluent (g VSS=m3); and Xr is the concentration of biomass in the return sludge (g VSS=m3).

Ks represents the substrate concentration at half the maximum specific substrate utilization rate, mm. Ks is a measure of the affinity of the micro-organism to the substrate. The lower the Ks value, the greater is the affinity of the organism to the substrate. When two organisms are competing for the same substrate in a limiting substrate condition, the organism with lower Ks value will have more success in growing. It should be recognized that with a mixed culture containing mixed substrate as in typical wastewater, the Monod

Fig. 2 Completely mixed activated sludge process schematic diagram. (Note: The aeration tank is a completely mixed reactor.)

Activated Sludge Process

15

equation described here gives only an approximation of the process kinetics. Another term used is the food to micro-organism ratio (F=M), which is given by the equation: F QSo ¼ M VX

The observed yield coefficient, especially at low growth rates, is less than that given by Eq. (9). This is because of the need for maintenance energy by the cells for nongrowth needs. Eq. (10) provides an expression for observed yield coefficient, Yobs:

ð4Þ Yobs ¼

Y 1 þ kd yc

ð10Þ

where F=M is the food to micro-organism (biomass) ratio (g BOD=g MLVSS day); and

where kd is the endogenous respiration coefficient (g VSS=g VSS day); and

So is the influent BOD or biodegradable sCOD (g=m3). F=M ratio is related to the term U, the specific substrate utilization rate as follows: U ¼

ðF=M ÞE 100

E ¼

X ¼



So
S 100 So

ð6Þ

By substituting the expressions for F=M and E, Eq. (5) can be rearranged as: U ¼

The reactor biomass (MLVSS) X is given by the following equation:

ð5Þ

E is the % BOD or sCOD removal efficiency in the process


yc is the solid retention time, or sludge age (time
1).

ðSo
SÞQ VX

ð7Þ

yc Y ðSo
SÞQ yc Y ð S o
S Þ ¼ V ð 1 þ k d yc Þ t ð 1 þ k d yc Þ

ð11Þ

where V=Q ¼ t is the hydraulic retention time (hr). The relationship between yc and U is shown in Eq. (12): 1 ¼ YU
kd yc

ð12Þ

If the process yc has been selected, then U will have a fixed value as Y and kd are constants. Substituting U from Eq. (8) we get:

U is also given by the equation: U ¼




 dS . kS X ¼ dt Ks þ S

ð8Þ

dS kSX ¼ dt Ks þ S

S ¼

where k is the maximum substrate utilization rate (g substrate (BOD)=g biomass day). The amount of cell synthesis or growth can be related to the substrate removed by the microorganisms. Cell yield Y is given by (g biomass produced=g substrate utilized). dX=dt m ¼ m dS=dt k

ð13Þ

The reactor substrate concentration (also the effluent substrate concentration) S is given by the equation:

because

Y ¼

1 YkS
kd ¼ yc Ks þ S

ð9Þ

where Y is the yield coefficient (g VSS=g BOD or sCOD).

Ks ð1 þ kd yc Þ yc ðYk
kd Þ
1

ð14Þ

The production of excess biomass or waste activated sludge per day can be calculated from the following equation: Px



Y ðS o
S Þ ¼ Q þ Xinv þ Xivn 1 þ k d yc



ð15Þ

This is the amount of excess sludge that is formed by the conversion of soluble and colloidal organic matter to settleable sludge (biomass) in the aeration tank. This sludge has to be properly treated and disposed of. Typical sludge treatment may consist of thickening,

A

16

Activated Sludge Process

stabilization (anaerobic digestion), followed by land application.[1] In Eq. (15) the last two terms, Xinv and Xivn, are the nonvolatile (inorganic) suspended solids and influent volatile nonbiodegradable solids, respectively, entering the secondary process. These solids pass through the process unchanged. The amount of RAS from the secondary settling tank can be obtained from the following equation:
 1 Q Xr 1 þ R
R ¼ yc V X

ð16Þ

1 YkSo ¼
kd yc m Ks þ So

where R ¼ sludge recycle ratio ¼ Qr=Q; 3

Qr is the rate of settled sludge recycle (m =day); Xr is the concentration of settled sludge in secondary settling tank (g=m3) ¼ concentration of recycled sludge (g=m3). The poor settling of the sludge in the secondary settling tank causes Xr to be lower, which requires a higher sludge recycle ratio to maintain the same mixed liquor concentration in the aeration tank. The oxygen required for the biooxidation of organic matter can be estimated from Eq. (17): Ro ¼ QðSo
SÞ
1:42Px

Process stability depends on the process solid retention time (yc). For a particular influent wastewater quality, reactor configuration, and a given microorganism community in the reactor, Y, Ks, kd, and k are relatively constant. For a selected process U value, the yc becomes fixed. If the operating yc is below a specific value, the process becomes unstable and biomass washout may occur. This minimum yc value (ycm ) can be determined from Eq. (13) by putting S ¼ So, the influent substrate concentration indicating no wastewater treatment:

ð17Þ

where Ro is the mass of oxygen required for oxidation of organic matter per day (kg=day). The factor 1.42 represents the oxygen equivalent of cell mass produced. For only BOD removal, the oxygen requirements can vary from 0.8 to 1.3 kg O2=kg BOD removed for most conventional activated sludge processes.

ð18Þ

In most situations, Ks is much smaller than So and can be ignored. Hence, Eq. (18) becomes: 1  Yk
kd ¼ mm
kd yc m

ð19Þ

Activated sludge treatment processes should not be designed with yc less than ycm . The details about the derivation of the equations in the preceding section can be found in several studies, notably Metcalf & Eddy.[1] Plug-Flow Reactor with Recycle: The flow schematic of a plug-flow reactor is shown in Fig. 3. The modeling of a plug-flow reactor is mathematically more challenging than the completely mixed activated sludge process. By assuming biomass change to be negligible compared to the total amount present (i.e., MLVSS in the tank is constant), the integration of the substrate mass balance equation yields:[5]
 Si mm Xav V ðSi
SÞ þ Ks ln ¼ Y ð 1 þ RÞ S

ð20Þ

Fig. 3 Plug-flow activated sludge process schematic. (Note: The aeration tank is a plug-flow reactor with an average biomass concentration of Xav and effluent substrate concentration of S.)

Activated Sludge Process

17

where Si is the concentration of substrate in the aeration tank after mixing with the recycle sludge flow, So þ RS Si ¼ 1 þ R

ð21Þ

where Xav is the average biomass concentration in the tank (g=m3); and S is the concentration of substrate in the effluent from the aeration tank (g=m3).

dM dC ¼
Dl A dt dyf

M is the mass of oxygen transported (g); ð22Þ

Dl is the diffusion coefficient for oxygen in water (m2=time);

where Xi is the concentration of biomass in the aeration tank after mixing with the recycle sludge flow, Xi ¼

Xo þ RXr 1 þ R

A is the cross-sectional area through which O2 transport occurs (m2); and

ð24Þ

yc can be approximated by the following equation: 1 YkðSo
SÞ
kd ¼ yc ðSo
SÞ þ ð1 þ RÞKs lnðSi =SÞ

C is the dissolved oxygen concentration (g=m3);

ð23Þ

where Xo is the concentration of biomass in the influent flow (g MLVSS=m3) ¼ usually 0. Eq. (23) becomes: RXr Xi ¼ 1 þ R

ð26Þ

where

The small change in biomass is given by: kd Xav V X
Xi ¼ Y ðSi
SÞ
Q ð 1 þ RÞ

theory, oxygen mass to the liquid is transported through air and water films at the interface because of the concentration gradients. For gases with low solubility such as oxygen, the diffusion through the liquid film is the rate limiting step.[5] Fig. 4 shows the oxygen concentration gradients at the water interface during mass transport operation. The mass transport process can be expressed by Fick’s equation:

yf is the liquid film thickness (m). As the liquid film thickness is quite small, the differential quantity dC=dyf can be replaced by linear approximation of the concentration gradient as: dC Cs
C  dyf yf

ð27Þ

ð25Þ

In actual practice, a truly plug-flow or completely mixed-flow regime in a reactor is not attained because of longitudinal dispersion and nonideal mixing conditions. The equations reported here approximate the actual conditions in the field.

AERATION AND MIXING REQUIREMENTS The aeration is a necessary part of the activated sludge treatment process as it supplies the dissolved oxygen (DO) needed for the biooxidation of the organic matter. If a liquid is unsaturated with respect to DO concentration, the natural diffusion process would transport enough oxygen from the air to bring the liquid to saturation value. This transport is based on Fick’s law of diffusion. The oxygen transfer to the liquid phase is best described by the two-film theory. According to this

Fig. 4 Two-film oxygen mass transfer theory at liquid interface. (Note: Pg ¼ partial pressure of oxygen in bulk gas phase; Pi ¼ partial pressure of oxygen at the interface; Cs ¼ saturation concentration of dissolved oxygen at the interface; C ¼ dissolved oxygen concentration in bulk liquid.)

A

18

Activated Sludge Process

where Cs is the saturation dissolved oxygen concentration at the interface layer (g=m3); and C is the dissolved oxygen concentration in the bulk liquid (g=m3). Dividing both sides of the Eq. (26) by V, the volume of the liquid in the container, and substituting the dC=dyf from Eq. (27), we get: dM dC Cs
C ¼ ¼
Dl A V dt dt Vyf

ð28Þ

As yf cannot be measured easily, it is combined with Dl to form another constant: Kl ¼

Dl ðm/timeÞ yf

measure the rate of aeration. The Kla can be determined from the slope of a plot of ln(Cs – C) vs. time. The value of Cs, the DO saturation, for the temperature and dissolved solid concentrations of the test water can be obtained from several studies including Ref.[1]. For extrapolation of the clean water Kla to field conditions with wastewater, some corrections must be applied. Wastewater constituents (especially surface active agents) can reduce the field Kla value. A factor a can be used to make a correction to the measured clean water Kla: a ¼

b ¼

ð29Þ

The area of the bubbles through which the oxygen mass transfer takes place is quite difficult to measure. Hence, the parameters A=V are combined and represented by a, the specific surface area. Eq. (28) is simplified as: dC ¼ Kl aðCs
C Þ dt

ð32Þ

Another factor b corrects for the difference in DO saturation value of clean water and wastewater:

Eq. (28) becomes: dC A ¼
K1 ðCs
CÞ dt V

K1 awastewater K1 acleanwater

Cswastewater Cscleanwater

ð33Þ

Eq. (34) shows a temperature correction factor for Kla Kl aT 0 ¼ Kl a200 yðT
20Þ

ð34Þ

where y is the temperature correction factor for reaeration, with a usual value of 1.024.

ð30Þ

KlaT  is the overall mass transfer coefficient at T  C (time
1); and

where Kla is the overall mass transfer coefficient (time
1). Under turbulent conditions in the bulk liquid, Eq. (30) can be integrated with boundary conditions: at time ¼ t0, C ¼ Co; at time ¼ t, C ¼ C:

Kla20 is the overall mass transfer coefficient at 20 C (time
1).

K1 a ¼

ln½ðCs
Co Þ=ðCs
CÞ t
t0

The standard oxygen transfer rate (SOTR) in clean water can be calculated by knowing the average Kla in the tank at 20 C and at zero DO concentrations:

ð31Þ

The coefficient Kla is a good measure of the efficiency of an aerator. It depends on many factors, such as temperature, wastewater composition, tank geometry, and turbulence. Measurement of Kla of aeration devices in clean water can be determined by standard procedures developed by the American Society of Civil Engineers.[6] In the standard procedure, the clean water (usually tap water) in a tank is deoxygenated by adding a reducing agent, sodium sulfite, and a catalyst, cobalt chloride. Thereafter, aeration is resumed with DO being measured periodically at several points in the tank to

SOTR ¼ V ðK1 aav Csav Þðkg O2 =hrÞ

ð35Þ

where V is the volume of the aeration tank; Klaav is the average Kla value in the tank (time
1); and Csav is the average DO saturation value in the tank (g=m3); The field oxygen transfer rate (OTR) in wastewater can be estimated from the SOTR by applying the

Activated Sludge Process

19

appropriate correction factors:


 bCs
C OTR ¼ a 1:024ðT
20Þ SOTR Cs;20

ð36Þ

where Cs is the DO saturation in clean water in the aeration tank at the prevailing pressure and temperature (g=m3); C is the operating DO concentration (g=m3); Cs,20 is the DO saturation in clean water at 20 C (g=m3). Details about different types of aeration system for the activated sludge process can be found in Metcalf and Eddy.[1] Mixing of the mixed liquor is also accomplished by the aeration system. The mixed liquor solids must be kept in suspension for proper operation of the process. The mixing requirement depends on types of aeration equipment, depth, and width of the tank. Typically, a mixed liquor velocity of 0.15 m=sec in the aeration tank provides adequate mixing. For mechanical aerators, power input of 16–30 W=m3 is often specified for mixing of biomass in the tank.[2]

DESIGN PARAMETERS The design of an activated sludge system requires the determination of the following items: aeration tank volume V, oxygen requirement, daily sludge wasting rate, and sludge recycle rate R. The selection of the process yc value usually depends on past experience rather than kinetic considerations, because use of kinetic equations gives a much smaller

yc value.[2] The yc is selected to provide mixed liquor that settles well in the settling tank. For a conventional activated sludge process, it varies from 5 to 15 days[7] (see Table 1). Once yc has been selected, the tank volume can be calculated from Eq. (11) for a completely mixed reactor or from Eq. (21) for a plug-flow reactor. The constants Y, kd, k, and Ks must be estimated for calculating the volume V, and X must also be selected. Typically, X in the aeration tank varies from 1500 to 3000 mg MLVSS=L. The selection of the MLVSS in the aeration tank depends on the influent BOD load and desired U or yc values. The oxygen requirement for carbonaceous BOD removal can be estimated from Eq. (17). The aeration equipment can be selected to supply the needed oxygen based on the manufacturer’s information. The sludge mass wasted per day can be calculated from Eq. (15). The volume of the wasted sludge will depend on the settled sludge concentration and specific gravity of the settled solids. The sludge recycle rate R depends on the settled sludge concentration as shown in Eq. (16). If the settled sludge concentration is low, more sludge has to be recycled, making R greater. Conventional activated sludge process loading rate in terms of F=M ratio (mass loading; g BOD=g MLVSS day) varies from 0.2 to 0.6. In some instances, the loading rate is based on the tank volume. The volumetric loading rate in terms of BOD applied per unit tank volume varies from 0.3 to 0.6 kg BOD=m3 day.[7] The hydraulic residence time (V=Q) varies from 4 to 8 hr.

PROCESS MODIFICATIONS Over the years, the conventional activated sludge process has been modified to improve or to suit a specific operational condition. Some of the more common modifications are extended aeration process, contact

Table 1 Design and operating parameters for activated sludge process and its modifications Hydraulic retention time, t (hr) Conventional plug-flow

F/M (kg BOD/ kg VSS day)

Solid retention time, hc (day)

Volumetric loading rate (kg BOD/m3 day)

MLSS (mg/L)

4–8

0.2–0.4

5–15

0.3–0.6

1500–3000

Extended aeration

18–36

0.05–0.15

20–30

0.1–0.4

1500–5000

Contact stabilization

0.5–1.0a

0.2–0.6

5–15

1.0–1.2

1000–3000a

2–4

—

—

—

4000–9000b

Step aeration

3–5

0.2–0.4

5–15

0.6–1.0

2000–3500

Tapered aeration

4–8

0.2–0.4

5–15

0.3–0.6

1500–3000

High rate

2–4

0.4–1.5

5–10

1.6–16

3000–6000

Pure oxygen

1–3

0.25–1.0

3–10

1.6–3.3

3000–8000

b

a

In contact tank. In stabilization tank. (From Ref.[7].) b

A

20

stabilization process, step aeration, tapered aeration, high rate, and pure oxygen process. Extended Aeration Process: This process is very much like the conventional process but the loading rates are much lower. The sludge age is high and the hydraulic residence time is higher than the conventional process. These conditions result in a better effluent quality, with some nitrification of the wastewater. In addition, the process is quite stable under varying loading rates, producing good settling sludge. The design parameters for this process are given in Table 1. The high hydraulic residence time increases the capital and operating cost for the process. A modified configuration of this process is known as oxidation ditch, where the mixed liquor is moved around a racetrack style reactor. The movement of the wastewater and aeration along the ditch is facilitated by a brush type or vertical rotor aerator, which ensures a fluid velocity of about 0.3 m=sec so that the mixed liquor solids are not settled out in the channels. Contact Stabilization Process: In this modification of the conventional process, the waste is contacted in the aeration tank with the return sludge (which has been previously aerated) for a relatively short time, about 30–60 min. The mixed liquor is then separated in a settling tank, with the settled sludge being aerated in a separate tank for 3–6 hr before it is returned to the aeration tank (see Fig. 5). The short contact time allows the colloidal and particulate waste constituents to adsorb onto the micro-organism flocs. The stabilization of the adsorbed organic matter occurs when the sludge is reaerated. By following this scheme, the overall tank volume requirement reduces by about 50%.[5] The process is only successful where a large fraction of the influent BOD is in colloidal or particulate form, but for normal domestic wastewater it does not provide equivalent secondary treatment effluent quality. Step Aeration: In this process, the influent feed is introduced into the plug-flow reactor at two or more

Activated Sludge Process

points, which distributes the organic load along the length of the tank (Fig. 6). Thus, the oxygen uptake rate along the tank length becomes more uniform rather than high at the start and low at the end as in the conventional system. This gives a better performance during most operating conditions. All other design parameters are the same as in the conventional system as seen in Table 1. Tapered Aeration: This process corrects the problem of unbalanced aeration supply in a plug-flow conventional activated sludge system by providing more diffusers at the head end, which decrease progressively along the length of the tank as the BOD concentration decreases. The loading rates are the same as in a conventional system (Table 1). This arrangement reduces blower capacity and operating costs, and provides a greater degree of operational flexibility.[8] Fig. 7 shows a schematic of the process. High Rate: This process is characterized by a shorter hydraulic retention time of the mixed liquor in the aeration tank and a higher loading rate than the conventional process. Consequently, the effluent leaving the process is not as high a quality as in the conventional process, i.e., it has a higher BOD and suspended solids. It often precedes a second-stage nitrification process. Pure Oxygen Process: This process uses compressed pure oxygen instead of air, resulting in increased DO in the mixed liquor. Advantages include reduced power for oxygen diffusion, faster rate of organic matter stabilization, better settling sludge, and the ability to treat higher BOD wastewater. The process uses covered, completely mixed tanks in three or four stages with oxygen gas and wastewater entering at the head end. Each stage is mixed with a surface aerator. The exiting offgas contains only about 10% oxygen as the rest is used up in the biochemical reactions inside the reactors. In recent practice, the MLSS concentrations in the tanks vary from 1000 to

Fig. 5 Contact stabilization process.

Activated Sludge Process

21

A

Fig. 6 Step aeration process.

3000 mg=L with DO levels typically in the range of 4–10 mg=L.[9] The hydraulic retention in the aeration tank is usually 1–3 hr. The loading rate for this process is higher than the conventional process, which requires less aeration tank volume than the conventional process for comparable wastewater quality. The oxygen needed for the process has to be produced in situ, which increases the construction cost for the process. The common methods for producing oxygen gas are cryogenic process, pressure swing adsorption, and vacuum swing adsorption. Details of these processes can be found in Ref.[9]. There have been debates whether these processes have significant advantages over the conventional process. The consensus is that the advantages, if any, are marginal at best.[5] Nitrification of the wastewater may occur as a result of higher DO levels. The process does create foaming because of Nocardia accumulation in the aeration tank. It can also depress the pH of the mixed liquor as the CO2 formed is not stripped by nitrogen gas as with the air diffusion system. The construction materials for the

tank and accessories have to be selected properly as the atmosphere inside is more corrosive.[2]

OPERATIONAL PROBLEMS The activated sludge plant often has operational problems, which could be attributable to unusual characteristics of the influent wastewater or could be because of improper design and operation. These problems can be characterized by two factors: low soluble BOD removal and poor settling of solids in the secondary settling tank. The low BOD removal could be caused by many reasons such as higher influent organic loading, influx of toxic or inhibitory chemicals, change of pH in the aeration tank beyond the acceptable range 6.5–8.5, insufficient aeration, and insufficient biomass in the aeration tank. An increased F=M ratio beyond the process design value caused by higher influent BOD concentration

Fig. 7 Tapered aeration process.

22

could eventually cause low BOD removal. Increased sludge recycle rate will increase process microorganism concentration and balance the F=M ratio, which will correct the poor performance results. Variable and shock influent organic loading rates could also cause poor BOD removals. An equalizing tank at the head end could minimize the effects of shock or variable BOD influent conditions. Equalizing tank equalizes the diurnal flow variations by providing storage for the excess flow during peak flow hours and a constant withdrawal throughout the day. Toxic and inhibitory chemicals come as a result of a spill or temporary problems at chemical or other industrial plants discharging effluents into the sewer system. Sewer use ordinance prohibits the discharge of these materials; but in case of accidents these chemicals can inhibit the activities of the micro-organisms in the process. If detected early the wastewater containing the toxic wastes can be diverted to a holding tank, if available, and the unaffected micro-organisms can be nurtured under proper environment till nontoxic wastewater flow is resumed. In some cases where most of the micro-organisms have been destroyed, later external seed sludge may need to be imported to make the process restart. Accidental discharge of acidic or basic wastes from industries may change the wastewater pH beyond the acceptable range of 6.5–8.5. Prolonged operation of the process at these pH values may affect the performance of the plant. Neutralizing chemical may be needed to correct the pH to a safer range. In a conventional activated sludge plug-flow plant with diffused aeration, the inlet end has a higher oxygen uptake rate than at the end. In cases where the influent BOD load exceeds the oxygen supply, oxygen deficiency may occur and the effluent performance could suffer. The effluent could be high in suspended solids as well as BOD. This type of condition can be remedied by changing the aeration system to the tapered aeration process mentioned earlier or changing the feed introduction by following the step aeration process. In addition, if the air blower capacity is limited, it can be replaced by a larger unit. Insufficient biomass in the aeration tank could occur if the return sludge pumps do not have enough capacity to supply the increased sludge flow needed at higher influent BOD loading conditions. It could also occur if the settled sludge has a low concentration because of poor settling properties of the sludge, which would require a much higher sludge return flow rate. Addition of larger sludge return pump could help this situation to some extent but may not be able to cope with poor settling sludge conditions. The problem of poorly settling sludge in the secondary settling tank could be caused by two separate

Activated Sludge Process

conditions. The first, known as ‘‘bulking sludge,’’ is the most common problem of the activated sludge plants. The sludge does not settle well in the tank, giving a low solids concentration to the settled sludge. This requires a much larger amount of the settled sludge recycle rate to maintain a set MLSS level in the aeration tank. In some extreme cases where the sludge is too bulky, no amount of sludge recycling can maintain the process performance. The escape of unsettled sludge through the settling tank weirs can cause higher than the required BOD and suspended solids concentrations. Common factors that cause bulking are nutrient (N and P) deficiency, addition of septage to the influent, insufficient aeration capacity, low pH conditions, and influx of toxic wastes. The nutrient deficiencies of the influent wastewater can be fixed by adding appropriate amounts of nitrogen and phosphorus compounds. A preaeration system included for the septage before it enters the aeration tank could reduce its adverse impacts. Adding additional aeration capacity or modifying the aeration system to tapered aeration could reduce the problems of bulking sludge caused by insufficient aeration capacity. Low pH could occur from the entry of some industrial acidic wastes with the influent wastewater. This can cause the growth of fungi, which are filamentous, and bulking. Proper pH control can overcome this problem. Inadvertent influx of toxic wastes to the plant could also cause a change in the biota of the process and eventually cause bulking. The control of entry of such wastes can correct this problem. In some situations where the bulking is caused by the growth of filamentous organisms, the addition of an oxidizing agent such as chlorine or hydrogen peroxide can selectively reduce their numbers and solve the problem. Chlorine doses from 0.1 to 2.5 g Cl2=kg of returned sludge dry mass have been successful in controlling sludge bulking.[8] The other sludge settling problem is called ‘‘rising sludge.’’ In this case, settled sludge flocs tend to float up to the top giving it the name. This occurs in situations where the wastewater is nitrified to a great extent with nitrate-N present in the liquid phase. The environment at the bottom of the settling tank is suitable for denitrification of the nitrate molecules, i.e., anoxic conditions with organic carbon available from sludge deposits. Under these conditions, nitrogen gas bubbles are formed from nitrate molecules that attach onto the sludge flocs to float them to the top. By reducing the length of time the settled sludge stays at the settling tank bottom and by increasing the sludge recycle rate to the aeration tank, denitrification can be reduced in the settling tank.[8] In addition, if nitrification is not needed, increased organic loading rate to the aeration tank can reduce nitrification and remove the rising sludge problem.

Activated Sludge Process

CONCLUSIONS Activated sludge process is one of the most common secondary treatment processes available for treating wastewater. It depends on the microbial metabolism of soluble and colloidal organic matter in the presence of dissolved oxygen and nutrients. The resulting biomass from the process is subsequently settled in a settling tank and a portion of the settled biomass (sludge) is recycled back to the aeration tank. Excess biomass from the process is further treated before disposal. The process in combination with the primary treatment can remove up to 90þ% of the incoming carbonaceous BOD and suspended solids. Under some conditions, it can also convert influent ammonia to nitrate, i.e., nitrification. The reactors used for the aeration process could be plug-flow or completely mixed type. In most cases, plug-flow configuration is used. The design of the reactor depends on the biokinetic parameters of the mixed micro-organisms developed in the aeration tank for the type of wastewater entering the system. The design of the system requires the determination of the following items: aeration tank volume, oxygen requirement, daily sludge wasting, and recycling rates. Many modifications for the conventional process have been proposed for improving the system or for treating a specific type of wastewater. These modified processes are extended aeration, contact stabilization, step aeration, tapered aeration, high rate, and pure oxygen process. Operational problems to activated sludge process can cause low soluble BOD removal and poor settling of the sludge in the secondary settling tank. The reasons for these problems could be because of unusual

23

characteristics of the influent wastewater, or improper design and operation.

REFERENCES 1. Metcalf & Eddy, Inc. Wastewater Engineering: Treatment and Reuse, 4th Ed.; McGraw-Hill: New York, 2003. 2. Water Environment Federation. Wastewater Treatment Plant Design; Vesilind, P.A., Ed.; Water Environment Federation: Alexandria, VA, 2003. 3. Ha¨nel, K. Biological Treatment of Sewage by Activated Sludge Process; Ellis Horwood Ltd.: Chichester, U.K., 1988. 4. Horan, N.J. Biological Wastewater Treatment Systems; John Wiley & Sons: Chichester, U.K., 1990. 5. Sundstrom, D.W.; Klei, H.E. Wastewater Treatment; Prentice-Hall, Inc.: Englewood Cliffs, NJ, 1979. 6. American Society of Civil Engineers. Measurement of Oxygen Transfer in Clean Water; ANSI=ASCE Standard 2-91, 2nd Ed.; American Society of Civil Engineers: Reston, VA, 1992. 7. Droste, R.L. Theory and Practice of Water and Wastewater Treatment; John Wiley & Sons: New York, 1997. 8. Gray, N.F. Activated Sludge—Theory and Practice; Oxford University Press: Oxford, U.K., 1990. 9. American Society of Civil Engineers (ASCE). Design of Municipal wastewater Treatment Plants, 4th Ed., ASCE Manual and Report on Engineering Practice No. 76; American Society of Civil Engineers: Reston, VA, 1998; Vol. 2.

A

Adsorption A Shivaji Sircar Chemical Engineering Department, Lehigh University, Bethlehem, Pennsylvania, U.S.A.

INTRODUCTION The separation and purification of fluid mixtures (gas or liquid) by adsorption is a major unit operation in the chemical, petrochemical, environmental, pharmaceutical, and electronic gas industries. A list of the key commercial applications of this technology is given in Table 1.[1] The phenomenal growth in the development of this technology is demonstrated by Fig. 1, which shows a year-by-year tally of U.S. patents issued between 1980 and 2000 on five different topics of adsorption.[1] The total number of patents is overwhelming.

ADSORPTION AS A SEPARATION PROCESS Adsorption is a surface phenomenon. When a multicomponent fluid mixture is contacted with a solid adsorbent, certain components of the mixture (adsorbates) are preferentially concentrated (selectively adsorbed) near the solid surface creating an adsorbed phase. This is because of the differences in the fluid–solid molecular forces of attraction between the components of the mixture. The difference in the compositions of the adsorbed and the bulk fluid phases forms the basis of separation by adsorption. It is a thermodynamically spontaneous process, which is exothermic in nature. The reverse process by which the adsorbed molecules are removed from the solid surface to the bulk fluid phase is called desorption. Energy must be supplied to carry out the endothermic desorption process. Both adsorption and desorption form two vital and integral steps of a practical adsorptive separation process where the adsorbent is repeatedly used. This concept of regenerative use of the adsorbent is key to the commercial and economic viability of this technology. Three generic adsorptive process schemes have been commercialized to serve most of the applications shown in Table 1. They include 1) temperature swing adsorption (TSA); 2) pressure swing adsorption (PSA); and 3) concentration swing adsorption (CSA).[2–9] The fluid mixture (feed) to be separated is passed over a regenerated adsorbent (contained in an adsorber vessel) to produce a stream enriched in the less strongly adsorbed components of the mixture, Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120007656 Copyright # 2006 by Taylor & Francis. All rights reserved.

followed by desorption of the adsorbed components, which produces a stream enriched in the more strongly adsorbed components of the mixture. The TSA processes are generally designed for removal of trace impurities from a mixture (gas or liquid), where the desorption is effected by heating the adsorbent. The PSA processes are designed for separation of the components of a bulk gas mixture or for removal of dilute impurities from a gas stream, where the desorption is effected by lowering the gas phase partial pressure of the adsorbed components within the adsorber. The CSA processes are designed for separation of bulk liquid mixtures, where the desorption is effected by flowing a less selectively adsorbed liquid (eluent or desorbent) over the adsorbent. Numerous variations of these processes have been developed to achieve different separation goals by using 1) different modes and conditions of operation of the adsorption and the desorption steps in conjunction with a multitude of other complementary steps (designed to improve separation efficiency and product quality); 2) different types of adsorbents; 3) different process hardware designs; 4) different process control logic, etc. Several families of micro- and mesoporous adsorbents offering a spectrum of adsorption characteristics are also available for these separations. Consequently, the technology has been a very versatile and flexible separation tool, which provides many different paths for a given separation need. This availability of multiple design choices is the driving force for innovations.[10] Commercial success, however, calls for a good marriage between the optimum adsorbent and an efficient process scheme. Several emerging concepts in this field can potentially expand its scope and scale of application. These include 1) rapid PSA processes; 2) novel adsorber configurations; 3) use of reversible chemisorbents; 4) adsorbent membranes; 5) simultaneous sorption and reaction, etc.[10] The design and optimization of adsorptive processes typically require simultaneous numerical solutions of coupled partial differential equations describing the mass, heat, and momentum balances for the process steps. Multicomponent adsorption equilibria, kinetics, and heat for the system of interest form the key fundamental input variables for the design.[11,12] Bench- and pilot-scale process performance data are generally needed to confirm design calculations. 25

26

Adsorption

Table 1 Key commercial applications of adsorption technology Gas separation Gas drying Trace impurity removal Air separation Carbon dioxide–methane separation Solvent vapor recovery Hydrogen and carbon dioxide recovery from steam-methane reformer off-gas Hydrogen recovery from refinery off-gas Carbon monoxide–hydrogen separation Alcohol dehydration Production of ammonia synthesis gas Normal–isoparaffin separation Ozone enrichment Liquid separation Liquid drying Trace impurity removal Xylene, cresol, cymene isomer separation Fructose–glucose separation Fatty chemicals separation Breaking azeotropes Carbohydrate separation Environmental separation Municipal and industrial waste treatment Ground and surface water treatment Air pollution control VOC removal Mercury vapor removal Bioseparation and pharmaceutical separation Recovery of antibiotics Purification and recovery of enzymes Purification of proteins Recovery of vitamins Separation of enantiomers of racemic compounds Removal of micro-organisms Home medical oxygen production Electronic gas purification Production of ultrahigh-purity N2, Ar. He, H2, O2 Purification of fluorinated gases NF3, CF4, C2F6, SiF4 Purification of hydrides NH3, PH3, ASH3, SIH4, Si2H6

ADSORBENT MATERIALS A key factor in the development of adsorption technology for the fluid separation has been the availability of appropriate adsorbents. The most frequently used categories include crystalline materials like zeolites, and amorphous materials like activated carbons, silica and alumina gels, polymeric sorbents, and ion-exchange resins. These materials exhibit a large spectrum of pore structures (networks of micro- and mesopores of different shapes and sizes) and surface chemistry (degrees of polarity), which provide a large choice of core adsorptive properties (equilibria, kinetics, and heat) to be utilized in

Fig. 1 U.S. patent survey of adsorption topics.

the design of the separation processes. Table 2 lists some of the physical properties of common adsorbents. The microporous alumino-silicate zeolites (Types A, X, and mordenite are frequently used) provide a vari˚ ), cavity and channel sizes, ety of pore openings (3–10 A and framework Si=Al ratios. They are also available in various cationic exchanged forms (Na, K, Li, Ag, Ca, Ba, Mg), which govern their pore openings and cationic adsorption site polarities. They are highly hydrophilic materials and must be dehydrated before use. The amorphous adsorbents contain an intricate network of micropores and mesopores of various shapes and sizes. The pore size distribution may vary over a wide range. The activated carbons and the polymeric sorbents are relatively hydrophobic in nature. The silica and alumina gels are more hydrophilic (less than zeolites) and they must also be dehydrated before use. Commercial adsorbents are generally produced in bound forms (0.5–6.0 mm diameters) in regular particle shapes (beads, pellets, extrudates, granules, etc.). The purpose is to reduce pressure drops in adsorbers. Clay, alumina, polymers, pitch, etc. are used as binders, which typically constitute 10–20% (by weight) of the final product. The binder phase usually contains a network (arteries) of meso- and macropores (0.5–50.0 mm diameters) to facilitate the transport of the adsorbate

Adsorption

27

Table 2 Physical properties of some adsorbents NaX Zeolite (Bayer, Germany)b

BPL Carbon (Calgon, U.S.A.)

BET area (m2=g)

Molecular Sieve Carbon (Takeda, Japan)

H151 Alumina (Alcoa, U.S.A.)

Silica Gel (Grace, U.S.A.)

—

1100

—

350

800

Pore volume (cm3=g)

0.54

0.70

0.43

0.43

0.45

Bulk density (g=cm3)

0.65

0.48

0.67

0.85

0.77

˚) Mean pore diameter (A

7.4a

30

3.5

43

22

a

Crystal pore aperture size. Manufacturer given in parentheses.

b

molecules from the bulk fluid phase to the adsorption sites (within zeolite crystals and micropores of amorphous adsorbents) and vice versa. Adsorption of fluid molecules on the binder material is generally very weak. Fig. 2 shows a schematic drawing of a bound zeolite pellet depicting the pathways for transport of the adsorbate molecules. The vast majority of fluid separation by adsorption is affected by the thermodynamic selectivity of the adsorbent for certain components of the fluid mixture over others. Physisorption is the dominant mechanism for separation. Thus, it is governed by the surface polarity of the adsorbent and the polarizability and permanent polarity of the adsorbate molecules. All adsorbate molecules, in this case, have access to the adsorption sites. The separation can also be based on a kinetic selectivity by the adsorbent where certain molecules of the fluid mixture diffuse into the adsorbent pores faster than the others because of their relative size differences. Size or steric exclusion of certain components of a fluid mixture from entering the adsorbent pores (typically for zeolites) is also possible. The last case is known as ‘‘molecular sieving.’’ Adsorbents may be energetically homogenous, containing adsorption sites of identical adsorption energy (heat of adsorption), or energetically heterogenous,

containing a distribution of sites of varying energies. The cause of adsorbent heterogeneity is generally physicochemical in nature. It is created by a distribution of micro- and mesopores of different sizes and shapes within the adsorbent particle as well as by a distribution of adsorption sites of different surface chemistry and polarity within the micropores. An adsorbent is often tailor-made to suit a separation need or a process can be designed to best fit the properties of an adsorbent. Special adsorbents are also available for specific applications (e.g., removal of mercury vapor, drying of reactive fluids, resistance to acids, etc). More recently, adsorbents have been produced that use reversible chemisorption as the mechanism for gas separation.[13] Creation of new adsorbents and modification of existing adsorbents continue to be an active area of research and development.

KEY ADSORPTIVE PROPERTIES FOR SEPARATION All practical adsorptive separation processes are carried out using a stationary packed bed (adsorber) of the adsorbent particles. Each particle is subjected to the adsorption, the desorption, and the complementary

Fig. 2 Schematic drawing of a bound adsorbent particle.

A

28

steps of the process in a cyclic fashion. The ad(de)sorption characteristics exhibited by the particle during different periods of the cycle are governed by the multicomponent adsorption equilibria, kinetics, and heat for the fluid mixture of interest under the local conditions (e.g., fluid phase pressure, temperature and composition, adsorbate loadings in the particle, and its temperature) that the particle experiences. As these conditions can vary over a wide range during a process cycle, it is imperative that those adsorptive properties be accurately known over that range for reliable process design.

ADSORPTION EQUILIBRIA Adsorption equilibria determine the thermodynamic limits of the specific amounts of adsorption (mol=g) of a pure gas or the components of a fluid mixture (gas or liquid) under a given set of conditions [pressure (P), temperature (T ), and mole function (yi or xi) of component i] of the bulk fluid phase. The simplest way to describe adsorption equilibria of pure gas i is in the form of adsorption isotherms where the amount adsorbed ðn0i Þ is plotted as a function of gas pressure (P) at a constant temperature (T ). The pure gas adsorption isotherms can have various shapes (Types I–V) by Brunauer classification depending on the porosity of the adsorbent (microporous, mesoporous, or nonporous) and the system temperature (below or above the critical temperature of the adsorbate).[9] However, the most common isotherm shape is Type I, which is depicted by most microporous adsorbents of practical use. These isotherms exhibit a linear section in the very low-pressure region (Henry’s law region) where the amount adsorbed is proportional to the gas pressure ½ðn0i Þ ¼ Ki P
: The proportionality constant is called

Adsorption

Henry’s law constant (Ki), which is a function of temperature only. The amount adsorbed monotonically increases with increasing pressure beyond the Henry’s law region with a progressively decreasing isotherm slope and finally the amount adsorbed asymptotically approaches the saturation adsorption capacity (mi) of the adsorbate. Figs. 3A and 3B show examples of Type I isotherms for adsorption of pure N2 and pure O2, respectively, on various zeolites at 25 C.[2] The figures demonstrate that N2 is more strongly adsorbed than O2 on all zeolites and their adsorption characteristics are significantly affected by the structure of the zeolite, as well as by the nature of the cation present in them. The LSX zeolites in Fig. 3 represent X zeolite structure with low Si=Al ratio. The amounts adsorbed of a pure gas at any given pressure decrease with increasing temperature because of the exothermic nature of the adsorption process. The equilibrium amounts adsorbed of component i from a binary gas mixture (ni) are generally described as functions of gas phase mole fractions (yi) at a constant system temperature (T ) and total gas pressure (P). An example is given in Fig. 4 for adsorption of binary N2–O2 mixtures on Na–mordenite at various temperatures where the total gas pressure was 1.0 atm.[2] These binary isotherm shapes are typical for Type I adsorption systems on microporous adsorbents. The relative adsorption between components i and j of a gas mixture is expressed in terms of the selectivity of adsorption (Sij ¼ niyj=njyi). Component i is more selective than component j if Sij > 1. The thermodynamic selectivity decreases with increasing T for any given values of ni. For adsorption on a homogenous adsorbent at constant T, Sij can be constant, increase, or decrease with adsorbate loading depending on the size differences between the molecules of components

Fig. 3 Pure gas adsorption isotherms for (A) nitrogen and (B) oxygen on various zeolites.

Adsorption

29

ni(t), in this case, is the amount of component i adsorbed at time t. Numerous models have been developed to describe pure and multicomponent gas adsorption on porous adsorbents. The analytical models are, however, most useful for process design. A few analytical models for Type I adsorption systems, which are thermodynamically consistent, are given below:[2] Langmuir :

bi Pyi ¼ yi =ð1  yi Þ

Multisite Langmuir : . X  ai bi Pyi ¼ yi 1  yi

ð1Þ ð2Þ

MartinezBasmadjian : n .  X  X ai o bi Pyi ¼ yi 1  yi exp  ai wi yi

ð3Þ

Toth :

Fig. 4 Binary gas adsorption isotherms for nitrogen (1) and oxygen (2) mixtures on sodium mordenite.

i and j.[14] For adsorption on a heterogenous adsorbent, Sij generally decreases with increasing adsorbate loading.[2] Table 3 gives a list of Henry’s law selectivity (Sij ¼ Ki=Kj) for several binary gas mixtures at 30 C on a zeolite and an activated carbon.[2] The firstmentioned gas of a pair is the more selectively adsorbed component. Separation of a gas mixture by a time dependent kinetic selectivity [Sij(t) ¼ ni(t)yj=nj(t)yi] has also been used in practice when there is a difference in the rates of adsorption of the components of the gas mixture.

bi Pyi ¼ yi



Gas mixture

5A Zeolite

BPL Carbon

CO2–CH4

195.6

2.5

CO2–CO

59.1

7.5

CO2–N2

330.7

11.1

CO2–H2

7400.0

90.8

CO–CH4

3.3

0.33

CO–N2

5.6

1.48

CO–H2

125.0

12.1

CH4–N2

1.7

4.5

CH4–H2

37.8

36.6

N2–H2

22.3

8.2

ð4Þ

The frequently used Langmuir model describes adsorption of equal-sized adsorbates (mi ¼ mj) on an energetically homogenous adsorbent. The multisite Langmuir model is an extension to include the effects of dissimilar adsorbate sizes (mi 6¼ mj). The Martinez–Basmadjian model is a further extension to include lateral interactions in the adsorbed phase. The Toth model is developed to describe adsorption of equal-sized molecules on an energetically heterogenous adsorbent. The variables of Eqs. (1)–(4) are the fractional coverage of component i of the gas mixture (yi ¼ ni=mi) at P, T, and yi, the number of adsorption sites occupied by the adsorbate type i (ai), the energy of lateral interactions between i molecules in the adsorbed phase (wi), the gas–solid interaction parameter for component i (bi), and the adsorbent heterogeneity parameter for all adsorbates (k < 1). The temperature coefficient of the parameter bi is given by: bi ¼ b0i expðqi =RTÞ

Table 3 Selectivities of binary gas mixtures

X k 1=k 1  yi

ð5Þ

where b0i is a constant. qi is the isosteric heat of adsorption of pure gas i in the Henry’s law region. R is the gas constant. The pure gas adsorption isotherms (y0i vs. P) for these models can be obtained by setting yi ¼ 1. The extent of specific equilibrium adsorption of component i from a liquid mixture having mole fraction xi for that component is expressed in terms of a variable called the Gibbsian surface excess [ni e (mol=g)], which is related to the actual amounts adsorbed by[15] X  nei ¼ ni  n i xi ð6Þ The surface excess of component i is equal to its actual amount adsorbed ðnei  ni Þ only when xi  1 and the

A

30

Adsorption

component i is very selectively adsorbed ðSij  1Þ over other components of the mixture.[15] The binary liquid phase surface excess adsorption isotherm (ne1 vs. x1) at a constant temperature (pressure is not a variable) on a microporous adsorbent is often Ushaped, as shown in Fig. 5A. Component 1 is selectively adsorbed if ne1 > 0. For adsorption of a dilute solute from a liquid mixture, on a microporous solid, the excess isotherm is similar in shape as Type I isotherm for vapor adsorption (as shown in Fig. 5B). Numerous analytical models have also been developed for binary liquid phase surface excess isotherms. A model equation that accounts for adsorbate size differences, bulk liquid phase nonideality, as well as a simplified description of adsorbent heterogeneity is given below:[16]

m1 ne1 ¼ ðS0H  S0L Þ



"

(

x2 a1

þ

 a2 S0H a1 ðb  1Þ SH

SH S0H

1=ðb1Þ

 SH ¼ ðSH a1 þ a2 Þðb1Þ=b ; S0H   SL ¼ ðSL a1 þ a2 Þðb1Þ=b S0L

#  SL 1=ðb1Þ  S0L ) S0L  ð7Þ SL 

ð8Þ

where the variable b ¼ (m1=m2), ai is the activity of component i in the bulk liquid phase, and S0L and S0H are the selectivities of adsorption of component 1 over component 2 at the limit of x1 ! 0 at the lowest and the highest energy sites of the adsorbent.

HEAT OF ADSORPTION The pertinent thermodynamic variable to quantitatively describe the thermal effects in the exo(endo) thermic gas ad(de)sorption process is called the isosteric heat of adsorption.[17] The isosteric heat of adsorption of component i of an ideal gas mixture (qi) at adsorbate loading of ni and temperature T is given by the following thermodynamic relationship:[17] qi ðni Þ ¼ RT 2



d lnðPyi Þ dT



ð9Þ ni

Eq. (9) is frequently used to obtain the isosteric heat of adsorption of a pure gas i ðq0i Þ as a function of n0i and T from measured isotherms at different temperatures (yi is equal to unity in that case). Estimation of isosteric heat of adsorption of component i of a gas mixture by using Eq. (9) is, however, not practical and they must be obtained by calorimetric measurements. In the absence of lateral interactions, the isosteric heat of adsorption of a pure gas on an energetically homogenous adsorbent is independent of adsorbate loadings. It remains constant at its value at the Henry’s law region ðqi Þ at all coverages. The presence of lateral interactions between adsorbed molecules (pronounced at higher coverages) can increase q0i with increasing n0i . The isosteric heat decreases with increasing adsorbate loading when the adsorbent is energetically heterogenous. The isosteric heat is generally a very weak function of T. Fig. 6 shows several calorimetrically measured examples of these behaviors for adsorption of pure SF6 on various micro- and mesoporous adsorbents.[18] The isosteric heat of adsorption of a component of a gas mixture is equal to that of the pure gas for adsorption on a homogenous adsorbent.

Fig. 5 Binary surface excess isotherms for adsorption of liquid mixtures: (A) benzene (1) þ cyclohexane (2) on silica gel and (B) pyridine (1) þ n-heptane (2) on silica gel and alumina.

Adsorption

31

A

Fig. 6 Isosteric heat of pure SF6 on various absorbents: (A) zeolites, (B) alumina, and (C) activated carbons.

However, the component isosteric heat of a mixture can be substantially different from that of the pure gas at the same loadings when the adsorbent is energetically heterogenous.[17] Furthermore, the dependence of component isosteric heat on the adsorbate loadings can be very complex in that situation.[17]

temperature coefficients of liquid phase adsorption isotherms are generally much smaller than those for gas phase adsorption. The temperature changes in a liquid phase adsorber because of the ad(de)sorption processes are also small owing to the high heat capacity of the liquid.

HEAT OF IMMERSION

ADSORPTION KINETICS

The thermodynamic property to quantify the heat effects for ad(de)sorption of a liquid mixture is the heat of immersion. Fig. 7 shows examples of heat of immersion of binary benzene–cycloxane mixtures on two activated carbons at 30 C.[15] The corresponding surface excess isotherms are U-shaped (e.g., see Fig. 5A). Benzene is the more selectively adsorbed species. The

The actual kinetics of the ad(de)sorption process is generally very fast (in the order of microseconds). However, a significant resistance may exist for transfer of the adsorbate molecules from the bulk fluid phase to the adsorption sites inside the micropores of the adsorbent (see Fig. 2). For gas adsorption, these resistances may be caused by 1) molecular diffusion through a fluid film outside the adsorbent particles (for mixtures only); 2) Knudsen diffusion through the meso- and macropores of the adsorbent and the binder material in parallel with surface diffusion of adsorbed molecules on the walls of those pores (if any); and 3) activated (hopping) diffusion of adsorbed molecules inside the micropores. The same resistances exist for adsorption of liquid mixtures except that the flow through the meso-macropores of the adsorbent and the binder is controlled by molecular and surface diffusion through liquid filled pores. Additional mass transfer resistance called ‘‘skin resistances’’ at the surface of the adsorbent particles or zeolite crystals has also been observed.[2] Some or all of these processes are strongly influenced by the local fluid phase concentration and temperature within the adsorbent particle. The magnitudes of gas diffusivities by different mechanisms follow the order: meso-macropore gas diffusivity (Dp)  surface diffusivity (Ds) in those pores  micropore diffusivity (Dm). For example, Dp

Fig. 7 Heat of immersion of binary liquid mixtures of benzene (1) þ cyclohexane (2) on activated carbons.

32

Adsorption

and Ds for adsorption of water vapor into the mesopores of an activated alumina sample were 5 10 102 cm2=sec and 3 106 cm2=sec, respectively.[4] The micropore diffusivity of water vapor into the crystals of NaA zeolite was 1 106 cm2=sec.[4] The diffusivity of a liquid adsorbate through the meso-macropores is much slower than that for a gas. For example, the diffusivity of bulk liquid phase water from ethanol into an alumina sample was 6 107 cm2=sec.[4] The time constant for meso-macropore diffusion into an adsorbent particle of radius Rp is given by ðDp =R2p Þ. The time constant for micropore diffusion into a pore (having a characteristic diffusional distance of Rm) is given by ðDm =R2m Þ. Because Rm (1–2 mm) is typically much smaller than Rp (1–2 mm), the mesomacropore diffusion generally controls the overall mass transfer into a practical adsorbent particle, even though Dm  Dp . An exception will be the case where the diameter of the adsorbate molecule is very close to that of the micropore, so that Dm is extraordinarily small, and the micropore diffusion becomes the controlling mechanism. Table 4 shows the micropore diffusivities of various gases in the Henry’s law region into the crystals of several zeolites at 300 K.[2] The table shows the remarkable decrease in the micropore diffusivity of a gas when its molecular diameter approaches that of the zeolite pore. The temperature coefficients of Ds and Dm are given by the Arrhenius relationship [Ds, Dm ¼ D0 exp(E=RT)] because these diffusions are activated processes. E is the activation energy for the diffusion process and D0 is a constant. These diffusivities can also be complex functions of adsorbate loadings and compositions.[19] The most rigorous formulation to describe adsorbate transport inside the adsorbent particle is the chemical potential driving force model. A special case of this model for an isothermal adsorption system is the Fickian diffusion (FD), model which is frequently used to estimate an effective diffusivity for adsorption of component i (Di) from experimental uptake data for pure gases.[2,19] The FD model, however, is not generally used for process design because of mathematical complexity. A simpler analytical model called linear driving force (LDF) model is often used.[20] According to this model, the rate of adsorption of component i of a gas mixture

into an adsorbent particle of radius R is given by:[11] X d
ni
i Þ þ
j Þ ¼ kii ð
ni  n kij ð
nj  n dt


i is the average adsorbate loading of component i where n
i represents the in the particle at time t. The variable n adsorbate loading of component i that would be in equilibrium with the superincumbent gas phase conditions at time t. kii and kij are straight and cross mass transfer coefficients for component i. Further simplifications are often made by assuming that kij ¼ 0 and kii is a function of T only. The relation between kii and Di is generally given by Kii ¼ O(Di=R2), where O is a constant.[20] A value of O ¼ 15 is often used even though other values are also possible.[20] A parallel formulation called the surface excess linear driving force (SELDF) model for describing adsorption kinetics from liquid mixtures using Gibbsian surface excess as the variable has also been used successfully.[4] Table 5 shows examples of LDF mass transfer coefficients for adsorption of several binary gas mixtures on BPL activated carbon particles (6–16 mesh) at 23–30 C.[21] The data show that the mass transfer coefficients are relatively large for these systems. There is a scarcity of multicomponent adsorption equilibria, kinetics, and heat data in the published literature. This often restricts extensive testing of theoretical models for prediction of multicomponent behavior.

DESCRIPTION OF SELECTED ADSORPTIVE SEPARATION PROCESSES Adsorption has become the state of the art technology for many separation applications as listed in Table 1. The more prolific areas include: a. Drying of gaseous and liquid mixtures. b. Production of oxygen and nitrogen enriched gases from air. c. Production of ultrapure hydrogen from various gas sources. d. Separation of bulk liquid mixtures where distillation is not convenient.

Table 4 Examples of pure gas diffusivities into zeolites at 300 Ka ˚) Gas Kinetic diameter (A NaA Zeolite (4A) (m2/sec) Na-CaA Zeolite (5A) (m2/sec) N2

3.70

14

3.1 10

17

10

1.1 10

Nax Zeolite (13X) (m2/sec) Fast

Kr

3.65

1.2 10

—

—

CH4

3.76

1.2 1015

6.0 1010

—

n-C4H10

4.69

9.6 1020

1.5 1013

2.4 1011

a

ð10Þ

˚ , respectively. Effective crystal pore openings of 4A, 5A, and 13X zeolites are 4.0, 4.9, and 7.6 A

Adsorption

33

Table 5 Examples of binary LDF mass transfer coefficients on BPL carbon k11 (sec1)

k22 (sec1)

CO2 (1) þ He (2)

0.44

—

Gas mixtures CH4 (1) þ He (2)

1.42

—

N2 (1) þ He (2)

3.33

—

CO2 (1) þ CH4 (2)

0.35

0.76

CO2 (1) þ N2 (2)

0.35

0.66

There has been extensive research and development in all of these areas during the last 30 yr. For example, topics (a)–(c) alone have generated more than 600 U.S. patents during the period 1980–2000.[1] They have been assigned to 160 different corporations around the world. For the sake of brevity, only a selected few commercial processes will be discussed here.

Adsorptive Drying Both TSA and PSA processes are commercially used for removal of trace or dilute water contamination from a gas. They are commercially designed to handle 1–40,000 ft3 of feed gas per minute. The basic steps of a conventional TSA process consists of: 1) flowing the contaminated gas over a packed bed of a desiccant (silica gel, alumina, zeolite) at a near-ambient temperature and withdrawing a dry product gas until the moisture concentration in the product gas rises to a preset level; 2) heating the adsorbent to 150–300 C by flowing a hot dry gas countercurrently through the dbed and rejecting the water laden effluent gas; and 3) cooling the bed to feed gas temperature by countercurrently flowing a dry gas through the bed at feed gas temperature while rejecting the effluent gas. The cycle is then repeated. A part of the dry gas (10–30%) produced in step 1 is generally used to supply the gas for steps 2 and 3. The effluent gas from step 3 is often

heated to supply the gas for step 2. Fig. 8A is a schematic diagram of a three-column TSA gas drying unit. The total cycle time (all steps) for TSA processes generally varies between 2 and 8 hr. A typical dynamic water removal capacity of an alumina dryer is 5–15% by weight. Product gas dew points of less than 40 C can be easily obtained. Many different process modifications like thermal pulsing, elimination of the cooling step, lower temperature regeneration, etc. are also used for decreasing the costs of drying.[4] The TSA dryers for liquid mixtures use similar process steps to those used for gas dryers except that the adsorbers are first drained to remove the void liquid before they are heated countercurrently using a hot gas. Heating vaporizes the adsorbed water as well as liquid films adhering to the adsorbent particles. The hot effluent gas is cooled to condense out the components of the feed liquid mixture. The adsorber is then cooled by countercurrently flowing a cold gas and refilled with dry liquid before starting a new cycle.[4] The basic steps of a conventional PSA gas drying process (Skarstrom cycle) consists of: 1) adsorption of water vapor from the feed gas by flowing the gas over a desiccant bed (silica gel, alumina, zeolite) at an elevated pressure (say 5–15 atm) and withdrawing a dry product gas at feed pressure; 2) countercurrently depressurizing the adsorber to near-ambient pressure and venting the effluent; 3) countercurrently purging the adsorber with a part of the dry gas produced by step 1 at near-ambient pressure while venting the effluent; and 4) countercurrently pressurizing the adsorber with a part of the product gas from step 1. Adsorption at a relatively lower feed gas pressure (1.3– 1.7 atm) and desorption under vacuum (both depressurization and purge steps) are also practiced. Fig. 8B is a schematic diagram of a two-bed PSA dryer. The process can be used to obtain a very dry product (say 60 C dew point). A typical process uses 15–30% of the dry product gas as purge. The total cycle times for PSA processes generally vary between 2 and 6 min.[4]

Fig. 8 Schematic drawings of (A) three-column TSA and (B) twocolumn PSA processes.

A

34

Adsorption

Air Fractionation

Table 6 Comparative performance of various zeolites for O2 production by a VSA process

A large variety of PSA process concepts have been commercialized for: 1) production of 23–95 mol% oxygen using a N2 selective (thermodynamic) zeolite; 2) production of 99þ mol% nitrogen using an O2 selective (kinetic) carbon molecular sieve; and 3) simultaneous production of 90þ mol% O2 and 98þ mol% N2 using a N2 selective zeolite from ambient air. Some of these concepts are called vacuum swing adsorption (VSA) because the final desorption pressure is subatmospheric. Commercial units are designed in the size range of 0.012–100 tons of oxygen per day (TPD).[2] A VSA process for production of 90% O2 from air using a LiX zeolite uses the cyclic steps of: 1) pressurizing the adsorber from an intermediate pressure level (PI) to the final adsorption pressure level of PA (1.43 atm) with compressed air; 2) flowing compressed air feed at PA through the adsorber and producing the 90% O2 enriched product gas; 3) countercurrently depressurizing the adsorber to nearambient pressure and venting the effluent gas; 4) countercurrently evacuating the adsorber to a vacuum level of PD (0.34 atm) and countercurrently purging the adsorber with a part of the O2 product gas from step 2 at that pressure while venting the effluent gas; and finally (5) countercurrently pressurizing the adsorber from PD to PI using a part of the product gas. The cycle is then repeated.[22] Using a total cycle time of 70 sec, the O2 productivity by the process, in terms of bed size factor (BSF), was 830 lb of zeolite per TPD of oxygen. The total power requirement was 11.6 kW=TPD when the product O2 gas was delivered at a pressure of 6.45 atm by recompression.[22] The LiX zeolite is currently favored for O2 production. Table 6 demonstrates its superiority over other zeolites for making 90 mol% O2 from air using a specific VSA cycle operating between final adsorption and desorption pressures of 1.48 and 0.25 atm, respectively.[23] The Carbon Molecular Sieves (CMS) contain constricted pore mouths that permit the slightly smaller O2 molecules to diffuse into the pores of the carbon faster than the N2 molecules from air. This produces an O2 enriched adsorbed phase based on kinetic selectively when the CMS is contacted with air for a short period of time. The material has practically no thermodynamic selectivity for either gas.[24] A simple four-step PSA process using a CMS consists of: 1) passing compressed air into the adsorber to pressurize it to PA, and then withdrawing a N2 enriched product gas at that pressure; 2) connecting the adsorber with another adsorber that has completed step 3 below, to pressure equalize the adsorbers; 3) countercurrently depressurizing the adsorber to near-ambient pressure and venting the O2 enriched gas, and finally 4) pressure equalizing the adsorber with the companion adsorber,

Zeolite

BSF

Energy of separation

NaX

1.00

1.00

CaX

1.28

1.78

LiX

0.51

0.88

which has completed step 1. The cycle is then repeated. Using a feed air pressure of 8.1 atm (¼PA), the process could produce a 99.0 mol% N2 product gas with a N2 recovery of 49.4% from the feed air. The N2 productivity was 92 ft3=hr per cubic foot of adsorbent.[24] The O2 enriched waste gas contained 33.8% O2. The N2 productivity and recovery were reduced to 73% and 39.2%, respectively, when its purity was raised to 99.5 mol%. Other process cycles have been designed using CMS adsorbents to raise the N2 product gas purity above 99.9 mol%.[25]

Production of Hydrogen The most common industrial method to make ultrapure hydrogen is by steam-methane reforming (SMR) using a catalyst at the temperature 890–950 C. The reformed gas is then subjected to a high temperature water gas shift (WGS) reaction at 300–400 C. The WGS reactor effluent typically contains 70–80% H2, 15–25% CO2, 1–3% CO, 3–6% CH4, and trace N2 (dry basis), which is fed to a PSA system at a pressure of 8–28 atm and a temperature of 20–40 C for production of an ultrapure (99.99þ mol%) hydrogen gas at the feed pressure. Various PSA systems have been designed for this purpose to produce 1–120 million cubic feet of H2 per day. A popular PSA cycle called polybed process consists of 11 cyclic steps.[26] They include: 1) passing the feed gas at pressure PF through a packed adsorbent column and withdrawing the high-purity H2 product gas; 2) cocurrently depressurizing the adsorber from pressure PF to PI while producing a stream of essentially pure H2; 3) further cocurrently depressurizing the adsorber from pressure PI and PII and withdrawing another stream of H2 enriched gas; 4) even further depressurizing the adsorber cocurrently from pressure PII and PIII and again withdrawing a stream of H2 enriched gas; 5) countercurrently depressurizing the adsorber from pressure PIII to a near-ambient pressure (PD) and venting the effluent gas; 6) countercurrently purging the adsorber with a stream of essentially pure H2 produced by a companion adsorber undergoing step 4; 7) countercurrently pressurizing the adsorber from PD to PII by introducing the effluent gas from

Adsorption

another adsorber carrying out step 3; 8) further pressurizing the adsorber countercurrently from pressure PII and PI using the gas produced by another adsorber undergoing step 2; and 9) finally pressurizing the adsorber countercurrently from pressure PI to PF using a part of the product gas produced by another adsorber carrying out step 1. The cycle is then repeated. Increasing the pressure ratio between the feed gas and the purge gas (PF=PD) generally improves the separation performance of the PSA process by: 1) providing higher specific adsorption capacities for the selectively adsorbed components of the feed gas during the adsorption step (thus reducing adsorbent inventory) and 2) lowering the quantity of the purge gas required for adsorbent regeneration (thus increasing product recovery). However, increased feed gas pressure also increases the void gas quantity inside the adsorber at the end of the adsorption step, which: 1) increases the amount of adsorbent needed to contain the impurities during the subsequent cocurrent depressurization steps and 2) increases product loss during the countercurrent depressurization step. Consequently, there is an upper feed gas pressure limit (typically NO
), which is capable of scavenging alkyl radicals in competition with oxygen, i.e., an effective CB-A antioxidant. Its photostabilizing mechanism also involves the regeneration of the nitroxyl radical from both the corresponding hydroxylamine (>NOH an effective CB-D) and alkylhydroxylamine,[22,40,57,58] see Scheme 3. Overall, the photoantioxidant activity of HALS can be ascribed to a regenerative donoracceptor (CB-A=CB-D) antioxidant mechanism involving >NO
and >NOH.

Table 2 Effect of different metal dithiolates on the photostability (embrittlement time, EMT) of PP processed at 190 C (For structures, see Table 1) Antioxidant

Concentration (mol/100 g  104)

Unstabilized PP

0

ZnDEC

2.5

NiDEC NiDEC

2.5 10

UV-EMT (hr) 90 175 740 840

FeDMC

0.25

85

FeDMC

2.5

150

FeDMC

5

336

CoOct X

3

1600

Chimassorb 81

3

245

NiDBP þ Chimassorb 81

6

2650

ANTIOXIDANTS: METAL DEACTIVATOR The main function of metal deactivators (MD) is to retard efficiently metal-catalyzed oxidation of polymers. Polymer contact with metals occur widely, for example, when certain fillers, reinforcements, and pigments are added to polymers, and, more importantly when polymers, such as polyolefins and PVC, are used as insulation materials for copper wires and power cables (copper is a pro-oxidant since it accelerates the decomposition of hydroperoxides to free radicals, which initiate polymer oxidation). The deactivators are normally polyfunctional chelating compounds with ligands containing atoms like N, O, S, and P (e.g., see Table 1, AOs 33 and 34) that can chelate with metals and decrease their catalytic activity. Depending on their chemical structures, many metal deactivators also function by other antioxidant mechanisms, e.g., AO 33 contains the hindered phenol moiety and would also function as CB-D antioxidants.

ANTIOXIDANTS: BIOLOGICAL The use of antioxidants in human-contact applications, e.g., food-contact, medical, and pharmaceutical, present a challenge in terms of their safety and level of migration into the contact media, e.g., food and body fluids. The biological antioxidant vitamin E, which is a suitable candidate for such areas of application, is a fat-soluble, and sterically hindered phenol antioxidant with the most bioactive form of the vitamin being the a-tocopherol (Table 1, AO 10). In-vitro rate studies on the antioxidant activity of atocopherol has shown that it is one of the most efficient alkylperoxyl radical traps, far better than the commercial hindered phenols, e.g., BHT (2,6-di-tert-butyl-4methylphenol, AO 1).[59] Its efficiency was attributed to stereo electronic effects: the electronic synergy between a fully methylated aromatic ring and the chroman moiety results in a highly stabilized tocopheroxyl radical, formed during the rate limiting step, Reaction 8, because of the interaction between the p-orbitals on the two oxygen atoms.[59]

A

92

Antioxidants

Fig. 2 Melt and color stability of PP processed in the presence of antioxidants. (View this art in color at www.dekker.com.)

a-tocopherol has been shown to be a very effective melt processing antioxidant for PP, especially at low concentration.[49,60,61] A comparison of the antioxidant efficiency of a-tocopherol with that of Irganox 1010 (AO 5) during melt extrusion of PP at 260 C shows (see Fig. 2A) clearly the superior performance of the former at all concentrations and, in particular, at very low concentrations. Tocopherol can, therefore, be used cost-effectively at only one-quarter of the concentration typically required for the stabilization of PP by synthetic hindered phenols, such as Irganox 1010. The activity of the former is attributed to the rapid rate of deactivation of radicals responsible for PP chain scission. A further important contribution to its antioxidant activity stems from its oxidation products, which are formed during polymer melt processing; these products were shown to be very effective antioxidants.[61] Tocopherol products formed during PP and PE melt processing consist mainly of direct coupling products, leading to the formation of dihydroxydimer,

DHD and quinonoid-type products, trimers, TRI, spirodimers, SPD, quinone methides, QM, together with some aldehydes, ALD,[62] see Fig. 3 for structures of these products. All the oxidation products were shown to be more highly colored than tocopherol itself, with the aldehydes being the most colored, and the trimers the least colored. In general, sterically hindered phenols contribute to some discoloration (yellowing) of polyolefins during processing. Yellowing of polyolefins containing hindered phenols has been attributed to a number of factors including the formation of colored oxidation products, e.g., quinonoid structures, and interactions between the phenols and transition metal ion catalyst residues from the polymerization stage.[60,63] The extent of discoloration depends on the chemical structure of the parent antioxidant, the oxidation products, and the type and amount of catalyst residues in the polymer. It was shown[49] that at low concentration, both tocopherol and Irganox 1010 cause comparable

Fig. 3 Structures of the main transformation products of a-tocopherol formed during melt processing of polyolefins.

Antioxidants

93

via a redox-type reaction, through the formation of a phosphite-phenol C–C coupled product (Scheme 6). ANTIOXIDANTS: REACTIVE

Fig. 4 Color stability of PP processed in presence of 300 ppm tocopherol in combination with the phosphate U-622. Total amount of products formed from tocopherol is also shown.

levels of discoloration during melt processing of PP. At increasingly higher concentrations, however, the extent of discoloration affected by tocopherol is higher (Fig. 2B). In order to reduce the extent of discoloration, very small concentrations, e.g., 300 ppm, of a phosphate antioxidant can give rise to a pronounced color suppression, together with higher levels of retention of the tocopherol in the polymer (Fig. 4). The higher retention of the tocopherol antioxidant observed when a small amount of the phosphite U626 was used has been attributed to[61] first, a reduction in the amount of the more intensely colored transformation products, and second to the regeneration of tocopherol in the presence of the phosphite,

A strategy that is based on the use of reactive antioxidants can also be explored to achieve stabilization of polymers suitable for human-contact applications. Reactive antioxidants that become an integral part of the macromolecular chain can result in nonmigratory stabilizer systems that would be unaffected by extractive hostile contact media. In general, reactive antioxidants are compounds that contain one or more antioxidant functions (the antioxidant, AO component) and one or more chemical functions that are capable of reacting either with monomers (same or different) or with polymers (the reactive, Rv, component). The AO function is based on any antioxidant moiety (see examples A–D in Scheme 7), whereas the reactive moiety can either be a polymerizable (e.g., Rv functions 1–4) or nonpolymerizable (e.g., Rv functions 5–7) groups, and may or may not contain a spacer (an inert flexible and short chemical link connecting the antioxidant moiety to the reactive function). Reactive antioxidants may either be copolymerized with monomers during polymer synthesis or grafted on preformed polymers; they are therefore linked to the polymer. Although the copolymerization route has been successfully exploited,[35,64,65] it has not received greater attention because of cost incurred in the synthesis and production of tailor-made ‘‘speciality’’ materials for low-volume specific applications. On the

Scheme 6 Redox reactions resulting in the regeneration of tocopherol.

A

94

Antioxidants

Scheme 7

other hand, grafting of antioxidants on preformed polymer melts can offer a more flexible and versatile approach where standard compounding and processing machines are used to conduct chemical grafting reactions. Both routes, however, offer tremendous advantages in terms of the physical persistence of antioxidants in the polymer. In both cases, the process of chemical attachment (target reaction) of antioxidants onto the polymer backbone proceeds in competition with other undesirable processes. The main prerequisite here, therefore, is to achieve the target reaction without detriment to the overall polymer properties and the fabrication process. Both thermal- and photo-antioxidant functions have been grafted on polyolefins during melt processing in the presence of a free radical initiators.[66–68] The practical success of in situ melt grafting of antioxidants on polyolefins, however, depends on the correct choice of chemical systems and processing variables that would reduce the interference of side reactions, without altering significantly the polymer characteristics, e.g., molar mass, morphology, and physical properties. The three different types of reactive antioxidant systems typically used for grafting reactions on polyolefins are briefly as described in the following.

Monofunctional Polymerizable Antioxidants The use of polymerizable monofunctional antioxidants with one reactive group per antioxidant molecule is

considered here. Production of these antioxidants is generally straightforward. Therefore, it can offer, a broad, versatile, and economic route for the production of a range of polymer-grafted antioxidants and antioxidant concentrates. Different monofunctional antioxidants have been reactively grafted on polyolefins, e.g., PP, LDPE, HDPE, poly (4-methyl-1-ene), in the presence of free radical initiators using single- or twin screw extruders, or internal mixers. It has been demonstrated, however, that the efficiency of chemical attachment of such monofunctional polymerizable antioxidants on polyolefins (Reaction 9) is always low.[66,69] This is mainly because of the highly competitive homopolymerization reaction of the reactive antioxidant (Reaction 10). For example, studies on the effect of processing variables on the extent of melt grafting on PP of different mono-acryloyl-containing hindered phenol (DBBA) and hindered amine (AOTP) antioxidant functions have shown that grafting efficiency is less than 50%.[69,70] The remaining ungrafted antioxidants were recovered, almost completely, as homopolymers of the parent antioxidants, which were incompatible with the host polymer and were readily removed by extractive solvents. Furthermore, the performance of these homopolymers, when incorporated in the polymer matrix as conventional antioxidants, is very poor. The problems of homopolymer formation and low efficiency of grafting of mono-functional polymerizable antioxidants in polyolefins were subsequently

Antioxidants

95

A

addressed by alternative approaches (see Reactions 9 and 10, above).

Monofunctional Nonpolymerizable Antioxidants Non-polymerizable monofunctional antioxidants were subsequently used to avoid the problem of homopolymerization of the antioxidant. For example, melt grafting of the two maleated antioxidants, BPM and APM (see below), on PP was shown to lead to high grafting efficiencies (up to 75% in the former and >90% in the latter), which were attributed to the nonpolymerizable nature of the maleate (maleimide) functions.[71] The performance of these antioxidants, especially under extractive organic solvent conditions, was also shown to far exceed that of conventional antioxidants with similar antioxidant function.

Bifunctional Polymerizable Antioxidants The use of reactive antioxidants containing two polymerizable polymer-reactive functions in the same antioxidant molecule is outlined here. Careful choice of the processing parameters, the type, and the amount of free radical initiator can lead to very high levels of antioxidant grafting.[71] For example, melt grafting of concentrates (e.g., 5–20 wt.%) of the di-acrylate hindered piperidine, AATP, on PP in the presence of a peroxide initiator has led to almost 100% grafting. This exceptional grafting efficiency of AATP is in marked contrast with the much lower grafting levels achieved with the mono-functional HALS analogues, e.g., MyATP and AmyTP.[70] Examination of the mechanisms involved in the grafting process of such bifunctional antioxidants has shown that the grafting reaction occurs through the intermediacy of a crosslinked structure, involving the polymer and the reactive antioxidant, lead finally to an antioxidant grafted polymer product, which remains comparable in its general characteristics, e.g., solubility, crystallinity, molar mass, to a conventionally stabilized sample.[70,71]

Monofunctional Polymerizable Antioxidants in the Presence of a Comonomer The use of a reactive di- or poly-functional comonomer (nonantioxidant), which can co-graft with a monofunctional polymerizable antioxidant on polymers, can improve the grafting efficiency from as low as 10– 40% to an excess of 80–90%. This strategy, however,

96

Antioxidants

presents immense challenges because of the presence of more than one polymerizable group in the comonomer, which could lead to additional undesirable (competing) side reactions, complicated by the possibility of comonomer-induced crosslinking reactions of the polymer. The success of this ‘‘one-pot’’ synthetic approach lies in the ability to achieve a delicate balance between the composition of the chemical system (antioxidant, comonomer, free radical initiator) and reaction conditions (e.g., temperature, residence time) with the aim of promoting the target grafting reaction at the expense of all competing side reactions.[69,70] In practice, the success of this method has been clearly illustrated.[69,70] The novelty of this approach lies in the fact that co-grafting of polymerizable polyfunctional agents (traditionally used as crosslinkers, e.g., the trimethylol propane triacrylate, Tris) with mono-vinyl antioxidants (and other additives) in extruders or mixers leads to the production of highly grafted antioxidants in a noncrosslinked polymer. This cografting method can be applied to a wide range of antioxidant functions (e.g., HAS, UVA, hindered phenols, aromatic amines) to achieve outstanding levels of antioxidant grafting. Table 3 shows an example, which illustrates the excellent performance, especially under extractive conditions, of a highly bound synergistic antioxidant system (hindered phenols þ UV absorber) produced by this method in PP compared to a conventional (unbound) commercial antioxidant system.

ANTIOXIDANT MIXTURES: SYNERGISM AND ANTAGONISM The interaction between two or more antioxidants (or antioxidant functions) in plastics formulations can

Table 3 Comparison of the antioxidant performance (accelerated UV aging) of synergistic mixture (melt grafted in presence of Tris) with a conventional antioxidant mixture based on the same antioxidant functions (at 1 : 1 w=w ratio) Antioxidant (0.4% in PP films) None

UV embrittlement time (hr) Unextracted

Extracted

75

70

a

205

80

HAEBb

330

70

1160

1130

DBBA

PP-g(DBBA-HAEB)Tris

lead to enhanced performance by more than the sum of their individual effects; i.e., synergistic effects. Synergism can result from the combined action of two chemically similar antioxidants (homosynergism), e.g., two hindered phenols, or from two different antioxidant functions present in the same stabilizer molecule (autosynergism), e.g., Irganox 1081 (AO 9, Table 1), or when separate stabilizer molecules that carry different antioxidant functions are physically blended in a stabilizer formulation (heterosynergism). Conversely, antioxidant combinations resulting in reduced performance, relative to the sum of their individual contribution, are called antagonistic. Highly effective UV stabilizing systems can be achieved by the use of synergistic mixtures of compounds acting by different mechanisms. Table 4 illustrates the synergism obtained from combinations of different metal dithiolates with the UV absorber Chimassorb 81 (AO 28, Table 1) in LDPE. Hindered phenol antioxidants combined with sulfur-containing compounds exhibit synergism during thermal stabilization of polyolefins. In contrast, similar combinations of Irganox 1076 (AO 4, Table 1) with different metal dithiolates lead to antagonism during photostabilization, see Table 4.[26,72] This antagonistic behavior has been attributed to sensitization leading to photolytic destruction of the dithiolates by oxidation products of phenols, particularly stilbenequinones. Stabilizers that contain two different antioxidant functions (e.g., PD and CB activities) in the same molecule, such as phenolic sulfides, show much higher molar intrinsic activity as thermal antioxidants (because of autosynergism) than conventional hindered phenols with only CB activity.[11] HALS exhibit a complex behavior when present in combination with other antioxidants and stabilizers. Effective synergism in both melt and thermal stabilization has been achieved when secondary and tertiary HALS were used in combination with both aromatic and aliphatic phosphites; the synergistic optimum depends on the structure of the phosphite.[73] HALS also synergize the action of UV-absorbers, e.g., benzotriazoles, in different polymers such as polypropylene, polystyrene, and ABS.[74]

UV531 is a commercial UV absorber, 2-hydroxy-4octyloxy benzophenone Tris is a triacrylate comonomer. a Unbound, processed alone (no Tris) as a conventional antioxidant. b See Table 2 for structures.

CONCLUSIONS Antioxidants and stabilizers are generally used in the chemical processing industry for the protection and preservation of properties of materials, including food, with the aim of prolonging and extending their shelf life. They are essential ingredients for the long-term durability of many polymers, such as

Antioxidants

97

Table 4 Synertistic and antagonistic effects on photostability of LDPE processed at 150 C

Stabilizer system

Concentration (l04 M l00 g1)

Photo-embrittlement time (hr)

A

Observed

Calculated effecta,b

References

Control PP (no antioxidant) ZnDEC NiDEC Tinuvin 770 Irganox l076 NiDBP NiBX CuDIP Chimassorb 81

— 3 3 3 3 2.5 2.5 2.5 3

l000 l400 l800 2400 l750 2800 2500 2300 l650

— — — — — — — — —

[26] [26] [26] [26] [26] [72] [72] [72]

Synergistic systems NiDEC þ Chimassorb 81 ZnDEC þ Chimassorb 81 NiDBP þ Chimassorb 81 NiBX þ Chimassorb 81 CuDIP þ Chimassorb 81

3 þ 3 3 þ 3 2.5 þ 2.5 2.5 þ 2.5 2.5 þ 2.5

— 4000 5500 4900 5350

— 3000 4500 4200 4000

[26] [26] [72] [72] [72]

Antagonistic systems NiDEC þ Irganox l076 ZnDEC þ Irganox l076 NiDEC þ Tinuvin 770 NiDBP þ Irganox l076

3 þ 3 3 þ 3 3 þ 3 2.5 þ 5

l580 l250 l850 —

3550 3150 4200 —

[26] [26] [26] [72]

Concentration of HOBP in this case was 5  l04M l00 g1. % Synergism ¼ f[(Es  Ec)  (E1  Ec) þ (E2  Ec)]=(E1  Ec þ E2  Ec)g  100, where Es is the embrittlement time of synergist, Ec the embrittlement time of control, E1 the embrittlement time of antioxidant l, and E2 the embrittlement time of antioxidant 2. a

b

polyolefin, and are crucial to the upgrading of their performance, and for achieving the benefits of sustainable development in polymers recycling programs. Antioxidants and stabilizers are chosen for target applications on the basis of chemical, physical, toxicological, and economic factors. The final selection of an antioxidant package must take into consideration the performance requirements of the end-use polymer article including toxicity, compatibility, appearance, and color. Issues of efficacy and safety have been the driving force behind much of the recent progress made in the areas of biological, reactive, and macromolecular antioxidants. More stringent regulations and legislations for certain applications of stabilized polymers, such as in food, toys, medicine, and other health-related areas, would promote further interest in the use of biological and naturally occurring antioxidants and reactive antioxidants for chemical processing, and for producing safe and ‘‘permanently’’ stabilized polymer compositions. Current emphasis on sustainable development and green chemistry approaches should lead to further exploration of the benefits of renewable resources and environmentally benign synthetic routes in the development and procurement of new antioxidants, or for replacing existing ones.

Compared to conventional antioxidants, reactive antioxidants that are capable of becoming covalently bound to the polymer backbone are not readily lost from polymers during fabrication and in-service. There is a lot of evidence that demonstrates the performance (in terms of polymer protection) of ‘‘immobilized’’ antioxidants in practice, especially when polymer products are subjected to harsh environment, e.g., exposure to high temperatures, UV-light and leaching solvents. It is clear from this that high mobility of low molar mass antioxidants is not a necessary prerequisite to achieving stabilization and attachment of antioxidants to polymers can be industrially beneficial. Reactive antioxidants grafted on polymer melts behave in a similar way to low molar mass conventional antioxidants, but offer many additional advantages. The polymer-linked antioxidants do not suffer from the problem of compatibility, volatility, and migration, i.e., they do not suffer physical loss even under highly aggressive and extractive environments. Such antioxidant systems would be much more riskfree and environmentally friendly. The ability to produce highly grafted antioxidant concentrates (master batches), which can be used in conventional (the same or different) polymers, as ‘‘normal’’ additives would extend the use of reactive antioxidants to new areas of application.

98

REFERENCES 1. Bolland, J.L.; Gee, G. Kinetic studies in the chemistry of rubber and related materials. II. The kinetics of oxidation of unconjugated olefins. Trans. Faraday Soc. 1946, 42, 236–243. 2. Bolland, J.L. Kinetic studies in the chemistry of rubber and related materials. VI. Benzyl peroxidecatalyzed oxidation of ethyl linoleate. Trans. Faraday Soc. 1948, 44, 669–677. 3. Bolland, J.L. Kinetics of olefin oxidation. Quart. Rev. 1949, 3, 1–21. 4. Bateman, L.; Morris, A.L. Initiation efficiencies in olefin autoxidation. Trans. Faraday Soc. 1952, 48, 1149–1155. 5. Al-Malaika, S. Autoxidation. In Atmospheric Oxidation and Antioxidants; Scott, G., Ed.; Elsevier Applied Science Publishers: London, 1993; 76, Vol. 1, 45–82. 6. Scott, G. Initiators, proxidants and sensitizers. In Atmospheric Oxidation and Antioxidants; Scott, G., Ed.; Elsevier Applied Science Publishers: London, 1993; 76, Vol. 1, 83–119. 7. Howard, J.A Absolute rate constants for reactions of oxyl radicals. Adv. Free Radical Chem. 1972, 4, 49–173. 8. Iring, M.; Tudos, F. Thermal oxidation of polyethylenes and polypropylene: effects of chemical structure and reaction conditions on the oxidation process. Prog. Polym. Sci. 1990, 15 (2), 217–262. 9. Hinsken, H.; Moss, S.; Pauquet, J.-R.; Zweifel, H. Degradation of polyolefins during melt processing. Polym. Deg. Stab. 1991, 34, 279–293. 10. Johnston, R.T.; Morrison, E.J. Thermal scission and cross-linking during polyethylene melt processing. In Polymer Durability: Degradation, Stabilisation and Lifetime Prediction; Clough, R.L., Billingham, N.C., Gillens, K.T., Eds.; Advances in Chemistry Series-249, A.C.S.: Washington, 1996; 651–682. 11. Al-Malaika, S.; Scott, G. Thermal stabilizers of polyolefins. In Degradation and Stabilisation of Polyolefins; Allen, N.S., Ed.; Applied Science: London, 1983; 247–281. 12. Scott, G. Antioxidants: chain breaking mechanisms. In Atmospheric Oxidation and Antioxidants; Scott, G., Ed.; Elsevier Applied Science: London, 1993; Vol. 1, 121–160. 13. Al-Malaika, S. Antioxidants: preventive mechanisms. In Atmospheric Oxidation and Antioxidants; Scott, G., Ed.; Elsevier Applied Science: London, 1993; Vol. 1, 161–224. 14. Pospisil, J. Chain-breaking antioxidants in polymer stabilization. In Developments in Polymer Stabilisation; Scott, G., Ed.; Applied Science: London, 1979; Vol. 1, 1–37.

Antioxidants

15. Pospisil, J. Aromatic amines antidegradants. In Developments in Polymer Stabilisation; Scott, G., Ed.; Applied Science: London, 1984; Vol. 7, 1–63. 16. Chakraborty, K.B.; Scott, G. Mechanisms of antioxidant action: the behavior of hindered piperidine UV stabilizers during the processing of LDPE. Chem. Ind. 1978, 7, 237. 17. Carlsson, D.J.; Garton, A.; Wiles, D.M. The photo-stabilization of polyolefins. In Developments in Polymer Stabilisation; Scott, G., Ed.; Applied Science: London, 1979; Vol. 1, 219–259. 18. Denisov, E.T. Inhibitor regeneration in oxidation. In Developments in Polymer Stabilisation; Scott, G., Ed.; Applied Science: London, 1980; Vol. 3, 1–20. 19. Bagheri, R.; Chakraborty, K.B.; Scott, G. Mechanisms of antioxidant action: behavior of a hindered piperidine and related oxidation products during processing and photo-oxidation of polypropylene. Polym. Deg. Stab. 1982, 4 (1), 1–16. 20. Shlyapintokh, V.Ya.; Ivanov, V.B. Antioxidant action of sterically hindered amines and related compounds. In Developments in Polymer Stabilisation; Scott, G., Ed.; Applied Science: London, 1982; Vol. 5, 41–70. 21. Berger, H.; Bolsman, T.A.B.M.; Brouwer, D.M. Catalytic inhibition of hydrocarbon autoxidation by secondary amines and nitroxyls. In Developments in Polymer Stabilisation; Scott, G., Ed.; Applied Science: London, 1983; Vol. 6, 1–27. 22. Al-Malaika, S.; Omikorede, E.O.; Scott, G. Mechanisms of antioxidant action: mechanochemical transformation products of 2,2,6,6-tetramethyl-4-hydroxypiperidinoxyl in polypropylene. J. Appl. Polym. Sci. 1987, 33, 703–713. 23. Schwetlick, K. Mechanisms of antioxidant action of phosphite and phosphonite esters. In Mechanisms of Polymer Degradation and Stabilisation; Scott, G., Ed.; Elsevier Science: New York, 1990; 23–60. 24. Pobedimskii, D.G.; Mukmeneva, N.A.; Kirpichnikov, P.A. Developments in Polymer Stabilisation; Scott, G., Ed.; Applied Science: London, 1980; Vol. 2, 125–184. 25. Al-Malaika, S. Antioxidant mechanisms of derivatives of dithiophosphoric acid. In Mechanisms of Polymer Degradation and Stabilisation; Scott, G., Ed.; Elsevier Science: New York, 1990; 61–107. 26. Al-Malaika, S.; Chakraborty, K.B.; Scott, G. Peroxidolytic antioxidants: metal complexes containing sulfur ligands. In Developments in Polymer Stabilisation; Scott, G., Ed.; Applied Science: London, 1983; Vol. 6, 73–120. 27. Scott, G. Peroxidolytic antioxidants: sulphur antioxidants and autosynergistic stabilizers based on

Antioxidants

28.

29.

30.

31.

32.

33.

34.

35.

36.

37.

38.

39. 40.

41.

alkyl and aryl sulphides. In Developments in Polymer Stabilisation; Scott, G., Ed.; Applied Science: London, 1983; Vol. 6, 29–71. Shelton, J.R. Organic sulphur compounds as preventive antioxidants. In Developments in Polymer Stabilisation; Scott, G., Ed.; Applied Science: London, 1981; Vol. 4, 23–69. Osawa, Z. Inhibition of metal-catalyzed degradation of polymers. In Developments in Polymer Stabilisation; Scott, G., Ed.; Applied Science: London, 1984; Vol. 7, 193–232. Chan, M.G. Metal deactivators. In Oxidation Inhibition of Organic Materials; Klemchuk, P., Pospisil, J., Eds.; CRC Press: Boca Raton, 1990; Vol. 1, 225–246. Muller, H. Metal deactivators. In Plastics Additives Handbook, 2nd Ed.; Gachter, R., Muller, H., Eds.; Hanser: Munich, 1987; 75–95. Gugumus, G. Light stabilizers. In Plastics Additives Handbook, 5th Ed.; Zweifel, H., Ed.; Hanser: Munich, 2001; 141–425. Scott, G. Substantive antioxidants. In Developments in Polymer Stabilisation; Scott, G., Ed.; Applied Science: London, 1981; Vol. 4, 181–221. Scott, G. Macromolecular and polymer-bound antioxidants. In Atmospheric Oxidation and Antioxidants; Scott, G., Ed.; Elsevier Applied Science: London, 1993; Vol. 2, 279–326. Fu, S.; Gupta, A.; Albertsson, A.C.; Vogl, O. New polymerizable 2(2-hydroxyphenyl)2H-benzotriazole ultraviolet absorbers: 2[2,4-dihydroxy-5-vinyl (isopropenyl)phenyl]2,3–2H-dibenzotriazole. In New Trends in the Photochemistry of Polymers; Allen, N.S., Rabek, J.F., Eds.; Elsevier Applied Science: London, 1985; 247–264. Pospisil, J. Stabilizers mixtures and polyfunctional stabilizers. In Oxidation Inhibition of Organic Materials; Klemchuk, P., Pospisil, J., Eds.; CRC Press: Boca Raton, 1990; Vol. 1, 173–224. Al-Malaika, S. Reactive modifiers for polymers. In Chemical Reactions on Polymers; Benham, J.L., Kinstle, J.F., Eds.; ACS Symposium Series364, ACS: Washington, 1988; 409–425. Al-Malaika, S. Reactive processing and polymer performance. Polym. Plast. Technol. Eng. 1990, 29 (12), 73–86. Al-Malaika, S. Tying additives down. Chemtech 1990, 6, 366–371. Glass, R.D.; Valange, B.M. Antioxidant ‘crossover effect’ in oven ageing of polypropylene. Polym Deg. Stab. 1988, 20 (3,4), 355–363. Gugumus, F. Stabilization of plastics against thermal oxidation. In Oxidation Inhibition of Organic Materials; Klemchuk, P., Pospisil, J., Eds.; CRC Press: Boca Raton, 1990; Vol. 1, 61–172.

99

42. Zweifel, H. Effect of stabilization of polypropylene during processing and its influence on longterm behavior under thermal stress. In Polymer Durability: Degradation, Stabilisation and Lifetime Prediction; Clough, R.L., Billingham, N.C., Gillens, K.T., Eds.; Advances in Chemistry Series-249, ACS: Washington, 1996; 373–396. 43. Billingham, N.C.; Calvert, P.D. The physical chemistry of oxidation and stabilization of polyolefins. In Developments in Polymer Stabilization; Scott, G., Ed.; Applied Science: London, 1980; Vol. 3, 139–190. 44. Billingham, N.C.; Prentice, P.; Walker, T.J. Some effects of morphology on oxidation and stabilization of polyolefins. J. Polym. Sci., Polym. Symp. 1976, Deg. Stab. Polyolefins 1977, 57, 287–297. 45. Luston, J. Physical loss of stabilizers from polymers. In Developments in Polymer Stabilisation; Scott, G., Ed.; Applied Science: London, 1980; Vol. 2, 185–240. 46. Al-Malaika, S.; Desai, P.; Scott, G. Mechanisms of antioxidant action: photostabilization of polypropylene using dithiophosphoryl compounds— effects of alkyl substituents. Plast. Rubber Proces. Appl. 1985, 5 (1), 15–18. 47. Al-Malaika, S.; Goonetileka, M.D.R.J.; Scott, G. Migration of 4-substituted 2-hydroxyphenones in low density polyethylene: Part I. Diffusion characteristics. Polym. Deg. Stab. 1991, 32 (2), 231–247. 48. Klemchuk, P.P.; Horng, P.L. Transformation products of hindered phenolic antioxidants and color development in polyolefins. Polym. Deg. Stab. 1991, 34, 333–346. 49. Al-Malaika, S.; Goodwin, C.; Issenhuth, S.; Burdick, D. The antioxidant role of a-tocopherol in polymers II- melt stabilizing effect in polypropylene. Polym. Deg. Stab. 1999, 64 (1), 141–156. 50. Yachigo, S.; Ida, K.; Saski, M.; Inoue, K.; Tanaka, S. Studies on polymer stabilizers. Part V. Influences of structural factors on oxidative discoloration and thermal stability of polymers. Polym. Deg. Stab. 1993, 39, 317–328. 51. Nesvadba, P.; Krohnke, C. A new class of highly active phosphorous free processing stabilizers for polymers. Proceedings of the Sixth International Conference Additives 97, ECM: New Orleans, 1997. 52. Zweifel, H. Stabilisation of Polymeric Materials; Springer-Verlag: Berlin, Germany, 1998; 53 pp. 53. Horsey, D. Hydroxylamines, a new class of low color stabilizers for polyolefins. Proceedings of the Fifth International Conference Additives 96, ECM: Houston, 1996.

A

100

54. Henman, T.J. Melt stabilization of polypropylene. In Developments in Polymer Stabilisation; Scott, G., Ed.; Applied Science: London, 1979; Vol. 1, 39–99. 55. Rabek, J.F. Mechanisms of Photophysical Processes and Photochemical Reactions in Polymers: Theory and Applications; Wiley: New York, USA, 1987; 594 pp. 56. Vink, P. Loss of UV stabilizers from polyolefins during photo-oxidation. In Developments in Polymer Stabilisation; Scott, G., Ed.; Applied Science: London, 1980; Vol. 3, 117–138. 57. Scott, G. Stable radicals as catalytic antioxidants in polymers. In Developments in Polymer Stabilisation; Scott, G., Ed.; Elsevier Applied Science: London, 1984; Vol. 7, 65–104. 58. Al-Malaika, S.; Scott, G. Photostabilization of polyolefins. In Degradation and Stabilisation of Polyolefins; Allen, N.S., Ed.; Applied Science: London, 1983; 283–335. 59. Burton, G.W.; Le Page, Y.; Gabe, E.J. et al. Antioxidant activity of vitamin E and related phenols: importance of stereoelectronic factors. J. Am. Chem. Soc. 1980, 102 (26), 7791–7792. 60. Al-Malaika, S.; Ashley, H.; Issenhuth, S. The antioxidant role of a-tocopherol in polymers. I The nature of transformation products of atocopherol formed during melt processing of LDPE. J. Polym. Sci. Part A, Polym. Chem. 1994, 32, 3099–3113. 61. Al-Malaika, S.; Issenhuth, S. The antioxidant role of a-tocopherol in polymers. III. Nature of transformation products during polyolefins extrusion. Polym. Deg. Stab. 1999, 65 (1), 143–151. 62. Al-Malaika, S.; Issenhuth, S.; Burdick, D. The antioxidant role of vitamin E in polymers Vseparation of stereoisomers and characterization of oxidation products of dl-a-tocopherol formed in polyolefins during melt processing. Polym. Deg. Stab. 2001, 73 (3), 491–503. 63. Pospisil, J. Antioxidants and related stabilizers. In Oxidation Inhibition of Organic Materials; Klemchuk, P., Pospisil, J., Eds.; CRC Press: Boca Raton, 1990; Vol. 1, 33–59. 64. Vogl, O.; Albertsson, A.C.; Janovic, Z. Polymerizable, polymeric, polymer-bound (ultraviolet) stabilizers. In Polymer Stabilisation and Degradation; Klemchuk, P., Ed.; ACS

Antioxidants

65.

66.

67.

68.

69.

70.

71.

72.

73.

74.

Symposium Series-280, American Chemical Society: Washington, 1985; 197–210. Bartus, J.; Goman, P.; Sustic, A. Polym. Prep., Div. Polym. Chem., Am. Chem. Soc. 1993, 34 (2), 158–159. Munteanu, D. Polyolefin stabilization by grafting. In Developments in Polymer Stabilisation; Scott, G., Ed.; Elsevier Applied Science: London, 1987; Vol. 8, 179–208. Al-Malaika, S. Reactive antioxidants for polymers. In Reactive Modifiers in Polymers ; AlMalaika, S., Ed.; Blackie Academic Professional: London, 1997; 266–302. Scott, G. Mechanochemical modification of polymers by antioxidants and stabilizers. In Developments in Polymer Stabilisation; Scott, G., Ed.; Elsevier Applied Science: London, 1987; Vol. 8, 209–238. Al-Malaika, S.; Suharty, N. Reactive processing of polymers: mechanisms of grafting reaction of functional antioxidants on polyolefins in the presence of a coagent. Polym. Deg. Stab. 1995, 49 (1), 77–89. Al-Malaika, S.; Scott, G.; Wirjosentono, B. Mechanisms of antioxidant action: polymerbound hindered amines by reactive processing. III. Effects of reactive antioxidant structure. Polym. Deg. Stab. 1993, 40 (2), 233–238. Al-Malaika, S.; Ibrahim, A.Q.; Rao, J.; Scott, G. Mechanisms of antioxidant action: photoantioxidant activity of polymer-bound hindered amines. II. bis acrylates. J. Appl. Polym. Sci. 1992, 44, 1287–1296. Bauer, I.; Habicher, W.D.; Rauteneberg, C.; Al-Malaika, S. Effects of antioxidants and stabilizers. In Comprehensive Polymer Science; Eastmond, G.C., Ledwith, A., Russo, S., Sigwalt, P., Eds.; Pergamon: New York, 1989; Vol. 6, 539–578. Bauer, I.; Habicher, W.D.; Rauteneberg, C.; et al. Antioxidant interaction between organic phosphites and hindered amine light stabilizers during processing and thermoxidation of polypropylene. Polym. Deg. Stab. 1995, 48 (3), 427–440. Gugumus, F. Developments in the uv stabilization of polymers. In Developments in Polymer Stabilisation; Scott, G., Ed.; Elsevier Applied Science: London, 1987; Vol. 8, 209–238.

Biocatalysis B Tyler Johannes Michael R. Simurdiak Huimin Zhao Department of Chemical and Biomolecular Engineering, University of Illinois, Urbana, Illinois, U.S.A.

INTRODUCTION

THE SCOPE OF BIOCATALYSIS

Biocatalysis may be broadly defined as the use of enzymes or whole cells as biocatalysts for industrial synthetic chemistry. They have been used for hundreds of years in the production of alcohol via fermentation, and cheese via enzymatic breakdown of milk proteins. Over the past few decades, major advances in our understanding of the protein structure–function relationship have increased the range of available biocatalytic applications. In particular, new developments in protein design tools such as rational design and directed evolution have enabled scientists to rapidly tailor the properties of biocatalysts for particular chemical processes. Rational design involves rational alterations of selected residues in a protein to cause predicted changes in function, whereas directed evolution, sometimes called irrational design, mimics the natural evolution process in the laboratory and involves repeated cycles of generating a library of different protein variants and selecting the variants with the desired functions (see the entry ‘‘Protein Design’’). Enzyme properties such as stability, activity, selectivity, and substrate specificity can now be routinely engineered in the laboratory. Presently, approximately 100 different biocatalytic processes are implemented in pharmaceutical, chemical, agricultural, and food industries.[1] The products range from research chemicals to commodity chemicals and the number of applications continue to grow very rapidly. In spite of these successes, however, the vast potential of biocatalysis has yet to be fully realized. In this entry, we briefly outline the scope of biocatalysis and discuss its advantages and disadvantages as compared to chemical catalysis. We then review such topics as enzyme and whole-cell based biocatalysis, biocatalysts used in nonaqueous media, biocatalyst immobilization, discovery and engineering of novel enzymes, and hybrid approaches combining chemical and biological synthesis. An overview of the six general classifications of enzymes along with their relative use in industry is discussed. Selected industrial applications of whole-cell based biocatalysis including the production of lactic acid and 1,3-propanediol are also studied.

Advantages and Disadvantages of Biocatalysis vs. Chemical Catalysis

Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120017565 Copyright # 2006 by Taylor & Francis. All rights reserved.

Similar to other catalysts, biocatalysts increase the speed in which a reaction takes place but do not affect the thermodynamics of the reaction. However, they offer some unique characteristics over conventional catalysts (Table 1). The most important advantage of a biocatalyst is its high selectivity. This selectivity is often chiral (i.e., stereo-selectivity), positional (i.e., regio-selectivity), and functional group specific (i.e., chemo-selectivity). Such high selectivity is very desirable in chemical synthesis as it may offer several benefits such as reduced or no use of protecting groups, minimized side reactions, easier separation, and fewer environmental problems. Other advantages, like high catalytic efficiency and mild operational conditions, are also very attractive in commercial applications. The characteristics of limited operating regions, substrate or product inhibition, and reactions in aqueous solutions have often been considered as the most serious drawbacks of biocatalysts. Many of these drawbacks, however, turn out to be misconceptions and prejudices.[2,3] For example, many commercially used enzymes show excellent stability with half-lives of months or even years under process conditions. In addition, there is an enzyme-catalyzed reaction equivalent to almost every type of known organic reaction. Many enzymes can accept non-natural substrates and convert them into desired products. More importantly, almost all of the biocatalyst characteristics can be tailored with protein engineering and metabolic engineering methods (refer to the section Biocatalyst Engineering and see also the entry ‘‘Protein Design’’) to meet the desired process conditions. Biocatalytic processes are similar to conventional chemical processes in many ways. However, when considering a biocatalytic process one must account for enzyme reaction kinetics and enzyme stability for single-step reactions, or metabolic pathways for multiple-step reactions.[4] Fig. 1 shows the key steps 101

102

Biocatalysis

Table 1 Advantages and disadvantages of biocatalysis in comparison with chemical catalysis Advantages

Disadvantages

Generally more efficient (lower concentration of enzyme needed)

Susceptible to substrate or product inhibition

Can be modified to increase selectivity, stability, and activity

Solvent usually water (high boiling point and heat of vaporization)

More selective (types of selectivity: chemo-selectivity, regio-selectivity, diastereo-selectivity, and enantio-selectivity)

Enzymes found in nature in only one enantiomeric form

Milder reaction conditions (typically in a pH range of 5–8 and temperature range of 20–40
C)

Limiting operating region (enzymes typically denatured at high temperature and pH)

Environment friendly (completely degraded in the environment)

Enzymes can cause allergic reactions

(From Ref.[2].)

in the development of a biocatalytic process. It usually starts with the identification of a target reaction, followed by biocatalyst discovery, characterization, engineering, and process modeling. In many cases, biocatalyst engineering is the most timeconsuming step, often involving two major approaches: rational design and directed evolution. In addition to biocatalyst development, product isolation is an important step. The overall process economics depends on all these factors, which needs to be demonstrated in a pilot-scale plant before scale-up. Biocatalysts can constitute a significant portion of the operating budget; however, their cost can be reduced by reusing them when immobilized (refer to the section Biocatalyst Immobilization).

Enzyme Based Biocatalysis vs. Whole-Cell Biocatalysis Both isolated enzymes and whole cells can be used as biocatalysts. Compared to whole cells, isolated enzymes offer several benefits, including simpler reaction apparatus, higher productivity owing to higher catalyst concentration, and simpler product purification.[2] Until recently, only enzymes that were abundantly produced by cells could be used in industrial applications. Now it is possible to produce large amounts of an enzyme through the use of recombinant DNA technology. In brief, the DNA sequence encoding a given enzyme is cloned into an expression vector and transferred into a production host such as

Fig. 1 Flowchart of the development of a biocatalytic process.

Biocatalysis

Escherichia coli or Saccharomyces cerevisiae for gene expression. The overexpressed enzymes are purified from the cell extracts based on their chemical and physical properties. The most commonly used enzyme purification techniques include electrophoresis, centrifugation, and chromatography. Centrifugation separates enzymes based on their differences in mass or shape, whereas electrophoresis separates enzymes based on their differences in charge. Liquid chromatography separates enzymes based on their differences in charge (ion-exchange chromatography), in mass (gel filtration chromatography), or in ligand-binding property (affinity chromatography). The whole-cell biocatalysis approach is typically used when a specific biotransformation requires multiple enzymes or when it is difficult to isolate the enzyme. A whole-cell system has an advantage over isolated enzymes in that it is not necessary to recycle the cofactors (nonprotein components involved in enzyme catalysis). In addition, it can carry out selective synthesis using cheap and abundant raw materials such as cornstarches. However, whole-cell systems require expensive equipment and tedious work-up because of large volumes, and have low productivity. More importantly, uncontrolled metabolic processes may result in undesirable side reactions during cell growth. The accumulation of these undesirable products as well as desirable products may be toxic to the cell, and these products can be difficult to separate from the rest of the cell culture. Another drawback to whole-cell systems is that the cell membrane may act as a mass transport barrier between the substrates and the enzymes.

Nonaqueous Biocatalysis Historically, enzymes have been used extensively in aqueous media. Enzymes are well suited to their natural aqueous environment; however, biotransformations in industrial synthesis often involve organic molecules insoluble in water. More importantly, because of its high boiling point and high heat of vaporization, water is usually the least desired solvent of choice for most organic reactions. Thus, shifting enzymatic reactions from an aqueous to an organic medium is highly desired. Over the past 15 yr, studies have shown that enzymes can work in organic solvents.[5] However, the enzymatic activity is quite low in an organic solvent compared to that in water. Recent advances in protein engineering and directed evolution have aided in the development of enzymes that show improved activity in organic solvents. Progress has also been made in developing simple, scalable, and low-cost techniques to produce highly active biocatalyst preparations for use

103

in organic solvents.[6] One such method improves enzyme activity in organic solvents by lyophilizing (freeze-drying) an aqueous biocatalyst solution in the presence of organic and inorganic molecules called excipients. These excipients include nonbuffer salts, crown ethers, cyclodextrins, and solid-state buffers.[7] Some remarkable results have also been achieved by using ionic liquids as solvents in biocatalytic reactions.[8] Biocatalyst Immobilization Immobilization is the process of adhering biocatalysts (isolated enzymes or whole cells) to a solid support. The solid support can be an organic or inorganic material, such as derivatized cellulose or glass, ceramics, metallic oxides, and a membrane. Immobilized biocatalysts offer several potential advantages over soluble biocatalysts, such as easier separation of the biocatalysts from the products, higher stability of the biocatalyst, and more flexible reactor configurations. In addition, there is no need for continuous replacement of the biocatalysts. As a result, immobilized biocatalysts are now employed in many biocatalytic processes. More than one hundred techniques for immobilizing enzymes have been developed which can be divided into five major groups summarized in Table 2.[9] Adsorption of the enzyme onto a surface is the easiest and the oldest method of immobilization. Entrapment and cross-linking tend to be more laborious enzyme fixation methods, but they do not require altering the enzyme as much as other techniques. The formation of the covalent linkage often requires harsh conditions, which can result in a loss of activity because of conformational changes of the enzyme. It is important to note that most of these techniques can also be used to immobilize whole cells. In addition, although these types of immobilization are considered to be relatively old and well established, the emerging field of nanotube biotechnology has created another possible means of immobilizing biocatalysts.[10] Biocatalyst Discovery: Sources and Techniques Traditionally, potentially commercial enzymes are identified by screening micro-organisms, which are frequently isolated from extreme environments, for biocatalytic activity. Commercial enzymes are selected by probing libraries of related enzymes for a range of properties, including activity, substrate specificity, stability over a temperature range, enantio-selectivity, or compatibility under various physical and chemical conditions. Unfortunately, most of the commercially viable enzymes have been isolated in only a few microbial species such as Bacillus and Pseudomonas because

B

104

Biocatalysis

Table 2 Methods of enzyme immobilization A. Covalent attachment Isolated enzymes usually attached through amino or carboxyl groups to a solid support Variety of supports such as porous glass, cellulose, ceramics, metallic oxides B. Adsorption Ion-exchangers frequently used in industry because of simplicity Industrial applications include anion-exchangers diethylaminoethyl cellulose (DEAE-cellulose) and the cation-exchanger carboxymethyl cellulose (CM-cellulose) C. Entrapment in polymeric gels Enzyme becomes trapped in gel volume by changing temperature or adding gel-inducing chemical Enzymes may be covalently bound to gel (for instance, polyacrylamide cross-linked with N,N 0 -methylenebisacrylamide) or noncovalently linked (calcium alginate) D. Intermolecular cross-linking Enzyme cross-linked with bifunctional reagents Popular cross-linkers are glutaraldehyde, dimethyl adipimidate, dimethyl suberimidate, and aliphatic diamines E. Encapsulation Enzymes enveloped in semipermeable membrane, which allows low molecular weight substrates and products to pass through the membrane Enclosed in a variety of devices: hollow fibers, cloth fibers, microcapsules, film (From Ref.[9].)

of the limitations in micro-organism cultivation techniques.[11] It has been widely acknowledged that the majority of microbial species (up to 99%) have never been cultivated and thus have never been investigated. To access this vast untapped microbial diversity, several companies such as Diversa Corporation (San Diego, California, U.S.A.) and TerraGen Discovery (Vancouver, British Columbia, Canada) have successfully developed modern bioprospecting techniques such as multiple metagenome cloning to isolate novel industrial enzymes.[12] New methods for exploring natural biodiversity have been greatly facilitated by high-throughput screening technologies and robust expression in recombinant organisms. Recombinant DNA technology makes it possible to produce enzymes at levels 100-fold greater than native expression and allows expression of genes from organisms that cannot be cultured. Although some problems may be resolved by screening larger libraries of DNA, this may not be the most efficient or expedient method of obtaining a viable biocatalyst. A more efficient means of obtaining a good biocatalyst may involve engineering the catalyst itself using various protein engineering and metabolic engineering techniques (refer to the section Biocatalyst Engineering and see also the entry ‘‘Protein Design’’).

Biocatalyst Engineering Nature has supplied us with a vast array of biocatalysts capable of catalyzing numerous biological reactions. Unfortunately, naturally occurring biocatalysts are often not optimal for many specific industrial applications,

such as low stability and activity. Moreover, naturally occurring biocatalysts may not catalyze the reaction with the desired non-natural substrates or produce the desired products. To address these limitations, molecular techniques have been developed to create improved or novel biocatalysts with altered industrial operating parameters. It should be noted that, for enzyme based biocatalysts, many molecular techniques have been developed for engineering enzymes with novel or improved characteristics. Readers are referred to the entry ‘‘Protein Design.’’ In this section, we mainly discuss the molecular techniques used for whole-cell based biocatalyst engineering, or metabolic engineering. Metabolic engineering is a rapidly growing area with great potential to impact biocatalysis.[13] It has been broadly defined as ‘‘the directed improvement of product formation or cellular properties through modifications of specific biochemical reaction(s) or the introduction of new one(s) with the use of recombinant DNA technology.’’[14] In an industrial context, the ultimate goal of metabolic engineering is the development of optimal biocatalysts. In the past two decades, metabolic engineering has been successfully used to engineer micro-organisms to produce a wide variety of products, including polymers, aromatics, carbohydrates, organic solvents, proteins, antibiotics, amino acids, and organic acids. According to the approach taken or the aim, these applications can be classified into seven groups: 1) expression of heterologous genes for protein production; 2) extension of the range of substrate for cell growth and product formation; 3) design and construction of pathways for the production of new products; 4) design and

Biocatalysis

105

construction of pathways for degradation of xenobiotics; 5) engineering of cellular physiology for process improvement; 6) elimination or reduction of byproduct formation; and 7) improvement of yield or productivity.[15] Several of these applications have been implemented at industrial-production scale (refer to the section Industrial Applications of Whole-Cell Based Biocatalysis) and the number of applications should continue to grow. In particular, with the recent advances in genomics, proteomics, and bioinformatics, many new genes and pathways will be discovered and the regulation of metabolic network will also be better understood, all of which will accelerate the development of more commercially viable bioprocesses through metabolic engineering.

Hybrid Approaches Combining Chemical Synthesis and Biocatalysis Biocatalysts exhibit exquisite catalytic efficiency that is often unmatched by conventional catalysis. Nonetheless, conventional organic synthesis will likely remain the staple of the chemical and pharmaceutical industries. In the future, the integration of these two approaches will probably offer the optimal route for industrial synthesis. An illustration of this principle can be found in the selective deprotection of reactive functional groups. Enzymes are unique deprotecting tools for combinatorial synthesis because of their remarkable selectivity and ability to operate under mild reaction conditions. A recent example is the synthesis of long multiply lipidated peptides containing various side-chain functional groups.[16] In this study, penicillin acylase was used for selective N-deprotection of a highly labile S-palmitoylated oligopeptide. After removal of the protecting group, the S-palmitoylated oligopeptide was used as a building block in further synthetic steps.

INDUSTRIAL APPLICATIONS OF ENZYME BASED BIOCATALYSIS With the rapid technical developments in gene discovery, optimization, and characterization, enzymes have been increasingly used as biocatalysts. According to the International Union of Biochemistry and Molecular Biology (IUBMB) nomenclature system, all enzymes are classified into six classes on the basis of the general type of reactions that they catalyze (Table 3). Within each class are subclasses and the enzymes themselves. The result is an ordered system of enzymes and the reaction(s) that each catalyzes. It is important to note that, in biological processes, every class of enzyme is utilized in the cell to a large extent. However, this is not the same in industrial processes, where certain classes of enzymes are used more often than others. As shown in Fig. 2, most of the enzymes that have been used as biocatalysts in industry are hydrolases (65%), even though oxidoreductases are typically much more useful than hydrolases as catalysts. The utility of an enzyme class depends on the relative commercial importance of the products that each enzyme produces, the accessibility of the enzymes, and the specific characteristics of the enzymes (e.g., stability, activity, and selectivity).

Oxidoreductases Oxidoreductases catalyze oxidation and reduction reactions that occur within the cell. They are very appealing for industrial uses because of the reactions that they are able to catalyze. However, they often need expensive cofactors such as nicotinamide adenine dinucleotides (e.g., NADþ=NADH) and flavines (e.g., FAD=FADH2) in the reactions. In fact, nicotinamide adenine dinucleotides are required by about 80% of oxidoreductases. Fortunately, several NAD(H)

Table 3 Classification of enzymes Enzymes

Type of reactions

Representative subclasses

Oxidoreductases

Catalyze the transfer of hydrogen or oxygen atoms or electrons from one substrate to another

Oxidases, oxygenases, peroxidase, dehydrogenases

Transferases

Catalyze the group transfer reactions

Glycosyltransferases, transketolases, methyltransferases, transaldolases, acyltransferases, transaminases

Hydrolases

Catalyze hydrolytic reactions

Esterases, lipases, proteases, glycosidases, phosphatases

Lyases

Catalyze the nonhydrolytic removal of groups

Decarboxylases, aldolases, ketolases, hydratases, dehydratases

Isomerases

Catalyze isomerization reactions

Racemases, epimerases, isomerases

Ligases

Catalyze the synthesis of various types of bonds with the aid of energy-containing molecules

Synthetases, carboxylases

B

106

Biocatalysis

Fig. 4 Enzymatic synthesis of L-tert-leucine.

process is carried out in a membrane reactor in which the cofactor NADþ is regenerated by FDH. This process has now reached ton-scale production at Degussa (Germany).[18]

Transferases Fig. 2 The relative use of enzyme classes in industry. (From Ref.[2].)

regeneration systems have been developed, the most widely used being the formate=formate dehydrogenase (FDH) system.[17] An example of a pharmaceutical synthesis reaction involving an oxidoreductase is the synthesis of 3,4-dihydroxylphenyl alanine (DOPA).[2] 3,4-Dihydroxylphenyl alanine is a chemical used in the treatment of Parkinson’s disease. The industrial process that synthesizes DOPA utilizes the oxidoreductase polyphenol oxidase. As shown in Fig. 3, the monohydroxy compound is oxidized by the regio-specific addition of a hydroxyl group. It is worth mentioning that epinephrine (adrenaline) can also be synthesized by a similar reaction path using the same enzyme.[2] Another example is the use of leucine dehydrogenase coupled with FDH for the reductive amination of trimethylpyruvate to L-tert-leucine (Fig. 4). The whole

Fig. 3 Enzymatic synthesis of 3,4-dihydroxylphenyl alanine (DOPA).

Transferases catalyze the transfer of functional groups such as methyl, hydroxymethyl, formal, glycosyl, acyl, alkyl, phosphate, and sulfate groups by means of a nucleophilic substitution reaction. They are not widely used in industrial processes; however, there are a few examples of industrial processes that utilize transferases. A classical example of industrial application of transferases is the use of various glycosyltransferases for the synthesis of oligosaccharides. Oligosaccharides and polysaccharides are important classes of naturally occurring compounds, which play vital roles in cellular recognition and communication processes.[19] Because of the required use of many protection and deprotection groups, chemical synthesis of complex oligosaccharides represents a daunting challenge in synthetic organic chemistry. By contrast, enzymatic synthesis of oligosaccharides by glycosyltransferases requires very few protection and deprotection steps because of the high regio- and stereoselectivity of glycosyltransferases, thus offering an attractive alternative.[2] Another example is the use of a glucokinase (a transferase) in combination with an acetate kinase for the production of glucose-6-phosphate (Fig. 5). This process is carried out in multikilogram scale by the Japanese company Unitika.[1]

Fig. 5 Enzymatic synthesis of glucose-6-phosphate.

Biocatalysis

107

Hydrolases Hydrolases catalyze the addition of water to a substrate by means of a nucleophilic substitution reaction. Hydrolases (hydrolytic enzymes) are the biocatalysts most commonly used in organic synthesis. They have been used to produce intermediates for pharmaceuticals and pesticides, and chiral synthons for asymmetric synthesis. Of particular interest among hydrolases are amidases, proteases, esterases, and lipases. These enzymes catalyze the hydrolysis and formation of ester and amide bonds. Lipases can hydrolyze triglycerides into fatty acids and glycerol. They have been used extensively to produce optically active alcohols, acids, esters, and lactones by kinetic resolution. Lipases are unique, in that they are usually used in two-phase systems. A classic example is the use of a lipase for the production of (S,R)-2,3-p-methoxyphenylglycyclic acid, an intermediate for diltiazem. In this process, methylp-methoxyphenylglycidate is stereospecifically hydrolyzed by a lipase immobilized in a hollow fiber membrane reactor. The enzyme is located at the interfacial layer between an organic and an aqueous phase.[1] Proteases such as a-chymotrypsin, papain, and subtilisin are also useful biocatalysts for regio-selective or stereoselective hydrolytic biotransformations. For example, dibenzyl esters of aspartic and glutamic acid can be selectively deprotected at the 1-position by subtilisin-catalyzed hydrolysis (Fig. 6).[2] In addition, a-chymotrypsin is used in the kinetic resolution of a-nitro-a-methyl carboxylates, which results in Lconfigured enantiomers of the unhydrolyzed esters with high optical purity (>95% e.e.).[2]

Lyases Lyases are the enzymes responsible for catalyzing addition and elimination reactions. Lyase-catalyzed reactions involve the breaking of a bond between a carbon atom and another atom such as oxygen, sulfur, or another carbon atom. They are found in cellular processes, such as the citric acid cycle, and in organic synthesis, such as in the production of cyanohydrins.[2]

Fig. 6 Regio-selective ester-hydrolysis catalyzed by subtilisin.

B Fig. 7 Enzymatic synthesis of (S)-malic acid.

Several industrial processes using lyases as catalysts have been reported. Perhaps the most prominent lyasecatalyzed process is the production of acrylamide from acrylnitrile. This process is carried out by the Nitto Chemical Company of Japan at a scale of more than 40,000 tons per year.[20] Another example is the use of a fumarase for the production of (S)-malic acid from fumaric acid. As shown in Fig. 7, a water molecule is added to the double bond in fumarate by means of an addition reaction. The result is a cleavage of the carbon–carbon double bond, and a formation of a new carbon–oxygen bond. A third example is biocatalytic production of a cyanohydrin from a ketone. This reaction is catalyzed by a lyase called oxynitrilase. It consists of the cleavage of one carbon–oxygen bond, and the addition of a HCN molecule. The chirality of the product is based on the form of the enzyme used (R-oxynitrilase or S-oxynitrilase).[2] Isomerases Isomerases catalyze isomerization reactions such as racemization and epimerization. They have not been used in many industrial applications. However, one of the most successful enzyme based biocatalytic processes involves an isomerase: the use of glucose isomerase for the production of high-fructose corn syrup (HFCS) (Fig. 8). High-fructose corn syrup is used as an alternative sweetener to sucrose in the food and beverage industry. The isomerization of glucose to HFCS on an industrial scale is carried out in continuous fixed-bed reactors using immobilized glucose isomerases. The total amount of HFCS produced by glucose isomerase exceeds a million tons per year.[21] Ligases Ligases catalyze reactions that involve the creation of chemical bonds with nucleotide triphosphates. They

Fig. 8 Enzymatic synthesis of high-fructose corn syrup.

108

are important in certain cellular processes, such as connecting nucleotides in DNA replication. However, similar to isomerases, ligases have very few industrial applications.[2] It is important to note that DNA ligases are essential tools in recombinant DNA technology and are used almost in every biology-related laboratory.

INDUSTRIAL APPLICATIONS OF WHOLE-CELL BASED BIOCATALYSIS Whole-cell based biocatalysis utilizes an entire microorganism for the production of the desired product. One of the oldest examples for industrial applications of whole-cell biocatalysis is the production of acetic acid from ethanol with an immobilized Acetobacter strain, which was developed nearly 200 yr ago.[1] The key advantage of whole-cell biocatalysis is the ability to use cheap and abundant raw materials and catalyze multistep reactions. Recent advances in metabolic engineering have brought a renaissance to whole-cell biocatalysis. In the following sections, two novel industrial processes that utilize whole-cell biocatalysis are discussed with emphasis on the important role played by metabolic engineering. Lactic Acid L-lactic

acid has long been used as a food additive and has recently received great attention because it can be used as an important feedstock for the production of other chemicals such as polylactic acid (PLA), acetaldehyde, polypropylene glycol, acrylic acid, and pentadione.[22] Among them, PLA is the most important product as it can be used to manufacture thermoformed containers, packaging, nonwovens, papercoated articles, and film products.[23] Lactic acid can be produced from sucrose, whey (lactose), and maltose or dextrose from hydrolyzed starch using Lactobacillus strains. Compared to other polymeric materials such as polyethylenes, polylactic acid has several advantages including an increased hydrophilicity, resistance to ultraviolet light, ability to be dyed with dispersion dyes, a range of melting temperatures between 120
C and 170
C, and low flammability and smoke generation. Most importantly, polylactic acid is biodegradable and is derived from renewable resources, utilizing energy from the sun and lowering the fossil fuel dependence for production.[23] In recognition of the superior properties and the huge potential market of polylactic acid, Cargill Inc. and The Dow Chemical Company started a joint venture Cargill Dow LLC to produce lactic acid using

Biocatalysis

fermentation (whole-cell biocatalysis) several years ago. A production plant was built in Blair, NE, in 2001 and it now produces 140,000 metric tons of polylactic acid per year. It is predicted that the eventual cost of polylactic acid will be between $0.50 and $0.75 per pound (http:==www.cargilldow.com). One of the drawbacks in the current commercial fermentation process is that the predominant form of the product is the deprotonated lactate rather than lactic acid, requiring more expensive and wasteful product purification steps. This is because the Lactobacillus fermentation operates at a minimum pH of 5.0–5.5 which is above the pKa of lactic acid (3.87). To overcome this limitation, a powerful strain improvement method, genome shuffling, was used to improve the acid tolerance of a poorly characterized industrial strain of Lactobacillus.[24] A population of strains with subtle improvement in pH tolerance was isolated using classical strain improvement methods such as chemostats, and were then shuffled by recursive pool-wise protoplast fusion to create mutant strains that grow at substantially lower pH than does the wild-type strain.

1,3-Propanediol 1,3-Propanediol is an intermediate that is widely used in the synthesis of polyesters and polyurethanes. Polymers based on 1,3-propanediol are very useful in the carpet and textile industry because of their good light stability and biodegradability.[25] The conventional methods for producing 1,3-propanediol rely on petroleum derivatives and are quite capital intensive and=or generate waste streams containing environmental pollutants. Thus, the use of micro-organisms to produce 1,3-propanediol from glucose represents an attractive alternative. Both the biological production of 1,3-propanediol from glycerol and that of glycerol from glucose have been known for many years.[13] However, there is no single micro-organism that could convert basic carbon sources such as glucose to the desired 1,3-propanediol end-product. Such a micro-organism is highly desired in the process as it requires less energy input and uses an inexpensive starting material. A team of researchers from DuPont and Genencor has successfully used metabolic engineering techniques to engineer such a micro-organism. The conversion of glucose to 1,3-propanediol requires the combination of two natural pathways: glucose to glycerol and glycerol to 1,3-propanediol. The best natural pathway for the production of glycerol from glucose was found in the yeast Saccharomyces cerevisiae, which consists of two enzymes: dihydroxyacetone-3-phosphate dehydrogenase and glycerol-3-phosphate phosphatase. The best natural

Biocatalysis

pathway for production of 1,3-propanediol from glycerol was found in Klebsiella pneumoniae, which consists of glycerol dehydratase and 1,3-propanediol dehydrogenase. The genes encoding these two natural pathways were cloned and expressed in E. coli. E. coli was chosen as the production strain because it has been used in large-scale production on an industrial level, it has many genetic tools, and its metabolism and physiology are well characterized. This engineered E. coli was found to produce over 120 g=L of 1,3-propanediol in 40 hr fed-batch fermentation.[13]

CONCLUSIONS Biocatalysis has become an important tool for industrial chemical synthesis and is on the verge of significant growth. In the past several decades, many biocatalytic processes have been implemented to produce a wide variety of products in various industries. Most of them use naturally occurring enzymes or micro-organisms as catalysts. With the help of innovative biocatalyst discovery methods and advances in protein engineering and metabolic engineering, the time and cost of developing new biocatalysts can be reduced significantly. Most importantly, the biocatalysts can be readily tailored to their specific applications and process conditions through protein engineering and metabolic engineering. It is possible that in the future they can be rationally designed to act specifically in any chemical reaction of interest, fulfilling the holy grail of catalysis: catalysis by design. In addition, the use of biocatalysts in organic solvents in combination with the integration of biocatalysis and chemical catalysis will continue to broaden the scope of the applications of biocatalysts. New advances in genomics, proteomics, and bioinformatics will fuel the development of biocatalysis as an integral part of industrial catalysis.

REFERENCES 1. Wandrey, C.; Liese, A.; Kihumbu, D. Industrial biocatalysis: past, present, and future. Organic Process Res. Dev. 2000, 4 (4), 286–290. 2. Faber, K. Biotransformations. In Organic Chemistry: A Textbook, 3rd Ed.; Springer-Verlag: Berlin, Germany, 1997. 3. Rozzell, J.D. Commercial scale biocatalysis: myths and realities. Bioorg. Med. Chem. 1999, 7 (10), 2253–2261. 4. Schmid, A.; Dordick, J.S.; Hauer, B.; Kiener, A.; Wubbolts, M.; Witholt, B. Industrial biocatalysis today and tomorrow. Nature 2001, 409 (6817), 258–268.

109

5. Klibanov, A.M. Improving enzymes by using them in organic solvents. Nature 2001, 409 (6817), 241–246. 6. Schoemaker, H.E.; Mink, D.; Wubbolts, M.G. Dispelling the myths—biocatalysis in industrial synthesis. Science 2003, 299 (5613), 1694–1697. 7. Lee, M.Y.; Dordick, J.S. Enzyme activation for nonaqueous media. Curr. Opin. Biotechnol. 2002, 13 (4), 376–384. 8. Kragl, U.; Eckstein, M.; Kaftzik, N. Enzyme catalysis in ionic liquids. Curr. Opin. Biotechnol. 2002, 13 (6), 565–571. 9. Klibanov, A.M. Immobilized enzymes and cells as practical catalysts. Science 1983, 219 (4585), 722–727. 10. Martin, C.R.; Kohli, P. The emerging field of nanotube biotechnology. Nat. Rev. Drug Discovery 2003, 2 (1), 29–37. 11. Dalboge, H.; Lange, L. Using molecular techniques to identify new microbial biocatalysts. Trends Biotechnol. 1998, 16 (6), 265–272. 12. Burton, S.G.; Cowan, D.A.; Woodley, J.M. The search for the ideal biocatalyst. Nat. Biotechnol. 2002, 20 (1), 37–45. 13. Chotani, G.; Dodge, T.; Hsu, A.; Kumar, M.; LaDuca, R.; Trimbur, D.; Weyler, W.; Sanford, K. The commercial production of chemicals using pathway engineering. Biochim. Biophys. Acta 2000, 1543 (2), 434–455. 14. Stephanopoulos, G.N.; Aristidou, A.A.; Nielsen, J. Metabolic Engineering: Principles and Methodologies; Academic Press: London, U.K., 1998. 15. Nielsen, J. Metabolic engineering. Appl. Microbiol. Biotechnol. 2001, 55 (3), 263–283. 16. Machauer, R.; Waldmann, H. Synthesis of lipidated eNOS peptides by combining enzymatic, noble metal- and acid-mediated protecting group techniques with solid phase peptide synthesis and fragment condensation in solution. Chemistry 2001, 7 (13), 2940–2956. 17. Chenault, H.K.; Whitesides, G.M. Regeneration of nicotinamide cofactors for use in organic synthesis. Appl. Biochem. Biotechnol. 1987, 14 (2), 147–97. 18. Kragl, U.; VasicRacki, D.; Wandrey, C. Continuous production of L-tert-leucine in series of two enzyme membrane reactors—modelling and computer simulation. Bioprocess Eng. 1996, 14 (6), 291–297. 19. Ginsburg, V., Robbins, P.W., Eds. Biology of Carbohydrates; Wiley: New York, 1984. 20. Zaks, A. Industrial biocatalysis. Curr. Opin. Chem. Biol. 2001, 5 (2), 130–136. 21. Gerhartz, W., Ed. Enzymes in Industry: Production and Applications; VCH: New York, 1990.

B

110

22. Varadarajan, S.; Miller, D.J. Catalytic upgrading of fermentation-derived organic acids. Biotechnol. Prog. 1999, 15 (5), 845–854. 23. Hunt, J. Large-scale production, properties and commercial applications of polylactic acid polymers. Polym. Degrad. Stab. 1998, 59 (1–3), 145–152. 24. Patnaik, R.; Louie, S.; Gavrilovic, V.; Perry, K.; Stemmer, W.P.; Ryan, C.M.; Del Cardayre, S.

Biocatalysis

Genome shuffling of Lactobacillus for improved acid tolerance. Nat. Biotechnol. 2002, 20 (7), 707–712. 25. Zhu, M.M.; Lawman, P.D.; Cameron, D.C. Improving 1,3-propanediol production from glycerol in a metabolically engineered Escherichia coli by reducing accumulation of sn-glycerol3-phosphate. Biotechnol. Prog. 2002, 18 (4), 694–699.

Biofilms B T. Reg. Bott School of Engineering, Chemical Engineering, University of Birmingham, Birmingham, U.K.

INTRODUCTION The term ‘‘biofilm’’ refers to a colony of living matter, usually microorganisms including algae, fungi, or bacteria, together with extracellular material attached to a solid surface. In natural conditions it will almost certainly contain several different microbial species, and also depending on the prevailing conditions, the biofilm may also contain adventitious particulate matter. In processing operations, biofilms can be an asset or a hindrance. In some cases, the biofilm is essential for achieving the objective of the process, e.g., the use of trickle filters for the removal of contamination to improve the quality of potable water.[1] On the other hand, the effectiveness of a cooling water system can be reduced by the presence of biofilms on heat transfer surfaces.[2] Biofilm formation is a natural phenomenon, so that wherever suitable conditions exist, it would be anticipated that a biofilm would develop. Essential, of course, are the presence of nutrients to sustain life and a suitable temperature. Biofilms are formed on surfaces; so the availability of a suitable surface that may be colonized by microorganisms is vital. The structure and the composition of the biofilm are greatly dependent on the environment in which it grows. The composition of the fluid (usually water) in contact with the biofilm, including organic matter, oxygen, and trace elements, its pH, and its flow conditions across solid surfaces, all affect the rate of growth and the robustness of the biofilm structure that develops. The removal of waste products from the vicinity of the biofilm will also depend on the structure of the biofilm, in terms of porosity and thickness, and the flow conditions internal and external to the biofilm. The structure of the biofilm will also influence the availability of nutrients within the biofilm, because of its effect on the penetration of the fluid carrying the nutrients within the biofilm. The biofilm activity may affect the surface on which it resides. For example, the pH near the base of the biofilm may be different from that of the bulk fluid external to the biofilm, because of the metabolism of the microorganisms. This may give rise to corrosion or deterioration of the surface. Clearly, this is an important consideration in relation to the integrity of processing equipment. The initiation and continued growth of a biofilm is an extremely complex process, subject to many internal Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120030621 Copyright # 2006 by Taylor & Francis. All rights reserved.

and external influences. A discussion of these factors forms the basis of this entry.

MICROBIAL ACTIVITY Bacteria may be regarded as more versatile than algae or fungi because they are not limited by the need for light or a consumable substrate. In addition, there is considerable variation among bacteria, which is due, in part at least, to the differences in the properties of cell surface polymers.[3] Fletcher[4] suggests that the bacteria attached to a surface appear to be metabolically different from their planktonic or ‘‘free swimming’’ counterparts. Algae utilize carbon dioxide and inorganic chemicals as their primary source of nutrients, and light (usually sunlight) to photosynthesize sugars. The associated biofilms are generally composed of single cells or filamentous organisms, but other forms can exist depending on the conditions. In natural conditions, algal biofilms are found on rocks or stones, in rivers, or on the seashore. Colonies can also reside on man-made structures such as bridge supports standing in water or offshore oil rigs. Fungi require a fixed organic source of carbon. The rigid cell wall limits them to being saprophytic on organic substrates or as parasites on animals, plants, algae, or even other fungi. In fact, fungi may be found on any solid that provides an organic substrate, provided that the local conditions are satisfactory. Microbial cells are surrounded by a cell wall, which retains the cell contents, and is the primary barrier between the cell surface and the environment in which it exists. The quality of the cell wall, in terms of selective permeability, maintains the necessary levels of nutrients, trace elements, and cell internal pH. The cell membrane is the site of transfer processes: water is able to pass through this membrane, in or out of the cell, depending on the thrust of the osmotic pressure. The chemistry of the cell wall affects its properties in terms of surface electric charge and the availability of binding ions. In many organisms, the cell wall is rigid, giving a characteristic shape to individual cells such as rod, filament, or sphere. The rigidity of the cell wall allows the development of structures that may be beneficial for the maintenance of a coherent biofilm. Some cells, however, do not have a rigid cell wall and, therefore, require an intrinsic mechanism to control osmotic 111

112

Biofilms

pressure for the prevention of damage by excess water intake. A few freshwater organisms can dispose of water or imbibe it through contractile vacuoles. Certain algae can control the condition of soluble metabolites to counter the effects of osmotic pressure. Microorganisms can produce extracellular materials, such as slimes of polysaccharides and mucilages, which may help to maintain attachment to the solid substrate, provide a source of nutrients if the nutrient availability declines for any reason, or enhance protection of the cells. Clearly, the availability of nutrients will determine whether or not a biofilm can form and develop. In common with all living matter, the elements that constitute microorganisms are associated with organic chemistry in the widest sense, including carbon, hydrogen, oxygen, nitrogen, sulfur, phosphorus, and other inorganic molecules. The requirements are a source of energy, carbon, and reducing chemicals. Although CO2 is an increasing component of the atmosphere, it is used as a nutrient source of carbon only by algae, via photosynthesis. The usual source of carbon for microorganisms is carbohydrates, especially polysaccharides because of their abundance. Nevertheless, other sources used include fats, hydrocarbons such as methane, and proteins. Cellulose is preferred by some fungi. The inertness of atmospheric nitrogen precludes its availability for microorganisms. The usual source is nitrogen-containing compounds such as amino acids, ammonia, nucleotides, uric acid, and urea. Sulfur is plentiful in naturally occurring compounds. Inorganic sulfates can be reduced via the sulfide and incorporated into amino acids. Hydrogen sulfide is used as a source of sulfur by some microorganisms. Organic sulfur may represent an alternative sulfur-containing nutrient. Trace elements, including potassium, magnesium, and iron, are required by microorganisms, while calcium, sodium, and silica may be necessary for the growth of some species. There may also be a requirement for traces of zinc, copper, cobalt, manganese, and molybdenum. The metabolism of microorganisms gives a basis for classification:[2] 1. Phototrophs obtain their energy directly from the sun. 2. Chemotrophs acquire their energy from the oxidation of organic as well as inorganic compounds. 3. Autotrophs are capable of synthesizing their cell carbon from simple compounds such as CO2. 4. Heterotrophs require fixed sources of carbon. 5. Lithotrophs produce the reducing equivalents required for cell synthesis from inorganic substances like H2S and ferrous iron. 6. Organotrophs obtain chemical reduction from the oxidation of organic molecules.

Movement to a favorable location for nutrient assimilation is achieved by the use of flagella, as with some bacteria, or by the use of fibrils, as with some algae cells to glide across surfaces. During metabolism, a single cell will take in nutrients allowing the synthesis of macromolecules that provide the basis for growth. As the cell increases in size, it will eventually divide. Cell division in bacteria is well ordered; the daughter cells that result from the division are uniform in dimension and genetic character. The time to reach full size can be very short (as low as a few hours) depending on the prevailing conditions, particularly temperature and nutrient availability. Cell division in some fungi starts with the production of a ‘‘bud’’ on the cell surface. Of course, the bud is much smaller than the original cell, but it grows till it reaches adult dimensions, and the reproductive process is repeated. There is a limit to the number of buds that can be produced by a single parent cell, as only one can be grown at a particular point on the surface of the cell. Fungi grow almost exclusively through what is known as apical growth, though they can also reproduce by means of sexual and asexual cycles. For the development of algal biofilms, as with bacteria, few cells are needed to attach to a surface, as cell division rapidly gives rise to colonies that eventually coalesce to form a compact biofilm.[5]

THE IMPORTANCE OF A SURFACE It would appear that many microorganisms tend to reside on a surface, in discrete colonies, or in a film. It has been suggested[3] that under natural conditions, i.e., in rivers or lakes, 90% of the microorganisms in the biosphere exist in biofilms. Of fundamental importance is that the microorganisms contained in a developing biofilm are exposed to a continuous supply of nutrients. The disturbance of the surface of the water source (e.g., a lake) in contact with the atmosphere ensures there is aeration, which would be advantageous for aerobic organisms. Furthermore, the movement of the water would assist the removal of waste products of the metabolic processes in the cells. The advantages experienced by the sessile cells (those attached to a surface) may be contrasted with the depravation of the planktonic cells (those suspended in water) where nutrient availability at the cell surface is largely because of Brownian diffusion, the cells moving with the water flow along the water streamlines. Other advantages that accrue for biofilms on surfaces[4] include protection from short-term fluctuations in pH, salt and biocide concentrations, and dehydration. Because of the mass-transfer resistance of the biofilm to the transport of nutrient, conditions will change throughout the biofilm depth. Where several

Biofilms

different microorganisms are present, it is possible for symbiosis to occur. Symbiosis may be defined as a mutually beneficial partnership between organisms of different kinds. Examples include the utilization and breakdown of less readily degradable substances by specialized organisms, which provides a source of nutrients for other microorganisms, e.g., the degradation of cellulose by fungi. The creation of ecological niches, such as anaerobic zones under aerobic biofilms, allows the growth of anaerobic cells in otherwise aerobic environments.

COLONIZATION OF SURFACES It is generally accepted that the colonization of a surface by microorganisms is a multistage process, as suggested by Characklis.[6] Five steps are involved. Formation of an Organic Conditioning Layer A conditioning layer on the virgin solid surface, newly immersed in water, is created by the adsorption of macromolecules. Chamberlain[7] has studied the importance of these adsorbed macromolecules in the attachment of microbes to solid surfaces. The adsorbed macromolecules on surfaces are generally organic in character, although there may be other compounds included in the adsorbed layer, such as metallic hydroxides and mineral particles. Because macromolecules frequently possess multiple attachment points, either functional groups or segments of more hydrophobic character, their attachment is likely to be irreversible. In freshwater systems, such as cooling water circuits, the macromolecules involved, including humic acids and polysaccharides, are likely to be derived from formally living matter. The humic components originate primarily from lignin-type materials and may contain carboxylic and phenolic residues that impart high reactivity. These components are generally highly aromatic. In seawater used for cooling purposes, humic compounds are somewhat different in character and generally comprise complex condensation molecules with polysaccharide, peptide, and lipid components. In contrast to fresh water systems, the marine humic compounds are generally aliphatic in character. In food-processing streams, the macromolecules are often proteins or glycoproteins, which may bring about the attachment of casein and other milk proteins to stainless steel, the usual material of construction in food processing equipment. The rate of adsorption of macromolecules is, in part at least, controlled by their concentration in the water, their relative affinities for the material with which the surface is made, and the hydrodynamics of the flow across the surface.

113

Transport of Microorganisms to a Surface The transport of microbial cells dispersed in the liquid stream toward the surface is strongly influenced by the regime of flow across the surface, i.e., laminar or turbulent (see item elsewhere in this encyclopedia). The liquid immediately in contact with the solid surface may be regarded as being stationary, because of the viscous drag exerted by the solid surface. The velocity of the liquid layers gradually increases at distances at right angles to the surface. When close to the surface, these layers are slow moving and constitute what is generally known as the viscous or laminar sublayer. In simple terms, if the bulk flow is turbulent, a boundary layer exists between the slow-moving laminar sublayer and the turbulent bulk flow. For microorganisms to approach the surface and become attached to the conditioning layer, they have to pass from the turbulent bulk liquid, through the boundary layer and into the viscous sublayer. Initially, the cells are carried by eddy diffusion (i.e., resulting from the turbulence in the bulk liquid), but as they approach and enter the viscous sublayer, the eddy diffusion is damped out. Within the viscous sublayer itself, the transport mechanism is caused by molecular diffusion or Brownian motion. The thickness of the viscous sublayer is strongly dependent on the velocity in the bulk liquid. The greater the bulk flow, the thinner the viscous sublayer; i.e., the resistance to the transport of the cells to the surface is lower, and hence a more rapid build-up of cells on the surface is possible. Frictional drag forces become significant in the viscous sublayer, which slows the cells as they move towards the surface. Complex interactions between the cells and the effects of the flow conditions are also likely to affect the approach velocity; for instance, the disturbance of the laminar sublayer by the turbulence in the bulk flow. Thermophoresis, where particles move in response to a temperature gradient, may also affect the movement of cells in relation to a surface. Because cells have a large water content, gravity will have little effect on settlement. As already stated, cells may respond to chemical stimuli that influence movement in relation to surfaces, and it is possible that the conditioning layer, because of its composition, attracts cells by chemotaxis, thereby facilitating colonization.

Attachment to a Surface The attachment of particles in a fluid suspension to a solid surface is extremely complex and involves longrange attraction forces to bring the particles to the surface and provide a basis for further interaction. The forces involved may include van der Waals forces

B

114

Biofilms

and electrostatic forces. Because of the complexity, simplifying assumptions are made in an attempt to provide a reasonable explanation of the phenomena involved. The majority of solid surfaces, when immersed in an aqueous solution, acquire an electrical surface charge. As a result, an electrical potential between the surface and the bulk fluid is created. There is an attraction between oppositely charged ions (counterions) in the fluid and the surface. At the same time, similarly charged ions (coions) are repelled away from the surface. The electric discontinuity created at the surface is generally known as the ‘‘electrostatic double layer.’’ Fig. 1 is a simplified illustration of the situation that applies to a spherical colloid particle immersed in water. The diffuse electrostatic double layer plays an important role in the interaction between the solid surface and microorganisms and colloids in suspension. Indeed, microorganisms in suspension have been called ‘‘living colloids,’’ the ‘‘bridging’’ may take the form of chemical bonding with the adsorbed macromolecules on the surface. The basic concepts of particle adhesion theory may be modified where microorganisms are the adhering particles. Mozes,[8] applying adhesion theory to the formation of biofilms, makes the following observations in relation to microbial cells in suspension:

1. It is not a homogeneous, rigid, smooth, and spherical ideal colloid particle. 2. It is not inert and in equilibrium with its environment. 3. It may be capable of independent movement. 4. It may respond physiologically to contact with a surface.

Both reversible and irreversible attachments are possible. Microbial cells can be held in close proximity to a surface by the long-range forces, but still be capable of Brownian motion. Under the flow conditions, mild shear forces, brought about by flow and disturbances in the viscous sublayer, may also be present. These effects may restrict or prevent attachment. It is anticipated that this could be the situation if no conditioning molecules were residing on the surface and would explain the reason for the observed delay in biofilm formation under flowing conditions. Any adhesion of cells under these conditions could be said to be reversible. Once there is a bonding between the conditioning molecules and the microbial cells, removal by shear forces becomes more difficult and the attachment can be regarded as irreversible. As the biofilm develops, however, shear forces caused by the fluid flow may be capable of removing cells from the surface. The extent of this removal is dependent on the quality of the biofilm and the bulk flow velocity. Furthermore, the establishment of irreversible attachment is dependent on the quality of the surface involved. Rough surfaces are known to facilitate the formation of biofilms. If the crevices on the surface are relatively large, a bacterium cell that is only 1 or 2 mm in size, could easily ‘‘hide’’ in a crevice and be unaffected by the shear forces. Under these circumstances, it is possible that there is a greater opportunity to form a bond with the conditioning molecules. Once these ‘‘survivors’’ are attached, the biofilm can begin to develop and irreversible attachment established. In the food industry, great care is taken to make sure that the surfaces of equipment in contact with food products are exceptionally smooth to reduce the risk of the formation of biofilms that could adversely affect food quality.

Nutrient Transport and Assimilation

Fig. 1 An idealized representation of an electrostatic double layer and a solid colloidal particle. (From Ref.[2].)

After colonization of the surface, the microorganisms utilize the available nutrients to grow, multiply, and synthesize both intracellular products and extracellular polymeric substances that constitute the substance of the biofilm. Significant amounts of biofilm can be produced under ideal conditions, and even if planktonic cells are no longer present in the flowing water, the sessile cells already on the surface can provide the basis for biofilm development.[9] The transport of nutrients from the bulk water to the developing biofilm depends on the concentration difference of nutrients between the bulk water and the surface of the biofilm. A concentration driving force is established by the removal of nutrients by the growing biofilm, assuming that the nutrient concentration in the bulk is maintained. As with the initial mass

Biofilms

transfer of cells to the surface, the boundary layers adjacent to the biofilm constitute a resistance to the transport of nutrients to the biofilm; the rate of growth may be limited by this resistance. The extent of the resistance will depend on the flow conditions outside the boundary layers. In broad terms, the higher the bulk velocity, the lower the resistance to mass transfer of nutrients. As the biofilm develops, the nutrient availability to the bulk biofilm may become affected. The biofilm, despite its voids and channels, offers a further resistance to mass transfer. The cells within the biofilm consume nutrients that diffuse through the biofilm in response to the difference in concentration between nutrients at the biofilm surface and the cells attached to the conditioning layer. As a consequence, it is entirely possible that cells in the region of the solid surface are likely to become starved of nutrients. The properties of the biofilm may be different, therefore, in the layers where nutrient is available compared with the regions where there is little or no nutrient. For instance, the lack of oxygen may encourage anaerobic species to develop (some bacteria can exist as aerobes or anaerobes), with attendant changes to the quality of the biofilm. Biofilm Removal As is apparent from the foregoing discussion, the growth of the biofilm is a complex interaction between the flowing fluid and the physiology of the biofilm structure. Furthermore, the properties of the biofilm may change with time as it grows, resulting in the removal of part of the biofilm by the shear forces acting on the outer layers of the biofilm. The process is generally referred to as ‘‘sloughing.’’

115

B Fig. 2 Schematic concept of flow within a biofilm. (From Ref.[10].)

‘‘streamers’’ oscillating rapidly in the flow.[11] These streamers, which oscillate in response to the effects of turbulence in the flowing water, may assist the assimilation of nutrients by the destruction of the laminar sublayer around the colony. In other words, the resistance to the mass transfer of the nutrients is reduced. The cell clusters can be composed of a single species that is likely to be the result of the initial colonization of the surface. It is possible to have groups of cells containing a mixture of various species, and this might be the result of synergy, i.e., an interactive association between two populations of microorganisms, not necessarily for survival, but for each group’s benefit. Extracellular polysaccharide may be present around the clusters or less densely distributed in the spaces between the microcolonies. The result is a random array of channels through which water can flow. Fig. 2 illustrates the possible flow pattern between clusters of cells. Clearly, this water flow has implications for the delivery of nutrients to the cells in the lower regions of the biofilm. Under high flow rates of fluid, when shear forces are strong, it is possible that the surviving cells lie so that the removal forces they experience are reduced to a minimum,[12] and the biofilm is less prone to sloughing.

THE DEVELOPMENT OF BIOFILMS THE STRUCTURE OF BIOFILMS Concepts on the detailed structure of biofilms prior to the invention of the confocal scanning laser microscope were largely the results of intuition and speculation, because, in general, the biofilm had to be removed from the environment in which it was formed. Such a procedure could lead to the damage of the biofilm and loss of moisture that would change its appearance. With the invention of newer methods of investigation, detailed examinations of fully hydrated biofilms, where water flows across the surface of the biofilm, could be made. The results of the observations revealed a heterogeneous structure consisting of cell clusters separated by interstitial voids and channels.[10] Realtime video imaging of biofilms growing on surfaces subject to fast flowing water revealed the presence of

The discussion so far has dealt with scientific background to the formation of biofilms. It is of interest therefore to examine, as far as possible, the pattern of development in industrial process operations. The work described here is related to fresh water fouling by biofilms and carried out in a pilot scale laboratory apparatus, but it gives an indication of the effects of the principal operating variables on biofilm development with time. Biofilms in cooling water systems reduce the cooling efficiency because they are prone to occur on heat transfer surfaces, where, in general, conditions are conducive to development. The organisms involved are usually bacteria. Unless corrective action is taken, the biofilms so produced can cause serious operating problems with attendant higher operating costs.

116

Biofilms

The change in biofilm accumulation on a surface with time, under flowing water conditions, would be expected to follow the idealized curve in Fig. 3. Three regions can be seen in the diagram: 1. Initiation and growth, which is related to the laying down of the conditioning macromolecules on the surface. 2. After the colonization, there is a period of rapid growth, where the conditions, in terms of temperature, availability of nutrients, and their transport to the biofilm, are conducive to sustained growth. 3. The rate of growth gradually falls off, till a stable thickness of biofilm is reached, which might vary from a few micrometers to several millimeters, depending on the prevailing conditions. 4. The leveling off of the biofilm growth is attributed to a balance between growth and removal, depending on nutrient availability and temperature on the one hand, and shear forces caused by water flow on the other. Under practical operating conditions, it will be seen that this ideal representation of biofilm development with time will be modified. Fig. 4 demonstrates that practical data roughly follow the idealized curve, but with considerable fluctuation of. The reason for this ‘‘saw tooth’’ appearance is attributed to the growth and partial sloughing of parts of the biofilm because of the shearing action of the flowing water at the surface of the biofilm. It is interesting to note that there is little effect on the development of the biofilm by the elimination of the planktonic bacteria in the flowing water. Once the surface has been colonized, development continues provided that a nutrient supply is maintained. An understanding of the effect of different variables on biofilm growth is essential for devising

Fig. 4 Practical biofilm growth curve on the inside of an aluminum tube. (From Ref.[9].)

methods for biofilm control where biofilms pose an operational problem, in cooling water circuits, for instance, or in enhancing growth where a biofilm is the key to the effectiveness of the process, such as water purification.

MEASUREMENT OF BIOFILM ACCUMULATION Very relevant to the representation of biofilm growth data is how to measure biofilm accumulation. There are various methods available, but a detailed discussion is outside the scope of this encyclopedia. For many years, direct weighing was employed, i.e., to determine the biofilm accumulation by weighing a tube or insert plate before and after contamination with microorganisms. The technique is very unsatisfactory for a number of rather obvious reasons. The removal of the test surface from the rest of the system for weighing may damage the biofilm, and the adventitious moisture associated with the biofilm is likely to give misleading information. Furthermore, the method does not lend itself to continuous tracking of the change in biofilm accumulation with time. Another method that has been used is to carry out a cell count on a unit area of test surface, but, apart from being labor intensive, this method also has the same problems of direct weighing. A number of methods have been developed to overcome the problem.[13] Some of the data reproduced here involve the use of infrared absorbance as a measure of biofilm thickness.[14]

THE AVAILABILITY OF NUTRIENTS Fig. 3 Idealized concept of biofilm growth on a surface with time. (From Ref.[2].)

It is to be expected that the availability of nutrients is crucial to the development of biofilms. The effects of

Biofilms

117

B

Fig. 5 The change in biofilm accumulation on the inside of a tube with three different nutrient concentrations expressed in terms of glucose concentration. (From Ref.[9].)

the changes in nutrient concentration on biofilm growth are shown in Fig. 5. Fig. 6 demonstrates the importance of trace elements on biofilm development. Until trace elements were added to the system, little or no growth was evident.

THE EFFECT OF BULK WATER VELOCITY The effects of changes in bulk water velocity are basically twofold. Higher velocities represent greater turbulence in the bulk flow and a reduction in the thickness of the boundary layers adjacent to the biofilm residing on the solid surface. As the velocity increases, the availability of nutrients, for a given concentration, to the biofilm increases because of the lower resistance to mass transfer of nutrients to the biofilm. It would be expected that this would be evident in the higher rate of biofilm growth. As the velocity increases, however, the attendant shear forces acting on the biofilm also increase. It would be

Fig. 7 The effect of velocity on biofilm growth at constant concentration. (From Ref.[9].)

expected that this would be manifest in a reduction of biofilm accumulation, till a new plateau is reached, i.e., the competing effects of increased growth because of greater nutrient availability and the removal of biofilm by the increased velocity. The result is apparent in Fig. 7.

THE EFFECT OF TEMPERATURE Microorganisms have an optimum growth temperature, when, provided that there are sufficient nutrients available, the growth is maximum. The optimum temperature is different for different species, on account of various metabolic characteristics. It is usually in the range of 20–50
C, with many in the range of 35– 40
C. Fig. 8 shows the very pronounced effect of a relatively small temperature change on the development of a biofilm of Escherichia coli.[16]

THE EFFECT OF SURFACE

Fig. 6 The effect of adding trace elements to the nutrient on the growth of a biofilm on the inside of a glass tube, using infrared monitoring. (From Ref.[15].)

Mott and Bott[17] illustrated the effect of different materials on the accumulation of Pseudomonas fluorescens biofilms on the inside of tubes under identical operating conditions (see Fig. 9). The differences between the effects of the materials occur for two reasons: roughness and surface electrical properties. The quality of the surface, in terms of roughness, on which microorganisms attach, can affect the biofilm accumulation as discussed earlier. The effect of roughness is illustrated in Fig. 9 by the difference of biofilm accumulation between electropolished and ‘‘as received’’ 316 stainless steel. The rougher stainless steel is seen to be more hospitable to biofilm growth.

118

Biofilms

CONTROL OF BIOFILMS

Fig. 8 The effect of temperature on the growth of a biofilm of E. coli. (From Ref.[16].)

In the light of these results, modification of the surface is a possible way of changing biofilm accumulation, i.e., to reduce roughness where the biofilm represents a nuisance or to use a rough surface where the biofilm performs a useful function.

Fig. 9 Development of a biofilm of P. fluorescens on different materials. (From Ref.[7].)

Reference in the discussion so far has been made to the need of controlling biofilm growth, a requirement that arises often. Where it is required to enhance growth, such as the treatment of wastewater, it is simply a question of ensuring that the operating conditions are optimum for the species present. There may be, of course, limited opportunity for this approach by virtue of the existing conditions of the process. It is unlikely that there would be much opportunity, for instance, to raise the temperature of the incoming stream because of the difficulties this would present and the costs involved. Of greater significance is the need to control the growth of biofilms, as encountered in the use of naturally occurring water for cooling purposes. Some mention has been made in the preceding discussion, but it is useful to summarize the opportunities available. At the onset, it is important to stress that the tactic adopted will very much depend on the operating conditions and particularly on the quality of the water utilized. The obvious choice of method for controlling biofilm growth is to ‘‘dose’’ the cooling water with a biocide that will kill the microorganisms present in the system. For many years, this was the technique employed and the preferred biocide was chlorine, as it was very effective, available, and relatively cheap. As the cooling water is usually discharged back after use to the source from which it was obtained, e.g., a lake, river, or canal for fresh water systems, or the sea where this was more convenient, it has led to concern for the environment. Although chlorine is still used in many cooling systems, it is coming under increasingly tighter control to reduce the threat to the environment. As a consequence, there has been considerable investment in the search for a reliable alternative. Perhaps the solution lies in the use of physical method, i.e., to avoid the use of chemicals altogether. The earlier discussion had indicated that velocity plays an important part in the development of biofilms, so that higher velocities might be used for biofilm control, as this would increase the removal forces. Although a realistic possibility, the technique is likely to be costly in terms of energy usage, because of the serious increase in pressure drop through the system that this would entail. In general terms, it has to be remembered that pressure drop increases as the square of velocity increases. A common technique in the power industry is to circulate sponge rubber balls with the cooling water through the steam condensers. It has to be said, however, that the opportunities for physical control have not been fully explored. Some physical methods for control that were described in a recent article are now being investigated.[18] They include the use of ultrasound and inserts in heat exchanger tubes, and the circulation of polymer fibers. The latter technique

Biofilms

would appear to show promise because the fibers reach all parts of the system in the same way as dissolved chemical biocides do, thereby keeping all parts of the system clean, provided that the fluid velocity is maintained. The other two techniques tend to give more localized control, in particular in heat exchangers, where the biofilm growth is likely to be most prevalent owing to the favorable temperature. An alternative physical strategy is to use modified surfaces to reduce the adherence of the microorganisms to the surface to facilitate removal by the shear forces. The extra cost of treatment of the surface such as electropolishing or ion implantation could be high. The alternative of coating the surfaces with, say, a polymer could invoke questions of its integrity over long periods of time. In response to the environmental issues, the socalled ‘‘environment friendly’’ biocides have been or are being developed. The concept is that after a relatively short time, they decompose to innocuous breakdown products, some of which could be nutrients for biological activity that gives safe disposal. Although this suggests a contradiction in terms, there need not be any difficulties, provided the problem is recognized and handled properly. The cost of these biocides is relatively high, partly because the dose required is, in general, much higher than the traditional chemicals such as chlorine. For this reason, the dosing strategy requires careful attention.[19] The advantages of this approach include optimizing the chemical use and minimizing the maintenance costs. Some biocides that have been used for a number of years are naturally environment friendly and include the oxidizing agents, ozone and hydrogen peroxide, which breakdown to oxygen and water, respectively. A major difficulty with ozone is that it has to be generated as required, because it cannot be stored. Hydrogen peroxide is sometimes a constituent of proprietary biocide formulations.

CONCLUSIONS Biofilms are common in natural environments and may include bacteria, algae, and fungi. Microorganisms prefer to reside on surfaces because the surface offers a certain degree of protection. But at the same time, they have a ready access to nutrients contained in the fluid passing over the surface. The important variables that affect the stability of the biofilm are temperature and the velocity of the fluid in contact with the biofilm. There is an optimum temperature and an optimum velocity to sustain the maximum biofilm thickness, provided that there is an adequate supply of nutrients. In processing, biocides can either be an aid or a disadvantage, for instance, as seen in wastewater treatment and fouling of heat exchangers in cooling water circuits, respectively. Where biofilms are an

119

impediment to processing, control is implemented by chemical or physical means. ARTICLE OF FURTHER INTEREST Fluid Flow, p. 975. REFERENCES 1. Hope, C.K.; Bott, T.R. Laboratory modelling of manganese biofiltration using biofilms of Leptothrix discophora. Water Res. 2004, 38, 1853–1861. 2. Bott, T.R. Fouling of heat exchangers. 26 Chemical Engineering Monographs; Elsevier: Amsterdam, Holland, 1995. 3. Costerton, J.W.; Geesey, G.G.; Cheng, K.J. How bacteria stick. Sci. Am. 1978, 238 (1), 86–96. 4. Fletcher, M. Bacterial metobolism in biofilms. In Biofilms—Science and Technology; Melo, L.F., Bott, T.R., Fletcher, M., Capdeville, B., Eds.; Kluwer Academic Publishers: Dordrecht, Holland, 1992; 113–124. 5. Leadbeater, B.S.C.; Callow, M.E. Formation, composition and physiology of algal biofilms. In Biofilms—Science and Technology; Melo, L.F., Bott, T.R., Fletcher, M., Capdeville, B., Eds.; Kluwer Academic Publishers: Dordrecht, Holland, 1992; 149–162. 6. Characklis, W.G. Microbial fouling – a process analysis. In Fouling of Heat Transfer Equipment; Somerscales, E.F.C., Knudsen, J.C., Eds.; Hemisphere Publishing: New York, U.S.A., 1981; 251–291. 7. Chamberlain, A.H.L. The role of adsorbed layers in bacterial adhesion. In Biofilms—Science and Technology; Melo, L.F., Bott, T.R., Fletcher, M., Capdevilee, B., Eds.; Kluwer Academic Publishers: Dordrecht, Holland, 1992; 59–67. 8. Mozes, N. The ways we study interfacial phenomena of living cells. In Adhesion des Microorganisms aux Surfaces; Bellon-Fontaine, M.W., Fourniat, J., Eds.; Chatenay-Malaby, France, June 2729, 1994; Lavosier TEC and DOC: Paris, 1995; 3–13. 9. Bott, T.R.; Miller, P.C. Mechanisms of biofilm formation on aluminium tubes. J. Chem. Technol. Biotechnol. 1983, 33B, 177–184. 10. Stoodley, P.; deBeer, D.; Lewandowski, Z. Liquid flow in biofilm systems. Appl. Environ. Microbiol. 1994, 60, 2711–2716. 11. Lewandowski, Z.; Stoodely, P. Flow induced vibrations, drag force and pressure drop in conduits covered with biofilm. Wat. Sci. Technol. 1995, 32, 19–26. 12. Santos, R.C.; Callow, M.E.; Bott, T.R. The structure of Pseudomonas fluorescens biofilms in

B

120

contact with flowing systems. Biofouling 1991, 4, 319–336. 13. Lens, P.; Moran, A.P.; Mahony, T.; Stoodley, P.; O’Flaherty, V. Biofilms in Medicine, Industry and Environmental Biotechnology; Part 3, Section 5; IWA Publishing: London, U.K., 2003; 441–470. 14. Bott, T.R. Monitoring biofouling using infrared absorbance. In Biofilms in Medicine, Industry and Environmental Biotechnology; Lens, P., Moran, A.P., Mahony, T., Stoodely, P., O’Flaherty, V., Eds.; IWA Publishing: London, U.K., 2003; 461–470. 15. Santos, R.C. Polymer coatings in relation to single and mixed population biofilms. Ph.D. Thesis, University of Birmingham, 1993.

Biofilms

16. Bott, T.R.; Pinheiro, M.M.V.P.S. Biological fouling—velocity and temperature effects. Can. J. Chem. Eng. 1977, 55, 473–474. 17. Mott, I.E.C.; Bott, T.R. The adhesion of biofilms to selected materials of construction for heat exchangers. Proceedings of the Ninth International Heat Transfer Conference, Jerusalem, Israel, 1991; 5, 21–26. 18. Bott, T.R. Potential physical methods for the control of biofouling in water systems. Trans. I. Chem. A: Chem. Eng. Res. Design 2002, 79, 484–490. 19. Grant, D.M.; Bott, T.R. Biocide dosing strategies for biofouling control. Heat Trans. Engineering 2005, 24 (1), 44–50.

Biofuels and Bioenergy B Dinesh Gera Fluent Incorporated, Morgantown, West Virginia, U.S.A.

INTRODUCTION Biofuel is any gas, liquid, or solid fuel derived either from recently living organisms or from their metabolic by-products, including dedicated energy crops and trees, agricultural food and feed crop residues, aquatic plants, wood and wood residues, animal wastes, and other waste materials. It is a renewable energy, unlike other fossil fuel sources such as coal, petroleum, or nuclear energy. Biomass (any plant-derived organic matter) is a subset of biofuels, and it is used repeatedly in this entry. The term ‘‘bioenergy’’ refers to the production of energy (liquid, solid, and gaseous fuels) and heat using the biofuels or biomass. Biomass is a very heterogeneous and chemically complex renewable resource. Understanding this natural variability and range of chemical compositions is essential for scientists and engineers conducting research and developing energy technologies using biomass resources. Biomass that is available for energy has a potential to produce an array of energy-related products, including liquid, solid, and gaseous fuels; electricity; heat; chemicals; and other materials. Most scientists now agree that temperatures around the world are rising, and global warming may be occurring. There will be high societal costs if unchecked growth in atmospheric concentrations of greenhouse gases continues. Bioenergy and bio-based industrial feedstocks offer sound, economically friendly, and environmentally beneficial ways to reduce the pace at which CO2 and other global warming gases accumulate in the atmosphere. The utility of bioenergy and its coproducts can play a major role in mitigation strategies for reducing greenhouse gases. Potential environmental benefits from biomass include offsetting these greenhouse gas emissions and sequestering carbon, improving water quality, and reducing soil erosion through the use of perennial cropping systems on marginal lands and by recovering wastes and capturing methane emissions. The relative performance of biomass energy systems in reducing net greenhouse gas emissions depends on the sustainability of the sources of biomass feedstock, the energy requirements of the conversion systems, and the overall conversion efficiencies. Unlike fossil fuels, biomass production systems recapture emissions of carbon dioxide. It has been well documented that Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120018027 Copyright # 2006 by Taylor & Francis. All rights reserved.

biomass energy and product systems have the potential to substantially reduce net greenhouse gas emissions. These vary greatly across biomass systems. A 1999 life cycle analysis indicated a 95% reduction in carbon dioxide emissions from a woody crop fired integrated gasification combined-cycle system relative to the average coal-fired power system. Direct-fired biomass systems, using residues that otherwise would have gone to landfills, generated even greater reductions in greenhouse gas emissions by avoiding methane production. The United States Department of Agriculture (USDA) and the Department of Energy (DOE) studies have shown that compared to gasoline, greenhouse gas emissions can be reduced on a per-gallon basis by 20–30% with the use of corn ethanol and 85–140% with that of cellulosic ethanol.[1] This entry is organized into three major parts. The first identifies the biomass resources in the form of conventional forestry, agricultural crops and residue, and oil-bearing plants, among others. The second describes the conversion processes of bioresources into biofuels, and it is followed by the end product usage of biofuels in producing electricity in power plants.

BIORESOURCES A variety of fuels can be made from biomass resources, including liquid fuels such as ethanol, methanol, biodiesel, Fischer–Tropsch diesel; gaseous fuels such as hydrogen and methane, and solid fuels such as switch grass, walnut shells, and sawdust, among others. Examples of biofuel include alcohol (from fermented sugar), black liquor from the paper manufacturing process, and soybean oil. Biofuel contains no petroleum, but it can be blended at any level with petroleum fuel to create a biofuel blend. It can be used in conventional heating equipment or diesel engines with no major modifications. Biofuel is simple to use, biodegradable, nontoxic, and essentially free of sulfur and aromatics. A typical blend of 20% biofuel with 80% of convention petroleum or fossil fuel has demonstrated significant environmental benefits. Biofuel is the only alternative fuel to have fully completed the health effects testing of the Clean Air Act. The use of biofuel in a conventional heating system or diesel engine results in a substantial reduction of unburned 121

122

Biofuels and Bioenergy

hydrocarbons, carbon monoxide, and particulate matter compared to the emissions from heating oil or diesel fuel. In addition, the exhaust emissions of sulfur oxides and sulfates (major components in acid rain) from biofuel are essentially eliminated compared to heating oil or diesel fuel. Bioresources include any organic matter available on a renewable basis, including dedicated energy crops and trees, agricultural food and feed crops, agricultural crop wastes and residues, wood wastes and residues, aquatic plants, animal wastes, municipal wastes, and other waste materials. Herbaceous energy crops: These are perennials that are harvested annually after taking 2–3 yr to reach full productivity. They include such grasses as switchgrass, miscanthus (also known as elephant grass or e-grass), bamboo, sweet sorghum, tall fescue, kochia, wheatgrass, and others. Woody energy crops: Short-rotation woody crops are fast-growing hardwood trees harvested within 5–8 yr after planting. These include hybrid poplar, hybrid willow, silver maple, eastern cottonwood, green ash, black walnut, sweetgum, and sycamore. Industrial crops: Industrial crops are developed and grown to produce specific industrial chemicals or materials. Examples include kenaf and straws for fiber, and castor for ricinoleic acid. New transgenic crops are being developed that produce the desired chemicals as part of the plant composition, requiring only extraction and purification of the product. Agricultural crops: These feedstocks include the currently available commodity products such as cornstarch and corn oil, soybean oil and meal, wheat starch, other vegetable oils, and any newly developed component of future commodity crops. They generally yield sugars, oils, and extractives, although they also can be used to produce plastics and other chemicals and products. Aquatic crops: A wide variety of aquatic biomass resources exist, such as algae, giant kelp, other seaweed, and marine microflora. Commercial examples include giant kelp extracts for thickeners and food additives, algal dyes, and novel biocatalysts for use in bioprocessing under extreme environments. Agriculture crop residues: Agriculture crop residues include biomass (primarily stalks and leaves) that are not harvested or removed from the fields in

commercial use. Examples include corn stover (stalks, leaves, husks, and cobs), wheat straw, and rice straw. With approximately 80 million acres of corn planted annually, corn stover is expected to become a major biomass resource for bioenergy applications. Forestry residues: Forestry residues include biomass that is not harvested or removed from logging sites in commercial hardwood and softwood stands, as well as material resulting from forest management operations, such as precommercial thinnings and removal of dead and dying trees. Municipal waste: Residential, commercial, and institutional postconsumer wastes contain a significant proportion of plant-derived organic material that constitutes a renewable energy resource. Waste paper, cardboard, wood waste, and yard wastes are examples of biomass resources in municipal wastes. Biomass processing residues: All processing of biomass yields by-products and waste streams collectively called residues, which have significant energy potential. Residues are simple to use because they have already been collected. For example, processing of wood for products or pulp produces sawdust and a collection of bark, branches, and leaves=needles. Animal wastes: Farms and animal processing operations create animal wastes that constitute a complex source of organic materials with environmental consequences. These wastes can be used to make many products, including energy. Biofuels can be converted into all three natural forms, i.e., liquid, gas, and solid, from bioresources, using the biochemical, photobiological, and thermochemical processes described in the following section.

BIOMASS CONVERSION PROCESSES Biochemical conversion processes: Enzymes and micro-organisms are frequently used as biocatalysts to convert biomass or biomass-derived compounds into desirable products. Cellulase and hemicellulase enzymes break down the carbohydrate fractions of biomass into fiveand six-carbon sugars, a process known as hydrolysis. Yeast and bacteria ferment the sugars into products such as ethanol. Biotechnology advances are expected to lead to dramatic biochemical conversion improvements.

Biofuels and Bioenergy

Photobiological conversion processes: Photobiological processes use the natural photosynthetic activity of organisms to produce biofuels directly from sunlight. For example, the photosynthetic activities of bacteria and green algae have been used to produce hydrogen from water and sunlight. Thermochemical conversion processes: Heat energy and chemical catalysts are used to break down biomass into intermediate compounds or products. In gasification, biomass is heated in an oxygen-starved environment to produce a gas composed primarily of hydrogen and carbon monoxide. In pyrolysis, biomass is exposed to high temperatures in the absence of air, causing it to decompose. Solvents, acids, and bases can be used to fractionate biomass into an array of products including sugars, cellulosic fibers, and lignin. Biofuels exist in all three natural forms, that is, gas, liquid, and solid: 1. Gas fuel a. Methane—Methane can be produced by the natural decay of garbage dumps over time. Additionally, biomass can be gasified to produce a synthesis gas composed primarily of hydrogen and carbon monoxide, also called syngas or biosyngas. Hydrogen can be recovered from this syngas, or syngas can be catalytically converted to methanol. It can also be converted using the Fischer– Tropsch catalyst into a liquid stream with properties similar to diesel fuel, called Fischer–Tropsch diesel. However, all of these fuels also can be produced from natural gas or coal using a similar process. 2. Liquid biofuels a. Bioalcohols are ethanol and methanol, when they are not produced from petroleum. A significant amount of ethanol is produced from sugar beets and corn using the fermentation process. Ethanol is the most widely used biofuel today, with a current capacity of 1.8 billion gallons per year based on starch crops such as corn. Ethanol produced from cellulosic biomass is currently the subject of extensive research and development, and demonstration efforts. It is currently being used as an automotive fuel and a gasoline additive. b. Straight vegetable oil (SVO) is a waste product of the food service industry, also

123

called fryer grease. It can be used to run a conventional diesel engine if the oil is clean and heated to the appropriate temperature before being injected into the engine. Using SVO as a fuel greatly lowers health and environmental hazards caused by using petroleum fuels while boosting energy security. c. Biodiesel is produced through a process in which organically derived oils are combined with alcohol (ethanol or methanol) in the presence of a catalyst to form ethyl or methyl ester using transesterification. The biomass-derived ethyl or methyl esters can be blended with conventional diesel fuel or used as a neat fuel (100% biodiesel). Biodiesel can be made from soybean or canola (rapeseed) oils, animal fats, waste vegetable oils, or microalgae oils. It can be used in unaltered diesel engines. Biodiesel can be mixed with petroleum diesel. Vehicle performance on biodiesel is almost similar to vehicle performance on conventional petroleum diesel. 3. Solid fuel a. Biomass, such as switch grass, walnut shells, and sawdust, is currently used as the cofiring fuel in industrial boilers to reduce the nitrogen oxide (NOx) emissions.

GLOBAL BENEFITS AND IMPACTS A 1999 biofuel life cycle study, jointly sponsored by the USDOE and the USDA concluded that biofuel reduces net CO2 emissions by 78% compared to petroleum fuels from biofuel’s closed carbon cycle.[1] The CO2 released into the atmosphere when biofuel is burned is recycled by growing plants, which are later processed into fuel. Scientific research confirms that biofuel has a less harmful effect on human health than petroleum fuel. Biofuel emissions have decreased levels of polycyclic aromatic hydrocarbons (PAHs) and nitrited PAH compounds (nPAH), which have been identified as cancer causing compounds. Test results indicate that PAH compounds were reduced by 50–85%. Targeted nPAH compounds were reduced by 90%. Biofuel is nontoxic and biodegradable. In addition, the flash point (the lowest temperature at which it can form an ignitable mix with air) is 300
F, well above petroleum fuel’s flashpoint of 125
F. Several developed countries have introduced policies encouraging use of biofuels made from grain, vegetable oil, or biomass to replace part of their fossil fuel use in transport. These initiatives generally have at

B

124

least three goals: 1) to prevent environmental degradation by using cleaner fuel; 2) to reduce dependence on imported, finite fossil fuel supplies by partially replacing them with renewable, possibly domestic, sources; and 3) to provide a new demand for crops to support producer incomes and rural economies. The use of biomass for power could have impacts on local air pollution. As an example, direct combustion and cofiring of biomass offers improvements in conventional air pollutants, NOx, sulfur dioxide (SO2), particulate matter (PM10 and PM2.5), carbon monoxide (CO), and volatile organic chemicals (VOC) relative to direct-firing of coal. It follows, then, that certain emissions from advanced coal energy systems, which are now under development by the USDOE, also may be moderated by biomass cofiring or cogasification. The USDOE is conducting research on biomass gasification systems that most likely would reduce the overall emissions relative to most conventional fossil systems. Because the impact of conventional pollutants is primarily local or regional, facility location is also a significant environmental consideration. Utilization of bio-based chemicals may help to reduce risks to human health from environmental releases and workplace exposure to toxic chemicals, particularly those derived from petroleum-based feedstocks. Substitution of bio-based products for petroleum-based end products has the potential to reduce pollution from virtually all stages of production, from extraction of the raw material to final product manufacturing and product disposal. Substituting bio-based products for inorganic-based products, e.g., wood for steel, biocement for cement, and cotton insulation (formaldehyde-free) for fiberglass, can have even more substantial effects on reducing greenhouse gas emissions, as the inorganic-based products are extremely energy intensive to produce. Substituting a biochemical for a petrochemical that has different properties can reduce the pollution generated by the production of the petrochemical, and also may reduce the environmental impact associated with using the chemical in manufacturing final consumer products. Finally, even substituting a biochemical for an identical petrochemical can reduce the upstream impacts associated with the extraction of the material. Using the advanced bioethanol technology available, it is possible to produce ethanol from any cellulose=hemicellulose material, which means any plant or plant-derived material. Many of these materials are not just underutilized and inexpensive, but also create disposal problems. For example, rice straw and wheat straw are often burned in the field, a practice that is becoming limited by air pollution concerns. Also, much of the material now going into landfills is cellulose=hemicellulose material and could be used

Biofuels and Bioenergy

for bioethanol production. Technology is now available to convert the municipal solid waste to ethanol. Wastes from many paper mills, food processing, and other industries also may be converted to bioethanol. Another possible bioethanol feedstock is corn stover (corn stalks and leaves). Because of the stover’s large volume and proximity to current ethanol production facilities and because it is already there for the taking, biofuels program analysts expect stover to be one of the primary feedstocks for advanced bioethanol production. Currently, stover is left in the fields in some cases and plowed under in others. Appropriate levels of stover harvesting will need to be carefully determined to avoid losing stover’s value for erosion control and soil enrichment. In many areas though—particularly in northern areas where springtime soil warming is of concern— stover is routinely plowed under—mostly to get rid of it. Harvesting part of that stover might allow farmers to switch to a no-till operation, which soil scientists recognize to be far better from a soil erosion and fertilizer use perspective. The total bioenergy potential of the base year, 1990, was estimated at 225 EJ (1 EJ ¼ 1 exajoule ¼ 1018 J) or 5.4 billion (109) tons of oil equivalent (Gtoe).[2] For comparison, the actual use of bioenergy in 1990 was 46 EJ or 1.1 Gtoe. By the year 2050, this potential was estimated to have grown to between 370 and 450 EJ (8.8 and 10.8 Gtoe). The slowest growth is expected to occur in the ‘‘crop residues’’ category (because of increasing the harvesting index). To put the estimated totals into perspective, the estimated global energy value from photosynthesis is 4000 EJ.[2]

SOME ISSUES IN BIOMASS HANDLING Material handling, collection logistics, and infrastructure are important aspects of the biomass resource supply chain. Biomass material handling: Materials handling systems for biomass constitute a significant portion of the capital investment and operating costs of a bioenergy conversion facility. Requirements depend on the type of biomass to be processed as well as the feedstock preparation requirements of the conversion technology. Biomass storage, handling, conveying, size reduction, cleaning, drying, and feeding equipment, and systems require special attention. Biofuels can be stored in existing fuel tanks. Biofuel is completely compatible with petroleum fuels. Biomass collection logistics and infrastructure: Harvesting biomass crops, collecting biomass residues, and storing and transporting

Biofuels and Bioenergy

125

biomass resources are critical elements in the biomass resource supply chain. Biomass feedstock production shares many of the potential environmental effects associated with other agricultural systems, including soil erosion, fertilizer, and pesticide runoff. However, biomass production systems using perennial crops, such as trees and grasses, that are being developed by the Agricultural Research Service, the Forest Service, the Oak Ridge National Laboratory, land grant universities, and others will lower overall chemical use and reduce soil erodable from rates associated with conventional monoculture crop production. Positive environmental effects have been documented where biomass (perennial crop) production replaces conventional crop production on marginal, highly erodable lands. Additional research is needed to optimize the management systems for a wide variety of soil and climate conditions. Utilization of biomass products such as compost also can serve as a component of an overall strategy to reduce, reuse, and recycle solid waste. For example, promoting composting is entirely consistent with efforts to enhance recycling. In providing incentives to promote biomass energy from waste, it will be important to understand and mitigate potential instances where diverting waste for energy might have impacts on programs and incentives to promote waste reduction and recycling. It is important to keep in mind that the established, national hierarchy for materials utilization places reuse and recycling above the use of materials for energy recovery. Improvements in handling and management of solid waste, such as the

capture and use of landfill gas, can reduce emissions of methane, a powerful greenhouse gas, as well as other pollutants.

THERMOCHEMICAL PROPERTIES OF BIOMASS FEEDSTOCKS AND FUEL Biomass feedstocks and fuels exhibit a wide range of physical, chemical, and agricultural=process engineering properties. Despite their wide range of possible sources, biomass feedstocks are remarkably uniform in many of their fuel properties (see Table 1), compared with competing feedstocks such as coal or petroleum. For example, there are many kinds of coals whose gross heating value ranges from 20 to 30 MJ=kg (million joules per kilogram; 8600–12,900 Btu=lb). However, nearly all kinds of biomass feedstocks destined for combustion fall in the range 15–19 MJ=Kg (6450–8200 Btu=lb). For most agricultural residues, the heating values are even more uniform—about 15– 17 MJ=Kg (6450–7300 Btu=lb); the values for most woody materials are 18–19 MJ=Kg (7750–8200 Btu=lb). Moisture content is probably the most important determinant of heating value. Air-dried biomass typically has about 15–20% moisture, whereas the moisture content for oven-dried biomass is around 0%. Moisture content is also an important characteristic of coals, varying in the range of 2–30%. However, the bulk density (and hence energy density) of most biomass feedstocks is generally low, even after densification—between about 10% and 40% of the bulk density of most fossil fuels— although liquid biofuels have comparable bulk densities.

Table 1 Thermochemical properties of various bioenergy feedstocks and fuels Heating value (MJ/kg) Bioenergy feedstocks

Liquid biofuels

Fossil fuels

(From Refs.[3,4].)

Ash (%)

Sulfur (%)

Potassium (%)

Ash melting temperature (
C)

0.73–0.97 0.04

900

Corn stover Sugarcane bagasse Hardwood Softwood Hybrid poplar Bamboo Switch grass Miscanthus

17.6 18.1 20.5 19.6 19.0 18.5–19.4 18.3 17.1–19.4

5.6 3.2–5.5 0.45 0.3 0.5–1.5 0.8–2.5 4.5–5.8 1.5–4.5

0.1–0.15 0.009 0.01 0.03 0.03–0.05 0.12 0.1

Bioethanol Biodiesel Black liquor

28 40 11–13

30%), homology modeling usually gives a reasonably accurate model. In cases of low sequence identities (90%), high molecular weight polymer with narrow MWD (Mn up to 35,000, Mw up to 70,000, and polydispersity index of 1.8–2.4) was obtained from the SDR under exposure to modest UV light intensities of 25 mW=cm2, at residence times of about 2 sec.[10] In contrast, only 30% conversion could be achieved in a thin static film, after a longer exposure time of 10 sec at the same UV intensity, with a lower molecular weight polymer being formed. These results demonstrate that the high-intensity mixing taking place with the thin SDR film allows the polymerization to proceed faster than under static thin film conditions and that a better-quality polymer can also be obtained from processing in the SDR. The vigorous mixing regime in the thin films of the SDR has also been exploited for crystallization processes.[12] Homogeneous nucleation in the thin films results in a narrow crystal size distribution for crystals formed in the SDR. Uniformly shaped barium sulfate crystals of 0.5–1 mm have been produced with a reasonably low power dissipation of 115 W=kg.[12]

Bioprocess and Chemical Process Intensification

An OFR can therefore be much more compact than conventional plug flow reactor designs, allowing reactor designs for longer residence times to be of practical dimensions. The main niche application envisaged is to allow the conversion of long residence time batch processes to continuous processing, when other continuous reactor designs are too expensive or impractical. This is not the only application of this technology: it can be used in a range of unit operations, including heat and mass transfer, multiphase mixing, particle suspension, bioreactions, and fermentations. Some selected examples are given below. Mass transfer rates have been shown to be increased in yeast cultures because of increased breakup of bubbles and increased gas holdup resulting in enhanced aeration. Three-phase heterogeneous photocatalytic OFR that suspends catalyst-coated particles while they are contacted with air and exposed to UV radiation from an axially located lamp represents a good example of the utilization of several processing fields to enhance reaction rates. In this reactor, particle disengagement occurs in a baffle-free zone at the top of the column. Conversion of a batch saponification reaction to continuous processing in an OFR has been shown to result in a 100-fold reduction in reactor size, as well as greater operational control and flexibility. The mixing in OFRs is very uniform, as the radial and axial velocities are of the same magnitude. This has been shown to be an advantage when handling shear-sensitive materials, such as certain pharmaceutical crystals and in flocculators. The scale-up and design of OFRs are well understood. Indeed, the understanding of the OFR’s scale-up is believed to be one of its advantages. A full review of the history and further details of the theory and applications of OFR technology are available.[15]

Controlled Deformation Static or Dynamic Mixers/Reactors Oscillatory Flow Reactors Oscillatory flows are known to increase the transfer processes.[13] In oscillatory flow reactors (OFRs), there is a controlled stage-wise mixing present, which is achieved using tubes fitted with low-constriction orifice plate baffles and oscillatory motion (range: 0.5–10 Hz) superimposed upon the net flow of the process fluid.[14,15] The combination of the baffles and the oscillatory motion creates a flow pattern conducive to efficient heat and mass transfer while maintaining plug flow. Unlike conventional tubular reactors where a minimum Reynolds number must be maintained, the degree of mixing is independent of the net flow, allowing long residence times to be achieved in a reactor of greatly reduced length-to-diameter ratio.

The above-described mixers are essentially low-viscosity devices. In many operations where the viscosity is high, when dealing with concentrated multiphase gas–liquid–solid binary or tertiary systems, or when liquid-to-solid phase transformation occurs during mixing, novel equipment designs are needed to intensify the heat=mass transfer processes. The multiphase fluids also represent an important class of materials that have microstructure developed during processing and subsequently ‘‘frozen-in,’’ ready for use as a product. To deliver certain desired functions, the control of microstructure in the product is important. This microstructure is developed in most cases by the interaction between the fluid flow and the fluid microstructure; hence, uniformity of the flow field is important.

Bioprocess and Chemical Process Intensification

Typical products with microstructure are detergents, emulsions, suspensions, paints, food emulsions, margarines, low-fat spreads, ice cream, cosmetics, personal care products, microcapsules, and agglomerated powders. The classical processing of these materials aims for low-viscosity operations, which usually result in either dilute systems or high operation temperature. Intensified processes involving microstructured products (also known as intensive structuring) deliberately use low temperature=high concentration (which results in high viscosity—non-Newtonian flow behavior), small processing volume (provides uniform deformation fields), and high=ultrahigh deformation rates under laminar flow conditions so that high=ultrahigh stresses can be generated and sustained to achieve a small microstructure at a fast production rate. These requirements therefore qualify this processing strategy as intensive processing. All flows can be decomposed into shear and extensional components. The effectiveness of the flow field is dependent on the deformation rate, the relative values of shear and extension, and the microstructure of the fluid.[16,17] Extensional flows are more effective in microstructure development, such as droplet breakup and mechanochemical reactions. However, such flows are difficult to generate and to maintain and therefore, in practical applications, capillary entrance=exit flows provide a suitable means of achieving extensional flows where the shear component of the flow field changes with the capillary entry–exit angle.[16,17] Indeed, OFRs also generate extensional flows, which result in efficient droplet breakup in emulsification. Mixers in which the deformation rate and the relative values of shear=extension rates can be controlled are known as controlled deformation mixers.[18–20] The multiple expansion–contraction static mixer

187

(MECS mixer), which is just a series of short capillaries separated by flow dividers, is an example of a controlled deformation static mixer and provides all of the above requirements because ultrahigh shear= extension rates approaching 106 sec
1 can be generated and maximum shear=extension rate ratio can be controlled.[6,17,21] Although MECS mixers are very useful in the processing of liquid=liquid systems (emulsions), they are not suitable for solid=liquid systems, in which case the controlled deformation dynamic mixer (CDDM) can be used.[19,20,22–25] Such mixers are suitable for highly viscous fluids and can also have pumping action and heat transfer facilities. Controlled deformation dynamic mixers can be regarded as stage-wise reactors in which each stage is connected to the neighboring stages in parallel and in series. The fluid motion is three-dimensional; when the fluid leaves a stage in series, it undergoes extensional deformation while parallel-stage transfer provides mixing. Such devices have two elements, a rotor and a stator, both of which have cavities machined on them. The rotor and stator cavities can match, thus providing a suitable geometry for extensional flow at the exit=entrance to a cavity. Shear deformation is dominant within the cavities, and shear and extensional flows are superimposed in the regions where there is minimum rotor=stator clearance. There are two such devices available. They are either in the form of two concentric cylinders (inner cylinder is the rotor) or as two (or four) disks as shown in Fig. 2.[19,20,22–25] The cavity arrangement in a disk CDDM is shown in Fig. 3. The rotor (Fig. 3A) and stator (Fig. 3B) disks can be further modified to have microporous disks for simultaneous reaction and separation or an electric field can be applied across the disk clearance.[22] Furthermore, the disks can have

Fig. 2 Diagrammatic view of a CDDM in rotor–stator disk configuration. (From Ref.[22].)

B

188

Bioprocess and Chemical Process Intensification

Fig. 3 Cavity arrangement in the upper rotor and lower stator disks of a CDDM in rotor–stator disk configuration. Elongated cavities are more efficient in pumping=conveying and smaller-aspect ratio cavities are for mixing. (A) Upper rotor. (B) Lower stator. (From Ref.[23].) (View this art in color at www.dekker.com.)

a temperature profile imposed on them. Controlled deformation (static and dynamic) mixers have been used in the intensive structuring of concentrated detergent formulations, agglomeration and microencapsulation of fine powders, and intensive separation of stable water-in-crude-oil emulsions.[19,20,22,26–31] PHENOMENON-BASED PI Miniaturization or reduced processing volume can result in physical PI by virtue of increasing surface area per unit volume. Miniaturized heat exchangers= reactors can be fabricated using microchannels, but such fabrication techniques are expensive and are limited by the availability of the fabrication technique for industrial scale applications. On the other hand, if the small processing volume results in enhanced selectivity, then the subsequent process has a greater degree of intensification. Therefore, in phenomenon-based PI, for a given reaction, both the reactor size=material and the nature=strength of the processing field must be considered carefully to achieve an even more impressive degree of intensification. Here, we review some of the phenomena that have been used in PI and already utilized in industry.

prevailing TSVs favor the formation of (A-in-B) dispersion, phase inversion to (B-in-A) dispersion can be achieved by changing the TSVs (thermodynamically driven process). Alternatively, the dispersion can be subjected to a well-prescribed deformation, characterized by its rate and type of deformation state variables (DSVs) to invert the dispersion under constant thermodynamic conditions; this phenomenon is known as FIPI. It is found that FIPI is not catastrophic and the dispersion goes through an unstable, cocontinuous state denoted as [AB], followed by a relatively stable

Flow Induced Phase Inversion Phenomenon and PI in Particle Technology Flow induced phase inversion (FIPI) has been observed by the author and applied to intensive materials structuring such as agglomeration, microencapsulation, detergent processing, emulsification, and latex production from polymer melt emulsification.[16–18,21,29,32–35] A diagrammatic illustration of FIPI is shown in Fig. 4. When material A is mixed with material B, in the absence of any significant deformation, the type of dispersion obtained [(A-in-B) or (B-in-A)] is dictated by the thermodynamic state variables (TSVs) (concentration, viscosity of components, surface activity, temperature, and pressure). If the

Fig. 4 Isothermal FIPI paths for the inversion of (A-in-B) or (B-in-A) emulsions through a cocontinuous unstable emulsion phase (AB). TSV, thermodynamic state variable; DSV, deformation state variable. (View this art in color at www.dekker.com.)

Bioprocess and Chemical Process Intensification

multidispersion state, denoted as [(A-in-B)-in-A], before complete phase inversion to (B-in-A). Therefore, the interchangeability of TSVs with DSVs forms the basis of FIPI processes. The importance of FIPI is twofold. It can be used to promote phase inversion without changing the thermodynamics of the system to obtain a higher entropy state, or it is possible to delay phase inversion while reducing the system entropy.[36] The characteristics of the microstructure formed (such as emulsion droplet size) are dependent on the type of microstructure and deformation (shear, extension, or combined), as well as the deformation rate. To maximize the fluid microstructure=flow field interactions, the flow field must be uniform, which requires the application of the flow field over a small processing volume, which can be achieved by using MECS mixers or CDDMs.

Isothermal FIPI Emulsification Isothermal emulsification of viscous oils to obtain concentrated oil-in-water emulsions is carried out by inverting a water-in-oil emulsion under flow conditions using an MECS mixer.[16,17,21] Initial water-in-oil emulsion is prepared using a batch mixer (typical shear rate of 100 sec
1), which is subsequently inverted to oil-inwater emulsion at extension rates approaching 106 sec
1. It is found that the FIPI starts at a critical extension rate and the extension rate has to be increased to achieve full inversion resulting in submicrometer size emulsion droplets with a very narrow size distribution (size span < 1).[17] In a novel process, FIPI was also applied to the emulsification of polymer melts in water, thus providing an alternative method to emulsion polymerization for the production of latexes.[18,21,32–34] In fact, some thermoplastic melts (such as polyethylene) cannot be obtained through the emulsion polymerization route; hence, the present technique is an example of PI providing a novel product form. To achieve the emulsification of thermoplastics, it is necessary to operate near or above 100 C and at elevated pressures, which necessitates the use of polymer processing equipment fitted with a MECS mixer at the outlet. It was found that molecular surfactants could not be used to obtain the initial (water-in-polymer melt) emulsion. Instead, hydrophobically modified water-soluble polymers were used as the surface active material. After the phase inversion in the MECS mixer, the resulting emulsion was diluted to the level required. This also freezes the molten latexes. The important attributes of FIPI emulsification include a low level of surfactant use, low temperature processing, production of submicrometer particles with a narrow size distribution, and production of novel products.

189

Nonisothermal FIPI and Intensive Granulation Technology Although the underlying manifestation of FIPI is the same, detailed molecular or microscopic mechanisms are different from one multiphase system to another. When FIPI is applied to granulation (i.e., agglomeration and microencapsulation of fine particles in the absence or presence of liquids) using polymers as binders, the dominant mechanism for phase inversion is based on the behavior of macromolecules in flow fields, which includes stress-induced crystallization, diffusion, macromolecular chain scission, macromolecular confirmation, deposition, and reaction. Therefore, high= ultrahigh stresses must be generated and sustained to achieve these phenomena. A combination of these phenomena results in FIPI in polymer melt=solid particle dispersions, which is further enhanced by operations at low temperatures as a result of high viscosity (i.e., high stress generation) and differential crystallization, because of the distribution of the molecular weight of polymeric binders.

Intensive granulation technology Agglomeration and encapsulation of fine particles are encountered in several industrial sectors such as pharmaceuticals, detergents, fertilizers, and animal feedstocks. The volume throughput in these processes can vary enormously. Batch processing is suitable for pharmaceutical applications where the volume of production is quite low and product values are very high, while a continuous process is essential in other sectors where the volume throughput is very high. Currently available agglomeration techniques are either based on the provision of a suitable processing environment for primary particle collisions (growth agglomeration) or on the compression of particles to form densified particles (pressure agglomeration) in the presence or absence of a binder. These techniques are well known and periodic reviews are available. The growth agglomeration technique is often conducted in batch mixers or fluidized beds in which particle breakup and agglomeration occur simultaneously. The residence times are relatively long and the variety of possible primary particles for agglomeration and binders is restricted. In pressure agglomeration, pressurization is confined to a small volume and there are restrictions on the raw materials. The so-called multinucleus-type capsules= microcapsules are essentially agglomerated particles with prescribed released characteristics, which can be controlled by the concentration and solubility of the binder. In agglomeration, the amount of binder is kept to a minimum, as long as the agglomerates are

B

190

sufficiently strong. Therefore, in principle, multinucleus microcapsules can be manufactured through an agglomeration process, where the binder concentration should be controlled. However, currently available agglomeration processes do not have this flexibility and therefore such capsules=microcapsules are produced through specific techniques. A novel agglomeration=encapsulation (granulation) technique based on isothermal FIPI has been disclosed recently and applied to several industrial processes. In an isothermal FIPI granulation, a concentrated suspension of powder in liquid is caused to crumble (phase invert) through the further addition of a powder (a ‘‘crumbling agent’’), i.e., a phase inversion from a solid phase suspended in liquid (a paste) to a liquid phase suspended in solid (an agglomerated granular material). The process requires a degree of mechanical energy input to disperse the two phases and, therefore, it must be carried out in some kind of mixing environment. Also, the critical deformation rate at which FIPI starts is dependent on the thermodynamic state of the structured fluid as well as the type of flow field, i.e., shear, extensional, or combined flows. This mixing flexibility allows FIPI in either a batch or a continuous mode of operation. The FIPI granulation technique combines both growth and pressure agglomeration processes. For a typical isothermal FIPI agglomeration=microencapsulation process, phase inversion takes place at a critical dispersed phase concentration (crumbling concentration), when the primary particle size and mixing conditions are kept constant. The use of temperature and deformation as processing parameters can therefore promote phase inversion in solid particle=polymer systems and can be utilized in PI in particle technology. The reduction in temperature in polymer melt=particle mixture (paste) results in a reduction of the continuous phase volume because of the crystallization and stress induced deposition of the higher molecular weight fractions of the polymer. Furthermore, at lower temperatures, because of the reduced polymer chain mobility and interactions with the large numbers of small primary particles, which make a three-dimensional cage, the effective mobilephase volume decreases further with increasing primary particle concentration. Consequently, the continuous phase can no longer sustain the increased phase volume of the dispersed phase and phase inversion takes place in the form of the melt fracture. This phenomenon now forms the basis of the present granulation technique.

Intensive granulator The complete granulation process is carried out in a system that consists of an extruder and a CDDM

Bioprocess and Chemical Process Intensification

in the form of rotor–stator disks with heat transfer facilities. A diagrammatic illustration of the granulator is shown in Fig. 2. It consists of two back-to-back rotors (referred to as upper and lower rotors) made from a single circular block and two stators (also referred to as upper and lower stators) facing the rotors. As shown in Fig. 3, the rotors and stators contain cavities on their surfaces, which are designed to pump=convey and=or achieve mixing. The cavities on the rotor and stator are offset by half a cavity length so that the rotor and stator cavities never exactly match with each other. The separation (gap) between the upper or lower rotor and the corresponding stators can be varied independently. The filled polymer melt from the extruder output flows into the gap between the upper stator and rotor. The rotor block (incorporating the upper and lower rotors) spins sandwiched between the upper stator and lower stator of the granulator. The upper rotor temperature is equal to the temperature when solidification of the binder starts, while the upper stator temperature is maintained at the temperature when the solidification is complete. Granulation takes place inside the gap between the upper stator and rotor. Once the granules are formed, they travel through the gap between the lower stator and the rotor and finally emerge and drop onto the collection tray. The upper stator also contains three auger powder feeders located symmetrically at various distances from the center. Additional powder can be added into the filled polymer melt after phase inversion. The addition of the extra powder is useful to increase the particle content of the agglomerates and to form a shell-and-core-type microcapsule. Samples from various regions were recovered and analyzed. It was found that granulation takes place over four steps. These steps are associated with the following zones: 1) melt flow zone; 2) granule nucleation zone; 3) crumbling (melt fracture and fragmentation) zone; and 4) granule transport zone. The granulation occurs over a short distance, within a distance of 2–5 mm. The location of the crumbling (phase inversion) is important as it dictates the size of the granulator. This location is dependent on the effectiveness of heat transfer, the temperature of the polymer melt from the extruder, the polymer concentration, and the clearance between the rotor and the stator. Fig. 5 illustrates the granule particle size distribution as a function of gap size between rotor and stator.[25] As can be seen, the average particle size D(50) is approximately equal to the gap size, and the particle size span (ca. 0.5) is very low and independent of gap size. Typically, the concentration of the binder is 25 wt%. In all of the known agglomeration and microencapsulation techniques, the concentration of

Bioprocess and Chemical Process Intensification

191

CHEMICAL PI USING NANOSTRUCTURED MICROPOROUS MATERIALS AND CATALYSTS Preparation of Novel Porous Materials for PI

Fig. 5 Granule size distribution as a function of gap (clearance) between the rotor and stator disks. Average granule size D(50) in mm and particle size span are also indicated. (From Ref.[24].) (View this art in color at www.dekker.com.)

the active material (filler) is dependent on the granule size, larger granules having a higher binder concentration. Because of the fact that the granulation is induced from a molten state where the filler concentration is uniform, it is therefore not surprising that the granules have a constant active (filler) concentration.

Functionalized microporous polymers (with a superimposed nanostructure in the form of nanopores within the walls of the micropores) have been prepared through a high internal phase emulsion (HIPE) polymerization route.[35,37–40] These polymers (known as PolyHIPE Polymers, PHP) were subsequently metallized by solution deposition followed by heat treatment.[39,41] Microporous polymers have been used in the intensification of several processes encountered in nuclear, petroleum, bioconversion, tissue engineering, chemical, and environmental technologies.[22,31,39–45] The advantages of PHP and its metallic form are associated with the accessibility of their pores, controllability of the pore and interconnect structures, versatility of fabrication, and chemical modification of their walls. The structure of these materials is shown in Figs. 6A–C. These polymers can be manufactured over a wide range of pore size (D) (0.5 mm < D < 5000 mm) and interconnect size (D) (0 < d=D < 0.5).

Fig. 6 Basic PHP structures: (A) primary pores with large interconnecting holes; (B) primary pores with nanosized interconnecting holes; (C) large coalescence pores (three such pores are partially shown) dispersed into the primary pores in the process of coalescence; and (D) detail of the coalescence pores. Note that these pore structures can be prepared over a wide size range.

B

192

Pores with size above
200 mm are obtained through a coalescence polymerization route.[35,39] Nanostructured microporous catalysts or catalyst supports offer intensified catalysis as they provide enhanced surface area accessible to the reactants and products. In nonstructured catalysts, although the surface area may be large, they are often inaccessible as a result of surface fouling and diffusion resistance can slow down the rate of reaction. In a recent development, microporous materials were used as templates for the solution deposition of metals, which were subsequently heat treated to obtain porous metallic structures, where the size of the pores ranged from 10 mm to 10 nm.[39,41] The relative phase volume of these two regions can be controlled and the overall porosity can be in excess of 50%. Fig. 7 illustrates the size scale of structures ranging from 10 mm to 10 nm. These attributes are used in devising intensified processes as well as in the discovery of several size dependent phenomena, especially in biology, which are subsequently utilized in BI. Currently, PHPs (both hydrophobic and hydrophilic versions) are used in the intensification of stable water-in-oil emulsion separation (demulsification), gas–liquid separation, as applied to tar and water removal from biogas produced through the gasification of biomass, BI, tissue engineering, and metal ion=toxin removal from

Bioprocess and Chemical Process Intensification

contaminated water.[8,22,31,35,39,40,42–45] The accessibility of the pores increases both the rate and the capacity of metal ion removal. In currently available ionexchange resins, only the outermost part of the resin beads is utilized, because as soon as the resin surface is saturated or fouled, metal ion removal rate and capacity are drastically reduced.

Intensification of Stable Water-in-Oil Emulsion Separation The breakdown of the stable emulsions and subsequent separation to oil and water (demulsification) are important in nuclear, petroleum, and environmental technologies. The emulsion stability is primarily induced by the use of surfactants and is enhanced by reduced size and narrow size distribution of the emulsion droplets. Disruption to low interfacial activity (hence instability) can be achieved by using demulsification agents, which are, however, costly and environmentally undesirable, as they are irrecoverable. Demulsification can also be achieved by electric and=or centrifugal fields, or by chemical treatment of the emulsion. During the reprocessing of depleted uranium, a highly stable and viscous water-in-oil emulsion is

Fig. 7 Nanostructured microporous nickel with accessible pores showing the hierarchy of the pore sizes: (A) micropores created by fused metal grains; (B) structure of the metal grains before heat treatment; (C) surface pore structure of the grains; and (D) inner structure of metal grains showing the existence of a porous inner core.

Bioprocess and Chemical Process Intensification

formed called interfacial crud (or crud). This emulsion deposits in the process lines, thus reducing the heat=mass transfer rates, making it necessary to periodically remove crud by solvent washing. The removal of water from crude oil and the subsequent treatment of the water produced is a very important process as the crude oil may contain up to 90% water, although it is still economically viable to separate crude from water. In offshore applications, if the separation is achieved at the seabed, massive savings in pumping and heating costs can be achieved. Such a separation process must be intensive, because of the volume of crude oil produced and the hostile environment, or limited space such as the seabed or offshore oil platform. However, such a separation technology at the source would be sustainable utilizing the potential energy of the crude oil from the well, i.e., high temperature and pressure, both of which increase the demulsification rate. Furthermore, the produced water must also be treated, if it is to be discharged into the sea. Recently, we have shown that sulfonated PHP can act as a demulsifier for highly stable emulsions like crud and water-in-crude oil emulsions.[31,44] The mechanism of demulsification is that sulfonated PHP removes selectively surface active species in the emulsion, causing destabilization. At the same time, it also adsorbs metal ions, thus achieving two functions at the same time. As a result, these materials are called demulsifier adsorbers. In highly stable emulsions where neither electric field nor demulsifier adsorbers are effective, the combination of these two methods appears to create synergy for separation. The intensification of crude oil-in-water emulsions was tested using BP Amoco crude (Harding Field, North Sea) with specific gravity of 0.8 g=cm3 and viscosity of 150 mPa sec at 25 C. The aqueous phase is a model seawater containing 28.1 g=L NaCl, 0.6 g=L CaCl2, 5 g=L MgCl2 in double-distilled water. The emulsion viscosity at 25 C and shear rate of 1000 sec
1 is 1030 mPa sec. These emulsions show no sign of separation after 4 weeks of standing. The tests were carried out using a flow-through electric field separator described in Refs.[44,46]. This cell contains a circular channel of 1 cm diameter with two electrodes separated by 10 cm in which the anode is electrically insulated and the counter electrode is earthed. Freshly prepared emulsion containing 0.5 g demulsifier per kilogram of 50=50 emulsion was fed from the center of the circular channel and oil-rich and water-rich phases were collected from the top (anode) and the bottom (cathode) and allowed to separate. The overall degree of separation was measured within 10 min of collection of the sample. The effect of PHP demulsifier adsorber is illustrated in Figs. 8 and 9. Fig. 8 shows the variation of percentage

193

B

Fig. 8 Process intensification in water-in-crude oil separation under electric field with or without PHP: variation of percentage separation immediately after passing through the electric field in the absence or presence of the sulfonated PHP. Electric field strength of 2.5 kV applied over a distance of 10 cm. (From Ref.[43].) (View this art in color at www.dekker.com.)

separation as a function of flow rate under a constant electric field, while Fig. 9 shows the variation of separation as a function of the electric field at a constant flow rate. Both figures clearly show the synergy present between the two types of intensification in oil–water emulsion separation. A continuous intensified separator based on the rotating disk in rotor–stator arrangement with electric field has been disclosed recently.[22]

Fig. 9 Process intensification in water-in-crude oil separation under electric field with or without PHP: variation of percentage of separation with electric field strength applied over a distance of 10 cm when the emulsion flow rate is kept constant at 60 ml=min. Percentage separation into oil– water layers is carried out either immediately (within 10 min) or after 1 hr of emulsion passing through the electric field. (From Ref.[43].) (View this art in color at www.dekker. com.)

194

Bioprocess and Chemical Process Intensification

CARRIER-MEDIATED INTENSIFICATION OF SEPARATION PROCESSES Toxins in the form of heavy metal ions or organic molecules in water (and indeed from gases) can be removed by using carriers with suitable surface chemistry, which can bind the toxic chemicals in their structure and are subsequently removed by filtration. These carriers include small microscopic particles or surfactants.[47–50] Because of their unique ability to form well-ordered molecular aggregates and structures, ranging from a few nanometers to several hundred micrometers, surfactants can be used in the intensification of separation processes, such as the removal of organic and heavy metal ion contaminants from water. Surfactants offer a very large surface area for mass transfer and surfactant assemblies can be modified to be selective.[49] Surfactant-mediated separations can be further enhanced by the orthogonal superimposition of cross-flow separation field and an electric field.[51,52] In such separation processes, both the permeate flux and the surfactant rejection can be enhanced by over 100-fold each.[51,52] However, this enhancement is by no means common, as it depends on the surfactant concentration, as well as surfactant head-group charge density and surfactant phase behavior. Furthermore, surfactants can degrade electrochemically, therefore the selection of surfactant becomes important in electric field enhanced mediation processes.[52] During the cross-flow microfiltration of dilute surfactant solutions, surfactants form highly stable gel phases within micropores. These gel phases do not dissolve even though they are in contact with surfactant-free water flowing at shear rates in excess of 5  104 sec
1. Such stable-strong surfactant phases are expected to form at concentrations over 30 wt%. This phenomenon, first observed by the author, is utilized in cross-flow filtration and forms the bases of the demulsification of highly stable water-in-oil mulsions as well as the removal of tar=water from gases.[22] These observations indicate that microporous materials (with a varying degree of hydrophilicity) preferentially adsorb surface-active species, thus forming the basis of oil–water separation using PHPs. In the intensification of chromate removal from water, a double-chain cationic surfactant, dioctadecyldimethylammonium chloride (DODDMAC), was used as a carrier and a cross-flow electrofiltration was used, in which both the transient and the steadystate fluxes and the rejection of metal ions and surfactant were measured.[51,52] Dioctadecyldimethylammonium chloride in water forms multilamellar droplets, even at very low concentrations. This structure is shown in Fig. 10. Metal ions are entrapped within the water layers and organic toxins can be immobilized within the surfactant bilayers. Under an electric field,

Fig. 10 Schematic representation of surfactant in multilamellar droplet phase with entrapped metal ions (M
n) in the aqueous phase (W) layers, which are separated by surfactant bilayers. The number of layers (hence the size of the surfactant droplet) is dependent on temperature and concentration. When the surfactant head group is positively charged thus encapsulating oppositely charged metal ions, under an electric field, surfactant lamellar droplets migrate to the anode and form a highly stable viscous gel in which the positively charged metal ions are concentrated at the anode only separated by the surfactant bilayer. (From Ref.[52].)

DDOMAC multilamellar droplets with entrapped anions migrate to the negative electrode, forming a stable gel. The electric field intensification in the surfactantmediated separation processes is shown in Table 2.[52] In the feed, surfactant and metal ion concentrations are denoted by CSF and CMF, respectively, while the corresponding concentrations in the permeate under steady-state conditions are denoted by CSP and CMP. Surfactant and metal ion rejections at a steady state are defined as RS ¼ (1
CSP=CSF) and RM ¼ (1
CMP=CMF). The molar ratio of metal ion to surfactant is denoted by YMS. The separation of the electrodes is 3 mm. In Table 2, the initial current, pH, and solution conductivity are also given. It shows that both metal and surfactant are separated effectively under an electric field in which the permeate flux, surfactant, and metal ion rejections are enhanced. Economic analysis of the process indicates that some 20- to 50-fold efficiency increase is achieved compared with the no electric field case. For the process to be economical, low-solubility surfactants that can form multilamellar droplets should be used as carriers.

Bioprocess and Chemical Process Intensification

195

Table 2 Process intensification in the removal of heavy metal ions from water using surfactants as carriers under electric fielda CSF (mM)

CMF (mM)

YMS

J
(L/hr/m2)

R
M (%)

R
S (%)

I0 (A)

pH0

K0 (lsec/cm)

0

3

0.2

0.067

254

99.4

—

—

6.70

65

E (V) 0

3

0.5

0.17

190

65.3

88.7

—

6.59

70

10

3

0.2

0.067

1500

—

—

0.1

6.75

61

30

3

0.2

0.067

2390

99.5

—

0.1

6.68

68

30

1

0.5

0.5

1060

60.0

76.5

0.3

7.55

120

30

3

0.5

0.17

2760

96.1

97.0

0.4

7.24

119

30

5

0.5

0.1

2490

97.9

—

0.4

6.25

—

30

10

0.5

0.05

2450

98.0

97.4

0.5

6.45

138

50

3

0.5

0.17

4450

99.1

98.5

0.7

7.24

93

a


R
M,

R
S,

J , steady state flux; metal rejection; surfactant rejection; pH0, initial feed pH; K0, conductivity; I0, initial cell currents; E, electric field; CSF, feed surfactant concentration; CMF, feed metal concentration; YMS, metal to surfactant molar ratio. (From Ref.[52].)

PHENOMENON-BASED BIOPROCESS INTENSIFICATION The most suitable driving force in BI is the reduction of the diffusion path that already operates in transport processes across biological bilayers. Consequently, biocatalyst membranes and specially designed bioreactors, such as jet loop and membrane reactors, are available to intensify biochemical reactions.[4–8,53] Supported biocatalysts are often employed to enhance catalytic activity and stability and to protect enzymes= microorganisms from mechanical degradation and deactivation.[5,6,8,54] Immobilization of the cells is one of the techniques employed to improve the productivity of bioreactors. Immobilized cells are defined as cells that are entrapped within or associated with an insoluble matrix. Various methods that are used for immobilization include covalent coupling, adsorption, entrapment in a three-dimensional scaffold, confinement in a liquid–liquid emulsion, and entrapment within a semipermeable membrane. Bioreactors with immobilized cells have several advantages over those operating with free cells or immobilized enzymes. Immobilized cell systems permit the operation of bioreactors at flow rates that are independent of the growth rate of the micro-organisms employed. Catalytic stability can be greater for immobilized cells than for free cells. Some immobilized micro-organisms tolerate higher concentrations of toxic compounds compared with free cells, when the cell support media act as a temporary sink for the excess toxin. However, in the current biocatalyst support technology, the presence of the support itself introduces mass transfer restrictions for the substrate=product=nutrient diffusion to and from the biocatalyst. These disadvantages are also valid when supports are used to grow animal cells in vitro.[35,40] In this case, well-designed cell support systems

are necessary for cell proliferation and viability. The efficiency of cell function (for example, production of collagen II during the in vitro growth of chondrocytes) is dependent on the microarchitecture, i.e., pore and interconnect size, and surface chemistry of the support.[35,40,54,55] Microporous polymers with a well-prescribed internal microstructure were prepared in monolithic form to form a flow-through microbioreactor in which phenol degrading bacteria, Pseudomonas syringae, were immobilized. Initially, bacteria were force-seeded within the pores and subsequently allowed to proliferate, followed by acclimatization and phenol degradation at various initial substrate concentrations and flow rates. Two types of microporous polymer were used as the monolithic support. These polymers differ with respect to their pore and interconnect sizes, macroscopic surface area for bacterial support, and phase volume. Poly-HIPE polymer with a nominal pore size of 100 mm with phase volume of 90% (with a highly open pore structure) yielded reduced bacterial proliferation while the polymer with nominal pore size of 25 mm with phase volume of 85% (with small interconnect size and large pore area for bacterial adhesion) yielded monolayer bacterial proliferation. Bacteria within the 25 mm polymer support remained monolayered without any apparent production of extracellular matrix during the 30-day continuous experimental period as shown in Fig. 11(A). The microbioreactor performance was characterized in terms of volumetric utilization rate and compared with the published data, including the case where the same bacteria were immobilized on the surface of microporous polymer beads and used in a packed bed during the continuous degradation of phenol.[42] It has been shown that at a similar initial substrate concentration, the volumetric utilization in the microreactor is at least 20-fold more efficient than the packed bed, depending on the

B

196

Bioprocess and Chemical Process Intensification

CONCLUSIONS The combination of process intensification and process miniaturization is an important element of sustainable technology and a platform for advancement in multidisciplinary science and engineering. There are now several examples of intensified processes being utilized in industry, as apparent from patent activity. Some of these intensified processes are confined to a single unit operation, while others integrate several intensified unit operations in the more efficient production of existing=improved=novel products. Physical PI is limited in the scope and degree of intensification that it can achieve, while phenomenonbased intensifications can result in more impressive levels of intensification. They also provide new opportunities in understanding the behavior of structured matter, macromolecules, and microorganisms within a confined environment with superimposed transport processes. The understanding of the transport processes in human body and organ structures is very useful in PIM, while the application of the techniques and principles of PIM to biology and medicine can result in advances in tissue engineering and biology. The ultimate aim of PIM is to establish intensified plants by integrating intensified unit operations that are already multifunctional. An example of such a plant is a ‘‘biomass power plant’’ in which an intensified gasifier is integrated with an intensified catalyst system and intensified environmental remediation (process water and gas cleaning) to produce electricity through several routes, including internal combustion engine and high-temperature fuel cells.[45,56] Fig. 11 Pore size dependence of bacterial behavior in microbioreactors. (A) Monolayer bacteria coverage in a small pore size (25 mm) microbioreactor. (B) Biofilm formation by the same bacteria in a large pore size (100 mm) microbioreactor. (From Ref.[8].)

flow rate of the substrate solution. The concentration of the bacteria within the pores of the microreactor decreases from 2.25 cells=mm2 on the top surface to about 0.4 cells=mm2 within 3 mm reactor depth. If the bacteria-depleted part of the microreactor is disregarded, the volumetric utilization increases by a factor of 30-fold compared with the packed bed. This efficiency increase is attributed to the reduction of the diffusion path for the substrate and nutrients and enhanced availability of the bacteria for bioconversion in the absence of biofilm formation, as well as the presence of flow over the surface of the monolayer bacteria. When 100 mm pore size PHP is used as a support, or when the flow through the support is absent (i.e., beads used in packed beds), biofilm formation occurs, as shown in Fig. 11(B).[42]

ACKNOWLEDGMENTS I wish to thank Drs K. Boodhoo, M. Dogru, and A. Harvey for helpful discussions. Bioprocess intensification and PI developed by the author have been supported by the United Kingdom Engineering and Physical Sciences Research Councils (EPSRC) and the Department of Trade and Industry (United Kingdom), as well as by the industry, notably, AstraZeneca, BLC Research, British Nuclear Fuels Ltd (BNFL), BP Amoco, Cytec, Exxon Mobil, ICI, Intensified Technologies Incorporated (ITI), Morecroft Engineers, Norsk-Hydro, Safety-Kleen Europe, Triton Chemicals, Unilever, and Willacy Oil Services. I am grateful for their support.

REFERENCES 1. Akay, G.; Mackley, M.R.; Ramshaw, C. Process intensification: opportunities for process and

Bioprocess and Chemical Process Intensification

2. 3. 4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

product innovation. In IChemE Research Event; Chameleon Press: London, 1997; Vol. 1, 695–703. Stankiewicz, A.; Moulijn, J.A. Process intensification. Ind. Eng. Chem. Res. 2002, 41, 1920–1924. Akay, G. Upping the ante in the process stakes. Chem. Eng. 2004, 752, 37–39. Chisti, Y.; Moo-Young, M. Bioprocess intensification through bioreactor engineering. Trans. IChemE. 1996, 74, 575–583. Giorno, L.; Drioli, E. Biocatalyst membrane reactors: applications and perspectives. TIBTECH 2000, 18, 339–349. Akay, G.; Erhan, E.; Keskinler, B.; Algur, O.F. Removal of phenol from wastewater by using membrane-immobilized enzymes: part 2. Crossflow filtration. J. Membr. Sci. 2002, 206, 61–68. Pekdemir, T.; Keskinler, B.; Yildiz, E.; Akay, G. Process intensification in wastewater treatment: ferrous iron removal by a sustainable membrane bioreactor system. J. Chem. Technol. Biotechnol. 2003, 78, 773–780. Akay, G.; Erhan, E.; Keskinler, B. Bioprocess intensification in flow through micro-reactors with immobilized bacteria. Bioeng. Biotech. 2005, 90, 180–190. Boodhoo, K.V.K.; Jachuck, R.J.J. Process intensification: spinning disc reactor for styrene polymerisation. Appl. Therm. Eng. 2000, 20, 1127–1146. Boodhoo, K.V.K.; Dunk, W.A.E.; Jachuck, R.J.J. Continuous photopolymerisation in a novel thin film spinning disc reactor. In Photoinitiated Polymerization; Crivello, J.V., Belfield, K.D., Eds.; ACS Symposium Series No. 847, Washington, DC, 2003; 437–450. Boodhoo, K.V.K.; Jachuck, R.J.J. Process intensification: spinning disc reactor for condensation polymerisation. Green Chem. 2000, 2, 235–244. Cafiero, L.M.; Baffi, G.; Chianese, A.; Jachuck, R.J. Process intensification: precipitation of barium sulphate using a spinning disc reactor. Ind. Eng. Chem. Res. 2002, 41, 5240–5246. Erkoc, E.; Yapici, S.; Keskinler, B.; Cakici, A.; Akay, G. Effect of pulsed flow on the performance of carbon felt electrode. Chem. Eng. J. 2002, 85, 153–160. Mackley, M.R.; Ni, X. Mixing and dispersion in a baffled tube for steady laminar and pulsatile flow. Chem. Eng. Sci. 1991, 46, 3139–3151. Baird, M.H.I.; Harvey, A.P.; Mackley, M.R.; Ni, X.; Rama Rao, N.V.; Stonestreet, P. Mixing through oscillations and pulsations—a guide to achieving process enhancements in the chemicals and process industries. Chem. Eng. Res. Des. 2003, 81, 373–383.

197

16. Akay, G. Flow induced phase inversion in powder structuring by polymers. In Polymer Powder Technology; Narkis, M., Rosenzweig, N., Eds.; Wiley: New York, 1995; 542–587. 17. Akay, G. Flow induced phase inversion in the intensive processing of concentrated emulsions. Chem. Eng. Sci. 1998, 53, 203–223. 18. Akay, G Stable Oil in Water Emulsions and a Process for Preparing Same. European Patent 649,867, Oct 17, 2001. 19. Akay, G.; Irving, G.N.; Kowalski, A.J.; Machin, D Process for the Production of Liquid Compositions. European Patent 799,303, Oct 31, 2001. 20. Akay, G.; Irving, G.N.; Kowalski, A.J.; Machin, D Dynamic Mixing Apparatus for the Production of Liquid Compositions. US Patent 6,345, 907, Feb 12, 2002. 21. Tong, L.; Akay, G. Process intensification in particle technology: flow induced phase inversion in the intensive emulsification of polymer melts in water. J. Mater. Sci. 2002, 37, 4985–4992. 22. Akay, G Method and Apparatus for Processing Flowable Materials and Microporous Polymers. International Patent Application; PCT Publication WO 2004=004880, Jan 15, 2004. 23. Akay, G.; Tong, L.; Addleman, R. Process intensification in particle technology: intensive granulation of powders by thermo-mechanically induced melt fracture. Ind. Eng. Chem. Res. 2002, 41, 5436–5446. 24. Akay, G.; Tong, L. Process intensification in polymer particle technology: granulation mechanism and granule characteristics. J. Mater. Sci. 2003, 38, 3169–3181. 25. Akay, G.; Tong, L. Process intensification in particle technology: intensive agglomeration and microencapsulation of powders by nonisothermal flow induced phase inversion process. Int. J. Transport Phenom. 2003, 5, 227–245. 26. Akay, G Agglomerated Abrasive Material Compositions Comprising Same, and Process for Its Manufacture. US Patent 4,988,369, Jan 29, 1991. 27. Akay, G Coating Process. European Patent 0,382,464, Jan 21, 1993. 28. Akay, G Process for the Manufacture of an Agglomerated Abrasive Material. European Patent 0,303,416, Feb 10, 1993. 29. Akay, G. Flow induced phase inversion agglomeration. Polym. Eng. Sci. 1994, 34, 865–880. 30. Akay, G Detergent Powders and Process for Preparing Them. European Patent 534,525, Dec 27, 1996. 31. Akay, G.; Vickers, J Method for Separating Oil in Water Emulsions. European Patent 1,307,402, May 7, 2003.

B

198

32. Akay, G.; Tong, L. Preparation of low-density polyethylene latexes by flow-induced phase inversion emulsification of polymer melt in water. J. Colloid Interface Sci. 2001, 239, 342–357. 33. Akay, G.; Tong, L.; Hounslow, M.J.; Burbidge, A.S. Intensive agglomeration and microencapsulation of powders. Colloid Polym. Sci. 2001, 279, 1118–1125. 34. Akay, G.; Tong, L.; Bakr, H.; Choudhery, R.A.; Murray, K.; Watkins, J. Preparation of ethylene vinyl acetate copolymer latex by flow induced phase inversion emulsification. J. Mater. Sci. 2002, 37, 4811–4818. 35. Akay, G.; Dawnes, S.; Price, V.J Microcellular Polymers as Cell Growth Media and Novel Polymers European Patent 1,183,328, Jun 3, 2002. 36. Akay, G. Flow induced phase inversion: mechanism and applications. In Recent Advances in Transport Phenomena; Dincer, I., Yardim, F., Eds.; Elsevier: Paris, 2000; 11–17. 37. Williams, J.; Wrobleski, D.A. Spatial distribution of the phases in water-in-oil emulsions: open and closed microcellular foams from cross-linked polystyrene. Langmuir, 1988, 4, 656–662. 38. Akay, G.; Bhumgara, Z.; Wakeman, R.J. Self-supported porous channel filtration modules: preparation, properties, and performance. Chem. Eng. Res. Des. 1995, 73, 782–797. 39. Akay, G Microporous Polymers. International Patent Application; PCT Publication WO 2004= 005355, Jan 15, 2004. 40. Akay, G.; Birch, M.A.; Bokhari, M.A. Microcellular polyhipe polymer (PHP) supports osteoblastic growth and bone formation in vitro. Biomaterials 2004, 25, 3991–4000. 41. Akay, G.; Dogru, M.; Calkan, B.; Calkan, O.F. Flow induced phase inversion phenomenon in process intensification and micro-reactor technology. Process intensification in water-in-crude oil emulsion separation by simultaneous application of electric field and polymeric demulsifiers. In Microreact Technology and Process Intensification; Wang, Y., Halladay, J., Eds.; Oxford University Press: Oxford, 2005; Chapter, 18. 42. Erhan, E.; Yer, E.; Akay, G.; Keskinler, B.; Keskinler, D. Phenol degradation in a fixed bed bioreactor using micro-cellular polymerimmobilized Pseudomonas syringae. J. Chem. Technol. Biotech. 2004, 79, 195–206. 43. Erhan, E.; Keskinler, B.; Akay, G.; Algur, O.F. Removal of phenol from wastewater by using membrane immobilized enzymes: part 1. Dead end filtration. J. Membr. Sci. 2002, 206, 361–373. 44. Akay, G.; Noor, Z.Z.; Dogru, M. Process intensification in water-in-crude oil emulsion separation

Bioprocess and Chemical Process Intensification

45.

46.

47.

48.

49.

50.

51.

52.

53.

54.

55.

56.

by simultaneous application of electric field and polymeric demulsifiers. In Microreactor Technology and Process Intensification; Wang, Y., Halladay, J., Eds.; Oxford University Press: Oxford, 2005; Chapter 23. Dogru, M.; Akay, G. Gasification International Patent Application; PCT Publication WO 2005/ 047435, 26 May, 2005. Pekdemir, T.; Akay, G.; Dogru, M.; Merrells, R.E.; Schleicher, B. Demulsification of highly stable water-in-oil emulsions. Sep. Sci. Technol. 2003, 38, 1161–1184. Akay, G.; Keskinler, B.; Cakici, A.; Danis, U. Phosphate removal from water by red mud using crossflow microfiltration. Water Res. 1998, 32, 717–726. Keskinler, B.; Akay, G.; Bayhan, Y.K.; Erhan, E. The effect of ionic environment on the crossflow microfiltration behaviour of yeast cell suspensions. J. Membr. Sci. 2002, 206, 351–360. Scamehorn, J.F.; Christian, S.D.; El-Sayed, D.A.; Uchiyama, H.; Younis, S.S. Removal of divalent metal cations and their mixtures from aqueous streams using micellar enhanced ultrafiltration. Sep. Sci. Technol. 1994, 26, 809–830. Akay, G.; Odirile, P.T.; Keskinler, B.; Wakeman, R.J. Cross microfiltration characteristics of surfactants. In Surfactant Based Separations: Science and Technology; Scamehorn, J.F., Harwell, J.H., Eds.; ACS Symposium Series: Washington, DC, 2000; Vol. 740, 175–200. Akay, G.; Wakeman, R.J. Electric field intensification of surfactant mediated separation processes. Chem. Eng. Res. Des. 1996, 74, 517–525. Akay, G.; Odirile, P.T. Metal ion removal using electric field intensified mediated crossflow filtration process. J. Membr. Sci. in press. Keskinler, B.; Akay, G.; Pekdemir, T.; Yildiz, E.; Nuhoglu, A. Process intensification in wastewater treatment: oxygen transfer characteristics of a jet loop reactor for aerobic biological wastewater treatment. Int. J. Environ. Technol. Manag. 2004, 4, 220–235. De Bartolo, L.; Morelli, S.; Bader, A.; Drioli, E. The influence of polymeric membrane surface free energy on cell metabolic functions. J. Mater. Sci. Mater. Med. 2001, 12, 959–963. Bokhari, M.; Birch, M.; Akay, G. Polyhipe polymer: a novel scaffold for in vitro bone tissue engineering. Adv. Exp. Med. Biol. 2003, 534, 247–254. Akay, G.; Dogru, M.; Calkan, O.F.; Calkan, B. Biomass processing in biofuel applications. In Biofuels for Fuel Cells; Lens, P., Westerman, P., Haberbauer, M., Menero, A., Eds.; IWA Publishing: London, 2005; Chapter 4.

Bioprocessing B Ryan G. Soderquist James M. Lee Department of Chemical Engineering, Washington State University, Pullman, Washington, U.S.A.

INTRODUCTION Bioprocessing utilizes biological processes to convert raw materials into useful substances. For centuries mankind has converted raw materials into desirable products by harnessing biological processes. In fact the use of baker’s yeast, or leaven, to enhance the quality and texture of bread, as well as the fermentation of grape or cereal extracts to generate ethanol, are common bioprocesses that have been utilized since ancient times. Very little was known about the mechanisms responsible for these important conversions, however, until 1856 when Pasteur characterized yeast as a living organism converting sugar to ethanol.[1] Scientific progress in more recent times has not only revealed the mechanisms of traditional bioprocesses, but has also improved the ability of bioprocessing to generate useful products. Advances in pure culture and other classical bacteriological methods initiated the first commercial production of lactic acid by fermentation in 1881.[1] The subsequent development of the first cholera, diphtheria, and tetanus vaccines from 1885 to 1890 greatly improved human health.[1] The production of penicillin and the subsequent emergence of recombinant DNA technology in the early 1970s, however, significantly enhanced the potential of bioprocessing. Bioprocessing is now a diverse technology, and the potential of bioprocessing seems to have very few limits. New applications of bioprocessing overshadow its traditional role as the fundamental technology for fermented food and beverage production. Pharmaceutical production is one of the fastest growing fields of bioprocessing. Biological processes facilitate the production of a wide array of pharmaceutical compounds, which are typically proteins that act as enzymes or antibodies. Biological processes are also critical in toxic waste degradation, enhanced oil recovery, and other environmental efforts.[2] Modern bioprocessing has several common components and fundamentals, which are the focus of this entry.

BIOPROCESSING TECHNOLOGY Biological processes use living cells (microbial, animal, or plant), or components extracted from living cells Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120025834 Copyright # 2006 by Taylor & Francis. All rights reserved.

such as enzymes, to create desired chemical or physical changes.[2] The major products of bioprocessing include bulk organics, organic acids, amino acids, antibiotics, extracellular polysaccharides, nucleotides, enzymes, vitamins, alkaloids, pigments, vaccines, therapeutic proteins, monoclonal antibodies, and insecticides.[3] Biological processes are also used in wastewater treatment, sludge processing, composting, and bioremediation (i.e., the degradation of crude oils and toxic organic chemicals).[1] These more recent applications of bioprocessing are in addition to the manufacture of traditional foods and beverages such as yogurt, bread, vinegar, soy sauce, beer, and wine. The use of bioprocessing to induce the desired physical or chemical changes has advantages and disadvantages over traditional chemical processes that do not employ living cells and their components. The advantages of biological processes include mild reaction conditions, specificity, and the use of renewable resources.[2] Most biological processes are conducted at temperatures in the range of 25–37
C, at a pH range of 5–8, and at atmospheric pressure. Therefore, the operation is less hazardous and the manufacturing facilities are less complex compared to a traditional chemical process. The use of biomass is also a noteworthy production advantage, as biomass is easily regenerated without deleterious effects on the environment. The disadvantages of biological processes for industrial-scale processes, however, include complex product mixtures, dilute aqueous environments, contamination, and variability.[2] When living cells are used to produce a desired product, multiple enzyme reactions typically occur, and the final product mixture contains a complex mixture of cell mass and its components, metabolic by-products, and remnants of the original nutrients. The desired product must be separated from this complex product mixture. Biological processes secrete small amounts of a product into an inherently dilute aqueous environment, which further complicates separation techniques. Contamination from unwanted environmental bacteria and molds is a paramount concern with biological processes, particularly when slow-growing plant and animal cells are used. Living cells tend to mutate over time and can lose characteristics vital for the success of a process.[2] 199

200

Some cellular components are also unstable at room temperatures for long periods of time, which can limit a continuous process. Thus, the advantages and disadvantages of biological processes should be compared against traditional chemical methods on a case-by-case basis.

ENZYME-MEDIATED PROCESSING To convert raw materials into a desired product, it is advantageous to utilize cellular components instead of complete cells in some instances. Enzymes are the most common cellular components used to catalyze biochemical reactions. They are a class of proteins that increase the rate of a reaction without undergoing permanent chemical changes themselves. Enzymes can be obtained either from a pure culture of micro-organisms or directly from plants and animals.[2] A small amount of enzyme can produce large amounts of a product because enzymes are highly specific and catalyze specific chemical reactions. They also increase the rate of a reaction much faster than nonbiological catalysts. A historical use of enzymes is the use of cow stomachs, containing the enzyme rennin, to curdle milk and initiate the cheese-making process. Enzymes are now applied commercially on both large and small scales. The three major categories for commercial production are industrial enzymes, analytical enzymes, and medical enzymes.[2] Industrial enzymes are applied in tons, while analytical and medical enzymes are applied in the range of milligrams to grams. There are several bulk industrial enzymes that are applied on a large scale, as outlined by Ward.[1] Carbohydrases are used to process starches. Cellulase can convert cellulose, the largest component of biomass, into sugars that can be further converted into fuel ethanol via fermentation. Proteases have several industrial applications, particularly in the food processing industry. Lipases and lactases are used in several food processing applications as well. Ward lists several key enzyme properties that are important for applications in bioprocessing.[1] The primary considerations are the specificity of an enzyme with respect to the type of reaction being carried out, the substrates used, and the nature of the product. Another vital aspect for efficient processing is the enzyme solubility and stability at different pH and temperature conditions. The structure, charge, and shape of an enzyme must also be known when purification and recovery of the enzyme are desirable. It is necessary to develop rate equations from kinetic studies that quantify how the rate of an enzymatic reaction is affected by various chemical and physical conditions to calculate reaction times, yields, and optimum economic conditions. These studies are typically

Bioprocessing

done by measuring the concentrations of product and substrate with respect to time at different enzyme concentrations. Lee describes the three most common approaches to derive the rate equation for an enzymatic reaction.[2] The Michaelis–Menton method assumes that there is an enzyme–substrate complex that forms much faster than the product-releasing step. In this approach, the rate equation is simplified by assuming that the slow step determines the rate, and that the other step is at equilibrium. The Briggs– Haldane method assumes that the change in the enzyme–substrate concentration with respect to time is negligible. This is also known as the pseudo–steadystate assumption. Finally, the numerical approach determines the overall rate expression for an enzymatic reaction by numerically solving the differential equations simultaneously for substrate and product concentrations. The reuse of an enzyme can be economically favorable when a high-cost enzyme is used. It can be difficult to separate and reuse an enzyme because enzymes are typically globular proteins that are highly soluble in water. A common technique to facilitate the reuse of a high-value enzyme is to immobilize the enzyme onto a surface, inside of an insoluble matrix or within a semipermeable membrane. Both chemical and physical means can be employed to immobilize enzymes. The former method involves the covalent attachment of enzymes to water-insoluble supports and is the most widely used method for enzyme immobilization.[2] Lee outlines three different physical methods that are commonly utilized for enzyme immobilization.[2] Enzymes can be adsorbed physically onto a surfaceactive adsorbent, and adsorption is the simplest and easiest method. They can also be entrapped within a cross-linked polymer matrix. Even though the enzyme is not chemically modified during such entrapment, the enzyme can become deactivated during gel formation and enzyme leakage can be problematic. The microencapsulation technique immobilizes the enzyme within semipermeable membrane microcapsules by interfacial polymerization. All of these methods for immobilization facilitate the reuse of high-value enzymes, but they can also introduce external and internal mass-transfer resistances that must be accounted for in design and economic considerations.

CELL-MEDIATED PROCESSING Bioprocessing also uses living cells as an alternative to enzymes and other cellular components. Each different cell species contains a variety of cellular components that uniquely transform raw materials into desirable products. The use of living cells thus eliminates the need for production and purification of a specific enzyme.

Bioprocessing

Living cells, however, can also produce by-products that are potentially harmful to certain products. Fig. 1 shows a typical biological process that utilizes living cells.[2] First, a culture of the desired cell line is used to inoculate a small flask, referred to as a shake flask. This culture is then used to inoculate a small bioreactor. The culture contents of this small bioreactor are then used to inoculate a larger bioreactor. Inoculums are subsequently transferred to reactors of increasing size until the culture volume is large enough to inoculate a production-scale bioreactor. The cells are then separated from the liquid medium when the culture growth in the production-scale bioreactor is complete. The subsequent recovery steps depend on whether the product of interest was secreted to the supernatant or retained within the cells during production. A series of purification steps are then typically required before a product is ready for packaging. Ward discusses several considerations in the selection of organisms utilized in a bioprocess.[1] The gaseous requirements of an organism, particularly oxygen and carbon dioxide requirements, can influence reactor design parameters. The nutritional requirements of different organisms are also a consideration. Bacteria, fungi, and yeasts can require nitrogen, minerals, and specific nutrients in addition to the carbohydrates used as a carbon source. Plant cell cultures require only a well-defined mixture of salts and carbohydrates. Mammalian cells, on the other hand, need a variety of complex nutritional requirements that can be very expensive. The strength of an organism’s cell wall is a key consideration, because the ability of an organism to withstand shear forces generated from agitation plays an important role in reactor design. The growth rate of a cell line also influences processing considerations. As a general rule, the average growth rate of a cell population decreases with increasing cell complexity. Hence, mammalian cells will grow slower than plant cells, which in turn will grow much slower than yeasts and bacteria. Small-scale trials are conducted to determine the growth and production characteristics of a cell line as a function of the culture environment once a suitable host organism is selected. A small-scale culture typically requires shake flasks with capacities in the

201

range of 125 ml to 1 L. The primary environmental conditions that are optimized at this stage of the process include the medium composition, pH, and temperature.[3] Other parameters that can be calculated from small-scale experiments include cell growth rates, specific productivities, and product yields. Understanding the growth kinetics of a cell system is particularly important in bioreactor design. Lee describes a typical growth cycle for the cultivation of cells in a batch process.[2] The lag phase of growth typically occurs immediately following inoculation, and is a period where cellular growth and division is minimal. This lag phase usually occurs because the cells must adjust to the new medium before growth can begin. The cells may increase in size during this period even though the cell density does not increase. The growth of a culture then accelerates and the cell density starts to increase until the cellular division rate approaches a maximum value. During the exponential growth phase, the cellular division rate remains at a maximum and constant value, and the cell density increases exponentially as the cells rapidly divide. At the close of the exponential growth phase, the growth decelerates because the cell density begins to approach a maximum value, and there is a consequent decrease in the cellular division rate. During the stationary phase of growth, the cell density remains at its maximum value and does not increase any further. Finally, during the death phase of growth toxic metabolic by-products accumulate and the cells deplete nutrients, which results in cellular lysis.

BIOREACTOR UTILIZATION A bioreactor is a device that transforms raw materials into products by means of biological processes conducted by enzymes or living cells.[2] Bioreactors are frequently called fermenters when they generate a product by means of microbial cells, because the growth of microbes in a vessel is commonly called fermentation. There are several different bioreactor designs that can be customized to a desired process.[3] The processing conditions in a reactor, such as temperature, pH, and aeration, depend on the aspects of a particular

Fig. 1 Steps in a typical bioprocess.

B

202

reactor. The mode of reactor operation, such as batchwise or continuous production, is also configured for a specific biological process. The main challenge in bioreactor design for aerobic cell lines is to provide adequate mixing and aeration. A stirred tank reactor is most commonly employed for large-scale aerobic bioprocesses.[2] In a stirred tank reactor, mechanical agitation by an impeller accomplishes the mixing and bubble dispersion, and there are a variety of impeller configurations available. A conventional stirred tank reactor also contains a number of baffles at the vessel perimeter.[1] When air is introduced at the base of the tank through the use of a ring sparger, the rotating impeller and baffle system facilitate the dispersion of air bubbles throughout the medium.[1] The agitator can, however, damage shearsensitive systems such as mammalian and plant cell cultures.[2] A common design assumption is that a stirred tank reactor is well mixed at each point in the vessel during all of the production phases. A bioreactor is referred to as a batch reactor when all of the substrates are added to the vessel before inoculation, and the products are removed only at the end of the process. This definition of batch operation refers to the addition and removal of liquid and solid components from the system, but does not include the continuous addition of oxygen that must be supplied to aerobic systems. Most commercial bioreactors are operated in the batch mode.[3] To comply with manufacturing regulations, a batch process is typically advantageous because contamination or genetic changes in a cell line can occur during a long-term process. Fed-batch operation incorporates intermittent or continuous feeding of nutrients during production. This type of operation can mitigate the effects of catabolite repression, reduce the viscosity of the medium, lessen the effects of toxic medium components, or extend the production phase of a process for as long as possible.[1] Examples of processes where fed-batch operation is particularly useful include the production of baker’s yeast and penicillin.[3] Stirred tank reactor systems can also operate in a continuous mode. In this configuration, fresh medium is continually supplied to the reactor and the desired products are continuously removed in the course of production. A continuous system is referred to as a chemostat when the flow rate is set to a constant value. It is further known as a turbidostat when the flow rate is set to maintain a constant turbidity or cellular concentration.[2] Continuous reactor systems are commonly abbreviated as CSTR or CSTF and they refer to continuous stirred tank reactor and continuous stirred tank fermenter, respectively. An important parameter to minimize in the design and operation of a continuous reactor is the residence

Bioprocessing

time, which is the ratio of the reactor volume over the reactor flow rate. When a product is formed during the exponential growth phase, continuous operation is advantageous over batch operation because the residence time is lower in the first scenario. Batch operation is preferred when most of the production occurs during the stationary phase of growth, because the residence time for a batch operation in this scenario is much smaller.[2] Concerns with continuous operation arise from foaming and plugging of the system with large cell aggregates, along with contamination and genetic instability of a cell line. Continuous reactors are particularly useful when a process employs immobilized cells and enzymes, or when cell recycling can be incorporated. The tubular flow reactor is another common reactor configuration. In this reactor, cells and medium are continuously added at one end of a cylindrical tube and biological processes occur as the reaction mixture flows along the tube. The fluid in a tubular reactor can flow either vertically or horizontally.[3] There is no fluid mixing in this method of operation, and thus the stream composition varies in both the radial and the longitudinal directions. Variations in the radial direction, however, are typically small compared to variations in the longitudinal direction. When radial variations are neglected, the reactor is known as a plug flow reactor, frequently abbreviated as PFR. Plug flow operation can be a simulation of batch operation in a continuous flow system because substrate and product conversions vary as the reaction proceeds along the tube in a manner analogous to batch operation. Doran lists several considerations between batch, continuous stirred tank, and plug flow reactor configurations.[3] A number of continuous stirred tank reactors configured in series will theoretically approach the same reaction characteristics of those in a plug flow or a batch reactor. For single reactors not in series, the reaction kinetics determine the benefits between reactor configurations. If the reaction is zero order, there is no difference between batch, continuous stirred tank, and plug flow reactors, as far as the reaction considerations are concerned. For first-order reactors, however, batch and plug flow reactor operations will lead to higher conversions than continuous stirred tank reactors with the same volume. This is because high reactant concentrations are optimal for first-order reaction kinetics and the reactants in a CSTR are instantly diluted as they enter the vessel. When a reaction is autocatalytic, which is the case with biological reactions where reaction rates increase as the amount of biomass increases, the rate of conversion in a continuous stirred tank reactor can be significantly greater than in a batch or plug flow reactor. In addition to the standard modes of reactor operation, alternative reactor configurations also facilitate

Bioprocessing

complicated bioprocesses. These alternative configurations are typically designed to improve on the disadvantages of traditional stirred tank reactors such as the high-power requirements of impeller operation and shear damage from impeller tips. Additionally, alternative reactor configurations can more efficiently meet the specific requirements of a system, such as better aeration, more effective heat removal, increased cell separation and cell retention, and cell immobilization.[2] Airlift reactors are an alternative bioreactor in which the agitation is accomplished by a clear cyclic pattern of air flow through both a riser and a downflow reactor compartment. Stirred tank and airlift reactors are most often used for aerobic cell cultures. Airlift reactors are often utilized to culture plant and animal cells because they typically have lower shear levels.[3] Another alternative bioreactor is the bubble column, which is composed of a long cylindrical vessel with a sparging device at the bottom. In a bubble column, the mixing action of rising bubbles fulfills the oxygen needs of the cells.[2] Bubble columns are used industrially for the production of baker’s yeast, beer, and vinegar, and for wastewater treatment.[3] While column reactors are the simplest type of reactor, they are inflexible and limited to a relatively narrow range of operating conditions.[2] For example, if the fluid has a high viscosity, a column reactor cannot provide adequate mixing and mass-transfer characteristics.[3] Apart from the aeration, mixing, and conversion parameters associated with a particular reactor configuration, there are also several practical considerations that must be accounted for in the design of an industrial bioreactor. Bioreactors must be able to withstand up to 3 atm of positive pressure and temperatures of up to 150–180
C, which is why most large vessels are typically made of steel.[3] The aseptic operation of a vessel is a key concern as 3–5% of industrial fermentations are lost because of a failure in sterilization procedures.[3] Bioreactor control is also an important consideration as deviations in reactor conditions can impact cellular growth and metabolism. Most bioreactors contain on-line measurement devices that determine the temperature, pressure, aeration rate, agitation rate, pH, and other key parameters for a designated biological process. Doran outlines several of the key parameters involved in the scale-up of a biological process from laboratory-scale shake flasks to production-scale bioreactors.[3] In the first stage of studies a bench-top bioreactor, typically 1–2 L, is used to determine the oxygen requirements of the cells, their shear sensitivity, foaming characteristics, and any limitations that the reactor imposes on the organism. The results of these early studies enable decisions regarding operation in the batch, fed-batch, or continuous mode. A pilot-scale

203

bioreactor, which typically has a capacity of 100– 1000 L, is used in the next phase of process development. The pilot-scale reactor maintains an equivalent geometry, method of aeration and mixing, impeller design, and other important features of the bench-top bioreactor. A loss or variation in the performance of a cell line often occurs in the conversion from a bench-top to a pilot-scale bioreactor, even though the described reactor aspects remain constant. Thus, the results of pilot-scale findings often necessitate the reassessment of economic projections and design considerations. The industrial-scale phase of process development commences after the successful completion of the bench-top and pilot-scale phases. During this phase of development, the production-scale bioreactor is tested, along with all of the auxiliary services, such as the air supply, sterilization process, steam and cooling water supply lines, and reactor control devices.

DOWNSTREAM PROCESSING After biological reactions have generated a product of interest, it is necessary to recover this product from a liquid mixture that typically contains several undesirable components. The treatment of any culture broth after bioreactor cultivation is known as downstream processing. Downstream processing can account for 60–80% of the total production cost, particularly in the production of modern recombinant proteins and monoclonal antibodies.[2,3] A typical downstream process requires several steps in the areas of solid–liquid separation, cell rupture, product recovery, and product purification. It is important to minimize the number of downstream processing steps required because significant product losses inevitably occur during each step.[3] The product from a desired fermentation is either secreted from the cells into the fermentation broth or is entrapped within the cells themselves. In either case, the first step in downstream processing is to separate the cells from the fermentation broth. This step is typically accomplished by filtration or centrifugation and is known as solid–liquid separation. Filtration separates cells from a fluid by forcing the fluid through a porous filter medium, which deposits solids as liquids pass through. Vacuum or positivepressure equipment is used to create the driving force for filtration. The main advantages of filtration include high rates of separation, low cost, mechanical simplicity, and relative ease of maintenance. However, it can have a low retention or poor containment, and can require the addition of a filter aid to ensure good filtration when solids accumulate on the membrane.[4]

B

204

Centrifugation is an alternative to filtration. In bioprocessing, centrifugation removes cells from the fermentation broth, eliminates cell debris, and collects precipitates.[3] The equipment used for this purpose is usually more expensive than filtration equipment. It, however, is more effective than filtration when particles are small and difficult to filter. There are two basic types of centrifuges that are employed on a large scale. A tubular centrifuge is a hollow cylindrical tube in which a suspension is fed through the bottom and the clarified liquid is removed from the top.[2] Solids are deposited on the wall of a tubular centrifuge during this process. The solids are removed manually from the wall, which hinders continuous operation. A disk centrifuge is more commonly utilized when continuous operation of the centrifuge is required. This kind of centrifuge is a short and wide bowl that spins on a vertical axis and contains closely spaced cone-shaped disks that increase the capture of particles on the surface. The clarified liquid flows toward the center of the bowl and is expelled through the annulus. The solids are thrown outward and can be discharged intermittently or continuously.[2] If the desired product is entrapped within the cells themselves, the cells are disrupted by mechanical or chemical means. The former include grinding with abrasives, high-speed agitation, high-pressure pumping, and ultrasonic techniques.[3] The latter include the treatment of cells with surfactants, alkalis, organic solvents, or osmotic shock.[2] In general, the mechanical methods of disruption have the widest application, while the chemical methods are generally cell and product specific and thus not applicable to all systems.[5] While cell rupture techniques have to be very powerful, care must taken not to damage the desired components. For example, low temperatures are often required during cellular disruption procedures to prevent product loss. The recovery phase of a bioprocess begins once the products are freely suspended in a solution, either in the supernatant from the solid–liquid separation or in the lysate from the cellular disruption. Extraction is the process of separating the desired components in a liquid solution by contact with another insoluble liquid into which the desired product will be selectively extracted. This process is also commonly known as liquid–liquid extraction. The solvent-rich phase is called the extract and the residual liquid from which the desired product has been removed is called the raffinate. Adsorption serves the same purpose as extraction in the recovery, or isolation, of the desired product from a dilute solution. In adsorption, a specific substance is adsorbed selectively by certain solids as a result of a specific physical or chemical interaction.[2] Three common adsorption methods include conventional

Bioprocessing

adsorption, ion-exchange adsorption, and affinity adsorption. The adsorption of organic chemicals onto charcoal or porous polymeric adsorbents is a broad application of conventional adsorption.[3] Conventional adsorption is a reversible process caused by intermolecular forces of attraction.[2] Ion-exchange adsorption occurs when charged ions reversibly interact with charged surfaces on the molecule of interest, and it is an established practice for the recovery of amino acids, proteins, antibiotics, and vitamins.[3] Affinity adsorption is a very selective process that depends on specific chemical interactions between the molecule of interest and a solid resin.[1] The product is often purified further after the recovery phase, particularly when the desired product is a protein of pharmaceutical interest. Precipitation is a widely used method that can be induced by the addition of salts, organic solvents, or heats.[2] It can both purify and concentrate a particular protein fraction, and is frequently accomplished by the addition of salts for a ‘‘salting out’’ effect.[1] The addition of a salt precipitates proteins because increasing salt concentrations reduce the solubility of a protein in a solution.[2] Even though precipitation is an effective and relatively inexpensive method, it is also a fairly crude step and is often followed by chromatographic separations.[6] Chromatography is a purification method similar to adsorption as both methods involve an interaction between a solute and a solid matrix. Many principles of adsorption also apply to chromatography. A solution composed of several different solutes is injected at one end of a chromatography column, and each solute moves at a rate dependent on its relative affinity for a solid and immobile resin, which is known as the stationary phase. Solutes with a higher affinity for the stationary phase will move through the column more slowly than those with a lower affinity for the stationary phase. Chromatography differs from adsorption because it is based on the different rates of movement for solutes in a column, rather than the capture of one solute by an adsorbent.[2] The stationary phase in chromatography is commonly packed in a cylindrical column and can be an adsorbent, an ion-exchange resin, an affinity resin, a porous solid, or a gel.[2] A chromatography process is typically scaled up by maintaining a constant column height, fluid velocity, and sample concentration, while increasing the column diameter and the volumetric flow rate in proportion to the column cross-sectional area.[7] In electrophoresis, the net surface charge and the size of molecules separate them from one another in an electric field.[1] Electrophoresis is one of the most effective methods of protein separation and purification. It has a very high resolving power to clearly separate charged protein molecules. This method of

Bioprocessing

separation, however, is not amenable to all mixtures as the components must have an ionic form and each component must migrate differently in an electric field.[2] Membrane separation is a purification step that uses polymeric membranes to separate components primarily on the basis of size. Pressure is the main driving force in membrane separations. Microporous membrane filters are frequently used to sterilize biological solutions and have pores ranging from 0.025 mm to several micrometers in diameter.[1] Reverse osmosis and ultrafiltration membranes can separate on the basis of molecular size with ranges of 500–1000 Da and 1000–1,000,000 Da, respectively.[1] A cross-flow configuration, in which recirculation of the liquid through thin channels provides a constant flow parallel to the membrane surface, frequently reduces the accumulation of molecules on the membrane surface.[2] After purification, product packaging and marketing complete a bioprocess. Lyophilization is a common packaging technique for long-term storage. Lyophilization utilizes the principle of sublimation to dry a product in a manner that facilitates rehydration.[1] Medical and clinical trials are required before a product can be released as a health-care application if the product is a new pharmaceutical. Most regulatory agencies also require a set of well-documented manufacturing protocols and standards during all of the production phases.[3]

CONCLUSIONS As the potential applications of biotechnology are vast, bioprocessing will be a key technology in the future. Scientific breakthroughs made in gene expression and protein engineering will rely on bioprocessing techniques to incorporate these breakthroughs into new products and services. Developments in biological

205

processing will not only continue to impact food and beverage production, but will also significantly advance health-care and environmental conservation efforts. A sound understanding of bioprocessing principles will be increasingly important to the scientists and engineers of the future to utilize new biological processes on a large scale.

REFERENCES 1. Ward, O.P. Bioprocessing; Bryant, J.A., Kennedy, J.F., Eds.; Open University Press Biotechnology Series; Open University Press: Buckingham, U.K., 1991. 2. Lee, J.M. Biochemical engineering. In Prentice Hall International Series in the Physical and Chemical Engineering Sciences; Amundson, N.R., Ed.; Prentice Hall: Englewood Cliffs, NJ, 1992. 3. Doran, P.M. Bioprocess Engineering Principles; Academic Press Limited: London, 1995. 4. Reed, I.; Mackay, D. Clarification techniques. In Protein Purification Techniques, 2nd Ed.; Roe, S., Ed.; Oxford University Press: Oxford, 2001; 51–82. 5. Cumming, R.H.; Iceton, G. Cell disintegration and extraction techniques. In Protein Purification Techniques, 2nd Ed.; Roe, S., Ed.; Oxford University Press: Oxford, 2001; 83–108. 6. Harris, E.L.V. Concentration of the extract. In Protein Purification Techniques, 2nd Ed.; Roe, S., Ed.; Oxford University Press: Oxford, 2001; 111–154. 7. Hagel, L. Separation on the basis of chemistry. In Protein Purification Techniques, 2nd Ed.; Roe, S., Ed.; Oxford University Press: Oxford, 2001; 155–183.

B

Bioremediation B Teresa J. Cutright Department of Civil Engineering, The University of Akron, Akron, Ohio, U.S.A.

INTRODUCTION Biologists agree that the natural biodegradation of decaying plants, animals, and other organics by organisms has occurred since the beginning of time. The natural degradation of organic matter is the direct result of the microorganism’s metabolism to protect itself from the environment, obtain nutrients, produce energy in a usable form, convert nutrients into cellular form, discard unnecessary products, and replicate genetic information. Mankind has been able to manipulate microbial metabolism for his own benefit. The first documented use of this dates back to the use of yeast for bread in ancient Egypt and fermentation of grapes in ancient Greece.[1] The implementation of microbiological processes evolved with man to encompass other food and pharmaceutical applications. By the beginning of the 20th century, bacteria applications had expanded to include the treatment of wastewater. However, it was not until the early 1970s that scientists investigated the viability of the bioremediation of contaminated soil.[2] Although the mechanisms behind bioremediation have existed since the dawn of time, its application for the treatment of hazardous contaminants is still considered a new technology. Bioremediation as a remedial technology has several advantages. It can:
Simultaneously treat subsurface soil and water.
Be completed in less time than traditional pumpand-treat operations.
Eliminate long-term liability.
Result in complete mineralization of the contaminants.
Be conducted under mild operating conditions.
Easily be combined in a treatment train.
Be completed with little or minimal stress on environmental and ecological systems. Because bioremediation is still considered a relatively new technology, it can be scientifically intensive at the beginning of the project. For instance, a thorough site characterization can be required to ascertain the extent of contamination. In addition, extensive monitoring can be required to gain an understanding of the chemical and biological process responsible, and to track degradation by-products. Overcoming Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120007679 Copyright # 2006 by Taylor & Francis. All rights reserved.

the mass transfer limitations during in situ applications and long treatment times are two other common disadvantages. However, the primary limitation to successful bioremediation is that it is often initiated without knowing the full picture. For instance, other unidentified contaminants, or even the degradation by-products may be toxic to the microbes. This can result in increasing the time required for the treatment or stopping remediation altogether. Bioremediation is still an extremely desirable treatment approach as the advantages still far outweigh the disadvantages. Bioremediation is desirable because it is very costeffective and the most environment-friendly remediation technology. This entry contains an overview of bioremediation including the common terms associated with bioremediation, the requirements for successful applications, the key steps for treatability studies, and an introduction to degradation pathways.

TERMINOLOGY AND CATEGORIES ASSOCIATED WITH BIOREMEDIATION The most basic definition of bioremediation is the ‘‘use of any living organism to convert contaminants into less harmful compounds.’’ Previously this definition was restricted to microorganisms or microbial processes (e.g., microbial mediated enzyme reactions), where microorganisms referred to eubacteria, archebacteria, and unicellular organisms such as bacteria, yeasts, fungi, and protozoa. Although ‘‘living organisms’’ encompass multicellular organisms, it was not until the early 1990s that bioremediation applications were extended to include plants (i.e., phytoremediation). As shown in Fig. 1, the bioremediation umbrella contains several classifications depending on the approach implemented, additive(s) introduced, and location of the treatment. The subdivision of organisms based on their metabolic characteristics (i.e., chemotrophs, phototrophs, mesophiles, etc.) are not presented here but can be found in detail in several texts.[3–5] The top of Fig. 1 contains the most common descriptors applied to bioremediation. Mineralization is the complete conversion of an organic compound into biomass, carbon dioxide, water, and salts. The difference between the degradation of natural organic 207

208

Bioremediation

Fig. 1 Descriptive terms for bioremediation applications.

matter and chemical contaminants is the ease of degradation, complexity of the structure to be degraded, and time required for achieving complete mineralization. Biodegradation is also used to describe any of the microbial mediated degradation steps prior to mineralization. It includes substitution or transformation reactions that cause a simplification in the compound structure. For instance, a substitution process would be the introduction of a hydroxyl group in place of chloride [Eq. (1)], while transformation would facilitate the breaking of intermolecular bonds [Eq. (2)]. The generic term of transformation does not differentiate between whether the primary structure of the parent compound is still evident or not. As shown in Eq. (2), the structure of the parent compound is no longer easily discernable. Conversely, biotransformation is the alteration of the molecular structure via microbial mediated enzymatic catalysis where the structure of the parent compound is still evident [Eq. (3)]. The goal of the intermediate steps in both biodegradation and biotransformation is to yield a compound that is more susceptible to microbial attack (i.e., less recalcitrant). Detoxification refers to both abiotic and biotic processes that render the compound less toxic. The biotransformation of 2,4,6-chlorophenol to 2,6chlorophenol as shown in Eq. (3) is an example of detoxification because toxicity generally increases with the increasing number of chloride substitutions.

The bottom of Fig. 1 contains the subcategories used to describe the application, approach, and location of the bioremediation process. In defining the different approaches, it is important to note that indigenous species are those microorganisms inherent to the contaminated media or site. The terms biostimulation and bioenhancement are often interchangable. Both refer to the addition of a terminal electron acceptor (TEA), supplemental nutrients, auxiliary carbon sources, or substrate analogs to stimulate the activity of the indigenous species. The subtle distinction occurs with regard to the location of the additives. Biostimulation is the delivery of the nutrients and=or TEA to the indigenous microbes. Occasionally, bioenhancement is used to distinguish when the stimulation of the ‘‘natural microbes’’ occurs in aboveground containers. There are three instances where the indigenous species cannot be stimulated, viz., the appropriate microbial population is not present, bacterial numbers are too low and cannot be increased in an acceptable time frame, and contaminant concentrations or type

Bioremediation

are toxic to the bacteria.[4] When the indigenous species cannot be stimulated, bioaugmentation can be tried. Bioaugmentation is the addition of foreign microorganisms for the remediation of a contaminated site. Foreign does not imply the use of genetically engineered species, but refers to any microbe not inherent to the contaminated site. In selecting the foreign species, it is important to ascertain whether the microbes will survive in the new environment, are effective for the target compounds, and can easily be distributed in the subsurface.[6] Often the addition of supplemental nutrients and TEAs (i.e., biostimulation) occurs in conjunction with bioaugmentation to assist with biomass generation. As previously mentioned, phytoremediation is the use of plants for the direct (phytoextraction) or indirect (stimulation of the rhizosphere) treatment of contaminated media. The physical location of the bioremediation process is divided into two primary categories, i.e., in situ and ex situ, which can also be used in conjunction with the descriptors defined above. In situ applications are those processes that take place in the contaminated media without excavation. ISB is the acronym commonly used for in situ bioremediation. Natural attenuation is also referred to as intrinsic bioremediation. It is the result of several natural processes (abiotic transformation, biodegradation, sorption, etc.) that reduce contaminant concentrations in the environment without human intervention. Bioventing and biosparging are used to distinguish the location of delivered nutrients=TEAs. Bioventing introduces the TEA to stimulate activity in the vadose zone (i.e., the top 2 ft of the unsaturated soil) at approximately 1=10 the delivery rate of soil vapor extraction. Biosparging introduces the material below the groundwater level. Although biosparging is predominantly associated with TEA delivery, it can also be used to describe the introduction of nutrients and=or microbes. Ex situ applications are processes that occur after the contaminated media have been excavated. Once excavated the media can be treated on-site or taken off-site for subsequent treatment. Composting is the oldest ex situ treatment that utilizes a mixture of mesophilic and thermophilic organisms. Windrows of soil are constructed to a height of 3–4 ft and length of 50–200 ft. Water is added weekly to provide the necessary moisture and regulate the internal temperature of the compost pile. Landfarming occurs on lined containers of various dimensions. The soil is applied at a maximum of 1 ft and is tilled one or two times a month to enhance aeration and nutrient delivery. Biopiles are 6 ft mounds of excavated soil that were premixed to provide a ‘‘uniform’’ dispersion of the contaminant. An array of air ducts is placed within the bottom of the pile to introduce the required TEA. Periodically, the piles are sprinkled with water and nutrients to

209

maintain the required moisture and nutrient levels. As with composting and landfarming applications, biopiles are constructed on engineered liners to contain any contaminant releases. Bioreactors are used to describe any contained ex situ biological treatment that does not fall into the categories mentioned above. These systems can be operated in batch, semibatch, or continuous modes with either attached growth or free suspensions of bacteria.

CONTAMINANT CLASSIFICATIONS APPLICABLE TO BIOREMEDIATION Bioremediation has been successfully demonstrated for a variety of contaminant classifications. The majority of the studies have focused on petroleum compounds (BTEX, gasoline, diesel, jet fuel, etc.) because of their widespread occurrence as a contaminant. The other major waste classifications where bioremediation has been successful are solvents (toluene, trichloroethylene, etc.), creosote, pulp and paper, pesticides, textiles, polycyclic aromatic hydrocarbons (PAHs), and polychlorinated biphenyls (PCBs). Table 1 contains a partial list of the microbial genus successfully implemented for these contaminants. For aqueous petroleum contaminants, bacteria and yeasts are the most prevalent degraders. In contaminated soil systems, bacteria and fungi are the microorganisms responsible for degradation.[7] Petroleum compounds have also been widely studied as a result of their susceptibility to microbial attack. There are several rules of thumb associated with the ease of degradation in terms of compound structure. In general, as the compound molecular weight increases or the compound’s structure becomes more complex, degradation becomes more difficult. The relative biodegradability of contaminants under aerobic conditions, in order of decreasing ease (i.e., the easiest first, the more recalcitrant last) is:[3,4,8]











Linear alkanes (C10–C19) Gases (C2–C4) Branched alkanes (to C12) Alkanes without substitutions Alkenes (C3–C11) Branched alkenes Aromatics Cycloalkanes PAHs PCBs.

Polychlorinated biphenyls and other halogens can often be reductively dehalogenated much faster anaerobically (i.e., in the absence of oxygen) as compared to their aerobic counterparts.

B

210

Bioremediation

Table 1 Select microbial genera with successful demonstration of treating different contaminant classifications Contaminant Aliphatics

Creosote

Microbial genus Achromobacter

Flavobacterium

Acinetobacter

Micrococcus

Pseudomonas

Arthrobacter

Mycobacterium

Vibrio

Arthrobacter

Cunninghmellaa

Penicilliuma

Aspergillus Pesticides, herbicides

Bacillus

a

Fusarium

a

Pleurotusa

Cladosporiuma

PAH degraders

Thauera

Achromobacter

Athiorocaceae

Penicilliuma

Alcaligenes

Corneybacteriuma

Methylomonas

Arthrobacter

Flavobacterium

Zyleriona

Phaneorchaetea

Trametesa

Pseudomonas Dyes

Food, dairy, and meat

PAHs

PCBs

Aeromonas Micrococcus

Pseudomonas

Klebsiella

Shigella

Acinetobacter

Brevibacterium

Rhodococcus

Arthrobacter

Nitrosomonas

Vibrio

Bacillus

Pseudomonas

Aeromonas

Brevibacterium

Phanerochaetea

Achromobacter

Cunninghmellaa

Pichiaa

Acinetobacter

Corneybacterium

Candidaa

Anthrobacter

Flavobacterium

Pseudomonas

Bacillus

Penicilliuma

Sporobolmycesa

Acinetobacter

Corneybacterium

Ralstonia

Arthrobacter

Pseudomonas

Vibrio

Arthrobacter

Sporotrichuma

Trichodermaa

Eisenia

Talaromycesa

Xanthomonas

Alcaligenesa Pulp and paper

Chromobacter Solvents

a

Alcaligenes

Morganeela

Pseudomonas

Citrobacter

Mycobacterium

Rhodococcus

Desulfomonite

Nitrosomonas xanthobacter

Enterobacter

Nocardia

Fungi.

It is important to note that there are specific bacteria-contaminant combinations that will be the exceptions to the trend. In addition, this list does not include the impact of bioavailability or concentration. For instance, normal alkanes are considered the most biodegradable of all petroleum hydrocarbons. However, at elevated concentrations the C5–C10 compounds inhibit the activity of several hydrocarbon degraders.[9] Furthermore, the number, type, and position of substitutions will also influence the ease of degradation in branched chains, aromatics, and multiple ringed compounds. As the complexity of the compound’s structure and the number of compounds present at a site increase, one microbial strain will

not be able to facilitate the complete bioremediation of target compounds and degradation by-products. In these situations, the use of a consortium, or mixed population of bacteria, should enable complete mineralization via cometabolism. Cometabolism is the indirect metabolism of a nongrowth substrate or recalcitrant compound during the transformation of the required growth substrate.

ELEMENTS ESSENTIAL TO BIOREMEDIATION There are several basic microbial requirements that must be in place if bioremediation is to be successful.

Bioremediation

They are essential for supporting any microbial ecosystems and include the appropriate microbes, a substrate for carbon and energy source, TEA, an inducer to facilitate enzyme production, nutrients for microbial maintenance and growth, mechanisms to degrade metabolic by-products, the absence of toxicity and competing organisms, and the appropriate environmental conditions.

211

Table 2 Common TEAs used for aerobic, anaerobic, and fermentation respiration Respiration mode Aerobic

Terminal electron acceptor Oxygen H2O2 Various organics

Anaerobic

NO3 SO42 Various organics

Adequate Microbial Population Methane fermentation

The primary requirement for any successful biological treatment is the presence of an adequate microbial population. As shown earlier in Table 1, there are a wide variety of microbial species with demonstrated success at remediating an array of contaminants. Soils that have favorable growth conditions will exhibit viable microbial numbers of greater than 104 colony forming units (CFU)=g of dry soil. Normal activities in ‘‘clean’’ groundwater are in the range of 106–108 CFU=L. Soils and groundwaters depicting activities 10,000 ppm). The exact impact of concentration is dependent on the micro-organism(s). Typically, biodegradation rates are modeled as first order but will change with time. Rate changes occurring over a long period of time (months and years) are often the result of the time required by the bacterium to alter its metabolism for inducing enzymes essential for that catabolic pathway.[12] Conversely, short time frame changes are often associated with those systems that can be represented as Michaelis–Menton kinetics.[9] The number of active species as well as the dominant genus in a microbial population can change (i.e., shift) significantly over time. For example, hydrocarbon degraders are typically 20% (w=v). Polyethylene glycol is an exception and is useful for precipitation of low-solubility proteins like globulins. With higher-solubility proteins like albumins, the difference in density between the aggregate and solution is negligible. This makes centrifugation impossible. Recovery of protein from the PEG–protein complex after precipitation is, however, very slow using dialysis. Crystallization differs from precipitation in that it produces a regular lattice of molecules of the product. As biological products can be obtained at very high purity by crystallization, it is also often used as a polishing step. Sometimes, crystallization as a purification step can obviate the need for expensive chromatographic steps. Crystallization can be induced by pH adjustment, lowering ionic strength, or adding seed crystals of the desired protein to the solution.[8]

Adsorption Adsorption is based on the principle of the differential affinity of dissolved solutes in a mobile liquid phase to a stationary solid phase. Adsorption columns usually have a smaller capacity but higher specificity than extraction. Various adsorbents have been in use, including activated charcoal (usually from vegetable sources), and more recently, customized resins that can have either positive or negative charges on them, or

Bioseparations

Fig. 3 Three different types of adsorption isotherms.

are hydrophobic in nature. Polystyrene, methacrylate, and acrylate based adsorbents are used for low-MW compounds, cellulose based adsorbents for proteins, poly(styrene-divinylbenzene) resins for nonpolar solutes, and acrylic esters for hydrophobic solutes.[2,12] Mathematical analysis of adsorption is beyond the scope of this entry, but it is important to point out that the equilibria used are often based on adsorption isotherms. The three isotherms shown in Fig. 3 occur in practice, although linear isotherms are very rare. The nonlinearity of the Langmuir and Freundlich isotherms makes scale-up and analysis tedious. The Freundlich isotherm is based on empirical data analysis and occurs more commonly in practice than the Langmuir isotherm, which has a theoretical basis. Adsorption is usually done in a batch mode in fixed, agitated, or expanded bed adsorption units. The process usually consists of cyclic operations of adsorption and desorption in two steps. Desorption or recovery of the adsorbed solutes from the adsorbents can be done by either changing the temperature or pH, or using a solvent, which also regenerates the adsorbents for use in the next operation cycle.

PROTEIN REFOLDING Formation of IBs during protein expression poses a bottleneck in the efficient downstream processing of therapeutic proteins. The reasons for IB formation are not fully known. Because translation is a slower process than protein folding, it is likely that the misfolding of translation intermediates plays some role. Further, since post-translational modifications, such as glycosylation and lyposylation, which are known to affect the secondary structure of proteins, are absent in bacteria, the non-modified protein structure may cause misfolding. The recovery of soluble

229

active protein from purified IBs requires the denaturation of the polypeptide by a denaturing agent such as 8 mol=L urea or 6 mol=L guanidine HCl and its subsequent refolding back to an active form. Fig. 4 depicts the protein configuration during the refolding step. There is no universal method to refold any protein. The conditions needed to properly refold the protein of interest remain an empirical science and need to be optimized for each protein.[13] In general, a solubilized protein solution is usually added into a large volume of a refolding buffer to reduce the concentration of the denaturing agent and also to avoid aggregate formation of protein molecules in the course of renaturation.[14] Refolding can also be done by dialysis against a refolding buffer and buffer exchange by gel chromatography.[15,16] The dilution method is usually the easiest but greatly decreases the concentration of the target protein. Also, the requirement of a large volume and the consequent inefficient mixing in large tanks lead to heterogeneity in folding conditions.[17] Often, the refolding conditions optimized for a protein in the laboratory scale might not work on a large scale.

PURIFICATION Chromatographic techniques are the sole method for high-resolution purification of proteins and many other bioproducts. As opposed to adsorption, where the feed is continuously passed through the column until breakthrough of the strongly adsorbed components occurs, chromatography separates components in samples that are injected in a pulsatile manner. Elution of the components is achieved either in an isocratic mode, where the same mobile phase is maintained throughout, or by the gradient elution mode. In the former, the separation of components depends on the differences in the retention time of each component in the column. In the latter, the sample is introduced as a pulse and is followed by a different mobile phase containing a displacer. Conventional batch chromatography is relatively simple and offers operating flexibility, but suffers from several drawbacks as well. To obtain high purity and yield (>99%), a large amount of solvent is needed and column utilization is inefficient.[18,19] To get around these disadvantages, continuous countercurrent chromatographic methods were developed. This mode of operation maximizes the mass transfer, but it is difficult to maintain a stable solid phase velocity.[19,20] An alternative to continuous countercurrent chromatography is to simulate the movement of the solid phase by periodically moving the inlet and outlet ports, while keeping the bed stationary. This mode of operation is called simulated moving bed chromatography and has the advantage of significantly higher

B

230

Bioseparations

Fig. 4 Schematic depiction of the protein refolding process. (View this art in color at www.dekker.com.)

throughput per bed volume and low solvent use.[19] Column switching, chromatography with recycle, and reciprocating size exclusion chromatography in a semicontinuous frontal mode are refinements to conventional methods and can further reduce solvent use, increase peak resolution, and facilitate faster recovery of large molecules.[21–23] The type of matrix in the column dictates the various kinds of chromatography techniques. All matrix materials need to be inert, chemically and physically stable to withstand the harsh cleaning procedures, be rigid to withstand the high flow rates, and have the porosity to provide a high surface area for adsorption. Matrices can be inorganic like porous silica, controlled pore glass, and hydroxyapatite, synthetic organic polymers like polyacrylamide, polymethyl methacrylate, or natural polysaccharides like cellulose, dextran, and

agarose. The type of matrix often defines the basis of chromatography, be it size exclusion, ion exchange, hydrophobic interaction, or affinity. Table 4 summarizes the different types of chromatography and their basis of separation. The pattern of separated proteins obtained on the detector is termed a chromatogram. The quality of a chromatographic separation can be quantitatively appreciated by calculating the hypothetical number of plates in the peak of a chromatogram (N). The theoretical plate height (H) is also often used as a parameter to quantify a chromatogram. Fig. 5 shows a typical peak in a chromatogram. N for such a peak and is given as:
2 tr N ¼ 16 ð1Þ tw

Table 4 Various types of liquid chromatography Types Ion exchange

Basis Ionic charge

Application Proteins

Gel filtration=size exclusion

Size

Desalting, large molecules

Affinity

Specific binding

Antibodies, antigens

Reverse phase

Hydrophobicity

Peptides

Hydrophobic interaction

Hydrophobicity

Proteins

Chromatofocusing

Isoelectric point

Proteins

Ligand exchange

Adsorption

Sugars

Bioseparations

231

B

Fig. 6 Graphical representation of the van Deemter equation showing the effect of eluent velocity on the plate height of a chromatography column. Fig. 5 Typical chromatogram peak characteristics.

A greater number of plates indicate higher efficiency of separation. This implies that a smaller base width (tw) will indicate a better resolution. Further, if each plate has a height H, then, H ¼ L=N

ð2Þ

where L is the length of the column. Obviously, a lower plate height indicates a better resolution. H also depends on the interstitial velocity v of the mobile phase. The relation between H and v is given by the Van Deemter equation: H ¼ A þ

B þ Cv v

ð3Þ

where A, B, and C are constants. Knowing the Van Deemter constants for a system can give an idea about the operating velocity at which resolution of the eluting peaks will be the best. As shown in the plot of H vs. v (Fig. 6), there is an optimum v at which H is minimum. Numerically, H is minimum when v ¼ (B=C)1=2.

Bulk biochemicals like organic acids and solvents are often formulated as concentrated solutions after removal of water from the stream exiting from the final separation step. Antibiotics are usually crystallized from solution to a very high purity. Recombinant proteins are usually sensitive and can lose bioactivity if not stored in a proper manner. Proteins are formulated mostly as dry powders, and a variety of stabilizing agents are added to preserve the bioactivity of the protein.[12] Enzymes are usually freeze dried or lyophilized to obtain a solid form. Lyophilization basically sublimes the water to vapor under vacuum. It is a nondenaturing process and is used for drying pharmaceutical products, therapeutic proteins, food items, viruses, and bacteria. If the product is not very sensitive to heat, it can be dried by spray dryers, vacuum shelf dryers, or rotary dryers. These are well-known and widely used unit operations in the chemical industry. Details about these unit operations can be found in many classic chemical engineering handbooks.

CASE STUDIES POLISHING After a series of chromatography steps yield a highly pure product in the liquid state, the product needs to undergo one or more steps to convert it to the form in which it will be sold. Drugs need to be formulated and mixed with binders, colors, and other excipients if they are sold in a solid form. They can also be sold as liquids or aerosols, suspensions or emulsions. In such a case, they need to be diluted and mixed at the correct concentration.

In this section, we try to develop downstream processing strategies for two different products: rhGDNF as an IB in E. coli and EPO as a protein secreted from CHO cells. The genomes of the two different organisms used have been genetically engineered to produce the proteins. The purpose of such an exercise is to learn to integrate the basic steps discussed so far in simulating an actual process. The cases will combine the spectrum of both intracellular and extracellular proteins as well as prokaryotic and eukaryotic expression.

232

Production of rhGDNF by E. coli GDNF is an approximately 20 kDa, glycosylated polypeptide that exists in its native form as a homodimer.[24] The gene for GDNF has been mapped to human chromosome 5 at p12–p13.1 and gives rise to two alternatively-spliced forms that code for prepropeptides of 211 and 185 amino acid (aa) residues, respectively.[25] Both the long (alpha) and the short (beta) forms yield equivalent 134-aa residue mature forms after proteolytic cleavage.[24,26] Although the native molecule contains two potential glycosylation sites, nonglycosylated rhGDNF has full biological activity.[24] GDNF has the ability to promote the survival and differentiation of dopaminergic neurons in primary cultures of the embryonic ventral midbrain.[24] It has also been found to have trophic effects on dopaminergic neurons (increasing the number surviving in culture, dopamine uptake, cell size, and extension of neurites). It also protects mecencephalic dopaminergic cells against the deleterious effects of the neurotoxin 1-methyl-4-phenylpyridinium (MPPþ).[27] This finding is central to the development of GDNF for the treatment of Parkinson’s disease because the primate model of Parkinson’s disease relies on the use of 1-methyl-4phenyl,1,2,3,6-etrahydropyridine, a precursor to the neurotoxin MPPþ. By virtue of its effects on cell culture and efficacy in animal studies, GDNF constitutes a powerful therapeutic candidate for several neurodegenerative diseases, especially Parkinson’s.[25,28] For large-scale production of GDNF, E. coli is a very good choice as a host. It has several advantages in this case. First, nonglycosylated GDNF, though highly potent as a therapeutic agent, has a relatively small size (134 aa, MW 15 kDa) and simple structure. Second, the recombinant nonglycosylated form is as functionally active as the glycosylated native form found in the human brain. In general, bacterial host systems cannot reproduce the authentic glycosylation of eukaryotic proteins, and in such cases where glycosylation is important for the protein to function, a more complex host system has to be chosen. For this case, therefore, E. coli is very suitable as a host for the production of rhGDNF. However, the recombinant protein often forms an IB in such a bacterial system. This causes problems in the extraction and downstream processing of the protein. A flowsheet of the steps in downstream processing is given in Fig. 7. At the end of the fermentation, the cells are transferred into a blending tank, which isolates the upstream from the downstream section of the plant. The broth is then fed into a disk stack centrifuge and centrifuged to harvest the cells. The supernatant is centrifuged again and the cell sludges from both streams are resuspended in ice-cold buffer. The buffer facilitates the separation of the cell debris from the IBs.

Bioseparations

The suspension is maintained at 4 C to prevent heating and subsequent proteolytic degradation of the target protein during the cell lysis steps, which is done in a high-pressure homogenizer. The cell lysate from the homogenizer is a viscous extract. To reduce the viscosity, the lysate is treated with lysozyme (added to a final concentration of 100 mg=ml) and then homogenized again. The second homogenized stream is then treated with half the volume of protease inhibitor cocktail (without ethylenediaminetetraacetic acid) to inhibit any proteolytic degradation of the target protein. It is then centrifuged and the IBs are recovered in the heavy phase while most of the cell debris remains in the supernatant. This is possible because the density (1.3 g=L) and the size (j: 1 mm) of the IBs are significantly larger than the cell debris. The IB sludge is then washed with wash buffer containing Triton X100 to remove any loosely associated contaminants. The mixture is centrifuged, and the IB sludge is then resuspended in phosphate buffer solution (PBS) in a blending tank. The PBS is drained in a microfilter, and the IBs are transferred to a glass-lined blending tank and mixed with a solubilizing buffer. The solubilizing buffer contains urea solution, a chaotropic agent that dissolves the denatured protein in the IBs. At the end of the solubilization, the solution is concentrated in a diafiltration unit. All the remaining fine particles (biomass debris and IBs) are removed using a polishing dead-end filter. This polishing filter protects the chromatographic units that are used next. The filtered stream is purified using anion exchange chromatography. Two quaternary amine Sepharose columns (Q-Sepharose, Pharmacia) work in parallel. The feed is applied with 10 mM Tris-HCl (pH 8.0) containing 4 M urea and the column is developed with a linear gradient of 0–0.5 M NaCl in this buffer.[24] The next step involves the removal of the SO3 moieties from the cystine side chains in the rhGDNF to allow proper disulfide bonding and correct refolding of the protein to its native form. Pooled fractions from the Q-Sepharose column are mixed with 19 volumes of refolding buffer in a blending tank.[24] Dilution is necessary both in the solubilization and in the refolding steps to minimize the intermolecular interactions that can lead to protein inactivation. At this stage, the stream contains the dissolved rhGDNF, which is then concentrated using a diafilter followed by a dead-end polishing filter to remove any non-native protein aggregates, and then subsequently fed with 50 mM Tris–HCl (pH 8.0) into two S-Sepharose (sulfonate based cation exchanger) columns working in parallel. The column is developed with a linear gradient of 0–1 M NaCl in this buffer.[24] Fractions containing rhGDNF are applied to a hydroxyapatite column in 25 mM sodium phosphate (pH 6.8). The purified protein

Bioseparations

233

B

Fig. 7 A process flowsheet for rhGDNF production.

is lyophilized to yield a powdered form. It is 99.99% pure and contains 0, a sessile profile, and e ¼
1 when rB
rA < 0, a pendent profile] and y ¼ (z
zo)= a þ yo, where yo is the dimensionless ordinate of the datum point (usually the apical point) and is given by yo ¼ e2Hoa. Then, by introducing y0 ¼ dy=dx ¼ tan j, the above dimensionless equation can be decomposed into a system of first-order nonlinear differential equations:[6,24,34] dx x cos f ¼ df exy
sin f dy x sin f ¼ df exy
sin f

ð6Þ

which is a convenient form for numerical integration in terms of the independent variable j and can be used with no difficulty for the entire range 0  j < 1 in the sessile case (e ¼ þ1) and up to the first inflexion point 0  f < f1in in pendent configurations (e ¼
1).

C

542

Contact Angles, Surface Tension, and Capillarity

To overcome the difficulty encountered at the inflexion points, we need a system of equations in terms of an independent variable that increases or decreases monotonically with arc length regardless of the profile of the meniscus. For this the natural choice is the dimensionless arc length, s, itself. It is defined as ds ¼ ½ðdxÞ2 þ ðdyÞ2
1=2

ð7Þ

The Young–Laplace equation can now be written as a system of three first-order nonlinear differential equations: dx ¼ cos f ds dy ¼ sin f ds

ð8Þ

df sin f ¼ ey  ds x Both sessile and pendent profiles form continuous families either of which can be generated by using these equations regardless of the value of e. The choice of e, given a convention for the changes of s and j, affects only the sessile or pendent character of the starting point and its neighborhood. The initial value chosen for y for any given abscissa, x0, sets the curvature at that point and fully determines the profile. The initial value chosen for s is irrelevant; the integration depends only on the step size, Ds, and its sign. The actual configuration of a meniscus is given by that member of the family of solutions of the Young–Laplace equation that satisfies a given set of boundary conditions. By using this method we effectively change a tedious two-point boundary-value problem into a much simpler initial-value problem and the final solution is obtained by interpolating or iterating over a sequence of profiles. Larkin used a similar initial-value artifice to integrate the twodimensional partial differential equation of Young and Laplace over a domain lacking rotational symmetry.[25] It should be noted though that the Young– Laplace equation is elliptic and its solution over a domain subject to a given set of boundary conditions has to be carried out by solving a system of algebraic equations defined over the mesh points of a finitedifference lattice.

BOUNDARY CONDITIONS AND INTEGRATION START-UP FOR POINTS OF ZERO SLOPE Starting the integration of the Young–Laplace equation at a point of zero slope is exceedingly convenient

whenever possible, the reason being that the solution in the vicinity of such a point behaves as a Bessel function or a modified Bessel function of zero order. Several types of profiles, real or hypothetical, can be considered. For the purposes of this article, only sessile drops and bubbles will be analyzed. Sessile capillary menisci in the vicinity of a point with zero slope are described by the modified Bessel equation y00 þ ð1=xÞy0  y ¼ 0

ð9Þ

which can be obtained by expanding (1 þ (y0 )2)n, n ¼ 1=2, 3=2, etc. about y0 ¼ 0, and neglecting all terms (y0 )n with n  2. When the point of zero slope lies on the axis of rotational symmetry, as is the case with axisymmetric drops and bubbles, the solution for points in the immediate vicinity of the axis is yðxÞ ¼ tan fþ

I0 ðxÞ I1 ðxþ Þ

0  x  xþ

ð10Þ

where xþ is an abscissa for which the above approximation still holds within a given tolerance. Normally, this is the abscissa corresponding to a jþ of up to 0.5 ; values of jþequal to 0.1 or smaller can be chosen for safety. This method automatically gives us the coordinates of the starting point of the curve at (xþ, yþ) yþ  yðxþ Þ ¼ tan fþ

I0 ðxþ Þ I1 ðxþ Þ

ð11Þ

and the dimensionless apical curvature, i.e., the curvature at the point x ¼ 0 y0  yð0Þ ¼ 2H o a ¼ tan fþ

1 I1 ðxþ Þ

ð12Þ

The choice of a sign for jþ determines whether we obtain a sessile bowl (jþ > 0) or a sessile dome (jþ < 0) according to our convention for curvature signs. Using the Bessel approximation as a start-up artifice always gives us a two-parameter family of solutions (or a one-parameter family for each initial abscissa x0) as is always the case with a second-order differential equation. The parameter xþ is directly related to the curvature y0 or yþ of a given profile. However, to singularize a profile that passes through a point (xo, yo) three parameters are necessary (xo, yo, jo), although solutions may not exist for some combination of parametric values (for example, if xo ¼ 0, no profiles with finite nonzero slope at xo exist). In all cases, once xþ and yþ have been determined, we may proceed

Contact Angles, Surface Tension, and Capillarity

543

C

Fig. 2 Axisymmetric sessile drops.

with the integration of the equations by making use of some standard Runge–Kutta subroutine or an equivalent method.

REPRESENTATIVE ROTATIONALLY SYMMETRIC MENISCI Figs. 2 and 3 show a family of bounded axisymmetric sessile configurations integrated all the way from jþ ¼ 0 (or j ¼ jþ) to j ¼ 180 . For any given angle, j > 0, the thickness of the meniscus increases rapidly with increasing radius, passes through a maximum, and decreases slightly as it tends to an asymptotic limit as the radius tends to infinity. The limit is given by the two-dimensional solution of a translationally symmetric configuration—the separating elastica. If the integration is continued beyond j ¼ p, the profile is found to coil over itself indefinitely, as

indicated in Fig. 4, where the first few loops are shown. If the integration is extended it will be noticed that the ‘‘centers’’ of the loops fall close to a straight line—the generatrix of a conical surface—which passes through the origin (x ¼ 0, y ¼ 0) (Fig. 5). This is characteristic of all solutions of the Young– Laplace equation that have a nodoidal (i.e., coil-like) character and appears also in many other representative cases. The curves in the figures have been drawn for sessile drops with the denser liquid—shaded area—below. Sessile bubbles have identical configurations except that they are turned upside down, yet have the heavier fluid still on the lower side. All of these curves are parametrized with respect to their curvature at the origin, y0, which is equivalent to parametrization by the value of their starting xþ. The family of so-called ‘‘bounded’’ sessile configurations, that is, sessile drops and bubbles centered at the axis, belongs to the one-parameter subfamily

Fig. 3 Axisymmetric sessile drops. Locus of various contact angles: 45 , 90 , 135 , 180 .

544

Contact Angles, Surface Tension, and Capillarity

Fig. 4 Axisymmetric sessile profile integrated beyond 180 .

of solutions mentioned above. So does the family of bounded pendent configurations, that is, pendent drops and bubbles centered at the axis. Figs. 6 and 7 show equilibrium configurations of pendent drops with increasing curvature (decreasing xþ).

Similar calculations can be done for families of unbounded sessile profiles. An unbounded sessile profile can be visualized physically as the configuration of an axisymmetric dry patch, or that of a meniscus around a cylindrical rod with its axis perpendicular

Fig. 5 Axisymmetric sessile profile with extended integration.

Contact Angles, Surface Tension, and Capillarity

545

C

Fig. 6 Axisymmetric pendent drop assumed to be in static equilibrium (for xþ ¼ 0:002; 0:0018; 0:0016, and 0.0014).

to the free liquid interface, or the external meniscus around the ring of a de Nou¨y tensiometer. In this case, the boundary condition is y0 ðxÞ ! 0

as x ! 1

ð13Þ

and the solution of the Bessel approximation becomes:

yðxÞ ¼ tan fþ

K0 ðxÞ K1 ðxþ Þ

xþ  x

ð14Þ

Likewise, the same treatment can be applied to the calculation of the shapes of pendent drops and bubbles. The behavior of pendent profiles about the apical point of zero slope, y0 (x0) ¼ 0 at x0 ¼ 0, is given by the Bessel equation: y00 þ ð1=xÞy0 þ y ¼ 0

ð15Þ

of which, the solution is

yðxÞ ¼ tan fþ

J0 ðxÞ J1 ðxþ Þ

0  x  xþ

ð16Þ

where again the starting point for numerical integration is located at yþ  yðxþ Þ ¼ tan fþ

J0 ðxþ Þ J1 ðxþ Þ

ð17Þ

such that the apical curvature is yo  y(0) ¼ a2Ho ¼ tan jþ(1=J1(xþ)). Not discussed here, but also equally feasible is the calculation of the equilibrium profiles of sessile and pendent lenticular configurations—for example, drops of an immiscible liquid floating on an interface (like oil on water) form a sessile lens as illustrated in Fig. 8.[36]

AXISYMMETRIC MENISCI AND THE DETERMINATION OF CONTACT ANGLES It may be observed in Figs. 2 and 3 that the dimensionless profiles provide a unique correlation between the diameter of the meniscus (or the diameter of the meniscus at the contact line) and the elevation of the apical point with respect to the plane on which the meniscus rests. Therefore, once we have established the apical curvature of the meniscus that corresponds to that profile, it is a fairly straightforward task

546

Contact Angles, Surface Tension, and Capillarity

Fig. 7 Axisymmetric pendent drop assumed to be in static equilibrium (for xþ ¼ 0:0012; 0:0010; and 0.0008).

to determine the values of the capillary constant, a2, and the surface tension, s. The dimensions of the meniscus should be established by taking precise microscopic measurements.[37–44] Alternatively, equivalent correlations can be derived for other menisci or capillary phenomena that

will lead to similar determinations of the surface tension s. These could depend on the relative elevations of the menisci inside thin capillary tubes or in between plates, or in more accurate measurements using ring tensiometers or other commercially available devices.

Fig. 8 Sessile lenticular configuration.

Contact Angles, Surface Tension, and Capillarity

CONCLUSIONS Capillarity phenomena are everyday occurrences that result from the existence of surface tension or interfacial tensions. In addition to the static phenomena discussed herein, surface tension and capillarity are also responsible for numerous dynamic phenomena that may result from localized gradients in temperatures or in compositions; the study of dynamic capillary phenomena (e.g., Marangoni flows, Be´nard cells) is the subject of much literature coverage and is beyond the scope of this survey. Static capillary phenomena lead to precisely determined geometrical shapes like sessile menisci, pendent menisci, minimal surfaces, which can be used for the physical determination and measurement of the surface tension or the interfacial tensions between fluids. In addition to the simple forms considered herein, more complex forms (e.g., sessile lenticular drops) can be studied.[35,36] Mathematical resolution of these shapes is a combination of the (numerical) solution of the highly nonlinear Young–Laplace equation together with an appropriate set of boundary conditions. For practical purposes, only axisymmetric forms are readily amenable to mathematical analysis.

REFERENCES 1. Udell, K.S. Heat transfer in porous media considering phase change and capillarity . . . the heat pipe effect. J. Heat Mass Transfer 1985, 28 (2), 485–495. 2. Maxwell, J.C. Capillary attraction. In Encyclopaedia Britannica, 9th Ed.; 1878; Vol. 5; Rayleigh’s revisions, 11th Ed.; 1911. 3. Young, T. A course of lectures on natural philosophy and the mechanical arts, 2 volumes; J. Johnson: London, 1807. 4. Laplace, P.S. (Marquis de). Theory of Capillary Attraction; Supplement to the 10th book of Celestial Mechanics, 1805; Bowditch, N., Transl., 1839. 5. Gauss, C.F. Principia generalia theoriae figurae fluidorum. Comment. Soc. Regiae Scient. Gottingensis Rec. 1830, 7. 6. Strutt, J. (Lord Rayleigh). On the theory of surface forces. Phil. Mag. 1890, 30 (5), 285–298, 456–475. 7. Strutt, J. (Lord Rayleigh). on the theory of the capillay tube. Proc. R. Soc. 1915, A92, 184–195. 8. van der Waals, J.D. Thermodynamische Theorie der Kapillarita¨t unter Voraussetzung stetiger Dichtea¨nderung. Ver. Kon. Ak. Wet. (Amsterdam) 1893, 1 (8). Zitschr. Phys. Chem. 1894, 13, 657.

547

9. Neumann, F. Von. Vorlesungen u¨ber die Theorie der Kapillarita€t; B. G. Teubner: Leipzig, 1894. 10. Bakker, G. The´orie de la couche capillaire plane des corps purs. Scientia 1911, 31. 11. Kirkwood, J.G. Collected Works—Theory of Liquids; Gordon and Breach: London, 1968. 12. Buff, F.P. The theory of capillarity. In Handbuch der Physik; Springer Verlag, 1960; Vol. 10. 13. Dzyaloshinskii, I.E.; Lifshitz, E.M.; Pitaevskii, L.P. The general theory of van der Waals forces. Adv. Phys. 1961, 10, 165–209. 14. Ninham, B.W.; Parsegian, V.A. van der Waals forces—special characteristics in lipid-water systems and a general method of calculation based on the Lifshitz theory. Biophys. J. 1969, 10, 646–663. 15. Ninham, B.W.; Parsegian, V.A. van der Waals forces across triple-layer films. J. Chem. Phys. 1970, 52, 4578–4587. 16. Adam, N.K. The Physics and Chemistry of Surfaces; Oxford University Press, 1941; Dover Publications, 1968. 17. Adamson, A.W. Physical Chemistry of Surfaces; Wiley-Interscience, 1967. 18. Eisenhart, L.P. A Treatise on the Differential Geometry of Curves and Surfaces; Ginn and Co., 1909; Dover Publications, 1960. 19. Zhmud, B.V.; Tiberg, F.; Hallstensson, K. Dynamics of capillary rise. J. Colloid Interface Sci. 2000, 228 (2), 263–269. 20. Wayner, P.C., Jr.; Wang, Y.X.; Zheng, L.; Plawsky, J.L. Optical evaluation of the effect of curvature and apparent contact angle in droplet condensate removal. J. Heat Transfer 2002, 124 (4), 729–738. 21. Greenhill, A.G. The Application of Elliptic Functions; McMillan & Co., 1892; Dover Publications, 1959. 22. Petrov, V.M.; Chernous’ko, F.L. Determining the equilibrium form of a liquid subject to gravity forces and surface tension. Fluid Dyn. 1966, 1 (5), 109–112. 23. Concus, P.; Finn, R. On the behavior of a capillary surface in a wedge. Proc. Natl. Acad. Sci. 1969, 63 (2), 292–299. 24. Concus, P.; Finn, R. On a class of capillary surfaces. J. Analyse Math. 1970, 23, 65–70. 25. Larkin, B.K. Numerical solutions of the equation of capillarity. J. Colloid Interface Sci. 1967, 23, 305–312. 26. Kim, H.Y.; Lee, H.J.; Kang, B.H. Sliding of liquid drops down an inclined solid surface. J. Colloid Interface Sci. 2002, 247 (2), 372–380. 27. Zhou, D.; Blunt, M.; Orr, F.M. Hydrocarbon drainage along corners of noncircular capillaries. J. Colloid Interface Sci. 1997, 187 (1), 11–21.

C

548

28. Young, T. Essay on the cohesion of fluids. Philos. Trans. R. Soc. 1805, 95, 68–86. 29. Thomson, W. (Lord Kelvin). Capillary attraction. Proc. R. Inst. 1886, 11 (3), 483–507; Nature 1886, 34, 270–272, 290–294, 366–369; Popular Lectures Addresses 1981, 1, 1–79. 30. Bashforth, F.; Adams, J.C. An Attempt to Test the Theory of Capillary Action; Cambridge University Press, 1883. 31. Bakker, G. Kapillarita¨t und Oberfla¨chenspannung. In Handbuch der Experimental Physik; Akademie Verlag: Leipzig, 1928; Vol. 6. 32. Padday, J.F. The profiles of axially symmetric menisci. Philos. Trans. R. Soc. 1971, A269, 265– 293. 33. Padday, J.F.; Pitt, A. Axisymmetric meniscus profiles. J. Colloid Interface Sci. 1972, 38 (2), 323–334. 34. Huh, C.; Scriven, L.E. Shapes of axisymmetric fluid interfaces of unbounded extent. J. Colloid Interface Sci. 1969, 30, 323–337. 35. Pujado´, P.R. Capillary Hydromechanics; Dissertation, University of Minnesota, 1972. 36. Seeto, Y.; Puig, J.E.; Scriven, L.E.; Davis, H.T. Interfacial tensions in systems of three liquid phases. J. Colloid Interface Sci. 1983, 96 (2), 360–372. 37. Jennings, W.J.; Pallas, N.R. An efficient (calculation) method for the determination of interfacial tensions from drop profiles (for pendent or sessile drops oriented up or down). 195th ACS National Meeting, 3rd Chemical Congress of North

Contact Angles, Surface Tension, and Capillarity

38.

39.

40.

41.

42.

43.

44.

America, Toronto, Jun 5–10, 1988; Abstr. Coll. 37, V1. Briant, J.; Theibot, B. Measuring very low interfacial tensions and contact angles at high pressures and temperatures. Institut Franc¸ais du Pe´trole, World Surface—Active Agents Congress, Munich, May 6–10, 1984; Rev. Inst. Fr. Pet. 1985, 40 (2), 241–250. Allain, C.; Ausserre, D.; Rondelez, F. A new method for contact angle measurements of sessile drops. J. Colloid Interface Sci. 1985, 107 (1), 5–13. Rotenberg, Y. The determination of the shape of nonaxisymmetric drops and the calculation of surface tension, contact angle, surface area, and volume of axysymmetric drops. Diss. Abstr. Int. B 1984, 44 (9). Malcolm, J.D.; Elliott, C.D. Interfacial tension from the height and diameter from a single sessile drop or captive bubble. Can. J. Chem. Eng. 1980, 58 (2), 151–153. Carroll, B.J. The accurate measurement of contact angle, phase contact areas, drop volume, and Laplace excess pressure in drop-on-fiber systems. J. Colloid Interface Sci. 1976, 57 (3), 488–495. Ortiz-Arroyo, A.; Larachi, F.; Iliuta, I. Method for inferring contact angle and for correlating static liquid hold-up in packed beds. Chem. Eng. Sci. 2003, 58 (13), 2835–2855. Flock, D.L.; Le, T.H.; Gibeau, J.P. The effect of temperature on the interfacial tension of heavy crude oils using the pendent drop apparatus. J. Can. Pet. Technol. 1986, 25 (2), 72–77.

Corrosion in the Process Industries C J. A. Richardson Anticorrosion Consulting, Durham, U.K.

R. A. Cottis School of Materials, Corrosion and Protection Centre, University of Manchester, Manchester, U.K.

INTRODUCTION Corrosion makes a large impact on the economics of chemical processing. A recent U.K. government study estimated that corrosion expenditure amounts to
4.5% of the turnover of the U.K. chemical= petrochemical industry. Much of this total expenditure represents investment in materials and protection practices to manage corrosion in new equipment. However, a significant proportion arises from the failure to identify and mitigate well-known corrosion risks at the design stage, and the U.K. study estimated that 15% of the total estimated cost may be saved by the application of existing corrosion prevention technology. This entry provides an introduction to the management of corrosion in the process industries. It is necessarily rather general in character, and it should not be relied on for decision making; rather it provides a background for nonspecialists to help them to appreciate the decisions and advice of specialists.

MATERIALS FOR PROCESS EQUIPMENT Process equipment has to operate over wide ranges of temperature, pressure, and fluid composition. Volatile hydrocarbons are stored at temperatures well below –100 C, and furnace tubes may be required to operate at temperatures above 1000 C. Crude oil distillation equipment operates commonly under vacuum, whereas supercritical processes operate at pressures of several hundred atmospheres. Aqueous solutions of mineral acids, alkalis, and salts can be extremely corrosive toward metallic materials, whereas plastic materials are much more vulnerable to organic solvents. The wide diversity of commercial chemical process conditions dictates that all classes of engineering materials find use in chemical process equipment. The safe design and operation of pressurized equipment demands levels of strength and toughness that are generally achievable only in metallic materials. Consequently, the great majority of process equipment Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120007716 Copyright # 2006 by Taylor & Francis. All rights reserved.

is of metallic construction. Economics dictates carbon and low alloy steels as preferred first option materials, but they lack useful strength above
600 C, have relatively poor toughness below 0 C, and have poor corrosion resistance. Stainless steels and nickel alloys have strength and toughness capabilities across much wider temperature ranges, and wide ranging corrosion resistance, depending on the alloy. Copper, aluminum, and lead alloys have limited temperature capabilities, but find niche applications based on their specific corrosion resistances. Titanium, zirconium, and tantalum alloys are amongst the most corrosion-resistant metals available, though at a price, and are commonly used as lining materials on cheaper, steel substrates. Cobalt alloys are used principally for their wear resistances in the handling of solids, slurries, etc. Plastic materials have limited strength and temperature capabilities, but provide cost-effective corrosion resistances toward many fluids that are highly corrosive to metallic materials. Thermoplastic materials such as polyethylene, polypropylene, polyvinyl chloride) and the fluoropolymers (PVDF, FEP, PTFE), and thermosetting polyester, epoxy and furane resins reinforced with glass or carbon fibers (FRPs), find uses in relatively small, low-pressure vessels and piping systems, and in valves and pumps. Elastomeric materials, ranging from natural rubbers to relatively expensive, synthetic fluoroelastomers, are widely used as joint sealing materials. All the polymeric materials find uses as corrosion-resistant lining materials for steels. Ceramic materials have excellent temperature capabilities and good corrosion resistances, but are brittle, and thus unsuitable for pressure containment. In volume terms, their major use is as refractories for the thermal protection of steels. At lower temperatures, chemical stoneware; acid-resistant bricks; and specific oxide, carbide, and nitride ceramics have niche applications as corrosion-resistant linings and surfacings, in valves and pumps, and in seals and bearings. Silicon carbide and impervious graphite find use as heat exchange materials for particularly corrosive duties. Solid borosilicate glass is used for relatively small vessels and piping systems operating at low pressures, and 549

550

Corrosion in the Process Industries

proprietary glass formulations are used as corrosionresistant linings for steels.

the solution, and this will tend to stifle the corrosion process (in this situation the metal is described as being passive, and the protective corrosion product is called a passive film).

CORROSION CHEMISTRY Corrosion is fundamentally a chemical reaction between a metal and its environment.a As such it is a heterogeneous reaction between a fluid and a solid. At higher temperatures (when the environment is a gas rather than a liquid), the reaction is typically a direct reaction between oxygen and the metal to form the metal oxide. The oxide will form as a solid on the metal surface,b and oxidation will be controlled by the transport of oxygen and metal ions through the corrosion product. In fluids that allow conduction of ionic charge, such as solutions of salts in water or molten salts, the corrosion reactions may split into two parts (which may occur in different places): the anodic reaction that results in the conversion of metal to metal cations and produces surplus electrons in the metal, and the cathodic reaction that consumes electrons and hence balances the anodic reaction: Anodic reaction : Fe ! Fe2þ þ 2e

iron corrosion

Cathodic reaction : 2Hþ þ 2e ! H2

hydrogen evolution

Cathodic reaction : O2 þ 2H2 O þ 4e ! 4OH

oxygen reduction

Note that the two cathodic reactions are given—the first tends to be more important in acids and for reactive metals such as aluminum, while the second tends to be more important in neutral solutions and for less-reactive metals such as steel and copper alloys. An important aspect of corrosion in liquids is the nature of the corrosion product. If this takes the form of a soluble salt, then the metal will remain in contact with the liquid, and rapid corrosion can be expected (this is known as active corrosion). If the corrosion product is solid (often an oxide or hydroxide), then it will tend to form a barrier between the metal and

a

The term ‘‘corrosion’’ may also be applied to nonmetals, and these are touched on in this article, but the chemistry in this case will be different, and the wide variety of nonmetallic materials precludes detailed coverage here. b At higher temperatures and in the presence of salt deposits, the corrosion product may be liquid, in which case the behavior may be more like that of aqueous corrosion. At very high temperatures, the oxide may be volatile, in which case very rapid corrosion can occur.

CORROSION ENVIRONMENTS There are an essentially infinite number of corrosion environments, but there are a number of key classes in the process industries: Atmospheric: Atmospheric corrosion due to the combined effects of rain and the deposition of salt and other pollutants will affect most equipment. Corrosion occurs while the metal surface is wet, and is strongly influenced by the composition of deposits (such as sulfates from industrial atmospheres and chlorides from marine atmospheres). External corrosion of steel and stainless steel process equipment beneath thermal insulation and fireproofing is of particular concern. Gases: While most corrosion is electrochemical, corrosion in noncondensing gases is chemical, because there is no possibility for the transport of ionic charge in the environment, or of dissolution of the corrosion product. This leads to a rather different character for corrosion in gases, which will consist of the growth of scales of corrosion product. This requires the transport of oxygen and metal ions through the scale, and this generally requires a high temperature to occur at a significant rate. Waters: Water plays an important part in corrosion at lower temperatures. It acts as a solvent for salts that modify the corrosion process; it ionizes these salts, and thereby allows the passage of ionic charge; it may dissolve the corrosion products; it can dissolve a moderate amount of oxygen; and it can act as an oxidant itself with the evolution of hydrogen. The content of soluble salts, and hence the ionic conductivity, is important. Fresh water (e.g., unpolluted river water and tap water) has a low salt content, and in consequence, it is relatively noncorrosive toward carbon steel. This is especially true at moderate temperatures when oxygen is excluded from the water; many heating and cooling systems (including most domestic central heating systems) rely on deaeration of tap water (often by reaction with the steel in a closed system) to control corrosion. Similarly, the major water treatment for lower pressure boiler waters is concerned with the maintenance of a low salt content and the removal of oxygen. Even in the presence of oxygen, clean river water gives an acceptable corrosion rate of carbon steel (many steel-hulled boats for use only in freshwater are used without any corrosion protection). As the salt content increases, the water becomes more corrosive, especially in the presence of oxygen, and seawater is very corrosive to many metals.

Corrosion in the Process Industries

Acids: Acids tend to produce high corrosion rates, both because protective oxides become soluble in acids and the reduction of hydrogen ions becomes an effective cathodic reaction. There is a major distinction between oxidizing acids (notably nitric acid), which provide an alternative cathodic (oxidation) reaction and can corrode even noble metals such as copper and silver, and socalled reducing acids (such as hydrochloric and sulfuric, though the term ‘‘nonoxidizing acids’’ might be more accurate), which require an alternative reactant (such as dissolved oxygen) to corrode the noble metals, but which can corrode steels and more reactive alloys with hydrogen evolution. Note that oxidizing acids can cause passivation of iron and steels; so iron will corrode rapidly in concentrated hydrochloric acid, but it will passivate in concentrated nitric acid. Alkalis: Most metals are protected by a passive oxide in mildly alkaline solutions, but the protective oxide will redissolve in strong alkali to form oxy-anions of the metal, allowing corrosion to occur. For carbon steels, the region of corrosion in alkali is very limited, but it can lead to the serious problem of caustic stress corrosion cracking (SCC). Salts: In general, the nature of the anion present in dissolved salts is important for corrosion, largely as a result of the interaction of the anion with protective passive films. An important property of the anion present in salts is the pKa of the corresponding acid, as many localized corrosion processes involve the production of local acidic conditions at anodic regions; anions of weak acids will act as buffers and limit the drop in pH. Halides tend to be particularly aggressive, both because they have a low pKa and support the formation of strong acids, and because metal halides tend to be soluble. In contrast, carbonates and phosphates have a high pKa, and therefore restrict any drop in pH and tend to inhibit corrosion. The cation present in the salt can also affect corrosion, most commonly because of the precipitation of hydroxides at local cathodic regions (which tend to become more alkaline than the bulk solution). Thus, calcium and zinc cations tend to act as cathodic inhibitors for iron and steels in neutral solutions.

TYPES OF CORROSION Corrosion can produce a range of morphologies, some of which can be particularly damaging for a given amount of metal loss. Thus uniform corrosion, in which the metal loss is distributed over the entire metal surface, usually causes relatively slow, manageable loss in section thickness of components, whereas some localized forms of corrosion can lead to failure of components in a matter of months, weeks, or even days.

551

Uniform or General Corrosion Uniform corrosion usually occurs in fairly aggressive environments that attack the whole surface. Examples include carbon steel in seawater or acids, or aluminum alloys in strong alkali. The rate of metal loss is usually rather high, but, because it is distributed over the whole surface, the performance can usually be predicted, and managed with corrosion allowances, in most situations. Thus, sheet steel piling is often used in seawater without any corrosion protection, the corrosion rate of around 0.1 mm=yr,c coupled with the relatively thick steel sections, giving an acceptable life.

Galvanic Corrosion When two dissimilar metals are electrically connected together, such that electrons can flow from one metal to the other, it is probable that the anodic, metal dissolution reaction will concentrate on one metal, while the cathodic reaction occurs on both. This accelerates the corrosion of the anodic component. The acceleration will be particularly severe if the area of the cathodic metal is much larger than that of the anodic metal (Fig. 1). While the anodic metal will corrode more, the cathodic metal will normally corrode less, and one form of galvanic corrosion provides a method of corrosion control (know as cathodic protection), in which the anodic, corroding metal is provided deliberately in order to restrict the corrosion of the cathodic metal.

Dealloying or Selective Attack Dealloying occurs when one component of an alloy is lost preferentially. Thus, brass is an alloy of zinc (a rather active metal) and copper (a rather noble metal). Consequently, the zinc tends to be lost in preference to the copper. Often the copper will form a ‘‘seal’’ over the surface, preventing further corrosion, but if conditions do not allow this, then the corrosion can penetrate into the component, removing most of the zinc. The result is a porous copper component, which has little mechanical strength, and the problem is often discovered when the component fractures. Similarly, one component of a two-phase alloy can

c

The corrosion rate quoted is a typical average value. The rate varies according to depth, and a number of other factors, with the region around high tide normally having the highest corrosion rate. Recently, the phenomenon of accelerated low water corrosion has been identified, with corrosion rates of 1 mm=yr or more, and corrosion protection should be considered for all new structures.

C

552

Corrosion in the Process Industries

Fig. 1 Galvanic corrosion of a steel blank flange around a threaded brass coupling from a liquid fertilizer storage tank. (View this art in color at www.dekker.com.)

corrode preferentially, when the process is better called selective attack (Fig. 2). Differential Aeration Corrosion When there is a difference in oxygen content between two parts of a component, the part with the higher concentration will tend to become more alkaline (and may passivate), while the part with the lower concentration will tend to become more acidic (and will corrode more readily). The acceleration in corrosion rate due to differential aeration is relatively limited, because there must be a relatively large anodic area compared to the cathodic area in order to produce enough alkalinity to passivate the steel.

produces stronger acidification and reinforces the process. Pitting corrosion occurs on a free surface, often initiated by a local defect in the passive oxide, such as a surface sulfide inclusion (though in service, it is often initiated by a local crevice under a particle of dirt or similar). Once the pit has initiated, it is stabilized in essentially the same way as crevice corrosion. While they also occur in other passive alloys, crevice corrosion and pitting corrosion are particular problems with stainless steels, particularly in chloride-containing solutions. The main approach to control of these problems is in alloy composition, where increasing chromium, molybdenum, and nitrogen contents improve the resistance to pitting and crevice corrosion.

Flow Effects Crevice Corrosion and Pitting Corrosion These forms of corrosion are similar to differential aeration corrosion, in that an oxygen-free region becomes acidic by virtue of the net anodic reaction and consequently corrodes rapidly when coupled to a region in aerated solution. However, a key difference is that these forms of corrosion occur on alloys that are initially passive;d so there is no limit to the area of the passive, cathodic region. Thus, the severity of the attack may be much greater (Fig. 3). Crevice corrosion occurs when there is a narrow gap between two pieces of metal or a piece of metal and an insulator. The oxygen is consumed by the slow passive corrosion in the crevice, causing the crevice to become sufficiently acidic that the passivity breaks down and active corrosion starts. This then

d It can be argued that any form of corrosion that occurs in a crevice is crevice corrosion, and any form of corrosion that leads to small local penetration is pitting corrosion.

Solution flow typically enhances corrosion rates, by increasing the transport of dissolved oxygen to the metal surface, by increasing the rate of removal of protective corrosion products, and, in extreme cases, by physically removing the corrosion products or even metal (in the case of erosion by suspended particles or cavitation) (Fig. 4). In a few situations flow can be beneficial; thus for stainless steels in chloride solutions, flow can prevent the development of the acidification that is necessary for pitting and crevice corrosion.

Stress Corrosion Cracking The term SCC can be defined as ‘‘the initiation and propagation of cracks in a metallic component as a result of the combined influence of a mechanical stress and a specific corrosive environment’’. The stresses involved may be externally applied working stresses or internal residual stresses (often produced by deformation

Corrosion in the Process Industries

553

C

Fig. 2 Selective attack of the black ferrite phase in a stainless steel weld bead. (View this art in color at www.dekker.com.)

or welding during fabrication). Most alloys will suffer from SCC in some environments, usually ones that give a relatively low rate of general corrosion. Crack growth rates vary over a wide range, with typical times to failure ranging from hours to years (Fig. 5).

Corrosion Fatigue Metal fatigue is the process of crack initiation and growth due to the action of a fluctuating mechanical stress. Corrosion fatigue is simply metal fatigue that

Fig. 3 Crevice and pitting corrosion of a stainless steel autoclave head. Note the crevice corrosion underneath the bolts (now removed) and in the gap between the two parts that are still assembled, and the pitting corrosion on the free surface. This corrosion was probably caused by chloride derived from thermal insulation. (View this art in color at www. dekker.com.)

554

Corrosion in the Process Industries

Fig. 4 Flow-induced corrosion of a pump impeller. (View this art in color at www. dekker.com.)

is accelerated by the action of a corrosive environment (Fig. 6). The typical effect of corrosion is to reduce the time to failure compared to testing in air, especially at lower stresses and cyclic load frequencies, and to reduce or eliminate the fatigue limit.

body-centered cubic metals, such as ferritic steels, are most susceptible to hydrogen embrittlement, while face-centered cubic metals, such as austenitic stainless steels, are much less susceptible. Additionally, the susceptibility increases as the strength of the material increases (Fig. 7).

Hydrogen Effects

Hydrogen-induced cracking (HIC) and stress-oriented HIC

Hydrogen embrittlement Hydrogen will dissolve easily in metals, and, once in solution, it can cause brittle fracture. In general,

These processes involve crack formation by the precipitation of dissolved hydrogen onto lamellar nonmetallic inclusions. They are a particular problem

Fig. 5 External SCC of a stainless steel pipe after contact with wet, chloride-containing, thermal insulation.

Corrosion in the Process Industries

555

C

Fig. 6 Environmentally assisted cracking of a weld in a deaerator, caused by a combination of static residual and cyclic operating stresses, and revealed by magnetic particle inspection.

in equipment handling ‘sour’ hydrocarbon fluids that contain hydrogen sulfide (Fig. 8). Hydrogen attack This is a high temperature process relevant to equipment in petroleum refineries and petrochemical plants containing hydrogen. It is associated with the formation of methane by the reaction of hydrogen and carbon inside the steel. This results in decarburization, and the methane precipitates at the nonmetallic inclusions to produce cracking in a similar way to HIC.

Fretting Corrosion Fretting corrosion occurs when two metal surfaces are rubbed together, usually in a relatively noncorrosive environment, such as moist air. The rubbing removes the protective oxide, allowing further oxidation to occur, and the oxide produced acts as an abrasive to accelerate this process. Microbial Corrosion Corrosion of steels and stainless steels may be induced by the metabolic products of microbial growth and

Fig. 7 Cracks in a high strength steel relief valve spring. The cracks have been revealed using a dye that penetrates into the crack then seeps back out into a porous white coating. The cracks were caused by hydrogen embrittlement of the excessively hard steel, coupled with tensile residual stresses and the storage environment on a chemical plant (with an occasional trace of H2S). (View this art in color at www.dekker.com.)

556

Corrosion in the Process Industries

Fig. 8 Hydrogen-induced cracking—note the internal split produced by the precipitation of hydrogen gas onto flattened sulfide inclusions.

reproduction. The classic case is the corrosion of steel by sulfate reducing bacteria under anaerobic conditions in soils, or under deposits and tubercles in waters. Stainless steels are also vulnerable to pitting in waters containing iron and iron manganese bacteria, in the event that hydrotest or cleaning waters are retained in equipment for long periods prior to recommissioning.

MANAGEMENT OF CORROSION IN NEW EQUIPMENT Management of Corrosion by Design The corrosion of an alloy in an environment is influenced by chemistry, temperature, stress, geometry, and galvanic effects, which are exacerbated in process equipment by heat transfer and fluid flow. All these factors can be varied by design, and it follows that the propensity of process equipment to corrosion can be influenced strongly by design detail.

chloride-induced, external SCC of austenitic stainless steels under thermal insulation. Extremely high concentration factors are possible under boiling heat transfer in crevices or under deposits, or under film boiling conditions at high heat flux regions, as in water side, on-load corrosion in process and utility boilers. There are practical design measures, which mitigate the risks of developing corrosive microenvironments in heat transfer equipment, e.g., preferring tubeside water and horizontal orientation, or venting of top tubesheets in vertical, stainless steel water coolers, and the attenuation of heat fluxes in highly rated boilers by tube inlet ferrules. Temperature definition Although normally straightforward for bulk process streams, it is important to select materials for skin rather than bulk temperatures in heat transfer equipment, and to evaluate the effects of mixing exotherms on local bulk fluid temperatures, such as that may occur in acid addition=dilution equipment.

Chemistry definition Control of stress and stress concentration The commonest cause of unpredicted corrosion problems is the failure to define, accurately, the chemistry of process streams, including startup, shutdown, and transient conditions, or to anticipate changes in chemistry at specific locations in equipment. The corrosivity of a process stream is often determined by its minor components, e.g., the presence of 10s–100s ppm chlorides can promote localized corrosion of stainless steels and other passive alloys. It is important that minor components are defined, quantified, and evaluated at the design stage, including their possible local concentration such as in distillation and separation equipment. Heat transfer is a particular promoter of the development of local chemistries that are very different from the bulk fluid chemistry, in particular, where phase changes occur. Early condensates from acid gas=vapor streams can be very concentrated and corrosive relative to bulk condensates as in the cases of carbonic acid from steam, sulfuric acid from flue gases, and hydrochloric acid from refinery overhead streams. Initially, benign condensates or cooling fluids can concentrate due to intermittent contact with surfaces hot enough to promote concentration or dryout, as in the

Design codes for process equipment specify maximum allowable stresses for materials, the basis for which varies, depending on the type of equipment and the class of material. Corrosion allowances are specified commonly in design to ensure that pressurized components remain thick enough throughout their service lives to maintain membrane stresses below the design maximum allowable levels. In practice, process equipment commonly contains regions with substantially higher stresses. The codes sometimes allow higher stresses in local regions of the equipment, design stresses can be concentrated by defects in materials, and fabrication processes such as cold working and welding introduce ‘‘residual’’ stresses of the order of the yield stress. Environmentally assisted cracking (EAC) problems such as SCC and corrosion fatigue commonly initiate at such high stress regions. Clearly, locally high stresses permitted by the design code are ‘‘given’’, and if they present a risk of EAC to a specific material, then an alternative must be specified. However, stress concentration effects arising from defects can be mitigated in design

Corrosion in the Process Industries

by specifying detail such as transitional radii at changes in cross section, machining=grinding of weld toes, quality of surface finish, etc. Residual stresses can be reduced in design by specifying processes such as thermal stress relief, temper bead welding, peening, etc. Control of geometry The major effects of geometry relate to vulnerability of process equipment to crevice corrosion and erosion– corrosion. Crevice corrosion in process equipment is associated most commonly with flanged joints with gaskets, heat exchanger joints, and weld defects. Screwed and socket welding flanges present crevices to the fluid, whereas slip-on welding and welding-neck flanges avoid crevices. Care is needed in the specification and sizing of gaskets to avoid crevices. The ubiquitous tube=tubeplate joint in tubular heat exchangers is inevitably at potential risk of crevice corrosion, particularly where high heat transfer rates into the crevice can superheat the crevice relative to the bulk fluid. Consideration might be given in design to locating fluids that might promote crevice corrosion on the tubeside of the exchanger, and=or to procedures for closing off crevices such as tubeplate back face bore welding. Plate and other compact heat exchangers, with very large areas of compression or welded joints, can be particularly vulnerable to crevice corrosion, and consideration might be given to upgrading the material to mitigate risks. Crevice corrosion can occur at welds where lack of sidewall fusion or root penetration creates a crevice, and appropriate fabrication and inspection procedures must be implemented to prevent such defects. Erosion–corrosion is avoided in design through the use of limiting velocities for fluids that have been established by testing and=or experience. Beyond that, it is important to design and fabricate fluid systems to minimize flow disturbance, by using where appropriate long radius bends=elbows and gradual changes in cross section, ensuring that flanges are aligned correctly, avoiding excessive weld root protrusion in fabrication, etc. It is also important to select, size, and locate components that disturb flows, such as probes, valves, and orifice plates, to minimize local and downstream turbulence. Thus, it is a good design practice to locate control valves and orifice plates away from bends, etc. Control of galvanic effects Inevitably, process systems are constructed in a variety of materials, and the potential for galvanic interaction needs to be addressed in the design. Although the

557

risk of galvanic corrosion is confined mostly to mixed metal systems, carbon=graphite (widely used in heat exchange and jointing systems) and silicon carbide (mostly a heat exchanger material) are sufficiently electrically conducting for their potential contributions to galvanic corrosion to require evaluation. Available galvanic series of materials have been obtained in specific environments, mostly seawater, and are of limited use in galvanic corrosion risks in chemical process fluids that may have very different conductivities, tendencies to promote passivity on specific materials, etc. However, as a general rule, it is a good design practice to avoid material combinations that are widely separated in such series, and in particular, to avoid unfavorable area ratios involving small areas of ‘‘active,’’ anodic metal coupled to large areas of ‘‘noble,’’ cathodic material, e.g., more active fasteners and weld filler metals. Galvanic effects may be mitigated through coatings. Preferably, both couple members are coated, but if only one can be coated, it should be the noble, cathodic member. Equipment items judged to be vulnerable can, in principle, be isolated electrically using insulated flange joints, but in practice this may be frustrated by remote earthing via steel supports, pipe hangers, etc. Galvanic effects at material breaks in piping systems can be mitigated by physical separation of couple members using insulated spool pieces made from FRP, or coated, more noble couple member. Galvanic effects can be overridden by applying cathodic protection to the couple, as in the protection of the water boxes of seawater coolers by sacrificial anodes.

Management of Corrosion by Materials Selection Materials selection for corrosion resistance is as reliable as the information upon which it is based. Corrosion information derives from two sources— experience or test results. Experience is sometimes codified in the form of industry guides that address generic, industry-wide corrosion problems. More commonly, experience takes the form of corporate experience of materials performance in specific applications, obtained from the operation and inspection of existing equipment. This information may be accessible across corporate boundaries through informal industry networks, depending on its commercial significance, and electronic networks, such as CORROS-L (see www.jiscmail.ac.uk) and the NACE Corrosion Network (see www.nace. org), which are increasingly important. In the absence of relevant experience, materials selection decisions have to be based on the results of corrosion tests, which range from relatively simple,

C

558

laboratory procedures, through semitechnical=pilot plant tests, to operating plant tests and trials. Corrosion test results are widely available in the public domain, usually as tabulated corrosion rate data for specific environments, from materials suppliers and professional corrosion organisations. Alternatively, new corrosion tests can be undertaken to inform materials selection decisions for applications where no relevant experience or test data is available. Corrosion tests have inevitable limitations in their capacities to mimic actual service conditions of equipment. Standard, ambient pressure, immersion test procedures, with intermittent fluid refreshment, are available for both metallic and nonmetallic materials, but are limited to the ambient pressure boiling point of the fluid, and provide limited scope to simulate the effects of stress, geometry, heat transfer, and fluid flow. Such test procedures can be conducted at plant pressures and temperatures in autoclaves, and can be upgraded to focus on specific factors such as fluid flow and heat transfer. Even so a laboratory test, however elaborate, is a poor substitute for a test in the plant itself. The key advantage of plant testing is direct exposure to the service fluid, thereby avoiding the risks of missing undefined, significant fluid constituents, or time-dependent chemistry changes, in unrefreshed laboratory liquors. Test coupons can be installed in plants in racks, or bolted directly onto trays, baffles, flanges, slip rings, etc, or onto retractable coupon holders. Care is needed to locate test coupons in the right places, particularly where phase and=or chemistry changes occur such as in separation and distillation equipment. Test plant components such as spool pieces, impellers, heat exchanger tubes, agitator paddles, column internals, etc., provide perfect simulation of all potential contributory factors to corrosion, as long as the attendant risks of premature failure can be accommodated. It is not uncommon for new processes to encounter significant corrosion problems during the early stages of commercial operation, reflecting the poor reliability of the information used to select materials. Public domain corrosion tables are useful guides to corrosion performance but, not least because of the common absence of test details, are relatively unreliable information sources. Laboratory and=or plant coupon testing are more reliable, but the most reliable information derives from field performance data from operating plant inspections and=or long-term plant component tests. It is also important that materials selection is an integral part of process design from the earliest, chemistry definition stages through process design iterations, and that any laboratory, semitechnical, pilot plant, and sidestream activities are used to evaluate candidate materials as an integral part of process development.

Corrosion in the Process Industries

Management of Corrosion by Inhibition Although its scientific basis is beyond the scope of this contribution, corrosion control by the addition of corrosion inhibiting chemicals is important in several areas of chemical processing. The major application is in the management of corrosion, principally of steel, in cooling water and steam raising systems. Well-established practices are available commercially, and the great majority of corrosion problems in process industry water systems arise from the failure to specify at the design stage, and=or manage in service, appropriate water treatment practices. Inhibition is practised in several other areas of chemical processing, including the control of overheads corrosion of steel and other alloy heat exchangers by condensing hydrochloric acid in oil refining, and the control of wet carbon dioxide corrosion of steel vessels and pipelines in oil=gas recovery and acid gas stripping.

Management of Corrosion by Coatings and Linings A wide variety of coatings and lings are available for the protection of steel in chemical process equipment, and they often provide cost-effective alternatives to more expensive, corrosion-resistant alloys. Chemical plant atmospheres can promote particularly rapid atmospheric corrosion. For structural steelwork, galvanized zinc and=or paint coatings are usually specified. As alternatives to galvanizing, a variety of zinc- or zinc phosphate-containing primers are specified routinely, overcoated with compatible paint systems, including micaceous iron oxide epoxy, acrylic polyurethanes, silicones, etc. Paint coatings are also used for the prevention of atmospheric corrosion of process equipment, although care is needed in the use of metallic zinc, because of concerns about polarity reversal on warm=hot surfaces. A particular concern is the control of external corrosion beneath thermal insulation and fireproofing, for which relatively highquality paint systems are required for the protection of steel, while thin aluminum foil is an alternative to paint coatings for reducing the risk of external SCC of stainless steel surfaces. High performance paint coatings can be used for the internal corrosion protection of steel process equipment such as storage tanks, pipelines, vessels, and even heat exchangers, although a detailed account of the options is beyond the scope of this contribution. In environments where the corrosion rate of steel would be >
0.5 mm=yr if exposed, thicker linings are preferred to relatively thin paint coatings. A wide range of thermoplastic, FRP, rubber, and glass linings are available, depending on the application.

Corrosion in the Process Industries

Corrosion-resistant metals and alloys, including stainless steels, nickel alloys, titanium, zirconium, and tantalum can be applied as linings or claddings to cheaper steel substrates. Most (>90%) are applied by roll bonding, but weld overlaying and explosive bonding are also used, as appropriate. Particularly expensive metals, such as tantalum, can be used as very thin (
0.5 mm) loose linings.

Management of Corrosion by Electrochemical Intervention Wet corrosion of metals being an electrochemical process, it can be controlled, in principle, by electrochemical intervention, and commercial practices are available for specific applications. The most widely used technique is the cathodic protection of steel in soils and waters, which involves lowering the potential of steel to levels at which the rate of corrosion is negligible using distributed sacrificial anodes of zinc, magnesium or aluminum alloys, or inert impressed current anodes driven by an external power supply. Cathodic protection is used commonly in association with coatings, to confine current demand to defect sites in an otherwise protective coating. It is used in the process industries for the external protection of buried= immersed pipelines and infrastructure, and occasionally the internal control of corrosion, including galvanic corrosion, in tubular water coolers. Anodic protection involves the promotion of passivity on steel or stainless steel equipment by raising and controlling the potential into the passive range in the process fluid, using auxiliary cathodes, or coupling to more noble materials, or by the addition of oxidizing species to the fluid (a special case of inhibition). Anodic protection is used mostly to extend the useful range of stainless steels in reducing sulfuric and phosphoric acids, where it finds practical usage in heat exchangers and storage tanks. It is used, not merely to control corrosion at acceptable rates, but also to prevent unacceptable contamination of process fluids with corrosion products.

MANAGEMENT OF CORROSION IN OPERATING EQUIPMENT The corrosion performance of equipment in service is determined largely by decisions taken at the design stage. However, design decisions often anticipate appropriate management and maintenance processes in service, and if these are not delivered, then problems may ensue. For example, if the plant is operated, knowingly or otherwise, outside the design envelope,

559

the risk of corrosion will increase, and it can only be mitigated by appropriate review of the original materials of construction under the revised process conditions, and upgrading if required. If corrosion is to be controlled by corrosion inhibition, or electrochemical intervention, then corrosion risks in service will be determined by the extent to which these processes are managed and maintained according to the design intent. In modern practice, corrosion risks in operating chemical process equipment are managed within the broader context of loss-of-containment risk mitigation, based mainly on risk-based inspection (RBI). The principles of RBI are beyond the scope of this contribution, but are based essentially on risk mitigation by optimizing inspection frequency and coverage, using appropriate tools, techniques, procedures, and practices. The results of these activities define whether the design intent is being realized in practice, or whether further risk mitigation is required in the forms of equipment replacement or repair, or equipment modification and redesign (increased corrosion allowances, changes in materials, addition of coatings=linings, addition of corrosion inhibitors, etc). They also define the need for additional monitoring of corrosion rates through appropriate inspection and=or monitoring techniques. The inspection of process equipment to detect and size corrosion damage is beyond the scope of this contribution, and the following summary is confined to techniques that can be used for on-line corrosion monitoring. Techniques such as radiography and ultrasonics can be used externally to track corrosion distribution, wall thickness, and defect size through periodic measurements, but there are limitations relating to sensitivity and conditions of measurement. Periodic or continuous fluid analysis for the presence= concentrations of corrosion products can provide useful information, but the locations and distributions of the corrosion from which the products arise can only be inferred. Otherwise, the available techniques for corrosion monitoring fall into three broad categories. Firstly, there are techniques that provide information on accumulated metal loss, from which average corrosion rates can be obtained by periodic measurements. The simplest approach involves the use of coupons, which can be retrieved at intervals from critical locations on the plant to provide average corrosion rates from weight or dimensional change. Alternatively, probes with elements comprising strips, wires, or thin-walled tubes of the test material can be inserted at critical locations, and the changes in their dimensions monitored by changes in their electrical resistance, which can be measured readily and intermittently online, the latest commercial technology being sensitive to nanometer changes in dimension. In a third technique, hydrogen that has been generated during

C

560

corrosion of the internal surface permeates the wall of a pipe or vessel and is collected in a chamber on the external surface, where the change of pressure can be related to metal loss on the internal surface. A second category of techniques based on electrochemistry, provide instantaneous measurements of corrosion rate=state. Linear polarisation resistance and electrochemical noise measurements require the introduction of probes with varying numbers of elements of test material. Measurements are made across individual pairs of elements, involving the detection of small changes in potential or current, which are generated spontaneously, or in response to the application of small potential or current perturbations. Corrosion rates, and information on the propensity to localized corrosion, can be generated from the measurements, the principles behind which are beyond the scope of this contribution. The corrosion potentials of plant components can be measured online, and although they contain no information as to corrosion rate, they can provide useful information as to whether components are passive or not, or close to potential ranges where there may be a vulnerability to localized corrosion. Finally, in an electrochemical variation to the technique referred to earlier, hydrogen that permeates through a pipe or vessel wall can be oxidized electrochemically in a cell strapped onto the pipe exterior, thereby providing an indirect measurement of corrosion rate on the interior surface. The final two techniques allow remote, direct monitoring of metal loss from plant components. Thin layer activation (TLA) uses isotopes created in a surface layer of the component by proton irradiation. Corrosion is monitored through periodic measurements of the loss in activity as the surface corrodes, which can be made outside the pressure envelope. The electric field signature method (FSM) measures changes in the electric field pattern in a metallic structure induced by corrosion, from voltage measurements amongst pairs of an array of pins distributed on the external surface. TLA and FSM are sensitive to around 1% of layer depth and wall thickness, respectively. They are relatively expensive techniques, but find use in the diagnosis and control of critical and expensive corrosion problems on operating plant.

CONCLUSIONS Corrosion will always remain a significant challenge to the process industries. Process technologies will continue to develop into areas where there is no relevant experience of materials performance. There are inevitable risks in the use of corrosion testing to predict materials performance. With the exception of

Corrosion in the Process Industries

a relatively few generic problems in the nuclear and oil=gas processing fields, corrosion models do not allow reliable prediction of materials performance. However, for the majority of process equipment, corrosion prevention is a relatively mature technology, based on major progress in the understanding of the mechanisms and control of corrosion phenomena over the last 40–50 yr. Much of this progress has been made through industry initiatives, and the accumulated knowledge and experience can be sourced in the activities and publications of bodies such as the National Association of Corrosion Engineers (NACE, http:== www.nace.org), American Petroleum Institute (API, http:==www.api.org), and the Materials Technology Institute (MTI, http:==www.mti-link.org) in the USA, and the European Federation of Corrosion (EFC, http:==www.efcweb.org) and DECHEMA (http:== www.dechema.de) in Europe. Of particular relevance are the NACE Special Technical Groups (STGs) on Water Treatment (STG11), Petroleum Refining and Gas Processing (STG34), Process Industry (STG36– 39) and Energy Generation (STG41), and the EFC Working Parties (WPs) on Corrosion and Scale Inhibition (WP1), Corrosion by Hot Gases and Corrosion Products (WP3), Nuclear Corrosion (WP4) and Corrosion in the Refinery Industry (WP15). Most process=chemical chemical engineers receive little or no training in corrosion and its management in their graduate education and early formative years. When corrosion issues arise in the design or the operation of process equipment, it is advisable to consult corrosion specialists on all but very routine matters, and the earlier the involvement of the specialist, the better.

BIBLIOGRAPHY Agree, M. Betz Handbook of Industrial Water Conditioning, 9th Ed.; Betz Laboratories Inc.: Trevose, 1991. Behrens, D. DECHEMA Corrosion Handbook, 12 Volumes, 4th Ed.; John Wiley and Sons, 1991. (Also available on CD-ROM: Elsevier Science; 2001 onwards.). Craig, D.B.; Anderson, D.S. Handbook of Corrosion Data, 2nd Ed.; ASM International: Ohio, 1995. Dillon, C.P. Corrosion Control in the Chemical Process Industries, 2nd Ed.; Materials Technology Institute: St Louis, 1994. Francis, R. Galvanic Corrosion: A Practical Guide for Engineers; NACE International: Houston, 2000. Gaverick, L. Corrosion in the Petrochemical Industry; ASM International: Ohio, 1994. Gravener, D.L. Corrosion Data Survey—Metals Edition, 6th Ed.; NACE International: Houston, 1985.

Corrosion in the Process Industries

Guides to Corrosion Control, Dept. of Industry: London. Available in print or online from the U.K. National Corrosion Service. (http:==www. npl.co.uk=ncs=). Hamner, N.E. Corrosion Data Survey—Nonmetals Edition, 5th Ed.; NACE International: Houston, 1975. Holman, R. Corrosion and Associated Costs in the U.K. Chemicals and Petrochemicals Sector; Paint Research Association: London, 2002 (http:== www.pra.org.uk=projects=costofcorrosion.htm). Kemmer, F.N. The Nalco Water Handbook, 2nd Ed.; McGraw Hill Inc.: New York, 1989.

561

Lai, G.Y. High Temperature Corrosion of Engineering Alloys; ASM International: Ohio, 1990. Moniz, B.J.; Pollack, W.I. Process Industries Corrosion—The Theory and Practice; NACE International: Houston, 1986. Revie, R.W. Uhlig’s Corrosion Handbook, 2nd Ed.; John Wiley and Sons Inc.: New York, 2000. Schweitzer, P.A. Corrosion Resistance Tables, 5th Ed.; Marcel Dekker Inc.: New York, 2004. White, R.A. Materials Selection for Petroleum Refineries and Gathering Systems; NACE International: Houston, 1998.

C

Critical Phase Behavior C J. Richard Elliott, Jr. Department of Chemical Engineering, University of Akron, Akron, Ohio, U.S.A.

INTRODUCTION The critical pressure is defined as the pressure above which no vapor can exist. The critical temperature is defined as the temperature above which no liquid can exist. The supercritical fluid (SCF) region is, therefore, the region above both the critical temperature and the critical pressure. The subject of critical phase behavior can be perplexing to the uninitiated and painfully obvious to the fully indoctrinated. This entry is directed primarily at the former, especially those who have an appreciation for general thermodynamics and chemical processing but lack exposure to the nuances of the critical region. We also include a broad survey of patents and technology that impact chemical processing and relate to critical phase behavior.

BACKGROUND For pure fluids, the critical region is complicated by behavior that cannot be easily correlated. The most obvious instance is the curvature of the temperature– density coexistence curve. Classical equations tend to overestimate the critical point when fit to data outside the critical region. In other words, the true curvature is flatter than the curvature of a classical equation. Mathematically, (rL
rV)  (Tc
T)0.325 instead of the classical curvature given by (rL
rV)  (Tc
T) 0.5. Remedies for this problem generally take the form of either scaling crossover equations[1] or multiparameter equations.[2] Extensions to mixtures further complicate the issue. Critical scaling tends to be a specialized subject with no single approach being recognized as a conventional standard. Fortunately, the qualitative behavior from classical equations is generally correct once allowance is made for deviations very near the critical point. Because most applications of critical phase behavior are related to mixtures, the focus is on mixtures in this entry. The critical region becomes very interesting when one considers the behavior of mixtures. At its most basic level, the critical point is the point at which the vapor and the liquid become indistinguishable at fixed overall composition. The path to the critical point follows an indirect route when the composition is rich in a supercritical component like CO2 and leans in a Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120007719 Copyright # 2006 by Taylor & Francis. All rights reserved.

heavy component like hexadecane. Above the critical temperature, one may easily find a dew point at low pressure, and an increasing fraction of liquid as one raises the pressure from that point. But increasing the pressure at fixed temperature may result in the fraction of liquid decreasing before one reachs the bubble point. How can this be possible? Raising the pressure at fixed temperature makes the vapor phase denser, making it a better ‘‘solvent.’’ The vapor phase then dissolves more of the heavy component. Ultimately, one reachs a point where all but one drop is vapor. With this preparation, the reader can appreciate the key elements of a subject that can be challenging. The critical region is a space where phases are barely distinguishable; hence subtle changes make large differences. The focus in this entry is on critical phase behavior as it pertains to extraction and separation. We omit coverage of supercritical reactions because the phase behavior is qualitatively similar with or without reaction. Reactions merely alter the compositions. The impact on reactions of high compressibility in the critical region is of special interest, however, and initial study can be facilitated through the compilation of Noyori.[3]

BASICS OF CRITICAL PHASE BEHAVIOR Fig. 1 provides the most logical starting point for any introduction to critical phase behavior. The solid curve represents the distinct boundary between the vapor and the liquid phases along the vapor pressure curve. The dashed lines, on the other hand, represent regions of the phase diagram that have been defined by convention for purposes of general reference. Physical properties like density, viscosity, and diffusivity vary continuously across these imaginary boundaries, but are generally similar within the designated regions. We focus especially on the properties near the critical point and in the SCF region because the properties in that region are sensitive to variables within the engineer’s control, not just temperature and pressure, but also composition. This sensitivity can provide engineers with the ability to tune the phase behavior. If one includes liquid–liquid and micellar critical behavior within the scope of this subject, advanced topics like nanoscale morphology can be contemplated as well as more conventional topics like extraction. 563

564

Critical Phase Behavior

150 125 SCF P (bar)

100 75

Liquid

50 Gas

25 Vapor 0

200 225 250 275 300 325 350 T (K) Fig. 1 Phase diagram for pure CO2.

The key is to recognize the engineering possibilities when the thermodynamics are indefinite. Thermodynamics can be difficult and critical phase thermodynamics excruciating, but opportunities abound and much can be accomplished with a simple, general understanding of what is occurring.

Phase Diagrams and Classifications of Phase Behavior Despite the many opportunities in the critical region, there are also constraints. Some of the most striking are the discontinuities that occur in the critical loci of mixtures. At first glance, one might assume that the critical points of mixtures should vary continuously from one pure component to the other as the composition varies. That does happen occasionally, but rarely in applications of interest. In mixtures like CO2 and polystyrene, the CO2 phase can hold only a limited amount of polymer before polymer-rich phase begins to precipitate. One might suggest increasing the pressure to increase the density and the carrying power of the CO2. But then a second liquid phase is more likely to form than a truly SCF phase. Or, at temperatures around 50–80 C, vapor–solid equilibrium may predominate. At the polymer-rich end of the critical locus, the critical temperature is extremely high. A small amount of CO2 can be dissolved under these conditions, but the upper critical locus diverges toward very high pressures instead of approaching the critical point of CO2 as the temperature is reduced. Such problems with polymers may seem relatively obvious in retrospect, but similar problems occur with lower molecular

weight components like vegetable oil. Hydrogenation of vegetable oil is a common application, but one should consider the difference in critical temperatures between hydrogen (33 K) and vegetable oil (>600 K). Clearly, we must carefully consider the constraints that arise when we combine components in the critical region. Fortunately, van Konynenburg and Scott[4] have analyzed these constraints in a very general manner and organized the analysis into a classification scheme like Aristotle’s genus and species. Their approach was remarkably simple but effective. They studied the van der Waals equation of state for mixtures with a range of choices for the parameters a and b appearing in the equation. Qualitatively, these parameters are capable of describing the critical phase behavior of nearly all binary mixtures. Illustrations of the phase diagram types are given in Fig. 2.[5] Note that type VI behavior is not mentioned by van Konynenburg and Scott. However, type VI behavior can be exhibited by the van der Waals equation.[6] This behavior appears to be exclusive to aqueous systems.[7] Other type variations have been suggested over the years,[8,9] but the current consensus holds that these six types are sufficient. Types II and III are the most commonly encountered in SCF applications. Classes I and V represent mixtures for which HE < 0, where HE is the excess heat of mixing. Generally, HE < 0 when solvation interactions occur in the absence of association, as in the acetone þ chloroform system. Strongly solvating mixtures like these are relatively uncommon. These diagrams illustrate the trends of the critical loci as the composition changes. Critical loci roughly correspond to the upper limit of temperature and pressure for a phase envelope at fixed composition. Fig. 3 illustrates in three dimensions the compositions leading to each critical point. Above the critical locus, we would expect to see only a single phase. The slashes along the critical locus indicate the side on which the phase envelopes open up into two phases. In types III–V, the critical locus is interrupted by a vapor– liquid critical end point (VLCEP). At the VLCEP, the liquid phase splits into two liquid phases. An upper critical end point (UCEP) is similar, but the vapor merges with one of the liquid phases. In types IV and V, the lower critical end point (LCEP) also plays a significant role. The LCEP is a condition at which a liquid phase merges with a vapor phase. The resulting liquid–liquid (LL) condition gradually evolves along the critical locus into a vapor–liquid (VL) condition as the two tangent points on the Gibbs energy curve shift toward the VL critical point of pure component 2. The work of van Konynenburg and Scott stands out as a classic throughout all of thermodynamics. A basic understanding can be achieved by noting how phase behavior depends on the strength of the molecular

Critical Phase Behavior

565

HE (T=0) < 0

HE (T=0) > 0 II

ll A

P

B

C

I

A

P

B

U llv T

T IV

V

ll A

P

K llv

A

P

llv L

L

U

K

B

B

llv T

T

T1

T1

III K

A

P

VI A

P

U B

llv

L T

T2

T3 T4

Component Critical Point Binary Critical Endpoint Component lv Binary llv

B

llv

T

T1

Binary Critical Locus U Upper Critical Endpoint (UCEP) L Lower Critical Endpoint (LCEP) K Vapor-Liquid Critical Endpoint

interactions. Considering the quadratic mixing rule: a ¼ x21 a11 þ 2x1 x2 a12 þ x22 a22 ; with a12 ¼ ð1
kij Þ ða11 a22 Þ1=2 we may observe that molecular attractions are weaker as kij becomes more positive. Strong molecular attractions are indicated by negative values for kij. van Konynenberg and Scott generalized their results by defining two dimensionless quantities describing the inherent difference between the two pure components, z  ða11 =b211
a22 =b222 Þ=ða11 =b211 þ a22 =b222 Þ, and the strength of their binary interaction by L  ða11 = b211
2ða12 =b1 b2 Þ þ a22 =b222 Þ=ða11 =b211 þ a22 =b222 Þ. By convention, the compound with the lower value of Tc is taken as component 1. Hence, a large value of z indicates that the components are very different in volatility. Note that L ¼ 0 when a12 ¼ (a11 þ a22)=2, i.e., the arithmetic mean. A large positive value of L generally indicates weak binary attractions, but a better characterization of binary interaction is indicated by the curve for kij ¼ 0. In the absence of any other

Fig. 2 Progression of binary phase behavior with increasing molecular asymmetry according to van Konynenburg and Scott. Arrows denote progressions of phase behavior expected by theory. Experimental progressions frequently differ. (From Ref.[77].)

guideline, the best approach is to follow the kij ¼ 0 line when assessing the critical phase behavior for any given mixture. The transition from I to II to III occurs as the binary interactions become weaker. This is especially clear when the pure components are similar in attractive interactions (z  0–0.4). Basically, the weakening binary attractions lead to less stable solutions. At high values for z, the differences between the pure components are so large that slight weakness in the binary attraction destabilizes the liquid phase. Indeed, types III–V all exhibit two liquid phases. This is surprising for the type V case, because the binary attractions are quite strong. It is tempting to ascribe the trend in z to a difference in sizes of the molecules, because the larger molecules tend to have larger values for a. On the other hand, Fig. 4 assumes that the molecules have equal size; hence, there must be some other explanation. To clarify,

566

Critical Phase Behavior

1 0.9 0.8

III

0.7

P B

K

A

kij = 0

0.6 0.5 0.4

III

0.3 x B , yB

0.2 T

V

T3

T2

T4

II

0.1 IV

0 –0.1 –0.2 A

P

IV

K B

–0.3 V

I

–0.4 –0.5

L

U

–0.6 –0.7 0

x B , yB

0.2

0.4

0.6

0.8

1

ζ T

T1

Fig. 3 Type III and IV phase behavior illustrated in threedimensional diagrams. Symbols and labels are the same as Fig. 2. (From Ref.[77].)

Elliott and Lira[5] show that a=b2  d2, where d is the solubility parameter. For large molecules, the solubility parameter varies little with respect to molecular weight. Thus, increases in z correspond primarily to increases in cohesive energy density, not molecular size. Taking pentane as an example of component 1, we can estimate z from solubility parameters to be 0.16, 0.27, 0.33, 0.40, 0.52, 0.62, 0.84 for cyclohexane, benzene, acetone, n-hexanol, ethanol, methanol, and water. This provides an idea of the range of chemical functionalities addressed in Fig. 4. The van der Waals equation was an excellent choice for van Konynenburg and Scott, but it does have its limitations. By choosing the van der Waals equation, the fundamental model was greatly simplified. Simplifying as much as possible is highly advisable given such a complex problem to begin with. Unfortunately, this choice results in a certain degree of ambiguity when it comes to classifying the phase behavior for a given mixture of interest. Because the van der Waals equation gives only qualitative accuracy, reliable values of the binary interaction parameters are rarely known

Fig. 4 van der Waals master diagram for equal-sized molecules. (From Ref.[4].)

for mixtures of interest. With these limitations, the accepted practice is to classify the phase behavior according to the experimental observations rather than relying on a strict interpretation of the van der Waals equation. For example, cyclohexane and benzene are best represented by kij > 0, but they are not believed to exhibit LL behavior. Hence, they are classified as type I. One might assume that reproducing the van Konynenburg and Scott analysis for any new equation should be straightforward given the advances in computers since the work was originally performed in 1968. Figs. 5 and 6 shed light on why this is not the case. These figures show how subtle changes in the Gibbs energy lead to substantial changes in the phase behavior. In some cases, the important values of Gibbs energy look like little more than slips of the pen. In Fig. 5, a bend in the Gibbs energy at low x2 indicates the existence of a vapor root, but the tangent line connecting this root to the liquid lies above the tangent line connecting the two liquid roots. As Gibbs energy must be minimized, the tangent line between the two liquid roots is favored if a sufficient amount of component 2 is present. Otherwise, the VL region is the most relevant. In Fig. 6, all three roots lie on a

Critical Phase Behavior

567

A

B

0.05

C

0.00

0.005

–0.05

0.000 –0.005 –0.010

–0.15

∆Gmix /RT

∆Gmix /RT

–0.10

–0.20 –0.25

–0.015 –0.020 –0.025 –0.030 –0.035

–0.30

–0.040 –0.35

–0.045 –0.050

–0.40 0

0.2

0.4

0.6

0.8

1

0

0.01

x2

0.02

0.03

x2

Fig. 5 Gibbs energy of a system inside the VLL region of a type IV or V system showing the LL tangent (A) and the VL tangent (B) at the same conditions of Tr ¼ 1.005, Pr ¼ 1.0. (View this art in color at www.dekker.com.)

single tangent line, indicating that this condition lies exactly on the liquid–liquid–vapor (LLV) line. Noting that Figs. 5 and 6 represent the same pressure at increasing temperatures, we may wonder what happens when the temperature is further increased at constant pressure. Basically, the middle root recedes from tangent line and we are left with equilibrium between the vapor

and lower liquid. Note that van Konynenburg and Scott show this same transition analysis over a wider range of conditions, but their figures are highly idealized, masking the subtlety of the changes in Gibbs energy. Figs. 5 and 6 illustrate instances of type IV or V behavior near the high-pressure LLV line. This LLV line is a short segment, in contrast to the low-pressure

0.05 0.00

0.005

–0.05

0.000 –0.005

–0.10 ∆Gmix /RT

∆Gmix /RT

–0.010 –0.15 –0.20 –0.25

–0.015 –0.020 –0.025 –0.030

–0.30 –0.035 –0.35

–0.040 –0.045

–0.40 0

0.2

0.4

x2

0.6

0.8

1

0

0.01

x2

0.02

0.03

Fig. 6 Gibbs energy of a type IV or V system showing the LLV three-point tangent at Tr ¼ 1.010, Pr ¼ 1.0. (View this art in color at www.dekker.com.)

568

1 0.9 III 0.8 IV

0.7 0.6 ξ

LLV line in types II–IV. The low-pressure LLV line is relatively easy to identify computationally, because the liquid binodal region increases in size as one lowers the temperature. Therefore, it is simple to find the LLV region at a low temperature, then steadily raise the temperature using the previous result as the initial guess for the next. Initially, locating the high-pressure LLV line is more difficult because it only exists over a narrow range of conditions. Kolafa et al.[10] have developed a modern program for mapping global phase diagrams. By incorporating hydrogen bonding into their equation of state, they were able to easily reproduce the qualitative features of closed loop (Type VI) diagrams. Finally, we should address the trend as the size ratio varies. The most substantial alteration is the expansion of type IV behavior. We can estimate the result at kij ¼ 0 by considering the critical loci (SRK) as computed by the Soave–Redlich–Kwong[11] equation of state. Soave’s SRK equation was shown to accurately correlate the critical loci of a large number of systems.[12] Considering only the behavior along kij ¼ 0, we obtain a plane with z on the abscissa and x on the ordinate, where x  (b2
b1)=(b2 þ b1). In this instance, we compute z and x from a and b parameters of the SRK equation, which are different from a and b parameters of the van der Waals equation. Fig. 7 is the result of this procedure. Several caveats should be noted while referring to Fig. 7. First, van Konynenburg and Scott observe a narrow range of type IV behavior from z ¼ 0.26–0.30 that is qualitatively consistent with Fig. 7, but their value of x was 0.333. Fig. 7 indicates a value of x ¼ 0.1 to match this z range. We may explain this by noting that the values of a and b parameters are different for the two equations of state. Second, types I and V are omitted from Fig. 7 although many phase diagrams have been classified as types I and V experimentally. Similar to van Konynenburg and Scott, we attribute this to interference from solid phase boundaries in the experimental systems, hypothesizing that an LL region must exist at T ! 0 when the geometric mean is applied. Hence, type II systems will be classified experimentally as type I if the LL region lies below the solid phase boundary, and similarly for type IV and V systems. Third, it should be noted that for the SRK equation, the x
z relation increases up and to the left along a homologous series; hence, the number of components in a series that overlap the type IV region may be greater than initially anticipated. On a related note, solute molecular weights greater than 500 all lie in a tiny portion at the upper left corner of Fig. 7. Thus, a vast amount of polymer solution experience lies in a region of Fig. 7 where types II–IV are barely distinguishable based on the current analysis, and small variations in kij would drastically alter the phase behavior. Finally, one should note that negative values of x are omitted in Fig. 7.

Critical Phase Behavior

0.5 0.4 0.3 II 0.2 0.1 0 –1

–0.8

–0.6

–0.4

–0.2

0

0.2

0.4

ζ Fig. 7 SRK master diagram for phase behavior at kij ¼ 0. X, D, & are the computed type for II, III, and IV regions. Computations are based on n-alkanes with N2, CH4, C2H6, CO2, CH3OH, and H2O as solvents.

A small number of the sampled binary combinations resulted in negative x values and no type IV behavior was observed, but the sampling was so sporadic as to be inconclusive. With this background, the reader should be ready to consider the trends in phase behavior for a number of experimental systems. Several compilations are available for experimental systems and their type classifications.[7,13–16] These are typically in terms of homologous series, similar to the experimental programs themselves. We prefer to organize the list according to the type behavior. The type behavior plays a larger role the chemical processing. Type I systems A large number of these have been compiled by Rainwater[17] and Luks and Miller.[15] For the most part, these are mixtures of components with similar molecular weights and chemical structures like N2 þ CH4, N2 þ O2, or Refrigerant22 þ Refrigerant114. The components may differ structurally to an extent that the solution behavior would be classified as nonideal,

Critical Phase Behavior

however, and azeotropes are possible. Some of the more unusual examples include: HCl þ dimethyl ether, methane þ pentane (and lower paraffins), ethane þ n-heptadecane, propane þ nonacosane, CO2 þ n-hexane, and N2O þ n-heptadecane[18] and ethyl ether þ n-butanol. Simple systems like these can play important roles in refrigeration, especially if they form azeotropes. Applications of azeotropes in refrigeration are complicated by small variations in the azeotropic composition with respect to pressure. The result in the presence of a small leak is that the composition of the refrigerant varies depending on whether the leak is at high pressure or low pressure. This necessitates a protocol of replacing all refrigerant of an azeotropic system if any leak is suspected. Hence, even the thermodynamically simplest systems can induce frustration in a particular application. Type II systems The primary examples include CO2 þ n-heptane through CO2 þ n-dodecane. The CO2 þ n-hexane system is a prime example of a system that would very likely be type II if not for the interference of the solid phase boundary. The SRK equation of Fig. 7 overestimates the range of type II behavior for CO2 systems, terminating with pentadecane instead of dodecane. Small positive values for kij would improve the likelihood of quantitative agreement in this regard, but distract from the utility of Fig. 7 as a general, qualitative guideline. CO2 mixtures with 2-hexanol and 2-octanol also exhibit type II behavior. NH3 þ n-butane provides another interesting example of type II behavior. Other mixtures of NH3 þ higher hydrocarbons are likely to exhibit this type of phase behavior as well. Type III systems Type III behavior indicates the most extreme asymmetry between the components of a binary mixture. Nearly all H2 systems supply striking examples of type III behavior. CO2 mixtures with 2,5-hexanediol and 1-dodecanol are also classified as type III. The system CO2 þ n-tridecane is peculiar because it was classified by van Konynenburg and Scott as type III, whereas Enick et al.[19] have classified it as type IV, owing to experimental identification of a three-phase region. The system CO2 þ n-tetradecane is a variation on type III, where the solute-rich locus terminates in a solid(wax)–liquid–liquid boundary. Several important systems fall into a similar category. For example, CO2 þ naphthalene is commonly used as a model system for supercritical extraction. The naphthalene system differs from the n-tetradecane system in that the solute-rich locus terminates at a higher temperature

569

than the solvent-rich locus, leaving an open gap. For CO2 þ tetradecane, on the other hand, the two loci overlap in temperature. Wax precipitation from natural gas (mostly CH4) probably has a similar gap. In the temperature range of the gap, there is a large region of vapor þ solid equilibrium, uninterrupted by liquid phases. The x-P plots commonly seen in the supercritical literature appear as P-x projections at constant temperature when considered in this light. These distinctions are clarified in Fig. 13.6 of Ref.[5]. Branching in the solute favors lower melting temperatures resulting in normal type III behavior, as exemplified by CO2 þ squalane.[20] Type IV and V systems We consider these two types simultaneously because they share their distinctive feature. That feature is an interruption in the critical locus where two liquid phases appear over a short range of compositions before the critical locus reappears as a liquid–liquid critical point. Unlike low-temperature LL behavior, varying the pressure has a strong impact on type IV or V liquid– liquid–equilibria, (LLE) making it appear or entirely disappear over a remarkably narrow range of pressures. Systems that exhibit type IV behavior include methane þ 1-hexene and benzene þ polyisobutylene, the only polymer solution mentioned by van Konynenburg and Scott. Peters has also speculated that methane and ethane mixed with alkylbenzenes will form type II– IV solutions, in contrast to the I, III, V solutions of the n-alkanes.[21] The alkane mixtures provide the prototypical examples of type I ! type V behavior. Methane þ hexane (and higher alkanes), ethane þ octadecane, and propane þ pentatriacontane are all type V. The upper LL regions of these systems are noteworthy in that the temperature difference between the UCEP and the LCEP seems to monotonically increase with increasing carbon number.[21] Ultimately, this trend must reverse as type III behavior sets in, but no indication of this reversal has been observed experimentally. Mixtures of methane with hexane isomers provide unusual examples of type V phase behavior. Type V behavior is exhibited for all isomers except 2,2dimethyl butane. Ternary mixtures of methane with the 2,2 and 2,3-isomers provide a rare example of tricritical behavior. Turning to another example, the type V LLV locus becomes extremely short as the asymmetry of the mixture increases to the point where transition to type III behavior is approached. Ethane þ p-dichlorobenzene provides an example of this phenomenon, with an LLV locus extending over a mere 0.6 K.[22] Such an odd effect may seem to have little practical significance, unless one considers the impact of an unexpected precipitation on a critical pipeline.

C

570

A prominent chemical processing application of type V behavior is exemplified in the Selexol process. The Selexol process was originally based on type V behavior in propane þ triglyceride systems and has also been applied to propane þ fatty acids. Dimerization of the acids makes them perform effectively like ‘‘diglycerides’’ thermodynamically. Hixson et al. have published extensively on this process.[23,24] They have also developed the concept of using temperature gradients within an extraction column to enhance separations, a concept recently discussed in the context of SCF CO2 extractions.[25] The glycerides partition favorably into the propane-rich phase. Downstream manipulations of the pressure and temperature lead to delicate fractionation of the glyceride constituents. As an example of a nutraceutical application, vitamin A has been produced in this way. Applications based on CO2 or dimethyl ether remain as possibilities. Summary of Phase Diagram Basics Summing up, we reiterate that critical phase diagrams describe an array of subtle, complex behaviors. Some of these create opportunities for chemical processing like the Selexol process. Some characterize potential problems like wax precipitation. Either way, the phase diagrams provide concise insight into a broad scope of possibilities that researchers in high pressure chemical processing need to be aware of. It should be noted that the discussion here is actually quite circumscribed in that only binary mixtures have been explicitly considered. Recent work shows a quantum leap of higher order complexity when it comes to treating ternary mixtures and cosolvent effects.[26]

RESEARCH TOPICS Beyond the basics, there are a number of efforts underway to advance the role of critical phase behavior in chemical processing. A few of these are already making an impact, but most remain as concepts under development. We briefly summarize key elements of these and provide links that should facilitate a broader understanding on subjects of interest. This list does not aspire to be comprehensive. Rather, it reflects the author’s biased impressions of subjects that might resonate with developers of chemical processes. It is hoped that this brief survey will improve with feedback from readers and the passage of time. Theory Much of the traditional theory behind critical phase behavior has been discussed in the context of phase

Critical Phase Behavior

diagrams. This section is devoted to a brief review of fundamental studies that have been less fully developed when it comes to chemical processing in the critical region. Among the early theoretical developments in SCF studies was the finding that solvent molecules tend to cluster around solute molecules,[27] giving rise to local density functionals that may deviate markedly from the bulk density. Similarly, local compositions differ markedly from bulk compositions. These composition enhancements have been demonstrated experimentally by fluorescence spectroscopy and theoretically by molecular simulation[28] and by integral equation theories.[29] This enhances the solubility of heavy solutes to a great extent. Another impact of the local composition is on reaction kinetics. One might suspect that reaction kinetics should follow the local composition rather than the bulk composition. In fact, this has been demonstrated experimentally and described theoretically.[30,31] There appears to be little recognition of these phenomena in chemical processing, but that may be because of lack of awareness on the part of reaction engineers. Alternatively, critical phase specialists may not be aware of high-pressure reactive processes that could benefit from this knowledge. Either way, this peculiar property of the critical region remains to be exploited in chemical processing. One theoretical subject that has received much attention is the subject of equations of state. Equations of state based on conventional mixing rules, like the Peng–Robinson[32] or SRK equations, tend to underestimate clustering effects. Clustering is essentially a local composition effect. Another strong contributor to local composition anomalies is hydrogen bonding. The local composition mixing rules are typified by the Wong–Sandler modification of the Peng–Robinson equation (PRWS).[33] Methods like this are capable of correlating complex phenomena, but they show signs of detachment from physical reality. For example, the kij parameters tend to be extremely large and erratic in value. Nevertheless, this flexibility can be applied to clustering that arises from critical fluctuations or from hydrogen bonding, obviating the need to draw distinctions. At present, the PRWS approach is probably the least likely to fail when it comes to correlating complex phase equilibria. This makes it the next logical model to consider after trying a simple equation like the Peng–Robinson model. Hydrogen bonding equations of state are typified by the statistically associating fluid theory (SAFT) equation.[34,35] These models apply Wertheim’s[36] perturbation theory to obtain an accurate characterization of the association and solvation interactions that are primarily responsible for the difficulties experienced in representing phase equilibria involving polar substances. These theories are very appealing from a chemical perspective and are promising for

Critical Phase Behavior

predictions. The solvation parameters follow trends that would be expected from correlations of spectroscopic measurements like the Kamlet–Taft parameters.[37] One of the more important advantages of the SAFT approach is that extension to polymer solutions is entirely straightforward, which cannot be said of the PRWS approach. Furthermore, the formalism lends itself to treatment of copolymers just as straightforwardly. For example, Feelly et al.[38] were able to accurately predict copolymer solubilities in CO2 over the range of 350–550 K and 1100–2100 bars using kij values calibrated for homopolymers. Optimization on the molecular scale is greatly facilitated by the SAFT approach, showing great promise for future development in chemical processing. A recent review of modifications and applications of the SAFT approach has been written by Muller and Gubbins.[39]

Petroleum Applications Oil and gas recoveries provide examples of chemical processing at high pressure that arise quite naturally. A short list of related research topics would include CO2 flooding for improved oil recovery, and wax and hydrate remediation in gas recovery. The temperature and the pressure at which oil recovery takes place are steadily increasing as reserves are sought at progressively greater depths. The added expense of operating at these depths also intensifies motivation to enhance the productivity of each well as much as possible. One strategy for this enhancement has been to recover or generate CO2 at the wellhead and inject it into the well to displace residual oil. The CO2 is generally miscible with the oil, reducing its density and viscosity and facilitating recovery. The problem with this approach is that some of the CO2 bypasses the oil and goes directly to the reservoir outlet. The method becomes ineffective when these ‘‘fingers’’ of bypass flow become numerous. To overcome the fingering problem, dissolving polymers into the CO2 can enhance the viscosity. The problem then becomes improving polymer solubility. This has been a longstanding problem. Early work showed that additives like toluene were effective, but the amount required (>10 wt%) was too large to be economical.[40] An alternative approach is to customize the molecular structure to enhance solubility. Beckman et al. have found that fluorination promotes solubility to a sufficient extent that viscosity can be increased by a factor of 3 with 4 wt% polymer.[41] Unfortunately, fluorinated polymers are expensive. When considering the immense volumes of materials being pumped around in an oil field, a small percentage is multiplied by a very large number. To be economical, the viscosity enhancing agents need to be nearly as inexpensive as the oil itself.

571

The subject of polymer solubility in CO2 has stimulated research with biomedical applications. These are discussed in a later section. Wax precipitation from natural gas is another longstanding problem being exacerbated by progressively deeper drilling. Assuming that natural gas is predominantly methane and the relevant phase diagram is similar to type III, it may be possible to surpass the critical locus at the elevated temperatures and pressures of reservoir conditions. As a related phenomenon, high-pressure synthesis of polyethylene from ethylene is known to operate in this upper region of the type III diagram.[6] All temperature and pressure drops would then result in precipitation in the recovery stream. Pedersen described one method of predicting solid–vapor precipitation based on the assumption of an ideal solution in the solid phase.[42] Recently, Coutinho et al.[43] have reported improved results when a predictive activity model is applied to the solid phase. The fluid phases have been accurately modeled with conventional equations and mixing rules in these instances. The availability of reasonable models for these systems provides opportunities to brainstorm through a number of prospective remedies. Promising strategies focus on precipitating at controlled locations and seeding heavy components that lower the viscosity of any wax that does form. With knowledge of the phase diagrams, it should be possible to accommodate wax precipitation problems with reasonable success. Gas hydrates form when small molecules like nitrogen, methane, CO2, or propane come in contact with water at pressures above 30 bars and temperatures of roughly 275  30 K. The composition of the gas species in the hydrate is in the order of 5 wt%. Technically, hydrate description requires an entirely new family of phase diagrams. We defer to existing literature for a detailed discussion.[44] Like waxes, hydrates are a nuisance to gas recovery. A commonly suggested treatment is to dope the gas stream with methanol. Methanol disrupts the formation of hydrates, necessitating higher pressures before the hydrates can form. Recently, Peters et al. have reported that additives like cyclohexane or cyclobutanone can have the opposite effect, reducing the pressure or elevating the temperature of hydrate formation.[45] At first, this may seem like an added nuisance, but it adds a remediation strategy. If precipitation can be effected at controlled locations, then the problem may be remedied.

Bio/Medical Applications The studies most commonly identified with supercritical fluids and critical phase behavior are those concerned with extraction of natural products. No review of this subject would be complete without

C

572

reference to the decaffeination of coffee.[46] A commercial process for many years, it still acts as a model system for treatment of natural products.[47] In this application, purification of the caffeine enhances the value of both the caffeine and the residue. More recent applications have focused on simply extracting the valuable component. Typical extracts have been alkaloids, triglycerides, fatty acids, and antioxidants. Any given issue of the Journal of Supercritical Fluids is likely to contain 3–4 articles on similar extractions. Fatty alcohols can be obtained from fatty acids by hydrogenolysis in near critical propane. The phase behavior of this system shows signs of type IV behavior with a pressure-dependent LL region near room temperature and 30 bars.[48] Apparently, these efforts are still largely academic. In a brief search, 72 patents since 1980 mentioned supercritical extraction in their titles. Many of the early patents were focused on upgrading fuels. Thirteen patents target specific natural products, with the earliest dating from 1989. Fatty acids and taxol related species are mentioned twice each. Three patents focused on flavors. The rest of the patents targeted esoteric species, with one recent patent focusing on a specific protein. None of the patents for natural products were from the United States. Three were from Europe, four from China, and six from Japan. This trend suggests that much of the chemical processing interest in natural products derives from overseas. Greater interest in the United States has been focused on the formation of microparticles for encapsulation and time release applications. DeBenedetti et al. have described two processes for microparticle production: one by rapid expansion of supercritical solutions (RESS)[49,50] and one based on a supercritical antisolvent (SAS), for which they hold a patent.[51] In the RESS process, a homogeneous polymer þ CO2 solution is expanded through a nozzle, dropping the pressure and precipitating polymeric particles whose size and shape depend on the precise conditions of the expansion. Nineteen patents identified the concept of expansion from supercritical solution since 1988, the substantial majority of which targeted production of microparticles. In the SAS process, a homogeneous solution is exposed to a supercritical phase that does not dissolve the polymer, inducing precipitation. Eight patents identified the concept of SAS, including the original in 1993. Four of these were international patents and three were US patents. Five of these patents targeted drugs or proteins. Recent references to these applications provide links to further study.[52,53] The attraction of CO2 in these applications is its biocompatibility. Any residual CO2 solvent will be readily metabolized. Use of CO2 as a biocompatible solvent has required several adaptations to enhance the solubility of proteins and polymers in the CO2

Critical Phase Behavior

phase. Most of these have focused on fluorination,[54] although recent efforts have turned toward cheaper esters and ethers.[55] One such adaptation of CO2 won a Green Chemistry Award for 2002.[56,57] Graft polymerization in CO2 is another proven means of altering the compatibility and solubility.[58] Johnston et al. have demonstrated how fluorinated surfactants can be used to dissolve proteins into supercritical micelles.[59,60] Johnston is named in seven patents on drug and microparticle preparations. Taken together, the bio=medical applications appear to be some of the most promising for chemical processing in the near term.

Green Chemistry Similar to CO2’s attractiveness from a biological perspective, it is also environmentally friendly relative to traditional solvents. This property suggests its extended use as a substitute in many applications where traditional solvents have been applied. Two examples can be cited from the paint and dry cleaning industries.[61] Donohue and collaborators have developed a process for spray painting with CO2 substituted in place of the usual solvents.[62,63] This is a fully developed, viable process, but it must compete with continually advancing technologies in the paint industry. The competing technologies are already reducing the proportions of solvent by lowering the molecular weight of the paint and making it reactive after coating.[64] Dry cleaning with CO2 is rapidly becoming a more competitive process. Two similar technologies were developed at Los Alamos and by DeSimone et al. These processes effectively use near-critical CO2 for washing and a thermal cycle for regenerating the CO2. This chemical process has received a surge of interest from ICI and Linde.[65] With this investment, it is likely that this process will increase market share over the near term. A related technology is dedicated to the use of CO2 as a cleaning agent in the electronics industry. Los Alamos and DeSimone, among others, are now competing in the development of that processing technology. Carbon dioxide is not the only nonflammable, biocompatible, and widely available solvent. Let us not forget about water. In particular, supercritical water oxidation and related reactive processes have shown a tremendous capacity for reducing toxic chemicals to innocuous constituents. Akiya and Savage have authored a recent review.[66] Of particular interest is the section on hydrolysis in SCF water. Numerous references are tabulated according to chemical family. With regard to the more toxic compounds, Klein et al. have been especially active for many years[67,68] and Tester et al. have focused on alkyl halides.[69]

Critical Phase Behavior

Modell et al. have patented this technology for nonincineratory destruction of toxic chemicals.[70,71] Early work in this area was related to the supercritical extraction=depolymerization of coal. Lee et al. have found it possible to depolymerize scrap polymer such that clean monomers can be recovered, turning a liability into an asset.[72,73] This approach is especially interesting in the case of polyvinyl chloride (PVC). Whereas PVC normally degrades into polychlorinated hydrocarbons, the presence of oxygen and water promotes decomposition into the monomer and hydrochloric acid. Potential problems in this developing technology include clogging and corrosion.[74] Summary of Research Topics Widespread applications of small-scale critical phase technology, like dry cleaning and paint spraying, are likely to have a subtle, autocatalytic effect on general implementation. High-pressure processing equipment is generally perceived as being inherently high in capital cost. In actuality, doubling the mass of steel in one or two key components of an overall process makes a small difference in the overall cost. Major costs are involved, however, in custom developed equipment that requires repeated testing and redesign and cannot be mass produced. One way of representing this is by the size exponent of a cost estimation chart. An exponent of 0.24 indicates the importance of economies of scale for SCF equipment.[75] On the other hand, global demand for a specialty flavor or a nutraceutical is not the sort of thing that can be scaled to large values. Hence, one key to economical critical phase chemical processing is mass production of interchangeable components for small-scale operation. The economics and the prevalence of critical phase behavior in chemical processing could change substantially if such a development were to take place.

CONCLUSIONS We have summarized briefly the key concepts and applications of a broad and complex subdiscipline of thermodynamics. Necessarily, selective judgments have been made as to what aspects of critical phase behavior are most important to chemical processing. Alternative perspectives on the role of critical phase behavior in chemical processing are readily available. For example, the role of reactive synthetic processes at critical conditions has received little attention here.[76] As one prospective source for further links, the article by Perrut provides an especially relevant survey.[75] Readers are encouraged to survey this literature for themselves and keep abreast of the latest developments.

573

It is a subject of current research and its status in chemical processing is constantly evolving.

REFERENCES 1. Kiselev, S.B.; Ely, J.F.; Lue, L.; Elliott, J.R. Computer simulations and crossover equationof-state of square-well fluids. Fluid Phase Eq. 2002, 200, 121–145. 2. Younglove, B.A.; Ely, J.F. Thermophysical properties of fluids. II. Methane, ethane, propane, isobutane and normal butane. J. Phys. Chem. Ref. Data 1987, 16, 577–798. 3. Noyori, R. Supercritical fluids: Introduction. Chem. Rev. 1999, 99, 353–634. 4. Van Konynenburg, P.H.; Scott, R.L. Critical lines and phase behavior in binary van der Waals mixtures. Phil. Trans. Roy. Soc. London Ser. 1980, A298 (1442), 495–540. 5. Elliott, J.R.; Lira, C.T. Introductory Chemical Engineering Thermodynamics; Prentice-Hall: Englewood Cliffs, NH, 1999. 6. De Loos, T. Supercritical Fluids; Kiran, E., Levelt Sengers, J.M.H., Eds.; Kluwer: Amsterdam, 1994; 65–89. 7. Streett, W.B. Chemical Engineering at Supercritical Fluid Conditions; Paulaitis, M.E., Penninger, J.M.L., Gray, R.D., Davidson, P., Eds.; Ann Arbor Science: Ann Arbor, MI, 1983; 3–30. 8. Ochel, H.; Becker, H.; Maag, K.; Schneider, G.M. J. Chem. Thermo. 1993, 25, 667. 9. Brunner, E. Fluid mixtures at high-pressures.9. phase-separation and critical phenomena in 23 (normal-alkane þ water) mixtures. J. Chem. Thermo. 1990, 22, 220. 10. Nezbeda, I.; Pavlicek, J.; Kolafa, J.; Galindo, A.; Jackson, G. Global phase behavior of model mixtures of water and n-alkanols. Fluid Phase Eq. 1999, 158, 193–199. 11. Soave, G. Equilibrium constants from a modified Redlich-Kwong equation of state. Chem. Eng. Sci. 1972, 27, 1197. 12. Elliott, J.R.; Daubert, T.E. Evaluation of the equation of state method for calculation of the critical properties of mixtures. Ind. Eng. Chem. 1987, 26, 1686–1691. 13. Rowlinson, J.S.; Swinton, F.L. Liquids and Liquid Mixtures, 3rd Ed.; Butterworths: Boston, 1982. 14. Schneider, G.M. High-pressure investigations on fluid systems — a challenge to experiment, theory, and application. J. Chem. Thermo. 1991, 23, 301–326. 15. Miller, M.M.; Luks, K.D. Observations on the multiphase equilibria behavior of CO2-rich and

C

574

16.

17.

18. 19.

20. 21.

22.

23. 24. 25.

26.

27. 28.

29.

30.

31.

Critical Phase Behavior

ethane-rich mixtures. Fluid Phase Eq. 1989, 44, 295. Brunner, E. Fluid mixtures at High Pressures. VII. Phase separations and critical phenomena in 12 binary mixtures containing ammonia. J. Chem. Thermo. 1988, 20, 1397. Rainwater, J.C. Supercritical Fluid Technology: Reviews in Modern Theory and Applications; Bruno, T.J., Ely, J.F., Eds.; CRC Press: Boca Raton, FL, 1991; 57–148. Jangkamolkulchai, A.; Lam, D.H.; Luks, K.D. Fluid Phase Eq. 1989, 50, 175–187. Enick, R.M.; Holder, G.I.; Morsi, B.I. Critical and 3 phase-behavior in the carbon-dioxide tridecane system. Fluid Phase Eq. 1985, 22, 209–224. Schneider, G.M. Angew. Chem. Int. Ed. Engl. 1978, 17, 716–727. Peters, C.J. Supercritical Fluids; Kiran, E., Levelt Sengers, J.M.H., Eds.; Kluwer: Amsterdam, 1994; 117–145. Diepen, G.A.M.; Scheffer, F.E.C. On critical phenomena of saturated solutions in binary systems. J. Am. Chem. Soc. 1948, 70, 4081. Hixson, A.N.; Miller, R. US, 2,219,652, 1940. Bogash, R.; Hixson, A.N. Chem. Eng. Progress 1949, 45, 597–601. Stahl, E.; Quirin, K.-W.; Gerard, D. Dense Gases for Extraction and Refining; Springer-Verlag: New York, 1988. Scheidgen, A.L.; Schneider, G.M. Fluid phase equilibria of (carbon dioxide plus a 1-alkanol plus an alkane) up to 100 MPa and T ¼ 393 K: cosolvency effect, miscibility windows, and holes in the critical surface. J. Chem. Thermo. 2000, 32, 1183–1201. Kim, S.; Johnston, K.P. Clustering in supercritical mixtures. AIChE J. 1987, 33, 1603. Knutson, B.L.; Tomasko, D.L.; Eckert, C.A.; Debenedetti, P.G.; Chialvo, A.A. Supercritical Fluid Technology: Theoretical and Applied Approaches in Analytical Chemistry; Bright, F.V., McNally, M.E.P., Eds.; American Chemical Society: Washington, DC, 1992; Vol. 488, pp. 60. Wu, R.S.; Lee, L.L.; Cochran, H.D. Structure of dilute supercritical solutions — clustering of solvent and solute molecules and the thermodynamic effects. Ind. Eng. Chem. Res. 1990, 29. Brennecke, J.F.; Chateauneuf, J.E. Homogeneous organic reactions as mechanistic probes in supercritical fluids. Chem. Rev. 1999, 99, 433. Roek, D.P.; Chateauneuf, J.E.; Brennecke, J.F. A fluorescence lifetime and integral equation study of the quenching of naphthalene fluorescenceby bromoethane in super-and subcritical ethane. Ind. Eng. Chem. Res. 2000, 39, 3090.

32. Peng, D.Y.; Robinson, D.B. A new two-constant equation of state. Ind. Eng. Chem. Fundam. 1976, 15, 59. 33. Wong, D.S.H.; Sandler, S.I. A theoretically correct mixing rule for cubic equations of state. AIChE J. 1992, 38, 671. 34. Chapman, W.G.; Gubbins, K.E.; Jackson, G.; Radosz, M. New reference equation of state for associating liquids. Ind. Chem. Eng. Res. 1990, 29, 1709. 35. Huang, S.; Radosz, M. Equation of state for small, large, polydisperse, and associating molecules: Extension to fluid mixtures. Ind. Eng. Chem. Res. 1991, 30, 1994. 36. Wertheim, M.S. Fluids with highly directional attractive forces. III. Multiple attraction sites. J. Stat. Phys. 1986, 42, 459. 37. Abraham, M.H.; Andonian-Haftvan, J.; Kamlet, M.J.; Whiting, G.S.; Leo, A.; Taft, R.W. J. Chem. Soc. Perkin Trans. 1994, 2, 1777. 38. Feelly, T.; Becker, F.; Latz, H.; Sadowski, G.; Buback, M. In Paper 82b, AIChE Fall National Mtg; AIChE: New York, Indianapolis, 2002. 39. Mu¨ller, E.A.; Gubbins, K.E. Molecular-based equations of state for associating fluids: a review of SAFT and related approaches. Ind. Eng. Chem. Res. 2001, 40, 2193. 40. Irani, C.; Zajac, J. Handling of high pour point West-African crude oils. J. Pet. Tech. 1982, 34, 289–298. 41. Shi, C.M.; Huang, Z.H.; Beckman, E.J.; Enick, R.M.; Kim, S.Y.; Curran, D.P. Semi-fluorinated trialkyltin fluorides and fluorinated telechelic ionomers as viscosity-enhancing agents for carbon dioxide. Ind. Eng. Chem. Res. 2001, 40, 908–913. 42. Pedersen, K.S. Prediction of cloud point temperatures and amount of wax precipitation. SPE Prod. and Fac. 1995, 46. 43. Coutinho, J.A.P.; Pauly, J.; Daridon, J.L. A thermodynamic model to predict wax formation in petroleum fluids. Braz. J. Chem. Eng. 2001, 18, 411–423. 44. Sloan, E.D.J. Clathrate Hydrates of Natural Gases; 2nd Ed.; Marcel Dekker: New York, 1998. 45. Mooijer-van den Heuvel, M.M.; Peters, C.J.; de Swaan Arons, J. Gas hydrate phase equilibria for propane in the presence of additive components. Fluid Phase Eq. 2002, 193, 245. 46. Prasad, R.; Gottesman, M.; Scarella, R.A. Decaffeination of aqueous extracts of roasted coffee. General Foods Corp. USA, US, 4,246,291, 1981. 47. Mohamed, R.S.; Saldana, M.D.A.; Mazzafera, P.; Zetzl, C.; Brunner, G. Extraction of caffeine, theobromine, and cocoa butter from Brazilian cocoa beans using supercritical CO2 and ethane. Ind. Eng. Chem. Res. 2002, 41, 6751–6758.

Critical Phase Behavior

48. Rovetto, L.J.; Bottini, S.B.; Peters, C.J. Phase equilibrium data on binary and ternary mixtures of methyl palmitate, hydrogen and propane. J. Supercrit. Fluids 2004, 31, 111–121. 49. Debenedetti, P.G.; Tom, J.W.; Kwauk, X.; Yeo, S.D. Rapid expansion of supercritical solutions (RESS). Fluid. Phase Eq. 1993, 82, 311–321. 50. Smith, R.D. Supercritical fluid molecular spray thin films and fine powders. Battelle Memorial Institute, USA. US, 4,734,451, 1988. 51. Debenedetti, P.G.; Lim, G.B.; Prud’homme, R.K. Formation of protein microparticles by antisolvent precipitation. Princeton University. Eur, 542314, 1993. 52. Weber, M.; Russell, L.M.; Debenedetti, P.G. Mathematical modeling of nucleation and growth of particles formed by the rapid expansion of a supercritical solution under subsonic conditions. J. Supercrit. Fluids 2002, 23, 65–80. 53. Werling, J.O.; Debenedetti, P.G. Numerical modeling of mass transfer in the supercritical antisolvent process: miscible conditions. J. Supercrit. Fluids 2000, 18, 11–24. 54. DeSimone, J.M.; Guan, Z.; Elsbernd, C.S. Science 1992, 257, 945. 55. Conway, S.E.; Byun, H.S.; McHugh, M.A.; Wang, J.D.; Mandel, F.S. Poly(lactide-co-glycolide) solution behavior in supercritical CO2, CHF3, and CHClF2. J. App. Polym. Sci. 80, 1155–1161. 56. Sarbu, T.; Styranec, T.J.; Beckman, E.J. Design and synthesis of low cost, sustainable CO2-philes. Ind. Eng. Chem. Res. 2000, 39, 4678–4683. 57. Ritter, S.K. Green Challenge. C&E News 2002, 80, 26–30. 58. Trivedi, A.H.; Kwak, S.; Lee, S. Grafting of poly (vinyl chloride) and polypropylene with styrene in a supercritical CO2 solvent medium: synthesis and characterization. Polym. Eng. Sci. 2001, 41, 1923–1937. 59. Johnston, K.P.; Harrison, K.L.; Clarke, M.J.; Howdle, S.; Heitz, M.P.; Bright, F.V.; Carlier, C.; Randolph, T.W. Science 1996, 271, 624. 60. Jacobson, G.B.; Tumas, W.; Johnston, K.P. Biphasic catalysis in water=carbon dioxide micellar systems. US, 6,479,708, 2002. 61. Tullo, A.H. Dry cleaning and paint spraying are CO2 ‘pioneers’. C&E News 1999, 77, 13. 62. Lee, C.; Hoy, K.L.; Donohue, M.D. Supercritical fluids as diluents in liquid spray application of coatings and apparatus therefor. Union Carbide Corp., USA, EP, 1989. 63. Argyropoulos, J.N.; Lear, J.J.; Hoy, K.L.; Donohue, M.D. Pressurized fluid composition and process for making same. Union Carbide and Plastics Technology Corp., USA, EP, 481431, 1992.

575

64. Tullo, A.H. Automotive coatings. C&E News 2002, 80, 27–30. 65. Tullo, A.H. ICI enters CO2 dry cleaning. C&E News 2002, 80, 12. 66. Akiya, N.; Savage, P.E. Roles of water for chemical reactions in high-temperature water. Chem. Rev. 2002, 102, 2725–2750. 67. Abraham, M.A.; Klein, M.T. Pyrolysis of benzylphenylamine neat and with tetralin, methanol, and water solvents. Ind. Eng. Chem. Res. 1985, 24, 300–306. 68. Izzo, B.; Klein, M.T.; LaMarca, C.; Scrivner, N.C. Hydrothermal reaction of saturated and unsaturated nitriles: Reactivity and reaction pathway analysis. Ind. Eng. Chem. Res. 1999, 38, 1183– 1191. 69. Salvatierra, D.; Taylor, J.D.; Marrone, P.A.; Tester, J.W. Kinetic study of hydrolysis of methylene chloride from 100 to 500
C. Ind. Eng. Chem. Res. 1999, 38, 4169–4174. 70. Modell, M. Treatment of organic material in supercritical water. Modar, Inc., USA. US, 4,338,199, 1982. 71. Modell, M.H.; Edward, G.; Gairns, Stuart, A. Method and apparatus for treating paper-mill effluents. Modell Environmental Corp., USA. WO, 9,510,486, 1995. 72. Lilac, W.D.; Lee, S. Kinetics and mechanisms of styrene monomer recovery from waste polystyrene by supercritical water partial oxidation. Adv. Envi. Res. 2001, 6, 9–16. 73. Lee, S.; Gencer, M.A.; Fullerton, K.L.; Azzam, F.O. Oxidation-depolymerization process for waste polymers. University of Akron, USA. US, 5,386,055, 1995. 74. Mitton, D.B.; Yoon, J.-H.; Cline, J.A.; Kim, H.-S.; Eliaz, N.; Latanision, R.M. Corrosion behavior of nickel-based alloys insupercritical water oxidation systems. Ind. Eng. Chem. Res. 2000, 39, 4689–4696. 75. Perrut, M. Supercritical fluid applications: Industrial developments and economic issues. Ind. Eng. Chem. Res. 2000, 39, 4531–4535. 76. McCoy, M. ‘Green’ processes based on supercritical carbon dioxide are moving out of the lab. C&E News 1999, 77, 11–13. 77. Lira, C.T. In Supercritical Fluid Technology in Oil and Lipid Chemistry; King, J.W., List, G.R., Eds.; AOCS Press: Champaign, IL, 1996. Reproduced with permission.

BIBLIOGRAPHY Chemical Reviews 1999, 99, 565–590. Ind. Eng. Chem. Res. 2000, 39, 4441–5048.

C

Cross-Linked Polyethylene C Carosena Meola Giovanni Maria Carlomagno Department of Energetics, Thermofluidynamics and Environmental Control (DETEC), University of Naples Federico II, Napoli, Italy

Giuseppe Giorleo Department of Materials and Production Engineering (DIMP), University of Naples Federico II, Napoli, Italy

INTRODUCTION In people’s opinion, plastic is a light and weak substance that easily melts when warmed. And yet, cross-linking the carbon atoms suffices to transform such material into a superior material that may be resistant to temperature, pressure, corrosion, and that can be used in a variety of applications. In fact, polyethylene (PE) once crosslinked is advantageously employed in the fabrication of blanket insulation for electrical and telephone wires, pipes for the transport of cold and hot liquids, prostheses for the human body, and so forth. Since the late 1960s, when the European scientist Engel first succeeded in cross-linking PE, there has been a proliferation of cross-linking methods with the intention of fabricating a type of PE suitable for a specific need. There are many cross-linking methods; each method has advantages and disadvantages and no one method works well for every product. In fact, a cross-linked polymer obtained with one method may be excellent for one application, but it may be very inadequate for another application. It is important to choose the most effective method for the specific need and to comply with quality standards. Otherwise, a fixed sequence of operations alone does not ensure the product quality, but it is also important to characterize the products through appropriate testing and nondestructive evaluation. Another important point is with regard to the hazards and risks related to the substances, or devices, employed; manufacturers must comply with safety regulations and ecosustainability. The intended purpose of this article is to expose the reader to an overview of the existing methods and to give indications and suggestions about the most appropriate method for a specific application with the existing legislation involving both quality and safety aspects.

bags to envelopes for alimentary packaging, cases to hold and protect electronic devices, medical prostheses, and so forth. Indeed, the list could go on without end! A long chain of carbon atoms (CH2) represented as: ð
CH2
CH2
Þ

ð1Þ

is also called linear, or high-density polyethylene (HDPE). However, the ethylene molecules do not always add on in a regular fashion, but sometimes, under high-pressure polymerization, ethylene molecules attach as short branches leading to low-density polyethylene (LDPE). Low-density polyethylene is much more economical to produce and can be used in many applications, which do not require specific material characteristics. Since its accidental discovery in the early 1930s in Great Britain from the failure of a chemical reaction under pressure, researchers’ efforts have been driven toward obtaining a PE with specific chemical, mechanical, and thermal characteristics for the fabrication of complex-shaped tools, or for use in adverse environmental conditions.[1] The fundamental way to improve material properties such as impact strength, chemical resistance, and thermal characteristics is via crosslinking. Indeed, the introduction of cross-linked polyethylene (PEX) in the early 1970s was another milestone in the plastic era. From that date, PEX has captured a giant share of the market because of its superior characteristics with respect to other plastics. Modifications in the polymeric structure can be brought about by several methods, which differ from each other in two main factors. One factor is the state (i.e., molten or solid) of the polymer during crosslinking. The other factor is the type of activator used to promote cross-linking. In this article the different methods are grouped into two main categories:

CROSS-LINKING METHODS Polyethylene is certainly the most used kind of plastic. In fact, there are many things made of PE, from shop Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120007720 Copyright # 2006 by Taylor & Francis. All rights reserved.

 Chemical processes, which require a chemical initiator (peroxide or silane) to induce links in the polymer chain. 577

578

Cross-Linked Polyethylene

 Radiation processes, which involve exposure to ionizing radiation from either radioactive sources or highly accelerated electrons, to liberate free radicals for cross-linking.

–O–O– group, as in dicumyl peroxide:

Chemical Processes

or two –O–O– groups as in 1,3-1,4bis(tert-butylperoxyisopropyl) benzene:

Cross-linking is activated by a chemical substance, which could be:  A peroxide—the method is called peroxide initiated cross-linking. The resulting PE is also called PEX-A in the European standards.  A silane—the method is called cross-linking via silane or moisture-based vinyl silane cross-linking. The resulting PE is also called PEX-C in the European standards. Peroxide processes The basic process consists of the decomposition of a peroxide at high temperature and the creation of carbon bonds along the PE chain; cross-linking occurs in the molten state. This process was developed in the late 1960s in central Europe by the German scientist Thomas Engel and it is called the Engel process. Indeed, there are several different peroxide-based processes. In the Engel process, a granulated blend of PE, peroxide, and stabilizers is sintered together under high pressure; cross-linking occurs during extrusion through a long heated die. This process is generally used for the production of HDPE. In the Pont a’Mousson process, PE is mixed with peroxide, extruded, and cross-linked in a salt bath at high temperature; the resulting product is a low, or medium, density PE. In the Daoplas process instead, the peroxide is incorporated after extrusion and activated in equipment downstream (the extruder) at high temperature and pressure. An organic peroxide is a carbon-based chemical that includes a minimum of two oxygen atoms bonded together (–O–O–).[2–4] The general formula is: R1
O
O
R2

ð2Þ

where R1 and R2 can be aryl, alkyl, or acyl groups. There are also peroxides with a second –O–O– bond and three R-groups. Owing to the chemical structure of such R-groups several different families such as alkyl, aryl, and acyl peroxides and peroxyketals are identified. The alkyl peroxides produce the most reactive free radicals and they are the most used for cross-linking.[4] Such peroxides may contain one

The peroxyketals contain two peroxy groups bonded on the same carbon atom; this peculiarity causes instability and high reactivity and thereby complicates peroxide transportation, handling, and storage operations. The peroxide cross-linking reaction occurs in three steps: Step I: The addition of heat causes peroxide thermal decomposition (i.e., the oxygen bonds break via homolysis). One unpaired electron remains in each oxygen atom and promotes the formation of peroxide radicals. Step II: Each peroxide radical reacts with the PE molecule, i.e., abstracts a hydrogen atom from the polymer chain, becoming a stable ROH species. The abstraction of hydrogen causes the formation of polymer radicals. Step III: Two polymer radicals react with each other forming stable PEX. Peroxides are mainly used for the production of HDPE pipes. The basic process consists of three steps.  Mixing: The peroxide, in liquid or molten phase, is sprayed on PE granules.  Extrusion: The compound (soaked granules plus eventual additives) is poured into the extruder where it is melted and shaped. The cross-linking reaction takes place either in the extruder (Engel) or in the downstream equipment.  Curing: The extruded product enters the equipment where cross-linking is completed under controlled temperature and pressure. The use of high-pressure tanks is necessary for degassing of volatiles and avoiding the formation of voids inside the PEX. Organic peroxides belong to the aromatic hydrocarbon=alkylbenzene=dialkylbenzene chemical family and are designated by the International Occupational Safety and Health Information Centre (CIS) with the symbols O (oxidizing), Xi (irritating), and N (environmentally

Cross-Linked Polyethylene

dangerous) and are collocated in the Hazard class 5.2. UN No. 3110. They are generally shock, heat and friction sensitive, and are incompatible with strong oxidizing agents, can react violently with reducing agents, heavy metals, concentrated acids, and concentrated bases, and may ignite organic materials on contact. Such substances can cause local irritations and burns of the skin and mucous membranes of the eyes and the respiratory tract. The risk and safety phrases are: R7-36-38-51-53 and S3-7-17-36-37-39. Thorough care and special technical regulations are required for transportation, handling, and storage.

579

produces stable copolymers with long shelf life. Conversely, polymerization in a high-pressure reactor results in highly branched products. b. Silane-grafted PEs are fabricated in compounding equipment involving a single- or twin-screw extruder. Silane may be grafted to any ethylene-polymer, and so PEX of specific characteristics can be obtained. The resulting silane-grafted PE also contains hydrolyzable alkoxy groups, which, in the presence of water, react to join adjacent grafted PE molecules to form stable three-dimensional cross-linked networks of siloxane linkages.[6]

Vinyl silane moisture In this process, the cross-linking is activated by silane coupling agents. The organic moieties of silanes can react with many chemicals, including polymers, via typical organic chemistry reactions.[4–7] The organosilane molecule includes a central silicon atom (Si) bounded to two different categories of groups (vinyl and alkoxy), which exhibit different reactivity. The vinyl groups allow silane grafting to the PE backbone. The hydrolyzable alkoxy groups in the presence of water or moisture react (via condensation and hydrolysis) and generate a three-dimensional network of siloxane linkages. The PE is cross-linked in the crystalline state. The use of silanes causes siloxane (Si–O–Si) bridges to form, which are less rigid than carbon-to-carbon (C–C) bonds induced by peroxides. On the other hand, because cross-linking is accomplished on the shaped product in the presence of water (70–95 C), there is a certain flexibility in the decision making about the desired degree of cross-linking. In fact, one may interrupt the process at a certain time to control the material rigidity. There are two main processes for cross-linking PE with moisture-cured vinyl silanes: 1. One-step (Monosil) process, which involves a continuous feeding of liquid silanes during extrusion. The extruder is equipped with a long barrier screw and an injection system. This speeds up production but poses safety problems owing to the volatile and flammable nature of silanes. 2. Two-step (Sioplas) process, which involves a step to prepare cross-linkable PE and another step to create the final product. Step 1: A silane blend containing peroxide is melt processed with PE and formed into pellets, which can be stored for several months. Such copolymers can be obtained by two methods: a. Reactor copolymers involving polymerization of ethylene with vinyl silane. Such technology

Step 2: The grafted PE pellets, combined with a catalyst masterbatch, is extruded and cured in a hightemperature water bath, or in a steam sauna, to reach the desired cross-linking level. Polyethylene can also be cross-linked in a single step by using dry silane masterbatches (a powder composed of porous polymer carriers) with the dry silane process.[4] Silane coupling agents possess the propensity to decompose in contact with atmospheric moisture and to produce hydrogen chloride, methyl alcohol, alkyl ethers of ethylene glycol, and other hydrolysis products. Under certain conditions autoignition can cause an explosion.[8] Thus, carefulness is essential in transportation, handling, and storage; personnel must wear protective equipment to avoid eye and skin contact. It is advised that silane coupling agents be kept away from fire and moisture.

Radiation Processes These processes do not require addition of any chemicals to the original PE compound. The main effect of radiation is triggering of the chemical ionization process, which results in irreparable damage to the life sustaining chemistry of living organisms and the initiation of cross-linking in polymeric materials. Thus, radiation, apart from improvement of polymer characteristics, is used also for sterilization purposes and is particularly convenient for the cross-linking of PE for medical implants. Ionizing radiation results in energy transfer to the electrons orbiting the polymer nuclei, which causes excitation of the PE molecules. Two effects may be observed:  Chain scission (radiolysis) with loss of hydrogen atoms and formation of free radicals; extensive chain scission entails reduction of molecular weight with material degradation.  Cross-linking with merging of two hydrogen atoms to form a hydrogen molecule and merging of two

C

580

Cross-Linked Polyethylene

polymer radicals (via formation of carbon-to-carbon bonds). Cross-linking transforms a linear network into a three-dimensional one with an increase of the molecular weight. Indeed, during radiation, both chain scission and cross-linking occur simultaneously and competitively.[9,10] The predominance of one, or the other, depends on many factors involving the sensitivity of the polymer to radiation effects (chemical yield), the irradiation dose, the dose rate, the presence of stabilizers or radical scavengers, steric hindrance effects, and the polymer radiation environment.[9] In the presence of oxygen, scission predominates over cross-linking, while in an inert environment, such as nitrogen, cross-linking predominates over scission. There are two basic radiation methods: gamma radiation (g) and electron beam radiation (EBR). The first involves exposure of products to a radioactive isotope as in Cobalt 60 (Co60) or Cesium 137 (Cs137). Electron beam radiation consists of bombarding the PE with high-energy electrons, which are accelerated to near light speed. Several kinds of accelerators are commercially available and classified as electrostatic direct current, electrodynamic direct current, radiofrequency linear accelerators, magnetic induction, and continuous wave machines. The target material is passed under the accelerator window by conveyors, reel-toreel equipment, or other handling means. The governing parameters are electron energy, pulse current, dimensions of the irradiation window, conveyor speed, dose rangeability, beam distribution, properties of the material such as density and hydrogen content, and the number of transits under the window.[11] The dose, D, or energy per mass unit, is calculated by the relationship: D ¼ K

I V

ð5Þ

with I being current intensity, V feed ratio, and K equipment factor. Radiation is potentially hazardous and requires ad hoc facilities and procedures. Both g-rays and electrons are highly penetrating; thus, a shielded environment with thick walls (i.e., 6 ft) constructed of high-density concrete and according to standard legislation is necessary. The product to be treated is loaded into carriers on a conveyor that passes into the cell through a labyrinth. Personnel in these facilities must comply with standard Health and Safety legislation. Cross-linking of thin-film PE can also be induced by excimer UV lamp irradiation, for coating purposes, to change a hydrophilic surface into a hydrophobic one. With this method a PE film can be bonded to another material without adhesive.

APPLICATIONS Cross-linked polyethylene is currently employed in civil engineering, electric=electronic fields, medicine, and the packaging industry, and as technology evolves it may be used in many other fields. However, each field has specific requirements and so a specific material must be fabricated with compliance to existing standards (i.e., ASTM F648 for Surgical Implants, ASTM F876 for PEX Tubing, ASTM F877 for PEX Hot and Cold Water Distribution Systems, ASTM D3555 for Wire and Cable Insulation, and so forth). Potentially, the thermomechanical characteristics of PEX can be tailored by appropriate selection of compound ingredients, cross-linking level, and sequence of manufacturing operations. Herein, the attention is focused on three main applications, pipes, electrical cable insulation, and medical implants.

Pipes Since the introduction of PEX in the early 1970s, plastic pipes became a viable alternative to traditional materials such as copper and clay and cement. Crosslinked polyethylene was immediately considered as a good compromise between safety and economy. On the one hand such material has high resistance to chemical and electrochemical corrosion (due to a bad electric conductor), low encrustation tendency and low head losses, long-term pressure resistance, noise dampening-elastic properties. On the other hand it is light, flexible, and easy to transport and install because it can be delivered in coils; this also minimizes the number of joints and, in turn, the leakage rates.[6,12] Polyethylene pipes are grouped in classes according to their lifetime rating. The currently used classes are PE80 and PE100, which should withstand a hoop stress of 8 and 10 Mpa, respectively, for 50 yr at room temperature and include a safety factor that accounts for the variability of the working conditions. Now, other classes are under development.[12] Cross-linked polyethylene pipes can fail in a ductile way under high stress, or in a brittle way at low stress. The second failure is caused by slow crack growth (SCG), which represents the weakness of PEX pipes and is most probably induced by defects, or solid particles, such as residual catalyst, attached to the inner pipe surface.[13] To improve stress cracking resistance, a third-generation PE, also called bimodal PE, has been developed.[13] A desired molar weight distribution is obtained with a combination of fractions of branched high- and linear low-molecular-weight PEs. The highermolecular-weight fraction enhances strength while the linear lower-molecular-weight fraction facilitates processing.

Cross-Linked Polyethylene

Pipes may be susceptible to gas permeability, which can lead to two adverse effects. One concerns permeability to oxygen, which, in closed circuits such as underflow heating, or radiation heating, causes the steel components to corrode. The other drawback is with regard to the fouling of potable water by carbon dioxide, nitrogen, and other gases from contaminated soil. A possible solution may be the adoption of multilayer structures involving a copolymer of ethylene and vinyl alcohol between two PEX layers. Thus, gas permeation is avoided by the hydrophilic barrier while permeation to liquids is prevented by the hydrophobic PEX.

Electrical Cable Insulation Application of PE in the insulation of electrical cables requires giving the material specific characteristics such as thermal stability under load, heat resistance, and heat shrinkability. This is attained through four main steps: compounding, forming, cross-linking, and expansion. In particular, fabrication of heat shrinkable cables is based on the shape-memory phenomenon. The technological process consists of conveying elongational stresses into the preform by expanding and then cooling it under stress.[14] The frozen stresses are released when the cable is reheated (during application) causing shrinkage. Generally, both LDPEs and HDPEs are used for cable insulation. When PEX was first introduced in the electric cable insulation field (almost 40 yr ago) it appeared as the solution to many problems because of its excellent dielectric strength, low dielectric permittivity, low loss factor, high resistance to chemicals, and good mechanical properties. Indeed, substitution of impregnated paper and mineral oil filled cables with PEX for insulation of underground high-voltage electric wires offers many advantages in manufacturing, transportation, installation, and environmental benefits because of the elimination of oil leakage. Unfortunately, many premature failures occurred before the expected duration of 40 yr. The failures were caused by electrical treeing initiated by water treeing. This problem raised a question of great concern among power utilities worldwide. Many studies have been developed to understand the complex mechanism of tree inception and propagation with insulation breakdown, and to search for any solution without changing material (i.e., without discarding PEX). It is supposed that, under normal operating voltage stress, breakdown channels originate at the tip of defects present in the form of microvoids, gas cavities, conducting inclusions, or intrusions in the insulation structure.[15,16] Water trees are then supposed to consist of tracks of oxidized (hydrophilic) polymer within the hydrophobic PEX. Such tracks do not cause insulation failure but under favorable conditions (lightning surges),

581

evolve into an electrical tree with deleterious consequences.[16] The growth of water trees may be contrasted by hydrophilic clusters, but additives, while improving water tree resistance also reduce hydrophobic and dielectric characteristics. Thus, technology is still being sought to increase tree resistance without affecting valuable PEX characteristics.

Medical Implants Ultra-high-molecular-weight polyethylene (UHMWPE) was introduced in hip arthroplasty in the early 1960s by the English surgeon Sir John Charnley as a solution to severe hip arthritis and is still the current material of choice for the bearing surface in total joint replacement prostheses. The majority of total hip prostheses implanted in the following three decades have included an acetabular UHMWPE cup articulating against a femoral ball of cobalt–chromium alloy. Wearing of UHMWPE components was observed with undesirable biological response and in 1994 the National Institutes of Health (NIH) officially assessed the tissue inflammation and bone resorption (osteolysis) as caused by the PE wear debris.[17,18] In 1998, the development of a highly cross-linked polyethylene covered by a U.S. patent was a milestone in arthroplasty.[19] The benefits in wear resistance have led to a proliferation of cross-linking technology into hip and knee replacements. Chemical cross-linking typically involves the generation of noxious fumes and sensitizing by-products of peroxide degradation. Thus, almost all manufacturers use ionizing radiation. Trade names like CrossfireTM (75 kGy gamma radiation and 25 kGy gamma radiation sterilization under nitrogen environment), DurasulTM (95 kGy EBR and ethylene oxide sterilization), LongevityTM (100 kGy EBR and gas plasma sterilization), and MarathonTM (50 kGy gamma radiation and gas plasma sterilization) are commercially available. It has to be noted that in addition to potentially improving the wear resistance, cross-linking can adversely affect other physical properties such as tensile strength and elongation with susceptibility to crack formation at the articulating surface.[20,21] Another great problem is oxidation induced by the free radicals, which remain entrapped within the crystalline phase of the UHMWPE during radiation. To minimize such undesirable effects it was suggested to remelt the material after gamma irradiation to extinguish the residual free radicals and then to sterilize with ethylene oxide or gas plasma to avoid the reintroduction of free radicals that would occur during radiation sterilization.[20] Another method was warm irradiation with adiabatic melting to make cross-linked UHMWPE with high-energy (10 MeV, 40 kW) EBR without sacrificing the mechanical properties.[21]

C

582

Cross-Linked Polyethylene

The method suitability was assessed through simulator studies that otherwise seemed incapable of predicting the in vivo performance.[22] In fact, there are many factors, apart from the material properties and manufacturing processes, such as the surgical technique and the patient response, which complicate the determination of a cause–effect relationship. Such considerations prompted in June 1999 a Safety Notice from the Medical Devices Agency (MDA SN1999) cautioning careful monitoring of postimplant patients with highly cross-linked UHMWPE components.[23] Despite many studies concerning material improvements in total joint replacement, failures continue to occur and some questions remain about the UHMWPE cross-linking level and procedure to prevent wear. Of course, the patient’s response is a primary factor, and ascertaining the benefits of a kind of prosthesis will need several years of clinical data.

TESTING A product of given characteristics can be obtained by mixing pellets of different polymers, additives such as colorants, stabilizers, antioxidants, flame-retardants, and so forth. Thermomechanical modifications can be brought about either by silanes or peroxides or by exposure to ionizing radiation. A product of good quality is certainly the result of experience, ability, and good knowledge of the interactions between the different substances, and also characterization of final products with the most effective techniques is of vital importance. Many techniques, collected in Table 1, are available for the characterization of PEX.[23–29]

Chemical Analysis This analysis consists basically of the evaluation of chemical modifications induced in PE by cross-linking methods. The most important parameter to measure is the gel content (fraction or gel%), which indicates the cross-linking degree. It is generally measured, following the ASTM D 2765 standard, as the percentage of the original weight of a sample after extraction for 24 hr in boiling toluene (or xylene) and successive drying in a vacuum oven at 90 C. The variation of gel% with curing time (silane cross-linking) and irradiation dose (EBR) for LDPE is shown in Fig. 1. Cross-linking of 70% can be obtained with exposure to a radiation dose of 90 kGy, or 50 hr curing in hot water with silane. However, the main difference between the two methods is that in EBR the cross-linking degree can be easily increased by increasing the dose. Instead, in the silane method, lengthening curing has no significant effect; an increase in the cross-linking degree requires the addition of an increased percentage of silane to the compound. Another important parameter is the swelling ratio. A sample is placed in a solvent and the variation of its height is monitored as a function of time through a contact probe. A relationship between sample and solvent allows for the evaluation of the cross-link density and the chain length between cross-links. The technique is named swell ratio testing (SRT) in the American ASTM F2214-02, or hot set test (OST) in the German DIN 57472 Standard.[23,24] Of vital importance, especially for medical implants, is quantifying the presence of free radicals, which can react with oxygen forming carbonyls and cause material

Table 1 A collection of techniques useful for the characterization of PEX Chemical analysis

Mechanical analysis

Physical analysis

Thermal analysis

Nondestructive evaluation

Gel content measurements ASTM D2765

Compression tests ASTM D2990

Specific gravity determination

Dynamic mechanical analysis (DMA)

Ultrasonics

Fourier transform infrared spectroscopy (FTIR) ASTM E1421

Elongation tests

Scanning electronic microscopy (SEM)

Differential scanning calorimetry (DSC) ASTM D3417

Scanning acoustic microscopy (SAM)

Resonance spectroscopy (RS)

Fatigue tests ASTM E647

Transmission electronic microscopy (TEM)

Thermogravimetric analysis (TGA)

Photothermal radiometry

Swell ratio testing (SRT) ASTM F2214-02 Hot set test (OST) DIN 57472

Small punch analysis

Thermomechanical analysis (TMA)

Dielectric spectroscopy

Trace element analysis (TEA) ASTM F648

Shore hardness testing (SHT) ASTM D2240 Tensile tests ASTM D2990, D638

Partial discharge (PD)

Cross-Linked Polyethylene

583

A novel method is the small punch analysis, or pin on disk test, which is used to evaluate the weight loss of friction material.[10] It is performed with a metallic pin in friction contact on a sample (small disk); the pin, moving cyclically, yields a stress–strain curve related to the specimen wear. This test is generally performed on retrieved, or aged, UHMWPE components.

Physical Analysis The density, rp, or specific gravity, can be evaluated by applying the Archimedean principle in water:[24] rp ¼ rw

Fig. 1 Variation of gel content with curing time and irradiation dose.

embrittlement. This is done with resonance spectroscopy. The chemical structure of a molecule can also be analyzed by Fourier transform infrared spectroscopy. Many radiolytic products are visible in the infrared spectrum and can be detected by an infrared detector like the mercury–cadmium–telluride detector. The absorbance is proportional to the concentration of the chemical species active at the selected frequency. Again, for medical implants a trace element analysis is performed to ensure absence, or allowed percentages, of some substances like titanium, calcium, chlorine, etc.

Mechanical Analysis The mechanical properties, such as tensile, compression, shear, and fatigue strengths are fundamental for the product’s lifetime. Each product must be tested by taking into account the effective operative stresses. A hip cup should allow a patient to take as many as a million steps a year. Pipes and cable insulation for underground conveyance must perform properly in severe soil conditions and seismic zones; pipes should resist scratches and stresses due to high pressure and temperature. Tensile, compression, elongation, and fatigue tests are commonly performed according to well-traced guidelines to evaluate the mechanical characteristics of materials and their suitability for different applications. The hardness is generally evaluated with the ProfilerP10 and with imprints obtained statically and dynamically. The hardness coupled with mechanical properties is helpful in predicting the material’s response to the interaction with other objects, or substances.

Wpa Wpa
Wpw

ð6Þ

where rw is the density of water and Wpa and Wpw are the weight of the polymer in air and water, respectively. Morphological modifications arising through crosslinking can be observed by scanning electronic microscopy (SEM). Three images are shown in Fig. 2 for pure LDPE (A), silane grafted and cured in hot water for 80 hr (B), and electron beam irradiated at a dose of 80 kGy (C). A fragment of each material was gold-coated and viewed by SEM. Changes in crystallinity can be observed by transmission electronic microscopy.

Thermal Analysis Several methods may be used to analyze the material behavior under controlled temperature variations. Differential scanning calorimetry (DSC) is the most utilized thermal analysis technique. It consists of observing and recording (thermogram) exothermic and endothermic phenomena that occur in a sample sealed in an aluminum sample chamber under temperature variations. It is possible to evaluate glass transition temperature (Tg), melting point (Tm), crystallinity degree from the fusion enthalpy, and crystallization temperature (Tc). Typical outputs are shown in Fig. 3 for thermo shrinking insulation of low-voltage cables cross-linked via silane grafting and curing for 91 hr in hot water (A) and electron beam irradiation at a dose of 129 kGy (B). The different heat flow levels are linked to the different cross-linking mechanisms. In fact, irradiation was performed at ambient temperature while the material was in the solid state; instead, curing in hot water, when the material was not completely solid and had poor structural stability, probably led to the formation of molecular defects. Indeed, the response to DSC measurements is strictly related to molecular structure; a low-density material behaves differently from a high-density one.[25]

C

584

Cross-Linked Polyethylene

Fig. 2 Scanning electron microscopy images: (A) pure LDPE; (B) silane grafted and cured in hot water; and (C) electron beam irradiated.

The dynamic mechanical analysis gives detailed information about the viscoelastic properties of a sample when heated, cooled, or held under isothermal conditions. The three a, b, and g peaks displayed by the material before melting can be used to evaluate the effects on the PE molecular structure of additives

and cross-linking methods. With the thermogravimetric analysis, information about weight loss, degradation onset temperature, and degradation rate of a sample when heated, or held isothermally, in nitrogen (or argon) atmosphere is attained. The dimensional variations of a sample when it is heated, cooled,

Fig. 3 Differential scanning calorimetry profiles: (A) electron beam irradiation and (B) silane grafting and curing in hot water.

Cross-Linked Polyethylene

585

or held under isothermal conditions are observed by the thermomechanical analysis technique.

C

Nondestructive Evaluation Cross-linked polyethylene may include impurities and voids from which the major causes of premature failures (SCG in pipes, treeing in electric cables, and cracking in medical implants) can originate. Thus, nondestructive evaluation with effective techniques that can discover defects at the incipient stage, before the component is put in operation, is of vital importance. Conventional ultrasound (up to 10 MHz), which is the current technique for detection of flaws in metal piping and vessels, is limited by the attenuated nature of polymers. A scanning acoustic microscope with operating frequencies up to 100–150 MHz has been found to be more effective.[27] Another technique that has proved to be practical for the evaluation of PEX is photothermal radiometry, commonly known as lockin thermography, which is the multiplexed version. Studies present in the literature demonstrate the success of such a technique when investigating the effect of molecular orientation in stretched PE, the loss of mechanical properties, local material inhomogeneities due to extrusion and crosslinking processes, as well as material differences linked to the different compounds.[28,29] The test procedure consists of acquiring phase (or amplitude) images while the specimen surface is thermally stimulated with a sinusoidal heat flux. Two phase images, taken at a heating frequency f ¼ 0.12 Hz, are shown in Fig. 4 for LDPE cross-linked with EBR. The first image (A) shows a piece of commercial (Megarad, Italy) lowvoltage electric cable insulation after expansion to three times its original diameter; it is possible to see yield tracks that are probably due to local material inhomogeneities or nonuniform distribution of forces during enlargement. The second image (B) refers to a sheet irradiated at increasing dose from right to left; it is possible to see dark zones on the left due to material degradation under overdose. Lockin thermography can be exploited for characterization of PEX through the variation of the phase angle. In fact, the local phase angle variation is well correlated to the local variation of the elastic modulus under variation of irradiation dose, as shown in Fig. 5; more specifically, a differential phase angle (with respect to the phase angle of untreated material) is considered. The modulus of elasticity is evaluated at 150 C above the melting point to relate it directly to the cross-linking degree.[26] Through the variation of the phase angle the variation of the material density can also be evaluated as demonstrated in Ref.[28] for a retrieved UHMWPE hip cup after 9 yr of implantation.

Fig. 4 Phase images for f ¼ 0.12 Hz. (A) post expansion cable insulation; and (B) sheet irradiated at increasing dose from 90 kGy (right) to overdose (left). (View this art in color at www.dekker.com.)

The electric utility industry is today forced by the Market Directive to adopt condition-based maintenance to ensure steady supply. The method universally adopted to ensure proper installation of cable system accessories and to periodically determine the state of the insulation for decision making about repairing or replacing before failure occurs is partial discharge (PD). This method is based on the time-of-flight principle of induced PD signals; a time variation is attributed to local faults. Calibration is needed to establish a correct relationship between the magnitude

Fig. 5 Differential phase angle against modulus of elasticity for LDPE for varying the electron beam dose.

Fig. 6 A summary of main points in three PEX applications.

586 Cross-Linked Polyethylene

Cross-Linked Polyethylene

of the discharge in the cable sample and the signal received; quantitative measurements are complicated by interference and noisy environments. Recently, a new on-site and on-line calibration method, which allows overcoming of the problems of previous methods, has been developed.[30] The insulation deterioration for water treeing is associated with a change in dielectric properties such as loss factor (tan d) and capacitance to be measured with dielectric spectroscopy.

CONCLUSIONS The benefits of PEX over other plastics has led to a proliferation of cross-linking technologies and, of course, each manufacturer claims the superior quality of its product. Indeed, PEX can be advantageously used in many applications owing to caution in:  Selection of the compound ingredients; fractions of LDPEs and HDPEs; and specific additives such as stabilizers, antioxidants, flame-retardants, and so forth.  Selection of the most suitable cross-linking method (peroxide, silane, or radiation) that gives the material the thermomechanical properties adequate for the specific application.  Complying with local and international standards (DIN, UNI, UNE, BS, ASTM) regarding mechanical and physical properties to ensure a safe life without unexpected failure and meet hygienic requirements (ANSI, NSF) to avoid health effects. Application of PEX in three main fields: piping, cable insulation, and medical implants has been examined. Many premature failures of PEX products in each field were reported in the literature; the origin was recognized in the presence of defects forming during fabrication. A summary with advantages and caution notes is sketched in Fig. 6. It seems that production standards are established mainly on the basis of destructive tests and statistical inference with little attention to in-process monitoring and nondestructive evaluation. In this context, infrared thermography, as a remote imaging system of temperature mapping and nondestructive evaluation, may be advantageously exploited.

ACKNOWLEDGMENTS The authors wish to express their gratitude to: Megarad s.r.l. (Italy) for supplying specimens and related information, Dr. Paolo Suriano for assistance in performing DSC and mechanical tests, and Dr. Benedetto De Vito for the SEM images.

587

REFERENCES 1. Gibson, R.O. The discovery of polythene. R. Inst. Chem. Lect. Ser. 1964, 1, 1–30. 2. Smedberg, A.; Hjertberg, T.; Gustafsson, B. Effect of molecular structure and topology on network formation in peroxide cross-linked polyethylene. Polymer 2003, 44, 3395–3405. 3. Anbarasan, R.; Babot, O.; Maillard, B. Crosslinking of high-density polyethylene in the presence of organic peroxides. J. Appl. Polym. Sci. 2004, 93, 75–81. 4. www.specialchem4polymers.com (accessed Jun 2004). 5. Shah, G.B.; Fuzail, M.; Answar, J. Aspects of cross-linking of polyethylene with vinyl silane. J. Appl. Polym. Sci. 2004, 92, 3796–3803. 6. Unidelta S.p.A. Crosslinked polyethylene pipes. In Technical Handbook T4001; 2000. 7. PolyOneTM Corporation. Technical Report N. 66; 2002. 8. Tamanini, F.; Chafee, J.L.; Jambar, R.L. Reactivity and ignition characteristics of silane=air mixtures. Process Saf. Prog. 1998, 17 (6), 243–258. 9. Ivanov, V.S. Radiation Chemistry of Polymers; VSP: The Netherlands, 1992. 10. Valenza, A.; Visco, A.M.; Torrisi, L.; Campo, N. Characterization of ultra-high-molecularweight polyethylene (UHMWPE) modified by ion implantation. Polymer 2004, 45, 1707–1715. 11. Becker, R.C.; Bly, J.H.; Cleland, M.R.; Farrell, J.P. Accelerator requirements for electron beam processing. Rad. Phys. Chem. 1979, 14, 353–375. 12. Scheelen, A. Recent Developments in PE Pipe Materials with Outlook to the Future, Technical Report; Solvay Polyolefins Europe: Belgium. 13. Hubert, L.; David, L.; Se´gue´la, R.; Vigier, G.; Degoulet, C.; Germain, Y. Physical and mechanical properties of polyethylene for pipes in relation to molecular architecture. I. Microstructure and crystallisation kinetics. Polymer 2001, 42, 8425–8434. 14. Morshedian, J.; Khonakdar, H.A.; Mehrabzadeh, M.; Eslami, H. Preparation and properties of heat-shrinkable cross-linked low-density polyethylene. Adv. Polym. Technol. 2003, 22 (2), 112–119. 15. Boggs, S.; Xu, J. Water treeing-filled versus unfilled cable insulation. IEEE Electrical Insulation Mag. 2001, 17 (1), 23–29. 16. Sarathi, R.; Das, S.; Kumar, C.R.A.; Velmurugan, R. Analysis of failure of crosslinked polyethylene cables because of electrical treeing: a physicochemical approach. J. Appl. Polym. Sci. 2004, 92, 2169–2178.

C

588

17. Hovie, D.W. Tissue response in relation to type of wear particles around failed hip arthroplasties. J. Arthropl. 1990, 5, 337–348. 18. Total Hip Replacement NIH Consensus Statement, Sep 12–14, 1994; Vol. 12 (5), 1–31. 19. Non-Oxidizing Polymeric Medical Implant U.S. Patent 5,414,049, May 9, 1998. 20. McKellop, H.; Shen, F.W.; Lu, B.; Salovey, R.; Campbell, P. The effect of sterilization method and other modifications on the wear resistance of acetabular cups of ultra-high molecular weight polyethylene. A hip simulator study. J. Bone and Joint Surgery 2000, 82 (12), 1708–1725. 21. Muratoglu, O.K.; Bragdon, C.R.; O’Connor, A.S.; Jasty, M.; Harris, W.H. A novel method of cross-linking ultra-high molecular-weight polyethylene to improve wear reduce oxidation, and retain mechanical properties. J. Arthoplasty 2001, 16, 149–160. 22. Greenwald, A.S.; Bauer, T.W.; Ries, M.D. New polys for old: contribution or caveat? J. Bone Joint Surg. 2001, 83 (2), 27–31. 23. Spiegelberg, S. Analytical techniques for assessing the effects of radiation on UHMWPE. Society for Biomaterials Annual Conference, St. Paul, MN, 2001. 24. Khonakdar, H.A.; Morshedian, J.; Wagenknecht, U.; Jafari, S.H. An investigation of chemical

Cross-Linked Polyethylene

25.

26.

27.

28.

29.

30.

crosslinking effect on properties of high-density polyethylene. Polymer 2003, 44, 4301–4309. Valle´s-Lluch, A.; Contat-Rodrigo, L.; Ribes-Greus, A. Differential scanning calorimetry studies on high- and low-density annealed and irradiated polyethylenes: influence of aging. J. Appl. Polym. Sci. 2003, 89, 3260–3271. Meola, C.; Nele, L.; Giuliani, M.; Suriano, P. Chemical and irradiation crosslinking of polyethylene. Technological performance over costs. Polym. Plast. Technol. Eng. 2004, 43 (3), 629–646. Avila, S.M.; Horvath, D.A. Microscopic void detection as a prelude to predicting remaining life in electric cable insulation. International Topical Meeting on Nuclear Plant Instrumentation, Controls and Human-Machine Interface Technologies (NPIC&HMIT 2000), Washington, DC, Nov 2000. Busse, G.; Eyerer, P. Thermal wave remote and nondestructive inspection of polymers. Appl. Phys. Lett. 1983, 43, 355–357. Meola, C.; Carlomagno, G.M.; Prisco, U.; Vitiello, A. Non-destructive control of polyethylene blanket insulation. Res. Nondestr. Eval. 2004, 15 (2), 55–63. Zhong, L.; Chen, G.; Xu, Y. A novel calibration method for PD measurements in power cables and joints using capacitive couplers. Meas. Sci. Technol. 2004, 15, 1892–1896.

Crystal Growth C C. W. Lan Department of Chemical Engineering, National Taiwan University, Taipei, Taiwan, ROC

W. C. Yu Department of Molecular Science and Engineering, National Taipei University of Technology, Taipei, Taiwan, ROC

W. C. Hsu Sino-American Silicon Product Inc., Hsinchu, Taiwan, ROC

INTRODUCTION Crystal growth is not only a physical process that is interesting to science, but also a technologically important subject in the chemical and material industry. In this entry, we give an introduction to crystal growth physics including the thermodynamics and kinetics for nucleation and growth. Growth mechanisms, surface roughing, and growth inhibitions are also reviewed. Finally, the technologies of crystal growth and the current issues are also discussed.

BACKGROUND Crystals are solids in which molecules are arranged in a regular, repetitive three-dimensional (3-D) pattern, and crystal growth is a process of building up this structure. In nature, the building units (unit cells) of the crystals are divided into only seven kinds, the so-called seven crystal systems. The molecules can be placed at cell corner, face, or center. Hence, a total of 14 so-called Bravais lattices can be produced. Because of the repetitive nature of the unit cells, crystals can take on characteristic and interesting forms, for example, the colorful and beautiful shine of various gem stones like diamond, ruby, and sapphire. Although the basic structure unit determines the intrinsic properties of a material, the ultimate form and properties of the crystal are determined by the growth process. For example, the formation of beautiful snowflakes can have all kinds of morphologies, and this is controlled by growth conditions. Even in biology, the formation of structures of oyster shells, corals, ivory, and bones is determined by crystal growth, which has evolved to a well-known field called biomineralization.[1,2] Crystal growth is not only a physical process that is interesting to science, but also a technologically important subject in the chemical and materials industry. Crystals, especially semiconductor single crystals, are Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120030516 Copyright # 2006 by Taylor & Francis. All rights reserved.

important materials in modern technology since the invention of transistors at Bell Laboratories.[3–8] Most of the modern electronic, optical, and optoelectronic devices, as well as integrated circuits and optics, are built upon on the substrate sliced from a single crystal. The miniaturization of the devices and circuits requires large and perfect single crystals, and this has promoted the fast development of crystal growth technology in the past two decades. Crystal production globally is estimated at more than 200,000 tons=yr, of which the largest fraction of about 60% is the semiconductors silicon, GaAs, InP, GaP, and its alloys. In fact, for the solar cells alone, more than 10,000 tons of silicon per year has been used globally. Quartz is the second largest commodity crystal and its production rate is more than 300 tons=yr in the world. Table 1 gives a brief view of the typical single crystals available in the market including their applications and growth methods, which will be discussed later.[3–6] Besides quartz, other oxide crystals such as LiNbO3 and LiTaO3 play an important role in wireless communication, particularly in the surface acoustic wave (SAW) filters.[9] Even for sapphire crystals, their use as substrates for the growth of GaN film used in blue and white light emitting diodes (LEDs) has increased dramatically in recent years.[10,11] Besides the bulk single-crystal growth, it is one of the myths of the thin-film technologies that the crystals are grown layer by layer on an oriented substrate, the so-called epitaxy. The epitaxial growth, especially from the vapor phase, has been routinely used for making LEDs, diode lasers, and quantum-well devices.[12] Crystal growth starts with the first-order phase transition from an either homogenous or heterogenous nucleation, followed by the construction (growth) of surfaces and morphologies. The growth can be carried out by solidification from a melt, growth from a supersaturated solution, condensation from a vapor phase, or grain growth from solid phases. In terms of thermodynamics, as illustrated in Fig. 1, the crystallized phase 589

590

Crystal Growth

Table 1 Typical commercial crystals: Their market share, applications, and growth methods Crystals (% market share)

Examples

Applications

Growth methods

Semiconductor (60%)

Si, SiGe, GaAs, InGaAs, InP

ICs, transistors, solar cells, diodes, etc.

Czochralsiki (Cz), Bridgman one melting

Scintillation crystals (12%)

Bi4Ge3O12, Lu2SiO5, BaF2, CaF2, PbWO4, Tl:CsI, Na:CsI, NaI

Radiation detectors, positron emission tomography (PET)

Cz, Bridgman

Optical crystals (10%)

Sapphire, CaF2, BaF2, MgF2, MgO, quartz, Si, Ge, ZnSe, ZnS, CaCO3, YVO4, LiNbO3, a-BaB2O4 (BBO), Y3Al5O12 (YAG)

Substrates, lenses, mirror, prisms, wave plate, polarizer, etc.

Bridgman, Cz, Kyropoulos

Acoustic crystals (10%)

LiNbO3, LiTaO3, quartz

SAW filter, sensor, etc.

Cz, Bridgman

Nonlinear optics and laser crystals (5%)

YAG, KH2PO4 (KDP), Bi12SiO20 (BSO), KTiOAsO4 (KTA), b-BBO, LiB3O5 (LBO), KTiOPO4 (KTP), AgGaS2, ZnGeP2, Mg:LiNbO3, Nd:YVO4, Nd:YAG, Nd:GdVO4, Ti:sapphire

Laser applications, optical communication, etc.

Cz, solution growth

Jewelry (3%)

Sapphire, ruby, amber, MgAl2O4, CaCO3

Decoration, watch window

Cz, Vernuil

should be more stable. In other words, the solution phase has a higher Gibbs free energy because of supersaturation or undercooling. The starting phase can be different for the same crystal. For industrial applications, growth from the melt remains the major method for semiconductors and oxides. If the molten state of the material is not thermodynamically stable or an undesirable phase is expected to be encountered during cooling down, growth from a solution or a vapor phase at a lower temperature is preferred. For a material having high vapor pressure, vapor growth can also be a convenient approach. For the growth of a single crystal, a seed is often required to avoid parasitic nucleation. Crystal growth is also a convenient way for purification, which is known as recrystallization, because of the different solubility of a solute in the solution and the solid phases. In metallurgy, purification of metals by melting and solidification is a common process. Organic chemists prefer the recrystallization from solution. By repeating the solidification or recrystallization process several times, the purity of the material could be greatly improved.[13] The best example is the zone refining process, a technique of fractional crystallization developed by Pfann.[14] This economical and continuous process has been widely used in industry. Materials with purity up to 6 or 7 N have been routinely produced. Without these high-purity materials,

the modern semiconductor industry would not be possible. Furthermore, because of its purification (or concentration) nature, crystal growth or crystallization is essentially a chemical process for the production of a wide range of specialty chemicals and pharmaceuticals. This kind of application is usually referred to as industrial or mass crystallization, which concerns the

Fig. 1 The Gibbs free energy as a function of temperature for solid and fluid phases.

Crystal Growth

591

nucleation and growth of a huge number of small crystals with sizes around 200–1000 micron.[15,16] Crystal growth is a science of great depth and breadth, and the technology for crystal growth has been developed for a century. Many extensive texts have been written, e.g., Refs.[3,5–8]. Several journals, as well as handbooks, have been published on the subject.[17] Therefore, this entry attempts to give a brief overall view of the subject for beginners. We shall start with some fundamental concepts for crystal growth. Based on the point of view of thermodynamics and kinetics, we will discuss when and how the growth can proceed and what form the crystal can take. The effect of impurities is discussed briefly. Segregation for melt growth is also introduced, followed by the role of dopant mixing. Last, some typical growth methods used in laboratories and industries for growing single crystals are introduced. However, vapor growth, industrial crystallization, and biomineralization are not discussed here.

THERMODYNAMICS AND KINETICS Driving Force for Nucleation and Crystal Growth

Nucleation is the first step of the first-order phase transition, but a barrier exists in the formation of embryos for the new phase. The interfacial energy is always a positive term to destabilize the nuclei. But once the nuclei grow large enough, the free energy drops rapidly with the new phase formation. The critical size enabling the nucleus to dissolve or to grow is at the local maximum of the Gibbs free energy. A schematic of the Gibbs free energy as a function of the nucleus size is shown in Fig. 2. It should be pointed out that the interfacial energy is closely related to the geometry of the nucleus. With foreign particles or substrates, the interfacial energy between the nucleus and the foreign media can be significantly reduced.[19] This can be interpreted by the larger bonding energy between the nucleus and the foreign molecules than the salvation energy. On the other hand, if a latticematched substrate is used, the barrier for forming nucleus is significantly reduced, and this is the reason that the seeded growth can be carried out at a small supercooling or supersaturation. The classic nucleation theories give the critical radius with the following form for a spherical nucleus:[19] rc ¼ 2V g=DG

As mentioned for Fig. 1, the first-order transition from a fluid phase to a solid phase requires an excess Gibbs free energy DG, in other words, the supersaturation s or supercooling DT, for nucleation and the following growth. The driving force per unit area, f, is proportional to the Gibbs free energy difference per molecule as shown below: f ¼
DG=V ¼ kT lnð1 þ sÞ=V  ðkT=V Þs

Nucleation

ð2Þ

where g is the interfacial energy; again, DG measures the free energy for phase change. The corresponding nucleation barrier is given as the following: DGn ¼ ð16=3Þpg3 ðV =kTsÞ2  g3 =s2

ð3Þ

ð1Þ

where V is the volume per molecule, k the Boltzmann constant, T the temperature, and s the dimensionless supersaturation defined as s ¼ P=Psat
1 for the gas phase or C=Csat
1 for the solution phase. P is the partial pressure and C the solute concentration. For the growth of a single crystal in applications, the supersaturation is usually kept small to avoid parasitic nucleation. Therefore, ln(1 þ s) approaches s and the driving force is proportional to supersaturation as shown in Eq. (1). For melt growth, f  ðDH=V Þs where DH is the latent heat per molecule and s ¼ DT=Tm, the dimensionless supercooling.[18] When a driving force is large enough to overcome the interfacial energy for forming the crystal, nucleation occurs and crystal growth can then continue.

Fig. 2 Gibbs free energy as a function of crystal size during nucleation.

C

592

Crystal Growth

Or, in general DGn ¼ Bg3=s2, where B is a factor depending on the nucleus geometries. With this energy barrier, the steady-state nucleation rate can be obtained with the form of Jn ¼ A expð
DGn =kTÞ ¼ A exp½
Bg3 =ðkTs2 Þ

ð4Þ

where A is a constant.[19] For 2-D nucleation, Liu et al. have shown that the nucleation barrier is proportional to g2=s.[20] Direct validation of the nucleation theories is difficult because of lack of direct observation and measurements. However, in an experimental model system of a colloidal monolayer, the 2-D theory has been found to be in good agreement with the nucleation data.[21] Further discussion can be found elsewhere.[2,22] The induction period for nucleation that takes the reciprocal of nucleation rate is often used.[15,16] For single-crystal growth, a longer induction period indicates a more stable solution which makes the growth easier.

Growth Kinetics After nucleation, crystal growth continues from the nucleus surface. When the crystal surface is rough, the growth kinetics is usually simple because the growth sites are randomly distributed. The growth rate is in general proportional to the local driving force. However, in reality, the crystal surface consists of steps, with terraces and kinks, as well as adatoms and vacancies, as illustrated in Fig. 3. The preferred growth sites are the kinks due to more bonding to grasp the adatoms. The growth requires adsorption and then surface diffusion of atoms to the growth sites. The simple step model with surface diffusion can lead to a simple linear law for the growth, i.e., the growth rate being

Fig. 3 A simple surface model on the crystal surface.

linearly proportional to the supersaturation s, similar to the one in the rough surface formed by kinks.[18] However, when the surface is smooth at the end of step growth, the kinetics depends on the growth mechanisms. A new 2-D nucleation is a way to start the next layer, as illustrated by Fig. 4A. However, this often requires a substantial supersaturation (10–30%), which is less common. An outlook of an as-grown 4-N,N-dimethylamino40 -N-methyl-stilbazolium tosylate (DAST) crystal from methanol shows several nucleated crystals on the surface under large supersaturation (10%). Very often the growth can still continue even with only 1% or less supersaturation, and this is possible through the spiral growth of a screw dislocation.[23] Based on the spiral growth, Burton, Cabrera, and Frank proposed a model (referred to as the BCF theory) considering the transport of solute molecules from the bulk solution to the surface, and then surface diffusion to the kinks of the spiral generated steps from a screw dislocation.[23] A schematic of the spiral generated step is shown in Fig. 4B. Such a mechanism requires a much smaller driving force than the 2-D nucleation and is a favorable growth mode in nature. The BCF theory also predicts a quadratic kinetics at low supersaturation. The BCF theory has been quite successful in comparison with experimental observation and measurements (detailed discussions can be found in, e.g., Refs.[24,25].). For melt growth, the growth

Fig. 4 Growth mechanisms: (A) surface nucleation and (B) BCF spiral growth; an as-grown DAST crystal showing the 2-D nucleation is also illustrated.

Crystal Growth

rate based on screw dislocations is also quadratic, and confirmed by many examples. The typical growth curves based on different mechanisms are illustrated in Fig. 5, where the growth on the kinks or a rough surface is referred to as the adhesive growth. The consideration of multiple screws can lead to a more complicated kinetic behavior. Nevertheless, the BCF theory lays the foundation for crystal growth theory, and has a strong impact in metallurgy, materials science, and the semiconductor industry. Moreover, the steps and dislocations appear randomly on the crystal surface, their growth rates (flow) in a particular direction do not occur at the same speed. In some cases, the step growth is inhibited by impurities. Then, step bunching can occur and lead to macrosteps, as illustrated in Fig. 6, which can be visible up to several millimeters. An SEM image of the macrosteps on the surface of a DAST crystal is also illustrated in Fig. 6. Similarly, growth hillocks can also be generated from the spiral growth because the growth speed near the screw center is faster. Furthermore, the step growth can be affected further by fluid flow leading to the formation or dissipation of the kinematic waves, thus roughening or smoothening the crystal surface.[26,27] An extensive and thorough review of the growth mechanisms has been given by van der Eerden.[28] Recently, the use of atomic force microscopy in the observation of the growth steps has led to significant progress in understanding the growth mechanisms.[29,30] Some computer modeling studies have also been presented based on the various models, and the simulated results give a deeper insight of the growth dynamics, e.g., Refs.[31,32]. In the melt growth, the driving force is usually high so that the equilibrium growth habit is suppressed, and in most cases the interface is rough. Nevertheless, if the growth front is curved, and the thermal gradient is low,

Fig. 5 Growth rate dependence on the supersaturation for various growth mechanisms.

593

C

Fig. 6 Step bunching mechanism forming macrosteps; the photograph shows the typical macrosteps (several microns) on the surface of a DAST single crystal.

facets, as the result of step or spiral growth, may appear in some parts of the interface. Because the segregation in the rough and faceted surface is different, faceting can affect significantly the crystal quality and dopant segregation.

Growth Inhibition The step growth can be inhibited by many factors, especially by impurities and foreign particles. Several mechanisms exist, such as step pinning, incorporation, kink blocking, and surfactant effects.[33] Step pinning, as sketched in Fig. 7 (top), and its effect on the growth rate is also shown schematically. As shown, once the impurities are adsorbed on the surface, the step needs to grow around the pinning points. If the impurity level is low, the average pinning distance is much greater than the radius of curvature, and the step can advance freely. However, with a high impurity level, if the supersaturation is low, the growth can be stopped almost completely forming the so-called dead zone.[34] The dead zone size increases with the impurity concentration. The step spinning can change the face speed, and the impurity pinning can be quite selective to faces depending on the bonding nature. Therefore, the final crystal shape, as a result of face competition, can be changed as well. Growth inhibition is often observed in biomineralization.[2] Because the incorporation of extraneous ions

594

Crystal Growth

of the solution and the surface. The a factor is defined as: a ¼ ðDH=kTÞfk

ð6Þ

where DH is the latent heat of the phase change and fk ( t1), the lower growing planes will survive at the final shape. Again, when the growth of some planes is inhibited by impurities or foreign macromolecules, the crystal morphology can be affected significantly. A recent review by Winn and Doherty, though focused on organic solution growth, is a good reference to learn more about the prediction of crystal morphology.[38]

Fig. 9 A schematic sketch of distribution and segregation of impurities during normal freezing. The dashed lines show the final concentration distribution for the cases of complete and no mixings in the melt, respectively.

Segregation and Morphological Instability In most of the industrial applications, dopants are added into the crystals for tailoring their electrical or optoelectronic properties. However, because of the different dopant solubility in the crystal and in the liquid, dopant segregation (inhomogeneity) during growth is inevitable. This is a problem particularly for a batch growth process. Let us take the normal freezing, as illustrated in Fig. 9, as an example. The crystal growth starts from one end to the other by solidification. With an initial dopant concentration in the melt at C0 and no solid-state diffusion, the axial dopant segregation with complete dopant mixing (without solid-state diffusion) can be described by the Schiel equation:[39,40] Cs ¼ C0 ð1
fs Þk
1

ð9Þ

where fs is the fraction of solidification and k the segregation coefficient. The segregation coefficient k is the ratio of the dopant solubility in the solid and that in the melt. The schematic dopant distribution for k < 1 having complete dopant mixing in the solid after solidification is illustrated by the lower dashed line in Fig. 9. However, in practice, the dopant is not completely mixed in the melt. A dopant boundary layer is established in front of the interface during solidification, as illustrated by the solid line in Fig. 9. The dopant concentration profile for the case of no

mixing is illustrated by the upper dashed line. The Schiel equation is often used to fit the dopant profile in the crystal by using the segregation coefficient as a parameter, even though, the dopant is not well mixed. The fitting value is often referred to as the effective segregation coefficient keff. Burton et al. further proposed a simple model (the so-called BPS theory) to correlate the effective segregation coefficient with convection, which is characterized by a boundary layer thickness d, as:[41] keff ¼

k k þ ð1
kÞe
V d=D

ð10Þ

where V is the solidification speed and D the dopant diffusivity in the melt. Although the BPS theory is not quite correct for the closed system, it still gives a good physical insight for the dopant segregation under convection. Apparently, from Eq. (10), to improve axial dopant uniformity, one can increase the solidification speed or reduce dopant mixing (having a thicker boundary thickness), so that keff can be closer to unity. In other words, if the solidification distance is long enough, the axial dopant distribution can be rather uniform. Hence, the reduction of mixing, such as the growth in space, has been an active research topic in crystal growth. Suppressing the flow by magnetic

596

Crystal Growth

fields, and thus reducing axial segregation, has also been extensively used in practice.[42] Furthermore, because of the dopant segregation, the dopant boundary layer affects the solidification temperature (TL) in front of the growth interface. If the thermal gradient GT is smaller than the gradient of the solidification temperature at the growth front, constitutional supercooling can occur.[43] This can be described by: GT  m dC=dz

ð11Þ

where m is the slope of the liquidus temperature in the phase diagram. Once the supercooling is large enough to overcome the interfacial energy, the microscopically planar interface can break down into cellular cells or dendrites. Mullins and Sekerka gave an excellent analysis on the morphological instability, which predicts the onset wavelength that forms a steady cellular array.[44] Despite the segregation in the axial direction, segregation can also occur in the lateral (or radial) direction because of the nonuniform dopant boundary layer. The control of lateral segregation is extremely important in industry to ensure that the wafers cut from the ingot have a good uniformity, especially for large wafers. The interface shape and the convection in the bulk phase are the key factors affecting the lateral segregation. Extensive research efforts through computer modeling have been made to control the transport processes and the interface shape for improving crystal uniformity.[45,46]

Fig. 10 Stability diagram for solution growth.

CRYSTAL GROWTH METHODS There are many ways to grow crystals. However, the growth of a bulk single crystal remains a special interest to industry because of the need of single-crystal substrates for device applications. As mentioned previously, a good growth method provides uniform supersaturation or supercooling around the seed crystal, while the other place remains undersaturated or superheated to avoid parasitic nucleation. Some growth techniques have been widely adopted in industry. They can be divided into three categories, namely, the solution growth, melt growth, and vapor growth. Here, we give

Fig. 11 (A) Schematic sketch of hydrothermal growth of quartz crystals. (B) The crystal basket after growth. (Courtesy of Hantek Inc., Taiwan.) (View this art in color at www. dekker.com.)

Crystal Growth

a brief introduction to the solution growth and melt growth methods through their major applications. The discussion of the growth from the vapor phase can be found elsewhere.[5–7]

Solution Growth Growth of a bulk crystal from the solution is straightforward. A seed can be selected from the crystallization of a supersaturated solution by slow cooling or slow evaporation of solvent. Then, the growth of a bulk crystal can be carried out from the seed crystal by controlling the solution at the metastable region, as shown in Fig. 10; outside the metastable zone, the growth of crystals from the seeds only is not possible. To maximize the growth speed (production rate), the solubility needs to be as large as possible. This could be achieved by using either a better solvent or a higher temperature

597

or pressure. The growth of quartz in an autoclave is a typical example. Because of its piezoelectric properties and low cost, quartz, next to silicon, is the second most widely used electronic material. More than 300 tons of single crystals have been used each year. Even at high temperature and pressure in an autoclave, to have enough solubility, the natural quartz nutrient, lascas, needs to be dissolved by forming a complex in alkali metal hydroxide or carbonate aqueous solution. A schematic of the hydrothermal growth is illustrated in Fig. 11A. In practice, hundreds of thin plate seed crystals are placed in the growth zone for growth, and the growth takes several months. A crystal basket taken out from the autoclave after growth is shown in Fig. 11B. The inner volume of the autoclave is up to 3.5 m3. The temperature difference between the two zones is the chief process control for the growth rate, and the baffle in between is to reduce the thermal mixing of both zones. Two similar processes, the low and

Fig. 12 Major melt growth methods. (View this art in color at www.dekker.com.)

C

598

high pressure, are widely used in industry.[47] The low-pressure process operated at about 70–100 MPa uses up to 1.0 M sodium carbonate solution, and the temperature at the growth chamber is around 345 C with a temperature difference of 10 C from the nutrient chamber. On the other hand, the high-pressure process operated from 100 to 150 MPa uses up about 1.0 M sodium hydroxide solution and grows the crystal at 360 C. The growth rate is about 1 mm=day. High-temperature solution growth has also been adopted to grow many kinds of crystals, but often at normal pressure. An extensive introduction is given by Elwell and Scheel.[48] Besides the growth at high temperature, there are a number of important crystals that can be grown from low-temperature solutions. The most well-known one is the growth of potassium dihydrogen phosphate (KDP), an important nonlinear optical crystal for laser applications. Significant progress has been made because of the development of the fast growth process using a careful control of solution purity and particles.[49,50] By removing particles and clusters through reheating and filtration of the circulated solution, the growth rate up to 60 mm=day is possible for KDP.[51]

Melt Growth The growth of a bulk crystal from the melt is the fastest approach because of the higher growth temperature

Crystal Growth

and the larger driving force. Typical melt growth methods are illustrated in Fig. 12. The zone melting method at the top of the figure is more popular for purification instead of growing crystals. For purification, multiple zones are usually adopted.[14] The Bridgman method, both vertical and horizontal, has several unique advantages as compared with other methods. This method is simple and does not require skillful crystal growers. Hence, a fully automatic operation can be easily achieved. As illustrated in the figure, this technique melts the material in an ampoule and solidifies it directionally from one end to the other by moving the ampoule (or heating profile). Because a very low thermal gradient can be used, this technique is particularly useful for the crystals vulnerable to thermal stresses, such as III–V or II–VI compound semiconductors. Large-size (up to 8 in. diameter) GaAs single crystals with low dislocation density have been grown.[52] Lately, the etched-pit density (EPD) has been reduced to 104=cm2 for 8 in.-diameter crystals.[53] The horizontal configuration is less popular in applications, because the D-shaped crystals obtained from the horizontal boat are more difficult in postprocessing. The most popular technique in melt growth, as shown in Fig. 12 (bottom), is the Czochralski (Cz) method; the Kyropulos method is quite similar but the growth is proceeded by slow cooling. Fig. 13A shows a typical industrial puller for the growth of 8 in.-diameter silicon; a photograph of a silicon crystal after growth is also shown in Fig. 13B. The Cz

Fig. 13 (A) An outlook of a commercial puller having magnetic fields and recharge tank. (B) An as-grown 8 in.-diameter silicon single crystal. (Courtesy of Taisil Electronic Materials Inc., Taiwan.) (View this art in color at www.dekker.com.)

Crystal Growth

process was first introduced by Czochralski for the kinetic study of crystallization.[54] The development of the technology for semiconductors was credited to Teal and Little.[55] They started using a seed crystal to define the crystal orientation, and introduced the concepts of diameter control and dopant distribution by controlling the temperature and the crystal=crucible rotation. However, a dislocation-free single crystal was not possible until a necking procedure was proposed by Dash in 1958.[56] Before the growth, when the seed crystal is dipped into the melt, the thermal shock can generate defects, especially in the form of dislocations. The Dash necking is to let the seed grow with a small diameter, so that the dislocation lines can extend and end at the surface quickly. After some distance of necking, a dislocation-free crystal is obtained. Then, the shouldering procedure is to reach the full diameter. The body, the part that can be sliced into wafers, is pulled up to a desired length. To finish the growth, the end coning is required before detaching the crystal from the melt surface. Again, this is to minimize the thermal shock during detachment. Over three decades of development, dislocation-free silicon single crystals up to 16 in. in diameter have been grown,[57] while 8– 12 in.-diameter silicon crystals have been produced routinely for integrated-circuit applications. Besides silicon, the Cz method has also been used extensively for compound semiconductors, such as GaAs and InP. Up to 8 in.-diameter GaAs and 6 in.diameter InP ingots with an excellent quality have been grown, and are now available in the market.[58] Furthermore, many oxide crystals used in solid-state lasers, SAW devices, and nonlinear optics are often grown by this technique as well. Other examples are summarized in Table 1. It should be pointed out that the Czochralski method is a meniscus control process in which the crystal shape is determined by the meniscus. Therefore, the on-line control of the crystal diameter is possible by either monitoring the optical ring (mainly for silicon) of the meniscus or weighting the growing crystal. The control variable can be either the pulling speed or the melt temperature.

CONCLUSIONS In this article, we give a brief overall view of the crystal growth. The fundamental and application aspects are discussed briefly. In general, crystal growth starts from the nucleation of a thermodynamically stable or metastable phase. Once the driving force overcomes the interfacial energy for forming a new crystal, crystal growth can proceed, but is still limited by the supply of the nutrient through mass transport and the growth kinetics on the crystal surface. The impurities and

599

foreign particles play a crucial role on the growth kinetics and thus the final morphologies. Indeed, how the foreign molecules attach to the crystal surface is a chemistry issue. Therefore, pH, temperature, and solvent are effective parameters to modify the crystal morphologies. By choosing a proper template the metastable form of the crystals could be obtained. In industrial applications, crystal growth is also an important process to purify the materials. Several economical continuous processes have been adopted, such as zone refining, fractional recrystallization, sublimation, etc. Furthermore, the growth of a high-quality bulk crystal is important for many modern technologies. In addition to a proper growth method, high-purity raw materials and a clean growth environment, skillful techniques, and process control are necessary. Because the growth takes time, it always needs to compromise between the quality and the yield. The beautiful appearance of a crystal also requires a perfect structure for applications. The requirement of the crystal quality increases dramatically as the device size diminishes. The stringent specifications for the substrates continue to challenge crystal growers to optimize the process and develop new processes in the future.

REFERENCES 1. Mann, S. Biomineralization Principle and Concepts in Bioorganic Materials Chemistry; Oxford University Press: Oxford, 2001. 2. Dove, P.M.; DeYoreo, J.J.; Weiner, S. Biomineralization; Mineralogical Society of America: Washington, DC, 2003. 3. Scheel, H.J. Historical introduction. In Handbook of Crystal Growth 1a: Thermodynamics and Kinetics; Hurle, D.T.J., Ed.; North-Holland: Amsterdam, 1994; Chapter 1, 1–42. 4. Scheel, H.J. The development of crystal growth technology. In Crystal Growth Technology; Scheel, H.J., Fukuda, T., Eds.; John Wiley & Sons: New York, 2003, Chapter 1. 5. Laudise, R.A. The Growth of Single Crystals; Prentice-Hall: Englewood Cliffs, NJ, 1970. 6. Brice, J.C. Crystal Growth Processes; John Wiley and Sons: New York, 1986. 7. Pamplin, B.A., Ed.; Crystal Growth, 2nd Ed.; Pergamon Press Inc.: Elmsford, New York, 1980. 8. Hurle, D.T.J. Crystal Pulling from the Melt; Springer-Verlag: Berlin, 1993. 9. Campell, C.K. Surface Acoustic Wave Devices for Mobile and Wireless Communications; Academic Press: San Diego, 1998. 10. Talbot, D. LEDs vs. the light bulb. Technol. Rev. 2003, 106, 30–36.

C

600

11. Zorpette, G. Blue chips. Sci. Am. 2000, 282, 30–21. 12. Stringfellow, G.B. Organometallic Vapor-Phase Epitaxy, 2nd Ed.; Academic Press: San Diego, 1999. 13. Zief, M.; Wilcox , W.R., Eds.; Fractional Solidification; Marcel Dekker Inc.: New York, 1967. 14. Pfann, W.G. Zone Melting; John Wiley & Sons Inc.: New York, 1958. 15. Myerson, A.S. Handbook of Industrial Crystallization; Butterworth-Heinemann: Boston, 1993. 16. van der Heijden, A.E.D.M.; van Rosmalen, G.M. Industrial mass crystallization. In Handbook of Crystal Growth 2a: Basic Techniques; Hurle, D.T.J., Ed.; North-Holland: Amsterdam, 1994; Chapter 7, 315–415. 17. Hurle, D.T.J. Handbooks of Crystal Growth; North-Holland: Amsterdam three volumes, 1993. 18. Mutaftschiev, B. Nucleation theory. In Handbook of Crystal Growth 1a: Thermodynamics and Kinetics; Hurle, D.T.J., Ed.; North-Holland: Amsterdam, 1993; Chapter 4, 307–475. 19. Markov, I.V. Crystal growth for beginners: fundamentals of nucleation. In Crystal Growth, and Epitaxy; World Scientific: Singapore, 1995. 20. Liu, X.Y.; Maiwa, K.; Tsukamoto, K. Heterogeneous two-dimensional nucleation and growth kinetics. J. Chem. Phys. 1997, 106, 1870–1879. 21. Zhang, K.-Q.; Liu, X.Y. In situ observation of colloidal monolayer nucleation driven by an alternating electric field. Nature 2004, 429, 740– 743. 22. Liu, X.Y. Generic mechanism of heterogeneous nucleation and molecular interfacial effects. In Advances in Crystal Growth Research; Sato, K., Nakajama, K., Furukawa, Y., Eds.; Elsevier Science: Amsterdam, 2001; 42–61. 23. Burton, W.K.; Cabrera, N.; Frank, F.C. The growth of crystals and the equilibrium structures of their surfaces. Philos. Trans. R. Soc. Lond. A 1951, 243, 243–299. 24. Bennema, P; Glimer, G.H. Kinetics of crystal growth. In Crystal Growth; Hartmann, P., Ed.; North-Holland: Amsterdam, 1973; 263–327. 25. Chernov, A.A. Formation of crystals in solutions. Contemp. Phys. 1989, 30, 251–276. 26. Frank, F.C. Growth and Perfection of Crystals; Roberts, B.W., Turnbull, D., Eds.; John Wiley: New York, 1958; 411. 27. Bennema, P.; Glimer, G.H. Kinetics of crystal growth. In Crystal Growth; Hartmann, P., Ed.; North-Holland: Amsterdam, 1973; 263–327. 28. van der Eerden, J.P. Crystal growth mechanisms. In Handbook of Crystal Growth 1a: Thermodynamics and Kinetics; Hurle, D.T.J., Ed.; North-Holland: Amsterdam, 1993; Chapter 6, 307–475.

Crystal Growth

29. DeYoreo, J.J.; Orme, C.A.; Land, T.A. Using atomic force microscopy to investigate solution crystal growth. In Advances in Crystal Growth Research; Sato, K., Nakajama, K., Furukawa, Y., Eds.; Elsevier Science: New York, 2001; 361–380. 30. DeYoreo, J.J. Eight years of AFM: what has it taught us about solution crystal growth. 13th International Conference on Crystal Growth, Hibiya, T., Mullin, J.B., Uwaha, M., Eds.; Elsevier: Kyoto, Japan, 2001. 31. Vekilov, P.G.; Lin, H.; Rosenberger, F. Unsteady crystal growth due to step-bunching cascading. Phys. Rev. E 1997, 55, 3202–3214. 32. Kwon, Y.I.; Derby, J.J. Modeling of the coupled effects of interfacial and bulk phenomena during solution crystal growth. J. Cryst. Growth 2001, 230, 328–335. 33. DeYoreo, J.J.; Vekilov, P.G. Principles of crystal nucleation and growth. In Biomineralization; Dove, P.M., DeYoreo, J.J., Weiner, S., Eds.; Mineralogical Society of America, 2003; 57–114. 34. Voronkov, V.V.; Rashkovich, L.N. Step kinetics in the presence of mobile adsorbed impurities. J. Cryst. Growth 1994, 144, 107–115. 35. Pimpinelli, A.; Villain, J. Physics of Crystal Growth; Cambridge University Press: Cambridge, 1998. 36. Jackson, J.A. Mechanism of growth. In Liquid Metals and Solidification; American Society for Metals: Cleveland, OH, 1958; 174. 37. Bennema, P. Growth and morphology of crystals: integration of theories of roughening and Hartman–Perdok theory. In Growth Mechanisms: Handbook of Crystal Growth 1a: Thermodynamics and Kinetics, Hurle, D.T.J., Ed.; North-Holland: Amsterdam, 1993; 477–581. 38. Winn, D.; Doherty, M.F. Modeling crystal shapes of organic materials grown from solution. AIChE J. 2000, 46, 1348–1367. 39. Tiller, W.A.; Jackson, K.A.; Rutter, J.W.; Chalmers, B. The redistribution of solute atoms during the solidification of metals. Acta Metall. 1953, 1, 428–437. 40. Flemings, M.C. Solidification Processing; McGraw-Hill: New York, 1974. 41. Burton, J.A.; Prim, R.C.; Slichter, W.P. The distribution of solute in crystals grown from the melt. Part I. Theoretical. J. Chem. Phys. 1953, 21, 1987–1991. 42. Hurle, D.T.J.; Series, R.W. Use of a magnetic field in melt growth. In Handbook of Crystal Growth 2a: Basic Techniques; Hurle, D.T.J., Ed.; NorthHolland: Amsterdam, 1994; Chapter 5, 259–285. 43. Rutter, J.W.; Chalmers, B. A prismatic substructure formed during solidification of metals. Can. J. Phys. 1953, 31, 15.

Crystal Growth

44. Mullins, W.W.; Sekerka, R.F. Stability of a planar interface during solidification of a dilute binary alloy. J. Appl. Phys. 1964, 35, 444–451. 45. Brown, R.A. Theory of transport processes in single crystal growth from the melt. AIChE J. 1988, 34, 881–911. 46. Lan, C.W. Recent progress in crystal growth modeling and growth control. Chem. Eng. Sci. 2004, 59, 1437–1457. 47. Hervey, P.R.; Foise, J.W. Synthetic quartz crystal— a review. Miner. Metall. Process. 2001, 18 (1), 1–4. 48. Elwell, D.; Scheel, H.J. Crystal Growth from High Temperature Solutions; Academic Press: London, 1975. 49. Zaitseva, Z.P.; Rashkovich, L.N.; Bogatyreva, S.V. Stability of KH2PO4 and K(H,D)2PO4 in fast crystal growth rates. J. Cryst. Growth 1995, 148, 276–292. 50. Zaitseva, N.; Atherton, J.; Rozsa, R.; Carman, L.; Smolsky, I.; Runkel, M.; Ryon, R.; James, L. Design and benefits of continuous filtration in rapid growth of large KDP and DKDP crystals. J. Cryst. Growth 1999, 197, 911–920. 51. Yang, S.; Su, G.; Li, Z.; Jiang, R. Rapid growth of KH2PO4 crystals in aqueous solution with additives. J. Cryst. Growth 1999, 197, 383–397.

601

52. Rudolph, P.; Jurisch, M. Bulk growth of GaAs: an overview. J. Cryst. Growth 1999, 198=199, 325–335. 53. Borner, F.; Bynger, Th.; Eichler, St.; Flade, T.; Hammer, R.; Jurisch, M.; Kretzer, U. 2nd Asian Conference on Crystal Growth and Crystal Technology, 2002; Paper A.20. 54. Czochralski, J. A new method for the measurement of crystallization rate of metals. Z. Phys. Chem. 1918, 92, 219–221. 55. Teal, G.K.; Little, J.B. Growth of germanium single crystals. Phys. Rev. 1950, 78, 647. 56. Dash, W.C. The growth of silicon crystals free from dislocations. In Growth and Perfection of Crystals; Doremus, R.H., Roberts, B.W., David Turnbull, Eds.; John Wiley and Sons, Inc.: New York, 1958; 361–385. 57. Shiraishi, Y.; Takano, K.; Matsubara, J.; Iida, T.; Takase, N.; Machida, N.; Kuramoto, M.; Yamagishi, H. Growth of silicon crystal with a diameter of 400 mm and weight of 400 kg. J. Cryst. Growth 2001, 229, 17–21. 58. Fujita, K. Past, present and future of the growth of compound semiconductor crystals. 2nd Asian Conference on Crystal Growth and Crystal Technology, 2002; Paper A.18.

C

Cumene Production C Robert J. Schmidt UOP LLC, Des Plaines, Illinois, U.S.A.

INTRODUCTION The history of the cumene market has been examined in great detail with much discussion regarding product usage, emerging markets, and process economics over the past 10–20 yr.[1] With more than 90% of the world’s phenol production technology currently based on the cumene hydroperoxide route, it is the focus of this entry to review the latest technology improvements in cumene processing made over the past 10 yr. Current state-of-the-art processes for the production of cumene as a feedstock for phenol involve zeolitic catalyst technology offerings from UOP, Badger Licensing (formerly ExxonMobil and the Washington Group), and CDTech based on zeolitic catalysis. Much of the improvements in these technologies relate to yield, stability, and operating costs. CUMENE PRODUCTION Cumene is produced commercially through the alkylation of benzene with propylene over an acid catalyst. Over the years, many different catalysts have been used for this alkylation reaction, including boron trifluoride, hydrogen fluoride, aluminum chloride, and phosphoric acid. Cumene processes were originally developed between 1939 and 1945 to meet the demand for highoctane aviation gasoline during World War II.[2,3] By 1989, about 95% of cumene demand was used as an intermediate for the production of phenol and acetone. Today, nearly all cumene is used for production of phenol and acetone with only a small percentage being used for the production of a-methylstyrene. The demand for cumene has risen at an average rate of 2–4% per year from 1970 to 2003.[4,5] This trend is expected to continue through at least 2010. Currently, over 80% of all cumene is produced by using zeolite-based processes. Early processes using zeolite-based catalyst systems were developed in the late 1980s and included Unocal’s technology based on a conventional fixed-bed system and CR&L’s catalytic distillation system based on an extension of the CR&L MTBE technology.[6–9] At present, the Q-Max2 process offered by UOP and the Badger Cumene Technology developed by ExxonMobil and offered by Badger Licensing represent the state-ofthe-art zeolite-based catalyst technologies. A limited Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120026490 Copyright # 2006 by Taylor & Francis. All rights reserved.

number of cumene units remain using the fixed-bed, kieselguhr-supported solid phosphoric acid (SPA) catalyst process developed by UOP and the homogenous AlCl3 and hydrogen chloride catalyst system developed by Monsanto. Solid Phosphoric Acid Catalyst Although SPA remains a viable catalyst for cumene synthesis, it has several important limitations: 1) cumene yield is limited to about 95% because of the oligomerization of propylene and the formation of heavy alkylate by-products; 2) the process requires a relatively high benzene=propylene (B=P) molar feed ratio on the order of 7=1 to maintain such a cumene yield; and 3) the catalyst is not regenerable and must be disposed of at the end of each short catalyst cycle. Also, in recent years, producers have been given increasing incentives for better cumene product quality to improve the quality of the phenol, acetone, and especially a-methylstyrene (e.g., cumene requires a low butylbenzene content) produced from the downstream phenol units. For the UOP SPA catalyst process, propylene feed, fresh benzene feed, and recycle benzene are charged upflow to a fixed-bed reactor, which operates at 3
4 MPa (400–600 psig) and at 200–260 C. The SPA catalyst provides an essentially complete conversion of propylene on a one-pass basis. A typical reactor effluent yield contains 94.8 wt% cumene and 3.1 wt% diisopropylbenzene (DIPB). The remaining 2.1% is primarily heavy aromatics. This high yield of cumene is achieved without transalkylation of DIPB and is a key advantage to the SPA catalyst process. The cumene product is 99.9 wt% pure. The heavy aromatics, which have a research octane number (RON) of about 109, can be either used as high-octane gasoline-blending components or combined with additional benzene and sent to a transalkylation section of the plant where DIPB is converted to cumene. The overall yield of cumene for this process based on benzene and propylene is typically 97–98 wt% if transalkylation is included or 94–96 wt% without transalkylation. Generally, it has been difficult to justify the addition of a transalkylation section to the SPA process based on the relatively low incremental yield improvement that it provides. 603

604

Cumene Production

ALCL3 AND HYDROGEN CHLORIDE CATALYST Historically, AlCl3 processes have been used more extensively for the production of ethylbenzene (EB) than for the production of cumene. In 1976, Monsanto developed an improved cumene process that uses an AlCl3 catalyst, and by the mid-1980s, the technology had been successfully commercialized. The overall yield of cumene for this process can be as high as 99 wt% based on benzene and 98 wt% based on propylene.[10] Detailed process flow information is widely published in the literature for this technology.[11] Dry benzene, both fresh and recycled, and propylene are mixed in an alkylation reaction zone with an AlCl3 and hydrogen chloride catalyst at a temperature of less than 135 C and a pressure of less than 0.4 MPa (50 psig).[11] The effluent from the alkylation zone is combined with recycled polyisopropylbenzene and fed to a transalkylation zone also using AlCl3 catalyst, where polyisopropylbenzenes are transalkylated to cumene. The strongly acidic catalyst is separated from the organic phase by washing the reactor effluent with water and caustic. The distillation section is designed to recover a high-purity cumene product. The unconverted benzene and polyisopropylbenzenes are separated and recycled to the reaction system. Propane in the propylene feed is recovered as liquid petroleum gas (LPG).

ZEOLITE CATALYSTS In the past decade, beta zeolite (given the universal BEA) has rapidly become the catalyst of choice for commercial production of EB and cumene. Mobil invented the basic beta zeolite composition of matter in 1967.[12] Since that time, catalysts utilizing beta have undergone a series of evolutionary steps leading to the development of the state-of-the-art catalysts such as QZ-20002 catalyst and QZ-20012 catalyst for cumene alkylation. Much of the effort between 1967 and the early 1980s involved characterization of the perplexing structure of beta zeolite. It was quickly recognized that the BEA zeolite structure has a large-pore, three-dimensional structure, and a high acidity capable of catalyzing many reactions. But it was not until early 1988 that scientists at Exxon finally determined the chiral nature of the BEA structure, which is shown in Fig. 1. While the structure of beta was being investigated, new uses for this zeolite were being discovered. A major breakthrough came in late 1988 when workers at Chevron invented a liquid phase alkylation process using beta zeolite catalyst. Chevron patented the process in 1990.[13] While Chevron had significant commercial experience with the use of Y (FAU) zeolite in liquid phase aromatic alkylation, Chevron quickly recognized the benefits of beta over Y as well and other

Fig. 1 Beta zeolite. (View this art in color at www.dekker.com.)

acidic zeolites used at that time, such as mordenite (MOR) or ZSM-5 (MFI). Fig. 2 shows a comparison of the main channels of these zeolites. Chevron discovered that the open 12-membered ring structure characteristic of beta zeolite coupled with its high acidity made it an excellent catalyst for aromatic alkylation. These properties were key in the production of alkyl aromatics such as EB and cumene in extremely high yields and with product purities approaching 100%. Moreover, Chevron discovered that the combination of high activity and porous structure imparted a high degree of tolerance to many typical feed contaminants. From a technical perspective, the process developed by Chevron was a breakthrough technology in that the high cumene yields and purities were not attainable by the other vapor phase or liquid phase processes of the day. Nevertheless, the manufacturing cost of beta zeolite was still too high for catalyst producers to make a commercially viable catalyst. UOP, however, developed new manufacturing technology to make a beta zeolite based catalyst a commercial reality. In 1991 a new cost-effective synthesis route was invented by Cannan and Hinchey at UOP.[14] The new synthesis route patented the substitution of diethanolamine, a much less expensive templating agent, for a substantial fraction of tetraethylammonium hydroxide, which had been used in the synthesis previously. Moreover, the route further enables the use of tetraethylammonium bromide (instead of the hydroxide) as an additional cost saving approach. Finally, the new synthesis route allows the practical synthesis of beta over a wider range of silica to alumina ratios, a factor that has a profound effect on the catalyst’s performance. Subsequently, UOP sought to develop zeolitic catalysts that would overcome the limitations of SPA including a catalyst that is regenerable, produces higher cumene yield, and decreases the cumene cost of production. More than 100 different catalyst materials were screened, including mordenites, MFIs, Y-zeolites,

Cumene Production

605

C

Fig. 2 Comparison of various zeolites. (View this art in color at www.dekker.com.)

amorphous silica–aluminas, and beta zeolite. The most promising materials were modified to improve their selectivity and then subjected to more rigorous testing. On the process side, Unocal developed an early liquid phase fixed-bed reactor system based on a Y-type zeolite catalyst in the 1980s.[15] The higher yields associated with the liquid phase based process were quickly recognized and adapted by the industry with the selectivity to cumene generally falling between 70 and 90 wt% based on converted benzene and propylene dependent on operating conditions. Major side products in the process are primarily polyisopropylbenzenes, which are transalkylated to cumene in a separate reaction zone to give an overall yield of cumene of about 99 wt%. The distillation requirements involve the separation of propane for LPG use, the recycle of excess benzene to the reaction zones, the separation of polyisopropylbenzene for transalkylation to cumene, and the production of a purified cumene product. By 1992, UOP had selected the most promising catalyst, based on beta zeolite, for cumene production and then began to optimize a liquid phase based process design around this new catalyst. The result of this work led to the commercialization of the UOP Q-Max process and the QZ-2000 catalyst in 1996. More recently in 2001, UOP commercialized a new alkylation catalyst, QZ-2001, which offers improved stability and operation as low as 2 B=P molar feed

ratio. The low B=P feed ratio (2) represents the lowest in the industry and affords cumene producers the option to expand capacity and=or revamp existing fractionation equipment with significant cost savings. The Q-Max process flow scheme based on a liquid phase process is shown in Fig. 3. The alkylation reactor is divided into four catalyst beds contained in a single reactor vessel. Fresh benzene feed is routed through the upper-mid section of the depropanizer column to remove excess water that may be present in the fresh benzene feed. Relatively dry benzene is withdrawn from the depropanizer for routing to the alkylation reactor. Recycle benzene to the alkylation and transalkylation reactors is recovered as a sidedraw from the benzene column. A mixture of fresh and recycle benzene is charged downflow into the alkylation reactor. Propylene feed is divided into portions and injected into the alkylation reactor between the catalyst beds and each portion is essentially completely consumed in each bed. An excess of benzene is used in the alkylation reactor to avoid polyalkylation and to help minimize olefin oligomerization. The alkylation reaction is highly exothermic and the temperature rise in the reactor is controlled by recycling a portion of the alkylation reactor effluent to the reactor inlet to act as a heat sink. In addition, the inlet temperature of each downstream bed is reduced to the same temperature as the first bed inlet by injecting a portion of cooled reactor effluent between beds.

606

Cumene Production

Fig. 3 Q-Max process. (View this art in color at www.dekker.com.)

Effluent from the alkylation reactor flows to the depropanizer column, which removes any propane that may have entered with the propylene feed along with excess water that may have entered with the fresh benzene feed. The bottom stream of the depropanizer column goes to the benzene column where excess benzene is collected overhead and recycled. Any trace nonaromatics that may have been in the fresh benzene feed can be purged from the benzene column to avoid an unacceptable accumulation of nonaromatic species in the benzene recycle stream. The benzene column bottom stream goes to the cumene column where the cumene product is recovered overhead. The cumene column bottoms stream, which contains predominantly DIPB, goes to the DIPB column. If the propylene feed contains an excessive amount of butylenes, or if the benzene feed contains an excessive amount of toluene, higher levels of butylbenzenes and=or cymenes can be formed in the alkylation reactor. These compounds are distilled out and purged from the overhead section of the DIPB column. The DIPB stream leaves the DIPB column by a sidedraw and is passed to the transalkylation reactor. The DIPB column bottom stream consists of heavy aromatic by-products, which are normally blended into fuel oil. Steam or hot oil typically provides the heat requirements of the product fractionation section. The sidedraw from the DIPB column containing mainly DIPB combines with a portion of the recycle benzene and is charged downflow into the transalkylation reactor. In the transalkylation reactor, DIPB and benzene are converted to additional cumene. The effluent from the transalkylation reactor is then sent to the benzene column. QZ-2000 or the newer QZ-2001 catalyst can be utilized in the alkylation reactor while QZ-2000 catalyst remains the catalyst of choice for the transalkylation

reactor. The expected catalyst cycle length is 2–4 yr, and the catalyst should not need replacement for at least three cycles with proper care. At the end of each cycle, the catalyst is typically regenerated ex situ via a simple carbon burn by a certified regeneration contractor. However, the unit can also be designed for in situ regeneration. Mild operating conditions and a corrosion-free process environment permit the use of carbon–steel construction and conventional process equipment. An alternative zeolite process was developed by CR&L and licensed by CDTech and is based on the concept of catalytic distillation, which is a combination of catalytic reaction and distillation in a single column.[6–9] Catalytic distillation uses the heat of reaction directly to supply heat for distillation of the reaction mixture. This concept has been applied commercially for producing not only cumene but also EB and methyl tert-butyl ether (MTBE). The use of a single column that performs both the reaction and the distillation functions has the potential of realizing substantial savings in capital cost by essentially eliminating the need for a separate reactor section. Unfortunately, many available zeolite catalysts that are ordinarily very effective in promoting alkylation in a fixed-bed environment are much less effective when used in the environment of the catalytic distillation column. Also, a separate fixed-bed finishing reactor may be required to ensure that complete conversion (100%) of the olefin occurs to avoid yield losses of propylene to the LPG product stream. As such, the amount of catalyst and physical size of the distillation column may be substantially larger than the benzene column used in the conventional fixed-bed process. Thus, the savings realized by the elimination of the reactor section may be more than offset by the increased catalytic distillation column size, catalyst cost, and addition of a finishing reactor. Depending on the relative values

Cumene Production

607

and trade-off of these considerations including utility values, catalytic distillation may still be an appropriate option for producers in certain circumstances, although market interest in this process option for cumene production has been low in recent years.

Zeolitic Alkylation Chemistry The production of cumene proceeds by a modified Friedel–Crafts alkylation of benzene with propylene. This reaction can be promoted with varying degrees of effectiveness using many different acid catalysts. The basic alkylation chemistry and reaction mechanism are shown in Fig. 4. The olefin forms a carbonium ion intermediate, which attacks the benzene ring in an electrophilic substitution. The addition of propylene to the benzene ring is at the middle carbon of C3 olefin double bond, in accordance with Markovnikov’s rule. The presence of the isopropyl group on the benzene ring weakly activates the ring toward further alkylation, producing DIPB and heavier polyalkylate by-products. Because new high-activity beta zeolite catalysts such as QZ-2000 catalyst are such strong acids, they can be used at lower reaction temperatures than SPA catalyst or other relatively lower-activity zeolites such as MCM-22 catalyst.[16,17] The lower reaction temperature in turn reduces the olefin oligomerization reaction rate, which is relatively high for SPA catalyst. The result is that beta zeolite catalysts tend to have higher selectivity to cumene and lower selectivity to both nonaromatics that distill with cumene (such as olefins, which are analyzed as Bromine Index, and saturates) and heavy by-products. For example, although butylbenzene is typically produced from traces of butylene

in the propylene feed, there is always the potential also for butylbenzene to form through the oligomerization of propylene to nonene, followed by cracking and alkylation to produce butylbenzenes and amylbenzenes. As a result of having relatively high activity and operating at relatively low temperature, beta zeolite catalyst systems tend to eliminate oligomerization. This results in essentially no butylbenzene formation other than that formed from the butylenes in the propylene feed. The cumene product from a beta zeolite based process such as the Q-Max process unit fed with a butylene-free propylene feedstock typically contains less than 15 wt ppm butylbenzenes. The Q-Max process typically produces nearequilibrium levels of cumene (between 85 and 95 mol%) and DIPB (between 5 and 15 mol%). The DIPB is fractionated from the cumene and reacted with recycle benzene at transalkylation conditions to produce additional cumene. The transalkylation reaction is believed to occur by the acid catalyzed transfer of one isopropyl group from DIPB to a benzene molecule to form two molecules of cumene, as shown in Fig. 5 Beta zeolite catalyst is also an extremely effective catalyst for the transalkylation of DIPB to produce cumene. Because of the high activity of beta zeolite, transalkylation promoted by beta zeolite can take place at very low temperature to achieve high conversion and minimum side products such as heavy aromatics and additional n-propylbenzene as highlighted in Fig. 6. Virtually no tri-isopropyl benzene is produced in the beta system owing to the shape selectivity of the three-dimensional beta zeolite structure, which inhibits compounds heavier than DIPB from forming. As a result of the high activity and selectivity properties of beta zeolite, a beta zeolite based catalyst

Fig. 4 Alkylation chemistry. (View this art in color at www.dekker.com.)

C

608

Cumene Production

Fig. 5 Transalkylation chemistry. (View this art in color at www.dekker.com.)

(e.g., QZ-2001 catalyst or QZ-2000 catalyst) is specified for both the alkylation and the transalkylation sections of the Q-Max process. With both alkylation and transalkylation reactors working together to take full advantage of the QZ-2001=QZ-2000 catalyst system, the overall yield of cumene based on benzene and propylene feed in

the Q-Max process can be at least 99.7 wt% or higher. Because the Q-Max process uses small, fixed-bed reactors and carbon–steel construction, the erected cost is relatively low. Also, because the QZ-2001=QZ-2000 catalyst system is more tolerant of feedstock impurities (such as water, p-dioxane, sulfur, etc.) compared to other catalysts available, the Q-Max process requires

Fig. 6 Possible alkylation side reactions. (View this art in color at www.dekker.com.)

Cumene Production

minimal pretreatment of the feeds, which further minimizes the capital costs. This is in distinct contrast to other technologies based on zeolites other than beta where extensive feed contaminant guard beds are required to protect the catalyst from rapid and precipitous deactivation and loss of conversion when exposed to trace amounts of sulfur, water, oxygenates, and nitrogen. Cumene Product Impurities Beta zeolite catalyst can be optimized to nearly eliminate all undesirable side reactions in the production of cumene. The improvement in beta zeolite catalyst quality has occurred to the point that any significant impurities in the cumene product are governed largely by trace impurities in the feeds. The selectivity of the catalyst typically reduces by-products to a level resulting in production of ultrahigh cumene product purities up to 99.97 wt%. At this level, the only significant byproduct is n-propylbenzene with the catalyst producing essentially no EB, butylbenzene, or cymene beyond precursors in the feed. Fig. 7 shows the reactions of some common feedstock impurities that produce these cumene impurities. Beta Catalyst Resistance to Feed Contaminants Beta zeolite is a large pore zeolite with optimized acid site densities that exhibits:  Good mass transfer properties.

609

 Significant reduction in undesirable polymerization and by-product side reactions.  High tolerance to feedstock impurities and poisons. The resistance of new zeolitic catalysts to temporary and permanent catalyst poisons is essential to the economic and commercial success of a zeolitic based cumene process. The following commercial data obtained using beta zeolite as a catalyst illustrates the outstanding ability of beta zeolite to cope with a wide range of feedstock contaminants: Cyclopropane n-Propylbenzene (nPB) is formed by alkylation of benzene with cyclopropane or n-propanol, and by anti-Markovnikov alkylation of benzene with propylene. Cyclopropane is a common impurity in propylene feed and approximately half of this species is converted to nPB in the alkylation reactor. Essentially, all alkylation catalysts produce some nPB by anti-Markovnikov alkylation of propylene. The tendency to form nPB rather than cumene decreases as the reaction temperature is lowered, making it possible to compensate for cyclopropane in the feed to some extent. As the operating temperature of zeolitic alkylation catalyst is decreased, the deactivation rate increases. However, because of the exceptional stability of the beta zeolite catalyst system, a unit operating with beta zeolite catalyst can be operated for extended cycle lengths and still maintain an acceptable level of nPB in the cumene product. For example, with FCC-grade propylene feed containing typical amounts of cyclopropane,

Fig. 7 Reactions of feed impurities. (View this art in color at www.dekker.com.)

C

610

a beta zeolite based process can produce an overall cumene product containing 250–300 wt ppm nPB while achieving an acceptable catalyst cycle length. Water Water can act in this environment as a Bronsted base to neutralize some of the weaker zeolite acid sites. This effect is not harmful to any appreciable extent to the beta zeolite catalyst at typical feed stock moisture levels and under normal alkylation and transalkylation conditions. This includes processing of feedstocks up to the normal water saturation condition (typically 500–1000 ppm) resulting in 10–150 ppm water in the feed to the alkylation reactor dependent on feed and=or recycle stream fractionation efficiency. Oxygenates Small quantities of methanol and ethanol are sometimes added to the C3s in pipelines to protect against freezing because of hydrate formation. Although the beta zeolite catalyst is tolerant of these alcohols, removing them from the feed by a water wash may still be desirable to achieve the lowest possible levels of EB or cymene in the cumene product. Cymene is formed by the alkylation of toluene with propylene. The toluene may already be present as an impurity in the benzene feed, or it may be formed in the alkylation reactor from methanol and benzene. Ethylbenzene is primarily formed from ethylene impurities in the propylene feed. However, similar to cymene, EB can also be formed from ethanol. p-Dioxane is sometimes present in benzene from extraction units that use ethylene glycol based solvents. It is reported to cause deactivation in some zeolitic alkylation catalysts even at very low ppm levels. However, beta zeolite catalyst appears to be tolerant to p-dioxane at levels typically found in benzene extraction processes and does not require costly removal of this impurity. Sulfur Sulfur has no significant effect on beta zeolite catalyst at the levels normally present in olefin and benzene feeds considered for cumene production. However, even though the beta zeolite catalyst is sulfur tolerant, trace sulfur that makes its way into the finished cumene unit product may be a feed quality concern for downstream phenol processors where the typical sulfur specification is quinoline > benzoquinoline > methylated benzoquinoline > carbazole > methylated carbazoles. Solid acid denitrogenation[15] A new method to remove the nitrogen compounds of lubricating base oils was suggested. A solid acid was employed as an effective reagent to remove the basic and nonbasic nitrogen compounds. However, the sulfur compound is relatively difficult to remove. Clay treating after the solid acid denitrogenation process can apparently improve the oxidation stability of denitrogenated base oils. Ultraviolet irradiation and liquid–liquid extraction[7] Denitrogenation for light oils is based on a combination of ultraviolet (UV) irradiation and liquid–liquid extraction. Two extraction systems, one oil=water and the other oil=acetonitrile, were used for the denitrogenation of three separate light oils of differing

Denitrogenation

631

nitrogen and hydrocarbon composition. Photodecomposition of carbazole was found to be suppressed by the presence of double-ring aromatic hydrocarbons. This adverse effect can be reduced by the addition of hydrogen peroxide to the water phase. In the presence of 30% H2O2 and 36 hr of photoirradiation, the nitrogen content of light oils was decreased to less than 20% of feed value. In the oil=acetonitrile system, the nitrogencontaining compounds in the light oils were extracted into the acetonitrile phase and were photodecomposed effectively, even in the presence of double-ring aromatics. Then, with 10 hr of photoirradiation, the nitrogen content of the light oils was decreased successfully to less than 3% of initial concentration. This method is time-consuming and is not cost-effective and therefore cannot treat high concentrations of pollutants. Alkylation and a subsequent precipitation method using alkylating agents (CH3I and AgBF4)[16] Nitrogen and sulfur are methylated by the addition of the alkylating agents under moderate conditions and are removed successfully as the precipitates of the corresponding S-methylsufonium and N,N-dimethlycarbasolium tetrafluoroborates. By these means, the sulfur and nitrogen concentration of vacuum gas oil were reduced simultaneously to less than 0.1 and 7.0% of initial concentration with a 20-fold molar excess of CH3I and a 10-fold molar excess of AgBF4. Analytic methods are by means of field ionization-mass spectrometry (FI-MS) and gas chromatography-atomic emission detection (GC-AED). This is a nondestruction method. Chemically assisted ultrasound process CAUP is an ultrasonic reaction that produces a considerable amount of energy and pressure, which causes large numbers of bubbles or cavitations. With this energy, the bounding of a pollutant can be broken. Most of conventional methods necessarily accompany high energy and pressure all the time, which is expensive. Ultrasound is an amazing source of energy and pressure with relatively low electricity. The mechanism of ultrasound can be summarized in three phenomena:[17–19] (1) solvent compression and rarefaction (rapid movement of fluids); (2) cavitation [high temperature: 20,000 F, high pressure: 75,000 psig, collision time: 1.25 (10
5 sec)]; and (3) microstream with little heating (great volume of vibrational energy confined to a small volume of reaction). In summary, the target compound is dissolved in a solvent. Surfactant entails one phase of emulsion by changing surface properties of target compound and solvent while stirring, then ultrasonic reactions attack the loose bonding of target compound through

D

Fig. 1 Diagram showing ultrasonic reaction. (View this art in color at www.dekker.com.)

a high degree of energy and pressure. The bond scission reaction occurs between N–H and C–H bonding. The concentration of the target compound is then noticeably decreased. Fig. 1 shows the mechanism of CAUP.

CONCLUSIONS The trend in fossil fuels shows that heavy bitumen and oil shale consumption will increase over the next century because high quality oil resources are limited. This also means there is no need to expect high quality fuel production only from petroleum or natural gas. Low value energy sources will be commercially viable only if denitrogenation technology is allowed to reduce the nitrogen content. Many profitable resources such as heavy bitumen and oil shale have commanded less attention due to their low commercial value, although the exploration of oil shale for production has become an important energy program. The U.S. reserves the largest amount of oil shale in the world, but can only produce high quality oil as long as the technology is available. Technology tends to control the industry. Denitrogenation technology is equally important as desulfurization and demetallization when it comes to upgrading low value energy resources. The theory and technology of denitrogenation will have a great impact on commercial applications in industry and scientific areas.

REFERENCES 1. Yen, T.F. Environmental Chemistry; Prentice Hall, Inc.: New Jersey, 1999; Vol. 4A, 157.

632

2. Pakdel, H.; Roy, C. Recovery of bitumen by vacuum pyrolysis of Alberta Tar Sands. Energy Fuels 2003, 17, 1145–1152. 3. Mei, H; Mei, B.W.; Yen, T.F. A new method for obtaining ultra-low sulfur diesel fuel via ultrasound assisted oxidative desulfurization. Fuel 2003, 82, 405–414. 4. Speight, J.G. Fuel Science and Technology Handbook; Marcel Dekker, Inc.: New York, 1990; Vol. 41, 81. 5. Yen, T.F. Environmental Chemistry; Prentice Hall, Inc.: New Jersey, 1999; Vol. 4A, 162. 6. Hsu-Chou, R.S.Y.; Mobashery, S.; Yen, T.F. Denitrogenation of shale oil by oxime formation from pyrroles. Energy Sources 1998, 20, 857–866. 7. Shiraishi, Y.; Hirai, T.; Komasawa, I. Photochemical denitrogenation processes for light oils effected by a combination of UV irradiation and liquid–liquid extraction. Ind. Eng. Chem. Res. 2000, 39 (8), 2826–2836. 8. Sumpter, W.C.; Miller, F.M. Heterocyclic Compounds with Indole and Carbazole Systems; 1954; 1–2, N.Y. 2. 9. Shue, F.-F.; Yen, T.F. Concentration and selective identification of nitrogen- and oxygencontaining compounds in shale oil. Anal. Chem. 1981, 53 (13), 2081–2084. 10. Williams, P.T.; Nazzal, J.M. Polycyclic aromatic compounds in shale oils: Influence of process conditions. Environ. Technol. 1998, 19 (8), 775–787. 11. Sugaya, K.; Nakayama, O.; Hinata, N.; Kamekura, K.; Ito, A.; Yamagiwa, K.; Ohkawa, A. Biodegradation of quinoline in crude oil. J. Chem. Technol. Biotechnol. 2001, 76 (6), 603–611.

Denitrogenation

12. Williams, P.T.; Nazzal, J.M. Pyrolysis of oil shales: influence of particle grain size on polycyclic aromatic compounds in the derived shale oils. J. Inst. Energy 1999, 72 (491), 48–55. 13. Yen, T.F.; Shue, F.-F.; Wu, W.-H.; Tzeng, D. Ferric chloride-clay complexation method: removal of nitrogen-containing compounds from shale oil and related fossil fuels. Anal Chem. 1983, 457–466. 14. Shin, S.H.; Sakanishi, K.; Mochida, I.; Grudoski, D.A.; Shinn, J.H. Identification and reactivity of nitrogen molecular species in gas oils. Energy Fuels 2000, 18 (7–8), 539–544. 15. Wang, Y.Z.; Li, R.L. Denitrogenation of lubricating base oils by solid acid. Petrol. Sci. Technol. 2000, 18 (7–8), 965–973. 16. Shiraishi, Y; Hirai, T; Komasawa, I. A novel desulfurization for fuel oils based on the formation and subsequent precipitation of S-alkylsulfonium salts. 4. Desulfurization and simultaneous denitrogenation of vacuum gas oil. Ind. Eng. Chem. Res. 2001, 40 (15), 3398–3405. 17. Tu, S.P.; Kim, D.; Yen, T.F. Decolorization and destruction of metallophthalocyanines in aqueous medium by ultrasound; A feasibility study. J. Environ. Eng. Sci. 2002, 1, 237–246. 18. Kim, D.; Tu, S.P.; Yen, T.F. Evaluation of versatile ultrasonic effects on degradation of organometallics from petroleum. Korean J. Environ. Eng. Res. 2003, 8 (2), 11–23. 19. Kim, D.; Shiu, F.J.Y.; Yen, T.F. Devulcanization of scarp tire through matrix modification and ultrasonication. Energy Sources 2003, 25 (11), 1099–1112.

Design of Extrusion Dies D Milivoje M. Kostic Department of Mechanical Engineering, Northern Illinois University, DeKalb, Illinois, U.S.A.

Louis G. Reifschneider Department of Technology, Illinois State University, Normal, Illinois, U.S.A.

INTRODUCTION The goal of this chapter is to introduce the reader to the importance of extrusion die design as well as the complexities inherent in the task. Extrusion is of vital importance to all plastics processing. In addition to providing raw stock such as sheet for thermoforming and pellets for injection molding and other extrusion processing, numerous end-use products are made with extrusion such as film, tubing, and a variety of profiles. Although the types of extruded products made can differ dramatically in shape, there are a set of common rules that govern basic die design. For example, it is important to streamline the flow from the inlet to the exit, and as a practical measure, to fine-tune the flow balance and product dimensions, flow adjustment devices could be included in the die design. Several unique products are made by extrusion and the dies needed to make these products are classified as: 1) sheet dies; 2) flat-film and blown-film dies; 3) pipe and tubing dies; 4) profile extrusion dies; and 5) co-extrusion dies. Furthermore, each product type has unique hardware downstream of the die to shape and cool the extruded melt. To aid the reader, detailed illustrations of the various die designs and the complementary downstream cooling and shaping hardware are shown. Predicting the required die profile to achieve the desired product dimensions is a very complex task and requires detailed knowledge of material characteristics and flow and heat transfer phenomena, and extensive experience with extrusion processing. Extrusion die design is still more an art than a science, even though the latter is becoming more and more relevant for design optimization because of recent advancement in the powerful computation and modeling of complex flow and heat transfer processes, before, through, and after the die.

DESIGN FUNDAMENTALS Extrusion is a continuous process where solid polymeric materials, either pellets or powders, are sheared Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120039324 Copyright # 2006 by Taylor & Francis. All rights reserved.

and heated as they are conveyed through either a single- or a twin-screw extruder (as described elsewhere) to become a pressurized melt. The pressurized melt flows through a properly shaped orifice, or extrusion die, and then is pulled (with a little pressure) as it is cooled and shaped to a final product called the extrudate. The proper design of an extrusion die is extremely important to achieve the desired shape and accurate dimensions of the extruded product. The function of an extrusion die is to shape the molten plastic exiting an extruder into the desired cross section depending on the product being made. The die provides a passage between the circular exit of the extrusion barrel and the more complex and often much thinner and wider die exit. A schematic of a common die, called a sheet die, is shown in Fig. 1A to illustrate this point. The extrusion process creates products of uniform cross section in a continuous fashion. An ideal passage will:[1,2]
Balance the melt flow by providing a more uniform exit velocity across the entire die exit.
Achieve this flow balance with a minimal pressure drop.
Streamline the flow to avoid abrupt changes in the flow passage that may cause stagnation areas. Stagnated flow may lead to thermal degradation of the plastic melt as the melt is exposed to high heats for long periods. As a practical measure, flow control devices should be incorporated into the die design to permit finetuning of the die passage shape to ensure a proper flow balance. In addition, the design of extrusion dies is complicated by two unique material properties of molten plastics:[3]
Melts exhibit shear thinning behavior (become less viscous) as they are sheared.
Melts exhibit viscoelastic behavior, which influences the ‘‘die swell’’ on exiting the die. 633

634

Design of Extrusion Dies

Fig. 1 Coat hanger-type sheet die concept (A): (1) central inlet port; (2) manifold (distributes melt); (3) island (along with manifold, provides uniform pressure drop from inlet to die lip; (4) die lip (die exit forms a wide slit); and schematic of sheet die (B): (1) upper die plate; (2) lower die plate; (3) manifold; (4) island; (5) choker bar; (6) choker bar adjustment bolt; (7) flex die lip; (8) flex lip adjustment bolt; (9) lower lip; (10) die bolt; (11) heater cartridge.

The shear thinning causes the volumetric flow to be very sensitive to slight changes in die geometry. For example, the flow for a typical polymer melt through a slit will vary with the cubic thickness of the gap.

Thus, a small change in the die gap along the contour of the die exit may cause considerable change of the melt flow. The term ‘‘die swell’’ refers to the enlargement in the direction orthogonal to the flow direction.

Design of Extrusion Dies

635

Swelling after exiting the die lip is due to two distinct phenomena:[4,5]
Velocity relaxation (unification) of the melt flow.
Viscoelastic relaxation of the strained polymer molecules. Velocity relaxation occurs because the melt is no longer under shear from the no-slip walls of the extrusion die. The melt assumes a uniform bulk velocity that causes the high-speed areas to slow down and the areas previously retarded from the wall to increase their speed. The net result is enlarging (swelling) of the melt cross-section bulk as it exits the die, while the stagnant outside region and, especially, the corners are stretched and shrunk. Newtonian and non-Newtonian fluids exhibit die swell owing to velocity relaxation. The swelling due to viscoelastic memory is a characteristic of polymeric fluids and occurs because the polymer molecules are stretched in the flow direction while passing through the high-shear area before the die exit. On exiting, the molecules recoil and shorten in the flow direction. The result is an expansion in the direction orthogonal to the flow, a swelling of the diameter of a round strand on exiting the die, for example. The amount of viscoelastic swelling is a combination of the material properties of the polymer as well as process conditions such as melt temperature, shear rate, and residence time under high shear, especially near the die exit.[6] The design of extrusion dies today is facilitated by computer-based simulation tools. The flow of Table 1 Typical extruded product shapes Films

t < 0.01 in.

Sheets

t > 0.01 in.

Profiles

Strand

Open

Hollow chamber

Tubes

d < 1.0 in.

Pipes

d > 1.0 in.

non-Newtonian fluids through complex passages is routinely performed by computational fluid dynamics (CFD) programs.[7–10] Factors such as shear thinning are readily accounted in die design. The viscoelastic behavior can also be modeled today, although this simulation requires extensive material testing to obtain the required material parameters for accurate simulation results. This is discussed in more detail later in this entry. Extrusion dies vary in shape and complexity to meet the demands of the product being manufactured. There are five basic shapes of products made with extrusion dies, as illustrated in Table 1.[11,12] Film and sheet dies are called slit dies as the basic shape of the die exit is a slit. Film is also made with annular dies as in the case of blown film. Strand dies make simple geometric shapes, such as circles, squares, or triangles. Pipe and tubing dies are called annular dies as the melt exits the die in the shape of an annulus. The inner wall of the annulus is supported with slight air pressure during extrusion. Open profile dies make irregular geometric shapes, such as ‘‘L’’ profiles or ‘‘U’’ profiles, and combinations of these. Hollow profile dies make irregular profiles that have at least one area that is completely surrounded by material. Examples of hollow profiles can be simple, such as concentric squares to make a box beam, or a very complex window profile. Each extruded product relies on a die to shape the moving melt followed by shaping and cooling devices downstream to form the extrudate into the final desired shape and size. A more complete treatment of these devices, which are typically water-cooled metallic

D

636

devices that contact the extrudate melt, is presented later in the section ‘‘Extrudate Cooling and Sizing Hardware.’’

SHEET DIES The most common extrusion die for sheet products is the coat hanger-type manifold die as shown schematically in Fig. 1A and as a section view in Fig. 1B. The key elements of Fig. 1A are:
Central inlet port: connects to the extruder barrel.
Manifold: provides a streamlined channel to evenly distribute the melt to the island.
Island: with the manifold serves to create an equal pressure drop from the die inlet to all points across the die exit.
Die lip: wide slit across the die that provides the final sizing of the melt. Commercial sheet dies typically employ four features to control the flow to the die lip. These are the combined shape of the manifold and the island as well as the following three features shown in Fig. 1B:
Choker bar: adjustable along with the width of the die and serves to tune the flow balance across the width of the die.
Lower lip: sets the nominal sheet thickness.
Flex-lip: adjustable along the width of the die and provides the final tuning to create uniform flow across the die. In addition, sheet dies have die bolts that hold the upper and lower die plates together. The die plates are normally heated with cartridge heaters spaced along the width of the die. This type of die is typically made for a specific type of polymer to account for the shear thinning behavior of that polymer.[11] Consequently, the flow distribution in the die will change with melt viscosity, i.e., when the power law viscosity index of the resin changes. As the polymer grades change, flow adjustments can be made at the choker bar and at the flex lip, which both span the width of the die and can be adjusted at numerous points along the width. Clam shelling, or die deflection, is another cause of nonuniform flow across the die.[13,14] The higher pressures along the centerline of the die coupled with the lack of bolting to keep the die plates together cause the centerline of the die gap to widen. Clam shelling can increase as the throughput of the die increases because of higher die pressures. Thus, the flow balance across the die will be sensitive to production rates. Innovations in automatic flow adjustments have been

Design of Extrusion Dies

made with designs like the Auto-Flex sheet die where the die lip gap at the flex lip is automatically adjusted by changing the length of the flex lip adjustment bolts.[15,16] The temperature-controlled bolts change length in response to cross-machine scanning of the sheet thickness.

FLAT FILM AND BLOWN FILM DIES Dies used to make film less than 0.01 in. thick include flat, slit-shaped dies called T-dies and annular dies for blown film (Figs. 2 and 3). The design of the T-die is similar to the coat hanger-type die with the exception that the manifold and the land length are constant along with the width of the die. Consequently, the use of T-dies is often limited to coating applications with low-viscosity resins that resist thermal degradation, as the ends of the manifold in the T-die create stagnation pockets.[13] A common application for a film die is to coat a substrate like paper. Blown film dies are the most common way of making commercial films. Because the blown film is so thin, weld lines are not tolerated, and the melt is typically introduced at the bottom of a spiral mandrel through a ring-shaped distribution system, as shown in Fig. 3. A series of spiral channels, cut into the mandrel-like multiple threads, smear the melt as it flows toward the die exit. This mixing action ensures that the melt is homogenous on exiting the die. Unlike other extrusion processes, blown film is sized and quenched from melt to solid film without contacting metallic cooling elements. The interior of the melt tube is pressurized with approximately 2 in. of water pressure. This pressure causes the tube to suddenly expand into a bubble as it exits the die. The tube forms a bubble because it is pinched overhead with nip rolls, which retain the air pressure. During the process of expanding, the melt tube undergoes an order of magnitude reduction in thickness and thus cools rapidly. This quenching moment occurs at the frost line of the bubble. The melt quenching occurs with a combination of external cooling air and internal bubble cooling air, as shown in Fig. 3. After the film passes through the nip rolls, it passes through a series of guide rollers to be wound up on to a roll.

PIPE AND TUBING DIES Both pipe and tubing are made in dies with an annular die exit. A pipe product is defined as being greater than 1 in. in outer diameter and a tube less than 1 in. Dies for these products are made in two styles: 1) in-line dies (also called spider dies) shown in Fig. 4A and

Design of Extrusion Dies

637

D

Fig. 2 Comparing designs of T-type die (A) [(1) constant cross-section manifold; (2) constant land length] to coat hanger-type die (B) [(1) manifold cross-section decreases as distance from centerline increases; (2) land length becomes shorter farther from the centerline of the die].

2) cross-head dies shown in Fig. 4B. The key elements of an in-line die are:
Housing: mounts onto the end of the extruder, provides a circular passage through which the melt flows; it supports the mandrel and retaining ring.
Mandrel (Torpedo): suspended in the center of the circular passage in the die body with metal bridges called spiders (typically three are used). One spider allows for passage of air into the center of the torpedo, is streamlined to avoid flow stagnation, and supports the die pin.
Die pin: mates with the torpedo to provide streamlined sizing to the final inner diameter of the melt tube leaving the die; it has an air hole running through it to allow air to pass through the die body to the interior of the melt tube. A slight positive air pressure may be used to keep the inner diameter of the tubular extrudate from collapsing on exiting the die.
Die land: forms the outer diameter of the tubular extrudate, held in place with a retaining ring and position adjusted with centering bolts. The die land can be changed to create a tube of a different diameter or wall thickness while keeping the original die pin.
Retaining plate: secures the alignment of the die land with the die pin, bolted to the die body.
Heater band: closely fitted to the housing (and for larger pipes to the exposed portion of the die pin) to ensure that the die is held at a temperature close to the required temperature of the melt.


Flange for extruder attachment: tapered flange to permit alignment and attachment to extruder with split locking collars. The in-line die is the least costly for manufacture of the two designs but can create defects called weld lines in the product. Weld lines occur because the melt is split and rejoined as it passes over the spider legs. A cross-head die can overcome this problem by eliminating the spiders. The melt enters the side of the die and turns 90 as it flows through a coat hanger-type passage that is wrapped around the mandrel. Key elements of a cross-head die that are different from an in-line tubing die are (see Fig. 4B):
Core tube: mandrel with coat hanger-type passage that splits the flow and uniformly distributes melt along the annulus between the die pin and the die land.
Side feed: melt enters from the side of the die and flows around the mandrel.
Air supply: in-line with the die pin support. Another application for the cross-head-style tubing die is wire coating. The following adjustments are made to a cross-head tubing die to perform wire coating:
First, instead of passing air through the core tube, a bare conductor wire is pulled through the die entering the core tube inlet and exiting the die pin.

638

Design of Extrusion Dies

10 9

8

7 13 6 5

11

4

2

1

12 14

3

Fig. 3 Schematic of spiral mandrel blown film die operation: (1) ring-shaped melt distribution; (2) die body; (3) spiral flow mandrel; (4) sizing ring; (5) spreader; (6) film bubble; (7) frost line; (8) solidified film; (9) bubble collapsing rollers; (10) nip rollers; (11) external bubble cooling air; (12) internal bubble cooling air inlet; (13) internal bubble cooling pipe; and (14) heated internal bubble air return.


Second, the length of the die pin is shortened to cause the wire to contact the melt tube before it exits the die land.

PROFILE EXTRUSION DIES Profile extrusions are the most difficult to make because changes in take-up speed or screw rotational speed alone are not enough to compensate for deficient product dimensions. In the case of sheet and film, if the edges of the sheet are not at the target thickness, they can be trimmed off and sent back to the extruder

to be reprocessed. Profile extrudates are significantly affected by nonuniform die swell unlike sheet and tube products. In the case of profiles with corners and other irregularities, like a square profile, the die exit needed to achieve a square profile is not square owing to the influence of die swell. Fig. 5 illustrates a die exit required to achieve a square extrudate. Note that the corners have an acute cusp shape and the side walls are not flat. A melt exiting this required but nonorthogonal shape will swell into a desired, orthogonal square shape. The design of nonorthogonal die exit required to achieve orthogonal profiles is addressed later in this entry in the section Modern Design Simulation and Computational Tools. Open profile dies are typically shapes, such as ‘‘Ushaped’’ or ‘‘L-shaped’’ channels, that are not axisymmetric, unlike tube shapes. Consequently, open profiles are more prone to cooling unevenly and thus may generate residual stresses in the solidified (frozen) extrudate that cause the product to bow. A critical design rule for open profiles is to maintain a uniform wall thickness throughout the product cross section. Examples of poor and better profile designs are shown in Fig. 6 with the poorly designed sections shown on the left-hand side and the improved designs on the right-hand side. The difficulty with the original designs of both profiles A and B is the nonuniform wall thickness. The thinner sections will solidify first with the thicker sections still remaining molten in some core area. The result will be additional thermal shrinkage in the thicker regions and thus warpage of the final product. Product A will warp downward and product B will warp toward the right. These warping problems can be alleviated by making the entire cross section with more uniform thickness. Then, the entire cross section will solidify more uniformly and a little residual stress will be trapped in the solid extrudate. Design A illustrates a case where a hollow profile is used to solve the warpage problem whereas design B illustrates the use of an open profile to replace the thick region. The revised product designs of A and B will require more expense to fabricate the dies for these products as a set of mandrels must be made to form the hollow chambers. However, there are key benefits derived by making these changes:
Better-quality products due to more uniform cooling and shrinkage: straighter products.
Less material use by removing thick, unnecessary regions: savings of material costs.
Faster cooling rates due to less hot plastic to cool: higher production rates. Profile dies are commonly made with a series of plates that are stacked together to form a complex

Design of Extrusion Dies

639

D

Fig. 4 Schematic of in-line tubing die (A): (1) housing; (2) mandrel (torpedo); (3) die pin (interchangeable); (4) die land (interchangeable); (5) retaining plate; (6) die centering bolt; (7) air hole; (8) mandrel support (spider leg); (9) die flange (mount to extruder with split clamp); (10) heater band; and cross-head tubing (or wire coating) die (B): (1) air or wire conductor inlet; (2) melt inflow (side inlet); (3) melt exit (annulus); (4) air or wire conductor exit; (5) core tube; (6) flow splitter; (7) housing; (8) die pin; (9) die land; (10) retaining plate; (11) retaining ring bolt; (12) die centering bolt; (13) heater band.

passage from the circular exit of the extruder to the required profile die exit. A stacked plate design makes for easier manufacture and permits adjustments to parts of the die assembly as needed during extrusion trials to fine-tune the die flow. An example of a stack plate die that makes a U-shaped profile is shown in Fig. 7. This figure illustrates an exploded view of a stack plate die, a cross-sectional view of the assembled

die, and a detail of the die exit compared to the target profile. Stacked plate profile dies typically have these elements:
Adapter plate: forms transition from circular extruder exit to approximate profile shape.
Transition plate: forms streamlined transition from adapter plate exit to preland plate inlet.

640

Design of Extrusion Dies

Fig. 5 Irregular die shapes required for regular extrudates.


Preland plate: imparts significant flow adjustment by reducing thickness in high-flow areas and increasing thickness in low-flow areas anticipated downstream in the die land to make flow more uniform.
Die land plate: provides a uniform cross-section passage that is typically 10 times longer than the thickness of the extrudate to relax the viscoelastic stresses in the melt before leaving the die (reduces die swell) and forms the shape of the extrudate leaving the die. The die land profile has the required shape to compensate for extrudate deformation after the die (die swell and drawdown) and yield the desired shape downstream. The die exit profile shown in Fig. 7 creates an extrudate that is a U-shaped profile with three sides

of uniform thickness and perpendicular walls to the bottom surface. The irregular shape of the die exit was generated with the aid of CFD as outlined in the section Modern Design Simulation and Computational Tools later in this entry.

COEXTRUSION DIES Another important product made with extrusion dies is the creation of multilayered materials. Multilayered sheet and film materials have two applications:
Making more economical material by sandwiching a less costly core material between two more expensive materials.

Fig. 6 Examples of poor (on left) and improved extrusion product designs (on right) to achieve uniform product thickness: (1) profile (A) made into a hollow profile and (2) profile (B) made into an open profile.

Design of Extrusion Dies

641

D

Fig. 7 U-Profile stack die: exploded view (top); section view (lower left); and end view (lower right): (1) extruder mounting plate; (2) die adapter plate; (3) transition plate; (4) preland plate; (5) die land plate; (6) die bolt hole; (7) alignment dowel pin hole; (8) thermocouple well; (9) pressure transducer port; (10) heater band; (11) breaker plate recess. Detail (lower right): (A) die exit profile and (B) product profile.


Creating a composite material with improved properties by combining two or more materials that each posses a desirable property.

multimanifolds within dies.[17] Two, three, or more melt streams may be combined with co-extrusion dies. The key elements of a feed block manifold as shown in Fig. 8A are:

Applications of coextruded material include:
Sheet stock made with an acrylic topcoat over acrylonitrile–butadiene–styrene (ABS). The acrylic provides UV resistance and gloss while the ABS provides impact resistance.
Blown film with special barrier properties: one layer limits oxygen migration through the film and another provides protection from UV radiation. There are two common methods of achieving co-extruded materials: feed block manifolds and


Inlet ports for the upper layer, middle layer, and lower layer.
Streamlined melt lamination area that channels separate flow streams into one laminated melt stream inside feed block.
Adapter plate between the feed block and the sheet die.
Sheet die, which is identical to a monolayer die. The laminated melt stream enters the center of the die and spreads out along the manifold flowing out of the die exit as a distinct multilayer extrudate.

642

Design of Extrusion Dies

Fig. 8 Coextrusion feed block manifold and sheet die (A): (1) sheet die with flow restriction; (2) adapter plate; (3) feed block asembly; (4) core material layer inlet; (5) upper material layer inlet; (6) lower material layer inlet; and coextrusion multimanifold sheet die (B): (1) lower melt channel; (2) upper melt channel; (3) lower choker bar; (4) lower choker bar adjustment bolt; (5) upper choker bar; (6) upper choker bar adjustment bolt; (7) flex lip.

An alternative to the feed block design is a multimanifold die as depicted in Fig. 8B. The key elements of this design are:
It is similar to a monolayer extrusion die, except that there is more than one feed channel.
Each melt channel has its own choker bar for flow control.


Melt streams converge inside the die near the die exit and emerge as a distinct multilayer extrudate. The feed block technique is cheaper to implement than the multimanifold approach, but because the melt streams travel some distance before reaching the die exit, irregular flow patterns can develop at the interface of the different melt streams.[18–20] This is especially

Design of Extrusion Dies

643

true when attempting to coextrude melts of significantly differing viscosities. Lower-viscosity melt tends to encapsulate the more viscous melt. The alternative method is to keep the melt streams separated until just before the die exit as is done with the multimanifold design. Multimanifold designs permit coextrusion of plastic melts having significantly differing viscosities. With any coextrusion process, however, there must exist basic chemical compatibility between the neighboring melt streams to ensure good cohesion between the layers.

EXTRUDATE COOLING AND SIZING HARDWARE With the exception of blown film and strand profiles, all extrudates require cooling and=or sizing by some metallic element. Table 2 summarizes the type of cooling and sizing hardware used for the various extrusion products.[21] In the case of sheet extrusion, the cooling is achieved with a chill roll stack, schematically illustrated in Fig. 9. The chill rolls are typically highly polished chrome-plated rollers that impart the surface gloss on sheet products and cool the extrudate while pulling the melt away from the die with a constant take-up speed. The average sheet thickness is achieved by the combination of extrusion screw rotational speeds and take-up speed adjustments. If the line speed taking the extrudate melt away from the die is greater than the average die exit speed, the thickness of the sheet decreases. This is called drawdown. In the case of pipe and tubing products, the nominal outer and inner diameters of the extrudate are made by selecting the appropriate size of the die pin and die land. The final outer dimension of tubing products is achieved with sizing rings that are typically placed in the vacuum water bath, shown schematically in Fig. 10. The outer diameter of the tube is set with the sizing ring as the vacuum, which combined with a slight positive pressure inside the tube, forces the Table 2 Cooling and sizing devices for various extruded products Product type

Cooling and sizing device

Film and sheet

Chill roll

Blown film

External and internal bubble air

Profiles—strand

Water tank

Profiles—open and hollow chamber

Vacuum calibrators and water tank

Tubing and pipes

Sizing rings and vacuum water tank

extrudate against the inner race of the ring. The desired inner diameter of the tube is achieved by adjusting the take-up speed of the extrudate relative to the average die exit speed. If the take-up speed is greater than the average die exit speed, the cross-sectional area decreases. Because the outer diameter of the tube is set with a sizing ring, the inner diameter will increase. Thus, the wall thickness of a tube is controlled this way. The sizing and cooling of profiles have special requirements because of their complex shape.[21,22] These devices are called calibrators and are often as complicated as the die. To maintain the shape of a profile, vacuum is applied while simultaneously cooling the extrudate. Some calibrators, called wet vacuum calibrators, alternately inject water between the extrudate and the calibrator to lubricate and augment the cooling. A schematic of the dry vacuum calibrator setup used to size and cool the U-shaped profile is shown in Fig. 11. Fig. 11B illustrates a partially disassembled vacuum calibrator to reveal the following details: the vacuum channels and the cooling lines that simultaneously hold the moving extrudate in the desired shape while cooling it. Vacuum calibrators for profiles are typically made of stainless steel to withstand the abrasive action of extrudates while in contact with the calibrator during the cooling process. For example, the U-profile shown in Fig. 11 will tend to shrink onto the core feature of the lower calibrator and pull away from the upper calibrator surfaces. This complicates the design of the calibrator as the extrudate will conform to the ideal calibrator shape as it deforms. The cooling of the extrudate also complicates the simulation and design of the calibrator as discussed in the next section.

MODERN DESIGN SIMULATION AND COMPUTATIONAL TOOLS The development of powerful computing hardware and proficient numerical techniques makes it possible now to simulate, analyze, and optimize three-dimensional extrusion processes with complex geometries, including nonlinear and viscoelastic polymer behavior. Numerical simulation has the potential to uncover important interior details of the extrusion process, such as velocity, shear stress, pressure, and temperature fields in the region of interest, which is not possible to do experimentally. A critical challenge for simulation methods is the ability to accurately represent the complex viscoelastic polymer material behavior that is dependent on process parameters, like shearing flow rate and temperature. Another challenge is to accurately model the complex geometry and the boundary conditions of extrusion dies and calibrators, especially for profile extrusions. Experts in the polymer processing field have cited that the increasing complexity of product designs, coupled

D

644

Design of Extrusion Dies

Fig. 9 Chill roll stack for sheet extrusion: (1) sheet die with flow restrictor; (2) molten sheet extrudate; (3) lower chill roll (all chill rolls temperature controlled); (4) middle chill roll (imparts gloss=surface texture to sheet); (5) upper chill roll; and (6) solidified sheet.

with the shorter development times, and a shortage of qualified engineers fuel the need for more process simulation in industry.[23] As already stated, several commercial polymer flow simulation programs are used for profile die design.[7–10] However, because the cooling rate of the extruded product determines the speed of the extrusion line, optimal design of a calibrator is also critical to productive operations. In addition, the design of the calibrator has an influence on the straightness of the final product because of uneven cooling results in unfavorable thermal deformations and warped products.[24]

Analytical solutions have been developed to aid the design of calibrators for simple shapes, such as sheets and pipes.[22] More complex shapes, such as window profiles, require the use of numerical finite element methods that can model arbitrary shapes.[25]

Computational Fluid Dynamics Simulation of Polymer Flow for Die Design The design process begins with a target product shape. The objective of the simulation is to determine

Design of Extrusion Dies

645

D

Fig. 10 Tubing vacuum water-bath calibration and take-off: (1) tubing=pipe die; (2) molten tube extrudate; (3) baffle; (4) vacuum water tank; (5) sizing ring; (6) solidified tube; and (7) puller.

the required die passage that results in a balanced mass flow exiting the die and an extrudate shape downstream of the die that matches the target profile. A commercial polymer flow simulation program was used to simulate the three-dimensional die flow and heat transfer through the U-profile die, shown in Fig. 12.[7,26] Because the last two die plates have the greatest influence on the extrudate profile shape, only

these two plates are designed with flow simulation. Simulation requires the following inputs:[7] 1. Geometric model of the die passage a. Two-dimensional profile of the inlet plane of the passage b. Two-dimensional profile of the target extrudate shape.

Fig. 11 Profile vacuum calibration and take-off. (A) Section view of calibration process: (1) melt enters profile die; (2) profile die stack; (3) molten profile extrudate; (4) calibrator (cools, shapes, and sizes extrudate); (5) solidified plastic; (6) puller; (7) orientation of profile. (B) Partially disassembled calibrator: (8) profile passing through calibrator; (9) upper calibrator stack; (10) lower calibrator stack; (11) upper vacuum channel; (12) lower vacuum channel; (13) core feature of lower calibrator; (14) cooling line.

646

Design of Extrusion Dies

Fig. 12 U-Profile die plates designed with CFD simulation.

c. Specification of the preland length, the die land length (note that die land has a constant cross-sectional profile), and the free surface flow length after the die exit. 2. Thermomechanical properties of the polymer melt: density, heat capacity, and thermal conductivity. 3. Rheological properties of the polymer melt: non-Newtonian viscosity as a function of shear rate and temperature and=or viscoelastic material characteristics. 4. Process conditions: inlet melt temperature, mass flow rate into the die passage (or pressure at the inlet), die wall temperature, and take-up speed of the extrudate downstream from the die. The computational domain resembles the real three-dimensional die geometry and a free surface flow after the die, where velocity redistribution (equalization) and stress relaxation take place in a short distance downstream from the die exit (Fig. 13A). Because of the symmetry of the die design, only half of the die passage is modeled (Fig. 13B). A finite element model of the die passage and the free surface region consists of 16,592 hexahedral elements and 19,530 nodes, as detailed elsewhere.[26] The computational domain must have appropriate boundary conditions to represent the realistic conditions present as the melt passes through the die and exits into a free surface flow (Fig. 13C). The used commercial CFD program implements an ‘‘inverse extrusion’’ solution algorithm, which computes the shape of the die exit (die land profile) required to achieve the target profile dimensions at the exit of the free surface domain.[7] The program solves for the shape of the

die land that will achieve the target profile after die swell occurs.[7,26]

Cooling Simulation and Calibrator Design For optimal design of the profile extrusion calibrators, cooling bath, and other cooling accessories, a comprehensive knowledge about the extrudate heat transfer process (cooling) is necessary. The biggest challenge in modeling of profile extrudate cooling is to specify properly the boundary conditions in every local part of the cooling equipment. It is possible to approximate heat transfer coefficients or determine the values experimentally.[27,28] It may be very difficult to estimate the heat transfer coefficient in a vacuum calibrator because it is not possible to predict, without experimental verification, where the polymer has a good contact with the cooling wall and what is the influence of a thin layer of cooling water being sucked in from the cooling bath. However, even estimated values can be used to get a good overall picture of the process, as the polymer materials have a fairly low thermal conductivity. This means that the obtained results are not exact, but they can be very useful for design. Therefore, the modeling and simulation of extrudate cooling is a useful tool for studying the profile extrusion cooling process, as well as for design improvement of the calibrators and other cooling equipment. Other researchers also indicated that calibration design can be done to estimate the cooling performance.[29,30] An illustration of the type of information that can be obtained by simulating the cooling of the extrudate is shown by cooling simulation of a U-profile extrudate using a commercial simulation program and

Design of Extrusion Dies

647

D

Fig. 13 Computational model for U-Profile die design: (A) preland, die land, and free surface as computational domain; (B) finite element mesh (symmetry exploited to reduce computational requirements); (C) boundary conditions for simulation of polymer flow through die and extrudate free surface; and (D) relevant profiles: (1) preland inlet; (2) die land (uniform along flow length); (3) final free surface (target extrudate profile); and (4) symmetry plane.

experimentally determined heat transfer coefficients in a vacuum calibrator.[26]

CONCLUSIONS AND DESIGN RECOMMENDATIONS As stated above, extrusion die design is a complex task because the extrudate product dimensions depend not only on the die design (die shape), but also on the polymer properties and extrusion process parameters. The following are general recommendations for extrusion die design:
Achieve a balanced melt flow exiting the die.


Minimize the pressure drop required to achieve a balanced flow to permit the maximum mass flow rate with the smallest-sized extruder required.
Provide flow control devices in the die to optimize the flow distribution.
Streamline the die flow passage to avoid flow stagnation areas. Such areas facilitate degradation of the polymer melt due to prolonged exposure at elevated temperatures.
Use modular design with stacked plates for manufacturability, convenient assembly, and disassembly, as well as convenient modifications and cleaning.
Die land length should be at least 10 times the product thickness (or gap) to facilitate the polymer melt stress relaxation within the die.

648


Avoid thick and nonuniform extrudate wall thickness to achieve better flow balance control in the die, minimize material use, reduce cooling times, and minimize postextrusion warping of the product.
Avoid or minimize hollow profiles as they increase die fabrication costs and complicate the cooling process of the extrudate. Except for circular dies, it is virtually impossible to design a die geometry to achieve a quality extrudate product for a wide range of polymers and extrusion process conditions. That is why a good die design must incorporate appropriate adjustment features to be set (or tuned) during the extrusion process to compensate for the deficiency of the final product, i.e., the cooled extrudate. For a fixed die geometry, adjustment of the deficient profile may be achieved by changing extrusion process parameters, like temperature, flow rate, cooling rate, and=or take-up speed. However, it is important to optimize the die design to make the necessary adjustments practically possible. This is why polymer extrusion die design has most often relied on experience, empirical data, and expensive trial and error adjustments to design and optimize a die and complementary process parameters. However, by integrating computational simulation with empirical data and by improving the extrusion monitoring instrumentation the die design process can be improved. A better die design method yields improved product quality and a reduction in the time to design and optimize the extrusion process, resulting in lower costs. It is important to state again that computational simulation and empirical extrusion engineering are synergistic in nature. They have their exclusive strengths and weaknesses that cannot replace each other, but, if properly integrated, may significantly improve extrusion die design.

REFERENCES 1. Tadmor, Z.; Gogos, C.G. Die forming. In Principles of Polymer Processing; John Wiley & Sons, 1979; 521–524. 2. Rosato, D.V. Die design and performance. In Extruding Plastics; Chapman & Hall, 1998; 228–282. 3. Rauwendaal, C. Die forming. In Understanding Extrusion; Hanser, 1998; 107–109. 4. Tadmor, Z.; Gogos, C.G. Polymer melt rheology. In Principles of Polymer Processing; John Wiley & Sons, 1979; 148–172. 5. Michaeli, W. Monoextrusion dies for thermoplastics. In Extrusion Dies for Plastics and Rubber, 2nd Ed.; Hanser, 1992; 195–198.

Design of Extrusion Dies

6. Woei-Shyong, L.; Hsueh-Yu, H. Experimental study on extrudate swell and die geometry of profile extrusion. J. Polym. Eng. Sci. 2000, 40 (5), 1085–1094. 7. Polyflow (application software); Fluent Inc.: Lebanon, NH; http:==www.fluent.com=software= polyflow (accessed Mar 2005). 8. Flow 2000; Compuplast International: Zlin, Czech Republic; http:==www.compuplast.com= FLOW_2000.htm (accessed Mar 2005). 9. Dieflow: Chippewa Falls WI; http:==www. dieflow.com (accesses Mar 2005). 10. HyperXtrude, Altair Engineering, Inc.: Troy, MI; http:==www.altair.com=software=hw_hx.htm (accessed Mar 2005). 11. Michaeli, W. Monoextrusion dies for thermoplastics. In Extrusion Dies for Plastics and Rubber, 2nd Ed.; Hanser, 1992; 128–194. 12. Levy, S.; Carley, J.F. Extrusion dies for specific product lines. In Plastics Extrusion Technology Handbook, 2nd Ed.; Industrial Press, Inc., 1989; 96–139. 13. Extrusion Dies Industries, LLC: Chippewa Falls, WI; http:==www.extrusiondies.com (accessed Mar 2005). 14. Michaeli, W. Monoextrusion dies for thermoplastics. In Extrusion Dies for Plastics and Rubber, 2nd Ed.; Hanser, 1992; 147–148. 15. Levy, S.; Carley, J.F. On-line and computer control of the extrusion process. In Plastics Extrusion Technology Handbook, 2nd Ed.; Industrial Press, Inc., 1989; 302–304. 16. Tadmor, Z.; Gogos, C.G. Die forming. In Principles of Polymer Processing; John Wiley & Sons, 1979; 533–537. 17. Michaeli, W. Coextrusion dies for thermoplastics. In Extrusion Dies for Plastics and Rubber, 2nd Ed.; Hanser, 1992; 234–238. 18. Levy, S.; Carley, J.F. Extrusion dies for specific product lines. In Plastics Extrusion Technology Handbook, 2nd Ed.; Industrial Press, Inc., 1989; 228–233. 19. Gifford, W.A. A three-dimensional analysis of coextrusion in a single manifold flat die. J. Polym. Eng. Sci. 2000, 40 (9), 2095–2100. 20. Michaeli, W. Coextrusion dies for thermoplastics. In Extrusion Dies for Plastics and Rubber, 2nd Ed.; Hanser, 1992; 215–221. 21. Levy, S.; Carley, J.F. Extrusion dies for specific product lines. In Plastics Extrusion Technology Handbook, 2nd Ed.; Industrial Press, Inc., 1989; 188–213. 22. Michaeli, W. Calibration of pipes and profiles. In Extrusion Dies for Plastics and Rubber, 2nd Ed.; Hanser, 1992; 311–326.

Design of Extrusion Dies

23. Michaeli, W.; Pfannschmidt, O.; Franz, A.; Vogt, N. Pre-computing developments—progress report on process simulation in industry. Kunststoffe 2001, 91 (7), 32–39. 24. Brown, R.J. Predicting How the Cooling and Resulting Shrinkage of Plastics Affect the Shape and Straightness of Extruded Products. Proceedings of the Annual Technical Conference of the Society of Plastics Engineers, May 4–8, 2000. 25. Sheehy, P.; Tanguy, P.A.; Blouin, D. A finite element model for complex profile calibration. J. Polym. Eng. Sci. 1994, 34 (8), 650–656. 26. Reifschneider, L.G.; Kostic, M.K.; Vaddiraju, S.R. Computational Design of a U-Profile Die and Calibrator. Proceedings of the Annual Technical Conference of the Society of Plastics Engineers, Chicago, May 16–20, 2004.

649

27. Michaeli, W. Calibration of pipes and profiles. In Extrusion Dies for Plastics and Rubber, 2nd Ed.; Hanser, 1992; 324–329. 28. Fredette, L.; Tanguy, P.A.; Hurez, P.; Blouin, D. On the determination of heat transfer coefficient between PVC and steel in vacuum extrusion calibrators. Int. J. Num. Methods Heat Fluid Flow 1996, 6, 3–12. 29. Placek, L.; Svabik, J.; Vlcek, J. Cooling of Extruded Plastic Profiles. Proceedings of the Annual Technical Conference of the Society of Plastics Engineers, May 4–8, 2000. 30. Carneiro, O.S.; Nobrega, J.M.; Covas, J.A.; Oliveria, P.J.; Pinho, F.T. A Study of the Thermal Performance of Calibrators. Proceedings of the Annual Technical Conference of the Society of Plastics Engineers, Chicago, May 16–20, 2004.

D

Desulfurization D Chunshan Song Uday T. Turaga Xiaoliang Ma Clean Fuels and Catalysis Program, The Energy Institute, and Department of Energy and Geo-Environmental Engineering, The Pennsylvania State University, University Park, Pennsylvania, U.S.A.

INTRODUCTION Desulfurization of hydrocarbon streams is an important process used in a petroleum refinery to reduce the sulfur concentration in fuels such as gasoline, jet fuel, kerosene, diesel, and heating oil so that the resulting fuels meet environmental protection standards.[1–7] Hydrotreating is one of the most popular and widely practiced desulfurization processes and refers to the catalytic removal of sulfur [hydrodesulfurization (HDS)], nitrogen [hydrodenitrogenation (HDN)], oxygen [hydrodeoxygenation (HDO)], and metals [hydrodemetallation (HDM)] from petroleum distillates in the presence of hydrogen. Hydrotreating first appeared in petroleum refineries as a finishing process in the 1930s.[2] Hydrotreating operates at conditions milder than those typically used in fluidized catalytic cracking (FCC) or hydrocracking. Typical hydrotreating process conditions vary with feedstock and are summarized in Table 1. In the past two decades petroleum refining has changed extensively and the fortunes of hydrotreating, in particular, have witnessed a sea change. Hydrotreaters now occupy a central role in modern refineries and more than 50% of all refinery streams now pass through hydrotreaters for conversion, finishing, and pretreatment purposes.[2] Hydrodesulfurization is the largest application of catalytic technology in terms of the volume of material processed.[8] On the basis of usage volume, HDS catalysts are ranked third behind catalysts used for automobile emission control and FCC.[8] Commercial hydrotreating catalysts are, typically, sulfides of Mo or tungsten (W) supported on g-Al2O3
and promoted by either Co or Ni. Nickel, known for its high hydrogenation activities, is preferred as a promoter when feedstocks containing high amounts of nitrogen and aromatics need to be processed. Table 2 provides the compositional range and physical properties of typical hydrotreating catalysts in the oxidic phase.

GROWING DEMAND FOR DESULFURIZATION The challenge of fulfilling the world’s growing transportation energy needs is no longer a simple issue of Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120007732 Copyright # 2006 by Taylor & Francis. All rights reserved.

producing enough liquid hydrocarbon fuels. This challenge is instead accentuated by a complex interplay of environmental and operational issues. Environmental issues include societal demands that liquid hydrocarbon fuels be clean and less polluting. The emergence of new refining processes and the increasing use of new forms of energy production, e.g., fuel cells, exemplify operational issues. Together, these trends are driving the need for deep desulfurization of diesel and jet fuels. As an example of the kind of regulations that are being specified and contemplated for hydrocarbon fuels, Table 3 lists the compositional and performance properties for diesel. The United States Environmental Protection Agency (USEPA) has mandated that diesel— whose automotive use is now growing at a pace faster than that of gasoline—have no more than 15 parts per million by weight (ppmw) of sulfur by 2006.[9,10] This represents a 97% reduction in the allowable sulfur concentration in diesel from 500 to 15 wppm. The United States now allows up to 3000 wppm of sulfur in jet fuel.[11] With the European Union now demanding no more than 1000 wppm of sulfur in jet fuel, a reduction in the permissible sulfur content of U.S. jet fuel can be expected.[12] New gasoline sulfur regulations will require most refiners to meet a 30 ppmw sulfur average with an 80 ppmw cap for both conventional and reformulated gasoline by January 1, 2006. Initially, fuel sulfur was regulated to reduce emissions of the oxides of sulfur, which contribute to acid rain, ozone, and smog. The recent and stricter round of sulfur specifications, however, are an effort to reduce automobile emissions of the oxides of nitrogen (NOx) and particulate matter (PM). For example, the 15 ppmw diesel sulfur limit follows from the USEPA’s parallel program of rule making that seeks to reduce automobile NOx and PM emissions by 95% and 90%, respectively, by 2007. Automobile manufacturers are demanding ultra-low-sulfur fuels because only then would their advanced, sulfur-sensitive after-treatment technologies achieve such drastic reductions in NOx and PM emissions.[13,14] Sulfur specifications are the more visible drivers for desulfurization research. Fig. 1 presents a qualitative 651

652

Desulfurization

Table 1 Typical hydrotreating process conditions for various feedstocks Feedstock

Temperature ( C)

Hydrogen pressure (atm)

LHSVa (hr
1)

Naphtha

320

15–30

3–8

Kerosene

330

30–45

2–5

Atmospheric gas oil

340

38–60

1.5–4

Vacuum gas oil

360

75–135

1–2

Atmospheric residue

370–410

120–195

0.2–0.5

Vacuum residue

400–440

150–225

0.2–0.5

a

LHSV, liquid hourly space velocity. (From Ref.[2].)

relationship between the type and size of sulfur molecules in various distillate fuel fractions and their relative reactivities.[5] Various refinery streams are used to produce three major types of transportation fuels, gasoline, jet fuels, and diesel fuels, that differ in composition and properties. Other fuel specifications are equally important, albeit less visible, reasons for continued research and development in desulfurization. For example, European countries typically require diesel fuel with sulfur less than 50 wppm, cetane number of 48–51, and density less than 0.825 g=cm3. These fuel specifications are more stringent than those enforced in the United States (see Table 3). To be active in the European diesel fuels market—larger than that in the United States—oil refiners will have to produce ultraclean, high-quality, premium diesel fuel through sophisticated refining

Table 2 Composition and properties of typical hydrotreating catalysts Composition and propertiesa

Range

Typical values

Active phase precursors (wt%) MoO3 CoO NiO

13–20 2.5–3.5 2.5–3.5

15 3.0 3.0

Promoters (wt%) SiO2 B, P

0.5–1 0.5–1

0.5 0.5

150–500 0.25–0.8

180–300 0.5–0.6

3–50 100–5000 0.8–4 2–4 0.5–1.0 1.0–2.5

7–20 600–1000 3 3 0.75 1.9

Physical properties Surface area (m2=g) Pore volume (cm3=g) Pore diameter (nm) Mesopores Macropores Extrudate diameter (mm) Extrudate length=diameter Bulk density (g=ml) Average crush strength= length (kg=mm) a

Active phase precursors and promoters are supported on a g-Al2O3 carrier.

processes such as the selective ring opening (SRO) of naphthenes.[15] Ring opening is commonly observed in the hydrocracking process where one or more carbon–carbon (C–C) bonds are broken. Selective ring opening, in contrast, breaks only one C–C bond open, thus preventing extensive reduction in molecular weight.[16] This process, shown for a model diaromatic molecule in Fig. 2, leads to extensive improvement in the quality of diesel fuel by increasing cetane number and decreasing fuel density simultaneously.[16] Hydrodesulfurization is important for the SRO process because the most effective SRO catalysts are based on noble metals such as iridium, which are highly sensitive to sulfur. Therefore, all SRO feedstocks will have to be extensively desulfurized and hydrogenated before being sent to an SRO reactor. Selective ring opening, while fulfilling important nonsulfur fuel specifications, is an emerging refining process that symbolizes a compelling operational issue that could require more active HDS catalysts. In recent times, there has been tremendous interest in fuel cells. This interest is set to intensify with the U.S. government’s new Freedom Cooperative Automotive Research (Freedom CAR) program that seeks to develop cars based on hydrogen-powered fuel cells.[17] Enough hydrogen to satisfy a fuel cell-based transportation system will have to be produced by processing hydrocarbon fuels through reforming and related processes all of which are sensitive to sulfur. While most fuel cells are also sulfur sensitive, some are intolerant to as little as 0.1 wppm sulfur, e.g., polymer electrolyte membrane fuel cell (PEMFC).[7] As the sulfur compounds in liquid hydrocarbon fuels and the H2S produced from these sulfur compounds in the hydrocarbon reforming process are poisonous to both the catalysts in hydrocarbon fuel processor and the electrode catalysts in fuel cell stack, the sulfur content in the liquid hydrocarbon fuels has to be reduced to a very low level [98% selectivity was reported.[26] The synthesis of cyclohexanol by the selective hydration of cyclohexene in a mixture of benzene, cyclohexene, and cyclohexane with very close boiling points was reported.[27] This process configuration involves feeding a slurry containing hydrophilic H-ZSM5 zeolite particles into an RD column. The catalyst particles were recovered in the aqueous phase. Dehydration The production of diisopropyl ether by the dehydration of isopropyl alcohol and the simultaneous removal of the product diisopropyl ether and water at the reaction zone was reported to increase the conversion of this equilibrium limited reaction.[21] The dehydration of methanol to produce dimethyl ether was also reported.[28] The dehydration of tert-butanol to produce isobutylene using b-zeolite, hydrofluorine (HF)-treated zeolite, and HF-treated montmorillonite has been reported to give a higher conversion of tert-butanol due to the in situ separation via CD.[29–31] It was reported that at a pressure of 80 kPa and 64–99
C using b-zeolite, conversion of 99% and selectivity of 94.5% were obtained.[31] The preparation of isobutene from a slurry CD process by feeding fine particles of acid cation-exchange resins with the tert-butanol was reported to give a 99.99% conversion.[32] Recently, the production of high-purity (>98%) tetrahydrofuran from the dehydration of 1,4butanediol using cation-exchange resins was reported. Furthermore, this process was carried out using a new technique involving the use of a dual column operating at different pressures.[33,34]

2603

A number of advantages of CD were obtained for the exothermic alkylation process and particularly noteworthy is the increased catalyst lifetime and enhanced selectivity to monoalkylated rather than dialkylated or trialkylated product. Catalytic Distillation Technology commercialized the production of ethylbenzene using the CD EBÕ technology in 1994 at the Mitsubishi Petrochemical in Yokkaichi, Japan. The CD CumeneÕ process was first brought onstream in 2000 at a capacity of 270,000 MTA by Formosa Chemicals and Fibre Corporation, Taiwan, and was expanded to double the capacity since 2004. A schematic of a CD process for cumene production is shown in Fig. 3. The CDTECH process uses a zeolite catalyst in one of its patented CD structures and the product yield exceeds 99.5% purity with 99.9% selectivity to cumene. The high selectivity to cumene is achieved by controlling a low propylene concentration in the reaction section using a combination of process parameters such as system pressure, location of catalyst zone, and feedpoint. A low propylene concentration will result in a low propylene oligomerization rate and hence will reduce the amount of diisopropylbenzene and triisopropylbenzene produced by the consecutive reactions. One interesting aspect of this process is to recycle the diisopropylbenzene and triisopropyl benzene where transalkylation with benzene produces more cumene. Recently, a novel CD process for the alkylation of benzene with propylene using suspended catalyst rather than encasing the catalyst in a rigid structure with 100% conversion for propylene with more than 90% selectivity to high-purity cumene was reported.[36] An improvement of the suspension CD by simultaneous alkylation and transalkylation for producing

Alkylation An important industrial application of CD is the alkylation of benzene with ethylene or propylene to produce ethylbenzene or cumene, respectively, using acidic ion-exchange resins such as Amberlyst or zeolites operating at 130–5065 kPa and 80–500
C.[35] Cumene is a chemical intermediate for the production of phenol, acetone, and alpha-methyl styrene, which are used to produce resins and solvents. Ethylbenzene is an intermediate for styrene, an important monomer for polymers. Alkylation of benzene could also be used to reduce the carcinogenic benzene content of gasoline.

Fig. 3 A schematic of a CD process for cumene production: 1) CD column and 2) distillation column.

R

2604

cumene was also reported using a modified zeolite catalyst.[37] Suspension CD involves the use of small catalyst particles inside the distillation column and the separated catalyst particles can be recycled and=or regenerated before recycling. Although suspension CD overcomes the difficulty of getting suitable CD packings, it requires the separation and recycling of the catalyst particles, which will increase the cost of the process. Currently, there is no commercial process using suspension CD. The alkylation of benzene with C2–C4 olefins has also been patented using the acid form of the b-zeolite.[38] Selective alkylation of toluene with C2–C4 olefins using acidic zeolites such as Y-, b-, or Ozeolites followed by transalkylation with benzene has also been patented.[39] A recent paper reported an increased catalyst lifetime for the alkylation of benzene with ethylene at a relatively high ethylene to benzene ratio using Y-zeolite contained in high-efficiency packing.[40] This paper clearly shows that many of the advantages of CD are due to the fact that selectivity greater than 99.5% to ethylbenzene at a benzene to ethylene molar ratio of 1.5 : 2 was achieved. A recent patent on the use of enriched or pure ethylene at a benzene= ethylene ratio of 1 : 1 for the production of ethylbenzene was granted.[41] The alkylation of FCC off-gas with benzene to produce ethylbenzene by CD was recently reported.[42] A rectified RD configuration could be used to increase the yield of monoalkylated product.[43] Besides the production of cumene and ethylbenzene, there are a number of recent reports on the production of linear alkylbenzene, precursors to detergents, via the alkylation of benzene with C6–C18 olefins. One process uses suspension CD and essentially 100% conversion of olefin at low temperatures of 90–100
C was obtained.[44] An HF-treated mordenite used in the alkylation of benzene and C10–C14 olefins was found to give a 74–84% selectivity to linear alkylbenzene containing 80% 2-phenyl isomer.[45] A new patent on the alkylation of aromatic hydrocarbons such as benzene and cumene with straight-chain C6–C20 olefins on acidic catalyst such as zeolites or fluorine-treated zeolite catalyst packed in a Katamax-type packing was granted.[46] A patent application on the manufacture of xylenes from reformate by RD also appeared and higher than equilibrium amounts of para-xylene were claimed.[47]

Hydrogenation Catalytic distillation hydrogenation is one of the more recent applications of CD that was commercialized by CDTECH for the selective hydrogenation of dienes in C4–C6 streams and the saturation of benzene in the

Recent Advances in Catalytic Distillation

aromatics under the trade name CD HydroÕ. Commercial applications of CD Hydro include selective hydrogenation of butadiene in a mixed C4 refinery stream, selective hydrogenation of pentadiene and hexadiene, hydrogenation of benzene to cyclohexane and hydrogenation of acetylene in a C4 stream.[48] A process for the hydrogenation of benzene in a reformate stream was commercialized in Texaco’s refinery at Bakersfield, TX, U.S.A., to produce gasoline with less than 10 ppm benzene in 1995. Conventional Ni or Pd catalyst supported on alumina placed in a distillation packing consisting of flexible, semirigid openmesh tubular element was used as CD packing. The process operates at 1378–1723 kPa and 149–204
C compared to 2412–3101 kPa and 260
C for a fixedbed hydrogenation process. The highly exothermic reaction is controlled by recycling the condensed overhead to the column and therefore safer than conventional fixed-bed hydrogenation processes, which require a high mass flow rate or efficient heat exchangers to remove the reaction heat. The heat of hydrogenation could also be used for vaporization of the feed in the catalyst zone resulting in a near-isothermal operation, which improves catalyst lifetime since the sintering of catalyst is reduced. It was reported that the CD Hydro process could be installed at about 25% less capital cost than a conventional reformate splitter followed by a fixed-bed hydrogenation unit. It should be noted that there are a number of benefits associated with a hydrogenation process carried out in a CD column. Besides significant energy savings and hence reduction of greenhouse gases such as carbon dioxide, there is improved process safety due to the lower temperature and pressure required for a CD hydrogenation process and the near-isothermal temperature profile in the reaction zone, which reduces the possibility of runaway reactions. The lower pressure requirement also eliminates the need for a hydrogen compressor. In a recent patent on the hydrogenation of cyclopentadiene, it was shown that the pressure required for hydrogenation in a CD column was 8 psig and 175
F whereas greater than 200 psig and 200–450
F were used in a conventional hydrogenation reactor to achieve 99% conversion.[49] It was also reported that the catalyst was found to be more stable in the selective hydrogenation of methylacetylene and propadiene in a C3 fraction in a CD column than in a trickle-bed reactor.[50] An extended catalyst lifetime in the hydrogenation of C3 fraction was also reported.[51] The phenomena of lower hydrogen pressure and temperature required for hydrogenation in a CD column is particularly noteworthy. Experimental and modeling studies are being carried out by our research group to understand this distinctive feature of CD hydrogenation. This is likely related to the enhanced mass transfer of hydrogen to the catalyst in a CD process.

Recent Advances in Catalytic Distillation

In 2001, a CD process for the production of 2,2,4trimethylpentane with 100 octane rating from the dimerization of isobutene and hydrogenation of the octenes using a Pd=cation-exchange resin was disclosed, but no data were given in the patent.[52] Hydrogenation in a CD column has been now applied to the hydrocracking and hydrotreating processes for vacuum gas oil to produce diesel and lighter distillates using different conventional hydrogenation catalysts such as Ni=Mo on alumina and Pd= alumina=zeolite.[53] Another related patent disclosed that hydrocracking together with posttreatment of the hydrocracked fraction via RD resulted in the reduction of hydrogen consumption and reduction of the overall reactor and catalyst volumes for a given level of performance.[54] It appears that the application of CD to the treatment of petroleum fractions will gain more attention in the near future. An improvement on the catalytic hydrogenation of acetylenes and dienes in the C2–C5 fraction in a thermally cracked feed stream without significantly hydrogenating the C2 and C3 olefins was achieved using a combination of CD and fixed-bed catalytic steps.[55]

Desulfurization With the new legislation on the reduction of S content in gasoline and diesel worldwide, production of low-S gasoline and diesel fuel is of great importance. Catalytic

2605

distillation was found to be particularly effective for the desulfurization of naphtha and gasoline produced from FCC.[4,56] In 2000, a new commercial CDTECH process for the desulfurization of FCC gasoline (CD HDSÕ) came onstream at Motiva in Texas, U.S.A. CD HDS takes advantage of the fact that the light olefins are fractionated to the top of the column while the benzothiophenes and substituted thiophenes are fractioned to the bottom of the column. Therefore, CD HDS minimizes the hydrogenation of light olefins and preserves the octane number of gasoline. The catalysts used are conventional hydrodesulfurization (HDS) catalysts such as Co=Mo and Ni=Mo on alumina supports, which are put inside catalyst structures composed of flexible, semirigid open-mesh tubular material such as stainless steel wire mesh.[57] A CD process, CD Hydro=CD HDS, utilizing two-distillation columns for the desulfurization of a naphtha stream with a minimum loss of olefins and octane number was disclosed.[58] The naphtha is fed to the first column containing hydrogenation catalysts such as Ni sulfide and Pd oxide to catalyze the reaction of mercaptans with some diolefins to form olefinic thioethers, which are then sent to the second distillation column together with the heavy sulfur compounds where desulfurization occurs (Fig. 4). This CD process produces gasoline with a minimal mercaptan content and eliminates the caustic treatment process used to remove mercaptans. It was reported that a combination of CD Hydro and CD HDS could reduce the FCC gasoline sulfur by 90% while the octane loss is very minimal.

Fig. 4 A schematic of a CD Hydro=CD HDS process for production of low-sulfur gasoline. FRCN, full-range catalytic naphtha; LCN, light catalytic naphtha; MCN, medium catalytic naphtha; HCN, heavy catalytic naphtha.

R

2606

This combined process could produce low-sulfur gasoline at half the current cost of desulfurization and won the Brian Davis Refining Technology award in 1999. The start-up of the first CD Hydro=CD HDS unit was at Irving Oil’s St John, New Brunswick Refinery, Canada, in 1999 and produced less than 150 ppm S.[59] In 2000, Irving Oil received a U.S. Environmental Award. In 2004, the CD Hydro=CD HDS process was used to process about 1.0 million barrels per day of gasoline for mercaptans and sulfur removal in refineries all over the world. Recently, a review of the production of ultralow-sulfur gasoline, a patent application for the desulfurization of full-range naphtha by thioetherification and HDS, and the sulfidation of HDS catalysts utilizing CD also have appeared.[60–62] A recent paper on the comparison between the HDS in a conventional trickle-bed reactor and CD for light gas oil HDS concluded that the CD process for HDS is more efficient because the distillation process reduces the inhibitory effects of the H2S and allows the HDS of the more refractory sulfur compounds to produce a low-sulfur diesel fuel and enhanced catalyst lifetime.[63] Application of CD for the HDS of diesel fuel is more challenging owing to the high boiling point of the diesel fraction although a design for the ultra-low-sulfur diesel production was published.[64] A stripped RD column could be applied in the hydrocracking and hydrotreating of petroleum to avoid excessive hydrocracking or hydrotreating of lighter products.[43] If the desired products are of higher molecular weight, a rectified RD column can be used.

Esterification There are many examples of the application of CD or RD for esterification.[4,11] Esterification of methanol or ethanol with acetic acid forms methyl acetate or ethyl acetate, respectively. Methyl acetate is important in the manufacture of polyesters and is an important solvent for cellulose while ethyl acetate is used in inks, fragrances, and pharmaceuticals. The manufacture of high-purity methyl acetate is difficult because of the equilibrium limitation and also the formation of azeotropes. The production of methyl acetate by Eastman Chemical Co. was the first commercial application of RD using a homogeneous liquid acid catalyst. Only one RD column and two smaller columns for processing sidestreams are required while in the conventional methyl acetate synthesis, two reactors and eight distillation columns are required. Catalytic distillation was also used to produce methyl acetate. A macroporous acidic ion-exchange resin fixed in an open cross-flow structure packing was used in the CD column where the acetic acid was

Recent Advances in Catalytic Distillation

fed above the catalyst zone and the methanol was fed below the catalyst zone.[65] Because the acidic ion-exchange resin provides at least 20 times more protons per unit volume of reacting liquid than a homogeneous catalyst, the residence time to achieve a given level of conversion can be reduced. The process operates at atmospheric pressure with virtually complete conversion of methanol and acetic acid in equimolar proportions and only 10 theoretical plates are required. The CD process also eliminates the need for downstream treatment of liquid acid. A pilot scale production of butyl acetate in a CD process using a Katapak in combination with a CY packing from Sulzer Chemtech containing an acidic ion-exchange resin was reported.[66] In the past few years, a large number of CD processes on esterification using Katapak packing such as Katapak-S or Multipak packing together with solid acid catalyst such as Amberlyst 15 have been reviewed in detail and hence these processes will not be reviewed here.[11] All these processes operate at mild temperature and pressure with conversions and selectivities in excess of 95% and in some cases close to 100%.

Dimerization and Oligomerization The oligomerization of olefins is an exothermic consecutive reaction, which benefits from the application of CD for enhanced selectivity to intermediate products. Catalytic distillation plays a particularly important role in enhancing the catalyst lifetime because in situ separation reduces the undesirable high-molecularweight oligomers or polymers, which will form coke and deactivate the catalyst. The use of reaction heat for distillation also reduces the formation of hot spots and catalyst deactivation due to sintering. Because of the phase out of MTBE, octenes will take on a more important role in the gasoline pool. The selective dimerization of C4 in a raffinate stream to octenes from the steam cracking of naphtha could be used to produce octane enhancers. In addition, there is an existing source of isobutylene due to the phase out of MTBE, which could be selectively dimerized to 2,4,4-trimethylpentene-1 and 2,4,4-trimethylpentene-2, which on hydrogenation will produce 2,2,4-trimethylpentane with an octane number of 100. Smith et al. have patented a CD process for dimerization using an acidic cation-exchange resin.[67] A patent application on the dimerization of light olefins using a combination of RD and conventional reactors was filed.[68] We have studied the oligomerization of butenes in a CD column. We found that when a Ni-zeolite was encased in a fiberglass bag as a CD packing, owing to the mass transfer resistance of the fiberglass bags, rapid catalyst deactivation due to the formation of

Recent Advances in Catalytic Distillation

higher oligomers on the catalyst was observed. A novel CD packing containing Ni was found to be very effective for the selective dimerization of butenes to octenes.[69] Such CD packing containing Group VIII metals such as Pd was also very active for the hydrogenation of the C8 alkenes. We found that a combination of dimerization and hydrogenation CD packing provides a high conversion of isobutylene with high selectivity to 2,2,4 trimethylpentane and long catalyst lifetime time (Fig. 5). It was also reported that the dimerization of isobutene was more selective in the presence of tert-butyl alcohol.[70] A patent application on the production of oligomer from isobutane with CD that combines dehydrogenation and oligomerization was reported.[71] A process for the dimerization of isobutene and the hydrogenation in the presence of S compounds was patented.[72] Recently, the use of CD to produce a lubebase stock from lower-molecularweight feedstock and acidic catalyst such as zeolites was disclosed.[73] A CD process for improving yields of higher-molecular-weight olefins from lower-molecularweight olefins using isomerization and disproportionation was also patented.[74]

Aldol Condensation and the Production of Methyl Isobutyl Ketone Aldol condensation of acetone produces diacetone alcohol (DAA), an environmentally friendlier solvent due to its low volatility and high boiling point, 165.6
C. Aldol condensation of acetone to produce DAA is strongly limited by equilibrium. At the boiling point of acetone, the equilibrium conversion of acetone

2607

is only 4.3%. Diacetone alcohol also undergoes dehydration to mesityl oxide (MO). HO

O 2 HC 3

H3C

CH3

CH3 CH3 O DAA

Acetone

CH3

H3C

+ H2O CH3

(2)

O

MO CH3

H3C CH3 MO

O

+H

H3C

CH3

(3)

2

CH3

O

MIBK

Experimental studies in our laboratory on the aldol condensation were carried out in a 1 in. CD column using Ambelite IRA-900 anion-exchange resin housed in fiberglass bags. The reboiler duty, which affected the flow rates, was found to play an important role in the selectivity to DAA.[75,76] A rate-based three-phase CD model was developed, which accurately predicts the yield and selectivity obtained under steady-state and transient conditions.[77–79] Model predictions and experimental data indicate that the production of DAA is external mass transfer controlled while the production of MO is kinetically controlled. The external mass transfer resistance was caused by the fiberglass bags. Recently,

Fig. 5 A schematic of a CD process for the production of isooctane. (A) Dimerization and hydrogenation in two separate zone. (B) Dimerization and hydrogenation in the same zone.

R

2608

we prepared beryl saddles coated with MgO and used it as CD packing without any fiberglass bags; a higher selectivity to DAA was obtained. A one step synthesis of methyl isobutyl ketone (MIBK) via the aldol condensation of acetone and the in situ hydrogenation of MO was patented.[80,81] However, it is clear that a CD process for a one-step synthesis of MIBK from acetone and hydrogen is very complicated because MO can undergo a large number of reactions such as the production of highermolecular-weight phorones besides being hydrogenated to MIBK. In addition, the presence of azeotropes and phase separations due to the production of water from the dehydration of DAA makes the one-step synthesis of MIBK from the aldol condensation of acetone very challenging.

OTHER CD APPLICATIONS Besides the above CD processes, some of the recent novel applications of CD are outlined below. The production of amines from the hydrogenation of aniline and the selective production of diethanolamine from the reaction of monoethanolamine and ethylene oxide have been reviewed.[4] A patent on the production of phenol from cumene hydroperoxide disclosed that solid acid catalysts such as zeolites, ion-exchange resins achieved 100% conversion with about 60% selectivity to phenol at 50–90
C and 0–10 psig.[82] This process utilizes the heat of the decomposition of cumene hydroperoxide to effect the separation of the lower boiling components and hence reduces the energy cost and carbon dioxide emissions. An interesting application of CD for the production of methanol from syngas (a mixture of CO and H2) was patented in 1999.[83] Recently, another process on the methanol production from syngas using a CD column and multiple distillation stages, at least three, was disclosed.[84] This process utilizes CD to increase the conversion of synthesis gas to methanol beyond the equilibrium limitation using catalysts such as Cu–Zn, or RaneyTM in the form of sponge. A patent on the application of CD to produce alpha-olefins via the isomerization of internal olefins was granted.[85] A new CD process for the production of vinyl acetate from acetic acid, ethylene, and oxygen using a Pd-type catalyst at 338–420 K, 2–5 bar was disclosed.[86] This illustrates the wide-ranging possibilities for the application of CD in a variety of processes for the chemical, petrochemical, and petroleum industry. The production of acetic acid from the carbonylation of dimethyl ether or methanol using RD and homogeneous catalyst was also patented.[87]

Recent Advances in Catalytic Distillation

Applications of CD for Separations Catalytic distillation can also be used for selective separations such as the separation of piperidine from n-amylamine, separation of isobutylene in a C4 stream, and removal of acetic acid from dilute aqueous streams.[4] The application of CD for separations will not be reviewed in this article. The potential use of RD for the separation of chiral compounds is very noteworthy although no corresponding CD process was reported.[88]

CONCLUSIONS Catalytic distillation is a rapidly developing field with applications in many processes for the chemical, petrochemical, and petroleum industry. It is emerging as a tool for green process and technology innovations because it utilizes process intensification to achieve energy efficiency and the reduction of greenhouse gases. With the advances in the fundamental understanding of the reaction engineering aspects of CD based on both experimental research and mathematical modeling, CD is becoming a valuable tool for the development of new green chemical and petrochemical processes. Some challenges for CD include the development of robust catalysts or CD packings that will have the mechanical strength to withstand high liquid=vapor flow rates in a distillation column and modeling the multiphase flow characteristics in a CD column using computational fluid dynamics. Another important aspect is to understand the behavior of noncondensable gases such as hydrogen in the CD hydrogenations. We are currently carrying out both experimental and mathematical modeling of a new CD process for the production of isooctane from isobutene.

REFERENCES 1. Anastas, P.T.; Zimmerman, J.B. Design through the 12 principles of green engineering. Environ. Sci. Technol. 2003, 37 (3), 94A–101A. 2. Dautzenberg, F.M.; Mukherjee, M. Process intensification using multifunctional reactors. Chem. Eng. Sci. 2001, 56, 251–267. 3. Stankiewicz, A. Reactive separations for process intensification: an industrial perspective. Chem. Eng. Process. 2003, 42, 137–144. 4. Ng, F.T.T.; Rempel, G.L. Catalytic distillation. In Encyclopedia of Catalysis; John Wiley & Sons Inc.: New York, 2002.

Recent Advances in Catalytic Distillation

5. Malone, M.F.; Huss, R.S.; Doherty, M.F. Green chemical engineering aspect of reactive distillation. Environ. Sci. Technol. 2003, 37, 5325–5329. 6. Podrebarac, G.G.; Ng, F.T.T.; Rempel, G.L. More uses for catalytic distillation. CHEMTECH 1997, 27, 37–45. 7. Smith, L.A., Jr. Catalytic Distillation Process U.S. Patent 4,232,177, Nov 4, 1980. 8. Smith, L.A., Jr. Catalytic Distillation Process U.S. Patent 4,307,254, Dec 22, 1981. 9. Smith, L.A., Jr. Catalytic Distillation Process and Catalyst, Canadian Patent 1,125,728, Jun 15, 1982 10. Smith, L.A., Jr. Catalytic Distillation Process U.S. Patent 4,336,407, Jun 22, 1982. 11. Hiwale, R.S.; Bhate, N.V.; Mahajan, Y.S.; Mahajani, S.M. Industrial applications of reactive distillation: recent trends. Int. J. Chem. Reactor Eng. 2004, http:=bepress.com=ijcre=vol2=R1. 12. Taylor, R.; Krishna, R. Modelling reactive distillation. Chem. Eng. Sci. 2000, 5183–5229. 13. Mohl, K.; Kienle, A.; Gilles, E.; Rapmund, P; Sundmacher, K.; Hoffmann, U. Steady-state multiplicities in reactive distillation columns for the production of fuel ethers MTBE and TAME: theoretical analysis and experimental verification. Chem. Eng. Sci. 1999, 54, 1029–1043. 14. Quitain, A.T.; Itoh, H; Goto, S. Reactive distillation for synthesizing ethyl tert-butyl ether from bioethanol. In Reaction Engineering for Pollution Prevention; Abraham, M.A., Hesketh, R.P., Eds.; Elsevier Science B.V.: Amsterdam, Netherlands, 2000; 237–246. 15. Quitain, A.; Itoh, H.; Goto, S. Industrial-scale simulation of proposed process for synthesizing ethyl tert-butyl ether from bioethanol. J. Chem. Eng. Jpn. 1999, 32, 539–543. 16. Adams, J.R.; Smith , L.A., Jr.; Hearn, D; Jones, E.M., Jr.; Arganbright, R.P. Integrated Process for the Production of Tame U.S. Patent 5,792,891, Aug 11, 1996. 17. Bakshi, A.; Hickey, T.P. Etherification Process U.S. Patent 5,919,989, Jul 6, 1999. 18. Hamid, S.H. Handbook of MTBE and Other Gasoline Oxygenates; Marcel Dekker, 2004. 19. Wang, L.; Li, J. Study on synthesis of propylene glycol monoethyl ether with b zeolite as catalysts by catalytic distillation. Jingxi Shiyou Huagong 2003, 5, 17–19. 20. Yu, S-B.; Li, Y-H.; Chen, H-F. ETBE reactive distillation on supported b zeolite membrane. Shiyou Zuebao Shiyou Jiagong 2003, 19 (5), 58–62. 21. Marker, T.L.; Frank, G.A.; Barger, P.T.; Hammershaimb, H.V. Two-Stage Process for Producing Diisopropyl Ether Using Catalytic Distillation U.S. Patent 5,744,645, Apr 28, 1998.

2609

22. Smith, L.A., Jr. Method for Operating a Catalytic Distillation Process U.S. Patent 5,221,441, Jul 22, 1993. 23. Reusch, D.; Beckmann, A.; Nierlich, F.; Tuchlenski, A. Catalytic Hydration Procedure for the Production of Tert-Butanol from Isobutylene and Water and using Reactive Distillation German Patent DE10260991(A1), Jul 8, 2004. 24. Zhang, C.M.; Adesina, A.A.; Wainwright, M.S. Isobutene hydration over Amberlyst-15 in a slurry reactor. Chem. Eng. Proc. 2003, 42 (12), 985–991. 25. Xu, Y.; Chuang, K.T.; Sanger, A.R. Design of a process for production of isopropyl alcohol by hydration of propylene in a catalytic distillation column. Inst. Chem. Eng. Trans. IChem. E 2002, 80, Part A, 686–694. 26. Liu, H.; Qu, Y.; Wang, W. Simulation of hydration process of water and ethylene oxide by reactive distillation to produce glycol. Beijing Huagong Daxue Xuebao Ziran Kexueban 2002, 29 (1), 18–20. 27. Frank, S.; Qi, Z.; Sundmacher, K. Synthesis of cyclohexanol by three-phase reactive distillation: influence of kinetics on phase equilibria. Chem. Eng. Sci. 2002, 57 (9), 1511–1520. 28. An, W.; Chuang, K.T.; Sanger, A.R. Dehydration of methanol to dimethyl ether by catalytic distillation. Can. J. Chem. Eng. 2004, 82 (5), 948–955. 29. Abella, L.C.; Gaspillo, P.D.; Itoh, H.; Goto, S. Dehydration of tert-butyl alcohol in reactive distillation. J. Chem. Eng. Jpn. 2000, 33 (2), 351. 30. Gotze, L.; Bailer, O.; Moritz, P.; Von, S. Reactive distillation with KATAPACK. Catal. Today 2001, 69 (1–4), 201–208. 31. Knifton, J.F.; Sanderson, J.R.; Stockton, M.E. Tert-butanol dehydration to isobutylene via reactive distillation. Catal. Lett. 2001, 73 (1), 55–57. 32. Xue, Y.; Xu, J.; Xu, C.; Zhou, M.; Sheng, Z.; Shang, J. Preparation of isobutene by slurry catalytic distillation process. Huaxue Gongye Yu Gongcheng (Tianjin, China) 2003, 20 (5), 251–255, 274. 33. Liu, Q.; Xiao, J. Synthesis of tetrahydrofuran from butanediol by pressure-sensitive reactive distillation. Jisuanji Yu Yingyong Huaxue 2001, 18 (2), 123–126. 34. Liu, Q; Zhang, F.; Gao, H. Synthesis of tetrahydrofuran from butanediol by reactive distillation. Huaxue Gongcheng 2002, 30 (2), 75–78. 35. Smith, L.A., Jr. Alkylation of Organic Aromatic Compounds U.S. Patent 4,849,569, Jul 19, 1989. 36. Wen, L.; Min, E.; Pang, G.; Yu, W. Synthesis of cumene by suspension catalytic distillation process. Huagong Xuebao (Chinese Ed.) 2000, 51 (1), 115–119.

R

2610

37. Lei, Z.; Li, C.; Li, J.; Chen, B. Suspension catalytic distillation of simultaneous alkylation and transalkylation for producing cumene. Sep. Purif. Technol. 2004, 34 (1–3), 265–271. 38. Hendirksen, D.E.; Lattner, J.R.; Johannes, M.; Janssen, M.J.G. Alkylation Process Using Zeolite Beta U.S. Patent 6,002,057, Dec 14, 1999. 39. Chen, J. Process for the Production of Alkyl Benzene U.S. Patent 5,866,736, Feb 2, 1999. 40. Qi, Z.; Zhang, R. Alkylation of benzene with ethylene in a packed reactive distillation column. Ind. Eng. Chem. Res. 2004, 43 (15), 4105–4111. 41. Zhang, J.; Li, D.; Fu, J.; Cao, G. Process and Apparatus for Preparation of Ethylbenzene by Alkylation of Benzene with Dilute Ethylene Contained in Dry Gas by Catalytic Distillation U.S. Patent 6,504,071, Jan 7, 2003. 42. Xu, L.; Wang, Q.; Liu, W.; Xie, S.; Sun, X.; Bai, J.; Zhang, S.; Wang, C. Thermodynamics of ethylbenzene synthesis through alkylation of FCC off-gas with benzene by catalytic distillation. Cuihua Xuebao 2003, 24 (1), 73–78. 43. Tung, P. Dimensions in Reactive Distillation Technology U.S. Patent 6,500,309, Dec 31, 2002. 44. Wang, E.; Li, C. Simulation of suspension catalytic distillation for the synthesis of linear alkyl benzene. Chin. J. Chem. Eng. 2003, 11 (5), 520–525. 45. Knifton, J.F.; Anantaneni, P.R.; Dai, P.E.; Stockton, M.E. Reactive distillation for sustainable, high 2-phenyl LAB production. Catal. Today 2003, 79–80, 77–82. 46. Winder, J.B.; Wharry, D.L.; Schell, J.R.; Brown, M.J.; Murray, J.L.; Howe, R.C.; Sorensen, W.L.; Szura, D.P. Reactive Distillation Process for the Alkylation of Aromatic Hydrocarbons U.S. Patent 6,642,425, Nov 4, 2003. 47. Feng, X.; Buchanan, J.S.; Crane, R.A.; Dakka, J.M.; Iaccino, L.L.; Mohr, G.D. Manufacture of xylenes by reactive distillation methylation of a reformate. PCT Int. 2003, WO2004000768(A1), Dec 31. 48. Rock, K.; Gildert, G.R.; McGuirk, T. Catalytic distillation extends its reach. Chem. Eng. (New York) 1997, 104 (7), 78–84. 49. Silvergerg, S.E.; Lattner, J.R.; Sanchez, L.E. Use of hydrogenative catalytic distillation to produce cyclopentane and=or cyclopentene from dicyclopentadiene. PCT Int. 2000, WO2000029358(A1), May 25. 50. Yu, Z.; Gao, B.; Zhang, J. Selective hydrogenation of C3 fraction in catalytic distillation column. Xiandai Huagong 2001, 21 (7), 23–26. 51. Li, B. Study on C3 catalytic distillation hydrogenation technology. Shiyou Huagong Sheji 2003, 20 (4), 33–35.

Recent Advances in Catalytic Distillation

52. Gildert, G.R.; Loescher, M.E. Catalytic Distillation Process for the Production of C8 Alkanes U.S. Patent 6,274,783, Aug 14, 2001. 53. Louie, W.S.; Mukherjee, U.K.; Hamilton, G.L. Hydrocracking of Vacuum Gas Oils and Other Gas Oils in Cocurrent-Countercurrent Reaction Flow System with Post-Treatment Reactive Distillation U.S. Patent 6,514,403, Feb 4, 2003. 54. Mukherjee, U.K.; Louie, W.S. Hydrocracking of Vacuum Gas and Other Oils Using a PostTreatment Reactive Distillation System U.S. Patent 6,547,956, Apr 15, 2003. 55. Gartside, R.J.; Haines, R.I.; Skourlis, T.; Sumner, C. Olefin Plant Recovery System Employing a Combination of Catalytic Distillation and Fixed Bed Catalytic Steps U.S. Patent 6,759,562, Jul 6, 2004. 56. Rock, K.L.; Foley, R.; Putman, H.M. Improvements in FCC gasoline desulfurization via catalytic distillation. 1998 Annual Meeting, National Petroleum Refiners Association, Mar 15–17, 1998. 57. Hearn, D. Catalytic Distillation Machine U.S. Patent 5,266,546, Nov 30, 1993. 58. Hearn, D.; Putman, H.M. Gasoline Desulfurization Process U.S. Patent 5,597,476, Jan 28, 1997. 59. Gardner, R.; Schwarz, E.A.; Rock, K.L. Start up of first CDHydro=CDHDS unit at Irving Oil’s Saint John, New Brunswick Refinery, NPRA, New Orleans, LA, Mar 18–20, 2001, AM-01-39. 60. Rock, K.L. Ultra low sulfur gasoline via catalytic distillation. In Pre-Print Archive, American Institute of Chemical Engineers (Spring National Meeting), New Orleans, LA, Mar 11–14, 2002; American Institute of Chemical Engineers: New York, 2002; 946–951. 61. Podrebarac, G.G.; Gildert, G.R. Catalytic distillation for desulfurization of full-range naphtha by thioetherification and hydrodesulfurization. PCT Int. Appl. 2002, WO2002066580(A1), Aug 29. 62. Loescher, M.E.; Podrebarac, G.G.; Ho, P.K. Reactive distillation column for sulfidation of petroleum hydrodesulfurization catalysts. PCT Int. Appl. 2002, WO2002062471A2, Aug 15. 63. Vargas-Villamil, F.D.; Marroquin, J.O.; de la Paz, C.; Rodriguez, E. A catalytic distillation process for light gas oil hydrodesulfurization. Chem. Eng. Process. 2004, 43 (10), 1309–1316. 64. Perez-Cisneros, E.S.; Granados-Aguilar, S.A.; Huitzil-Melendez, P.; Viveros-Garcia, T. Design of a reactive distillation process for ultra-low sulfur diesel production. Comput. Aided Chem. Eng. 2002, 10, 301–306. 65. Krafczyk, J.; Gmehling, J. Application of catalyst packings for the manufacture of methyl acetate

Recent Advances in Catalytic Distillation

66.

67.

68.

69.

70. 71.

72.

73.

74.

75.

76.

by reactive rectification. Chem. Ing. Techn. 1994, 66 (10), 1372–1375. Hanika, J.; Kolena, J.; Smejkal, Q. Butylacetate via reactive distillation—modelling and experiment. Chem. Eng. Sci. 1999, 54 (21), 5205–5209. Smith, L.A., Jr.; Hearn, D; Jones, E.M., Jr. Oligomerization Process U.S. Patent 5,003,124, Mar 26, 1991. Nurminen, M.; Pyhalahti, A.; Siira, P.; Tiitta, M. Process for Dimerizing Light Olefins U.S. Patent 2004=0181106 A1, Sep 16, 2004. Ng, F.T.T.; Huang, C.; Podrebarac, G; Nkosi, Rempel; G.L.. Selective dimerization of butenes to octenes via catalytic distillation. Conference Preceedings of the 7th World Congress of Chemical Engineering, Glasgow, Scotland, Jul 10–14, 2005. Loescher, M.E. Recovery of Tertiary Butyl Alcohol U.S. Patent 6,596,913, Jul 22, 2003. Vora, B.V.; Hammershaimb, H.U. Oligomer Production from Isobutene With Catalytic Distillation U.S. Patent 6,025,533 (A), Feb 15, 2000. O’Rear, R.; Dennis, J. Process for Making a Lube Base Stock from a Lower Molecular Weight Feedstock in a Catalytic Distillation Unit U.S. Patent 6,398,946, Jun 4, 2002. Abazajian, A.N. Process for Improved Yields of Higher Molecular Weight-Olefins from LowerOlefins Using Isomerization and Disproportionation in a Reactive Distillation Column U.S. Patent 6,518,469, Feb 11, 2003. Di Gerolamo, M.; Catani, R.; Marchionna, M. Process for the Hydrogenation of Branched Olefins Deriving from the Dimerization of Isobutene U.S. Patent 0078462A1, Apr 24, 2003. Podrebarac, G.G.; Ng, F.T.T.; Rempel, G.L. The production of diacetone alcohol with catalytic distillation. Part I. Catalytic distillation experiments. Chem. Eng. Sci. 1998, 53 (5), 1067–1075. Podrebarac, G.G.; Ng, F.T.T.; Rempel, G.L. The production of diacetone alcohol with catalytic distillation. Part II. A rate-based catalytic distillation model for the reaction zone. Chem. Eng. Sci. 1998, 53 (5), 1077–1088.

2611

77. Huang, C.; Yang, L.; Ng, F.T.T.; Rempel, G.L. Application of catalytic distillation for the aldol condensation of acetone: a rate-based model in simulating the catalytic distillation performance under steady-state operations. Chem. Eng. Sci. 1998, 53 (19), 3489–3499. 78. Huang, C.; Ng, F.T.T.; Rempel, G.L. Application of catalytic distillation for the aldol condensation of acetone: the effect of the mass transfer and kinetic rates on the yield and selectivity. Chem. Eng. Sci. 2000, 55 (23), 5919–5931. 79. Zheng, Y.; Ng, F.T.T.; Rempel, G.L. Catalytic distillation: a three-phase nonequilibrium model for the simulation of the aldol condensation of acetone. Ind. Eng. Chem. Res. 2001, 40 (23), 5342–5349. 80. Lawson, K.H.; Nkosi, B. Production of MIBK Using Catalytic Distillation Technology U.S. Patent 6,008,416, Dec 28, 1999. 81. Saayman, N.; Lund, G.J.; Kindersmans, S. Process for Production of MIBK Using CD Technology U.S. Patent 6,518,462, Feb 11, 2003. 82. Levin, D.; Santiesteban, J.G. Production of phenol using reactive distillation. PCT Int. Appl. 2001, WO02001046102A1, Jun 28. 83. Memphos, S.P.; Groten, W.A.; Adams, J.K. Process for Production of Methanol U.S. Patent 5,886,055, Mar 23, 1999. 84. Allison, J.D.; Wright, H.A.; Harkins, T.H.; Jack, D.S. Use of Catalytic Distillation Reactors for Alcohol Manufacture U.S. Patent 6,723,886, Apr 20, 2004. 85. Powers, D.H. Alpha Olefin Production U.S. Patent 6,768,038, Jun 24, 2004. 86. Adams, J; Groten, W.; Nemphos, S. Process for Vinyl Acetate U.S. Patent 6,620,965, Sep 16, 2003. 87. Voss, B. Acetic Acid Reactive Distillation Based on DME=Methanol Carbonylation U.S. Patent 6,175,039, Jan 16, 2001. 88. Okasinksi, M.J.; Doherty, M.F. Simultaneous kinetic resolution of chiral propylene oxide and propylene glycol in a continuous reactive distillation column. Chem. Eng. Sci. 2003, 58, 1289–1300.

R

Recycling of Spent Tires R Roger N. Beers David A. Benko The Goodyear Tire & Rubber Company, Akron, Ohio, U.S.A.

INTRODUCTION Automobiles are an important part of our modern life providing mobility, status, enjoyment, and numerous other things. The tires from these cars and other vehicles are mostly discarded after their first useful life is consumed, even though a substantial amount of the original tire remains. Because tires are vulcanized under heat and pressure to form an extremely durable thermoset composite of high-molecular weight polymers, organic and inorganic fillers, steel, fabric, and rubber chemicals, it is very difficult to reverse the process and recover the original materials in the same form as when they started. Simply discarding the scrap tires is a waste of resources and causes environmental problems if landfilled or kept in tire stockpiles. Through the 1990s U.S. motor vehicle registrations have risen, greatly increasing the number of tires in use. Advancements in tire durability and treadwear have increased the tire mileage expectations, but large numbers of tires are still being removed from service. A great amount of effort is being exerted by the tire and rubber companies, government bodies, waste management companies, recyclers, and other interested parties to address this scrap tire challenge. New and innovative uses for scrap tires are being developed, and traditional uses are being expanded by improved quality products, improved processing, and changing government and consumer attitudes toward recycled products. This entry shows the major uses for scrap tires now and the amount of scrap tires disposed of by those processes. Reviews of rubber recycling, technical aspects of reprocessing vulcanized rubber, and recovery and reuse of rubber have been published previously and provide more information on this topic.[1–6]

COMPOSITION The number of scrap tires generated annually is about 281 million.[7] Passenger tires account for 84% of the tires scrapped in a year, while light and heavy trucks contribute 15%, and the remaining 1% is generated from aircraft, heavy equipment, and off-road tires. The approximate weight of scrap tires is 5.86 million tons.[7] The average scrap passenger tire weighs 20 lb, Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120007939 Copyright # 2006 by Taylor & Francis. All rights reserved.

light truck tires weigh 30 lb, heavy truck tires weigh 100 lb, with the largest weighing up to 10,000 lb. For a typical spent passenger tire, 12–13 lb of rubber can be recovered. This rubber is composed of about 35% natural rubber and 65% synthetic polymers, while in truck tires the percentages are reversed with about 35% synthetic polymers and 65% natural rubber.[8] Rubber polymers found in tires are polyisoprene (mostly natural and some synthetic), polybutadiene, styrene–butadiene (solution and emulsion), and butyl. The other major components of scrap tires are carbon black, inorganic fillers, polyester or nylon fabric for ply, wire for belts and beads, and a variety of rubber chemicals comprising sulfur, accelerators, antidegradents, oils, and waxes. See Fig. 1 for the amounts of these materials.[9] A tire is a highly engineered and a complex composite that is difficult to physically degrade and nearly impossible to devulcanize to reclaim the original starting materials. Major emphasis has been put on using the cured rubber in some form (whole tires, cut tires, crumb) or to burn the tires to recover the energy value of all the components. ENVIRONMENTAL CONCERNS Scrap tires account for about 1.8% of the solid waste generated annually in the United States.[7] Besides, the tires taken out of service each year, approximately 300 million tires, are stockpiled around the country as of 2001.[7] These stockpiles are breeding grounds for mosquitoes, fire hazards, unsightly, and are liabilities. Fortunately, since the early 1990s this number has been reduced from the 1 billion tires reported stockpiled. This decline in stockpile estimates was based on three factors. Aggressive programs in some states to clean up stockpiles, improvements in estimating stockpile size, and the loss of tires in tire fires. This large number of tires provides a huge challenge for tire companies, environmentalists, government agencies, and businesses to utilize efficiently. HISTORY Rubber reclaiming or recycling is nearly as old as the rubber industry itself. In the early 1900s rubber was 2613

2614

Recycling of Spent Tires

Fig. 1 Composition of average radial passenger tire by weight. (View this art in color at www. dekker.com.)

scarce and very expensive, so it was reused as much as was practical. The tire splitting industry dates back to about 1915.[3] Tires were split and made into shims, gaskets, mats, dock fenders, and other useful items. Before World War II all the rubber used was natural rubber, which was reclaimed with heat and chemicals and reused at levels up to 20% in new rubber compounds. Retreading of passenger tires was used extensively through the 1930s and 1940s. In the 1960s, recycling dropped to about 20%.[3] Cheap oil imports, larger uses of synthetic rubber, and the advent of steel-belted radials along with changing consumer attitudes led to a steady decline in tire recycling. In the 1970s interest in recycling spent tires increased as tire companies and others realized that discarding these high-energy containing materials was economically and environmentally unsound. Whole scrap tires were used as floating breakwaters, tire reefs, and highway crash barriers to take advantage of their energy absorbing properties and resistance to environmental degradation. Emphasis was placed on manufacturing increased quantities of shredded and ground rubber of different sizes and composition for use in engineering applications, such as road bases, erosion control, and asphalt modifiers. Finely ground crumb rubber was also used in tires and automotive applications. This was also the time when efforts began to use the high energy content of tires to generate steam and electricity. Pyrolysis of scrap tires was conducted under different processing conditions to degrade the tires to recover a mixture of gas, oil, and carbon black for sale and reuse. Although much work and investigation of the pyrolysis process was performed, it was not economically viable. Some of these recycling processes are still being used and have been enhanced by new technology, while others have declined significantly or are no longer used. The major uses for scrap tires today will be described in the remainder of this entry.

UTILIZATION OF RESOURCES There are many new and innovative ways to utilize the resources offered by scrap tires. The reuse=recovery rate for scrap tires is about 78% with another 8% being retreaded.[7] There are five methods available to solve the problems associated with scrap tires: Reduce: The most desirable option is to reduce the number of tires that enter the waste stream. Manufacturers have reduced the number of scrap tires by selling more durable extended mileage tires, with average tread life of 40,000–50,000 mi. Compared to the average mileage expectations of 25,000–27,000 in the 1970s, the industry has prevented millions of tires annually from becoming scrap tires. Reuse: Reusing the tire is the next best alternative. Tire casings can be retreaded or made into various miscellaneous products. Aircraft tires can be retreaded up to 12 times and the average truck tire can be retreaded two or three times. About 16.2 million tires were retreaded in 2001 if passenger, light, medium, and heavy trucks are included. Casings can be used for products, such as muffler hangers, snow blower blades, plant holders, and other miscellaneous engineering applications. A large amount of casings are shredded, the wire and the fabric removed, and the rubber ground into various size crumbs. Larger-size tire shreds and tire chips are used in civil engineering uses. The smaller size crumb rubber is used in rubber and plastic articles and in asphalt. Recycle: Today there is a great emphasis on devulcanization methods to recycle the rubber in tires and other rubber products. Some of these new devulcanization methods use supercritical fluid technology, ultrasonic techniques, microwave energy, and biological modification.[10–28] These methods are explained in detail in the entry, which deals primarily with devulcanization. Recover: Much emphasis is being directed toward recovering the energy from the materials remaining

Recycling of Spent Tires

2615

Table 1 2001 Annual utilization of scrap tires Usage for scrap tires

Table 2 2001 North American crumb rubber markets

Millions of tires

Market

Millions of pounds

All reuse

218

Asphalt modifications

292

Energy recovery, TDF

115

Molded products

307

Civil engineering

40

Sports surfacing

141

Ground rubber

33

Tires=automotive

112

Exports

18

Surface modified=reclaim

36

Fabricated products

8

Plastic blends

38

Miscellaneous

7

Animal bedding

37

Construction

28

(From Ref.[7].)

in the spent tires. The use of old tires to provide electricity and energy for cement kilns and papermaking are economically and environmentally sound solutions. Landfill: Whole tires create problems in landfills, but shredded tires do not. They take up less space, do not readily burn, and do not provide habitat for mosquitoes. Landfilling of tires is a poor alternative to the options presented above and is simply disposal with no intent to get any value from the tires. The major uses of scrap tires are divided into categories as shown in Table 1. About 218 million tires were reused in 2001, with the majority going to energy recovery. The next two largest usage categories were for civil engineering applications and ground rubber applications. All these uses will be described in detail.

GROUND OR CRUMB RUBBER Tires are shredded and reduced in size for most reuse and recycle applications. The larger sizes are called tire shreds and tire chips. Tire shreds are basically flat irregularly shaped tire chunks ranging in size from 18 to 1 in. with the majority being between 4 and 8 in.[29]. These are used mainly for construction and civil engineering applications. Tire chips are more finely and uniformly sized, ranging from 3 to 0.5 in. in size. Tire chips are used extensively in civil engineering and tire-derived fuel (TDF). Ground rubber and crumb rubber are used interchangeably, but actually ground rubber particles are intermediate in size between tire chips and crumb rubber, ranging from 3=8 in. to 20 mesh. Crumb rubber particles range from 4 to about 200 mesh, with 20 to 80 mesh particles used the most.[30] Ground and crumb rubber are free of fabric, wire, and other contaminants. The smaller the particle size, the more expensive the rubber. In 2001, the average price of 10 mesh crumb rubber was about 13 cents=lb, and 80 mesh crumb rubber averaged about 31 cents=lb.[31] There are three main methods for producing crumb rubber: dry ambient grinding, wet grinding, and

Other

5

Total

996

(From Ref.[33].)

cryogenic grinding.[4] Dry ambient grinding uses serrated grinders to grind various types of scrap and spent rubber articles. Normally, any wire or steel has been separated previously. The size of the rubber particles from this process usually ranges between 10 and 30 mesh. Even though this process is considered an ambient process, the rubber particles are exposed to considerable heat that is generated in the grinding process. Ambient grinding produces irregular-shaped particles with small appendages caused by the softening and tearing during grinding. Wet or solution grinding uses water and starts with 10–20 mesh rubber. These particles are ground between two wheels with the water enhancing the lubricity and cooling. Particles as fine as 200 mesh are obtained and size is related to the grinding time. This process gives fairly smooth particles. Scrap rubber can also be reduced in size by cryogenic grinding. Small pieces of rubber are placed in liquid nitrogen and ground into fine powder generally ranging from 30 to 100 mesh. Surface morphology is different from ambient grinding because the particles are fractured and surface oxidation is believed to be less.[32] Table 2 shows the amount of ground rubber used in North America in 2001.[33] This amount includes the market for the United States and Canada. The ground-rubber market consists of ground scrap

Table 3 Tire crumb use in the United States (millions of pounds) Market

1995

2000

Asphalt modification

97

261

Rubber products

74

250

Surfaces

35

113

Automotive

52

100

11

143

269

867

Miscellaneous Total (From Ref.[1].)

R

2616

tires and tire buffings. Table 3 shows the amounts of crumb rubber used in the United States only for 1995 and 2000 and shows the growth taking place in this market.[1] Although estimates vary slightly, at least 33 million scrap tires were used in this market in 2001 and the rest was obtained from tire buffings. No attempt is made to differentiate between tire buffings and scrap tires in identifying markets.

Recycling of Spent Tires

demonstrated that it is quieter, smoother, reducing overall cost, and a good method of using scrap tires. California, Arizona, Texas, and Florida promote the use of ground rubber in asphalt. Several other states are evaluating the use of RMA and may utilize this method. Growth in this area for crumb rubber depends on a number of factors: DOT support, additional performance and cost data, and a readily available supply of crumb rubber.

Rubber Modified Asphalt Molded Products Ground tire rubber is used as an additive in various types of asphalt pavement constructions. There are two main processes in use: the dry and the wet process, and modifications of both exist. In the dry process, ground rubber ranging in size from 0.25 in. to 20 mesh can be used as an aggregate substitute at a loading of 3–4 wt%.[34] The wet process uses 30–100 mesh crumb rubber, which is added to modify the physical and the chemical properties of the binder. Levels of crumb rubber in the wet process can vary, but as the level rises above 5–10%, the cost increases significantly compared to conventional pavement.[35] A recent article shows that 80 mesh crumb rubber is very good for use in both wet and dry processes.[36] The small particle size reacts faster with the asphalt to form a homogenous mix and is compatible in all types of asphalt equipment. A study conducted in Canada used three types of crumb rubber at 10% by weight with two commonly used asphalt binders.[37] The rubber-modified asphalts (RMAs) prepared were evaluated for resistance to forming ruts and low-temperature cracking. Compared to conventional asphalt, the crumb rubber modifications significantly improved the high-temperature performance and moderately improved the lowtemperature cracking. Both low- and high-temperature responses revealed improvement as the particle size of the rubber was reduced from 30 to 80 mesh and lower. The wet process modified binder is commonly referred to as asphalt rubber and is also used in seal coats, stress absorbing membranes, waterproofing membranes, and hot mix binders. The level of use of ground rubber varies according to the method, but in 2001 about 290 million pounds was used for asphalt construction.[33] The rubber improves asphalt ductility and increases the temperature at which asphalt softens, thus improving durability and fatigue resistance. The rubber also improves the skid resistance of the asphalt pavement. A review paper by Van Kirk indicated that there is sufficient field experience to show the advantages of crumb rubber in RMA.[38] The main advantages are that pavement thickness can be reduced with no loss in service life, pavement cracking is decreased, and maintenance cost is reduced. The use of RMA has

Molded products are another large market for ground and crumb rubber from scrap tires. The size-reduced rubber can be molded, extruded, or processed in various ways with other thermosets or virgin rubber. Automotive products, tires, and plastic blends are not included in this category and will be discussed as separate categories. Crumb rubber extends and modifies the molded products, in particular lowering cost, improving impact resistance, lowering shrinkage, and improving processing. This market should experience relative growth because of improved product quality, more manufacturer and consumer interest, campaigns to buy recycled materials, and state tire program incentives. In 2001, about 300 million pounds was used in this area.[33] Some typical products are floor tile, mats, water hoses, sound proofing panels, bin liners, and belt covers.

Tires and Automotive Applications Great effort is being expended to find ways to use increasing amounts of scrap tires in the manufacture of new tires. Automotive companies are strongly encouraging tire producers to increase the recycle content in the new tires they purchase. Ford is currently trying to have the tire recycle content raised to 10% by weight with the ultimate goal being 25% recycle. Many new technologies are being investigated to enable higher use of tire-derived rubber without compromising the stringent physical properties needed in a radial tire. Currently, about 10% is the maximum amount of reprocessed rubber used in certain compounds, and the overall tire content is around 3%.[7] Recycle is used in passenger, farm, and light and heavy truck tires. Tire components that commonly contain reprocessed rubber are the innerliner, treads, sidewall, ply, and barrier compounds. Currently, only about 25% of the total reprocessed rubber used in tires comes from scrap tires. The remainder comes from curing bladders, inner tubes, and tread peels. The curing bladders and inner tubes are used almost exclusively in the tire liners because they are butyl and provide excellent

Recycling of Spent Tires

2617

Table 4 Tread compound properties containing crumb rubber Tensile

b

Control (0 crumb)

20 phra crumb

20 phr crumb

40 phr crumb

100

85

80

75

300% mod

100

91

82

77

100C rebound

100

94

93

91

Mooney scorch

100

93

81

79

Cure amount

100

92

79

78

Abrasion

100

90

83

68

Heat buildup

100

89

86

78

Viscosity

100

73

73

56

a

phr is parts by weight per hundred parts of rubber in compound. b Lower figures indicate worse performance. (From Ref.[39].)

air retention properties. Also, butyl is incompatible with other tire elastomers. Typical particle sizes used range from 20 to 80 mesh. The 80 mesh particles provide smoother compounds and possibly slightly improved properties. The effect of ground rubber recycle on most rubber compound properties is negative, which prevents higher usage. Ground rubber is used as an extender or filler to reduce costs and help processing. Properties that are impacted negatively are compound viscosity, tensile strength, modulus, elongation, scorch, abrasion resistance, cure, and hysteresis.[39] Typical effects of loading on key compound properties are shown in Table 4. The table shows how compound properties change as the amount of 40 mesh ground scrap tire rubber is added to a typical passenger tire tread.[39] The control contains no crumb rubber and 20, 30, and 40 parts of crumb rubber are added to the other tread formulations. The data are normalized with the control given a value of 100, and the other compounds compared to that value. Processing improvements generally noted are improved mold flow, better air bleed, and reduced shrinkage. There are several reasons for the reduced properties. The hard ground rubber particles contain accelerator fragments from being cured previously and these fragments migrate into the virgin rubber and cause it to cure faster with less scorch protection.[40] Sulfur from the virgin rubber compound also migrates into the

cured particles making them harder and causing the matrix to be cured to a lower amount, reducing properties. Another reason attributed to the reduced strength is poor interfacial bonding between the crumb rubber and the virgin rubber matrix.[41] Various surface treatments, both chemical and physical, additives, new processing methods, and compound modifications are being developed to improve the physical properties.[42–45] Table 5 shows the results of treated and untreated crumb when added at 20% to a natural rubber compound. In this case, the crumb is also natural rubber and the treated portion was surface activated with a two-step process developed by Vredestein rubber recycling.[44] The treated crumb compound was significantly better than the untreated crumb compound. Automotive uses besides tires include belts, hoses, friction materials, mud flaps, air deflectors, brake pedal covers, caulks, and coatings. Like tires, the industry would like to see more recycled content in these applications. A total of about 112 million pounds of ground rubber is consumed in these applications.[33]

Sports Surfacing This market segment has undergone a huge growth rate in the past several years. Examples are the use of

Table 5 Comparison of treated and untreated crumb rubber in NR compound Control (no crumb)

20% treated crumb

20% untreated crumb

93.5

83.2

63.8

Abrasion (mm lost)

85

80.8

99.5

Compression set (%)

39

43.7

52.3

Tear strength (N) 3

Rebound (%)

51

49

24.4

Tensile (MPa)

24

20.4

17

(From Ref.[44].)

R

2618

rubber in running track material, in grass surfaced playing areas, in stadium playing surfaces, and as turf topdressing. The main reasons for using ground or crumb rubber for these type surfaces is to reduce injuries, improve drainage, reduce damage to fields, and improve turf viability. Crumb rubber also retards weed growth, does not attract insects or rodents, and has a higher cushioning effect than other fill materials. The Detroit Lions football team’s practice field contains a rubber soil material that has allowed practice in all weather conditions and has reduced player injuries. Another use is in horse arenas, where it gives the animals better footing and reduces injuries to the horse and to the rider. One company produces a granular product 3=8
1=8 in. in size for the playground market.[46] The material is put down in a layer 6 in. thick and provides greater shock absorbing characteristics than other playground surfaces, such as sand or mulch. Approximately 140 million pounds of ground rubber was used for some type of sport surfacing in 2001.[33] The increased use was mostly because of the improved quality of the surfacing, better application techniques, and the safety benefits. These type materials are finding use in golf courses and public parks.

Plastic Blends Thermoplastic materials are manufactured using many processes, such as injection molding, extrusion, hot stamping, and others, that use heat and pressure to form a particular product or shape. In most cases, the thermoplastic is pelletized and crumb rubber can be readily blended in the composition during processing. The crumb rubber is used to improve impact resistance, improve processing, extend the material, or provide some performance improvement to the finished product. In many cases, the crumb rubber surfaces are treated with silanes, chemicals, such as ethylene–glycidyl methacrylate, compatibilizing agents, high-energy radiation like plasma etching, or other novel techniques to increase the adhesion with the thermoplastic matrix and improve compatibility.[47–50] Plastic blends with rubber are used for a great number of products, such as body panels for golf carts and lawn tractors, sports equipment, toys, cable and wire insulation, and truck bed liners. Crumb rubber is even used to make thermoplastic elastomers (TPE) by alloying 120 mesh rubber with an olefinic matrix.[51,52] Before alloying, the rubber powder undergoes a proprietary surface devulcanization process and compatibilization technique to enhance the blending of the rubber with a polyethylene=polypropylene mix to form the TPE. These applications accounted for about 38 million pounds in 2001 and there appears to be significant market potential for this application because of

Recycling of Spent Tires

increased research and development on surface treated rubber.[33]

Construction Ground rubber finds many uses in the construction industry because of its excellent weathering, chemical resistance, and nonbiodegradable properties. Some uses are paving or tiles for driveways, shingles and roofing materials, thermal and acoustic insulation panels, plastic lumber, flooring materials, and other building materials. There has been substantial growth in the manufacture of floor tiles and sidewalk pads, where crumb rubber is used with urethane binders for cost and injury reduction. About 28 million pounds of ground rubber was used for these applications in 2001.

Surface Modification/Reclaim This category involves processing used rubber products into a form that can be incorporated into virgin rubber compounds. Ground rubber is the starting material for the chemical process that severs the cross-links and polymer chains by chemical or steam digestion to make reclaim. Other reprocessed rubber uses crumb rubber and modifies the surface by chemical or physical means to enhance the adhesion and compatibility to improve properties when mixed with virgin rubber compounds. About 36 million pounds of ground rubber is used for these applications.[33] The major benefit of using reclaim is lower cost and the improvements in processing it can add to the compounds. Compounds containing reclaim have lower nerve, which translates into better extrusion, increased calendar speed, and improved flow and mold filling. Properties like green strength and tensile are lowered when reclaim is used. Today some whole tire reclaim is used in low performance requirements like mats and extruded rubber products. The major rubber reclaimed is butyl that comes from truck inner tubes. Crumb rubber can be ground into small particles from 30 to 200 mesh and used in virgin rubber compounds primarily as a filler. Smaller particle sizes give optimum surface smoothness and slightly higher physicals. Mesh sizes from 30 to 80 can lower compound cost, improve mold flow, help air bleed, and reduce shrinkage. However, crumb rubber generally increases compound viscosity and reduces strength and modulus. In low performance requirements, levels from 5% to 10% are commonly used. To increase properties, the crumb rubber is treated to enable the virgin rubber to bond better. New surface treatments are gaining, particularly in the tire market.

Recycling of Spent Tires

Animal Bedding Ground rubber is used in agriculture for animal bedding. Ground rubber is used because it does not hold moisture, is relatively cheap, is not readily biodegradable, and does not harbor insects and pests. Mats are also made with crumb rubber for use by animals. Over 35 million pounds of ground rubber is used for this application.[33] These are the main uses for ground and crumb rubber. In addition, another half-million tires or less are used for miscellaneous applications as diverse as can be imagined. The exact numbers of tires used for each application are as up-to-date as possible and as accurate as the data available. Overlaps in category use and some estimates in the number of actual tire units consumed may be slightly different from other published figures. The source of most of the information is the Scrap Tire Management Council that was formed to monitor and promote disposal and reuse of scrap tires.

CIVIL ENGINEERING Civil engineering includes a wide range of uses for scrap tires. The five main applications include lightweight fill for embankments, subgrade insulation for roads, backfill for walls and bridge abutments, landfill construction and operation, and septic system drain fields. In nearly all the applications, the tire chips or tire shreds typically replace currently used construction materials, such as soil, clean fill, drainage aggregates and lightweight fill materials. The benefits of using the tire chips and shreds are lower density, improved drainage properties, better thermal insulation, and lower cost. Tire stockpiles are good for these applications, as the presence of dirt on the tires is not a problem. The civil engineering area is growing and about 40 million tires were used in 2001.[7] Lightweight Fill for Embankments Work has been done in a number of states using tire shreds as a subgrade fill in the construction of highway embankments and other fill projects. Tire shreds weighs considerably less than conventional soil fill and this allows construction on weak compression foundation. The use of tire shreds for most projects is considerably cheaper than alternatives, such as expanded shale aggregate and polystyrene insulation blocks.[29,53] Scrap tire materials have also been used to retain forest roads, protect coastal roads from erosion, enhance the stability of steep slopes along highways, and reinforce shoulder areas. The importance

2619

of this application is illustrated by a project in Maine, which used 1.2 million tires as lightweight fill for the construction of two highway embankments on weak clay and saved about $300,000.[7] Subgrade Insulation for Roads In northern climates, one of the problems affecting roads is the water released when the subgrade soils thaw in the spring. To prevent this, tire shreds are used as subgrade insulation. The insulation provided by a 6–12 in. layer of tire shreds keeps the subgrade soils thawed during the winter. Tire shreds also allow excess water to drain from beneath the road.[54] Backfill for Walls and Bridge Abutments Several projects have been constructed using tire shreds as backfill for walls and bridge abutments. The weight of the tire shreds produces low horizontal pressure on the wall allowing thinner walls at lower cost. Another advantage is that tire shreds provide better drainage and good thermal insulation, which eliminates problems with water or frost buildup behind the wall.[55] Landfill Construction and Operation Many landfill operators have found that scrap tires can be used beneficially in the construction and operation of the landfill. Tire shreds are used in building leachate collection basins in new landfills and they can also be used as the drainage layer in landfill caps. Another application is mixing tire shreds with soil as daily cover material. Septic System Drain Fields Tire shreds and chips are used as a replacement for stone in septic system leach fields. The lower density of the tire shreds greatly reduces the expense and the labor to construct leach fields, while providing equal performance to stone and gravel. The number of states allowing scrap tires in septic systems is increasing. As in all the civil engineering applications, there is concern about the environmental impact of using scrap tires. One of the major issues is groundwater contaminants and the effect of tire shreds on water quality. Studies have shown that tire shreds placed above or below the water table pose no significant health or environmental risk.[7,56] Neutral pH groundwater does not increase the concentration of metals and no detectable organic leachates are found from the tire shreds. Under some conditions, the steel belts contained in the tire

R

2620

Recycling of Spent Tires

shreds may increase the levels of iron and manganese, but these are not harmful. These concerns affect the laws related to scrap tire use, and along with other issues like cost and availability, determine the use and growth. Each potential civil engineering use brings with it a particular set of technical, environmental, and economic constraints that must be fully evaluated before the application is readily acceptable. The potential use of scrap tires in civil engineering is very substantial and this will continue to grow.

ENERGY MARKET Tires have a tremendous market potential as fuel. Energy recovery was the largest market in 2001, representing about 115 million scrap tires annually.[7] On average, 97% of a tire’s content contributes directly to energy value. The average radial passenger tire derives its energy content from the following ingredients:  Synthetic rubber is produced from crude oil, a high-energy fuel much cleaner than coal.  Natural rubber is a renewable energy source that is harvested from trees.  Carbon black also comes from crude oil and has a high energy value.  Rubber chemicals, oils, and organic fibers are all derived from crude oil and contribute to the energy value.  Steel belts and beads oxidize at high temperatures to produce 3500 Btu=lb. Energy from scrap tires is being recovered in the United States and many other countries worldwide with beneficial results. Burning one passenger tire produces approximately 300,000 Btu, which is roughly equivalent to 2 gal of fuel oil or 23 lb of coal. Whole scrap tires average 15,000 Btu=lb and most coal averages 12,000–12,500 Btu=lb (Table 6).[57] Tires contain less nitrogen than coal, and there is less sulfur in tires than most types of coal. Tires burned under proper conditions reduce solid waste and air emissions, and produce little or no smoke or fumes.

Table 6 Comparative energy values Fuel

Btu/lb

Petroleum coke

13,700

Bituminous coal

12,750

Subbituminous coal

10,500

Lignite coal TDF (From Ref.[57].)

7,300 14,000–15,500

Cement, electricity, and paper producers burn scrap tires as a supplemental fuel, because tires have more energy value per pound than most coal and burn more evenly. Scrap tire combustion decreases the emission of particulates, oxides of nitrogen, and carbon dioxide compared to coal. Whole tires can be burned in some cases or TDF is created by shredding scrap tires and processing them into tire chips. Typically, these tire chips range in size between 2 and 4 in., but many facilities are shifting to smaller fuel chips 2 in. by 2 in. or slightly smaller. These smaller chips contain less steel and can reduce handling problems and lower ash disposal. Boiler design and the production process determine the allowable wire content level. Tire-derived fuel is easy to handle, store, and feed into combustion devices. It is also easily blended with conventional fuels. Using scrap tires as fuel totally eliminates the scrap tire, while it decreases the dependence on foreign oil supplies and conserves precious natural resources. The total use of scrap tires for energy recovery has declined over the past several years from 1996 when approximately 135 million scrap tires were consumed.[7] The use for energy recovery seems to have stabilized with the potential for slow growth over the next several years. It must be remembered that the energy market is very complex and many other factors affect the use of scrap tires for energy recovery. Factors that impact this area are air pollution laws, states subsidies for use of scrap tires, energy deregulation, economic conditions, product quality, and costs. All of these issues plus others can affect the economics and help or hinder the use of scrap tires in these energy applications.

Cement Industry There are 36 cement kiln locations that processed tires for fuel in 2001 with the majority burning whole tires and the rest tire chips.[7] The original reason cement kilns began using scrap tires was to reduce fuel costs. A big advantage of using scrap tires in cement kilns is that there is no solid waste disposal. The tires are completely consumed and become a part of the final product. Other advantages are less carbon dioxide, nitrogen oxides, and sulfur dioxide. The wire of the tire provides iron oxide, which is one of the raw materials required. Also, limestone, which is a raw material for the cement, neutralizes sulfur from the tires. The amount of scrap tires used as a percentage of the total fuel consumed can range from 5% to 25%. The cement industry is the largest user of scrap tires for energy and will continue to consume significant amounts. Two important issues that will affect the amount of scrap tires used are USEPA focus on nitrogen oxides and the demand for cement.

Recycling of Spent Tires

Paper and Pulp Mills There are 18 facilities that use scrap tires for pulp and paper production.[7] Tire-derived fuel is mixed with wood waste to produce steam. The TDF improves the combustion efficiency and displaces fossil fuels. A significant portion of Oregon’s scrap tires goes to this market. The volume of scrap tires used for this application has fallen from its high in 1996 much like cement industry use. It should be noted that the decrease in TDF consumption was never attributed to increased air pollution or environmental degradation. Utility Boilers There are 11 plants using scrap tires to produce electricity for sale and they consume the second largest amount after the cement industry.[7] Two factors that have impacted the use of TDF and will continue to impact its use are the electric industry deregulation and the implementation of Clean Air Act amendments. Until all the issues are sorted out, further expansion in this area is unlikely. Industrial Boilers Industrial boilers use TDF to generate steam. There are 17 facilities that use scrap tires, although it is usually in a limited degree.[7] In this market segment, the use of TDF is primarily a function of the solid wastes available and represents about 2–5% of the fuel supply. Still, tires are used because the economics and the emissions are favorable. A small increase in the use of TDF should occur in the industrial boiler market.

EXPORT OF TIRES Tires with adequate tread and=or retreadable tires are regularly exported from the United States. Used tires are regularly sold in these markets in many other parts of the world. Routinely, slightly more than 1 million tires=mo are exported or about 15 million tires=yr by best estimate.[7]

FABRICATED PRODUCTS The fabricated products market encompasses those products made by cutting, punching, or stamping from scrap tire carcasses. This is one of the oldest methods of reuse and is not changing. This market has a large array of products, all of which take advantage of the toughness and durability of tire carcass materials. Examples of these products are dock bumpers, muffler

2621

hangers, mat components, and snowblower blades. These entire applications generally only use bias ply tires, or tires with no steel belts. This market will remain the same and uses about 8 million scrap tires.[7]

MISCELLANEOUS USES There are a wide variety of uses for scrap tires that do not fit neatly into any of the preceding categories. The total in this catchall category is about 7 million scrap tires. These include agricultural uses where tires are used to construct stock feeders, protect fence posts, weigh down covers, erosion control, and other uses. Usage can also include swings, planters, or other more imaginative and innovative uses.

CONCLUSIONS Among all the methods available to manage the increasing number of scrap tires, reuse, recycle, and recovery offer the most potential to arrive at a program that consumes all the scrap tires generated each year and steadily decreases the existing tire piles. Reuse includes the civil engineering and ground rubber markets that are steadily increasing the volumes of tires they consume in highway construction projects, asphalt modifications, sports surfaces, and molded products. New technology and innovations are driving the increased use of scrap tires, but there are other factors that impact use of scrap tires. State regulations on landfills, incentive programs to use tires, establishment of an infrastructure to collect, market and distribute scrap tires, and consumer attitudes also affect the number consumed. Reclaim or recovery of the polymers and the carbon black in the tires has great potential for providing new sources of materials to be used again to manufacture tires or other rubber products. Scientists are working on new ways to devulcanize rubber and success has been reported by using ultrasonic, microwave, supercritical fluids, and mechanical means. True devulcanization would allow much more of the recaptured polymers to be used in tires and=or other demanding applications. Recovery of energy is still the largest market for scrap tires today, even though this important market that has shown some decline in volume since 1996. New boiler technology and processing enhancements to burn scrap tires are emerging and should solve any technical problems related to this application. Again, air pollution laws, energy deregulation, and economic issues are just as important as new technology in expanding the energy recovery market. All these forces will continue to challenge the industry for ways to consume spent tires. Currently,

R

2622

about 78% of the tires removed from service each year are being reused and that number should continue to grow as new technology and greater acceptance of recycled products occurs.

REFERENCES 1. Myhre, M.; MacKillop, D.A. Rubber recycling. Rubber Chem. Technol. 2002, 75, 429–474. 2. Crane, G.; Elefritz, R.A.; Kay, E.L.; Laman, J.R. Scrap tire disposal procedures. Rubber Chem. Technol. 1978, 51, 577. 3. Beckman, J.A.; Crane, J.; Kay, E.L.; Laman, J.R. Scrap tire disposal. Rubber Chem. Technol. 1974, 47, 597–623. 4. Klingensmith, W. Recycling, production, and use of reprocessed rubbers. Rubber World 1991, 203 (6), 16. 5. Warner, W.C. Methods of devulcanization. Rubber Chem. Technol. 1994, 67, 559–566. 6. Adhikari, B.; De, D.; Maiti, S. Reclamation and recycling of waste rubber. Prog. Polym. Sci. 2000, 25, 909–948. 7. Rubber Manufacturers Association; http://www. rma.org/scraptires/facts_figures.html (accessed July 2004). 8. Schnormeier, R. Recycled Tire Rubber in Asphalt; 71st Annual Meeting Transportation Research Board: Washington, DC, 1992. 9. Goodyear Tire & Rubber Company, Internal Document, 1999. 10. Beers, R.N.; Benko, D.A. U.S. Patent 6,387,965, May 14, 2002. 11. Isayev, A.I.; Chen, J.; Tukachinsky, A. Novel ultrasonic technology for devulcanization of waste rubbers. Rubber Chem. Technol. 1995, 68, 267. 12. Isayev, A.I.; Kim, S.H.; Yushanov, P. Ultrasonic devulcanization of SBR rubber: experimentation and modeling based on cavitation and percolation theories. Rubber Chem. Technol. 1998, 71, 168. 13. Isayev, A.I.; Yashin, V.V. A model for rubber degradation under ultrasonic treatment: part I. Acoustic cavitation in viscoelastic solid. Rubber Chem. Technol. 1999, 72, 741. 14. Boron, T.; Klingensmith, W.; Roberson, P. Ultrasonic devulcanization of tire compounds. Tire Technol. Int. 1996, 82–84. 15. Isayev, A.I.; Kim, S.H.; Levin, V.Y. Superior mechanical properties of reclaimed SBR with bimodal network. Rubber Chem. Technol. 1997, 70, 194. 16. Pelofsky, A.H. U.S. Patent 3,725,314, 1973. 17. Mangaraj, D.; Senapati, N. U.S. Patent 4,548,771, 1985. 18. Isayev, A.I. U.S. Patent 5,258,413, 1993.

Recycling of Spent Tires

19. Isayev, A.I.; Chen, J. U.S. Patent 5,284,625, 1994. 20. Isayev, A.I.; Schworm, D.; Tukachinsky, A. Devulcanization of waste tire rubber by powerful ultrasound. Rubber Chem. Technol. 1996, 69, 92. 21. Isayev, A.I.; Chen, J.; Yushanov, S.P. Ultrasonic devulcanization of rubber vulcanizates. II. Simulation and experiment. J. Appl. Polym. Sci. 1997, 59, 815. 22. Makrov, V.M.; Drozdovski, V.F. Reprocessing of Tires and Rubber Wastes; Ellis Horwood: New York, 1991. 23. Novotny, D.S.; Marsh, R.L.; Masters, F.C.; Tally, D.N. U.S. Patent 4,104,205, 1978. 24. Clifford, M.L. U.S. Patent 4,130,616, 1978. 25. Tyler, K.A.; Cerny, G.L. U.S. Patent 4,459,450, 1984. 26. Hunt, J.R.; Hall, D. U.S. Patent 5,362,759, 1994. 27. Wicks, G.G.; Schulz, R.L.; Clark, D.E.; Folz, D.C. U.S. Patent 6,420,457, 2002. 28. Romine, R.A.; Snowden-Swan, L.J. U.S. Patent 5,597,851, Jan 28, 1997. 29. Read, J.; Dodson, T.; Thomas, J. Experimental Project—Use of Shredded Tires for Lightweight Fill; Oregon Department of Transportation: Salem, OR, 1991. 30. Heitzman, M. Design and Construction of Asphalt Paving Materials with Crumb Rubber; Transportation Research Board: Washington, DC, 1992. 31. Price and Market Surveys; Recycling Research Institute. 32. Kohler, R.; O’Neill, J. Rubber Division Meeting, ACS, Cleveland, OH, Oct 17–20, 1995. 33. Recycle Research Institute, 2002 Scrap Tire & Rubber Users Directory, 2002. 34. Morrison, G.R.; Hesp, S.A.M. New look at rubber-modified asphalt binders. J. Mater. Sci. 1995, 30, 2584. 35. Tarricone, P. Recycled roads. Civil Eng. 1993, 63 (4), 46. 36. Rouse, M.W. Application of CRM in asphaltic materials. Rubber World 1995, 212 (2), 23. 37. Coomarasamy, A.; Hesp, S.A.M. Performance of scrap tire rubber modified asphalt paving mixes. Rubber World 1998, 218 (2), 26–32. 38. Van, K.J. Rubber Division Meeting, ACS, Orlando, FL, Sep 21–24, 1999. 39. Goodyear Tire & Rubber Company, Internal Technical Evaluation, 2001. 40. Hamed, G.R.; Gibala, D.; Zhao, J. Tensile behavior of an sbr vulcanizate containing a single rubber particle. Rubber Chem. Technol. 1998, 71, 861. 41. Myhre, M.; MacKillop, D.A. Modification of crumb rubber to enhance physical properties of recycled rubber products. Rubber World 1996, 214 (2), 42–46.

Recycling of Spent Tires

42. Peter, J.; Schmidt, P.; Mahlke, D. U.S. Patent 5,844,043, 1998. 43. Kim, J.K.; Burford, R.P. Study on powder utilization of waste tires as a filler in rubber compounding. Rubber Chem. Technol. 1998, 71, 1028. 44. Dierkes, W. Rubber World 1996, 214 (2), 25. 45. Fuhrmann, I. Karger-Kocsis. Promising approach to functionalisation of ground tyre rubberphotochemically induced grafting: Short Communication. J. Plast. Rubber Compos. 1999, 28, 500. 46. http:==www.gtrcrumbrubber.com=products.html (accessed Oct 2002). 47. Yasuda, H.; Marsh, B.; Brandt, S.; Reilly, C.N. ESCA study of polymer surfaces treated by plasma. J. Polym. Sci. Chem. Ed. 1977, 15, 991. 48. Evans, J.M. Nitrogen corona activation of polyethylene. J. Adhesion 1973, 5, 1. 49. Baker, W.E.; Rajalingham, P. The role of functional polymers in ground rubber tire-polyethylene composite. Rubber Chem. Technol. 1992, 65, 908. 50. Xu, Z.; Losur, N.S.; Gardner, S.D. Epoxy resin filled with tire rubber particles modified by plasma surface treatment. J. Adv. Mater. 1998, 30, 11.

2623

51. Jury, J.R.; Chien, A. U.S. Patent 6,262,175, Jul 17, 2001. 52. Burgoyne, J.; Fisher, J.; Jury, J. U.S.Patent 5,510,419, Apr 23, 1996. 53. Bosscher, P.J.; Edil, T.B.; Eldin, N.N. Construction and Performance of a Shredded Waste Tire Test Embankment, 71st Annual Meeting; Transportation Research Board: Washington, DC, Jan, 1992. 54. Humprey, D.N.; Eaton, R.A. Tire Chips as Subgrade Insulation—Field Trial, Proceedings of the Symposium on Recovery and Effective Reuse of Discarded Materials and By-Products for Construction of Highway Facilities, Federal Highway Administration, Denver, CO, 1993. 55. Humprey, D.N.; Sandford, T.C.; Cribbs, M.M.; Manion, W.P. Shear Strength and Compressibility of Tire Chips for Use as Retaining Wall Backfill, 72nd Annual Meeting; Transportation Research Board: Washington, DC, Jan 1993. 56. Sengupta, S; Miller, J. Investigation of Tire Shreds for Use in Residential Subsurface Leaching Field Systems: A Field Scale Study; Department of Civil Engineering, UMass Dartmouth: Dartmouth, MA, 2000. 57. Rouse Rubber Industries, TDF Fact Sheet.

R

Reformulated Gasoline R A. K. Dalai D. Ferdous Catalysis and Chemical Reaction Engineering Laboratories, Department of Chemical Engineering, University of Saskatchewan, Saskatoon, Canada

INTRODUCTION The purpose of introducing reformulated gasoline (RFG) is to improve the air quality to reduce motor vehicle emissions of toxic and tropospheric ozoneforming compounds. The RFG regulations involve the greatest reduction in emissions of ozone-forming volatile organic compounds (VOCs) and toxic air pollutants (TAP) through the reformulation of conventional gasoline. They also take into consideration the cost of achieving such reduction.

BACKGROUND Prior to 1995, around 85 million people (one-third U.S. population) lived in urban areas. The air quality of these areas violated federal public health standards, largely because of automotive pollutants. Gasoline and diesel-fueled cars, trucks, and buses produced half of all air pollution in the United States. This air pollution includes 66% of airborne carbon monoxide, 31% of smog-forming hydrocarbons, and 43% of lungdamaging nitrogen oxide. On January 1, 1995, nearly one-third of the motorists in the United States found something different at the gas pump, RFG. This gasoline is designed to reduce the environmental impact of burning the fuel so that air quality meets public health standards of the national Clean Air Act amendments of 1990. The CAA was designed to address the issues related to fuel quality improving gasoline. This required that the RFG to be used in cities with the worst smog pollution should contain oxygen to reduce harmful emissions of ozone. ARCO first introduced RFG in 1989 in Southern California. In the RFG, benzene and aromatic levels were reduced, vapor pressure (or volatility) decreased, and the oxygenates were added to improve combustion and other fuel properties. Reformulated and oxygenated fuels have been already used at different regions of the United States where the ozone level limit has been exceeded. As indicated earlier, one way to reduce air pollution from cars and trucks is to use an RFG that is designed to burn cleaner, while the CAA requires cities with the worst Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120007943 Copyright # 2006 by Taylor & Francis. All rights reserved.

smog pollution to use RFG, but other cities with smog problems may choose to use RFG. The RFG program is a significant step toward cleaning the air we breathe, and a significant component of the country’s smog reduction strategy. The RFG is currently used in 17 states and the District of Columbia in the United States. Nowadays, about 30% of gasoline sold in the United States is reformulated. Each oil company prepares its own formula to meet federal emission reduction standards. The RFG’s air quality benefits and other industrial and transportation controls aimed at smog reduction are responsible for the long-term downward trend in U.S. smog. In North America and other countries, increasingly stringent fuel specifications for RFG are mandated.

REFORMULATED GASOLINE AND ITS GOAL Reformulated gasoline, referred to as ‘‘clean gasoline’’ is designed to reduce both exhaust and evaporative emissions from vehicles. This fuel offers a way to reduce a variety of gasoline emissions without developing alternative fuel technologies to replace oil derived fuels and internal combustion engine. The definition of RFG includes the following key factors:[1]
A 15% reduction in toxic chemicals, including VOCs.
A maximum of 1% of the carcinogen benzene (a 50% reduction) and a maximum of 25% for aromatic hydrocarbons (compared with 40% previously).
A minimum content of 2% oxygen, by weight, to promote cleaner combustion, especially to reduce carbon monoxide emission. The sunlight driven complex reactions involving VOCs and oxides of nitrogen (NOx) are responsible for atmospheric ozone formation. The purpose of using reformulated fuels is to reduce both ozone-forming VOC emissions and air toxic emissions from vehicles. The key constraints that are introduced by Phase I in 1995 and II in 2000 are reduction in VOCs, NOx, and toxics. While each of theses variables has complex 2625

2626

interactions with other gasoline properties, each is affected by a dominating component.[2] Volatile organic compounds, NOx, and toxics are the functions of Reid vapor pressures (RVP), nitrogen content, and benzene content, respectively. Beginning in January 1995, Phase I RFG was intended to provide a 15–17% reduction in both ozone-forming VOC emissions and air toxic emissions from vehicles. In the beginning of January 2000, Phase II RFG intended to provide a 25–29% reduction in VOC emissions, 20–22% reduction in air toxics emissions, and 5–7% reduction in nitrogen oxide emissions from vehicles.[3] This phase-wise introduction of RFG indicated that fuel parameters reduced a vehicle’s pollutant emission. Phase I of California’s RFG program began in 1992 and required reduced gasoline vapor pressure during the summer ozone season, use of detergent additives to control engine deposits, and elimination of lead-based antiknock additives.[4] The current EPA summer vapor pressure maximum for most areas of the United States is 9.0 psi. A lower summer maximum of 7.8 psi has been in effect in ozone nonattainment areas in VOC Control Region 1, such as California, since 1992.[5] However, Phase 2 of the California RFG program that took effect in the first half of 1996 required more extensive changes to gasoline properties.[6] A number of programs have assessed the effect of the use of RFG on vehicle emissions; the most comprehensive of these is the Auto=Oil Air Quality Improvement Research Program.[4] Their task of assessing the air quality benefits of RFGs was relatively difficult, compared to the other evaluations of control measures that have typically been undertaken for regulatory purposes.[7] The tailpipe emissions of the major pollutants regulated under the CAA such as carbon monoxide, hydrocarbons, and nitrogen oxide were reduced by 10–20% using RFG in conventional automobiles compared to that with conventional gasoline. The changes in gasoline properties included an increase in oxygen content, decrease in alkenes, aromatics, benzene, and sulfur content, reduction in volatility, and decrease in distillation temperatures.[4] Evaporative VOC emissions can be reduced by limiting the vapor pressure of RFG during the summer months. The limit is lower for the southern part of states because higher ambient air temperatures increase evaporative losses. The RFG vapor pressure limits are lower than the current Environmental Protection Agency (EPA) limits. Lowering RVP increased the refiner’s cost of producing gasoline because low-cost normal butane must be removed from the gasoline pool. Phase 2 required approximately a 1.3 psi reduction in RVP (from 8 to 6.7 psi) in northern control areas (Region 2) and a 0.4 psi reduction (from 7.1 to 6.7 psi) in southern areas

Reformulated Gasoline

(Region 1) from Phase 1 RFG levels during the summer months. Reid vapor pressure plays an important role in reducing VOC emissions. However, reduction in RVP alone is not enough to achieve the required Phase 2 VOC reduction. A reduction in VP to 6.7 psi will reduce VOC emissions by about 24% in Region 1 and 22% in Region 2, well below the 29% and 27.4% required in Regions 1 and 2, respectively. Reduction in sulfur from 300 to 140 ppm will yield an additional reduction of 1.9%. Lowering aromatics from 32 to 26 vol% in Phase 1 provided an additional 1.5% VOC reduction that is still not enough to meet Phase 2 VOC reduction. The final necessary emission reductions must come from increasing E200, E300, and olefins, without violating the NOx emission reduction requirement.[6]

TOTAL OXYGEN CONTENTS AND TYPE OF OXYGENATES Oxygenates represent a key component of RFG. This fuel is required to contain an average of 2.1% oxygen by weight. The source of this oxygen is oxygenates organic combustible liquids containing an oxygen atom in their structure. While methyl tertiary-butyl ether (MTBE) and ethyl alcohol (ethanol) have been the oxygenates most commonly used in gasoline, ethyl tertiary-butyl ether (ETBE) and tertiary-amyl methyl ether (TAME) are likely to be used in RFG as well.[5] As the different oxygenated compounds contain different amounts of oxygen, the amount of oxygenate needed to meet the RFG oxygen requirement depends on which oxygenate the refiner=blender uses. Initially, it was believed that the oxygen requirement would advance the RFG goals but lately on subsequent testing has shown that oxygen content has very little effect on VOC or toxic emissions. Oxygenates are either alcohol or ether. The purpose of oxygenates in gasoline is to reduce CO and hydrocarbons emission by assisting fuel combustion. Since 1979, oxygenates have been added to the fuel in the form of MTBE and ethanol as octane enhancers to replace lead in limited areas of the country. During the 1980s, oxygenates were used on a large scale because of the implementation of oxygenated gasoline programs to control emissions of CO in cold weather. In 1990, the CAA Amendments required the use of a high percentage of oxygenates in gasoline. For example, seasonally 15% MTBE by volume or year-round 11% MTBE is added in parts of the United States where CO levels in the winter or ozone concentrations in the summer exceed the National Ambient Air Quality Standards. Several researchers have studied the effect of oxygenates in gasoline on pollutant emissions. Kirchstetter et al. reported no change in NOx emissions

Reformulated Gasoline

and decreases in CO and VOC for the oxygenated gasoline.[8] On the other hand, Keller et al. reported no significant change in exhaust emissions from advanced technology vehicles when comparing oxygenated and nonoxygenated gasolines that meet all other California Phase 2 RFG standards.[9] The emission reduction benefit of oxygenated gasolines is highest for older vehicles with open loop control. Newer vehicles with computerized closed-loop control get a lesser benefit, and there is only a small benefit for advanced technology vehicles. These vehicles have a feedback emission control system that takes away the majority of the effect of the oxygenate.[10] However, the oxygenate requirement remains because of the following reasons: 1) it is an octane booster; 2) it is helpful to older vehicles; and 3) it is domestically produced, displacing imported petroleum. The most commonly used oxygenate in the United States was MTBE because of its compatible blending properties, high octane, low vapor pressure, availability, and low cost. However, this oxygenate is being phased out because of its detection at low levels in groundwater and possible toxicity.[10] Several studies have been conducted to find an alternative to MTBE.[11,12] At this moment, ethanol is the only approved replacement.

DIFFERENT PHASES OF REFORMULATED GASOLINE PROGRAM Phase I of RFG Phase I of RFG was introduced on January 1, 1995. It contains both formula standards and performance standard.[13] These standards for RFG are described below. Formula standards The formula standards for RFG are as follows: maximum of 1 wt% benzene, maximum of 25 wt% total aromatic content, 2 wt% minimum of oxygen, no metals and detergent additives. Performance standards Because of the various needs for suppliers, consumers, regulators and the complexity of predicting fuel emissions, the development of performance standard involved different groups such as the EPA along with the petroleum industry, auto manufacturers, oxygenate producers, the Department of Energy, various environmental groups, and state air pollution directors, etc.[13] Two models were developed by the EPA to predict emission performance from fuel composition. The models are described below:

2627

Simple Model. The simple model was available for use from 1995 to 1997. In the simple model it was believed that RVP, oxygen, benzene, and aromatics will have a significant effect on emissions. However, the effects of aromatics on VOC and NOx emissions were unclear. During the ‘‘high-ozone season’’ (summertime), the performance characteristics of the RFG were estimated to reduce total car VOCs and total toxic emissions by 15% of the 1990 baseline levels, or to meet the equivalence of a formula fuel performance, whichever is more stringent.[13] The simple model was not directly used for monitoring VOC emissions. The EPA determined that a fuel with an RVP of 8.1 psi and 2.0 wt% oxygen would be sufficient to achieve the minimum 15% VOC emission reduction. Baseline emission levels are from 1990 model year (MY) vehicles operated on a baseline gasoline (Table 1). The baseline fuel, derived from average survey data of the compositions of gasolines in the United States in 1989, is shown in Table 1. This baseline fuel was defined to indicate a performance standard and not a formula composition standard. To achieve equivalency certification, refiner and importer’s gasoline must comply with the CAA requirements for emissions as shown in Table 2.[14] The simple model allows a wide variety of gasolines to comply with the regulations without specifying or limiting any other gasoline properties. For example, the formula states that oxygen content be at least 2 wt% but it does not stipulate the form of oxygen. This leaves the choice to the refiner as long as the fuel is blended to contain 2 wt% oxygen. Also, the RFG must produce no increase in NOx emissions because NOx and hydrocarbons react in the lower atmosphere or troposphere to form ozone in the presence of sunlight.[13] Table 1 Baseline gasoline properties defined by the EPA in 1989 Property

Summer

Winter

API gravity

57.4

60.4

RVP (psi)

8.7

11.7

IBP ( F)

91

87

10% ( F)

128

111

50% ( F)

218

199

90% ( F)

330

332







EBP ( F)

415

404

Aromatics (vol%)

32

26.4

Olefins (vol%)

9.2

11.9

Saturates (vol%)

58.8

61.7

Sulfur (ppm)

339

338

Benzene (vol%)

1.53

1.62

Octane (R þ M)=2

87.3

88.1

R, rated; M, measured. (From Ref.[13].)

R

2628

Reformulated Gasoline

Table 2 Simple model baseline emissions Summer Region 1

Region 2

Winter

Exhaust VOCs (g=mL)

0.444

0.444

0.656

Nonexhaust VOCs (g=mL)

0.858

0.766

0

Total VOCs (g=mL)

1.3

1.21

0.656

Exhaust benzene (mg=mL)

30.1

30.1

40.9

Evaporated benzene

4.3

3.8

0.0

Running loss benzene

4.9

4.5

0.0

Refueling benzene

0.4

0.4

0.0

1,3-Butadiene

2.5

2.5

3.6

Formaldehyde

5.6

5.6

5.6

Acetaldehyde

4.0

4.0

4.0

POMs

1.4

1.4

1.4

TAPs (mg=mL)

53.2

52.3

55.5

Region 1 is southern areas of the United States typically covered by ASTM class B during summer. Region 2 is northern areas of the United States typically covered by ASTM class C during summer. (From Ref.[13].)

Complex Model. A complex model is a set of statistically derived equations that relate fuel properties to vehicle emissions. This model became mandatory in 1998. The simple model calculates emission based on a fuel’s RVP, oxygen, aromatic, and benzene content whereas the complex model adds four more variables (sulfur, olefin, and the 200 F and 300 F distillation volume fractions) to the equation. This model is based on the data collected from programs conducted around the United States. The database was made up of over 200 test fuels, 500 automobiles, and 5000 emission testings.[15] The complex model can be divided into two portions: exhaust and nonexhaust. The nonexhaust VOC was derived directly from the simple model approach where the nonexhaust benzene was modeled as a weight fraction of nonexhaust VOC from the headspace model of General Motors. The exhaust model was based on 19 different test programs.[13]

Phase II of RFG Phase II of the RFG program began at refineries on December 1, 1999, and at retails outlets beginning January 1, 2000. This part of the RFG requirements used the complex model with stricter standards, as follows: a 25% reduction of VOCs and toxic emissions from baseline levels, with a latitude granted to the EPA encompassing consideration of technological feasibility

and cost.[13] The Phase II complex emissions model uses the same variables as the Phase I complex emissions model. However, the estimated emissions using the Phase II model are different from those predicted by the Phase I model. An approximate comparison between Phase I and Phase II is given in Table 3. The VOC, NOx, and TAP emissions reduction performance standards under Phase I using Phase I complex emissions model and under Phase II using Phase II complex emissions model are not directly comparable because of the differences between the Phase I and Phase II complex emissions models.[6] From Table 3, it was observed that Phase I winter RFG comes very close to meeting the Phase II RFG requirement.

AIR QUALITY BENEFITS OF USING RFG The emissions from vehicles impact our environment in three different ways. They impact our air quality by contributing toxins to the air, by interacting with other emissions and sunlight to form ground-level ozone (smog) and by staying in the atmosphere as greenhouse gases.[16] The literature of the actual emissions from oxygenated fuels is quite limited. In 1987, the EPA found that oxyfuels, especially gashols, can reduce CO emissions by 10–30% in high-altitude areas, whereas in 1988 the EPA reported that CO reductions depend on the percentage of oxygen in the gasoline along with engine and exhaust-system technology. Analysis of fuel data submitted to the EPA by industry for compliance purposes shows that emission reductions from the RFG program have been more than the program requires each year since 1995. The Auto= Oil Air Quality Improvement Research Program (Auto=Oil) study highlights changes in fuel formulations that reduce automotive pollutants, especially photochemical ozone precursors and mobile air toxic emissions. This study concluded that California Phase 2 RFG reduces fleet average hydrocarbon emissions by 10–27% compared to industry average gasoline.[17] The estimated reductions in total air toxics emissions from the use of California Phase 2 RFG relative to conventional gasoline range from 9% to 32%, largely because of the reduced benzene and aromatic content of California Phase 2 RFG.[18] This gasoline is associated with significant decreases in benzene emissions and increases in formaldehyde emissions, as well as minor decreases in 1,3-butadiane and acetaldehyde emissions.[19] The Auto=Oil study compares the emissions from vehicles using nonoxygenated RFG to vehicles using 11 vol% RFG.[18,20] This study showed only a 13% decrease in formaldehyde tailpipe emissions with the oxygenate-free fuel. The Auto=Oil study also evaluates engine-out and tailpipe (exhaust) emissions for 157

Reformulated Gasoline

2629

Table 3 Reformulated gasoline averaging standards RFG Phase 1 (January 1995–December 1999)

RFG Phase 2 (January 2000)

R

Summer region 1

Summer region 2

Winter

Summer region 1

Summer region 2

Winter

Oxygen (wt% min)

2.1

2.1

2.1

2.1

2.1

2.1

Benzene (vol% max)

0.95

0.95

0.95

0.95

0.95

0.95

Product quality standard

Performance standards (using Phase II complex emissions model), percent reduction required Toxic air pollutants (TAP) (%)

18.5

17.8

17.3

21.5

21.5

21.5

Volatile organic compounds (%)

20.8

10.5

n.aa

29.0

27.4

n.aa

Nitrogen oxides (%)

1.4

1.6

1.7

6.8

6.8

1.5

a

Not applicable. (From Refs.[6,16].)

hydrocarbons for different fuel formations such as California Phase 2 RFG with MTBE, nonoxygenated California Phase 2 RFG, and a conventional nonoxygenated gasoline.[20] This study shows that California Phase 2 RFG with MTBE is associated with large increases in MTBE emissions and a lesser increase in isobutylene emissions and smaller increases in formaldehyde, acetone, and propadiene emissions compared to nonoxygenated RFG. Kirchstetter et al. compared the on-road emissions associated with California Phase 2 RFG and a California Phase 1 RFG.[4,21] The comparison was made by measuring the tunnel emissions for two sequential years when different statewide gasoline formulation standards were in effect. On the basis of representative samples of service stations, California Phase 2 RFG contained, on an average, 10.7 vol% MTBE in the summer of 1996 compared to 1.0 vol% MTBE for the Phase 1 formulation in the summer of 1995. Kirchstetter et al. also observed an 18% reduction in CO emissions, a 22% reduction in nonmethane organic carbon emissions, and a 6% decrease in NOx emissions from the summer of 1995 (California Phase 1 RFG) to the summer of 1996 (California Phase 2 RFG).[4,21] Gertler et al. performed a tunnel study in the Sepulveda Tunnel in Los Angeles in October 1995 (Phase 1 California RFG) and July 1996 (California Phase 2 RFG).[22] A significant decrease in CO and NOx vehicle emission was observed for 1996 California Phase 2 RFG compared to the California Phase 1 RFG measurements in 1995. In their study, the most important change in the hydrocarbons measured from the 2 yr was a nearly 50% increase in the faction of MTBE emissions. Case Study 1 In Mexico, 76.7% of the population lives in urban areas. It has three mega cities with a population

ranging from 1.5 to 8.0 million. In particular, Metropolitan Area of Mexico City (MAMC) alone has more than 18 million people. This urban region is expected to be one of the 20 largest urban regions of the world by the year 2015.[23,24] In the year of 2001 Mexico started a program to limit the exhaust emissions on new gasoline vehicles. Schifter et al. studied the fuel reformulation on the vehicle exhaust emissions in MAMC.[24] In their work, they prepared 15 gasoline blends using the current refinery streams available in the country. The RVP, sulfur, aromatic, olefin, and oxygenated molecule (MTBE or ethanol) were the main parameters considered in their studies. The United States Federal Test Procedure and FTP were used to evaluate total hydrocarbons, nitrogen oxides, and toxic exhaust emissions. The statistical design and analysis methods similar to those employed in the Auto=Oil Air Quality Research program were used in this work. In their work, different ranges of MYs were used, which acted as a surrogate for different emission control technologies. Thirty vehicles such as GT-1 (GM: Cutlass-89; Chrysler: Spirit-90; VW: Sedan-90; Nissan: Tsuru-90), GT-2 (Chrysler: Spirit-93; Nissan: Tsuru-93; GM: Cavalier-94; VW: Jetta-98; Nissan: Sentra-98; GM: Monza-98; Ford: Pick-up-98; Ford: Escort-98; Chrysler: Stratus-98), and GT-3 (Nissan: Tsuru-90; Chrysler: Neon-99; GM: Chevy-99; Chrysler: Stratus99; Dodge: Pick-up-99; Nissan: Pick-up-2000; VW: Pointer-2000; Nissan: Sentra-2001; GM: Monza-2001; Ford: Fiesta-2002; GM: Astra-2002; Ford: Focus-2001; VW: Jetta-2001; VW: Sedan-2001) were randomly recruited in areas of the MAMC of different socioeconomic indicators. Addition of MTBE deceased CO emissions to some extent in all cases. However, it was statistically significant in the case of the GT1 fleet. Substitution of MTBE by ethanol decreased CO emissions in the entire fleet and it was statistically significant only for GT-1 and GT-2 vehicles. Methyl

2630

Reformulated Gasoline

tertiary butyl ether had a statistically significant impact on total hydrocarbon emissions for the entire fleet. A similar impact on GT1 and GT2 was observed when ethanol was added, whereas its effect on GT3 was significant. Increase in RVP resulted in lower emissions, which was statistically significant on GT-2 and GT-3 fleets. Increase in olefins and aromatics did not have any significant impact. However, concerning NOx emissions, the only statistically significant impact was that increasing olefins raises emissions on GT-1. For selected fuels (see Table 4) a comparison of calculated straight average values of VOCs, TAP, and NOx was made using the EPA Phase 1 and Phase 2 complex model. For complex model predictions, see Table 5. From Table 5 it was observed that F-10 had the highest reduction among the tested fuels with the 1990 MY’s American fleet.

Case Study 2 The effects of the use of reformulated and oxygenated gasoline fuel blends on the air quality of European city were studied by Vinuesa et al.[25] The urban region of the Strasbourg–Kehl area (GSKA) in the middle of the upper Rhine valley was chosen as the area of investigation. The part of the valley used in this work was flat, 29 km wide, and 33 km long. The chosen period of investigation was May 9–15, 1998, which corresponded to a photochemical ozone pollution episode with low synoptic wind and high temperature (3–5 C above Table 4 Characteristics of gasoline motor vehicle fuels Fuel Properties

I/Aa

MAa

F-10a

F-13a

Aromatics (vol%)

28.0

24.1

19.8

40.3

Olefins (vol%)

13.5

9.0

5.0

4.8

Oxygen, as MTBE (wt%)

0.34

1.21

1.03

1.14

Benzene (vol%)

1.1

1.0

0.6

1.1

RVP (psi)

8.88

7.66

6.62

8.06

RON

91

91

91

93

MON

84

84

84

85

(RON þ MON)=2

88

87

88

89

35.9

37.9

40.5

39.8

10% ( C)

57.9

59.5

68.4

65.9

50% ( C)

97

101

104

111

D-86 distillation IBP ( C) 



a

90% ( C)

167

163

168

162

EP% ( C)

203

205

205

200

Sulfur (ppm)

724

403

89

34

Fuel code names. (From Ref.[24].)

Table 5 Comparison of Schifter et al.’s work with the EPA complex model Fuel RFG-1

a

RFG-2

b

I=A MA F-10 F-13 a

Total VOC

NOx

17.0

1.5

27.0

3.82

17.44

29.68 5.74

Phase 1 complex model. b Phase 2 complex model. (From Ref.[24].)

7.0

9.36

Total toxics 17.0 22.0

1.66

0.13

19.86

12.93

12.84

12.8

33.94

the seasonal average). Their study focused mainly on May 11 when the maximum value of 193 mg=m3 of ozone was measured in the center of Strasbourg. Ethyl-t-butyl ether was used as an oxygenated compound in their work. The daily emissions of CO, VOC, and NOx (in milligrams or tons) for May 11, 1998, are given in Table 6. The goal of their work was to examine the differences in pollutant concentrations predicted by the model when the gasoline vehicles used different fuel blends. The wind fields were exactly the same in the reference case and in different scenarios and consequently the concentration levels were explained in a similar way. The effects of alternative fuels on the urban air quality were studied by building emission scenarios based on available emission factors. From their work they observed that the reformulation directly decreased the aromatic fraction in the composition of the fuel and the oxygenated compounds decreased this part by dilution. They also observed that the use of alternative fuels considerably reduced the direct emissions of total VOC by up to 30% and 45%, respectively. For NO and NO2 the situation was less clear. For all the scenarios, the simulated ozone levels were slightly lowered in the range of 1– 5% and the maximum effect was obtained when all the vehicles used these alternative fuel blends. When the percentage fell to 80%, the reduction of ozone became less than 3%. However, from their work they concluded that the use of reformulated and oxygenated fuel blends led to some improvement of the air quality at a local scale but it did not lead to the drastic changes even when all the fleets use various alternative fuels. Impacts of Oxygenated Compounds on the Environment In RFG, the most commonly used oxygenates are MTBE (methyl tertiary butyl ether) and ethanol. As a component of RFG, MTBE plays a role in reducing total hydrocarbons, air toxics, and CO.[18] However, it is also responsible for certain pollutant emissions.

Reformulated Gasoline

2631

Table 6 Daily emissions of CO, VOC, and NOx for 11 May 1998 CO emisssions Sources

mg

% CO

Biogenic (forest) Residential housing

12.9

18.7

Residential solvent use

VOC emissions mg

% VOC

4.14

15.0

1.52

5.5

NOx emission mg

% NOx

0.77

2.4

5.3

Ternary sector housing

0.17

0.3

0.02

0.1

0.36

1.1

Industries (point sources)

0.81

1.2

6.05

22.0

5.25

16.5

Other industries

0.17

0.2

0.04

0.1

1.10

3.5

0.85

3.1

Petrol station (distribution) Petrol station (storage) Road traffic (line sources) Road traffic (surface sources) Road traffic evaporation Air traffic

0.69

2.5

44.6

64.7

7.07

25.7

19.8

62.2

6.0

8.7

1.00

3.6

2.1

6.4

2.28

8.3

3.7

5.4

2.18

7.9

0.75

2.4

0.07

0.1

0.03

0.1

0.14

0.4

Railway traffic

0.53

0.8

0.22

0.8

1.62

5.1

Fluvial traffic

68.9

100

27.5

100

31.9

100

Total (From Ref.[25].)

Methyl tertiary butyl ether enters the environment in different ways such as auto emissions, evaporative losses from gasoline stations and vehicles, storage tank releases, pipeline leaks, and accidental spills, and refinery stock releases. When gasoline is released into air, a significant portion exists in air and a small portion enters into soil and water.[26] When it is released into water a significant portion remains dissolved in surface water, with some partitioning into air and a much smaller amount into soil.[27] Because of its high solubility, it moves through the soil and into the groundwater more rapidly than other chemicals.[26] Time spent at the service station, driving cars, at the parking garage and in homes with attached garage are the nonoccupational sources of gasoline emission to the environment and human exposure occurs through inhalation during fueling of automobiles.[28] Also, leakage from stationary sources (under ground storage tank) to soil or groundwater can contaminate the water supply. Methyl tertiary butyl ether has a strong taste and odor. A small amount of MTBE in water can make the water supply distasteful. Dermal contact (exposure of skin to toxic materials) of gasoline may occur though accidental spills of MTBE-blended gasoline or through the use of gasoline as a solvent.[29] The estimated arithmetic mean occupational dose via air is in the range of 0.1–1.0 mg=kg=day, whereas doses from residential exposures, communicating, and refueling are in the range of 0.0004–0.006 mg=kg=day.[30] During refueling of a car, the concentrations of MTBE

range from less than 1 ppm to 4 ppm (within the breathing zone) and from 0.01 ppm to 0.1 ppm inside cars (1 ppm ¼ 3.57 mg=m3 at 25 C, 1 atm).[31] However, the reference dose imposed by the EPA is 42 mg=kg=day. The data on the presence of MTBE in drinking water are very limited. The most extensive monitoring data on MTBE in drinking water are available from California. In January 2000, 1444 systems had tested 6492 sources of drinking water. Methyl tertiary butyl ether was detected in 52 (0.8%) of these sources, including 31 of 6076 groundwater sources (0.5%) and 21 of 416 surface water sources (5%). Overall, 30 (2.1%) of the 1444 public water systems reported detection of MTBE in at least one of their drinking water sources. Although the state database did not include some contaminated wells that have been closed, very few sources had MTBE concentrations exceeding the EPA taste and odor drinking water recommendation of 20–40 mg=L.[26]

ALTERNATIVES OF MTBE Because of the adverse effects of MTBE, ethanol is now considered as a promising alternative oxygenate. Ethanol is now marketed and distributed in every state in the United States. Ethanol is safe, biodegradable, and renewable. It also does not harm drinking water resources. Currently, around 80% of gasoline in California is blended with ethanol. At this time, the total ethanol consumption in California is 950 million

R

2632

Reformulated Gasoline

gallons per year, whereas in 2002 the consumption was 100 million gallons.[32] But other concern also arises in the case of ethanol. It raises vapor pressure of the ethanol–gasoline mixture, resulting in increased evaporative emissions.[10] Also, there are cost and supply issues if a large amount of ethanol is required.[11,12]

CONCLUSIONS The RFG program is a significant step toward cleaning the air we breathe. Evaporative VOC emissions can be reduced by limiting the vapor pressure of RFG during the summer months. The RFG program reduces emissions of TAP such as benzene, a known human carcinogen. It is possible to reduce the tailpipe emissions of the major pollutants regulated under the CAAcarbon monoxide, hydrocarbons, and nitrogen oxide by 10–20% using RFG. Methyl tertiary butyl ether oxygenate is being phased out because of its adverse effects on the environment. Ethanol is now considered as a promising alternative to MTBE.

NOMENCLATURE API ASTM CAA EBP ECS ETBE E200 E300 MTBE IBP POM RFG RVP TAP TAME VOC

American Petroleum Institute American Standards for Testing and Materials Clean Air Act End boiling point Emission control system Ethyl tertiary butyl ether Percentage of fuel evaporated at 200 F Percentage of fuel evaporated at 300 F Methyl tertiary butyl ether Initial boiling point Particulate organic matter Reformulated gasoline Reid vapor pressure Toxic air pollutants Tertiary amyl methyl ether Volatile organic compound

REFERENCES 1. Cannon, J.S.; Azimi, L.S. Reformulated gasoline: cleaner air on the road to nowhere. Int. Assoc. Hydrogen Energy 1995, 20 (12), 987–994. 2. Treiber, S.; McLeod, R.S.; Faitakis, Y.; Hutchings, R.L. The challenge to conventional blending technology. Hydrocarbon Process. 1998, 77 (6), 101–113.

3. Bowman, F.M.; Seinfeld, J.H. Atmospheric chemistry of alternate fuels and reformulated gasoline components. Prog. Energy Combust. Sci. 1995, 21, 387–417. 4. Cohen, J.P.; Yarwood, G.; Noda, A.M.; Pollak, A.K.; Morris, R.E. Auto=Oil Air Quality Improvement Research Program: Development of Emissions Reactivity Values for Phase II Results, SYSAPP94-94; Systems Applications International: San Rafael, CA, July 1994. 5. http:==www.chevron.com=prodserv=fuels=bulletin= fed-refm=fg-char.shtml. 6. Lidderdale, T.; Bohn, A. Demand and price outlook for Phase 2 reformulated gasoline, 2000. Energy Inf. Admin.=Pet. Supply Monthly, 1999 Apr. 7. Yang, Y.-J.; Milford, J.B. Quantification of uncertainty in reactivity adjustment fcators from reformulated gasolines and methanol fuels. Environ. Sci. Technol. 1998, 30, 196–203. 8. Kirchstetter, T.W.; Singer, B.C.; Harley, R.A.; Kendall, G.R.; Chan, W. Impact of oxygenated gasoline use on California light duty vehicle emissions. ES & T 1996, 30, 661–670. 9. Keller, A.; Froines, J.; Koshland, C.; Reuter, J.; Suffet, I.; Last, J. Health and Environmental Assessment of MTBE in Gasoline, Report to the Governor and Legislature of the State of California as Sponsored by SB521, Vol. 1; Summary and Recommendations, Nov 1998. 10. Maclean, H.L.; Lave, L.B. Evaluating automobile fuel=propulsion system technologies. Prog. Energy Combust. Sci. 2003, 29, 1–69. 11. California Energy Commission. Supply and Cost of Alternatives to MTBE in Gasoline, Staff Report, Oct 1998; P300-98-013. 12. California Energy Commission. Supply and Cost of Alternatives to MTBE in Gasoline, Technical Appendices, Staff Report, October 1998; P30098-013B. 13. Khan, R.M.; Popiel, E.; Reynolds, G.J. Designing fuel for the environment: reformulated gasoline. Energy Sour. 1996, 18, 513–523. 14. Federal Register. Regulation of fuels and fuels additives; standards for reformulated and conventional gasoline. Fed. Reg. 1994, 59 (32), 7716 15. Korotney, D. The complex model made simple, Paper PETR-4. American Chemical Society National Meeting, Washington, DC, Aug 1994. 16. Rask, K. Clean air policy and oxygenated fuels: do we get what we pay for. Energy Econ. 2004, 26, 161–177. 17. Auto=Oil Air Quality Research Improvement Program (Auto=Oil). Gasoline Reformulation and Vehicle Technology Effects on Exhaust Emissions, Technical Bulletin No. 17, 1995.

Reformulated Gasoline

18. Auto=Oil Air Quality Research Improvement Program (Auto=Oil). Program Final Report, 1997. 19. Franklin, P.M.; Koshland, C.P.; Lucas, D.; Sawyer, R.F. Evaluation of combustion byproducts of MTBE as a component of reformulated gasoline. Chemosphere 2001, 42, 861–872. 20. Auto=Oil Air Quality Research Improvement Program (Auto=Oil). Phase I and Phase II Test Data Public Release on Compact Disc; Systems Application International, Inc.: San Rafael, CA, 1996. 21. Kirchstetter, T.W.; Singer, B.C.; Harley, R.A. Impact of California reformulated gasoline on motor vehicle emissions. 1. Mass emission rates. Environ. Sci. Technol. 1999, 33, 318–328. 22. Gertler, A.W.; Sagebiel, J.C.; Dippel, W.A.; O’Connor, C.M. The impact of California phase 2 reformulated gasoline on real world vehicle emissions. J. Air Waste Manage. Assoc. 1999, 49, 1339–1346. 23. Bravo, H.A.; Torres, R.J. The usefulness of air quality monitoring and air quality impact studies before the introduction of reformulated gasolines in developing countries. Mexico city, a real case study. Atmos. Environ. 2000, 34, 499–506. 24. Schifter, I.; Diaz, L.; Vera, M.; Guzman, E.; Lopez-S, E. Fuel formulation and vehicle exhaust emissions in Mexico. Fuel 2004, 83, 2065–2074.

2633

25. Vinuesa, J.-F.; Mirabel, P.; Pnche, J.-L. Air quality effects of using reformulated and oxygenated gasoline fuel blends: application to the Strasbourg area (F). Atmos. Environ. 2003, 37, 1757–1774. 26. Ahmed, F.E. Toxicology and human health effects following exposure to oxygenated or reformulated gasoline. Toxicol. Lett. 2001, 123, 89– 113. 27. WHO (World Health Organization of the United Nations). Environmental Health Criteria 206: Methyl Tertiary-Butyl Ether; WHO: Geneva, Switzerland, 1998. 28. Dourson, M.L.; Felter, S.P. Route-to route extrapolation of the toxic potency of MTBE. Risk Anal. 1997, 17, 717–725. 29. NRC (National Research Council). Toxicological and Performance Aspects of Oxygenated Motor Vehicle Fuels; National Academy Press: Washington, DC, 1996. 30. Brown, S.L. Atmospheric and potable water exposures to methyl tert-butyl ether (MTBE). Reg. Toxicol. Pharmacol. 1997, 25, 256–276. 31. Hartle, R. Exposure to methyl tert-butyl ether and benzene among station attendants and operators. Environ. Health Perspect. 1993, 101 (6), 23–26. 32. http:==www.ethanolrfa.org=leg_position_mtbe. shtml.

R

Renewable Energy R Gareth P. Harrison Institute for Energy Systems, School of Engineering and Electronics, University of Edinburgh, Edinburgh, U.K.

INTRODUCTION Climate change is widely regarded as one of the most significant global challenges in the 21st century. With conventional fossil fuel generation a major contributor, the expansion of renewable energy use is a critical element of the strategy to lower emissions of greenhouse gases. Many countries have set challenging targets for renewable use and deployment driven by the environmental, sustainability, and security benefits that may be attributed to renewables. Here, these arguments are reviewed along with the renewable technologies available. The challenges for integrating these variable and often intermittent generating sources are highlighted and alternative applications in creating new energy vectors and in direct use are explored. Finally, the economics of renewables are examined with particular reference to the limitations of traditional comparisons with conventional sources and the impact of more robust techniques on the relative cost of fossil fuels and renewables.

THE DRIVE FOR RENEWABLE ENERGY The case for promoting renewable energy revolves around the three key benefits associated with it: 1. Nonpolluting 2. Infinite reserves 3. Security of supply Pollution The primary argument in favor of renewable energy is that it does not entail the release of chemical pollutants to convert the energy. This is in contrast to fossil fuels, which emit a range of pollutants from carbon dioxide (CO2) to the components of acid rain to ash. Table 1 shows the pollutants from typical 2000 MW plants. The combustion of fossil fuels produces the majority of anthropogenic CO2 with transport and power generation the largest sources. CO2 is widely accepted as the major cause of climate change, which, if left unchecked, is predicted to lead to a global temperature Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120007945 Copyright # 2006 by Taylor & Francis. All rights reserved.

rise of between 1.5
C and 5.8
C by the end of the century (Fig. 1).[2] This, and accompanying changes in other climate variables (precipitation, wind speed, etc), will have impacts on many sectors ranging from agriculture to human health that will be seen at local, regional, and global scales. In responding to this, many industrialized countries have signed the Kyoto Protocol requiring cuts in CO2 and other gases by 2010, which became legally binding on participants in February 2005 following ratification by Russia. A difficulty for the Kyoto agreement is the refusal of the U.S.A. and Australia to ratify it, partly because large developing nations like China and India escape emissions limits. In July 2005, the U.S.A. and five AsiaPacific states announced a voluntary pact to reduce emissions by developing new technology like ‘clean coal;’[3] the effectiveness of this nonbinding agreement is disputed, however. These aggrements are the beginning of a longer process, with the United Kingdom Royal Commission on Environmental Pollution recommending CO2 cuts of 60% by 2050 to limit the eventual rise of greenhouse gas concentrations to twice the preindustrial level.[4] Achieving the reductions required by Kyoto and the longer-term targets will not be straightforward. To date, modest CO2 emissions have been achieved through switching to less carbonintensive fuels like natural gas (albeit justified on a cost basis) but, ultimately, emission-free energy is required. While continued fossil fuel use will be possible if largescale capture and sequestration of carbon can be achieved, the development and deployment of truly carbon-free energy sources like hydrogen, nuclear, and renewables will be critical. Fossil-fueled power stations produce significant quantities of sulfur dioxide (SO2) and nitrogen oxides (NOx), which are precipitated as ‘‘acid rain’’ across wide areas and legislation such as the European Union Large Combustion Plant Directive has imposed emission limits in response. SO2 emissions can be reduced by a range of techniques including the treatment of coal before ignition, combustion processes that collect sulfur in the ash, and SO2 removal from the combustion product gases, known as flue gas desulfurization (FGD). Although effective, FGD is expensive, particularly so for retrofit, and reduces station efficiency.[5] While road transport is the major cause of NOx emissions, power 2635

2636

Renewable Energy

Table 1 Typical emissions from 2000 MW fossil-fueled power stations (in ktons=yr) Pollutant

Conventional coal (no FGD)

Carbon dioxide Sulfur dioxide Nitrogen oxides Airborne particulates Solid waste and ash Ionizing radiation (Bq)

Conventional oil

Combined-cycle gas turbine

11,000

9,000

6,000

150

170

0

45

32

10

7

3

0

840

0

0

10 9

1012

10 11

[1]

(From Ref. .)

stations still contribute a significant fraction. A wide variety of NOx control technologies have been developed (e.g., flue gas scrubbing with alkalis) but again are expensive. Despite not deploying NOx scrubbing, the United Kingdom has achieved reductions with greater use of gas-fired plant and the retrofit of low-NOx burners to existing coal stations. Use of renewables can avoid these costly removal processes. Radiation is emitted from nuclear and, because of the presence of trace elements, from fossil-fueled power stations. The amounts are generally small, with public exposure similar to that of background radiation levels. However, the major issues surrounding nuclear power are how to deal with radioactive waste products and how to avoid proliferation of radioactive material. Until these issues are resolved and a clear lead given by governments, new nuclear plant is unlikely to be built, particularly not by the private sector. Once again, renewables can begin to fill the gap. Finite Resources There is also the issue of sustainability of fuel supplies. Fossil fuels are the product of sedimentation processes over millions of years, but they will have been consumed in a matter of centuries. Table 2 shows recent

estimates of global proven fossil fuel reserves and an estimate of their remaining lifetime assuming current usage rates. While there is uncertainty over the remaining reserves, particularly given that further oil and gas fields will be identified, there is little doubt that with rapidly increasing demand for oil and gas and increasing exploitation costs (in financial and energy terms) supplies will be limited to several decades. Fortunately, there are sufficient renewable energy resources to meet our energy needs many times over, although in many cases the means of doing this in a technically and economically efficient manner is still under development. Table 3 shows an estimate of the power available from a range of renewables. To put these in perspective, global installed electrical capacity is currently just over 3000 GW (3 TW) and is anticipated to grow by some 120 GW=yr. Although biomass is currently the most heavily used renewable in overall energy terms, very little is used to generate electricity. Accordingly, hydropower is the number one renewable in electricity terms and contributes around 19% of global supply. Installed capacity is currently around 650 GW with a further 100 GW under construction. In recent years wind has seen the largest growth and, with 8 GW added in 2003 alone, global installed capacity is now almost 40 GW; forecasts suggest an installed capacity of 1250 GW by 2020 meeting 12% of electrical needs.[8] Security An increasingly strong argument in favor of renewables is their positive effects on security by spreading risk. At the crudest level this can be interpreted as Table 2 Proven fossil fuel reserves and reserve=production ratio

Fig. 1 History and range of future temperature rise. (From Ref.[2].) (View this art in color at www.dekker.com.)

Fuel

Proven reserve

Reserve to production ratio

Oil

156 billion tonnes

41

3

Gas

176 trillion m

Coal

984 billion tonnes

(From Ref.[6].)

67 192

Renewable Energy

2637

Table 3 Estimate of global renewable energy resource Estimate of recoverable resource

Resource base

1,000 TW

90,000 TW

Wind

10 TW

1,200 TW

Wave

0.5 TW

3 TW

Tides

0.1 TW

30 TW

—

30 TW

1,150 TW

450 TW yr

Resource Solar radiation

Geothermal (heat flow) Biomass (From Ref.[7].)

increasing energy self-sufficiency and lowering the risk of supply interruptions by accident, terrorist activity, or the actions of politically unstable fossil fuel exporting nations. In the U.K., fuel diversity is often indicated using the Shannon–Weiner index, which measures the logarithmic weighting of fuel technologies. Hence, by increasing the variety of fuel sources through new types and increasing volumes of renewables, diversity and, by implication, security increase. Security can also be measured in terms of the economic implications of exposure to volatile fossil fuel prices. Studies suggests that fossil fuel price volatility has a range of negative effects on economic activity including impacts on employment levels and asset values; for example, a 10% oil price spike is estimated to reduce economic growth in the United States alone by as much as $200 billion over the following year.[9] This dwarfs current and future investment to commercialize renewables estimated at around $125 billion between 2001 and 2010.[10] The deployment of renewables reduces exposure to fossil price risk and this can be shown to lower overall cost.[11] This perhaps surprising result arises as the costs of renewables do not have any correlation with fossil fuel price changes. The addition of renewables creates a diversified generation portfolio that serves to lower overall generating costs for a given level of risk.

Hydropower Hydropower is the conversion of the gravitational potential energy of water into electricity. The power generated depends on the water flow rate and the height through which the water falls—the ‘‘head’’—and is roughly 10 times the product of the two quantities. Global installed capacity is approximately 650 GW and produced 2700 TW hr in 2003 (around 19% of primary energy).[6] The estimated technical potential is around 14,000 TW hr=yr but the economically exploitable potential is between 40% and 65% of that, with much of the more economic plant already installed. The pattern of availability and exploitation varies significantly with countries like Norway producing 99% of their electricity from it while others little or none. The more developed regions have exploited far greater proportions of their resource while only around a small fraction of Asia’s much larger potential has been tapped. Hydropower is characterized by high initial capital costs offset by a long lifetime (civil engineering works often last more than 50 yr), high reliability, and low operational costs. Hydro is generally defined on the basis of its installed capacity as Table 4 indicates. Large hydro is often defined as a grid-connected scheme in excess of 20 MW capacity and would normally possess a dam and a storage reservoir (Fig. 2). The largest scheme, the Three Gorges project on the Yangtze, will shortly have a capacity of 18 GW. Most large-scale schemes were developed prior to 1990 and the potential for new large schemes is now rather limited given that there are fewer commercially attractive sites still available, but perhaps more importantly, because of opposition on environmental grounds mainly as a result of the flooding of land to create the storage reservoir.

BRIEF REVIEW OF RENEWABLE ENERGY There are a range of renewable energy sources currently in use in electricity systems or with the potential to contribute significantly in the future. The following energy sources are briefly reviewed and their merits explored:       

Hydropower Solar energy Wind energy Geothermal energy Biomass Tidal power Wave energy

Fig. 2 Typical embankment dam in Texas, U.S.A. (View this art in color at www.dekker.com.)

R

2638

Renewable Energy

Table 4 Hydro capacity definitions Type

Capacity

Micro hydro

1000 the catalyst is severely contaminated and may actually have to be replaced. Nevertheless, it is possible to process the heavier feedstocks in catalytic crackers and bring about some degree of desulfurization in the process. The degree of desulfurization depends on the amount of sulfur in the feedstock. It is generally believed that to optimize use of a catalytic cracking unit, feedstocks should be treated to remove excess high molecular weight material and metals by processes such as visbreaking, coking, or deasphalting to prolong catalyst activity. Sulfur compounds are changed in such a way that the cumulative sulfur content of the liquid and nonvolatile products is lower than the sulfur content of the original feedstock. The decomposition of sulfur constituents into hydrocarbons and hydrogen sulfide (or other gaseous sulfur products) occurs.

Hydroconversion Hydroconversion (variously referred to as hydrotreating and hydrocracking) is a group of refining processes in which the feedstock is heated with hydrogen under pressure. The outcome is the conversion of resids to a range of products. Hydrotreating is generally used for the purpose of improving product quality without appreciable alteration of the boiling range. Mild processing conditions are employed so that only the more unstable materials are attacked. Thus, nitrogen, sulfur, and oxygen compounds undergo hydrogenolysis to split ammonia, hydrogen sulfide, and water, respectively. Olefins are saturated, and unstable compounds, such as di-olefins, which might lead to the formation of gums or insoluble materials, are converted to more stable compounds. Heavy metals present in the feedstock are also usually removed during hydrogen processing. On the other hand, hydrocracking is a process in which thermal decomposition is extensive and the hydrogen assists in the removal of the heteroatoms as well as mitigates the coke formation that usually accompanies thermal cracking of high molecular weight polar constituents. Hydrocracking is similar to catalytic cracking, with hydrogenation superimposed and with the reactions taking place either simultaneously or sequentially. Hydrocracking was initially

Resid Conversion

used to upgrade low-value distillate feedstocks, such as cycle oils (highly aromatic products from a catalytic cracker that usually are not recycled to extinction for economic reasons), thermal and coker gas oils, and heavy-cracked and straight-run naphtha. These feedstocks are difficult to process by either catalytic cracking or reforming, because they are usually characterized by a high polycyclic aromatic content and=or by high concentrations of the two principal catalyst poisons, sulfur, and nitrogen compounds. Hydrocracking is perhaps the single most significant advance in petroleum refining technology over the last several decades. It is essentially an efficient thermal catalytic method of converting resids to lower boiling products. However, hydrocracking should not be regarded as a competitor for catalytic cracking. Catalytic cracking units normally use virgin gas oils as feedstocks; hydrocracking feedstock usually consists of refractive gas oils derived from cracking and coking operations. Hydrocracking is a supplement to, rather than a replacement for, catalytic cracking. The problems encountered in hydrocracking resids can be directly equated to the amount of complex, higher boiling constituents that may require pretreatment. Processing these feedstocks is not merely a matter of applying know-how derived from refining conventional crude oils but also requires knowledge of composition. The choice of processing schemes for a given hydrocracking application depends on the nature of the feedstock as well as the product requirements. The single-stage process can be used to produce gasoline but is more often used to produce middle distillate from heavy vacuum gas oils. The two-stage process was developed primarily to produce high yields of gasoline from straight-run gas oil, and the first stage may actually be a purification step to remove sulfur-containing (as well as nitrogen-containing) organic materials. Both processes use an extinction–recycling technique to maximize the yields of the desired product. Significant conversion of resids can be accomplished by hydrocracking at high severity. For some applications, the products boiling up to 340
C (650
F) can be blended to give the desired final product. Product yields and hydrogen consumption vary with the feedstock but the process can be used in many different ways. One of the major advantages of hydrocracking is that it may be used to process the higher-boiling refractory feedstocks that may be produced by catalytic cracking or by any of the coking processes. Hydrocracking can also be applied to the conversion of the more difficult feedstocks, such as residua and resides, and there are a variety of hydrocracking processes that are designed specifically for this particular use. Hydrocracking is an extremely versatile process that can be utilized in many different ways. One of the

Resid Conversion

advantages of hydrocracking is its ability to break down high-boiling aromatic stocks produced by catalytic cracking or coking. This is particularly desirable when maximum gasoline and minimum fuel oil must be made. However, it must not be forgotten that product distribution and quality vary considerably depending on the nature of the feedstock constituents as well as on the process. In modern refineries, hydrocracking is one of several process options that can be applied to the production of liquid fuels from the heavier feedstocks. The most important aspect of the modern refinery operation is the desired product slate, which dictates the matching of a process with any particular feedstock to overcome differences in feedstock composition. The reactions that are used to chemically define the processes (i.e., cracking and subsequent hydrogenation of the fragments, hydrogenation of unsaturated material, hydrodesulfurization, hydrodenitrogenation) may all occur. Hydrocracking a feedstock will, in all likelihood, be accompanied by hydrodesulfurization, thereby producing not only low-boiling products but also low-boiling products that are low in sulfur. Hydroprocesses offer direct desulfurization of resids and high conversion. These processes were not originally designed for resids and yet the evolution of the various processes and catalysts has seen their application to resids and residua. Application of a hydroprocess as a pretreatment process has some merit. Changing the character of the feedstock constituents can make for easier conversion in a later stage of the refining sequence. The advantages are: 1. The products require less finishing. 2. Sulfur is removed from the catalytic cracking feedstock, and corrosion is reduced in the cracking unit. 3. Carbon formation during cracking is reduced and higher conversions result. 4. The catalytic cracking quality of the gas oil fraction is improved. The downside to the direct application of hydroprocesses to the resids is always the cost of hydrogen and the short lifetime of the catalyst (which in effect is also a cost). There is the potentially wasteful use of hydrogen with hydrogen sinks within the feedstock whereupon hydrogen is used, but with little, if any, effect on the product character.

2661

catalysts, it is common to have a guard bed in front of the fixed bed. When insufficient demetallization activity occurs in the guard bed, the feed is switched to a second guard bed with fresh catalyst and the catalyst is replaced in the first guard bed. Thus, the fixed bed is protected from deposition of metals. To hydrogenate the largest macromolecules in the resid, the asphaltenes, some or all of the catalysts need to have pores 50–100 mm in diameter. Even with these precautions it is difficult to get longer than 1 yr run lengths on fixed-bed hydroconversion units with vacuum resid feeds and conversions to volatile liquids of 50% or more. This is due to catalyst deactivation with coke or to coke and sediment formation downstream of the reactor.

Ebullating Bed Processes The LC Finer and H-Oil units mechanically ebullate the catalyst so that it can be mixed and replaced onstream. The conversion is greatly dependent on the feed but conversions of vacuum resids to volatile liquids of the order of 70% is possible. Often these units are limited by the deposition of coke and sediment downstream of the reactor in hot and cold separators. Dispersed Catalysts Processes If one cannot diffuse the asphaltenes to the catalyst, why not diffuse the catalyst to the asphaltenes? Dispersed catalysts also can be continuously added in sufficiently low enough amounts (i.e., 100 ppm) to consider them throwaway catalysts with the carbonaceous by-product. However, economics usually dictate some form of catalyst recycle to minimize catalyst cost. Nevertheless, by designing the reactor to maximize the solubility of the converted asphaltenes, the conversion of vacuum resids to gas and volatile liquids can be above 95% with greater than 85% volatile liquids. However, the last 5–10% conversion may not be worth the cost of hydrogen and reactor volume to produce hydrocarbon gases and very aromatic liquids from this incremental conversion. The answer depends on the value and use of the unconverted carbonaceous liquid by-product.

CONCLUSIONS Fixed-Bed Processes Because metal removal is one of the fastest reactions and the metals accumulate in the pores of supported

Resid conversion is now in a significant transition period as the demand for transportation fuels increases. To satisfy the changing pattern of product demand, significant investments in resid conversion processes

R

2662

will be necessary and technologies are needed that will take resid conversion beyond current limits and, at the same time, reduce the amount of coke and other nonessential products. More options are now being sought to increase process efficiency in terms of the yields of the desired products. New processes for resid conversion probably will be used in conjunction with visbreaking and coking options with some degree of hydroprocessing as a primary conversion step. In addition, other processes such as asphalt coking technology, the Cherry-P, or the eureka process may replace or, more likely, augment the deasphalting units in many refineries. There remains room for improving coking and hydroconversion processes by reducing hydrocarbon gas formation, inhibiting the formation of polynuclear aromatic compounds not originally present in the resid, and separating an intermediate quality (low in cores) fraction before or during conversion. In addition, the challenge for hydroconversion is to take advantage of the nickel and vanadium in the resid to generate an in situ dispersed catalyst and to eliminate catalyst cost. Finally, resid catalytic cracking needs to

Resid Conversion

move to poorer quality and lower-cost feeds by making more tolerant catalysts.

BIBLIOGRAPHY Dolbear, G.E. Petroleum Chemistry and Refining; Speight, J.G., Ed.; Taylor & Francis: Washington, DC, 1998; 7 pp. Gray, M.R. Upgrading Petroleum Residues and Resids; Marcel Dekker Inc.: New York, 1994. Speight, J.G. The Chemistry and Technology of Petroleum, 3rd Ed.; Marcel Dekker Inc.: New York, 1999. Speight, J.G. The Desulfurization of Heavy Oils and Residua, 2nd Ed.; Marcel Dekker Inc.: New York, 1999. Speight, J.G.; Ozum, B. Petroleum Refining Processes; Marcel Dekker Inc.: New York, 2002; 538–587. Wiehe, I.A. A solvent-resid phase diagram for tracking resid conversion. Ind. Eng. Chem. Res. 1992, 31, 530. Wiehe, I.A.; Liang, K.S. Fuel Sci. Technol. Int. 1996, 14, 289.

Rheology of Cellulose Liquid Crystalline Polymers R Qizhou Dai Faculty of Foresty, Biomaterials Chemistry, The University of British Columbia, Vancouver, British Columbia, Canada

Richard Gilbert Department of Wood and Paper Science, North Carolina State University, Raleigh, North Carolina, U.S.A.

John F. Kadla Faculty of Foresty, Biomaterials Chemistry, The University of British Columbia, Vancouver, British Columbia, Canada

INTRODUCTION Cellulose and many of its derivatives form liquid crystalline phases in solutions and melts. Because of the chirality of the cellulose backbone, cellulosic liquid crystalline phases form chiral nematic structures. Chiral nematic mesophases possess a unique structure in which the alignment of molecular sheets is at a slight angle to one another resulting in a helicoidal supramolecular structure. It can be described by a pitch p (or its inverse, the twist p
1); p ¼ l0=n~, where l0 is the reflection wavelength and n~ is the mean refractive index of a sheet, and the corresponding handedness of the twist; right-handed helicoidal structures being assigned to a positive pitch (p > 0) and left-handed helicoidal structures to a negative pitch (p < 0). This type of supramolecular arrangement results in unique optical and physical properties. As a result, many lyotropic cellulosic systems have been developed with potential applications ranging from liquid crystalline displays to high modulus, high strength regenerated cellulose fibers. In recent years, systems that display a reversal of their helical twisting sense by solvent or chemical modification have been of particular interest. The viscosity of these chiral nematic mesophases is concentration-dependent and strongly influenced by shear rate. When the shear rate is high enough, the polymer chains will orient along the shear direction. The chiral nematic structure changes to a flow-aligned nematic-like phase and the dependence of viscosity on concentration decreases. Undoubtedly during the shear process, shear forces cause an orientation in the shear direction of the twisted macromolecules of the chiral nematic structure. A quasi-nematic phase is formed with some dispersed chiral nematic domains. Upon removal of the shear stress, the thermodynamically more stable chiral nematic phase will reform leading to a disruption in the macromolecular orientation Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120040351 Copyright # 2006 by Taylor & Francis. All rights reserved.

within the forming products, and as a result lower than expected modulus and strength. In this entry, we report on the transient rheological behavior of lyotropic cellulosic mesophases with varying pitch and handedness and discuss the relationship between the chiro-optical properties and relaxation behavior.

CELLULOSIC LIQUID CRYSTALS Cellulose is the most abundant natural polymer. Cellulose and its derivatives (cellulosics) have played an important role in developing and establishing the current concepts and industrial applications of polymer science. Current interests are based on their versatile properties, biodegradability, and status as a renewable resource (Fig. 1). Discovery of the formation of liquid crystalline solutions by cellulosics in the mid-1970s[1] has resulted in attempts to develop new cellulosics products with properties superior to those of conventional cellulosic. Following the first observation of mesophases formed in aqueous solutions of hydroxypropyl cellulose (HPC),[1] a variety of other cellulose derivatives have been reported to form liquid crystals.[2–7] Liquid crystalline solutions of cellulose and its derivatives[8–12] provide a potential route to high-modulus and hightenacity cellulosic fibers, films, and other highperformance products. Efforts to make high-performance fibers and films from cellulosic mesophases have been made.[12–17] For example, cellulose fibers produced from cellulosic mesophases show properties superior to those of commercially available fibers.[16,18,19] Although these fibers are superior to commercial products, their physical properties are lower than theoretically predicted. This is in part because of that the ordered structures 2663

2664

Rheology of Cellulose Liquid Crystalline Polymers

Fig. 1 Structural cellulose.

are not well preserved because of the orientation relaxation during the processing steps. Liquid crystalline (LC) solutions of cellulose derivatives form chiral nematic (cholesteric) phases.[6,20] Chiral nematic phases are formed when optically active molecules are incorporated into the nematic state. A fingerprint texture is generally observed under crossed polarizers for chiral nematic liquid crystals when the axis of the helicoidal structure is perpendicular to the incident light (Fig. 2). In general, chiral nematic polymer liquid crystals (LCP) cannot form monodomains in which the rodlike polymers have a spatially uniform orientation within the sample. Typically, because of the high density of orientational defects, the LCPs are textured, with a distribution of polymer orientation. Microscopically, the polymer chains have a preferred orientation with a relatively narrow distribution around the average orientation. Macroscopically, the variation in space of the orientation results in a domain structure. Defects and orientational variations give rise to the polydomain texture and the overall LCP sample may be randomly ordered (Fig. 3).

representation

of

Lyotropic liquid crystalline cellulose derivatives exhibit unique optical properties because of their helicoidal supramolecular structure.[21] The chiro-optical properties of the helicoidal structure can be described by a pitch p (or its inverse, the twist p
1); p ¼ l0=n~, where l0 is the reflection wavelength and n~ is the mean refractive index of a sheet, and the corresponding handedness of the twist; right-handed helicoidal structure being assigned to a positive pitch (p > 0) and left-handed helicoidal structures to a negative pitch (p < 0).[5,6,22] The nematic mesophase can be considered as a critical state of chiral nematic phase, in which the helicoidal structure has infinite pitch and no handedness. The supramolecular structure of chiral nematic mesophase indicates the effects of polymer–solvent and polymer–polymer interactions in the mesophase (Fig. 4). The effect of polymer–solvent interactions on the mesophase can be derived from the rigidity of the polymer chain, the critical concentration to form liquid crystalline phase, and relaxation studies. After shearing a rigid polymeric liquid crystal, a banded texture is formed in which the direction of the bands

Fig. 2 Fingerprint pattern for chiral nematic mesophase of 40% EC in m-cresol. (View this art in color at www.dekker.com.)

Rheology of Cellulose Liquid Crystalline Polymers

A

2665

B

R

Fig. 3 Hierarchy of the distribution of the orientation of the molecules in a polydomain LCP. (A) Microscopic and (B) mesoscopic.

is perpendicular to the shearing direction after removing the shear force.[23,24] The banded structure forms because of a periodic packing of rigid=semirigid polymer chains in a zigzag fashion, which is believed to be owing to the relaxation of the polymer orientation.[25] For lyotropic LC solutions the banded texture is a transient phenomenon. After a period of time, the polymer chains fully relax and the banded texture disappears. The bandwidth and the relaxation time of the bands depend on the rigidity and pitch of the polymer chain resulting from specific polymer–solvent interactions. In fact, at the same concentration ethyl cellulose (EC) chains are more rigid in dichloroacetic acid (DCA) than in acrylic acid or glacial acetic acid and the pitch of EC=DCA mesophase is much higher than that of the EC=acetic acid[26] or acrylic acid,[27] resulting in wider bands and longer relaxation times for EC=DCA mesophase (Fig. 5).[28] However, the rigidity of the polymer chain is not the only factor to determine the pitch of the mesophase. Guo and Gray[29] found that for acetylethylcellulose

Left-handed

p

(AEC) in chloroform, with the increase of degree of acetylation, the rigidity of the cellulose chain increases monotonically. However, the pitch of the mesophase first increases to infinity and then decreases with an inversion of the handedness of the chiral nematic phase, from left-handed to right-handed. As cellulosics are chiral, there must also be a chiral contribution to the interactions between the rods in the mesophase leading to the chiral nematic structure. Several theories have been proposed to account for the ‘‘twisting’’ force between chiral rodlike mesogens in the liquid crystalline phase. Goossens was the first to propose that the chiral nematic structure is the result of anisotropic dispersion energy between chiral mesogens.[30] Samulski and Samulski proposed that the macroscopic twist sense was dependent on the dielectric constant of the medium and the chirality of the constituent molecules.[31] They determined that the introduction of an asymmetric dispersion energy results in adjacent mesogens having a slight twist relative to one another, and that increasing temperature

Right-handed

Fig. 4 Helicoidal structure of chiral nematic mesophases.

2666

Rheology of Cellulose Liquid Crystalline Polymers

Fig. 5 Banded texture for 45% EC in m-cresol: (A) 1 min after shear and (B) 75 min after shear. (View this art in color at www. dekker.com.)

increases the frequency of molecular rotation, causing a decrease in the dispersion energy and an increase in the pitch. The observed handedness and pitch of the helicoidal structure is not only sensitive to temperature, but also depends on the concentration and structure of the components.[26,32–37] According to Lin-Liu, Shih, and Woo[38] the temperature dependence of pitch for chiral nematic polymers does not seem to follow any particular pattern. It is believed that as temperature is increased, specific interactions, e.g., hydrogen bonding, whether inter- or intramolecular or polymer–solvent interactions are destroyed. The polymer chains become more flexible and the side groups more easily relaxed, thereby changing the physical properties of the chiral nematic structure. Similarly, an increase in concentration leads to a decrease in pitch for most lyotropic cellulosic liquid crystals with the exception of cellulose tricarbanilate (CTC) in ethyl methyl ketone, 2-pentanone,[35] or triethylene glycol monoether[34] and the chlorophenyl urethane derivative in diethylene glycol monoether.[39] Numerous investigators have proposed theories for the temperature dependence of pitch.[40–44] However, none completely explain the experimental results obtained for the various cellulosic systems. Of those proposed, Osipov’s approach, which is based on a molecular statistical theory,[42] takes into account steric and chiral interactions in solution to predict the influence of temperature and solvent on the pitch and twist sense of cellulose derivatives.[43] Assuming the cellulosic chain adopts a ‘‘twisted belt’’ as opposed to a helix, and the persistence length, l, of the cellulose chains is much smaller than that of a rigid chain, the twisting power could be expressed by 2p=P ¼
r2 ðw
lkTÞ=2K22

ð1Þ

where w is related to the attractive interaction between chains, P the pitch, k the Boltzmann constant, K22 the

twist elastic constant, r the number density of rigid segments, r ¼ cL=l0, where c is the number density of the macromolecular chains, l0 the length of the segments, L the total length of the chain (L > l > l0), and the pseudo-scalar parameter l is determined by the steric repulsion between chains. The theory predicts that the handedness of cellulosic liquid crystalline solutions, designated by the sign of the pitch, depends not only on temperature (T ) and on steric repulsion of the chain (l), but also on an attractive interaction parameter, w, which depends on the nature of the solvent. The chiral forces are balanced when (w
lkT ) ¼ 0. In this compensated condition, the pitch of the mesophase should become infinite, and the mesophase resembles a normal nematic phase. Besides degree of substitution, nature of substituents, solvent, concentration, and temperature, other factors that change the polymer–solvent interactions can affect the pitch of lyotropic cellulosic mesophases. Doping inorganic salts[45,46] or small chiral molecules[47] into the lyotropic mesophase changes the polymer–solvent interactions. As the results, the pitch of the chiral nematic mesophase changes accordingly.

RHEOLOGY OF CELLULOSE LIQUID CRYSTALLINE SOLUTIONS The microstructure and polymer–solvent interactions of lyotropic cellulosic mesophases can be derived from rheological studies. The lyotropic LCP solution is a complicated system and a wide range of unusual rheological phenomena have been observed. Because of the wide range of technological applications and complexity of liquid crystal polymers, there is a strong need for further advance in the understanding of their rheological phenomena. Several theories have been introduced to predict the rheological

Rheology of Cellulose Liquid Crystalline Polymers

apparent viscosity curve,[62] oscillatory underdamped stress relaxation,[62] helix uncoiling,[63] and the higher apparent viscosity, as compared to nematic LCPs.[64] However, cellulosics are semiflexible polymers. Experimental results have shown that semiflexible HPC=H2O system[65] has different rheological behavior from that of the more rigid poly(g-benzyl-glutamate)=m-cresol system.[66] So far, no theory has been proposed to model the rheology of semiflexible chiral nematic LCPs. Following the first observation of mesophase formation of HPC=H2O, the rheology of this system has been widely investigated. Rheological studies of other cellulosic mesophases, e.g., cellulose,[67] ethylcellulose,[26,68,69] and cellulose tricarbanilate,[70] have also been reported.

Concentration Dependence of Viscosity Isotropic solutions exhibit a monotonic increase in shear viscosity with increasing concentration. The viscosity increases to a maximum when the isotropic to anisotropic transition is approached. Upon formation of the anisotropic phase, the viscosity begins to decrease, after which the viscosity increases strongly as the concentration continues to increase (Fig. 6). In the isotropic state, the hydrodynamic volume is large because of the random polymer orientation. This restricts the polymer diffusivity and causes an increase in viscosity. In the anisotropic phase, the aligned polymer leads to a small hydrodynamic volume and a decrease in viscosity as rotational diffusion is much easier with a net orientation. Zugenmaier found that the correlation of the viscosity maximum with the formation of the anisotropic phase was only valid when the shear rate was low

Viscosity

behavior of LCPs. The classical Leslie–Ericksen (LE) continuum theory uses a rigid rodlike director to describe the local molecular orientation.[48,49] The rheological behavior is the result of the interaction between the director field and applied flow field. Leslie–Ericksen theory is applicable when the flow field only affects the direction of molecular orientation without changing the degree of molecular alignment. This theory has been successful in discussing the flow of low molecular weight nematic liquid crystals. For LCP, it is applicable only at very low shear rates in defect-free samples.[50] Doi molecular theory adds a probability density function of molecular orientation to model rigid rodlike polymer molecules.[51] This model is capable of describing the local molecular orientation distribution and nonlinear viscoelastic phenomena. Doi theory successfully predicts director tumbling in the linear regime and two sign changes in the first normal stress difference,[52] as will be discussed later. However, because this theory assumes a uniform spatial structure, it is unable to describe textured LCPs. Larson and Doi introduced a mesoscopic polydomain model based on LE theory. This model includes a domain orientation distribution function and incorporates director tumbling, distortional elasticity, and texture size.[53] Larson–Doi model can qualitatively predict the steady flow behavior and transient behavior. However, discrepancies between the theoretical predictions and the experiments of model systems were observed, especially when the shear history includes rest periods.[50] This model is restricted to low shear rates without perturbing the molecular orientation distribution function in each domain.[52] The theory for LCP rheology is still evolving. Rey and Tsuji[54] proposed a complete tensor theory to take into account three major effects of LCP flow: 1) short range order elasticity; 2) long range order elasticity; and 3) viscous flow. This theory is based on the unification of the classical nonequilibrium theories. The complete theory can predict more complex phenomena of LCP behavior, e.g., banded texture, defect generation, and coarsening. Extensive and detailed reviews of LCP rheology theory were provided by Larson,[55,56] Burghardt,[52] Rey and Denn,[57] and Marrucci and Greco.[58] All the above theories are derived for rigid rod nematic liquid crystal systems. The rheological behavior of chiral nematic liquid crystals is more complex and less understood than that of nematic systems. Rey introduced a model based on rigid rod chiral nematic liquid crystals to describe permeation shear flow and small amplitude oscillatory shear flow.[59–61] The model can predict some common phenomena of chiral nematic liquid crystals, e.g., the three-region

2667

Isotropic

Anisotropic Concentration

Fig. 6 Viscosity as a function of concentration for lyotropic LCPs.

R

2668

Rheology of Cellulose Liquid Crystalline Polymers

(shear rate ! 0).[70] At high shear rate, the concentration at which the maximum in viscosity occurs decreases. When the shear rate is high enough to cause shear-induced orientation (pseudo-nematic phase), entanglements of the random distributed polymers are released. The viscosity maximum disappears and a steady increase of viscosity vs. concentration is observed. This indicates that the viscosity of the nematic or pseudo-nematic mesophase is less sensitive to concentration than that of the chiral nematic phase.[70]

Steady Flow Behavior

log η

Steady flow behavior is one of the most thoroughly studied rheological properties. Onogi and Asada[71] hypothesized the universal existence of three shear flow regimes to describe the viscosity of LCPs: a shear thinning regime at low shear rates (regime I), a Newtonian plateau at intermediate shear rates (regime II), and another shear thinning regime at high shear rates (regime III) (Fig. 7). The three-regime curve was observed in the anisotropic aqueous HPC solution[72] and HPC in acetic acid.[73] Asada proposed that some LCP systems do not exhibit all three regimes because not every regime lies in the accessible shear rate range. In those systems that exhibit all three regimes, regime I is generally believed to reflect a defect structure, or stacked polydomain texture. The texture is a supramolecular disorder of the nematic structure and is composed of nematic domains with little or no macroscopic orientation.[72,74] This region is characterized by distortional elasticity associated with spatial variation in the director (average local molecular orientation) field. Regime II reflects a ‘‘dispersed polydomain’’ structure and regime III is believed to be characterized by nonlinear effects of flow on the molecular orientation. In regimes I and II, the flow is not strong enough to affect the molecular orientation.

In regime III, the flow field is very strong and shear-induced molecular orientation becomes important. According to birefringence measurements for anisotropic HPC=H2O solutions[65] and HPC=m-cresol solutions,[75] the molecular orientation is a monotonically increasing function of the steady state shear rate. We found that the steady state flow of lyotropic AEC in acrylic acid (AA) showed three-regime curve, as shown in Fig. 8. Different AECs had different DA and the mesophases had different chiro-optical properties. The properties of different AECs are listed in Table 1. All the mesophases showed a typical threeregime curve. With the increase of shear rate, a shear-thinning is followed by a Newtonian plateau, and then followed by another shear-thinning region at high shear rate. The Newtonian plateau is in the shear rate range 1–5 S
1. We also found that with the increase of pitch, the flow curve becomes flat and the three regimes are not as distinct as the one with smaller pitch. Even though several cellulosic mesophase systems show a three-regime flow curve, in fact, it appears that all three regimes are not typically accessible for all LCPs including cellulosics. Whether the three-regime flow curve is indeed universal or not remains a subject of debate. The most striking phenomenon in the steady flow of LCPs is the occurrence of a negative first normal stress difference (N1)—two sign changes in N1 as a function of shear rate. In contrast, isotropic solutions only exhibit positive N1 at all shear rates. Negative N1 has been reported in HPC=H2O[76–78] and HPC=m-cresol systems.[79] The negative N1 is the result of the coupling of molecular tumbling under flow and the local molecular-orientation distribution.[76] At low shear rates, the director tumbles with the flow and N1 will be positive. At intermediate shear rates, nonlinear viscoelastic effects are important. The director tumbling competes with the steady director alignment along

regime I shear thinning

regime II Newtonian plateau •

log γ

regime III shear thinning Fig. 7 Three-regime steady state shear viscosity for LCPs.

Rheology of Cellulose Liquid Crystalline Polymers

2669

R

Fig. 8 Viscosity as a function of shear rate of AEC=AA solutions. (The properties of AEC samples are given in Table 1.)

the flow and the director oscillates about a steady value, N1 becomes negative. At very high shear rates, the director aligns along the flow and N1 is positive again. It is generally believed that the shear rate at which N1 is minimum halts director tumbling and aligns the molecules.[79] However, negative N1 behavior disappears at very high polymer concentration because of polymer–polymer friction interactions.[76] Relaxation Behavior The relaxation behavior, or the transient behavior of cellulosic liquid crystalline solutions upon the cessation of steady flow, is unique for LCPs. There are two kinds of relaxation. The bulk stresses relax quickly while the structures relax over a much longer time. Mewis and Moldenaers suggested that two levels of structures exist.[80] Stress relaxation reflects fast relaxation at the molecule level and is independent of the previous shear rate. Structure relaxation reflects the gradual change of the textures. This slow process is unique in

Table 1 Chiro-optical properties of lyotropic 50% AEC= AA solutions with different degrees of acetylation AEC sample

Sample description

Degree of acetylation

AEC-1

Low DA

0.13

AEC-2

Medium DA

0.34

AEC-3

AEC=EC mixture

0.34

AEC-4

High DA

0.50

Pitch
360 nm


7.4 mm

þ1.2 mm

þ370 nm

that its time scale is inversely proportional to the previous shear rate. When the strain recovery (or recoil) after cessation of steady state flow is plotted against time multiplied by the previous shear rate, the curves superimpose one another.[50] It is generally believed that under high flow rate, the chiral nematic mesophase aligns along the flow direction and uncoils to form a nematic structure. Upon cessation of flow, the chiral nematic phase will reform and the molecular orientation will decrease.[81] By using evolution of the dynamic moduli[82] and birefringence[52] as a function of time, the structural change can be investigated. For lyotropic HPC=H2O solutions, upon flow cessation, molecular orientation decreases to a globally isotropic condition at all rates[52] and the dynamic moduli increase to a maximum.[82] However, the final relaxed state depends on the shear history. After high shear rates, the solutions evolve toward an ‘‘equilibrium’’ state with a high modulus, while after low shear rates, the solutions relax to the ‘‘equilibrium’’ state with a low modulus.[65] The low-modulus state is ordered and evolves out of a state that has no macroscopic order upon cessation of flow. The high-modulus state is much less ordered, although it evolves from a rather well flow-aligned state upon the cessation of flow.[83] At high shear rates the chiral nematic structure changes to a flow-induced nematic phase. However, the shear-oriented phase is easy to disrupt after removing the shear force. This is because of the driving force for the liquid crystalline solution to form the more thermodynamically stable chiral nematic structure.[82] Relaxation in a pseudo-nematic lyotropic

2670

solution, after being oriented along the high shear rate direction, will persist for a long period of time, as the driving force to reform helicoidal structure may be limited. So far, there are a few reports that discuss the relationship between the relaxation behavior and chiro-optical properties of the mesophases. In the case of a lyotropic polypeptide system, poly(g-benzylglutamate) (PBG) solutions containing a single optical isomer of either poly(g-benzyl-l-glutamate) (PBLG) or poly(g-benzyl-d-glutamate) (PBDG) or in the case of a racemic mixture of these two, there is no support for the idea that the chiro-optical properties of the chiral nematic PBG=m-cresol solution has any significant effects on the relaxation behavior.[52] This may be because of the fact that the mesophases of both optical isomers and the racemic mixture have a large pitch and are nematic-like. In our recent study, relaxation behaviors upon shear force removal were investigated for lyotropic solutions of AEC in AA with different chiro-optical properties.[84] Ethylcellulose and AEC can form chiral nematic liquid crystalline solutions in many solvents and were found to change handedness as well as pitch with the change of degree of acetylation. In acrylic acid, EC forms a left-handed mesophase, while fully acetylated AEC forms a right-handed mesophase.[85] At a certain critical degree of acetylation (DA ) the pitch becomes infinite and an untwisted, nematic-like structure forms. When the degree of acetylation of AEC increases across DA , the handedness of cholesteric liquid crystalline solutions changes from lefthandedness to right-handedness. The pitch of the lyotropic solution of EC or fully acetylated AEC in AA is about one-tenth that of PBG solutions. Therefore, the effect of chiro-optical properties on relaxation behavior is expected to be more pronounced in AEC system than in PBG system. The values of pitch for lyotropic AEC solutions in AA (50% w=w) are given in Table 1, where AEC-2 is a pure AEC with medium DA and AEC-3 is the mixture of EC and fully acetated AEC, which has the same average DA as AEC-2. However, in the liquid crystalline solutions, even at the same concentration, AEC-2 and AEC-3 have different pitch and handedness. This phenomenon was also observed in the lyotropic AEC=chloroform system.[86] The difference in chiro-optical properties may come from the complex interactions of multiple chiral centers present in each repeating unit of the cellulose chain, not from simple racemic mixtures as in the PBG system. At the same concentration, the viscosities of AEC solutions are different such that the larger the pitch, the smaller the viscosity.[29] To eliminate the influence of viscosity of LCP solutions on the relaxation behaviors, the concentrations of AEC solutions were adjusted so that all solutions had similar viscosities

Rheology of Cellulose Liquid Crystalline Polymers

under shear. As shown in Fig. 8, when the shear rate is between 0.5 and 6 s
1, the four AEC solutions had the same viscosity.

Stress relaxation Stress relaxation reflects fast relaxation at the molecular level and is independent of the previous shear rate. In the tumbling region (low pre-shear rate region), the final part of the stress relaxation curve is determined by the textural contribution because its time of evolution scales with the previous shear rate. This reflects the decrease in domain size with shear rate, which controls the scaling relations. In the flow align region, the textural contribution disappears and fast relaxation becomes nearly independent of shear rate.[87] As shown in Fig. 9, the stress relaxation curves of all AEC=AA solutions collapse into one curve when the solutions were presheared with the same rate. Because the stress relaxation is at the molecular level and the chiro-optical properties reflect the suprastructural level, it is expected that the lyotropic solutions with different chiro-optical properties have the same stress relaxation behavior in both the tumbling and flow-align regions.

Dynamic modulus evolution upon cessation of flow The development of storage and loss moduli after flow cessation is a useful tool to analyze structural relaxations on LCPs. Upon flow cessation, the flow-induced orientation is lost. The evolution of the moduli with time is because of the reformation of a chiral nematic phase that had become nematic under flow. In Fig. 10, the relaxation in complex modulus with time for various AEC samples is shown. As seen in Fig. 10, the larger the pitch, the slower the rate the modulus evolution. The modulus of the right-handed mesophase (AEC-4) developed faster than that of the left-handed mesophase (AEC-1) with similar pitch. The difference in relaxation behavior of the AEC solutions may be because of the smaller driving force to reform helicoidal structures of chiral nematic phases from flow-induced nematic mesophases for the mesophases with larger pitch (nematic-like). The reformation of chiral nematic phase from nematic phase sometimes goes through the transient state of band texture. The formation of band texture is also a relaxation phenomenon. For HPC=H2O solutions, the band texture is only observed when the molecules have been well orientated in the shear direction.[76] A critical lower shear rate limit exists because of the stability of chiral nematic textures.

Rheology of Cellulose Liquid Crystalline Polymers

2671

R

Fig. 9 Reduced shear stress s (s ¼ [s(t)
sfin]=(sin - sfin)) vs. time for all AEC=AA solutions.

An upper shear rate limit also exists above which no bands are formed.[88] The upper critical shear rate is associated with the flow-aligning region of molecular dynamics. Band texture formation is driven by the release of energy, which has been stored in the mesophase during shear.[76] The evolution of the band texture depends on the previous shear rate applied to mesophase. The rate of evolution of the band texture

will increase, remain constant, or decrease according to whether the mesophase is sheared at low, intermediate, or high rates.[89] When the band texture appears during recoil, the appearance of the band texture stops the strain recovery process until the band texture disappears, and then recovery continues. Therefore, the presence of the band texture during recoil enhances the strain recovery.[90]

Fig. 10 Complex modulus evolution of AEC solutions vs. time.

2672

Rheology of Cellulose Liquid Crystalline Polymers

CONCLUSIONS Over the last three decades extensive information pertaining to the properties and behaviors of liquid crystalline cellulose derivatives has been obtained. However, the fact remains that fibers and films produced from cellulosic mesophases are inferior to those predicted by theory. A number of questions still remain surrounding these unique polymer systems. For example, what role, if any, do the pitch and handedness of the helicoidal structure play in determining the transient rheological behavior of lyotropic cellulosic mesophases. A correlation between the chiro-optical properties and relaxation behavior should be established. From our recent study, the relaxation behavior of lyotropic liquid crystalline solutions of (acetyl) (ethyl) cellulose was affected by the chiro-optical properties of the mesophase. In stress relaxation, handedness and pitch had no obvious effect. However, the development of complex modulus upon flow cessation was affected by pitch and handedness; the larger the pitch, the slower the rate the moduli evolved. Interestingly, however, the moduli of the right-handed mesophase developed faster than that of the left-handed one. Under flow, the cellulosic mesophases become nematic (shear aligned). Therefore, in samples with large pitch (nematic-like), the driving force to reform the helicoidal structures of chiral nematic phases upon removal of shear will be smaller than those with smaller pitch.

7.

8.

9.

10.

11.

12.

13.

REFERENCES 1. Werbowyj, R.S.; Gray, D.G. Liquid crystalline structure in aqueous hydroxypropyl cellulose solutions. Molec. Crystals Liquid Crystals 1976, 34 (4), 97–103. 2. Patton, P.; Gilbert, R.D. Anisotropic solutions of methylol cellulose. Polym. Preprints (American Chemical Society, Division of Polymer Chemistry) 1983, 24 (2), 266–267. 3. Bhadani, S.N.; Gray, D.G. Cellulose-based liquidcrystalline polymers; esters of (hydroxypropyl) cellulose. Molec. Crystals Liquid Crystals 1983, 99 (1–4), 29–38. 4. Laivins, G.V.; Gray, D.G. Liquid crystalline phase transition of a semiflexible polymer: acetoxypropyl cellulose. Macromolecules 1985, 18 (9), 1753–1759. 5. Gilbert, R.D. Cellulose and cellulose derivatives as liquid crystals. ACS Symp. Ser. 1990, 433, 259–272. 6. Gray, D.G.; Harkness, B.R. Chiral nematic mesophase of lyotropic and thermotropic cellulose derivatives. In Liquid Crystalline and

14.

15. 16.

17.

18. 19. 20.

Mesomorphic Polymers; Shibev, V.P., Lam, L., Eds.; Springer-Verlag: New York, 1993; 298–323. Sixou, P.; Bosch, A.T. Lyotropic liquid crystal solutions of cellulose derivatives. In Cellulose: Structure, Modification and Hydrolysis; Young, R.A., Rowell, R.M., Eds.; Wiley: New York, 1986; 205–219. Chanzy, H.; Chaunis, S.; Monzie, P. Oriented cellulose films and fibers from a mesophase system. J. Polym. Sci. Polym. Phys. Ed. 1980, 18 (5), 1137–1144. Patel, D.L.; Gilbert, R.D. Mesomorphic solutions of cellulose triacetate in halogenated organic acids and mixtures of trifluoroacetic acid and dichloromethane. J. Polym. Sci. Polym. Phys. Ed. 1981, 19 (9), 1449–1460. Patel, D.L.; Gilbert, R.D. Lyotropic mesomorphic formation of cellulose in trifluoroacetic acid=chlorinated-alkane solvent mixtures at room temperature. J. Polym. Sci. Polym. Phys. Ed. 1981, 19 (8), 1231–1236. Chen, Y.S.; Cuculo, J.A. Lyotropic mesophase of cellulose in ammonia=ammonium thiocyanate solution. J. Polym. Sci. A: Polym. Chem. 1986, 24 (9), 2075–2084. Conio, G.; Corazza, P.; Bianchi, E.; Tealdi, A.; Ciferri, A. Phase equilibria of cellulose in N,Ndimethylacetamide=lithium chloride solutions. J. Polym. Sci. Polym. Lett. 1984, 22 (5), 273–277. Yang, K.S.; Theil, M.H.; Cuculo, J.A. Lyotropic mesophases of cellulose in the ammonia– ammonium thiocyanate solvent system. Effects of system composition on phase types. ACS Symp. Ser. 1989, 384, 156–183. Yang, K.S.; Cuculo, J.A. Formation and characterization of the fibers and films from mesophase solutions of cellulose in ammonia=ammonium thiocyanate solvent. Polymer 1992, 33 (1), 170–174. Gilbert, R.D. Making strong cellulose fibers. Chemtech 1995, 25 (11), 44–48. Gilbert, R.D.; Hu, X.; Fornes, R.E. Preparation of high-strength=high-modulus regenerated cellulose fibers from lyotropic mesophases. J. Appl. Polym. Sci. 1995, 58 (8), 1365–1370. Bianchi, E.; Ciferri, A.; Conio, G.; Tealdi, A. Fiber formation from liquid-crystalline precursors. II. Cellulose in N,N-dimethylacetamide-lithium chloride. J. Polym. Sci. B: Polym. Phys. 1989, 27 (7), 1477–1484. O’Brien, J.P. Cellulosic Fibers from Anisotropic Solutions. US Patent 4,464,323, 1984. O’Brien, J.P. High Strength Cellulosic Fibers. US Patent 4,501,886, 1986. Gilbert, R.D. Polymeric Materials Encyclopedia; Salamone, J.C., Ed.; CRC Press: Boca Raton, 1996.

Rheology of Cellulose Liquid Crystalline Polymers

21. Guo, J.X.; Gray, D.G. Lyotropic cellulosics liquid crystals. In Cellulose Polymers, Blends and Composites; Gilbert, R.D., Ed.; Hanser: New York, 1994; 25–46. 22. Zugenmaier, P. Structural investigation on some cellulose derivatives in the crystalline and liquid crystalline state. In Cellulose: Structure, Modification, and Hydrolysis; Young, R.A., Rowell, R.M., Eds.; Wiley: New York, 1986; 221–245. 23. Suto, S.; Tateyama, S. Transient shear response of liquid crystal-forming hydroxypropyl cellulose solution in dimethylacetamide. I. Stress growth and relaxation behavior. J. Appl. Polym. Sci. 1994, 53 (2), 161–168. 24. Suto, S.; Kohmoto, K.; Abe, A. Transient shear response of liquid crystal-forming hydroxypropyl cellulose solution in dimethylacetamide. II. Correlation between band formation and stress relaxation. J. Appl. Polym. Sci. 1994, 53 (2), 169–178. 25. Nishio, Y.; Yamane, Y.; Takahashi, T. Morphological studies of liquid-crystalline cellulose derivatives. II. Hydroxypropyl cellulose films prepared from liquid-crystalline aqueous solutions. J. Polym. Sci. Polym. Phys. Ed. 1985, 23 (5), 1053–1064. 26. Zugenmaier, P.; Haurand, P. Structural and rheological investigations on the lyotropic, liquid-crystalline system: O-ethylcellulose-glacial acetic acid-dichloroacetic acid. Carbohydr. Res. 1987, 160, 369–380. 27. Shimamoto, S.; Gray, D.G. A method to preserve the chiral nematic order of lyotropic ethylcellulose and (acetyl)(ethyl)cellulose mesophases in solid films. Chem. Mater. 1998, 10 (6), 1720–1726. 28. Zhao, C.T.; Zhang, G.L.; Cai, B.L.; Xu, M. Solvent composition dependence of band morphology in sheared lyotropic ethyl cellulose liquid crystals. Macromol. Chem. Phys. 1998, 199 (8), 1485–1488. 29. Guo, J.X.; Gray, D.G. Effect of degree of acetylation and solvent on the chiroptical properties of lyotropic (acetyl)(ethyl) cellulose solutions. J. Polym. Sci. B: Polym. Phys. 1994, 32 (15), 2529–2537. 30. Goossens, W.J. Molecular theory of cholesteric phase and of twisting power of optically active molecules in a nematic liquid crystal. Molec. Crystals Liquid Crystals 1971, 12 (3), 237. 31. Samulski, T.V.; Samulski, E.T. van der Waals– Lifshitz forces in lyotropic polypeptide liquid crystals. J. Chem. Phys. 1977, 67 (2), 824–830. 32. Siekmeyer, M.; Steinmeier, H.; Zugenmaier, P. Supermolecular liquid-crystalline structures from highly concentrated cellulose derivative solutions.

2673

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

44.

Angew. Makromol. Chem. 1989, 166=167, 131–138. Siekmeyer, M.; Zugenmaier, P. Investigations of molar mass dependence of the lyotropic liquidcrystalline system: cellulose tricarbanilate= diethylene glycol monoethyl ether. Makromol. Chem. Rapid Commun. 1987, 8 (10), 511–517. Siekmeyer, M.; Zugenmaier, P. Solvent dependence of lyotropic liquid-crystalline phases of cellulose tricarbanilate. Makromol. Chem. 1990, 191 (5), 1177–1196. Vogt, U.; Zugenmaier, P. Structural models for some liquid crystalline cellulose derivatives. Berichte der Bunsen-Gesellschaft 1985, 89 (11), 1217–1224. Guo, J.X.; Gray, D.G. Preparation, characterization, and mesophase formation of esters of ethyl cellulose and methyl cellulose. J. Polym. Sci. A: Polym. Chem. 1994, 32 (5), 889–896. Harkness, B.R.; Gray, D.G. Left- and righthanded chiral nematic mesophase of (trityl) (alkyl)cellulose derivatives. Can. J. Chem. 1990, 68 (7), 1135–1139. Lin-Liu, Y.R.; Shih, Y.M.; Woo, C.W. Molecular theory of cholesteric liquid crystals and cholesteric mixtures. Phys. Rev. A: Atomic, Mol. Optical Phys. 1977, 15 (6), 2550–2557. Siekmeyer, M.; Steinmeier, H.; Zugenmaier, P. Structural investigations and phase behavior of a ternary lyotropic liquid-crystalline cellulosic system: cellulose tricarbanilate=3-chlorophenylurethane of cellulose=triethylene glycol monomethyl ether. Makromol. Chem. 1989, 190 (5), 1037–1045. Kimura, H.; Nakano, H. Orientational phase transition in the system of flexible molecules. J. Phys. Soc. Jpn. 1979, 46 (6), 1695–1700. Kimura, H.; Hosino, M.; Nakano, H. Statistical theory of cholesteric ordering in hard-rod fluids and liquid crystalline properties of polypeptide solutions. J. Phys. Soc. Jpn. 1982, 51 (5), 1584– 1590. Osipov, M.A. Molecular theory of solvent effect on cholesteric ordering in lyotropic polypeptide liquid crystals. Chem. Phys. 1985, 96 (2), 259–270. Osipov, M.A. Theory for cholesteric ordering in lyotropic liquid crystals. Nuovo Cimento della Societa Italiana di Fisica, D. Condensed Matter, Atomic. Molec. Chem. Phys. Fluids, Plasmas, Biophys. 1988, 10D (11), 1249–1262. Osipov, M.A. Molecular theory of cholesteric polymers. In Liquid Crystalline and Mesophoric Polymers; Shibev, V.P., Lam, L., Eds.; SpringerVerlag: New York, 1994; 1–25.

R

2674

45. Chiba, R.; Nishio, Y.; Miyashita, Y. Electrooptical behavior of liquid-crystalline (hydroxypropyl)cellulose=inorganic salt aqueous solutions. Macromolecules 2003, 36 (5), 1706–1712. 46. Nishio, Y.; Chiba, R.; Miyashita, Y.; Oshima, K.; Miyajima, T.; Kimura, N.; Suzuki, H. Salt addition effects on mesophase structure and optical properties of aqueous hydroxypropyl cellulose solutions. Polym. J. 2002, 34 (3), 149–157. 47. Shiau, C.C.; Labes, M.M. Control of the pitch of synthetic polypeptide lyotropic phases utilizing a chiral solvent. Molec. Crystals Liquid Crystals 1985, 124 (1–4), 125–130. 48. Ericksen, J.L. Equilibrium theories of crystals. In Advances in Liquid Crystals 2; Brown, G.H., Ed.; Academic Press: London, 1976; 233–298. 49. Leslie, F.M. Theory of flow phenomena in liquid crystals. In Advances in Liquid Crystals 4; Brown, G.H., Ed.; Academic Press: London, 1979; 1–81. 50. Vermant, J.M.; Moldenaers, P.; Mewis, J. An evaluation of the Larson–Doi model for liquid crystalline polymers using recoil. Rheol. Acta 1999, 38 (6), 537–547. 51. Doi, M.; Edwards, S.F. The Theory of Polymer Dynamics; Oxford University Press: London, 1986. 52. Burghardt, W.R. Molecular orientation and rheology in sheared lyotropic liquid crystalline polymers. Macromol. Chem. Phys. 1998, 199 (4), 471–488. 53. Larson, R.G.; Doi, M. Mesoscopic domain theory for textured liquid-crystalline polymers. J. Rheol. 1991, 35 (4), 539–563. 54. Rey, A.D.; Tsuji, T. Recent advances in theoretical liquid crystal rheology. Macromol. Theory Simul. 1998, 7 (6), 623–639. 55. Larson, R.G. Constitutive Equations for Polymer Melts and Solutions; Boston: Butterworths, 1988. 56. Larson, R.G. The Structure and Rheology of Complex Fluids; Oxford University Press: New York, 1999. 57. Rey, A.D.; Denn, M.M. Dynamic phenomena in liquid-crystalline materials. Annu. Rev. Fluid Mech. 2002, 34, 233–266. 58. Marrucci, G.; Greco, F. Flow behavior of liquid crystalline polymers. Adv. Chem. Phys. 1993, 86, 331–303. 59. Rey, A.D. Generalized cholesteric permeation flows. Phys. Rev. E 2002, 65 (2), No. 022701. 60. Rey, A.D. Theory of linear viscoelasticity of cholesteric liquid crystals. J. Rheol. 2000, 44 (4), 855–869. 61. Rey, A.D. Theory of linear viscoelasticity of chiral liquid crystals. Rheol. Acta 1996, 35 (5), 400–409.

Rheology of Cellulose Liquid Crystalline Polymers

62. Rey, A.D. Structural transformations and viscoelastic response of sheared fingerprint cholesteric textures. J. Non-Newtonian Fluid Mech. 1996, 64 (2–3), 207–227. 63. Rey, A.D. Flow alignment in the helix uncoiling of sheared cholesteric liquid crystals. Phys. Rev. E 1996, 53 (4), 4198–4201. 64. Rey, A.D. Simple shear and small amplitude oscillatory rectilinear shear permeation flows of cholesteric liquid crystals. J. Rheol. 2002, 46 (1), 225–240. 65. Hongladarom, K.; Secakusuma, V.; Burghardt, W.R. Relation between molecular orientation and rheology in lyotropic hydroxypropyl cellulose solutions. J. Rheol. 1994, 38 (5), 505–523. 66. Hongladarom, K.; Burghardt, W.R. Molecular alignment of polymer liquid crystals in shear flows. 2. Transient flow behavior in poly(benzyl glutamate) solutions. Macromolecules 1993, 26 (4), 785–794. 67. Navard, P.; Haudin, J.M. Rheological behavior of isotropic and lyotropic solutions of cellulose. In Cellulose: Structure, Modification, and Hydrolysis; Young, R.A., Rowell, R.M., Eds.; Wiley: New York, 1985; 247–261. 68. Santamaria, A.; Lizaso, M.I.; Munoz, M.E. Rheology of ethyl cellulose solutions. Macromol. Symp. 1997, 114 (Polymer–Solvent Complexes), 109–119. 69. Lizaso, I.; Munoz, M.E.; Santamaria, A. Transient rheological behavior of lyotropic solutions of ethyl cellulose in m-cresol. Rheol. Acta 1999, 38 (2), 108–116. 70. Zugenmaier, P. Polymer solvent interaction in lyotropic liquid crystalline cellulose derivative systems. In Cellulose Polymers, Blends and Composites; Gilbert, R.D., Ed.; Hanser: New York, 1994; 71–94. 71. Onogi, S.; Asada, T. Rheology and rheooptics of polymer liquid crystals. In Rheology, Vol. 1: Principles; Astarita, G., Marrucci, G., Nicolais, L., Eds.; Plenum: New York, 1980; 127–147. 72. Walker, L.; Wagner, N. Rheology of region I flow in a lyotropic liquid-crystal polymer: the effects of defect texture. J. Rheol. 1994, 38 (5), 1525–1547. 73. Metzner, A.B.; Prilutski, G.M. Rheological properties of polymeric liquid crystals. J. Rheol. 1986, 30 (3), 661–691. 74. Ugaz, V.M.; Cinader, D.K.; Burghardt, W.R. Origins of region I shear thinning in model lyotropic liquid crystalline polymers. Macromolecules 1997, 30 (5), 1527–1530. 75. Caputo, F.E.; Burghardt, W.R. Real-time 1-2 plane SAXS measurements of molecular orienta-

Rheology of Cellulose Liquid Crystalline Polymers

76.

77.

78.

79.

80.

81.

82.

83.

tion in sheared liquid crystalline polymers. Macromolecules 2001, 34 (19), 6684–6694. Ernst, B.; Navard, P. Band textures in mesomorphic (hydroxypropyl) cellulose solutions. Macromolecules 1989, 22 (3), 1419–1422. Grizzuti, N.; Cavella, S.; Cicarelli, P. Transient and steady-state rheology of liquid crystalline hydroxypropylcellulose solution. J. Rheol. 1990, 34 (8), 1293–1310. Baek, S.G.; Magda, J.J.; Larson, R.G.; Hudson, S.D. Rheological differences among liquidcrystalline polymers. II. Disappearance of negative N1 in densely packed lyotropes and thermotropes. J. Rheol. 1994, 38 (5), 1473–1503. Baek, S.G.; Magda, J.J.; Cementwala, S. Normal stress difference in liquid-crystalline hydroxypropylcellulose solutions. J. Rheol. 1993, 37 (5), 935–945. Mewis, J.; Moldenaers, P. Transient rheological behavior of a lyotropic polymeric liquid crystal. Molec. Crystals Liquid Crystals 1987, 153 (A), 291–300. Asada, T.; Toda, K.; Onogi, S. Deformation and structural re-formation of lyotropic cholesteric liquid crystal of hydroxylpropeycellulose þ water system. Molec. Crystals Liquid Crystals 1981, 68, 231–246. Grizzuti, N.; Moldenaers, P.; Mortier, M.; Mewis, J. On the time-dependency of the flow-induced dynamic moduli of a liquid-crystalline hydroxypropyl cellulose solution. Rheol. Acta 1993, 32 (3), 218–226. Godinho, M.H.; van der Klink, J.J.; Martins, A.F. Shear-history dependent ‘‘equilibrium’’ states of liquid-crystalline hydroxypropylcellulose solutions

2675

84.

85.

86.

87.

88.

89.

90.

detected by rheo-nuclear magnetic resonance. J. Phys.: Condensed Matter 2003, 15 (32), 5461–5468. Dai, Q.G.; Gilbert, R.D.; Khan, S.A.; Kadla, J.F. Relaxation behaviors of lyotropic (acetyl)(ethyl) cellulose=acrylic acid solutions with different chiro-optical properties. Polym. Preprints (ACS, Div Polym. Chem.) 2004, 45 (1), 824–825. Shimamoto, S.; Uraki, Y.; Sano, Y. Optical properties and photopolymerization of liquid crystalline (acetyl) (ethyl) cellulose=acrylic acid system. Cellulose 2000, 7 (4), 347–358. Cowie, J.M.G.; Rodden, G.I. Blending as a method of tuning reflection wavelength and helical twisting sense in films and composites of liquid crystalline cellulose derivatives. Polymer 2002, 43 (12), 3415–3419. Mewis, J.M.; Vermant, J.; Moldenaers, P. Experimental evidence for the existence of a wagging regime in polymeric liquid crystals. Macromolecules 1997, 30 (5), 1323–1328. Vermant, J.M.; Mewis, J.; Picken, S.J. Band formation upon cessation of flow in liquidcrystalline polymers. J. Rheol. 1994, 38 (5), 1571–1589. Harrison, P.; Navard, P.; Cidade, M.T. Investigation of the band texture occurring in acetoxypropyl cellulose thermotropic liquid crystalline polymer using rheo-optical, rheological, and light scattering techniques. Rheol. Acta 1999, 38 (6), 594–605. Riti, J.B.; Navard, P. Textures during recoil of anisotropic hydroxypropyl cellulose solutions. J. Rheol. 1998, 42 (2), 225–237.

R

Rotational Molding of Polymers R Ce´line T. Bellehumeur Department of Chemical and Petroleum Engineering, University of Calgary, Calgary, Alberta, Canada

INTRODUCTION

ROTATIONAL MOLDING PROCESS

Rotational molding, also known as rotomolding or rotocasting, is used to fabricate hollow plastic parts such as tanks used for chemical, agricultural, and automotive applications, toys, boat shells, and outdoor furniture. This process offers advantages over other processes in the fabrication of extremely large containers and of parts with intricate contours and free of weld lines. It is economically suited for short-run production and very large hollow articles, or for parts that have special constraints, which cannot be molded using other techniques. On the other hand, there are several disadvantages related to this process. The cycle time is long and typically ranges from 10 to 40 min, depending on the product size and molding conditions. The material cost is usually high due to the need for pulverization, and the choice of resins used in rotational molding is quite limited with polyolefins dominating the market. The process remains labor intensive, although some progress has been seen over the years in its automation. The rotational molding process has caught the attention of several major resin suppliers as well as mold and machine suppliers. The last few decades have also seen growing interest among researchers and designers, leading to significant technological advances through process modification, process control, part design, and the development of new materials. In this entry, we focus on fundamental phenomena that control the quality of the end product from a processing and material perspective. The information will provide the reader with a better understanding of the process, which is necessary for the formulation of successful strategies in the development and processing of new resins for rotational molding applications. First, a detailed description of the rotational molding process is presented. This is followed by a discussion about particular features of the process, namely, melt densification and melt solidification phenomena. Recent technological advances in the development of resins new to rotational molding are also reviewed and presented in the last section of this entry.

Process Description

Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120007948 Copyright # 2006 by Taylor & Francis. All rights reserved.

The concept of rotational molding is illustrated in Fig. 1. The polymer is first loaded into a mold, which is then heated and rotated about its two primary axes (Figs. 1A and 1B). During the heating process, the tumbling powder gradually melts and sticks to the mold surface. Heating is continued after the powder has melted until complete densification is achieved. The mold is then cooled and the molded part is removed once it has reached a temperature safe for manual handling (Figs. 1C and 1D). Unlike most other polymer processes, rotational molding does not utilize pressure to force the melt into shape but relies primarily on the gradual deposition and adhesion of the polymer onto the mold. The mold rotation speed is relatively low (4–20 rpm) and the uniform coverage of the mold surface is ensured by the biaxial motion of the mold. The heating and cooling of the mold rely mostly on convective transport of energy. The most common method of heating is by means of gas combustion, while cooling relies on exposing the mold to forced air flow, water mist, water spray, or a combination of these methods. In the rotational molding process, the resin is used mostly in powder form, the majority of which is produced through grinding between rotating metal plates. The quality of the powder depends on the pulverization process variables, such as the gap size between the plates, the temperature and blade conditions. Past studies have shown that the powder characteristics, namely, the shape, size, and size distribution, influence both the rotational molding cycle and the final product performance.[1,2] The quality of the powder affects heat transfer, the coverage of the mold cavity, as well as the initial size of bubbles during the melt deposition process. Simple screening procedures for the quality assessment of polymeric powder for rotational molding include particle size distribution, dry flow index, and bulk density. Typically, 35-mesh powder with a particle size distribution ranging from 100 to 500 mm is used in rotational molding as it provides a good balance between pulverization cost, powder flow characteristics, and powder packing

2677

2678

Rotational Molding of Polymers

Fig. 1 The rotational molding process.

density. Both the flow index, which quantifies the ease of flow, and the bulk density are dependent on particle shape and size distribution. Devices have been commercialized that allow for monitoring the temperature profile of the mold and the mold cavity during the molding process.[3] Such capability has led to important advances in the development of control protocols and process design. During the rotational molding cycle, all polymers exhibit a characteristic mold inside air temperature profile (see Fig. 2). Changes in the temperature profile are characteristic of key transitions in the processing of polymers. In Fig. 2, the region between points A and B represents the induction stage, where the solid polymer particles tumble freely in the mold and are heated primarily through conduction. Point B corresponds to the onset of powder adherence onto the mold surface and coalescence between individual particles, while point C marks the end of the melting phase. During the stage between B and C, semicrystalline polymers will undergo a melting transition that results in a reduction in the heat transfer to the mold inside air cavity and thus a reduction in the slope of the temperature profile. Such a change in the slope of the temperature profile is not as significant for amorphous polymers. The insulating effect of the deposited layer of polymer on the mold surface also causes a reduction in the heat transfer to the inner air. As the polymer particles adhere to the mold surface, pockets of air are trapped within the melt and form bubbles. The region from C to D corresponds to the fusion stage, where the bubbles undergo dissolution. Point D is often referred to as the peak internal air temperature (PIAT). The region between D and E characterizes the melt cooling stage. Point E indicates the residence time at which the melt undergoes recrystallization.

Molding Cycle Time The rotational molding process is inherently transient, and the selection of optimal molding conditions will

depend on several factors such as mold size, shape, and material, charge of the polymer, polymer thermal and rheological properties, convection conditions inside the oven, and cooling conditions. The measurement of the mold inside air temperature has been established as one of the means to control the molding cycle. Correlations between the PIAT and the mechanical properties of the molded part were first recognized by Crawford and Nugent.[4] Over the years, it has become a standard procedure to use the inside air temperature of the mold as an indicator in the determination of the time required for the completion of the heating cycle (melt densification) prior to the occurrence of polymer degradation upon exposure to high temperatures. One important challenge in rotational molding is maximizing the mechanical properties of the end product while minimizing the molding cycle time. In many rotational molding applications, stiffness, dimensional stability, chemical resistance, and impact properties determine the performance of the final product. Processability encompasses the ability of a given resin to produce a final part with adequate properties when processed under a wide range of conditions (e.g., oven temperature, heating time, cooling conditions). The coalescence and melt solidification phenomena play a major role in both the molding cycle and the properties of the final parts. Resins with poor coalescence behavior require a longer heating time in the molding cycle, or the result will be the production of parts with large and numerous bubbles, which negatively impact the properties and appearance of the final product. The melt solidification behavior of the resins dictates the development of residual stresses and morphological features in the molded parts, which in turn contribute to the performance of the final products in commercial applications. Over the years, guidelines that have to do with processing conditions, process control, powder quality, and material properties have been defined to optimize the process. A great deal of study has been carried out by Crawford’s group, resulting in a number of

Rotational Molding of Polymers

2679

R

Fig. 2 Typical inside air temperature profile of mold.

patented technologies. The continuous monitoring of the inside air temperature as well as the inside pressure of the mold has been shown to provide useful information about the state of the material during the molding process and has been implemented into control strategies.[3,5] Strategies to reduce the molding time include preheating the powder and the mold, as well as using internal mold pressure in the late stage of the heating cycle for the removal of bubbles formed.[5] Along the same lines, mold pressurization during the cooling stage can significantly help reduce the cooling cycle time and also lead to improved dimensional stability of the molded product.[5,6] The identification of key material properties and their effect on the molding cycle and the performance of the final product are discussed in more detail in the following sections.

MELT DENSIFICATION In rotational molding, two bulk movements dominate the melt deposition and densification: 1) the adhesion of particles on the mold surface and 2) the melting and collapse of the particulate structure.[7,8] The latter

movement causes the entrapment of air into the melt and, thus, the formation of bubbles. Two distinct stages have been identified in the densification process: 1) particle coalescence and 2) bubble dissolution, as illustrated in Fig. 2.

Powder Coalescence The coalescence of polymers is driven by the work of surface tension, which counteracts the viscous dissipation associated with the molecular diffusion within the coalescing domain. This phenomenon is often referred to in the literature as polymer sintering. In the rotational molding process, coalescence occurs at temperatures above that of the material melting point when dealing with semicrystalline polymers, or above the glass transition temperature for amorphous resins. The first analytical model describing the coalescence process was proposed by Frenkel:[9]
1=2 y Gt ¼ ð1Þ a Za where y, a, G, t, and Z are the neck radius formed between the particles, the radius of the particles,

2680

material surface tension, time, and material viscosity, respectively. The model is based on the balance of the work of surface tension and the viscous dissipation. All other forces, including gravity, are neglected. The shape of the two spheres evolves, as shown in Fig. 3. The validity of Frenkel’s model is limited to Newtonian flow and can only be used to predict the early stage of the coalescence process, when the diameter of the two spherical particles remains nearly unchanged. The inadequacy of a Newtonian model in describing the coalescence of polymers was also demonstrated in other studies, as reviewed by Mazur, and has led to the development of models as well as alternative methods for the characterization of the coalescence behavior of polymers for rotational molding applications.[10] Based on theoretical and experimental analyses of the coalescence phenomenon, the material properties of primary interest in the evaluation of resin coalescence behavior in rotational molding have been identified as the resin viscosity, surface tension, and elasticity. Most practitioners define the flow behavior of polymers based on the melt flow index; however, this property is not entirely relevant to the rotational molding process because it is essentially a shear-free and pressure-free process. The use of zero-shear viscosity has been proposed as a better way to assess the coalescence behavior of resins. Resins with lower zeroshear viscosity coalesce at a faster rate and can thus be processed using a shorter molding cycle.[11] The coalescence of individual powder particles is initiated as the particles stick and melt onto the mold surface or melt front. As the melt deposition process continues, pockets of air remain trapped between partially fused particles and lead to the formation of bubbles. In the rotational molding process, the coalescence of particles occurs at a temperature range close to the melting point of the material; thus, from a processing standpoint, low values of zero-shear viscosity at low temperatures (i.e., close to the temperature at which the particles adhere to the mold surface) are preferable. In addition to zero-shear viscosity, material elasticity plays an important role in determining coalescence behavior in the rotational molding process.[10,12,13] For rotational molding grade material, this effect was first recognized from the experimental results obtained with impact modified polypropylenes and is illustrated in Fig. 4.[12] In this work, the relative elasticity of resins

Rotational Molding of Polymers

was evaluated based on stress relaxation measurements. The time required for the complete relaxation of the material was used as a measure of the melt elasticity. Resins with increased elasticity, but otherwise comparable zero-shear viscosities, produced parts with higher bubble content and, thus, lower density (see Fig. 4). In more recent studies, the relative elasticity of a resin has been evaluated from rheological oscillatory measurement, by comparing the ratio of the loss modulus over the elastic modulus (G00 =G0 ). When comparing resins with similar molecular weight, low values of G00 =G0 at low frequencies are indicative of high elasticity, which translates into lower coalescence rates. The relative elasticity alone rarely explains poor coalescence behavior because resins with increased elasticity will often show variations in thermal and diffusion properties that are critical in the densification process. For instance, changes in the molecular structures leading to the broadening of the melting range can be detrimental to the coalescence process. Particles that soften at low temperatures and adhere to each other prior to a substantial movement of the bulk of the material cause the entrapment of large bubbles, as illustrated in the results presented in the literature.[8,14] Moreover, the presence of heterogeneities in the molecular formulation that cause the formation of segregated amorphous regions can seriously affect molecular interdiffusion and, thus, the coalescence process. Molecular structure has an evident influence on a resin’s rheological characteristics and expected coalescence behavior. Polymers with higher molecular weight have a slower coalescing rate because viscosity increases with molecular weight. Narrow molecular weight distribution results in lower elasticity and lower viscosity at low deformation rates, which is beneficial to the coalescence process. The comonomer content in polyethylene copolymers has mixed effects on the coalescence process. The increase of comonomer content generally results in a reduction in the melting temperature and the heat of fusion, which lead to earlier melting and onset of coalescence in transient molding processes. However, the incorporation of short-chain branches on linear polymer chains causes a reduction in the coalescing rate, as illustrated in Fig. 5. The effect of short-chain branches content on the coalescence rate may originate with the differences in the chain mobility due to variations of the chain linearity. The more linear the chains, the faster they are expected to diffuse.

Fig. 3 Shape evolution of two coalescing spheres.

Rotational Molding of Polymers

2681

R

Fig. 4 (A) Stress relaxation curve of rotational molding grade ethylene copolymers. (B) Part density, which is indicative of the bubble content, measured after interrupting the rotational molding heating cycle at different heating times. (From Ref.[12].)

Variations in chemical composition and molecular structure can also have important repercussions on the surface energy of the material. Several techniques have been proposed for the experimental determination of surface tension, the sessile and pendant drop methods being the most promising and commonly used

techniques. The common limitation of these techniques is that the melt surface energy is not directly measured; thus, the validity of the results depends on the underlining assumptions of the models used to fit the experimental observations. While surface tension has long been recognized as a controlling parameter in polymer

2682

Rotational Molding of Polymers

Fig. 5 Effect of comonomer content on the coalescence behavior of polyethylene copolymers at ramped temperature (111–226.5 C at 11 C=min). The relative methyl content for PE9-O, PE10-O, and PE11-O is 0.95, 0.56, and 0.15, respectively. (From Ref.[15].)

coalescence, limited work has been done in determining its effect on the processability of the resin in the rotational molding process.[16] The primary reason is that the measurement of the interfacial energy of polymer melts is challenging, owing to the viscous nature of the material and the potential for thermal degradation of the material upon exposure to high temperature and the long time scale of the experiments that have been defined for this type of measurement. Dissolution of Bubbles The level of coalescence between particles, the size of the particles, and the packing arrangement dictate the size of air cavities and, thus, the size of the bubble initially formed in the melt. Once formed, the bubbles remain stationary in the melt.[7] A relatively small bubble diameter, combined with the high viscosity of the melt, prevents the movement of the bubbles into the melt. The bubble removal is known to be a diffusion-controlled process. The identification of key parameters in the dissolution of bubbles formed in the melt has been done using a theoretical model that describes this process. The disappearance of the air bubble formed into the melt was modeled based on

the analytical solution presented by Gogos:[17]  
R þ a2 R20
R2 þ a1 ðR0
RÞ þ a1 a2 ln R0 þ a 2 cs;P1
c1 ¼ 2D t r1

ð2Þ

where R0 and R are the initial bubble radius and the bubble radius for the time interval t, respectively, and D is the diffusion coefficient of the gas into the melt (see Fig. 6). The terms a1 and a2 in Eq. (2) are defined as follows: a1 ¼

  4G 2 cs;P1
Rg T 3 cs;P1
c1

ð3Þ

a2 ¼

2G cs;P1 Rg T r1 ðcs;P1
c1 Þ

ð4Þ

with Rg and T being the ideal gas law constant and the temperature of the melt–bubble system, respectively. The terms rb and r1 symbolize the density of gas at the bubble=polymer melt interface for nonzero and zero surface tensions, respectively. The terms c1 and cs,P1, which symbolize the dissolved gas concentration

Rotational Molding of Polymers

2683

R

Fig. 6 Schematic of gas bubble dissolving into polymer melt with Cs,Pb being the dissolved gas concentration at the bubble=melt interface, c1 the dissolved gas concentration in the melt when partially saturated, x the degree of saturation in the melt, H Henry’s law constant, Pb the bubble pressure, R the bubble radius, T the temperature, and rb the density of the gas in the bubble.

in the melt when partially saturated and saturated for zero surface tension, respectively, are determined as follows: cs;Pb
cs;P1 ¼ c1 ¼ xcs;Pb

2G cs;P1 Rg TP Rr1

ð5Þ ð6Þ

The dissolved gas concentration at the bubble=melt interface, cs,Pb, can be related to the bubble pressure Pb through Henry’s law. Gogos compared his model’s predictions with the experimental data produced by Spence.[18] The model predictions fit very well with experimental data when selecting a degree of saturation close to 100% (91.9–99.6%). An alternative approach has been proposed by Kontopoulou and Vlachopoulos, who modeled the dynamics of bubble dissolution into the melt using conservation of mass and momentum.[19] Factors that can affect the rate of bubble dissolution are the initial size of the bubbles, the temperature, the melt diffusion characteristics, the surface tension, and the solubility characteristics. The rate of consumption of oxygen due to degradation reactions probably contributes to the disappearance of bubbles, but this phenomenon has not yet been considered in the analysis of the process. It has been shown that within the range of viscosity typical to polyethylene rotational molding grade resins, the effect of viscosity on the dissolution of bubbles is marginal.[19] The level of air saturation in the polymer melt, however, was found to be crucial. While it cannot be determined accurately in the process, small variations in the level of saturation were found to cause large changes in the dynamics

of the bubble dissolution.[17,19] Interfacial tension was shown to be important under conditions where the melt saturation level is close to unity.[17] The model proposed by Gogos was also useful in providing an explanation for the effect of mold pressurization on the bubble dissolution process. A sudden increase in the melt pressure causes a sudden shrinkage of the bubble. It also affects the level of melt saturation and causes an instep concentration gradient at the bubble=melt interface (see Fig. 6), leading to an increased rate of dissolution and faster removal of bubbles, as illustrated in Fig. 7. Experimental and theoretical results clearly showed that a small increase in the mold cavity pressure can lead to a significant reduction in the heating time required for complete melt densification in the rotational molding process.

Nonisothermal Densification The densification process in rotational molding has been studied from both fundamental and practical perspectives. The models presented in the previous sections have furthered the understanding of the densification process in rotational molding; however, their use has been limited to the prediction of the densification process carried out under isothermal conditions. It is well known that heat transfer, powder coalescence, bubble formation, and bubble dissolution are collectively important in rotational molding; however, very few studies have addressed all aspects in modeling the densification process in rotational molding. Heat transfer in the rotational molding process was first modeled by Rao and Throne.[20] Since this initial attempt at modeling the heating cycle, several studies

2684

Rotational Molding of Polymers

Fig. 7 Effect of mold cavity pressurization on the bubble dissolution time, based on experimental data presented by Spence. (From Ref.[18].)

have been undertaken to improve the model with consideration of different powder flow patterns, multidirectional heat transfer, complex mold geometry, internal heating, and the temperature dependence of the material properties. The thermal resistance within the mold wall was usually neglected. Moreover, it was generally assumed that the powder remains well mixed under conditions typically used in the rotational molding process. This assumption greatly simplified the modeling of the process, and model predictions have been found to be in good agreement with experimental results published in the literature.[21] However, the powder flow pattern may be better modeled using an avalanche or rolling type of flow, based on experimental evidence of the development of segregation patterns.[22,23] In most models presented in the literature, the mold curvature was neglected and only unidirectional heat transfer was considered. However, the trend toward the fabrication of more technically challenging parts with tighter specifications has provided some impetus to develop models that consider multidirectional heat transfer.[24] Moreover, the consideration of the kinematic of the mold rotation has been an important milestone in the development of commercial simulations that are used as design tools in the selection of rotation speed and heating conditions to ensure uniform melt deposition onto the mold surface for given mold geometry and polymer thermal properties. The combined effect of heat transfer and sintering on the nonisothermal densification of polymer powder in rotational molding has been examined by Bellehumeur and Tiang.[25] For modeling purposes, the heating cycle was divided into three regimes: 1) heating

of the mold and its contents until the mold temperature reaches the polymer melting point; 2) the polymer powder gradually melts and deposits on the mold surface; and 3) starts once all the powder has disappeared. In modeling the early stage of melt densification, the powder deposition process was treated as occurring in a layer-by-layer manner. The coalescence time allocated for each layer was determined by the rate of melt deposition. The initial size of the bubbles formed in the melt was determined by the neck growth achieved between particles, the size of the particles, and their packing arrangement. Bellehumeur and Tiang examined the relative importance of the powder characteristics and the material rheological properties on the melt densification process.[25] Their results showed that the initial size of the bubbles formed into the melt is primarily controlled by the powder particle size and packing arrangement (Fig. 8). This observation highlights the importance of powder quality on the molding cycle and the quality of the final product. The time required by the melt deposition process is roughly proportional to the square of the molded part thickness and is significantly faster for polymers with lower melting points. A decrease in the heat of fusion also accelerates the deposition process, though not as much as a decrease in the melting temperature. A consequence of a fast melt deposition is that larger bubbles are formed in the melt, making them more difficult to remove during the heating cycle and negatively affecting the densification process. This problem can be surmounted with a reduction in the oven temperature; however, this reduction results in an increase in the total molding cycle time.

Rotational Molding of Polymers

2685

R

Fig. 8 Effect of the particle packing arrangement on the relative density of a rotational molding grade polyethylene with heating time. (From Ref.[25].)

MELT SOLIDIFICATION The solidification of the polymer melt in rotational molding is relatively slow, in comparison to other processes, and is estimated to be in the range of 10– 30 C=min. Moreover, the melt solidification is gradual and nonuniform across the molded part thickness, leading to important variations in the morphological features, as illustrated in Fig. 9, and dictating the properties and overall performance of the final product. The effects are more dramatic for resins with slower crystallization rates, such as polypropylene, compared to that observed with polyethylene. Many thermoplastics used in rotational molding applications are semicrystalline in nature. The temperature and rate of crystallization vary with the material composition and molecular structure. Polymers that tend to crystallize have flexible backbones, regularly ordered atomic structure, and either small or no pendant groups. Crystallization occurring at higher temperatures is associated with a lower nucleation rate and the formation of coarse morphological structures. This usually results in the formation of structures that have a higher degree of stability and perfection but a lower failure strain, because the

increase in the size of the spherulitic structure (common in polyethylene and polypropylene) is accompanied by a reduction in the interspherulitic boundary links. This, in turn, allows for the easier transmission of energy through the material and thus causes a loss in the ductility. This problem can be alleviated by using a faster cooling rate, which, on the other hand, leads to the generation of residual stresses in the parts. Because the material in contact with the mold is cooled at a faster rate than that near the inner surface, an asymmetrical residual stress profile is generated across the molded part thickness, inducing a bending moment of the part. If these induced residual stresses are high enough to overcome the structural integrity of the part, the molded part will deform and warp. Other factors that affect the development of residual stresses and the dimensional stability of the molded part include the material density, molded part thickness, mold material and thickness, rotation speed, and the application of mold release. The presence of residual stresses is associated not only with dimensional stability but also with the overall integrity of the molded part and is known to most severely affect the environmental stress cracking resistance of the product.

2686

Rotational Molding of Polymers

Fig. 9 Microphotograph of rotationally molded polypropylene parts subjected to water spray cooling (A, mold surface and B, inner surface) and parts initially subjected to water spray and subsequently air flow in the range of temperatures where crystallization occurs (C, mold surface and D, inner surface).[26]

In the rotational molding process, the polymer is subjected to high temperatures for relatively long periods of time to allow for melt deposition and full densification of the powder particles. As a consequence, thermo-oxidative degradation reactions can be severe if the material is not adequately stabilized. Degradation can often be determined by visual inspection of the molded part (change in color). Subtle changes can be detected through polarized or fluorescence microscopy (illustrated in Fig. 10) or infrared spectroscopy for the detection of carbonyl groups. The rheological characterization of specimens collected from the

molded part can also provide useful information for degradation reactions can cause either chain scission or cross-linking reaction within the materials. Signs of degradation are usually first seen on the inner surface of the molded part for it is exposed to air during the entire molding process. Degradation reactions can be minimized with the use of an inert atmosphere in the mold. Excessive degradation of the material due to long exposure at high temperatures leads to the deterioration of the morphology (spherulitic structure) and the reduction in the mechanical properties of molded parts.[26,27]

Fig. 10 Microstructure of rotationally molded polypropylene samples: (A) undercured specimen viewed under polarized light microscopy; (B) overcured specimen viewed under polarized light microscopy; and (C) overcured specimen viewed under fluorescence microscopy. In these pictures, the degraded layer shows higher birefringence. (From Ref.[26].)

Rotational Molding of Polymers

Despite the importance of the melt solidification phenomena in rotational molding, only a few studies have been conducted with respect to modeling the cooling stage of the rotational molding process. One of the challenges resides in the nonlinearity of the model, which arises due to the unknown position of the moving solid=melt interface, and which is very similar to that encountered in modeling the heating stage of the rotational molding process. Throne pioneered this topic and investigated the various factors influencing the cooling rates.[28] Further work has been conducted to model this stage of the process with consideration of shrinkage, which can also greatly affect the molding cycle time.[24] Methodologies have also been proposed for the development of morphological features during processing based on half-time crystallization data, determined from differential scanning calorimetry, and spherulitic growth rate, which was obtained from optical microscopy experiments.[29] These are important steps toward the establishment of material–structure–properties–processing relationships for the rotational molding process.

TECHNOLOGICAL ADVANCES IN THE DEVELOPMENT OF NEW MATERIALS Polyethylenes account for over 80% of all rotational molding production. They dominate the rotational molding market because of their thermal stability, availability in powder form, and relatively low cost. Over the last decade, polyolefins based on metallocene and singlesite catalyst technologies have found their way into the rotational molding market. These technologies allow for better control of the molecular structure, which can have important repercussions on the processing behavior of the material as well as on the physical, chemical, and mechanical performance of the molded parts. Early results obtained with metallocene polyolefins suggested that these resins could be processed using a shorter heating time in rotational molding. In most studies, the reduction in the heating cycle was attributed to differences in the thermal properties (melting point and heat of fusion) between metallocene and Ziegler– Natta resins. However, warpage was commonly reported as a problem associated with using metallocene-based polyethylene. A more recent study showed that, with careful control of the molecular structure and the introduction of a certain level of heterogeneities, a significant reduction in the molding cycle time (up to 30%) could be achieved while maintaining adequate physical and mechanical properties.[30] One important limitation in expanding the markets for rotational molding applications is the small range of resins suitable for the process. One cause for this limited choice of resins used commercially resides in

2687

poor processability behavior and the cost of pulverization, which is particularly important for resins that require cryogenic conditions. Alternatives to the use of powder particles in rotational molding have been proposed, thus eliminating the need for pulverization. The underwater pelletizing technology can be used to produce small cylindrical particles called micropellets, which are being used increasingly for rotational molding applications. It is well established that micropellets have better flow properties than powder particles. This characteristic has a significant impact on the thickness uniformity of molded parts and mold filling for complex geometries. However, one disadvantage of using micropellets is their narrow particle size distribution and their larger particle size. The packing of micropellets during the melt deposition process is such that larger bubbles are formed, compared to that seen when using a 35-mesh powder. These larger bubbles require a longer exposure to high temperatures for their complete dissolution. Recent years have seen increased interest in the development of new rotational molding grade resins, which include polypropylene, acrylonitrile butadiene styrene, acetals, polyamides, thermoplastic foams, polyolefin blends, polyolefin plastomers, thermotropic liquid crystal polymers and nanocomposites. Two major concerns exist for the development of new resins: 1) obtaining adequate product performance and 2) achieving adequate processability. The most important roadblock in the development of many new resins, such as styrenic copolymers and polypropylene, resides in the fact that good impact properties can only be obtained with the incorporation of an impact modifier, often taking the form of a copolymer. Unfortunately, such a change in the material formulation often results in an increase in the material’s relative elasticity, which is detrimental to the densification process and results in the production of parts with high bubble content and poor surface finish. Similarly, while the successful incorporation of nanoparticles in a polymer matrix can lead to the improvements of properties such as barrier resistance, it can also result in a dramatic increase in the linear viscoelastic properties. This proves to be an important drawback because increased zero-shear viscosity and elasticity are detrimental to the processability of the material. The traditional approach to solving this problem has consisted in varying the parameters known to affect properties that are key to the coalescence process (i.e. viscosity, elasticity, surface tension) with the incorporation of lubricants and other additives.[31] Promising results were also obtained for processing polyolefin plastomers with the uniform incorporation of the copolymer using metallocene catalyst technologies.[14] Alternative approaches have also included the careful selection of processing conditions that favor the completion of the coalescence

R

2688

process between individual particles prior to the formation of bubbles into the melt.[32]

CONCLUSIONS The rotational molding industry has seen a steady growth in the last two decades with the development of new process control strategies, machinery, and, to some extent, automatization. Yet, the development of new rotational molding applications is restricted for the processing cycle times are long and the choice of resins that can economically be used in this process is shockingly limited. The development of new resins is possible only if adequate product performance can be obtained while maintaining adequate processability. This can only be achieved with a good understanding of the phenomena that govern the quality of the end product. From a material formulation perspective, the key aspects to consider are the melt coalescence and melt solidification. While there has been significant progress in the development of methods for the rapid assessment of the processability and performance of resins, further work is required in defining strategies to overcome many of the process limitations.

REFERENCES 1. Throne, J.L.; Sohn, M.-S. Characterization of rotational molding grade polyethylene powders. Adv. Polym. Technol. 1989, 9 (3), 181–192. 2. McDaid, J.; Crawford, R.J. The grinding of polyethylene powders for use in rotational molding. Proceedings of the Society of Plastics Engineers Annual Technical Conference, Atlanta, GA, Apr 26–30, 1998; The Society of Plastics Engineers: Brookfield, 1998. 3. Crawford, R.J.; Nugent, P.J. A new process control system for rotational moulding. Plast. Rubber Comp. Process. Appl. 1992, 17 (1), 23–31. 4. Crawford, R.J.; Nugent, P.J. Impact strength of rotationally moulded polyethylene articles. Plast. Rubber Comp. Process. Appl. 1992, 17 (1), 33–41. 5. Crawford, R.; Cramez, M.C.; Oliveira, J.J.; Spence, A. The importance of monitoring mold pressure during rotational molding. Proceedings of the Society of Plastics Engineers Annual Technical Conference, San Francisco, CA, May 5–9, 2002; The Society of Plastics Engineers: Brookfield, 2002. 6. Chen, C.H.; White, J.L.; Ohta, Y. Mold pressurization as a method to reduce warpage in rotational molding of polyethylene. Polym. Eng. Sci. 1990, 30 (23), 1523–1528.

Rotational Molding of Polymers

7. Crawford, R.J.; Scott, J.A. An experimental study of heat transfer during rotational moulding of plastics. Plast. Rubber Process. Appl. 1985, 5 (3), 239–248. 8. Kontopoulou, M.; Vlachopoulos, J. Melting and densification of thermoplastic powders. Polym. Eng. Sci. 2001, 41, 155–169. 9. Frenkel, J. Viscous flow of crystalline bodies under the action of surface tension. J. Phys. 1945, 9 (5), 385–391. 10. Mazur, S. Coalescence of polymer particles. In Polymer Powder Technology, 1st Ed.; German, R.M., Messing, G.L., Cornwall, R.G., Eds.; John Wiley & Sons: New York, 1995; 157–216. 11. Bellehumeur, C.T.; Bisaria, M.K.; Vlachopoulos, J. An experimental study and model assessment of polymer sintering. Polym. Eng. Sci. 1996, 36 (17), 2198–2207. 12. Kontopoulou, M.; Bisaria, M.; Vlachopoulos, J. An experimental study of rotational molding of polypropylene=polyethylene copolymer. Int. Polym. Process. 1997, 12, 165–173. 13. Bellehumeur, C.T.; Kontopoulou, M.; Vlanchopoulos, J. The role of viscoelasticity in polymer sintering. Rheol. Acta 1998, 37 (3), 270–278. 14. Wang, W.Q.; Kontopoulou, M. Effect of molecular structure on the rotational molding characteristics of ultra-low-density-ethylene-a-olefin copolymers. Polym. Eng. Sci. 2004, 44 (3), 496–508. 15. Guille´n-Castellanos, S.A.; Bellehumeur, C.T.; Weber, M. The effect of molecular structure on the coalescence of ethylene-a-olefin copolymers. Proceedings of the Society of Plastics Engineers Annual Technical Conference, San Francisco, CA, May 5–9, 2002; The Society of Plastics Engineers: Brookfield, 2002. 16. Tinson, A.; Takacs, E.; Vlachopoulos, J. The effect of surface tension on the sintering of polyethylene copolymers and blends in rotational molding. Proceedings of the Society of Plastics Engineers Annual Technical Conference, Chicago, IL, May 16–20, 2004; The Society of Plastics Engineers: Brookfield, 2004. 17. Gogos, G. Bubble removal in rotational molding. Polym. Eng. Sci. 2004, 44 (2), 388–394. 18. Spence, A. Analysis of Bubble Formation and Removal in Rotationally Moulded Products. PhD thesis, Department of Mechanical and Manufacturing Engineering, Queen’s University in Belfast, Belfast, Northern Ireland, 1994. 19. Kontopoulou, M.; Vlachopoulos, J. Bubble dissolution in molten polymers and its role in rotational molding. Polym. Eng. Sci. 1999, 39 (7), 1189–1198. 20. Rao, M.A.; Throne, J.L. Principles of rotational molding. Polym. Eng. Sci. 1972, 12 (4), 237–264.

Rotational Molding of Polymers

21. Nugent, P. A Study of Heat Transfer and Process Control in the Rotational Moulding of Polymer Powders. PhD thesis, Department of Mechanical and Manufacturing Engineering, Queen’s University in Belfast, Belfast, Northern Ireland, 1990. 22. Pop-Iliev, R.; Park, C.B. Single-step rotational foam molding of skin-surrounded polyethylene foams. J. Cell. Plast. 2003, 39 (1), 49–58. 23. Olinek, J.; Anand, C.; Bellehumeur, C.T. Experimental study on the flow of polyethylene powder particles in rotational molding. Polym. Eng. Sci. 2005, 45 (1), 62–73. 24. Gogos, G.; Liu, X.; Olson, L.G. Cycle time predictions for the rotational molding process with and without mold=part separation. Polym. Eng. Sci. 1999, 39 (4), 617–629. 25. Bellehumeur, C.T.; Tiang, J.S. Simulation of nonisothermal melt densification of polyethylene in rotational molding. Polym. Eng. Sci. 2002, 42 (1), 215–229. 26. Cramez, M.C.; Oliveira, M.J.; Crawford, R.J. Effect of nucleating agents and cooling rate on the microstructure and properties of a rotational moulding grade polypropylene. J. Mater. Sci. 2001, 36 (9), 2151–2161.

2689

27. Cramez, M.C.; Oliveira, M.J.; Crawford, R.J.; Apostolov, A.A.; Krumova, M. Rotationally molded polyethylene: structural characterization by x-ray and microhardness measurements. Adv. Polym. Technol. 2001, 20 (2), 116–124. 28. Throne, J.L. Some factors influencing cooling rates of rotational molding parts. Polym. Eng. Sci. 1972, 12 (5), 335–339. 29. Martin, J.A.; Cramez, M.C.; Oliveira, M.J.; Crawford, R.J. Prediction of spherulite size in rotationally molded polypropylene. J. Macromol. Sci. B 2003, B42 (2), 367–385. 30. Hay, H.; Weber, M.; Donaldson, R.; Gibbons, I.; Bellehumeur, C. Single site polyethylene resins with enhanced processability for rotational molding applications. Proceedings of the Society of Plastics Engineers Annual Technical Conference, Chicago, IL, May 16–20, 2004; The Society of Plastics Engineers: Brookfield, 2004. 31. Chaudhary, B.I.; Taka´cs, E.; Vlachopoulos, J. Processing enhancers for rotational molding of polyethylene. Polym. Eng. Sci. 2001, 41 (10), 1731–1742. 32. Scribben, E.; Baird, D. The rotational molding of a thermotropic liquid crystalline polymer. Polym. Eng. Sci. 2005, 45 (3), 410–423.

R

Rubber Devulcanization R David A. Benko Roger N. Beers The Goodyear Tire & Rubber Company, Akron, Ohio, U.S.A.

INTRODUCTION During the preparation of many rubber articles such as tires a variety of additives including sulfur, accelerators, and activators are used to induce sulfur crosslinks between the rubber chains. These are formed during the heating or curing process and together with reinforcing fillers give the article its strength. Table 1 shows the composition of a rubber compound typical of that used in a body ply of a radial passenger tire. The cross-links are composed of structures containing various sulfide lengths, usually described as monosulfidic, disulfidic, or polysulfidic, which are determined by the type, amount, and proportion of additives used in the rubber compound. Devulcanization in a sulfur cured rubber is defined as the cleavage of the mono-, di-, and polysulfidic cross-links formed during vulcanization (Fig. 1). The vulcanization process is irreversible and additional heating induces changes in the network with a shift toward shorter cross-links but does not devulcanize the compound. Other methods are therefore needed to induce devulcanization.

BACKGROUND INFORMATION Because of the large numbers of used tires existing in stockpiles throughout the United States it is desirable to recycle these tires back into new products. Unfortunately, simply grinding up the used tire and adding the resultant material to virgin rubber causes a significant drop in properties including reducing the strength and flexibility of the compound. The need for a commercially viable devulcanization process has existed for years and continues to be a goal. Devulcanization without rubber chain degradation offers the potential for recycling used tires back into new products without sacrificing performance. Considerable effort has been directed to solving this problem.[1,2] Although chemical probes have been developed that selectively cleave carbon–sulfur and sulfur–sulfur bonds but not carbon–carbon bonds, most of the effort on devulcanization processes has been focused on providing a usable form of rubber suitable for use as a reclaimed material in new articles.[3] Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120008034 Copyright # 2006 by Taylor & Francis. All rights reserved.

In most cases direct evidence for actual devulcanization, i.e., breaking sulfur–sulfur and carbon–sulfur bonds without polymer chain scission, is lacking. However, in many instances the so-called devulcanization process increases the suitability for reuse (Fig. 2). This article will review the classical chemical methods of devulcanization and their inherent limitations. The newer methods of devulcanization currently under development will also be reviewed. These include devulcanization using microbes, microwave, ultrasonically induced devulcanization, and devulcanization using supercritical fluids.

CHEMICAL DEVULCANIZATION A large number of chemical devulcanization agents for natural and synthetic rubbers have been developed. These include phosphines and phosphates, numerous sulfides and mercaptans, metal salts such as methyl iodide, phenyl lithium, lithium aluminum hydride, and phase-transfer catalysts.[3–25] Included in the list of sulfides and mercaptans are diphenyl disulfide, dibenzyl disulfide, diamyl disulfide, bis(alkoxy aryl) disulfides, butyl mercaptan and thiophenols, xylene thiols and other mercaptans, phenol sulfides and disulfides.[6–13,26,27] Cook and coworkers reported the preparation, evaluation, and structural correlation of alkyl phenol sulfides as devulcanizing agents for styrene–butadiene rubber (SBR).[13] The effect of these alkyl phenol sulfides as reclaiming agent was compared with that of many aromatic thiols. Some N,N-dialkyl aryl amine sulfides were shown to be highly active reclaiming agents for vulcanized SBR in both neutral and alkaline reclaiming processes.[12] A review of the science and technology of reclaimed rubber was published by Le Beau in 1967.[14] Knorr has shown the action of diaryl disulfide on the natural and synthetic rubber scraps of technical goods.[15] A reaction mechanism for the breaking of polysulfide bonds in the presence of propane thiol=piperidine probe was proposed by Saville and Watson.[28] The thiol-amine combination gives an associate, possibly piperidinium propane-2-thiolate ion pair where sulfur atoms enhance the nucleophilic character that is responsible for cleaving 2691

2692

Rubber Devulcanization

Table 1 Typical tire compound Ingredient

a

Amount

Diene rubberb

100

Reinforcing filler

50–60

Oils

5–15

Antioxidants

1–2

Zinc oxide

5

Accelerators

1–2

Sulfur

2–3

a

Amounts are parts by weight per 100 parts by weight rubber. Diene rubber includes natural rubber, polyisoprene rubber, butadiene rubber, and styrene–butadiene rubber. b

organic polysulfide linkages. The cleavage of polysulfide cross-links takes place possibly as a result of pp–dp delocalization of the displaced sigma electron pair of RSS
. Campbell found hexane-1-thiol to be more reactive and thus capable of cleaving both polysulfide and disulfide linkages while leaving the monosulfides intact (Fig. 3).[29] Moore and Trego described the use of triphenyl phosphine and di-N-butyl phosphate as chemical probes to establish a cross-link network structure in rubber vulcanizates.[4,5] Triphenyl phosphine and trialkyl phosphates cleave di- and polysulfide, as illustrated in Fig. 4. Studebaker identified the use of lithium aluminium hydride as a chemical probe.[23,24] Under the right conditions it cleaves poly- and disulfide bonds, leaving the monosulfide intact. Lithium aluminum hydride reacts with polysulfides in an etheral solvent at moderate temperatures and then with a weak acid, the terminal groups are liberated as thiols and interior sulfur atoms are converted to hydrogen sulfide.[30–39] Lithium

aluminum hydride under appropriate reaction conditions also cleaves disulfide bonds in organic disulfide, which is structurally related to cross-links, into two thiol groups.[32] Similarly Gregg and Katrenick found that phenyl lithium will cleave polysulfides and disulfides but not monosulfide linkages.[21,22] Methyl iodide can be employed to estimate monosulfide linkages in vulcanized natural rubber (NR).[33,34] After swelling the rubber in methyl iodide for several days the level of network bound iodine after reaction would reflect the concentration of monosulfide groups because the simple saturated monosulfide group reacts as shown in Fig. 5. Simple disulfides reacted very slowly with methyl iodide but their reaction and those of monosulfides could be catalyzed by mercuric iodide.[35,36] Anderson patented the reclaiming of sulfur vulcanized rubber in the presence of oil, water vapor, and a mixture of aryl disulfides (diphenyl disulfide, dicresyl disulfide, and dixylyl disulfide) at elevated temperature and pressure.[40] Desulfurization of suspended rubber vulcanizate crumb (10–30 mesh) was carried out in a solvent such as toluene, naphtha, benzene, or cyclohexane in the presence of sodium.[41] The alkali metal cleaves mono, di- and polysulfidic cross-linkages of the swelled and suspended vulcanized rubber crumb at around 300 C in the absence of oxygen. As claimed by the authors, such treatment yielded a rubber polymer having a molecular weight substantially equal to that of rubber prior to vulcanization. 2-Mercaptobenzothiazole was also found to be effective as a reclaiming agent.[42] In this process

Fig. 1 Vulcanization.

Rubber Devulcanization

2693

R

Fig. 2 Devulcanization.

powder rubber from waste tires was kneaded with process oil in the presence of 2-mercaptobenzothiazole or its cyclohexylamine salt to give reclaim rubber. Vehicle tire scrap containing polyisoprene rubber, SBR, and butadiene rubber was devulcanized by low-temperature phase-transfer catalyst. Both the devulcanizing agent composition and the process were patented. The novelty of this process lies in the use of low-temperature phase-transfer catalyst and a process temperature lower than 150 C. The devulcanized rubber of this invention is distinguishable from conventional reclaimed rubber in that the devulcanized rubber is substantially free from polysulfide crosslinks, which are selectively broken during the process with negligible main chain scission.[25,43]

Microwave Method The microwave method is useful because it provides an economical, ecologically sound method of reusing elastomeric waste to return it to the same process and products in which it was originally generated. In this technique a controlled dose of microwave energy is used at a specified frequency and energy level to cleave carbon–carbon bonds.[44–46] Thus, in this process elastomer waste can be reclaimed without complete depolymerization to a material capable of being recompounded and revulcanized having physical properties essentially equivalent to the original vulcanizate. In typical commercial processes the ‘‘devulcanized’’ rubber is not degraded. In this process it is claimed that

Fig. 3 Polysulfide bond breaking with thiols.

2694

Rubber Devulcanization

Fig. 4 Phosphine and phosphite devulcanization.

sulfur vulcanized elastomer containing polar groups is suitable for microwave devulcanization. Tyler and Cerny claimed their microwave devulcanization process as a method of pollution controlled reclaiming of sulfur vulcanized elastomer containing polar groups.[47] The microwave energy devulcanization device generates heat at a temperature in excess of 260
C to yield a mass, which is fed to an extruder that extrudes the rubber at a temperature of 90–125
C. The extrudate can be used per se as a compounding stock. Another process was developed for reclaiming waste elastomers by microwave radiation. The process involves the impregnation of the waste rubber with an essential oil and then heat treating the impregnated material under reduced pressure with microwave radiation.[48] The waste material must be polar so that the microwave energy will generate the heat necessary to devulcanize. Microwave energy between 915 and 2450 MHz and between 41 and 177 W=hr=lb is sufficient to sever all cross-link bonds but insufficient to sever polymer chain degradation. The cost of devulcanized

Fig. 5 Lithium aluminum hydride and methyl iodide devulcanization.

hose and inner tube material by the microwave method is only a fraction of the cost of the original compound. The transformation from waste to refined stock ready for remixing takes place in only 5 min with usually 90–95% recovery of the rubber. Therefore, it appears that this microwave technique is a unique method of reclaiming in terms of properties and fastness of the process.

Ultrasonic Method After the microwave techniques, ultrasonic energy was used for the devulcanization of cross-linked rubber. The first work with ultrasonic energy was reported by Pelofsky in 1973, which was patented.[49] In this process solid rubber articles such as tires are immersed into a liquid, which is then kept under a source of ultrasonic energy whereby the bulk rubber effectively disintegrates upon contact and dissolves into liquid. In this process ultrasonic irradiation is in the range of about 20 kHz and at a power intensity of greater than 100 W. Ultrasonic reclaiming of NR was reported by Okuda and Hatano in 1987, which was also patented.[50] They subjected the NR vulcanizate to 50 kHz ultrasonic energy for 20 min to achieve devulcanization followed by revulcanization and obtained reclaimed rubber with similar properties to those of original rubber. Isayev and coworkers reported in a number of publications the phenomenon of devulcanization by ultrasound energy and they also patented their developments.[51–59]

Rubber Devulcanization

The devulcanization process requires a high energy level to break carbon–sulfur and sulfur–sulfur bonds.[1] An ultrasonic field creates high-frequency extension– contraction stresses in various media.[1] Isayev and his group also made a percolation simulation of the network degradation during ultrasound devulcanization in which they claimed an excellent agreement of experimental data for SBR and ground rubber from used tires (GRT) with the predicted dependence of the gel fraction of devulcanized rubber on cross-link density. Curing behavior, rheological properties, structural characteristics of devulcanized rubber from model SBR and GRT rubbers, as well as mechanical properties of vulcanized rubber samples were studied and a possible mechanism of devulcanization was also discussed. They characterized the degree of devulcanization by the measurement of cross-link density and gel fraction of the devulcanized rubber. Later, they published on the ultrasound devulcanization of sulfur vulcanized SBR and on vulcanization of ultrasonically devulcanized SBR elastomers.[60] Isayev and coworkers studied the devulcanization of SBR at various temperatures, viz., 121 C, 149 C, and 176 C, different clearances at various flow rates, and the ultrasonic oscillation amplitudes.[51–53] The extent of devulcanization was studied by measuring percentage and cross-link density of the gel fraction. It was reported that both the cross-link density and the gel fraction decrease in the devulcanization process. For original ground rubber tire the measured gel fraction is 83% and cross-link density of gel is 0.21 kmol=m3, but after ultrasound treatment at 121 C barrel temperature it reduces to 64–65% with cross-link density of 0.02 kmol=m3. The cross-link density also decreases with higher residence time in the treatment zone and with higher specific ultrasonic energy. The mechanical properties of the revulcanized sample were also studied. With decrease in the crosslink density of the devulcanized rubber, the tensile strength of revulcanized samples varies from 1.5 to 10.5 MPa and elongation at break varies from 130% to 250%. Based on the results of mechanical properties, Isayev et al. proposed that the devulcanized sample having a cross-link density lower than 0.06 kmol=m3 can be regarded as overtreated, and samples with cross-link density higher than 0.10 kmol=m3 can be regarded as undertreated.[52] Thus, overtreatment causes main chain breakage and undertreatment causes insufficient devulcanization. They also reported that ultrasound treatment of SBR results in low molecular weights of the sol fraction: Mn ¼ 2 – 4  103.[53] Ultrasonic devulcanization, therefore, causes significant degradation of polymer chains. A simple model based on a purely topological consideration was proposed and simulation of the process was carried out.[61–64] In the model they have assumed a breakup

2695

of the main chain bond and cross-link bonds as independent random events. Such random scission of cross-links and main chain results in the formation of soluble branched rubber chains regarded as fragmented gel structure or microgel. It is found that during ultrasound devulcanization the molecular weight of sol fraction decreases, from which it may be understood that during ultrasound treatment not only C–S or S–S bonds but also C–C bonds break. Isayev et al. suggested a revulcanization scheme.[55] They concluded that devulcanized rubber contained a larger amount of sulfidized molecules that were responsible for cross-linking during revulcanization.

Biotechnological Processes Biodegradation of NR was recognized as early as 1914.[65] But an effective process was not discovered because rubber is a hydrophobic substance that is subject to attack only at the surface, rubber vulcanizates are highly cross-linked and highly branched, and rubber vulcanizates contain a large number of biologically active additives that retard biodegradation. Various biodegradation processes were tried over the years since, with some success reported, particularly after about 1985. Fermentation methods were studied extensively at Rutgers University by Nickerson and coworkers and by Elmer in the 1970s on ground scrap tires.[65] Their work was aimed at producing commercially valuable products via the fermentation process from scrap tires. This work led to many important findings and showed that although technically feasible, this was not a solution to get rid of scrap tires. Even under ideal conditions only minor reduction of the rubber occurred. An interesting recent approach was reported in a patent application to utilize a chemolithiotrope bacterium in aqueous suspension for attacking powder elastomers on the surface only, so that after mixing with virgin rubber, diffusion of soluble polymer chains is facilitated and bonding during vulcanization becomes again possible.[66,67] A biotechnological process was developed by Straube et al. for the devulcanization of scrap rubber by holding the comminuted scrap rubber in a bacterial suspension of chemolithotropic microorganisms with a supply of air until elemental sulfur or sulfuric acid is separated.[68] This seems to be an interesting process, which obtains reclaim rubber and sulfur in a simplified manner. The biodegradation of the cis-1,4-polyisoprene chain was achieved by Tsuchi, Suzuki, and Takeda.[69–71] They used bacterium that belonged to the genus Nacardia and led to considerable weight loss of different soft-type NR-vulcanizates. The microbial

R

2696

desulfurization or devulcanization of particle surfaces was investigated to increase the possibility of producing high-quality rubber products containing a larger percentage of recycled rubber.[72] In a typical process rubber powder, mainly SBR of old tires with 1.6% sulfur, was treated with different species of Thiobacillus, i.e., Thiobacillus ferrooxidans, Thiobacillus thiooxidans, and Thiobacillus thioparus in shake flasks and in a laboratory reactor. The sulfur oxidation depends to a large extent on the particle size. The best results were obtained with T. thioparus with a particle size of 100–200 mm. Of the total sulfur of the rubber powder, 4.7% was oxidized to sulfate within 40 days.[72] In a recent paper Steinbuchel studied the biological attack of microorganisms on rubber materials to evaluate the possible contributions of biotechnology for the development and recycling of used rubber products.[73] Adaptation of microbial enrichment cultures with tire crumb material for several months resulted in enhanced growth of microorganisms, especially for NR and SBR. Romine and Snowden-Swan developed a process for exposing fine powdered rubber to an enzyme that attacks the sulfur cross-links on the surface of the rubber particles.[74] This process is carried out at ambient temperature and takes approximately 3 days. The devulcanization reaction is halted at the sulfone (R–SO2–R) or sulfoxide (R–SO–R) stages and not allowed to proceed to the unwanted sulfate (R– OSO2–R) form. The sulfone and sulfoxide forms are reactive with virgin rubber and allow more of the devulcanized rubber to be used. The specific thiophyllic microbes used include T. ferrooxidans, T. thiooxidans, Rhodococcus rhodochrous, and Sulfolobus acidocaldarius. About 10–20% by weight of this devulcanized rubber can be added to a virgin rubber matrix. Rubber particles processed with S. acidocaldarius exhibited an increase of 15% in the modulus of elasticity when added at 15% by weight to a virgin rubber compound. A patent was issued to Fliermans and Wicks in 2002 for combining microwave and biological techniques to treat vulcanized rubber particles.[75] Samples of 40 mesh crumb rubber were incubated with bacillus-type bacterium. The best results were obtained with a 10=40 volume ratio of bacterium to crumb rubber at a temperature of 60–65 C for a 20 hr treatment period. The devulcanized rubber was evaluated in a tire tread compound at 20% with definite improvement over untreated crumb in terms of Mooney viscosity, tensile strength, and elongation. It was found that by treating the biologically devulcanized rubber with microwave energy, the overall properties of the resulting cured rubber are improved over the comparable control mix using either untreated crumb rubber or rubber treated solely with the biotreatment or microwave protocols.

Rubber Devulcanization

Supercritical Fluid Devulcanization Hunt and Kovolak discovered that cured rubber could be devulcanized by heating with 2-butanol under supercritical conditions.[76] By heating rubber compounds to at least 150 C under a pressure of 3.4  106 Pa in the presence of 2-butanol the molecular weight of the rubber was maintained at a relatively high level and its microstructure was unchanged. For example, sulfur-cured SBR samples that contained no filler, carbon black, silica, or a combination of carbon black and silica were heated with the 2-butanol under supercritical conditions. The SBR had an original weight average molecular weight of about 400,000. The weight average molecular weights of the devulcanized SBR samples recovered are reported in Table 2. Several related alcohols were also investigated and were found to be less effective, although most did induce devulcanization. Table 3 shows the cumulative amounts of SBR polymer recovered from a cured sample after heating at various temperatures. It can be seen from Table 3 that 2-butanol was far better than any of the other alcohols evaluated. Benko and Beers found that by treating ground rubber obtained from used tires in a similar process they could devulcanize the surface of the ground particle.[77–79] This enabled its reuse as new rubber compounds with only minimal loss of properties. Although 2-butanol is a preferred solvent, a number of other alcohols and ketones are described. The alcohol or ketone employed as the solvent will have a critical temperature that is within the range of about 200 C to about 350 C. It is preferred for the alcohol or the ketone used as the solvent to have a critical temperature that is within the range of about 250 C to about 320 C. The term ‘‘critical temperature’’ is defined as the temperature above which the gas of a compound (the alcohol or the ketone) cannot be liquefied by the application of pressure. Some representative

Table 2 Polymer molecular weights after devulcanization Example

Filler

Molecular weighta

11

No filler

181,000

12

No filler

186,000

13

Silica

244,000

14

Silica

293,000

15

Carbon black

197,000

16

Carbon black

216,000

17

Carbon black=silica

177,000

18

Carbon black=silica

177,000

a

The molecular weights reported are weight average molecular weights.

Rubber Devulcanization

2697

Table 3 Polymer recovery with alcohols Cumulative polymer recovery (%) Example

Alcohol

150 C

200 C

250 C

300 C

1

2-Butanol

38

82

90

93

2

2-Butanol

40

70

85

92

3

Methanol

2

3

4

7

4

Ethanol

2

4

9

20

5

1-Propanol

3

16

43

69

6

2-Propanol

2

7

13

25

7

1-Butanol

4

19

57

86

8

Isobutyl alcohol

2

10

44

74

9

1-Pentanol

3

11

42

89

4-Methyl-2-pentanol

2

11

33

68

10 (From Ref.[76].)

examples of alcohols that can be used include methanol, ethanol, allyl alcohol, 1-propanol, isopropyl alcohol, n-butanol, iso-butanol, 2-butanol, tert-butanol, 1-pentanol, 2-methyl-1-butanol, 3-methyl-1-butanol, 3-methyl-2-butanol, 2,2-dimethyl-1-propanol, methyl isobutyl ketone, and 1-hexanol. Some representative examples of ketones that can be used include acetone, methyl ethyl ketone, methyl n-propyl ketone, methyl isopropyl ketone, and diethyl ketone. Mixtures of such alcohols and ketones can be utilized as the solvent. In this series of experiments, whole tire reclaim rubber was ground to a particle size of 40 mesh (about 420 mm) and the surface of the ground crumb rubber was then devulcanized. The surface devulcanization was carried out in 2-butanol under the conditions of time, pressure, and temperature specified in Table 3. Then, the samples of surface devulcanized reclaimed rubber made were analyzed by a thermogravimetric

technique to determine the volatile content and the polymer content. The results of this analysis are also reported in Table 4 along with the analysis of a control that was not subjected to the devulcanization procedure. The samples of surface devulcanized reclaimed rubber made in this series of experiments were then compounded with a blend of virgin rubbers and cured. The blends were made by mixing 20 parts per 100 parts of rubber (phr) of the surface devulcanized reclaimed rubber samples with 70 phr of PlioflexÕ 1712 SBR, 30 phr of BudeneÕ 1254 polybutadiene rubber, about 9 phr of aromatic oil, about 70 phr of carbon black, about 2 phr of stearic acid, about 4 phr of wax, about 1 phr of accelerator, about 2 phr of zinc oxide, about 1.5 phr of sulfur, and about 1 phr of antioxidant. The PlioflexÕ 1712 has a bound styrene content of about 28.5% and was oil extended with about 37.5% of an

Table 4 Surface devulcanization Temperature ( C)

Pressure (psig)

Time (min)

Volatiles (%)

Polymer (%)

19

270

900

20

41.61

11.8

20

270

900

40

38.75

9.16

21

270

1500

20

15.06

31.24

22

270

1500

40

21.36

23.07

23

300

900

20

38.09

18.54

24

300

900

40

46.08

10.96

25

300

1500

20

36.78

16.58

26

300

1500

40

37.44

7.28

27

285

1200

30

35.77

19.56

28

285

1200

30

36.68

20.23

12.26

53.28

Example

Control (From Ref.[77].)

R

2698

Rubber Devulcanization

Table 5 Cure properties of devulcanized rubber Example

Torque (dN)a

Ts1 (min)b

T25 (min)c

T90 (min)d 18.5

19

17

7.3

9.1

20

18.2

7

8.7

16.7

21

16.3

7

9

19

22

16.3

7

8.7

19.5

23

16.4

6.7

8.3

17.7

24

16.9

6.8

8.4

19.5

25

18

6.5

8.3

18.5

26

17

6.5

8.1

17

27

17.5

6.5

8.2

16.5

28

16.6

6.8

8.6

19

Original

16.6

5.3

6.6

14.9

None

19

6.4

8.3

7.2

Properties are obtained from a cure rheometer at 150 C. a Minimum torque to maximum torque. b Time to a 1 dN increase in torque. c Time to 25% of maximum torque. d Time to 90% of maximim torque. (From Ref.[77].)

aromatic oil. The blends were then cured at 150 C for 20 min. The cure properties of the blends are reported in Table 5 and the physical properties of the cured blends are reported in Table 6. As can be seen from Table 5, the blends made with the surface devulcanized reclaimed rubber crumb did not have cure characteristics that differed substantially from the blend made without including any of the reclaimed rubber (the series labeled ‘‘None’’). In fact, the blends made with the surface devulcanized reclaimed rubber crumb had cure characteristics that

were more similar to those made with no ground rubber than they were to those made with untreated whole tire reclaim rubber (the series labeled ‘‘Original’’). As can be seen from Table 6, the physical properties of some of the cured blends made with the surface devulcanized reclaimed rubber crumb samples were equivalent to those made with only virgin rubber. For instance, the 100% modulus, 300% modulus, and percent elongation measured in Examples 19 and 20 were very similar to those found in the control using only virgin rubber (the series labeled ‘‘None’’).

Table 6 Physical properties of devulcanized rubber Tensile strength (MPa)

Elongation (%)

100% Modulus (MPa)

300% Modulus (MPa)

19

15.4

747

1.23

4.45

20

15.3

701

1.33

4.93

21

13

775

1.10

3.39

22

13.7

798

1.11

3.46

23

15.1

814

1.09

3.54

24

15.2

778

1.11

3.88

25

16.1

764

1.19

4.49

26

15.4

738

1.21

4.43

27

15.2

789

1.13

3.88

28

14.8

791

1.11

3.74

Original

14.5

661

1.16

4.10

None

18.6

757

1.28

4.93

Example

(From Ref.[77].)

Rubber Devulcanization

CONCLUSIONS

2699

12.

Because of the large numbers of used tires existing in stockpiles it is desirable to recycle these tires back into new products. Unfortunately, simply grinding up the used tire and adding the resultant material into virgin rubber causes a significant drop in properties including reducing the strength and flexibility of the compound. The need for a commercially viable devulcanization process has existed for years and continues to be a goal. Devulcanization without rubber chain degradation offers the potential for recycling used tires back into new products without sacrificing performance. Chemical processes have been known for many years but are either ineffective in selectively cleaving cross-links or commercially not viable. Recent developments may eventually provide a true devulcanization process that is commercially viable. These include ultrasonic devulcanization, biotechnological devulcanization, and supercritical fluid devulcanization.

19. 20.

REFERENCES

21.

1. Warner, W.C. Methods of devulcanization. Rubber Chem. Technol. 1994, 67, 559. 2. Adhikari, B.; De, D.; Maiti, S. Reclamation and recycling of waste rubber. Prog. Polym. Sci. 2000, 25, 909. 3. Schnecko, H. Kautschuk Gummi Kunststoffe 1994, 47, 885. 4. Moore, C.G.; Trego, B.R. Structural characterization of vulcanizates. Part IV. Use of triphenylphosphine and sodium di-n-butyl phosphite to determine the structures of sulfur linkages in natural rubber, cis-1,4-polyisoprene, and ethylenepropylene rubber vulcanizate networks. J. Appl. Polym. Sci. 1964, 8, 1957. 5. Moore, C.G.; Trego, B.R. Structural characterization of vulcanizates. Part II. Use of triphenylphosphine to determine the structures of sulfur linkages in unaccelerated natural rubber-sulfur vulcanizate networks. J. Appl. Polym. Sci. 1961, 5, 299. 6. Tewksbury, L.B., Jr.; Howland, LH. Canadian Patent 2,469,529, 1949. 7. Rebmann, A. Swiss Patent 215,952, 1948. 8. Sverdrup, E.F.; Elgin, J.C. U.S. Patent 2,415,449, 1947. 9. Sverdrup, E.F.; Elgin, J.C. Canadian Patent 452,085, 1948. 10. Cotton, F.H.; Gibbons, P.A. U.S. Patent 2,408,296, 1946. 11. U.S. Patents 2,211,592, 1940; 2,414,145, 1947; 2,467,789, 1949; 2,471,866, 1949.

13.

14. 15. 16. 17. 18.

22.

23.

24.

25.

26. 27. 28.

29.

30.

31. 32.

U.S. Patents 2,193,624, 1940; 2,359,122, 1944; 2,333,810, 1943; 2,363,873, 1944; 2,372,584, 1945; 2,469,529, 1949. Webb, F.J.; Cook, W.S.; Albert, H.E.; Smith, G.E.P., Jr. Arylamine sulfide catalysts in reclaiming GR-S vulcanizates. Ind. Eng. Chem. 1954, 46, 1711. Le Beau, D.S. Rubber Chem. Technol. 1967, 40, 217. Knorr, K. Kautschuk Gummi Kunststoffe 1994, 47, 54. Selker, M.L.; Kemp, A.R. Sulfur linkage in vulcanized rubbers. Ind. Eng. Chem. 1944, 36, 16. Selker, M.L. Reaction of methyl iodide with sulfur compounds. Ind. Eng. Chem. 1948, 40, 1467. Selker, M.L.; Kemp, A.R. Sulfur linkage in vulcanized rubber acetone extraction of vulcanizates. Rubber Chem. Technol. 1949, 22, 8. Moore, C.G. J. Polym. Sci. 1958, 32, 503. Manik, S.P.; Banerjee, S. Sulfenamide accelerated sulfur vulcanization of natural rubber in presence and absence of dicumyl peroxide. Rubber Chem. Technol. 1970, 40, 1311. Gregg E.C., Jr.; Katrenick, S.E. Chemical structures in cis-1,4-polybutadiene vulcanizates. Model compound approach. Rubber Chem. Technol. 1970, 43, 549. Gregg E.C., Jr. Sulfur crosslinks in polybutadiene vulcanizates. Rubber Chem. Technol. 1969, 42, 1136. Studebaker, M.L. Lithium aluminum hydride analysis of sulfur—cured vulcanizates. Rubber Chem. Technol. 1970, 43, 624. Studebaker, M.L.; Nabors, L.G. Sulfur group analyses in natural rubber vulcanizates. Rubber Chem. Technol. 1959, 32, 941. Nichols, P.P. The scission of polysulfide crosslinks in scrap rubber particles through phase transfer catalysis. Rubber Chem. Technol. 1982, 55, 1499. Elgin, J.C. Canadian Patent 456,789, 1949. Sverdrup, E.F. U.S. Patent 2,494,593, 1949. Saville, B.; Watson, A.A. Structural characterization of sulfur-vulcanized rubber networks. Rubber Chem. Technol. 1967, 40, 100. Campbell, D.S. Structural characterization of vulcanizates part X. Thiol-disulfide interchange for cleaving disulfide crosslinks in natural rubber vulcanizates. Rubber Chem. Technol. 1970, 43, 210. Arnold, R.C.; Lien, A.P.; Alm, R.M. The action of lithium aluminum hydride on organic disulfides. J. Am. Chem. Soc. 1950, 72, 731. Farmer, R.H.; Ford, J.F.; Lyons, J.A. J. Appl. Chem. 1954, 4, 554. Porter, M.; Saville, B.; Watson, A.A. J. Chem. Soc. 1963, 346.

R

2700

33. Meyer, K.H.; Hohenemser, W. Helv. Chim. Acta 1935, 18, 1061. 34. Meyer, K.H.; Hohenemser, W. Contribution to the study of the vulcanization reaction. Rubber Chem. Technol. 1936, 9, 201. 35. Hilditch, T.P.; Smiles, S. J. Chem. Soc. 1907, 91, 1394. 36. Steinkopf, W.; Muller, S. Ber 1923, 56B, 1926. 37. Klinger, H.; Maason, A. Ann. 1898, 252, 241. 38. Brjuchonenko, A. Ber. 1899, 31, 3176. 39. Bateman, L.; Glazebrook, R.W.; Moore, C.G.; Porter, M.; Ross, G.W.; Saville, R.W. J. Chem. Soc. 1958, 2838. 40. Anderson, E., Jr. U.S. Patent 4,544,675, 1985. 41. Myers, R.D.; Nicholson, P.; MacLeod, J.B.; Moir, M.E. U.S. Patent 5,602,186, 1997. 42. Okamoto, H.; Inagaki, S.; Onauchi, Y.; Furukawa, J. Nippon Gomu Kyokaishi 1979, 52 (12), 774. 43. Nicholas, P.P. U.S. Patent 4,161,464, 1979. 44. Novotny, D.S.; Marsh, R.L.; Masters, F.C.; Tally, D.N. U.S. Patent 4,104,205, 1978. 45. Fix, S.R. Elastomerics 1980, 112 (6), 38. 46. Makrov, V.M.; Drozdovski, V.F. Reprocessing of Tires and Rubber Wastes; Ellis Horwood: New York, 1991. 47. Tyler, K.A.; Cerny, G.L. U.S. Patent 4,459,450, 1984. 48. Hunt, J.R.; Hall, D. U.S. Patent 5,362,759, 1994. 49. Pelofsky, A.H. U.S. Patent 3,725,314, 1973. 50. Okuda, M; Hatano, Y. Japanese Patent 62,121,741, 1987. 51. Isayev, A.I.; Chen, J.; Tukachinsky, A. Novel ultrasonic technology for devulcanization of waste rubbers. Rubber Chem. Technol. 1995, 68, 267. 52. Tukachinsky, A.; Schworm, D.; Isayev, A.I. Devulcanization of waste tire rubber by powerful ultrasound. Rubber Chem. Technol. 1996, 69, 92. 53. Levin, V.Yu.; Kim, S.H.; Isayev, A.I.; Massey, J.; Von Meerwall, E. Ultrasound devulcanization of sulfur vulcanized SBR: crosslink density and molecular mobility. Rubber Chem. Technol. 1996, 69, 104. 54. Isayev, A.I.; Yushanov, S.P.; Kim, S.H.; Levin, V.Yu. Rheo Acta 1996, 35, 616. 55. Isayev, A.I.; Kim, S.H.; Levin, V.Yu. Superior mechanical properties of reclaimed sbr with bimodal network. Rubber Chem. Technol. 1997, 70, 194. 56. Levin, V.Yu.; Kim, S.H.; Isayev, A.I. Vulcanization of ultrasonically devulcanized SBR elastomers. Rubber Chem. Technol. 1997, 70, 120. 57. Johnston, S.T.; Massey, J.; VonMeerwall, E.; Kim, S.H.; Levin, V.Yu.; Isayev, A.I. Ultrasound

Rubber Devulcanization

58. 59. 60.

61.

62.

63.

64.

65.

66. 67. 68.

69. 70.

71. 72.

73. 74. 75. 76. 77. 78. 79.

devulcanization of SBR: molecular mobility of gel and sol. Rubber Chem. Technol. 1997, 70, 183. Isayev, A.I. U.S. Patent 5,258,413, 1993. Isayev, A.I.; Chen, J. U.S. Patent 5,284,625, 1994. Tapale, M.; Isayev, A.I. Continuous ultrasonic devulcanization of unfilled NR vulcanizates. J. Appl. Polym. Sci. 1998, 70, 2007. Isayev, A.I.; Yushanov, S.P.; Chen, J. Ultrasonic devulcanization of rubber vulcanizates. I. Process model. J. Appl. Polym. Sci. 1996, 59, 803. Isayev, A.I.; Yushanov, S.P.; Chen, J. Ultrasonic devulcanization of rubber vulcanizates. II. Simulation and experiment. J. Appl. Polym. Sci. 1997, 59, 815. Isayev, A.I.; Yushanov, S.P.; Schworm, D.; Tukachinsky, A. Plast. Rubber Compos. Process Appl. 1996, 25, 1. Yushanov, S.P.; Isayev, A.I.; Levin, V.Yu. Percolation simulation of the network degradation during ultrasonic devulcanization. J. Polym. Sci. Phys. Ed. 1996, 34, 2409. Beckman, J.A.; Crane, G.; Kay, E.L.; Laman, J.R. Scrap tire disposal. Rubber Chem. Technol. 1974, 47, 597. Merseburg, T.H.; Neumann, W., et al. DEO 4042009, German Patent, Jun 6, 1992. Loffler, M. Deutsche Kautschuk Tagung dkt ’94, Jun 27–30, 1994. Straube, G.; Straube, E.; Neumann, W.; Ruckauf, H.; Forkmann, R.; Loffler, M. U.S. Patent 5,275,948, 1994. Tsuchii, A.; Suzuki, T.; Takeda, K. Appl. Environ. Microbiol. 1985, 50, 965. Tsuchii, A.; Takeda, K.; Tokiwa, Y. Degradation of the rubber in truck tires by a strain of Nocardia. Biodegradation 1997, 7, 405. Tsuchii, A.; Takeda, K. Appl. Environ. Microbiol. 1990, 56, 269. Loffler, M.; Straube, G.; Straube, E. Biohydrometall. Technol. Proc. Int. Biohydrometall. Symp. 1993, 2, 673. Linos, A.; Steinbuchel, A. Kautschuk Gummi Kunststoffe 1998, 51, 496. Romine, R.A.; Snowden-Swan, L.J. U.S. Patent 5597851, Jan 28, 1970. Fliermans, C.B.; Wicks, G.G. U.S. Patent 6,407,144, Jun 18, 2002. Hunt, L.K.; Kovalak, R.R. U.S. Patent 5891926, 1999. Benko, D.A.; Beers, R.N. U.S. Patent 6380269, 2002. Benko, D.A.; Beers, R.N. U.S. Patent 6387965, 2002. Benko, D.A.; Beers, R.N. U.S. Patent 6462099, 2002.

Scrubbers S S. Komar Kawatra Department of Chemical Engineering, Michigan Technological University, Houghton, Michigan, U.S.A.

INTRODUCTION Scrubbers are pollution control devices, which remove pollutants from gas streams, particularly from the combustion gases produced by facilities such as coalfired power plants. Scrubbers may use absorbents in slurry or solution form (wet scrubbers), or in powder form (dry scrubbers), with wet scrubbers being more often used. Modern scrubbers can consist of several cleaning steps such as dust separation, sulfur removal, capture of mercury, lead, or other heavy metals, and breakdown of nitrogen oxides. Scrubbers of the future will also need to capture and sequester carbon dioxide. Currently, the most common application of scrubbers is still the removal of sulfur, frequently combined with capture of fly-ash.

POLLUTANTS REMOVED BY SCRUBBERS Sulfur In sulfur scrubbers, the sulfur oxides that are produced during coal combustion are removed from the combustion gases using an absorptive chemical, such as calcium oxide. As nitrogen oxides are also produced during combustion, it is possible to develop combined desulfurization methods that can remove both sulfur dioxides and nitrogen oxides.[1] Postcombustion absorbers capture the sulfur in a solid form for disposal. There are two basic approaches to capturing the sulfur: 1) divert the combustion gases through a scrubber unit that is separate from the combustor and 2) design the combustor so that the sulfur absorbent can be injected directly along with the fuel and sulfur is captured in the combustion chamber. Sulfur scrubbers have been widely used by the industry for some time and have been highly developed. Separate sulfur scrubber units have the advantage that they can be used to retrofit existing plants.[2] There are several types of scrubber available, and many different chemical reactions have been used for the extraction of sulfur oxides, as shown in Table 1. Nitrogen Oxides Nitrogen oxides are one of the primary sources of acid rain that are emitted during combustion. There are Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120007955 Copyright # 2006 by Taylor & Francis. All rights reserved.

several different oxides produced, which are collectively referred to as NOx. These oxides are generated from two mechanisms:[16] 1. Thermal NOx. This is formed directly from the nitrogen in the combustion air because of the high temperatures and the presence of oxygen. Formation by this mechanism is strongly affected by temperature and residence time, with significant amounts produced at temperatures above 1200
C. Unfortunately, the factors that generally lead to complete combustion (high temperatures, long residence times, and thorough mixing of fuel and air) all tend to promote thermal NOx production. NOx from this source is usually controlled by some combination of reductions in flame temperatures, staged combustion, and flue-gas recirculation, which tend to slightly degrade the combustion efficiency. 2. Fuel NOx. This comes from combustion of nitrogen that is contained in the chemical structure of the fuel. This can account for up to 50% of the NOx from oil, and as much as 80% of the NOx from coal. Conversion of the nitrogen that was chemically combined with the fuel into NOx is strongly dependent on the fuel=air stoichiometry, but is relatively unaffected by combustion temperature, unlike thermal NOx, which is strongly affected by combustion temperature. Fuel NOx emissions are reduced by burning the fuel with low oxygen availability, causing the nitrogen to form N2 in preference to NOx. Once the NOx has been formed in the combustion process, if it is not removed it largely converts to NO2. This is the source of the brownish plume often seen from power plant stack discharges. Unlike sulfur, it is not easily reacted with absorbents to form a solid sludge. Instead, it is catalytically reduced with ammonia or urea to form N2 and water. This catalytic reduction requires injection of the reductant into the flue gases within a particular temperature window, with adequate residence time and catalytic surfaces to complete the reduction process. This can be integrated with scrubbers for other materials, such as sulfur and 2701

2702

Scrubbers

Table 1 Absorbents that have been studied for removal of sulfur oxides from coal combustion gases Absorbent CaCO3

Regenerable?

Product

No

CaSO3 or CaSO4

Ca(OH)2

No

CaSO3 or CaSO4

CaO

No

CaSO3 or CaSO4

Reference [3]

MgO

Yes, heat to release SO2

SO2

[4]

Na2SO3

Yes, heat to release SO2

SO2

[3]

CeO2 þ Al2O3

Yes, heat to release SO2

SO2

[5]

Ca(OH)2 þ fly-ash

No

Calcium silicate sulfates

[6]




CuO þ MnO2

Yes, heat with H2 at 200–560 C

H2S

[7]

Ca(OH)2 þ methanol

No

CaSO3 or CaSO4

[8]

CaO þ MgO

No

CaSO3 and MgSO3

NaHCO3

Yes, react with Ca(OH)2

CaSO4

[10]

Gaseous oxidation

Nothing to regenerate

H2SO4

[11]

CaCO3 þ NaCl

No

CaSO3 or CaSO4

[12]

Chalk

No

CaSO3 or CaSO4

[13]

Cement flue dust

No

Calcium silicate sulfates

Alkali þ Al2O3

Yes, heat with CO at 700–800
C

S, SO2

[14]

Ca–Mg Acetate

No

CaSO3 or CaSO4

[15]

[9]

Some are low-cost throw-away absorbents, while others are regenerable and convert the sulfur oxides into marketable products.

fly-ash, which provide the necessary residence time and control of process conditions. Carbon Dioxide A variety of methods for sequestration of CO2 from combustion of fossil fuels have been suggested in the literature to deal with the estimated 1583 million metric tons carbon equivalent annually released into the atmosphere in the U.S.A.[17,18] These methods include use of photoautotrophic organisms to convert it to biomass, deep-ocean disposal as a liquid, reaction with minerals to form stable carbonates, and industrial utilization. All of these sequestration methods can be made more efficient if the carbon dioxide is first separated from the flue gases and concentrated. The ideal approach would be for carbon dioxide to be absorbed from the flue gas, with the CO2-laden absorbent then regenerated by a rapid, low-energy process, releasing the CO2 in a purified, concentrated form. This CO2 could then be sequestered or utilized by any appropriate method. To date, all commercial plants for separation and concentration of CO2 use chemical absorption with a monoethanolamine solvent. This solvent was developed over 60 years ago to remove acid gases, such as CO2 and H2S, from natural gas streams. Monoethanolamine absorption is popular for the existing markets for high-purity CO2, because it produces a very highgrade product. The process has also been used to remove CO2 from flue gases, but it had to be modified

to resist solvent degradation and equipment corrosion. Also, to minimize reagent costs, the solvent strength was kept relatively low. This resulted in large equipment sizes and high regeneration energy requirements. Other CO2-absorption strategies have tended to make use of advanced technologies such as pressure-swing absorption, membrane technology, and hollow-fiber permeators. These are all highly efficient at producing a high-purity CO2 stream, but are expensive to apply on a large scale. As a result, the technologies that are receiving the most attention are not sufficiently economical for wholesale capture of CO2 from fossil fuel combustion, and a lower-cost absorbent is needed for flue-gas CO2 fixation applications. A survey of the technologies used for capturing sulfur oxides has indicated that certain low-cost commodity chemicals used for capturing sulfur could also be useful for capturing carbon dioxide.[19] One of these absorbents can be used to produce an aqueous solution which will efficiently capture carbon dioxide at temperatures less than approximately 25
C, as shown in Fig. 1. This solution can then be completely regenerated at only 100
C, releasing concentrated carbon dioxide that can be easily utilized or permanently sequestered.[20] Mercury/Heavy Metals Even though coal is not enriched in mercury relative to the components of the rest of the Earth’s crust,

Scrubbers

2703

S

Fig. 1 Capture of carbon dioxide by a 50 g=L solution of absorbent in a 500 ml batch absorber. The gas being treated was ambient air, at a carbon dioxide concentration of 403 mmoles=mole and a flow rate of 0.96 slpm. The fact that this absorbent can reduce CO2 levels to less than that of ambient air shows that a properly designed countercurrent scrubbing unit could use this solution to keep CO2 levels in a combustion gas stream from exceeding the normal levels present in the atmosphere. It should be noted that, under these conditions, the ability of pure water to absorb carbon dioxide was negligible.

combustion of coal is nevertheless a significant source of mercury emissions. This is because of the high volatility of metallic mercury and its compounds, which are efficiently vaporized by the heat of coal combustion. As a result, mercury is emitted as a trace component at levels of a few parts per billion in coal combustion gases. As mercury is highly toxic and tends to accumulate in the ecosystem, it has become clear that coal-fired power plants must prevent even these highly diluted mercury emissions from escaping into the environment. However, it is highly uneconomical to have a dedicated scrubber that captures only mercury. This is because the need to contact the entire volume of gases with mercury absorbent results in the need for an absorbent mass that is thousands of times greater than the mass of the mercury captured. It is much more feasible to ‘‘piggy-back’’ the mercury capture capacity with scrubbers for other, higherconcentration pollutants produced by the plant. Possibilities include 1. Injection of activated carbon particles into the flue gas to absorb the mercury or operation of the coal combustion to convert a portion of the coal into activated charcoal. The mercuryladen charcoal is then captured by the fly-ash removal system (which may be an electrostatic precipitator, filter, or dust scrubber). 2. Capture of mercury in the sludge from a sulfur oxide scrubber.

3. Capture and sequestration along with carbon dioxide. To capture mercury simultaneously with capture of sulfur oxides and carbon dioxide, it is necessary to ensure that the mercury is in a form that will be absorbed by the scrubber. The reducing environment of coal combustion results in a large fraction of the vaporized mercury being in the elemental state, which is unreactive, volatile, and can be transported worldwide over a period of years before it finally oxidizes and precipitates as a contaminant. It has been determined that, if mercury is in an oxidized state, it is readily captured by many existing scrubbers. This is because of oxidized mercury having a much higher solubility in water than elemental mercury, and a greatly increased reactivity with scrubbing agents. Therefore, the key to efficient capture of mercury is to ensure that it is in the oxidized state while it passes through the scrubber. This will require that the scrubber actively promote the oxidation of mercury. However, such oxidation cannot be achieved by the current generation of scrubbers.

WET SCRUBBERS Wet scrubbers use a slurry or solution of a sulfur absorbent in water, which is generally contacted with the flue gases using a scrubber tower such as that

2704

shown in Fig. 2. Flue gases rise through the tower through a falling spray of absorbent. The absorbent spray removes the sulfur oxides from the gases and collects in the base of the tower, where it is removed. The mist eliminator at the top of the column is to prevent fine droplets of absorbent from being carried up the stack, where they could cause corrosion or deposition problems in the stack or be released as particulate pollution. In general, when designing a wet scrubber for capturing pollutants from gases, there are several steps that should be taken into consideration: 1) selection of absorbent material; 2) equilibrium data evaluation; 3) estimation of operating data, including mass and energy balances; 4) absorption column selection; 5) column diameter calculation; 6) estimation of column height and=or number of plates or other transfer units; 7) determination of pressure drop through the column.[21] In the case of wet flue-gas scrubbers, the units are fairly standardized and much of this work has already been done. Absorbents

Scrubbers

are used as the sulfur absorbents, the following reactions are believed to occur:[24]  sulfur dioxide hydration: SO2 ðgÞ þ H2 OðlÞ ¼) H2 SO3 ðaqÞ

ð1Þ

H2 SO3 ðaqÞ ¼) Hþ ðaqÞ þ HSO3 ðaqÞ

ð2Þ

 lime reactions: CaðOHÞ2 ðsÞ ¼) Ca2þ ðaqÞ þ 2OH ðaqÞ Ca2þ ðaqÞ þ HSO3 ðaqÞ þ

ð3Þ

1 H2 OðlÞ 2

1 ¼) CaSO3  H2 OðsÞ þ Hþ ðaqÞ 2 Hþ ðaqÞ þ OH ðaqÞ ¼) H2 OðlÞ

ð4Þ ð5Þ

overall: CaðOHÞ2 ðsÞ þ SO2 ðgÞ 1 1 ¼) CaSO3  H2 OðsÞ þ H2 OðlÞ 2 2

ð6Þ

 limestone reactions: The most common absorbents are lime (calcium hydroxide) and limestone (calcium carbonate) slurries. Limestone is the preferred absorbent in many modern scrubbers, because of its low cost compared with lime and other absorbents.[22,23] However, lime is also used because of its higher reactivity, which allows it to absorb sulfur more rapidly. This makes it possible to use smaller scrubbers to treat a given quantity of gas when lime is the absorbent. When lime or limestone

Hþ ðaqÞ þ CaCO3 ðsÞ ¼) Ca2þ ðaqÞ þ HCO3 ðaqÞ 2þ

Ca ðaqÞ þ

HSO3 ðaqÞ

1 þ H2 OðlÞ 2

1 ¼) CaSO3  H2 OðsÞ þ Hþ ðaqÞ 2 Hþ ðaqÞ þ HCO3 ðaqÞ ¼) H2 CO3 ðaqÞ

ð7Þ

ð8Þ

ð9Þ

Fig. 2 Basic schematic of a wet scrubber column. Absorbent slurry percolates down through the packing, while the flue gases flow upward. The most common absorbents for sulfur oxides are limestone (calcium carbonate), lime (calcium hydroxide), and magnesium-enhanced lime made from dolomite. The sulfur-bearing sludge for some scrubbers is market-grade gypsum, but for other scrubbers it is a waste product that must be landfilled.

Scrubbers

2705

H2 CO3 ðaqÞ ¼) CO2 ðgÞ þ H2 OðlÞ

ð10Þ

overall: CaCO3 ðsÞ þ SO2 ðgÞ þ

1 H2 OðlÞ 2

1 ¼) CaSO3  H2 OðsÞ þ CO2 ðgÞ 2

ð11Þ

The solid product from each of these sets of reactions is primarily calcium sulfite hemihydrate (CaSO3 1 2 H2O), which has been confirmed by x-ray diffraction analysis of scrubber sludges.[25,26] A similar set of reactions collects sulfur trioxide (SO3) from the flue gases, forming gypsum (CaSO4  2H2O) as the solid product, but under normal boiler conditions sulfur trioxide makes up only about 0.5% of the total sulfur oxides, and so its removal is less important than the removal of sulfur dioxide.[24,27]

Equilibrium and Operating Data

where G is the molar gas flow rate (moles=s), y the mole fraction of sulfur dioxide in flue gas, kg the gas film mass transfer coefficient (moles=m2s), a the interfacial surface area (m2=m3), y* the equilibrium sulfur dioxide concentration at the gas=liquid interface, V the volume of the gas=liquid intermixing region (m3), and Ng the number of gas-phase transfer units (dimensionless). As not all of the parameters needed can be calculated in advance (kg can only be approximated, and a must be determined experimentally), the gas-phase mass transfer rate for a given scrubber design must be determined experimentally by operating the scrubber under conditions where y* approaches zero. Under these conditions, Eq. (12) can be integrated to give: V G

d½CaCO3  ¼ kc ð½Hþ   ½Hþ eq Þ  Sspc ½CaCO3  dt

ð14Þ

where [CaCO3] is the calcium carbonate concentration in the slurry (moles=l), [Hþ] the hydrogen ion concentration (moles=l), [Hþ]eq the equilibrium hydrogen ion concentration at the limestone surface (moles=l), Sspc the specific surface area of the limestone in the slurry, and kc the reaction rate constant. The limestone dissolution rates at various pH values and partial pressures of carbon dioxide are shown in Fig. 3. Slurry–Gas Contact

The factor that determines the size of a wet scrubber needed to produce a given capacity is the mass transfer rate of sulfur dioxide from the gas phase to the liquid phase, as shown in Eq. (1).[16] This mass transfer rate can be expressed as: Z Z kg a dy ¼ dV ð12Þ Ng ¼ y  y G

Ng ¼  lnð1  EÞ ¼ kg a

residence time; 5) wall effects; and 6) gas flow distribution.[16] In a limestone-based wet scrubber, the dissolution of calcium carbonate [Eq. (7)] is the primary rate-limiting reaction, because of the low solubility of calcium carbonate. The rate for the dissolution reaction can be expressed as:

ð13Þ

where E is the overall sulfur dioxide absorption fractional efficiency. The number of gas-phase transfer units that exist in a given column design depends on a number of factors, including: 1) slurry spray rate; 2) droplet size and distribution; 3) gas-phase residence time, which is controlled by the height of the spray zone; 4) liquid

The heart of a scrubber column is the slurry–gas contact zone, where gases are intimately combined with the absorbent slurry so that the pollutants can be captured by the reactions given previously. There are a number of possible methods for designing the contacting zone, including sprays, crossflow plates, baffle plates, counterflow plates, and packed columns. These all have the purpose of maximizing the interfacial area between the gas phase and the liquid phase, to allow rapid transport of gases across the surface.[29] A serious problem occurs in many wet scrubbers if the sulfur dioxide is partially oxidized to sulfur trioxide. In this case, the main precipitate is calcium sulfite hemihydrate, with up to 15% calcium sulfate in solid solution in the sulfite particles. If more than 15% is as calcium sulfate, then it can no longer precipitate with the sulfite crystals, and instead precipitates as separate crystals of gypsum. However, there is a shortage of gypsum seed crystals in the slurry in this situation, and so much of the gypsum crystallizes in the scrubber, particularly in the slurry–gas contact zone. This can rapidly plug the scrubber and must be avoided. Many plants prevent this problem by adding thiosulfate (S2O42) to the scrubber slurry as a reducing agent. This prevents the oxidation of sulfur dioxide and thus eliminates the formation of gypsum and the buildup of gypsum scale in the scrubber. A second solution to the plugging problem is to completely oxidize the calcium sulfite to gypsum, which provides more seed crystals for the gypsum and also prevents plugging. When the solid sludge is removed from the scrubber as unoxidized calcium sulfite, as is done in many older

S

2706

Scrubbers

Fig. 3 Limestone dissolution rates as a function of pH and carbon dioxide partial pressure, at a temperature of 25
C. (From Ref.[28].)

scrubbers, it has no market value and must be disposed of by landfilling. The more advanced scrubbers include an oxidation step, which converts the sulfite to sulfate, and the solid product is then gypsum.[23] If it is sufficiently pure, this synthetic gypsum can be marketed to make plaster, wallboard, cement, and other construction products. The major barrier to widespread marketing of scrubber sludge is that it contains a number of impurities, and is of uneven quality, which makes it unattractive for most purposes in its raw form. Potential users are therefore not eager to purchase the material, even with significant price breaks.[30,31] Typical quality requirements for gypsum for use in wallboard manufacture are given in Table 2. For the gypsum to be salable, it should meet or exceed these requirements. A complete circuit for an advanced scrubber is shown in Fig. 4, which includes oxidation of the sludge to form gypsum.[9] In this circuit, limestone is first reduced to a fine particle size by a grinding mill, producing a slurry. The slurry is then added to the absorber tank, and pumped into the scrubber tower. A portion of the descending absorbent is diverted back to the absorber tank, which provides more time for the sulfur dioxide and limestone to react. The remaining absorbent collects in the base of the tower, where it is oxidized by injected air while being recirculated in the lower portion of the scrubber. A portion of the absorbent is continuously drawn off to a hydrocyclone,

which separates the gypsum particles from the absorbent slurry, and returns the liquid to the scrubber. The most efficient wet-scrubber technology, from the standpoint of sulfur removal efficiency and equipment size, is the magnesium-enhanced lime process. This type of scrubber uses lime that contains up to 12% magnesium, which increases the absorption capacity of the lime to approximately 10–15 times that of the limestone scrubbers described previously. The principal advantage is that the magnesium-enhanced lime is soluble enough that the SO2 removal is governed by the degree of gas–liquid contact in the

Table 2 Impurity limits for gypsum for use in plaster or wallboard manufacture Impurity Fe2O3

Maximum wt.% 1.5

SiO2

1.0

MgO

0.1

K2 O

0.1

Na2O

0.04

Cl

0.01

CO3

1.5

SO2

0.25

Moisture

8.0

(From Ref.[32].)

Scrubbers

2707

S

Fig. 4 Circuit for a wet limestone scrubber, with oxidation of the solids to gypsum. The absorbent tank simplifies control of the process, while the hydrocyclone and filter remove coarse gypsum particles.

scrubber, and not by the degree of absorbent dissolution as is the case with limestone. A disadvantage is that the magnesium-enhanced lime is comparatively expensive because it must be calcined by heating before use. Also, the magnesium specifically inhibits the formation of gypsum, which helps to prevent plugging and scaling, but also results in the sludge being an unmarketable sulfite sludge instead of gypsum. Finally, the magnesium content of the sludge is too high for use in most synthetic gypsum markets.

Other Types of Wet Scrubber In addition to lime and limestone, a number of other absorbents have been used to improve the efficiency of sulfur removal or to recover the sulfur in a marketable form while regenerating the absorbent. Nextgeneration scrubbers are therefore under development to improve the efficiency and reduce the quantity of unmarketable waste products.[33–35] Several of the scrubber technologies that use other absorbents are listed here:

Power Consumption Example Assume a 180 MW boiler burning coal with 2.5% sulfur by weight, and a heating value of 12,767 BTU=lb. An appropriate limestone scrubber with forced oxidation would operate with a liquid=gas ratio of 130 gal. liquid per 1000 ft.3 of flue gas, and a pressure drop of 5 in. water. Such a scrubber would consume 2.549 MW to operate, with the breakdown as shown in Table 3. This corresponds to 1.42% of the total power output of the plant. Such a scrubber would remove approximately 93% of the sulfur, while consuming approximately 13,000 lb=hr of limestone being added at 35% solids.[16]


Dual alkali process. In this process, the absorption of the sulfur dioxide is first carried out using a solution of a sodium alkali, such as NaOH, Na2CO3, or Na2SO3. Because these are all very soluble in water, they can absorb the sulfur dioxide very rapidly and completely, and can be easily oxidized afterward. Also, the absorbent is a clear liquid rather than a slurry, and so the problems with scaling and plugging of the scrubber are much reduced.[10] The oxidized sulfur-bearing alkali is then circulated to a vessel where it reacts with lime or limestone, which precipitates the sulfur as calcium sulfate and regenerates the sodium alkali. A flow diagram of the

2708

Scrubbers

Table 3 Typical power requirements for limestone scrubber with forced oxidation of sludge to gypsum Average power (KW) Absorber system Oxidation air blower Absorber recirculation pump 1 Absorber recirculation pump 2 Absorber recirculation pump 3 Absorber recirculation tank agitators Mist eliminator wash water pump Misc. pumps and agitators Subtotal Dewatering area Vacuum pump for filter Filter wash water tank heater Reclaim water pump Hydrocyclone overflow pump Filter feed tank agitator Clarifier overflow sump pump Misc. pumps and agitators

375 312 367 380 60 19 24 1537 55 16 14 15 7 6 13

Subtotal

126

Reagent preparation Ball mill drive Mill product tank pump Limestone feed tank agitator Misc. pumps and agitators

220 5 25 6

Subtotal

256

Other systems General instrument air Differential induced draft fan power

50 580

Subtotal

630

Total power used by scrubber system

2549

(From Ref.[16].)

process is shown in Fig. 5. The dual-alkali process is reported to be stable and resistant to disturbances, and to be capable of removing more than 99% of the sulfur dioxide from flue gases.[10,36]
Wellman–Lord process. This is a regenerablesorbent process, producing SO2 gas, which can be sold for industrial uses. It uses a solution of sodium sulfite (Na2SO3), which absorbs SO2 and becomes a sodium bisulfite solution (NaHSO3). The sodium bisulfite is then decomposed in a forced circulation evaporator, releasing the SO2 at sufficiently high concentration to be compressed and sold as SO2 gas, or used for producing elemental sulfur or sulfuric acid.[3]
Magnesium oxide process. The magnesium oxide slurry is used to collect SO2, and the resulting magnesium sulfite is thermally treated to release the SO2 and regenerate the absorbent, as shown in Fig. 6. Like the Wellman–Lord process, this process

is relatively complex and has a capital cost about 14% higher than that of limestone scrubbers.[4] It is therefore only economically viable when there is a reliable market for the by-products.[38]

SPRAY-DRY SCRUBBERS Spray-dry scrubbers are an alternative to conventional wet scrubbers. In this type of scrubber, an alkaline slurry or solution is sprayed in fine droplets into a reaction vessel, along with the flue gas. The droplets rapidly react with the sulfur dioxide while drying to a fine powder of sulfite salts. This powder is entrained in the gas stream, and is carried to a dust precipitator where it is collected, as shown in Fig. 7. Most of the sulfur dioxide is collected in liquid-phase reactions while the droplets are drying, but 10–15% additional sulfur dioxide can be absorbed in gas=solid reactions, as the absorbent powder is swept through the ductwork and particulate collector. These are cocurrent devices, and so the limestone utilization and sulfur removal efficiency are inherently lower than those of countercurrent devices such as wet scrubbers. Partial recycle of the sorbent is often used to improve the sorbent utilization. It is typical to install the spray-dryer before the plant fly-ash collector, so that the existing dust control equipment can be used to collect the used absorbent. Slaked lime [Ca(OH)2] is the most common absorbent, although sodium carbonate (Na2CO3) is used in some plants. Spray-dryers have also been used with regenerable magnesium oxide absorbent.[4] Spray-dryers are simpler and more compact than conventional wet scrubbers and have a lower capital and operating cost. Also, they do not produce large quantities of wastewater, and the spent absorbent is dry, thereby eliminating the need for thickening and filtration of the sludge. However, if the same dust precipitator is used for both the fly-ash and the spray-dryer product, the mixture of fly-ash and spent absorbent that they produce is unmarketable, and must be disposed of. Also, they require more expensive absorbents than conventional wet scrubbers. They are most suitable for retrofitting small plants that burn medium-sulfur coals, where capital costs and space restrictions are more of a consideration.[38]

VENTURI SCRUBBERS Venturi scrubbers are mainly used for collecting fine particulates (such as fly-ash) from gas streams,[39] but they have also been adapted for absorption of sulfur dioxide.[40] These units are mechanically very simple, consisting of a reducing inlet with liquid sprays,

Scrubbers

2709

S

Fig. 5 Flow diagram of the dual-alkali scrubber process, using lime to regenerate the sodium alkali. The clarified liquid from the thickener contains dissolved calcium sulfate, which would produce calcium carbonate scale in the scrubber when it contacts the carbon dioxide in the flue gas. It is therefore precipitated in the softening reactor by a combination of carbon dioxide and sodium carbonate, and the resulting calcium carbonate precipitate is removed by the hydrocyclone.

a narrow throat where the gas=liquid contact occurs, and an expanding region, as shown in Fig. 8. The flue gases are injected into the venturi at the contracting inlet, along with an absorbent, such as lime slurry. The gas accelerates to high speed as it enters the throat, and atomizes the absorbent, providing good gas=liquid contact. The gas and the atomized liquid then expand and slow in the expander region and are then diverted to a mist eliminator to separate the liquid droplets from the scrubbed gases. Venturi scrubbers have lower capital cost than other types of scrubber because they are mechanically simple; but they have a high energy consumption because of the need for pressurizing the gas to force it through the venturi. They also double as a fly-ash collection device, and so there is no need for separate scrubbers and fly-ash collectors when these units are used.[40] Because venturi scrubbers are cocurrent devices, with both the flue gas and absorbent traveling in the same direction, they cannot remove sulfur dioxide as completely as countercurrent devices, such as wet-scrubber towers, do.[29,41] They also produce a wet mixture of

fly-ash and alkaline absorbent, which is unmarketable and can form a cement-like substance upon disposal. A number of theoretical studies of venturi performance have been made to produce theoretical models that can predict performance from first principles. One of the key areas of uncertainty has been the droplet size formed by the venturi. Typically, this is estimated using the Nukiyama and Tanasawa equation to estimate the surface-mean droplet diameter:[39]

D0 ¼

pffiffiffiffiffiffi 1920 sL pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi V0 rL =62:3

mL ffi þ 75:4 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi sL rL =62:3

!0:45 

1000QL QG

1:5

ð15Þ

where D0 is the drop diameter (mm), V0 the gas velocity (ft=s), sL the liquid surface tension (dynes=cm), rL the liquid density (lb=ft3), mL the liquid viscosity (centipoises), QL the liquid flow rate (ft3=s), and QG the gas flow rate (ft3=s).

2710

Scrubbers

Fig. 6 Magnesium oxide regeneration and sulfur dioxide recovery section for a magnesium oxide scrubber. (From Ref.[37].)

This equation unfortunately has a number of severe shortcomings, including its lack of dimensional homogeneity and the fact that the effect of nozzle size on the drop diameter is not defined. It was originally derived for small diameter atomizing nozzles, and drop sizes reported by various investigators have varied from the equation prediction by as much as two to three times, and so this approach has not been highly successful.

In general, the ability of these scrubbers to collect particles improves, as the energy input to the system increases, and so collection of very fine particles requires increased flow rates and operating pressures. This has lead to a fairly useful scrubber design method based on the dissipation of power in the gas–liquid contactor. A number of studies have concluded that the collection efficiency of a scrubber on a given dust

Fig. 7 Basic configuration for a single-stage spray-dry sulfur absorber, with no recycle of absorbent.

Scrubbers

2711

S

Fig. 8 Schematic diagram of a venturi scrubber with a cyclonic mist eliminator. The wetted inlet, throat, and expander make up the venturi section.

is dependent only on the contacting power, with only minor effects from the size or geometry of the scrubber.[39] The contacting power is commonly given in units of MJ=1000 m3, and may conveniently be broken into three parts: 1. Gas-phase contacting power, PG. In SI units, this is equal to the effective friction loss of the gas as it flows through the contactor, in kPa. 2. Liquid-phase contacting power. This is given by the equation PL ¼ pf(QL=QG), where pf is the nozzle feed pressure in kPa, and QL and QG are the liquid and gas flow rates, respectively, in m3=s. 3. Additional power supplied separately, PM, such as by a power-driven rotor. The total contacting power is then PT ¼ PG þ PL þ PM. The collection efficiency, expressed as the

number of equivalent contacting units (Nt), then becomes: N t ¼ aPTg

ð16Þ

where a and g are empirical constants that depend on the character of the particles being collected.[39]

DRY SORBENT INJECTION Dry sorbent injection is very similar to the use of spray-dryers, except that the sorbent is injected as a dry powder rather than as an atomized slurry.[42,43] The most common sorbent is hydrated lime, but other sorbents can also be used. The sorbent is usually injected directly into existing ductwork, and so the amount of space required is negligible compared with that of other flue-gas desulfurization processes. This makes dry sorbent injection a very low-cost option.

2712

Scrubbers

Fig. 9 Integrated dry injection process utilizing both calcium hydroxide and sodium bicarbonate to reduce sulfur dioxide and nitrogen oxide emissions. Calcium hydroxide is added to the hot flue gases before they are cooled in the economizer and combustion air heater, while sodium bicarbonate is added to the cooled gases before they enter the electrostatic precipitator. (From Ref.[44].)

Unfortunately, the reactivity of dry absorbents is much lower than that of absorbent slurries or solutions, and so dry sorbent injection is only suitable for applications where less than 70% of the sulfur dioxide needs to be removed from the flue gases.[38] In cases where hydrated lime does not remove enough of the sulfur dioxide, but economics makes more efficient absorbents impractical, a two-stage absorbent injection scheme can be used, as shown in Fig. 9. Here, the relatively low-cost calcium hydroxide is used to remove

the bulk of the sulfur dioxide, and is then followed by a spray of more effective (but higher-cost) sodium bicarbonate. In addition to further reduction of the sulfur dioxide content, the sodium bicarbonate spray also reduces the content of nitrogen oxides. It is also possible to use limestone in dry sorbent injection, as is done in the limestone injection multistage burner (LIMB) system (Fig. 10). In this system, pulverized limestone is injected into the boiler directly, where the temperature is high enough to flash-calcine

Fig. 10 The basic integrated LIMB dry sorbent injection system.

Scrubbers

2713

S

Fig. 11 Schematic of the SNRB catalytic baghouse. (From Ref.[46].)

the CaCO3 to CaO. The CaO dust is carried off with the flue gases until the temperature drops enough for CaSO3 to become stable. The CaO then captures the SO2 to form CaSO3, which is removed by the electrostatic precipitators, along with the fly-ash. If necessary, a second, wet SO2 scrubber is used to finish the sulfur removal.[37] A related technology, the SOx-NOx-Rox BoxTM, or SNRB, also destroys nitrogen oxides while removing sulfur dioxide.[45,46] This unit is a replacement for electrostatic precipitators and is installed in the fluegas stream between the economizer and the combustion air heater, where the flue gases are still hot. An alkali is injected as sulfur absorbent, along with anhydrous ammonia. The gases then enter the hightemperature catalytic baghouse, which consists of catalyst-impregnated ceramic filter ‘‘bags,’’ as shown in Fig. 11. The ceramic filters capture and remove the sulfur-loaded absorbent, and the catalyst in the filters catalyzes the reduction of the nitrogen oxides by the ammonia, producing nitrogen gas and water. SNOX SYSTEM The SNOX system is designed for removing both sulfur oxides and nitrogen oxides from flue gases, and is

unusual in that it does not use an alkali as an absorbent to collect the sulfur dioxide. Instead, it oxidizes the sulfur dioxide to sulfur trioxide and uses a special condenser to collect the sulfur trioxide as marketable sulfuric acid. This is combined with catalytic destruction of the nitrogen oxides with ammonia, producing the overall circuit shown in Fig. 12. The system is reported to be capable of better than 90% removal of both SO2 and NOx, while producing sulfuric acid at 93–95% concentration.[11]

RELATIVE COSTS OF SCRUBBERS A comparison of the estimated sulfur removal ability and costs for various postcombustion processes is given in Table 4. It can be seen from this table that there is considerable variation in costs, depending on factors such as the size of the plant, fraction of the time that the plant operates at full capacity, ability to retrofit existing facilities, differences in fuel prices, sulfur content of the coal, and ability of existing equipment such as electrostatic precipitators to cope with changes in the process. Other techniques for reducing sulfur emissions, such as fuel switching or advanced combustion technologies, are included for comparison.

2714

Scrubbers

Fig. 12 Diagram of the SNOX system for production of sulfuric acid from flue gases while simultaneously destroying nitrogen oxides. (From Ref.[11].)

gasification combined cycle (IGCC) and fluidized bed combustion are drastically more expensive than scrubbers, but it should be remembered that the comparison is between retrofitting existing plants (for the scrubbers) and building new plants (for the IGCC and fluidized bed). The price differential between the two types of pollution control process is much less for completely new plants or for repowering existing plants.[38] In general, the performance of a scrubber process will be improved if the sulfur content of the feed coal

In general, fuel switching is currently the cheapest option at the current price difference of $10=ton between high-sulfur and low-sulfur coals. However, as low-sulfur coal prices increase, this will become less economical. Dry processes are the next most economical option, but they may not be able to meet emission control standards. The various types of scrubber have broadly similar costs, and the choice will depend on the specific plant under consideration. It appears from Table 4 that advanced combustion technologies such as integrated Table 4 Cost comparison of various SO2 control options

Technology Fuel switching=blending

% SO2 removal

Operating costs (mills/kWhr)

Capital costs ($/kW) Low

Base

High

Low

Base

High

2–80

20

28

30

3

6

13

Most sensitive parameters %S, CF, FPD, SCA

Lime=limestone FGD

90

120

240

520

5

16

150

MW, RF, CF, %S

Lime spray drying with existing ESP

76

70

170

540

3

10

130

MW, RF, CF, %S, SCA

Lime spray drying with new fabric filter

86

140

240

620

5

13

150

MW, RF, CF, %S

Integrated gasification combined cycle

95

1710

2100

2800

44

91

605

MW, CF, heat rate

Atmospheric fluid bed combustion

90

1360

1680

2250

40

80

480

MW, CF, heat rate

Dry sorbent injection

70

25

50

110

2

6

40

MW, CF, %S, SCA

CF, capacity factor; FPD, fuel price differential; MW, size of plant(MW); %S, fuel sulfur content; RF, retrofit factor; SCA, specific collection area of electrostatic precipitator (ESP). (From Ref.[47].)

Scrubbers

is reduced. A lower sulfur feed provides the following benefits: 1. Gas-scrubbing processes require less absorbent. 2. A lower percentage of the total SO2 can be removed and still reach emission targets, allowing the less-efficient, lower-cost technologies to be used. 3. The quantity of desulfurization waste which must be disposed of is reduced. It is therefore beneficial to pretreat the coal to remove as much sulfur as is practical before combustion, so that the expense of postcombustion desulfurization can be reduced. Combined with the other benefits of coal cleaning, it is evident that precombustion coal treatment is valuable even when it is not sufficient to completely desulfurize the coal by itself.

CONCLUSIONS A wide range of postcombustion technologies are available for reducing the emissions of sulfur and other pollutants from coal-fired power plants. These include both wet and dry scrubbers, which produce either sulfur-bearing wastes or marketable by-products. In general, the less-expensive techniques are also the least effective for reducing emissions, and so the choice of which process to use depends on the quantity of pollutants in the fuel, the types of pollutants, and the emissions target, as well as on many other factors such as plant size, whether it is a retrofit of an existing plant or a new plant, availability of cleaner fuel, cost of waste disposal, and availability of markets for by-products. Spray-dryers or dry sorbent injection are a good choice if there is little space in the plant for installing new equipment, there is no market for by-products, and the plant emissions can be brought into regulatory compliance with relatively modest reductions in emissions. Wet scrubbers are well suited for plants with significant room for expansion, and which need to remove a large proportion of the pollutants from their emissions. Scrubbers with throw-away absorbents are the best choice when waste disposal is cheap, while those with regenerable absorbents and=or marketable by-products are the best choice when waste disposal is expensive and markets for by-products are nearby. Precombustion and postcombustion pollutant removal technologies should not be considered to be in competition. Rather, they are complementary technologies that should be used together for maximum benefit. Many precombustion processes are low in cost, but cannot remove all of the pollutants, while

2715

postcombustion treatment can often capture nearly all of the emissions from the combustion gases, but their costs increase as the coal pollutant content increases. By using both types of process together, the maximum reduction of emissions can be achieved at the minimum cost.

REFERENCES 1. Anonymous. Will combined SO2=NOx processes find a niche in the market? Power 1990, 134, 26–28. 2. Ellison, W. Today’s FGD systems satisfy retrofit needs for 1990s. Power 1991, 135, 101–106. 3. Couch, G.R. Power from Coal—Where to Remove Impurities? Report No. IEACR=82; IEA Coal Research: London, 1995. 4. Burnett, T.A.; Wells, W.L. Conceptual design and economics of and improved magnesium oxide flue gas desulfurization process. In Flue Gas Desulfurization; Hudson, Wells, Eds.; American Chemical Society: Washington, DC, 1982; 381–411. 5. Hedges, S.W.; Yeh, J.T. Kinetics of sulfur dioxide uptake on supported cerium oxide sorbents. Environ. Progr. 1992, 11 (2), 98–103. 6. Kind, K.K.; Wasserman, P.D.; Rochelle, G.T. Effects of salts on preparation and use of calcium silicates for flue gas desulfurization. Environ. Sci. Technol. 1994, 28 (2), 277–283. 7. Bjornbom, E.N.; Druesne, S.; Zwinkels, M.F.M.; Jaras, S.G. Study on the regeneration of copper– manganese sorbent for removal of sulfur dioxide from flue gases. Ind. Eng. Chem. Res. 1995, 34 (5), 1853–1858. 8. Withum, J.A.; Yoon, H. Treatment of hydrated lime with methanol for in-duct desulfurization sorbent improvement. Environ. Sci. Technol. 1989, 23 (7), 821–827. 9. Makansi, J. Controlling SO2 emissions. Power 1993, 137, 23–56. 10. Valencia, J.A. The limestone dual alkali process for flue gas desulfurization. In Flue Gas Desulfurization; Hudson, Wells, Eds.; American Chemical Society: Washington, DC, 1982; 325–347. 11. Durrani, S.M. The SNOX process: a success story. Environ. Sci. Technol. 1994, 28 (2), 88A–90A. 12. Bulewicz, E.M.; Janicka, E. Catalytic effect of NaCl on flue-gas desulphurization by limestonebased sorbents during the FB combustion of coal. J. Inst. Energy 1990, 63, 124–130. 13. Dennis, J.S.; Hayhurst, A.N. Alternative sorbents for flue-gas desulphurization, especially in fluidized-bed combustors. J. Inst. Energy 1989, 62, 202–207.

S

2716

14. Gavalas, G.R.; Edelstein, S.; FlytzaniStephanopoulous, M.; Weston, T.A. Alkalialumina sorbents for high-temperature removal of SO2. AIChE J. 1987, 33 (2), 258–266. 15. Levendis, Y.A.; Zhu, W.; Wise, D.L.; Simons, G.A. Effectiveness of calcium magnesium acetate as an SOx sorbent in coal combustion. AIChE J. 1993, 39 (5), 761–773. 16. Babcock; Wilcox. Nitrogen oxides control, Chapter 34, Sulfur dioxide control, Chapter 35. In Steam, Its Generation and Use, 40th Ed.; The Babcock and Wilcox Company, 1992; 35-1–35-15. 17. EIA. Emissions of Greenhouse Gases in the United States 2000. Energy Information Administration, DOE=EIA-0573 (2000), 2001. ftp:==ftp. eia.doe.gov=pub=oiaf=1605=cdrom=pdf=ggrpt= 057300.pdf. 18. Herzog, H.; Drake, E.; Adams, E. CO2 Capture, Reuse, and Storage Technologies for Mitigating Global Climate Changes. White Paper—Final Report, DOE Order No. DE-AF22-96PC01257, 1997. 19. Kawatra, S.K.; Eisele, T.C. Coal Desulfurization; Taylor and Francis: London, 2001. 20. Seidell, A. Solubilities of Inorganic and Organic Compounds: A Compilation of Quantitative Solubility Data from the Periodical Literature; D. Van Nostrand Company: New York; 1919. 21. Theodore, L.; Buonicore, A.J.; McKenna, J.D.; Kugelman, I.J.; Jeris, J.S.; Santoleri, J.J.; McGowan, T.F. Waste management, Section 25. In Perry’s Chemical Engineers’ Handbook, 7th Ed.; Perry, R.H., Green, D.W., Maloney, J.O., Eds.; McGraw-Hill: New York, 1999. 22. Bryan, R.R.; Smith, A.A.; Farmer, C. Texas plant demonstrates viability of coal option. Power 1993, 137, 57–64. 23. Dettmer, R. Sans sulphur: the Drax FGD project. IEE Rev. 1994, 40 (2), 88–89. 24. Fellman, R.T.; Cheremisinoff, P.N. Lime=limestone scrubbing for SO2 removal, Chapter 12. In Air Pollution Control and Design for Industry; Cheremisinoff, Ed.; Marcel Dekker: New York, 1993; 339–357. 25. Kawatra, S.K.; Eisele, T.C. Separation of Flue-Gas Scrubber Sludge into Marketable Products, First Quarterly Technical Progress Report; U.S. Department of Energy, Pittsburgh Energy Technology Center, Contract No. DE-FG2293PC93214, 1993. 26. Kawatra, S.K.; Eisele, T.C.; Banerjee, D.D. Recovery of gypsum and limestone from scrubber sludge by water-only cyclone and conventional froth flotation. In New Remediation Technology in the Changing Environmental Arena; Scheiner, et al., Eds.; Society for Mining,

Scrubbers

27.

28.

29.

30.

31.

32.

33. 34.

35.

36.

37.

38. 39.

Metallurgy, and Exploration, Inc.: Littleton, CO, 1995; 99–104. Eisele, T.C.; Kawatra, S.K. Separation of the components of flue gas scrubber sludge by froth flotation. In Proceedings of the 19th International Mineral Processing Congress; Vol. 4, Chapter 33; Society of Mining, Metallurgy, and Exploration: Littleton, CO, 1995; 163–166. Plummer, L.N.; Wigley, T.M.L.; Parkhurst, D.L. The kinetics of calcite dissolution in CO2–water systems at 5
C to 60
C and 0.0 to 1.0 atm CO2. Am. J. Sci. 1978, 278, 179–216. Fair, J.D.; Steinmeyer, D.E.; Penney, W.R.; Crocker, B.B. Gas absorption and gas–liquid system design, Section 14. In Perry’s Chemical Engineers’ Handbook, 7th Ed.; Perry, R.H., Green, D.W., Maloney, J.O., Eds.; McGraw-Hill: New York, 1999. Ellison, W.; Hammer, E.L. FGD gypsum use penetrates U.S. wallboard industry. Power 1988, 132, 29–33. Van der Brugghen, F.W.; Koppins-Odink, J.M. Flue Gas cleaning in power stations in the Netherlands. In Proceedings: First Combined Flue Gas Desulfurization and Dry SO2 Control Symposium; Electric Power Research Institute, Paper 1-5, 1989, 1=69–1=94. Shoop, K.J.; Blystone, S.S.; Kawatra, S.K. Zeta potential measurements of the components of wet flue-gas scrubber sludge. In Proceedings of the 1996 SME Annual Meeting, Phoenix, AZ, Preprint No. 96–99, 1996. Feeney, S. Upgrade scrubbers to improve performance. Power 1995, 139, 32–37. Anonymous. DOE kicks off clean air program to boost performance of scrubbers. Fossil Energy Rev.: U.S. Department of Energy, 1992; October– December, 18–19. Anonymous. Southern company begins testing 2nd generation clean coal scrubber. Fossil Energy Rev.: U.S. Department of Energy, 1992; October– December, 28–29, 39. Hodges, G.J.; Roset, G.K.; Woodland, L.R.; Stevenson, J.A. Dual-alkali scrubbing at stillwater mining company. Mining Eng. 1992, 44 (10), 1269–1271. Chiang, S.-H.; Cobb, J.T., Jr. Coal conversion processes (desulfurization). In Kirk-Othmer Encyclopedia of Chemical Technology, 4th Ed.; John Wiley & Sons, 1993; Vol. 6, 511–540. Vernon, J.L.; Jones, T. Sulphur and Coal, Report No. IEACR=57; IEA Coal Research: London, 1993. Pell, M.; Dunson, J.B. Gas–solid operations and equipment, Section 17. In Perry’s Chemical Engineers’ Handbook, 7th Ed.; Perry, R.H., Green, D.W., Maloney, J.O., Eds.; McGraw-Hill: New York, 1999.

Scrubbers

40. Brady, J.D.; Legatski, L.K. Venturi scrubbers, Chapter 11. In Air Pollution Control and Design for Industry; Cheremisinoff, Ed.; Marcel Dekker: New York, 1993; 339–357. 41. Porter, H.F. Solids drying and gas–solid systems. In Perry’s Chemical Engineers’ Handbook, 6th Ed.; McGraw-Hill, 1984; 20=93. 42. Yeh, J.T.; Demski, R.J.; Joubert, J.I. Control of SO2 emissions by dry sorbent injection. In Flue Gas Desulfurization; Hudson, Wells, Eds.; American Chemical Society: Washington, DC, 1982; 349–368. 43. Stouffer, M.R.; Rosenhoover, W.A.; Withum, J.A. Advanced coolside desulfurization process. Environ. Progr. 1993, 12 (2), 133–139.

2717

44. Helfritch, D.; Bortz, S.; Beittel, R.; Bergman, P.; Toole-O’Neil, B. Combined SO2 and NOx removal by means of dry sorbent injection. Environ. Progr. 1992, 11 (1), 7–10. 45. Makansi, J. DOE’s clean coal program: what industry has learned. Power 1992, 136, 56–61. 46. Kudlac, G.A.; Farthing, G.A.; Szymanski, T.; Corbett, R. SNRB catalytic baghouse laboratory pilot testing. Environ. Progr. 1992, 11 (1), 33–38. 47. White, D.M.; Maibodi, M. Assessment of Control Technologies fro Reducing Emissions of SO2 and NOx from Existing Coal-fired Utility Boilers, Project Summary, Environmental Protection Agency, Report No. EPA=600=S7-90=018, 1991.

S

Six Sigma Design: An Overview of Design for Six Sigma (DFSS)

S

Sean A. Curran Kwok-Wai Lem Steve Sund Mina Gabriel Honeywell International, Inc., Morristown, New Jersey, U.S.A.

INTRODUCTION

What Is Sigma?

Traditional six sigma methodology has become a standard process optimization tool for process industries. However, it has become clear that the ‘‘Holy Grail’’ of six sigma, 3.4 defects per million opportunities (DPMO), is simply unachievable after the fact. Consequently, there has been a growing movement to implement six sigma design usually called design for six sigma (DFSS). This methodology begins with defining customer needs and leads to the development of robust processes to deliver those needs. In this entry, we introduce the DFSS approach to product=process development. This is based on a systematic application of powerful tools to define customer requirements and relate them to producer capability. This is followed by a detailed discussion of the various tools and their use. We show how the linkage of customer needs to product requirements and subsequent linkage of product requirements to process requirements drive to the development of robust processes.

The simple answer is that it is a measure of variability. Generally, in chemical processing we deal with normally distributed variables such as temperatures, flow rates, pressures, purity, mechanical strength, etc. These variables will have a nominal value or a specific setting and some level of variation. The average of all values is referred to as the mean (m) and the spread is defined by the standard deviation, sigma (s). We can describe these distributions statistically and predict the probability of a certain value occurring as illustrated in Fig. 1. In cases where the variables are not normally distributed, such as high purity materials where the distribution becomes highly skewed as one approaches the physical limits of 100% purity or 0% impurity, the data must be handled appropriately. The correct distribution can be modeled, or the data can be transformed into a normal distribution by using averages of multiple measurements. The central limit theorem predicts that such averages will be normally distributed. Strictly speaking, one cannot know the true mean and standard deviation unless the entire population is measured. In practice, we use a sample to predict the true mean and standard deviation. Consequently, there is always a finite probability that the sample chosen does not represent the true population.

DEFINITIONS AND THEORY What Does DFSS Mean? Ultimately any supplier’s objective is to deliver on customer expectations 100% of the time.[1,2] DFSS is a systematic approach to develop processes that are capable of delivering on those expectations.[3–5] Ideally, all products would behave the same way all the time. In the real world, errors (i.e., defects) can and do occur because all processes exhibit variation. Consequently, all products manufactured via those processes will exhibit variation.[6] DFSS is a systematic methodology to predict that variability and the ability to meet customer needs. The methodology applies a suite of tools to define customer expectations and couples them directly with manufacturing capabilities so that the customer is always satisfied, and the manufacturer can earn adequate returns. Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120016185 Copyright # 2006 by Taylor & Francis. All rights reserved.

What Is Six Sigma? The discussion of sigma (s) given earlier deals with the natural variation in the process.[1,7–10] Consequently, we denote the measurement of s as a measure of the voice of the process. This is clearly an internally focused assessment of process capability, i.e., the ‘‘voice of the process.’’ The probability of a defect in the process, i.e., failure to meet customer expectations, is commonly defined in terms of DPMO. The focus of DFSS is to achieve a minimum of six standard deviations between the mean and the nearest specification (
6s), corresponding to 3.4 DPMO. Because the specifications are set to meet customer needs, six sigma in this 2719

2720

Six Sigma Design: An Overview of Design for Six Sigma (DFSS)

Fig. 1 Normal probability curve. (View this art in color at www.dekker.com.)

context is the ratio between the voice of the customer and the voice of the process. This ratio is the sigma score, commonly referred to as ‘‘Z’’ score and is calculated as shown in Eq. (1), where m is the process mean, s is the process standard deviation, USL is the upper specification limit, LSL is the lower specification limit, and the sigma score (Z) is the smaller of the calculated values. Brief inspection of the equation shows that Z is dimensionless. Z ¼

min ðUSL  m; m  LSLÞ s

ð1Þ

In effect this calculation normalizes all processes to a number of sigmas rather than absolute values. The relationship between DPMO and Z score is illustrated in Fig. 2. Z refers to the white area under the curve and the shaded area shows the area where there is a probability of failure. DPMO is the integration of the shaded areas, i.e., the proportion of the results beyond the calculated Z value. Obviously, as Z increases the defective part of the distribution shrinks. The exact probability associated with a specific Z score can be easily obtained from Z score tables or calculated with common software packages such as ExcelÕ or statistical packages such as MinitabÕ.a In the case of Excel, such calculations are not part of a standard package, but macros can easily be written to perform the needed calculations.

cost. Infinite repetitions of product enhancements are only useful if they provide value to the customer. Further improvements add cost, without adding value. Brief inspection of a Z score table shows that a Z score of 6 actually corresponds to a probability of 109 for a failure. However, most studies of process capabilities are relatively short term in nature. After all, the supplier’s interest is in making product, not continuously evaluating the process. Experience has shown that over time, processes will drift off the original mean. Typically this drift amounts to 1.5s (for example, see p. 512 of Ref.[5]), so one compensates for the long-term variation by subtracting 1.5 from the short-term sigma score.[1,5] In effect, a short-term 6s process becomes a long-term 4.5s process, and this corresponds to the often quoted 3.4 DPMO.

What is a CTQ? The acronym CTQ stands for critical to quality. At the highest level, CTQs are those benefits that accrue to the customer when they use a product=process. Top level CTQs are not specific to either product or process.

Why Six Sigma? Naturally, all producers wish to minimize their production cost while maintaining quality. Likewise, all customers desire 100% defect free products at minimum

a ExcelÕ is a registered trademark of Microsoft, Inc. and MinitabÕ is a registered trademark of MinitabÕ, Inc.

Fig. 2 Relationship of Z score with DPMO. (View this art in color at www.dekker.com.)

Six Sigma Design: An Overview of Design for Six Sigma (DFSS)

These are the performance attributes that the customer is actually paying for such as load bearing capacity, stability in a harsh environment, etc. In the following section, we discuss requirement flow down. As those requirements flow down, we establish lower level CTQs. These are those process and=or product requirements that are critical for delivering the customer CTQs, i.e., those parameters one must control in order to deliver the desired performance.

DFSS TOOLS Quality Function Deployment Once the customer needs (top level CTQs) have been defined a DFSS team needs to determine how process parameters that can be addressed, such as raw materials, process control parameters, etc. are related to the ability to deliver customer CTQs. This is done by generating a series of quality functional deployment (QFD) matrices as sketched in Fig. 3. There is no hard and fast rule on how many levels a QFD needs. Frequently, polymer and chemical QFDs have only three levels as there are no subassemblies, and consequently, there is no need for a ‘‘parts’’ QFD because process specifications relate directly to the process design CTQs.

2721

We discuss QFD at some length because it is one of the pillars of a successful DFSS effort. A proper QFD relates customer CTQs to product features and assesses various competing alternatives for delivering the desired features.[4,5] Quality functional deployment then relates product features to product design, product design to process design, and finally process design to process specifications. Naturally, any product has an array of features and customer needs are rarely monofunctional. A successful DFSS effort focuses on the most critical features. Quality functional deployment provides a mechanism for defining product design CTQs based on customer CTQs, then defining process design CTQs based on product design requirements and process specification based on process CTQs, as illustrated in Fig. 4. Usually, the most difficult part of a QFD is the definition of customer needs. Rarely, if ever, is the first attempt at a QFD complete. Customers frequently do not know their total needs, and can only detect them as deficiencies, well after the initial QFD definition. The only way to compensate for this initial failing is to maintain close customer contact and treat the QFD as a living document. Once initial designs are developed, customer feedback must be used to update and refine the QFD. The first matrix (QFD1) relates the customer needs, i.e., top level CTQs, to the required functions and features a product needs to deliver those CTQs.

Fig. 3 Linkage of customer needs to process controls via QFD. (View this art in color at www.dekker.com.)

S

2722

Six Sigma Design: An Overview of Design for Six Sigma (DFSS)

Fig. 4 Flow down of customer requirements to process specifications. (View this art in color at www.dekker.com.)

For example, if the customer CTQ is stability in harsh environments, the related features might be the ability to withstand 100 C exposure at pH 1 for 1000 hr. Note that the features still have not defined a specific product or process, but they have defined measurable criteria that relate to the desired CTQ. At this stage, the DFSS team assesses what competitive alternatives exist that might deliver the same features, and proceeds to flow the functional requirements downward to eventually define process specifications. We do not have the space here to explore all the aspects of QFD, but there are excellent sources available in print and online that provide this kind of detail.[4,5,11–13] We now delve briefly into the mechanics of the QFD. This is a tool that is simple to understand, but can be very difficult in practice. An illustrative example of a typical QFD is given in Figs. 5 and 6. A customer desires high purity alcohol, at reasonable cost with just in time delivery. The five steps in a QFD1 (Fig. 5A) are: 1. Assess customer needs, typically through customer interviews, surveys, trade groups, marketing contacts, etc. Once the needs have been identified they are prioritized via numerical rankings. A variety of approaches to these rankings can be used. In Fig. 5A, we have used a 1,3,5 ranking for denoting importance. The customer must absolutely have high purity and reasonable cost; hence these needs are rated 5. The delivery schedule is important, but there is some flexibility generating a rating of 3. The exact ranking scheme is not critical, provided the DFSS team understands how important these performance CTQs are to the customer.

2. Determine what types of existing solutions and competitive approaches already exist to deliver the desired product features and contrast them with the DFSS team’s potential solution. Again a numerical ranking is used to assess how well each of the alternatives addresses the customer CTQs. In Fig. 5A, we have used a logarithmic 1,3,9 scale for these ratings. Competitor 1 can deliver high purity alcohol, but at a high cost and poor delivery schedule. Competitor 2 has low cost, but lower purity. The rating of each alternative is multiplied by their importance rating and their products are summed to provide an overall weighted performance rating. Thus Competitor 1’s overall rating is 53 f(9  5) þ (1  5) þ (1  3)g. This exercise allows the team to quickly assess how big the gap is between potential solutions, as well as defining potential strong points. 3. The DFSS team assesses what types of product features are needed to deliver the required customer benefit, and how they will measure those features—measurement system, targets, and ranges. Those items are listed as functional product requirements. 4. The DFSS team evaluates how strong the relationship is between the functional requirements and each customer CTQ. The DFSS team defines the rating scale for the relationship strength, although in our experience a logarithmic 0,1,3,9 scale, with 9 being the strongest relationship, works quite well. This type of ranking helps to rapidly make an assessment rather than arguing about which relationships are slightly stronger than the others.

Six Sigma Design: An Overview of Design for Six Sigma (DFSS)

2723

S

Fig. 5 QFD1 and QFD2 for high purity alcohol. (View this art in color at www.dekker.com.)

5. Each relationship strength is multiplied by each CTQ importance rating and the products are summed to give a weighted relative importance for each feature. Examination of Fig. 5 shows how all these different pieces of information are correlated to generate a numerical ranking of functional product requirement importance. Clearly, QFD is at best a ‘‘semi’’ quantitative tool, particularly at the top level. CTQ importance

ratings are based on the customer’s opinion. The strengths of the relationships are based on the DFSS team’s experience and knowledge of what types of features will drive performance. Nevertheless, this is an extremely powerful tool because it focuses the team on what the product should be designed to do for the customer rather than what their existing product can do. Once the team has completed the prioritization of functional requirements, they proceed to QFD2 (Fig. 5B) where the process is repeated, but this time

2724

Six Sigma Design: An Overview of Design for Six Sigma (DFSS)

Fig. 6 QFD3 and QFD4 for high purity alcohol. (View this art in color at www.dekker.com.)

linking functional product requirements with product design requirements. The prioritized functional requirements are reranked, typically on a 1–5 scale; the product design requirements to deliver those functions are defined; potential competitive approaches are defined; and the numerical evaluations in terms of competitive performance and relationship strength are evaluated. Once the product design requirements are defined and prioritized, they are linked to the process requirements via QFD3 as shown in Fig. 6A. The process is repeated in QFD4 to link process requirements to process controls (Fig. 6B). Note that there are no operating ranges specified in Fig. 6. If one is designing a new process or product, the design requirements may well be known while the required process parameter settings to reach those design requirements need to be evaluated. The strength of the QFD is that it helps illustrate those

process parameters that need to be defined. The approach for defining them is discussed in the sections on transfer functions, design of experiments (DOE), and setting specifications. Transfer Functions Simply put, the DFSS definition of a transfer function is: Y ¼ f ðx1 ; x2 ; x3 ; . . . ; xn Þ;

SY ¼ gðSx1 ; x1; Sx2 ; x2; . . .Þ

ð2Þ

This shows that there is a definable relationship between the output of a process and its inputs, and between the variability of the inputs and variability

Six Sigma Design: An Overview of Design for Six Sigma (DFSS)

in the output. A simple illustration is given in Fig. 7. Two aspects of a given feature are important: first its direct effect on the output (Y ), second the variability of the feature and the consequent variability of the output (SY). Variability is unavoidable as raw material lots change, processes change, and environmental influences affect a process. Therefore, simply defining the targets is only half the story. As with QFD, transfer functions are an essential pillar of the DFSS methodology. In fact once QFD has defined what is required, then the transfer functions are a mathematical model of how the process can deliver customer CTQs. If we understand how process settings affect product features and how features affect performance, then it is feasible to design a product that will meet customer requirements, with yields matching model predictions. If we do not understand these relationships, then the probability of success diminishes rapidly. There are two sets of transfer functions one should be concerned with: target settings and effects of variability.[4,14] These are illustrated in Fig. 8, which shows the relationship between controllable inputs and their variability with the desired output and its variability. Thus, QFD defines how one flows the customer CTQs downwards, while the transfer functions define how one predicts process capability and defines the critical control points.[4,5] The transfer functions allow the team to establish this linkage early in the development, rather than try to do it once products are in production. In addition when customer requirements change, the team does not have to begin a new project. With the transfer functions in hand, they can quickly evaluate capability and predicted reliability for the required process changes. The most difficult part of using transfer functions is defining them in the first place. Once they exist, they are simple to use. The process map in Fig. 9 illustrates schematically how one can define transfer functions. These functions can take a variety of forms and arise

2725

from a variety of sources as illustrated in Fig. 10. The functions can be derived from known mechanistic models, e.g., based on the ideal gas laws, one can predict gas pressure at a given temperature, or an Arrhenius equation can be used to predict kinetics. The appropriate model parameters may be known a priori or can be determined by experimentation. One can also develop empirical models based on experimental data. Mechanistic models are preferable, because they explain why the output depends on the input and there is a clear understanding of the mechanism. Empirical models are also quite useful with the caveat that they are based on an observation of correlation. There is always a possibility that the observed correlation is serendipitous; so the DFSS team must carefully review the model to be sure that the observed function is reasonable. In practice, empirical models are more prevalent in complex chemical systems, but it is essential that the model only be used for the same experimental space that was used in the model. Extrapolation outside this area is at best highly questionable. If the team wishes to extrapolate, then additional experimentation is required to make sure the model fits in the new region. We illustrate transfer function development with the high purity alcohol example as mentioned earlier. Assuming we have an empirical function, with known parameters, we can formulate a mathematical model to product performance and cost. Using the approach of Chang,[15] we can solve the short cut distillation equations of Hengstebeck, Geddes, Fenske, Underwood, and Gilliland.[16–21] This provides the transfer functions for technical requirements. We then need to incorporate the financial requirements. We have developed a Monte Carlo simulation program to couple the transfer functions for each level of the QFD leading to a final result, which optimizes for technical specifications while minimizing the cost. The program obtains a standard normal value and transforms based on the input mean and standard deviation for each of the input variables. All these

Fig. 7 Schematic illustration of transfer functions for targets and variability in a key variable. (View this art in color at www.dekker. com.)

S

2726

Six Sigma Design: An Overview of Design for Six Sigma (DFSS)

Fig. 8 Linkage of transfer functions from process parameters through customer CTQs. (View this art in color at www.dekker.com.)

are sent to the trays subroutine to calculate ideal stages (IS). These are then stored in an array of responses, and statistics such as mean, standard deviation, skewness, and kurtosis are calculated on the results. In addi-

tion, a factor response summary is calculated for each of the four input variables. The method used is that found in Refs.[22,23] The results of this simulation are shown in Fig. 11. We then extended the model to incor-

Fig. 9 Process map for creating transfer functions. (View this art in color at www.dekker.com.)

Six Sigma Design: An Overview of Design for Six Sigma (DFSS)

2727

S

Fig. 10 Types of transfer functions. (View this art in color at www.dekker.com.)

porate those factors related to cost. This was of more interest, because it incorporates multiple levels of transfer function. In effect, we can manipulate the input variables and see how the ultimate outputs, which are four levels higher in the QFD, are affected. This comprehensive model uses the same method described earlier; transforming a standard normal value using the first four means and standard deviations as depicted in Fig. 11. Submission to the trays subroutine returns the number of IS and reflux ratio (RR). These are then used to calculate total cost (TC) of the column based on the algorithm (i.e., transfer functions) shown below. Capital cost (CC) [$=yr] is calculated by Eq. (3), which is a function of the number of IS and RR.

Variable cost (VC) is calculated by Eq. (6) and is a function of YSC.

CC ¼ 1353278:3 þ 44483:4  IS

Simply knowing the required controls is insufficient to guarantee consistent performance. The DFSS team must design controls and safeguards into the system. The primary tool in this effort is the design failure modes and effects analysis (DFMEA). Failure mode and effects analysis (FMEA) is a well-known tool, which has been extensively used in a wide range of industries.[1–5,7–8] It is an approach that allows prioritization based on the highest failure risks in the development process. Failure mode and effects analysis couples the severity, frequency, and detectability of failure to meet customer CTQs to assess overall risk of a given failure type. There are different types of FMEA for manufacturing processes, administrative processes, and design processes. A DFSS team will

 949840:14=RR

ð3Þ

Fixed cost (FC) [$=yr] is calculated by Eq. (4), which is a function of CC. FC ¼ ð0:0000759 þ 0:0933  CC  CCÞ1=2

ð4Þ

Yearly steam cost (YSC) [$=yr] is calculated based on 8000 hr in a year and Eq. (5) is a function of steam rate (SR) and steam cost (SC).

YSC ¼ SR  8000  SC

ð5Þ

VC ¼ 327322 þ 1:31  YSC

ð6Þ

Finally, TC [$=yr] is a function of FC and VC, and is calculated by Eq. (7). TC ¼ VC þ FC

ð7Þ

The results of this simulation are shown in Fig. 12.

Design FMEA

2728

Six Sigma Design: An Overview of Design for Six Sigma (DFSS)

Fig. 11 Monte Carlo simulation of separation efficiency for alcohol purification. (View this art in color at www.dekker.com.)

be involved in the DFMEA and at the very least a preliminary manufacturing FMEA. A typical design FMEA is given in Fig. 13 for the alcohol example discussed earlier. The headings across the top of the FMEA define the types of information it requires. The DFSS team must evaluate the risk associated with failure of the CTQs. We discuss the functional requirement FMEA, but the same approach

applies to any FMEA. The first column specifies the requirement that is being evaluated. The second column specifies the failure mode. A failure mode is the way in which the requirement is not met. The third column specifies the effect on the customer if that requirement is not been met. The next column is a numerical rating of severity, i.e., how badly does it affect the customer. The fifth column addresses causes

Six Sigma Design: An Overview of Design for Six Sigma (DFSS)

2729

S

Fig. 12 Monte Carlo simulation for separation efficiency, incorporating cost transfer function. (View this art in color at www.dekker.com.)

for the failure mode. The next is a numerical rating of the likelihood for the occurrence of this cause. The seventh column is an assessment of the types of systems being employed to determine the occurrence of failure and to eliminate it before product release. The next is a numerical rating of the effectiveness of that particular detection system. The total risk is computed

by multiplying the severity rating by the frequency rating by the detectability rating, generating the next column titled the risk priority number (RPN). The remaining columns are for tracking what corrective actions are needed, who is responsible, and the effect on the RPN after taking the required actions. Inspection of the RPN values immediately focuses the DFSS

2730

Six Sigma Design: An Overview of Design for Six Sigma (DFSS)

Fig. 13 Design failure mode and effects analysis for alcohol purification. (View this art in color at www.dekker.com.)

team on the highest risk failure modes, which in this illustration are rise in energy cost and poor model. This is not to imply that the other failure modes are unimportant. It is simply a mechanism to reduce or eliminate the highest risks. Failure mode and effects analysis is an iterative process. Once the high RPNs are reduced, the DFSS team needs to determine if the overall risk is sufficiently small to justify product manufacturing. If not, then the ongoing efforts taken must be aimed at reducing the other high risk items. Developing an FMEA is a DFSS team effort. The numerical ratings for severity, detectability and frequency must be agreed on by the team, so that everyone agrees on the definitions. Otherwise one team member may consider the effect to be mild, another moderate, and yet another severe. As with QFD, the numbers here are ‘‘semi’’ quantitative, and round table team discussions help to achieve a more accurate picture of the true risks associated with the specific failure modes. If possible, the team should formalize the rating system as much as possible. A sample of such a formalized scale is given in Fig. 14.

Experimental Design and Process Simulation Design of experiments is a fundamental tool in DFSS.[4,5] Systematic experimentation is frequently needed to determine the functional form of a transfer function.[14,24–29] This applies in particular to chemical reactions and polymer processing, where predicting yields of complex, multistep processes is difficult. DOE has been widely used to determine optimized settings for chemical processes. In DFSS one must also evaluate the effect of input variation on the output. There are various types of DOE that address this. Taguchi designs systematically evaluate the effect of noise variables on the outputs. Schematically, this approach is illustrated in Fig. 15A for a two-factor design, with two levels for each factor and two noise variables. The corners of the large square (the outer array) define the factor settings. Around each corner is a smaller square (the inner array) with the various settings for the noise factors. For example, the factors DV1 and DV2 might be RR and residence time. The noise variable might be steam feed rate and lot to lot

Six Sigma Design: An Overview of Design for Six Sigma (DFSS)

2731

S

Fig. 14 Typical rating scale used for FMEA. (View this art in color at www.dekker.com.)

changes in raw materials. Minimizing the influence of noise variables then makes the design inherently more robust, i.e., stable to external influences. The major advantage for the Taguchi designs is the direct assessment of the effects of noise variables. Frequently, experimenters refrain from using these designs because they require a large number of experimental runs. Note that this two level two-factor design requires 16 experiments to define a main effects model. An alternative approach is the use of response surface or mixture designs to determine the transfer function. A schematic of a two-factor response surface design is given in Fig. 15B. This is accomplished in nine experimental runs and evaluates main effects, as well as factor interactions. If the variability of the inputs is known, then one can model the predicted

variability using propagation of error techniques or Monte Carlo simulations based on the transfer function. The major advantage relative to the Taguchi approach is the need for significantly fewer experimental runs.[24,25] However, this approach assumes that the experiment has adequately mapped the experimental space, and that the input variations are known.

TRANSITION TO MANUFACTURING Setting Specifications The achievement of six sigma is typically a balance of the cost and customer requirement. Consequently, there are two ways to achieve six sigma. One can develop a

Fig. 15 Schematic illustration of different robust experimental designs: (A) Taguchi design and (B) response surface design. (View this art in color at www.dekker.com.)

2732

Six Sigma Design: An Overview of Design for Six Sigma (DFSS)

Fig. 16 Sample control plan for high purity alcohol production. (View this art in color at www.dekker.com.)

very narrow process, which typically drives up the cost. Alternatively one can set wide specifications, but the customer CTQs must still be met. In DFSS one attempts to predict the variability as discussed earlier. This allows a prediction of achievable specifications.[30] In practice, variation in customer CTQs is driven by multiple factors. For example, if the customer CTQ is a water white polymer (i.e., a transparent, colorless polymer), variations might depend on melt temperature, vacuum levels, quench speed, stabilizer content, etc. The strength of DFSS is that one will predict the set of conditions that minimize the variation around the target level of yellowing. Thus one might opt for rapid quenching and higher stabilizer because the oxygen level cannot be maintained at a sufficiently low level. Finally, one can set specifications based on known process capability. If a sufficient number of lots have been produced, one can statistically evaluate the variation and set the specification limits outside the statistical control limits. Typical DFSS efforts do not lead to a sufficient number of sample lots for such a statistical evaluation. However, once the product is scaled up and in production, the specifications should be re-evaluated to make sure they will remain at six sigma performance. Establishing Control Plans The final critical piece of the DFSS puzzle is a control plan.[1–5,8–10] This is a document that specifies which are the critical inputs, where their limits need to be, and how they will be evaluated. Such a document should be part of any transfer to manufacturing. It needs to specify standard operating procedures that are

required and what the capability of the measurement system is. DFSS teams should preferably include, but at the very least work with, manufacturing to make sure that any product or process designed can, in fact, be produced. Consequently, a preliminary control plan should be developed in parallel with the assessment of transfer functions and the FMEA. It should then be updated as more knowledge is developed about the control points and product performance. A good control plan has all the elements required to assess performance and take action to prevent failures. It includes the process points to be monitored, the specification limits, the measurement technique and measurement capabilities, as well as responsibility to take action. A typical control plan for our alcohol example is shown in Fig. 16.

CONCLUSIONS Design for six sigma is not a radical new idea. It is in fact simply a codification of the scientific method. DFSS is a systematic application of good scientific and engineering practices. The tools employed are very powerful if used correctly. Fundamentally, DFSS is a road map and set of tools to provide the researcher with clear insight into customer needs and the customer with clear insight into producer capabilities.

ACKNOWLEDGMENTS The authors thank Tony Signorelli, Janice Sund, Ross Rapoport, Alan Levy, Bill Hill, Dick Johnson, and

Six Sigma Design: An Overview of Design for Six Sigma (DFSS)

Pa´udraig Curran for their helpful comments and suggestions.

ARTICLE OF FURTHER INTEREST Thermosets: Materials, Processes, and Waste Minimization, p. 3031.

REFERENCES 1. Harry, M.J.; Lawson, J.R. Six Sigma Producibility Analysis and Process Characterization; Addison-Wesley: New York, 1992. 2. Harry, M.J.; Schroeder, R. Six Sigma: The Breakthrough Management Strategy Revolutionizing the World’s Top Corporations; Doubleday & Company: New York, 1999. 3. Crosby, P. Quality Is Free: The Art of Making Quality Certain; Reissue Ed.; Mentor Books, 1992. 4. Berryman, M.L. Transform your organization into one that’s world class. Six Sigma Forum Mag. 2002, 2 (1). 5. Creveling, C.M.; Slutsky, J.L.; Antis, D., Jr. Design for Six Sigma—In Technology and Product Development; Pearson Education, Inc.: New Jersey, 2003. 6. Senge, P.M.; Carstedt, G. Innovating our way to the next industrial revolution. MIT Sloan Manage. Rev., (Winter), 24–38. 7. Pande, P.S.; Holpp, L.H. What is Six Sigma? 1st Ed.; McGraw-Hill Trade, 2001. 8. Chowdhury, S. Design for Six Sigma—The Revolutionary Process for Achieving Extraordinary Profits; Dearborn Trade Publishing: Chicago, 2002. 9. Grant, E.L.; Leavenworth, R.S. Statistical Quality Control, 4th Ed.; McGraw-Hill: New York, 1988. 10. Pyzdek, T. The Six Sigma Handbook, 1st Ed.; McGraw-Hill Trade, 2000. 11. Cohen, L. Quality Function Deployment: How to Make QFD Work for You; Addison-Wesley, 1995. 12. Terninko, J. Step-by-Step QFD: Customer-Driven Product Design; CRC Press, 1997. 13. ReVelle, J.B.; Moran, J.W.; Cox, C.A. The QFD Handbook; Wiley & Sons: New York, 1997. 14. Haugen, E.B. Probabilistic Mechanical Design; Wiley & Sons: New York, 1980.

2733

15. Chang, H.-Y. Computer aids short-cut distillation design. Hydrocarbon Process. 1980, 59 (8), 79–82. 16. Hengstebeck, R.J. A simplified method for solving multicomponent distillation problems. Trans. AIChE 1946, 42, 309–329. 17. Geddes, R.L. Computation of petroleum fractionation—estimation of A.S.T.M. distillation curves from true boiling-point distillation analyses. Ind. Eng. Chem. 1941, 33, 795–801. 18. Thiele, E.W.; Geddes, R.L. Computation of distillation apparatus for hydrocarbon mixtures. Journal of Industrial and Engineering Chemistry 1993, 25, 289–295. 19. Fenske, M.R. Fractionation of straight-run Pennsylvania gasoline. Ind. Eng. Chem. 1932, 24, 482–485. 20. Underwood, A.J.V. Fractional distillation of multicomponent mixtures. Chem. Eng. Prog. 1948, 44 (8), 603–614. 21. Gilliland, E.R. Multicomponent rectification: minimum reflux ratio. Ind. Eng. Chem. 1940, 32, 1101–1106. 22. Beyer, W.H., Ed. CRC Handbook of Tables for Probability and Statistics; The Chemical Rubber Co.: Cleveland, OH, 1966; 228 pp. 23. Suhir, E. Applied Probability for Engineers and Scientists; McGraw-Hill: New York, 1997. 24. Montgomery, D.C. Design and Analysis of Experiments, 3rd Ed.; Wiley & Sons: New York, 1991. 25. Del Vecchio, R.J. Understanding Design of Experiments: A Primer for Technologists; Hanser Publishers: Munich, 1997. 26. Box, G.E.P.; Hunter, W.G.; Hunter, J.S. Statistics for Experimenters—An Introduction to Design, Data Analysis, and Model Building; Wiley & Sons: New York, 1978. 27. Hicks, C.R. Fundamental Concepts in Design of Experiments, 4th Ed.; Saunders College Publishing: New York, 1993. 28. Ross, P.J. Taguchi Techniques for Quality Engineering, 2nd Ed.; McGraw-Hill Professional, 1995. 29. Roy, R.K. Design of Experiments Using The Taguchi Approach: 16 Steps to Product and Process Improvement; Interscience, 2001. 30. Brown, B. Chair Specifications for the Chemical and Process Industries: A Manual for Development and Use; ASQC Chemical and Process Industries Division Chemical Interest Committee ASQC, Quality Press: Wisconsin, 1996.

S

Size Reduction S Sunil Kesavan Akebono Corporation, Farmington Hills, Michigan, U.S.A.

INTRODUCTION The comminution or size reduction process is an important unit operation in the process industries in which solid materials are broken or cut into smaller sizes by the application of mechanical stress. Solids can be reduced in size by crushers and grinders which employ compression, impact, attrition, shear, or combinations thereof. Depending on the degree of particle size reduction desired, the end result is achieved in one or several steps. Devices that are used for size reduction operations can be classified into primary and secondary crushers, grinders, and pulverizers. Explosive blasting is used in many instances for primary size reduction of ore formations into sizes workable by primary crushing machines. The primary crushers are slow-speed machines that reduce the run-of-mine product into 15–25 cm lumps. A secondary crusher reduces these lumps to 5 mm product. Grinders reduce the products of crushing operations into powder. An intermediate grinder typically produces a product that passes a 40-mesh screen. Fine grinders reduce feed into product that passes a 200-mesh screen. Ultrafine grinders can convert the product of secondary crushers into 1–10 m product. Cutting machines produce particles with a definite size and shape in the 2–10 mm size range. Size reduction machines can be classified as follows: a. Primary and secondary crushers 1. Gyratory crushers 2. Jaw crushers 3. Roll crushers b. Intermediate and fine grinders 1. Impact mills a. Hammer mills b. Centrifugal pin mill 2. Attrition mills 3. Tumbling mills a. Ball and pebble mills b. Rod mills c. Tube mills; compartment mills 4. Rolling-compression mills a. Ring roll mills b. Bowl mills c. Ultrafine grinders 1. Fluid-energy mills Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120018807 Copyright # 2006 by Taylor & Francis. All rights reserved.

2. Agitated mills 3. Impact mills with internal classification d. Cutting machines 1. Knife cutters, dicers, slitters.

GRINDING ENERGY REQUIREMENTS Given the relatively large power consumption accompanying size reduction, it is important to quantify the energy requirements of these unit processes. Comminution theory dates back to the works of Kick[1] and Rittinger[2] that were done in Germany. A generalized differential equation that governs crushing can be written as: dW ¼
KðdD=Dn Þ

ð1Þ

where W is the energy input, K and n are constants, and D is the particle size. Eq. (1) represents Kick’s law when n ¼ 1 and Rittinger’s law when n ¼ 2. Kick’s law[1] essentially states that the work required to obtain a given reduction ratio is the same irrespective of starting size. According to Rittinger’s law[2], work is proportional to surface created. Rittinger’s and Kick’s laws are only useful over a limited particle size range and are not utilized today. Bond[3] proposed that work is inversely proportional to the square root of particle diameter. Bond’s theory of comminution is represented by Eq. (1) when n ¼ 1.5 and can be written as: W ¼ 100Wi ð1=DP0:5
1=DF0:5 Þ where Wi is the Bond work index or work required to reduce a unit weight of product from a theoretical infinite size to a product with 80% passing 100 m. Extensive data are available on the work index and the Bond law is widely used today. PRIMARY AND SECONDARY CRUSHERS Gyratory Cone Crushers These types of crushers employ a conical crushing element that gyrates in an eccentric manner in a shell 2735

2736

resembling an inverted cone (Fig. 1). The material to be crushed enters the top of the crusher where the crushing surfaces are most widely spaced. The product becomes wedged and squeezed between the mantle and the hopper and is progressively broken down until it discharges through the narrow opening at the bottom of the crusher. Gyratory crushers are available in large sizes used for primary crushing and smaller sizes for secondary crushing of soft to medium-hard materials. These continuous-discharge crushers are available in capacities up to 3500 ton=hr and are more cost effective to operate than jaw crushers.

Jaw Crushers These devices suited for coarse and intermediate crushing of large volumes of hard and semihard materials employ swinging jaws that work against a hardened stationary surface. The jaws, which are essentially flat, form a V-shaped crushing chamber with a wide inlet at the top and a narrow discharge at the bottom. The large feed material gets progressively reduced in size and falls down toward the narrow throat section. There are three types of jaw crushers (Fig. 2)—the Blake, the Dodge, and the single-toggle type. In the popular double-toggle Blake jaw crusher (Fig. 3), the moving jaw is hinged at the top with the maximum movement at the bottom of this jaw. In the Dodge crusher, the swinging jaw is hinged at the bottom. This gives a fixed discharge opening, giving a more

Fig. 1 Schematic representation of the operating principle of a gyratory cone crusher.

Size Reduction

A

B

C

Fig. 2 The essential features of (A) Dodge, (B) Blake, and (C) single-toggle jaw crushers.

uniform product than the Blake crusher. The greatest movement acts on the large feed at the top of the jaw. The Dodge crusher is not used much in commercial large volume crushing operations. The single-toggle jaw crusher is pivoted at the top like the Blake crusher but has an additional movement in the downward direction. The Blake crusher costs more than the single-toggle machines but the operating costs are less for hard feed materials. Taggart[4] proposed a rule-of-thumb that a jaw crusher would be the economically preferred size reduction equipment if the required throughput in tons per hour was less than the square of the crusher gap in inches. Depending on size, jaw crushers can handle up to a 48-in. size feed and create product smaller than half an inch. Jaw crushers are available in a variety of sizes up to about 1000 ton=hr and can be stationary, portable, or skid mounted. Roll Crushers These tertiary crushers employ smooth or toothed heavy-duty impact and abrasion-resistant steel-rimmed rolls. The rolls are mounted inline in a horizontal manner and turn toward each other at equal speeds to create a nip into which a friable feed material is introduced (Fig. 4). Heavy-duty compression springs with automatic reset are used to dampen crushing shock and to protect the crusher from tramp iron and oversize material. An adjustable screw that adjusts spring tension changes the crusher opening. A flywheel is used to even out pulses and economize on power consumption. These crushers have a theoretical maximum reduction ratio of 4 : 1 and will only crush materials to about 10 mesh. Roll crushers produce a controlled product size distribution without a lot of fines. The narrow particle size distribution is achieved by controlling a combination of variables including roll speed, gap measure, differential speed, feed rate, and roll surface. Toothed single roll crushers that crush materials by working on a breaker plate are also available. The crushing teeth are set in segments to facilitate replacement.

Size Reduction

2737

S

Fig. 3 Cross section of a Blake-type jaw crusher showing the nonchoking design of the swing jaw.

INTERMEDIATE AND FINE GRINDERS Impact Mills Hammer mills Impact mills like the hammer mill use swinging hammers (typically running at 750–1800 rpm) to pulverize the solid feed by impact and attrition. As shown in Fig. 5, these hammers are mounted by pins on the periphery of a number of disks mounted on a horizontal rotating shaft and are free to swing. The hammers force the product against a rugged breaker plate. The

product gets broken down until it is small enough to fall through a discharge grate at the bottom. For some applications like the grinding of dried animal byproducts, a vertical hammer mill that uses a vertical drive shaft with horizontal hammers and screens is more efficient. Hammer mills are used in primary, secondary, and tertiary crushing operations. These mills are relatively inexpensive but tend to produce a lot of fine material. They are best suited for soft or semihard materials, as harder materials tend to rapidly wear out the hammers. Hammer mills can handle feed sizes of up to 10–20 in. and throughputs up to 500 ton=hr. Air-swept hammer

Fig. 4 Schematic crusher.

of

a

smooth-roll

2738

Size Reduction

force directs the feed outward on to intermeshing pins or blocks mounted on the plate. Pin mills are available in capacities up to 200 ton=hr. These mills are economical, easy to operate, produce a uniform grind, are suitable for wet or dry grinding, and provide high throughput with low energy consumption. Attrition/Disk Mills

Fig. 5 Schematic representation of the operating principle of a hammer mill. (Courtesy of Sturtevant, Inc., Boston, MA.)

These mills use a combination of impact, shear, and cutting action to grind materials between the replaceable wearing surfaces of two grinding disks. One or both of the disks can rotate; if both, they are run counter to each other. The distance between the disks can be adjusted to vary the product particle size. Fig. 7 is a schematic representation of a typical attrition mill. Liquid cooling of the disks is used with heat-sensitive materials to prevent degradation. Air is sometimes drawn through the mill to help remove product and prevent choking. Attrition mills can grind up to 8 ton=hr of product to a particle size passing 200 mesh. Tumbling Mills

mills use air to help convey the particle product out of the mill.

Ball mills

An example of a pin mill is shown in Fig. 6. It applies centrifugal forces to grind feed particles by impact. Feed entering the mill is divided into two streams that drop down on to a rotating plate. Centrifugal

These mills employ attrition and impact to grind the product by tumbling in cylindrical mills partly filled with grinding media. The media can be round metal balls or nonmetallic pebbles that are 1=2 in. or larger in size. Ball mills for grinding hard materials are usually lined with heavy-duty steel alloy liners. Some ball mills employ internal baffles to prevent slippage of the grinding media over the internal shell surface.

Fig. 6 Simpactor centrifugal pin-type impact mill. (Courtesy of Sturtevant, Inc., Boston, MA.)

Fig. 7 Schematic of a disk-type attrition mill. (Courtesy of Sturtevant, Inc., Boston, MA.)

Centrifugal pin mills

Size Reduction

Mills for dry grinding are available with either full or semi air-swept capability. Ball mills can operate either in a batch or a continuous mode. Batch mills typically use a charge of 50– 55% of the mill volume, while continuously operating mills use a media charge of 40–45%. In the continuous mode, the mill can operate in such a mode that the product can leave the mill through a discharge grate. When operating in a continuous mode, the effluent of the grinding mill is sent through a classifier to separate out the oversize product to be reprocessed through the mill. Commonly used types of grinding media are carbon steel, stainless steel, chrome steel, tungsten carbide, ceramic, or zirconia. Ball mills produce up to 50 ton=hr of powder substantially passing a 200-mesh screen. Temperature control can be achieved by the use of jacketed ball mills through which a heat transfer fluid is circulated. Critical and Operating Speed. In operation, the balls in the mill are carried up in contact with each other and with the walls until the centrifugal force is overcome by the centripetal force. The critical speed Ncr, of a ball mill is the speed in rpm above which the grinding media will centrifuge and all milling effectively stops. The critical speed of a mill is given by:

2739

The grinding media in the mill fall through different heights depending on their relative location. This provides a classifying action within the mill resulting in increased crushing energy efficiency. Ball mills with centrifugal and planetary action are also available. Retsch manufactures a planetary ball mill in which multiple grinding chambers are turned around their own axes, and, in the opposite direction, around the common axis of a sun wheel that is driven by a rugged motor. This movement results in the superimposition of centrifugal forces that change constantly (Coriolis motion). The grinding media describe a semicircular motion, separate from the inside wall and collide with the opposite surface at high impact energy. Fig. 9 illustrates the forces encountered in the operation of this grinding mill that can reduce particles to submicron sizes through the imposition of impact and friction forces. The Retsch centrifugal ball mill uses grinding chambers that move in a horizontal plane at speeds of 100–580 rpm. The centrifugal forces that are generated propel the grinding media against the inside walls of the mill where they roll over the product (Fig. 10). Size reduction is achieved by a combination of impact and friction. The mill is furnished with an automatic reversal system to counter any agglomeration effects and to enhance homogenization.

Ncr ¼ 42:3=D0:5 where D is the internal diameter of the mill in meters. The operating speed of ball mills is usually 55–75% of the critical speed. Operating close to Ncr drastically reduces the effectiveness of the grinding action. Conical ball mills like the Hardinge mill shown in Fig. 8 have their larger diameter closer to the feed inlet.

Fig. 8 Hardinge conical ball mill. (Courtesy of Metso Minerals, York, PA.)

Fig. 9 Schematic of forces operating in a planetary ball mill. (Courtesy of Retsch, Inc., Newtown, PA.)

S

2740

Size Reduction

Autogenous and semiautogenous mills Autogenous mills are mechanically similar to ball mills but employ the product to be crushed as the grinding medium. These mills can be used for hard, soft, as well as sticky products. The product is discharged through a discharge grate designed to retain oversize product. Autogenous grinding reduces metal wear, eliminates secondary and tertiary crushing stages, and offers savings in capital and operating costs. A variation called air autogenous milling uses air to transport and classify the product and uses no discharge screens or grills that can become blocked. Semiautogenous mills use a combination of the feed, product, and grinding media to achieve the desired size reduction. Rolling Compression Mills

Fig. 10 Schematic of centrifugal ball mill action. (Courtesy of Retsch, Inc., Newtown, PA.)

These mills resemble a mortar and pestle in principle and use a rolling ball or roller member moving against the face of a ring or a casing to grind soft to medium-hard product at rates of up to 50 ton=hr. Pressure is applied by either using heavy springs or by the centrifugal force of the roller. Either the roller or ring may be stationary. Built-in air classification is used to improve the grinding efficiency of these mills. The common types of rolling compression mills are rolling-ring pulverizers, bowl mills, and roller mills. These mills are widely used for grinding coal, cement clinker, and limestone.

Tube mills

ULTRAFINE GRINDERS

Tube mills are basically finish-grinding ball mills with mill length several times the diameter. Tube mills can be made with a uniform diameter throughout the mill length or may be made with several compartments with different section diameters and lengths. Compartment mills are tube mills with slotted transverse partitions that separate grinding media of different sizes with the larger media working on coarser product.

Fluid-Energy Mills

Rod mills Rod mills are regarded as intermediate grinding mills and are basically tube mills that employ grinding rods that are about as long as the mill. Rod mills employ rolling compression and attrition and produce very small amounts of oversize or fines. Feed sizes up to 50 mm (2 in.) can be reduced to product in the 5–10 mesh range in open circuit grinding and to a minus 35 mesh range in closed circuit grinding with a classifying device. These mills require operator attention to minimize rod misalignment or entangling during grinding.

These energy-intensive mills have no moving parts and use pressurized air, steam, or inert gases to grind particles to ultrafine sizes. The particles to be ground are suspended in a high-velocity fluid stream that carries the particles around in a circular or an elliptical path. The majority of the particle size reduction is accomplished by interparticle attrition with additional grinding caused by particles rubbing against the walls. The particles get classified as they go around the closed path provided by the mill. The larger particles are thrown toward the outer walls and stay in the mill, while the finer particles stay on the inner walls and are removed from the mill through an exit port. The effluent from the mill is sent to a particle separator to collect the ground product. No one jet mill is suitable for all process applications. Several configurations are available to serve specific needs: fluid bed, opposed jet, and multiple-port types. The Jet-O-Mizer (Fig. 11) from Fluid Energy Processing can produce product with a 1–50 m average size and a narrow particle size distribution. This

Size Reduction

2741

S

Fig. 13 Cutaway of the Micro-Jet mill showing fluid jets. (Courtesy of Fluid Energy Processing, Telford, PA.) (View this art in color at www.dekker.com.)

Fig. 11 Cutaway of Jet-O-Mizer showing fluid jets. (Courtesy of Fluid Energy Processing, Telford, PA.) (View this art in color at www.dekker.com.)

vertical jet mill can also be used for combined operations like grinding=blending and grinding=coating. For finer finished product down to the submicron size range, the Sturtevant Micronizer jet mill or the

Fluid Energy Processing Micro-Jet grinders are suitable. The mills use a circular grinding chamber. The feed metered into the mill enters an air or a gas vortex created by precisely aligned jet nozzles along the mill periphery (Figs. 12 and 13). The tangential angle of the fluid flow causes interparticle impact resulting in size reduction. The desired product and oversize material are separated by air classification in the mill. The mill provides narrow particle size distribution with uniform shape and no heat build-up. Micronizers are available in capacities up to 5 ton=hr. The Roto-Jet fluid-bed jet mill (Fig. 14) supplied by Fluid Energy Processing is designed to grind products to the 0.5–40 m average size range with specific top

Fig. 12 Operating schematic of the Micronizer jet mill. (Courtesy of Sturtevant, Inc., Boston, MA.) (View this art in color at www.dekker.com.)

2742

Size Reduction

Pump

Premix Holding Tank

Fig. 16 Circulation attritor. (Courtesy of Union Process, Inc., Akron, Ohio.)

Attritors

Fig. 14 Cutaway of Roto-Jet showing fluid jets. (Courtesy of Fluid Energy Processing, Telford, PA.) (View this art in color at www.dekker.com.)

and=or bottom size requirements. The machine uses a variable-speed rotor for tight control of product size.

The Szegvari attritor uses 1=8–3=8 in. grinding media agitated at speeds up to 350 rpm. Attritors are available that operate in three different modes: 1. Batch attritor shown in Fig. 15. 2. Circulation attritors that use an attritor in association with a holding tank that can hold about

Agitated Mills Agitated mills are stirred vertical ball mills in which the grinding media are agitated by vibratory energy or by a rotating impeller that can run at speeds up to 1700 rpm. These mills are suited for both wet and dry grinding in batch or continuous mode. The fast grinding machines are energy efficient, compact, easy to operate, and are best suited for reduction of particles to submicron sizes. Particle size reduction is achieved by both shear and impact.

Cooling water out Discharge Pump

Cooling water in

Fig. 15 Batch attritor. (Courtesy of Union Process, Akron, Ohio.)

Fig. 17 Continuous attritor. (Courtesy of Union Process, Inc., Akron, Ohio.)

Size Reduction

2743

S

Fig. 18 Cross section of the Sweco vibro-energy mill. (Courtesy of Sweco, Florence, KY.) (View this art in color at www. dekker.com.)

10 times the volume of the attritor (Fig. 16). The contents of the holding tank are passed through the attritor multiple times until the desired reduction in size is achieved. This combination allows the use of small attritors for large grinding jobs through the use of high circulation rates. 4. Continuous attritors with the grinding time controlled by the feed pumping rate (Fig. 17). A high-speed attritor that uses 0.5–3 mm grinding media and impeller speeds of up to 1700 rpm has been developed recently for dry grinding.

The Vibra-Drum vibratory grinding mill supplied by General Kinematics uses a static grinding drum containing grinding media akin to a ball mill. Vibration energy is imparted to the drum from an external mechanism. A subresonant two-mass drive and spring system alternately stores and releases grinding power. Once in motion, the natural frequency design ensures that energy is only needed to move the grinding media as a fluid mass. This mill has lower capital investment, installation, maintenance, and energy costs as compared to conventional rotational mills.

Vibratory mills

Impact Mills with Internal Classification

The Sweco vibro-energy mill (Fig. 18) applies high frequency, three-dimensional vibrations to the grinding chamber that contains small cylindrical grinding media. This helps produce an ultrafine product with a narrow particle size distribution.

An example of a classifying impact mill is the Powderizer marketed by Sturtevant, Inc. (Fig. 19). The feed to the mill enters the grinding chamber where it is broken by the action of impactors=pins mounted on a rotating disk. A column of air sweeps up the pulverized product

2744

Size Reduction

Fig. 19 Schematic of Powderizer air-swept impact mill. (Courtesy of Sturtevant, Inc., Boston, MA.)

through a rotating classifier that rejects oversize products that are returned back for further size reduction. Particle size can be adjusted by changing the classifier speed without shutting down the mill. The product can be pulverized to sizes below 10 m with narrow size distribution. Pulverizers are available in sizes that can handle 20–7000 kg=hr. The Powderizer is also suitable for heat-sensitive materials.

operate in an open circuit mode, where the product being crushed passes through the mill only once. This mode of operation is inefficient as a considerable amount of energy is wasted in regrinding product that has reached the desired product size. A more efficient mode of operation is called closed circuit grinding (Fig. 21). In closed circuit grinding, the mill discharge is sent to size classification equipment that separates out the product that has reached the desired size.

ROTARY CUTTERS Cutting mills are used for reducing the size of soft, medium-hard, tough, elastic, fibrous, and temperaturesensitive materials. Examples of products handled in such mills are electronic scrap, film, rubber, foil, foodstuffs, paper and paper products, textiles, and domestic waste. The mills cut, chop, or tear feed using a rotormounted set of blades or cutters. Fig. 20 shows the cutting action of the SM2000 cutting mill supplied by Retsch. The feed material is taken up by the rotor and is crushed by the stainless steel cutting strips inserted in the housing. Helically arranged reversible cutting plates of hard metal operate by successive cutting.

EQUIPMENT OPERATION Open and Closed Circuit Grinding Comminution is one of the most inefficient unit operations from the viewpoint of energy consumption. Continuous crushing and grinding equipment can

Fig. 20 Schematic showing the cutting action of the SM2000 cutting mill. (Courtesy of Retsch, Inc., Newtown, PA.)

Size Reduction

2745

Surge Hopper

Classifier

Oversize Raw Material Fines Hopper And Feeder

High Speed Attritor

Pneumatic Conveyor Belt

Fig. 21 Representation of a closed-circuit grinding operation employing an attritor. (Courtesy of Union Process, Inc., Akron, Ohio.)

Oversize material is returned to the mill for further size reduction. Closed circuit grinding is suitable for either wet or dry grinding of materials.

CONCLUSIONS The greatest advancements in grinding technology in the recent past have been made in ultrafine grinding

equipment like fluid-energy mills that can generate products in the micron and submicron ranges. This trend is expected to continue as comminution processes evolve to address technical issues in the production of ultrafine particles for high technology applications. As comminution is a very important unit operation in the conversion of run-of-mine raw materials, a considerable amount of effort was spent in the early and middle part of the last century to develop efficient size reduction equipment. The basic principles employed in the operation of crushing and grinding machines have not changed substantially over the past few decades. Saving energy, reducing costs, and cutting pollution have been the main areas targeted for continuous improvement in size reduction operations. The technical areas for future improvement of comminution processes fall into the following general categories: 1) Advanced sensors to provide integrated online physical and chemical characterization of feed and product to help in automated process control; 2) Utilization of the knowledge of real-time feed characteristics to optimize comminution processes; 3) Improved modeling of grinding mill operations like three-dimensional simulation of charge motion in tumbling mills; 4) Advanced abrasion-resistant milling media and surfaces; and 5) Novel or improved physical separation processes to be used in conjunction with size reduction equipment. As in the past, it is expected that future developments in conventional comminution will be evolutionary rather than revolutionary.

REFERENCES 1. Kick, F. Das Gasetz der propertionalen Widerstande und siene Anwendung; Arthur Felix: Leipzig, 1885. 2. Rittinger, P.R. Lehrbuch der Aufbereitungskunde; Ernst and Korn: Berlin, 1867. 3. Bond, F.C. The theory of comminution—Meeting of AIME, Mexico City, October, 1951. Trans. Am. Inst. Min. Metall. Pet. Eng. 1952, 193, 484–494. 4. Taggart, A.F. Handbook of Mineral Dressing; Wiley: New York, 1945.

S

Soave’s Modified Redlich–Kwong Equation of State S J. Richard Elliott, Jr. Department of Chemical Engineering, University of Akron, Akron, Ohio, U.S.A.

INTRODUCTION Throughout the 1970s and 1980s, chemical engineering process design underwent a virtual revolution. By increasing computational speed and the accuracy of thermodynamic models, computer simulation transformed the design process from hand calculations with charts and tables to a modern engineering methodology. In 1996, the Chief Executive Officer of Dow Chemical, Frank Popoff, proclaimed, ‘‘Process modeling is the single technology that has had the biggest impact on our business in the last decade.’’ Soave’s equation played a central role in that revolution. It was the first thermodynamic model to provide feasible accuracy for physical properties of a wide variety of compounds, over a wide range of conditions. Even now, applications and adaptations of Soave’s equation abound. What follows is a short presentation of the nature of these applications and adaptations, how Soave’s equation relates to other equations of state, and why Soave’s approach has stood the test of time. An excellent reference on this subject is provided in Soave’s own words with a greater emphasis on the historical context of Soave’s development.[1]

THREE-PARAMETER CORRESPONDING STATES The primary key to Soave’s early success was the recognition of the need to extend from two-parameter to three-parameter corresponding states. With two parameters, it is possible to fit the critical temperature and critical pressure of any compound, but a general description of the vapor pressure curve is unattainable. The slope of the reduced vapor pressure ðPrsat
P sat =Pc Þ curve varies substantially from one compound to another, but a two-parameter equation of state assumes that it is invariant. Because the vapor pressure varies as an exponential of temperature, small errors near the critical temperature are greatly amplified at reduced temperatures ðTr
T=Tc Þ near 0.45. Vapor pressure is a key property in modeling phase behavior because it represents the vapor–liquid equilibria (VLE) of the pure fluid. Accurate vapor pressure characterization is essential to VLE correlation. As the most common operation in chemical processing is Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120007869 Copyright # 2006 by Taylor & Francis. All rights reserved.

distillation, accurate modeling of VLE had a tremendous impact on process modeling. Characterization of the vapor pressure curve was readily accessible in the form of Pitzer’s acentric factor. The definition of the acentric factor is o
1  log10 ðP sat =Pc ÞjTr¼0:7 Hence, the third parameter, o, implicitly contains information about the vapor pressure, making vapor pressure prediction something like a circular loop. But Soave went beyond this simple observation. Wilson had previously recognized these issues,[2,3] but his equation met with limited success, especially at low reduced temperatures. Soave was careful to analyze the temperature dependence of his equation of state in great detail at the outset. He achieved this by introducing an adjustable parameter into the attractive contribution of the Redlich–Kwong[4] equation. PV ¼ Zig þ Zrep þ Zatt RT br a br  ¼ 1 þ 1  br bRT 1 þ br

Z


ð1Þ

where r is the molar density, Z the compressibility factor, Zig the ideal gas contribution, Zrep the repulsive contribution, Zatt the attractive contribution, P absolute pressure, T the absolute temperature, R the gas constant, and a and b are parameters characterizing each compound. The repulsive contribution approaches infinity as the density approaches a close-packed value. The attractive part depends on temperature and approaches zero at infinite temperature. For pure fluids, a and b can be characterized by matching the critical criteria: ð@P=@rÞT ¼ 0; ð@ 2 P=@r2 ÞT ¼ 0: The resulting equations are aR2 Tc2 a ¼ Pc




1 9ð21=3  1Þ

 ;

RTc b ¼ Pc




ð21=3  1Þ 3



ð2Þ

where a is a special temperature dependent adjustable parameter. Soave’s principal modification of the Redlich–Kwong equation was to redefine a. Whereas, 2747

2748

Soave’s Modified Redlich–Kwong Equation of State

the original Redlich–Kwong definition was effectively: a


pffiffiffiffiffiffiffiffiffiffiffi Tc =T

ð3Þ

Soave modified the definition of a to be: a


h

 pffiffiffiffiffiffiffiffiffiffiffii2 1 þ k 1  T=Tc

ð4Þ

where k ¼ 0:480 þ 1:574o  0:176o2

ð5Þ

Note that the discussion so far relates only to the model for pure fluids, while the primary application is for VLE of mixtures. Extending the model to mixtures requires characterization of the a and b parameters for the mixture. These characteristic equations are called mixing rules. They are discussed in the section titled ‘‘Mixing Rules.’’ Soave’s method of developing this definition is instructive. First, he tabulated values of a that would exactly match the experimental vapor pressures of methane through n-hexane for Tr 2 [1.0, 0.3]. Then, he plotted these with respect to temperature. As that plot did not generate linear correlations, he plotted several other candidate relations. By shortening the temperature pffiffiffiffiffi range to Tr 2 [1.0, 0.45] and plotting pffiffiffi a vs. Tr , Soave obtained a simple linear trend. He noted that a was constrained to unity at the critical temperature by its definition, and that the acentric factor establishes a second point for this linear trend, resulting in the form of Eq. (4). This form is sufficient for ~3% accuracy in the vapor pressure of hydrocarbons, if the critical properties and acentric factor are accurately known. Note that the term ‘‘critical properties’’ in the present context refers to the critical temperature and pressure, not including the critical density. Matching the critical density might be considered an alternative version of three-parameter corresponding states that deserves a brief review in the present context. Hougen and coworkers explored this alternative for a number of years, but they focused on density in the critical region rather than vapor pressure. Density in the critical region is difficult to measure and fraught with theoretical challenges related to the nonclassical behavior of the coexisting densities in the critical region. Although relatively unimportant for most process simulations, density is important in the critical region for supercritical extractions. Despite these challenges, Lee and Kesler were able to develop an equation of state that fits the critical density and the vapor pressure curve. The Lee–Kesler equation interpolated between the equations of state for

methane and octane in the form of: Z ¼ ZCH4 þ oðZnC8  ZCH4 Þ

ð6Þ

By adapting the multiparameter Benedict–Webb– Rubin (BWR) equation for methane and n-octane as their basis, they approached an effective implementation of what might be called as four-parameter corresponding states. Unfortunately, their equation never achieved the widespread application of the Soave equation, despite having greater accuracy for liquid density and comparable accuracy for VLE. This observation may shed some light on an additional strength of the Soave equation. Soave’s equation was much simpler than the multiparameter BWR two-compound interpolation form of the Lee–Kesler equation. Soave’s equation can be rearranged into the form of a cubic polynomial, which can be solved analytically, eliminating the convergence difficulties that plagued early programmers. Thus, the popularity Soave’s equation relative to the Lee–Kesler equation is a testament to the merits of a simpler form. APPLICATIONS AND ADAPTATIONS OF SOAVE’S EQUATION Applications of the Soave equation are too numerous to list comprehensively. Soave’s original paper has been cited in roughly 1600 publications since 1980, and 88 times in 2003 alone. These citations include applications to petroleum recovery and refining, natural gas production, coal liquefaction, cryogenic gas separation, refrigeration, wax precipitation, hydrate formation, polymerization, interfacial tension, supercritical fluid processing, and petrochemical production. Interested readers should perform a ‘‘cited reference’’ search of the citation index to obtain a detailed list. Restricting the search to 2003 suffices to indicate the scope of coverage listed above. Beyond applications reported in the literature, process simulation is ubiquitous throughout the chemical industry and academia. Every time users select the Soave, or SRK, thermodynamics model, they apply the Soave equation. One motivation for this selection is the long experience with the model and the compilation of correction factors, when the basic model is deficient. For example, binary interaction parameters (kij’s) have been compiled for a large number of binary mixtures to improve VLE correlation.[5] It is also possible to compensate for inaccuracies in density through volume translations.[6] In light of the broad scope of applications and adaptations, the discussion here is restricted to general observations about the nature of adaptations and the accuracy to be expected. The Peng–Robinson equation

Soave’s Modified Redlich–Kwong Equation of State

2749

is included as an early adaptation as given below: Z


MIXING RULES

PV br a br ¼ 1 þ  RT 1  br bRT 1 þ 2br þ ðbrÞ2

ð7Þ

Some might consider the Peng–Robinson equation to be entirely independent of the Soave equation, but they have used exactly the same form for a that Soave developed in Eq. (4). Therefore, the only difference is the form of the denominator. The Peng–Robinson denominator results in slightly improved density correlations, especially for the low-boiling compounds of interest to the Gas Processors Association, sponsors of the Peng–Robinson research. In general however, density predictions are inaccurate, such that volume translations like those of Ref.[6] are still worthwhile. Moreover, comparisons generally indicate only marginal distinctions between the accuracy of the two equations for VLE correlations.[7,8] Hence adaptations like extended a correlations and advanced mixing rules are equally applicable to the original Soave form or the Peng–Robinson form with little distinction. We draw the line at the Peng–Robinson equation, however. Other equations of state might also be considered as incremental adaptations. For example, replacing the repulsive term with a more realistic hard sphere term might be considered to be a small adaptation, but that would be considered a major revision by this author. One further adaptation of Soave’s equation for pure fluids played a major role in maintaining the relevance of his model over the years. Soave himself was aware that his a correlation was deficient for mixtures involving hydrogen. This deficiency is not surprising when it is realized that, at room temperature, hydrogen’s reduced temperature approaches Tr  10. Graboski proposed an improved a correlation for hydrogen mixtures. Graboski also proposed a modified form of Eq. (5) based on optimizing the coefficients with a larger database than Soave’s. Graboski’s modification is occasionally referred to as the API equation. Soave’s original a correlation is also deficient for polar compounds. Whereas the slope of the vapor pressure curve for hydrocarbons tends to drop at low temperatures, it remains relatively constant for polar compounds. Several authors proposed a adaptations for the Soave and Peng– Robinson equations. Notable references should include Stryjek and Vera,[9] Patel and Teja,[10] Soave,[1] and references cited therein. The disadvantage of these adapted a correlations is that they are no longer predictive. Whereas the acentric factor and Eq. (5) suffice for the original model, specific regressions must be performed for each compound of interest, when applying adapted a correlations. Stryjek and Vera[9] have tabulated the parameters for roughly 50 common compounds in their adaptation for the Peng–Robinson equation.

Another major adaptation of Soave’s equation involved application to asymmetric nonideal mixtures. The original mixing rules proposed by Soave were in the quadratic form originally suggested by van der Waals as given below: b ¼ aij ¼

NC X

xi bi ;

i¼1

a ¼

NC X NC X

xi xj aij ;

i ¼ 1j ¼ 1

pffiffiffiffiffiffiffiffiffiffiffi aii ajj ð1  kij Þ

ð8Þ

Quadratic mixing rules are sufficient for symmetric deviations from ideality, but not for highly asymmetric mixtures like alcohols mixed with hydrocarbons. In symmetric mixtures, the kij parameter can be adjusted to characterize large nonidealities, but it primarily adjusts the magnitude of the Gibbs excess energy, and not the skewness. Negative values of kij indicate strong solvation interactions, as in inorganic acid þ water mixtures. Positive values of kij indicate weak solvation interactions, as in acetone þ hydrocarbon mixtures. In mixtures like alcohols þ hydrocarbons, both the magnitude and skewness of the Gibbs excess curve are affected, because the alcohols associate except at infinite dilution. Huron and Vidal[11] showed that equating the infinite pressure Gibbs energy of mixing to that of an activity model like the NRTL[12] or UNIQUAC[13] models provided a mixing rule that was sufficiently flexible to describe very complex phase behavior. With this modification, simple cubic equations like Soave’s could be applied to nearly any kind of mixture at any conditions, including supercritical conditions. The Huron–Vidal mixing rule combined with NRTL activity model is illustrated below. NC P

3 x G C j ji ji 7 NC 6a X a 1 j¼1 7 6 i ¼ xi 6  7; N C 5 4 P b ln 2 b i i¼1 xk Gki 2

b ¼

NC X

k¼1

xi bi

ð9Þ


 Cji Gji ¼ bj exp aji RT

ð10Þ

i¼1

where Cji, Cij, and aji are adjustable binary parameters. (There is no relation between aji of the mixing rule and Soave’s pure component a.) A notable adaptation that combines complex mixing rules, the Stryjek–Vera a correlation, and the Peng–Robinson equation is the PRWS model of Wong

S

2750

Soave’s Modified Redlich–Kwong Equation of State

and Sandler.[14] The principal difference from the Huron–Vidal rule is that the Wong–Sandler rule degenerates to the quadratic mixing rule at low density. Technically, quadratic mixing at low density is required for thermodynamic consistency, but the small magnitude of nonidealities in the vapor phase means that the inconsistencies there have small impact on the accuracy of the correlation. The Wong–Sandler mixing rule is

b ¼

Q 1  D

ð11Þ

a QD ¼ RT 1  D Q ¼

XX i

j

2

 xi xj b 

ð12Þ a  RT ij

pffiffiffi NC 6 a X 2 6 i pffiffiffi  D ¼ xi 6 4 bi RT lnð 2  1Þ i¼1 

ð13Þ NC P

3 xj Gji Cji 7 j¼1 7 7 N C 5 P xk Gki k¼1

a  b  RT ij   h  ai  aj i 1  kij ¼ bi  þ bj  2 RT RT

ð14Þ

ð15Þ

Note that kij values vary over large ranges (1, 0.5) in the PRWS model. An implementation of the PRWS model is available through the author’s website. A number of related efforts have studied the impact of applying the infinite pressure limit, or the zero pressure limit.[15] The zero pressure limit would appear to be closer to the basis applied in developing activity models. Once again, the more complex model is technically more correct, but the improvement in accuracy is small. A reasonable compromise in accuracy and simplicity is offered by the PRWS mixing rule. In general, the Huron–Vidal mixing rule provides flexibility but the optimal parameters for the activity model alone are not optimal when adapted to the advanced mixing rule, necessitating regression of each specific binary mixture. Michelsen and Dahl,[16] on the other hand, showed how UNIFAC predictions could be implemented directly into the mixing rules, circumventing the regression step. This is the basis of the predictive SRK model of Holderbaum and Gmehling.[17] Gmehling has published a recent review that favors adaptation of a volume translated Peng–Robinson

equation with temperature-dependent parameters in the complex mixing rules to improve excess enthalpy correlation.[18] His model is also capable of treating electrolytes and polymer solutions[19] within this general methodology. The inclusion of such a large number of adjustable parameters is made feasible by maintaining an extremely large database with perpetual global optimization of all parameters for all data. An alternative to incorporating empirical activity models into the mixing rules is to seek an alternative physical explanation of solution nonidealities. Hydrogen bonding models are noteworthy in this regard. Most of the nonideality in the interaction between water and oil, for example, can be attributed to water hydrogen bonding while oil cannot. A hydrogenbonding model would account for this observation explicitly, reducing the magnitude of the parameters in the mixing rules. Most inclusions of hydrogen bonding have also included corrections of the Van der Waals repulsive contribution,[20,21] and elimination of any parameter resembling Soave’s a correlation. Such extensive generalizations are judged to be outside the scope of the this entry. On the other hand, Tassios and coworkers[22] have managed to include a hydrogen bonding contribution while maintaining the Soave model for the remainder of the equation of state. Unfortunately, including the hydrogen bonding contribution does not guarantee elimination of the need for complex mixing rules. Hence, the benefit of including an additional term in the model is not readily apparent. Ultimately, this benefit may be recognized as molecular modeling becomes integrated with chemical process and product design. Its future prospects are discussed below.

FUTURE It is reasonable to ask, ‘‘What next? Is there anything left to be done?’’ Of course, the need for research is never ending. The primary criticism against the current incarnation of Soave’s initiative is the lack of a strong relation between the form of Soave’s equation and the rigorous physics that can be determined from statistical mechanics. For example, the van der Waals repulsive contribution is known to be inconsistent with molecular simulations.[23] Furthermore, pffiffiffiffiffi there is no basis in statistical mechanics for the Tr dependency in Soave’s a correlation. The presumption is that an improved physical basis would improve the capability for extrapolating the model to conditions at which experimental measurements are unavailable. So far, this presumption has been impossible to prove. By perpetually introducing new parameters and new data and reoptimizing, Gmehling and coworkers have demonstrated that adaptation of Soave’s equation is

Soave’s Modified Redlich–Kwong Equation of State

sufficiently flexible to correlate the vast majority of what is known about phase equilibria. One caveat in this regard would be variations related to ‘‘proximity effects.’’ These are changes that depend on details of the molecular structure that first order group contributions cannot address. One example would be differences between p-xylene and m-xylene. One obvious answer to this problem is higher order corrections for proximity effects, and these are on the way. By proceeding in this manner, it will be many years before such adaptations of Soave’s equation are supplanted. Alternatives to Soave’s approach and group contribution adaptations would be better focused on capabilities not offered by such approaches. For example, pure component properties like vapor pressure are assumed to be available when applying Soave’s methodology. In the coming world of chemical product design, this assumption may not be satisfactory. Molecular simulation offers the prospect of being able to make these predictions for transport properties as well as equilibrium properties.[24–27] Proximity effects would also be naturally included within molecular modeling. While the National Research Council has estimated that such predictive capability may not be available for a ‘‘decade or two,’’[28] viable preliminary versions may come much sooner than that. The rise of molecular product design will ultimately lead to questions about molecular simulations of mixtures, which make it difficult to rationalize distinctions between molecular scale models and models that derive their justification primarily from macroscopic measurements. When this happens, systematic analysis that can be rationalized on both the molecular and macroscopic scales will take precedence. It is in this context that hydrogen-bonding models will begin to play a more significant role. Such a development would present little difficulty to researchers like Gmehling because they could easily add new terms to their model equations. The effort will be motivated by complementary goals that can be achieved through a consistent molecular perspective applied to phase equilibria, transport phenomena, adsorption, membrane permeation, ion exchange, protein folding, and a host of other diverse applications. Thus, phase equilibrium can teach us things about molecular interactions that can be applied in fields that are far removed from traditional phase equilibrium applications. This is where the future lies and Soave’s equation has paved the way to such an ambitious goal.

CONCLUSIONS The SRK equation has played a major role in the development of chemical process design and will continue to play a similar role. Adaptations may obscure

2751

the role of the original model, but it can be discerned through the recollection of a brief history. These adaptations include the Peng–Robinson equation, extended vapor pressure correlations, the Huron–Vidal and Wong–Sandler mixing rules, and group contribution extensions by Gmehling and coworkers. The strength of the SRK approach has been a heavy emphasis on correlating key engineering data. As future adaptations continue to recognize which data are key and accurately correlate larger databases, the SRK approach will be perpetuated for many years to come.

REFERENCES 1. Soave, G. 20 years of Redlich–Kwong equations of state. Fluid Phase Equilibr. 1993, 82, 345–359. 2. Wilson, G.M. Vapor–liquid equilibria correlated by means of a modified Redlich-Kwong equation of state. Adv. Cryog. Eng. 1964, 9, 168–176. 3. Wilson, G.M. Calculation of enthalpy data from a modified Redlich-Kwong equation of state. Adv. Cryog. Eng. 1966, 11, 392–400. 4. Redlich, O.; Kwong, N.S. On the thermodynamics of solutions. V. An equation of state. Fugacities of gaseous solutions. Chem. Rev. 1949, 44, 233. 5. Elliott, J.R.; Daubert, T.E. Revised procedures for phase equilibrium calculations with the Soave equation of state. Ind. Eng. Chem. Proc. Des. Dev. 1985, 24, 743. 6. Peneloux, A.; Rauzy, E.; Freze, R. A consistent correction for Redlich–Kwong–Soave volumes. Fluid Phase Equilibr. 1982, 8, 7. 7. Graboski, M.S.; Daubert, T.E. A modified Soave equation of state for phase equilibrium calculations. I. Hydrocarbon systems. Ind. Eng. Chem. Proc. Des. Dev. 1978, 17, 443. 8. Yang, J.; Griffiths, P.R.; Goodwin, A.R.H. Comparison of methods for calculating thermodynamic properties of binary mixtures in the sub and super critical state: Lee–Kesler and cubic equations of state for binary mixtures containing either CO2 or H2S. J. Chem. Thermo. 2003, 35, 1521–1539. 9. Stryjek, R.; Vera, J.H. PRSV—An improved Peng–Robinson equation of state for pure compounds and mixtures. Can. J. Chem. Eng. 1986, 64, 323–333. 10. Patel, N.C.; Teja, A.S. New cubic equation of state for fluids and fluid mixtures. Chem. Eng. Sci. 1982, 37, 463–473. 11. Huron, M.J.; Vidal, J. New mixing rules in simple equations of state for representing vapour–liquid equilibria of strongly non-ideal mixtures. Fluid Phase Equilibr. 1979, 3, 255.

S

2752

12. Renon, H.; Prausnitz, J.M. Estimation of parameters for the NRTL equation for excess Gibbs energies of strongly nonideal liquid mixtures. Ind. Eng. Chem. Proc. Des. Dev. 1969, 8, 413. 13. Abrams, D.S.; Prausnitz, J.M. Statistical thermodynamics of liquid mixtures: a new expression for the excess Gibbs energy of partly or completely miscible systems. AIChE J. 1975, 21, 116. 14. Wong, D.S.H.; Sandler, S.I. A Theoretically correct mixing rule for cubic equations of state. AIChE J. 1992, 38, 671. 15. Twu, C.H.; Coon, J.E.; Bluck, D.; Tilton, B. CEOS=AE mixing rules from infinite pressure to zero pressure and then to no reference pressure. Fluid Phase Equilibr. 1999, 158, 271. 16. Michelsen, M.; Dahl, S. High-pressure vapor– liquid equilibrium with a UNIFAC-based equation of state. AIChE J. 1990, 36, 1829–1836. 17. Holderbaum, T.; Gmehling, J. PSRK: a group contribution equation of state based on UNIFAC. AIChE J. 1991, 70, 251–265. 18. Gmehling, J. Potential of group contribution methods for the prediction of phase equilibria and excess properties of complex mixtures. Pure Appl. Chem. 2003, 75, 875–888. 19. Wang, L.S.; Ahlers, J.; Gmehling, J. Development of a universal group contribution equation of state. 4. Prediction of vapor–liquid equilibria of polymer solutions with the volume translated group contribution equation of state. Ind. Eng. Chem. Res. 2003, 42, 6205–6211. 20. Chapman, W.G.; Gubbins, K.E.; Jackson, G.; Radosz, M. New reference equation of state for associating liquids. Ind. Chem. Eng. Res. 1990, 29, 1709.

Soave’s Modified Redlich–Kwong Equation of State

21. Elliott, J.R.; Suresh, S.J.; Donohue, M.D. A simple equation of state for nonspherical and associating molecules. Ind. Eng. Chem. Res. 1990, 29, 1476. 22. Voutsas, E.C.; Kontogeorgis, G.M.; Yakoumis, I.V.; Tassios, D.P. Correlation of liquid–liquid equilibria for alcohol=hydrocarbon mixtures using the CPA equation of state. Fluid Phase Eq. 1997, 132, 61–75. 23. Honnnell, K.G.; Hall, C.K. A new equation of state for athermal chains. J. Chem. Phys. 1989, 90, 1841. 24. Chen, B.; Potoff, J.J.; Siepmann, J.I. Monte Carlo calculations for alcohols and their mixtures with alkanes. Transferable potentials for phase equilibria. 5. United-atom description of primary, secondary, and tertiary alcohols. J. Phys. Chem. B. 2001, 105, 3093–3104. 25. Martin, M.G.; Siepmann, J.I. Transferable potentials for phase equilibria. 1. United-atom description of n-alkanes. J. Phys. Chem. B. 1998, 102, 2569–2577. 26. Bourasseau, E.; Haboudou, M.; Boutin, A.; Fuchs, A.H.; Ungerer, P. New optimization method for intermolecular potentials: optimization of a new anisotropic united atoms potential for olefins: prediction of equilibrium properties. J. Chem. Phys. 2003, 118, 3020–3034. 27. Unlu, O.; Gray, N.; Gerek, Z.N.; Elliott, J.R. Transferable step potentials for the straight chain alkanes, alkenes, alkynes, ethers, and alcohols. Ind. Eng. Chem. Res. 2004, 43, 1788–1793. 28. Breslow, R.; Tirrell, M.V., Eds. Beyond the Molecular Frontier—Challenges for Chemistry and Chemical Engineering; The National Academies Press: Washington, DC, 2003.

Solid–Liquid Mixing: Numerical Simulation and Physical Experiments

S

Philippe A. Tanguy Francis Thibault Gabriel Ascanio URPEI, Department of Chemical Engineering, Ecole Polytechnique, Montreal, Quebec, Canada

Edmundo Brito-de La Fuente Departamento de Alimentos y Biotecnologı´a, UNAM, Me´xico, Me´xico

INTRODUCTION Solid–liquid mixing processes find numerous applications in industry. The modeling of these mixing operations using first principles is still limited, although some progress has been made in recent years. This study assesses the real capability of the networkof-zone approach in the case of a complex mixing problem involving a coaxial mixer. Coaxial mixers are very popular for the preparation of pastes and slurries in the chemical, food, and coating industries. The mixer considered is composed of two rotating shafts: a fast-driven shaft supporting an open impeller and a slow shaft driving a scraping anchor arm.

LITERATURE SURVEY The dispersion of solids in liquids, the preparation of solid suspension in liquid media, and the make-down of pigment slurries in agitated vessels are typical solid–liquid mixing problems that find numerous applications in the process industries. For instance, they are involved in the preparation of paints and coatings, the manufacturing of food products, as well as suspension polymerization. In solid–liquid mixing design problems, the main features to be determined are the flow patterns in the vessel, the impeller power draw, and the solid concentration profile versus the solid concentration. In principle, they could be readily obtained by resorting to the CFD (computational fluid dynamics) resolution of the appropriate multiphase fluid mechanics equations. Historically, simplified methods have first been proposed in the literature, which do not use numerical intensive computation. The most common approach is the dispersion–sedimentation phenomenological model. It postulates equilibrium between the particle flux due to sedimentation and the particle flux resuspended by the turbulent diffusion created by the rotating impeller. Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120018056 Copyright # 2006 by Taylor & Francis. All rights reserved.

Based on this concept and assuming a one-dimensional distribution of particles along the vessel height, an analytical expression for the concentration profile can be derived: f Pes ¼ expð
Pes z=HÞ
1
expð
Pe f sÞ

ð1Þ


represents the average volume fraction of the where f suspension, z the vertical coordinate, H the fluid height in the vessel, and Pes the Peclet number of the solid particles. The Peclet number is defined as Pes ¼ vt  H=De;p

ð2Þ

where vt is the settling velocity at equilibrium and De,p the turbulent diffusion coefficient. This dimensionless number is a model fitting parameter, and it is generally correlated with the operating conditions and the physical properties of the suspension.[6] The model has been verified experimentally in a tank provided with four Rushton turbines (radial discharge impeller) at a very low solids concentration.[1,2] Several improvements have been proposed in the literature to make the model more general. Ferreira, Rasteiro, and Figueiredo[3] introduced new terms in Eq. (1) to take into account the radial variation of the particle concentration. Rasteiro, Figueiredo, and Friere[4] suggested the use of the Richardson and Zaki’s expression[5] for the settling velocity, as this velocity is strongly dependent on the solid volume fraction in the suspension. The application of CFD in the modeling of solid– liquid mixing is fairly recent. In 1994, Bakker et al.[6] developed a two-dimensional computational approach to predict the particle concentration distribution in stirred vessels. In their model, the velocity field of the liquid phase is first simulated taking into account the flow turbulence. Then, using a finite volume approach, the diffusion–sedimentation equation along with the convective terms is solved, which includes Ds, a 2753

2754

Solid–Liquid Mixing: Numerical Simulation and Physical Experiments

turbulent diffusion coefficient of the particles, defined as: Ds ¼

pffiffiffiffi kt 3pnp D2p

ð3Þ

where np is the overall volume of the particles, Dp their diameter, and kt the turbulent kinetic energy density. As the authors did not use iterative coupling between the computation of the flow field and that of the solids concentration, the effect of the solids on the fluid mechanics in the vessel was not taken into account. As an example, they investigated the suspension of 20 mm particles with several radial discharge impellers. They showed that their approach was capable of predicting the solid–liquid interface in the vessel and the effect of the position and number of turbines on the solid distribution. Unfortunately, no experimental validation of this approach was carried out. All the above models do not consider the particle– particle interactions, although these interactions influence the settling velocity and ignore the effect of the solid phase on the hydrodynamics in the vessel. As a consequence, the practical range of application is restricted to low solids concentrations. A particle migration model was proposed by Gadala-Maria and Acrivos[7] to describe experimental shear-induced migration observations. This model allows for a better understanding of the shear effects on particle diffusion for concentrated suspensions.[8–11] Based on these studies, a conservation equation for the solid phase was established by Phillips, Amstrong, and Brown,[12] which takes into account convective transport, diffusion due to particle–particle interactions, and the variation of viscosity within the suspension, namely:   @f þ v  Hf ¼ H  Kc a2 fHðj_gjfÞ þ H @t   a2 2 @Z Hf  KZ j_gjf @f Z

ð4Þ

where f is the volume fraction, v the suspension velocity, a the particle radius, jg_ j the magnitude of the rate-ofstrain tensor. Kc and KZ are empirical constants equal to 0.41 and 0.62, respectively. In order to describe the variation of the viscosity Z with the particle volume fraction, the authors suggested the use of the Krieger– Dougherty phenomenological model.[13] This approach was applied to investigate the diffusion of suspensions consisting of poly(methylmethacrylate) monodisperse particles (Dp ¼ 675 mm) at high concentration (f > 45%) in Newtonian silicon oil. Two geometries were tested: a Couette flow (flow between parallel plates with one plate in motion) and a Poiseuille flow (flow in a cylindrical channel). For these two cases, as the concentration

varies only in the radial direction, several analytical expressions could be established for the solid volume fraction and the suspension velocity profile. The computational results were compared to concentration measurements based on nuclear magnetic resonance and a qualitative agreement was obtained. It should be noted here that in principle Eq. (4) can be applied at any solids concentration, but it is however restricted to noncolloidal systems. A different numerical strategy to simulate multiphase mixing was introduced by Mann [14] and Mann and Hackett.[15] The idea of the method, called the network-of-zone, is to subdivide the flow domain in a set of small cells assumed to be mixed perfectly. The cells are allowed to exchange momentum and mass with their neighboring cells by convective and diffusive fluxes. Brucato and Rizzuti[16] and Brucato et al.[17] applied this idea to the modeling of solid–liquid mixing. An unsteady mass balance for the particles was derived to estimate the solid distribution in the vessel, namely:

Vc

2 X dCc ¼ QðCp
Cc Þ þ aQðCi
Cc Þ dt i¼1 5 X ðvt Si Ci
vt Sc Cc Þ þ i¼3

ð5Þ

where the subscript c is the cell on which the balance is applied, the subscript i refers to the adjacent cells, and the subscript p denotes the feeding cell, i.e., the one yielding convective momentum, V is the cell volume, C the concentration in the cell, S the cell surface area where the particles settle from, Q the volume flow rate of fluid entering the cell, a an adjustable parameter that describes the turbulent diffusion, and vt the particle settling velocity. In this model, the sedimentation and the diffusive flow were in the vertical direction, and the convective flow was radially oriented. These assumptions were justified on the basis of the radial discharge flow generated by the impellers. Brucato et al.[17] verified the model prediction with the experimental data of Fajner et al.[2] and Magelli et al.[1] It was shown that the axial concentration profile was very well predicted, however the validation was limited again to extremely low concentration suspensions. The attractiveness of the network-of-zone method to compute solid–liquid mixing flows resides in its relative simplicity while being capable of capturing the main flow phenomena for a wide range of concentrations. The objective of the present work is to assess the real capability of this approach in the case of a complex mixing problem involving a coaxial mixer. Coaxial mixers are very popular for the preparation of pastes and slurries in the chemical, food, and coating industries. Another mixer setup is also tested,

Solid–Liquid Mixing: Numerical Simulation and Physical Experiments

which consists of a single marine propeller rotating in a vessel (see description in the appropriate section below). This setup is used to validate the numerical model.

DESCRIPTION OF THE COAXIAL MIXER SETUP The experimental setup is shown in Fig. 1. It consists of the following items: (1) an AC drive with a nominal speed of 1760 rpm; (2) a gearbox with a reduction ratio 3.53:1 yielding a rotating speed of 500 rpm; (3) a torquemeter with a measurement range between 0 and 22.6 Nm (accuracy of  0.1% at full scale); and (4) a transparent vessel with a hemispherical bottom.

2755

Table 1 Operating range of the agitators Speed ratio Nc /Na

Na (rpm)

Nc (rpm)

0 (anchor only)

(0–125)

0

4

(0–125)

(0–500)

8

(0–62.5)

(0–500)

24

(0–20.8)

(0–500)

1

0

(0–500)

The vessel diameter Dc and the height of the cylindrical section Hc are both equal to 40.64 cm, yielding a maximum volume of fluid of about 60 L. The impeller configuration is the following:

VARIABLE SPEED: ANCHOR 0-125 RPM GEARBOX

TACHOMETER TORQUEMETER

GEARBOX MOTOR 3 Hp, 1760 RPM

VESSEL

Fig. 1 Coaxial mixer setup.

S

2756

Solid–Liquid Mixing: Numerical Simulation and Physical Experiments

Fig. 2 Marine propeller mixer.

An anchor arm:  Diameter Da ¼ 36.83 cm  Width Wa ¼ 3.81 cm  Wall clearance Cw ¼ 1.9 cm Four rigid rods at 90 used for pigment wetting:[18]  Length Dt ¼ 23.77 cm  Cross-section diameter Dst ¼ 0.95 cm A pitched blade turbine with two blades at 45 :    

Diameter Dp ¼ 20.2 cm Width Wp ¼ 5 cm Length Lp ¼ 7 cm Bottom clearance Cb ¼ 20.32 cm.

This configuration yields the following dimensionless ratios Cw=Dc ¼ 0.047, Da=Dc ¼ 0.906, and ANCHOR

Wa=Da ¼ 0.103 for the anchor, Dt=Dc ¼ 0.585 for the rods; and Dp=Dc ¼ 0.5 and Cb=Dc ¼ 0.5 for the pitched blade turbine. In this coaxial mixer, the primary role of the anchor is to clean up the wall from any accumulated solid lumps and reincorporate them back in the bulk. It also acts as a moving baffle, hampering the creation of a vortex at the liquid free surface. The purpose of the pitched blade turbine is to provide axial pumping so as to promote the resuspension of the solids, and radial dispersion to avoid solids reagglomeration. Finally, the aim of the wetting rods is to ease hydrophobic pigment incorporation by avoiding the creation of surface lumps.[19] Two driving shafts are installed in this mixer: a fast rotating shaft drives the four rods and the turbine in a counterclockwise direction at speed Nc, whereas a slow rotating shaft entrains the scraping anchor in the clockwise direction at speed Na. The operating range used in this work was as described in Table 1. In order

ROD + PBT

CONTROL POINTS

Fig. 3 Virtual finite elements method concept in 2-D.

Solid–Liquid Mixing: Numerical Simulation and Physical Experiments

2757

to investigate the particle motion inside the coaxial mixer, a Newtonian solution of corn syrup with a viscosity of 1.05 Pa.s and a density of 1360 kg=m3 was used in conjunction with red Ballotini glass beads. The beads had an average diameter of 1 mm and a density of 2500 kg=m3. The maximum packing factor (fm) was 0.6.

resuspending mechanism in such a mixer. The operating conditions were as follows: N ¼ 173 rpm
¼ 2.8%; N ¼ 230 rpm and f
¼ 7.1%; and and f
N ¼ 350 rpm and f ¼ 11.9%. For all the experiments, the particles were initially at rest on the tank bottom and the stirrer was suddenly set in motion.

MARINE PROPELLER MIXER

Computational Model

This simple mixing system (Fig. 2) involves a marine propeller in an unbaffled vessel (actually a laboratory beaker). The geometrical characteristics are the following:

Let us consider the incompressible flow of a suspension in a given domain O. The governing equations are:   @v þ v  grad v þ grad p þ div t rm @t

 Distance shaft-impeller edge ¼ 2.6 cm  Blade diameter ¼ 1.8 cm  Bottom clearance ¼ 2.6 cm

¼ rm g

in O

ð6Þ

div v ¼ 0

in O

ð7Þ

where v, p, g, and r are the velocity, pressure, gravity, and specific gravity, respectively. For a solid–liquid medium, the density rm can be expressed as:

The propeller rotates clockwise in a down-pumping mode. The vessel (glass beaker) has a diameter of 7.2 cm and the fluid height is 6.5 cm corresponding to a stirred volume of 264 cm3. The same fluid as the one considered in the coaxial mixer experiments was used. The rotating speed and the volume concentration of the particles were varied in order to investigate the

rm ¼ rl ð1
fÞ þ rs f

where rl and rs represent the density of the liquid phase and the solid phase, respectively.

Vs

i=3

i=2

j

i=4 i=1

z V (x,y,z)

y x

ð8Þ

Fig. 4 Tetrahedral finite volumes.

S

2758

Solid–Liquid Mixing: Numerical Simulation and Physical Experiments

The stress tensor t in Eq. (6) is related to the rate-ofstrain tensor by a rheological equation of state such as: t ¼
2Zs g_

ð9Þ

where Zs is a function of jg_ j and f, g_ ¼

1 2 ½grad v

þ ðgrad vÞT 

ð10Þ

The finite element method was used for the discretization of the flow equations. Considering the complex kinematics in the coaxial mixer and the associated change of topology at each time step, a new mesh should a priori be built for every topology considered in the time discretization. As a large number of time steps would be required to depict the agitator kinematics accurately, this approach would be a tremendous chore. To alleviate this difficulty, several alternatives have been proposed in the literature:

and jg_ j ¼ ð_g : g_ Þ

1 2

ð11Þ

The suspension viscosity Zs may or may not be a function of jg_ j depending on the rheological behavior of the suspending medium. In a nondilute suspension, it is, however, always a function of the particle volume fraction f. In this work, the Krieger–Dougherty model for a Newtonian suspension was used: Zs ðfÞ ¼ Zl ð1
f=fm Þ
1;82

ð12Þ

where Zl is the viscosity of the suspending liquid and fm the maximum packing factor. For mathematical convenience, boundary conditions and initial conditions must be prescribed. For the simple marine propeller problem, a Lagrangian viewpoint was adopted. The frame of reference was attached to the propeller so that the propeller was fixed but the vessel was rotating. The boundary condition was then a zero velocity on the impeller, while the vessel wall rotated at
Oimpeller. The free surface was considered to be flat, therefore the normal velocity was zero and a shear-free condition was assumed. It should be noted that in the Lagrangian viewpoint, the frame of reference is in rotation. The fluid is therefore subjected to a constant acceleration and the momentum conservation equation [Eq. (6)] must be modified to account for centrifugal forces and Coriolis forces.[20] An advantage is, however, that the flow can be solved numerically at steady state provided the flow is fully periodic, which limits the computational efforts significantly. In the case of the coaxial mixer, the rotation kinematics is much more complex since the two sets of agitators counter-rotate at different speeds. For the sake of simplicity, we decided to simulate the flow using the frame of reference of the anchor. In this Lagrangian viewpoint, the anchor is fixed but the vessel wall rotates at
Oanchor and the turbine rotates at Oanchor þ Oturbine. In such a situation, contrary to the simple propeller problem, the resolution of the flow equations is time-dependent as the position of the central agitator changes with time.

1. The description of the agitator by momentum sources or sinks inside the domain.[21] The major drawback of this method is the evaluation of a force equivalent to the representation of a body in rotation. 2. The arbitrary Euler–Lagrange method. It consists of moving the finite element mesh nodes with time.[22] This method works well as long as the mesh is not too distorted. In practice, remeshing is usually required after a few time steps. 3. Domain decomposition with sliding meshes.[23] This strategy is very popular in the finite volume literature and implemented in finite volume-based software (FluentTM, CFXTM, and Star-CDTM). The idea is to decompose the flow domain into several concentric cylindrical meshes and allow slipping of the meshes between the partitions. The continuity of the solution at the mesh interface is imposed by

Fig. 5 Mesh of the propeller.

Solid–Liquid Mixing: Numerical Simulation and Physical Experiments

2759

S

Fig. 6 Suspending mechanism as a function of time.

conservative interpolation. This method is powerful but limited to simple agitator configurations. It seems to work when very fine meshes are used although no error analysis has been published so far. 4. Mesh superimposition. This technique used in commercial finite element CFD software, like FIDAPTM and PolyFlowTM, consists of generating a volume mesh without the moving impellers

and a surface mesh of each impeller. At each time step, the surface mesh is projected in the volume mesh and a procedure has been developed to determine if the nodes of the volume mesh are located inside the surface mesh. When it is the case, the velocity of the impeller is imposed on these nodes. This technique is fairly simple to implement, however, it does not allow a precise representation of the impeller shape.

2760

Solid–Liquid Mixing: Numerical Simulation and Physical Experiments

Table 2 Variation of the average concentration with the rotation speed Average volume fraction

Predicted volume fraction

173

2.8%

0.5%

230

7.1%

1.7%

350

11.9%

7.3%

N (rpm)

[24]

5. The virtual finite element method VFEM. The method that was chosen for this work has been developed specifically to simulate flows around moving bodies in fixed enclosure. The principle of VFEM is the following: a volume mesh of the vessel with the anchor arm is first generated. Then, the surface of the moving bodies (in our case, the impeller and the rods) is meshed; the discretization nodes generated being stored as control nodes. Knowing the kinematics of the impellers, the velocity is imposed on these control nodes as constraints in the momentum equation and their position is updated with time according to the impeller rotation. A constrained optimization technique based on Lagrange multipliers and a penalty strategy is used to impose the velocity constraints. The VFEM approach was applied by Bertrand, Tanguy, and Thibault[24] to investigate the complex flow patterns in a planetary kneader and by Tanguy and Thibault[25] for the characterization of the hydrodynamics in a coaxial mixer provided by a helical ribbon and a Rushton turbine. Contrary to the mesh superimposition technique, the VFEM does not require that the control nodes coincide with the volume mesh nodes. They can be located inside finite elements as they are treated like external solicitations or optimization constraints. Fig. 3 illustrates the VFEM concept in 2-D. The above method was implemented in our POLY3-DTM CFD finite element code using unstructured meshes made of tetrahedral elements. The reader is referred to Bertrand, Tanguy, and Thibault[24] for detailed information. The modeling of solid–liquid mixing requires an additional equation to predict the dispersion of the solid phase in the vessel. As mentioned before, the network-of-zones approach was used in this work, which is based on unsteady mass balances on the solid phase carried out on a set of cells. In the literature, such mass balances are predominantly made on regular cells (finite volumes) in structured grids. The cells typically consist of quadrangles in 2-D and hexahedra in 3-D. In the present work, due to the use of unstructured

grids, the mass balances were performed on the same elements as those used for the resolution of the flow equations, i.e., tetrahedral finite volumes (Fig. 4). Let us consider the following mass balance on a tetrahedral element subjected to the velocity field v(x,y,z) of the suspension (as computed from the solution of the flow equations) and a sedimentation velocity, namely: Vj

X dfj ¼ Qcj;i fcj;i þ Qsj;i fsj;i dt i

ð13Þ

where the subscript i represents the neighboring finite elements adjacent to the four sides of finite element j, Vj the volume of finite element j, Qcj,i the convective flux of solid particles going through a common face to elements i and j, Qsj,i the sedimentation flux going

Fig. 7 Surface mesh of the moving impellers.

Solid–Liquid Mixing: Numerical Simulation and Physical Experiments

through a common face to elements i and j, and fcj ,i and fsj ,i the volume fraction of solid particles defined as follows:fcj ,i ¼ fI, if Qcj,i > 0 is the convective flux of particles leaving element i and entering element jfc j ,i ¼ fj, if Qcj,i < 0 is the convective flux of particles leaving element j and entering element i fs j ,i ¼ fi, if Qsj,i > 0 is the sedimentation flux of particles leaving element i and entering element j fs j ,i ¼ fj, if Qsj,i < 0 is the sedimentation flux of particles leaving element j and entering element i. This model assumes perfect mixing inside the element, as the particles enter the cell with different concentrations but leave the cell with a homogeneous concentration. Moreover, inertia is neglected, as well as slipping, between the solid and liquid phases. In Eq. (13), the fluxes can be expressed using the following equations: Qcj;i ¼

Z

Qsj;i ¼

Z

 vcj;i ðx; y; zÞ  ! n j;i dS

ð14Þ

 vsj;i  ! n j;i dS

ð15Þ

Sj;i



Sj;i



where ~ n represents the unit vector of each face pointing to inward element j, S the surface area of the face, vc the suspension velocity, and vs the settling velocity determined using the classical sedimentation relations,[26] namely: Ar ¼ 24 Rep

ðAr < 4; 8Þ

ð16Þ

Ar ¼ 24 Rep þ 3; 6 Rep 1:687 ð4; 8 < Ar < 105 Þ

ð17Þ

Ar ¼ 4=9 Re2p

ð18Þ

ðAr > 105 Þ

where Ar is the Archimedes number defined as: Ar ¼

4 3 rg Dp ðrs
rl Þ l2 3 ml

ð19Þ

and Rep is the particle Reynolds number defined as: Rep ¼

Dp vl rl ml

Table 3 Characteristics of the surface meshes Number of elements

Number of control points

400

208

Mixing rods

5022

2279

PBT

1968

884

Surface mesh Shaft

and ct ¼ 1=2 for the subsequent steps (second order Gear scheme). After substitution in Eq. (13), the following implicit numerical problem is obtained: " # at ftþ1 þ bt ft þ ct ft
1 j Vj Dt X tþ1 ¼ Qcj;i fcj;i þ Qsj;i ftþ1 ð22Þ sj;i i

This problem can then be resolved with the appropriate initial conditions of the mixing problem considered. Several mixing cases were considered in this work:

1. Simple solid–liquid mixing experiments for the validation of the numerical model 2. Hydrodynamic studies in the coaxial mixer without solid particles 3. Coaxial solid–liquid mixing experiments. VALIDATION OF THE NUMERICAL MODEL The propeller setup was used for this purpose. From a computational standpoint, a mesh of the vessel– propeller set was created containing 8746 elements yielding 54,333 velocity equations and 8746 concentration equations. The surface mesh of the propeller (Fig. 5) comprised 964 control points. A maximum of three control points per element was used to avoid locking. Unsteady state flow simulations were performed with a 1-s time step and three coupling iterations between the Navier–Stokes equations and the solid transport equation were required per time step. Steady state was deemed obtained when the solids concentration coefficient of variation did not change. Fig. 6 illustrates the suspending mechanism versus time. At 350 rpm, all the particles are suspended

ð20Þ

In order to compute the unsteady term df=dt, the following expression can be used: df at ftþ1 þ bt ft þ ct ft
1 ¼ dt Dt

2761

ð21Þ

where at ¼ 1, bt ¼ –1 and ct ¼ 0 at the first time step (Euler implicit scheme), and at ¼ 3=2, bt ¼ –2

Table 4 Numerical and experimental values of Kp for an anchor Kp (numerical) Kp (experimental) This work

256

253

Tanguy, Thibault, and Brito de la Fuente[31]

206

199

Ho and Kwong[32]

—

215

Rieger and Novak[33]

—

206

S

2762

Solid–Liquid Mixing: Numerical Simulation and Physical Experiments

Fig. 8 Numerical and experimental power curve for three different speed ratios.

and uniformly distributed. An attempt has been made to evaluate the minimum suspending speed Njs using the work of Armenante and Nagamine.[27] We found that Njs ¼ 855 rpm for an average particle concentration of 11.9% v=v, which seems to contradict our visual observations. However, as mentioned by Ibrahim and Nienow,[28] published Njs correlations largely overestimate the suspending velocity when the suspending medium viscosity is larger than 0.1 Pa.s. Fig. 6 also shows the time evolution of the computed volume fraction until stability is reached. As we do not know the experimental solids concentration distribution in the vessel, the numerical results can only be compared with the experimental results on a qualitative basis. We noted that the computation allows the prediction of the solid accumulation below the agitator. In agreement with the experiments, the network-of-zone model also predicts an increase in the resuspension mechanism when the rotation speed and=or the average concentration in the vessel increase (Table 2). The accumulation of solids in the vessel bottom at equilibrium was also well captured. For instance at N ¼ 230 rpm and a volume fraction of 7.1%, a solid layer with a volume fraction greater than 60% accumulates at the bottom and at the wall, which has been observed experimentally. At N ¼ 350 rpm and a volume fraction of 11.9%, almost all the solid particles are suspended, except in a small region close to the bottom edge, again in agreement with the experiments. The performance of

the computational model appears therefore satisfactory as fine hydrodynamic details can be predicted for intermediate concentration values. At very low concentration, as the solid layer thickness is reduced, a finer mesh would be required to enhance the model precision.

COAXIAL MIXER RESULTS AND DISCUSSION Considering the prediction of the power consumption, classically, the power drawn by the impeller is expressed with power curves, i.e., plots of the power number Np versus the Reynolds number, Re, where: Np ¼

P rN 3 D5

ð23Þ

Re ¼

rND2 m

ð24Þ

Table 5 Numerical and experimental values of Kp for the coaxial mixer Speed ratio, RN

Kp (numerical)

Kp (experimental)

0

256

253

4

1003

817

8

2651

2284

24

17,411

16,486

Solid–Liquid Mixing: Numerical Simulation and Physical Experiments

2763

S

Fig. 9 Effect of speed ratio on the flow field (upper part of the vessel).

P is the mixing power, N and D the rotation speed (in rps) and the impeller diameter respectively, and m and r the fluid viscosity and density. The following relations apply for the laminar and turbulent regimes, respectively: Kp ¼ Np Np ¼ C

Re ¼

P ; mN 2 D3

Re > 300

Re < 10

ð25Þ ð26Þ

where Kp and C depend only on the mixer geometry for a given fluid. In a coaxial mixer, the selection of the characteristic speed and dimension that appear in the expression of these dimensionless numbers is ambiguous, as we have

two different speeds (Na and Nc) and three impeller diameters that can be considered. In this work, we used Na and Da as the characteristic parameters. The reader is referred to the discussion of this particular choice in Refs.[29,30]. The coaxial mixer power curve and the value of Kp have been obtained by numerical simulation, varying the impeller speed and the speed ratio. For each simulation, the velocity field was used to compute the power by a macroscopic energy balance, namely: P ¼


ZZZ

O

t : HvdO

ð27Þ

Numerical simulations employed the virtual finite element method described above combined with a

2764

Lagrangian flow description (anchor frame of reference). The finite element mesh (generated by IDEASTM software from SDRC) included 17,083 tetrahedral elements, yielding 106,295 velocities degrees of freedom. Twenty time steps per revolution (angular displacement of 18 degrees per time step) were used, the value of the time step depending on the revolution speed. Typically, the time step was in the range 0.03–0.135 s. For each time step, the control points located on the surface mesh of the moving impellers (Fig. 7 had to be updated. The number of control points required is given in Table 3. The flow simulations carried out for several revolutions showed that the flow was periodic. Moreover, it was found that only one single revolution was enough to obtain a stable, converged solution at low Reynolds number. Finally each simulation

Solid–Liquid Mixing: Numerical Simulation and Physical Experiments

required between 12 and 24 CPU hours on an IBM RISC6000=550 server. Table 4 shows a comparison of the computed and experimental values of Kp for the anchor only. An excellent agreement is obtained. The comparison with literature data shows that the computed values are larger than the data published. The difference is believed to originate from the shape of the vessel bottom (hemispherical in the present work, flat in the literature results). We show in Fig. 8, a comparison of the numerical and experimental power curves for three different speed ratios (RN ¼ Nc=Na ¼ 4, 8 and 24). These results have been obtained with m ¼ 15 Pa.s and r ¼ 1500 kg=m3. It can be seen that there is good agreement between the predicted and experimental

Fig. 10 Effect of speed ratio on the flow field (lower part of the vessel).

Solid–Liquid Mixing: Numerical Simulation and Physical Experiments

2765

1.00

S

0.90 0.80 0.70

z/H

0.60 0.50 0.40 0.30 0.20

Rn Rn Rn Rn

0.10

= = = =

0 4 8 24

0.00 0.000

0.005

0.010

0.015

0.020

0.025

0.030

(m/s)

Fig. 11 Axial velocity in the upward direction as a function of the speed ratio.

values. The slope of –1 in the laminar region is captured by the simulation. The onset of the transition regime (decrease of the slope of the Np versus Re curve) occurs for a Re value below 10, and the threshold value is sensitive to the value of RN. For a given Re value, the power increases with RN. This result seems logical as the Reynolds number has been defined with the anchor parameters and when RN increases, the central shaft speed rotates faster. This speed increase enhances the average shear rate in the tank, which entrains an increase in the power draw. The values of the constant Kp of the coaxial mixer versus RN were also established. Results are shown in Table 5. Here again, the agreement between the predictions and the experimental data is very good. It is usual in laminar mixing simulations to represent the flow using tracer trajectories. The computation of such flow trajectories in a coaxial mixer is more complex than in traditional stirred tank modelling due to the intrinsic unsteady nature of the problem (evolving topology, flow field known at a discrete number of time steps in a Lagrangian frame of reference). Since the flow solution is periodic, a node-by-node interpolation using a fast Fourier transform of the velocity field has been used, which allowed a time continuous representation of the flow to be obtained. In other words, the velocity at node i was approximated

with a Fourier series taking the following form: vi ¼ ai0 þ ain cos nt þ

n
1 X k¼1

ðaik cos kt þ bik sin ktÞ ð28Þ

where n is the number of harmonics and coefficients aki, k ¼ 1, 2, . . . ,n–1, and bki, k ¼ 1,2, . . . , n–1 are obtained from the whole set of velocity values at node i during an impeller revolution. In practice, 10 harmonics were employed according to the Shannon sampling theorem.[34] We show in Figs. 9 and 10 the effect of the speed ratio on the flow field for an anchor speed of 4.43 rpm. It can be seen that for the two injection points considered (upper part and lower part of the vessel), the radial dispersion increases significantly with the speed ratio. The axial dispersion is also enhanced but less dramatically. A speed ratio of four does not lead to good dispersion and therefore should yield longer mixing times. These results show the synergy between the anchor and the turbine, the axial pumping increasing with the speed ratio. To quantify this axial pumping, the axial velocity in the upward direction is plotted in Fig. 11 versus the speed ratio. An additional advantage of the coaxial mixer can be seen as the axial

2766

Solid–Liquid Mixing: Numerical Simulation and Physical Experiments

Fig. 12 Predicted and experimental solid volume fraction at equilibrium.

pumping in the upper part of the vessel is enhanced when the speed ratio is augmented. We now turn to the prediction of the suspension mechanism of the ballotini versus the speed ratio in the coaxial mixer. The average volume concentration is 1%, and the solids are initially at rest in the tank bottom. The first case investigated corresponds to the motion of the sole anchor arm at a speed of 40 rpm. Simulations are carried out in the Lagrangian frame of reference (fixed anchor, rotating vessel). Fig. 12 shows the predicted and experimental solid volume fraction at equilibrium. The computation of the solid–liquid interface at the bottom is fairly well

captured by the numerical model as in the previous case dealing with the propeller. The next case considered is the resuspension with the only central shaft in rotation at 160 rpm. This simulation has been carried out with the virtual finite element method described before. Steady state was achieved after 20 revolutions. In Fig. 13, we compare the numerical and experimental distribution of particles in the tank. The agreement is again noteworthy, and the computation predicts that no particle has been resuspended. In fact, although the rotation is counterclockwise, particles have moved in the clockwise direction and accumulated behind the anchor arm. The Mann model is therefore capable of

Fig. 13 Numerical and experimental distribution of particles.

Solid–Liquid Mixing: Numerical Simulation and Physical Experiments

2767

time but it has not yet reached its eventual position after 60 revolutions.

CONCLUSIONS Solid–liquid mixing processes can be simulated with good precision when sound CFD methods are used. The application of a combination of the virtual finite element method and the network-of-zone approach was used in this work to analyze the complex flow and suspension mechanisms in a coaxial mixer. Experiments carried on the laboratory scale confirmed the validity of the predictions. Coaxial mixing shows strong performance capabilities in the case of tough mixing problems involving complex rheology, which should prove more and more useful in industry. Tools are now available to design these systems without resorting to empirical rules.

REFERENCES

Fig. 14 Resuspending of ballotini a central shaft speed of 160 r.p.m. and a speed ratio of 4.

predicting this odd motion phenomenon at the bottom. Finally, we show in Fig. 14 the resuspending of ballotini for a rotation speed of the central shaft of 160 rpm and a speed ratio RN of 4 (Nc=Na ¼ 4) after 60 revolutions. This number of revolutions is the maximum that we were capable of computing in a reasonable timeframe (a few days of CPU time). Interesting information can already be obtained from this snapshot result. First, an overconcentration is noticeable in the anchor arm wake in agreement with the experiment. This overconcentration decreases with time as the tank bottom becomes leaner in particle, thereby decreasing the number of particles that can be entrained in the wake. As far as the solid–fluid interface is concerned, the interface rises in the bulk with

1. Magelli, F.; Fajner, D.; Nonentini, M.; Pasquali, G. Solid distribution in vessels stirred with multiple impellers. Chem. Eng. Sci. 1990, 45, 615–625. 2. Fajner, D.; Magelli, F.; Nocentini, M.; Pasquali, G. Solids concentration profiles in a mechanically stirred and staged column slurry reactor. Chem. Eng. Res. Des. 1985, 63, 235–240. 3. Ferreira, P.J.; Rasteiro, M.G.; Figueiredo, M.M. A new approach to measuring solids concentration in mixing tanks. Adv. Powder Tech. 1994, 5 (1), 15–24. 4. Rasteiro, M.G.; Figueiredo, M.M.; Freire, C. Modelling slurry mixing tanks. Adv. Powder Tech. 1994, 5 (1), 1–14. 5. Richardson, J.F.; Zaki, W.N. Sedimentation and fluidisation. Part I. Trans. Instn. Chem. Engrs. 1954, 32, 35–53. 6. Bakker, A.; Fasano, J.B.; Myers, K.J. Effects of flow pattern on the solids distribution in a stirred tank. I. Chem. E. Symp. Ser. 1994, 136, 1–8. 7. Gadala-Maria, F.; Acrivos, A. Shear-induced structure in a concentrated suspension of solid spheres. J. Rheol. 1980, 24, 799–814. 8. Leighton, D.; Acrivos, A. Viscous resuspension. Chem. Eng. Sci. 1986, 41 (6), 1377–1384. 9. Leighton, D.; Acrivos, A. Measurement of shearinduced self-diffusion in concentrated suspensions of spheres. J. Fluid Mech. 1987, 177, 109–131. 10. Leighton, D.; Acrivos, A. The shear-induced migration of particles in concentrated suspensions. J. Fluid Mech. 1987, 181, 415–439. 11. Altobelli, S.A.; Givler, R.C.; Fukushima, E. Velocity and concentration measurements of

S

2768

12.

13.

14.

15.

16.

17.

18.

19.

20. 21.

22.

Solid–Liquid Mixing: Numerical Simulation and Physical Experiments

suspensions by nuclear magnetic resonance imaging. J. Rheol. 1991, 35 (5), 721–735. Phillips, R.J.; Armstrong, R.C.; Brown, R.A. A constitutive equation for concentrated suspensions that accounts for shear-induced particle migration. Phys. Fluids. 1992, A4 (1), 30–40. Krieger, I.M.; Dougherty, T.J. A mechanism for non-Newtonian flow in suspension of rigid spheres. Trans. Soc. Rheol. 1959, 3, 137–152. Mann, R. Gas-liquid stirred vessel mixers: toward a unified theory based on network-of-zones. Chem. Eng. Res. Des. 1986, 64, 23–34. Mann, R.; Hackett, L.A. Fundamentals of Gas– Liquid Mixing in a Stirred Vessel: An Analysis using Networks of Backmixed Zones, Proceedings 6th European Conference on Mixing, Pavia, Italy, 1988; 321–328. Brucato, A.; Rizzuti, L. In The Application of the Network-of-Zones Model to Solid-Liquid Suspensions, Proceedings 6th European Conference on Mixing, Pavia, Italy, 1988; 273–280. Brucato, A.; Magelli, F.; Nocentini, M.; Rizzuti, L. An application of the network-of-zones model to solids suspension in multiple impeller mixers. Trans. Instn. Chem. Engrs. 1990, 69, Part A, 43–52. Duquesnoy, J.A.; Thibault, F.; Tanguy, P.A. Dispersion of clay suspensions at high solids content. Private communication, 1995, MacMillan Bloedel, British Columbia. Duquesnoy, J.A.; Tanguy, P.A.; Thibault, F.; Leuliet, J.C. A new pigment disperser for high solids paper coating colors. Chem. Eng. Technol. 1997, 20, 424–428. Tritton, D.J. Physical Fluid Dynamics; Clarendon Press: Oxford, 1988; 325 pp. Pelletier, D.H.; Schetz, J.A. Finite element Navier–Stokes calculation of three-dimensional turbulent flow near a propeller. AIAA 1986, 24, 1409–1416. Dermidzic, I.; Peric, M. Finite volume method for the prediction of fluid flow in arbitrarily shaped

23.

24.

25.

26.

27.

28.

29.

30.

31.

32. 33.

34.

domains with moving boundaries. Int. J. Numer. Methods. Fluids 1990, 10, 771–790. Perng, C.Y.; Murthy, J. A sliding-mesh technique for simulation of flow in mixing tranks. ASME 93-WA-HT-33 1993, Old SRN049. Bertrand, F.; Tanguy, P.A.; Thibault, F. A threedimensional fictitious domain method for incompressible flow problems. Int. J. Num. Meth. Fluids 1997, 25, 719–736. Tanguy, P.A.; Bertrand, F.; Labrie, R.; Brito-de la Fuente, E. Numerical modelling of the mixing of viscoplastic slurries in a twin-blade planetary mixer. Chem. Eng. Res. Des. 1997, 74, 499–504. Coulson, J.M.; Richardson, J.F. Chemical Engineering, 3rd Ed.; Pergamon Press: New York, 1978. Armenante, P.M.; Nagamine, E.U. Effect of low off-bottom impeller clearance on the minimum agitation speed for complete suspension of solids in stirred tanks. Chem. Eng. Sci. 1998, 53 (9), 1757–1775. Ibrahim, S.B.; Nienow, A.W. The effect of viscosity on mixing pattern and solid suspension in stirred vessels. I. Chem. E. Symp. Ser. 1994, 136, 25–32. Tanguy, P.A.; Thibault, F. Power consumption in the turbulent regime for a coaxial mixer. Can. J. Chem. Eng. 2002, 80, 601–603. Thibault, F.; Tanguy, P.A. Power draw characterization of coaxial mixer with Newtonian and non-Newtonian fluids. Chem. Eng. Sci. 2002, 57, 3861–3872. Tanguy, P.A.; Thibault, F.; Brito de la Fuente, E. A new investigation of the Metzner–Otto concept for anchor mixing impellers. Can. J. Chem. Eng. 1996, 74, 222–228. Ho, F.; Kwong, A. A guide to designing special agitators. Chem. Eng. 1973, July issue, 94–104. Rieger, F.; Novak, V. Power consumption of agitators in highly viscous non-Newtonian liquids. Trans. Instn. Chem. Engrs. 1973, 51, 105–111. Ljung, L. System Identification Theo for the User; Prentice-Hall: Englewood, Cliffs, NJ, 1987.

Solid–Liquid Separation S Frank M. Tiller Wenping Li Department of Chemical Engineering, University of Houston, Houston, Texas, U.S.A.

INTRODUCTION Solid–liquid separation (SLS) involves operation of solid and liquid systems with the objectives of:[1] 1. 2. 3. 4.

Recovering solids (the liquid being discarded). Recovering liquid (the solids being discarded). Recovering both solids and liquid. Removing pollutants, solutes, micro-organisms, etc.

Solid–liquid separation is encountered in all stages of manufacturing processes ranging from raw material purification through product separation to waste management. Solid–liquid separation operations include screening, cake and deep-bed filtration, gravitational sedimentation, sedimenting and filtering centrifugation, expression (cake squeezing), hydro-cycloning, washing, membrane separation, flotation, etc. Selection of proper equipment and optimum operating conditions are among the most important problems faced by engineers involved in SLS. Because of the complex nature of fluid=particle systems and a general lack of fundamental training, SLS operations are often a problem area, or bottle neck in a plant. Various SLS operations are described as follows.[2]

cake and permits the liquid to pass through under pressure, vacuum or centrifugal forces. Once the cake is formed, it becomes the primary filter medium and particles finer than the openings of the medium can be separated.

CROSS-FLOW FILTRATION In cross-flow filtration (Fig. 1B), shear forces are introduced at the cake surface to reduce cake thickness and total cake resistance. It is exclusively used in membrane separation applications to prevent fouling on membranes.

SCREENING Screening (Fig. 1C) is a type of surface filtration to separate solids and liquid by screens with openings smaller than particle size. It is usually employed among the first few steps of a SLS process. Vibrations and other mechanism are often applied to avoid blinding of the screens.

DEEP-BED FILTRATION SEDIMENTATION Separation involving sedimentation is dependent upon settling velocity, which requires a difference in density between solid particles and the suspending liquid. Gravitational sedimentation operations are divided into clarification and thickening. Clarification involves dilute suspensions and frequently has the objective of liquid recovery. Thickening refers to solid recovery by forming more concentrated slurries. Particle size, liquid and particle densities, and liquid viscosity are important factors in sedimentation processes.

CAKE FILTRATION In cake filtration (Fig. 1A), solids and liquid are separated by filter medium, which retains the solids as a Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120007766 Copyright # 2006 by Taylor & Francis. All rights reserved.

In deep-bed filtration (Fig. 1D), particles are caught inside the filter medium. Examples of deep-bed filters are granular beds and some cartridge filters. Deep-bed filtration is used for dilute suspensions ( 1

d ¼ 0

d  0.6–0.8

d  0.9–1.0

d > 1.0

The empirical constitutive parameters for five materials with different compactibilities are shown in Table 2. The 9-mm spheres with n ¼ d ¼ 0, and K ¼ K0 ¼ constant correspond to an incompressible material. The Kaolin flat D[12] is moderately compactible and the Mierlo biosolids,[13] water treatment residue,[14] activated sludge (AS)[14] with n and d exceeding unity are classified as super-compactible materials. For super-compactible cakes, when d > 1.0 and n > 1.0, Eqs. (16)–(18) can be written better, respectively, as q ¼ q



h i K0 Pa 1
1=ð1 þ Dpc =pa Þd
1 mLðd
1Þ

Pa a0 ð1
nÞ

esav ¼ eso




As Dpc increases indefinitely in Eqs. (19)–(21), the term with Dpc approaches zero, and the flow rate and the average solidosity reach constant. Plots of calculated values of flow rate and average solidosity as functions of pressure drop across cakes with oc ¼ 0.02 m3=m2 for carbonyl iron, Kaolin flat D, and AS (in Table 2) are shown in Figs. 4 and 5. Whereas the flow rate q increases linearly with Dpc for incompressible carbonyl iron, the flow rate increases with a power of Dpc for compactible kaolin. For supercompactible materials with n > 1, and d > 1, increasing the pressure drop beyond some low value has negligible effect on the flow rate and average solidosity. The behavior of super-compactible materials that has little or no effect of pressure on either the flow rate or the average solidosity, when pressure is beyond some critical value, is unique and different compared to incompressible and moderately compactible materials. Special attention should be paid for filtration design and operation of this type of materials. SLS SYSTEM[2,5,16] Four Stages in SLS System

ð19Þ

h i 1
1=ð1 þ Dpc =pa Þn
1

ð20Þ

 d
1 1
1=ð1 þ Dpc =pa Þn
1 n
1 1
1=ð1 þ Dpc =pa Þd
1

ð21Þ

Fig. 5 Variation of esav against Dpc. (View this art in color at www.dekker.com.)

Solid–liquid separation systems generally consist of four stages, which are: 1) pretreatment to increase particle size; 2) solid concentration in thickeners; 3) solid separation in filters and centrifuges; and 4) posttreatment to remove solubles and reduce liquid content. Fig. 6 shows the relationship among these stages. Solid–liquid suspension pretreatments are employed to improve performance of the following SLS processes. Coagulation and flocculation pretreatment for colloidal systems aim at increasing the effective particle size and particle settling velocity in sedimentation operations. Addition of FeCl3, Al2(SO4)3, acidifying, cationic polyelectrolytes, and other methods are used to produce flocculation of suspensions. Aging to allow slow reactions to occur, freezing, and addition of filter aids are representative of other pretreatment methods. It is relatively difficult to determine the best pretreatment practice. Large flocs generally mean a higher chemical cost and a lower capital investment. In the second stage, the suspension is concentrated in a thickener. Although gravity thickeners dominate the field, cross-flow filter thickeners, hydrocyclones, and electrophoretic devices can be also used. A large fraction of the liquid can be removed economically in thickeners, thereby leading to smaller units in the third stage. In the third stage, cakes are produced in diverse types of filters and centrifuges. Wet cakes and filtrate or centrate are the products of filters and centrifuges. The fourth (post-treatment) stage involves further processing of both the filtrate and the wet solids from

Solid–Liquid Separation

2775

Although the term dewatering is widely used, deliquoring is more general and covers both aqueous and nonaqueous liquids. Deliquoring operations consist of expression (squeezing), blowing, sucking with a vacuum, and gravitational and centrifugal drainage. Fractional Liquid (Solid) Recovery of a SLS System In any separation process, three streams are involved as shown in Fig. 7 for a filter with cake formation, a cross flow filter without cake formation, and a thickener or clarifier based upon gravitational sedimentation. The filter is assumed to have a fixed volume into which slurry is fed and cake is formed as filtrate flows through the filter medium. The fraction of filtrate or overflow, compared to the liquid in the slurry, is called the fractional recovery of liquid, which can be obtained based upon material balance. The fractional recovery for cake filter, cross-flow filter, and thicker or clarifier are as follows:

Fig. 6 Stages of solid–liquid separation.

stage 3. A small fraction of colloidal particles which migrate through the pores of both the cake and the medium appear as turbidity in the filtrate or centrate, and require some sort of deep-bed operation for further clarification in some applications. Specifications on the product liquid determine the need for additional processing. For the cake, if the mother liquor in the slurry contains soluble substances, it may be necessary to wash the cake. Deliquoring is a process after cake washing. Decreasing the liquid content of cakes has become a major objective of many SLS processes.

V Fractional recovery ¼ VF ð1
js Þ 9 8 > > = < ð1
js =esav Þ=ð1
js Þ ðFilterÞ ¼ ð1
js =jsc Þ=ð1
js Þ ðCross-flowÞ > > ; : ð1
js =esu Þ=ð1
js Þ ðThickener, clarifierÞ

ð22Þ

The fractional removal, which is defined as the fraction of liquid removed in the cake (sediment) or concentrate then becomes Fractional removal ¼ 9 8 Vc ð1
esav ÞðFilterÞ > > > > = < Vc ð1
jc ÞðCross-flowÞ > > > > ; : Vc ð1
esu ÞðThickener, clarifierÞ VF ð1
js Þ

Fig. 7 Three streams involved in SLS systems.

ð23Þ

S

2776

Solid–Liquid Separation

Table 3 Fractional removal of liquid

Example

Stage

Example 1, feed vol.% solids ¼ 1%

Feed

Example 2, feed vol.% solids ¼ 5%

Feed

Thickener Filter, centrifuge Deliquoring

Thickener Filter, centrifuge Deliquoring

Vol.% solids

Void ratio (volume of liquid/ volume of solid)

Liquid removed (volume of liquid/ volume of solid)

1

99

—

—

—

10 25 50

9 3 1

90 6 2

90.9 6.1 2.0

90.9 97 99

5

19

—

—

—

10 25 50

9 3 1

10 6 2

52.6 31.6 10.5

52.6 84.2 94.7

Liquid removed (%)

Fig. 8 Equipment used for the four stages of solid–liquid separation.

Total liquid removed (%)

Solid–Liquid Separation

2777

S

Fig. 9 Operations of nutsche filters. (Courtesy of the Rosenmund Guedu Group.) (View this art in color at www. dekker.com.)

Eqs. (22) and (23) are equivalent as ‘‘fractional recovery ¼ 1
fractional removal.’’ Liquid Recovery of Different Stages of SLS

FEED

FILTRATE

It is instructive to consider the fraction of liquid removed in the second through the fourth stages

shown in Fig. 6. Two examples with feeds containing 1.0 and 5.0% solids (volume), and going through sedimentation, filtration, and deliquoring will be considered. It is assumed that 1) the thickener produces a 10% by volume slurry; 2) a cake containing 25% by volume; and 3) the deliquored cake contains 50% by volume after expression, blowing with gas, or spinning in a centrifuge. To facilitate comparisons, the concentrations have been converted to void ratios. The results are shown in Table 3.

CAKE DISCHARGE

Fig. 10 Vertical tank-vertical leaf filter.

Fig. 11 Candle filter. (View this art in color at www.dekker. com.)

2778

Solid–Liquid Separation

Frame

FILTER CAKE

FILTER CLOTH

RECESSED PLATE

Filtrate out

Cake formation Plate

FEED

Filter Cloth

Slurry in Fig. 12 Schematic of a plate-and-frame filter chamber.

FILTRATE

From Table 3, the preponderance of liquid removed in the thickener is apparent. As slurries become dilute, larger fractions of the liquid are removed in the thickener operation. Small changes in the concentration of the stream from the thickener can have a large effect on the size of the filter or centrifuge. If the thickener effluent could be increased to 11% solids, the corresponding void ratio would be 89=11 ¼ 8.09. The amount of liquid to be removed in the filter or centrifuge would change from 6 (9–3) to 5.09 (8.09–3) volumes of liquid per unit volume of solid. The size of the filter or centrifuge could then be reduced about 15%. Recognizing the importance of removing as much liquid as possible in the thickener, John Chandler pioneered a process in which the height of the red mud in the thickener was increased from the usual value of 1.0 to 10–15 m. The additional stress produced by the super-thick bed led to a much higher underflow concentrations. In some cases, the filter was actually eliminated.

Fig. 14 The recessed plate provides space for cakes and eliminates the need for frames.

SLS EQUIPMENT[2,5,16] An overview of some of the major types of equipment in relation to the four stages of SLS is provided in Fig. 8. After inorganic salts and polymers are used in the pretreatment, the resulting slurry is fed to a gravity thickener. The overflow from the thickener passes through a deep-bed for removal of fine particles. The underflow goes to the solids separation operations that are classified according to the driving force, i.e., gravity, vacuum, pressure, or centrifugal. Gravity separation involving large particles is usually accomplished with screens that may be stationary, vibrating, or rotating cylinders. Large screens have slot openings greater than 0.5 in and small screens have openings of less than 0.5 in. Slot openings in rotary

Fig. 13 Plate-and-frame filter.

Solid–Liquid Separation

2779

 Continuous filters. a. Rotary drum and disk filters. b. Horizontal belt filters. c. Indexing belt filters.

Air Cake Discharge

       Fig. 15 Cake discharge in a membrane filter press.

screens run from 0.01 to 0.1 in. (254–2540 mm). Microscreens frequently fall in the 15–60 mm range. In vacuum filtration, the driving force (20 in Hg ¼ 10 psi, 1 psi ¼ 6895 Pa) is slightly higher than the gravity. The vacuum operation is frequently tied with continuous equipment such as drum, disc or belt filters, in which cake is removed continuously. As the permeability of cakes diminishes, pressure becomes an important element in producing a satisfactory flow rate. Pressure filters are operated in the range of 30–60 psi and some times up to 100 psi. In contrast to vacuum filters, pressure filters normally operate in batch mode. The rate of cake buildup in batch operation is slow in comparison with continuous drum filters. The time required to dump the cake and clean and reassemble a pressure filter is called ‘‘dead time.’’ It is a significant element in determining the capacity of a pressure filter. In addition to vacuum and pressure, centrifugal forces are also used to increase driving force in separation of particles from liquids. Solid–liquid separation equipment can also be classified according to principle of each unit operation. A list of equipment based on operating principle is given as follows and will be discussed.  Batch pressure filters. a. Pressure vessel filters. b. Filter press.

S

Deep-bed filters. Croos-flow filters. Membrane filters. Thickeners and clarifiers. Centrifuges. Hydrocyclones. Expression equipment.

Batch Pressure Filters[2,5] Pressure filters are usually operated batch-wise. The batch pressure filters can be classified as tank (pressure vessel) filters or presses. Tank filters have different types of filter elements in pressure vessels. Presses consist of a series of filter surfaces (plates). The elements are mounted on a frame and are pressed together mechanically. Tank filters Tank (pressure vessel) filters are useful where noxious vapors are involved, and a completely closed system is desirable. Pressure vessel filters are divided into the following types: 1. Pressure nutsche filters (Fig. 9). 2. Leaf filters (Fig. 10). 3. Candle or tubular filters (Fig. 11). Pressure Nutsche. Nutsche filters contain a single horizontal filtering surface in a pressure vessel. Gas is used to provide pressure for filtration. Because of the limited filtration area, they are often operated with thick cakes and are suitable for small batches. Automatic nutsches are available to perform reaction, crystallization, filtration, reslurry washing, drying, and cake discharging in the same vessel. Operations of Nutsche filter are shown in Fig. 9. Leaf Filter. In comparison to nutsche filters, leaf filters provide more filtration area in the same volume of pressure vessel. They are more suitable for handling

Table 4 Brine sludge dewatering—Comparison of recessed press and membrane filter Press type

Cake % solids

Cake structure

Cake thickness (mm)

Cake compr. strength (tpf)

Average rate (pph/ft2)

100 psig std. recess

59

Soft core

30

3.5

1.70

25 psig membrane

67

Very firm

25

>5.0

2.04

(From Ref.

[20]

.)

2780

Solid–Liquid Separation ing Dry

D

n tio ta Ro

ing ry

Wa sh g in

10

5˚

Cake

Disch

arge

rge

Sucti

cha

Dis

on

φ° ve

cti

ffe

a ti o

n

1 2 8˚

˚

˚

15

C ake for m

Ine

ve

cti

ffe

1 2 6˚

25

Ine

Slurry level Cake

DRUM

DISK

larger quantities of slurry. Fig. 10 shows a vertical tank–vertical leaves filter. In general, cake can be discharged more easily from vertical leaf-type filters. During cake formation, to prevent dropping cake, the thickness is normally restricted to 3.5–4.0 cm. A space of about 2.0 cm must be maintained between the cakes to prevent arching and facilitate discharge. Candle Filter.[17] Tubular filter elements contained in a matching vessel are known as candle filters. The actual filter vessel may contain one or more filter candles, and may be used as pressure or suction filters for the filtration of liquids and gases. A particular advantage offered is that candles may readily be changed to different types, to suit particular requirements or applications. A typical candle filter[18] is shown in Fig. 11. Materials of candles are selected to fit a particular process.

A

Fig. 16 Cycles for disc and drum filters show cake formation, suction, washing, drainage, and discharge areas. (From Ref.[21].)

Filter presses The filter presses play a significant role in the SLS industry, where it is unnecessary to operate in a closed atmosphere. The most typical filter presses is the plate-and-frame filter press. The recessed filter press and the membrane filter press are revised, based on the plate-and-frame filter press. Plate-and-Frame Filter Press. The major elements of the plate-and-frame version of the press are illustrated in Fig. 12. Feed to a plate-and-frame press is generally through openings at the bottom to reduce sedimenting tendencies. A filter medium (usually cloth or paper) placed over a grooved plate serves as the support for the cake, which is deposited in an adjoining frame. Plates and frames are alternated as shown in Fig. 13. The medium serves as a gasket when mechanical closure is affected. Where washing is desired, every other plate in a plate-and-frame press is constructed, so that liquid can be passed from one plate through the cake contained in the frame to the opposite plate. Thus, the wash passes through two cakes. Recessed Plate Press.[19] Recessed plate press shown in Fig. 14 does not require a frame. The edges of the plate are extended outward leaving a space for a cake.

B

C

D Fig. 17 Isometric view of disc filter.

Membrane Filter Press. Membrane filter presses use impermeable, flexible membranes, or diaphragms to squeeze the cake for further cake deliquoring, as shown in Fig. 15. This type of filter provides less dead time in a filtration cycle, better washing, and drier cake compared to traditional plate-and-frame and recessed plate filter presses. A comparison of a recessed press operated at 100 psi and a membrane filter operated at 25 psi for sludge dewatering is shown in Table 4.[20]

Solid–Liquid Separation

2781

S

Fig. 18 Horizontal belt filter illustrating filtration, washing, and drainage stages.

Firmer cake with higher % cake solids, and faster filtrate rate were obtained for the membrane filter. Continuous Filters[21] Continuous filters work best on medium sized particles in the range of 5–50 mm. The larger particles generally encountered exert minor capillary forces, and cake drying or ‘‘drainage’’ can be accomplished by sucking air through the cakes under vacuum. Continuous filters are normally used for materials that are relatively concentrated and easy to filter with a cake buildup rate at cm=min. Types of continuous filter are characterized by different type of traveling filter surfaces. Rotary drum (Fig. 16), disc (Figs. 16 and 17), and horizontal belt filters (Fig. 18) will be discussed. With drum filters (Fig. 16), the feed is below the filter surface; the slurry flows upward; and the cake

Fig. 19 Automatic backwash sand filter (Centra-floTM): 1, Overflow; 2, filter influent; 3, coarse media; 4, fine media; 5, filtrate nozzles; 6, filtrate chamber; 7, level controller; 8, filter reject; 9, washbox; 10, counter-current washer; 11, airlift; 12, central feed chamber. (View this art in color at www. dekker.com.)

faces downward. Disc filters (Figs. 16 and 17) have a vertical surface. Operations of drum and disk filters with different cycles including cake formation, washing, drying, and cake discharge are shown in Fig. 16. A horizontal belt filter is shown in Fig. 18. It occupies more space than drum and disc filters for the same filter area. When settling is acute, it must be used in preference to the drum the disc filter. Unlike the drum and disk filters with strict requirements on cake formation, washing, drainage, and discharge time, the various stages on the horizontal filter are completely adjustable, and the entire filter surface can be utilized in contrast to the limitations of the disc and drum types. Deep-Bed Filter Deep-bed filters are employed for slurries with very dilute concentration less than 1000 ppm (parts per million by weight). The deep-bed has pores in which the fine particles are caught. Capture of particles in deep-beds depends upon transport mechanisms that carry the particles to the surface of the medium. As the deposit builds up, the permeability ultimately drops to a point where the bed must be regenerated or discarded. The deep beds are in the form of granular media (sand, crushed anthracite coal, garnet, usually backwashable)

Fig. 20 Johns–Manville cartridge.

2782

Solid–Liquid Separation

Fig. 21 Cartridge filters: (A) Wound cartridge and (B) melt blown cartridges. (From Ref.[22].) (View this art in color at www.dekker.com.)

(Figs. 1D and 19) or cartridges (usually disposable) (Figs. 20 and 21), which are cylinders containing a variety of materials for trapping the particles. An example of the wide application of cartridge filters includes removing particles from lubricating oils in automobiles and trucks.

Cross-Flow Filters In cross-flow filtration (Fig. 1B) or delayed cake filtration, the slurry flows parallel to the cake surface with sufficient velocity to prevent partially or entirely the deposition of cake. It is used successfully to increase flow rate in membrane filtration. It is also employed for concentrating and recovering very fine particles in dilute suspensions when deep-bed or cake filtration would not applicable. There are basically two types of cross-flow filter. One is without rotating element in which slurry is usually pumped into the filter in the direction parallel to the filter media to produce a cross-flow. Another type of cross-flow filter is equipped with rotating elements, such as rotary filter press with agitators or turbines attached to a rotary shaft as shown in Fig. 22.[23]

Membrane Filters[22] Membrane separation has advantages of low energy consumption, effective multiple fine particle removal, and small waste stream. Developing of high performance and low cost membranes is a major factor in hindering the advances of membrane processes. Based on the size of particles separated, membrane filtrations are categorized as MF (0.1–2.0 mm), UF (0.005–0.1 mm), NF (0.0005–0.005 mm) and RO ( 1.3 show very limited miscibility with water. Solvents with intermediate values (0 < log P < 1.3) are somewhat miscible with water. Mixtures of nonpolar solvents are normally characterized by the term ‘‘solubility parameter’’ (d). The difference in solubility parameters of mixture components provides a measure of solution nonideality.[3] Mixtures of aliphatic hydrocarbons are nearly ideal, whereas mixtures of aliphatic hydrocarbon with aromatics show appreciable nonideality. Sometimes, it is difficult to predict the behavior of highly nonideal mixtures. Thermodynamic properties of binary and multicomponent mixtures have been dealt with extensively in the literature.[3,4]

PROPERTIES OF SOLVENTS The proper choice of a solvent for a particular application depends on several factors, among which its physical properties are of prime importance. In most applications, the solvent is preferred to be in its liquid state under the temperature and pressure conditions at which it is been employed. Other properties, such as density, vapor pressure, heat capacity, surface tension, and transport properties, are also very important. Electrical, magnetic, and optical properties are, in some cases, relevant to the application. All these are physical properties of the solvents. Chemical properties of solvents, like the physical properties, are very important. Solubility parameters or the solvency factors provides an idea of how good a solvent is to solubilize certain types of materials. Polarity, acid–base properties, hydrogen bonding ability, and water miscibility are important chemical properties of solvents and are somewhat related. Physical Properties Under ambient conditions, solvents are normally liquid, unless a supercritical solvent is considered. The freezing=melting points and boiling points of some widely used solvents are listed in Table 4.[1,2] A solvent could form solids if it is stored outside during winter, and the temperature drops below its freezing point. Several freeze thaw cycles can damage the performance of paints and coatings. Solvents with low boiling points

Solvents

have high vapor pressure, which leads to a quick drying or separation time. But sometimes, slow drying is necessary for coating applications where a solvent with high boiling point may be utilized. Specific heat capacity is another important property of solvents. This is defined by the energy required to raise one unit mass of the solvent by one degree. This term becomes important if a solvent is used in a reaction process, which is either exothermic or endothermic. Solvents with high specific heat capacity can absorb more heat per unit temperature rise, compared to those with low specific heat capacity. The response of a solvent to an electric field depends on the intrinsic dipole moment of its molecules, and also on cooperative effects of adjacent dipoles, when these are correlated in the liquid. The dipole moment m is the measure of the separation of the positive and negative centers of charge in the molecule, and is measured best for the solvent vapor, where such cooperative effects are absent.[1] When this is impractical because of low volatility of the solvent, then the dipole moment may be measured for a dilute solution with the solvent being the solute in an inert solvent. The inert solvent is normally chosen from a group of solvents with very low polarity, such as benzene, hexane, and tetrachloromethane. Solvents with highly symmetrical molecules have zero dipole moments, but electronegative atoms connected to aliphatic or aromatic skeletons cause the molecules to have finite dipole moments. Table 5 lists dipole moments for some common solvents.[1,2] It also lists dielectric constants (e) for these solvents. The higher is the dielectric constant, the more polar is the solvent and vice versa. Low dielectric constant fluids, also known as ‘‘dielectric fluid,’’ are used in electronics industry as cleaning solvents for circuit boards. Examples are fluorocarbons (because of their nonflammability), high flash point hydrocarbons, and silicones. Surface and transport properties of solvents are very important for solvents. Surface tension of a solvent shows how easy or difficult it would be to wet the surface on which the solvent is being applied. Low surface tension implies better wetting ability and vice versa. Water and other polar organic solvents have very high surface tension, whereas silicones, fluorocarbons, and aliphatic hydrocarbons have low surface tension. Solvents with low surface tension are easier to leak through threaded joints compared to those with high surface tension. Viscosity is another important property of solvents. High viscosity implies high power requirement for the flow of the solvent. Low viscosity means the solvent will flow easily; however, it may not be desirable in many applications. For example, a paint may need a solvent with optimum viscosity, which will neither drip out of the brush easily nor will stick to the brush so

2805

Table 5 Dipole moment (m) and dielectric constants (e) of common solvents Name

l

e

n-Hexane Benzene

0.09 0.00

1.88 2.27

Toluene

0.31

2.38

o-Xylene

0.45

2.57

m-Xylene

0.30

2.37

p-Xylene

0.00

2.27

Ethylbenzene

0.37

2.40

Cumene

0.39

2.38

Methanol

2.87

32.66

Ethanol

1.66

24.55

n-Propanol

3.09

20.45

i-Propanol

1.66

19.92

n-Octanol

1.76

10.34

1,2-Ethanediol

2.31

37.70

1,2-Propanediol

2.25

32.00

1,3-Propanediol 1,2-Butanediol 2,3-Butanediol (meso)

2.55 2.18 2.1

35.00 — 21.53

Diethyl ether

1.15

4.20

Tetrahydrofuran

1.75

7.58

Dioxane

0.45

2.21

Acetone

2.69

20.56

Methyl-i-butyl ketone

2.77

15.87

Formic acid

1.82

58.50

Acetic acid

1.68

6.15

Propanoic acid

1.68

3.37

Methyl acetate

1.68

6.68

Ethyl acetate

1.78

6.02

Propyl acetate

1.78

6.00

Chlorobenzene

1.69

5.62

Chloroform

1.15

4.89

1,1,1-Trichloroethane

1.70

7.25

1,1,2-Trichloroethane

1.55

7.29

Trichloroethylene

0.80

3.42

Morpholine

1.56

7.42

Triethylamine

0.66

2.42

Aniline

1.51

6.98

Diethanolamine

2.81

25.19

Acetonitrile

3.92

35.94

Nitrobenzene

4.22

37.78

Formamide

3.37

109.50

Dimethylformamide

3.82

36.71

Dimethylacetamide

3.72

37.78

Dimethyl sulfoxide Ammonia

4.06 1.47

46.45 22.38

m is Debey unit and e is dimensionless. (From Refs.[1,2].)

S

2806

hard that it will be difficult to apply on a surface. Both surface tension and viscosity are temperature dependent. In formulations involving solvents, the viscosity behavior of the final formulation may be quite different from the solvent itself. Many times, solvent-based formulations such as paints exhibit non-Newtonian behavior with the viscosity being a strong function of shear stress. However, most of the pure solvents exhibit Newtonian behavior.

Chemical Properties Chemical properties of solvents affect their usefulness in various applications. The solvent should selectively dissolve the desired solutes, should be inactive=inert in the chemical reactions, and solvate the transition states and products really well. This can be achieved by the proper blend of chemical properties such as solvency, polarity, hydrogen bond donation or acceptance ability, acidity or basicity, hydrophilicity, and redox properties. A particularly common test for ranking hydrocarbon solvent strength is the kauri-butanol test. The kauributanol value (KB) of a solvent represents the maximum amount of the solvent that can be added to a stock solution of kauri resin (a fossil copal) in butyl alcohol without causing cloudiness. Because kauri resin is readily soluble in butyl alcohol but not in hydrocarbon solvents, the resin solution will tolerate only a certain amount of dilution. ‘‘Stronger’’ solvents such as toluene can be added in a greater amount (and thus have a higher KB value) than ‘‘weaker’’ solvents like hexane. Another common test for solvency is called aniline point test. The aniline point is called the ‘‘aniline point temperature,’’ which is the lowest temperature ( F or C) at which equal volumes of aniline (C6H5NH2) and the oil form a single phase. The aniline point (AP) correlates roughly with the amount and the type of aromatic hydrocarbons in an oil sample. A low AP is indicative of higher aromatics, while a high AP is indicative of lower aromatics content. Diesel oil with AP below 120 F (49 C) is probably risky to use in oil-base mud. In general, the lower the aniline point, the more the number of unsaturants that are present and the higher the potential for swelling certain rubber compounds. The American Petroleum Institute has developed test procedures that are the standard for the industry. Solvent polarity is related to the dipole moment as discussed in the earlier section. A solvent without a permanent dipole must be classified as nonpolar. However, a solvent may exhibit local polarity, if it possesses two mutually canceling dipoles. One such example is

Solvents

1,4-dioxane, where the two oxygen atoms can participate in electron-pair donation to nearby acceptor atoms in solutes, although the molecule as a whole does not have a permanent dipole moment. Furthermore, highly polarizable molecules may interact via induced dipoles, so that polarizability may contribute to the chemical aspect of polarity. The acidity and basicity properties of solvents play an important role in various processes, especially in chemical reactions. A solvent may accept or donate a proton, thereby increasing or decreasing the rate of a chemical reaction. This behavior is similar to that of a catalyst. Sometimes, a solvent mixture may be a better alternative than the single components. Aqueous solubility of solvents is related to polarity and dipole moment. Many solvents are hygroscopic and need drying agents, such as molecular sieves to remove moisture from them. Solubility of solvents in water and vice versa are listed in Table 6. It is observed that the solvents with –OH, –C(O)Me, –COOH, –NH2, and –CN groups have high degree of miscibility with water, whereas hydrocarbons, esters, and chlorinated compounds have a low solubility in water. As discussed earlier, solubility parameter is important when nonpolar solvents are mixed. Table 7 provides a list of molar liquid volume and solubility parameter for some common solvents.[3] Aromatics have a higher value of this parameter compared to the aliphatics.

APPLICATIONS OF SOLVENTS Major applications of solvents are found in paints= coating and cleaning applications. However, they are also widely used in chemical processes (as solvent as well as raw material), inks, solvent extraction, heat transfer systems, and electrochemistry. Some solvents are also used as catalysts in a chemical reaction.

Paints/Coatings Paints and coatings consume the largest amount of solvent among all the applications.[5] Because of regulatory concerns, the consumption of organic solvents in coating applications is being reduced and replaced with water-based as well as nonsolvent coatings (powder coating). Use of green solvents is being encouraged in recent years. Solvents in paint or a coating serve multiple purposes. These include solubilization of resins and other ingredients, wetting, viscosity reduction, adhesion promotion, and gloss enhancement. Initially, the resin or polymer is dissolved in the solvent to form a continuous phase.

Solvents

2807

Table 6 Water miscibility of different solvents Name

In water

Water in

n-Hexane

2.57e-6

5.31e-4

log P O/W

Benzene

4.13e-4

2.75e-3

2.13

Toluene

1.01e-4

171e-3

2.69

o-Xylene

2.97e-5

2.53e-3

3.12

m-Xylene

2.48e5

2.36e-3

3.20

3.90

p-Xylene

2.65e-5

2.68e-3

3.15

Ethylbenzene

2.58e-5

2.53e-3

3.15

Cumene

9.79e-6

2.01e-3

Methanol

miscible

miscible

Ethanol

miscible

miscible

n-Propanol

miscible

miscible

3.66
0.70


0.25 0.28

i-Propanol

miscible

miscible

0.13

n-Octanol

4.4e-5

2.75e-1

3.15

1,2-Ethanediol

miscible

miscible

1,2-Propanediol

miscible

miscible

1,3-Propanediol

miscible

miscible

1,2-Butanediol

—

—


2.27


1.41 —

—
0.92

2,3-Butanediol (meso)

miscible

miscible

Diethyl ether

1.54e-2

5.76e-2

Tetrahydrofuran

miscible

miscible

0.46

1,4-Dioxane

miscible

miscible


0.42

Acetone

miscible

miscible

Methyl-i-butyl ketone

3.10e-3

9.72e-2

Formic acid

miscible

miscible

Acetic acid

miscible

miscible

Propanoic acid

miscible

miscible

Methyl acetate

7.31e-2

2.69e-1

0.89


0.24 1.31


0.54


0.24 0.32

0.18

Ethyl acetate

1.77e-2

1.29e-1

0.73

Propyl acetate

4.1e-3

1.45e-1

1.24

Chlorobenzene

7.83e-5

2.03e-03

2.84

Chloroform

1.24e-3

6.1e-3

1.94

1,1,1-Trichloroethane

1.78e-4

2.51e-3

2.36

1,1,2-Trichloroethane

5.96e-4

8.67e-3

1.89

Trichloroethylene

1.88e-4

2.29e-2

2.35

Morpholine

miscible

miscible

Triethylamine

1.03e-2

2.13e-1

Aniline

6.72e-3

2.05e-1

Diethanolamine

7.8e-1

miscible

Acetonitrile

miscible

miscible

Nitrobenzene

2.78e-4

1.62e-2

Formamide

miscible

miscible

Dimethylformamide

miscible

miscible

Dimethylacetamide

miscible

miscible

Dimethyl sulfoxide

miscible

miscible

Ammonia

miscible

miscible

All the quantities are dimensionless. (From Refs.[1,2].)


1.08

1.36 0.90


1.43


0.34

1.85


0.97


1.01


0.77


1.35


1.49

S

2808

Solvents

Table 7 Molar volume and Hildebrand solubility parameter for solvents n (cm3 mol
1)

d (J cm
3)1/2

Perfluoro-n-heptane

226

12.3

Neopentane

122

12.7

Isopentane

117

13.9

n-Pentane

116

14.5

n-Hexane

132

14.9

Solvents

1-Hexene

126

14.9

n-Octane

164

15.3

n-Hexadecane

294

16.3

Cyclohexane

109

16.8

Carbon tetrachloride

97

17.6

Ethyl benzene

123

18.0

Toluene

107

18.2

Benzene

89

18.8

Styrene

116

19.0

Tetrachloroethylene

103

19.0

Carbon disulfide

61

20.5

Bromine

51

23.5

At the same time, many plants have switched to hydrocarbon-based solvents. Another cleaning application involving solvents is the degreasing of metal parts and other objects in manufacturing plants and automotive repair facilities. Mineral oils and other high flash point hydrocarbon solvents are used in these applications.

Reaction Solvents Chemical reactions normally require a solvent that can dissolve all the raw materials. Reaction kinetics and selectivity may be affected by the choice of solvent. Thermophysical properties of solvents are also very important for endothermic and exothermic reactions. Water, hydrocarbons, alcohols, ketones, chlorinated solvents, and amines are the most widely used solvents in the chemical process industry. In many applications, the solvent is consumed as a raw material. One example is the manufacture of esters, where alcohol takes part in the reaction.

(From Ref.[3].)

Dyes and pigments are then added to this solution to provide color. When a coating is applied, the solvent starts to evaporate leaving a thin continuous film on the surface. The solvent can also increase adhesion by softening the primer coat. In most of the coating applications, a solvent blend is used instead of a single solvent. Thus, each solvent component serves a particular purpose. In water-based paints, the dominant solvent is water. There may be a small amount of other solvents to carry out certain functions. There are two types of water-based paints—one with latexes (composed of fine polymer particles dispersed in the solvent) and the other with water-soluble polymers. Latex paints are very common in the architectural market with flat, semigloss, and gloss coatings. More and more industrial coating applications are switching to water-based paints for environmental reasons.

Cleaning The dry cleaning industry is one of the largest consumers of solvents. Perchloroethylene (PCE) is used in as high as 80% of the dry cleaning applications. Although this solvent is nonflammable, easy to recover, and exhibits good solvency, its environmental impact is severe. There is work underway to reduce the emission of this solvent from the dry cleaning plants.

Solvent Extraction Solvent extraction is widely used in pharmaceutical and food processing industries. Oil seed extraction, manufacturing of neutraceuticals, decaffeinated coffee, intermediates, and some reactive-separation processes utilize solvent extraction. Hydrocarbons are common solvents for oil seed extraction. Supercritical solvents are gaining popularity in producing neutraceuticals and other active ingredients.

CONCLUSIONS Innumerable production processes and applications rely on the proper selection of solvents. Organic solvents, although most widely used, are being slowly replaced with aqueous and supercritical solvents. Ionic liquids are still in the research stage and commercial applications are just beginning to appear. Solvent mixtures are used in many applications where a single solvent cannot do the job. Both physical and chemical properties of the solvents are important. In addition, toxicity, flammability, and environmental friendliness should not be overlooked while selecting a solvent. All the properties are assessed to design the most optimum solvent for a particular need. The most widely used applications for solvents are paints=coatings, cleaning, reaction medium, and solvent extraction.

Solvents

REFERENCES 1. Marcus, Y. The Properties of Solvents; John Wiley & Sons Ltd: New York, 1999. 2. Riddick, J.A.; Bunger, W.B.; Sakano, T.K. Organic Solvents; Wiley-Interscience: New York, 1986. 3. Prausnitz, J.M.; Lichtenthaler, R.N; de Azevedo, E.G Molecular Thermodynamics of Fluid-phase Equilibria; Prentice Hall: Upper Saddle River, NJ, 1999.

2809

4. Marcus, Y. Solvent Mixtures; Marcel Dekker, Inc: New York, 2002. 5. Sullivan, D.A. Solvents, industrial. In Kirk-Othmer Encyclopedia of Chemical Technology, 3rd Ed.; Wiley: New York, 1980; Vol. 22, 529–571. 6. McClain, J. Processing with supercritical solvents. Chemical Engineering 2000, Feb. 7. Crabb, C. Exploring ionic liquids. Chemical Engineering 2001, Mar. 8. Gorman, J. Faster, better, cleaner? Science News 2001, 160 (10), Sept.

S

Sonochemical Reaction Engineering S David A. Bruce Amarnath Nareddy Department of Chemical and Biomolecular Engineering, Clemson University, Clemson, South Carolina, U.S.A.

INTRODUCTION Sonochemical reaction processes, which are relatively new, use the transmission of ultrasonic waves (20 kHz to 1 MHz) through liquid media to initiate chemical transformations and enhance mass transfer. These processes are similar to other chemical treatment methods, such as those employing plasma, flame, and thermo chemistries, in that a large amount of energy is introduced to the material in a very short period of time. However, the highly reactive zones in a sonochemical process are micrometer sized and the bulk fluid, which is generally at ambient conditions, is not exposed to these harsh conditions for extended periods. The chemical effect of these sound waves is derived from the formation, growth, and sudden collapse of micrometer size bubbles via a process known as acoustic cavitation. The microbubbles formed during cavitation are slowly filled with vaporized liquids until they reach a critical size, whereupon they undergo violent collapse in less than a microsecond. This rapid implosion causes the gases inside the bubble to be adiabatically compressed, which leads to significant increases in temperature and pressure inside the cavity. In fact, the temperature inside a bubble during implosion can be as high as 5000 K and the pressure greater than 500 atm, which equates to heating rates of more than a billion degrees per second during bubble collapse. Such extreme conditions lead to the formation of highly reactive species that can readily react with chemicals in the surrounding liquid phase. Cavitation phenomena can also produce reactive radicals and ions via localized electrical discharges that result from positively and negatively charged ions becoming separated during microbubble oscillation and collapse. Interestingly, this entire reaction process occurs repeatedly in a fluid medium whose temperature and pressure are only slightly above ambient conditions. Thus, ultrasound irradiation can significantly improve reactions rates, while simultaneously allowing the overall process to operate at milder reaction conditions. The unique reaction environment generated in sonochemical reactors has been shown to greatly enhance reaction rates for a variety of chemical transformations. However, the commercial scale application Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120039788 Copyright # 2006 by Taylor & Francis. All rights reserved.

of sonochemical technologies is still limited. It is only recently that advances in electromechanical transducers have enabled high powered ultrasonic waves to be efficiently generated in industrial scale reactors. To date, sonication has been used to fabricate nanomaterials, epoxidize and oxidize unsaturated hydrocarbons, couple halogenated aromatics, and cause a variety of degradation reactions as well as a number of other organic and inorganic chemical reactions. Many of the reactive species formed by acoustic cavitation are free radicals; hence, reaction pathways that include the formation of free radicals are often accelerated by sonochemical processing. However, there are many factors, some of which are still not fully understood, which determine whether a given reaction process will be accelerated by ultrasonic irradiation. Therefore, much of the research and process development in this field is still Edisonian in nature.

HISTORY Though sonochemistry is a relatively new field of study, cavitation phenomena have been of interest for over 200 years. Leonhard Euler first mentioned the effects of cavitation in 1754 and the first article on cavitation by Thornycroft and Barnaby was printed in 1895.[1] However, it was not until the work of Lord Rayleigh in 1917 and Loomis and others in 1927 that the physical, chemical, and biological effects of cavitation were fully described.[2–4] Later studies by Harvey et al. defined the underlying mass transfer phenomena controlling bubble growth (i.e., rectified diffusion), and shortly thereafter the first computer simulations of a cavitating bubble were reported by Noltingk and Neppiras.[5,6] A decade later, Naude and Ellis suggested that bubbles collapsing near solid surfaces undergo asymmetric implosions that lead to the formation of microjets (see Fig. 1).[7] These microjets are believed to cause solid surface pitting and can readily explain the microscale surface erosion that is observed with dispersed particulates (e.g., oxide catalysts), which leads to their decreasing in size during sonication. Since the 1980s, numerous studies have provided insights on the physical environment (temperature 2811

2812

Sonochemical Reaction Engineering

Fig. 1 Ultrasound induced collapse of a microbubble on a surface. (Courtesy of Crum, L.A. J. Physique Colloque 1979, 40, 285–288.)

and pressure) inside an imploding microbubble, as well as information about the types of reactive species formed in this environment (e.g., free radicals such as H
and OH
).[8–10] During the same period, the production of low cost sonication equipment has led chemists to discover a wide array of chemical reactions whose rate is greatly enhanced or whose product distribution is significantly altered by the cavitation process.[11–16]

GENERATION OF ULTRASOUND WAVES Ultrasound is sound energy with a frequency beyond human hearing. It is a form of acoustic radiation with a frequency range between 20 kHz and 500 MHz. Depending on the frequency, it is divided into three regions: power ultrasound (20–100 kHz), high frequency ultrasound (100 kHz–1 MHz), and diagnostic ultrasound (1–500 MHz). Diagnostic ultrasound has a low intensity and has long been used in the medical field for therapeutic, operative, and diagnostic procedures. Power and high frequency ultrasound, commonly called destructive ultrasound, are used in a variety of industries. Common industrial applications include: welding plastics, cleaning and decontamination, beneficiation of coal, soldering, and processing that includes emulsification, extraction, crystallization, and filtration. In sonochemistry, physical and chemical changes are brought about by the usage of power ultrasound. Three types of transducers are used to generate power ultrasound: gas driven transducers, liquid driven

transducers, and electromechanical transducers. Of these, electromechanical transducers are the most widely used and include piezoelectric transducers and magnetostrictive transducers.[12,17] Piezoelectric transducers convert electric energy into kinetic energy or sound waves. When piezoelectric materials, such as quartz or BaTiO3, are subjected to an alternating electric field, corresponding dimensional changes in the crystal generate high energy mechanical vibrations (sound energy). These transducers are highly efficient and can be used over the entire range of ultrasonic frequencies, but they are generally restricted to small volume processes for short terms and often operate at a fixed frequency; hence, they are more commonly used for laboratory scale experiments or processes. The magnetostrictive transducers, on the other hand, are used when large volumes are to be processed and when the reaction times are long. The magnetostrictive transducers utilize the Joule effect in which a ferromagnetic material (e.g., nickel or iron) in an external magnetic field alternately expands and contracts producing sound waves. These transducers are very robust, have high driving force, and can be operated at elevated temperatures (250 C) for extended periods of time. These characteristics make the magnetostrictive transducers a good choice for largescale industrial applications.

ACOUSTIC CAVITATION Acoustic cavitation is a nonlinear process that effectively concentrates the diffuse energy of sound in

Sonochemical Reaction Engineering

liquids. The microbubbles formed during cavitation result from sound wave generated compression and rarefaction cycles, which cause the local fluid density to momentarily increase or decrease, respectively. When the local density decreases during a rarefaction cycle to a sufficiently low value, such that the spacing between molecules exceeds the cohesive forces in the liquid, voids or cavities are formed. These voids, during the subsequent compression and rarefaction cycles, grow because of inertial effects and the rectified diffusion of mass from the bulk liquid. The growing cavity eventually reaches a critical size, where the combination of internal pressure and surface tension effects no longer balance the compression forces generated by the next sound wave. At this instant, the bubble implodes. This bubble collapse leads to a highly reactive environment, localized in a micrometer sized region in the bulk fluid. To gain a better understanding, the three discrete stages of cavitation, viz., nucleation, bubble growth, and implosive collapse, are described below with respect to homogenous (liquid) sonochemical reaction systems.

Nucleation Nucleation refers to the formation of bubbles or cavities and is the origin of all cavitation processes. These cavities are formed when the acoustic energy applied to the liquid is sufficient to overcome the attractive forces binding neighboring liquid molecules. This energy, which equates to the tensile strength of the liquid is of the order of 1000 atm for pure water. However, the presence of dissolved gases and suspended particulates can reduce this tensile strength to the order of 1 atm. These gas pockets serve as nucleation sites for ultrasound cavitation bubbles and are often called ‘‘weak spots.’’ Ultrasound transmitted through a liquid consists of alternate compression and rarefaction cycles. When the acoustic amplitude

2813

during the rarefaction cycle is sufficiently large and exceeds the average intermolecular bonding forces in the liquid, neighboring liquid molecules are pulled apart forming voids or cavities.[18] Depending on the vapor pressure of dissolved organics and inorganics, these voids can be rapidly filled with vapor content and form bubbles. Hence, bubble formation and the chemical effects of ultrasound can only be seen when there are impurities, such as dissolved gases, in the liquid and when the sound intensity exceeds the cavitation threshold (i.e., the surface tension of the liquid near weak spots).[12] Fig. 2 provides an illustration of the relationship between the acoustic pressure (which varies) and bubble creation, growth, and collapse processes.

Bubble Growth and Dynamics Bubble growth is very complex, in that it is highly nonlinear with time and also different for each of the millions of bubbles formed in the liquid. Fundamentally, the gas bubbles generated almost instantaneously during nucleation are inherently unstable.[19] In the absence of further acoustic irradiation, large bubbles will float to the surface and burst, while small bubbles will slowly dissolve owing to the excess internal gas pressure. In the presence of ultrasound, a bubble can grow many times larger than its initial size over a period of time much longer than bubble nucleation or collapse. Depending on the acoustic intensity and frequency, these bubbles will form a stable cavity and oscillate around a mean radius for several acoustic cycles or they will continue to grow until they reach a critical size and collapse. In general, this bubble growth process occurs at sufficiently slow rates so that it can be considered isothermal; hence, extremely little or no reaction is thought to occur during the growth phase of the cavitation process. Bubble growth with high intensity ultrasound arises primarily from inertial effects. At this high intensity,

Fig. 2 The cavitation process: growth and collapse of a microbubble as a function of time and ultrasound frequency. (View this art in color at www. dekker.com.)

S

2814

bubbles gain so much momentum during the negative pressure cycle that they will have no time to recompress during the positive pressure cycle and grow rapidly in the course of a single cycle of sound.[19] These bubbles, no longer in phase with the ultrasonic field, become unstable and implode in the subsequent compression cycle. At low ultrasound intensities, bubble growth primarily occurs via rectified diffusion, which is the unequal mass transfer of species into the bubble during rarefaction and compression cycles. This phenomenon was first recognized by Harvey et al. during their experiments on animals, and Leighton has recently expounded on a well established theory that describes rectified diffusion in terms of an ‘‘area effect’’ and a ‘‘shell effect.’’[5,20] These two effects derive directly from basic mass transfer principles, which demonstrate that the rate of mass transfer is directly related to the surface area across which transfer can occur and the concentration (or more exactly, chemical potential) driving force. The ‘‘area effect’’ can be described by the following. During rarefaction, when the bubble expands, concentrations and pressure inside the bubble decrease, and so dissolved gases and other volatile species diffuse from the liquid into the bubble. However, bubble compression yields high pressure and concentrations inside the bubble, which causes gas to diffuse from the

Sonochemical Reaction Engineering

bubble’s interior into the bulk liquid. Because the surface area of the bubble is greater during the expansion phase than in the compression phase, the amount of gas diffused is greater during the expansion cycle. Therefore, over a complete cycle, there will be a net inflow of gas into the bubble causing it to grow over time, as depicted in Fig. 2. Similarly, the ‘‘shell effect’’ arises because the diffusion rates of dissolved volatile species are proportional to their liquid phase concentration gradients. Consider a thin spherical shell of liquid surrounding an acoustically generated bubble. This liquid shell will change volume as the bubble pulsates, which in turn causes the concentration gradients of dissolved species to change, as shown in Fig. 3. Note that in Fig. 3, xi and yi equal the bulk concentration of the diffusing species i in the liquid shell and gas phases, xi,int and yi,int equal the concentration of species i at the gas– liquid interface, x
i equals the concentration of species i that would exist in the liquid phase if it were in equilibrium with a gas having concentration yi, and y
i equals the concentration of species i that would exist in the gas phase if it were in equilibrium with a liquid having concentration xi. At equilibrium conditions, there is no concentration gradient in the liquid surrounding the bubble; hence, no net mass transfer occurs. During bubble expansion (rarefaction), the shell contracts and the concentration of volatile species in the

Fig. 3 Concentration profiles for species i in the gas and liquid phases of a microbubble undergoing rectified diffusion: the shell effect.

Sonochemical Reaction Engineering

2815

gas and surrounding liquid decreases. However, the concentration gradient is high because of the thinner shell and so the rate of diffusion of gas into the bubble from the bulk liquid is high. The opposite trend occurs when the bubble contracts. During bubble compression, the liquid shell expands and the concentration of volatile species in the gas and liquid shell increases. The increased shell thickness causes the concentration gradient to be small, thereby lowering the rate of gas diffusion. Hence, the amount of gas diffused is greater during the expansion phase because the high concentration gradient drives the gas a shorter distance, whereas in the compression phase the lower concentration gradient is driving gas a longer distance. Thus, both the area and shell effects lead to a net mass transfer of material into the bubble during each acoustic cycle. Various mathematical models have been put forth to describe the rate of bubble growth and the threshold pressure for rectified diffusion.[14,20–25] The most widely used model quantifies the extent of rectified diffusion (i.e., the convection effect and bubble wall motion) by separately solving the equation of motion, the equation of state for the gas, and the diffusion equation. To further simplify the derivation, Crum and others made two assumptions: 1) the amplitude of the pressure oscillation is small, i.e., the solution is restricted to small sinusoidal oscillations, and 2) the gas in the bubble remains isothermal throughout the oscillations.[23,24] Given these assumptions, the wall motion of a bubble in an ultrasonic field with an angular frequency of o ¼ 2pf can be described by the Rayleigh–Plesset equation:     r 3g  P1 þ 2s r0 3 2 0 r€r þ ðr_ Þ þ 1  2 r r PA  sin ot þ r0 or b_r ¼ 0 r

ð1Þ

In this expression, r and r0 are, respectively, the instantaneous and equilibrium (i.e., when no sound field is acting on the liquid) values of the bubble radius; and r˙ and r¨ represent, respectively, the first and second order time derivatives of the instantaneous bubble radius; r is the liquid density; g is the polytropic exponent of the gas inside the bubble (i.e., the ratio of heat capacities, Cp=Cv); PA is the acoustic pressure amplitude; P1 is the hydrostatic (ambient) pressure; b is the bubble pulsation damping term that accounts for thermal, viscous, and radiation effects; s is the liquid surface tension; t is time; and or is the resonance frequency of the bubble, which is defined by the equation below: o2r

¼



1 rr02

  2sð3g  1Þ 3gP1 þ r0

ð2Þ

This equation of motion for bubble growth includes inertial terms that lead to a nonlinear solution; hence, the pressure threshold of bubble growth is dependent on the frequency of the sound field. The diffusion of gas in and out of the bubble results from a linear response to the concentration driving force and can be adequately described by Fick’s law of mass transfer:[26] dC @C ¼ þ n  HC ¼ DH2 C dt @t

ð3Þ

where C is the molar concentration of gas in the liquid, n is the liquid velocity, and D is the gas diffusivity in the liquid. By simultaneously solving the equation of state for the gas, the equation of motion for the bubble wall [Eq. (1)], and the equation for the diffusion of dissolved gas [Eq. (3)], it is possible to derive an expression for the mean rate of gas flow into the bubble: 2 !0:5 3 4 Þ i dn hðr r = 0 5 ¼ 4pDC1 r0 H 4hr =r0 i þ r0 pDt dt

ð4Þ

where the angled brackets represent time averages of the enclosed variables, n is the number of moles of gas in the bubble, C1 is the concentration of dissolved gas in the bulk liquid for the mean ambient pressure P1, and H is defined by the equation: H ¼ ðCi C0 Þ 

hðR=R0 Þ4  3g ð1 þ 2s=r0 P1 Þi hðR=R0 Þ4 i

ð5Þ

Further refinements to theories involving bubble growth account for the fact that in a high energy cavitating system the cavitation effects are a result of the entire bubble population rather than a single bubble.[20] The acoustic effects of power ultrasound largely depend on the size distribution of bubbles in the liquid. Depending on the characteristics of the local field, there is a critical size range in which a free floating bubble must lie to undergo transient cavitation. The bubble nuclei formed can grow to provide intense cavitational effects or grow to a resonant size where it will oscillate stably for many cycles. Church predicted that the number of bubble nuclei growing by rectified diffusion to transient collapse decreased with increasing frequency.[25] Leighton also observed that the measured growth rates were often much greater than the predicted values and he attributed this to a phenomenon called ‘‘microstreaming.’’[20] Microstreaming continuously refreshes the liquid at the bubble wall by bringing the liquid from further out close to the bubble wall. This helps to counteract the depletion of gas occurring during rectified diffusion. Microstreaming

S

2816

Sonochemical Reaction Engineering

also plays a role in dissolution by removing the excess dissolved gas concentration from the region adjacent to the bubble wall.

Bubble Collapse and Splitting The splitting or implosion of a transient cavity occurs when the bubble has reached a sufficiently large size so that the compression forces of an incoming sound wave exceed the stabilizing forces generated by the internal pressure and the surface tension of the bubble. The time required for bubble collapse is inversely related to the frequency of the ultrasonic wave and often occurs in only a few microseconds. Because the time required for thermal transport is significantly longer than that required for bubble collapse, the gases within the bubble undergo adiabatic compression, generating a localized ‘‘hot spot,’’ where temperatures can be as high as 5000 K. Further, the heating rates during bubble collapse are more than a billion degrees per second, and the sudden inrush of the surrounding fluid compresses the bubble contents to pressures of hundreds of atmospheres.[8,19] The high temperature is responsible for some of the production of highly reactive radicals, while the high pressure accounts for additional increases in chemical reactivity.[18] The maximum temperature and pressure reached during adiabatic collapse are given by the following equations:[27]

oppositely charged ions in the outer Helmholtz layer yielding local unbalanced electrical charges having a field strength above the critical electric field strength, which induces electrical microdischarges in the fluid. The discharge of these high energy electrons in the vicinity of other chemical species leads to the formation of highly reactive ions and free radicals. During cavitation, both hot spot and electrical discharge phenomena lead to the formation of active intermediates (ions and free radicals) that can survive for extended periods and react with species in the liquid layer surrounding the recently collapsed microbubble.

REACTION ZONES OF SONOCHEMICAL REACTIONS

ð7Þ

Electron paramagnetic resonance (EPR) and spin trapping studies have shown that there are three regions of sonochemical activity in liquid systems undergoing cavitation, as shown in Fig. 4.[28] Zone 1 contains a mixture of gas and vapor, where the amount of vapor in the bubble is directly related to the vapor pressure of species in the liquid phase. The high temperature and pressure generated during bubble collapse cause organic compounds to undergo pyrolysis reactions. Also, highly reactive radicals (e.g., OH
, H
, and O
) are generated in this zone because of the thermal dissociation of solvent (water) molecules in the vapor. The interfacial region, Zone 2, is also called the supercritical fluid region because the temperature and pressure in this zone during bubble collapse are above the critical temperature and pressure of the liquid (e.g., for water, 647 K and 221 bar). In this region, less

where T0 is the ambient temperature and P is the pressure in the bubble at its maximum size (sum of the vapor pressure of the liquid and the pressure of the gas in the bubble). There also exists a second mechanism for producing reactive ions and radicals via cavitation, and it involves the concepts of localized charging and electrical discharge.[14] There is a tendency for a surface potential to be developed at the vapor–liquid interface of a microbubble because of orientational effects and the presence of charged species in the liquid (e.g., low concentration impurities). Following the Helmholtz model, this surface potential leads to the formation of an electrochemical double layer, which has an inner layer of weakly solvated ions and an outer layer of oppositely charged, fully solvated species that can more freely diffuse through the liquid. During bubble collapse or splitting, the ions at the gas–liquid interface (inner Helmholtz layer) become separated from the

Fig. 4 Three reaction zones generated by acoustic cavitation. Zone 1 is the vapor phase, Zone 2 consists of the supercritical fluid region, and Zone 3 represents the bulk liquid. (View this art in color at www.dekker.com.)

Tmax ¼ T0 Pmax





ðPA þ P1 Þðg  1Þ P

ðPA þ P1 Þðg  1Þ ¼ P P



g=ðg  1Þ

ð6Þ

Sonochemical Reaction Engineering

volatile and less polar organic molecules are oxidized by the reactive radical species (e.g., OH
and H
) generated in Zone 1. Zone 3 consists of the bulk liquid phase outside the bubble. This zone is near ambient temperatures and no primary sonochemical activity takes place in this zone. The type of sonochemistry a given species will undergo is determined by its vapor pressure, polarity, and chemical composition (i.e., the nature and strength of its bonds). For example, hydrophobic compounds with high vapor pressure readily diffuse into the gas phase of growing microbubbles (assuming a polar solvent is used for the sonication reaction). Thus, these species very often undergo pyrolysis reactions in Zone 1 during bubble collapse, whereas less volatile hydrophobic species tend to accumulate at the vapor liquid interface during bubble growth and undergo reactions with reactive radical species generated in Zone 1 during bubble collapse.[29] Finally, nonvolatile, hydrophilic compounds do not readily diffuse into the vapor phase nor do they accumulate at the bubble interface; thus, they tend to be less affected by cavitation processes because they can react only with longlived radical species that diffuse out of Zone 1.

FACTORS AFFECTING SONICATION CHEMISTRY This section briefly discusses the factors that significantly affect cavity formation and cavitational intensity (i.e., the temperature and pressure generated during collapse). Acoustic Frequency Put simply, acoustic energy manifests itself as an oscillating pressure wave. The frequency of this pressure fluctuation (defined as the number of oscillations per second) is known to directly affect the critical size and number of cavitation bubbles. At very high frequencies (1 MHz), cavitation reduces or ceases to occur because of the very short rarefaction and compression cycles that either reduce cavity formation or prevent bubble collapse. At high frequencies (100 kHz–1 MHz) more cavities are formed, but the intensity of cavitation is lower because the average bubble volume before collapse is small (i.e., less material compressed during collapse). At low frequencies (20–100 kHz) fewer cavities are formed, but the resonance bubble size before collapse is larger; hence, cavitation is more violent, leading to higher localized temperatures and pressures at the cavitation sites.[12] For example, sonic irradiation of water at 20 kHz creates resonant bubbles with a radius of 161 mm, whereas the bubble radius is

2817

6.77 mm at 520 kHz.[30] Also, bubble lifetimes in water are shorter at higher frequencies (e.g., 3  107 sec at 514 kHz as compared to 3  105 sec at 20 kHz).[12] Thus, as acoustic frequency increases, production of cavities in the liquid increases (up to 1 MHz) and the intensity of cavitation decreases.[18] Acoustic Intensity Acoustic intensity (I ) is a measure of the amount of energy transmitted to the liquid and is defined as the rate at which the acoustic energy passes across a unit area perpendicular to the direction of the propagating sound wave. It can be shown that the acoustic intensity is proportional to the square of the amplitude (PA) of the acoustic wave divided by the density of the liquid (r) and the speed of sound in the liquid (c) (e.g., c ¼ 1500 m=sec for water):[20] I ¼

PA2 2rc

ð8Þ

To achieve any significant sonochemical effect, the sound intensity must be sufficiently high to overcome the cohesive forces in the liquid (i.e., break the intermolecular bonds) so that significant quantities of cavities can be formed. For degassed water at 25 C, the energy necessary to overcome the cohesive forces in the liquid is 6000 W=cm2 at an acoustic frequency of 500 kHz. With increases in the intensity value, the sonochemical rate increases because of the large number of cavities formed. Also, the range of bubble sizes undergoing transient cavitation increases, increasing reaction rates. However, several experiments have shown that there is an upper limit to the power input. When the intensity is increased beyond this upper limit, bubbles grow so large in a single rarefaction cycle that they have insufficient time to collapse before the next rarefaction cycle.[16] Additionally, these large bubbles shroud the surface of the acoustic generator, reducing the coupling of sound energy to the liquid. Thus, sonochemical reaction rates increase with increasing acoustic intensity up to some limit where mechanical damage to the sound generator can occur or the presence of large bubbles reduces the overall efficiency of the cavitation process.[16,12] External Pressure An increase in the ambient (or hydraulic) pressure has two effects. First, there is an increase in the minimum acoustic intensity required to initiate cavitation, which occurs when the acoustic wave has sufficient amplitude to overcome both the tensile strength of the liquid and the external pressure. The second effect is the increase

S

2818

in the intensity of cavitation collapse, leading to higher collapse temperatures and pressures. This is because the total pressure inside the cavitation bubble, just before collapse, is higher owing to the increased external pressure and applied ultrasound power. Even though the temperature and pressure in the final stages of collapse will increase, the overall sonochemical rate may or may not increase depending on the types of active intermediates formed and which reaction zone they are formed in. In contrast, as the external pressure is reduced, the severity of bubble collapse decreases and the dissolved gas content in the liquid decreases, eventually leading to a smaller number of microbubble nucleation sites (i.e., fewer cavitation bubbles). These effects combine to reduce the overall sonochemical reaction rate. Thus, the liquid surface tension and the nature of the specific chemical reaction determine the optimal external pressure for a sonochemical reaction.[12,16,18]

Bulk Temperature Most homogenous reactions are accelerated by moderate increases in temperature; however, sonochemical reactions are often slowed down by increases in the bulk fluid temperature. This counterintuitive observation is easily explained once one considers the cavitation processes that lead to the formation of reactive radicals and ions. As the fluid temperature increases, the vapor pressure of the solvent and dissolved species increases, which leads to greater quantities of vapor entering into the gas phase of cavitation microbubbles. This increased quantity of vapor in the bubble cushions the collapse, lowering the temperature and pressure generated during bubble implosion, thereby lowering the number of reactive intermediates formed. Additionally, increased temperatures decrease the cavitational threshold for the liquid, which leads to the formation of large numbers of microbubbles that serve to dampen the passage of ultrasound energy through the liquid medium.[17] Hence, sonochemical reactions are usually carried out within a temperature range of 10–35 C. Presence and Nature of Dissolved Gases The presence of dissolved gas is essential for cavitation to occur in a liquid. The dissolved gas molecules disrupt intermolecular bonding between solvent molecules and hence, serve as nucleation sites for cavitation. There are three properties of dissolved gases that have significant influence on the degree of nucleation and cavitational intensity: solubility of gas in the liquid, ratio of specific heats (g or CP=CV), and thermal conductivity (l). More soluble gases reduce the cavitational effects because the bubbles formed redissolve

Sonochemical Reaction Engineering

before collapse can occur.[12] Also, the greater the solubility of the gas, the greater the amount that penetrates the bubble, thereby inducing the cushioning effect and lowering the intensity of shock waves released after collapse.[17,18] As the gas content in the liquid increases, both the cavitation threshold and the cavitational intensity are reduced.[18] High degradation of organic compounds is achieved when a gas with high average specific heat ratio is employed. From Eqs. (6) and (7), the maximum temperature and pressure generated during adiabatic collapse increase with an increase in the value of g, leading to higher degradation. This has been confirmed by sonochemical degradation studies of aqueous solutions of carbon tetrachloride, which showed that the initial rate of formation of free chlorine was less when the solution was saturated with nitrogen (g ¼ 1.40) instead of argon (g ¼ 1.66).[31] Solutions saturated with monatomic (inert) gases (Ar, Ne, He) give the highest rate with diatomic gases (O2 and N2) proceeding at an intermediate rate, and the lowest rate occurs for polyatomic gases (CO, etc.). The thermal conductivity of the gas also has an important effect on cavitation intensity. Dissolved gases with high thermal conductivity can more rapidly dissipate heat generated during collapse to the surroundings, effectively reducing the maximum temperature (Tmax) attained. Hence, even though helium (g ¼ 1.66), being an inert gas, has a high g value, the maximum temperature attained during collapse is lower (with all the other conditions maintained the same) than that of nitrogen (g ¼ 1.4) because of the higher l value of the former (lHe ¼ 14.30  and lN2 ¼ 2.52  102 W=m=K). 102 W=m=K Thus, greater cavitation intensity is achieved when the liquid is saturated with a high g and low l inert gas.

Solvent The nature and strength of solvent intermolecular interactions (especially the presence of hydrogen bonding) can greatly influence the physical and chemical outcomes of ultrasound irradiation. These interactions determine the intensity of bubble collapse and the ease with which bubble nucleation can occur (recall that nucleation occurs when the pressure amplitude of the ultrasound wave exceeds the natural cohesive forces in the liquid). Increases in solvent viscosity and surface tension reduce the rate of bubble nucleation (i.e., fewer microbubbles are formed) but increase the intensity of bubble collapse (i.e., higher temperatures and pressures).[17,18] Some of the adverse effects of high surface tension can be overcome with the addition of small amounts of surfactants, which reduce the solvent surface tension and facilitate bubble nucleation.[12,17]

Sonochemical Reaction Engineering

Solvent vapor pressure also has a significant effect on the cavitation phenomenon because the intensity of cavitation decreases as the vapor pressure of the solvent increases. This is because more vapor is enclosed in the microbubble, which cushions the collapse, leading to lower collapse temperatures and pressures. On the other hand, solvents with low vapor pressure tend not to diffuse into the growing microbubble; thereby reducing the size of the bubble, which lessens the intensity of bubble collapse.[17] Thus, a delicate balance of solvent properties must be achieved to attain the desired sonication conditions.

TYPES OF CAVITATION REACTIONS Ultrasound irradiation can affect chemical reactions in two basic ways. It will either activate a new reaction or enhance the rate of an existing reaction pathway. Most primary activation processes involve reactions that occur in Zone 1 during bubble collapse, which are generally thought to be thermally initiated processes. During collapse, vaporized molecules are dissociated into smaller fragments (e.g., H
, Cl , CH
3 ), which diffuse into the surrounding liquid media and react with other compounds, degrade into even smaller reactive species, or recombine to form low molecular weight compounds, such as CO2, H2O, and HCl. Also, solvent molecules are commonly dissociated into highly reactive radical species (e.g., CH
3 , OH
, H
). This reaction mechanism is the main degradation path for volatile organics with high vapor pressure and is predominant at lower ultrasound frequencies. The concentration of these high energy dissociation products is relatively small and their lifetimes short, but it is possible for them to diffuse to Zones 2 and 3 where they can react with other nonactivated species. Secondary reactions involving free radicals (e.g., OH
and H
), which were created by pyrolysis reactions in Zone 1, take place in the supercritical fluid region (Zone 2) and in the bulk liquid (Zone 3). These reactions involve species that strongly associate with the solvent or low volatility organics that cannot easily diffuse into the bubble. Higher ultrasound frequencies generally enhance these types of reactions. This is because at low ultrasound frequencies, the decomposition of the vapor gas mixture inside the microbubble is very efficient, yielding numerous free radicals. However, the proportion of radicals undergoing recombination to form unreactive species is also very high, whereas, at higher ultrasound frequencies fewer radicals are formed, but a greater percentage escape from the bubble; hence, the concentration of radical species exiting Zone 1 is greater at higher ultrasound frequencies.[32]

2819

The number of organic reactions that have been initiated or enhanced by sonochemical methods is vast. Table 1 provides a brief listing of some of these. More detailed and thorough reviews of this subject can be found in the literature.[11,12,16,19,30–36] Sonochemical methods have proven to be efficient means to achieve a wide array of addition, isomerization, and oxidation reactions, while slightly fewer studies have examined the use of ultrasound for reduction and substitution reactions. In general, ultrasound irradiation has been shown to play varying roles in homogeneous and heterogeneous synthetic chemistry. For some reaction systems, the extreme conditions encountered during bubble collapse cause the initiation of a new reaction (e.g., the reduction of alkoxysilanes using LiAlH4), whereas other reaction systems simply experience an increase in reaction rates that results from enhanced mixing and, in some cases, an increase in the number of highly reactive free radical intermediates (e.g., the oxidation of organics using KMnO4). Most industrial applications of sonochemistry have involved the degradation of organic compounds, especially the destruction (pyrolysis or oxidation) of organics, such as aromatics and chlorinated organics, dissolved in water. These compounds are major contaminants in industrial and agricultural wastewater, and nearcomplete destruction (conversion to CO2 and H2O) of low molecular weight organics having high volatilities has been achieved. Further, dehalogenation reactions have been shown to occur with organic species having much lower volatilities (chlorobenzene, polychlorobiphenyls, etc.).

SONOLYSIS OF WATER Though there are many important reactions initiated by ultrasound, water dissociation is by far the most important, and hence, is discussed here in detail. Weiss proposed and many have confirmed the mechanism shown below for the formation of H
and OH
radicals in Zone 1 during bubble collapse:[37] H2 O ! H
þ OH
The formation of these free radical intermediates and subsequent sonochemical reactions carried out by them depend largely on the nature of the dissolved gases in the aqueous medium. Studies have shown that the sonolysis of distilled water saturated with different pure gases and their mixtures has different hydrogen peroxide yields and cavitation threshold values.[31] In general, these studies show that a gas with a high ratio of heat capacities and low thermal conductivity yields higher rates of production of hydrogen peroxide, indicating increased levels of free radical production

S

2820

Sonochemical Reaction Engineering

Table 1 A representative list of organic reactions that are initiated or enhanced via irradiation with ultrasound Homogenous organic reactions Addition reactions Aldol condensations

Hydroboration reactions

Alkene hydrosilations

Knoevenagel condensations

Alkylation reactions

Michael addition reactions

Claisen–Schmidt condensations

Reformatsky reactions

Diels–Alder cyclizations

Ring opening polymerizations

Enantioselective Barbier reactions

Ullam coupling reactions

Esterification reactions

Wittig reactions to form olefin

Nucleophilic and electrophilic substitution reactions Ester hydrolysis

Halogenation of aromatics

Friedel–Crafts acylations

Nitration of aromatics

Friedel–Crafts alkylations

Solvolysis of halogenated alkanes

Oxidation reactions Alcohol oxidation to ketones and carboxylic acids

Oxidation of halogenated alkanes=aromatics

Epoxidation of olefins

Oxidation of arylalkyls

Reduction reactions Clemmensen reduction of carbonyls

Reduction of alkoxy silanes

Hydrogenation of carbonyls

Selective reduction of olefins

Degradation reactions Complete pyrolysis of organics

Polymer degradation

Dehalogenation reactions

Ozonolysis

via the thermal decomposition of water. The following free radical reactions were proposed for water upon ultrasonic irradiation under argon:[16] H2 O $ H
þ OH


H
þ H
! H2 OH
þ OH
! H2 O2 Hydrogen peroxide is more readily formed when water contains oxygen, but the presence of oxygen is not necessary for hydrogen peroxide formation.[31] In the presence of oxygen, more H atoms are scavenged via the following reaction: H
þ O2 ! HOO
This reaction prevents the recombination of H
and OH
radicals to form water molecules and hence, increases the rate of other oxidation processes. In general, free radical reactions similar to those shown for water could occur with any organic or inorganic species capable of being present as a vapor during bubble collapse.

TYPES OF SONOCHEMICAL REACTORS There are a variety of sonochemical reactor designs that have been developed for laboratory use as well as large scale production units. The method of ultrasound generation depends largely on the size of the reactor with piezoelectric transducers being used for small systems and magnetostrictive transducers more commonly found in large scale reactors. There are three basic types of sonochemical reactors: 1) reactors having the ultrasonic source in direct contact with the reacting fluid; 2) reactors with the ultrasound source mounted on the exterior walls of the reactor (i.e., a vibrating wall reactor); and 3) reactors that receive sound energy from an exterior coupling fluid that is in direct contact with the ultrasound source. There are several recent reviews of sonochemical reactors in the literature, and these works should be consulted before deciding on a particular sonochemical reactor configuration.[11,12,16] There are three major factors that must be taken into consideration when selecting the appropriate reactor type: cost, level of contaminants (from the erosion of the ultrasound source), and ultrasound energy density required.

Sonochemical Reaction Engineering

Fig. 5 Flow through reactor with direct sonication via an amplifying horn.

The first reactor type, which employs direct sonication, is represented by the lab scale system shown in Fig. 5. This flow reactor assembly consists of a titanium horn assembly, a piezoelectric transducer capable of variable energy output, and a flow through reaction chamber. This design provides for maximum

2821

utilization of the sound energy output from the transducer, but at a price. The replaceable tip of the titanium horn is prone to erosion and pitting, which can lead to a reduction in energy utilization as well as contamination of the process stream. The second type of sonochemical reactor normally consists of a flat walled vessel that has multiple piezoelectric transducers mounted on the exterior surface of the reactor (in much the same way ultrasonic baths are arranged). These reactors have noncylindrical geometries because of difficulties associated with mounting the ceramic transducers on a curved surface (the transducer is normally glued to the exterior surface of the reactor). More optimized reactor designs of this type have multiple transducers located on opposing sides of the reactor, which creates a more uniform sound field for cavitation. An advantage of this reactor type is that corrosion byproducts from the ultrasound source cannot contaminate the reacting media; however, the coupling of the transducers to the reactor wall can be inefficient, and it is difficult to transmit high energy fluxes through the reactor walls. The final reactor configuration type usually has the reacting fluid pass through a metal tube that is surrounded by a coupling fluid that is in direct contact with an ultrasound source. An efficient variant of this reactor type is the Branson sonochemical reactor shown in Fig. 6. This reactor configuration consists of a straight pipe for the reacting media and a series of external pipe sections that are arranged perpendicular to the reactant flow. This design is nonintrusive, hence there is little chance of contamination of the reagents; however, any design that uses probes

Fig. 6 Branson sonochemical reactor employing indirect sonication in a tubular configuration.

S

2822

or horns to focus the acoustic energy is at a disadvantage because of tip erosion, which ultimately leads to greater maintenance costs. CONCLUSIONS Acoustic cavitation provides a relatively simple means for accessing high energy species in a reactor assembly that operates at near-ambient conditions. All sonochemical reaction processes are initiated by the introduction of high intensity sound waves into a liquid with moderately high surface tension. These high frequency pressure waves generate micrometer size bubbles that grow until they suddenly collapse (acoustic cavitation), generating extremely high pressures and temperatures (up to 1000 atm and 5000
C) in localized regions of the fluid. Such extreme thermal conditions lead to the formation of a variety of highly reactive free radicals and charged species. The interaction of these reactive intermediates with neighboring (cooler, low energy) species results in unique homogeneous, and in some cases heterogeneous, chemical conversions. To date, sonochemical transformations have been effectively employed in a wide variety of wastewater treatment applications; however, a number of efficiency and contamination issues have limited their use in the large scale production of organic chemicals and inorganic materials. As the technology for introducing high energy sound waves into liquids continues to advance, it is likely that sonochemical processing will become an ever more widely used tool to prepare high value added chemicals and materials. REFERENCES 1. Thornycroft, J.I.; Barnaby, S.W. Torpedo-boat destroyers. Min. Proc. Inst. Chem. Eng. 1895, 122 (4), 51–69. 2. Rayleigh, Lord. On the pressure develop in a liquid during collapse of a spherical cavity. Philos. Mag. Ser. 6, 1917, 34 (200), 94–98. 3. Wood, R.W.; Loomis, A.L. The physical and biological effects of high frequency sound waves of great intensity. Philos. Mag. Ser. 7, 1927, 4 (22), 417–436. 4. Richards, W.T.; Loomis, A.L. The chemical effects of high frequency sound waves. I. A preliminary study. J. Am. Chem. Soc. 1927, 49, 3086–3100. 5. Harvey, E.N.; Barnes, D.K.; McElroy, W.D.; Whitely, A.H.; Pease, D.C.; Cooper, K.W. Bubble formation in animals. J. Cell. Comp. Physiol. 1944, 24, 1–22. 6. Noltingk, B.E.; Neppiras, E.A. Cavitation produced by ultrasonics. Proc. Phys. Soc. 1950, 63B, 674–685.

Sonochemical Reaction Engineering

7. Naude, C.F.; Ellis, A.T. On the mechanism of cavitation damage by non-hemispherical cavities in contact with a solid boundary. J. Basic Eng. 1961, 83, 648–656. 8. Suslick, K.S.; Hammerton, D.A.; Cline, R.E. The sonochemical hot spot. J. Am. Chem. Soc. 1986, 108, 5641–5642. 9. Suslick, K.S. Sonochemistry. Science 1990, 247 (4949), 1439–1441. 10. Makino, K.; Mossoba, M.M.; Reisz, P. Chemical effects of ultrasound on aqueous solutions. Formation of hydroxyl radicals and hydrogen atoms. J. Phys. Chem. 1983, 87, 1369– 1377. 11. Adewuyi, Y.G. Sonochemistry: environmental science and engineering applications. Ind. Eng. Chem. Res. 2001, 40, 4681–4715. 12. Thompson, L.H.; Doraiswamy, L.K. Sonochemistry: science and engineering. Ind. Eng. Chem. Res. 1999, 38, 1215–1249. 13. Gedanken, A. Using sonochemistry for the fabrication of nanomaterials. Ultrason. Sonochem. 2004, 11, 47–55. 14. Margulis, M.A.; Margulis, I.M. Mechanism of sonochemical reactions and sonoluminescence. High Energ. Chem. 2004, 38 (5), 285–294. 15. Pandit, A.B.; Moholkar, V.S. Harness cavitation to improve processing. Chem. Eng. Prog. 1996, 7, 57–68. 16. Shah, Y.T.; Pandit, A.B.; Moholkar, V.S. Cavitation Reaction Engineering; Kluwer Academic= Plenum Publishers: New York, 1999. 17. Moholkar, V.S.; Shirgaonkar, I.Z.; Pandit, A.B. Cavitation and sonochemistry in the eyes of a chemical engineer. Indian Chem. Eng. B. 1996, 38 (2), 81–93. 18. Lorimer, J.P.; Mason, T.J. Sonochemistry: part 1— the physical aspects. Chem. Soc. Rev. 1987, 16, 239–274. 19. Suslick, K.S. The chemical effects of ultrasound. Sci. Am. 1989, 260, 80–86. 20. Leighton, T.G. Bubble population phenomena in acoustic cavitation. Ultrason. Sonochem. 1995, 2 (2), S123–S135. 21. Hsieh, D.Y.; Plesset, M.S. Theory of rectified diffusion of mass into gas bubbles. J. Acoust. Soc. Am. 1961, 33 (2), 206–215. 22. Eller, A.; Flynn, H.G. Rectified diffusion during nonlinear pulsations of cavitation bubbles. J. Acoust. Soc. Am. 1965, 37 (3), 493–503. 23. Crum, L.A.; Hansen, G.M. Generalized equations for rectified diffusion. J. Acoust. Soc. Am. 1982, 72 (5), 1586–1592. 24. Crum, L.A. Rectified diffusion. Ultrasonics 1984, 22L, 215–223.

Sonochemical Reaction Engineering

25. Church, C.C. A method to account for acoustic microstreaming when predicting bubble-growth rates produced by rectified diffusion. J. Acoust. Soc. Am. 1988, 84 (5), 1758–1764. 26. Bird, R.B.; Stewart, W.E.; Lightfoot, E.N. Transport Phenomena; John Wiley & Sons: New York, 1960. 27. Neppiras, E.A. Acoustic cavitation. Phys. Rep. 1980, 61 (3), 159–251. 28. Riesz, P.; Kondo, T.; Krishna, C.M. Sonochemistry of volatile and non-volatile solutes in aqueous solutions: E.P.R and spin trapping studies. Ultrasonics 1990, 28, 295–303. 29. Drijvers, D.; Van Langenhove, H.; Vervaet, K. Sonolysis of chlorobenzene in aqueous solution: organic intermediates. Ultrason. Sonochem. 1998, 5 (1), 13–19. 30. De Visscher, A.; Van Langenhove, H. Sonochemistry of organic compounds in homogeneous aqueous oxidising systems. Ultrason. Sonochem. 1998, 5 (3), 87–92.

2823

31. Fitzgerald, M.E.; Griffing, V.; Sullivan, J. Chemical effects of ultrasonics—‘‘Hot Spot’’ chemistry. J. Chem. Phys. 1956, 25 (5), 926–933. 32. Mason, T.J. Sonochemistry: current uses and future prospects in the chemical and processing industries. Philos. Trans. R. Soc. Lond. A. 1999, 357, 355–369. 33. Bremner, D.H. Recent advances in organic synthesis utilizing ultrasound. Ultrason. Sonochem. 1994, 1 (2), S119–S124. 34. Low, C.M.R. Ultrasound in synthesis: natural products and supersonic reactions? Ultrason. Sonochem. 1995, 2 (2), S153–S163. 35. Einhorn, C.; Einhorn, J.; Luche, J.L. Sonochemistry: the use of ultrasonic waves in synthetic organic chemistry. Synthesis 1989, Nov, 787–813. 36. Ley, S.V.; Low, C.M.R. Ultrasound in Synthesis; Springer-Verlag: Berlin, 1989. 37. Weiss, J. Radiochemistry of aqueous solutions. Nature 1944, 153, 748–750.

S

Sorbent Technology S Shuguang Deng Chemical Engineering Department, New Mexico State University, Las Cruces, New Mexico, U.S.A.

INTRODUCTION This article covers the fundamentals, status, and future developments of sorbent materials and their applications in adsorptive separation and purification processes. A sorbent is usually a solid substance that adsorbs or absorbs another type of substance. It is the sorbent that makes a sorption process a unique and different separation and purification process from others. With the rapid development in novel sorbent materials and innovative cyclic adsorption processes, sorption has become a key separation process in many process industries including chemical, petrochemical, environmental, pharmaceutical, and electronic gases. A brief review of the fundamentals of adsorption and the basic requirements for sorbent materials is presented, followed with a summary of the status of commercial sorbents and their applications. The focus of this article is placed on recent advances in novel sorbent materials including oxide molecular sieves, sol–gel derived xerogels and aerogels, metal organic framework, hydrogen storage media, p-complexation and composite sorbents, and high-temperature sorbents for oxygen or carbon dioxide sorption. A concluding section outlines the future research needs and opportunities in sorbent technology development for new energy and environmental applications.

ADSORPTION MECHANISMS AND SORBENT MATERIALS According to King, a mass separating agent is needed to facilitate separation for many separation processes.[1] The mass separating agent for adsorption process is the adsorbent, or the sorbent. Therefore, the characteristic of the sorbent directly decides the performance of any adsorptive separation or purification process. The basic definitions of adsorption-related terminologies are given in the following to clarify and standardize these widely used terms in this field. Adsorption: The adhesion of molecules (as of gases, solutes, or liquids) to the surfaces of solid bodies or liquids with which they are in contact. Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120007963 Copyright # 2006 by Taylor & Francis. All rights reserved.

Absorption: The absorbing of molecules (as of gases, solutes, or liquids) into the solid bodies or liquids with which they are in contact. Sorption: Formation from adsorption and absorption. Adsorbent: A usually solid substance that adsorbs another substance on its surface. Sorbent: A usually solid substance that adsorbs and absorbs another substance. Adsorbate: Molecules (as of gases, solutes, or liquids) that are adsorbed on adsorbent surfaces. ˚. Microporous: Pore size smaller than 20 A ˚. Mesoporous: Pore size between 20 and 500 A ˚. Macroporous: Pore size larger than 500 A Adsorptive separation can be achieved through one of the following mechanisms. Understanding the fundamentals of adsorptive separation mechanisms will allow us to better design or modify sorbent materials to achieve their best possible separation performance.[2–4] Adsorption equilibrium effect is because of the difference in the thermodynamic equilibria for each adsorbate–adsorbent interaction. The majority of adsorptive separation and purification processes are based on equilibrium effect. One example is to generate oxygen-enriched air or relatively pure oxygen (95%) from air using a zeolite molecular sieve 5A or 13X in either a pressure swing adsorption (PSA) or a vacuum swing adsorption (VSA) process. In this case, nitrogen is selectively adsorbed by the zeolite adsorbent, and oxygen is collected from the adsorption effluent stream. Adsorption kinetics effect arises because of the difference of rates at which different adsorbate molecules travel into the internal structure of the adsorbent. There are only a few commercial successes using adsorption kinetic difference to achieve adsorptive separation of gases. The typical example is separation of nitrogen from air using a carbon molecular sieve (CMS). The CMS adsorbent has a similar adsorption equilibrium capacity for both nitrogen and oxygen, but the diffusivity of oxygen in CMS is at least 30 times larger than that of nitrogen in CMS.[5] High-purity 2825

2826

Sorbent Technology

nitrogen can be recovered from the adsorption effluent stream in a PSA process because oxygen moves much faster than nitrogen into the micropores of CMS adsorbent. However, the cycle time of this CMS-based PSA process is much shorter than that of a typical PSA process based on adsorption equilibrium effect. This is because there will be no separation if both nitrogen and oxygen are allowed to reach adsorption equilibrium with the CMS adsorbent. Molecular sieving effect, also called steric effect, is derived from the molecular sieving properties of some adsorbents with a microporous structure. In this case, the pore openings of the adsorbent structure are small enough to exclude large adsorbate molecules from penetrating the micropores of the adsorbent. This is the extreme case of the kinetic effect. There are several commercial applications based on this mechanism in adsorptive separation processes. One typical example is separating normal paraffin from iso-paraffin and aromatics in an adsorption process using zeolite 5A as an adsorbent. n-Paraffin, with a long straight chain, has a smaller effective diameter than the well-defined aperture of zeolite 5A. Therefore it adsorbs in the micropores of the adsorbent during the adsorption step, and is recovered from the adsorbed phase in the desorption step. A representative process for n-paraffin separation from naphtha and kerosene is UOP’s Molex process that employs a simulated moving bed with binderless zeolite 5A as an adsorbent and light paraffin as a desorbent.[6] We can define separation factor and selectivity as the ability of an adsorbent to separate molecule A from molecule B as:[7] Separation factor : aAB ¼

XA =YA XB =YB

ð1Þ

Here XA, YA are strictly equilibrium mole fractions for component A in the adsorbed phase and adsorbate (fluid) phase, respectively; as are XB, YB for component B. For equilibrium-based adsorptive separation process, the adsorbent selectivity is the same as the separation factor as defined in Eq. (1). Apparently, this definition is not applicable to other processes based on kinetic and steric effects. In a kinetically controlled adsorption process, the adsorbent selectivity depends on both equilibrium and kinetic effects. A simplified definition for adsorbent separation factor is given by Ruthven et al.:[8]

SAB

KA ¼ KB

rffiffiffiffiffiffiffi DA DB

ð2Þ

where SAB is the adsorbent selectivity, K is the adsorption equilibrium constant or isotherm slope, and D is the

effective diffusivity. Although the above equation is strictly valid under the assumptions that components A and B have independent linear adsorption isotherms and independent diffusion process, it provides a good estimate of adsorbent selectivity for kinetically controlled processes. Theoretically speaking, selectivity for adsorbents with a molecular sieving effect should be infinitely large because the larger molecules are excluded from getting into the adsorbent micropores. In reality, the adsorbent selectivity for steric effect is somewhat reduced by combining with the equilibrium effect from adsorption on the surface of large pores. So adsorption processes based on molecular sieving are usually considered as adsorption equilibrium effect. Another very important adsorbent property affecting the adsorption process is the adsorption capacity because it determines the size of an adsorbent vessel, the amount of adsorbents required, and the related capital and operating costs. The requirements for commercial sorbents are discussed briefly as follows.

Characteristics of Sorbent Materials Commercial sorbents used in cyclic adsorption processes should ideally meet the following requirements:
Large selectivity derived from equilibrium, kinetic, or steric effect;
Large adsorption capacity;
Fast adsorption kinetics;
Easily regenerable;
Good mechanical strength;
Low cost. The above adsorbent performance requirements can simply transfer to adsorbent characteristic requirements as follows:
Large internal pore volume;
Large internal surface area;
Controlled surface properties through selected functional groups;
Controlled pore size distribution, preferably in micropore range;
Weak interactions between adsorbate and adsorbent (mostly on physical sorbents);
Inorganic or ceramic materials to enhance chemical and mechanical stability;
Low-cost raw materials. These basic requirements are usually proposed for adsorbents used in cyclic adsorption processes that are based on physical adsorption. There is an increasing demand for strong chemical adsorbents used in

Sorbent Technology

2827

purification processes to remove trace contaminants from main stream fluids such as the removal of very toxic contaminants from electronic process gas streams, and the removal of toxic, or radioactive species from contaminated water. In these cases, the sorbents are used as getter materials; no regeneration is needed, and instead, the spent sorbent materials are disposed of in designated areas regulated by government environmental policies.

COMMERCIAL SORBENTS AND APPLICATIONS An excellent review and detailed coverage on commercial adsorbents and new adsorbent materials has been presented by Yang in his newly published monograph on adsorbents.[2] A very brief overview of existing commercial adsorbents is given here. Commercial sorbents that have been used in large-scale adsorptive separation and purification processes include activated carbon, zeolites, activated alumina, silica gel, and polymeric adsorbents. Although the worldwide sales of sorbent materials are relatively small as compared with other chemical commodities, sorbents and adsorption processes play a very important role in many process industries. The estimated worldwide sales of these sorbents are as follows:[2]
Activated carbon: $1 billion
Zeolite: $1.07 billion
Activated alumina: $63 million

Cumulative pore volume, cm3/100 gm

60


Silica gel: $71 million
Polymeric adsorbents: $50 million Activated Carbon Activated carbons are unique and versatile adsorbents because of their large surface area, microporous and mesoporous structure, universal adsorption effect, high adsorption capacity for many nonpolar molecules including organic molecules, and high degree of surface reactivity. They are used widely in industrial applications that include decolorizing sugar solutions, personnel protection, solvent recovery, volatile organic compound removal from air and water, water treatment, hydrogen and synthesis gas separation, and natural gas storage.[4,9,10] Activated carbons are produced in two main steps: carbonization of the carbonaceous raw materials at temperatures below 800  C in the absence of oxygen, and activation of the carbonized products.[10] The properties of activated carbon depend largely on the nature of the raw materials, the activating agents and activation conditions. For gas-phase applications, activated carbons are usually made in pellets with mostly micropores; while for liquid-phase applications, activated carbon is produced in powder form with relatively large mesopores to enhance mass transfer rate in the carbons. Fig. 1 compares the pore size distributions of major commercial adsorbents discussed in this section. Activated carbons have a broad pore size distribution like activated alumina and silica gel. Although activated carbon is thought to be ‘‘hydrophobic,’’ it does adsorb

Activated carbon

50 Silica gel 40

30 Zeolite 5A 20 Activated alumina

10

0

MSC MSC 2

5

10

20

Pore diameter, Å

50

Fig. 1 Pore size distributions for activated carbon, silica gel, activated alumina, two molecular sieve carbons (MSCs), and zeolite 5A. (From Ref.[3].)

S

2828

Sorbent Technology

quite significant amount of water (>30 wt%) when relative humidity is higher than 50%. An example isotherm of water on activated carbon, along with water isotherms on other commercial adsorbents, is plotted in Fig. 2. The change from ‘‘hydrophobic’’ to ‘‘hydrophilic’’ on the activated carbon surface is attributed to the initial adsorbed water film on the carbon surface. This occurs because when the carbon surface is fully covered with a layer of water molecules, the adsorbed water molecules exhibit strong affinity to other polar molecules including water. Carbon molecular sieve (CMS) is a specially made carbonaceous material with ˚ ). The major very narrow pore size distribution (4–9 A application of CMS is in the generation of high-purity nitrogen from air in a PSA process. The representative physical properties of commercial adsorbents and their major applications are summarized in Tables 1 and 2, respectively. Zeolites Zeolites are porous crystalline aluminosilicates that are made of assemblies of SiO4 and AlO4 tetrahedra joined together through shared oxygen atoms. The general chemical formula for zeolites is: Mx=n ½ðAlO2 Þx ðSiO2 Þy zH2 O

ð3Þ

where x and y are integers with y=x (Si=Al ratio) equal or larger than 1; n is the valance of cation M, and z is the number of water molecules in each unit cell. The tetrahedra can be arranged in many different ways to form different crystalline structures. Some zeolites

exist as minerals in nature, but all commercially important zeolites are synthetic. Zeolites are unique adsorbents owing to their special surface chemistries and crystalline pore structures. It should be pointed out that probably only 10% of $1 billion worldwide sales of zeolite is used as adsorbents; the majority of commercial zeolites are used as detergent additives (zeolite 4A), animal food additives (zeolite 4A), ion exchange, and catalyst supports. Among all commercial sorbents zeolites are probably the most extensively investigated and documented. Many excellent monographs and review articles are available.[2,11–13] Please refer to Tables 1 and 2 for properties and major applications of zeolites.

Activated Alumina Activated alumina is a porous high-surface area form of aluminum oxide with the formula of Al2O3nH2O. Commercially, it is prepared either from thermal dehydration of aluminum trihydrate, Al(OH)3, or directly from bauxite (Al2O33H2O), as a by-product of the Bayer process for alumina extraction from bauxite. Its surface is more polar than that of silica gel and, reflecting the amphoteric nature of aluminum, has both acidic and basic characteristics. Surface areas are in the range 250–350 m2=g depending on the activation temperature and the source of raw materials. Because activated alumina has a higher capacity for water than silica gel at elevated temperatures it is used mainly as a desiccant for warm gases including air, but in many commercial applications it has now been replaced by zeolitic materials in a thermal swing

Adsorption, kg H2O/100 kg adsorbent

40 E 30

20

D C B

10

A

0

0

20

40 60 Relative humidity, %

80

100

Fig. 2 Equilibrium sorption of water vapor from atmospheric air at 25  C on: (A) alumina (granular), (B) alumina (spherical), (C) silica gel, (D) 5A zeolite, and (E) activated carbon. The vapor pressure at 100% relative humidity is 23.6 torr. (From Ref.[3].)

Sorbent Technology

2829

Table 1 Representative physical properties of commercial adsorbents Adsorbent Activated carbon

Nature

Specific surface area (m2/g)

Pore ˚) diameter (A

Porosity

Particle density (g/cm3)

Hydrophobic amorphous

Small pore

400–1200

10–25

0.4–0.6

0.5–0.9

Large pore

200–600

> 30

0.5

0.6–0.8

Zeolite

Hydrophilic=hydrophobic crystalline

600–700

3–10

0.6

1.0

Activated alumina

Hydrophilic crystalline=x-ray amorphous

200–350

10–75

0.5

1.25

Silica gel

Hydrophilic=hydrophobic amorphous 750–850

22–26

0.47

1.09

Small pore Large pore

300–350

100–150

0.71

1.62

Polymeric adsorbent

Hydrophilic=hydrophobic

450–1100

25–90

0.5

1.25

Carbon molecular sieve

Hydrophilic

400

3–9

0.5

1.0

adsorption (TSA) process. However, activated alumina has a low adsorption heat for water and other polar molecules as compared with zeolite; it is possible to regenerate activated alumina under PSA conditions. Activated alumina also demonstrates moderate adsorption affinity for carbon dioxide, which makes it a suitable sorbent for removing water and carbon dioxide from air in a PSA process. These adsorption properties of activated alumina have been explored extensively for air purification applications by industrial gas companies.[14–17] This is a perfect example to demonstrate the importance of sorbent regenerability over sorption capacity and selectivity in pressure swing adsorption processes. Activated alumina is also an excellent catalyst support. More applications and representative properties of activated alumina are listed in Tables 1 and 2.

Silica Gels Silica gel is the most widely used desiccant because of its large adsorption capacity for water (40 wt%), as shown in Fig. 2, and easy for regeneration (150 C, compared with 350 C for zeolites). Silica is a partially dehydrated polymeric form of colloidal silicic acid with the formula of SiO2nH2O. Its water content, which is typically about 5 wt%, is presented in the chemically bonded hydroxyl groups. Silica is an amorphous mate˚ in size, rial comprising spherical particles of 20–200 A which aggregate to form the sorbent with pore sizes ˚ and surface areas of 100– in the range of 60–250 A 850 m2=g, depending on gel density. Its surface has mainly Si–OH and Si–O–Si polar groups; this is why it can be used to adsorb water, alcohols, phenols,

amines, etc. by hydrogen bonding mechanisms. Other commercial applications include the separation of aromatics from paraffins, the chromatographic separation of organic molecules, and modified silica in chromatography columns.[2,18–20] Polymeric Adsorbents A wide range of synthetic, nonionic polymers are available for use as sorbents, ion-exchange resins, and particularly for analytical chromatography applications. Commercially available resins in bead form (typically 0.5 mm in diameter) are based usually on copolymers of styrene=divinyl benzene (DVB) and acrylic acid esters=divinyl benzene, and have a wide range of surface polarities, porosities, and macropore sizes. The porosities can be built through emulsion polymerization of relevant monomers in the presence of a solvent that dissolves the monomers and serves as a poor swelling agent for the polymer. This creates a polymer matrix with surface areas ranging up to 1100 m2=g.[2,4] The major application of polymeric adsorbents is in water treatment. The macroporous polymeric resins can be modified by attaching different functional groups to mimic activated carbon, and to replace activated carbons for certain specific applications in food and pharmaceutical industries where color contamination by the black carbons of the final products is a major concern.

NEW DEVELOPMENTS IN SORBENT MATERIALS AND APPLICATIONS The past two decades have witnessed major advances in new nanostructured sorbent materials including

S

2830

Sorbent Technology

Table 2 Selected applications of commercial sorbents Adsorbent

Applications (the first molecule is the product)

Activated carbon

Hydrogen separation from syngas and hydrogenation processes Ethylene from methane and hydrogen Vinyl chloride monomer (VCM) from air Removal of odors from gases Recovery of solvent vapors Removal of SOx, and NOx Purification of helium Clean-up of nuclear off-gases Decolorizing of syrups, sugars, and molasses Water purification, including removal of phenol, halogenated compounds, pesticides, caprolactam, chlorine

Carbon molecular sieve

Nitrogen separation from air

Zeolite

Oxygen from air Drying of gases Removing water from azeotropes Sweetening sour gases and liquids Purification of hydrogen Separation of ammonia and hydrogen Recovery of carbon dioxide Separation of oxygen and argon Removal of acetylene, propane, and butane from air Separation of xylenes and ethyl benzene Separation of normal from branched paraffins Separation of olefins and aromatics from paraffins Recovery of carbon monoxide from methane and hydrogen Purification of nuclear off-gases Separation of cresols Drying of refrigerants and organic liquids Separation of solvent systems Pollution control, including removal of Hg, NOx, and SOx from gases Recovery of fructose from corn syrup

Activated alumina

Drying of gases, organic solvents, transformer oils Removal of HCl from hydrogen Removal of fluorine and boron–fluorine compounds in alkylation processes Removing of water and carbon dioxide from air in a PSA process

Silica gel

Drying of gases, refrigerants, organic solvents, transformer oils Desiccant in packings and double glazing Dew point control of natural gas

Polymeric adsorbents

Water purification, including removal of phenol, chlorophenols, ketones, alcohols, aromatics, aniline, indene, polynuclear aromatics, nitro- and chlor-aromatics, polychlorinated biphenyls (PCBs), pesticides, antibiotics, detergents, emulsifiers, wetting agents, kraftmill effluents, dyestuffs, and radionuclides Recovery and purification of steroids, amino acids and polypeptides Separation of fatty acids from water and toluene Separation of aromatics from aliphatics Separation of hydroquinone from monomers Recovery of proteins and enzymes Removal of colors from syrups Removal of organics from hydrogen peroxide

Clays (acid treated and pillared)

Removal of organic pigments Refining of mineral oils Removal of PCBs

(From Ref.[4].)

Sorbent Technology

mesoporous molecular sieves, sol–gel-derived metal oxide xerogels and aerogels, metal organic framework, p-complexation and composite adsorbents, new carbonaceous materials (carbon nanotubes, carbon fibers, superactivated carbons), high-temperature ceramic sorbents, and strong chemical sorbent materials. Although these new sorbent materials have demonstrated promising sorption properties for many existing and new applications, systematic studies on synthesis methods and characterization of these new materials are necessary to fully explore and realize their potential as commercial sorbents. The review that follows aims at attracting more research efforts to develop novel sorbent materials to meet the increasing needs of new energy, environmental, and other emerging technologies.

Oxide Molecular Sieves Microporous and mesoporous oxide molecular sieves that have the characteristics of large internal surface area and pore volume are ideal candidates for use as sorbent materials and catalyst supports of many heterogeneous catalysts. Oxide molecular sieves are generally synthesized by hydrothermal methods that involve both chemical and physical transformations within an amorphous oxide gel, often in the presence of a template species. The gel eventually converts to a crystalline material in which the template species and=or solvent molecules are guests within the channels and cages of an oxide host framework. A porous material is obtained upon removal of the guest molecules from the oxide framework. By manipulating the synthesis parameters, including starting precursors, synthesis temperature, pH, template species, drying, and calcination conditions, it is possible to tailor the pore size and shape of these porous materials for different applications. However, tailoring of porosity in oxide molecular sieves in terms of a priori structural design is extremely difficult because of the inherent complexity of the synthetic procedures employed.[21] Recent advances and applications of oxide molecular sieves have been summarized in several review articles.[2,21–23] Microporous zeolite materials synthesized with molecular templates and their applications in host–guest chemistry have been covered elsewhere.[13] A new class of silicate=aluminosilicate mesoporous molecular sieves designated as M41S was discovered in the former Mobil research laboratory by extending the concept of zeolite templating with small organic molecules to large long-chain surfactant molecules.[24] A representative member of this family is MCM-41, which has a honeycomb-shaped hexagonal arrange˚, ment of uniform mesopores in the range of 15–100 A specific surface area of 1040 m2=g, pore volume above 0.7 cm3=g, and significantly high sorption capacity for

2831

hydrocarbons (49 wt% for n-hexane at 40 torr and 21 C, and 67 wt% for benzene at 50 torr and 25 C).[24] Other significant members of the M41S family include MCM-48 (cubic phase), MCM-50 (stabilized lamellar phase), SBA-1 (cubic phase), and SBA-2 (cubic phase).[21] Although M41S type mesoporous oxide molecular sieves have exhibited unique properties of large surface area and exceptionally large pore volume (> 0.7 cm3=g), their large pore volume may not be attractive for gas sorption because the adsorbate–adsorbent interactions are not enhanced inside the internal pores of these materials.[2] Therefore, M14S type mesoporous oxide molecular sieves without surface modification are rarely used as sorbents. Significant research efforts were devoted to surface modification of M41S materials for different applications.[2] An amine-grafted MCM-48 sorbent, synthesized from tetraethoxysilane (TEOS), has been shown to have a surface area of 1389 m2=g, a silanol number of 8, higher thermal stability than MCM-41, high adsorption selectivity, and high capacity for both carbon dioxide and hydrogen sulfide.[25]

Sol–Gel-Derived Xerogels and Aerogels Sol–gel processing refers to the fabrication process of ceramic materials by preparation of a sol, gelation of the sol, and removal of the solvent.[26] Sols are dispersions of colloidal particles in a liquid solvent, and a gel is a solid matrix encapsulating a solvent. In a sol–gel process, the sol can be formed from a solution of colloidal powders or hydrolysis and condensation of alkoxides or salt precursors. In the latter approach, which is much more popular, primary particles of uniform size are formed and grow in a sol and connect to each other to form aggregates during gelation. These aggregates forming the network of the gel are broken apart into the primary particles in the drying step. Upon calcination and sintering, these primary particles are bound together strongly to form a very rigid solid network, and large interparticle space with uniform nanoscale pores is formed. Xerogels are obtained by drying the gels through evaporation at normal conditions under which capillary pressure causes shrinkage of the gel network, while areogels are produced by drying the wet gels at supercritical conditions where the liquid–vapor interface is eliminated, and relatively little shrinkage of the gel network occurs. Xerogels and aerogels typically have relatively large surface area, high porosity, and internal pore volume, and are ideal candidates as sorbent and catalyst support materials for many applications. The sol–gel process offers a very high flexibility to tailor xerogels and areogels for specific applications by manipulating the synthesis conditions.

S

2832

Sorbent Technology

Silica xerogel is probably the most studied and documented porous material in the sol–gel system.[27,28] Although silica has several crystalline forms, only amorphous silica gel is used as a desiccant (sorbent). A microporous silica that was synthesized with ˚, TEOS as precursor has an average pore size of 6.4 A 3 pore volume of 0.24 cm =g, and Brunauer–Emmett– Teller (BET) surface area of 588 m2=g.[29] However, this material lost about 90% of its microporosity when it was heated at 600 C for 30 hr. By doping with 1.5% of alumina, the thermal stability of this microporous silica was significantly improved.[29] Crystalline sorbent materials including g-alumina, zirconia, and titania were also synthesized using the sol–gel process in Lin’s group;[29] the representative pore size distribution and pore texture data of xerogels of g-alumina, zirconia, and titania are summarized in Fig. 3 and Table 3, respectively. As shown in Fig. 3, the pore size distributions of these materials are rather narrow, with an average pore diameter of about 3 nm. Such narrow size distribution and nanoscale average pore size are determined by the primary crystallite particles. The particles of the sol–gel-derived alumina, titania, and zirconia, owing to the Ostwald ripening mechanism,[26] are usually of nanoscale size; the uniform particle size distributions of g-alumina crystallites are plate-shaped with size ranging from about 5 to 20 nm. The sol–gel-derived g-alumina consists of such plate-shaped crystallite particles, which give rise to a relatively large surface area. Crystallites of tetragonal zirconia and rutile are of more spherical shape, with a crystallite size of about 15 and 11 nm, respectively.[29]

γ-Alumina Titania Zirconia

1.6

dV/dlog(D)

1.2

0.8

0.4

0.0 2

1

3

4

5

6 7 8 9

2

3

10 Pore Diameter (nm)

Fig. 3 Pore size distribution of sol–gel-derived alumina, zirconia, and titania. (From Ref.[29].)

Table 3 Pore texture sol–gel-derived alumina, zirconia, and titania (calcined at 450 C for 3 hr) Xerogel

Average ˚) pore size (A

Pore volume (cm3/g)

BET surface area (m2/g)

g-Al2O3

28

0.33

373

ZrO2

38

0.11

57

TiO2

34

0.21

147

One of the outstanding characteristics of sol– gel-derived g-alumina xerogel is its excellent mechanical properties. Preparation of porous g-alumina granules with good mechanical properties and desirable pore structure is of great importance in the development of novel catalysts and sorbents for various applications. The superior mechanical properties can be derived from the unique microstructure of the granule, which is defined by compacting small g-alumina crystallite particles bound together by the bridges of the same material formed through coarsening or sintering. Such nanostructured g-alumina can be prepared by combining the Yoldas process and the ‘‘oil-drop’’ method.[30–34] Table 4 compares the crush strength and attrition rate of sol–gel-derived g-alumina xerogel granules with those of several commercial sorbents. It is clearly shown in Table 4 that the sol–gel-derived galumina xerogel granules have excellent mechanical properties as compared with commercial sorbents. The excellent mechanical properties makes sol– gel-derived alumina granules very suitable for fluidized bed and other applications including separation and purification process for food and healthcare products that have very strict regulations on sorbent power contamination. Sol–gel-derived xeorgel sorbents have been investigated for gas separation, purification, and environmental applications. g-Alumina sorbents and membranes doped with cuprous and silver ions have been studied for selective adsorption or transfer of CO and ethylene through p-complexation.[35–37] Significant efforts have been devoted to explore the possibility of using CuO-doped g-alumina sorbents for removing SOx and NOx from flue gas.[38–44] The sol–gel-derived CuO=g-alumina sorbents have demonstrated high sorption capacity, high reactivity for SO2, and high thermal and chemical stability. The excellent mechanical and desulfurization properties of sol–gel-derived sorbents make them ideal sorbent candidates for fluidized bed desulfurization process. However, the relatively high cost of sol– gel-derived alumina xerogels may prevent them from being used in many large-scale adsorption processes. Research efforts are needed to look for less expensive precursors to replace alkaoxides used in the Yoldas process.

Sorbent Technology

2833

Table 4 Comparison of crush strength and attrition rate of sol–gel-derived g-alumina xerogel granules with commercial sorbents Sorbents Sol–gel alumina

Granular shape

Granular size (mm)

Average crush strength (N/granule)

Attrition rate (wt%/hr)

2.0–2.5

160

0.033

Spherical

Sol–gel alumina

Spherical

2.6–2.8

190

Alcoa alumina (LD-350)

Spherical

4.0–4.6

42

0.177

UOP silicalite

Cylindrical

1.4–1.6

16

0.575

Degussa DAY zeolite

Cylindrical

3.5–3.5

40

0.073

Sol–gel-derived metal oxide xerogels were also investigated for water adsorption because most of these metal oxides are good sorbent candidates for desiccant applications.[45–47] Significant research works have been carried out to study the adsorption= complexation properties of heavy metal ions including mercury, Cu(II), CdCl2, etc. in waste water on different sol–gel-derived xerogels.[48–54] The sol–gel-derived xerogels seem to be promising sorbent candidates for waste water treatment. Modified xerogel sorbents also showed promising adsorption properties for removing acid gas CO2 and H2S from natural gas, or as CO2 storage sorbents.[25,55] There are several advantages of using xerogels for enzyme immobilization, including the opportunity to produce them in defined shapes or thin films and the ability to manipulate their physical characteristics including porosity, hydrophobicity, and optical properties.[56,57] Metal oxide composite xerogels can also adsorb methyl orange.[58] There are also reports on microporous and mesoporous carbon xerogels for gas separation and purification.[59,60] As compared with xerogels, aerogels have larger surface area, larger pore volume, and higher porosity.[61–64] Alumina aerogels with a specific surface area as high as 1000 m2=g, and a pore volume as high as 17.3 cm3=g have been synthesized by supercritical carbon dioxide drying, but a very limited information on their adsorption properties was found.[61,62] A super water adsorbent consisting of 17–30% of CaCl2 doped on SiO2 aerogel showed an effective reversible adsorption capacity of 100 wt%; the adsorption capacity of hydrophilic silica aerogels can be fully recovered after regeneration.[64–66] CaO- and MgO-modified SiO2 aerogel sorbents can be used to capture pollution gases including CO2, SO2, CO, and NOx emitted from power plants based on fossil fuels.[67] Several studies reported the use of aerogels as destructive sorbents for toxic gases and radionuclide removal from contaminated environments.[64,68–70] Carbon aerogels can also be made from carbon materials under supercritical carbon dioxide drying conditions; these carbon aerogels were studied for removing uranium and other inorganic ions from contaminated water.[71–73] Aerogels are special sorbent candidates with excellent pore texture, which may play a

major role in environmental protection. However, more studies on their synthesis and adsorption properties are needed.

Metal Organic Framework (MOF) Recently, Yaghi’s group reported a novel crystalline nanoporous material that consists of metal atoms occupying the vertices of a lattice, with the lattice size, porosity, and chemical environment defined by the organic linker molecules that bind the metal atoms into a robust periodic structure.[74–76] These so-called metal organic framework (MOF) materials have been demonstrated to have an exceptionally high specific surface area of 4526 m2=g, and find use as adsorbents for methane and as hydrogen storage materials.[74,77–80] A reticular synthesis method was developed to realize the bottom-up synthesis through top-down design logic by using inorganic, metal organic, and organic molecules to build frameworks and large molecules.[81] Well-defined molecular building blocks that will maintain their structural integrity throughout the construction process were used to build the MOF molecules. It allows remarkable control over composition and structure of the material formed and employs the full range of the molecular synthetic methods and compounds in the preparation of this new type of porous sorbent materials. The ability to molecularly engineer the lattice size, chemical environment, and possibly structure by careful choice of the metal centers and organic linkers offers the opportunity for the development of new types of sorbents that could potentially meet the Department of Energy (DOE) target for hydrogen storage and that can be used for other applications in separation and purification. It is reported that metal organic framework-5 (MOF-5) of composition Zn4O(BDC)3 (BDC: 1,4benzenedicarboxylate) with a cubic three-dimensional extended porous structure and octahedral Zn–O–C clusters with benzene links can adsorb hydrogen up to 4.5 wt% at 78 K, and 1.0 wt% at room temperature and pressure of 20 bar.[74,79] It is identified by inelastic neutron scattering spectroscopy of the rotational

S

2834

transitions of the adsorbed hydrogen molecules that zinc and the BDC linker in MOF-5 are the two hydrogen binding sites responsible for hydrogen adsorption on this material. Higher hydrogen adsorption capacity at ambient temperature and 10 bar were observed on similar isoreticular metal organic framework-6 and -8 (IRMOF-6 and -8) having cyclobutylbenzene and naphthalene linkers.[79] A different microporous MOF sorbent [microporous metal coordination materials (MMOM)] was reported to have hydrogen sorption capacities (1.0 wt% at room temperature and 48 bar) similar to those of the best single-wall carbon nanotubes.[80] The adsorbed hydrogen can be released when the gas pressure is reduced. MOF sorbents have also been investigated for methane adsorption.[77] The reported methane storage capacity of MOF-6 is 155 cm3 (STP)=cm3 at 298 K and 36 atm, which is significantly higher than that of zeolite 5A (87 cm3 (STP)=cm3) and other coordination framework (213 cm3 (STP)=cm3).[77] Adsorption and desorption of carbon dioxide, nitrogen, and argon on a microporous manganese-based MOF sorbent has been reported.[78] Another interesting porous MOF sorbent, Cu-BTC (polymeric copper(II) benzene-1,3,5tricarboxylate) with molecular sieve character, was studied for its sorption properties of various adsorbates including nitrogen, oxygen, carbon monoxide, carbon dioxide, nitrous oxide, methane, ethylene, ethane, and n-dodecane.[82,83] A detailed investigation of sorption thermodynamics was performed for carbon dioxide by a sorption-isosteric method. It was demonstrated that Cu-BTC sorbent can be used for the separation of carbon dioxide–carbon monoxide, carbon dioxide– methane, and ethylene–ethane mixtures. In addition, this sorbent can also be used to remove carbon dioxide, nitrous oxide, high molecular weight hydrocarbons, and moisture from ambient air before cryogenic separation to produce oxygen and nitrogen.[82]

Sorbent Technology

Table 5 USDOE FreedomCAR hydrogen storage system targets Year Target factor

2005

2010

2015

Specific energy (MJ=kg)

5.4

7.2

10.8

Hydrogen (wt%)

4.5

6.0

9.0

Energy density (MJ=L)

4.3

5.4

9.72

System cost ($=kg=system)

9

6

3

Operating temperature ( C) Cycle life-time (adsorption= desorption cycles)

20=50

500

20=50

1000

20=50

1500

Flow rate (g=sec)

3

4

5

Delivery pressure (bar)

2.5

2.5

2.5

Transient response (sec)

0.5

0.5

0.5

Refueling rate (kg H2=min)

0.5

1.5

2.0

(From Ref.[84].)

Hydrogen can be stored both physically and chemically in a confined vessel with or without the assistance of a storage media. The most commonly used methods for hydrogen storage are: gaseous and liquid hydrogen storage, solid state storage in complex metal hydrides, chemical storage materials, and in nanostructured materials.[2,85] The representative hydrogen storage capacities, hydrogen storage, and release conditions in various materials are summarized in Table 6. Carbon nanotubes are probably the most investigated and documented hydrogen storage sorbent materials. Several excellent reviews on carbon nanotubes for hydrogen storage are available.[2,86] As shown in Table 6, the hydrogen storage capacities on representative carbon nanotubes are below 6 wt%, the most referred DOE target for 2010.[84,87,88] The following concerns about carbon nanotubes as hydrogen storage materials have driven research in this area to other directions:[85]

Hydrogen Storage Media The development of hydrogen-fueled transportation system and portable electronics will demand new materials that can store large amounts of hydrogen at ambient temperature and relatively low pressures with small volume, light weight, fast charging and discharging time, cyclic stability, and low cost. Table 5 summarizes the targets for hydrogen storage system for automotive applications set by USDOE. The hydrogen storage capacities are calculated as both weight and volume percentage of the storage system.[84] To achieve these goals, the hydrogen storage media (sorbent) should have a high reversible hydrogen sorption capacity, low weight and high packing density as well as fast sorption=desorption kinetics, and low cost.

1. Difficult to meet the DOE’s long-term target (9 wt%); 2. Mechanisms for hydrogen sorption in carbon nanotubes are not well understood; 3. Part of the adsorbed hydrogen can only be recovered at high temperatures; 4. Preparation and purification of carbon nanotubes involve complicated and expensive processes, which leads to high cost of carbon nanotubes; 5. Hydrogen storage capacity is quite sensitive to sorbent preparation conditions; 6. Mixed results on hydrogen adsorption capacity have been reported.

Sorbent Technology

2835

Table 6 Summary of hydrogen storage capacity of various nanostructured materials Materials

H2 storage capacity (wt%)

H2 storage conditions

H2 release conditions

Reference

Carbon nanotubes Single-walled

4.2

10 MPa, 300 K

1 bar, 300 K

[87]

Multi-walled

3.6

7 MPa, 298 K

1 bar, 298 K

[88]

BN

4.2

108 bar, 298 K

1 bar, 298 K

[89]

TiS2

2.5

40 bar, 298 K

1 bar, 298 K

[90]

Non-carbonaceous nanotubes

Microporous MOF 4.5

0.75 bar, 77 K

1 bar, > 77 K

[74]

1.0

48 bar, 298 K

1 bar, 298 K

[80]

Mg2NiH4

3.6

1 bar, > 528 K

[91]

NaAlH4

8.0

1 bar, 500 K

Mg(AlH4)2

6.6

1 bar, > 436 K

[93]

13.3

1 bar, > 473 K

[94]

MOF-5 MMOM Metal hydrides

LiBH4

90 bar, 403 K

[92]

Nitrides 9.3

1 bar, 443–473 K

1 bar, > 700 K

[95]

H2(H2O)2

5.3

1 bar, > 77 K

[97]

H2(H2O)

10.0

1 bar, 77 K

< 6000 bar, > 190 K

[97]

(H2)4(CH4)

33.3

2000 bar, 77 K

< 2000 bar, > 77 K

[97]

Li3N Clathrate=molecular compounds

Noncarbonaceous nanotubes including boron nitride (BN) and titanium sulfide (TiS2) have been prepared and studied for hydrogen sorption.[89,90] Hydrogen storage capacity (2.5–4.5 wt%) similar to those for carbon nanotubes have been obtained on these noncarbonaceous materials. MOF-based sorbents for hydrogen sorption was discussed in the previous section. As suggested in Table 6, the hydrogen sorption capacities on MOF-5 and MMOM are lower than those on carbon nanotubes. However, MOF sorbents look more promising than carbon nanotubes as hydrogen storage media for the following reasons: 1. MOF is easy to make and is less expensive; 2. Sorption sites for hydrogen on MOF are better defined; 3. MOF sorbents may have extremely high specific surface area (> 4000 m2=g); 4. It is possible to tailor the interaction between hydrogen and MOF by manipulating synthesis parameters including different building blocks. Metal hydrides were widely investigated for hydrogen storage, and are believed to be ideal hydrogen

6000 bar, 190 K

storage system because they have the following characteristics:[2,84,91–94] 1. Relatively high hydrogen storage capacity at modest pressures as indicated in Table 6; 2. Fast hydrogen charging and discharging rates; and 3. Moderate temperature for hydrogen desorption. However, metal hydrides also suffer from the following disadvantages as hydrogen storage materials: 1. High sensitivity to impurities in hydrogen (CO, H2O, O2, CO2, and H2S); 2. Storage capacity and rates decay with hydrogen charge–discharge cycles; and 3. Relatively high cost as compared with gaseous and liquid hydrogen storage methods. Another interesting hydrogen storage material is lithium nitride (Li3N), which shows 9.3 wt% useful hydrogen storage capacity between thermal swing cycles (473–700 K).[95] The requirement for high-temperature desorption will greatly limit its applications. Most recently, hydrogen clathrate hydrate and other

S

2836

Sorbent Technology

molecular compounds were found to have hydrogen storage capacities as high as 33.3 wt%.[96–98] This is a very innovative way to store hydrogen with exceptionally high capacity to meet the DOE long-term target. However, these clathrate and hydrogen storage compounds were synthesized at extremely high pressures and at liquid nitrogen temperature. It is unlikely these clathrate hydrates will be used for hydrogen storage until we find new clathrate hydrate compounds that can be synthesized and are stable at much lower pressures.

p-Complexation Sorbents and Composite Sorbents A very good review article based on a panel study of status, future research needs, and opportunities for porous sorbent materials was published several years ago.[99] It was pointed out that very significant advances have been made in tailoring the porosity of porous sorbent materials in terms of size and shape selectivity. Relatively little progress has been achieved in terms of chemoselectivity of sorbents based on specific interactions between adsorbate molecules and functional groups in the sorbents. Incorporation of active sites into sorbents is of high priority in the development of sorbents. The p-complexation bond is a weak chemical bond that is slightly stronger than van der Waals interaction, which governs physical sorption processes. Sorbents with p-complexation capability tend to have higher selectivity than other physical sorbents for certain adsorbate molecules. Several different types of p-complexation sorbents with Cuþ or Agþ ions supported on different supports (SiO2, g-Al2O3, TiO2, variety of zeolites, polymer resin, and activated carbon) were synthesized using different methods including thermal dispersion, wet-impregnation, sol–gel, microwave heating, ionexchange zeolite, and ion-exchange resin.[34–36,99–105] It was found that the CO adsorption capacity increases with Cuþ loading in an activated alumina supported sorbent.[100,101] To achieve the highest sorption capacity, the active species should be dispersed as a monolayer form.[99] The potential applications of these p-complexation sorbents include:[2] 1. 2. 3. 4. 5. 6.

Desulfurization of gasoline and diesel fuels; Separation of olefins and paraffins; CO separation from synthesis gases; CO removal from hydrogen; Removal of aromatics; and Removal of volatile organic compounds (VOCs).

A p-complexation sorbent can also be viewed as a composite sorbent especially when the sorbent support

contributes significantly to the adsorption. Composite sorbents are typically made by physically mixing the powders of constituent sorbents with different sorption properties; they tend to have multiple sorption sites for different adsorbate molecules. One example of a composite sorbent is a mixture of activated alumina and zeolites for removing moisture, carbon dioxide, and other trace components from air in an air-purification process prior to cryogenic air separation.[106–108] Conventionally, moisture is removed by activated alumina, carbon dioxide by zeolite 13X, and hydrocarbons by zeolite 5A.[107,108] Traditional air-purification processes employ multiple layers consisting of activated alumina, zeolite 13X, and optional zeolite 5A sorbents in a single vessel to achieve significant removal of moisture, carbon dioxide, and hydrocarbons from air. The major disadvantages of layered bed are nonuniform sorbent packing for a short sorbent layer, very significant temperature variation (> 30 C, sometimes called cold spots) between the zeolite and the activated alumina sorbent layers. The large temperature difference could upset the sorption process operation if it is designed to be operated isothermally. It is beneficial to have a single sorbent with multiple sorption features for different impurities and eliminate sorbent layering and temperature variations.

High-Temperature Ceramic O2 Sorbents Lin et al. disclosed in a U.S. patent a new group of sorbents for air separation and oxygen removal using oxygen-deficient perovskite-type ceramics as sorbents.[109] Perovskite-type ceramics are a group of metal oxides having the general formula of ABO3. The ideal perovskite structure for ABO3 is shown in Fig. 4. It consists of cubic array of corner-sharing BO6 octahedra, where B is a transition metal ion. The A-site ion, interstitial between the BO6 octahedra, may be occupied by an alkali, an alkaline earth, or a rare earth ion. Alternatively, the perovskite structure may be regarded as a cubic close packing of layers of AO3 with B cations placed in the interlayer octahedral interstices.[110] This group of the sorbents can be viewed as chemisorbents that can selectively adsorb a considerable amount of oxygen at high temperatures (> 300 C), and theoretically has an infinitely high selectivity for oxygen over nitrogen or other nonoxygen species. The presence of other gases has negligible effect on the separation properties of these new sorbents. High-temperature membrane separation of oxygen has also received increasing interest from other industrial gas companies.[111–113] Development of high-temperature oxygen separation technology opens up several high-temperature applications of oxygen including syngas production, hydrogen production, and partial oxidation fuel reforming processes.

Sorbent Technology

2837

A A

B

0.5

S

0.45

δ

0.4 0.35 LSCF-1 0.3 0.25 0.0001

The oxygen equilibrium and kinetic properties of perovskite-type ceramics have been extensively studied primarily for applications as fuel cell electrodes and oxygen permeable membranes,[110] and only a few for oxygen sorption.[114–117] Oxygen nonstoichiometry (d) occurs in some perovskite-type ceramics with B-site cations of variable oxidation states and A-site cations partially substituted by another cation with a lower oxidation state. Oxygen nonstoichiometry, or oxygen content, for a perovskite-type ceramic of a given composition is a function of temperature and oxygen partial pressure. Therefore, by changing temperature or oxygen partial pressure, the value of oxygen nonstoichiometry or the degree of oxygen vacancy in the material changes. Within a certain range of temperature and oxygen partial pressure the change of the oxygen nonstoichiometry does not affect the perovskite structure, and the change of the oxygen content in the material is a reversible process. The oxygen nonstoichiometry of the perovskite sorbents can be measured gravimetrically at different temperatures and oxygen partial pressures. Oxygen sorption capacity on the sorbent can then be calculated from the oxygen nonstoichiometry data once the initial state (zero sorption capacity) of the sorbent material is defined.[114] Figs. 5 and 6 are examples of oxygen nonstoichiometry of La1–xSrxCo1–yFeyO3–d perovskite oxide sorbents as a function of oxygen partial pressure or temperature, respectively.[114] The corresponding oxygen sorption isotherm of La1–xSrxCo1–yFeyO3–d perovskite oxide sorbents that were calculated from the oxygen nonstoichiometry data are shown in Fig. 7.[114] From these oxygen isotherms we can conceive a high-temperature vacuum swing sorption or temperature swing sorption process for oxygen separation or oxygen removing applications by using the La1–xSrxCo1–yFeyO3–d perovskite oxide sorbents. Future studies on

0.01 PO2 (atm)

0.1

1

B 0.5 1

2

3

0.45

δ

Fig. 4 Ideal perovskite structure for ABO3 type oxides.

0.001

0.4

0.35

0.3 0.0001

LSCF-2

0.001

0.01 PO2 (atm)

0.1

1

Fig. 5 Change of oxygen nonstoichiometry d with oxygen partial pressure (LSCF-1. La0.1Sr0.9Co0.5Fe0.5O3–d; LSCF-2, La0.1Sr0.9Co0.9Fe0.1O3–d). (From Ref.[114].)

perovskite oxide sorbents are needed to address the issues of slow desorption rate, potential sorbent structure stability in cyclic processes, and effective regeneration methods. High-Temperature CO2 Sorbents Increased awareness of the global warming trend has led to worldwide concerns regarding ‘‘greenhouse gas’’ emissions. Greenhouse gases include CO2, CH4, and N2O and are mostly associated with the production and utilization of fossil fuels, with CO2 being the single greatest contributor to global warming. Significant research efforts are being devoted worldwide on looking for economical ways of mitigating CO2 emission problem.[118–122] Carbon capture and sequestration costs can be considered in terms of four components: capture, compression, transport, and injection. Typically about 75% of this cost is attributable to capture

2838

Sorbent Technology

0.45 0.4

PO2 = 0.209 atm

δ

0.35 0.3 LSCF-2 LSCF-1

0.25 0.2 300

400

500 T(˚C)

600

700

800

Fig. 6 Change of oxygen nonstoichiometry d with temperature (LSCF-1, La0.1Sr0.9Co0.5Fe0.5O3–d; LSCF-2, La0.1Sr0.9 Co0.9Fe0.1O3–d). (From Ref.[114].)

A Amount Adsorbed (mmol/g)

0.5 0.4

500˚C 600˚C

0.3 0.2 0.1 0 0.0001

0.001

0.01 PO2 (atm)

0.1

1

B Amount Adsorbed (mmol/g)

0.6 0.5

500˚C 600˚C

Li2 ZrO3 þ CO2 ! Li2 CO3 þ ZrO2

0.4 0.3 0.2 0.1 0 0.001

and compression processes. Sorption of carbon dioxide on solid sorbents is receiving increased attention in view of the importance of both the removal and the recovery of carbon dioxide from flue gases.[123,124] Physical sorbents for carbon dioxide separation and removal were extensively studied by industrial gas companies.[125–127] Zeolite 13X, activated alumina, and their improved versions are typically used for removing carbon dioxide and moisture from air in either a TSA or a PSA process.[125–128] The sorption temperatures for these applications are usually close to ambient temperature. There are a few studies on adsorption of carbon dioxide at high temperatures. The carbon dioxide adsorption isotherms on two commercial sorbents hydrotalcite-like compounds, EXM911 and activated alumina made by LaRoche Industries, are displayed in Fig. 8.[123,124] As shown in Fig. 8, LaRoche activated alumina has a higher carbon dioxide capacity than the EXM911 at 300 C. However, the adsorption capacities on both sorbents are too low for any practical applications in carbon dioxide sorption at high temperature. Conventional physical sorbents are basically not effective for carbon dioxide capture at flue gas temperature (> 400 C). There is a need to develop effective sorbents that can adsorb carbon dioxide at flue gas temperature to significantly reduce the gas volume to be treated for carbon sequestration. Only a handful of studies on high-temperature carbon dioxide sorbents have been published in the past few years.[123,124,129–133] It is believed that lithium zirconate (Li2ZrO3) is one of the most promising sorbent materials for carbon dioxide separation from flue gas at high temperature because it can absorb a large amount of carbon dioxide at around 400–700 C.[130,131] The carbon dioxide adsorption and desorption uptake curves on lithium zirconate are shown in Fig. 9.[131] As shown in this figure, about 20% carbon dioxide was captured by the lithium zirconate sorbent during sorption step at 500 C based on the following reaction:

0.01

0.1

1

PO2 (atm) Fig. 7 Sorption isotherms of: (A) La0.1Sr0.9Co0.5Fe0.5O3–d and (B) La0.1Sr0.9Co0.9Fe0.1O3–d at 500 and 600 C. (From Ref.[114].)

ð4Þ

About 80% of adsorbed carbon dioxide can be desorbed with hot air 780 C. Addition of potassium carbonate (K2CO3) and Li2CO3 into Li2ZrO3 remarkably improves the CO2 sorption rate of the Li2ZrO3-based sorbent materials. X-ray diffraction (XRD) analysis for phase and structural changes during the sorption= desorption process shows that the reaction between Li2ZrO3 and CO2 is reversible.[131] Based on this work, a TSA process can be developed for carbon dioxide removal from flue gas using Li2ZrO3-type sorbent materials. High-temperature carbon dioxide sorbents can also find applications in fuel reforming process to enhance fuel to hydrogen conversion efficiency. It was reported

Sorbent Technology

2839

A 0.55

S

0.50 EXM911

0.45

20˚C

Q (mmol/g)

0.40 0.35 0.30 0.25 0.20 0.15

200˚C

0.10 0.05 0.00 0.0

300˚C 0.1

0.2

0.3

0.4 0.5 0.6 Pressure (bar)

0.7

0.8

0.9

1.0

B 1.1 1.0 Activated Alumina

0.9

20˚C

0.8 0.7 200˚C

0.6 0.5 0.4 0.3

300˚C

0.2 0.1 0.0 0.0

0.1

0.2

0.3

0.4 0.5 0.6 Pressure (bar)

0.7

that sorption of carbon dioxide can enhance the production of hydrogen for a steam–methane reforming process using a mixture of Ni-based reforming catalyst and a Ca-based sorbent. The rates of the reforming, water-gas shift, and carbon dioxide removal reactions are sufficiently fast that combined reaction equilibrium was closely approached, allowing for >95 mol% hydrogen to be produced in a single step.[134]

CONCLUSIONS AND FUTURE DIRECTIONS Existing commercial sorbents including activated carbon, zeolites, activated alumina, and silica gels will continue to play important roles in adsorptive separation and purification for current process industries in the near future. However, they cannot meet the needs

0.8

0.9

1.0

Fig. 8 Adsorption isotherms of carbon dioxide on commercial sorbents. (A) Hydrotalcite-like compound, EXM911; (B) LaRoche Industries activated alumina at 20, 200, and 300 C. (From Refs.[123,124].)

of future technological developments in the new energy economy and the stringent environmental regulations. The newly developed nanostructured sorbent materials have shown some very promising features, but they are basically unexplored and systematic investigations are needed on both synthesis methods and adsorption characteristic studies. The following are the author’s views on future research needs in both sorbent synthesis and applications: 1. Explore entirely new sorbent synthesis routes to better control of both sorbent pore texture and surface property. 2. Design new sorbent materials from basic building blocks and introduce active sorption sites according to sorbent–adsorbate interaction requirements. MOF material syntheses using

2840

Sorbent Technology

REFERENCES

800

10

600

Temperature ( C)

Weight change (%)

20

0

0

200

400

600

400 800

Time (min) Fig. 9 CO2 sorption and regeneration on the modified Li2ZrO3. Sorption process: 50% CO2 balanced by dry air at 500
C. Desorption process: 50% CO2 balanced by dry air at 780
C ! dry air at 780
C. Gas flow rate: 150 ml=min. (From Ref.[131].)

the isoreticular method and sol–gel technique are two examples of this approach. 8. A better understanding of the relationship between sorbent–adsorbate interaction, sorption equilibrium, and kinetics through molecular simulation, and provide guidance for sorbent synthesis. In terms of applications, new sorbents should be developed to meet the following pressing needs: 1. Deep desulfurization of fossil fuels for fuel cell application. 2. Hydrogen purification (H2S, CO, and CO2 removal). 3. Hydrogen and methane storage sorbents and processes. 4. Water treatment (arsenic, radionuclides and heavy metal ions and anions removal). 5. Air pollution control (SOx, NOx, and other toxic gases removal). 6. Chemisorbents as effective getter materials for toxic process gas and liquid streams. 7. Effective high-temperature carbon dioxide sorbents for carbon dioxide sequestration. ACKNOWLEDGMENT Professor Y.S. Lin is acknowledged for providing his publications and comments on high-temperature sorbents discussed in this entry.

1. King, C.J. Separation Processes, 2nd Ed.; McGraw-Hill: New York, 1980. 2. Yang, R.T. Adsorbents: Fundamentals and Applications; John Wiley & Sons: Hoboken, NJ, 2003. 3. Yang, R.T. Gas Separation by Adsorption Processes; Butterworths: Boston, MA, 1987; 1–48. 4. Thomas, W.J.; Crittenden, B. Adsorption Technology & Design; Butterworth Heinemann: Oxford, 1998; 8–30. 5. Karger, J.; Ruthven, D.M. Diffusion in Zeolites and Other Microporous Solids; John Wiley & Sons: New York, 1992; 416–418. 6. Ruthven, D.M. Principles of Adsorption and Adsorption Processes; John Wiley & Sons: New York, 1984; 396–405. 7. Myers, A.L.; Prausnitz, J.M. Thermodynamics of mixed-gas adsorption. AIChE J. 1965, 11 (1), 121–127. 8. Ruthven, D.M.; Farooq, S.; Knaebel, K.S. Pressure Swing Adsorption; VCH Publishers, Inc.: New York, 1994. 9. Patrick, J.W., Ed. Porosity in Carbons: Characterization and Applications; Halsted Press, an imprint of John Wiley and Sons: London, 1995. 10. Bansal, R.C.; Donnet, J.-B.; Stockli, F. Activated Carbon; Marcel Dekker, Inc.: New York, 1988. 11. Breck, D.W. Zeolite Molecular Sieve; Krieger Publishing Company: Marlbar, FL, 1974. 12. Barrer, R.M. Zeolites and Clay Minerals as Sorbents and Molecular Sieves; Academic Press: New York, 1978. 13. Auerbach, S.M.; Carrado, K.A.; Dutta, P.K., Eds. Handbook of Zeolite Science and Technology; Marcel Dekker, Inc.: New York, 2003. 14. Kumar, R Removal of Water and Carbon Dioxide from Atmospheric Air. U.S. Patent 4,711,645, Dec 8, 1987. 15. Jain, R Pre-purification of Air for Separation. U.S. Patent 5,232,474, Aug 3, 1993. 16. Jain, R.; Tseng, J.K Method and Apparatus for Producing Clean Dry Air having Application to Air Separation. U.S. Patent 6,077,488, Jun 20, 2000. 17. Deng, S.; Kumar, R.; Jain, R Air Purification Process with Thermal Regeneration. U.S. Patent 5,931,022, Aug 3, 1999. 18. Iler, R.K. The Chemistry of Silica; John Wiley & Sons: New York, 1979. 19. Unger, K.K. Porous Silica, Its Properties and Use as a Support in Column Liquid Chromatography; Elsevier: Amsterdam, The Netherlands, 1979.

Sorbent Technology

20. Vansant, E.F.; Van Der Voort, P.; Vrancken, K.V. Characterization and Chemical Modification of the Silica Surface; Elsevier: Amsterdam, The Netherlands, 1995. 21. Barton, T.J.; Bull, L.M.; Klemperer, W.G.; Loy, D.A.; McEnaney, B.; Misono, M.; Monson, P.A.; Pez, G.; Scherer, G.W.; Vartuli, J.C.; Yaghi, O.M. Tailored porous materials. Chem. Mater. 1999, 11 (10), 2633–2656. 22. Zhao, X.S.; Lu, G.Q. (Max) and Millar, G.J. Advances in mesoporous molecular sieve MCM41. Ind. Eng. Chem. Res. 1996, 35, 2075–2090. 23. Guliants, V.V.; Carreon, M.A.; Lin, Y.S. Ordered mesoporous and macroporous inorganic films and membranes. J. Membr. Sci. 2004, 235 (1=2), 53–72. 24. Beck, J.S.; Vartuli, J.C.; Roth, W.J.; Leonowicz, M.E.; Kresge, C.T.; Schmitt, K.D.; Chu, C.T.-W.; Olson, D.H.; Sheppard, E.W.; McCullen, S.B.; Higgins, J.B.; Schlenkert, J.L. A new family of mesoporous molecular sieves prepared with liquid crystal templates. J. Am. Chem. Soc. 1992, 114, 10834–10843. 25. Huang, H.Y.; Yang, R.T.; Chinn, D.; Munson, C.L. Amine-grafted MCM-48 and silica xerogel as superior sorbents for acidic gas removal from natural gas. Ind. Eng. Chem. Res. 2003, 42 (12), 2427–2433. 26. Brinker, C.J.; Scherer, G.W. Sol–Gel Science; Academic Press, Inc.: San Diego, CA, 1990. 27. Iler, R.K. The Chemistry of Silica; John Wiley & Sons: New York, 1979. 28. Hench, L.L. Sol–Gel Silica; Noyes Publications: Westwood, NJ, 1998. 29. Lin, Y.S.; Deng, S. Sol–gel preparation of nanostructured adsorbents. In Adsorption and Its Application in Industry and Environmental Protection; Dabrowski, A., Ed.; Elsevier: Amsterdam, 1998; vol.120A, 653–686. 30. Yoldas, B.E. Transparent porous alumina. Am. Ceram. Soc. Bull. 1975, 54, 286–288. 31. Deng, S.G. Synthesis and Properties of Nanostructured Adsorbents for Gas Separation and Environmental Applications, Ph.D. dissertation, University of Cincinnati, Cincinnati, OH, 1996. 32. Deng, S.G.; Lin, Y.S. Granulation of sol–gel derived nano-scale alumina. AIChE J. 1997, 43 (2), 505–514. 33. Buelna, G.; Lin, Y.S. Sol–gel derived nanoporous g-alumina granules. Microporous Mesoporous Mater. 1999, 30, 359–369. 34. Buelna, G.; Lin, Y.S. Preparation of spherical alumina and copper oxide coated alumina sorbents by improved sol–gel granulation process. Microporous Mesoporous Mater. 2001, 42, 67–76.

2841

35. Deng, S.; Lin, Y.S. Sol–gel preparation and properties of alumina adsorbents for gas separation. AIChE J. 1995, 41 (3), 559–570. 36. Wang, Y.; Lin, Y.S. Sol–gel synthesis and gas adsorption properties of CuCl modified mesoporous alumina. J. Sol–Gel Sci. Technol. 1998, 1, 185–195. 37. Lin, Y.S.; Ji, W.; Wang, Y.; Higgins, R.J. Cuprous chloride modified nanoporous alumina membranes for ethylene–ethane separation. Ind. Eng. Chem. Res. 1999, 38, 2292–2298. 38. Deng, S.; Lin, Y.S. Sulfation and regeneration of sol-gel derived regenerative sorbents for flue gas desulfurization. In AIChE Symp.; Cohen, Y., Peters, R.W., Eds.; 1995; 91 (309), 32–39. 39. Deng, S.G.; Lin, Y.S. Synthesis, stability and sulfation properties of sol–gel derived regenerative sorbents for flue gas desulfurization. Ind. Eng. Chem. Res. 1996, 35 (4), 1429–1437. 40. Lin, Y.S.; Deng, S.G. Removal of trace sulfur dioxide from a gas stream by regenerative sorption processes. Sep. Purif. Technol. 1998, 13 (4), 65–77. 41. Wang, Z.-M.; Lin, Y.S. Sol–gel derived alumina alumina supported copper oxide sorbent for flue gas desulfurization. Ind. Eng. Chem. Res. 1998, 37, 4675–4681. 42. Yang, Z.; Lin, Y.S. Sol–gel synthesis of silicalite=g-alumina granules. Ind. Eng. Chem. Res. 2000, 39, 4944–4948. 43. Buelna, G.; Lin, Y.S.; Liu, L.; Litster, J.D. Structural and mechanical properties of nanostructured granular alumina catalysts. Ind. Eng. Chem. Res. 2003, 42, 442–447. 44. Buelna, G.; Lin, Y.S. Combined removal of SO2 and NO using sol–gel derived copper oxide coated alumina sorbents=catalysts. Environ. Technol. 2003, 24 (9), 1087–1095. 45. Gordeeva, L.G.; Mrowiec-Bialon, J.; Jarzebski, A.B.; Lachowski, A.I.; Malinowski, J.; Aristov, Y.I. Selective water sorbents for multiple applications, 8. Sorption properties of CaCl2–SiO2 sol–gel composites. React. Kinet. Catal. Lett. 1999, 66 (1), 113–120. 46. Rogojevic, S.; Jain, A.; Gill, W.N.; Plawsky, J. Moisture adsorption in nanoporous silica xerogels. Electrochem. Solid State Lett. 2002, 5 (7), F22–F23. 47. Kittaka, S.; Uchida, N.; Kihara, T.; Suetsugi, T.; Sasaki, T. Interlayer water-molecules in vanadium pentoxide hydrate. 2. Effect of intercalated metal-ions on the adsorbability of water molecules.. Langmuir 1992, 8 (1), 45–48. 48. Vaghetti, J.C.P.; Zat, M.; Bentes, K.R.S.; Ferreira, S.; Benvenutti, E.V.; Lima, E.C. 4-Phenylenediaminepropyl silica xerogel as a sorbent

S

2842

49.

50.

51.

52.

53.

54.

55.

56.

57.

58.

59.

Sorbent Technology

for copper determination in waters by slurrysampling ETAAS. Anal. At. Spectrom. 2003, 8 (4), 376–380. Burleigh, M.C.; Dai, S.; Barnes, C.E.; Xue, Z.L. Enhanced ionic recognition by a functionalized mesoporous sol–gel: synthesis and metal ion selectivity of diaminoethane derivative. Sep. Sci. Technol. 2001, 36 (15), 3395–3409. Pavan, F.A.; Lucho, A.M.S.; Goncalves, R.S.; Costa, T.M.H.; Benvenutti, E.V. Anilinepropylsilica xerogel used as a selective Cu(II) adsorbent in aqueous solution. J. Colloid Interface Sci. 2003, 263 (2), 688–691. Khan, A.; Ahmad, S.; Zaidi, S.A.R.; Mahmood, F.; Khokhar, M.Y. Removal of mercury by 1naphthylthiocarbamide doped xerogel using radiotracer technique. Sep. Sci. Technol. 2002, 37 (13), 3099–3107. Arenas, L.T.; Vaghetti, J.C.P.; Moro, C.C.; Lima, E.C.; Benvenutti, E.V.; Costa, T.M.H. Dabco=silica sol–gel hybrid material. The influence of the morphology on the CdCl2 adsorption capacity. Mater. Lett. 2004, 5 (6), 895–898. Pavan, F.A.; Costa, T.M.H.; Benvenutti, E.V. Adsorption of CoCl2, ZnCl2 and CdCl2 on aniline=silica hybrid material obtained by sol–gel method. Colloids Surf. A—Physicochem. Eng. Aspects 2003, 226 (1–3), 95–100. Pavan, F.A.; Lima, I.S.; Benvenutti, E.V.; Gushikem, Y.; Airoldi, C. Hybrid aniline=silica xerogel cation adsorption and thermodynamics of interaction. J. Colloid Interface Sci. 2004, 275 (2), 386–391. Deshpande, R.S.; Sharp-Goldman, S.L.; Bocarsly, A.B. Thermodynamics and kinetics of CO2 adsorption on dehydrated palladium= cobalt-based cyanogels: a highly selective, fully reversible system for CO2 storage. Langmuir 2002, 18 (20), 7694–7698. Aucoin, M.G.; Erhardt, F.A.; Legge, R.L. Hyperactivation of Rhizomucor miehei lipase by hydrophobic xerogels. Biotechnol. Bioeng. 2004, 85 (6), 647–655. Hayashi, J.; Watada, Y.; Muroyama, K. Preparation of mesoporous material having a hydrophobic surface by combining silica xerogel with resin using sol–gel method. Mater. Lett. 2001, 50 (2=3), 87–91. Wu, Z.J.; Ahn, I.S.; Lin, Y.X.; Huang, L.Y.; Lan, X.R.; Lee, K. Methyl orange adsorption by microporous and mesoporous TiO2–SiO2 and TiO2–SiO2–Al2O3 composite xerogels. Compos. Interfaces 2004, 11 (2), 205–212. Fuertes, A.B. Low-cost synthetic route to mesoporous carbons with narrow pore size distribu-

60.

61.

62.

63.

64.

65.

66.

67.

68.

69.

70.

71.

72.

tions and tunable porosity through silica xerogel templates. Chem. Mater. 2004, 1 (3), 449–455. Yamamoto, T.; Endo, A.; Ohmori, T.; Nakaiwa, M. Porous properties of carbon gel microspheres as adsorbents for gas separation. Carbon 2004, 42 (8=9), 1671–1676. Teichner, S.L.; Nicolaon, G.A.; Vicarini, M.A.; Grades, G.E.E. Inorganic oxide areogels. Adv. Colloid Interface Sci. 1976, 5, 245–273. Ayen, R.J.; Iacobucci, P.A. Metal oxide aerogel preparation by supercritical extraction. Rev. Chem. Eng. 1988, 5 (1–4), 157–198. Fanelli, A.J.; Verma, S.; Engelmann, T.; Burlew, J.V. Scale-up studies of alumina aerogel catalyst support. Ind. Eng. Chem. Res. 1991, 30 (1), 126–129. Stengl, V.; Bakardjieva, S.; Marikova, M.; Subrt, J.; Oplustil, F.; Olsanska, M. Aerogel nanoscale aluminum oxides as a destructive sorbent for mustard gas. Ceramics-Silikaty; 2003, 47 (4), 175–180. MrowiecBialon, J.; Jarzebski, A.B.; Lachowski, A.J.; Malinowski, J.J.; Aristov, Y.I. Effective inorganic hybrid adsorbents of water vapor by the sol–gel method. Chem. Mater. 1997, 9 (11), 2486–2490. Bostain, D.A.; Brenizer, J.S.; Norris, P.M. Neutron radioscopic measurement of water adsorption coefficients in aerogels. Res. Nondestruct. Eval. 2002, 14 (1), 47–57. Ahmed, M.S.; Attia, Y.A. Multi-metal oxide aerogel for capture of pollution gases from air. Appl. Therm. Eng. 1998, 18 (9=10), 787– 797. Stengl, V.; Bakardjieva, S.; Marikova, M.; Subrt, J.; Oplustil, F.; Olsanska, M. Aerogel nanoscale magnesium oxides as a destructive sorbent for toxic chemical agents. Cent. Eur. J. Chem. 2004, 2 (1), 16–33. Khaleel, A.; Dellinger, B. FTIR investigation of adsorption and chemical decomposition of CCl4 by high surface-area aluminum oxide. Environ. Sci. Technol. 2002, 36 (7), 1620–1624. Shaban, I.S.; Mikulaj, V. Sorption–desorption of radiocesium on various sorbents in presence of humic acid. J. Radioanal. Nucl. Chem. 1996, 208 (2), 593–603. Yamashita, J.; Ojima, T.; Shioya, M.; Hatori, H.; Yamada, Y. Organic and carbon aerogels derived from poly(vinyl chloride). Carbon 2003, 41 (2), 285–294. Coleman, S.J.; Coronado, P.R.; Maxwell, R.S.; Reynolds, J.G. Granulated activated carbon modified with hydrophobic silica aerogel— potential composite materials for the removal

Sorbent Technology

73.

74.

75.

76.

77.

78.

79.

80.

81.

82.

83.

of uranium from aqueous solutions. Environ. Sci. Technol. 2003, 37 (10), 2286–2290. Gabelich, C.J.; Tran, T.D.; Suffet, I.H. Electrosorption of inorganic salts from aqueous solution using carbon aerogels. Environ. Sci. Technol. 2002, 36 (13), 3010–3019. Rowsell, J.L.C.; Millward, A.R.; Park, K.S.; Yaghi, O.M. Hydrogen sorption in functionalized metal-organic frameworks. J. Am. Chem. Soc. 2004, 126 (18), 5666–5667. Eddaoudi, M.; Li, H.; Yaghi, O.M. Highly porous and stable metal-organic frameworks: structure design and sorption properties. J. Am. Chem. Soc. 2000, 122, 1391–1397. Eddaoudi, M.; Moler, D.; Li, H.; Chen, B.; Reineke, T.; O’Keeffe, M.; Yaghi, O.M. Modular chemistry: secondary building units as a basis for the design of highly porous and robust metal-organic carboxylate frameworks. Acc. Chem. Res. 2001, 34, 319–330. Eddaoudi, M.; Kim, J.; Rosi, N.; Vodak, D.; Watcher, J.; O’Keeffe, M.; Yaghi, O.M. Systematic design of pore size and functionality in isoreticular MOFs and their application in methane storage. Science 2002, 295, 469–472. Dybtsev, D.N.; Chun, H.; Yoon, S.H.; Kim, D.; Kim, K. Microporous manganese formate: a simple metal-organic porous material with high framework stability and highly selective gas sorption properties. J. Am. Chem. Soc. 2004, 126 (1), 32–33. Rosi, N.; Eckert, J.; Eddaoudi, M.; Vodak, D.; Kim, J.; O’Keeffe, M.; Yaghi, O.M. Hydrogen storage in microporous metal-organic frameworks. Science 2003, 300, 1127–1129. Pan, L.; Sander, M.B.; Huang, X.Y.; Li, J.; Smith, M.; Bittner, E.; Bockrath, B.; Johnson, J.K. Microporous metal organic materials: promising candidates as sorbents for hydrogen storage. J. Am. Chem. Soc. 2004, 126 (5), 1308–1309. Yaghi, O.M.; O’Keeffe, M.; Ockwig, N.W.; Chae, H.K.; Eddaoudi, M.; Kim, J. Reticular synthesis and the design of new materials. Nature 2003, 423 (6941), 705–714. Wang, Q.-M.; Shen, D.; Bu¨low, M.; Deng, S.; Fitch, F.R.; Lemcoff, N.O.; Semanscin, J. Metallorganic molecular sieve for gas separation and purification. Microporous Mesoporous Mater. 2002, 55, 217–230. Wang, Q.-M.; Shen, D.; Bulow, M.; Lau, M.-L.; Fitch, F.R.; Deng, S. Metallo-organic Polymers for Gas Separation and Purification. U.S. Patent 6,491,740, , Dec 10, 2002.

2843

84. Office of Science, U.S. Department of Energy, Basic Research Needs for the Hydrogen Economy, Feb. 2004. 85. Seayad, A.M.; Antonelli, D.M. Recent advances in hydrogen storage in metal-containing inorganic nanostructures and related materials. Adv. Mater. 2004, 16 (9=10), 765–777. 86. Ding, R.G.; Lu, G.Q.; Yan, Z.F.; Wilson, M.A. Recent advances in the preparation and utilization of carbon nanotubes for hydrogen storage. J. Nanosci. Nanotechnol. 2001, 1 (1), 7–29. 87. Liu, C.; Fan, Y.Y.; Liu, M.; Cong, H.T.; Cheng, H.M.; Dresselhaus, M.S. Hydrogen storage in single-walled carbon nanotubes at room temperature. Science 1999, 286 (5442), 1127–1129. 88. Lueking, A.; Yang, R.T. Hydrogen storage in carbon nanotubes: residual metal content and pretreatment temperature. AIChE J. 2003, 49 (6), 1556–1568. 89. Tang, C.C.; Bando, Y.; Ding, X.X.; Qi, S.R.; Golberg, D. Catalyzed collapse and enhanced hydrogen storage of BN nanotubes. J. Am. Chem. Soc. 2002, 124 (49), 14550–14551. 90. Chen, J.; Li, S.L.; Tao, L.; Shen, Y.T.; Cui, C.X. Titanium disulfide nanotubes as hydrogenstorage materials. J. Am. Chem. Soc. 2003, 125 (18), 5284–5285. 91. Sandrock, G. A panoramic overview of hydrogen storage alloys from a gas reaction point of view. J. Alloys Compd. 1999, 887, 293–295. 92. Zaluska, A.; Zaluski, L.; Strom-Olsen, J.O. Structure, catalysis and atomic reactions on the nano-scale: a systematic approach to metal hydrides for hydrogen storage. Appl. Phys. A—Mater. Sci. Process. 2002, 72 (2), 157–165. 93. Fichtner, M.; Fuhr, O.; Kircher, O. Magnesium alanate—a material for reversible hydrogen storage? J. Alloys Compd. 2003, 356, 418–422. 94. Zuttel, A.; Wenger, P.; Rentsch, S.; Sudan, P.; Mauron, P.; Emmenegger, C. LiBH4: a new hydrogen storage material. J. Power Sources 2003, 118 (1=2), 1–7. 95. Chen, P.; Xiong, Z.T.; Luo, J.Z.; Lin, J.Y.; Tan, K.L. Interaction of hydrogen with metal nitrides and imides. Nature 2002, 420 (6913), 302–304. 96. Mao, W.L.; Mao, H.-K.; Goncharov, A.F.; Struzhkin, V.V.; Guo, Q.; Hu, J.; Shu, J.; Hemley, R.J.; Somayazulu, M.; Zhao, Y. Hydrogen cluster in clathrate. Science 2002, 29 (5590), 2247–2249. 97. Mao, W.L.; Mao, H.-K. Hydrogen storage in molecular compounds. Proc. Natl. Acad. Sci. USA 2004, 101 (3), 708–710.

S

2844

98. Mao, W.L.; Mao, H.-K. Composition and Method for Hydrogen Storage. U.S. Patent 6,735,960 B2, 2004. 99. Xie, Y.C.; Tang, Y.Q. Spontaneous monolayer dispersion of oxides and salts onto surface of supports: applications to heterogeneous catalysis. Adv. Catal. 1990, 37, 1–43. 100. Golden, T.C.; Kratz, W.C.; Wilhelm, F.C Highly Dispersed Cuprous Compositions. U.S. Patent 5,126,310Jun 30, 1992. 101. Golden, T.C.; Kratz, W.C.; Wilhelm, F.C.; Pierantozzi, R.; Rokicki, A Highly Dispersed Cuprous Compositions. U.S. Patent 5,175,137, Dec 29, 1992. 102. Deng, S.G.; Lin, Y.S. Microwave heating synthesis of supported sorbents. Chem. Eng. Sci. 1997, 52 (10), 1563–1575. 103. Takahashi, A.; Yang, R.T.; Munso, C.L.; Chinn, D. Cu(I)-Y-zeolite as a superior adsorbent for diene=olefin separation. Langmuir 2001, 17, 8405–8413. 104. Takahashi, A.; Yang, R.T.; Munso, C.L.; Chinn, D. Influence of Ag content and H2S exposure on 1,3-butadiene=1-butene adsorption by Ag ionexchanged Y-zeolites (Ag-Y). Ind. Eng. Chem. Res. 2001, 40, 3979–3988. 105. Yang, R.T.; Kikkinides, E.S. New sorbents for olefin paraffin separations by adsorption via p-complexation. AIChE J. 1995, 41 (3), 509–517. 106. Golden, T.C.; Kalbassi, M.A.; Taylor, F.W.; Allam, R.J Use of Zeolites and Alumina in Adsorption Processes. U.S. Patent 5,779,767, Jul 14, 1998. 107. Deng, S.; Kumar, R.; Wolf, R.; Andrecovich, M.J Purification of Gases Using Multi-Composite Adsorbent. U.S. Patent 6,358,302, Mar 19, 2002. 108. Kumar, R.; Huggahalli, M.; Deng, S.G.; Andrecovich, M. Trace impurity removal from air. Adsorption 2003, 9 (3), 243–250. 109. Lin, Y.S.; MacLean, D.L.; Zeng, Y High Temperature Adsorption Process. U.S. Patent 6,059,858, May 9, 2000. 110. Bouwmeester, H.J.M.; Burgraaf, A.J. Dense ceramic membranes for oxygen separation. In Solid State Electrochemistry; Gellings, P.J., Bouwmeester, H.J.M., Eds.; CRC Press: Boca Raton, FL, 1997; 481–553. 111. Dyer, P.N.; Richards, R.E.; Russek, S.L.; Taylor, D.M. Ion transport membrane technology for oxygen separation and syngas production. Solid State Ionics 2000, 134, 21–33. 112. Kang, D.; Srinivasan, R.S.; Thorogood, R.M.; Foster, E.P Integrated High Temperature

Sorbent Technology

113.

114.

115.

116.

117.

118.

119. 120.

121.

122.

123.

124.

125.

126.

127.

Method for Oxygen ProductionU.S. Patent 5,516,359, May 16, 1996. Prasad, R Advanced Membrane System for Separating Gaseous Mixtures. U.S. Patent 5,709,732, Jan 20, 1998. Yang, Z.; Lin, Y.S.; Zeng, Y. High-temperature sorption process for air separation and oxygen removal. Ind. Eng. Chem. Res. 2002, 41, 2775–2784. Yang, Z.; Lin, Y.S. A semi-empirical equation for oxygen nonstoichiometry of perovskitetype ceramics. Solid State Ionics 2002, 150, 245–254. Yang, Z.; Lin, Y.S. Equilibrium of oxygen sorption on perovskite type ceramic sorbents. AIChE J. 2003, 49, 793–798. Yang, Z.H.; Lin, Y.S. High temperature oxygen sorption in fixed-bed packed with perovskitetype ceramic sorbents. Ind. Eng. Chem. Res. 2003, 42, 4376–4381. Draper, E.L.; Becker, R.A. Research and Development Needs for the Sequestration of Carbon Dioxide as Part of a Carbon Management Strategy; The National Coal Council: Washington, DC, 2000. Azar, C.; Rodhe, H. Targets for stabilizing of atmospheric CO2. Science 1997, 276, 1818–1819. Bruant, R.G. Jr.; Guswa, A.J.; Celia, M.A.; Peters, C.A. Safe storage of CO2 in deep saline aquifers. Environ. Sci. Technol. 2002, 36, 240–245. Herzog, H.J. What future for carbon capture and sequestration? Environ. Sci. Technol. 2001, 35, 148–153. Rao, A.B.; Rubin, E.A. Technical, economic, and environmental assessment of amine-based CO2 capture. Environ. Sci. Technol. 2002, 36, 4467–4475. Zou, Y.; Mata, V.; Rodrigues, A.E. Adsorption of carbon dioxide on basic alumina at high temperatures. J. Chem. Eng. Data 2000, 45, 1093–1095. Zou, Y.; Mata, V.; Rodrigues, A.E. Adsorption of carbon dioxide onto hydrotalcite-like compounds (HTlcs) at high temperatures. Ind. Eng. Chem. Res. 2001, 40, 204–209. Golden, T.C.; Taylor, F.W.; Johnson, L.M.; Malik, N.H.; Raiswell, C.J Purification of Air. U.S. Patent 610,659, Aug 22, 2000. Monereau, C Use of an Activated Alumina for Removing the CO2 from a Gas. U.S. Patent 6,379,430, Apr 30, 2002. Kumar, R.; Huggahalli, M.; Bulow, M Thermal Swing Adsorption Process. U.S. Patent 6,432,171, Aug 13, 2002.

Sorbent Technology

128. Ko, D.; Siriwardane, R.; Biegler, L.T. Optimization of a pressure-swing adsorption process using Zeolite 13X for CO2 sequestration. Ind. Eng. Chem. Res. 2003, 42, 339–348. 129. Yong, Z.; Mata, V.; Rodrigues, A.E. Adsorption of carbon dioxide at high temperature–a review. Sep. Purif. Technol. 2002, 26 (2=3), 195–205. 130. Nakagawa, K.; Ohashi, T. A novel method of CO2 capture from high temperature gases. J. Electrochem. Soc. 1998, 145 (4), 1344–1346. 131. Ida, J.-I.; Lin, Y.S. Mechanism of hightemperature CO2 sorption on lithium zirconate.

2845

Environ. Sci. Technol. 2003, 37 (9), 1999– 2004. 132. Xiong, R.; Ida, J.-I.; Lin, Y.S. Kinetics of carbon dioxide sorption on potassium doped lithium zirconate. Chem. Eng. Sci. 2003, 58, 4377–4385. 133. Ida, J.-I.; Xiong, R.; Lin, Y.S. Synthesis and CO2 sorption properties of pure and modified lithium zirconate. Sep. Purif. Technol. 2004, 36, 41–51. 134. Ortiz, A.L.; Harrison, D.P. Hydrogen production using sorption-enhanced reaction. Ind. Eng. Chem. Res. 2001, 40, 5102–5109.

S

Spinning Disk Reactor S R. J. J. Jachuck Process Intensification and Clean Technology (PICT) Group, Department of Chemical Engineering, Clarkson University, Potsdam, New York, U.S.A.

J. R. Burns Protensive Ltd., Bioscience Centre, Centre for Life, Newcastle Upon Tyne, U.K.

INTRODUCTION The concept of process intensification aims to achieve enhancement in transport rates by orders of magnitude to develop multifunctional modules with a view to provide manufacturing flexibility in process plants. In recent years, advancement in the field of reactor technology has seen the development of catalytic plate reactors, oscillatory baffled reactors, microreactors, membrane reactors, and trickle-bed reactors. One such reactor that is truly multifunctional in characteristics is the spinning disk reactor (SDR). This reactor has the potential to provide reactions, separations, and good heat transfer characteristics.

 High solid=liquid heat=mass transfer.  High liquid=vapor heat=mass transfer. A schematic diagram of an SDR is shown in Fig. 2 For the purpose of clarity and continuity this entry has been subdivided into the following sections: hydrodynamics of liquid flow on a rotating surface, variation in SDR configuration, performance estimators, and process application of SDRs.

HYDRODYNAMICS OF LIQUID FLOW ON A ROTATING SURFACE Synchronized Flow Model

SPINNING DISK REACTOR The SDR technology utilizes the effects of centrifugal force, which is capable of producing highly sheared thin films (Fig. 1) on the surfaces of rotating disks or cones. Extensive heat and mass transfer studies using SDRs have shown that convective film heat transfer coefficients as high as 14 kW=m2, K, and mass transfer coefficients, KL, with values as high as 30
105 m=sec, and KG values as high as 12
108 m=sec, can be achieved while providing micromixing and appropriate fluid dynamic environment for achieving faster reaction kinetics.[1] The size of the disk may range from 60 to 500 mm in diameter and the surface characteristic may be smooth, grooved, or meshed depending on the application and the throughput requirement. The rotational speeds may range from 100 to about 6000 rpm (typically around 1500 rpm). The SDR, which has been successfully used to perform free radical as well as condensation polymerizations, fast precipitation reactions for the production of mono-dispersed particles and catalyzed organic reactions, has the following characteristics:[2–4]  Intense mixing in the thin liquid film.  Short liquid residence time (may allow the use of higher processing temperatures).  Plug flow characteristics. Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120022537 Copyright # 2006 by Taylor & Francis. All rights reserved.

The simplest model for flow over a rotating disk surface assumes that the liquid is rotating at the same speed as the disk itself and is thus fully synchronized with the disk rotation. Under these conditions, the centrifugal acceleration driving the liquid film across the disk surface at radius r can be simply estimated as ro2. With this assumption, the flow over the disk can be made analogous to flow over an inclined surface. Nusselt provided a simple model for laminar liquid flow down an inclined plane.[5] This assumed that the liquid had reached fully developed conditions in which drag due to viscous shear exactly balanced the weight of the film. Under these conditions, Nusselt showed that for a Newtonian fluid of kinematic viscosity, n, film thickness, f, could be written in terms of the liquid flow rate, Q, moving over a vertically inclined surface of width, w, under a gravitational acceleration, g, using the following relationship:

f ¼




3nQ wg

1=3

ð1Þ

By substitution of gravitational acceleration g with centrifugal acceleration ro2 and the width of the inclined surface w with the perimeter of the disk at radius r, the following equation for film thickness on 2847

2848

Spinning Disk Reactor

using the following equation: 3u ¼ S ¼ f




3 2p

1=3


Qro4 n2

1=3

ð4Þ

Integration of the equations for film thickness and velocity can be used to provide calculations for other global measures of conditions on the disk surface. The first of these is the measurement of the volume of liquid on the disk surface. This can be calculated by integration of Eq. (2) to give

Fig. 1 View of sheared liquid films on an SDR. (View this art in color at www.dekker.com.)

a rotating disk surface can be written as f ¼




3nQ 2pr 2 o2

1=3




1 12p2

[6]

ð2Þ

1=3


Q2 o2 rn

1=3




81nQp2 16o2

1=3

R4=3

ð5Þ

Residence time of the liquid on the disk can be estimated from the volume calculation above and the liquid flow rate Q using the following equation:

This equation assumes that the liquid velocity can instantaneously adjust to the balance drag against centrifugal force. This assumption has been shown to be a reasonable one. Burns et al. and Eq. (2) can be used as the starting point for film thickness modeling.[7] The fully synchronized flow model can be used to provide reasonable estimates for a wide range of measures that characterize the flow over the spinning disk. The first of these is the radial velocity of the film at radius r that can be calculated from Q u ¼ ¼ 2prf

V ¼

ð3Þ

For a fully developed flow profile the radial velocity can be used to provide an estimate of surface shear, S,

t ¼

V ¼ Q




81p2 R4 n 16Q2 o2

1=3

 3:68




R4 n Q2 o2

1=3

ð6Þ

A measure of average film thickness and velocity can also be generated from Eq. (5). In the case of average film thickness, fAV, it is defined as the volume of liquid on the disk per unit disk area and characterizes film thickness for the system. In the case of average radial velocity it is defined as the radius divided by the residence time and again characterizes conditions for the process. These two measures are given by 


81 1=3 Qn 1=3 Qn 1=3 fAV ¼  1:17 ð7Þ 16p R2 o2 R2 o2 uAV ¼


2 2 1=3 R Q o  0:27 t Rn

ð8Þ

Eqs. (7) and (8) allow for a characteristic radial surface shear rate to be calculated as SAV


1=3 3uAV QRo4  0:69 ¼ hAV n2

ð9Þ

Wetting of the disk surface

Fig. 2 A schematic view of an SDR. (View this art in color at www.dekker.com.)

Wetting is an important aspect of the SDR. If the disk is not wetted then dry spots are created and rivulets are formed, which significantly reduce the transport rates achieved on the disk. Hartley and Murgatroyd provided a list of theoretical models for calculating the wetting film for liquid flows under gravity.[8] The equations derived for these models were based on physical principles rather than empirical data but have compared favorably with experimental results. Because

Spinning Disk Reactor

2849

of this these models can be used for initial calculations of wetting of smooth spinning disk surfaces. Two models for wetting of falling films were provided by Hartley and Murgatroyd.[8] Both these are based on the stability of dry patches rather than wetted films with the assumption that if a dry patch is not stable a wetting film should form. The first model is derived from minimizing surface energy and the second is from force balance at a contact line surrounding a dry spot. These are given by Surface Energy Criterion
1=5
2=5 s n fmin ¼ 1:34 r g Force Criterion

 sð1
cos yÞ 1=5 n 2=5 fmin ¼ 1:72 r g

Nonsynchronized Flow Model—Spin-up Zone ð10Þ

ð11Þ

These equations produced very similar results with the exception that the force criterion approach included details of contact angle y between the liquid and the surface and was therefore a more complete description. This theory can be related to the conditions on a spinning disk by the substitution of gravitation acceleration g with centrifugal acceleration ro2. A stability parameter b for the film can then be given by comparing the film thickness at the edge of the disk with the minimum wetting film thickness above. This can be written as b ¼

f fmin

at r ¼ R with g ¼ ro2

Using Eqs. (10) and (11) with Eq. (2) gives !1=15 r3 Q5 o2 b ¼ 0:454 s3 ð1
cos yÞ3 nR4

Ramshaw, and Burns et al.) has shown that wetting films for b values within the ‘‘metastable film’’ region can be obtained.[7,9,10] In particular, the use of preflooding of the disk or the use of surface roughening can push the limit of the wetting film thickness down to the lower end of the metastable film region, further details of which will be discussed in the later sections. However, the above wetting parameter calculations can be used to examine how likely it is for a process to achieve a stable wetted disk surface.

ð12Þ

ð13Þ

Experimental data on falling films have suggested that this theory provides a very conservative measurement of minimum wetting film thickness as it is derived from the principle of the stability of dry patches rather than the breakdown of films. In general, it was observed that films could be maintained down to close to an order of magnitude lower than the above equations suggest, although under very contrived conditions.[8] Based on this, it can be assumed that b > 1 ) Stable wetting film 1 > b > 0:1 ) Metastable film b < 0:1 ) Rivulet flow

Examination of conditions used during a range of experimental work on SDR (Woods, Auone and

The synchronized flow model discussed in the previous sections can be used to provide estimates for general conditions on the spinning disk surface. However, to examine conditions close to the central feed requires a more complex two-dimensional model for fluid flow. In particular, this is required to provide information on the radial distance needed to achieve synchronization with the disk rotation, commonly referred to as the spin-up zone.[7] A two-dimensional model for flow over a rotating disk, incorporating inertial and viscous influences, was described by Wood and Watts.[6] This was termed the Pigford model, owing to its origin. Burns et al. gave a comparison of the model with measurements of flow over a spinning disk.[7] A review of those findings and the model derivation are given below. The twodimensional Pigford model can be written in terms of the radial and tangential flow velocity, u and v, respectively, relative to the disk using the following equations.


@u u @r

u




@v @r




 v2 12p2 r 2 K1 n 3 ¼
u
Q2 r



þ

uv ¼ r




 12p2 r 2 K2 n 2 u ðro
vÞ Q2

ð14Þ

ð15Þ

where K1 and K2 were used as empirical correction factors to adjust the viscous terms in the equation. No specific data were given on the values of the constants K1 and K2, although it was assumed that they were close to unity. The solution of these equations was shown by Burns et al. to be strongly linked to the Eckman length scale l, which was defined as[7] l ¼




Q2 on

1=4

ð16Þ

This scale was derived from the Eckman number that describes the ratio of viscous momentum transfer

S

2850

Spinning Disk Reactor

through the film to that of the angular momentum of the film and can be used to gauge the ability for synchronized rotating flow to be achieved. Burns et al. found that K1 had an unclear link with other experimental parameters and generally was in the region of 0.5–0.7 with an average of 0.61. In contrast, the value of parameter K2 was found to be strongly linked to the Eckman length scale l. The best fit of experimental results to the Pigford model was reported by Burns et al. as K1 ¼ 0:61 K2 ¼




l l0

ð17Þ



ð18Þ

where l0 ¼ 10.8 cm and is an empirical constant whose units served only to balance the dimensions in Eq. (26). It is likely, however, that l0 may depend on other parameters not examined within the scope of the reported experiments, such as nozzle diameter, and may be a more complex function of several other dimensionless groups. The experimentally determined values of parameters K1 and K2 allow the Pigford model to be used to examine the extent of the spin-up zone on the disk surface. Before this can be done, however, an unambiguous mathematical definition for the spin-up zone is required. This was provided by Burns et al. as the location at which the radial velocity stops accelerating and starts following the decelerating profile described by the synchronized flow model as shown in Eq. (3).[7] Mathematically, this was defined as du ¼ 0 dr

and

d2 u < 0 at r ¼ rS dr 2

ð19Þ

Numerical analysis of the Pigford model, combined with Eqs. (17) and (18), implied that the spin-up zone rS based on this definition should be given by

rS




d ¼ 0:88l 2l

0:025


l l0

0:37

ð20Þ

Variation in Configuration Surface structuring Surface structuring can be used on SDR to achieve a variety of effects on the processing conditions. These can be categorized by the scale of the features being made to the disk. At the film-thickness scale or lower, surface structuring or roughening can be used to increase the wetting of the disk surface by the interference with the liquid–solid–gas interface contact. It can also be used to generate greater surface area at the base of the film and sites to promote small-scale convective mixing for situations with low-viscosity fluids and high flow rates. Structuring at the larger than film-thickness scale can include surface grooves and channels and possible incorporation of surface meshes. These act to influence the velocity distribution across the disk allowing the liquid in some cases to detach and reimpact on the surface many times as it crosses the disk. This can be used to introduce a greater variation in the transport rates for the liquid flowing over the disk compared to the more steady conditions on the smooth disk surface. At the large end of the scale changes in the configuration can be a change in the whole inclination of the surface to the centrifugal field the influence of which is discussed in the next few sections. Spinning cone reactor The SDR can be described as just one of the many reactor designs based on rotating surfaces. In particular, the SDR can be treated as a special case of the more general spinning cone reactor. These devices generally have a flat spinning disk close to the center followed by a conical section inclined to the centrifugal field. The equations describing the flow over these devices can, however, be shown to be more general versions of the previous equations given for the spinning disk reactor. For the case of a spinning cone inclined at an angle a to the plane of rotation, film thickness, velocity, residence time, and surface shear can be related to that for the spinning disk using the following scaling relationships:

where d is the nozzle diameter used to inject the liquid. This result implies that the spin-up zone is approximately proportional to l2=3. A criterion for the accuracy of the synchronized flow model can also be derived from this analysis by comparison of the extent of the spin-up zone with the radius of the disk. This can be written simply as

fCONE ¼ ðcos aÞ2=3 fDISK

ð22Þ

uCONE ¼ ðcos aÞ2=3 uDISK

ð23Þ

tCONE ¼ ðcos aÞ5=3 tDISK

ð24Þ

SCONE ¼ ðcos aÞ4=3 SDISK

ð25Þ

rS  1 ) Synchronized flow R

It should be noted that the relationships described above assume that the cone extends to the same radius

ð21Þ

Spinning Disk Reactor

2851

as the disk rather than the same length in the direction of flow and, hence, results in a strong influence on residence time due to slower flow combined with a longer distance to travel. It should also be noted that the flow velocity is given in Eq. (23) rather than the strict radial velocity and is therefore in the direction of the inclined surface. Eq. (24) gives an equivalent residence time scaling assuming that the cone extends to the axis of rotation. In reality, this is not often the case and the device will have a flat disk surface close to the center that leads to a cone further out. A more complete calculation for residence time can be given using the following equation: t ¼




81p2 n 16Q2 o2

1=3

4=3

r2

4=3

 r1 cos5=3 a

4=3

þ r1

!

ð26Þ

where flow is over a flat disk up to a radius r1 and then a cone inclined at an angle a up to a radius r2. It should be noted that Eqs. (22)–(26) neglect the influence of gravity on the flow over the cone. Under most operating conditions, this will probably be a reasonable assumption; however, for systems with a vertical axis of rotation and low rotational speeds the validity of the previous equations may break down. A limit for the applicability of Eqs. (22)–(26) can be expressed as cos a 

Ro2 ) Gravity may be neglected g

ð27Þ

Nested cone reactor A further modification of the rotating surface concept is that of the nested cone. This allows the use of inclined surfaces to the centrifugal acceleration in a similar manner to the cone but with two major differences. The first is that the design is more compact as the axial component of flow alternates as the liquid moves from conical surface to conical surface, which reduces the axial space required for the device. The second difference is that the liquid detaches from the surface as it moves from surface to surface, leading to periodic impacts and the potential for increased mixing at the point of impact. This can further be enhanced by the use of nested cones where the upper and the lower sections are moving at different rotational speeds, or even in opposite directions of rotation at the expense of increased power demand on the shaft. Modeling of the flow on the conical surfaces can be approximated by the same scaling rules as shown in Eqs. (22), (23), and (25). However, residence time cannot be modeled in such a simple manner as it includes contributions both from flow over the conical sections and from detached flow between surfaces. In modeling the latter, the following equations can be used.

If it is assumed that the liquid leaving one conical section at a radius r1 has a radial velocity component that is small compared to that of the rotational speed r1o of the surface, then the liquid can be assumed to leave the surface at a velocity r1o traveling at a tangent to the cone in the same direction as its rotation. Under these circumstances, the distance x traveled by the liquid before impacting on a second cone surface at radius r2, assuming deflection due to the surrounding gas flow or gravity, can be written as x ¼

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi r22  r12

ð28Þ

Assuming that the liquid travels between the surfaces without any change in velocity, as viewed from a nonrotating Cartesian frame of reference, the time required to travel between the cone surfaces can be written as sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi x 1 r22 t ¼ ¼  1 ð29Þ r1 o o r12 In comparison, the estimated time spent on the conical surface prior to this can be estimated from the following equation, which is a modified version of Eq. (26): t ¼




81p2 n 16Q2 o2

1=3

4=3

r1

4=3

 r0 cos5=3 a

!

ð30Þ

This assumes that the conical surface extends from radius r0 to a radius r1. This equation neglects the influence of gravity and is therefore subject to the same limitations as those expressed in Eq. (27). A total residence time for the nested cone system can therefore be calculated by a sum of Eqs. (28) and (29) for each conical surface. The velocity of the liquid hitting the second surface can also be calculated using the previous assumptions. The perpendicular component of velocity relative to the rotating surface as radius r2 can be written as

uimp

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi r2 ¼ r1 o 1  12 r2

ð31Þ

The tangential component of impact velocity on the second surface will also depend on whether it is moving at the same or a different rotational speed. Assuming the first surface is rotating at o1 and the second surface at o2 the tangential impact velocity can be written as vimp ¼

r12 o1  r22 o2 r2

ð32Þ

Thus, it can be seen that substantial impact velocities, and hence energy available for mixing, can be

S

2852

Spinning Disk Reactor

Table 1 SDR—scaling influences Film thickness

Residence time

Flow velocity

Surface shear

26% increase

37% decrease

59% increase

26% increase

Double disk rotation

37% decrease

37% decrease

59% increase

152% increase

Ten times viscosity

115% increase

115% increase

54% decrease

78% decrease

Double liquid flow

Double disk radius

37% decrease

152% increase

21% decrease

26% increase

Disk to 30 cone

10% increase

27% increase

9% decrease

17% decrease

Disk to 75 cone

146% increase

851% increase

59% decrease

84% decrease

obtained by the use of surfaces moving at different rotational speeds even when the surfaces are in close proximity. Scaling Rules for Hydrodynamic Properties In the previous sections equations have been given to estimate a range of properties of the fluid flow over the rotating systems, spinning disks, and more generally spinning cones. These equations can be used to estimate general rules for examining the scaling of conditions from one rotating device to another and also from one set of operating conditions to another. Table 1 summarizes these implied scaling influences. It should be noted that these rules assume that the liquid flowing over the system has Newtonian properties and is fully synchronized with the disk rotation. The influence of gravity is also neglected and it is assumed that the surface is completely wetted.

Estimating transport by diffusion Owing to the nature of thin films generated by both spinning disks and spinning cones, transport by diffusion can become a significant method of transport of both heat and mass within the film. In the case of heat transfer this is especially true and will be the dominant mode of transport within the film. The equations for transport by diffusion have been studied and solved for many simple systems. One of the most comprehensive studies of this process was given by Crank in which the equations for diffusion are examined for several generalized systems.[11] In the solution of the diffusion equation one parameter is of paramount importance: that is, the Fourier number, Fo. This can be expressed as Dt DL ¼ 2 d ud2

Fo ¼

ZR

D dr uf 2

ð34Þ

0

Combining Eqs. (2), (3), (22), (23), and (34) gives Fo ¼




9p4 32

1=3


o 2 R 8 D3 nQ4

1=3

cos2=3 a

ð35Þ

It should be noted that the diffusivity D in Eq. (35) is the molecular diffusivity in the case of mass transfer, which is typically of the order of 109 m2=sec or thermal diffusivity defined as

Performance Estimators

Fo ¼

for diffusion. This can alternatively be written in terms of flow velocity u perpendicular to the process and the length traversed, L, in the direction of flow. In the case of thin films, the diffusion path length can be replaced by the film thickness f. However, for flow over a rotating surface film thickness and velocity are constantly changing and so an integrated form of Eq. (33) is required to examine the potential for diffusive transport through the rotating film. Fourier number for an SDR can be developed based on the following integration:

ð33Þ

where D is the diffusion coefficient for the process, t is the exposure time, and d is the characteristic path length

DHEAT ¼

k cP r

ð36Þ

in the case of heat transfer. For heat transfer, Eq. (34) will provide a measure of the capability of the film to reach the temperature of the solid surface assuming that it is maintained at a constant temperature. Burns and Jachuck showed that this parameter correlated well for a mass transfer limited process using the conversion of calcium hydroxide into calcium carbonate by diffusion of carbon dioxide through the liquid film on an SDR.[7] Results showed complete conversion for Fo in the region of 0.1–0.2. For nonreactive processes and heat transfer, however, efficient transport by diffusion should be expected for Fo in excess of 0.3 and preferably in excess of 1. It can be seen from Eq. (35) that the performance for increased flows can be maintained by use of either higher rotational speed or increased surface area.

Spinning Disk Reactor

2853

Eq. (34) also shows that inclined surfaces should yield slowly, decreasing diffusive transport capabilities compared to the spinning disk, owing to the thicker films. However, for reactions with some kinetic limitations the increased residence time may be more beneficial. It should also be noted that the results of mixing by collisions, motion over structured surfaces and through surface waves, are convective processes not considered in this formulation. Estimating mass transfer coefficients

It may be suggested that the transport performance is proportional to Fo defined for the spinning surface given by Eq. (35) as the assumptions used for the diffusive processing are the same.

Liquid–solid mass transfer by diffusion Mass transfer coefficients for the lower surface of a laminar film are strongly influenced by the shear at the liquid–solid interface. A solution for liquid–solid mass transfer coefficients for a diffusive process in a laminar film was provided by Bird et al. as the following equation:[12]

Correlations for mass transfer can be divided into diffusive and nondiffusive processes. The simpler of the two to predict is that of diffusive transfer and, in particular, the penetration of a species at either the solid–liquid or liquid–gas interface.

kLS

Gas–liquid mass transfer by diffusion

where the function G(7=3) is defined as

Estimation of liquid-side mass transfer coefficients for gas–liquid transport through the upper surface of the film can be most readily made by assumption of a purely diffusive transfer. Under these conditions, a mass transfer coefficient can be approximated using the following equation for mass transfer:[12] kLG

rffiffiffiffiffiffiffi 4D ¼ pt

ð37Þ

This assumes that a chemical species is penetrating into a static film over a time period of t with a diffusivity in the liquid of D. If it is assumed that the exposure time constant t in the equation is equal to the residence time of the liquid on a spinning disk surface, given by Eq. (6), then the liquid-side mass transfer coefficient kLG for diffusion into the film can be estimated as

ð38Þ

A measure of the transport performance can be given by considering the flux through the surface compared to the flux passing over the surface, which is the liquid flow rate. This produces the following dimensionless group:

Transport performance ¼
2 3 8 1=6 o D R ¼ 1:85 nQ4

Gð7=3Þ ¼

kLG pR2 Q ð39Þ

Z

0

1

ð40Þ

x4=3 ex dx ¼ 1:1906

ð41Þ

For a spinning disk, the standard model for falling film flow is complicated by the changing thickness and shear as the liquid flows over the disk. An approximation of this to conditions on a spinning disk surface can, however, be made by substitution of Eq. (9) for average liquid–solid surface shear into the above equation for mass transfer. If it is also assumed that the characteristic distance L traveled by the liquid is equal to that of the disk radius then an equation for the liquid–solid mass transfer coefficient kLS can be written for an SDR as

kLS

 3 1=6

1024Q2 o2 D 81p5 nR4
2 2 3 1=6 Q oD  0:59 nR4

kLG ¼





 2D S 1=3 ¼ Gð7=3Þ 9DL

o4 D6 Q ¼ 0:71 R2 n 2


1=9

ð42Þ

A similar transport performance estimate can be provided for the liquid–solid process through the following equation: Transport performance ¼
4 6 16 1=9 o D R ¼ 2:23 n 2 Q8

kLS pR2 Q ð43Þ

It should be noted that the grouping shown here is not identical to the Fo shown in Eq. (35) because of the strong influence of surface shear as well as molecular diffusion on the overall transport rates.

S

2854

Spinning Disk Reactor

Mass transfer with surface wave convection—liquid=gas The influence of surface waves on transport within a film has been shown to lead to enhanced mass transfer above that expected from pure diffusion.[10] This enhancement can be represented in terms of an improvement in the Sherwood number (Sh) as defined by Sh ¼

kLG fC D

ð44Þ

where fC is a characteristic film thickness for the process. A series of publications have been produced modeling the structure of surface waves produced by liquid flow over a rotating disk surface. Sisoev et al. produced a series of publications examining the fluid dynamics of surface waves on an SDR.[13] Data describing surface waves produced by the disk rotation combined with the mass transfer results produced by Auone and Ramshaw were used to formulate a mathematical model for the mass transfer process.[9,10] Their results indicated that (G. M. Sisoev, O. K. Matar, and C. J. Lawrence, personal communication) 1=2

Sh ¼ 0:154PeK

PeK ¼

Pe ¼ k


4

o2 fC nD

ro2 R sfC

 4 1=3


 2 Qn 1=3 ¼ fAV ¼ 0:54 2 2 3 R o

ð46Þ

ð47Þ

Combining these equations mass transfer coefficient kLG can be estimated as kLG ¼ 0:154




In scaling up of an SDR process it is assumed that liquid properties, such as density and viscosity, are constant and that radius R is altered in the scaling to allow for greater throughput Q. The three parameters in the scaling are therefore liquid flow rate Q, rotational speed o, and disk radius R. To examine the implications of scaling, the geometry scaling factor F for the system is used and is defined by F ¼

R2 R1

ð50Þ

where a process on a disk radius R1 is to be scaled to a disk of radius R2. The following sections provide scaling methods that link the two remaining parameters to the scaling factor F.

Scaling with Residence Time and Film Thickness Constant

and the characteristic film thickness is defined as

fC

SCALING CALCULATIONS

ð45Þ

where


in the scale-up of spinning disk processes as transport rates are likely to improve in reality if scaling is based on the more simplistic diffusion-only models shown by Eqs. (38) and (39).

fC5 o8 R4 rD n3s

 3 1=6




R41 Q21 o21




Q1 R21 o21



¼



¼

ð48Þ

Transport performance can therefore be estimated as kLG pR2 Transport performance ¼ Q
14 38 3 9 1=18 o R rD ¼ 0:29 n 4 s3 Q13

The first method of scaling is to preserve residence time and average film thickness. Using Eqs. (6) and (7) implies that

ð49Þ

It can be seen that performance under this model has a stronger positive dependence on rotational speed and a weaker negative influence from liquid flow rate owing to the increased surface wave generation under conditions of high speed and high flow. This can aid





R42 Q22 o22

Q2 R22 o22



) Constant residence time



) Constant film thickness

ð51Þ

ð52Þ

Therefore, scaling for this requirement would be Q2 ¼ F 2 Q1

ð53Þ

o2 ¼ o1

ð54Þ

This implies that flow rate is proportional to disk area while rotational speed should remain constant. Centrifugal acceleration would be increased in proportion to F for this system, as would radial velocity. Mass transfer performance based on the previous correlation would indicate increased performance from this scaling.

Spinning Disk Reactor

2855

Scaling with Residence Time and Mass Transfer Performance Constant If it is desired to scale while maintaining the same theoretical mass transfer performance and residence time then the following conditions must be met:


R41 Q21 o21



¼




R42 Q22 o22



) Constant residence time ð55Þ

This means that flow is scaled according to Q2 ¼ Q1




 o1 F2 o2

ð56Þ

Fixed mass transfer performance implies



4  R41 o1 R2 o2 ¼ Q21 Q22 ðModel 1: liquidgas diffusiveÞ

ð57Þ





4  R41 o1 R2 o2 ¼ Q21 Q22 ðModel 2: liquidsolid shear/diffusion)

ð58Þ





38 14  14 R38 R2 o2 1 o1 ¼ Q13 Q13 1 2 ðModel 3: liquidgas convectiveÞ

ð59Þ

Therefore, the scaling rules for these three mass transfer models are  Liquid–gas diffusive Q2 ¼ F 2 Q1

ð60Þ

o2 ¼ o1

ð61Þ

 Liquid–solid shear=diffusion Q2 ¼ F 2 Q1

ð62Þ

o2 ¼ o1

ð63Þ

 Liquid-gas with surface wave convection Q2 ¼ F

22=9

Q1

ð64Þ

o2 ¼ F 4=9 o1

ð65Þ

This implies that mass transfer by shear and by diffusion scales flow in proportion with disk area, while keeping rotational speed constant. If, however, convective transport due to surface waves is included the rotational speed is reduced as disk size is increased and

flow is increased slightly more than proportional with disk area. However, wetting characteristics and absolute applied acceleration are not kept constant under these laws and will change as the disk is scaled up for a given mass transfer and residence time performance. Ultimately, their effects on the process will need to be studied before the scaling laws can be used; however, these can be used to provide a rough engineering guide to process capabilities. Estimating Heat Transfer Performance for an SDR In the previous sections we have discussed mass transfer into the thin centrifuged film occurring on a disk surface. For many processes diffusive transport can be seen to be a significant, if not the most significant, mechanism for transport. In the case of heat, transport by diffusion, that is, conduction, can be considered even more effective as diffusion coefficients for that process are substantially higher. Therefore, transport by conduction can be used as a good estimate for heat transfer performance through the film. Substituting thermal diffusion for molecular diffusion allows the computation provided for liquid–solid mass transfer to be transferred to thermal diffusion using the following equation: hFILM ¼ kLS;THERM cP r

ð66Þ

where kLS,THERM is the equivalent transport coefficient for thermal diffusion as defined by Eq. (36). Combining this with Eq. (42) for transfer assuming a constant surface temperature gives an effective heat transfer coefficient for the film as hFILM ¼ 0:71




o4 k6 Qc3P r3 R2 n 2

1=9

ð67Þ

If the disk surface temperature is assumed to be held constant then the thermal power transfer capability into the film can be estimated using the following equation:
4 6 3 3 1=9 o k QcP r Power ¼ 0:71 pR2 DTLMTD R2 n 2

ð68Þ

where DTLMTD is the log-mean temperature difference between the film and the disk surface. Heat transfer in practice, however, is limited by the design of the spinning disk system below the film. One common design route used for the reactors is the supply of heating or cooling using a heat transfer fluid. If it is assumed that this fluid can transfer heat to the underside disk of thickness d with a performance of hHTF, then the overall heat transfer

S

2856

Spinning Disk Reactor

coefficient U for the system can be estimated using U ¼




1 hFILM

þ

kDISK 1 þ hHTF d


1

Process Application of SDRs ð69Þ

where kDISK is the thermal conductivity of the spinning disk material.

Quantifying Power Input to the Fluid Power consumed by a spinning disk and cone processes can be divided into various contributions. These include the power losses to bearing friction as well as drive inefficiencies, neither of which contributes directly to the power input of the fluid. A large component of power may also be required to initially spin-up the disk surface, this being a function of the disk mass and start-up time allowable. Power input to the fluid itself, however, may only be a fraction of the total power demand but is that which can be related to the processing conditions. For a spinning disk and spinning cone reactor, power input to the fluid can be represented as that given to accelerate the liquid up to the tip speed of the disk. This can be written as Power ¼

rQR2 o2 2

(spinning disk/cone)

ð70Þ

Power is input to the fluid from a nested cone system that has nonsynchronized upper and lower rotating surfaces. An equation to estimate the power input from that design can be written as ! n X u2imp;i þ v2imp;i rQR2 o2 Power ¼ þ rQ 2 2 i¼1 (nested cone)

ð71Þ

where the nested cone has n impacts with relative velocities of uimp and vimp as given by Eqs. (31) and (32) before the liquid leaves at the edge of the system at a radius R. Energy input to the liquid per unit volume can also be calculated from the above equations using Energy Power ¼ Volume Q

ð72Þ

This can be used to compare these reactors with that of a batch stirred reactor by using the following equation: Energy Power  Batch time ¼ Volume Batch volume

ð73Þ

The ability to handle viscous and inviscid fluids with considerable ease makes the SDR ideally suited for performing a range of process operations. It has considerable use in food, fine chemicals, polymers, energy, household products, water treatment, and the pharmaceutical industry. It is a unique treatment of fluids under the centrifugal field and has been shown to enhance rates of reaction by orders of magnitude. The SDR polymerization of styrene, butyl acrylate, and esters has been successfully demonstrated by Boodhoo and Jachuck.[2,3] Thin films on the disk also offer the opportunity to use ultraviolet radiation to trigger fast chemical reactions. Intense mixing within the films has been used to generate micro to submicrometer sized particles of barium sulfate and calcium carbonate.[14,15] Nano composites and continuous production of fine particles for high-value application are attractive possibilities with SDRs. The high mass transfer rates coupled with the ability to provide short sharp bursts of high temperature have been exploited by Protensive Limited to perform devolatilization of polymers or removal of lighter organic molecules from viscous polymer melts. In conclusion, it may be stated that SDR technology offers several opportunities for the processing sector, such as process flexibility, improved product quality, speed to market, just-in-time manufacturing, reduced footprint, improved inherent safety and energy efficiency, distributed manufacturing capability, and ability to use reactants at higher concentrations.

CONCLUSIONS This chapter has provided a summary of the characteristics, potential applications, hydrodynamic equations, and transport expressions that govern the operation of SDR. The reader is encouraged to refer to the References particularly for more information on the application of SDR. Currently, these reactors are being studied extensively by industry under pilot-plant programs and it is expected that details of the case studies will be published soon.

NOTATION D E F g H K1

Diffusivity of transferring quantity (molecule= heat=momentum) (m2=sec) Ekman number Scaling factor Acceleration due to applied field (gravitational=centrifugal) (m=sec2) Film thickness (m) Empirical constant for Pigford model

Spinning Disk Reactor

K2 Q r R RF RS S t u v x a b l s r y o n

Empirical constant for Pigform model Volumetric flow rate over the disk surface (m3=sec) Radial position (m) Radius of spinning disk (m) Radius of central feed pipe (m) Radius of spin-up zone (m) Surface shear (sec
1) Residence time on the disk (sec) Velocity in the direction of flow or radial velocity (m=sec) Tangential flow velocity (m=sec) Mesh pore size (m) Angle of inclination of the cone (rad) Wetting factor Ekman length scale (m) Interfacial tension (kg=sec2) Liquid density (kg=m3) Contact angle between liquid and disk surface (rad) Rotational speed of the disk (rad=sec) Kinematic viscosity (m2=sec)

Common Subscripts AV A CONE DISK 1 2

Average over the disk Average through the film Conditions for a spinning cone Conditions for a spinning disk Initial condition Scaled condition

REFERENCES 1. Jachuck, R.J.J.; Ramshaw, C. Process intensification: heat transfer characteristics of tailored rotating surfaces. In Heat Recovery Systems & CHP; Elsevier, 1994; Vol. 14, No. 5, 475–491. 2. Boodhoo, K.V.K.; Jachuck, R.J.J. Process intensification: spinning disk reactor for styrene polymerisation. In Applied Thermal Engineering; Elsevier, 2000; Vol. 20, 1127–1146. 3. Boodhoo, K.V.K.; Jachuck, R.J.J. Process intensification: spinning disk reactor for condensation polymerisation. In Green Chemistry; RSC Publishing, 2000; Vol. 4, 235–244.

2857

4. Vicevic, M; Jachuck, R.J.J.; Scott, K. Process intensification for green chemistry: rearrangement of alpha-pinene oxide on the spinning disk. 4th International Conference on Process Intensification for the Chemical Industry Brugge, Belgium, Sep 10–12, 2001. 5. Nusselt, W. Die oberfla¨chenkondensation des wasserdampfes. Z. Ver. Deut. Ing. 1916, 60, 541. 6. Wood, R.M.; Watts, B.E. The flow, heat, and mass transfer characteristics of liquid films on rotating disks. Trans. Inst. Chem. Eng. 1973, 51, 315–322. 7. Burns, J.R.; Ramshaw, C.; Jachuck, R.J.J. Measurement of liquid film thickness and the determination of spin-up radius on a rotating disk using an electrical resistance technique. Chem. Eng. Sci. 2003, 58 (11), 2245–2253. 8. Hartley, D.E.; Murgatroyd, W. Criteria for the break-up of thin liquid layers flowing isothermally over solid surfaces. Int. J. Heat Mass Transfer 1964, 7, 1003–1015. 9. Woods, W.P. The Hydrodynamics of Thin Films Flowing over a Rotating Disk. Ph.D. Thesis, University of Newcastle-upon-Tyne, U.K. 1995. 10. Aoune, A.; Ramshaw, C. Process intensification: heat and mass transfer characteristics of liquid films on rotating disks. Int. J. Heat Mass Transfer 1999, 42, 2543–2556. 11. Crank, J. The Mathematics of Diffusion, 2nd Ed.; Clarendon Press: Oxford, 1975; 44–68. 12. Bird, R.B.; Stewart, W.E.; Lightfoot, E.N. Transport Phenomena; Wiley: New York, London, 1960; 551–552. 13. Sisoev, G.M.; Matar, O.K.; Lawrence, C.J. Axisymmetric wave regimes in viscous liquid film flow over a spinning disk. J. Fluid Mech. 2003, 495, 385–411. 14. Cafiero, L.M.; Chianese, A.; Baffi, G.; Jachuck, R.J.J. Process intensification: precipitation of barium sulfate using a spinning disk reactor. Ind. Eng. Chem. Res. 2002, 41 (21), 5240–5246. 15. Trippa, G.; Jachuck, R.J.J. Process intensification: precipitation of calcium carbonate from carbonation reaction of lime water using a spinning disk reactor. ICheaP-6, Pisa, Italy, Jun 8–11, 2003.

S

Styrene S Guy B. Woodle UOP LLC, Des Plaines, Illinois, U.S.A.

INTRODUCTION Styrene is one of the most important aromatic monomers used for the manufacture of plastics. Small-scale commercial production of styrene began in the 1930s. Demand for styrene-based plastics has grown significantly, and in 2003 the worldwide annual production capacity was approximately 24.5 million metric tons.[1] About 65% of styrene is used to produce polystyrene. Polystyrene is used in the manufacture of many commonly used products such as toys, household and kitchen appliances, plastic drinking cups, housings for computers and electronics, foam packaging, and insulation. Polystyrene finds such widespread use because it is relatively inexpensive to produce and is easy to polymerize and copolymerize, resulting in plastics with a broad range of characteristics. In addition to polystyrene, styrene is used to produce acrylonitrile–butadiene– styrene polymer, styrene–acrylonitrile polymer, and styrene–butadiene synthetic rubber (SBR). The development of styrene technologies was mainly driven by demand for cheap synthetic rubber during and immediately after World War II. Between 5% and 10% of total styrene produced becomes a component of synthetic rubbers, which are copolymers of styrene and butadiene (SBR). Styrene copolymers containing acrylonitrile are specialty materials that are used for specific applications. Demand for styrene for the period 2004–2009 is estimated to grow at a rate of approximately 4% per year.[1]

PHYSICAL AND CHEMICAL PROPERTIES Styrene is a colorless aromatic liquid. It is only very slightly soluble in water, but infinitely soluble in alcohol and ether. Additional properties are listed in Table 1. Styrene is chemically reactive with the most important reaction being its polymerization to form polystyrene. Styrene can also copolymerize with other monomers, such as butadiene and acrylonitrile, to produce a variety of industrially important copolymers. In addition to polymerization, styrene can undergo other types of reactions due to the chemical nature of Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120007970 Copyright # 2006 by Taylor & Francis. All rights reserved.

its unsaturated side chain and aromatic ring. For example, styrene can be oxidized to form benzoic acid, benzaldehyde, styrene oxide, and other oxygenated compounds. Styrene oxide is used in the production of various cosmetics, perfumes, agricultural and biological chemicals.

REACTION KINETICS AND THERMODYNAMICS Essentially all commercially produced styrene uses ethylbenzene (EB) as a feedstock. Between 85% and 90% of worldwide styrene production is based on EB dehydrogenation. The remaining 10–15% of styrene is obtained as a coproduct in a process to produce propylene oxide.

Ethylbenzene Dehydrogenation Ethylbenzene is catalytically dehydrogenated in the presence of steam according to the equation:

CH3

CH2

+

H2

The reaction is highly endothermic and conversion is limited in extent by equilibrium. The reaction equilibrium constant is defined as: Keq ¼ ðPsty
Ph2Þ=Peb where Psty is the partial pressure of styrene, Ph2 is the partial pressure of hydrogen; and Peb is the partial pressure of ethylbenzene. High temperature, steam dilution, and low system pressure produce an equilibrium more favorable to styrene. For endothermic vapor-phase reactions, the equilibrium constant increases with temperature and 2859

2860

Styrene

Table 1 Physical properties of styrene Molecular weight

104.152

Specific gravitya

0.903




30.628

Melting point, C Boiling point,  C

145.2



Critical temperature, C Critical pressure, atm 

Vapor pressure, mm Hg at T  C a

373 46.1 1
7.0

5

20

10

40

60

100

200

400

760

18.0

30.8

44.6

59.8

69.5

82.0

101.3

122.5

145.2

Density is at 20 C referred to water at 4 C. (From Perry, R.H., Green, D.W., Eds.; Perry’s Chemical Engineers Handbook, 6th Ed.; McGraw-Hill: New York, 1984; 3-60 and Miller, S.A., Ed.; Ethylene and Its Industrial Derivatives; 901 pp.)

can be determined equation:[2]

according

to

the

following

ln Keq ¼ 16:12
ð15; 350=TÞ where Keq is the equilibrium constant in atmospheres and T is the temperature in K. The equilibrium constant has the dimension of pressure since two moles of products are formed for each mole of EB converted. Therefore, a higher total pressure will shift the reaction equilibrium to the left and reduce EB conversion. Lower pressure results in greater EB conversion without an accompanying significant decrease in styrene selectivity. Another method to create a positive shift in equilibrium is the use of steam dilution to reduce the partial pressures of EB, styrene, and hydrogen. Steam dilution provides the same effect as a reduction in total pressure. Steam dilution has several other important benefits. First, steam supplies heat to the reacting mixture. Consequently, the drop in temperature for a given EB conversion is lower, allowing greater EB conversions to be obtained with the same inlet temperature. Second, a minimum amount of steam appears to keep the catalyst in the required oxidation state for high activity. The actual quantity of steam varies with the type of catalyst used. Third, steam is believed to suppress the deposition of carbonaceous material on the catalyst. If the carbonaceous material is allowed to accumulate, the catalyst will become fouled and its activity will decline to unacceptable levels. The reaction feed mixture undergoes certain other reactions that are not equilibrium limited under typical operating conditions. Most important among these are the dealkylation reactions that result in the formation of benzene and ethylene or toluene and methane. Other reactions produce small amountsof a-methylstyrene and other high boiling components.

The key dealkylation reactions can be described by the following equations: CH3

+

CH3

H2C

CH2

CH3

+

H2

+

CH4

Both methane and ethylene undergo steam reforming reactions according to the following equations: CH4 þ H2 O ! CO þ 3H2 C2 H4 þ 2H2 O ! 2CO þ 4H2 The water–gas shift reaction also occurs and is generally near equilibrium at the reaction temperature: CO þ H2 O $ CO2 þ H2 The combination of dealkylation, steam reforming, and water–gas shift side reactions should be avoided, if possible. In addition to losing valuable EB feed by dealkylation, the resultant net formation of carbon dioxide and hydrogen by this combination of reactions inhibits the primary dehydrogenation reaction. The net hydrogen formation gives an unfavorable shift in equilibrium, while the presence of carbon dioxide has a negative effect on dehydrogenation catalyst activity.[3]

Styrene

Typically, there is less methane and ethylene present in the effluent of a reactor than would be expected from the benzene and toluene formation. Carbon monoxide is generally about 10 mol% of the total carbon oxides. The critical operating and design parameters for EB dehydrogenation are discussed in the following paragraphs. Reaction temperature Because the dehydrogenation reaction is endothermic, the reaction mixture temperature decreases as the reaction proceeds. The reaction rate slows because of the closer approach to equilibrium and the decrease in kinetic reaction rate with the decreasing temperature. Furthermore, the equilibrium constant is less favorable at lower temperature. Therefore, in a normal design, about 80% of the temperature drop occurs in approximately the first third of the catalyst bed. As a result, a high inlet catalyst temperature is required. However, high temperature also increases the rates of nonselective thermal reactions and dealkylation reactions, which form benzene and toluene by-products. In particular, as temperature is increased, the rate of benzene formation increases significantly relative to the rate of styrene formation. This means there is an effective upper limit to the inlet temperature if high styrene selectivity is a required criterion. Reaction temperature is generally adjusted by changing either the steam temperature or the steam-to-oil ratio. Catalyst quantity The amount of catalyst relative to EB feed is an important parameter for optimum reactor performance. Too little catalyst will prevent a close approach to equilibrium. If EB conversion is low, then distillation costs associated with recovery and recycle of the unconverted EB can become significant. With too much catalyst, the EB conversion reaches equilibrium before the outlet of the catalyst bed, while the side reactions continue leading to loss of selectivity. The optimum catalyst quantity is achieved by balancing the EB conversion level and the styrene yield. Catalysts typically lose activity with time onstream, which has the effect of decreasing the effective active catalyst quantity for reaction. Compensation for aging catalyst is achieved by adjusting other operating parameters, in particular, the reaction temperature. Reaction pressure Ethylbenzene dehydrogenation results in a significant increase in the volume of reactants due to the reaction stoichiometry. Lower pressure favors higher

2861

equilibrium conversion to styrene. Reaction pressure is established during the plant design at the lowest practical level. Modern commercial reactors operate below atmospheric pressure. Pressures as low as 300 mm Hg or lower are common. The key side reactions are largely independent of reaction pressure; hence, operating at lower pressures also provides higher styrene yield. Steam dilution or steam-to-oil ratio The main functions of steam dilution are to act as a diluent to reduce the hydrocarbon partial pressures, providing heat for the endothermic dehydrogenation reaction, and maintaining the catalyst’s active surface in a desirable state. Increasing the steam-to-oil ratio has the net effect of improving the EB conversion and styrene yield. However, costs associated with generating and superheating the dilution steam also increase and eventually offset the reaction advantages. Catalyst type and properties Ethylbenzene dehydrogenation is generally catalyzed by a potassium-promoted iron oxide catalyst. The most widely used catalysts are composed of iron oxide, potassium carbonate, and various metal oxide promoters. Examples of metal oxide promoters include chromium oxide, cerium oxide, molybdenum oxide, and vanadium oxide.[4] The potassium component substantially increases catalyst activity relative to an unpromoted iron oxide catalyst. Potassium has been shown to provide other benefits. In particular, it reduces the formation of carbonaceous deposits on the catalyst surface, which prolongs catalyst life. Properties such as catalyst size and shape also impact performance. In theory, smaller sized catalyst will increase reaction rates by providing more available catalyst surface area than larger sized catalyst. Small catalyst particles, however, have a disadvantage in that they result in greater pressure drop through a reactor and higher overall reaction pressures. To address this, catalyst developers have used specialized shapes, such as ribbed extrudates, to gain the advantage of increased surface area without incurring the penalty of increased pressure drop and reaction pressure. The Sud-Chemie Group and Criterion Catalysts are the major catalyst developers and manufacturers for the styrene industry. Both companies offer a wide range of catalysts to suit individual processing needs. Ethylbenzene conversion, styrene selectivity, catalyst activity, and catalyst stability can be optimized by selecting the best catalyst or a combination of catalysts for a particular application. Dow and BASF manufacture proprietary catalysts, which have been mainly for use in their own respective technologies.

S

2862

Styrene

Propylene Oxide with Styrene Co-production

COMMERCIAL PRODUCTION

In the late 1960s, a method was discovered to produce propylene oxide by the epoxidation of propylene using organic hydroperoxides as the epoxidizing agent.[5] During the epoxidation reaction, the hydroperoxide is essentially converted to the corresponding alcohol, which in turn can be dehydrated to a more desirable coproduct. Styrene is coproduced in the form of this process that uses EB hydroperoxide as the epoxidizing agent. The chemistry of this process can be broken down into three main reactions as shown in Fig. 1. The first step is oxidation of EB to form EB hydroperoxide. The oxidation is carried out in the liquid phase with a target EB conversion of approximately 13%.[6] Although higher conversions are attractive from an EB recovery and recycle standpoint, there is a significant disadvantage because the EB hydroperoxide selectivity declines sharply. The second step is epoxidation of propylene to form propylene oxide product and 1-phenylethanol. In the last step, the 1-phenylethanol is dehydrated to styrene and water. The dehydrated reaction mixture is typically stripped of light components and rerun in a styrene column to remove heavy by-products, resulting in a purified styrene product. The design and operation of a propylene oxide= styrene process plant is complicated and includes numerous pieces of equipment. As a result, the total investment cost for a commercial-scale plant is about four times that of an EB dehydrogenation plant to produce the same quantity of styrene product.

Reactor Design One important aspect of modern day EB dehydrogenation reactor design is managing the operating conditions to minimize thermal reactions. The major by-product from the thermal reaction of EB to styrene is benzene with significant subsequent conversions to a complex mixture of higher aromatics, such as anthracene and=or pyrene, as well as coke. Thermal reactions do not occur at a significant level below about 600 C, but become a considerable factor affecting overall yield when temperatures rise above 655 C. One technique to reduce thermal reactions is to delay heating the EB to the reaction inlet temperature until the last possible moment before being exposed to the catalyst. The method involves superheating EB vapor, along with a portion of the dilution steam, to a temperature below approximately 580 C. The EB is vaporized with a certain amount of steam—commonly called primary steam—to suppress coking. The EB primary steam is combined with the major part of the dilution steam immediately prior to entering the dehydrogenation catalyst bed. The major portion of the dilution steam is generally referred to as main steam. The main steam is superheated to a temperature such that, when it is mixed with the EB and the primary steam, the total combined feed mixture reaches the desired catalyst inlet temperature. Reactor design and catalyst bed configuration are key factors for controlling thermal reactions.

Fig. 1 Propylene oxide–styrene process chemistry.

Styrene

Commercial adiabatic reactors are typically of radial flow construction with the flow path moving from in to out. This radial outflow geometry requires a much lower inlet volume to obtain proper distribution of the feed vapor through the catalyst bed than either an axial flow or a radial inflow reactor configuration. The radial flow reactor design also provides the advantage of low pressure drop since the flow path through the catalyst is much shorter relative to an axial flow reactor. To minimize thermal reactions, the reactor centerpipe diameter should be as small as possible to minimize residence time at the highest temperature throughout the reactor. However, too small a diameter will produce a high pressure drop through the centerpipe, potentially causing flow maldistribution and causing the feed vapor to enter the catalyst bed with a velocity that can result in erosion and attrition of catalyst particles. A single-stage reactor with practical limits of temperature, pressure, and steam dilution is limited to 40–50% per pass conversion of EB. If the singlestage reactor effluent is reheated, the reaction mixture moves away from equilibrium allowing for higher EB conversion. When the reheated reaction mixture is fed to a second stage of catalyst, then total EB conversions of 60–75% per pass can be achieved. This process of reheating and adding catalyst stages can be repeated as frequently as economically feasible. With each additional reaction stage, however, a progressively smaller incremental EB conversion is achieved, generally with a corresponding decrease in styrene selectivity. To obtain high EB conversions, typically two or three reactors are used in series with some type of reheating between the reactors to raise the temperature of the reaction mixture. Modern day commercial

2863

reactors are highly engineered. Designers use specialized computational fluid dynamics programs to study flow characteristics throughout a reactor.

Commercial Adiabatic Dehydrogenation Processes Most commercial styrene plants are based on either the Lummus=UOP technology or the Fina=Badger technology. Dow Chemical is a major styrene producer and uses its own technology. These technologies are generally similar, but there are key differences in the details. Lummus/UOP Classic SMTM Process The first commercial plant based on the Lummus= Monsanto technology, which later became the Lummus= UOP technology, was commissioned in 1972. Since that time, more than 50 projects have been licensed with more than 40 plants in commercial operation as of 2004. A typical Lummus=UOP Classic SM process flow diagram is shown in Fig. 2. Fresh and recycled EB are combined with steam and fed to the dehydrogenation reaction section of the plant. The reactor effluent is condensed and separated into off-gas, process condensate, and a dehydrogenated mixture. The hydrogen rich off-gas stream is recovered through an off-gas compressor for use as a fuel gas. The process condensate is stripped of organics and either recycled for use within the styrene plant or exported. The dehydrogenated mixture, consisting mainly of unconverted EB, styrene product, benzene, and toluene, is fed to the distillation section of the plant.

Fig. 2 Lummus=UOP classic SM process.

S

2864

Styrene

Fig. 3 Lummus=UOP classic SM process dehydrogenation section.

The main equipment in the dehydrogenation reaction section of a Lummus=UOP Classic SM plant includes a steam superheater, two dehydrogenation reactors, a series of waste heat exchangers, and an off-gas compressor (Fig. 3). The equipment is designed to minimize pressure drop from the dehydrogenation reactors inlet to the off-gas compressor. The main steam is superheated and used to reheat the reaction mixture for the second stage dehydrogenator. The reaction mixture is reheated in a specially designed interchanger located inside the second stage dehydrogenator vessel shell. The cooled steam exiting the interchanger is reheated in the steam superheater prior to being fed to the first stage dehydrogenator. The superheated steam can range from 700 C to as high as approximately 850 C to achieve the desired inlet temperature for the first stage dehydrogenator. Superheated main steam is mixed with the EB and the primary steam immediately before entering the first stage dehydrogenator. The reactor is designed to provide a uniform reaction mixture while minimizing residence time in the centerpipe to avoid thermal reactions. The reactor effluent is cooled in a series of three waste heat exchangers before final cooling and condensing. The first stage of waste heat recovery is used to superheat the EB and the primary steam. Subsequent stages are used to generate steam at different pressures. Typically intermediate pressure steam and low pressure steam are generated, which are directed for use elsewhere in the styrene plant or larger EB–styrene complex. Hydrogen and light hydrocarbons removed from the condensed reactor effluent are compressed and used as fuel gas in the steam superheater. The process steam from the reactor effluent stream is condensed and separated by gravity from the liquid hydrocarbon components. The condensate is stripped of hydrocarbons and revaporized for use as process steam. The distillation section of a Lummus=UOP Classic SM plant consists of four distillation columns. The first

column in the sequence splits the EB and the lighter components from styrene. The EB=styrene monomer (EB=SM) splitter is operated under vacuum and uses structured packing, such as Sulzer Mellapak Plus packing, to minimize temperature and polymer formation.[7] Polymerization inhibitors are injected into the splitter to restrict polymer formation, in particular into the bottom section of the column. The overhead product from the EB=SM splitter is fed to an EB recovery column. The EB recovery column net bottoms’ stream is recycled to the dehydrogenation section. Benzene and toluene by-products in the recovery column overhead stream are separated in a benzene=toluene splitter. Oftentimes, the benzene recovered in this scheme is recycled as feed to the upstream EB plant. The EB=SM splitter bottoms’ stream is fed to the SM column where the styrene is purified by removal of any heavy residual tars. Tertiary-butyl catechol (TBC) is injected into the overhead of the SM column, and the column is operated under vacuum to minimize polymer formation. A unique feature of the Lummus=UOP Classic SM process is the noncompressive azeotropic heat recovery option.[8] In this option, the EB=SM splitter overhead vapor is used to boil an EB–water azeotrope mixture, which is then fed to the dehydrogenation reactors. The condensation of the splitter overhead vapor produces approximately 500 kcal=kg styrene. This energy savings potential makes the azeotropic heat recovery option economically attractive, in particular, in regions with moderate to high steam costs. Lummus/UOP Smart SMTM Process The Lummus=UOP Smart SM process is based on an oxidative reheat technology invented by UOP.[9] Although this technology can be used in the design of

Styrene

2865

S

Fig. 4 Lummus=UOP smart SM process dehydrogenation section.

a grassroots plant, it is most commonly used in a revamp of an existing plant to increase styrene production by as much as 60% with minimal capital investment cost. The Lummus=UOP Smart SM technology uses a specially designed reactor that contains two concentric catalyst zones. A cross-sectional view of the concentric oxidation and dehydrogenation catalyst beds is also shown in Fig. 4. In the first zone, hydrogen is selectively oxidized across a noble metalcontaining catalyst. The direct combustion of hydrogen reheats the reaction mixture, which is directly fed into the second zone where the standard EB dehydrogenation reaction occurs. In addition to providing the full reheating requirement, another benefit of this technology is it shifts the reaction equilibrium in a favorable direction by removing the hydrogen byproduct. This shift in equilibrium allows for higher EB conversion without a corresponding decrease in styrene yield.

The Lummus=UOP Smart SM technology was first commercialized in 1995 at Mitsubishi Chemical in Kashima, Japan. The Mitsubishi Chemical plant was designed with a dehydrogenation section containing two combination oxidation–dehydrogenation reactors as shown in Fig. 4. The temperature rise in the oxidation zone is proportional to the amount of oxygen reacted across the catalyst bed. The oxygen is diluted in steam and the oxygen=steam mixture is well mixed to ensure the reaction mixture remains outside the flammability envelope at all times.

Fina/Badger Styrene Process The Fina=Badger styrene process has evolved through many generations. The most recent design uses a flow diagram as shown in Fig. 5. Recycled and fresh EB

Fig. 5 Fina=Badger styrene process.

2866

Styrene

Fig. 6 Fina=Badger styrene process dehydrogenation section.

are mixed with steam and fed to the primary and the secondary dehydrogenation reactors. The reactor effluent is condensed and separated into vent gas, condensate, and hydrocarbon. The vent gas, the majority of which is hydrogen, is used as fuel gas. The condensate is stripped and used as feed water for steam generation. The hydrocarbon portion of the reactor effluent is fed to the distillation section of the plant, which consists of three distillation columns. The main types of equipment in the dehydrogenation section of the plant are the steam superheater, the primary and secondary dehydrogenation reactors, and a series of feed=effluent exchangers (Fig. 6). High pressure steam is also generated by the recovery of heat from the reactor effluent stream. The major portion of steam is superheated and used to reheat the reaction mixture for the secondary dehydrogenation reactor. As the cooled steam exits the reheater it is superheated again in the steam superheater, prior to being fed to the primary dehydrogenation reactor. The dehydrogenation reactors are designed to provide low pressure drop and uniform flow distribution. The reactor effluent is cooled in a series of three heat exchangers that heat the EB and steam feed to the reactors and generate steam. The Fina=Badger distillation section consists of three distillation columns. All the columns are designed to operate under vacuum to minimize temperature and polymer formation. The first column in the sequence splits the benzene and toluene byproducts from the unconverted EB and styrene product. The benzene and toluene mixture is typically sent to an integrated EB plant where it is further fractionated. In this case, the benzene by-product is ultimately consumed in the EB unit and the toluene becomes a by-product stream from the EB plant.

The EB recycle column separates the unconverted EB for recycle to the dehydrogenation reactors. Recent EB recovery columns use high efficiency packing to obtain minimum pressure drop through the column. This allows the column bottoms’ temperature to be maintained below 100 C. This is an important aspect of the design as styrene polymerization becomes significant at temperatures higher than approximately 100 C. The EB recovery column bottoms’ stream is fed to a finishing column where the styrene is purified by the removal of any heavy residue. Tertiary-butyl catechol is injected into the overhead of the finishing column to prevent polymerization. Tertiary-butyl catechol is widely used to prevent styrene polymerization during storage. In 1997, Fina=Badger joined with Shell Technology Ventures, a subsidiary of Shell Oil Company, to develop a reheating technology called Flameless Distributed Combustion (FDC) for application in EB dehydrogenation.[10] Flameless Distributed Combustion technology is patented by Shell Oil Company and was originally used as a heat injector for enhanced recovery of hydrocarbons from subterranean formations. Flameless Distributed Combustion technology enables specific constraints in the conventional dehydrogenation system to be overcome, in particular designing for low steam-to-oil ratios. A low steam-tooil ratio is desirable because of the substantial energy savings associated with superheating less steam. However, a practical lower steam-to-oil ratio limit exists due to the metallurgy of the steam superheater, steam transfer lines, and interstage reheater. Flameless Distributed Combustion allows for operation at molar steam-to-oil ratios less than 7 : 1 without a costly metallurgy upgrade. This is accomplished by heating the reaction mixture more directly through a combustion and convective heat transfer process.

Styrene

2867

Flameless Distributed Combustion technology, unlike the Lummus=UOP Smart SM technology, does not directly combust hydrogen from the reaction mixture; hence it does not obtain the benefit of a favorable shift in equilibrium.

identified catalyst formulations that provide more than 90% conversion of 4-vinylcyclohexene with approximately 92% selectivity to styrene.[11]

Other Processes

Preventing polymerization is the key to successful styrene storage. Special handling and storage procedures are required to maintain the styrene product quality and to avoid a potentially dangerous situation involving uncontrolled polymerization. During storage, styrene polymerization is prevented by maintaining low temperature and using an appropriate polymerization inhibitor. The industry standard styrene storage inhibitor is TBC and is typically used at concentrations between 10 ppm and 15 ppm. To be effective, TBC requires dissolved oxygen to be present in concentrations roughly equal to the TBC concentration. In addition to adding TBC inhibitor, maintaining the styrene at the lowest practical temperature is critical to preserving product quality. Styrene storage facilities are generally maintained at temperatures below about 20
C, which allows for storage times of around 10 weeks. Even a 5
C increase in the storage temperature to 25
C can reduce the storage time to less than 4 weeks.[12] Tertiary-butyl catechol is added occasionally during storage to maintain the concentration in the desired range.

Propylene oxide/styrene process Aside from EB dehydrogenation, the only other commercial-scale production of styrene is through a propylene oxide=styrene process that produces roughly 15% of worldwide styrene. This technology was developed as an alternative to the chlorohydrin method for producing propylene oxide. Styrene from Butadiene Because the conventional EB dehydrogenation technologies are relatively mature, there is little room for significant additional reduction in production costs. This situation has motivated a lot of research toward using alternative, lower cost feedstocks for styrene production. One area that has been examined involves a two-step process to convert butadiene to styrene. The first step of the process involves the cyclodimerization of butadiene to 4-vinylcyclohexene. The reaction is exothermic and can be catalyzed by either a copper-containing zeolite catalyst or an iron dinitrosyl chloride catalyst complex. Although both vapor-phase and liquid-phase processes have been studied, it appears that liquid-phase reactions are preferred because they achieve higher butadiene conversion levels. The second step is oxidative dehydrogenation of the 4-vinylcyclohexene to produce styrene. Dow has led the research effort in this area and has

Storage

ECONOMICS The cost of styrene production can be broken down into three main components: raw materials, utilities, and the fixed cost associated with the plant. The utilities cost includes fuel, electricity, steam, cooling water, catalyst, and chemical costs required to

Table 2 Styrene economics for conventional EB dehydrogenation process UNIT

Quantity UNIT/MT

Price $/UNIT

Cost $/MT

Produce Styrene

MT

1.0000

751

751.0

Raw materials Ethylene Benzene

MT MT

0.2912 0.7898

629 453

183.2 357.8

By-product credits Toluene Light ends

MT MT

0.0401 0.0401

378 289

(15.1) (11.6)

Net feedstock costs Utilities Fixed cost Total cost of production Basis: North America, 2003

514.2 95.0 35 644.2

S

2868

Styrene

Fig. 7 Distribution of styrene production cost components.

operate the plant. The major cost components for styrene production using conventional adiabatic dehydrogenation process are listed in Table 2. The major cost of production is for the ethylene and benzene raw materials, which account for approximately 80% of the

total cost of production. The benzene cost is the largest cost component; hence, the economics of styrene production are highly dependent on benzene price. The raw materials cost has two components—one dictated by the stoichiometry and the other caused by

Table 3 Styrene economics for propylene oxide-styrene process UNIT

Quantity UNIT/MT

Price $/UNIT

Product Styrene

MT

1.0000

751

751.0

Raw materials Ethylene Benzene Propylene Oxygen

MT MT MT MT

0.3135 0.8194 0.3541 0.2529

629 453 465 43

197.2 371.2 164.6 10.9

By-product credits Light ends Propylene oxide

MT MT

0.1000 0.4500

289 1227

(28.9) (552.1)

Tars

MT

0.0400

257

(10.3)

Net feedstock costs

Cost $/MT

152.6

Utilities

65.0

Fixed cost

95

Total cost of production Basis: North America, 2003

312.6

Styrene

yield losses occurring as a result of the process technology. If the unalterable stoichiometric raw material consumption is removed from the cost of production, the resultant distribution of cost components appears very different, as shown in Fig. 7. From this perspective, the raw materials’ cost is only about 15% of the incremental cost of production and the utilities and fixed costs become dominant. Recent catalyst and process design improvements have reduced the variable costs of styrene production, while ever-increasing complexity and more stringent regulations have greatly increased the fixed costs. Other recent trends, such as larger plant capacities and globalization of the styrene market, have also resulted in higher fixed costs.[13] The result of the shift of focus from variable to fixed costs is that plants are being designed for larger capacities. For example in 2003, typical new styrene plants in the Asia Pacific Region produced an average of 350 KMTA styrene per year, nearly double the capacity of typical plants started up just 5 years earlier. The drive to reduce fixed costs has led to numerous revamps of existing plants to substantially increase capacity. In many cases, capacity expansions on the order of 50% are being implemented. The propylene oxide=styrene process, the only other commercial process for production of styrene, is a growing influence on the overall styrene market economics. When viewed from the perspective that styrene is the primary product and propylene oxide is a by-product, the economics of this process appear encouraging (Table 3). Depending on the credit value assigned to the propylene oxide coproduct, the total cost of styrene production can be approximately 50% of conventional EB dehydrogenation technology. Approximately 33% of the styrene capacity added between 1998 and 2003 was produced using propylene oxide=styrene technology. More recently, the trend appears to be reversing and propylene oxide=styrene processes are accounting for less of newly installed capacity. Although propylene oxide=styrene plants are built to produce propylene oxide, there is a profound impact on the styrene market supply=demand balance.

CONCLUSIONS Since the first commercial-scale production in the 1930s, styrene, mainly through its derivatives, has become an integral part of life. Most people come in contact with numerous styrene-based products throughout the course of a normal day. Demand for styrene is expected to continue growing at a rate comparable to the gross domestic product growth rate. The chemical processing technologies that have been developed are sophisticated, producing styrene to meet the demand at low cost. Research and development

2869

efforts are aimed at further improvements in existing technologies and identification of new technologies for styrene production opportunities.

REFERENCES 1. Lidback, A. Styrene—This is Not a Drill, 2004 World Petrochemical Conference, Chemical Marketing Associates, Inc.: Houston: TX, March 23–25, 2004. 2. Carra, S.; Forni, L. Kinetics of catalytic dehydrogenation of ethylbenzene to styrene. Ind Eng Chem Process Des. Dev. 1965, 4 (3), 281–285. 3. Matsui, J.; Sodesawa, T.; Nozaki, F. Influence of carbon dioxide addition upon decay of activity of a potassium-promoted iron oxide catalyst for dehydrogenation of ethylbenzene. Appl. Catal. 1991, 67, 179–188. 4. Hirano, T. Roles of potassium in potassium-promoted iron oxide catalyst for dehydrogenation of ethylbenzene. Appl. Catal. 1986, 26, 65–79. 5. Kollar, J. Epoxidation Process. US Patent 3,351,635, November 7, 1967. 6. Chem Systems, Propylene Oxide 97=98–7, Tarrytown, New York, 1998. 7. Mullen, P. Enhancements in EB=SM technology. In AIChE 2003 Spring Meeting Proceedings; AIChE Spring National Meeting. Houston: TX, April 22–26, 2001. 8. Sardina, H. Dehydrogenation Process for Production of Styrene from Ethylbenzene Comprising Low Temperature Heat Recovery and Modification of the Ethylbenzene–Styrene Feed therewith. US Patent 4,628,136, December 9, 1986. 9. Imai, T. Dehydrogenation of Dehydrogenatable Hydrocarbons. US Patent 4,435,607, March 6, 1984. 10. Welch, V. Advanced styrene dehydrogenation with flameless distributed combustion. In AIChE 2003 Spring Meeting Proceedings; AIChE Spring National Meeting. Houston: TX, April 22–26, 2001. 11. Chem Systems PERP Report, Styrene from Butadiene, 93S3, Tarrytown, New York, 1995. 12. Technical Bulletin on Safe Handling & Storage of Styrene Monomer http:==www.sterlingchemicals. com=SCI=WEBSITE=scihome.nsf=(WebContentByDocID)=0AF6F53F1881D0AA8626CD90082 E36A?OpenDocument. 13. Ram, S. EB-SM splitter energy recovery options. In Styrene Conference General Session, Styrene Conference, Prague, Czech Republic, June 22– 25, 2003; ABB Lummus Global, Sud-Chemie AG, and UOP LLC, 2003.

S

Styrene–Butadiene Rubber S Jing Peng Department of Applied Chemistry, College of Chemistry, Peking University, Beijing, People’s Republic of China

INTRODUCTION

Synthesis

Styrene–butadiene rubber (SBR) is a random polymer made from butadiene and styrene monomers. It possesses good mechanical property, processing behavior, and can be used like natural rubber. Moreover, some properties such as wear and heat resistance, aging, and curing property are even better than in natural rubber. Styrene–butadiene rubber was the first major synthetic rubber to be produced commercially. Now it has become the most common rubber with the largest production and consumption in the synthetic rubber industry. It can be widely used in tire, adhesive tape, cables, medical instruments, and all kinds of rubberware. Styrene–butadiene rubber could be produced by using emulsion and solution process, thus it can be divided into emulsion-polymerized styrene–butadiene rubber (E-SBR) and solution-polymerized styrene– butadiene rubber (S-SBR). In this entry, we will describe their development and introduce their synthesis process, relationship between structure and property, processing property, blends, and applications.

The emulsion polymerization process has several advantages. It is normally carried out under mild reaction conditions that are tolerant to water in the absence of oxygen. The process is relatively resistant to impurities and amenable to using a range of functionalized and nonfunctionalized monomers. Additional benefits include the fact that emulsion polymerization gives high solid contents with low reaction viscosity and is a cost-effective process. The physical state of the emulsion (colloidal) system makes it easy to control the process. Thermal and viscosity problems are much less significant than in bulk polymerization. Table 1 shows the raw materials required in the polymerization of E-SBR, which include monomers (styrene and butadiene), water, emulsifier, initiator system, modifier, terminal agent, and a stabilizer system. The original polymerization reactions were carried out in batch reactors in which all the ingredients were loaded to the reactor and the reaction was terminated after it had reached the desired conversion. Current commercial productions are run continuously by feeding reactants and polymerizing through a chain of reactors before terminating at the desired monomer conversion. The monomers are continuously metered into the reactor chains and emulsified with the emulsifiers and catalyst agents. In cold polymerization, the most widely used initiator system is the redox reaction between chelated iron and organic peroxide using sodium formaldehyde sulfoxide (SFS) as a reducing agent [see Eqs. (1) and 2]. In hot polymerization, potassium persulfate is used as an initiator.

EMULSION-POLYMERIZED STYRENE– BUTADIENE RUBBER History In the 1930s, I. G. Farbenindustrie in Germany prepared the first E-SBR known as Buna S. The American government in 1940 established the Rubber Reserve Company to start storage of natural rubber and a synthetic rubber program. These programs were expanded when the United States entered World War II. These E-SBR grades were called GR-S (government rubber styrene). Initially, the synthesis of E-SBR was focused on a hot polymerized (41
C) E-SBR. Production of a 23.5% styrene and 76.5% butadiene copolymer began in 1942. Cold polymerized E-SBR (5
C), which has significantly better physical properties than hot polymerized SBR, was developed in 1947. Thereafter, the oil extended E-SBR was produced in 1951. Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120019881 Copyright # 2006 by Taylor & Francis. All rights reserved.

FeðIIÞEDTA þ ROOH ! FeðIIIÞEDTA þ RO þ OH

ð1Þ

FeðIIIÞEDTA þ SFS ! FeðIIÞEDTA

ð2Þ

Mercaptan is added to provide free radicals and to control the molecular weight distribution by terminating existing growing chains while initiating a new chain. The thiol group acts as a chain transfer agent to prevent the molecular weight from attaining the 2871

2872

Styrene–Butadiene Rubber

Table 1 Typical recipe for SBR emulsion polymerization Parts by weight Component

Cold

Hot

Styrene

25

25

Butadiene

75

75

Water

180

180

Emulsifier (FA, RA, MA)

5

5

Dodecyl mercaptan

0.2

0.8

Cumene hydroperoxide

0.17

—

FeSO4

0.017

—

EDTA

0.06

—

Na4P2O710H2O

1.5

Potassium persulfate

0.3

SFS

0.1

—

Stabilizer

Varying amount

Varying amount

FA, fatty acid; RA, rosin acid; MA, rosin acid=fatty acid. (From Ref.[1].)

excessively high values possible in emulsion systems. The sulfur–hydrogen bond in the thiol group is extremely susceptible to attack by the growing polymer radical and thus loses a hydrogen atom by reacting with polymer radicals [Eq. (3)]. The RS formed will continue to initiate the growing of a new chain as shown in Eq. (4) below. The thiol prevents gel formation and improves the processability of the rubber.

aluminum sulfate, or amine coagulation aid. The type of coagulation system is selected depending on the end use of the product. Sulfuric acid=sodium chloride is used for general purposes. Glue=sulfuric acid is used for electrical grade and low water sensitivity SBR. Sulfuric acid is used for coagulations where low-ash polymer is required. Amine coagulating aids are used to improve coagulation efficiency and reduce production plant pollution. The coagulated crumb is then washed, dewatered, dried, baled, and packaged. On the other hand, due to the difference in producing E-SBR, E-SBR grades have different properties. Hot emulsion polymerization is the original SBR process. The major characteristic of this process is that these grades have excellent processing behavior in terms of low mill shrinkage, good dimensional stability, and good extrusion characteristics. However, high content of microgels is also produced in hot polymerization; therefore, there is a trend toward using cold emulsion grades in many applications. Nevertheless, hot rubbers are still used in applications such as adhesives and flow modifiers for other elastomers where good flow properties are required.

Properties of E-SBR




P
þ RSH ! PH þ RS


ð3Þ

RS
þ M ! RSM


ð4Þ

During polymerization, parameters such as temperature, flow rate, and agitation speed must be controlled carefully to get the right conversion. Polymerization is normally allowed to proceed to about 60% conversion in cold polymerization and 70% in hot polymerization before it is stopped with a terminal agent that reacts rapidly with the free radicals. Common terminal agents include sodium dimethyldithiocarbamate and diethyl hydroxylamine. Once the latex is properly terminated, the unreacted monomers are removed from the latex. Butadiene is stripped by degassing the latex by means of flash distillation and reduction of system pressure. Styrene is removed by steam stripping the latex in a column. The latex is then stabilized with the appropriate antioxidant and transferred to blend tanks. In the case of oil-extended polymers or carbon black master batches, these materials are added as dispersions to the stripped latex. The latex is then transferred to finishing lines to be coagulated with sulfuric acid, sulfuric acid=sodium chloride, glue=sulfuric acid,

E-SBR is commercially available in Mooney viscosities ranging from 30 to about 120 [ML (1 þ 4) 125 C]. Lower Mooney viscosity E-SBR grades adhere more easily to the mill, incorporate fillers and oil more readily, show less heat generation during mixing, and are calendered more easily, shrink less, give higher extrusion rates, and have superior extrudate appearance than the higher Mooney viscosity grades. On the other hand, the high Mooney viscosity SBR grades have better green strength, less porosity in the vulcanizate, and accept higher filler and oil loadings. As the molecular weight of the SBR increases, the vulcanizate resilience and the mechanical properties, particularly tensile strength and compression set, improve. The processability of SBR can be improved as its molecular weight distribution broadens. However, the formation of high molecular weight fractions with the increase in the average molecular weight can prevent improvements in the processability. This is due to the fact that the tendency for gel formation also increases at higher molecular weights. In addition to the polymer viscosity, polymerization temperature also plays an important role in shaping the processability. Emulsion-polymerized SBR grades produced at low polymerization temperatures have less chain branching than those produced at higher temperatures. At an equivalent viscosity, cold polymerized E-SBR is normally easier to process than hot polymerized E-SBR, and this applies particularly

Styrene–Butadiene Rubber

to a better banding on mills, less shrinkage after calendering, and a superior surface of green tire compounds. Hot rubbers give better green strength because they have more chain branching. The styrene content of most emulsion SBR varies from 0% to 50%. The percentage of styrene in most commercially available E-SBR grades is 23.5%. In vulcanizates of SBR, as styrene content increases, dynamic properties and abrasion resistance decrease while traction and hardness increase. Polymerization temperature also affects the microstructure of E-SBR. In the cold polymerized E-SBR grades, the butadiene component has, on average, about 9% cis-1.4, 54.5% trans-1.4, and 13% of vinyl-1.2 structure. At the 23.5% styrene level, the glass transition temperature, Tg, of SBR is about –50
C. As the styrene content in the SBR increases, the glass transition temperature also increases. Rubbers with very low Tg values posses a high resilience and good abrasion resistance, but have poor wet traction. By contrast, those rubbers with high Tg (for instance, SBR 1721) exhibit a low resilience and poor abrasion resistance with an excellent wet traction. The emulsifier remains in the rubber after coagulation and can also have an influence on the processability. Rosin acid emulsifiers impart better knitting, tack, and adhesion to the SBR polymer. Generally, polymers emulsified with rosin acid have better extrusion rates, slower cure rates, poorer heat resistance, and can cause mold fouling and polymer discoloration. Fatty acid emulsified SBR polymers generally have less tack, faster curing, and high tensile properties. A compromise of the above properties is obtained by using a mixed rosin acid=fatty acid emulsifier system. Chemical Activity of E-SBR It has been proved that incorporation of carboxylic acid groups in the polymeric chain has a significant effect on colloidal properties of latex, processability, and end-use property. Carboxylated styrene–butadiene latexes (XSBR) are prepared via batch emulsion copolymerization with different amounts of acrylic acid in the absence of emulsifier. They are among the most important polymeric colloids, and can be used as binder in paper coatings, carpet backing, paints, and nonwoven. There are several studies on the preparation and properties of XSBR latexes. To improve the aging property of E-SBR, Parker and Roberts used the diimide reduction method to prepare hydrogenated E-SBR. In the system containing hydrazine hydrate, oxidant, and a metal-ion catalyst, hydrogenated E-SBR with 97% of hydrogenation could be obtained.[2] The hydrogenated E-SBR not only exhibited excellent ozone, oxidation, and UV resistance as expected, but also showed better mechanical properties in some circumstances than unhydrogenated E-SBR.

2873

Moreover, E-SBR can be grafted with some polar monomers such as acrylic acid or organic chlorium to give modified E-SBR with good heat resistance and tensile strength. Cure Properties and Processing Styrene–butadiene rubber can be cured with a variety of cure systems including sulfur (accelerators and sulfur), peroxides, and phenolic resins. In addition, some papers have reported that the SBR could be vulcanized using gamma irradiation in the presence of polyfunctional monomers. The efficiencies of seven functional monomers toward radiation vulcanization of SBR are in the following order: tertamethylol methane tetraacrylate (ATMMT) ¼ toluene diisocyanate (UA306T) > trimethylol propane trimethacrylate (TMPT) > diethylene glycol dimethacrylate (2G) > dipentaerthritol hexaacrylate (DPE6A) > hexamethylene diisocyanate (UA101H) > triallyl cyanurate (TAC).[3] Processing of SBR compounds can be performed in a mill, internal mixers, or mixing extruders. Styrene– butadiene rubber compounds are cured in a variety of ways by compression, injection molding, hot air or steam autoclaves, hot air ovens, microwave ovens, and combinations of these techniques. Types of E-SBR There are a large variety of E-SBR types based on the styrene content, polymerization temperature, antioxidants, oil and carbon black content. Each of these basic classifications includes a variety of SBR polymer variations with respect to Mooney viscosities, coagulation types, emulsifier type, oil levels, and carbon black types and levels. Table 2 shows the basic series of E-SBR.

Table 2 The types of E-SBR Series

Comments

1000

Hot polymerized polymers

1500

Nonextended cold polymerized polymers

1600

Non-oil-extended cold carbon black masterbatches

1700

Cold oil-extended polymers

1800

Cold oil-extended carbon black masterbatches

1900

Miscellaneous high styrene resin masterbatches

(From Ref.[1].)

S

2874

Styrene–Butadiene Rubber

SOLUTION-POLYMERIZED STYRENE– BUTADIENE RUBBER

Applications of E-SBR Emulsion-polymerized SBR is predominantly used for the production of car and light truck tires and truck tire retread compounds. A complete list of the uses of SBR includes houseware mats, drain board trays, shoe soles and heels, chewing gum, food container sealants, tires, conveyor belts, sponge articles, adhesives and caulks, automobile mats, brake and clutch pads, hose, V-belts, flooring, military tank pads, hard rubber battery box cases, extruded gaskets, rubber toys, molded rubber goods, shoe soling, cable insulation and jacketing, pharmaceutical, surgical, and sanitary products, food packaging, etc. The typical applications of E-SBR polymers are presented in Table 3.

History During the 1960s, Phillips produced the first solutionpolymerized random SBR grades, which are named as Solprene X-40, commercially. In 1969, Firestone produced the commercial S-SBR grades named as Duradone. The original aim to produce SBR with lower styrene content than products made by emulsion polymerization is to counteract the increase in styrene price. Standard S-SBR grades now have comparable styrene content to emulsion types. These grades have superior mechanical properties to E-SBR.

Table 3 Typical applications of E-SBR polymers Cold E-SBR

Hot E-SBR grades

Unextended

Oil extended
















High styrene master batch

Black master batch

Adhesives Type and label Caulking Laminating Mastic Panel Pressure sensitive Sealant Sprayable (cross-linked) Wall tile Automotive













Tire treads Apex=rim=flange Bead Carcass Retread Racing tires













Miscellaneous Rolls




Belts=hoses Hard rubber goods Floor tiles Footwear Sponge Wire and cable (low ash) (From Ref.[1].)































































Gaskets

Cove base







Mats Mechanical goods









































Styrene–Butadiene Rubber

In 1983, the high trans-SBR grades were synthesized in America. Solution-polymerized SBR grades have superior mechanical properties, particularly tensile strength, low rolling resistance, and handling, when used in tire applications. The ratio of butadiene configurations varies. Generally speaking, S-SBR grades have a lower trans and vinyl and a higher cis butadiene content than E-SBR grades. In initially making S-SBR grades, producers attempted to replicate the stereochemistry of E-SBR grades. However, solution polymerization differs from emulsion polymerization because of its flexibility and enables SBR grades with varying styrene–butadiene ratios and cis, trans, and vinyl contents to be produced by changing the catalyst and monomer ratios and reaction conditions. This enables S-SBR producers to synthesize grades specifically tailored for individual applications.

Synthesis Solution-polymerized SBR is made by terminationfree, anionic=live polymerization initiated by alkyl lithium compounds. Other lithium compounds are suitable (such as aryl, alkaryl, aralkyl, tolyl, xylyl lithium, and a=b-naphtyl lithium as well as their blends), but alkyl lithium compounds are the most commonly used in industry. The absence of a spontaneous termination step enables the synthesis of polymers possessing a very narrow molecular weight distribution and less branching. Carbon dioxide, water, oxygen, ethanol, mercaptans, and primary=secondary amines interfere with the activity of alkyl lithium catalysts, so the polymerization must be carried out in clean, nearanhydrous conditions. Stirred bed or agitated stainless steel reactors are widely used commercially. Polymerization is carried out in a solution of inert aliphatic or aromatic solvent. The polymerization rate of butadiene in the presence of lithium-based catalysts is lower than styrene. However, when butadiene and styrene monomers are mixed, the rate of polymerization is reversed, resulting in the production of block copolymer with a high proportion of butadiene blocks. Block formation must be suppressed because the property requirements of traditional SBR markets cannot be met by block copolymers. Random copolymerization is encouraged by incorporating into the solution ‘‘randomizing’’ agents such as dialkyl and heterocyclic ethers, which act as a Lewis base on the catalyst, or by controlled monomer charging (i.e., some of the styrene is added later in the polymerization cycle). The resulting copolymer is precipitated, separated, dried, and baled. By adjusting the reactivity of the initiator=modifier system toward metallation of the polymer backbone, Kerns and Henning have developed synthetic strategies

2875

that effectively control the relative level of branching and polydispersity in S-SBR. This development has allowed for the optimization of both the microstructure and the macrostructure of solution elastomers to meet the demands of a given application. They have taken the approach to control the polymer macrostructure by adjusting the initiator=modifier system so as to mediate the propensity for backbone metallation to occur. Several synthetic strategies were employed to impose control over the levels of branching in solution SBR grades. They found that the macrostructure is the independent variable and subtle differences can have profound effects on the rheology and, thus, mixing, extruding, and even physical properties of the resultant compound.[4] Compared with the E-SBR, S-SBR has different structures and properties. Some typical characteristics of S-SBR grades are shown in Table 4. Properties of S-SBR The optimization of property has been achieved by conventional solution SBR technology. By modifying the way in which monomers are added, the polymerization conditions, the use of cocatalysts and randomizing agents, the proportion of cis and vinyl isomers, the chain structure of the resulting ‘‘tailored’’ polymer can be varied. Generally, the Mooney viscosity of S-SBR is higher than E-SBR. Thus, it accepts higher filler and oil loadings. The effect of composition on the property of S-SBR is similar to that of E-SBR. As the styrene content increases, the rolling resistance, traction, and hardness increase, while the wear resistance decreases. With the increase of vinyl content, the wet traction increases. Mechanical Property Styrene–butadiene rubber is generally compounded with a vulcanization system, reinforcing filler (usually carbon black), processing=extending oil, and an antioxidant=stabilizer package prior to molding=fabrication. Standard S-SBR grades have comparable styrene content to emulsion types. These grades have superior mechanical properties to E-SBR. The typical mechanical properties of some commercial S-SBR grades are listed in Table 5. Cerveny investigated the development of the dynamic glass transition in styrene–butadiene copolymers by dielectric spectroscopy in the frequency range from 102 to 106 Hz. Two processes were detected and attributed to the alpha- and beta-relaxations. The alpha relaxation time has a non-Arrhenius temperature behavior that is highly dependent on styrene content

S

2876

Styrene–Butadiene Rubber

Table 4 The typical characteristics of some S-SBR grades Types S-SBR Low vinyl-1,2 structure

Medium vinyl-1,2 structure

High vinyl-1,2 structure

Tufdene 2000R

NS 118

Solprene 1204

SL 552

NS 114

SL574

NS110

E-SBR 1500

Initiator

Alkyl lithium

Alkyl lithium

Alkyl lithium

Alkyl lithium

Alkyl lithium

Alkyl lithium

Alkyl lithium

Redox reaction

Styrene content (%)

25

17.5

25

24

23

15

12

23.5

11.5 Two peaks

24 40.5 35.5 Two peaks

20 40 30 Two peaks

38 Two peaks

16 27 57 Two peaks

72.5 Two peaks

75

50

64

53

55

28

Entry

Structure of butadiene cis-1,4 (%) trans-1,4 (%) vinyl-1,2 (%) Molecular weight distribution

35 52 13 Narrow

Glass transition temperature (
C)

70

while the beta relaxation time shows an Arrhenius behavior with an activation energy that is independent of styrene-content. Furthermore, the shape of the alpha-relaxation is strongly influenced by the styrene content while the shape of beta-relaxation is independent of styrene content. Someone interprets these results as follows. The observed beta-relaxation is primarily due to local motions of butadiene monomers and is therefore not affected by the presence of styrene. The alpharelaxation, on the other hand, is highly sensitive to the styrene content owing to its cooperative character.[6] Chemical Activity of S-SBR Using chemical modification such as cyclization, chlorination, hydrogenation, and epoxidation, the different S-SBR derivatives could be obtained.

Tensile strength (MPa)

Stress at 300% elongation (MPa)

Elongation at break (%)

SE SLR-4601

19

15

350

SE SLR-4400

22

13

430

SE SLR-4610

19

9

550

SE SLR-4610H

21

9.4

570

SE SLR-4630

20

10.5

540

SE SLR-6410

23

9

630

(From Ref.[5].)

60

Controlled cyclization of SBR was achieved with the aid of cationic catalyst system based on diethylaluminum chloride (AlEt2Cl) and benzyl chloride (C6H5CH2Cl) and by working in xylene solution at high temperature (T > 100
C). Elastomers with low intrinsic viscosity, ready solubility, and free gel were produced by Wang et al. The cyclized products have been expected to be photoresists with high photosensitivity.[7] Dichlorocarbene-modified SBR prepared by the alkaline hydrolysis of chloroform using cetyltrimethylammonium bromide as a phase-transfer agent resulted in a product that showed good mechanical properties, excellent flame and solvent resistance, and good thermal stability. The molecular weight of the polymers, determined by gel permeation chromatography, showed that chemical modification was accompanied by an increase in molecular weight. Proton nuclear magnetic

Table 5 Typical mechanical properties of S-SBR grades Product

12 68.5 19.5 Wide

Styrene–Butadiene Rubber

resonance and Fourier transform infrared (FTIR) studies revealed the attachment of chlorine through cyclopropyl rings to the double bond of butadiene.[8] Because of the existence of unsaturation bonding in the SBR backbone, the material is susceptible to degradation under oxygen and ozone atmosphere. This drawback can be overcome by hydrogenating the copolymer to give hydrogenated styrene–butadiene copolymer. The hydrogenation of a copolymer of styrene and butadiene (SBR) catalyzed by Ru(CH¼CHPh) Cl(CO)(PCy3)2 was experimentally investigated within the temperature range of 120–160
C at PH2 of 300–1200 psi, and a catalyst concentration of 1.0– 7.8  105 M. Special attention was paid to minimizing the catalyst metal residue and cross-linking in the product. The results indicated that high-quality hydrogenated SBR could be achieved without cross-linking and the metal residue was less than 7 ppm without posttreatment.[9] In addition, the hydrogenated SBR under certain conditions is more stable than its unsaturated precursors. They can be used as model compounds for predicting the unperturbed dimensions of polymer chains. Cure Properties and Processing Compared with E-SBR, the curing rate of S-SBR grades is faster by about 10–20%. Generally, the processing of S-SBR compounds is the same as for E-SBR. The difference is that the molecular chain of S-SBR cannot be broken in the internal mixer with high power; therefore, all the components can be added into the mixer simultaneously without precalendering. Usually, the mechanical strength of raw S-SBR is lower than that of E-SBR; it is necessary to reduce the distance between two rolls by 10–20% to avoid the peeling off of the roll. On the other hand, S-SBR bonds easily on the cold roll. It will be better to ensure that the temperature difference between two rolls is about 5–10
C during the processing of S-SBR. Applications of S-SBR S-SBR grades have excellent balance between wet traction and rolling resistance; therefore, they can be used for low fuel consumption tire treads in all-season tires and high-quality rubber goods. In applications of S-SBR, carbon black and silica will often be added to enhance the property. Blending Rubber goods usually require a combination of properties that cannot be provided by one elastomer only and two or more polymer components have to be

2877

mixed to meet specific requirements such as lowering the compound cost, for ease of fabrication and to improve the performance of the industrial rubber. Natural rubber (NR) and SBR have been blended for a long time for these reasons. The tensile, hardness, and wear properties of the NR=SBR blends with the increment of NR percentages were greatly enhanced.[10] For the system of natural rubber (NR) and styrene butadiene rubber (SBR) in a 1:1 ratio, the best properties were found when the NR=SBR mixture was prepared in such a way as to favor the vulcanization of the SBR phase while preserving the NR phase from excessive vulcanization.[11] The morphology and mechanical and viscoelastic properties of a series of blends of NR and SBR latex blends were studied in the uncross-linked and crosslinked states. The morphology of the blends indicated a two-phase structure in which SBR is dispersed as domains in the continuous NR matrix when its content is less than 50%. A cocontinuous morphology was obtained at a 50=50 NR=SBR ratio and phase inversion was seen beyond 50% SBR when NR formed the dispersed phase. As the NR content and time of prevulcanization increased, the mechanical properties such as the tensile strength, modulus, elongation at break, and hardness increased. This was due to the increased degree of cross-linking that leads to the strengthening of the three-dimensional network. In most cases the tear strength values increased with the prevulcanization time. The effects of the blend ratio and prevulcanization on the dynamic mechanical properties of the blends were investigated at different temperatures and frequencies. All the blends showed two distinct glass transition temperatures, indicating that the system is immiscible. It was also found that the glass transition temperatures of vulcanized blends are higher than those of unvulcanized blends.[12] Styrene butadiene rubber=epoxidized natural rubber (ENR) blends were prepared with an internal mixer, Haake Rheomix. The scorch time, t2, and curing time, t90, were found to decrease with increasing ENR composition in the blends. The mechanical properties such as tensile strength, tear strength, and tensile modulus, M300 (modulus at 300% elongation), increase with increasing ENR composition in the blends. However, elongation at break shows an opposite trend. At similar immersion times, SBR=ENR blends containing a higher ENR content exhibit better oil resistance.[13] The mixing of incompatible polymers such as polyethylene terephthalate (PET) and SBR produces a blend with poor mechanical and impact properties, because polymeric phases interact weakly with each other and segregate. The use of SBR grafted with maleic anhydride (MAH) could increase the compatibility of the SBR–PET system by generating higher interactions and chemical bonds between the ingredients of the

S

2878

blend. The induced compatibility is reflected in the 2.5fold increase in the impact resistance of the blend as compared to that of pure PET. The grafting reaction to produce SBR-g-MAH is carried out by reactive extrusion using a reaction initiator benzoyl peroxide (BPO), and the extent of the reaction depends on the concentration of MAH and BPO. Results indicated the close relationship between processing conditions and microstructural parameters, such as particle diameter and interparticle distances of the dispersed rubber phase, which is necessary to achieve the best impact resistance.[14] The morphology of ternary polystyrene=SBR= polyethylene (PS=SBR=PE) blends has been investigated in the limits of a constant content of the major component (PS: 75 wt%) while changing the weight ratio of the two minor constitutive polymers. A coreshell structure for the dispersed phase has been predicted from the spreading coefficients and observed by transmission electron microscopy. Actually, with the increase in the relative content of PE with respect to SBR, the structure of the dispersed phase changes from a multicore structure to a PE=SBR core-shell morphology. The size of the PE subphase in the mixed dispersed phase increases sharply at a PE content that corresponds to phase inversion in the parent SBR=PE binary blends. The ultimate mechanical properties of these blends are sensitive to the strength of the SBR interphase between PS and PE. Some synergism has been observed in the PE=SBR composition dependence of the tensile strengths at yield and break.[15] Because SBR lacks the self-reinforcing qualities of natural rubber due to stress-induced crystallization, gum vulcanizates of SBR have lower tensile properties. The tensile property of E-SBR vulcanizates depends significantly on the type and amount of filler in the compound. Cured gum stocks have only 2.8–4.2 MPa in tensile strength, while fine particle carbon black loadings can produce tensile strength of 27.6 MPa. Though the compression set of some of the common E-SBR compounds is high, by proper compounding and blending, it is possible to obtain E-SBR vulcanizates with a low compression set. Some studies also indicated that filler affects the physical and mechanical properties of the blends. In such cases, the additives normally employed in rubber formulations are unevenly distributed, depending on the affinity of each compound to each polymeric phase. Thus, the dispersion of each one of these ingredients in the different rubbers will influence the rate and degree of vulcanization and, in consequence, the performance of the final composite. Recently, nanocomposites were prepared with different grades of nitrile rubber with acrylonitrile contents of 19%, 34%, and 50%, with SBR (23% styrene content), and with polybutadiene rubber with Namontmorillonite clay. The clay was modified with

Styrene–Butadiene Rubber

stearyl amine and was characterized by x-ray diffraction (XRD), FTIR spectroscopy, and transmission electron microscopy (TEM). At the degree of filler loading up to a certain level, mechanical properties of nanocomposites could be improved.[16] A new kind of rubber powder with ‘‘salami’’ structure (RPS) was prepared by spray drying the mixture of SBR latex and nano-CaCO3 slurry. It was found that RPS is an effective toughener with synergistic toughening effect on polypropylene (PP). The Izod impact strength of PP=RPS blend is not only higher than that of PP=rubber powder or PP=nano-CaCO3 blends, but also higher than that of a PP=rubber powder=CaCO3 blend. Transmission electron microscopy images showed that the microstructure of the PP=RPS blend is an ‘‘island-sea’’ structure with ‘‘salami’’ structure in RPS, in which nano-CaCO3 particles are embedded in SBR particles. Perhaps this morphology resulted in the improvement of mechanical properties of blends.[17]

Technical Development of SBR In recent years, SBR developments have been predominantly initiated by the tire industry. This is not surprising when several of the SBR producers have put their assets up for sale in recent times such as Polimeri in Europe, DSM in the United States, and others. This is partly due to the commoditization of SBR and also to butadiene price rises, which are difficult to pass for the tire producers. This has created severe pressure on margins for SBR producers. Therefore, the majority of recent developments have been undertaken by the tire companies themselves rather than the SBR producers who were more active in the past. Property optimization of SBR has been achieved, to some extent, by conventional S-SBR technology.[18] The automotive industry is under continued pressure to improve the environmental performance and useful life of automotive components. S-SBR producers are responding to develop SBR grades with optimum combination of rolling resistance, wear resistance, blow-out resistance, chipping=chunking resistance, road traction under a variety of weather conditions, handling, noise transmission, and other performance properties for different tire applications. There are many ways in which some improvements are being achieved:[18]  Reformulating compounds and using highperformance additives in conjunction with ‘‘tailored’’ S-SBR grades.  Developing novel additives=modifiers that can be added to the SBR at the compound stage.  Further modification of polymerization conditions to enable both block and random copolymerization.

Styrene–Butadiene Rubber

 Introduction of postpolymerization steps to obtain the copolymer with end group structure, thus significantly altering the properties of the resulting SBR grades.  Introduction of postpolymerization steps to facilitate better interaction with the reinforcement system. This is one of the most radical developments affecting the rubber industry because it enables silica to significantly displace carbon black as the favored reinforcement for the applications.

CONCLUSIONS With global demand at around 3.7 million tons per year, SBR is one of most important synthetic elastomers. Nearly 70% of SBR is consumed by the automobile industry for tires and tire products. There is growing competition between emulsion SBR and solution SBR grades. S-SBR has superior physical properties (particularly wear resistance) and blendability with other rubbers. However, they have been failing to make significant inroads, owing to processing difficulties and higher price. Therefore, E-SBR grades are used widely in the markets due to their low cost. While S-SBR has been paid more and more attention, people are striving to reduce the cost of preparation of S-SBR. New synthesis procedure and more effective catalysts are required in the future.

REFERENCES 1. http:==www.Azom.com=. 2. He, Y.; Daniels, E.S.; Klein, A.; El-Aasser, M.S. Hydrogenation of styrene-butadiene rubber (SBR) latexes. J. Appl. Polym. Sci. 1997, 64, 2047–2056. 3. Abdel-Aziz, M.M.; Youssef, H.A.; El-Miligy, A.A.; Yoshii, F.; Makuuchi, K. Effect of polyfunctional monomers on radiation vulcanization of styrene-butadiene rubber. J. Elastomers Plast. 1996, 28, 288–305. 4. Kerns, M.L.; Henning, S.K. Synthesis and rheological characterization of branched versus linear solution styrene-butadiene rubber. Rubber Chem. Technol. 2002, 75, 299–308. 5. http:==www.dow.com=. 6. Cerveny, S.; Bergman, R.; Schwartz, G.A.; Jacobsson, P. Dielectric alpha- and beta-relaxations in uncured styrene butadiene rubber. Macromolecular 2002, 35, 337–342.

2879

7. Wang, C.Y.; Huang, X.P.; Yang, J.H. Cationic cyclization of styrene-butadiene rubber. Eur. Polym. J. 2001, 37, 1895–1899. 8. Ramesan, M.T.; Alex, R. Dichlorocarbene modification of styrene-butadiene rubber. J. Appl. Polym. Sci. 1998, 68, 153–160. 9. Pan, Q.M.; Rempel, G.L. Hydroenation of styrene butadiene rubber catalyzed by RU(CH¼CHPh)Cl (CO)(PCy3)2. Macromol. Rapid Commun. 2004, 25 (8), 843–847. 10. Findik, F.; Yilmaz, R.; Koksal, T. Investigation of mechanical and physical properties of several industrial rubbers. Mater. Des. 2004, 25 (4), 269–276. 11. Visconte, L.L.Y.; Martins, A.F.; Suarez, J.C.M.; Nunes, R.C.R. Different preparative modes for the incorporation of additives in NR=SBR blends. J. Appl. Polym. Sci. 2004, 93 (2), 483–489. 12. Varkey, J.T.; Augustine, S.; Groeninckx, G.; Bhagawan, S.S.; Rao, S.S.; Thomas, S. Morphology and mechanical and viscoelastic properties of natural rubber and styrene butadiene rubber latex blends. J. Polym. Sci. B Polym. Phys. 2000, 38 (16), 2189–2211. 13. Ismail, H.; Suzaimah, S.; Hairunezam, H.M. Curing characteristics, mechanical properties and oil resistance of styrene butadiene rubber=epoxidized natural rubber blends. J. Elastomers Plast. 2002, 34 (2), 119–130. 14. Sanchez-Solis, A.; Estrada, M.R.; Cruz, J.; Manero, O. On the properties and processing of polyethylene terephthalate=styrene-butadiene rubber blend. Polym. Eng. Sci. 2000, 40 (5), 1216–1225. 15. Luzinov, I.; Xi, K.; Pagnoulle, C.; Huynh-Ba, G.; Jerome, R. Composition effect on the core-shell morphology and mechanical properties of ternary polystyrene=styrene butadiene rubber polyethylene blends. Polymer 1999, 40 (10), 2511–2520. 16. Sadhu, S.; Bhowmick, A.K. Preparation and properties of nanocomposites based on acrylonitrilebutadiene rubber, styrene-butadiene rubber, and polybutadiene rubber. J. Polym. Sci. B Polym. Phys. 2004, 42 (9), 1573–1585. 17. Su, X.Q.; Hua, Y.Q.; Qiao, J.L.; Liu, Y.Q.; Zhang, X.H.; Gao, J.M.; Song, Z.H.; Huang, F.; Zhang, M.L. The relationship between microstructure and properties in PP=rubber powder= nano-CaCO3 ternary blends. Macromol. Mater. Eng. 2004, 289 (3), 275–280. 18. Alert PERP Program—New Report; http:== www.Nexant.comAlert (accessed Jan 2004).

S

Superabsorbents S Takamasa Nonaka Faculty of Engineering, Department of Applied Chemistry and Biochemistry, Kumamoto University, Kurokami, Kumamoto-shi, Japan

INTRODUCTION Woven cloth, cotton wadding, cellulose fiber batt, papers, and foamed polyurethane have been used as traditional absorbent materials for water. These materials can absorb 1–20 g of water per gram material and the water absorbed is easily removed from the materials by applying low pressure. In recent years, superabsorbent polymers, which can absorb up to 1000 g of water per gram of polymer and up to about 100 g of dilute salt solution per gram of polymer and the water absorbed can hardly be removed from the polymers even by applying high pressure, have been prepared and commercially used in many applications. Superabsorbents are superabsorbent polymers, which are loosely crosslinked hydrophilic polymers and have ionic charges in the polymers. They absorb water, swell, and retain aqueous solution above hundred times their own weight. Because of the excellent characteristics, the superabsorbents have been widely used in many applications such as disposable diapers, feminine napkins, soil additives in agricultural or horticultural applications, gel actuators, water-blocking tapes, materials for drug delivery system, absorbent pads, etc., where water absorbency or water retention is important.

HYDROPHILIC POLYMERS FOR WATER ABSORPTION It is known that papers, cotton, and cellulose are natural hydrophilic polymers, which can absorb water, but are insoluble in water. Therefore, they have been used as disposable diapers, feminine napkins, etc. They absorb water by capillary action. Therefore, they absorb only a small amount of water and the water absorbed is easily removed by applying low pressure. Synthetic crosslinked poly(vinyl alcohol) and crosslinked poly(oxyethylene) are also hydrophilic polymers and they have almost the same water-absorption capacity of about 1–20 g water per gram of polymer as natural polymers. Those natural polymers and synthetic polymers do not essentially have ionic groups. On the other hand, superabsorbent polymers having anionic groups that were made from low crosslinked starch or synthetic polymers were found to have high Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120007973 Copyright # 2006 by Taylor & Francis. All rights reserved.

water-absorption capacity of more than 100 g water per gram of polymer. The water absorbed can hardly be removed even by applying high pressure. Thus, superabsorbent polymers should be hydrophilic polymers and low crosslinked and have ionic groups such as anionic groups, cationic groups, or betaine in the polymers.

CLASSIFICATION OF SUPERABSORBENTS[1] The first polymer hydrogels called as superabsorbents were prepared by hydrolysis of starch-graft-poly(acrylonitrile) at Northern Research Institute of Department of Agriculture of the U.S.A. in 1974. Since then, many superabsorbents consisting of modified natural hydrophilic polymers such as starch, cellulose, and alginic acids have been prepared. Most of them have anionic moieties such as poly(sodium acrylate) or poly(vinyl sulfonate). Now, many types of superabsorbents having not only anionic groups but also cationic groups or betaine have been prepared from both natural hydrophilic polymers and synthetic hydrophilic polymers. The classification of superabsorbents is listed in Table 1. As given in Table 1, superabsorbents can be classified in four kinds of parts:[1] 1. Superabsorbent polymers should be essentially hydrophilic polymers. Therefore, natural polymers such as starch and cellulose or synthetic polymers such as poly(acrylate), poly(vinyl alcohol), poly(acrylamide), or poly(oxyethylene) are used as hydrophilic base polymers. 2. Superabsorbents are low crosslinked to prevent to dissolve in water. Several methods for crosslinking are used: a. Crosslinked polymers can be prepared by copolymerization of hydrophilic vinyl monomers with crosslinking monomers such as methylenebisacrylamide, polyethyleneglycol dimethacrylate, divinylbenzene, etc. b. Water-soluble polymers such as poly(vinyl alcohol) or poly(oxyethylene) can be crosslinked by annealing or the reaction with crosslinking reagent such as glutaraldehyde. 2881

2882

Superabsorbents

Table 1 Classification of superabsorbents Base polymer

Starch—graft copolymerization, carboxymethylation Cellulose—graft copolymerization, carboxymethylation Synthetic polymers—poly(acrylate), poly(vinyl alcohol), poly(acrylamide), poly(oxyethylene), cationic polymer

Method of crosslinking

Copolymerization with crosslinking monomers Crosslinking of water-soluble polymer Irradiation of radioactive ray Self-crosslinking Introduction of crystal structure

Method of introduction of ionic groups

Polymerization of ionic monomer Caboxymethylation of hydrophilic nonionic polymers Graft copolymerization of hydrophilic monomer to hydrophobic polymer Hydrolysis of nitrile or ester group

Shape of products

Powder—spherical Film—amorphous Fiber—long fiber, short fiber, nonwoven cloth

c. Many polymers can be crosslinked by irradiation of radioactive rays such as g-ray. d. Poly(sodium acrylates) are partially crosslinked by self-crosslinking during polymerization at high monomer concentration. e. Water-soluble polymers such as poly(vinyl alcohol) can be crosslinked by crystallization by annealing or freeze-drying. 3. Superabsorbents should have ionic groups in the polymers. a. Ionic groups can be introduced by polymerization of monomers having ionic groups. b. Ionic groups can be introduced by carboxymethylation of nonionic polymers such as starch and cellulose. c. Ionic groups can be introduced by graft copolymerization of monomers having ionic groups on nonionic polymers such as starch, cellulose, or poly(vinyl alcohol). d. Ionic groups can be introduced by graft copolymerization of monomers having nitrile groups or ester groups on starch or cellulose, followed by hydrolysis. 4. Superabsorbents are commercialized in many shapes of products. Most superabsorbents are amorphous powders. Film and fiber types of superabsorbents are also used.

PREPARATION OF SUPERABSORBENTS[2] Up to the present, many types of superabsorbent polymers have been prepared and commercialized. Some

commercial superabsorbents and their manufacturers are listed in Table 2. Several examples are described here: 1. Starch=acrylonitrile copolymers. These are the first superabsorbents, which were prepared at Northern Research Institute of Department of Agriculture of the U.S.A. in 1974. These were prepared by graft copolymerization of acrylonitrile on starch using ammonium cerium(IV) nitrate as initiator, followed by hydrolysis with sodium hydroxide aqueous solution. Therefore, these have anionic charges of carboxylic group. The copolymers are easily subject to disintegrate with microorganism, because basic polymers are natural polymer. Therefore, these polymers do not retain water for a long period. 2. Starch=acrylic acid copolymers. These were prepared by graft copolymerization of acrylic acid on starch. The crosslinking of the polymers was carried out by simultaneous graft copolymerization of crosslinking monomers or postcrosslinking with crosslinking reagent such as epichlorohydrin or glutaraldehyde. The polymers cannot retain water for a long time, because the copolymers are subject to disintegration with microorganism. The types of polymers are mostly amorphous powder. 3. Carboxymethylated celluloses. These were prepared by the reaction of celluloses with monochloroacetic acid, followed by crosslinking. These are mostly used in fiber form. 4. Poly(sodium acrylates). Poly(acrylic acids) or poly(sodium acrylates) were obtained by i) solution polymerization of acrylic acid or sodium

Superabsorbents

2883

Table 2 Commercial superabsorbents and their manufacturers Manufacturer

Commercial name

Compositions

Europe

BASF Enka Stockhausen Unilever

Luquasorb Akucell Favor Lyogel

Poly(sodium acrylate) Carboxymethylcellulose Poly(sodium acrylate) Hydrolyzed starch=acrylonitrile copolymer

U.S.A.

Buckeye Dow Chemical Grain Processing Henkel Hercules National Starch Super Absorbent

CLD DWAL GPC SGP Aqualon Permasorb Magic Water Gel

Carboxymethylcellulose Poly(sodium acrylate) Hydrolyzed starch=acrylonitrile copolymer Hydrolyzed starch=acrylonitrile copolymer Carboxymethylcellulose Poly(sodium acrylate) Hydrolyzed starch=acrylonitrile copolymer

Japan

Nippon Shokubai Sanyo Kasei Daicel Nippon Exlan Sumitomo Seika

Aqualic Sanfresh CMC Daicel Espec Aquakeep

Poly(sodium acrylate) Starch=acrylic acid copolymer Carboxymethylcellulose Acryl fiber=sodium acrylate composite fiber Poly(sodium acrylate)

acrylate in water or by ii) reverse suspension polymerization in organic solvent. Crosslinked poly(acrylic acids) or poly(sodium acrylates) were prepared by copolymerization with crosslinking reagent such as methylenebisacrylamide or ethylene glycol dimethacrylate or post-crosslinking with diglycidyl compounds or higher alcohols. Crosslinked poly(acrylic acids) or poly(sodium acrylates) can also be obtained, when the polymerization is carried out at high concentration of the monomer in an aqueous solution. This crosslinking is called self-crosslinking. Crosslinked poly(sodium acrylates) are mostly prepared by solution polymerization and used as superabsorbents in hygienic application. 5. Vinyl alcohol=acrylic acid copolymers. Vinyl alcohol=sodium acrylate copolymers were obtained by hydrolysis of vinyl acetate= methylacrylate copolymers with alkaline solution. The vinyl alcohol=sodium acrylate copolymers obtained are insoluble in water, although they are not crosslinked with crosslinking monomer. This is because of crystal structure of vinyl alcohol moieties in the copolymers. Therefore, the vinyl alcohol=sodium acrylate copolymers have higher mechanical strength than crosslinked poly(sodium acrylates) in water. 6. 3-Methacrylamidepropyl trimethylammonium chloride=N,N 0 -methylenebisacrylamide (MBAAm) copolymers.[3] This type of crosslinked 3-methacrylamidepropyl trimethylammonium chloride= N,N 0 -methylenebisacrylamide copolymers were

prepared by copolymerization of 3-methacrylamidepropyl trimethylammonium chloride and N,N 0 -methylenebisacrylamide (crosslinking monomer). The copolymers have ammonium groups, which exist as cationic polymers in water. 7. Trialkyl-4-vinylbenzyl phosphonium chloride= acrylamide copolymers.[4] This type of crosslinked trialkyl-4-vinylbenzyl phosphonium chloride (TRVB)=acrylamide (AAm) copolymers were prepared by copolymerization of TRVB, AAm, and MBAAm (crosslinking monomer) in dimethyl sulfoxide. Three TRVBs with different alkyl chain lengths (butyl, hexyl, and octyl) in phosphonium groups were used. They are abbreviated as TBVB, THVB, and TOVB, respectively. The copolymers obtained have phosphonium groups. Therefore, they are crosslinked cationic polymers. 8. Trialkyl-4-vinylbenzyl phosphonium chloride= N-isopropylacrylamide (NIPAAm) copolymers.[5] These types of crosslinked TRVB=NIPAAm copolymers were prepared by copolymerization of TRVB, NIPAAm, and MBAAm in dimethyl sulfoxide. Poly(NIPAAm) is a thermosensitive polymer, which has a lower critical solution temperature (LCST) at around 33
C. Therefore, the copolymers are crosslinked cationic polymers and thermosensitive polymers, which swell and deswell below and above the LCST (about 33
C) of poly(NIPAAm). In Fig. 1, the structures of superabsorbent polymers (a)–(d), (g), and (h) are shown.

S

2884

PROPERTIES OF SUPERABSORBENTS Water-Absorption Capacity of Superabsorbent Polymers The water-absorption capacity of superabsorbent polymers is usually measured by the following methods. The water-absorption capacity is greatly affected by not only the chemical structures of the polymers but also the external conditions such as kinds of solutions, temperature, pressure, etc. under which it is measured. Tea bag method[6] The water content of superabsorbent polymers is usually measured by a tea bag method as follows: The dried copolymers are put into a tea bag made of nonwoven fabric. Then they are soaked into excess deionized water or salt solutions for 24 hr at desired temperature. After the tea bag containing polymers are soaked in aqueous solution, the water unabsorbed into gels is removed by placing the tea bag in air for a

Superabsorbents

short time. The water on the surface of the gels and the tea bag is immediately wiped with filter paper, and the weight (Ww) of the tea bag containing superabsorbent polymers is measured. The water content (Q) (H2Og= g-polymer) of the polymers is calculated using Eq. (1): Q ¼

Ww  Wt  Wd Wd

ð1Þ

where Wt and Wd are the weights of the wet tea bag and the dried polymers, respectively. In some cases, the water unabsorbed into superabsorbents can be removed by filtration or centrifugation. UV absorption method[7] A prescribed amount of superabsorbent polymer is soaked in a prescribed aqueous solution containing blue dextrin, which does not penetrate into superabsorbent polymers. The concentration of blue dextrin in outer solution after equilibrium swelling of the polymers is measured by UV spectrophotometer. And the

Fig. 1 The structure of super absorbent polymer (a)–(d), (g), and (h).

Superabsorbents

2885

water-absorption capacity of the polymers is calculated from the difference of the concentration of dextrin in the presence of polymers and that in the absence of polymers. Vortex time method[8] A prescribed amount of superabsorbent polymer is mixed with a prescribed amount of the desired aqueous solution, which is stirred by means of a magnetic stirring bar in a small beaker. As the water absorption proceeds, the viscosity of the suspension increases until the stirring vortex disappears at the ‘‘sorption time,’’ tv. The swelling capacity at equilibrium (Qmax) is calculated by using Eq. (2):

tv ¼


 1 Q ln 1  k Qmax

ð2Þ

where k is the first-order rate constant and Q is the swelling capacity (H2Og=g-polymer) at any time t.

Mechanism of Water Absorption with Superabsorbents As mentioned before, superabsobents should be essentially low crosslinked hydrophilic polymers and have ionic groups such as anionic groups, cationic groups, or betaine in the polymers. They usually have a high water-absorption capacity of more than 100 g-water=g-polymer even in dilute salt solution. In Fig. 2, swollen crosslinked poly (sodium acrylates) are illustrated. In general, the swelling ratio of hydrogels, which corresponds to water content, can be expressed by

Eq. (3):[9] Q5=3 ¼

 1 i 1 2   1=2 2 Vu S , ð1=2  X1 Þ n þ V1 V0



S ð3Þ

where Q is the swelling ratio, i=Vu the charge density attached to polymer matrix, ((1=2)  X1) the affinity between polymer matrix and water, S1=2 the ionic strength of outer solution, and n=V0 the crosslinking density. The first term [(1=2)  (i=Vu)  (1=S1=2)] in this equation represents the osmotic pressure because of ions, the second term f[(1=2)  X1]=V1g represents the affinity of polyelectrolyte for water, and the third term (V0=n) represents the degree of crosslinking. Therefore, this equation indicates that the water content of the superabsorbent polymers depends on the hydrophilicity, crosslinking density, charges of the crosslinked polymers, and the concentration of neutral salts in an aqueous solution. The various effects on the water-absorption ability of the polymers were investigated with superabsorbent polymers having anionic or cationic groups. Several effects on the water-absorption capacity of the superabsorbent polymers having carboxyl groups(1) or phosphonium groups(4) are described in the following section. Effect of ionic groups on the water-absorption ability of superabsorbent polymers In Figs. 3A and B the effect of the content of ionic groups on the water-absorption ability of starch= acrylic acid copolymers [polymer (d)] and the

Fig. 2 Illustration of swollen crosslinked poly(sodium acrylate).

2886

Superabsorbents

Fig. 3 Effect of ionic groups on the water absorption of (A) starch=acrylic acid and (B) TRVB=AAm=MBAAm [X : (97 - X) : 3] copolymer. Measured at room temperature in 0.9% NaCl aqueous solution. Copolymer: ( ) TBVB–AAm–MBAAm, ( ) THVB–AAm–MBAAm, and (`) TOVB–AAm–MBAAm. Measured at room temperature in deionized water.




TRVB=AAm=MBAAm copolymers [polymer (g)] containing phosphonium groups is shown, respectively. It is shown in Fig. 3A that the water content of the starch=acrylic acid copolymers increased with increasing content of acrylic acid, and then decreased above 90 wt.% of acrylic acid in the copolymers. In Fig. 3B, the water content of the TRVB=AAm=MBAAm copolymers having various contents of phosphonium groups in deionized water is shown. The water content increased with increasing phosphonium content in the copolymers, although the water contents of copolymers were different depending on the chain length of alkyl groups in phosphonium groups. This increase in water content with increasing content of ionic groups in the copolymers is because of the increasing osmotic pressure between the inner and outer sides of the polymer gels. It is also shown in Fig. 3B that the water content increased in the order of TBVB=AAm=MBAAm > THVB=AAm=MBAAm > TOVB=AAm=MBAAm copolymer. This order is inversely proportional to the chain length of alkyl chain in the phosphonium groups in the copolymers. This indicates that hydrophilic polymers with shorter alkyl chain in phosphonium groups absorb more water, although they have almost the same ionic groups in the copolymers. Effect of the degree of crosslinking on the waterabsorption ability of superabsorbent polymers In Figs. 4A and B the effect of the degree of crosslinking on the water-absorption ability of starch=acrylic acid copolymers [polymer (d)] and the TRVB=AAm= MBAAm copolymers [polymer (g)] containing phosphonium groups is shown, respectively.



It is shown in Fig. 4A that the water content of the starch=acrylic acid copolymers increased first, and then decreased with increasing content of crosslinking reagent, although the water content of starch=acrylic acid copolymers is different depending on the kind of crosslinking reagents. This indicates that hydrophilic copolymers should be crosslinked to some extent to retain high water content, and then the water content of the copolymers decreased with increasing degree of crosslinking of the copolymers. In Fig. 4B, the water content of the TBVB=AAm= MBAAm copolymers having different degree of crosslinking and almost the same phosphonium groups in deionized water is shown. The water contents of AAm–MBAAm copolymer carrying no phosphonium groups are also shown for comparison. The water content of the copolymers having phosphonium groups was fairly high and it decreased remarkably with increasing degree of crosslinking in the copolymers. However, the water content of the AAm–MBAAm copolymer carrying no phosphonium groups was fairly low compared with that of the copolymers having phosphonium groups and it decreased slightly with increasing degree of crosslinking of the copolymers. This result indicates that the introduction of ionic groups such as phosphonium groups into AAm–MBAAm zcopolymer is necessary to retain high water content. In recent years, it is reported that the water-absorption capacity of crosslinked poly(sodium acrylates) can be increased by increasing the crosslinking density near the surface and by decreasing the crosslinking density at the inner side of superabsorbent copolymers.[10] It is also reported that this type of superabsorbent polymer has high mechanical strength.

Superabsorbents

2887

S

Fig. 4. Effect of degree of crosslinking on the water absorption of (A) starch=acrylic acid and (B) TBVB=AAm=MBAAm copolymer. (A) Crosslinking reagent: 1) MBAAm; 2) ethyleneglycol diglycidyl ether; and 3) poly(ethyleneglycol) diglycidyl ether. Measured in 0.9%NaCl solution. (B) ( ) TBVB=AAm=MBAAm [2 : (98 - Y) : Y], (&) AAm=MBAAm. Measured in deionized water.




Effect of inorganic compounds in water on the water-absorption ability of superabsorbent polymers In Figs. 5A and B the effect of inorganic compounds in water on the water-absorption ability of starch=acrylic acid copolymers [polymer (b)] and the TRVB=AAm= MBAAm copolymers [polymer (g)] containing phosphonium groups is shown, respectively. It is shown in Fig. 5A that the water content of the starch=acrylic acid copolymers decreased sharply with increasing concentration of neutral salts (NaCl, MgCl2) or NaOH in aqueous solutions up to about 1 wt.% and then decreased gradually.[11] The order of effect is as follows: MgCl2 > NaCl > NaOH

This result indicates that MgCl2 containing divalent Mg2þ has a more pronounced effect than does NaCl containing monovalent Naþ. The decrease in the water absorption by an addition of neutral salts is because of the decrease in the osmotic pressure between inner and outer sides of the polymer hydrogels in their aqueous solution. In addition, it is known that the expansion of polymer chains because of the electrolytic repulsion between ionic groups of the polymers reduces, because an addition of these inorganic electrolytes shields the ionic charges of the polymers. This also results in the reduction of the water absorption of the superabsorbent polymers. In NaOH solution, parts of acrylic acid groups in starch=acrylic acid copolymers are neutralized with NaOH, and sodium acrylates formed dissociate easily

Fig. 5. Effect of addition of inorganic compounds on the water absorption of (A) starch=acrylic acid and (B) TBVB=AAm= MBAAm [X : (97 - X) : 3] copolymer. (A) Inorganic compounds: 1) NaOH; 2) NaCl; and 3) MgCl2. (B) NaCl concentration (wt.%): ( ) 0.009, ( ) 0.09, and (&) 0.9.






2888

in water. Therefore, starch=sodium acrylate copolymers become more polar hydrophilic polymers. This is the reason for the higher water content of the copolymers in NaOH solution than in other neutral salt solutions. However, the water content decreased also with increasing concentration of NaOH even in this aqueous solution. In Fig. 5B, the water content of the TRVB=AAm= MBAAm [X : (97 - X) : 3] copolymers having different phosphonium groups and constant degree (3 mol%) of crosslinking in 0.009, 0.09, and 0.9 wt.% NaCl solutions is shown. The water content of TBVB= AAm=MBAAm copolymer increased with increasing TBVB content in the copolymers and decreased with increasing concentration of NaCl in solutions and they were considerably depressed in NaCl solutions above 0.09 wt.%. The degree of the decrease in water content by the addition of NaCl increased with increasing content in phosphonium groups in the copolymers. Therefore, the decrease in water content of the copolymers by an addition of inorganic compounds is because of both the decrease in the difference of osmotic pressure between the inner side and the outer side of the copolymer hydrogels and shielding effect of ionic charges of the copolymers by an addition of inorganic compounds. Effect of physical structure of superabsorbent polymers on the water-absorption rate or water-absorption capacity The water-absorption rate of superabsorbent polymers is affected by: 1) the specific surface area of the polymers; 2) the capillary action; and 3) the formation of fish-eyes in the polymer hydrogels. 1. Specific surface area of polymer hydrogels can usually be increased by decreasing the particle size of the polymers. In Fig. 6A, the relationship

Superabsorbents

between water adsorption rate or absorption capacity and flaky particle size of crosslinked poly(sodium acrylates) is shown. The water adsorption rate increased with decreasing particle size in the range of particle size from 800 to 100 mm and then decreased. The decrease in the water-absorption rate with decreasing particle size (less than 100 mm) is because of the formation of fish-eyes in polymer hydrogels. The water-absorption rate can also be increased with decreasing apparent density (g=ml) of flaky polymer particles [Fig. 6B]. The decrease in apparent density of the polymers results in the increase in specific surface area of the polymers. This indicates that the water-absorption rate of the polymers increases with the specific surface areas of the polymers. 2. The water-absorption rate of the polymers can be increased by use of capillary action. It is reported that fibrous carboxymethyl celluloses, which were crosslinked with epichlorohydrin and had the carboxymethylation degree of 0.2–0.4, had high water-absorption rate of 20–40 ml= g-polymer=min. 3. The water-absorption rate of the polymers can be increased by mixing inorganic particles such as kaolin, talc, etc. with superabsorbent polymers. This mixing procedure results in the decrease in the formation of fish-eyes in polymer gels. In addition, the increase in the water-absorption rate is because of the fact that inorganic particles adsorbed on superabsorbent copolymers give space between polymer particles. Effect of temperature or pH on water-absorption rate or -absorption capacity The water-absorption rate of usual superabsorbent polymers can be increased by increasing the temperature, but the water-absorption capacity after equilibrium at

Fig. 6 Effect of particle size (A) and apparent density (B) on the water-absorption rate and water content of crosslinked poly(sodium acrylate).

Superabsorbents

each temperature is almost constant. The effect of pH on the water-absorption rate or -absorption capacity depends on the kind of ionic groups in the superabsorbent polymers. Usually the water-absorption capacity of superabsorbent polymers having weak acids such as acrylic acid increases with increasing pH. This is because of the formation of poly(acrylic salt) such as poly(sodium acrylate). On the other hand, the waterabsorption capacity of superabsorbent polymers having weak bases such as amino groups increases with decreasing pH. This is because of the formation of ammonium groups by protonation of amino groups. In both cases, the water-absorption capacity decreases at too high pH or at too low pH. The water-absorption capacity of superabsorbent polymers having strong acids such as sulfonic acid groups or strong bases such as ammonium groups or phosphonium groups essentially does not depend on pH of outer solution, but even in this case, the water-absorption capacity decreases at too low and too high pH. The decrease in the water-absorption capacity at too high and low pH is because of both the decrease in the difference of osmotic pressure between the inner side and the outer side of the polymer hydrogels and shielding effect of ionic charges of the polymers.

2889

in other organic solvents such as acetone=water mixtures.

OTHER PROPERTIES OF SUPERABSORBENT POLYMERS Evaporation of Water Absorbed in the Polymers[2] Water imbibed with superabsorbent polymers can hardly be removed even by applying high pressure, but the water absorbed is gradually removed by drying in air. In Fig. 8, the time dependence of decrease of water absorbed with vinyl alcohol=acrylic acid copolymers by drying at the temperatures of 20
C, 50
C, and 100
C is shown. It takes long time for water to evaporate from the superabsorbent polymers, although the rate of evaporation of water increased with increasing temperature. This result suggests that superabsorbent polymers can be used to retain water for a long time in soils for the agriculture in dry land. Stability of Superabsorbent Polymers Against UV Light[1]

Effect of organic solvent on the water absorption of superabsorbent polymers In Fig. 7, the water absorption of crosslinked poly(sodium acrylate) in various alcohols=water mixtures is shown. The water absorption decreases with increasing alcohol content and decreases abruptly at above certain limit of alcohol content. The content at which the abrupt decrease occurs depends on the kinds of alcohols. The similar phenomena are also observed

The stability of superabsorbent polymers against UV irradiation was evaluated by measuring the decrease of viscosity of swollen gels of poly(acrylate). In Fig. 9, the changes of the retention of the viscosity of the polymer gels UV-irradiated as a function of the degree of neutralization of acrylic acid moiety in crosslinked poly(acrylic acids) with alkali are shown. The degree of the retention of the viscosity of swollen gels increased with increasing degree of

Fig. 7 Absorption of water in alcohol=water mixtures with crosslinked poly(sodium acrylate): EtOH, ethanol; MeOH, methanol; PG, propylene glycol; EG, ethylene glycol.

S

2890

Superabsorbents

Fig. 8 Time dependence of the decrease of water absorbed with vinyl alcohol=acrylic acid copolymer at different temperatures.

neutralization of acid moiety in crosslinked poly (acrylic acids). It was found that poly(sodium acrylates), which was completely neutralized with alkali, are quite stable against UV light. It is reported that the significant decrease in the viscosity of crosslinked poly(acrylic acid) having a degree of neutralization less than 90% is because of the fact that radicals formed by UV irradiation to hydrogen of carboxyl group decomposed the main chain of poly (acrylic acids). Adsorption of Ammonia with Superabsorbent Polymers Containing Carboxyl Groups[1] Superabsorbent polymers consisting of poly(sodium acrylates) have partially carboxyl groups, which are

weak acid groups. Therefore, they can adsorb ammonia, which is a weak base. It is shown in Fig. 10 that crosslinked poly(acrylic acid) adsorbs ammonia, which is formed by the decomposition of urine with urease. Antibacterial Ability of Superabsorbent Polymers Containing Cationic Groups[5] It is reported that cationic polymers having ammonium groups, pyridinium groups, or phosphonium groups exhibit antibacterial activity against bacteria such as Escherichia coli or Staphylococcus aureus.[12–14] It is shown in Fig. 11 that TBVB–NIPAAm–MBAAm copolymer [polymer (h)], which has phosphonium groups and high water-absorption ability, had high antibacterial activity against S. aureus.

Fig. 9 Stability of swollen crosslinked poly (acrylate) neutralized with alkali against UV irradiation gel swollen about 50 times its own weight in water was used for this experiment.

Superabsorbents

2891

S

Fig. 10 Adsorption of ammonia with crosslinked poly(acrylic) acid. Conditions: urine, 50 ml; urease, 10 mg; 37
C; 2 hr.

It is also shown that the polymer having higher content of TBVB had higher antibacterial activity than did the copolymer with lower content of TBVB. They had the maximum antibacterial activity at 30
C and the water-absorption capacity of the copolymers decreased with increasing temperature and, in particular, decreased rapidly above 33
C, because the copolymers have LCST at around 33
C. The copolymers remaining in water can be separated by raising the temperature of aqueous solution above the LCST of the copolymers after using as an antibacterial reagent, because they are insoluble in water above the LCST of the copolymers, and they can be reused.

APPLICATION[15] Superabsorbent polymers have been used in a variety of fields. In Table 3, the applications of superabsorbent polymers are listed. Personal Hygiene Products[16–18] The most widely spread use of superabsorbent polymers is in personal hygiene products such as disposable infant diapers, feminine sanitary napkins, and adult incontinence articles. In particular, over 90% of the total superabsorbent polymers are sold as infant

Fig. 11 Antibacterial activity of TBVB– NIPAAm–MBAAm copolymer against S. aureus in deionized water at 30
C: ( ) blank, (G) TBVB–NIPAAm–MBAAm (1 : 100 : 3), and (&) TBVB–NIPAAm–MBAAm (3 : 100 : 3) copolymer.



2892

Superabsorbents

Table 3 Application of superabsorbent polymers Hygiene products

Disposable infant diapers Feminine sanitary napkins Adult incontinence articles

Agriculture and horticulture

Water retention of water in soils Sheet for cultivating paddy Seed coating Artificial sphagnum

Food packing

Fresh maintenance of foods Drip absorption Water removal from foods Cold insulator

Civil engineering and construction industry

Dew prevention Sealing water of leakage Hardening sluge Drilling

Toiletry

Fragrant gel Sweat band Portable toilet Disposable pocket heater Milky liquid pack

Controlled release and medical field

Sheet for surgical operation Drug delivery Stupe Dressing for protection of wound

Electronics and cabling

Papers for ink jet recorder Prevention from water penetration into communication and light fiber cable

Others

Water sensor Artificial snow Fire extinguishing water Water swellable toys Paint for prevention of water leakage Removal of water in oil

diapers. In Fig. 12, the schematic diagram of diapers is given. A diaper consists of an absorbent core that is sandwiched between a porous top-sheet and impermeable back-sheet. The top-sheet is made of a porous, hydrophobic substance, e.g., polyester or propylene nonwoven fabric, and the back-sheet is a nonporous, hydrophobic substance, e.g., polyethylene film. The absorbent core has a composite structure, with the cellulose pulp fluff and superabsorbent polymers randomly mixed. The superabsorbent polymers are also used for feminine hygiene products or adult incontinence products in the same manner as in infant diapers.

Agricultural and Horticultural Application[19] Superabsorbents can also help conserve water in a variety of agricultural and horticultural applications. Mainly, the polymers are used in the same way as

much, to help the soil retain moisture as soil additives. The polymers are mixed into soil at a concentration of about 1 wt.%. The resulting soil mixture retains moisture longer, and plants live longer after germination.

Food Packaging[20] Superabsorbent polymers are used as a liquid-absorber in food packaging systems. In these systems, the superabsorbent polymers absorb juice or water from fresh foods such as raw chicken, shellfish, and other meats or from frozen foods as they thaw. Chilled superabsorbent polymer gels may also be used as a dry-cooling medium. The water-swollen gel, contained in a durable plastic bag, is frozen and used to keep perishable foods cold. In addition to its liquid-water-absorption characteristic, superabsorbent polymers absorb water from the vapor state and therefore may be used to control humidity.

Superabsorbents

2893

S

Fig. 12 Schematic diagrams of infant diapers.

Civil Engineering and Construction Industry[21–24] Superabsorbent polymers have a number of uses in the civil engineering and the construction industry. One advantage of using superabsorbents in construction is to use the volume increase of the gel to form a barrier to further the water flow. Sealing composites made by blending a superabsorbent into a rubber such as chloroprene or thermoplastic elastomer such as poly (ethylene-co-vinyl acetate) have been developed for sealing around the joints of various building materials. The sealing composite may be used like mortar between the concrete blocks that make up the walls of the structure. If any gaps were left during construction or created by shifting after construction, the superabsorbent swells in any leaking water to fill any gaps and prevents water from leaking through the joint. A water-blocking construction filler that is composed of cement, water-absorbing polymer, and an asphalt emulsion has also been developed.

Electronics and Cabling[25,26] The swelling property of superabsorbents is also applied to protect communication cables from water damages. Leaking water degrades the performance of fiberoptic communication cables and power transmission cables. Water-blocking tapes prevent intrusion of water into the cables. A flexible, water-blocking tape may be made by mixing a superabsorbent polymer and a polymeric binder, and then spreading the mixture on to a nonwoven fabric. Alternatively, the superabsorbent is mixed with a rubber such as butyl rubber and solvent and then coated onto a polyester tape. The tape is wrapped around the cable, beneath the plastic covering, as shown in Fig. 13. Controlled Release and Medical Field[27,28] Superabsorbent polymers may also be used to control the release of substances that are initially dissolved or trapped within the polymer phase, such as pesticides, fertilizers, and pharmaceuticals. Just as the absorption of water by superabsorbent polymer is caused by unequal activities of water inside the gel and in the external liquid phase, substances that are initially at a higher activity in the polymer will diffuse out of the particle and into the surroundings. The swelling of crosslinked poly(acrylic acid) and other acidic or basic polymers depends on the pH and ionic strength of the swelling medium, and the changes in pH or ionic strength may serve as switches for controlled release. Superabsorbent polymers are also used as materials in the sheet for surgical operation, stupe, or dressing for protection of wound. Toiletry[29,30]

Fig. 13 Wire protected with a water-blocking tape containing superabsorbent polymer.

Superabsorbent polymers have been used in several cosmetic formulations. A skin-cream emulsion that

2894

was prepared with a crosslinked superabsorbent in addition to branched poly(acrylic acid) thickener for cosmetic exhibited a long-term moisturizing effect than did the emulsion without the superabsorbent polymer. Superabsorbents have also been used as a component in a gel-form cosmetic face mask. Superabsorbent polymers are also used as materials for fragment gel, sweat band, portable toilet, etc. Others[31–33] 1. Sensor: The swellability of superabsorbent polymer gels, their mechanical modulus and rubbery character, and their sensitivity to changes in water content, pH, and ionic strength make them suitable for use in various sensing systems. 2. Artificial snow: The fake snow can be made by mixing a superabsorbent polymer with up to 100 times its mass of water and then aerating the gel whole freezing it. The artificial snow is maintained with refrigeration system in indoor ski arenas. The frozen gel layer groomed to yield snow with a realistic feel, similar to ‘‘powder’’ snow. By using superabsorbent polymer in this way, the air temperature in the building can be at least 10
C higher than when using snow made from only water and it is more comfortable to the skiers. 3. In addition to this, superabsorbent polymers are used as additives in fire extinguishing water, materials for water-swellable toys, paint for prevention of water leakage, for removal of water in oil, etc.

CONCLUSIONS Superabsorbents are superabsorbent polymers, which are loosely crosslinked hydrophilic polymers and have ionic charges. They absorb water, swell, and retain aqueous solutions up to 100 times their own weight. The water absorbed can hardly be removed even by applying high pressure. Many types of superabsorbents have been prepared from various materials, by various methods, and in different shapes. They are modified natural hydrophilic polymers such as starch, cellulose, alginic acids, etc. and synthetic hydrophilic polymers such as poly (acrylic acid) and poly(sodium acrylate). Most of them have anionic moieties such as poly(sodium acrylate) and poly(vinyl sulfonate). Now, many types of superabsorbents having not only anionic groups but also cationic groups such as ammonium groups, phosphonium groups, or betaine have been prepared from both

Superabsorbents

natural hydrophilic materials and synthetic hydrophilic polymers. Superabsorbents can be made in a variety of shapes such as powder, granules, fiber, sheet, etc. It has been clarified that, to have high waterabsorption ability, superabsorbent polymers should be essentially low crosslinked hydrophilic polymers having ionic charges. The water-absorption ability of the superabsorbent polymers depends on the hydrophilicity, crosslinking density, charges of the polymers, and the concentration of neutral salts in an aqueous solution. This suggests that much water invade into polymers because of high osmotic pressure between inside and outside of polymer gels in aqueous solutions and the electrostatic repulsive expansion of polymers. The synthesis of new type of superabsorbents, the modification of the superabsorbents synthesized to enhance their absorbency, gel strength, and absorption rate, and the development of new application of the superabsorbent have been investigated by many researchers in recent years.

REFERENCES 1. Masuda, F. What are superabsorbents? In Superabsorbent Polymers; Kyoritsu Shuppan: Tokyo, 1988; 1–11 (written in Japanese). 2. The Society of High Polymer Japan Ed., Superabsorbent polymers. In Kobunshi Sinsozai Binran; Maruzen: Tokyo, 1989; 228–235, (written in Japanese). 3. Farahani, E.V.; Vera, J.H.; Cooper, D.G.; Weber, M.E. Swelling of ionic gels in electrolyte solution. Ind. Eng. Chem. Res. 1990, 29, 554. 4. Nonaka, T.; Yamada, K.; Watanabe, T.; Kurihara, S. Preparation of superabsorbent polymer hydrogels from trialkyl-4-vinylbenzyl phosphonoium chloride-acrylamide-methylenebisacrylamide copolymer and their properties. J. Appl. Polym. Sci. 2000, 78, 1883–1884. 5. Nonaka, T.; Watanabe, T.; Kawabata, T.; Kurihara, S. Preparation of thermosensitive and superabsorbent polymer hydrogels from trialkyl-4-vinylbenzyl phosphonium chloride-N -isopropylacrylamideN ,N 0 -methylenebisacrylamide copolymers and their properties. J. Appl. Polym. Sci. 2001, 79, 115–124. 6. The Japan Industry Standards (JIS). Testing Method for Water Absorption Capacity of Super Absorbent Polymers, K7223, 1996. 7. Masuda, F. Properties of superabsorbent polymers. In Superabsorbent Polymers; Kyoritsu Shuppan: Tokyo, 1988; 51–80 (written in Japanese). 8. Makita, M.; Tanioku, S. Water-Absorbing Resins. U.S. Patent 4,587,308, May 6, 1986. 9. Flory, P.J. Principle of Polymer Chemistry; Cornell University Press: Ithaca, NY, 1953.

Superabsorbents

10. Tsubakimoto, T.; Shimomura, T.; Irie, Y. Absorbent Articles from Powdered Resins. U.S. Patent 4,666,983, May 19, 1987. 11. Masuda, Y. Absorbent Polymer Gels, Gel Technology; Abe, M., Murase, N., Suzuki, T., Eds.; Science Forum: Tokyo, 1997; 18–25. 12. Nakagawa, Y.; Hayashi, H.; Tawaratani, T.; Kourai, H.; Horie, T.; Shibasaki, I. Disinfection of water with quaternary ammonium salts insolubilized on a porous glass surface. Appl. Environ. Microbiol. 1984, 47, 513. 13. Nonaka, T.; Uemurra, Y.; Kurihara, S. Preparation of the resins containing quaternary ammonium groups from glycidyl methacrylate-1,4-divinylbenzene copolymer beads and antibacterial activity of the resins. Nippon Kagaku Kaishi 1994, 12, 1097–1106 (written in Japanese). 14. Nonaka, T.; Ohtsuka, T.; Kurihara, S. Preparation of the resins containing phosphonium groups from glycidyl methacrylate-1,4-divinylbenzene copolymer beads and antibacterial activity of the resins. Nippon Kagaku Kaishi 1995, 1995 (7), 529–539 (written in Japanese). 15. Buchholz, F.L. Application of superabsorbent polymers. In Modern Soperabsorbent Polymer Technology; Buchholz, F.L., Graham, A.T., Eds.; Wiley-VCH: New York, 1997; 251–272. 16. Nishizawa, K.; Shirase, T.; Mizutani, H. MoisturePermeable Disposable Diapers. U.S. Patent 4,306,559, December 22, 1981. 17. Harper, H.; Bashaw, R.; Atkins, B. Absorbent Product Containing a Hydrocelloidal Composition. U.S. Patent 3,669,103, June 13, 1972. 18. Kellenberger, S.R. Absorbent Products Containing Hydrogels with Ability to Swell Against Pressure. U.S. Patent 5,147,343, September 15, 1992. 19. Kazanskii, K.S.; Dubrovskii, S.A. Chemistry and physics of agricultural hydrogels. Adv. Polym. Sci. 1992, 104, 97–133. 20. DeGouw, A.M.; Prins, J.; Dingermas, L. Package for Food Stuffs, such as Shell Fish, Which While in the Packaged State Will Exude Liquid, and a Packaging Method. Eur. Patent 68,530, January 5, 1983.

2895

21. Tsubakimoto, T.; Shimomura, T.; Kobayashi, H. Water absorbents. Jpn. Kokai Tokkyo Koho 1987, July 3, 62–149, 335. 22. Suetsugu, M.; Sezaki, E.; Nakazato, T.; Isono, M. Water-absorbing compositions containing polyolefin thermoplastic elastomers. Jpn. Kokai Tokkyo Koho 1994, June 07, 06–157, 839. 23. Shimomura, T.; Namba, T. Superabsorbent Polymers. In Superasorbent Polymers, Science and Technology, Symposium Series 573; Buchholz, F.L., Peppas, N.A., Eds.; American Chemical Society: Washington, DC, 1994; 112–115. 24. Moriyoshi, A.; Fukai, I.; Takeuchi, M. A composite material that solidifies in water. Nature 1990, 344, 230–232. 25. Bow, K.E. Electric Cable with Improved WaterBlock. Eur. Patent. 24,631, March 11, 1981. 26. Hogari, K.; Ashiya, F. Superabsorbent Polymres, Science and Technology, Symposium Series 573; Buchholz, F.L., Peppas, N.A., Eds.; American Chemical Society: Washington, DC, 1994; 128–140. 27. Peppas, N.A. Hydrogels in Medicine and Pharmacy; CRC Press, Inc.: Boca Raton, FL, 1987; Vol. 1, 1–180. 28. Bronsted, H.; Kopecek, J. Polyelectrolyte Gels, Properties, Preparation and Application; Harland, R.S., Prud’homme, R.K., Eds.; American Chemical Society: Washington, DC, 1992; 285–304. 29. Kendall, J.M.; Maes, D.H.; Figueroa, R. Jr. Improved Skin-Moisturizing Emulsions Containing Water-Absorbent Resin Polymeritate. Eur. Patent 281,395, September 07, 1988. 30. Gueret, J.L.; Contamin, J.C.; Ayache, L. SheetLike Materials for the Treatment of Skin or Hair. Eur. Patent 309,309, March 29, 1989. 31. Sawahata, K.; Gong, J.P.; Osada, Y. Soft and wet touch-sensing system made of hydrogel. Macromol. Rapid Commun. 1995, 16, 713–716. 32. Bucceri, A. Method for Making Artificial Snow. U.S. Patent 4,742,958, May 10, 1988. 33. Morioka, K.; Nakahigashi, S. Construction plan of Tsudanuma skiing arenas with artificial snow and technology of making artificial snow. Refrigeration 1992, 67, 28 (written in Japanese).

S

Supercritical CO2-Assisted Surface Coating Injection Molding

S

Masahiro Ohshima Department of Chemical Engineering, Kyoto University, Nishikyo-ku, Kyoto, Japan

INTRODUCTION Carbon dioxide (CO2) becomes supercritical at pressures and temperatures above 7.38 MPa and 31.1
C. At the supercritical condition, the CO2 has a higher diffusivity and a lower viscosity than liquids and the surface tension becomes absent, which allows a rapid penetration of molecule into the pores of heterogeneous matrices.[1] Taking advantage of these characteristics together with its nontoxicity, nonflammability, and inexpensiveness, CO2 has been used in a variety of industries, i.e., food, pharmaceutical, fiber, and chemical and plastic industries. Moreover, the fact that CO2 is a gas under ambient condition makes its removal from the plastic product possible without using the costly drying and solvent removal processes.[2] Because of this feature, it also has the potential of replacing many environmentally harmful solvents in industry. A successful application reported in the polymer processing was the microcellular plastic foams,[3,4] where CO2 was used as a physical foaming agent (PFA) to create microscale size cellular structure in plastics. After creating the cell structure in plastics, the CO2 diffuses out and does not stay long in the plastics, unlike chemical browning agents. This feature gives some benefits of plastic recycling. In addition to the microcellular foaming, many potential applications of CO2, such as polymer blending, additive impregnation, and surface modifications, have been investigated for polymer processing purposes. In the polymer processing field, the usage of supercritical CO2 (scCO2) offers several exciting possibilities.[5] One major advantage of applying scCO2 to polymer processing is that the processing conditions and the morphology of polymers can be controlled by scCO2. This is because the presence of CO2 in polymer changes the rheological and thermal properties in both the molten and solid states. For example, when CO2 dissolves in polymer, the surface tension, shear viscosity, and the glass transition temperature are reduced and the crystallization rate is changed. This is called plasticization effect of CO2. This effect provides many opportunities of improving the processing condition and quality of plastic products. Recently, by utilizing the plasticization effects of dissolved CO2, an injection molding process called Asahi Mold Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120040332 Copyright # 2006 by Taylor & Francis. All rights reserved.

Technology with CO2 (AMOTEC) was invented to reduce surface roughness and prevent the surface defection in solidified plastic products,[6,7] where pressurized CO2 is introduced in a mold cavity, the CO2 is dissolved into polymer at the melt front of polymer running in the mold cavity, and the dissolved CO2 reduces the polymer viscosity at the surface of the plastics owing to the plasticization effect. The viscosity reduction at the plastic surface increases the transferability of the mold shape on the plastics and improves the surface roughness. Extending and twisting the idea of the AMOTEC, a CO2-assisted surface modification injection-molding technique was developed.[8] It utilizes the plasticization effect of CO2 on polymer, solubility of low molecular compounds in scCO2 and infusion mechanism of CO2, and the low molecular compounds mixture into the injected polymer from its surface. By using a dye pigment as a low molecule to be dissolved in scCO2, the surface of the solidified plastic products in the mold cavity can be dyed.[9] In this entry, fundamental properties of CO2=polymer systems are briefly reviewed with some explanation of recent advancements in property measurements. The principle of the developed CO2-assisted surface modification injection molding is then explained with a validation experiment.

FUNDAMENTAL PROPERTIES Solubility of CO2 in Polymer Solubility and diffusivity of CO2 in molten polymer are two fundamental properties for the proposed CO2-assisted polymer processing. Several researchers have been studying the solubility of gases in polymers in conjunction with polymeric membrane and microcellular foaming. Durill and Griskey studied the solubility of CO2 and nitrogen (N2) gases into several polymers such as polypropylene (PP), low density polyethylene (LDPE), high density polyethylene (HDPE), and polystyrene (PS).[10,11] Liu and Prausnitz also investigated the solubility of several gases including CO2 in PE and in ethylene-vinyl acetate (EVA). Their experiments were carried out in the temperature range from 125
C to 200
C under a fairly low-pressure 2897

2898

Supercritical CO2-Assisted Surface Coating Injection Molding

Solubility (g-gas/g-polymer)

pressure increases and it follows the Henry’s law. Thus, the Henry’s constant is used to evaluate the solubility of low molecular gases in molten polymer systems as listed in Table 1. The solubility of CO2 increases as temperature decreases. The van’t Hoff equation is often used to describe temperature dependence of the solubility:
 Ea HP ðTÞ ¼ HO exp  ð1Þ RT

PP

0.10

LDPE HDPE EEA copolymer

0.08

PS

0.06

0.04

0.02

0.00

0

5

10 Pressure (MPa)

15

Fig. 1 Solubility of CO2 in several polymers (at 200
C, a gravimetric method). (From Ref.[27].)

region (approximately 2.5 MPa).[12] Lee and Flumerfelt studied the solubility of N2 in LDPE at the range from 120
C to 177
C and pressure up to 12.5 MPa.[13] Sato et al. studied the solubility of CO2 and N2 in PS, PP, and HDPE at 100
C to 200
C and pressure up to 20 MPa and in PVAc at 40
C to 100
C, pressure up to 17.5 MPa.[14,15,16,17] Fig. 1 shows the solubility of CO2 in LDPE, HDPE, PP, ethylene-ethylacrylate copolymer (EEA), and PS, which were measured at temperatures from 150
C to 200
C and pressure up to 12 MPa using a magnetic suspension balance (MSB).[18] As shown in Fig. 1, the solubility of gases in polymers increases as the saturation

Table 1 Henry’s constant 150 [Ref.] PS PS

180 [Ref.]

200 [Ref.]

[17]

3.200[17]

7.645[16]

6.604[16]

7.180a

6.847

6.353[16]

5.710[12]

4.102 4.710

3.320

PP PP [16]

HDPE

7.313

HDPE

6.807

5.885

LDPE

6.620

5.804

LDPE

6.715

PLA PBS

5.890 5.786

6.861

5.342

5.639

PC

3.467

EPDM

9.511

Henry constant (g-CO2=g-polymer=MPa)  103. a 185
C.

where Ea is the heat of solution, R is the gas constant, and T denotes temperature. Note that the temperature dependency of the N2 solubility in polymer shows a different behavior, i.e., the solubility of N2 in molten polymer increases as temperature increases in a certain temperature range.[15] Furthermore, the Henry’s law could not hold when the temperature is below the glass transition temperature due to the fact that the solution mechanism is changed to a combination of adsorption and dissolution. Diffusivity of CO2 in Polymer In addition to the solubility, many researchers have also investigated the diffusivity of gas into polymer and several models have been proposed for estimating the diffusivity of gas in polymer. The most widely accepted concept for developing the models is based on the free volume. The free volume concept for diffusion has an origin at the theoretical work of Cohen–Turnbull, where the thermodynamic diffusion coefficient of the solute was given by an exponential function of inversed free volume fraction. Recently, Thran et al. analyzed the correlation between the free volume fraction and diffusion for six gases, O2, N2, CO2, CH4, He, and H2, and 71 polymers.[19] They derived the exponential relationship between the specific free volume and the diffusion coefficient and reported empirical constants of the exponential correlation model. Extending the free volume model, Areerat et al. proposed a model for estimating the diffusivity of CO2 in polymer,[18,27] where pressure, temperature, and concentration dependences of diffusivity were described by
 x2 Dself @m1 @ ln x1 T;P RT " #
 @m1 B ¼A x2 exp 0 @ ln x1 T;P Þ ðV^mix  V^mix

Dmutual ¼

0 0 0 V^mix ¼ ð1  mCO2 ÞV^poly þ mCO2 V^CO 2

ð2Þ ð3Þ

where A and B are parameters; x2 is the mole fraction of polymer; V^mix is the specific volume of polymer=CO2

Supercritical CO2-Assisted Surface Coating Injection Molding 0 mixture at a given temperature and pressure and V^mix is the occupied specific volume of the mixture at the absolute zero temperature, which is estimated by 0 0 Eq. (3). V^poly are the occupied specific volume and V^CO 2 of polymer alone and pure gas, respectively. mCO2 is the 0 weight fraction of CO2 in the solution. V^mix  V^mix is an estimate of the free volume at the given temperature, pressure, and CO2 concentration. They proposed a scheme of calculating the V^mix from the swelling of polymer=CO2 mixture, which could be estimated using an equation of state from the solubility data, pressure– volume–temperature (PVT) data of polymer alone. Fig. 2 shows a temporal change in the weight of four different polymers, PP, LDPE, PS, and ethylenepropylene-diene rubber (EPDM) during sorption measurement of CO2 at 200
C. Those data were acquired using a gravimetric measurement scheme, i.e., the magnetic suspension balance (MSB), which was the same apparatus used for solubility measurement. When the diffusivity of CO2 is measured, CO2 pressure in the MSB cell is increased in a stepwise manner 1.0 MPa from a saturated state, the temporal change in weight, as shown in Fig. 2, was obtained and the diffusion coefficients were then determined assuming the diffusion process follows the Fick’s second law of diffusion. Fig. 3 shows the diffusivity of CO2 in PS at different temperatures and its estimates. The diffusivity is not only a function of temperature and pressure, but also concentration of dissolved CO2. The diffusivity increases as the temperature increases and decreases as the pressure increases. When CO2 dissolves, the polymer swells and its free volume increases. The diffusion coefficient then becomes larger

2899

as the dissolved CO2 concentration becomes higher. Although the diffusivity does not change drastically with the concentration at temperature high enough to keep the polymer molten, the effect of dissolved CO2 concentration on diffusivity becomes prominent around the glass transition temperature. Koga et al. investigated the CO2 sorption into poly(methyl methacrylate) (PMMA) as well as PS polymers using in situ neutron reflectivity.[20] They observed that both PMMA and PS swell by approximately 30%, and at the same time, they observed an anomalous diffusion of CO2 in these polymers at temperatures below their original glass transition temperature.

Diffusivity of scCO2 and Low Molecular Compounds Mixture in Polymer There are several papers discussing the diffusivity of one single penetrant in polymer. However, few papers could be found for measuring diffusivity of scCO2 and low molecular compounds mixture into polymer, i.e., diffusivity of multicomponents in scCO2, which is the key factor for determining the operating condition of the proposed surface coating injection molding techniques. In order to analyze the diffusion behavior of scCO2 and low molecule mixture into polymer, an online measurement system using near infrared (NIR) spectroscopy was developed.[21] Fig. 4 shows a schematic diagram of the NIR measurement system. The fiber optic probes of on-line Fourier transform NIR unit (FIR1000L, Yokogawa Electric CO.) were equipped with a high-pressure

1

Relative amount of dissolved CO2 [-]

0.9 0.8 EPDM

0.7

PP

0.6

PS LDPE

0.5 0.4 0.3 0.2 0.1 0 0.0E+00

2.0E+06

4.0E+06

6.0E+06 2

2

t/L [min/m ]

8.0E+06

1.0E+07

Fig. 2 Diffusivity of CO2 in polymers (temperature at 200
C and pressure step 10–11 MPa for EPDM, PP, PS, 10.5–11 MPa for LDPE. L is the thickness of the sample).

S

Supercritical CO2-Assisted Surface Coating Injection Molding

2900

× 10–5

[×10–9]

Exp. @ 200

Diffusion Coefficient [m2/s]

Mutual Diffusion Coefficient, D (cm2/s)

5

Exp. @ 150

4

estimates

3

2

1

5 CO2 only CO2 in CO2/PrOH system PrOH in CO2/PrOH system

1 b) PS/CO2 solutions

0

10

0

2

4

6

CO2 wt.%

0

5 Gas Pressure [MPa]

10

Fig. 5 Diffusion coefficients of CO2 in LDPE measure by NIR at 175
C.

Fig. 3 Mutual diffusion coefficients of CO2 in PS (experiments and estimates). (From Ref.[27].)

autoclave. The sample cell of the autoclave takes a cylindrical shape, in which a cylindrical sample polymer was charged and melted in the cell by increasing the temperature, so as to leave no space between the sample and the wall, while making an open space over the sample polymer. CO2 alone, or a mixture of CO2 and low molecule, was introduced to the open space and diffused into the molten polymer. The NIR light was transmitted through the molten polymer at a certain distance from the interface between the polymer and CO2 or mixture. When some NIR light passed through a layer of solution having a concentration of chemical species, the power of the light is absorbed and attenuated by

Fig. 4 NIR measurement system.

the chemical species due to the interaction between the photons and absorbing chemical bonds. The wavelength at which adsorption occurs depends on the species of chemical bonds. For example, CO2 can be detected at 2019 nm. The absorbance, which is defined by logarithmic fraction of the light transmitted by the solution, is directly proportional to the path length through the solution and the concentration of the absorbing chemical bond in the solution. This is called the Lambert–Beer law. Nagata et al. clarified that the Lambert–Beer law could be established at the polymer CO2 systems in the transmitted NIR measurements.[21] Applying the law, the change in concentration of CO2 and low molecular compounds in molten polymer could be monitored by measuring the NIR absorbance on-line at the autoclave. Fig. 5 shows the NIR measured diffusion coefficients of CO2 alone, and those of CO2 as well as propanol, of their mixture to LDPE (MI ¼ 8.0 g=10 min; Mw ¼ 1.05  105 g=mol; Mw=Mn ¼ 6.94) at 175
C by changing the pressure from 0.1 MPa to a specified level. As can be seen in Fig. 5, both propanol (PrOH) and CO2 of mixture could diffuse into the polymer even under the critical pressure of CO2, i.e., 7.38 MPa. Since the propanol is volatile, it mixes with CO2 in gas phase at 175
C. Therefore, the propanol could diffuse into polymer even at the low pressure. It could be worthwhile to note that the diffusivity of propanol was lower than that of CO2 at low-pressure level, but it increased close to that of CO2 over the critical pressure. Nonvolatile molecules, such as dye pigments, cannot be infused into polymer without using scCO2. Fig. 6 shows the change in absorbance at 1493 nm when CO2 and dye pigment (disperse blue) mixture was introduced to the autoclave and diffused into the LDPE at 175
C by changing the CO2 pressure from

Supercritical CO2-Assisted Surface Coating Injection Molding

12

2901

×10–4

S

Change in Absorbance at 1493 nm

10 8 6 4 2 0 –2 –4 –6 –8

0

0.5

1

1.5 2 Time [sec.]

2.5

3

×104

3.5

Fig. 6 Diffusion behavior disperse blue with CO2 into LDPE measured by NIR at 175 C. (View this art in color at www.dekker.com.)

0.1 to 10 MPa. Although the signal over noise ratio was low, it was clearly observed that the blue pigment was diffusing into the polymer. As shown by the experimental data, nonvolatile molecules could be infused into polymer using scCO2 as far as the molecule could be dissolved into scCO2.

developed by integrating the Cross–Carreau model with Doolittle’s equation by means of a variable free volume. The Cross–Carreau model was given by

CO2 Induced Viscosity Reduction

where Z and ZO are the shear and zero-shear viscosities, respectively; g_ is the shear rate; and n, t, and a are parameters which determine the shape of the viscosity-shear rate curve.   _ At high shear rates where ZOt g  1 can hold, Eq. (4) can be simplified into

The other factor of controlling the proposed injection molding is the plasticization effect of CO2, i.e., polymer viscosity reduction by CO2 dissolution. During the last decade, the study of CO2 induced viscosity reduction has been advanced drastically. Gerhardt et al. measured viscosity reduction of polydimethylsiloxane (PDMS)= CO2 solutions by using a high-pressure plunger-type capillary rheometer.[22] Kwag et al. also measured the viscosity of PS=CO2 solutions by employing the same technique as Gerhardt.[23] Lee et al. measured the viscosity of PS=CO2 solutions.[24] Gendron et al. measured the viscosity of PS=hydrofluorocarbon (HFC) solutions, as well as PS=CO2 solutions, and developed a viscosity prediction model for these mixtures by employing the William–Landle–Ferry (WLF) equation with a variable glass transition temperature, Tg.[25] They calculated Tg as a function of dissolved gas concentration by extending the Chow model. Royer et al. modified Gendron’s model so that the pressure effect could be taken into account when the viscosity of PS=CO2 mixtures was predicted.[26] Areerat et al. proposed the following practical model for predicting the viscosity of the polymer=CO2 solutions.[27,28] The model was

Zð_g; Þ ¼ 

Zð_gÞ ffi Zno

ZO 1þ



ZO g_ t

a  ð1
nÞ a

 ðn
1Þ g_ t

ð4Þ

ð5Þ

It is assumed that the parameters n, t, and a are constant and the viscosity reduction due to the CO2 dissolution could be described only by the changes in zero-shear viscosity. The zero-shear viscosity, ZO, is given as a function of free volume fraction as described by Zo

 ¼ A exp

B fðT; P; wg Þ



ð6Þ

where A and B are unique constant parameters for the polymer and they are determined when the polymer is given. f(T, P, wg) denotes the free volume fraction. Then, the free volume fraction is assumed to be a function of temperature, T, pressure, P, and weight percentage of dissolved CO2, wg.

Supercritical CO2-Assisted Surface Coating Injection Molding

2902

Substituting Eq. (6) into Eq. (5) results in
 C2 Z ¼ exp C3 þ g_ C1 fðT; P; wg Þ

where fr is a reference free volume fraction given at a reference temperature, Tr, pressure, Pr and weight per
centage of dissolved CO2, wg r: að:¼ V1 @V @T Pr ;wgr Þ is the
thermal expansion coefficient, bð:¼  V1 @V @P Pr ;wgr Þ is

ð7Þ

where

the isothermal compressibility coefficient and f is the gas expansion coefficient. Since zero-CO2 concentration is taken as a reference condition, the parameter fr is not a function of CO2. Then, fr, a, and b can be determined from the PVT measurement of the neat polymer. There remains only gas concentration coefficient, j, as a variable affected by CO2 dissolution. The gas expansion coefficient, j, is determined by solubility measurements, the models with these parameter values could predict the viscosity of polymer=CO2 single-phase mixtures.[27] Fig. 7 shows the CO2 induced viscosity reductions of PP and PS polymers. The viscosities of the polymer=CO2 mixtures were measured by a capillary rheometer equipped with a foaming extruder. As can be seen in Fig. 7, the viscosity of PS was reduced by 40% by dissolving CO2 3.5 wt% and that of PP was reduced by 25%. The solid lines in Fig. 7 represent the estimates of the aforementioned models.

C1 ¼ ðn  1Þ

C2 ¼ nB C3 ¼ ½n ln A  ðn  1Þ ln t Assuming that C1, C2, and C3 are not affected by temperature, pressure, and dissolution of CO2, they can be determined from a viscosity-shear rate curve of the neat polymer. Namely, the coefficient, C1, which is equivalent to n – 1, can be determined by the slope of the viscosity and shear rate curve. The values of C2 and C3 can be determined from data of viscosity vs. free volume fraction of the neat polymer. The data of free volume fraction required for determining C2 and C3 can be obtained from PVT data of the neat polymer at temperatures and pressures where the viscosity measurements of the neat polymer are performed. To predict the viscosity reduction, the change in the free volume fraction caused by CO2 dissolution has to be calculated. Extending the definition of free volume fraction, the free volume fraction is given as a function of temperature, pressure, and dissolution of gas: fðT; P; wg Þ ¼ fr þ




þ




@f @P

þ




@f @wg

@f @T 



Pr ;Tr ;wgr

Pr ;Tr ;wgr



CO2-ASSISTED SURFACE MODIFICATION INJECTION MOLDING As shown in the previous section, CO2 enhances the mobility of polymer and plasticizes the polymers. Furthermore, it can dissolve a variety of low molecular compounds, such as dye pigments, additive agents, monomer, and crystal nucleation agents, when it becomes a supercritical state. Integrating these characteristics of CO2 can create an scCO2-assisted surface coating injection molding technique. Two different injection schemes of coating the plastic surface are described in the following sections.

ðT  Tr Þ

ðP  Pr Þ

Pr ;Tr ;wgr

ðwg  wgr Þ

ð8Þ

¼ fr þ ð1  fr ÞaðT  Tr Þ  ð1  fr ÞbðP  Pr Þ þ fðwg  wgr Þ

6.0

PP/CO2 0%

Shear Viscosity ln(η) (Pa.s)

Shear Viscosity ln(η) (Pa.s)

6.0

0.53%

5.5

1.35% 2.06%

5.0

4.5

5.5

5.0

PP/CO2 0% 1.09%

4.5

1.63% 2.30% 3.49%

5.0 6.0 Shear rate ln(γ) (1/s)

7.0

5.0 6.0 Shear rate ln(γ) (1/s)

7.0

Fig. 7 CO2-induced viscosity reduction at PS=CO2 and PS= CO2 systems at 200
C.

Supercritical CO2-Assisted Surface Coating Injection Molding

2903

Direct Injection Scheme Mold wall

One of the scCO2-assisted surface coating schemes is illustrated in Fig. 8. scCO2 is introduced into a mold cavity of an injection molder. On the line of introducing the scCO2 from a high pressure CO2 generator to the cavity, low molecular solutes, such as dye pigments, alcohols, and monomers, low molecular weight polymers are dissolved into the scCO2. The cavity is pressurized and filled with the scCO2 dissolving the substance (Fig. 8A). Then, the molten polymer is injected into the pressurized cavity (Fig. 8B). When the polymer is flowing in the cavity, the CO2 and the low molecular solutes diffuse from the melt front of the flowing polymer. In the mold cavity, the injected polymer shows a so-called fountain flow. As shown in Fig. 9, the polymer at the melt front moves from the centerline of the stream to the cavity wall when it runs in the cavity. The polymer surface, where CO2 as well as low molecular solute are dissolved, is carried toward the cavity walls on the fountain flow and is solidified at the wall, which normally remains below the transition temperature (Fig. 8C). The viscosity of the polymer surface containing the CO2 is reduced owing to the plasticization effect. The reduced viscosity increases the transferability of polymer against the mold shape and reduces surface roughness. Furthermore, the dissolved CO2 in polymer eventually diffuses out to atmosphere while low molecular substance remains in the products. This produces solidified

S

CO2 + Dye Additives Nucleation agent

Polymer

Mold wall

Fig. 9 Fountain flow behaviors of polymer and diffusion of CO2 and additives into polymer melt front in a mold cavity. (From Ref.[9].)

plastic products whose surface is modified by the low molecular substances.

Core-Back Molding Scheme In the aforementioned direct injection scheme, the contact time, which is a period that the polymer front contacts scCO2, becomes a critical factor of determining the thickness of modified polymer layer. The contact time, which is almost equivalent to the injection period at the method, cannot be made longer due to

Fig. 8 Direct injection scheme of surface coating with scCO2. (From Ref.[9].)

2904

Supercritical CO2-Assisted Surface Coating Injection Molding

Fig. 10 Core-back injection molding of surface coating with scCO2. (From Ref.[9].)

the fact that the lower injection speed reduces processability of the polymer. In order to manipulate the contact time of CO2, solutes to polymer, and control the thickness of the polymer layer modified by a low molecular solute, the other scheme, which is named the core-back molding scheme, is developed as shown in Fig. 10. The polymer is injected into the cavity (Fig. 10A). At a certain moment in time, part of the mold is shifted back so as to make a narrow space while keeping the pressure sealing, and the scCO2 containing the low molecular substance is introduced into the space (Fig. 10B). Then, part of the mold is clamped back and the pressure of CO2 in the space is increased so as to dissolve both the CO2 and the low molecular substance in the plastic in the mold

(Fig. 10C). By varying the volume of the space and the pressurizing time, the thickness of the modifying layer can be changed.

EXPERIMENTAL RESULTS Fig. 11 shows the experimental setup of the CO2assisted surface coating injection molding. A 35-ton injection molder was used. The molder cavity was sealed so that the cavity pressure can be kept at a higher value than critical pressure 7.38 MPa. The high pressure CO2 was generated, by pumping the liquid CO2, and introduced into a buffer tank. A low molecular compound to be dissolved in scCO2, which

Fig. 11 Experimental setup for a surface coating injection molding. (From Ref.[9].)

Supercritical CO2-Assisted Surface Coating Injection Molding

2905

0.45 520

y 0.8

540

S

LDPE 560

PP

0.6 500

X Value [–]

580 0.4

0.4

600 620 650 770mm

0.2 480 470 450 380 0.2

0

0

0.4

X value

0.35

0.6

0.8 x

1

0.3 0 x

10 Buffer tank Pressure [MPa] y

20 z

0MPa

15MPa

18MPa

in this experiment was Oil Red-O dye pigment, was placed inside the buffer tank. By keeping the buffer tank temperature higher than critical temperature and pressure of CO2, the low molecular compound is dissolved in scCO2 in the tank. A gas line for supplying scCO2 and low molecule mixture is connected from the buffer tank to the mold. The direct injection scheme of surface coating was performed by this experimental setup. Low density polyethylene (LDPE) and polypropylene (PP) were used as sample polymers and a disperse dye pigment, Oil Red-O, whose melting point is 120
C, was used as a low molecule to be dissolved in scCO2 and coat the surface of LDPE and PP of injection products. The buffer tank temperature was kept at 60
C and the molder barrel temperature was 220
C, which is a normal temperature setup. By varying scCO2 pressure at the buffer tank, the direct injection scheme was performed. The color of injected plastic product was evaluated using the x-y-z color system as shown in Fig. 12, which also shows a result of experiments. The X-value indicates the degree of red color. As can be seen, the red color of the plastic products’ surface becomes darker as the pressure of scCO2 increases over critical temperature.

CONCLUSIONS In this entry, the fundamental aspects of CO2=polymer systems, such as solubility, diffusivity of CO2 in poly-

Fig. 12 The darkness of the red-colored surface of plastics. (View this art in color at www.dekker.com.)

mers, and CO2 induced viscosity reduction, as well as the basic principles of supercritical CO2 technique in polymer processing, were discussed. It was shown that utilization of these fundamental aspects of scCO2 and polymer systems is one of the promising schemes and has great potential for inventing a new polymer processing technology. As an example, a novel CO2assisted surface coating injection-molding technique was introduced, which utilizes the plasticization and the solvation effects of scCO2 for processing several plastic products. Two supercritical CO2-assisted surface coating injection-molding schemes were proposed. By the developed technique, the surface of plastic products could be modified with low molecular substances and the new functionality can be added to plastic products.

REFERENCES 1. Kajimoto, O. Solution structure in supercritical fluids. In Supercritical Fluids; Arai, Y., Sako, T., Takebayashi, Y., Eds.; Materials Processing, Springer: Berlin, 2001; 1–65. 2. Adshiri, T.; Arai, K.; Kitamura, M.; Masuoka, H.; Sako, Y.; Takishima, S. Material processing using supercritical fluids. In Supercritical Fluids; Arai, Y., Sako, T., Takebayashi, Y., Eds.; Materials Processing, Springer, 2001; 280–345. 3. Okamoto, K.T. Microcellular Processing; Hanser: Munich, 2003.

2906

4. Baldwin, D.F.; Park, C.B.; Suh, N.P. A microcellular processing study of poly(ethylene terephthalate) in the amorphous and semi-crystalline states. Polym. Eng. Sci. 1996, 35 (4), 1446–1453. 5. Kazarian, S.G. Polymer processing with supercritical fluids. Polym. Sci. Ser. 2000, 42 (1), 78–101. 6. Yamaki, H.; Matsuura, Y. Method for InjectionMolding of Amorphous Resin JP Patent. 1-245252 A, 1999. 7. Yasuda, K.; Yamaki, H. Injection molding technology with CO2 as a plasticize of resin. Proc. of international workshop of foam processing and supercritical fluid aided polymer processing. Tokyo, 2003; 113–116. 8. Ohshima, M.; Yasuda, K. Surface Modifying Injection Molding Method, JP Patent 2003320556, 2003. 9. Ohshima, M. Supercritical CO2-assisted surface coating injection molding. Proceeding of ANTEC, Chicago, 2004; CD-ROM. 10. Durill, P.L.; Griskey, R.G. Diffusion and solution of gases in thermally softened or moten polymers; Part I. Development of technique and determination of data. AIChE J. 1966, 12, 1147–1151. 11. Durill, P.L.; Griskey, R.G. Diffusion and solution of gases in thermally softened or molten polymers; Part II. Relation of diffusivities and solubilities with temperature pressure and structural characteristics. AIChE J. 1969, 15, 106. 12. Liu, D.; Prausnitz, J.M. Solubility of gases and volatile liquids in polyethylene and in ethylenevinyl acetate copolymers in the region 124– 225
C. Ind. Eng. Chem. Fundam. 1976, 15 (4), 330–335. 13. Lee, J.G.; Flumerfelt, R.W. Nitrogen solubilities in low-density polyethylene at high temperatures and high pressures. J. Appl. Polym. Sci. 1995, 58 (12), 2213–2219. 14. Sato, Y.; Wang, M.; Takishima, S.; Masuoka, M.; Watanabe, T.; Fukasawa, Y. Solubility of butane and isobutane in molten polypropylene and polystyrene. Proceedings of 15th International Congress of Chemical and Process Engineering, Praha, Czech, Aug. 2002. 15. Sato, Y.; Yurugi, M.; Fujiwara, K.; Takishima, S.; Masuoka, H. Solubilities of carbon dioxide and nitrogen in polystyrene under high temperature and pressure. Fluid Phase Equilib. 1996, 125, 129–138. 16. Sato, Y.; Fujiwara, K.; Takikawa, T.; Sumarno; Takishima, S.; Masuoka, H. Solubilities and diffusion coefficients of carbon dioxide and nitrogen in polypropylene, high-density polyethylene, and polystyrene under high pressures and temperatures. Fluid Phase Equilib. 1999, 162, 261–276.

Supercritical CO2-Assisted Surface Coating Injection Molding

17. Sato, Y.; Takikawa, T.; Takishima, S.; Masuoka, H. Solubility and diffusion coefficients of carbon dioxide in poly(vinyl acetate) and polystyrene. Journal of Supercritical Fluids 2001, 19, 187–198. 18. Areerat, S.; Hayata, Y.; Katsumoto, R.; Kegasawa, T.; Egami, H.; Ohshima, M. Solubility of carbon dioxide in polyethylene=titanium dioxide composite under high pressure and temperature. J. Appl. Polym. Sci. 2002, 86 (2), 282–288. 19. Thran, A.; Kroll, G.; Faupel, F. Correlation between fractional free volume and diffusivity of gas molecules in glassy polymers. J. Polym. Sci. Part B: Polym. Phys. 1999, 37 (23), 3344– 3358. 20. Koga, T.; Seo, Y.E.; Hu, X.; Shin, K.; Zhang, Y.; Rafailovich, M.H.; Sokolov, J.C.; Chu, B.; Satija, S.K. Dynamics of polymer thin films in supercritical carbon dioxide. Euro Phys Lett. 2002, 60 (4), 559–565. 21. Nagata, T.; Tanigaki, M.; Ohshima, M. On-line NIR sensing of CO2 Concentration for polymer extrusion foaming processes. Polym. Eng. Sci. 2000, 40 (8), 1843–1849. 22. Gerhardt, L.J.; Manke, C.W.; Gulari, E. Rheology of polydimethylsiloxane swollen with supercritical carbon dioxide. J. Polym. Sci. Part B: Polym. Phys. 1997, 35 (3), 523–534. 23. Kwag, C.; Manke, C.W.; Gulari, E. Rheology of molten polystyrene with dissolved supercritical and near-critical gases. J. Polym. Sci. Part B: Polym. Phys. 1999, 37 (19), 2771–2781. 24. Lee, M.; Park, C.B.; Tzoganakis, T. Measurements and modeling of PS=supercritical CO2 solution viscosity. Polym. Eng. Sci. 1999, 39 (1), 99–109. 25. Gendron, R.; Daigneault, L. Rheology of Thermoplastic Foam Extrusion Process In Foam Extrusion; Lee, S.-T., Ed.; Technomic: Pennsylvania, 2000; 35–80. 26. Royer, J.R.; Gay, Y.J.; Desimone, J.M.; Khan, S.A. High-pressure rheology of polystyrene melts plasticized with CO2: experimental measurement and predictive scaling relationships. J. Polym. Sci. Part B: Polym. Phys. 2000, 38 (23), 3168– 3180. 27. Areerat, S.; Funami, E.; Hayata, Y.; Nakagawa, D.; Ohshima, M. Measurement and prediction of diffusion coefficients of supercritical CO2 in molten polymers. Polym. Eng. Sci. 2004, 44 (10), 1915–1924. 28. Nagata, T.; Areerat, S.; Ohshima, M.; Tanigaki, M. Measurement and prediction of CO2-induced viscosity reduction of polypropylene. Kagaku Kogaku Ronbunshu 2002, 28 (6), 739–745.

Supercritical Fluid Extraction (SFE) S Ram B. Gupta Department of Chemical Engineering, Auburn University, Auburn, Alabama, U.S.A.

INTRODUCTION A fluid is supercritical when it is compressed beyond its critical pressure (Pc) and heated beyond its critical temperature (Tc). Supercritical fluid technology has emerged as an important technique for supercritical fluid extraction (SFE). In many of the industrial applications, it has replaced conventional solvent-based or steam extraction processes, mainly due to the quality and the purity of the final product and environmental benefits.

SUPERCRITICAL FLUIDS AND THEIR PROPERTIES There are several fluids of interest as listed in Table 1. In this list, carbon dioxide is the supercritical fluid of choice because it is nonflammable, nontoxic, inexpensive, and has mild critical temperature. Hence, much of the attention has been given to supercritical carbon dioxide for practical extraction applications. No amount of compression can liquefy the supercritical fluid. In fact pressure can be used to continuously change the density from gas-like conditions to liquidlike conditions. Near the critical region, small changes in the pressure can give rise to large changes in the density. Fig. 1 shows how density of carbon dioxide is varied by pressure at different isotherms. In addition to density, diffusivity of the supercritical fluids is higher than that of liquid solvents, and can be easily varied. For typical conditions, diffusivity in supercritical fluids is of the order of 10
3 cm2=sec as compared to 10
1 for gases and 10
5 for liquids. Typical viscosity of supercritical fluids is of the order of 10
4 g=cm=sec, similar to that of gases, and about 100-fold lower than that of liquids. High diffusivity and low viscosity provide rapid equilibration of the fluid to the mixture to be extracted, hence extraction can be achieved close to the thermodynamic limits. However, the main extraction benefit of supercritical fluids is their adjustable density that provides adjustable solvent strength. The compounds of choice can be dissolved=extracted in the supercritical fluid at high pressure and then this fluid mixture is carried to another vessel where simple lowering of the pressure Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120007975 Copyright # 2006 by Taylor & Francis. All rights reserved.

precipitates the compound. A simple extraction scheme is shown in Fig. 2.

SUPERCRITICAL FLUID EXTRACTION SCHEMES Most of the industrial SFE is for the extraction from a solid matrix. Because of the challenge in the pumping of the solids to high-pressure extractor, the process is usually carried out in a semibatch mode. A number of high-pressure extractors are used in parallel; the solid mixture is filled, extracted, and emptied in a sequential fashion, allowing almost a continuous stream of the extract product. The lowering of pressure to precipitate the extract is not always necessary. For example, in the case of decaffeination of coffee beans, water can be used to extract caffeine from the CO2=caffeine mixture, as caffeine is readily soluble in water (Fig. 3). Additional benefits of the supercritical carbon dioxide extraction are: 1) oxygen-free system prevents oxidation of the extract; 2) low temperature minimizes the thermal degradation; 3) microbes or their spores are not soluble, hence aseptic extracts are obtained; and 4) solvent-free extract is obtained because CO2 is gas at ambient and is not retained by the extract. When designing SFE, the most important factor to consider is the solubility of the desired compound in supercritical carbon dioxide. Because of the benign (noninteracting) molecular nature of carbon dioxide, solubilities are usually small. Nonetheless, with the recycle of CO2, multiple passes can be used to achieve the desired extraction. For illustration, solubilites of two compounds in orange skin, linalool, and limonene are shown in Figs. 4 and 5.[1] The solubility is highly dependent on the pressure and temperature of the supercritical carbon dioxide. Solubility increases with the pressure owing to increase in the solvation power of the supercritical fluid. An increase in the temperature causes a drop in the density of the fluid (hence the solvation power); however, the substance itself is more volatile at the increased temperature. For the SFE, density, rather than pressure or temperature, appears to be the more natural variable for describing the solubilities. The solubility data in Fig. 4 are replotted vs. density in Fig. 5. 2907

2908

Supercritical Fluid Extraction (SFE)

Table 1 Critical constants and safety data for various supercritical solvents Supercritical fluid Ethylene

Tc ( C) 9.3

Pc (atm) 49.7

Safety hazard Flammable gas

Trifluoromethane (fluoroform)

25.9

46.9

Chlorotrifluoromethane

28.9

38.7

—

Ethane

32.3

48.2

Flammable gas

Carbon dioxide

31.1

72.8

—

Dinitrogen monoxide (laughing gas)

36.5

71.7

Not combustible but enhances combustion of other substances

Sulfur hexafluoride

45.5

37.1

—

Chlorodifluoromethane (HCFC 22; R 22)

96.4

48.5

Combustible under specific conditions

Propane

96.8

42.4

Extremely flammable

Ammonia

132.4

111.3

Flammable and toxic

Dimethyl ether (wood ether)

126.8

51.7

Extremely flammable

Trichlorofluoromethane (CFC 11; R 11)

198.0

43.5

—

Isopropanol

235.2

47.0

Highly flammable

Cyclohexane

280.3

40.2

Highly flammable

Toluene

318.6

40.6

Water

374.0

217.7

SOLUBILITY ENHANCEMENT USING COSOLVENTS The problem of the poor solubility of the extractant in supercritical carbon dioxide, can be overcome by the addition of a small amount of the cosolvent, such as methanol, ethanol, or acetone.[2] The carbon dioxide by itself is almost nonpolar (it has some polarity owing to its quadrupole moment); addition of a polar

Highly flammable —

cosolvent that is fully miscible with supercritical carbon dioxide can provide a more polar solubilizing environment needed for the extraction of polar substances. For example, solubility of fish liver oil in supercritical carbon dioxide increases by addition of small amounts of ethanol (Fig. 6).[3] Depending on the amount of the cosolvent added, critical temperature and critical pressure of the binary mixture also change.

Fig. 1 Density dependence of carbon dioxide.

Supercritical Fluid Extraction (SFE)

2909

S

Fig. 2 Schematic of a typical supercritical fluid extraction process.

RAPID EXPANSION OF SUPERCRITICAL SOLUTION TO OBTAIN MICROPARTICLES

APPLICATIONS OF SUPERCRITICAL FLUID EXTRACTION

Once the solute is dissolved in the supercritical fluid, if the recovery of the solute is done by rapidly expanding the fluid, the solute can be recovered as fine particles or droplets. The pressure can be reduced as fast as the speed of sound, hence very high supersaturation can be achieved in a fraction of seconds. The solute precipitates out as a microparticles (if solid) or microdroplets (if liquid). Hence, it is possible to perform both extraction and micronization in a single step.

The SFE has been applied in a variety of industrial applications including: 1) foods: flavor extraction and concentration of flavors and fragrances, processing essential oils and edible oils; infusion of flavors and fragrances into the solid matrices; 2) neutraceuticals: vitamin extraction, antioxidant extraction, and the concentration of active ingredients; 3) petroleum: propane deasphalting, residuum oil extraction; 4) polymers: removal of monomers and oligomers, infusion of compounds into the polymeric matrix, and removal

Fig. 3 Decaffeination of coffee beans using supercritical fluid extraction.

2910

Supercritical Fluid Extraction (SFE)

Fig. 4 Solubility of limonene and linalool in supercritical carbon dioxide. (From Ref.[1].)

of binder from powdered metals; 5) cleaning: precision machined components, silicon wafers, medical implants, and electronic components; 6) analytical: for extraction of the analytes from samples of food, cosmetics, polymers, and pharmaceuticals. Only a few of these applications are described here in detail.

DECAFFEINATION Out of the above six major application areas, the applications in food, flavor, and fragrance have been heavily adopted by the industry. A good example is the coffee industry. Caffeine contents of coffee and

tea are 3.5–5.0% and 1–3.5%, respectively, and depend on the soil, altitude, and climate of the plantation. There is a significant consumer demand for decaffeinated coffee that contains less than 0.4% caffeine. The SFE to decaffeinate coffee has been in industrial practice since more than two decades. Nearly 90% of the coffee consumed in the United States is now decaffeinated by supercritical CO2 extraction.[4] Caffeine is extracted from the green coffee beans and then the aroma is simply developed later by roasting. In the case of decaffeination of tea, the extraction is conventionally performed on black tea. Because the enzymes in the green tea are to be protected, as they are needed in the development of the flavor and color at a later

Fig. 5 Log of solubility vs. density shows a more linear behavior.

Supercritical Fluid Extraction (SFE)

2911

S

Fig. 6 Enhancement of fish oil solubility in supercritical carbon dioxide by addition of ethanol cosolvent, at 60
C and 250 bar. (From Ref.[3].)

stage, there were some concerns that supercritical CO2 may inactivate the enzymes. EXTRACTION OF FLAVORS AND FRAGRANCES The flavors and fragrances extracted using supercritical carbon dioxide are significantly different from those extracted using steam distillation or solvent extraction. The SFE extract can almost be viewed as a new product, as usually the amount extracted is higher (Table 2) and the composition of the extract is somewhat different (Table 3), as more aromatic molecules Table 2 Comparison of percent yields of flavor and fragrance extracts from various natural products Natural substance

Steam distillation (% yield)

Aniseed

2.1–2.8

7

Carrot

0.2–0.5

3.3

Cardamom

Supercritical CO2 (% yield)

4–6

7

Clovebud

15–17

22

Cumin

2.3–3.6

14

Ginger

1.5–3.0

4.6

Garlic

0.06–0.4

4.6

Oregano

3–4

5

Pepper

1.0–2.6

8–18

Rosemary

0.5–1.1

7.5

Sage

0.5–1.1

4.3

Vetiver

0.5–1.0

1.0

(From Refs.[5,6].)

are extracted. In many instances the extract using supercritical carbon dioxide is closer to the natural fragrance or flavor.[5–7] The higher SFE equipment cost can be easily offset by the higher yield of the product and the lower operating cost, as compared to steam distillation or solvent extraction.

EXTRACTION OF EDIBLE OIL There are several published studies on the use of supercritical carbon dioxide to extract oils from potato chips and other snack foods, motivated because of the increasing consumer awareness toward low-fat foods. About one-half of the oils in a potato chip containing about 40% oil can be extracted while maintaining the original flavor and texture. The extracted oil can be recovered by depressurization and reused in a subsequent frying operation. Supercritical carbon dioxide evenly dispersed the excess oil on the surface, thus removing the greasy taste. The processed fried snacks were found to have improved flavor and taste.[8]

POLYMER FRACTIONATION Supercritical fluid can be used to extract low-molecularweight polymers=oligomers from the given mixture, leaving the high-molecular-weight polymer behind. The pressure of the extract fluid is lowered sequentially in different vessels causing the precipitation of the polymer depending on its molecular weight. The lowestmolecular-weight polymer is precipitated the last. Thus, the broad-molecular-weight polymer mixture

2912

Supercritical Fluid Extraction (SFE)

Table 3 Composition of lavender extract by supercritical CO2 and steam distillation Chemical component

Supercritical CO2 extract (%)

1,8-Cineole

Steam distillation product (%)

Table 4 Fractionation of hydroxyl-terminated polybutadiene using supercritical propane at 130 C Fraction Original mixture

Pressure range (bar)

Mw

—

6,250

2.12

970

1.24

Mw/Mn

5.83

6.75

Linalool

25.29

35.31

1

34–35

Camphor

7.90

7.81

2

124–165

1,690

1.20

Borneol

2.30

2.98

3

152–207

2,540

1.23

4-Terpineol

3.79

3.34

4

193–207

3,300

1.17

34.69

12.09

5

207–234

4,110

1.19

3,7-Dimethyl acetate

3.08

4.38

6

234–262

5,010

1.16

b-Farnescene

2.23

1.00

7

248–276

6,010

1.28

a-Bisabolol

2.09

3.76

8

276–317

7,420

1.27

9

303–338

9,050

1.30

10

517–552

21,540

1.83

Linalyl acetate

(From Ref.[7].)

can be separated into the cuts of narrower-molecularweight fractions. Carbon dioxide usually has extremely poor solubility for the polymers; however, it can be easily used to remove residual solvent or monomers. Also, the fractionation of softer polymers such as poly(ethylene glycol) is feasible using carbon dioxide.[9] Propane, ethylene, ethane fluids are needed for the fractionation of commercial polymers including poly(methyl methacrylate), polyethylene using high pressures.[10] Isothermal pressure profiling in the liquid polyethylene=supercritical propane regime yielded 14 fractions with narrow polydispersity. This process can be used as an alternative to temperature rising elution fractionation, which was developed to fractionate linear low-density polyethylene on the basis of short-chain branching. Supercritical fluid fractionation of hydroxylterminated polybutadiene was investigated by Watkins and Krukonis, as this polymer is used as a binder in solid propellants and the molecular weight plays an important role in the performance of the explosives.[11] The original polymer had a high polydispersity (ratio of weight average molecular weight to the number average molecular weight). On fractionation, 10 different cuts of narrower distribution were obtained in the pressure range 34–552 bar (Table 4).

(From Ref.[11].)

is a more attractive option for the fractionation of medium to heavy waxes.

EXTRACTION FROM ALCOHOLIC BEVERAGES In addition to extraction from solids, supercritical fluids can be used to extract aromatic molecules from liquids. Sen˜ora´ns et al. have utilized carbon dioxide to extract high-quality brandy aroma using a countercurrent supercritical fluid extractor.[13] The aroma quality is influenced by the extraction conditions. Medina and Martinez studied alcohol removal from beverages using supercritical carbon dioxide, to produce beverages with low-alcohol content but sufficient flavor, because of three key benefits: 1) water and salts are not appreciably removed by the carbon dioxide; 2) proteins and carbohydrates are not extracted or denatured; and 3) there is a good control in the aroma recovery. The alcohol removal efficiency increases with the extraction pressure; raffinate alcohol concentration can be reduced up to 3 wt.% at 250 bar and 40 C, from 6.2 wt.% in the feed.[14]

SAFETY AND HEALTH ISSUES WAX FRACTIONATION Supercritical carbon dioxide, ethane, and propane have been examined for the fractionation of paraffin wax.[12] The original feed contained wax molecules with 10–35 carbon atoms. A narrower carbon distribution is needed in the printing ink, cosmetics, and pharmaceutical applications. Based on the cost analyses, vacuum distillation was proposed to be a cheaper option for light paraffin wax, whereas supercritical fractionation

When dealing with supercritical carbon dioxide, there are two safety and health issues that are to be kept in mind when designing and operating the extractor: 1) the high pressure involved requires that the personnel are protected from the plant by proper isolating walls and 2) if carbon dioxide is released in the closed atmosphere, it can lead to asphyxiation, as it can replace the oxygen in the surroundings. If one is using a more flammable supercritical fluid (e.g., propane),

Supercritical Fluid Extraction (SFE)

then the flammability becomes an additional concern. The concentrations should be not be allowed to fall between the explosive limits. In the case of ammonia, toxicity hazard also exists.

CONCLUSIONS Supercritical fluid extraction offers several advantages over conventional extraction processes. The extraction is carried out at high pressures and then the extract is usually recovered by lowering the pressure, as the solubility is a strong function of fluid pressure. The compositions of the extracts are different from those from the liquid extraction. Supercritical fluid extraction has been well accepted for coffee decaffeination and is being applied in other food, cosmetics, and pharmaceutical applications. Supercritical carbon dioxide is an environmentally benign nonflammable fluid.

REFERENCES 1. Berna, A.; Chafer, A.; Monton, J.B. Solubilities of essential oil components of orange in supercritical carbon dioxide. J. Chem. Eng. Data 2000, 45 (5), 724–727. 2. Dobbs, J.M.; Wong, J.M.; Lahiere, R.J.; Johnston, K.P. Modification of supercritical fluid phase behavior using polar cosolvents. Ind. Eng. Chem. Res. 1987, 26 (1), 56–65. 3. Catchpole, O.J.; Grey, J.B.; Noermark, K.A. Fractionation of fish oils using supercritical CO2 and CO2 þ ethanol mixtures. J. Supercrit. Fluids 2000, 19 (1), 25–37. 4. McHugh, M.A.; Krukonis, V.J. Supercritical Fluid Extraction: Principles and Practice; Butterworth–Heinemann: Boston, 1994. 5. Mukhopadhyay, M. Natural Extracts Using Supercritical Carbon dioxide; CRC Press: Boca Raton, FL, 2000.

2913

6. Moyler, D.A. Extraction of flavours and fragrances with compressed CO2. In Extraction of Natural Products Using Near-Critical Solvents; King, M.B., Bott, T.R., Eds.; Blackie Academic & Professional (an imprint of Chapman Hall): Glasgow, U.K., 1994. 7. Reverchon, E.; Porta, G.D.; Senatore, F. Supercritical CO2 extraction and fractionation of lavender essential oil and waxes. J. Agric. Food Chem. 1995, 43 (6), 1654–1658. 8. Daneshvar, M.; Gulari, E. Supercritical-fluid fractionation of poly(ethylene glycols). J. Supercrit. Fluids 1992, 5 (2), 143–150. 9. Inada, S.; Joji, O.; Giichi, T.; Schoichi, T.; Toshio, I.; Katsuyoshi, K.; Hirofumi, O.; Norio, Y. Improvement of taste of fried snack with subcritical or supercritical carbon dioxide. Jpn. Kokai Tokkyo Koho 1989, 4. CODEN: JKXXAF JP 01243944 A2 19890928 Heisei. 10. Watkins, J.J.; Krukonis, V.J.; Condo P.D., Jr.; Pradhan, D.; Ehrlich, P. Fractionation of high density polyethylene in propane by isothermal pressure profiling and isobaric temperature profiling. J. Supercrit. Fluids 1991, 4 (1), 24–31. 11. Crause, J.C.; Nieuwoudt, I. Paraffin wax fractionation: state of the art vs. supercritical fluid fractionation. J. Supercrit. Fluids 2003, 27 (1), 39–54. 12. Watkins, J.J.; Krukonis, V.J. Supercritical Fluid Processing of Propellants, Technical Report PL-TR-91-3003; OLAC, Phillips Laboratory (AFSC): Edward Air Force Base, CA, December 1990. 13. Sen˜ora´ns, F.J.; Ruiz-Rodrı´guez, A.; Iba´n˜ez, E.; Tabera, J.; Reglero, G. Isolation of brandy aroma by countercurrent supercritical fluid extraction. J. Supercrit. Fluids 2003, 26 (2), 129–135. 14. Medina, I.; Martinez, J.L. Dealcoholisation of cider by supercritical extraction with carbon dioxide. J. Chem. Technol. Biotechnol. 1997, 68, 14–18.

S

Supercritical Fluid Technology: Reactions S Aydin K. Sunol Sermin G. Sunol Naveed Aslam University of South Florida, Tampa, Florida, U.S.A.

INTRODUCTION Significant and steady inroads toward wider and more effective utilization of supercritical fluids have been made over the past three decades. Although the widely stated suitability for synthesis of highvalue-added differentiated products is on the mark, continued interest in the displacement of key basic chemical processes with greener and superior supercritical ones is equally worthy owing to the longevity of such processes and the resulting socially responsible manufacturing practice in an era when the new major technology shifts and construction of new plants are far in between. Furthermore, a new wave of second-generation supercritical technologies started to emerge, particularly in materials, bringing forth new roles for dense gases. The motivation of this entry is to assess the current status of the reactions in supercritical fluids in an effort to extricate the challenges both in the current practice and in the potential areas into which we have not as yet ventured.

BACKGROUND One of the significant paradigm shifts in chemical processing for the new millennium is the increased use of environmentally benign technology. The effectiveness of green solvents such as supercritical water and supercritical carbon dioxide for carrying out reactions, difficult separations, and materials processing is naturally very promising. These solvents are preferred due to their low cost, nontoxicity, nonflammability, and thermal stability. The effectiveness of the supercritical solvents is related to their state, critical temperature, and pressure. Supercritical fluids have the mobility of gases and the dissolving power of liquid solvents resulting in efficient penetration into porous matrices, high mass transfer and reaction rates, and high solvency. Furthermore, these properties are extremely sensitive to perturbations in temperature, pressure, and composition resulting in innovative processing concepts with tunable Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120039329 Copyright # 2006 by Taylor & Francis. All rights reserved.

performance parameters suitable for devising creative processing strategies. Obviously, synergy between the physical characteristics of solvent and the conditions favorable for the desired chemistry is of paramount importance for the success of the application. Thus, over the last three decades, a spate of supercritical processes have been developed, particularly for manufacture of high-value-added products that are superior in performance and exhibit conscious regard for a more socially responsible manufacturing practice. Despite the higher capital charges associated with relatively high pressures, the necessity to often add a new component into the processing environment, and operational challenges at conditions foreign to most process engineers, the interest in supercritical fluids had been growing steadily since the early 1980s beyond the select number of areas. We see applications in the food and beverage industries, pharmaceuticals, biomedical, microelectronic industries, textiles, forest products, petrochemicals, chemicals, environmental cleanup, syn-fuel production, polymeric materials, ceramics, auto industry, coatings and paint industry, energetic materials, and fuels. More polymerization reactions carried out at supercritical conditions, select biomass conversion supercritical fluid technologies for hydrogen production, wider use of supercritical water oxidation processes, portfolio of self-assembly applications, a spate of opportunities in process intensification, many supercritical fluid aided materials synthesis applications, and numerous reactions for synthesis of specialty chemicals are expected for years to come. The entry will start with fundamentals, focusing on the unique properties and the resulting opportunities covered in a generic fashion. Applications in different domains and reactive environments will follow. Some concluding remarks will summarize reactions in supercritical fluids and provide some perspectives for the years to come. Supercritical as well as other reactions can be facilitated through supercritical fluid aided synthesis and functionalization of catalysts. These are discussed elsewhere in the Encyclopedia. 2915

2916

Supercritical Fluid Technology: Reactions

FUNDAMENTALS

Enhanced Reaction Rate

Properties and Opportunities

Reactions in supercritical media utilize high pressures. Therefore, the effect of pressure on reaction equilibrium as well as reaction rate plays an important role in supercritical phase reactions. Supercritical fluids that exhibit very high negative activation volumes for certain reactions will improve the rate and equilibrium conversion of the reaction. The kinetics of the reaction can be explained in terms of the transition state theory. According to the theory, the reaction occurs via a transition state species M
and the generic elementary reaction can be written as:

The unique properties of supercritical fluids make them attractive as a reaction medium as well. Although reactive supercritical fluids processes such as extraction of coal and supercritical water oxidation started to emerge in the 1970s and early 1980s, the first review on reactions in supercritical fluids was presented by Subramanian and McHugh, followed by many more recent and thorough reviews.[1–3] Earlier work on reactions at high pressure provided both fundamental understanding and technical readiness.[4–10] The motivations for using supercritical fluids solvents in chemical reactions are many and most of the opportunities are due to the unique properties including density tuning as shown in Fig. 1.[5,11] In this section, these unique properties will be briefly discussed in the context of reactions, followed by the resulting opportunities and other related issues. There are several reasons for carrying out the reactions at supercritical conditions. Naturally, some of the reasons are coupled. Nevertheless, they, in general, relate to favorable transport properties, unique solvency characteristics, favorable kinetic considerations, and their sensitivity to operating conditions (manipulated variables). These unique properties lead to opportunities in process synthesis, process intensification, and controllability. These advantages, coupled with the environmentally friendly nature of these processes, make reactions in supercritical fluids attractive. The effect of these properties on opportunities for these favorable reaction environments is summarized in Table 1.

aA þ bB þ    $ M
! Products

ð1:1Þ

The effect of pressure on the rate constant is given as:


@ ln kx @P



¼ T

¼




@ ln Kx
@P


DV þ RT



þ

T







@ ln k @P

@ ln k @P 



T

T

where kx , rate constant; P, pressure; T, temperature; Kx
mole fraction based equilibrium constant for reaction involving reactants and transition state kT ,
isothermal compressibility; DV , activation volume (difference between partial molar volume of activated

complex and reactants), DV ¼ V  aV A  bV B     ; R, universal gas constant. The rate constant in the above equation is expressed in terms of pressure independent units (mole fraction). If the rate constant is expressed in terms of pressure

Fig. 1 Supercritical fluid region and density tuning.

Supercritical Fluid Technology: Reactions

2917

Table 1 Effect of properties on opportunities for favorable reaction environments in supercritical media

Homogenization

Tunability and control

Increased catalyst activity

Enhanced mass transfer

Enhanced reaction rate

Significant

Significant

Most significant

Not significant

Significant

Not significant

Enhanced solubility and selective solvation

Most significant

Significant

Significant

Significant

Significant

Significant

Transport properties (mobility)

Significant

Significant

Significant

Most significant

Significant

Significant

Sensitivity to operating conditions

Significant

Most significant

Significant

Significant

Significant

Significant

dependent units (such as concentration), the relevant equation is:


@ ln k @P



T




DV þ ¼  RT




@ lnk @P



T

þ kT ð1  a  b    Þ

If the volume of activation is positive, the reaction is hindered by pressure. However, for high negative values of the volume of activation, the pressure enhances the rate of the reaction. Therefore, supercritical fluids that exhibit very high negative activation volumes for certain reactions will improve the rate of the reaction. The volume of reaction, rather than activation, is crucial in determining the effect of pressure on the equilibrium constant:


@ ln Kx @P



T;x

¼ 

DV f RT

where DV f is the reaction volume (difference between partial molar volume of products and reactants). If the equilibrium constant is expressed in terms of pressure dependent units (such as concentration), the relevant equation is:


@ ln K @P



T;x

¼ 

X DV f þ kT ui RT

where ui is the stoichiometric coefficient. As the above equation implies, supercritical fluids that exhibit very high negative activation volumes for certain reactions will improve the equilibrium conversion of the reaction. Enhanced Solubility The enhancement factors for solubility of compounds, over ideal solubility, in supercritical fluids is typically

Increased selectivity

Ease of downstream separation

around 104–106 and is discussed elsewhere. This enhancement implies that very heavy macromolecules can be solubilized in supercritical fluids to react or be selectively removed from the reaction environment. In another class of reactions, light noncondensable components such as hydrogen or oxygen will dissolve in supercritical fluids for effective hydrogenation or oxidation. Supercritical fluids can also dissolve in and expand liquids for a more favorable reaction and processing environment. This solubility behavior is synchronized with favorable reaction conditions through judicious selection of solvents and=or cosolvents.

Transport Properties (Mobility) The supercritical fluids exhibit gas-like viscosities, diffusivities, and liquid-like densities. These favorable transport properties lead to enhanced mass transfer, permeation, and wetting characteristics. The mass transfer limited multiphase reactions will benefit from reduction of a number of phases, as in the case of most oxidation, hydrogenation, or replacement of the more viscous liquid phase with a supercritical or a less viscous expanded liquid phase. The mobility combined with tunability results in effective maintenance of catalyst activity in heterogeneous catalysis. Sensitivity to operating conditions The solubility including retrograde behavior, number and nature of coexisting phases, rate of phase creation, reaction rates, and transport properties are very sensitive to the operating conditions such as temperature, pressure, composition including cosolvents, and other external fields such as electric and magnetic as well as the rate of change of these operating conditions. The sensitivities to manipulated variables are usually coupled leading to novel processes and enable tuning of the process to impart the desired product properties.

S

2918

Dramatic reduction of surface tension and reduction of dielectric constant affect the interfacial transport and solubility as well as reaction rate tuning, respectively.

Supercritical Fluid Technology: Reactions

combination also tunes the reaction to adjust the product distribution.[12]

Enhanced Mass Transfer Homogenization Reactions that otherwise would be carried out in more than one phase (heterogeneous reactions) can be transformed to homogeneous ones, with the aid of supercritical fluids, where interphase transport limitations are eliminated. This is realized due to enhanced solubilities of the supercritical fluids. Typical examples are reactions in water (supercritical water can solubilize organic compounds), homogeneous catalytic reactions, reactions of organometallic compounds. Homogenizing one compound more than the other may also affect relative rates in complex reactions and enhance the selectivity. In supercritical solutions, in a microscopic sense, molecules are nonuniformly distributed. There is aggregation of the solvent molecules around the solute and clusters are formed. The local clustering of solutes or solvents under supercritical condition increases the local concentration of the substrate or the catalyst in the solution and may result in enhanced reaction rates. Supercritical fluids aid rapid diffusion of solutes or weakening of the solvation around reacting species. This may result in changes of the reaction pathways. The homogenization could be effectively coupled with nucleation=crystallization to allow in-unit separation and phase transfer catalysis. Increased Catalyst Activity Some heterogeneous catalytic reactions are carried out in supercritical phase, to increase catalyst activity and life through in situ regeneration of surfaces with tuning of operation conditions. For example, supercritical fluids are capable of dissolving carbon, which may be irreversibly deposited on the catalyst otherwise. Tunability and Control Some properties of supercritical fluids can be monitored (manipulated) continuously by adjusting the temperature and pressure or density of the fluid. Dielectric constant is such a property and the solvent’s dielectric constant can influence the rate of the reaction. Supercritical fluids can be combined with polar cosolvent, which enhances the solubility selectivity and sensitivity of the environment to manipulated variables such as temperature and pressure. This

In many instances, reaction rates are limited by diffusion in the liquid phase. The rate of these reactions can be increased if the reaction is carried out in the supercritical phase. Typical examples are enzymecatalyzed reactions as well as very fast reactions such as some free radical reactions. Selectivity considerations usually dominate in complex reactions. If some steps of the complex reaction are controlled by diffusion, changing the diffusivity changes relative rates of the reaction steps and affects the selectivity. The mass transfer rates are also enhanced in porous media allowing effective in situ regeneration and removal of products=reactants as they become solubilized. The reduced viscosities play an important role here.

Ease of Downstream Separation Another utility of supercritical fluids as the reaction medium is fractionation and purification of the products or removal of unreacted reactants from the product stream. Supercritical fluids can be used as a solvent or as an antisolvent in this instance. In the case of homogeneous catalysis, employing supercritical fluids enables complete recovery of the expensive transition metal species. Also, these species may have environmentally unfriendly effects if not recovered completely. The combination of ionic liquids with the supercritical fluid can lead to product isolation as well as catalyst immobilization. Thus, catalysts can be recycled batchwise.[7]

Induced Reaction Selectivity Supercritical carbon dioxide introduces changes in the selectivity of the reactions. These changes include chemical selectivity, as well as stereoselectivity. The selectivity changes originate from the solvent as well as the modification of the catalysts with the carbon supercritical carbon dioxide.

Energy Demand Reduction The mechanical and thermal energy requirements depend on operating conditions, excursions in these conditions throughout the process, and heat integration. Naturally, we do have pressurization requirements

Supercritical Fluid Technology: Reactions

for all supercritical reactions, particularly high for supercritical water. Thermal energy loads for reactions at elevated conditions such as reactions in supercritical water are rather high as well. The energy demand reduction opportunities are due to favorable transport properties and enhanced rates as well as sensitivity to operating conditions bringing efficient recovery. In instances where supercritical reactions displace mass transfer limited reactions, less stringent agitation or no agitation results in reduction of mechanical energy requirements. When enhanced reaction rates are coupled with higher mass and heat transfer rates, energy required and volume per unit product decrease substantially. Furthermore, efficient recycling and ease in downstream separation imply reduction in energy consumption. More systematic studies in heat and mass integration of these processes will eventually be done as industrial deployment of supercritical fluids reaches a critical threshold. Safety Safety issues often relate to process safety as well as end-product safety. The inherent process safety concern with supercritical fluids=reactions is due to pressurized inventory. For operation of these processes, effective pressurization, depressurization, and relief systems are a prerequisite. When these safety concerns are adequately handled, one sees the safety advantages that supercritical reactions and processes bring about. In most supercritical environments, which are water or carbon dioxide based, we replace flammable and toxic alternatives. The enhanced rates and increased throughputs imply smaller equipment and inventories, further reducing the safety hazards. Furthermore, gases like hydrogen, carbon monoxide, and oxygen can be handled more safely in the presence of supercritical carbon dioxide. This is crucial in hydrogenation as well as oxidation reactions. In the case of product safety, the end products are inherently contaminant free, posing less of a safety concern. Naturally, a systematic study on safety of highpressure reactions, particularly for runaway reactions is well overdue. APPLICATIONS Supercritical fluids have been utilized as a reaction medium from basic industries such as syn-fuels, biomass conversion, environmental remediation to high-value-added specialty chemicals and materials. An exposition of the underlying mechanism along with application domain will follow in this section. Homogeneous and heterogeneous reactions will be followed by biochemical and polymerization reactions, all

2919

with supercritical fluids. The typical examples for which original references are too numerous to cite individually are summarized in Tables 2 to 4. Homogeneous Reactions in Supercritical Solvents Homogeneous reactions carried out in supercritical fluids can be either catalytic or noncatalytic. The objective of carrying catalytic and noncatalytic reactions in supercritical fluids is to increase the overall rate of the reaction by eliminating the interfacial transport effects.

Homogeneous Reactions Catalyzed by Organometallic Compounds Homogeneous catalysts have advantages over heterogeneous catalysts such as the possibility of carrying out the reaction under milder conditions, higher activity, and selectivity, ease of spectroscopic monitoring, and controlling the tunable reaction sites. A recent review by Noyori et al. discusses homogeneous catalytic reactions under supercritical conditions.[13] Examples of homogeneous catalytic reactions carried out under supercritical conditions are summarized in Table 2. Because most organic reactants and products are not soluble in water, it cannot be utilized as the solvent in most catalytic reactions, although most catalytic materials are soluble in water. Therefore, liquid solvents for homogenization of catalytic reactions are usually organic solvents. An alternative to environmentally unacceptable organic solvents is a supercritical solvent that has several advantages over organic solvents. The aforementioned advantages are increased reaction rate, higher selectivity, and ease in separation of reactants, products, and the catalyst after the reaction. Because the properties of supercritical solvents are sensitive to operating conditions, reaction rate and selectivity are more readily tunable when the reaction is carried out in supercritical solvents. Carbon dioxide is the supercritical solvent that is used most in homogeneous catalytic reactions. In addition to being environmentally acceptable (nontoxic, nonflammable), carbon dioxide is inert in most reactions, is inexpensive, and is available in large quantities. Its critical temperature is near ambient. Supercritical carbon dioxide dissolves nonpolar, nonionic, and low molecular mass compounds. However, addition of cosolvents enhances the solubility of many compounds in supercritical carbon dioxide. When homogeneous reactions are carried out under supercritical conditions, liquid gas interfacial transport

S

2920

Supercritical Fluid Technology: Reactions

Table 2 Homogeneous reactions in supecritical fluids Reaction Hydrogenation CO2 to formic acid Asymmetric hydrogenation of tiglic acid Asymmetric hydrogenation of enamides Asymmetric hydrogenation of imines Cyclopropene 3,3-Dimethyl-1,2-diphenyl cyclopropene Isoprene

Supercritical medium

Catalyst

CO2 CO2 CO2 CO2 CO2 CO2 CO2

Ruthenium(II) phosphine complex Ruthenium catalyst Cationic rhodium complex Iridium complex Manganese catalyst Manganese catalyst Rhodium catalyst

CO2 CO2 CO2 CO2 CO2 CO2 CO2 CO2

Ruthenium catalyst Ruthenium catalyst Cobalt catalyst Palladium catalyst Nickel catalyst Nickel catalyst Cobalt catalyst Cobalt catalyst

Water Water Water

None None=NaOH Acid

Oxidation Cyclohexane Methylacrylate Alkene epoxidation 2,3-Dimethylbutene epoxidation Cyclooctene epoxidation Cyclohexene epoxidation 2,3-Dimethylbutene epoxidation Phenols Ethanol 2-Propanol 2-Butanol Chlorinated hydrocarbons

CO2 CO2 CO2 CO2 CO2 CO2 CO2 Water Water Water Water Water

Iron catalyst Palladium catalyst Molybdenum catalyst Molybdenum catalyst Molybdenum catalyst Molybdenum catalyst Molybdenum catalyst None None None None None

Hydroformylation 1-Octene Propylene Styrene Carbonylation of arylhalides Carbonylation of alkynes and alkenes

CO2 CO2 CO2 CO2 CO2

Rhodium catalyst Cobalt carbonyl Rhodium catalyst Palladium catalyst Cobalt catalyst

Isomerization 1-Hexene to 2-hexene

CO2

Iron catalyst

Water Water

None=H2SO4=NaOH Acid

Hydrolysis Esters to carboxylic acids and alcohols Nitriles to amides and then to acid Butyronitrile Polyethyleneterephthalate and polyurethane Diaryl ether to hydroxyarene Triglycerides into fatty acid

Water Water Water Water Water Water

Autocatalytic Autocatalytic Autocatalytic None None None

Decomposition Cellulose and glucose decomposition Nitrobenzene 4-Nitroaniline 4-Nitrotoluene

Water Water Water Water

None None None None

C–C bond formation Ring opening metathesis polymerization Ring closing metathesis of dienes to cyclic olefins Cyclotrimerization of alkynes Olefination of arylhalides Hydrovinylation of styrenes Synthesis of 2-pyrones Synthesis of cyclopentenes Cyclotrimerization of alkynes to substituted benzene derivatives Phenol and p-cresol alkylation Diels–Alder cycloaddition Ring opening of 2,5-dimethylfuran

Hydration=dehydration Conversion of tert-butyl alcohol to isobutylene Dehydration of cyclohexanol, 2-methyl cyclohexanol, 2-phenylethanol

Supercritical Fluid Technology: Reactions

2921

Table 3 Heterogeneous catalytic reactions in supercritical solvents Reaction

Supercritical medium

Catalyst

Hydrogenation Fats and oils Fats and oils Acetophenone, cyclohexene Ethyl-pyruvate Fischer–Tropsch synthesis Dibenzothiophene hydrodesulfurization

Propane, CO2 CO2 CO2 CO2 n-Hexane, n-pentane, propane Water

Supported platinum, palladium catalysts Nickel catalyst Palladium on polysiloxane Pt=Al2O3 Fe, Ru, Co=Al2O3, SiO2 Ni–Mo=Al2O3

Dehydrogenation Cyclohexanol

Water

Pt catalyst

Oxidation Toluene Propylene Benzyl alcohol Ethanol Methanol Propene Isobutane 1,4-Dichlorobenzene NH3 Acetic acid Pyridine Phenol Chlorophenol

CO2 CO2 CO2 CO2 CO2 SC reactant SC reactant Water Water Water Water Water Water

Co=Al2O3 CuI=Cu2O=MnO2 on Al2O3 Pd=Al2O3 Pt=TiO2 Iron oxide on Mo aerogel CaI2, CuI, Cu2=MgO on Al2O3 SiO2, TiO2, Pd=carbon V2O5 MnO2=CeO2 Cu=Zn=Co oxide supported catalyst MnO2=CeO2, Pt=Al2O3, MnO2=Al2O3 Cu=Zn=Co oxide, MnO2=CuO, MnO2 CuO supported on zeolites

Cracking Heptane

CO2

Zeolite

Rearrangement and isomerization 1-Hexene Xylene Cyclohexene to methyl-cyclopentene

CO2 and cosolvents SC reactant Water

Pt=Al2O3 Solid acid catalyst Solid acid catalyst

CO2 or SC reactant

Zeolite

CO2

Acid catalyst

SC reactant Butane, pentane

Zeolite Zeolite

Coupling Alkene-arene coupling

Water

Palladium catalyst

Esterification Oleic acid Glycerol and CO2

CO2 CO2

Biocatalyst Zeolite

Hydroformylation Oct-1-ene Propylene 1-Hexene

CO2 CO2 CO2

Supported rhodium catalyst Supported rhodium and iron catalyst Rhodium catalyst

Alkylation Benzene, ethylene, isopentane, isobutene isobutane Metislene propene, propan2-ol Disproportionation Toluene to p-xylene, benzene Ethylbenzene to benzene and diethylbenzene

is eliminated, which is an advantage for reactions such as hydrogenation, where diffusion of gas into the liquid may be limiting the reaction rate. In asymmetric hydrogenation reactions, hydrogen and the supercritical solvent are miscible and this results in better enantioselectivity.[6] In Diels–Alder reactions, the advantage

of the supercritical solvent is higher selectivity obtained rather than increased rate of the reaction due to the solvent. Although most of the oxidation reactions are carried out in supercritical water, recently, homogeneously catalyzed reactions in supercritical carbon dioxide are increasingly reported.

S

2922

Supercritical Fluid Technology: Reactions

Table 4 Polymerization reactions in supercritical solvents Polymerization mechanism

Substrate

Transition metal catalyzed, ring opening polymerization

Norbornene polymer, polycarbonate

Dispersion, cationic polymerization

Isobutylene polymer

Homogeneous=precipitation, cationic polymerization

Vinyl ether polymer

Homogeneous, free radical=cationic polymerization

Amorphous fluoropolymers

Precipitation, free radical polymerization

Vinyl polymer, semicrystalline fluoropolymers

Dispersion, free radical polymerization

Polyvinyl acetate and ethylene vinyl acetate copolymer

Homogeneous Reactions of Supercritical Water Homogeneous reactions carried out at supercritical conditions within water are organo-metallic reactions and Diels–Alder reactions. Reactions in supercritical water are well studied and will be described in the following section.[14] Supercritical water has a low dielectric constant compared to liquid water. The dielectric constant changes significantly with the density of the supercritical fluid. Also, the effect of hydrogen bonding is less pronounced at supercritical conditions, one consequence of which is high solubility of organics in supercritical water. When the homogeneous reaction is carried out in supercritical water, we get a high reactant concentration and the reaction can proceed in the absence of interphase mass transfer resistances. Also, ion dissociation constant of water is higher in the critical region and is lower as supercritical conditions are accessed. These properties also vary continuously in the supercritical region, so that they can be tuned during the reaction by changing temperature and=or pressure. Examples of the homogeneous reactions in supercritical water are included in Table 2. Use of acid or base catalysts enhances the rates of some of the reactions.[14] Homogeneous Noncatalytic Reactions in Supercritical Solvents The use of supercritical fluids as reaction media for organometallic species is also investigated.[15] Reactions include photochemical replacement of carbon monoxide with N2 and H2 in metal carbonyls, where the reaction medium is supercritical xenon. Also, photochemical activation of C–H bonds by organometallic complexes in supercritical carbon dioxide is investigated. More recent studies on photochemical reactions also include laser flash photolysis of metal carbonyls in supercritical carbon dioxide and ethane and laser flash photolysis of hydrogen abstraction reaction of triplet benzophenone in supercritical ethane and CHF3.[16,17]

Heterogeneous Reactions in Supercritical Solvents Heterogeneous reactions in supercritical fluids can be catalytic or noncatalytic. Catalytic heterogeneous reactions are carried out on solid catalysts and are of great importance in the chemical process industries. The advantages of carrying out these reactions in a supercritical medium include enhanced interphase and intraparticle mass and heat transfer and in situ regeneration of catalyst and are described in the next section. Catalytic supercritical water oxidation also utilizes favorable properties of supercritical water, which are also discussed. Other heterogeneous reactions that will be described are fuels processing reactions and conversion and treatment of biomass.

Heterogeneous Catalytic Reactions in Supercritical Solvents Obviously, a solid-catalyzed reaction takes place only on the active sites of the porous catalyst with the implication of some mass and heat transport steps prior to and after the reaction. The first step is the diffusion of the reactants through the film surrounding the catalyst particle to the external surface of the catalyst and diffusion of the reactants in the catalyst pore to the active site in the pores. These steps are limited by the diffusivity and viscosity of the reactants. In the case of a supercritical fluid phase reaction, the diffusivity is higher than the liquid diffusivity, the viscosity is less than the liquid viscosity, and, therefore, the rate of transfer to the active site will be higher. After adsorption, reaction, and desorption steps, the products have to diffuse out of the pore, and through the film surrounding the particle into the bulk fluid. The rates of these steps can be accelerated utilizing a supercritical medium for the reaction. Heat transfer effects are also important in a solid-catalyzed reaction. Higher thermal conductivity of supercritical fluids is an advantage as well.[18] For two-phase reactions (typically hydrogenation and oxidation reactions), the first step is the diffusion

Supercritical Fluid Technology: Reactions

of the gas reactant to and through the gas–liquid interface and then into the bulk liquid. This mass transfer limitation is also eliminated if the reaction is carried out in a supercritical medium since the reactants are going to be in a single phase. Supercritical fluids bring other benefits to solidcatalyzed reaction rate through eliminating or minimizing mass and heat transfer resistances. Supercritical solvents have the ability to regenerate the catalyst during the course of the reaction, which increases the catalyst life and activity, because undesirable deposits on the catalyst, such as carbon deposits, are soluble in the supercritical fluids. The rate of the intrinsic reaction is increased in supercritical fluids and by tuning the properties of the supercritical medium, one can control the selectivity.[18–20] Supercritical fluids may also bring opportunities in downstream separation of the reactants and products. Examples of solid-catalyzed reactions in supercritical fluids are given in Table 3. Catalytic supercritical water oxidation is an important class of solid-catalyzed reaction that utilizes advantageous solution properties of supercritical water (dielectric constant, electrolytic conductance, dissociation constant, hydrogen bonding) as well as the superior transport properties of the supercritical medium (viscosity, heat capacity, diffusion coefficient, and density). The most commonly encountered oxidation reaction carried out in supercritical water is the oxidation of alcohols, acetic acid, ammonia, benzene, benzoic acid, butanol, chlorophenol, dichlorobenzene, phenol, 2-propanol (catalyzed by metal oxide catalysts such as CuO=ZnO, TiO2, MnO2, KMnO4, V2O5, and Cr2O3), 2,4-dichlorophenol, methyl ethyl ketone, and pyridine (catalyzed by supported noble metal catalysts such as supported platinum).[21,22]

Heterogeneous Noncatalytic Reactions in Supercritical Solvents Use of the supercritical fluids as the reaction medium in syn-fuel processing is one of the earlier applications in the field. The advantage of the supercritical fluid as the reaction medium is again threefold. During thermal degradation of fuels (oil-shale, coal), primary pyrolysis products usually undergo secondary reactions yielding to repolymerization (coking) or cracking into gas phase. Both reactions decrease the yield of the desired product (oil). To overcome this problem, dense (supercritical) hydrogen donor (tetralin) or nonhydrogen donor (toluene) or inorganic (water) medium is used. Also, supercritical medium provides ease of transport in and out of the porous coal matrix. Finally, downstream processing (separation of the products) becomes an easier task, when supercritical medium is

2923

used. A review of the use of supercritical fluids in coal processing is given by Kershaw, while the mechanism is discussed by Sunol.[23,24] The forest product applications in this category include biomass conversion and delignification for pulping purposes. Both provide unique opportunities, the first for hydrogen generation and later to obtain pulp within reaction times that are reduced almost two orders of magnitude.[25,26]

Polymerization Reactions in Supercritical Solvents Supercritical carbon dioxide is a promising green alternative to traditional solvents in polymer synthesis because of gas-like transport properties and liquid-like solubility. It can be removed easily from the polymer solution by depressurization during drying of the polymer. It provides easy separation of the polymer from the unreacted monomers and catalysts. Finally, it also exhibits Lewis acid–base interactions with electron donating functional groups of polymer chains.[27] Examples of homogeneous and heterogeneous polymerization reactions carried out in supercritical carbon dioxide are given in Table 4. Butane, pentane, and propane are also used as the reaction medium in polymer synthesis because carbon dioxide is not a strong solvent for most polymers.[28] Furthermore, some polymerization reactions (such as polyethylene synthesis) are carried out under supercritical conditions of the monomer.

Biochemical Reactions in Supercritical Solvents Because of their tunable properties, supercritical solvents provide a useful medium for enzyme-catalyzed reactions.[29] The mechanism of enzyme-catalyzed reactions is similar to the mechanism described for solid-catalyzed reactions. External as well as internal transport effects may limit the reaction rate. Utilizing supercritical fluids enhances external transport rate due to increase in the diffusivity and therefore mass transfer coefficient. Internal transport rate depends on the fluid medium as well as the morphology of the enzyme. Supercritical fluids can alter both. Water is known to be essential for the enzyme activity.[30] Small amounts of water enhance enzyme activity; however, excess water hinders the rate of some enzyme-catalyzed reactions. The active site concentration on enzymes, hence the enzyme activity, is found to be higher in the presence of hydrophobic supercritical fluids (ethane, ethylene) compared to hydrophilic supercritical carbon dioxide.

S

2924

The effect of pressure on enzyme-catalyzed reactions can be explained in terms of the transition theory. Supercritical fluids that exhibit very high negative activation volumes for certain reactions are expected to improve the rate of the reaction. Although, supercritical carbon dioxide has the advantage of being nontoxic and abundant, it is practically immiscible with water. Therefore, supercritical fluids used as the reaction medium in enzyme-catalyzed reactions include fluoroform, sulfur hexafluoride, and ethane, while lipases are the enzymes utilized in such reactions.[31] In addition to advantages like higher initial reaction rate and higher conversion, supercritical fluids provide an easy separation of products and unreacted substrates. This ecologically safe recovery of the products is a unique advantage provided by supercritical carbon dioxide.[29]

CONCLUSIONS A variety of chemical and biological reactions involving supercritical fluid technology are being explored and developed. They include polymerization reactions, biomass conversion, hydrogen production, applications of supercritical water oxidation, self-assembly applications, synthesis of specialty chemicals, manufacture of materials with tailored properties, and much more. These developments and new ones are expected to mature and be commercially deployed in years to come.

REFERENCES 1. Bartle, K.D.; Martin, T.G.; Williams, D.F. Chemical nature of a supercritical-gas extract of coal at 350 degrees C. Fuel 1975, 54 (4), 226–235. 2. Modell, M. Processing Methods for the Oxidation of Organics in Supercritical Water. U.S. Patent 4,338,199, 1982. 3. Subramanian, B; McHugh, M.A. Reactions in supercritical fluids—a review. Ind. Eng. Chem. Res. 1986, 25, 1–12. 4. Savage, P.E.; Gopalan, S; Mizan, T.I.; Martino, C.J.; Brock, E.E. Reactions at supercritical conditions: applications and fundamentals. AIChE J. 1995, 41 (7), 1723–1778. 5. Clifford, A.A. Reactions in supercritical fluids. In Supercritical Fluids, Fundamentals and Applications; Kiran, E., Levent Sengers, J.M.H., Eds.; Kluvert Academic Publishers: Dordrecht, 1994; 449–479.

Supercritical Fluid Technology: Reactions

6. Ikariya, T.; Kayaki, Y. Supercritical fluids as reaction media or molecular catalysis. Catal. Surv. Jpn. 2000, 4, 39–50. 7. Leitner, W. Supercritical carbon dioxide as a green reaction medium for catalysis. Acc. Chem. Res. 2002, 35, 746–756. 8. Leitner, W. Carbon dioxide as an environmentally benign reaction medium for chemical synthesis. Appl. Organomet. Chem. 2000, 14, 809–814. 9. Oakes, R.S.; Clifford, A.A.; Rayner, C.M. The use of supercritical fluids in synthetic organic chemistry. J. Chem. Soc. Perkin Trans. 2001, 1, 917–941. 10. Sunol, A.K.; Sunol, S.G. Substitution of solvents by safer products and processes. In Handbook of Solvents; Wypych, G., Ed.; Chem Tec Publishing: Toronto, 2001; 1419–1459. 11. Wu, B.C.; Paspek, S.C.; Klein, M.T.; LaMarka, C. Reactions in and with supercritical fluids—a review. In Supercritical Fluid Technology; Bruno, T.J., Ely, J.F., Eds.; CRS Press: London, 1991; 511–524. 12. Eckert, C.A.; Bush, D.; Brown, J.S.; Liotta, C.L. Tuning solvents for sustainable technology. Ind. Eng. Chem. Res. 2000, 39, 4615–4621. 13. Jessop, G.P.; Ikariya, T.; Noyori, R. Homogeneous catalysis in supercritical fluids. Chem. Rev. 1999, 99, 475–493. 14. Savage, P.E. Organic chemical reactions in supercritical water. Chem. Rev. 1999, 99, 603–621. 15. Poliakoff, M.; George, M.W.; Howdle, S.M. Inorganic and related chemical reactions in supercritical fluids. In Chemistry under Extreme or Non-Classical Conditions; van Eldik, R., Hubbard, D.C., Eds.; Wiley: New York, 1997; 189–218. 16. Ji, Q.; Eyring, E.M.; van Eldik, R.; Johnston, K.P. Laser flash photolysis studies of metal carbonyls in supercritical CO2 and ethane. J. Phys. Chem. 1995, 99, 13461–13466. 17. Roberts, C.B.; Brennecke, J.F.; Chateauneuf, J.E. Solvation effects of reactions of triplet benzophenone in supercritical fluids. AIChE J. 1995, 41, 1306–1318. 18. Baiker, A. Supercritical fluids in heterogeneous catalysis. Chem. Rev. 1999, 99, 453–473. 19. Subramanian, B. Enhancing the stability of porous catalysts with supercritical reaction media. Appl. Catal. A Gen. 2001, 212, 199–213. 20. Subramanian, B.; Lyon, C.J.; Arunajatesan, V. Environmentally benign multiphase catalysis with dense phase carbon dioxide. Appl. Catal. B Environ. 2002, 37, 279–292.

Supercritical Fluid Technology: Reactions

21. Ding, Z.Y.; Frish, M.A.; Li, L.; Gloyna, E.F. Catalytic oxidation in supercritical water. Ind. Eng. Chem. Res. 1996, 35, 3257–3279. 22. Savage, P.E. Heterogeneous catalysis in supercritical water. Catal. Today 2000, 62, 167–173. 23. Kershaw, J.R. Supercritical fluids in coal processing. J. Supercrit. Fluids. 1989, 2, 35–45. 24. Sunol, A.K.; Beyer, G.H. Mechanism of supercritical extraction of coal. Ind. Eng. Chem. Res. 1990, 29, 842–849. 25. Vick Roy, J.R.; Converse, A.O. Biomass hydrolysis with sulfur dioxide and water in the region of the critical point. In Supercritical Fluid Technology; Penniger, J.M.L., Radosz, M., McHugh, M.A., Krukonis, V.J., Eds.; Elsevier: Amsterdam, 1985; 397–414. 26. Sunol, A.K. Supercritical Delignification of Wood. U.S. Patent 5,041,192, Aug 20, 1991.

2925

27. Kendall, J.L.; Canelas, D.A.; Young, J.L.; DeSimone, J.M. Polymerization in supercritical carbon dioxide. Chem. Rev. 1999, 99, 543–563. 28. Srinivasan, G.; Elliot, J.R. Microcellular materials via polymerization in supercritical fluids. Ind. Eng. Chem. Res. 1992, 31, 1414–1417. 29. Nakamura, K. Biochemical reactions in supercritical fluids. In Supercritical Fluid Processing of Food and Biomaterials; Rizvi, S.S.H., Ed.; Academic & Professional: London, 1994; 54–61. 30. Mesiano, A.J.; Beckman, E.J.; Russel, A.J. Supercritical biocatalysis. Chem. Rev. 1999, 99, 623–633. 31. Knez, Z.; Habulin, M. Compressed gases as alternative enzymatic-reaction solvents: a short review. J. Supercrit. Fluids. 2002, 23, 29–42.

S

Supercritical Water Oxidation S Ram B. Gupta Department of Chemical Engineering, Auburn University, Auburn, Alabama, U.S.A.

INTRODUCTION A fluid is supercritical when it is compressed beyond its critical pressure and heated beyond its critical temperature. Hence, water is supercritical at >374
C temperature and >22.1 MPa pressure. Supercritical water has liquid-like density and gas-like transport properties, and behaves very differently than it does at room temperature. For example, it is highly nonpolar, permitting complete solubilization of most organic compounds and oxygen. The resulting singlephase mixture does not have many of the conventional transport limitations that are encountered in multiphase reactors. However, the polar species present, such as inorganic salts, are no longer soluble and start precipitating. The physiochemical properties of water, such as viscosity, ion product, density, and heat capacity, also change dramatically in the supercritical region with only a small change in the temperature or pressure, resulting in a substantial increase in the rates of chemical reactions. For example, Fig. 1 shows, how density, dielectric constant, and ionic product of water vary with temperature at 24 MPa. From the above figure, it is interesting to see that the dielectric behavior of 200
C water is similar to that of ambient methanol, for 300
C water it is similar to that of ambient acetone, for 370
C water it is similar to that of methylene chloride, and for 500
C water it is similar to that of ambient hexane. In addition to the unusual dielectric behavior, transport properties of water are significantly different from the ambient water as shown in Table 1.

waste streams and has received enormous interest during the last decade. Supercritical water oxidation (SWO) can be seen as a further development of the well-established wet air oxidation (WAO) process, running at temperatures up to 320
C and pressures up to 20 MPa. Treatment times in the WAO process are normally as high as several hours, and complete destruction of the organic material is seldom achieved, so a further waste treatment is necessary. In SWO, treatment times are in the range of seconds to a few minutes and more than 99% destruction is achieved in most cases. Incineration, WAO, and SWO are further compared in Table 2. In a typical SWO operation, organic waste and high-pressure oxygen (or air, or H2O2) are fed to the reactor. The reaction products are then cooled and depressurized to collect the benign products (Fig. 2). Owing to the use of excess oxygen, various elemental species from waste convert into their highest oxidation form. For example, carbon converts to CO2, hydrogen converts to H2O, nitrogen converts to HNO3, sulfur converts to H2SO4, etc. Early reactors were simple tube reactors as shown in Fig. 3. The above tube reactor is also designed to have intermittent ports for the entry of oxidant (e.g., oxygen) and=or quench water. This reactor design is cheaper than the other designs mainly owing to the ease of fabrication and installation. However, if the corrosive species are present then severe wall corrosion can occur. Hence, tube reactors are recommended for use in oxidation of waste that does not contain heteroatomic molecules (e.g., chlorine, sulfur, nitrogen, phosphorous). A typical flow diagram for SWO reactor is shown in Fig. 3.

WASTE OXIDATION Problems of SWO Owing to the lack of mass transfer limitations and high thermal energy, oxidation in supercritical water is very fast. Usually, in less than 1 min of reaction time, complete oxidation of organics to CO2 and mineral acids is achieved. Supercritical water has been successfully used to completely oxidize chemicals including polychlorinated biphenyls, organic solvents, and other industrial wastes. In fact, it is emerging a promising alternative to the incineration of aqueous organic Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120007978 Copyright # 2006 by Taylor & Francis. All rights reserved.

Supercritical water oxidation has been successfully tested for the oxidation of a variety of waste including radioactive waste, rocket propellants, chemical warfare agents, pulp and paper waste sludge, polymer plant waste, organic waste, and municipal waste sludge. Oxidants used in the process include pure oxygen, compressed air, hydrogen peroxide, permanganates, and other industrial oxidants. The SWO process has had some success in 2927

2928

Supercritical Water Oxidation

1400

-10

Ionic product

-12 -14

Density

-16 -18

800

-20 -22

Dielectric constant

200

-24

Hexane

400

CH2Cl2

Acetone

600

Ionic product

1000

MeoH

Dielectric constant x 10 Density [kg m-3]

1200

-26

Fig. 1 Physical properties of water vs. temperature at 24 MPa. Dielectric constants of typical organic solvents at room temperature are also indicated. (From Ref.[2].)

-28 -30

0 0

100

200

300

400

500

600

Temperature [°C]

military applications and some commercial installations. The lack of widespread industrial adaptation can be attributed to three major factors:[1,2]

The corrosion of the heat exchanger is sometime more than that of the reactor, owing to intermediate temperatures in the heat exchanger. To avoid acid formation, a base can be added with the feed stream. The acids and base combine to yield salts, which then precipitate out causing the plugging problem. Supercritical water oxidation poses a unique corrosion problem that has not been experienced in other chemical processes. Development of new materials is needed to address the problem. The material needs to withstand a chemically harsh environment along with the high temperature and pressure conditions. Even if some current materials are available for

1. Severe corrosion problems due to the formation of acids when elements such as S, P, Cl are present in the waste streams. 2. Serious plugging of the reactor, valves, and pipes caused by the precipitation of salts. Most waste streams contain salts that are not soluble in supercritical water. 3. Difficulty of scale-up and cost evaluation due to the lack of data on industrial scale SWO plants.

Table 1 Comparison of ambient and supercritical waters Ambient water

Supercritical water

Dielectric constant

78

30% by volume). Continuous oriented fibers are required for high performance composites. In both cases, performance depends strongly on the structure and performance of the component parts. Table 3 summarizes the key properties of selected major types of thermosetting resins.[1,23–25] Their crosslinking reactions can occur by radical chain addition

Thermosets–Materials, Processes, and Waste Minimization

3033

Table 1 Typical molding processes in thermoset composites Molding process

Description

Prepreg molding

Prepreg molding is in many respects the next step up from hand lay up. With prepreg molding the resin content of the finished component can be accurately controlled, which cannot always be said for hand lay up. Also, woven or unidirectional fiber reinforcements are used, rather than chopped strand mat. The reason is that they be aligned in the required orientation.

Compression molding

Compression molding is the most common method of forming thermosetting materials. It is simply the squeezing of a thermosetting preform into a desired shape by application of heat and pressure to the material in a mold at a desired duration time. Molding preform mixed with fillers and chopped fibers to strengthen the finished product, is put directly into the open mold cavity. The mold is then closed, pressing down on the preform and causing it to flow throughout the mold. It is while the heated mold is closed that the thermosetting material undergoes a chemical change which permanently hardens it into the shape of the mold.

Transfer molding

Transfer molding is most generally used for thermosets. This method is like compression molding in that the preform is cured into an infusible state in a mold under heat and pressure. It differs from compression molding in that the thermosetting preform is heated to a point of plasticity before it reaches the mold and is forced into a closed mold by means of a hydraulically operated plunger. The molten molding material in transfer molding flows around these metal parts without causing them to shift position. Transfer molding was developed to facilitate the molding of intricate products with small deep holes or numerous metal inserts.

Vacuum-assisted molding

In conventional SMC processes, charge loading and press closure speeds are selected with the primary aim of forcing out entrapped air. With vacuum assistance, the charge is selected to rapidly cover the mold surface and provide faster closure speeds without entrapping air. The increase in mold surface coverage by the charge has several additional benefits such as elimination of wave patterns and flow lines, better localized strength of components due to better retention of fiber orientation.

Injection molding

Injection molding is a major process in making thermoplastic materials. With proper modifications, the injection process is sometimes used for thermosetting plastics. In injection molding, material is put into a hopper which feeds into a heating chamber. A plunger pushes the material through this long heating chamber, where the material is softened to a fluid state. At the end of this chamber there is a nozzle which abuts firmly against an opening into a cool, closed mold. The fluid material is forced at high pressure through this nozzle into the cold mold. As soon as the material cools (in thermoplastics) or heated (in thermosets) to a solid state, the mold opens and the finished part is ejected from the press.

Resin transfer molding

In the resin transfer molding (RTM) process, a low viscosity resin is transferred into a closed mold containing all the appropriate reinforcements and inserts as a preform. The air is normally evacuated from the mold, allowing the use of low resin injection pressures and epoxy molds. Manhole covers, compressor casings, car doors, and propeller blades have all been manufactured by RTM.

Reaction injection molding

Reaction injection molding (RIM) is a relatively new processing technique that has rapidly taken its place alongside more traditional methods. Unlike liquid casting, the two liquid components, polyols and isocyanates, are mixed in a chamber at relatively low temperatures (75
–140
F) before being injected into a closed mold. An exothermic reaction occurs, and consequently RIM requires far less energy usage than any other injection molding system. The three major types of polyurethane RIM systems are rigid structural foam, low-modulus elastomers and high-modulus elastomers. Reinforced RIM (R-RIM) consists of the addition of such materials as chopped or milled glass fiber to the polyurethane to enhance stiffness and to increase modulus, thus expanding the range of applications. (Continued)

T

3034

Thermosets: Materials, Processes, and Waste Minimization

Table 1 Typical molding processes in thermoset composites (Continued) Molding process

Description

Pultrusion

In the pultrusion process, dry reinforcements are impregnated with a specially prepared low viscosity liquid resin system and drawn through a die heated to about 120–150
C where curing occurs. The solid laminate, which has assumed the shape of the die, is withdrawn by a series of haul off grippers, and is cut to length or coiled. Pultrusion is unique among the processes under consideration in that it is capable of producing complex components on a continuous basis. The process can basically produce any shape that can be extruded. It is also not allied to any one industry and applications range from civil engineering to electrical. These factors combine to give pultrusion one of the highest predicted growth rates of all composite processes.

Filament winding

Filament winding is a process of high speed precise laying down of resin impregnated continuous fibers on to a mandrel. The mandrel can be any shape. Pressure vessels, pipes, and drive shafts have all been manufactured using filament winding. Multiaxis winding machines can also be used. The process is usually computer controlled and the reinforcement can be oriented to match the design loads. The fibers may be impregnated with resin before winding (wet winding), pre-impregnated (dry winding), or post impregnated. Wet winding has the advantages of using the lowest cost materials with long storage life and low viscosity. The prepreg systems produce parts with more consistent resin content and can often be wound faster.

Fig. 1 Schematic diagram of the various typical composite manufacturing processes.

Thermosets: Materials, Processes, and Waste Minimization

3035

T

Fig. 2 Design methodology. (From Refs.[4,5].)

(unsaturated polyester resins, vinyl ester resins, benzocyclobutene, etc.), ionic ring opening (epoxies), condensation (urethanes, phenolics, polyimides, such as bismaleimides), and cyclotrimerization (cyanates, isocyanurates, etc.). Table 4 compares the mechanical properties of various organic fibers, graphite/carbon fibers, ceramic fibers, and glass fibers, and lists commercially important applications. One major application of organic fibers, such as SpectraÕ and aramid fibers, is in ballistic armor.[26,27] In contrast, graphite, ceramic fibers, and glass fibers are primarily used in structural applications. One of the reasons that organic fibers find little

structure applications is because of their poor compressive strength. (see Table 4 in Ref.[28].)

CHARACTERIZATION The final morphology of a network system depends greatly on the flow properties and phase separation of the reactive components as seen in Fig. 5.[16–18,29] Naturally, design of appropriate processes requires the ability to directly investigate the progress of reaction. This includes monitoring the chemistry of cure with spectroscopic techniques such as FTIR, NMR,

3036

Thermosets–Materials, Processes, and Waste Minimization

Fig. 3 Interrelationship of reinforcement-preform/prepreg-matrix. (From Refs.[4,29].)

UV-Vis, thermal analysis such as DMA, TMA, or rheological measurements. Fig. 5 illustrates the application of rheological techniques to monitor cure. Fig. 5A shows the relation of viscosity versus extent of reaction for a thermoset system, whereas Fig. 5B gives the relation of temperature versus extent of reaction in a liquid crystal thermoset. As an illustration, the effect of curing temperature on the structure of a cured liquid crystalline thermoset is given in Table 5.[30] Gillham[31] has postulated that the curing of a thermoset resin can be expressed in terms of a time–temperature transformation diagram (TTT), in which the entire curing cycle is represented by the resin rheology as a function of time and isothermal temperature. Similar to Fig. 5, the TTT diagram is divided into

four sections: liquid, gelled rubber, ungelled glass, and gelled glass. This concept has found application to describe the structure–property relationship of thermoset epoxy resins using torsional braid analysis.

RECENT ADVANCES With recent advances in the mixing of different types of thermosets and=or thermoplastics, the technology of IPN has provided very unique properties and performance.[32–34] IPN is a method to produce very special network polymer blends. Two materials which do not react with each other are blended and subsequently cured in place. The blended network polymers

Thermosets–Materials, Processes, and Waste Minimization

3037

Table 2 A combinatory approach to establish the matrix-fiber-process relationship

Combination route

Process

Further processing to final part

(Intermediate) products

Potential process to make helmets (given in the example)

(A1 þ D1)

Melt compounding Melt impregnation Powder impregnation

Chopped fiberreinforced granule UD prepreg, tape UD fiber tow

Injection, extrusion Molding Tape laying and winding Filament winding

Yes

(A2 þ E2)

Commingling Interweaving, braiding

Mixed yarn Textile fabric

Filament winding, pultrusion Hot pressing, pultrusion

Yes

(A3 þ C3)

Chopped fiber

No

Yes

—

Reinforced reaction Injection molding, RRIM

(B1 þ E1)

Film stacking

Glass mat-reinforced sheet GMT, SMC, TMC

Thermoforming Hot pressing

Yes

(A2 þ D2)

Powder or slurry impregnation

Impregnated fleece

Thermoforming

Yes

Impregnated textile fabric

Hot pressing

Hybrid textile fabric

Thermoforming

(B2 þ D2) (B2 þ E2)

Tufting, needle-punching

(B3 þ C3)

Structural RIM

—

No

Molding of self-reinforced polymers

—

No (unsolved process for molecular composites)

(C3 þ D3)

Yes

Hot pressing Yes ??

UD, unidirectional; GMT, glass mat thermoplastic; TMC, thick molding compound; SMC, sheet molding compound. (From Refs.[4,29].)

share the same region of space (macroscopic volume) of the sample. Thus one has an intimate mixture of two normally immiscible systems. Consequently, the IPN exhibits a combination of properties from the two networks. This approach can generate unique advantages such as crack resistance,[34] improved modulus, strength, and hydrolytic resistance.[35] One can even generate organic/inorganic hybrid IPNs to impart specific properties such as permeability reduction.[36] Advances in liquid crystalline thermosets in recent years,[30,37–40] have resulted in some fascinating thermosetting systems. Control of the liquid crystal type and morphology allows one to generate a wide range of properties. For example, the properties can be tuned from electrical insulators to conductors with addition of conductive fillers, while maintaining desirable high temperature performance.[39] In addition, can provide control of the degree of shrinkage on cure.[40] Another area of current research is development of nanocomposites. Incorporation of nanometer-sized fillers in thermosets allows even broader application of thermoset technology. The processing techniques are quite similar to conventional thermosets, but the

effect of filler can be quite dramatic. For example, incorporation of only 2.5% by weight of nanometersized silicates in a cyanate resin led to a marked improvement in physical and thermal properties while imparting a 30% increase in both the modulus and toughness.[29,41] As discussed earlier, while the scale of the fillers is substantially different, nanocomposite materials concepts and technology are very similar to those of conventional composite materials. This is clearly demonstrated in the case of new thermosets for nonlinear optical (NLO) applications,[37,38,42,43] where nanocomposite of liquid crystalline thermosets, IPNs, and simple filled thermosets are evaluated. Tripathy et al.[42] discussed four different ways to prepare nonlinear optical polymers. (1) The polymer matrix is doped with NLO moieties in a guest=host system; (2) In side-chain polymer systems, NLO polymers with active moieties are covalently bonded as pendant groups; (3) In the main chain polymer, the chromophores are incorporated as parts of the main polymer backbone to enhance the temporal stability of the NLO properties; and (4) Stability of the optical noninearity in sol–gel-based thermosets is related to

T

3038

Thermosets: Materials, Processes, and Waste Minimization

Fig. 4 Classification of composites. (From Ref.[29].)

the increasing crosslinking density. It was found that sol-gel technology can be used to prepare threedimensional network glasses of optically clear and low loss properties at high temperatures. The final structure can be further manipulated by employing IPN technology and=or using the AC electric field technique.[38] It has been reported that the combination of high Tg (glass transition temperature) polymers such as polyimides, high crosslink density, and permanent chain entanglement results in high temporal stability at elevated temperatures.[42,43] By varying the structure of various component groups of the molecule, one can understand how to control the liquid crystallinity and the resultant properties of the materials (see the Fig. 5B).[44] A successful development of a practical device for NLO application will require optimization of properties in all aspects in materials and processes, such as optical loss, optical power handling, processability, and reproducibility.

WASTE MINIMIZATION Finally, waste generation is a critical limitation for any industrial process.[45] Waste generation is unavoidable

so waste minimization becomes a fundamental requirement for economic feasibility. Waste generation inherently decreases the overall value of any material. A clear understanding of the materials and processes is central to waste reduction. For example, development of (1) Laboratory definition of a robust model for cure kinetics lead to; (2) robust on-line cure monitors, which coupled to; (3) robust process controllers for manufacturing can significantly reduce the waste in production of glass filled epoxies. Senge and Carstedt[45] recently offered a view in Fig. 6 of why industry produces waste. They suggested that a synthetic process can emulate nature to reduce the waste using a cyclic industrial system. This is accomplished by focusing on three key aspects of the manufacturing process: (1) resource productivity; (2) cleaning products; and (3) remanufacturing, recycle, and compost. Clearly in Fig. 6, preventing the waste generation from production, use, and disposal in the first place can eliminate the waste. One example is the industrial recycling of nylon 6 carpet patented by Honeywell (formally AlliedSignal), Inc. and DSM.[46,47] This process has addressed and overcome the economic, technical, and logistical barriers to commercialize a closed-loop recycling process to recover caprolactam

Epoxies

Very versatile. Wide range of formulations available for whole range of processes and applications. Excellent balance of mechanical, electrical, thermal, and environmental resistance properties. Poor toughness without impact modification

2. Resoles–Selfcondensation, thermally activated or acid catalyzed

1. Novolacs–require additional formaldehyde for further curing

Low cost, good high temperature resistant, fair mechanical properties

Phenolics

Stepwise reactions with curing agents such as amines, acids or anhydrides, and phenols

Ionic ring-opening chain polymerization catalyzed by tertiary amine (anionic) or BF3 complexes (cationic)

180

180

120

Radical

Similar to unsaturated polyester, but with improved properties and chemical resistance, and have higher cost

Vinyl ester, acrylics

Condensation

120

UUT,
C

Radical

Curing reaction

Low viscosity, easy processing, lowest cost, limited mechanical and thermal properties

Features

Unsaturated polyester

Resin

Table 3 Typical thermosetting resins of commercial applications

130

110

MCUT,
C

170–206

—

100–160

80–140

Tg,
C

0.88

0.35

TC, W/m.k

Thermal properties

1.1–3.5

3–5

CTE, 105 oK

5600– 12,000

E, MPa

69–207

r, MPa

(Continued)

54

KIC J/m2

Mechanical properties

Thermosets–Materials, Processes, and Waste Minimization 3039

T

290–315

Condensation of amine with acid ester to form amide-acid, then imide, followed by crosslinking through norbornene end-group

PMR: Excellent high temperature resistance, more brittle to BMI, uses of toxic component

TC, W/m.k

CTE, 105 oK

3700

E, MPa

r, MPa

200

400

60–210

KIC J/m2

Mechanical properties

Upper use temperature, UUT (
C); maximum continous use temperature, MCUT (
C); thermal conductivity, TC (W/m.k); coefficient of thermal expansion, CTE (105
K); tensile modulus, E (Mpa).

400

240–300

Polyimides

200–230

Epoxy-like processing at higher temperatures. Higher cost, low dielectric, good mechanical properties and toughness Chain polymerization through the double bonds in maleimides. Stepwise additions with, or allyl or propenyl modifiers

Tg,
C

BMI: Convenient processing, but at high temperatures. Very brittle without impact modification. Very good hot-wet performance

MCUT,
C 254–280

UUT,
C 180–220

Curing reaction

Thermal properties

Self-cyclotrimerization catalyzed by transition metal carboxylates and nonylphenol

Features

Cyanates

Resin

Table 3 Typical thermosetting resins of commercial applications (Continued)

3040 Thermosets: Materials, Processes, and Waste Minimization

Thermosets: Materials, Processes, and Waste Minimization

3041

Table 4 Typical properties of reinforcement fibers and their commercial applications Aramid SpectraÕ (PE)a 900

Filament diameter, m

Carbon

LM

HM

HS

HM

1000

1500=1000

1500=1000

1730=5100

1630=3000

38

27

12

12

7

7

7

Density, g=cm3

0.97

0.97

1.44

1.44

1.81

1.81

2.50

2.5–2.6

3.7

Tensile strength, Gpa

2.59

3.00

2.80

2.80

3.10

2.40

4.6

3.5

415

Tensile modulus, GPa

117

172

62

124

228

379

90

1.7

350–380

Tensile elongation,%

3.5

2.7

3.6

2.8

1.20

0.60

5.40

Specific tensile strength, 106 in.

10.7

12.4

7.8

7.8

6.8

5.4

7.4

4.3–4.5

1.8

Specific tensile modulus, 108 in.

486

714

173

346

507

846

140

309–321

380–412

1.1

5.9

6.9

Very significant

Significant

Very significant

Yarn specifications, d=fil

Glass-S

Boron

Al2O3

0.32–0.46

0.48–2.8

??

Significant

Very significant

Very significant

Very significant

??

Significant

??

??

??

Very significant

??

Significant

??

??

Very significant

Very significant

Very significant

Very significant

Compressive strength, GPa Commercial applications Structural Armor Tire=elastomer Others

OK

Very significant

a

SpectraÕ is a registered trademark of Honeywell International. (From Refs.[26,33].)

Fig. 5 Schematic diagrams of effect of extent of reaction on viscosity and morphology in curing of thermosets. (A), Effect of extent of reaction on viscosity. (From Refs.[11–14].) (B) Effect of extent of reaction on morphology. (From Refs.[30,37,38].)

T

3042

Thermosets–Materials, Processes, and Waste Minimization

Table 5 Effect of curing temperature on the structure Liquid crystalline thermoset system: 1. Liquid crystalline thermoset prepolymer: dihydroxy methylstilbene epoxy (Mn, 3600) 2. Curing agent: 4,40 -methylene dianiline Curing temperature,
C

Molecular order

Clearing transition temperature,
C

Liquid crystalline thermoset epoxy prepolymer

Not applicable

Nematic to isotropic

176

Cured isotropic thermoset

190

Thermoset

157

Cured liquid crystalline thermoset

140

Nematic

192

Sample

[30]

(From Ref.

.)

from waste nylon 6 articles. Nylon carpet is a composite structure containing 45% nylon fiber on a calcium carbonatefilled polypropylene backing. The fiber and backing are held together by a cure thermoset styrene

butadiene (SBR) latex adhesive. The waste product derived from the medium-pressure depolymerization of nylon 6 carpet from this process is a composite of 65% calcium carbonate and 18% polypropylene dispersed in

Fig. 6 Waste generation and reduction from current industry. (From Ref.[45].)

Thermosets–Materials, Processes, and Waste Minimization

3043

a styrene butadiene thermoset. This residue material can be used as a filler in road asphalt, asphalt roof membranes, molding compounds, and plastic lumber.[48,49] Thus, we have a real example of the closed loop recycle process as postulated by Senge and Carstedt. Gutowski[50] has examined the product induced material flows through the product manufacturing system, and has suggested several research strategies to reduce material-related environmental loads. Fig. 7A gives a typical schematic of a materials processing flow map. Fig. 7B gives a typical fabrication process of flat laminate sheet found commonly in composite manufacturing.[50] In Fig. 7B, we can see that waste can be generated from the raw materials supply, from each step in the process, and from scrap. Sometimes, the total waste generated from these processes could be as high as 25% based on the raw materials. The net flow balance in Fig. 7A, can be described by Eq. (1): Performance ¼ afðcritical component selection  its wastesÞ þ bfðprocess  its wastesÞ

þ cfðstructure  its wastesÞ

þ dfðproducts  its wastesÞ  performance wastes

ð1Þ

Performance ¼ afðcritical component selectionÞ þ bfðprocessÞ þ cfðstructureÞ

þ dfðproductsÞ  Swastes all sources

ð2Þ

where a, b, c, d are the constants. In term of a continuous flow process, we can rewrite Eqs. (1) and (2) into Eq. (3) Z Z PðxÞ ¼ lj V ðxj Þdxj  oj Wðxj Þdxj ð3Þ where, P(x) is a value performance function, V(xj) is the value generating function at component xj stage or phase j, and W(xj) is the waste generating function at component xj, and lj, and oj are constants. The variation of V(xj), W(xj), lj, and oj greatly affects the value of P(x). From Eq. (3), we can see that wastes generated from the materials flow negatively to impact a wide range of factors. These range from the selection of materials and from innovation design to environmental concerns as seen in Figs. 6 and 7. V(xj) is not restricted with any limitations in Eq. (3) and it is valid to include the feedback loops in Fig. 6.

Thus, both processes in Figs. 6 and 7 can be described by Eq. (3) with different forms of V(xj) and W(xj), and different values of lj and oj. Furthermore, this form of expression is not new and it has found uses in many applications in mathematics and science.[51–62] The definition of wastes can be viewed as the gap(s) between the full (theoretical) potential of the resources supplied and the actual delivered. As can be seen in the examples in Table 6, the defects found in a fiber structure impair our ability to obtain much less than half of its theoretical tensile modulus.[28] We need to point out that the concepts in Figs. 6 and 7, and Eqs. (1)–(3) have been the background of the so called ‘‘Six Sigma’’ methodology used to minimize the wastes in manufacturing industries over the last decade.[6–10] Six Sigma is a strategy that was developed to accelerate improvements in processes, products, and services, radically reduce manufacturing and=or administrative costs, and improve quality by relentlessly focusing on eliminating waste and reducing defects and variation. Since the material processing is a very broad field, we have limited our focus in the discussion to thermosets and their fiber-reinforced composites. However, the same principle will apply for any materials development effort.

PROPERTY OPTIMIZATION CASE STUDY We discussed ‘‘in silico’’ experimentation above. In particularly demanding applications, it is far more efficient to design the composite for the application before spending excessive time on trial and error. We will demonstrate this with a case study. On the basis of the design methodology illustrated in Fig. 2 and the processes given in Table 2, we will discuss the design of a helmet for ballistic protection. Figs. 8–10 show the design and simulated deformation of a helmet under ballistic impact. Honeywell International Inc.’s Spectra ShieldÕ composite has been used as the ballistic material for a lightweight helmet system since the mid-1990’s. Spectra Shield composite is made from Honeywell’s SpectraÕ fiber, one of the world’s lightest and strongest fibers (see Table 4), which is, pound-for-pound 10 times stronger than steel. The use of this material in helmet construction results in substantially increased ballistic protection, while reducing weight. Before field use of Spectra ShieldÕ material for this application, modeling of the composite helmet under ballistic impact was performed. The modeling was carried out to highlight the increased protection under the ballistic impact of Spectra fiber compared to metal. Finite element analysis (FEA) of the helmet configuration impacted by a standard projectile round to the front

T

3044

Thermosets: Materials, Processes, and Waste Minimization

Fig. 7 Materials flow relationship. (A) Relationship of ‘‘component section–process–structure–product–performance.’’ (B) schematic of a flat laminate sheet production process. (From Ref.[50].)

Table 6 Tensile modulus of several ordered polymers

No

Molecular structure

Materials 0

Theoretical (Gpa)

Gap Actual (Gpa)

GPa

%

1

Poly(p-phenylene-2,6-benzo[1,2-d:45-d ] bisoxazole (PBO)

Cis Trans

730–670 707–620

360

370–310 347–260

55–43 60–37

2

Poly(p-phenylene-2,6-benzo[1,2-d:45-d0 ] bisthiazole (PBZT)

Cis Trans

610–600 605–525

325

285–275 280–200

48–45 53–33

3

Polyethylene

360–320

172–117

243–148

76–41

4

Graphite

1500

600–70

1430–900

95–60

(From Ref.[28].)

Thermosets–Materials, Processes, and Waste Minimization

3045

T

Fig. 9 Helmet design for finite element analysis.

Fig. 8 Typical layout of a helmet.

of the helmet was performed using the Abaqus FEA program (Abaqus Inc. Pawtucket, Rhode Island, U.S.A.). A FEA analysis can be linear or nonlinear, but for our application nonlinear effects are significant and have to be taken into account. In a finite element analysis there are at least four primary causes of nonlinearity. 1. Material nonlinearity. This nonlinearity is present because of the fiber orientation in the composite. The effect of the Spectra fiber reinforcement on the composite structure was taken into account by a user subroutine which added the oriented fibers effect to the base material of the composite. 2. Geometric nonlinearity. This nonlinearity is applicable due to the relatively large deformations anticipated. As the helmet is a curved

structure, the effect of this curvature on the stress distribution has to be taken into account. 3. Force application nonlinearity. This nonlinearity is applicable due to the expected nonlinear displacement loading as a function of time. A FEA contact analysis was required because we idealized the system as a nondeformable projectile impacting a deformable helmet. 4. Boundary condition nonlinearity. This nonlinearity is also applicable in FEA contact problems. This nonlinearity is expected during the projectile impact on the helmet due to the boundary conditions changing as a function of time and loading. The model used consisted of a half helmet with three element layers of composite material through the thickness. A half model was used to take advantage of the symmetry present in the problem. The elements used were eight-noded brick elements because these elements can represent the deformation sufficiently. Anisotropic material properties were defined for each element using a user subroutine and impact was designated at the front of the helmet. After the solution was obtained the results were analyzed. Each of the three principal stresses (s1, s2,

Fig. 10 Finite element analysis results for stress during ballistic impact on the helmet.

3046

s3) are available in most FEA software packages and stresses are usually averaged by the FEA software packages to provide more accurate stress values when mapped (contoured) on to the mesh. A good first cut to the understanding of analysis results is the use of the von Mises stress (effective or equivalent stress). Fig. 10 shows the von Mises stress contour mapped to the FEA mesh in pounds=square inch (PSI). The simulation shows that the helmet would deform at the location of impact, but the stresses would be distributed out and around the impact location thereby reducing penetration probability as compared with a metal helmet. Thus, we have used a combination of an understanding of material properties with computer simulation to deliver the required performance.

Thermosets–Materials, Processes, and Waste Minimization

capable measurement systems must be extended to the whole field of thermoset manufacturing so that the design engineer can reliably predict the effect of waste. If we want to minimize the waste generation function, and maximize the value generating function in materials processing, we need to develop a clear, consistent way to predict and minimize the waste generation function.

ACKNOWLEDGMENTS The authors thank Tony Signorelli, Alan Levy, Richard Wilson, Janice Sund, Virginia Szigeti, Lori Wagner, Tom Izod, and Sunil Kasavan for their helpful comments and suggestions.

CONCLUSIONS REFERENCES Environmental and economic drivers will accelerate the development of products generating minimal wastes. As Senge and Carstedt have pointed out, an understanding of the utilization of resource productivity is essential to the development of an industry that can reduce waste. An understanding of thermoset processing is of vital importance in many industries such as integrated chip manufacturers, aerospace technologies, automotive manufacturers, etc. Traditionally, design and control of these processes has relied on trial and error methods due to the complexity of the reacting systems. With the recent advances of modeling, kinetic tests, and chemorheological measurement, characterization of these complex reacting systems have found a foundation in the prediction of the processing-structure and performance of thermosetting systems. However, there is still a long way to go. Most of the test methods described in the paper by Halley and Mackay[15] have one thing in common: there are very few data to determine the waste generation function in material control. Some of the key drivers for waste are well known. In thermosets for example, compounded resin consistency is a key input variable for waste minimization. Since the final structure of the thermoset controls the performance, there is a need to have a measurement system to evaluate resin reactivity reliably. One such measurement system has been reported by Kranbuehl et al.[63] They developed an in-line frequency-dependent electromagnetic sensor to monitor cure in polyimide and epoxy thermosets. The sensor was then incorporated into a closed loop controller to monitor and control cure during manufacturing. Similarly, we have shown that phenolic resin composition directly affects carbon composite performance.[64] Control of the base resin system allowed production of carbon-carbon composites with minimal failures due to cracking. These types of

1. Brydson, J.A. Plastics Materials, 6th Ed.; Butterworth—Heinemann: New York, 1999. 2. Newman, S. Introduction to composite materials technology: mass production techniques. In Composite Materials Technology–Processes and Properties; Mallick, P.K., Newman, S., Eds.; Hanser Publishers: New York, 1990; 10–24. 3. Crawford, R.J. Plastic Engineering, 3rd Ed.; Butterworth-Heinemann: New York, 1998. 4. Karger-Kocsis, J. Composites (structure, properties, and manufacturing). In Polymeric Materials Encyclopedia; Salamone, J.C., Ed.; CRC Press: New York, 1996; Vol. 2, 1378–1383. 5. Ashby, M.F.; Jones, D.R.H. Engineering Materials 2–An Introduction to Microstructures, Processing, and Design; Pergamon Press: New York, 1988. 6. Harry, M.J.; Lawson, J.R. Six Sigma Producibility Analysis and Process Characterization; Addison-Wesley: New York, 1992. 7. Curran, S.; Lem, K.W.; Sund, S.; Gabriel, M. Six sigma design–an overview of design for six sigma. In Encyclopedia of Chemical Processing; Lee, S., Ed.; Marcel Dekker, Inc.: New York, 2006 (in press) 8. Berryman, M.L. Transform your organization into one that’s world class. Six Sigma Forum Magazine 2002, 2 (1). 9. Creveling, C.M.; Slutsky, J.L.; Antis, D., Jr. Design for Six Sigma – In Technology and Product Development; Pearson Education, Inc.: New Jersey, 2003. 10. Bicerano, J. Prediction of Polymer Properties; Marcel Dekker, Inc.: New York, 1993. 11. Mark, J.E. Physical Properties of Polymer Handbook; American Institute of Physics: New York, 1996.

Thermosets–Materials, Processes, and Waste Minimization

12. Rajan, K. An informatics infrastructure for in silico materials science. NIST Combinatorial Methods Center NCMC3, May 23, 2003. 13. Iden, R.; Schrof, W.; Hadeler, J.; Lehmann, S. Combinatorial materials research in the polymer industry: speed versus flexibility. Macromolec. Rapid Commun. 2003, 24 (1), 63–72. 14. Yarovsky, I.; Evans, E. Computer simulation of structure and properties of crosslinked polymers: application to epoxy resins Polymer 2001, 43 (3), 963–969. 15. Halley, P.J.; Mackay, M.E. Chemorheology of thermosets–an overview. Polym. Eng. Sci. 1996, 36 (5), 593–609. 16. Lem, K.W. Rheology and Curing Behavior of Thermosetting Polyester Resins. Ph.D. Dissertation, Polytech. Inst. New York, Brooklyn: New York, U.S.A., 1983. 17. Lem, K.W.; Han, C.D. Thermokinetics of unsaturated polyester and vinyl ester resins. Polym. Eng. Sci. 1984, 24 (3), 175–184. 18. Mallick, P.K. Compression molding. In Composite Materials Technology–Processes and Properties; Mallick, P.K., Newman, S., Eds.; Hanser Publishers: New York, 1990; 67–102. 19. Li, M.; Tucker, C.L. Optimal curing of thermoset matrix composites: thermochemical and consolidation consideration. Polym. Compos. 2002, 23, 739–757. 20. Racich, J.L.; Koutsky, J.A. Boundary layers in thermosets. In Chem. Prop. Crosslinked Polym, Proc. ACS Symp; Labana, S.S., Ed.; 1977; 303– 323. 21. Mijovic, J.; Tsay, L.L. Correlations between dynamic mechanical properties and nodular morphology of cured epoxy resins. Polymer 1981, 22 (7), 902–906. 22. Dusek, K. Are cured thermoset resins inhomogeneous? Angew. Makromol. Chemie 1996, 240, 1–15. 23. Feldman, D. Composites–thermosetting polymers. In Polymeric Materials Encyclopedia; Salamone, J.C., Ed.; CRC Press: New York, 1996; 2, 1383–1389. 24. Bigg, D.M. Thermoplastic matrix composites. In International Encyclopedia of Composites; Lee, S.M., Ed.; VCH Publishers: New York, 1991; Vol. 6, 30–33. 25. Scola, D.A. Thermoset composites, advanced. In International Encyclopedia of Composites; Lee, S.M., Ed.; VCH Publishers: New York, 1991; Vol. 6, 34–48. 26. Prevorsek, D.C. SpectraÕ: The latest entry in the field of high-performance fibers. In Handbook of Fiber Science and Technology; Lewin, M., Ed.; Marcel Dekker, Inc.: New York, 1996; Vol. 3, 1–170.

3047

27. Gray, W.; Zabel, P.H. Armor, composite. In International Encyclopedia of Composites; Lee, S.M., Ed.; VCH Publishers: New York, 1991; Vol. 6, 125–140. 28. Kumar, S. Ordered polymer fibers. In International Encyclopedia of Composites; Lee, S.M., Ed.; VCH Publishers: New York, 1991; Vol. 4, 51–74. 29. Ganguli, S.; Dean, D.; Derrick, J.; Kelvin, P.; Price, G.; Vaia, R. Mechanical properties of intercalated cyanate ester-layered silicate nanocomposites. Polymer 2003, 44 (4), 1315–1319. 30. Su, W.-F.A. Thermosets (main chain liquid crystalline polymers). In Polymeric Materials Encyclopedia; Salamone, J.C., Ed.; CRC Press: New York, 1996; Vol. 11, 8375–8380. 31. Gillham, J.K. The Role of Polymer Matrix in Processing and Structural Properties of Composites; Seferis, J. C., Nicolais, L., Eds.; Plenum Press: New York, 1983; 127–145. 32. Sperling, L.H.; Mishra, V. Interpenetrating polymer networks (overview). In Polymeric Materials Encyclopedia; Salamone, J. C., Ed.; CRC Press: New York, 1996; Vol. 5, 3292–3301. 33. Frisch, H.L.; Xue, Y.P. Interpenetrating polymer networks (rubber-based). In Polymeric Materials Encyclopedia; Salamone, J. C., Ed.; CRC Press: New York, 1996; Vol. 5, 3302–3308. 34. Bascom, W.; Gweon, S.Y.; Grande, G. Multiphase matrix for carbon fiber composites. Adv. Chem. Ser. 1993, 233, 519–537. 35. Cook, W.D.; Dean, K.; Forsythe, J. Cure, rheology and properties of IPN thermosets for composite applications. Materials Forum 2001, 25, 30–59. 36. Yoshihara, T. Substrates Having Gas-barrier Organic-Inorganic Hybrid Polymer Layers and their Manufacture. Patent JP 2003094572, 2003. 37. Qin, H.; Mather, P.T. Optical rheology of new liquid crystalline thermosets (lcts): influence of shear on disclination texture. Mat. Res. Soc. Symp. Proc. 2002, 709, CC8.4.1–7. 38. Korner, H.; Shiota, A.; Ober, C.K. The processing of LC thermosets in orienting external fields. Mat. Res. Soc. Symp. Proc. 1996, 425, 149–160. 39. Wadahara, E.; Ishibashi, S.; Nagashima, Y. Polymer Conductor Composition and Molded Materials thereof. Patent JP 2001067933, 2001. 40. Dershem, S.; Yang, K. Low Shrinkage Thermosetting Resin Compositions and use in Low Shrinkage Die Attach Pastes. Patent WO 2002028813, 2002. 41. Ritzenthaler, S.; Court, F.; David, L.; GirardReydet, E.; Leibler, L.; Pascault, J.P. ABC Triblock copolymers/epoxy-diamine blends. 1. Keys to achieve nanostructured thermosets. Macromolecules 2002, 35 (16), 6245–6254.

T

3048

42. Tripathy, S.; Chen, J-I.; Marturunkakul, S.; Kumar, J. Nonlinear optical materials. In Polymeric Materials Encyclopedia; Salamone, J.C., Ed.; CRC Press: New York, 1996; 6, 4587–4596. 43. Shannon, P.J.; Gibbons, W.M.; Sun, S.T. Nonlinear optical polymers, thermosets. In Polymeric Materials Encyclopedia; Salamone, J.C., Ed.; CRC Press: New York, 1996; 6, 4605–4611. 44. Sek, D. Structural variations of liquid crystalline polymer macromolecules: review. Acta Polymerica 1988, 39 (11), 599–607. 45. Senge, P.M.; Carstedt, G. Innovating our way to the next industrial revolution. MIT Sloan Management Rev. 2001(Winter), 24–38. 46. AlliedSignal, Inc.; DSM. Innovative Green Chemistry for Sustainable Manufacture of Caprolactam. Program Proposal to U.S. Environmental Protection Agency0 s Presidential Green Chemistry Challenge Awards Program. 1999 (December). 47. Sifniades, S; Levy, A.B.; Hendrix, J.A.J. Processes for Depolymerization Nylon-Containing Whole Carpet to Form Caprolactam. US Patent 5,929,234, July 27, 1999, US Patent 5,932,724, August 3, 1999. 48. Pollution Prevention Assistance Division, Dept. of Natural Resources; AlliedSignal, Inc.; Georgia Department of Transportation; Georgia Environmental Facilities Authority. Demonstration of Polymer By-Product Utilization as an Asphalt Modifier. A proposal Submitted to U.S. Department of Energy for National Industrial Competitiveness Through Energy, Environment, and Economics. Solicitation No. DE-PS3698GO10294. October 20, 1998. 49. Lem, K.W.; Letton, A.; Izod, T.P.J.; Lupton, F.S.; Bedwell, W.B. Composition Containing Caprolactam-Free Residue from Depolymerization of Nylon 6 Carpet and Use Thereof in Paving Asphalt, Plastic Lumber and Crack Sealants. US Patent 6,214,908, April, 10, 2001, USP 6,414,066, July 2, 2002. 50. Gutowski, T.G. Environmentally benign manufacturing and ecomaterials; product induced material flows. Mater Trans. 2002, 43 (3), 359–363. 51. Vaklieva-Bancheva, N.G.; Shopova, E.G.; Ivanov, B.B. Application of Fourier transformation for waste minimization in batch plants. 1. Analysis of production recipes. Hung J. Ind. Chem. 2002, 30 (3), 199–206.

Thermosets–Materials, Processes, and Waste Minimization

52. O’Reilly, A.J. Batch reactor optimization, profitability vs. waste minimization. Chem. Eng. Res. Des. 2002, 80 (A6), 587–596. 53. Zhang, Q.-Y. Multiple objectives application approach to waste minimization. J. Zhejiang Univ. Sci. 2002, 3 (4), 405–411. 54. Ciantar, C.; Hadfield, M.; Howarth, G. Case studies to assist integrating waste prevention in product design. MechE Conference Transactions; Engineering for Profit from Waste 2001, 9, 201–210. 55. Cochrane, T.; Smith, J.A. Designing processes and products to minimize wastes produced. MechE Conference Transactions; Engineering for Profit from Waste 2001, 9, 137–148. 56. Henningsson, S.; Smith, A.; Hyde, K. Minimizing material flows and utility use to increase profitability in the food and drink industry. Trends Food Sci. Technol. 2001, 12 (2), 75–82. 57. Page, P.G. Efficient cost management through chemical conservation and waste minimization for the electroplating industry. Proc. AESF Ann. Tech. Conf. 1997, 84, 321–327. 58. Han, C.; Stephanopoulos, G.; Liu, Y.A. Knowledge-based approach in process synthesis. Recents Progres en Genie des Procedes-Simulation, Optimisation et Commande, SIMO’96. 1996, 10 (49), 1–13. 59. Basta, N. Design hazards out with process simulation. Environ. Eng. World 1996, 2 (3), 28–29. 60. Hilaly, A.K.; Sikdar, S.K. Process simulation tools for pollution prevention. Chem. Eng. 1996, 103 (2), 98–105. 61. Edgar, T.F.; Huang, Y.L. Artificial intelligence approach to synthesis of a process for waste minimization. In Emerging Technologies in Hazardous Waste Management IV; ACS Symp Ser, 1994; Vol. 554, 96–113. 62. Berglund, R.L.; Snyder, G.E. Minimize waste during design. Hydrocarb. Proc., Int. Ed. 1990, 69 (4), 39–42. 63. Kranbuehl, D.E.; Hood, D.K.; Rogozinski, J.; Barksdale, R.; Loos, A.C.; McRae, D. FDEMS sensing for automated intelligent processing of polyimides. Proc. ASME Mat. Div. 1995, 2, 1017–1046. 64. Curran, S.; Walker, T.B.; Brambilla, R. Characterization of phenolic resins for carbon composites. In POLY-471, 213th ACS National Meeting, San Francisco, April 13–17, 1997.

Thin Film Processes in MEMS and NEMS Technologies T W. R. Ashurst C. Carraro R. Maboudian University of California–Berkeley, Berkeley, California, U.S.A.

INTRODUCTION The integration of miniaturized mechanical components with microelectronic components has resulted in a new technology, known as microelectromechanical systems (MEMS), which extends the benefits of microelectronic fabrication to sensing and actuating functions. Several MEMS devices are in commercial use, such as accelerometers, pressure sensors, digital mirror displays, and gyroscopes. The first fully integrated single-chip MEMS accelerometer was manufactured in 1991 by Analog Devices. By September 2002, Analog Devices had sold more than 100 million MEMS accelerometers and gyroscopes. With size shrinking below the micrometer scale, nanoelectromechanical systems (NEMS) are also being realized. Generally, MEMS and NEMS technologies are able to exploit properties that scale favorably with decreasing size. While this ability provides a unique capability, it also poses challenges to the fabrication and reliability of these devices. Conventional machining and assembly techniques cannot be easily applied to microsystems. Moreover, because of the dominance of surface effects on the micro- and nanoscale, strong adhesion, friction, electrostatic charging, and wear have been shown to be issues crucial to this technology. In this article, we present an overview of MEMS and NEMS fabrication processes, reliability issues, and ways to improve reliability with special emphasis on the chemical processes involved.

FABRICATION PROCESSES FOR MEMS AND NEMS Historically, silicon and silicon based materials have formed the basis for the well-established integrated circuit (IC) technology. As a consequence, polycrystalline silicon (polysilicon) is one of the most commonly used structural materials for MEMS. Comparatively, silicon is a good structural material for the microscale.[1] However, the inherent mechanical nature of MEMS devices brings about a new level of complexity to their production and reliability compared to standard integrated circuits. Micromechanical structures can be Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120037168 Copyright # 2006 by Taylor & Francis. All rights reserved.

fabricated from a variety of methods, including bulk micromachining, LIGA (a German acronym for lithography, electroplating, and molding), and surface micromachining.[2] Surface micromachining is one of the most common methods for MEMS fabrication, and it involves the deposition, patterning, and etching of thin films–processes that are commonplace in the IC industry.[1,3] As such, surface micromachining leverages heavily from the technology of the IC industry, and allows for the production of sophisticated microstructures with parallel fabrication and high fabrication yield. It should be noted that silicon is not always chosen as the structural material. Some commercialized MEMS products use aluminum as the structural material, e.g., the Texas Instruments DMD.2 A simplified surface micromachining process diagram is illustrated in Fig. 1. The process begins with a silicon wafer, which has an electrical isolation layer (typically silicon nitride) deposited on the surface. Next, the first polysilicon (poly) layer is deposited. This will form the ground plane and actuation pad for the device in this illustration, a cantilever beam. This poly layer is then lithographically patterned and anisotropically etched (using, e.g., plasma etching). The sacrificial (spacer) layer, typically silicon oxide, is then deposited over the patterned poly layer. The sacrificial layer is also lithographically patterned, and anisotropically etched to reveal access windows to the underlying poly layer. The structural poly layer is deposited, lithographically patterned, and anisotropically etched to complete the structure. Now, although the fabrication of the device is complete, it is not usable until the sacrificial layer is removed. This is accomplished by using an isotropic etch, to dissolve the sacrificial layer and leave a ‘‘released’’ device. The general process for MEMS fabrication can also be applied to NEMS. Owing to the dimensions involved in NEMS, the patterning processes used differ from that of MEMS. For NEMS patterning, e-beam lithography or atomic force microscopy (AFM) writing can be used. E-beam lithography uses a highly focused electron beam to ‘‘write’’ on a thin film. The action of the electron beam alters the film material to the extent that it can be removed while leaving unaltered material intact. In this respect e-beam lithography is 3049

3050

Thin Film Processes in MEMS and NEMS Technologies

Fig. 1 Basic diagram of the surface micromachining process. (View this art in color at www. dekker.com.)

similar to conventional lithography. However, the e-beam must be traced over the mask (a serial process) and is not flash exposed as in conventional lithography. Atomic Force Microscopy writing can be done in several ways. One way is to apply a bias to the sharp tip and alter the material under the tip by exposing it to the high electric field. This may lead to localized oxidation, which can be used to transfer the pattern to the substrate. Alternatively, the tip can be scraped across the surface (especially on soft coatings)

and mechanically remove material from the desired areas. Similar to e-beam lithography, AFM writing is a serial process. In addition to nanoscale writing, the so-called ‘‘spacer lithography’’ can be used to precisely define nanometer-wide gaps. This process consists of the conformal deposition of a nanometer thin film over an exposed step surface. Afterward, the plan surfaces are anisotropically etched, leaving only the nanometer thin ‘‘spacer’’ material on the sidewalls of the vertical step. Unlike e-beam lithography

Thin Film Processes in MEMS and NEMS Technologies

3051

and AFM writing, spacer lithography is a parallel process and is carried out over the whole surface of the sample at once. CHEMICAL PROCESSES USED IN FABRICATION There are a variety of chemical processes used in the manufacturing of MEMS and ICs. These include deposition and etching processes. Some of the more common examples of these processes are described briefly in the sections that follow. Chemical Vapor Deposition One of the most common deposition processes encountered in the fabrication of microsystems is chemical vapor deposition (CVD) (see, e.g., Ref.[2]). This broad category encompasses processes including low-pressure (LP), atmospheric-pressure (AP), plasma-assisted (PA), ion beam-assisted (IBA), and laser-assisted (LA) CVD. Regardless of the particular type of CVD, the process consists of exposing a target substrate to a supply of a gas (or gas mixture) and activating that gas to produce a chemical reaction at the surface of the substrate. This activation method may include heating, striking a plasma, laser irradiation, ion beams, or combinations thereof. The process parameters are carefully controlled so that the deposited material films are of high quality, and have the desired stress and composition properties. In silicon based MEMS processing, common CVD films include polysilicon, silicon oxide, and silicon nitride. For polysilicon films (usually the structural layer), an LPCVD pyrolysis method is generally used with silane (SiH4) as the source gas [see Eq. (1)]. To obtain a uniform film across the wafer, the process is carried out at low pressure to ensure that the deposition is surface reaction controlled and not diffusion limited. Typical process temperatures are in the range of 580–650
C, and pressures between 0.1 and 0.4 Torr.

There are many source gases that can be used to produce silicon oxide. These include mixtures of silane and oxygen, tetraethoxysilane and oxygen, dichlorosilane and N2O, and silane and N2O. An example reaction is shown in Eq. (3). Because there are many options for source gases, the process parameters may range from 400 to 900
C and from 0.1 to 0.5 Torr. D

SiH4 þ O2 ! SiO2 þ 2H2 Etching

To create micromachines, films that have been deposited must be patterned and etched to reveal the desired structures. Often, it is important to etch these structures with vertical sidewalls (anisotropic etching). In this case, most pattern transfer operations (lithography and etch) are carried out using plasma etching. Conceptually, this process is the reverse of deposition. The etching process consists of exposure of the patterned and masked substrate to a low-pressure plasma. The reactive species and ions preferentially etch those areas that are not masked, resulting in the definition of features on the surface. The key to plasma etching is that the products of the reaction of the activated gas and the material to be etched must be volatile (see e.g., Ref.[2]). Reactive ion etching (RIE) and deep reactive ion etching (DRIE) are common examples of plasma etching. In this incarnation, the substrate to be etched is placed on a powered electrode in a plasma chamber. Process gases are admitted into the chamber and a plasma is struck. Because the substrate is directly in the ion flux of the plasma, the ions impinge on the surface and may participate in chemistry. For example, RIE of polysilicon may use SF6 as the reactive gas, and etches the Si by a reaction with fluoride ions to form the volatile product SiF4, as illustrated by Eqs. (4) and (5).a 

SF6 ! SF5þ þ F 

D

SiH4 ! Si þ 2H2

ð1Þ

Another important material is silicon nitride, which is usually employed as an electrical isolation layer. It is typically deposited by the LPCVD or PECVD method. The process gases are dichlorosilane SiCl2H2 and ammonia NH3 [see Eq. (2)] and the process parameters are in the range of 700–900
C and 0.2–0.5 Torr. D

SiCl2 H2 þ NH3 ! Si3 N4 þ 2HCl þ

3 H2 2

ð2Þ

Silicon oxide, typically a sacrificial layer, can be deposited using PECVD and LPCVD methods.

ð3Þ

Si þ 4F ! SF4

ð4Þ ð5Þ

The anisotropic nature of the etching by RIE and DRIE is a result of the directionality of the impinging ions. Sometimes it is useful to etch very deep channels or holes. Using conventional DRIE, some tapering of the sidewall profile is expected. However, an etching process has been developed where sequential etch=passivate steps are performed. This etching process, called the Bosch process, uses conventional DRIE methods as process gases for a short period of time, interrupts

a

Plasma chemistries are complex and these equations are meant only to indicate an example process.

T

3052

Thin Film Processes in MEMS and NEMS Technologies

this process with a fluorocarbon polymer deposit phase, and resumes DRIE. Because of the directionality of the etching process, the polymer on the plan surfaces is quickly removed and the etching process resumes. However, the material removal on the sidewalls is slower, and therefore sidewall surfaces are not readily etched. This process allows very deep features to be etched with little tapering. Because of the way this process is carried out, however, the sidewall surfaces are generally scalloped, and this is a characteristic feature of the Bosch etch process.

RELEASE PROCESSES While MEMS technologies make much use of the preexisting integrated circuit fabrication processes, there is at least one critical step that is unique to MEMS processing. This step, called the release step, refers to the processing stage that frees the microstructure from the sacrificial layers that were used during the fabrication of the device. This is usually accomplished by exposing the micromachine to an etchant fluid that will selectively and completely remove the sacrificial material. In the case of polysilicon based microstructures, the sacrificial layers are typically silicon oxide, and the etchant fluid is an aqueous solution of HF. In some cases, an etching solution of concentrated HF and HCl is used. The presence of HCl helps to protect the silicon nitride from etching by HF. Next, the etchant is completely rinsed away with deionized water. If, at this point, the released structures are simply removed from the rinse liquid and air-dried, they are almost inevitably found adhering to each other. This phenomenon is called ‘‘release stiction’’ and is caused by the action of the curved receding air–liquid interface as it passes over the structures. Alternate processes are typically carried out to avoid the problem of release stiction. One alternate approach is to change the water meniscus shape from concave (wetting) to convex (nonwetting), so that capillary forces are reduced.[4] This is accomplished by altering the composition of the surface by chemically grafting hydrophobic molecules to the surface. This approach is discussed in greater detail in the sections that follow. Other approaches involve the avoidance of liquid–vapor interfaces altogether through supercritical fluid, freeze sublimation drying, and dry-release methods such as the use of vapor HF etching. Fig. 2 shows a P–T diagram that depicts the concepts of supercritical drying and freeze sublimation. Supercritical drying is a process that begins with released microstructures, which have been kept submerged in liquid since their sacrificial layer etch, and then placed in a high-pressure chamber filled with a short alcohol (methanol, ethanol, and isopropanol) solvent. This solvent is then completely

Fig. 2 Diagram depicting the phase paths of critical point drying and freeze- sublimation drying. Note that neither of these processes involve crossing the liquid–vapor line. (View this art in color at www.dekker.com.)

displaced by liquid carbon dioxide (CO2). The chamber is pressurized and heated to cause the liquid CO2 to become supercritical. The chamber is then carefully vented and the structures are removed.[5] Freeze sublimation drying is similar to supercritical drying. This process also begins with released microstructures, which are maintained in a solvent, such as isopropanol. This solvent is then displaced by another solvent (water= isopropanol mixture), which is then frozen. The vapor above this solid is pumped away by a vacuum pump until the solid is completely sublimated and the process is complete.[6] Vapor HF etching is a process that completely avoids the issue of solvent displacements, as the sacrificial layer etch is carried out by a gaseous isotropic etchant. Usually, the samples are placed above a solution of HF and water and the vapors above that solution are used to perform the etch.[7] Polymer ashing is another process related method. This process is fairly complex and involves patterning of a polymer layer during the release. First, structures are partially released by a timed etch. Next, a polymer film is deposited onto the partially released structures. This film is patterned into support posts that hold the structure in position as the remainder of the sacrificial layer is etched away. Because the polymer support structures hold the devices in place, there is no concern for special drying techniques. Finally, the polymer supports are burned away, typically by ashing in an oxygen plasma.[8] This leaves behind fully released and free-standing microstructures.

SURFACE PROCESSING FOR INCREASED RELIABILITY Surface microstructures typically have lateral dimensions of 50–500 mm with thicknesses of 0.1–2.5 mm, and are offset 0.1–2 mm from the substrate. The large

Thin Film Processes in MEMS and NEMS Technologies

surface-area-to-volume ratios of surface and bulk micromachined micromechanisms lead to the dominance of surface and interfacial forces over body forces.[8–12] For commercial viability and industrial growth to continue, micromechanical systems must be built with high yields and reproducible device properties and must exhibit reliability over the expected device lifetime. The traditional high yields experienced in the IC industry extend to MEMS production only up to the point of microstructure release. The release techniques discussed here do not prevent adhesion from occurring during micromachine operation. Microstructure surfaces may come into contact unintentionally through acceleration or electrostatic forces, or intentionally in applications where surfaces impact or shear against each other. When adhesive attractions exceed restoring forces, surfaces permanently adhere to each other causing device failure—a phenomenon known as ‘‘in-use stiction.’’ Stiction is not the only reliability issue that plagues microsystems. Owing to the dimensions of the devices and roughness of the surfaces involved, the mechanical properties of the structural material are very important. Although silicon is a good structural material for the microscale, owing to its high elastic modulus and low density, small area contacts (as in contact points between two rough surfaces) can generate enormous contact pressures, which can create plastic deformations in silicon. For example, the hubs on microgears can become visibly cluttered with wear debris after a few hundred thousand cycles. These microgears seldom last more than a few million cycles before the buildup of wear debris and increased friction irreversibly bind the gear.[13] Current research on the issue of wear seeks to understand the fundamental mechanisms responsible, as well as to address the practical issues by integrating hard materials into the micromachining process. Adhesion, friction, and wear are collectively referred to as tribology. Owing to the extensive infrastructure related to the chosen materials base (silicon), and the relative ease in which silicon based microsystems can be produced, a convenient approach to overcoming some of these reliability issues is to apply coatings for stiction, friction, and wear control to the devices after they are produced. This is analogous to painting buildings or automobiles after they are constructed to make them more weather resistant. Therefore, facile and effective coating processes that satisfy the needs of the microsystems for reliable performance have been developed.

Methods to Quantify MEMS Tribology To quantitatively study stiction, friction, and wear, micromechanical test devices have been designed and

3053

fabricated. One of the most common test devices for stiction quantification is the cantilever beam array.[8] With this device, stiction is tested by electrostatically bringing the beam into contact with the substrate and releasing the electrostatic force. By examination of the behavior of the interaction of the beam with the substrate, one can determine an apparent work of adhesion, which is related to stiction. Other devices have been designed for friction and wear testing. Although there are many types of these devices, they generally consist of a movable structure and a fixed structure. To test friction, the movable structure is contacted against the fixed structure under a prescribed contact (normal) load and a tangential force is ramped up. The device is carefully monitored until a slip occurs and the tangential force at slippage is known. The coefficient of static friction can be determined by knowledge of the applied normal load and the tangential force required to produce a slip. Generally, devices that are used for friction testing can also be used for wear testing. Instead of a single slip event, the movable part is cycled against the fixed part under a given normal load. At the end of the test, scanning electron microscopy or AFM can be used to inspect the contacting parts to gage the amount of wear.

Self-Assembled Monolayer Coatings to Control MEMS Tribology Coatings that are to be applied to micro- and nanostructures must satisfy a series of constraints related to their properties and the process used to generate them. They must be conformal and uniform because all surfaces of the device must be coated with the same amount of coating to avoid the introduction of extraneous stresses. Coatings must be very thin for design fidelity reasons, and so that it does not adversely influence the operation of the device. (For example, a coating must not adversely affect the resonance frequency or quality factor for a MEMS resonator.) They must be thermally and chemically stable because the device is expected to last a long time and endure packaging steps. The coating process must be compatible with the device and should be easily implemented. In light of these constraints, an effective chemical modification for anti-stiction treatments involves the application of a molecular film to the micromachine surface. This is most often accomplished through a process known as self-assembled monolayer (SAM) deposition.[14] Self-assembled monolayers are molecular films that spontaneously form on a (usually pretreated) surface upon exposure to a reactive precursor molecule. SAM precursors generally consist of three main parts: a terminal group, a backbone, and a head group. Fig. 3 shows these parts on a model SAM

T

3054

precursor molecule, octadecyltrichlorosilane (OTS). The head group is a chemically reactive group that is chosen to bind the molecule to the MEMS surface. The backbone is often an n-alkyl chain (i.e., –(CH2)n–) whose function is to assist in the assembly of a wellpacked monolayer film. The terminal group is the part of the molecule that comprises the new surface after monolayer self assembly and imparts the desired surface functionality to the coating. For adhesion control, a highly hydrophobic surface is desired, and the terminal groups of choice are hydrophobic (water disliking) –CH3 or –CF3 groups (the –CF3 terminal groups result in monolayer coatings that have nonstick behavior similar to TeflonÕ). It has been demonstrated that, when properly integrated into the microstructure release process, SAM coatings deposited from solution can eliminate release stiction, reduce in-use stiction, and decrease both friction and wear in microengines.[11,12] Several classes of organic films have been explored as antistiction agents. For aluminum surfaces, perfluoroalkanoic acids have been successfully employed. On silicon surfaces, chloroilane based films, which include alkyl- and perfluoroalkyl-trichlorosilane SAMs, dialkyldichlorosilane, tris-dimethylaminosilanes, and primary alkene, alcohol and aminebased molecular films have been tested. Perfluoroalkanoic acids Perhaps the best example of antistiction technology for MEMS to date is the coating process employed by Texas Instruments on the DMD2 device. The DMD

Fig. 3 An example SAM precursor molecule (OTS) with the major parts of the molecule labeled. Alternatively, the terminal group and backbone may be collectively referred to as the ‘‘pendant’’ group. (View this art in color at www.dekker.com.)

Thin Film Processes in MEMS and NEMS Technologies

is a MEMS device, which consists of an array of a million or more rotatable aluminum mirrors. Photoresist is used as the sacrificial layer and is removed at the release step using a remote plasma containing oxygen and fluorine species. The device has contacting surfaces in relative motion, which are susceptible to adhesion, friction, and wear, and require lubrication. About 50 different lubrication schemes were investigated for DMD, ranging from SAMs to fluids to solid lubricants.[15] The most successful ones reportedly are perfluorinated n-alkanoic acids (CnF2n1O2H), which form selfassembled monolayers on aluminum oxide surfaces (see Fig. 4). Within this class of SAMs, perfluorodecanoic acid (PFDA, n ¼ 10) was found to be the lubricant of choice to minimize the friction coefficient and the possibility of thermal decomposition. To keep moisture out and create a background pressure of PFDA, hermetic chip package is used. When properly lubricated, devices have operated for more than 350 billion cycles.

Chlorosilane based monolayers Among the SAMs suitable for coating silicon, the OTS based variety is the most widely used. Some of the properties for the OTS SAM, of relevance to MEMS, are listed in Table 1. Although there is much debate concerning the true structure of the OTS monolayer on silicon oxide, Fig. 5 illustrates a simplified conceptual model of the film. In addition to the OTS precursor molecule, there are many other molecules of the form RSiCl3 and R,R0 SiCl2 that are used to produce oriented hydrophobic monolayers on silicon surfaces, with R and R0 denoting an aliphatic carbon chain. It has been demonstrated that the most effective chlorosilane reagents to produce hydrophobic coatings on oxidized silicon surfaces are perfluorinated alkyltrichlorosilanes.[19] Indeed, lower values for the apparent work of adhesion were reported for 1H,1H,2H,2H-perfluorodecyltrichlorosilane [CF3(CF2)7(CH2)2SiCl3; FDTS] in comparison to OTS.[17] Table 1 shows an interfacial property comparison among different surface preparations. Although these SAM coatings have been shown to effectively alleviate both release and in-use stiction, they possess a number of limitations intrinsically related to their chemistry. A serious limitation arises from the ability of the precursor molecule to polymerize.[20] As long as the precursor molecule has a functionality greater than one, bulk polymerization can occur, and the higher the functionality, the greater the likelihood for polymerization. This is potentially dangerous for micromachines in that large particulates, such as polymerized clusters of SAM precursor molecules (which can be several micrometers

Thin Film Processes in MEMS and NEMS Technologies

3055

T

Fig. 4 Structural formulas for perfluorodecanoic acid (PFDA) perfluorooctyltrichlorosilane (FOTS), dimethyldichlorosilane (DDMS), 1-octa decene, and perfluorodecyl-tris-(dimethylamino)silane (PF10TAS).

in diameter), can mechanically interfere with the device operation. Unfortunately, there is no satisfactory method for removal of the polymerized clusters once they have agglomerated on the surfaces of the substrate or micromachines. Another limitation of longchain chlorosilane SAM coatings (such as OTS) arises from the coating procedure, which must be performed from the liquid phase because of the extremely low vapor pressure of the precursor. Recent developments in the chlorosilane based monolayer technology address some of these issues by performing the coating process in the vapor phase. Vapor phase processing eliminates the use of organic solvents and greatly simplifies handling of the samples. Moreover, the stoichiometry of the precursor molecules can be more precisely controlled. It has been demonstrated that monolayer films that are produced from the precursor tridecafluoro-1,1,2,2-tetrahydrooctyltrichlorosilane [CF3(CF2)5(CH2)2SiCl3; FOTS] in

a low-pressure CVD style reactor exhibit low adhesion energies.[19] Fig. 4 shows the structure of FOTS. Additionally, in situ plasma cleaning of the sample as well as in situ measurement of the film growth provide excellent process uniformity, reproducibility, and monitoring capability.[21] This approach demands that the microstructures undergo some form of dry release process, such as critical point drying or vapor HF etching, before receiving the monolayer coating. The dimethyldichlorosilane [(CH3)2SiCl2; DDMS)] monolayer has also been proposed as a promising surface coating for MEMS. Fig. 4 shows the structure of DDMS. The monolayer has been compared to the OTS SAM with respect to the film properties and their effectiveness as antistiction coatings for micromechanical structures.[16,22] While water and hexadecane contact angles are comparable, the apparent work of adhesion for the DDMS monolayer is somewhat higher than that for OTS (Table 1). Furthermore,

Table 1 Physical property data for various surface treatments Contact angle (
)

Coefficient of static friction

Thermal stability in air (
C)

Water

Hexadecane

Work of adhesion (mJ/m2)

OTS

110

38

0.012

0.07

225

FDTS

115

68

0.005

0.10

DDMS

103

38

0.045

0.28

Surface treatment

Selective to Si

Reference

High

No

[16]

400

Very high

No

[17]

400

Low

No

[16]

Negligible

Yes

[18]

—

—

[16]

Octadecene

104

35

0.009

0.05

200

Oxide

0–30

0–20

20

1.1

—

Particulate formation

3056

Thin Film Processes in MEMS and NEMS Technologies

Fig. 5 A simplified diagram of the formation of an OTS monolayer. The first reaction (not shown) is the complete hydrolysis of the OTS molecule, resulting in the formation of 3HCl and the trisilanol form of OTS shown in the figure. Subsequently, water elimination reactions result in Si–O–Si linkages between adjacent OTS molecules and=or the oxidized substrate.

coefficient of static friction data indicate that the DDMS films are not as effective at lubrication as the OTS SAMs are, although both exhibit much improvement over chemical oxide. However, AFM data show that the samples that receive DDMS films accumulate fewer particles during processing than those that get the OTS SAM treatment. The thermal stability of the DDMS film in air far exceeds that of the OTS SAM, as the DDMS remains very hydrophobic to temperatures well upward of 400
C (Fig. 6).

Alkene based monolayers Another surface modification technique involves the free radical reaction of a primary alkene (e.g., C16H33CH¼CH2, 1-octadecene) with hydrogen terminated silicon.[23] Fig. 4 shows the structure of 1-octadecene while Fig. 7 displays a simplified diagram

of the resulting monolayer structure. This monolayer coating has several key advantages over OTS and FDTS based SAMs: 1. The coating does not produce HCl at any stage in the monolayer formation whereas chlorosilanebased chemistry does. 2. The coating does not require the formation of an intervening oxide layer. 3. The film formation procedure for alkene based monolayers is simpler than for chlorosilane based SAMs for two main reasons. First, the surface oxidation step is eliminated. Second, the coating solution does not need to be conditioned before use, as water is not a reagent in this process. 4. The coating process is much more robust because it is essentially insensitive to relative humidity.

Fig. 6 Water contact angle as a function of substrate temperature. The substrate was exposed to high temperature for 5 min in laboratory air. Note the extreme stability of the DDMS monolayer in comparison to the OTS and octadecene monolayers.

Thin Film Processes in MEMS and NEMS Technologies

3057

T

Fig. 7 A simplified diagram of the formation of an octadecene monolayer. The precursor molecule (1-octadecene) is bound directly to the silicon, with no oxide layer.

5. The coated surfaces have much fewer particulates in comparison to those coated with OTS. 6. The coating process can be made selective to coat only exposed silicon by generating radicals using a radical initiator. These improvements have been achieved without sacrificing the antistiction characteristics of the film, in that water and hexadecane contact angles, apparent work of adhesion, and coefficient of static friction data are found to be similar to those of OTS, (see Table 1).[18] In some devices, the lubrication properties of a film may be more critical than in others. Fig. 8 shows a complex torque multiplying transmission. In this example, the large number of gears brings the issue of friction and lubrication to the foreground. When the device is in operation, every gear experiences sliding contact at its hub, as well as at every tooth

Fig. 8 Micrograph of a complex micromechanical system. Here, there are a large number of sliding contacts (gear hubs and intermeshing gear teeth) that must be properly lubricated for the system to function.

intermeshing location. Initial releases of the device that employed oxidized surfaces were not viable. Attempts at surface modification with OTS were marginally successful, but devices ultimately failed because of the high degree of particle contamination and binding friction on the gear hubs. To date, the only devices of this type to successfully and reliably operate are those that receive octadecene films as part of their release process. tris-Dimethylaminosilane based monolayers Another binding chemistry that has been used to successfully attach perfluoroalkyl groups to micromachines surfaces is based on the precursors (tridecafluoro-1,1,2,2,-tetrahydrooctyl)tris-dimethylamino silane (PF8TAS) and (heptadecafluoro-1,1,2,2,-tetrahydrodecyl)tris-dimethylamino silane (PF10TAS).[24] Fig. 4 shows a structural diagram of PF10TAS. These precursors are not commercially available, but can be synthesized from their corresponding trichlorosilanes (FOTS and FDTS, respectively) and dimethylamine. It should be noted that the aminosilane precursors are extremely sensitive to water and must be kept rigorously anhydrous. The process for applying the aminosilane to micromachines is essentially the same as that used for the chlorosilanes. However, an important distinction is that there is no water vapor added to the chamber during the deposition. Although detailed process parameters are not given, the apparatus used to deposit the aminosilane based monolayers is also similar to that which is used for chlorosilane deposition.[24] The aminosilane coating PF8TAS has been characterized on Si(100) and on microengines. The AFM analysis on coated Si(100) confirms that the process does not generate particles. Ellipsometry measurements conducted on coated Si(100) in controlled humidity

3058

Thin Film Processes in MEMS and NEMS Technologies

environment revealed that the PF8TAS films were effective at preventing water sorption.[24]

Thin hard coatings for wear resistance Although much progress has been made in the area of monolayer films for stiction and friction control, these techniques are not effective at controlling wear. Currently, the mechanism for wear on the micro- and nanoscale is not fully understood, and research has shown that wear in micromachines depends on a number of factors, including several related to the environmental conditions of the device. Although the full details about the mechanism of wear may not be well known, it is generally accepted that one approach to combat the problem of wear is to integrate harder materials at the contacting surfaces. This means that, for silicon structures, materials like diamond, SiC, W, Al2O3, or others must be coated on the micromachine surfaces. Of these materials, diamond is the hardest. However, for some applications diamond may not be the best choice as exposure to moderate temperatures in oxidizing ambients can etch diamond. A more chemically inert material with very high hardness is silicon carbide. Silicon carbide is well known as an attractive material for demanding mechanical and high-temperature applications. Thin SiC coatings share many of the desirable properties of bulk SiC, such as exceptional tribological properties and corrosion resistance. Unfortunately, the conventional method for heteroepitaxial growth of 3CSiC on Si (based on a dual-precursor CVD) requires high growth temperatures (typically > 1000
C).[25] Because released polycrystalline silicon (polysilicon) microstructures cannot withstand such high temperatures without some deformation or warpage, the conventional SiC CVD method cannot be used to coat existing polysilicon devices. Another CVD process based on the single source precursor 1,3-disilabutane (DSB) has recently been developed.[26] The primary advantage of this method is that the deposition temperature can be much lower than in the conventional CVD method. In fact, the pyrolysis reaction of DSB [shown in Eq. (6) below] can be carried out at temperatures as low as 650
C. However, a deposition temperature of 800
C is chosen for micromachine coatings because it leads to reasonable growth rate, good uniformity, and low film stress.[27] Here, it is demonstrated that this CVD method is suitable for postprocessing of released polysilicon micromachine devices and leads to enhanced reliability and lifetime of microstructures. D

SiH3 CH2 SiH2 CH3 ! 2SiC þ 5H2

ð6Þ

The integration of hard material coatings into the micromachining process scheme is also complicated by the need to coat all surfaces of the device. This requirement leads to deposition processes that occur after the micromachines are fabricated and released. It therefore becomes very important that the coating process be completely uniform and conformal as application of uneven film thickness can alter the stresses on a device and cause it to deform. Recently, much research has focused on atomic layer deposition (ALD) methods. The ALD method is, in principle, capable of depositing perfectly uniform and conformal thin coatings with atomic level control of thickness. Briefly, the method utilizes self-limiting, surface absorption reactions of alternating species to form the desired film one atomic layer at a time (see, e.g., Ref.[28]). The viability of applying ALD films of Al2O3 to released micromachines has been shown.[28,29] It has been indicated that ALD Al2O3 films show promise for reduced friction and wear and reduced electrical shorting.[28,29] Another hard material of interest is titania (TiO2). Owing to its optical, photochemical, and catalytic properties, as well as its biocompatibility and activity in certain sensor applications, TiO2 has received attention in materials research. TiO2 is also reported to have low friction and wear, and is chemically stable.[30] In addition to these methods, the selective deposition of tungsten metal has been explored as an antiwear coating on polysilicon microstructures.[31] This coating is accomplished by heating polysilicon microstructures in a tungsten hexafluoride (WF6) gas at about 450
C [see Eqs. (7) and (8)]. In this manner, only exposed silicon is coated with W, and the reaction is self-limiting because the deposition of W obscures the underlying Si. W coated micromachines have been shown to exhibit lower wear and improved lifetime vs. uncoated microstructures.[31] D

2WF6 þ 3Si ! 2W þ 3SiF4 D

WF6 þ 3Si ! W þ 3SiF2

ð7Þ ð8Þ

CONCLUSIONS Thin film processes and the chemistry of surfaces play important roles in the technological implementation of MEMS and NEMS devices. Thin film deposition of a variety of materials is a mature technology, whereas MEMS surface coating technologies have only recently come of age. Unlike the commercially employed Si and Al, the development of structural layers that can perform

Thin Film Processes in MEMS and NEMS Technologies

reliably without further coatings is still in its infancy. Silicon carbide and diamond are emerging as the most promising technologies in this regard.

REFERENCES 1. Petersen, K. Silicon as a mechanical material. Proc. IEEE Electron. Devices 1982, 70, 420–457. 2. Madou, M.J. Fundamentals of Microfabrication; CRC Press: Boca Raton, 1997. 3. Howe, R.T. Surface micromachining for microsensors and microactuators. J. Vacuum Sci. Technol. B 1988, 6, 1809–1813. 4. Abe, T.; Messner, W.C.; Reed, M.L. Effects of elevated temperature treatments in microstructure release procedures. J. Microelectromech. Syst. 1995, 4, 66–75. 5. Mulhern, G.T.; Soane, D.S.; Howe, R.T. Supercritical carbon dioxide drying of microstructures. In Technical Digest, 7th International Conference on Solid-State Sensors and Actuators, Jun 1993; 296–299. 6. Guckel, H.; Sniegowski, J.J.; Christenson, T.R.; Raissi, F. The application of fine-grained polysilicon to mechanically resonant transducers. Sens. Actuators A 1990, 346–351. 7. Anguita, J.; Briones, F. HF=H2O vapor etching of SiO2 sacrificial layer for large-area surfacemicromachined membranes. Sens. Actuators A 1998, 64 (3), 247–251. 8. Mastrangelo, C.H. Adhesion-related failure mechanisms in micromechanical devices. Tribology Lett. 1997, 3 (3), 223–238. 9. Komvopoulos, K. Surface engineering and microtribology for microelectromechanical systems. Wear 1996, 200, 305–327. 10. Tas, N.; Sonnenberg, T.; Jansen, H.; Legtenberg, R.; Elwenspoek, M. Stiction in surface micromachining. J. Micromech. Microeng. 1996, 6, 385–397. 11. Maboudian, R.; Ashurst, W.R.; Carraro, C. Tribological challenges in micromechanical systems. Tribology Lett. 2002, 12 (2), 95–100. 12. Maboudian, R.; Carraro, C. Surface chemistry and tribology of MEMS. Annu. Rev. Phys. Chem. 2004, 55, 35–54. 13. Tanner, D.M. Reliability of surface micromachined microelectromechanical actuators. 22nd International Conference in Microelectronics, IEEE Electron Devices Society: Nis, Yugoslavia, May 2000; 97–104. 14. Ulman, A. An Introduction to Ultrathin Organic Films; Academic Press, Inc: San Diego, 1991. 15. Henck, S.A. Lubrication of Digital Micromirror Devices2. Tribology Lett. 1997, 3 (3), 239–247.

3059

16. Ashurst, W.R.; Yau, C.; Carraro, C.; Maboudian, R.; Dugger, M.T. Dichlorodimethylsilane as an anti-stiction monolayer for MEMS: a comparison to the octadecyltrichlorosilane self assembled monolayer. J. Microelectromech. Syst. 2001, 9 (4), 41–49. 17. Srinivasan, U.; Houston, M.R.; Howe, R.T.; Maboudian, R. Alkyltrichlorosilane-based self assembled monolayer films for stiction reduction in silicon micromachines. J. Microelectromechan. Syst. 1998, 7 (2), 252–260. 18. Ashurst, W.R.; Yau, C.; Carraro, C.; Lee, C.; Kluth, G.J.; Howe, R.T.; Maboudian, R. Alkene based monolayer films as anti-stiction coatings for polysilicon MEMS. Sens. Actuators A 2001, 91, 239–248. 19. Banga, R; Yarwood, J; Morgan, A.M.; Evans, B.; Kells, J. FTIR and AFM studies of the kinetics and self-assembly of alkyltrichlorosilanes and (perfluoroalkyl)trichlorosilanes onto glass and silicon. Langmuir 1995, 11 (11), 4393–4399. 20. Maboudian, R.; Ashurst, W.R.; Carraro, C. Selfassembled monolayers as anti-stiction coating for mems: characteristics and recent progress. Sens. Actuators A. 2000, 82, 219–223. 21. Mayer, T.M.; de Boer, M.P.; Shinn, N.D.; Clews, P.J.; Michalske, T.A. Chemical vapor deposition of fluoroalkylsilane monolayer films for adhesion control in microelectromechanical systems. J. Vac. Sci. Technol. B. 2000, 18 (5), 2433–2440. 22. Kim, B.H.; Chung, T.D.; Oh, C.H.; Chun, K. A new organic modifier for anti-stiction. J. Microelectromech. Syst. 2001, 10 (1), 33–40. 23. Sung, M.M.; Kluth, G.J.; Yauw, O.W.; Maboudian, R. Thermal behavior of alkyl monolayers on silicon surfaces. Langmuir 1997, 13 (23), 6164–6168. 24. Hankins, M.G.; Resnick, P.J.; Clews, P.J.; Mayer, T.M.; Wheeler, D.R.; Tanner, D.M.; Plass, R.A. Vapor deposition of amino-functionalized self-assembled monolayers on MEMS. In Proceedings of SPIE: Reliability, Testing, and Characterization of MEMS=MOEMS II; Ramesham, R., Tanner, D.M., Eds.; SPIE; 2003; Vol. 4980, 238–247. 25. Mehregany, M.; Zorman, C.A.; Roy, S.; Fleischman, A.J.; Wu, C.H.; Rajan, N. Silicon carbide for microelectromechanical systems. Int. Mater. Rev. 2000, 45 (3), 85–108. 26. Stoldt, C.R.; Carraro, C.; Ashurst, W.R.; Gao, D.; Howe, R.T.; Maboudian, R. A low-temperature CVD process for silicon carbide MEMS. Sens. Actuators A 2002, 97–98, 410–415. 27. Gao, D.; Wijesundara, M.B.J.; Howe, R.T.; Maboudian, R. Characterization of residual

T

3060

strain in SiC films deposited using 1,3-disilabutane for MEMS application. J. Microlithography Microfabrication Microsyst. 2003, 2 (4), 259–264. 28. Mayer, T.M.; Elam, J.W.; George, S.M.; Kotula, P.G.; Goeke, R.S. Atomic-layer deposition of wear-resistant coatings for microelectromechanical devices. Appl. Phys. Lett. 2003, 82 (17), 2883–2885. 29. Hoivik, N.D.; Elam, J.W.; Linderman, R.J.; Bright, V.M.; George, S.M.; Lee, Y.C. Atomic layer deposited protective coatings for microelectromechanical systems. Sens. Actuators A 2003, 103 (1–2), 100–108.

Thin Film Processes in MEMS and NEMS Technologies

30. Ashurst, W.R.; Jang, Y.J.; Magagnin, L.; Carraro, C.; Sung, M.M.; Maboudian, R. Nanometer-thin titania films with SAM-level stiction and superior wear resistance for reliable MEMS performance. Proceedings of the 17th IEEE International Conference on MEMS, 2004, 153–156. 31. Mani, S.S.; Fleming, J.G.; Sniegowski, J.J.; de Boer, M.P.; Irwin, L.W.; Walraven, J.A.; Tanner, D.M.; Dugger, M.T. Chemical vapor deposition coating for micromachines. In New Methods, Mechanisms and Models of Vapor Deposition; Materials Research Society Symposium Proceedings; 2000; Vol. 616, 21–26.

Thin Film Science and Technology T T. L. Alford Department of Chemical and Materials Engineering, Arizona State University, Tempe, Arizona, U.S.A.

J. Kouvetakis Department of Chemistry and Biochemistry, Arizona State University, Tempe, Arizona, U.S.A.

J. W. Mayer Department of Chemical and Materials Engineering, Arizona State University, Tempe, Arizona, U.S.A.

INTRODUCTION Thin film science and technology is the deposition and characterization of layered structures, typically less than a micron in thickness, which are tailored from the atomic scale upwards to achieve desired functional properties. Deposition is the synthesis and processing of thin films under controlled conditions of chemical processing. Chemical vapor deposition (CVD) and gasphase molecular beam epitaxy (MBE) are two processes that allow control of the composition and structure of the films. Characterization is the instrumentation that use electrons, X-ray, and ion beams to probe the properties of the film. Epitaxial films of semiconductors are used for their electronic properties to emit light in the infrared (IR) and the ultraviolet rays. The remaining work discusses two techniques in thin film analysis, Rutherford backscattering spectrometry (RBS) and X-ray diffractrometry with emphasis on strain measurements. Rutherford backscattering spectrometry is illustrated with analysis of silicide formation as an example of thin film reactions. Silicon– germanium–carbon films serve as an example of strain calculations. The critical aspect of modern thin film technology is the growth of epitaxial layers.[1] We illustrate the importance of epitaxial growth and strainengineering in producing light emitting films on silicon substrates.

film analysis. The techniques provide film thickness and composition, lattice structure, and epitaxy and strain. Rutherford backscattering spectrometry analysis determines how the composition varies as a function of depth and is used to characterize thin films and thin film reactions. During ion-beam analysis, the incident particle (typically a proton or helium ion) penetrates into the thin film and undergoes inelastic collisions, with target electrons, and loses energy as it transverses the sample.

Scattering Kinematics During the penetration of the helium ions a small fraction undergo elastic collisions with the target atom, which defines the backscattering signal. Fig. 2 shows a schematic representation of the geometry of an elastic collision between a projectile of mass M1 and energy E0 with a target atom of mass M2 initially at rest. After collision, the incident ion is scattered back through an angle W and emerges from the sample with an energy E1. The target atom after collision has a recoil energy E2. There is no change in target mass, because nuclear reactions are not involved and energies are nonrelativistic. For M1 < M2, the ratio of the projectile energies, the kinematic factor (K) is given by the following: E1 E "0 #2 1 ðM22
M12 sin2 WÞ 2 þ M1 cos W ¼ M2 þ M1

K ¼ RUTHERFORD BACKSCATTERING SPECTROMETRY The characterization of thin film structures is now a well-described laboratory technique.[2,3] Fig. 1 is a schematic representation of sources and detectors used in Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120018083 Copyright # 2006 by Taylor & Francis. All rights reserved.

ð1Þ

Eq. (1) shows that the energy of the backscattered particle is a function of the incident particle and target atom masses, the scattering angle, and incident energy. 3061

3062

Thin Film Science and Technology

Fig. 1 Schematic of radiation sources and detectors in thin film analysis techniques. Analytical probes are represented by almost any combination of source and detected radiation, i.e., photons in and photons out or ions in and photon out.

Fig. 2 Schematic representation of an elastic collision between a particle of mass M1 and initial energy E0 and a target atom of mass M2. After the collision, the projectile and target atoms have energies of E1 and E2, respectively.

Thin Film Science and Technology

3063

Scattering Cross-Section The identity of target elements is established by the energy of the scattered particles after an elastic collision. The number of target atoms is given by measuring the yield Y, the number of backscattered particles for a given value of incident particles Q. The detector’s solid angle is given as O. The areal density, the number of atoms per unit area, NS, is determined from the scattering crosssection s(W) by:

The cross-sections are relatively large, such that one can detect submonolayers of most heavy mass elements on silicon. For example, the yield of 2.0 MeV helium ions from 1014 cm
2 silver atoms (approximately 1=10th of a monolayer) is 800 counts for a current of 100 nano-amperes for 15 min and detector area of 5 msr. This represents a large signal for a small amount of atoms on the surface.

Depth Scale NS

Y ¼ sðWÞOQ

ð2Þ

This is shown schematically in Fig. 3. A narrow beam of fast particles impinges on a thin uniform target that is wider than the beam. At a scattering angle W from the direction of incidence, an ideal detector is located that counts each particle scattered in the differential solid angle dO. In the simplest approximation, the scattering cross section is given by: sðWÞ ¼




Z 1 Z 2 e2 4E

2



1 sin4 W2

ð3Þ

Light ions, such as helium, lose energy through inelastic collision with atomic electrons. In backscattering spectrometry, where the elastic collision takes place at depth t below the surface, there is energy loss along the inward path and on the outward path as shown in Fig. 4. The total is given by the relationship: 
dE 1 dE DE ¼ Dt K þ  ¼ Dt½S dX in cos W dX out

ð4Þ

where dE=dX is the rate of energy loss with distance and [S] is the energy loss factor. The particle loses

Fig. 3 Schematic of a typical scattering geometry. Only particles that are scattered within the solid angle O spanned by the solid-state detector are detected.

T

3064

Fig. 4 Energy loss components for a projectile that scatters from depth t. The particle loses energy DEin via inelastic collisions with electrons along the inward path. There is energy loss DEs in the elastic scattering process at depth t. There is energy lost to inelastic collisions DEout along the outward path. For an incident energy E0, the energy of the exiting particle is E1 ¼ E0
DEin
DEs
DEout.

energy DEin via inelastic collisions with electrons along the inward path. There is energy loss DEs in the elastic scattering process at depth t. There is energy loss due to inelastic collisions DEout along the outward path. For an incident energy E0 the energy of the exiting particle is E1 ¼ E0
DEin
DEs
DEout. Thin Film Reactions Heat treatment of deposited thin metal films on Si leads to the formation of silicides. The analysis of these silicide layers is a strong demonstration of the power of Rutherford backscattering thin film technology. The thin metal films react with the Si at temperatures substantially below those indicated in equilibrium phase diagraph. The phase Ni2Si forms at temperatures around 300 C, where the equilibrium phase diagram shows high temperatures above 800 C for this phase. This is clearly an interface reaction. Rutherford backscattering is a pioneering tool in the silicide formation studies. We illustrate the analysis of the formation of Nisilicide in Fig. 5, which plots backscattering yield versus the kinetic energy of the backscattered particles whose incident energy was 2.0 MeV. The solid line is the spectrum from a 200 nm layer of Ni deposited on Si. The thickness is indicated by the width of the energy signal.

Thin Film Science and Technology

After heat treatment at 250 C, there is a step in the rear edge of the Ni signal and the forward edge in the Si signal. Analysis of the ratios of the heights of the Ni to Si signals indicates a composition of Ni2Si. The phase Ni2Si was confirmed by X-ray diffraction measurements (not shown). Upon further heat treatment, the width of the energy steps increased indicating that the compound has grown in thickness. By measurement of the thickness of the film as function of time and temperature, one can determine the growth mechanism (diffusion limited with 1 t =2 growth rate). Other silicides, such as CrSi2, exhibit limited interfacial interaction (growth proportional to t). One can also implant inert markers of Ar markers into the films and measure the displacement of the markers during heat treatment. If the marker displaces toward the surface, the metal is the moving species. If the marker displaces deeper into the sample, then the silicon is the moving species. Further examples of these studies are given in Tu, Mayer, and Feldman.[1]

ION CHANNELING When a single-crystal sample is mounted on a goniometer such that a major crystallographic axis of the sample is aligned within approximately 0.1 or 0.5 of the incident beam, the crystal lattice can steer the trajectories of the incident ions penetrating into the crystal. It is this steering of the incident energetic beam that is known as ion channeling; the atomic rows and planes are guides that steer the energetic ions along the channels between rows and planes. The channeled ions do not approach close to the lattice atoms to be backscattered. Hence, the range (the total distance that a ion penetrates a solid) is increased by several fold. The reduced probability of scattering results in a two-orders of magnitude reduction in the yield by an aligned spectrum compared to that when the incident ions are misaligned from the lattice atoms, a random spectrum. Fig. 6 shows schematically a random and aligned spectrum for MeV helium ions incident on silicon. The characteristic feature of the aligned spectrum is the peak at the high energy end of the spectrum. This peak is a result of ions scattered from the outermost layer of atoms directly exposed to the incident beam. This peak is referred to as the surface peak. Behind the surface peak, at lower energies, the aligned spectra drops to a value of 1=40th of the silicon random spectrum indicating that nearly 97% of the incident ions are channeled and do not make close impact collisions with the lattice atoms. The rise in the aligned spectrum at lower energies represent the ions that are dechanneled, deflected from the steering by the lattice

Thin Film Science and Technology

3065

T

Fig. 5 Rutherford backscattering spectra of a 200-nm Ni thin film on a silicon substrate before and after annealing. (View this art in color at www.dekker.com.)

atoms, which can then collide in close impact collisions with the lattice atoms and hence directly contribute to the backscattering spectra. The ratio between the aligned minimum yield and random yield for the same

channel number gives the wmin value. The crystalline quality is given by the values of wmin at a specific energy, e.g., it is equal to 0.028 for Si when using 2.0 MeV helium ions.

Fig. 6 Random and aligned (channeled) backscattering spectrum from a single crystal. The surface peak corresponds to the small peak at the high energy end of the spectrum signal. The yield behind the peak is reduced because the atoms are shielded from close encounter elastic collisions from the ion beam that is channeled along the axial rows of the crystal.

3066

The application of channeling to RBS is used to determine the amount of damage in ion-implanted single-crystal silicon and the lattice location of ion-implanted dopant atoms. One important example of the contribution of channeling to integrated circuit technology is the analysis of damage evolution during thin film deposition. Modern day ion channeling analysis is done in unison with transmission electron microscopy and=or high resolution X-ray diffraction analysis.

Thin Film Science and Technology

for the reflected waves equals integer (n) multiples of l: nl ¼ 2 DP ¼ 2 dhkl sin W

ð5Þ

Hence, Bragg’s Law, nl ¼ 2dhkl sin W defines the condition for diffraction. The typical X-ray spectrum is a plot of intensity verses angle, e.g., 2W. The phase can be indentified by comparing the spectrum to the Powder Diffraction File compiled by the international Center for Diffraction Data, formerly known as Joint Committee on Powder Diffraction Standards.

X-RAY DIFFRACTOMETRY Bragg derived a description for coherent scattering from an array of periodic scattering sites, i.e., atoms in a crystalline solid. The scalar description of diffraction considers the case of monochromatic radiation impinging on two sheets of atoms in the crystal spaced dhkl between reflecting planes. The wavelength l of the radiation is smaller than the interatomic spacing dhkl of the specific (hkl) planes. Bragg invoked the Law of Reflectivity (or Reflections) that states that the scattering incident angle and exiting angle must be equal, Win ¼ Wout under the condition of coherent scattering. The wavelets scattered by the atoms combine to produce constructive inference if the total path difference 2 DP

Strain Measurements The strain in a layer is determined by comparing the perpendicular and parallel lattice spacings (a) of the film to that of the underlying substrate asub, determining if they are larger or smaller and by how much. Fig. 7 shows a Si film containing Ge (large circles) on a substrate of Si to illustrate the perpendicular and parallel lattice constant. The film is psedomorphic because the lattice lines up with that of the substrate atoms. These values are expressed as [Da=a]? and [Da=a]jj, for the difference in the perpendicular and parallel lattice constants, respectively. Since these differences are measured relative to the substrate, the

Fig. 7 Schematic of a Si film containing Ge (large circles) on a substrate of Si to illustrate the perpendicular and parallel lattice constant.

Thin Film Science and Technology

substrate lattice spacings must be known. From these parameters, a strain value is calculated by determining the difference in the lattice spacings of the film from its bulk relaxed state.

Silicon–Germanium–Carbon Films on Silicon The application of X-ray diffraction in thin film samples is illustrated by diffraction measurement of epitaxial layers of Si–Ge–C on silicon. Initially, a scan is taken about a plane parallel to the surface; this is called a symmetrical scan, since Win and Wout are identical. In Fig. 8, the Si substrate has an (0 0 1) orientation and the first allowed reflection that is parallel to the surface is the (0 0 4) reflection. The symmetrical scan in this case provides information about the perpendicular spacing (a?) between the places. To obtain information about both a? and parallel spacing (ajj), an asymmetrical scan is taken. This involves scanning a plane that is not parallel to the surface. In Si this is typically the ð2 2
4Þ and (2 2 4) reflection as shown in Fig. 9. In one case the spectrum is a glancing incidence (2 2 4) reflection and the other is a ð2 2
4Þ glancing exiting reflection. In each of these scans, note the existence of a peak from the Si substrate and from the Si–Ge–C film. Satellite peaks above and below the primary reflection are due to thickness interference fringes. In strain calculations, the substrate is assumed to be unstrained and hence all measurements are made relative to the substrate’s lattice constant.

Fig. 8 Rocking curve XRD of an epitaxial layers of Si–Ge–C layers on silicon. The symmetrical scan is of the first allowed reflection that is parallel to the surface, i.e., the (0 0 4) reflection. Note the existence of a peak from the Si substrate and from the Si–Ge–C film.

3067

Calculation of Strain The perpendicular and parallel lattice constants are calculated by first measuring the angular peak separation from glancing incident f2 2 4g reflections, o1, and the glancing exit ð
2
2 4Þ reflections, o2. The deviation of the Bragg angle between the substrate and layer, DW, is calculated from DW ¼ 0:5ðo1
o2 Þ

ð6Þ

Due to tetragonal distortion in the epitaxial layer, the angle between the (2 2 4) planes and the surface in the substrate will not be the same as this angle in the film. This difference, Dc was calculated from: Dc ¼ 0:5ðo1
o2 Þ

ð7Þ

The perpendicular lattice constant of the film is calculated by ½Da=a? ¼ DW cotðWB Þ þ Dc tanðcÞ

ð8Þ

where WB is the Bragg angle for the reflection measured and c is the angle between the (2 2 4) planes and the surface. Both angles refer to the substrate. The parallel lattice constant of the film is calculated from: ½Da=ajj ¼
DW cotðWB Þ
Dc cotðcÞ

ð9Þ

Negative values for [Da=a]? and [Da=a]jj indicate that the dimension in the film is larger than that of the substrate. When the perpendicular lattice constant

Fig. 9 A glancing incidence (2 2 4) reflection and a glancing existing reflection (2 2 4) of an epitaxial layers of Si–Ge–C layers on silicon. Note the existence of a peak from the Si substrate and from the Si–Ge–C film.

T

3068

for a pseudomorphic film is larger than the substrate, the film has a larger unit cell and is in compression for [Da=a]? ¼ 0.016638 and [Da=a]jj ¼ 0.016618. GROWTH OF THIN FILMS BASED ON Si–Ge–Sn SYSTEM BY CVD AND GAS PHASE MBE Silicon is the most desirable substrate for the growth of semiconductor materials. Virtually defect-free Si wafers are available at low cost, and the range of applications of any semiconductor grown on Si can be enhanced by integration with silicon-based microelectronics. The growth of Si1-xGex films on Si(1 0 0) has been the subject of intensive studies over the past two decades owing to their many important applications in high speed microelectronic devices.[4–7] A number of different methods have been used for the heteroepitaxial growth of Si1-xGex=Si, but the two most commonly employed techniques are molecular beam epitaxy utilizing solid Si and Ge, and ultrahigh vacuum chemical vapor deposition (UHV-CVD) or gas-source molecular beam epitaxy (GSMBE) utilizing SiH4 and GeH4 or Si2H6 and Ge2H6. There are two ultimate, but also diverse, objectives in the growth of Si1-xGex=Si. The first is the achievement of defectfree Si1-xGex layers, which may take the form of strained layer superlattices, while the second is the growth of self-assembled coherent Si1-xGex islands or quantum dots. It has been known for many years—on theoretical grounds—that the Si–Ge–Sn alloy system should have very interesting properties, especially as high efficiency IR devices. This has stimulated intense experimental efforts to grow such compounds, but for many years the resulting material quality has been incompatible with device applications. Recently, we have achieved growth of device quality Sn–Ge and Si–Ge–Sn films using novel CVD methods. This is an important development for several reasons. First, SnxGe1-x alloys have been predicted to undergo a transition from indirect to direct gap semiconductors so that this material may lead to the first direct-gap semiconductor fully integrated with Si technology.[7–9] Second, device-quality SnxGe1-x layers of arbitrary thickness can be deposited directly on Si and these can be used as ‘‘virtual substrates’’ for the growth of Ge1-x-ySixSny ternary analogs. The fabrication of Ge1-x-ySixSny makes it possible to decouple strain and band gap engineering to achieve unique systems cover a wide range of operating wavelengths in the IR and new device structures that lead to novel photonic devices based entirely on group IV materials.[10] In the following sections, we present the fabrication and discuss properties of device quality, strain-free SnxGe1-x films, as well as strained engineered

Thin Film Science and Technology

Ge1-x-ySixSny layers grown over time directly on Si wafers. We also describe synthesis of films and nanometer-scale islands of Si1-xGex grown on Si(1 0 0) substrates via a unique single-source molecular precursor method. This new approach allows precise control of concentration and structure at the atomic level and it is particularly useful for development of compositionally homogeneous and uniform assemblies of nanoscale structures.

Growth of Si–Ge–Sn on Si for Strain-Balanced Heterostructures Via CVD The binary SnxGe1-x alloys are grown by a specially developed CVD method.[11–17] The combination of SnD4 with high-purity H2 (15–20% by volume) remains remarkably stable at 22 C for extended time periods. This formulation provides the simplest possible CVD source of Sn atoms for the growth of novel Sn–Ge systems. Depositions were conducted in a custom-built ultra-high-vacuum CVD reactor on Si(0 0 1) wafers. Growth temperatures between 250 and 350 produced thick films (50–500 nm) with Sn concentrations up to 20%, as measured by Rutherford backscattering. Fig. 10 shows a comparison between random and aligned RBS spectra for a Sn0.02Ge0.98 sample and a Sn0.12Ge0.88 sample. The ratio wmin between the aligned and random peak heights is 4% in the x ¼ 0.02 sample and about 30% in the x ¼ 0.12 sample for both Ge and Sn. This provides proof that Sn occupies substitutional sites in the average diamond structure. The wmin ¼ 4% value closely approaches the practical limit of about 3% for structurally perfect Si, which is unprecedented for a binary crystal grown directly on Si. The microstructural properties of the films were investigated by XTEM. Electron micrographs demonstrating nearly defect-free growth of Sn0.06Ge0.94 are shown in Fig. 11. The images show that the predominant defects accommodating the large misfit between the alloys and the Si substrate are Lomer edge dislocations at the interface with no dislocation cores propagating to the film surface. These are parallel to the interface plane and do not degrade the film quality. The surfaces of the films are very smooth, continuous, and atomically flat. Electron diffraction and high-resolution X-ray studies (including rocking curves and reciprocal space maps) show a monotonically increasing average lattice constant as a function of the Sn-concentration, with no evidence for a significant tetragonal distortion or strain. The full width at half maximum of the X-ray peaks range from 0.25 to 0.50 indicating tightly aligned crystal mosaics. Reciprocal space maps of the (0 0 4) reflection show that there is no epilayer tilt between the Si and the GeSn (0 0 4) Bragg planes. Comparisons of the (2 2 4) grazing incidence and

Thin Film Science and Technology

3069

and 0.8 nm, respectively. The presence of a thin oxide layer on the film surface reduces the amplitude of the reflectivity at larger sample angles. The ternary Ge1-x-ySnxSiy are grown on Si (1 0 0) via Ge1-xSnx alloy buffer layers.[18,19] The depositions are carried out via ultra-high-vacuum chemical vapor deposition of unimolecular hydrides with direct Si–Ge bonds [SiH3GeH3, Si(GeH3)4, (GeH3)2SiH2] with SnD4. The crystal structure, elemental distribution, and morphological and bonding properties of the Ge1-x-ySnxSiy= Ge1-xSnx heterostructures are characterized by highresolution TEM, including electron energy loss nanospectroscopy, high resolution X-ray diffraction, AFM and Raman. These techniques demonstrate growth of perfectly epitaxial, uniform, and highly aligned layers with atomically smooth surfaces and monocrystalline structures. The Raman spectral shifts are consistent with lattice expansion produced by the Sn incorporation into Si–Ge tetrahedral sites. The films possess a variable and controllable range of compositions (10– 20% Si, 60–85% Ge, and 1–15% Sn), and exhibit lattice constants above and below that of bulk. A material with a tunable lattice constant above and below that of Ge is synthesized directly onto Si substrates. Representative RBS spectra and high-resolution TEM micrographs are shown in Fig. 13. The data are obtained from a sample with concentrations Si0.20Ge0.72Sn0.08 grown on Si(1 0 0) via a Ge0.98Sn0.02

Fig. 10 (A) RBS aligned and random spectra of Ge0.98Sn0.02 with near perfect crystallinity. Channeling of Sn and Ge approaches the theoretical limit of pure Si. (B) Aligned and random spectra for Ge0.88Sn0.12 show the same wmin for both Ge and Sn indicating that the entire Sn content is substitutional. (From Ref.[11].)

grazing exit o=2y scans, show that the GeSn layers are 100 relaxed and this is confirmed with reciprocal space maps of the (0 0 4) peak and the off axis and (2 2 4) reflection. Specular X-ray reflectivity scans (reflectivity vs. sample angle o) were obtained to characterize the density and layer thickness and determine the average roughness of the substrate film interface. The data for samples with 5% Sn content are shown by the curve in Fig. 12. Note that the reflectivity drops off steeply at 1100 sec, yielding the density of the Ge–Sn layer to be 5.32 g=cm3,which is close to Ge. The fringes indicate that the thickness of the Ge–Sn layer is 57.8 nm, which is virtually identical to the RBS value of 58 nm. The fringes are reduced in amplitude at larger angles and disappear at about 6000 sec due to minor surface roughness. The best fit shows a roughness thickness for the substrate Si and the Ge–Sn layer to be 1.2 nm

Fig. 11 Cross-sectional electron micrographs of Ge0.94 Sn0.06. Top panel shows atomically flat film surface morphology, middle panel shows the exceptional uniformity of the film thickness. Bottom panel is a high-resolution electron micrograph of the interface region showing virtually perfect epitaxial growth. Arrows indicate the location of misfit dislocations. (From Ref.[11].) (View this art in color at www. dekker.com.)

T

3070

Thin Film Science and Technology

Fig. 12 High-resolution X-ray reflectivity of a Ge0.95Sn0.05 film grown on Si(1 0 0). The reflectivity data give a layer thickness of 58–60 nm and density close to that of Ge (2% larger). (Courtesy of Dr. Stefan Zollner.) (View this art in color at www. dekker.com.)

buffer layer at 320
C. The high degree of RBS He ion channeling indicates aligned, single-phase material in which the constituent elements in the heterostructure occupy random substitutional sites in the same average diamond cubic lattice. Note that the extent of channeling is identical for Si, Ge, and Sn indicating substitutionality of the elements in both the Ge–Sn buffer layer and the Si–Ge–Sn film of the sample. Fig. 13 shows a dark field image of the entire Si0.20Ge0.72Sn0.08=Ge0.98Sn0.02=Si heterostructure indicating highly uniform layers free from threading defects. High-resolution micrographs of the Si0.20Ge0.72Sn0.08=Ge0.98Sn0.02=Si(1 0 0) interfaces show highly commensurate microstructures. The Si0.20Ge0.72Sn0.08 and Ge0.98Sn0.02 layers are nearly

lattice matched and their interface is virtually defect-free as shown Fig. 13D. The large lattice mismatch between Ge0.98Sn0.02 and Si(1 0 0) is accommodated by periodic edge dislocations located at the interface as shown in Fig. 13C. The strain properties of these materials have been investigated by high resolution XRD. Studies involving samples that span a wide range of concentrations show fully strained (tensile and compressively strained), as well as relaxed Si–Ge–Sn films are obtained on strain-free Ge–Sn buffer layers. These results show that strain engineering can be achieved in Si–Ge–Sn heterostructures and multilayers by tuning the lattice parameter of the Ge–Sn buffer layer.

Fig. 13 RBS aligned and random spectra of Si0.20Sn0.08Ge0.72 epilayer and Sn0.02Ge0.98 buffer layer showing a highly aligned heterostructure. Inset: (A) magnified view of the Si peak indicating complete substitutionality of Si in the Sn–Ge lattice; (B) XTEM of the entire heterostructure; (C) Si=Sn0.02Ge0.98 interface (indicated by arrow); and (D) Si0.20 Sn0.08Ge0.72=Sn0.02Ge0.98 interface. (From Ref.[19].)

Thin Film Science and Technology

3071

A host of novel strained engineered optical and electronic devices have been designed based on this concept and are currently being fabricated and tested. It is interesting to note that the strain is reliably robust up to at least 400–500
C (400
C is the growth temperature of the films).

T

Growth of SiGe and Si4Ge Epitaxial Materials by Gas Phase MBE of H3SiGeH3 and Ge(SiH3)4 Growth of Si1-xGex films and nanometer scale islands has been demonstrated by using single source molecular hydrides containing direct Si–Ge bonds.[20] The growth process occurs via single source GSMBE directly on Si(1 0 0) and the concentration x of the film is predetermined by tailoring the composition of the molecular precursor. The technique was demonstrated by growing Si0.5Ge0.5 and Si0.8Ge0.2 epitaxial films and coherent islands on Si(1 0 0) via thermal dehydrogenation of H3SiGeH3 and Ge(SiH3)4, respectively, between 475 and 700
C. Note that the entire content of Si and Ge of the precursors is incorporated in the film. The major advantages of using single molecular sources with preformed Si–Ge bonds for the MBE growth of Si1-xGex lie not only in the a priori control over the composition, but also the film morphology can be controlled by the adjustment of a single kinetic parameter, i.e., the flux rate of the precursor, at a given temperature. The growth of the Si0.5Ge0.5 and Si0.8Ge0.2 films on Si(1 0 0) substrates is conducted in the sample chamber of a low-energy electron microscope (LEEM) where in situ observation of the GSMBE growth process takes place in real time. Fig. 14 shows a sequence of framecaptured LEEM video images over a field of view of 8 mm of Si0.5Ge0.5 growth on Si(100)-(2x1) at 550
C using H3Si-GeH3. The images were taken with the (1=2 , 0) diffraction beam such that the (2  1) and (1  2) terraces separated by single-height atomic steps alternate in contrast from dark to bright due to the rotation of the dimer reconstruction across a step. The last LEEM frame taken at 46 min showing island formation was a bright-field image acquired using the (0, 0) beam. In Fig. 14, contrast reversal in the (2  1) and (1  2) terraces was observed during the time lapse from 0 to 190 sec as growth of Si0.5Ge0.5 proceeded. Such contrast reversal is indicative of layer-by-layer growth. The growth of a full Si0.5Ge0.5 monolayer (ML) is completed every 30 sec. The layerby-layer growth continued after the completion of 6 ML, after which the LEEM contrast of the surface became very diffuse. The loss of contrast was due to new layers growing on top of incomplete layers, and this kind of growth results in a rough surface even though the domains are two-dimensional.

Fig. 14 Frame-captured LEEM video images showing Stranski–Krastanov growth of Si0.5Ge0.5. The elapsed time during growth is indicated under each frame. Approximate Si–Ge coverage for each frame: 0 ML at 0 sec; 1 ML at 40 sec; 2 ML at 70 sec; 3 ML at 100 sec; 4 ML at 130 sec; 5 ML at 160 sec; 6 ML at 190 sec; and 3D islands at 46 min. Field of view is 8 mm. (From Ref.[20].)

Three-dimensional (3D) islands began to appear after about 18 min of growth. The last frame shown in Fig. 14 is a bright-field LEEM image taken after 46 min of growth with the 3D islands appearing as bright spots. The LEEM results indicate that the growth mode of the Si0.5Ge0.5 is Stranski–Krastanov. AFM images show that the Si0.5Ge0.5 islands are primarily large dome-shaped structures. The dome shape of the islands with (1 1 3) and (15 3 23) facets is clearly demonstrated in the XTEM images, which also show that the islands are completely coherent. Similar islands, grown at 600 and at 700
C are also found to be completely coherent demonstrating the fact that the single source approach is capable of producing large coherently strained dome-shaped islands at a wide range of conditions. Growth of Si0.8Ge0.2 films on Si(1 0 0) was conducted using the Ge(SiH3)4 gaseous precursor. In situ real-time LEEM observations indicated that the Stranski–Krastanov growth mode was also in operation in this case. AFM and XTEM images of films

3072

grown at 500
C indicate growth of completely coherent islands with uniform size and spatial distribution. The images of films gown at 600 and 700
C show that large (200 and 400 nm) faceted dome-shaped islands are produced at both temperatures. These are both strain-free and dislocated. Selected area electron diffraction patterns show a lattice constant of 0.548  0.001 nm, very close to the ideal 0.5476 nm for Si0.8Ge0.2. High spatial resolution electron energy loss spectroscopy (EELS) used to verify that the composition of the Si1-xGex films indeed reflects the stoichiometry of the unimolecular precursor used for growth. Typical EELS composition line-scans across the interface from the substrate to the island using a electron beam with 1 nm diameter are given in Fig. 15. The EELS composition profile of a Si0.5Ge0.5 island grown at 700
C in Fig. 15 shows an almost constant 50% Ge within the island. Similarly, the EELS composition profile of a Si0.8Ge0.2 island shows a nearly constant 20% Ge. No segregation of Ge is observed in any of our EELS composition profiles of the islands. The single-source MBE method growth of Si1-xGex films permit control of composition x at the atomic level via the design of the single-source gaseous precursor containing precise atomic arrangements with direct Si–Ge bonds. Uniform composition reflecting the stoichiometry of the precursor is observed in all cases without any segregation of either Ge or Si. The growth of Si1-xGex films proceeds via the Stranski–Krastanov growth mode. Morphological control of the size and shape of the islands is achieved by simple adjustments of the flux rate of the precursor and the growth temperature.

Fig. 15 Ge composition line scans determined by high-resolution EELS on (A) a Si0.5Ge0.5 island grown at 700
C, and (B) a Si0.8Ge0.2 island grown at 500
C. The profiles shown below each XTEM image correspond to the line across the substrate=island interface. Scan direction is from the substrate to the vacuum. (From Ref.[20].)

Thin Film Science and Technology

CONCLUSIONS Thin film science and technology is a dynamic field. Illustrated by growth of thin films for silicon technology for optical and high speed microelectronics devices. Accepted technology is readily available. The key is control of the composition and structure of thin films and islands fully integrated with Si technology. The Ge–Si–Sn system makes it possible to decouple strain and ban-gap engineering to each achieve unique photonic devices. In this work, the focus has been based entirely on group IV materials. This excitement can also be realized in photonic and high speed microelectronics. Growth methods and analysis are also readily available. ACKNOWLEDGMENT The authors acknowledge support from NSF in their investigations. REFERENCES 1. Tu, K.-N.; Mayer, J.W.; Feldman, L.-C. Electronic Thin Film Science; Macmillan Publishing Company: New York, 1992. 2. Feldman, L.C.; Mayer, J.W. Fundamental of Surface and Thin Film Analysis; PTR Prentice Hall: New Jersey, 1986. 3. Nastasi, M.; Mayer, J.W.; Hirvonen, J. Ion–Solid Interactions: Fundamentals and Applications; Cambridge University Press: UK, 1996. 4. Patton, G.L.; Harame, D.L.; Strock, J.-M.; Meyerson, B.S.; Scilla, G.-S. Graded-SiGe-base, poly emitter heterojunction bipolar transistors. IEEE Electron Device Lett. 1989, 10, 534–536. 5. Mooney, P.M.; Chu, J.O. SiGe Technology: Heteroepitaxy and high-speed microelectronics. Ann. Rev. Mater. Sci. 2000, 30, 335–362. 6. Tromp, R.M.; Ross, F.M. Advances in situ electron microscopy: Growth of SiGe on Si. Ann. Rev. Mater. Sci. 2000, 30, 431–449. 7. Jenkins, D.W.; Dow, J.D. Electronic properties of metastable GexSn1-x alloys. Phys. Rev. B. 1987, 36, 7994. 8. Mader, K.-A.; Baldereschi, A.; von Kanel, H. Band structure and instability of GexSn1-x alloys. Solid State Commun. 1989, 69, 1123. 9. Soref, R.-A.; Friedman, L. Direct-gap Ge=GeSn= Si and GeSn=Ge=Si heterostructures, Superlattice Microstruct 1993, 14, 189. 10. Soref, R.A.; Menendez, J.; Kouvetakis, J. Strained Engineered Direct-gap Ge=SnxGe1-x Heterodiode and Mult-quantum Well Photodetectors, Lasers, Emitters, and Modulators Grown

Thin Film Science and Technology

11.

12.

13.

14.

15.

on SnySixGe1-y-z Buffered Silicon. US Patent 6,897,471, 2005. Bauer, M.R.; Taraci, J.; Tolle, J.; Chizmeshya, A.V.G.; Zollner, S.; Smith, D.-J.; Menendez, J.; Hu, C.W.; Kouvetakis, J. Ge-Sn semiconductors for band-gap and lattice engineering. Appl. Phys. Lett. 2002, 81, 2992–2994. Taraci, J.; Zollner, S.; McCartney, M.-R.; Mene´ndez, J.; Santana, M.-A.; Smith, D.-J.; Haaland, A.; Tutukin, A.V.; Gundersen, G.; Wolf, G.; Kouvetakis, J. Synthesis of siliconbased infrared semiconductors in the Ge-Sn system using molecular chemistry methods. J. Am. Chem. Soc. 2001, 123 (44), 10980–10987. Chizmeshya, A.V.G.; Bauer, M.; Kouvetakis, J. Experimental and theoretical study of deviations from Vegards Law in the Ge1-xSnx system. Chem. Mater. 2003, 15, 2511–2519. Li, S.F.; Bauer, M.R.; Mene´ndez, J.; Kouvetakis, J. Scaling law for the compositional dependence of Raman frequencies in GeSi and SnGe alloys. Appl. Phys. Lett. 2004, 84, 867–869. Bauer, M.R.; Chizmeshya, A.V.G.; Menendez, J.; Kouvetakis, J. Tunable band structure in

3073

16.

17.

18.

19.

20.

diamond cubic tin germanium alloys grown on Si. Solid State Commun. 2003, 127, 355–359. Cook, C.S.; Zollner, S.; Kouvetakis, J.; Tolle, J.; Menendez, J. Optical constants and interband transitions of Ge1-xSnx alloys (x < 0.2) grown on Si, Thin Solid Films 2003, 455–456, 217–221. Bauer, M.R.; Cook, C.S.; Zollner, S.; Crozier, P.; Chizmeshya, A.V.G.; Kouvetakis, J. GeSn superstructured materials for Si-based optoelectronics. Appl. Phys. Lett. 2003, 83, 3489–3491. Bauer, M.R.; Ritter, C.; Crozier, P.; Menendez, J.; Ren, J.; Kouvetakis, J. Synthesis of ternary Si-GeSn semiconductors on Si(1 0 0) via SnxGe1-x buffer layers. Appl. Phys. Lett. 2003, 83 (9), 2163–2165. Aella, P.; Cook, C.; Tolle, J.; Zollner, S.; Chizmeshya, A.; Kouvetakis, J. Structural and optical properties of SnxSiyGe1-x-y alloys. Appl. Phys. Lett. 2004, 84, 888–890. Hu, C.W.; Taraci, J.L.; Tolle, J.; Bauer, M.R.; Crozier, P.A.; Tsong, I.S.T.; Kouvetakis, J. Synthesis of highly coherent SiGe and Si4Ge nanostructures by single-source molecular beam epitaxy of H3SiGeH3 and Ge(SiH3)4. Chem. Mater. 2003, 15 (19), 3569–3572.

T

Thin Liquid Film Deposition T Myung S. Jhon Department of Chemical Engineering and Data Storage Systems Center, Carnegie Mellon University, Pittsburgh, Pennsylvania, U.S.A.

Thomas E. Karis Hitachi Global Storage Technologies, San Jose Research Center, San Jose, California, U.S.A.

INTRODUCTION Nano-films play a vital role in all aspects of human existence. This encompasses the spectrum extending from biological systems to small and large-scale electromechanical devices. Cell membranes encapsulate cells, while a film of tears lubricates eye movement. Grease lubricates the automobile wheel bearings by gradually releasing a thin film of oil into the ball-race interface. Due to the pervasiveness of nano-films in our life, many aspects of their physics and chemistry are being revealed through increasingly sophisticated methods of scientific investigation. This entry focuses on the state-of-the-art in two related aspects of liquid nano-film physics, spreading and dewetting. The solid=thin liquid film system must be designed to either dewet and evaporate, as with water on automotive car finish, or to wet and spread as in well-designed lubrication systems. For example, when molecularly thin liquid films are employed to lubricate gears in a micro-electromechanical system, or in the head (slider) disk interface of a hard disk drive, certain properties are essential for them to function without an oil change for many years of continuous operation. These devices cannot be flooded with lubricant, or the small parts stick together. Lubricant must flow to where it is needed in asperity contacts by spreading, while at the same time the viscosity must be high enough to provide boundary lubrication and to avoid evaporation and spattering by dewetting or spin-off from the rapidly moving parts. This is especially true for open systems such as a hard disk drive. The lubricant spreads and the vapor pressure is suppressed by energetically favorable interaction with the surfaces, so there is little or even no evaporation at elevated temperature. Here, we employ examples primarily from the magnetic recording industry to show how the physics of nano-films is derived from the properties of bulk liquids through the thermodynamics of interaction with the solid surface. The nano-film spreading rate is derived from the film thickness dependence of the surface energy. The nano-film viscosity is related to the nano-film vapor pressure, both of which increase Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120008083 Copyright # 2006 by Taylor & Francis. All rights reserved.

due to dispersion force in the limit of molecularly thin film. Vapor pressure and evaporation are related to dewetting and capillary wave roughness through the surface energy for a typical perfluoropolyether (PFPE) lubricant Zdol which has a polar hydroxyl group on each end of the chain. Finally, the thermodynamic model is tied together by a description of the contemporary molecular dynamics (MD) and Monte Carlo (MC) simulation which graphically elucidates images of the chain conformation.

BACKGROUND AND MATERIALS A fundamental understanding of the spreading behavior of the liquid film is required for the proper design of a number of engineering systems. Micro=nanoscopic spreading behavior is quite different from macroscopic behavior due to the difference in the driving forces between these two cases. Forces, such as surface tension gradient and gravity, drive spreading in macroscopic films.[1] However, a disjoining pressure gradient is the driving force for spreading in thin liquid films that have a thickness on the order of nanometer.[2] Although the spreading of liquid films on solid surfaces in the macroscopic regime has been studied extensively, the spreading phenomenon in thin films remains to be clearly understood and is the subject of this entry. Nanoscale confined liquids are an important subject due to their ubiquitous nature and future industrial applications. The functionalities of polymer chain and solid surfaces, for example, are key control factors in determining the material design of lubricant used in nanotribology. Materials having constituents with dimensions on the nanometer scale behave remarkably different than when in bulk state, which has led to a new paradigm we now refer to as nanotechnology. Due to broad technological interests, numerous studies on nanoscale confined liquids have been investigated, both theoretically and experimentally, by scientists and engineers from a variety of backgrounds, including data storage, semiconductor devices, catalysis, polymer engineering and science, tribology, robotics, and medicine.[3] 3075

3076

Thin Liquid Film Deposition

The spreading behavior of small drops of polydimethylsiloxane (PDMS) on silica surfaces has been studied by many investigators including Cazabat et al.[4] Novotny[5] investigated the spreading of polyperfluoropropylene oxides on silica surfaces using scanning microellipsometry (SME) and scanning photoemission spectroscopy. The results revealed that the surface diffusion coefficient increases as the film thickness decreases down to 1 nm and remained constant below that thickness. The spreading characteristics of PFPEs have been investigated using SME equipped with temperature and humidity controls. They studied the effects of initial film thickness, end-group functionality, and molecular weight on silica surfaces[6,7] and on amorphous carbon surfaces.[8,9] Molecularly thin PFPE films are mainly referred to in this entry which examines and illustrates the essence of the thin liquid film, especially the role of moleculesurface coupling. In particular, two examples of these materials are the commercially available FomblinÕ PFPE Z and Zdol random copolymers with the linear backbone chain structure X

½ðOCF2 CF2 Þp

ðOCF2 Þq


O

X ðp=q ffi 2=3Þ

where X (the ends-group) is CF3 in PFPE Z, CF2CH2OH in Zdol (Fig. 1A). Notice that Zdol has a hydroxyl group at the end of the chain, which exhibits moderate interactions with solid surfaces.

EXPERIMENTATION AND MESOSCOPIC ANALYSIS In this section, experiments on spreading properties using SME for various PFPE=solid surface pairs are discussed. The rheological characterization of PFPEs is also given. Further interrelationships among SME spreading profiles, rheology, surface energy or disjoining pressure, and dewetting, as well as a qualitative interpretation based on mesoscopic thermodynamics and transport phenomena, are provided. Scanning Microellipsometry To demonstrate the essence of the spreading phenomena, we invoke the ‘‘thought experiments’’ for different lubricant=surface and lubricant=lubricant interactions, which are shown in Fig. 1B. In the experiments of O’Connor et al.,[7] monodisperse PFPE Z and Zdol, which were fractionated via supercritical fluid extraction in CO2, were dip-coated onto the surface of silicon wafers as shown in Fig. 2A. Film thickness was controlled by altering the PFPE concentration and draw rate. An SME apparatus (Fig. 2B)

Fig. 1 (A) The molecular structure of PFPEs and its simplified model and (B) The spreading profiles as time progresses (t ¼ 0 to t1) from ‘‘thought experiment’’: (i) Z and (ii) Zdol. The coordinate system is represented in B(i). (View this art in color at www.dekker.com.)

was used to measure the thickness of the film as it spread with time. The coated wafer was placed on a pedestallike plate housed in an environmental chamber with slits for passage of the incident and reflected beams. The chamber was mounted on a stage, which translated the sample across the beam area, and the thickness profiles were measured in the controlled temperature and humidity environment. The spreading profile obtained from SME strongly depended on the PFPE molecule–surface interactions, and provided a fingerprint for each pairing. Typical SME thickness profiles for monodisperse Z and Zdol, as examined by O’Connor et al.,[6,7] are shown in Figs. 3A and B. As the film spreads with time, the spreading front travels along the surface of the silicon wafers. The sharp ‘‘spike’’ for the PFPE Z profile in Figs. 3A was observed to decay over time. The ‘‘spike’’ at the step did not affect the spreading rate, and could be avoided by eliminating free surface vibrations and decreasing the solvent evaporation rate by dip-coating with the solution in the bottom of a tall and thin container. The spreading of PFPE Zdol exhibits a characteristic layering structure or shoulder with a height on the order of the radius of gyration for the PFPE molecules and the separation of PFPE molecules from the initial film layer at a sharp boundary. Thicker films of Zdol appear

Thin Liquid Film Deposition

3077

T

Fig. 2 (A) Partially dip-coating a disk in a PFPE solution and (B) schematic of the SME apparatus. The PFPE film is shaded. (View this art in color at www.dekker.com.)

to experience partial ‘‘dewetting’’ as indicated by the rough appearance in the SME scan (Fig. 3B). Dewetting phenomena (investigated for mogul dynamics[10] in magnetic recording disk lubrication) are important for the nanoscale, multiphase system design and will be discussed in the section entitled Capillary Waves and Dewetting. Similar PFPE spreading experiments on carbon coated magnetic recording disks were performed by Ma et al.,[8,9] a few years after the observations made by O’Connor et al. In their measurement, however, mass build-up for PFPE Z was not observed, and dewetting characteristics for Zdol were somewhat suppressed. These phenomena clearly demonstrated the relevance of surface energy driven flow effects in nanoscale spreading processes, and will play a critical role in design criteria for nanodevices and=or sensors. Ma et al.[11] examined spreading characteristics for various carbon surfaces (a-C=hydrogenated=nitrogenated), illustrating the fingerprint for the various surface=PFPE molecular coupling.

Mesoscopic Thermodynamic and Transport Models Disjoining pressure The remarkable similarity of the PDMS and PFPE spreading profiles suggests that interactions between

Fig. 3 SME spreading profiles for PFPE on silicon wafers: (A) Z (Mw ¼ 13,800 g=mol) and (B) Zdol (Mw ¼ 3100 g=mol). Both have the initial thickness of 5 nm for times of 20 min, 1.5, 3, 12, and 24 h at 26 C.

the polar entities of the liquid and the surface are at the root of the ‘‘anomalous’’ spreading in these liquids. In the case of PFPE terminated with functional end-groups, evidence for strong interactions between the end-groups and the polar active sites on the carbon surface has been reported. The free energy function provides a mesoscopic framework for understanding the terraced spreading phenomena. Since the surface energy is defined as the free energy per unit area, the total disjoining pressure ðPÞ for these fluids can be derived from the experimental surface energy data by: P ¼ 


@ d g þ gp @h

ð1Þ

where, gd and gp are the dispersive and polar components of the surface energy, respectively, and h is the film thickness. The regression fit to the surface energy data, shown as the smooth curves in Fig. 4A and B, were numerically differentiated to obtain the disjoining pressure. The total disjoining pressure, as well as the individual contributions from the dispersive and polar components, is shown in Fig. 4C. Notice that

3078

Thin Liquid Film Deposition

Fig. 4 The components of the surface energy measured on hydrogenated carbon overcoated thin film magnetic recording media: (A) The dispersive component of the surface energy for PFPE Z and Zdol; (B) the polar component of the surface energy for PFPE Zdol with molecular weight (MW) of 1100 (`) 1600 (), and 3100 (&) g=mol; and (C) the disjoining pressure as a function of film thickness for PFPE Zdol (MW is 3100 g=mol).

gd decreases monotonically with h. Below film thickness of approximately 0.5 nm, P at each molecular weight is dominated by gd , which increases rapidly with decreasing film thickness and is largely independent of molecular weight. The gp , however, oscillates with film thickness and becomes larger in magnitude than gd as h increases. Oscillations in gp provide an additional contribution to P for PFPE Zdol that produces alternating regions of stable and unstable film thickness. The sum of the two contributions gives rise to oscillations in the total disjoining pressure. The shoulder height in the Zdol spreading profile corresponds to the film thickness at which @P=@h < 0 and P changes sign from positive to negative. The shoulder heights for two of the Zdol fractions were close to those expected from analysis of the disjoining pressure isotherm.[12] Spreading characteristics from the mesoscopic transport model To quantify the spreading characteristics, we examined the thickness-dependent diffusion coefficient D(h), extracted from the SME via two methods. The first method utilizes the spreading data at constant height (isoheight), while the second method utilizes the entire

spreading profile. The first method does not assume that the spreading is a diffusive process, whereas the second method does. The length L, which the leading edge of the advancing lubricant front traveled after time t, is defined as the difference between the initial position of the leading edge and its position in each of the subsequent profiles. To quantify the spreading rate, L and the corresponding times t (obtained from the profiles in Fig. 3B) are plotted. As shown in Fig. 5A, the L–t plot (in log) can be fit with the two piecewise straight lines for the entire range of t, i.e., L / ta. We found that a ffi 1 at short times (Regime I), and after a gradual transition, a ffi 1=2 (Regime II; diffusion region). Therefore, the surface diffusion coefficient D is estimated from the data in Regime II (L / t1=2) alone via the relationship D ¼ L2=t. We adopted a second method[9] called the Boltzmann–Matano interface method [see Eq. (5)], which is more accurate for the purely diffusive process. Notice that Ma et al.[8,9] successfully employed this method, since the spreading profiles they measured exhibit little mass buildup at the front and negligible dewetting. The shape of spreading profiles can be derived from the diffusion coefficient D as follows: Since the flux vector j is proportional to the concentration (or related

Thin Liquid Film Deposition

3079

B 10-2

A

T

10-2 Experimental data 10-3

L (m)

L (m)

10-3 Regime I Slope =1

10-4

Regime II

τ=0 10-4 τ = 102 s

Slope =1/2

10-5 102

103

104

105

10-5 2 10

τ = 103 s

103

104

105

t (s)

t (s)

Fig. 5 (A) Travel length (L) as a function of time (t) obtained from Fig. 3B for monodisperse Zdol; T ¼ 26 C, under dry nitrogen and (B) L–t plot results using the modified diffusion equation (Eq. 7) for several values of t. (View this art in color at www.dekker.com.)

to film thickness, h) gradient Hh we obtain, jðr; tÞ ¼ DðhÞHhðr; tÞ

ð2Þ

Further, the mass conservation equation gives:

(or continuity)

@hðr; tÞ þ H  jðr; tÞ ¼ 0 @t

@

jðr; t þ tÞ ¼ et @t jðr; tÞ ¼ DðhÞHhðr; tÞ ð3Þ

By combining Eqs. (2) and (3), we obtain @hðx; tÞ ¼ H  ½DðhÞHhðx; tÞ
@t

below. (The more rigorous molecular simulation is described later.) By introducing a time lag denoted as t (which may be related to the relaxation time of the PFPEs), the Fick’s constitutive equation can be modified as:[14]

ð4Þ

where H is the ‘‘del’’ operator and h(x,t) is the film thickness. On the other hand, the diffusion coefficient can be evaluated from Eq. (4). This 1-D analysis is called the Boltzmann–Matano technique:[9]   Z h 1 dx DðhÞ ¼  x dz; 2t dz z¼h 0 Z h0 with the constraint x dz ¼ 0 ð5Þ 0

where h0 is the initial film thickness far from the step. Note that h and the coordinates x and z are shown in Fig. 1B(i). However, as shown from the L–t plot, the spreading cannot be described by the diffusive concept alone. Since the theories developed by Novotny[5] [Eq. (4)] and Mate[13] (hydrodynamic model) fail to describe the L / t behavior exhibited in Regime I, a simple mesoscopic model for the overall L–t behavior is described

ð6Þ

To derive Eq. (6), we imposed causality into the relationship between j and Hh inspired by the microscale heat transfer theory or Cattaneo equation.[15] By combining Eqs. (3) and (6), and assuming t@=@t to be small, we obtained the following modified diffusion equation: @

et @t

@h @ 2 hðr; tÞ @hðr; tÞ ffi t þ 2 @t @t @t ¼ H  ½DðhÞHhðr; tÞ
;

ð7Þ

where t characterizes the crossover behavior between Regimes I and II. Notice that Eq. (7) reduces to Eq. (4) when t ¼ 0. We found that the transition between these two regimes depends on the value of t. Increasing t shifts the transition to a later time. By setting t ¼ 103 s in Eq. (7), we attained an excellent fit for the experimental L–t data (Fig. 5B). Relationship between diffusion coefficient and disjoining pressure As will be shown later, the measured disjoining pressure P can be used to qualitatively describe the spreading profile. Before that, the relationship between P and D is established. Note that P and D can also be calculated by MD simulation

3080

Thin Liquid Film Deposition

The relationship between position dependent viscosity Z(z), P, and D(h) is developed by generalizing the hydrodynamic model (Reynolds equation): D ð hÞ ¼

Z

h

0

  ðh  zÞ2 @P dz @h ZðzÞ

ð8Þ

h3 @P 3Z1 @h

ð9Þ

In deriving Eq. (9), we used the no-slip boundary condition at the lubricant=solid boundary. If we were to generalize the above analysis with the partial-slip boundary condition, e.g., @v=@z ¼ bv (b  slip parameter) instead of no-slip condition with ZðzÞ ¼ Z1 , Eq. (9) could be modified as: 1 DðhÞ ¼  Z1



h2 h3 þ b 3



@P @h

ð10Þ

We could use a simplified form of ZðzÞ ¼ Z1 fðzÞ. Here Z1 is the bulk viscosity, and fðzÞ can be experimentally determined. Here, z is the distance normal to the solid surface. A partial justification for the above functional form can be drawn from the temperature dependence of the surface diffusion coefficient and the bulk viscosity,[7] or the fly stiction correlation with the bulk viscosity. To develop a rigorous hydrodynamic model, we need better rheological data and more details are given in the following section on Rheology Measurement. An alternative approach to describe thin film spreading is the jump diffusion model developed by Karis and Tyndall.[16] In their analysis, the flow rate is calculated by integrating the velocity v throughout the depth of the film

j ¼

Z

h

v dz 0

ð11Þ

The velocity (v) is taken to be the drift velocity of the fluid particles in a disjoining pressure gradient, HP. The velocity is therefore proportional to the disjoining pressure gradient with the proportionality constant m representing the mobility [the units of m are (m=sec)=(Pa=m)], and v ¼ mHP

j ¼ mhHP ¼ mh or

or by substituting constant viscosity assumption, i.e., ZðzÞ ¼ Z1 in Eq. (8), we obtain Eq. (9) developed by Mate:[13]

DðhÞ ¼ 

The flow rate is calculated from the velocity by integrating through the film thickness according to Eq. (11)

ð12Þ

DðhÞ ¼ mh

@P Hh @h

@P ðin 1-D form) @h

ð13Þ ð13Þ0

In the integration of Eq. (11) to derive Eq. (13), the mobility was assumed to be independent of the distance from the surface between z ¼ 0 and z ¼ h, and therefore should be considered an ‘‘effective’’ mobility. In general, m is a function of end group chemisorption, and m could be rigorously calculated with the MD simulation. The differential equation that defines the 1–D spreading profile is obtained via substitution of Eq. (13)0 into the continuity Eq. (4).      @h @ @h @ @P @h ¼ DðhÞ mh ¼ @t @x @x @x @h @x

ð14Þ

Eq. (14) is solved numerically to calculate spreading profiles of nonpolar and polar PFPEs from the disjoining pressure measured from contact angles.[16] The calculated profiles, shown in Fig. 6A for Zdol, are in quantitative agreement with those measured by SME, Fig. 6B. The only adjustable parameter was mg1 =d0 , where g1 is the bulk surface tension of the lubricant, and d0 is a characteristic dimension of the PFPE chain. Here d0  0:4 nm, which is approximately the van der Waals diameter of the PFPE chain. For Zdol, g1  24 mN=m. A previously undescribed region for thin film spreading, which is particularly relevant to dewetting, is the region of film thickness h where P > 0 and @P=@h > 0, is predicted from this analysis. This region does not always show up in the spatially averaged spreading profiles measured by SME. It is referred to as the reverse flow region, where the free energy is lowered by decreasing the local film thickness. Thus, the flow in this region tends to decrease rather than increase the film thickness. A similar effect is observed in spinodal decomposition and is referred to as ‘‘up-hill’’ diffusion.[17]

Rheology Measurement The bulk rheological properties of the PFPEs, including the melt viscosity, storage modulus, and loss modulus at several different temperatures, have been widely reported via steady shear and dynamic oscillation tests.[18] In this entry, the focus is on the confined geometry effects on viscosity.

Thin Liquid Film Deposition

3081

diameter. At ambient pressure (100 kPa), DðpV Þvis  6:2 J=mol, near ambient conditions, DHvis  DEvis . Therefore, the viscosity is given by: Z ¼



Nhp Vl





ðDEvis  TDSvis Þ exp RT



ð15Þ

A regression fit to the bulk viscosity as a function of temperature,[19] provided DEvis ¼ 34.7 kJ=mol) and DSvis ¼ 9:87 J=mol-K. The flow-activation energy is close to that reported for bulk Zdol with a molecular weight of 3100 in Refs.[6,7]. A positive value for the flow activation entropy of bulk Zdol means that the entropy of the flow unit increases on going into the flow-activated state. Changes in the lubricant flow-activation energy and entropy near the solid surface cause changes in the viscosity with decreasing film thickness. The flowactivation energy near a solid surface is estimated from the thin film vaporization energy as follows. In an ideal gas, the chemical potential m (or partial molar Gibbs free energy) is given by: dm ¼ RTd ln P Fig. 6 (A) Calculated and (B) measured spreading profiles for Zdol. MW is 3100 g=mol. (View this art in color at www.dekker.com.)

Dispersive interaction has a dramatic effect on the molecular layers closest to the surface, and can be explained in terms of the rate theory for viscous flow.[19] Within the rate theory, a flow event comprises the transition of a flow unit from its normal or quiescent state, through a flow-activated state, to a region of lower free energy in an external stress field. For small molecules, the flow unit is the whole molecule, while for longer chains, the flow unit is a segment of the entire molecule. By analogy with chemical reaction rate theory, there is a flow-activation enthalpy, DHvis , and entropy, DSvis , for transition into the flowactivated state. A flow unit is approximated by a particle in a box, with the energy being partitioned among rotational and translational degrees of freedom, which govern the transition probability. On this basis, the viscosity Z ¼ ðNhp =Vl Þ expðDGvis =RTÞ, where N is Avogadro’s number, hp is the Planck constant, Vl is the molar volume of the liquid, and DGvis ¼ DHvis  TDSvis is the flow-activation free energy. Here, DHvis ¼ DEvis þ DðpV Þvis , where DEvis is the flow-activation energy and DðpV Þvis is the pressure–volume work. At constant pressure, DðpV Þ ¼ pDVvis . For PFPE Z, the flow-activation volume DVvis  0.1 nm3,[20] which is equivalent to a spherical region  0.6 nm in

ð16Þ

where P is the partial pressure of the lubricant in the vapor phase and R is the gas constant. The chemical potential per unit volume in the lubricant film m=Vl ¼ P. The ratio of the film surface vapor pressure, P 0 ðhÞ to the vapor pressure of the bulk lubricant, P 0 ðhÞ=P 0 ð1Þ; is derived by integrating Eq. (16):   mðhÞ  mð1Þ ¼ RTln P 0 ðhÞ=P 0 ð1Þ ð17Þ The reference state is taken to be the chemical potential and vapor pressure of the bulk lubricant: uð1Þ ¼ 0 and P 0 ð1Þ is the vapor pressure of the bulk liquid. The surface chemical potential is approximated by the unretarded atom-slab dispersive interaction energy: m ¼ 

Vl A 6ph3

ð18Þ

The dispersive interaction coefficient A is also referred to as the Hamaker constant, and A ¼ 1019 J for Zdol. The dispersive component of the vaporization energy near a solid surface is approximately given by Eq. (18). The vaporization energy is the energy required to remove a molecule from the liquid without leaving a hole behind. The free volume needed for a flow unit to make a transition into the flow-activated state is less than the size of the entire molecule. It is found that the ratio n  DEvap;1 =DEvis;1 > 3; where DEvap;1 and DEvis;1 are the vaporization- and

T

3082

Thin Liquid Film Deposition

flow-activation energy of the bulk liquid, respectively. Thus, the flow-activation energy near the surface is approximated by: DEvis ¼ DEvis;1  m=n

ð19Þ

For linear chains longer than 5 or 10 carbon atoms, n increases due to the onset of segmental flow. In practice, n is experimentally determined from the measured values of DEvap and DEvis . For PFPE Zdol 4000, DEvap;1 ¼ 166 kJ=mol, giving n  4:8, which is consistent with segmental flow. In order to calculate the thin film viscosity with Eq. (15), the flow-activation entropy near the surface is also needed. An experimental flow-activation entropy is calculated from the spin-off data[19] with Eqs. (15) and (19) as follows. The experimental Z vs. h is determined from the dh=dt during air shear induced flow on a rotating disk. Eq. (19) is then solved for DSvis versus h using Eq. (19) for DEvis . Below 2.3 nm, TDSvis  1:9 kJ=mol, which corresponds to the critical configurational entropy change for flow (R ln 2   5:76 J=mol-K).

been found to produce a fluctuating surface roughness, which is detected by X-ray reflectivity and diffuse scattering. For hexane, the whole molecule is the flow unit. Thus, for hexane, density fluctuations near the surface involve molecular rearrangement. In contrast, for PFPE, density fluctuations near the surface involve segmental rearrangements. Similar X-ray reflectivity and diffuse scattering measurements of PFPEs on carbon overcoats and silica have been reported by Toney, Mate, and Leach.[21] The surface smoothens as the lubricant thickness on carbon is increased, which probably reflects the lubricant filling in porosity of the overcoat.[22] Surface density fluctuations observed with fully bonded end-groups show that these fluctuations are segmental rather than cooperative motions of the entire chain. Surface roughness induced by surface density fluctuations as a function of film thickness is estimated for the PFPE Zdol 3100 on amorphous carbon for which the spreading profile is shown in Fig. 3. The rms-roughness is given by the capillary wave continuum theory[23] as:

srms Capillary Waves and Dewetting Liquids are permeated by thermal fluctuations at finite temperatures. In quiescent bulk liquids, thermal fluctuations give rise to density fluctuations and diffusion, and these enable fluids to spread and flow. Molecular drift and shear flow, as discussed above, are present when an energy gradient is superimposed on the thermal fluctuations. According to the rate theory of flow, the likelihood that a thermal fluctuation will lead to a displacement event is determined by the ratio of the free energy change of the event taking place to the thermal energy. The same analogy holds for vaporization. Thermal fluctuations are present at the liquid vapor interface, as well as in the bulk, and governed by the same thermodynamic principles as diffusion and flow. However, the present discussion focuses on molecularly structured liquids with at least 20 monomer units. Thermal fluctuations leading to cooperative motion in polymers pertain to segments consisting of several monomer units. A measure of the degree of segmental flow is provided by the ratio of the vaporization energy to the flow-activation energy, which is approximately four to five for Zdol 4000. If the entire molecule is the flow unit, then this ratio is closer to three. Therefore, with typical PFPE lubricant molecules, the size of the cooperative groups participating in thermal fluctuations is less than the entire chain. Thermal fluctuations at the free liquid surface of molecularly thin hexane films on solid surfaces have

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi   kB T 2p3 g1 h4 ln 1 þ ¼ 4pg1 a2 A

ð20Þ

where a is a molecular length scale and kB is the Boltzmann constant. The length scale a  0:75 nm is roughly estimated from 60% of the maximum stable film thickness in the thin film region of Fig. p 6.ffiffiThe peak ffi to valley roughness, denoted spp ¼ ð2 2Þsrms , is more relevant to magnetic recording slider disk spacing considerations. Fig. 7 shows the roughness due to thermal fluctuations at the free surface. The stable and unstable regions for PFPE Zdol with molecular weight of 3100 are superimposed in the roughness curve. The dashed region of the roughness curve denotes the unstable region, which is the step height in Fig. 6, and the region between the zero crossing and the first maximum of P in Fig. 4C. Separation into two stable regions of film thickness is not always observed when the mean film thickness is within the unstable region, even though the spreading film will not spontaneously flow to form a film with thickness in the unstable region. Films are routinely dip-coated and remain at the metastable thickness within the unstable region during measurement of the contact angle to determine the surface energy shown in Fig. 4A and B. No transition was observed in the viscosity during spin-off as the film thickness made a transition through the unstable region. As noted by Toney, Mate, and Leach,[21] thermal fluctuations of the free surface, at least at room temperature, do not appear to initiate the phase transition into two stable

Thin Liquid Film Deposition

3083

entry, the simple lattice-based, simple reactive sphere (SRS) MC techniques for examining the fundamentals of PFPE film structure are introduced first. An off lattice-based, bead-spring MC model is introduced later to capture the deailed internal structure of the PFPE molecules, and the molecular dynamics method is implemented for a full-scale nanostructural dynamic analysis of thin PFPE liquid films.

SRS Model

Fig. 7 Calculated peak to valley surface roughness due to thermal fluctuations at the liquid vapor interface of Zdol (MW ¼ 3100 g=mol) as a function of the average film thickness.

regions of film thickness (i.e., dewetting) from the metastable state. This is because the fluctuations represent only a small perturbation to the overall molecular configuration of the PFPE chain. For PFPE Zdol with a molecular weight of 3100, the number average degree of polymerization is approximately 31, and the contour length is 12 nm. By analogy with the number of atoms in a flow unit of linear alkanes, the fluctuation comprises two to four chain monomers, or 6–12% of the PFPE chain contour length. Much larger cooperative fluctuations, or external perturbations, are needed to nucleate the spinodal decomposition. However, thermal fluctuations in the lower stable region of film thickness may play a role in the critical flying height for the onset of bridge formation.[24]

SIMULATION The immediate goal of the simulation is to construct reliable tools that accurately describe the static and dynamic behavior of thin liquid films. As should be apparent from the preceding sections, it is extremely difficult to measure the flow-activation energy, disjoining pressure, and dewetting due to thermal fluctuations accurately or directly. The molecular simulation provides a complementary tool and is described in detail in this section. A multitude of references deal with the general methodology behind molecular simulations.[25] In this

We adopted a spin analogy=lattice gas model, or SRS model, as shown in Fig. 8A, which illustrates the oversimplified molecular structure, while still capturing the essence of the molecule=surface interactions describing SME profiles. Notice that similar techniques using the Ising model have been previously investigated to study other physical systems. MC simulations based on the SRS model were pioneered by Ma et al.[26] to explain the peculiar spreading profiles of PFPEs on amorphous carbon surfaces shown in Fig. 3 via the adoption of four different interactions: molecule=molecule, molecule=surface, end-group=end-group, and end-group=surface (a molecule is denoted as a backbone in the absence of polar end-groups). Molecules are approximated as reactive spheres, where a spin S ¼ 1 is assigned to an occupied lattice site (for a vacant site, S ¼ 0). A comparison between MC simulation results for molecules with non-polar (e.g., PFPE Z) and polar (e.g., PFPE Zdol) end-groups is illustrated in Fig. 1. A drastic difference is apparent between Fig. 1B(i) and B(ii): the PFPE Z profile is relatively smooth and spreads more rapidly, while the Zdol profile exhibits a complex layering structure. These results are qualitatively consistent with the experimental observations given in Fig. 3, which reveal that molecules with polar end-groups demonstrate a first layer that is one molecule thick (the thickness of the first layer is on the order of the PFPE diameter of gyration in the bulk state), with subsequent layers approximately twice the thickness of the first layer. This simulation qualitatively explained the experimental spreading data on hydrogenated and nitrogenated carbon surfaces[11] by adjusting the screening length (descriptive for end-group=endgroup coupling), which is related to the hydrogen and nitrogen content. Using the SRS model, the L–t plot was constructed by plotting the distance traveled from the initial sharp boundary versus the number of MC steps (the iso-height in the spreading profile). The simulated L–t response shows a distinct transition between short and long times, which agrees with the experimentally observed L–t behavior shown in Fig. 5. The long-term behavior exhibits L / t1=2, thus meeting the criteria for the surface diffusion assumption.[27]

T

3084

Thin Liquid Film Deposition

Fig. 8 Molecular model of PFPE Zdol: (A) SRS model with a discrete spin state; (B) bead-spring model, where gray spheres represent the backbone and black spheres represent the polar end-groups; and (C) SRS model with a continuous spin state (small end group sphere with the large backbone sphere). (View this art in color at www.dekker. com.)

The 3D results for PFPE Zdol, Fig. 9A provides more insight into the layer coupling. Due to the simplicity in this SRS model with lattice sites, an occupancy spin, and an end-group orientation spin (toward or away from the surface), we can qualitatively investigate the 3D steady state surface structure of PFPE films, describe its dewetting phenomenon and complex surface morphology, and examine its properties as a function of interaction parameters, surface loading, and initial surface structure. To illustrate the dewetting phenomenon as a function of end-group=end-group and end-group=surface interaction energies, simulations were conducted on a 3D lattice.[28] For an initial condition, we randomly distributed a layer with 90,

100, and 110% loading onto the surface to represent a very rough, highly damaged layer as seen for the 100% initial loading condition in Figs. 9B and C. If the end-group=end-group interactions are too strong, the layer does not relax, and the amount that a layer will relax is determined by the end-group interaction strength. The fractal dimension along with the simulated surface morphology described above can have a potential use for fingerprint analyses of PFPE molecule= surface pairs. To better examine the disjoining pressure driving force behind the nanoscale spreading process, it is more advantageous to adopt a continuous spin state SRS model for an off-lattice scenario. The end-groups in this case are

Fig. 9 (A) A typical simulation results for spreading 3-D dewetting simulation with (B) initial condition for 100% layer loading; and (C) relaxed film with a rough, dewetted surface. Black indicates a bare surface and white indicates peaks.

Thin Liquid Film Deposition

represented as smaller, internal spheres on the molecular backbone sphere surface, as shown in Fig. 8C.

Monte Carlo Simulation with Bead-Spring Model To represent the molecular structure with reasonable accuracy, as well as to reduce computational time, the coarse-grained, bead-spring model (Fig. 8B) was employed to model a PFPE molecule. This simplifies the detailed atomistic information while preserving the essence of the internal molecular structure. The off-lattice MC technique was used to examine nanoscale PFPE lubricant film structures and stability with internal degrees of freedom.[29] In this model, a PFPE molecule is composed of a finite number of beads with different physical or chemical properties. For simplicity, we assume that all the beads, including the end-beads, have the same radius. Lennard-Jones and van der Waals potentials were used for nonpolar bead–bead and bead–wall interactions, respectively. For polar interactions, exponential potential functions were added to both end-bead end-bead and end-bead wall interactions. For the bonding potential between adjacent beads in the chain, a finitely extensible nonlinear elastic model was used. For example, PFPE Zdol can be characterized differently from PFPE Z by assigning the end-bead a polarity originating from the hydroxyl group in the chain end. The steady-state nanoscopic properties were examined and the result agrees qualitatively with the simulation results obtained via the SRS model. Our results provide a direct interpretation of the experimental surface energy data for PFPE films with functional end-groups.[12] The most remarkable achievement of MC simulation is its ability to predict equilibrium properties, including the surface morphology visualization. For example, in our simulation, stable films did not experience dewetting or film rupture. However, a rougher surface morphology was observed for smaller molecular weights (Fig. 10B) and strong end-bead functionalities (Fig. 10C). To establish a qualitative relationship between the orientation and the layer structure, we examined the number of layers and the end-bead density. We found that the adsorption of functional end-beads results in an alternate ordering in the subsequent layers, i.e., upward orientation in the second layer, downward orientation in the third, etc. Our result provides a direct interpretation of experimental surface energy data for PFPE films with functional end-groups. The nondispersive component of surface energy exhibited an oscillatory pattern with increasing film thickness and was shown to be approximately proportional to end-group density, as demonstrated in our study.[29]

3085

Our simulation results also suggest that the density variation of the end-groups is related to the characteristic behavior of the surface energy of PFPE films. Not only we can interpret experimental data, but we can also predict the nanostructure which is unobtainable experimentally. For example, we found that the observed expansion of layer thickness is attributed not to the bond stretching, but to the temperature dependence of the intermolecular interaction and excluded volume effect. Molecular Dynamics Simulation So far, we have demonstrated that the MC simulation (lattice-based SRS model and off-lattice bead-spring model) results are in qualitative agreement with the experimental results. A complementary approach is MD simulation using the bead-spring model. Our preliminary MD simulations provided similar results as the MC method for the calculation of the static properties of confined PFPE nanofilms, especially the radius of gyration and end-bead density profiles. The anisotropic molecular conformation and experimental layering structures in the film were also verified. MD simulations[30] provide a powerful tool for examining the dynamics of nanofilms through correlation functions by tracking the trajectories of molecules, including the space and velocity coordinates. MD simulations, therefore, are suitable for

Fig. 10 Morphology of PFPE films from MC simulations: (A) schematic of simulated surface roughness for; (B) molecular weight dependence [Np ¼ 10 (upper), 15 (middle), 20 (bottom)]; and (C) end-bead functionality dependence [epw ¼ ep ¼1 (upper) and epw ¼ ep ¼ 4 (bottom)]. Here, Np is the number of monomers and epw and ep are polar energy parameters.

T

3086

calculating the transport coefficients from the correlation functions. In our preliminary study, self-diffusion coefficients of PFPE nanofilms were investigated with the Einstein relationship or Green–Kubo formula, to examine the effects of molecular weight, end-group functionality, external surface interaction strength, polydispersity, and temperature. Several autocorrelation functions were examined to study molecular motions. The Kohlrausch–Williams–Watts function and wavelet analysis were employed to interpret the molecular structure of PFPE nanofilms. Insight into physics of the disjoining pressure was revealed by calculating the internal pressure tensor during the simulation of nanofilms.

CONCLUSIONS We reviewed fundamental scientific tools as well as potential applications relevant to the thin liquid film technology. We focused on understanding the behavior of molecularly-thin lubricant films relevant to the emerging field of nanotechnology, especially for achieving durability and reliability in nanoscale devices. The topics on the experimentation and theory for the physicochemical properties of ultra-thin PFPE films were reviewed by examining liquid film thickness from submonolayer to multilayer. By systematically tuning the end-group strength for PFPE, we can examine and control physicochemical properties for thin liquid films in various PFPE=solid surface combinations. Methods for extracting spreading properties from the SME and rheological properties (we only discussed surface-enhanced viscosity) of PFPEs are shown. The interrelationships among SME spreading profiles, surface energy or disjoining pressure, rheology, and tribology (dewetting and density= thermal fluctuations) are discussed from the viewpoint of thermodynamics. Mesoscopic theories, including microscale mass transfer, are introduced to interpret qualitatively thin PFPE film dynamics. Simulation tools, including a lattice-based SRS model, the offlattice bead-spring MC, and MD methods were reviewed. These tools accurately describe the static and dynamic behaviors of liquid nano-films. The simulation results are consistent with experimental findings and are thus suitable for describing nanoscale molecular mechanisms in thin film fluid dynamics.

REFERENCES 1. Teletzke, G.F.; Davis, H.T.; Scriven, L.E. How liquids spread on solids. Chem. Eng. Commun. 1987, 55, 41–82.

Thin Liquid Film Deposition

2. Mate, C.M.; Marchon, B. Shear response of molecularly thin liquid films to an applied air stress. Phys. Rev. Lett. 2000, 85, 3902–3905. 3. Roco, M. C.; Bainbridge, W. S., Eds. Converging Technologies for Improving Human Performance; Kluwer Academic Publisher: Norwell, MA, 2003. 4. Cazabat, A.M.; Fraysse, N.; Heslot, F.; Carles, P. Spreading at the microscopic scale. J. Phys. Chem. 1990, 94, 7581–7585. 5. Novotny, V.J. Migration of liquid polymers on solid surfaces. J. Chem. Phys. 1990, 92, 3189–3196. 6. Min, B.G.; Choi, J.W.; Brown, H.R.; Yoon, D.Y.; O’Connor, T.M.; Jhon, M.S. Spreading characteristics of thin liquid films of perfluoropolyalkylethers on solid surfaces: effects of chain-end functionality and humidity. Trib. Lett. 1995, 1, 225–232. 7. O’Connor, T.M.; Jhon, M.S.; Bauer, C.L.; Min, B.G.; Yoon, D.Y.; Karis, T.E. Surface diffusion and flow activation energies of perfluoropolyalkylethers. Trib. Lett., 1995, 1, 219–223. 8. Ma, X.; Gui, J.; Smoliar, L.; Grannen, K.; Marchon, B.; Jhon, M.S.; Bauer, C.L. Spreading of perfluoropolyalkylether films on amorphous carbon surfaces. J. Chem. Phys. 1999, 110 (6), 3129–3137. 9. Ma, X.; Gui, J.; Smoliar, L.; Grannen, K.; Marchon, B.; Bauer, C.L.; Jhon, M.S. Complex terraced spreading of perfluoropolyalkylether films on carbon surfaces. Phys. Rev. E 1999, 59 (1), 722–727. 10. Pit, R.; Marchon, B.; Meeks, S.; Velidandla, V. Formation of lubricant ‘‘moguls’’ at the head= disk interface. Trib. Lett. 2001, 10, 133–142. 11. Ma, X.; Gui, J.; Grannen, K.J.; Smoliar, L.A.; Marchon, B.; Jhon, M.S.; Bauer, C.L. Spreading of pfpe lubricants on carbon surfaces: effect of hydrogen and nitrogen content. Trib. Lett. 1999, 6 (1), 9–14. 12. Tyndall, G.W.; Karis, T.E.; Jhon, M.S. Spreading profiles of molecularly thin perfluoropolyether films. Trib. Trans. 1999, 42 (3), 463–470. 13. Mate, C.M. Application of disjoining and capillary pressure to liquid lubricant films in magnetic recording. J. Appl. Phys. 1992, 72, 3084–3090. 14. O’Connor, T.M.; Back, Y.-R.; Jhon, M.S.; Min, B.G.; Yoon, D.Y.; Karis, T.E. Surface diffusion of thin perfluoropolyalkylether films. J. Appl. Phys. 1996, 79 (8), 5788–5790. 15. Dolak, Y.; Hillen, T. Cattaneo models for chemosensitive movement—numerical solution and pattern formation. J. Mathematical Biology 2003, 46 (5), 460–478.

Thin Liquid Film Deposition

16. Karis, T.E.; Tyndall, G.W. Calculation of spreading profiles for molecularly-thin films from surface energy gradients. J. Non-Newtonian Fluid Mech. 1999, 82, 287–302. 17. Porter, D.A.; Easterling K.E., Eds. Phase Transformations in Metals and Alloys; Van Nostrand Reinhold Co. Ltd.: Workingham, Berkshire, England, 1981; Chapter 2. 18. Ferry, J. D. Viscoelastic Properties of Polymers, 3rd Ed.; John Wiley & Sons: New York, NY, 1980; 241–254. 19. Karis, T.E.; Marchon, B.; Flores, V; Scarpulla, M. Lubricant spin-off from magnetic recording disks. Trib. Lett. 2001, 11 (3–4), 151–159. 20. Cantow, M.J.R.; Barrall, E.M., Jr.; Wolf, B.A.; Geerissen, H. Temperature and pressure depenence of the viscosities of perfluoropolyether fluids. J. Polym. Sci. Polym. Phys. 1987, 25, 603–609. 21. Toney, M.F.; Mate, C.M.; Leach, K.A. Roughness of molecularly thin perfluoropolyether polymer films. Appl. Phys. Lett. 2000, 77 (20), 3296–3298. 22. Karis, T.E. Water adsorption on thin film media. J. Colloid Interface Sci. 2000, 225, 196–203. 23. Braslau, A.; Pershan, P.S.; Swislow, G.; Ocko, B.M.; Als-Nielsen, J. X-ray reflectivity studies of

3087

24.

25.

26.

27.

28.

29.

30.

a microemulsion surface. Phys. Rev. A 1988, 38, 2457–2470. Karis, T.E.; Nyak, U.V. Liquid nanodroplets on thin film magnetic recording disks. Trib. Trans. 2004, 47 (1), 103–110. Frenkel, D.; Smit, B. Understanding Molecular Simulation: From Algorithm to Applications, 2nd Ed.; Academic Press: New York, NY, 2002. Ma, X.; Bauer, C.L.; Jhon, M.S.; Gui, J.; Marchon, B. Monte Carlo simulations of liquid spreading on a solid surface. Phys. Rev. E 1999, 60 (5), 5795–5801. Vinay, S.J.; Phillips, D.M.; Lee, Y.S.; Schroeder, C.M.; Ma, X.; Kim, M.C.; Jhon, M.S. Simulation of ultra-thin lubricant films spreading over various carbon overcoats. J. Appl. Phys. 2000, 87 (9), 6164–6166. Phillips, D.M.; Jhon, M.S. Dynamic simulation of nanoscale lubricant films. J. Appl. Phys. 2002, 91 (10), 7577–7579. Izumisawa, S.; Jhon, M.S. Molecular simulation of thin polymer films with functional endgroups. J. Chem. Phys. 2002, 117 (8), 3972–3977. Jhon, M.S.; Izumisawa, S.; Guo, Q.; Phillips, D.M.; Hsia, Y.T. Simulation of nanostructured lubricant films. IEEE Trans. Mag. 2003, 39 (2), 754–758.

T

Thiochemicals: Mercaptans, Sulfides, and Polysulfides T Jeffrey H. Yen Vijay R. Srinivas Gary S. Smith Arkema lnc., King of Prussia, Pennsylvania, U.S.A.

INTRODUCTION

Sulfur (VI) compounds

The word ‘‘thiochemicals’’ is derived from the Greek word ‘‘theion.’’ Generally speaking, thiochemicals are sulfur analogs of oxygen-containing compounds; for example, thiochemicals containing sulfur in the þ2 oxidation state, mercaptans and alkyl sulfides, are sulfur analogs of alcohols and ethers, respectively. As the sulfur atom is a homolog of the oxygen atom, present in the VI Group of the Periodic Table, there are some similarities in character and occurrence in nature. The differences, however, are dramatic— oxygen being a colorless gas essential to sustain human life, while sulfur is a yellow solid. The most fundamental molecules that contain oxygen and sulfur, respectively, are water and hydrogen sulfide. Again, water is essential for all living beings and hydrogen sulfide is a deadly and poisonous gas. Many complex molecules of biological interest contain sulfur, and the capability of the sulfur atom to both oxidize and reduce is used extensively in the normal everyday biological processes. Generally, thiochemicals can be classified as follows:


Sulfones: R-SO2-R; sulfolane, dimethyl sulfone, etc.
Sulfonic acids and derivatives; RSO2OH, RSO2X; methanesulfonic acid, p-toluenesulfonic acid, methanesulfonyl chloride, methanesulfonamide, etc.

Sulfur (II) compounds
Basic thiochemicals: examples are sulfur dioxide, hydrogen sulfide, sodium sulfide, sodium thiosulfate, etc.
Mercaptans (thiols): RSH; examples are methyl mercaptan, ethyl mercaptan, etc.
Sulfides: RSR0 ; examples are dimethyl sulfide, diethyl sulfide, etc.
Polysulfides: RSx R0 ; dimethyl disulfide, diethyldisulfide, t-nonyl polysulfide, etc.
Sulfenic acids RSOH and derivatives: benzenesulfenic acids=salts, benzenesulfenyl chloride. Sulfur (IV) compounds
Sulfoxides RS(O)R and derivatives; dimethylsulfoxide.
Sulfinic acids and derivatives: RS(O)OH, RS(O)X; benzenesulfinic acid, p-toluenesufinyl chloride, other arenesulfinate esters and salts.
Sulfite esters: (RO)2SO; ethylene sulfite. Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120033979 Copyright # 2006 by Taylor & Francis. All rights reserved.

To describe the full range of thiochemicals is beyond the scope of this entry, which will, therefore, place emphasis on the most common thiochemicals. This entry will cover mercaptans, sulfides, and polysulfides, while the next entry will cover sulfonic acids and its derivatives, some sulfoxides, mercaptoacids and their derivatives. The safety, health, and environmental issues for both entries are discussed at the end of this chapter.

MERCAPTANS Mercaptans and alkyl sulfides are the sulfur analogs of alcohols and ethers, respectively. They can be characterized by their extremely unpleasant odor. These compounds play an important role in biological systems as well as in the application of chemistry to everyday life. Some of the alkyl sulfides are found in many plant and animal oils, and are minor components of petroleum distillates, shale oil, and coal tar. Chemical and Physical Properties The properties of mercaptans and alcohols are quite different, although they appear to be similar in nature. The bond dissociation energy of the S–H bond is over 10=kcal=mol lower than the corresponding O–H bond. The ease of free-radical hydrogen abstraction from a mercaptan supports this fact and permits these compounds to be included in preparative free-radical chemistry. The above fact, along with the different boiling points observed for the mercaptans compared to the corresponding alcohols and the differences in their acidities, helps explain why the chemistry of these two classes of compounds differs in so many ways. 3089

3090

Thiochemicals: Mercaptans, Sulfides, and Polysulfides

Table 1 lists some properties for various straightchain alphatic mercaptans (C2–C12). The boiling points range from 34 C to 277 C, while their solubility in water is the greatest for ethyl mercaptan (6.76=g=L) and decreases as the aliphatic chain length increases. Manufacturing Technology There are several methods for the preparation of mercaptans and sulfides.[1] It is our intention here to emphasize only the most facile methods that are safe to operate commercially, use inexpensive raw materials, produce minimum by-products, and allow for simple and easy by-product handling. While the sources of sulfur for thiolation can be many, the most common ones are hydrogen sulfide (H2S), elemental sulfur, and carbon disulfide (CS2). H2S is a component of sour natural gas and can be separated for use in the manufacture of mercaptans and other sulfur chemicals. Additionally, H2S is generated in the many hydrodesulfurizing units in a refinery. With minimal purification this could be used for the manufacture of thiochemicals. In many instances H2S is manufactured by either the Girdler or the Folkins processes: Girdler process Folkins process

450 C

H2 þ S ƒƒƒ! H2 S

CH4 þ S ! 2H2 S þ CS2

Mercaptans and sulfides are manufactured mainly by the thiolation of alcohols and by the addition of H2S to olefins. From alcohols via heterogenous catalysis Alcohols react with H2S in the presence of acid or base catalysts to give mercaptans. The nucleophilic substitution of alcohols by H2S occurs at around 300 C on an alumina-type catalyst impregnated with alkali metal oxides such as Na2O, K2O or transition metal oxides such as WO3. Phosphotungstate alkaline salts on alumina have also been used. ca: 300 C

ROH þ H2 S ƒƒƒƒƒ! RSH þ H2 O catalyst

Mercaptans can further react with alcohols, under the same conditions, to produce sulfides. Therefore, the H2S=ROH ratio must be high to optimize mercaptan selectivity over sulfide selectivity. The temperature also plays a key role in the selectivity and conversion. As the molecular weight of the alcohol increases, the temperature of the reaction needs to be minimized and controlled in the range of 270–400 C, to avoid dehydration of the alcohol to an olefin. Mercaptans ranging from methylmercaptan to dodecylmercaptan

have been industrially synthesized by thiolation of primary alcohols. This route is generally used to make primary mercaptans from primary alcohols. Thiolation of secondary alcohols is not the preferred route to synthesize secondary mercaptans, because of the facile dehydration that occurs under reaction conditions of the secondary alcohols. H2S can also react with aliphatic or cyclic oxides, viz., ethers; to produce the corresponding thioethers or sulfides. This is one of the preferred ways to make tetrahydrothiophene or thiophene. From olefins The catalytic addition of H2S to olefins follows the Markovnikov rule, with the –SH moeity attaching itself to the carbon atom connected to the least number of hydrogen atoms. With linear olefins, having terminal or internal double bonds, acid catalysis using zeolites, cation-exchange resins, and silica–alumina yields secondary mercaptans. The preferred route for secondary mercaptans is the H2S addition to olefinic compounds. SH j

RCH ¼ CHR0 þ H2 S ! RH C CH2 R0 With branched olefins, one usually obtains tertiary mercaptans as the major product. For example, tertiary mercaptans such as t-dodecyl mercaptan and t-nonyl mercaptan are manufactured starting from propylene oligomers and using the acid-catalyzed addition of H2S. In the presence of strong acids, electrophilic addition of H2S and mercaptans to olefins takes place, yielding Markovnikov adducts. Because divalent sulfur compounds are stronger nucleophiles than alcohol or water, the addition of mercaptans to an intermediate carbonium ion takes place quite readily.[2] This type of reaction yields secondary mercaptans and secondary thioethers from alpha olefins. Since 1965, the ready availability of alpha olefins has opened the possibility of utilizing several of their basic reactions known for a long time. Friedel Crafts catalysts such as anhydrous aluminium chloride, fluoboric acid, mixtures of hydrogen fluoride and boron trifiuoride, and their hydrocarbon complexes have been used for the addition of H2S to olefins—specifically decene.[3,4] Silica–alumina has been used to make high-molecular-weight mercaptans, which have found use as rubber modifiers.[5] Elemental sulfur along with the bases ammonia or alkylamine or rubber vulcanizing agents such as mercaptobenzothiazole, thiurams, and dithiocarbamates have been used as catalysts to make mercaptans and sulfides from olefins and hydrogen sulfides.[6–12] Acid cation exchangers (e.g., wet sulfonated styrene-divinylbenzene copolymers in the acid form or more advantageously in combination

98.0

C6H11SH n-C8H17SH

4802-20-4

D-Limonene

C10H20S2

n-C14H29SH

n-C10H21SH

(CH3)2CHC5H10SH

Mixture

t-C12H25SH

n-C12H25SH

t-C9H19SH

b

Determined by gas chromatography or titrimetric analysis. APHA by ASTM D-1209. c ASTM D-1078. d Tag closed cup.

dimercaptan

n-Tetradecyl mercaptan

a

143-10-2 2079-95-0

n-Decyl mercaptan

Mixture 3001-66-9

sec-Octyl mercaptan

t-Dodecyl mercaptan

Pennfloatd

112-55-0 25103-58-6

n-Dodecyl mercaptan

111-88-6

n-C6H13SH

111-31-9 1569-69-3

n-Hexyl mercaptan

Cyclohexyl mercaptan 25360-10-5

t-C4H9SH

109-79-5 75-66-1

n-Butyl mercaptan

t-Butyl mercaptan

n-Octyl mercaptan

96.0 99.0

(CH3)2CHSH

75-33-2

t-Nonyl mercaptan

99.0

n-C3H7SH

107-03-09

n-Propyl mercaptan

Isopropyl mercaptan

95.0

96.04

96

—

—

98.5

98.0

97.0

98.0

98.0 97.0

C2H5SH

75-08-1

Ethyl mercaptan

n-C4H9SH

99.5 99.3

CH3SH

74-93-1

Methyl mercaptan

Formula

CAS No.

Puritya (% min.)

Product

Table 1 Properties of mercaptans

100

40

15

—

N=A

15

20

15

15

15

15

15

15

—

298

235

—

N=A

227.0

269.0

188.7

194.0

155.0

149.0

62.1

96.3

66.2 51.0

20

34.4

4.5

IBP min.

15

40

15

Colorb (APHA max.)

—

320

245

—

N=A

248.0

285.0

201.0

203.0

161.0

158.0

67.8

100.0

56.1

69.4

36.1

7.5

95% max.

Distillationc range( C)

204

230

174.3

—

Mixture

202.4

202.4

160.3

146.3

116.2

118.2

90.2

90.2

76.2

76.2

62.1

48.1

Molecular wL (calc.)

1.034

0.850

0.849

—

0.873

0.857

0.844

0.848

0.845

0.946

0.841

0.801

0.843

0.814

0.840

0.837

0.870

Density at 20 (Kg/L)

575

95

65

—

>95

87

93

51

52

29

205

7

200 C Sulfur content  26.4%  28.4% Softening point  75 C  95 C Stearic acid content  9%  1.1%

but the higher-rank –S4– and –S5– polysulfides are generally produced as mixtures containing the tri–, tetra–, and pentasulfides. The higher dialkyl pentasulfides, such as tert-nonyl-S5-tert-nonyl (TNPS), on standing, tend to precipitate some sulfur and equilibrate at an average sulfur ‘‘rank’’ of about 4.7.

Applications The higher dialkyl polysulfides, such TNPS, find use as extreme-pressure lube additives, especially for cutting oils; the lower dialkyl polysulfides, such as (CH3)2Sx, as sulfur-donor agents, for example, to convert phosphites to thiophosphates: ðROÞ3 P þ R0 SX R0
! ðROÞ3 P ¼ S þ R0 SX
1 R0 Disulfides and polysulfides are useful for sulfiding hydrotreating catalysts, used in petroleum refining to convert metal oxides to the preferred metal sulfides. Hydrotreating is an essential process in the refining of petroleum. It removes heteroatoms, nitrogen, and sulfur from crude oil and its fractions, formulated into gasoline and diesel (Table 5).

Vulcanizing agents

SAFETY, HEALTH, AND ENVIRONMENTAL ISSUES 6.1 Material Handling[47] Mercaptans are flammable. Normal precautions used in handling any flammable chemical are applicable. Flares used to burn vapors to reduce pressure on a storage tank should be situated as far away from the general storage area as practical. The flare should have a suitable flame arrestor. All potential sources of fire, flame, and sparks in the immediate area should be sought out and eliminated. Plastic gloves and chemical goggles should be worn to protect the skin and eyes when handling mercaptans. If mercaptan contacts the skin or eyes, flush the affected area immediately and thoroughly with water. In handling mercaptans, avoid spills and leaks of liquid or vapor, avoid prolonged breathing of mercaptan vapors. Although the odor of mercaptan will normally become extremely disagreeable before the vapors reach dangerous concentrations, the nose may become temporarily desensitized after exposure. A self-contained breathertype mask is recommended for protection when working in areas of high vapor concentration or for prolonged exposure to lower concentrations of vapor.

Thiochemicals: Mercaptans, Sulfides, and Polysulfides

3097

Table 5 Chemical and physical properties of polysulfides Commercial product

Physical state

Physical and chemical properties 

Uses

TPS 20 Di-tert-dodecyl trisulfide CAS: 68425-15-0 EINECS: 270-335-7

Pale yellow and slightly odorous liquid

BP > 200 C (decomposition) dð20=4Þ ¼ 0:95

Extreme pressure additive for lubricating and cutting oils

TPS 32 Di-tert-decyl pentasulfide CAS: 68425-15-0 EINECS: 270-335-7

Pale yellow and slightly odorous liquid

BP > 200 C (decomposition) dð20=4Þ ¼ 1:01

Extreme pressure additive for lubricating and cutting oils; cross-linking agent for elastomers.

TPS 37 Di-tert-nonyl pentasulfide CAS: 68425-16-1 EINECS: 270-336-2

Yellow and slightly odorous liquid

BP > 200 C (decomposition) dð20=4Þ ¼ 1:024

Extreme pressure additive for lubricating and cutting oils; cross-linking agent for elastomers

TPS 44 Di-tert-butyl polysulfide CAS: 68937-96-2 EINECS: 273-103-3

Yellow and slightly odorous liquid

BP > 178 C (decomposition) dð20=4Þ ¼ 1:007

Extreme pressure additive for gear oils and cutting oils

TPS 54 Di-tert-butyl polysulfide CAS : 68937-96-2 EINECS: 273-103-3

Yellow and slightly odorous liquid

BP > 170 C (decomposition) dð20=4Þ ¼ 1:082

Extreme pressure additive for cutting oils

(From Refs.[47,48].)

To prevent accidental ignition of mercaptan vapors, employees working in handling and storage areas should not wear metal heel- or toe-plates on shoes. Nonsparking tools should be used when working on mercaptan equipment or containers. The general precautions discussed in handling mercaptans are applicable to Thiophene and tetrahydrothiophene. Di-t-nonyl polysulfide is a relatively safe nontoxic material. It can be handled in the same manner as any fuel oil. Materials of Construction[47] Steel, stainless steel, and copper-free steel alloys are the preferred materials of construction for mercaptan service. In particular, stainless steel should be used for any vessel or line that is to be open and exposed to air frequently. Aluminum is also suitable for mercaptan use provided the pressure rating of aluminum equipment or piping is sufficient to meet the pressure requirements of the application. Iron or carbon steel is less acceptable than stainless steels or aluminum although it can be used if appropriate measures are taken to condition the iron or carbon steel equipment before putting it into service. Allowing a small amount of mercaptan to stand in it for a period of time and subsequently keeping the equipment under a dry, inert atmosphere can prevent corrosion of the equipment. The hazard in using iron or carbon steel is the formation of iron–sulfur complexes, which are pyrophoric

and constitute a potential fire hazard because of the flammability of mercaptans. If iron or carbon steel is used in mercaptan service, thorough washing with water during disassembly and cleaning is absolutely necessary. Sulfonic acids are corrosive. Do not use copper or copper-bearing alloys for mercaptan service. Mercaptans readily attack these metals and are contaminated by them. Mercaptans are odorous; therefore, storage tanks should not be vented to the atmosphere. Venting should be to a flare or a scrubber to remove the mercaptan. Toxicity[47] All alkyl mercaptans have distinctive mercaptan odors, and their presence in air is quite evident. While this may present a problem in community relations, it is a built-in warning system. It is almost impossible to have an undetected mercaptan leak. However, there is a danger in ignoring the mercaptan odor. The lowermolecular-weight mercaptans, in high concentrations, tend to deaden the sense of smell; therefore, dangerous concentrations of vapors could go undetected. The lower-molecular-weight mercaptans have the strongest odors. In very low concentrations (ppm range) they are recognized as gas leaks, because they are commonly used to odorize fuel gas. In intermediate concentrations their pungent odor has been likened to that of garlic, rotting cabbage, and other decomposed

T

3098

organic matter. In higher concentrations, the desensitization effect takes place, and an alcoholic or somewhat pleasant odor may be experienced. Although some oral toxicity data are available, inhalation is the most likely route of entry into the body. Most mercaptans have an LC50 greater than 2000 ppm. This is above the normally accepted ‘‘Toxic Substance’’ classification. The exception, t-octyl mercaptan, has been tested recently and has been found to be a ‘‘highly toxic substance’’ with LC50 of less than 200 ppm. No cumulative or permanent sensitization effects have been reported from exposure to mercaptans. Long-term exposure to high concentrations of the lower-molecular-weight mercaptans can result in headache, nausea, loss of consciousness, and even death. This is possible because of olfactory desensitization. However, death from mercaptan inhalation is extremely rare, and if the victim is removed to fresh air, even if unconscious, recovery is complete. CONCLUSIONS Thiochemicals are among one of the more important intermediates and end-user chemicals, starting from a relatively simple molecule such as H2S to complex molecules such as polysulfides. Many thiochemicals that are commercially manufactured provide building blocks for more specialized ones. Thiochemicals enjoy applications as intermediates in the manufacture of agricultural chemicals, pharmaceuticals, detergents, lubricants, etc. Some thiochemicals have played important roles, as solvents, chain transfer agents in polymer synthesis, electroplating solutions, presulfiding agents for hydrotreating catalysts, coke and CO inhibitors in the steam cracking process, etc. ACKNOWLEDGMENT The authors acknowledge the support and information provided by Arkema Inc. REFERENCES 1. Reid, F.F. Organic Chemistry of Bivalent Sulfur; Chemical Publishing Company: New York, 1958; Vol. 1. 2. Ohno, A.; Oae, S. Thiols. In Organic Chemistry of Sulfur; Oae, S., Ed.; Plenum Press: New York, 1977. 3. Bell, R.T.; Thacker, C.M U.S. Patent 2,498,872, 1950. 4. Bell, R.T U.S. Patent 2,531,602, 1950. 5. Schulze, W.A U.S. Patent 2,502,596, 1950.

Thiochemicals: Mercaptans, Sulfides, and Polysulfides

6. Jones, S.O.; Reid, E. J. Am. Chem. Soc. 1938, 60, 2452. 7. Louthan, R.P U.S. Patent 3,221,056, 1965. 8. Doss, R.C U.S. Patent 3,419,614, 1968. 9. Warner, P.F U.S. Patent 3,114,776, 1963. 10. Lang, A.; Vannel, P U.S. Patent 3,333,008, 1967. 11. Pennsalt Chem Corp. U.S. Patent 2,951,875, 1960. 12. Elf Aquitaine European Patents 101,356, 1982. 13. Ipatieff, V.N.; Pines, H.; Friedman, B.S. J. Am. Chem. Soc. 1938, 60, 273. 14. Macho, V Czech Patent 185,469, 1980. 15. Arretz, E.; Mirassou, A.; Landoussy, C.; Auge, P. Fr. Demande FR. 2,531,426, 1984. 16. Kubicek, D.H Belgium Patent 886,261, 1981. 17. Onyestyak, O.; KaHo, D.; Papp, J.; Detrekoy, E. Hung. Teijes Hungarian Patent 29,972, 1984. 18. Fried, H.E Fur. Patent 122,654, 1984. 19. Hahn, W German Patent 1,110,631, 1961. 20. Fields, E.K U.S. Patent 4,347,384, 1982. 21. Kharasch, M.S.; Read, A.T.; Mayo, F.R. Chem. Md. 1938, 752. 22. Posner, H. Ber. Dtsch. Chem. 1905, 38, 646. 23. Ashworth, F.; Burkhardt, G.N. J. Chem. Soc. 1928, 1791–1802. 24. Grattan, D.W.; Locke, J.M.; Wallis, S.R. J. Chem. Soc. Perkin Trans. 1973, 1, 2264. 25. Mayo, F.R.; Walling, C. Chem. Rev. 1940, 27, 387. 26. Evans, E.A.; Vaughan, W.E.; Rust, F.F U.S. Patent 2,376,675, 1945; U.S. Patent 2,411,961, 1946; British Patent 567,524, 1946. 27. Dimmig, D.A U.S. Patent 4,140,604, 1979; European Patent 5400, 1979. 28. Olivier, J.; Souloumiac, J.; Suberlucq, J U.S. Patents 4,233,128, 1980, 4,443,310, 1984. 29. Buchholz, B U.S. Patent 3,652,680, 1972. 30. Bierker, G.L.; Kiovnick, A U.S. Patent 4,043,886, 1977. 31. Pennwalt Corporation Publication S-222B; Brandrup, J., Immergut, E.H., Polymer Handbook, 2nd Ed.; Wiley–Interscience: New York, 1975. 32. Mayo, F.R. J. Am. Chem. Soc. 1943, 65, 2324. 33. Pryor, W.A. Mechanisms of Sulfur Reactions; McGraw-Hill: New York, 1962; 82–88. 34. Allesandrin, C.G U.S. Patent 4,119,659, 1978. 35. Punja, N U.S. Patents 4,429,153, 1984, 4,370,346, 1983. 36. Mathew, C.T.; Ulmer, H.E U.S. Patent 3,393,331, 1976. 37. Fankhauser, E.; Sturm, E U.S. Patent 4,459,152, 1984. 38. Lemaire, H. Nonionic Surfactants; Schick, M.J., Ed.; Marcel Dekker: New York, 1966. 39. Onopchenko, A.; Schultz, J U.S. Patents 4,102,932, 1978, 4,009,211, 1977.

Thiochemicals: Mercaptans, Sulfides, and Polysulfides

40. Suzuki, S U.S. Patent 4,172,211, 1979. 41. Mawn, P.E. Outlook for lube oil additives. Impact (L760704), 1976, July 29. 42. Carson, J U.S. Patent 4,213,905, 1980. 43. Fujimoto, T.; Kondo, K.; Suda, M.; Tunemoto, D U.S. Patent 4,268,442, 1981. 44. Pennwalt Corporation Publication S-222B; Brandrup, J.; Immergut, E.H. Polymer Handbook, 2nd Ed.; Wiley–Interscience: New York, 1975. 45. Mayo, F.R. J. Am. Chem. Soc. 1943, 65, 2324. 46. Nevers, A.D U.S. Patent 3,826,631, 1974. 47. Arkema Corp. (formerly Pennwalt Corporation). Organosulfur Intermediates, 2000; Arkema corp.: Philadelphia, PA, 1979. 48. Arkema Corp. (formerly Pennwalt Corporation) Organic Chemicals Division Product Catalog 2000; Arkema Corp.: Philadelphia, PA, 1989.

3099

49. Pennwalt European Patents 171,092, 1986, 202, 420, 1986. 50. Elf Aquitaine European Patents 269,517, 1988, 337,837, 1989; Lubrizol Corp. U.S. Patent 4,344,854, 1982. 51. EI. Du Pont de Nemours U.S. Patents 4,792,633, 1988, 4,599,451, 1986. 52. Vottero, C.; Labat, Y.; Poirier, J.-M. European Patent 318,394, 1989. 53. Lindstrom, M Atochem North America. U.S. Patent 5,028,343, 1991. 54. Clark, P.D.; Lesage, K.L. Alberta Sulfur Res. Ltd. Q. Bull. 1987, 24 (1–2), 23–43, 27 (3), 41–43 1990. 55. Labat, Y. U.S. Patents 4,728,447, 1998; 4,827,040, 1989. 56. Halle, H. Oil Gas J. 1980, 70. 57. Wiechers, A. U.S. Patent 4,211,644, 1980.

T

Thiochemicals: Mercapto Acids and Organosulfur (IV/VI) Compounds

T

Jeffrey H. Yen Gary S. Smith Vijay R. Srinivas Arkema Inc., King of Prussia, Pennsylvania, U.S.A.

INTRODUCTION As discussed in the previous entry, ‘‘Thiochemicals: Mercaptans, Sulfides, and Polysulfides,’’ thiochemical is derived from the Greek word ‘‘theion.’’ Generally speaking, thiochemicals containing sulfur formally in the þ2 oxidation state are analogs of oxygen-containing compounds; for example, mercaptans and alkyl sulfides are sulfur analogs of alcohols and ethers, respectively. Unlike oxygen, sulfur can also exist in the formal þ4 and þ6 oxidation states, thus affording a vast array of functionalities. Generally, thiochemicals can be classified into the following categories: Sulfur (II) compounds
Basic thiochemicals: Examples are sulfur dioxide, hydrogen sulfide, sodium sulfide, sodium thiosulfate, etc.
Mercaptans (thiols): RSH; examples are methyl mercaptan, ethyl mercaptan, etc.
Sulfides: RSR0 ; examples are dimethyl sulfide, diethyl sulfide, etc.
Polysulfides: RSxR0 ; dimethyl disulfide, diethyldisulfide, t-nonyl polysulfide, etc.
Sulfenic acids RSOH and derivatives: benzenesulfenic acids=salts, benzenesulfenyl chloride. Sulfur (IV) compounds
Sulfoxides: RS(O)R and derivatives; dimethylsulfoxide.
Sulfinic acids and derivatives: RS(O)OH, RS(O)X; benzenesulfinic acid, p-toluenesufinyl chloride, other arenesulfinate esters, and salts.
Sulfite esters: (RO)2SO; ethylene sulfite. Sulfur (VI) compounds
Sulfones: R-SO2-R; sulfolane, dimethyl sulfone, etc.
Sulfonic acids and derivatives: RSO2OH, RSO2X; methanesulfonic acid (MSA), p-toluenesulfonic acid (PTSA), methanesulfonyl chloride (MSC), methanesulfonamide, etc. Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120016066 Copyright # 2006 by Taylor & Francis. All rights reserved.

To describe the full range of thiochemicals is beyond the scope of this entry, which will therefore emphasize information on the most common thiochemicals of commerce. This entry will cover mercaptoacids, sulfonic acids and their derivatives, some sulfoxides, etc., while the previous entry covers mercaptans, sulfides, and polysulfides. The safety, health, and environmental issues for both entries are discussed at the end of the previous entry.

MERCAPTO ACIDS, SALTS, ESTERS Chemical and Physical Properties Mercaptocarboxylic acids are difunctional molecules with both carboxyl and thiol functionality. The general formula is HS-R-CO2H, with R being alkyl or aryl. Both the carboxyl and thiol groups are acidic. For mercaptoacetic acid and thioglycolic acid the pKa values are 3.6 and 10.2, respectively. The physical properties of commercial mercaptoacids, esters, and salts are given in Table 1. The carboxylic acid group can be readily converted to afford salts or esters. Mild oxidants such as I2 convert the thiol to the disulfide. The generated disulfides can undergo thiol=disulfide exchange to afford unsymmetrical disulfides. The thiol group also reacts readily with aldehydes or ketones to afford dithioketals, with acyl chlorides or anhydride to afford thioesters, and with a,b-unsaturated carbonyl and nitrile compounds. As discussed below, the thiol readily reacts with alkyl halides to form dialkyl sulfides.

Manufacturing Technology Mercaptoacetic acid, also known as thioglycolic acid, is prepared from the continuous reaction of chloroacetic acid with two molar equivalents of aqueous ammonium or metal hydrosulfide. The reaction is run under autogenous H2S overpressure, derived from the neutralization of the chloroacetic acid with one equivalent 3101

152–156

166=760 101=12 85–87=2 95=2.2 137=2 101=0.3

120.2 162.3 218.4 218.4 210.3 238.3

Methyl 3-mercaptopropionate

Butyl 3-mercaptopropionate

2-Ethylhexyl 3-mercaptopropionate

Isooctyl 3-mercaptopropionate (mixture of isomers)

Ethylene glycol dimercaptoacetate

Ethylene glycol dimercaptopropionate

2-Mercaptoethyl tallate

106=3 107=4

204.3 204.3

2-Ethylhexyl mercaptoacetate

Isooctyl mercaptoacetate (mixture of isomers)

68440-24-4

22504-50-3

123-81-9

30774-01-7

50448-95-8

16215-21-7

2935-90-2

25103-09-7

7659-86-1

623-51-8 10047-28-6

155=760 110=40

120.2 148.2

Ethyl mercaptoacetate

2365-48-2

1119-62-6

111-17-1

505-73-7

123-93-3

107-96-0

Butyl mercaptoacetate

HSCH2CH2O2C–R (R ¼ unsaturated C18–C20)

148=760

210.3

(SCH2CH2CO2H)2

Dithiodipropionic acid

130

100–102

106.1

178.2

Thiodipropionic acid

128

111=15

79-42-5

85=5 15.5

68-11-1

123=29

16

CAS Reg. No.

Boiling point ( C/torr)

Melting point ( C)

Methyl mercaptoacetate

182.2

(SCH2CO2H)2 S(CH2CH2CO2H)2

Dithiodiglycolic acid

106.1 150.2

HS(CH2)2CO2H S(CH2CO2H)2

3-Mercaptopropionic acid

106.1

HSCH(CH3)CO2H

2-Mercaptopropionic acid (thiolactic acid)

Thiodiglycolic acid

92.12

MW

HSCH2CO2H

Structure

Mercaptoacetic acid (thioglycolic acid)

Name

Table 1 Physical properties of selected mercapto carboxylic acids, esters, and derivatives

3102 Thiochemicals: Mercapto Acids and Organosulfur (IV/VI) Compounds

Thiochemicals: Mercapto Acids and Organosulfur (IV/VI) Compounds

of hydrosulfide. Because H2S (pKa1 ¼ 7.0) has approximately 1000 times the acidity of the thiol group on the formed mercaptoacetic acid, the presence of the H2S inhibits formation of the mercaptoacetate dianion and subsequent reaction with chloroacetate to form the sulfide. The crude mercaptoacetate salt from the reaction is acidified, extracted into organic solvents, and ultimately recovered by vacuum distillation. Principal by-products are the halide salts. C1CH2 CO2 H þ 2MSH

! HSCH2 CO2 M þ MC1 þ H2 S

ð1Þ

HSCH2 CO2 M þ HC1 ! HSCH2 CO2 H þ MC1 M ¼ NH4 ; Na ; K; Ca

ð2Þ

HSCH2 CO2  þ HS ) *



SCH2 CO2 þ H2 S

C1CH2 CO2 þ  SCH2 CO2

! SðCH2 CO2 Þ2 þ C1

ð3Þ ð4Þ

The product mercaptoacetic acid will self-esterify on storage or heating. The self-esterification products are primarily the linear dimer (x ¼ 1), along with some cyclic dimer, oligoesters (x ¼ 2þ), and solid orthothioester. These self-esterification reactions, with the exception of the formation of the orthothioesters, can be readily reversed in water with dilute acid or base catalyst.

Most mercaptoacetic acid is sold in the form of esters, primarily for use in organotin stabilizers for poly(vinyl chloride) (PVC). The esterification is performed in batch or continuous mode, primarily using 2-ethylhexanol or isooctanol. Hþ

HSCH2 CO2 H þ ROH ! HSCH2 CO2 R þ H2 O R ¼ 2-ethylhexyl; isooctyl; butyl; etc:

ð6Þ

Purified mercaptoacetic acid is sold in anhydrous or 85% aqueous forms, most often for subsequent conversion to the esters or to the ammonium, sodium, potassium, or calcium salts. 3-Mercaptopropionic acid is produced from metal hydrosulfides and either acrylic acid or acrylonitrile. Mercaptoethyl tallate is another mercapto-ester used in commercial organotin stabilizers. It is manufactured by a standard esterification of mercaptoethanol and tall oil, a mixture of fatty acids.

3103

Applications Mercaptoacetate salts have extensive applications in personal care, particularly in cold-waving of human hair, hair straightening, and as depilatories. Essentially, the dianionic mercaptide reacts with the cystine (disulfide) cross-links between the keratin polypeptide chains via mercaptan–disulfide exchange. This allows deformation of the keratin with resultant curling or straightening of the hair. The cystine disulfide linkages are then restored by use of a ‘‘neutralizer,’’ actually an oxidant such as H2O2, sodium perborate, or sodium or potassium perbromate. Similarly, mercaptoacetic acid has found applications in leather treatments, including dehairing and softening of hides, and in the treatment of wool fibers. Mercaptoacetate and mercaptopropionate esters are key raw materials in the manufacture of organotin stabilizers for use in PVC. The mercapto-esters are reacted with organotin chlorides under two-phase aqueous conditions, with the liquid products being isolated by phase separation and stripping of residual water.[1,2] The generated organotin mercaptides are the workhorse thermal stabilizers for processing of PVC. Methyl and octyltin mercaptide stabilizers are used in PVC for food-contact applications because these materials offer very low migration from the PVC and low overall toxicity. Butyltin mercaptides are employed in the production of PVC films, sheet, injection moldings, floor tiles, and wall coverings, as well as for pipe extrusions and siding containing high levels of TiO2. Other applications for mercaptoacids and esters include the following. Mercaptoacetic acid serves as a cocatalyst with strong mineral acid or sulfonic acid resins for the condensation of phenol and acetone in the manufacture of bisphenol A. The acid and its ammonium and ethanolamine salts are components in descaling compositions for removal of iron oxides. Mercaptoacetic acid is used in oil drilling where it acts as an iron sequesterant and an acidizing agent. In concrete and cement applications, polymeric melamines and mercaptoacetic acid have been combined in a new class of ‘‘superplasticizers’’ that impart high fluidity, reduced water content, and improved mechanical properties. In this application, it is replacing sulfananilic acid. Alkanolamine salts of the dithiodipropionic acid are available as extreme pressure lubricants. Similar salts of dithiodiglycolic acid are also available.

SULFOXIDES AND SULFONES Chemical and Physical Properties The physical properties of the predominant alkyl sulfoxides and sulfones of commerce are listed in Table 2.

T

3104

Thiochemicals: Mercapto Acids and Organosulfur (IV/VI) Compounds

Table 2 Physical properties of selected sulfoxides and sulfones Name

Melting point ( C)

Boiling point ( C/torr)

Formula

MW

Dimethyl sulfoxide

CH3SOCH3

78.1

18.6

189.0=760

67-68-5

Dimethyl sulfone

CH3SO2CH3

94.1

108–110

238=760, sublimes 90–100=13

67-71-0

Sulfolane

120.2

28.5

287.3=760

126-33-0

3-Sulfolene

118.2

65

>110=760 (dec.)

77-79-2

Of these, dimethyl sulfoxide (DMSO) and sulfolane (tetramethylene sulfone) represent the largest volumes. The physical, chemical, and solvent properties of DMSO and sulfolane are well documented.[3–5,6] Dimethyl sulfoxide decomposes only slowly at its normal boiling point (189 C) in air, exhibiting only 2% decomposition to (CH3)2S, (CH3)2SO2, CH3SH, CH3SSCH3, and (CH3S)2CH2. Its decomposition (vide infra) is accelerated by acids, anhydrides, acid chlorides, amides, and glycols. Dimethyl sulfoxide has many applications as a solvent and reactant (vide infra). Sulfones possess extraordinary thermal and chemical stability. Sulfolane is thermally stable to 220 C decomposing to SO2 and polymeric materials. Dimethyl sulfone decomposes by C–S fission at 510– 640 C at 0.7 torr. Sulfones are generally nonreactive materials except under extreme conditions. Nonetheless, sulfones with multiple chlorine or fluorine substituents can be hydrolyzed to sulfonic acids or cleaved by halogen to afford the sulfonyl halides. Manufacturing Technology While a number of synthetic routes are available to various sulfoxides, the primary methods for commercial production of DMSO involve oxidation of dimethyl sulfide by oxides of nitrogen or by air in the presence of NOx catalyst.[3–5] Dimethyl sulfoxide is both the product and the reaction solvent. To alleviate the potential for exothermic, and potentially explosive, runaway reactions in these oxidations, the feed rate for dimethyl sulfide is adjusted to ensure complete conversion and, thus, low instantaneous concentrations at any time. Alternate oxidants for the conversion of sulfides to sulfoxides include nitric acid, H2O2=acetic acid, peracids, and halogen=water.[7]

In all oxidations of sulfides to sulfoxides, yield loss because of overoxidation to form the sulfones is common unless temperatures are low. Another source of yield loss in aliphatic sulfoxides with a-hydrogens arises from the Pummerer reactions, which occur at temperatures greater than 80 C under acidic conditions or in the presence of acid anhydrides=halides. The Pummerer reactions are quite general, leading to products of the general formula RSCR20 Z, with Z being acetate, chloride, hydroxide, etc. Because sulfoxide sulfur is in an intermediate oxidation state, formally þ4, it undergoes a small degree of thermal disproportionation to the corresponding sulfone and sulfide, at elevated temperatures near or above the normal boiling point. The formation of dimethyl sulfide is in part responsible for the garlic-like odor in aged DMSO, and can be minimized by storage under air. Ozone, H2O2, HNO3, and O2 with NOx catalysts are all effective oxidants for converting sulfides to sulfones, albeit much more slowly and at higher temperatures than were used to make sulfoxides. Alkyl aryl sulfones are typically produced by the Friedel–Crafts reaction of an aromatic hydrocarbon with an alkanesulfonyl halide or anhydride. Symmetrical aromatic sulfones are generally prepared by the sulfonation of aromatic hydrocarbon with SO3, H2SO4=SO3, Cl2SO2, ClSO3H, etc.[3–5] The corresponding aromatic sulfonic acids or sulfonyl chlorides are coproducts of the latter reaction. ½O

R2 SO ! R2 SO2 Lewis Acid Catalyst

RSO2 X þ ArH ƒƒƒƒƒƒƒƒƒ! RSO2 Ar þ HX X ¼ C1;Br;OSO2 R

D

ðCH3 Þ2 S þ NO2 ! ðCH3 Þ2 SO þ NO ðCH3 Þ2 S þ

1 2 O2

cat:NOX

ƒƒƒƒƒ! ðCH3 Þ2 SO

ð7Þ ð8Þ

CAS Reg. No.

R2 SO ! R2 SO2 þ R2 S

ð9Þ ð10Þ ð11Þ

3-Sulfolene is produced from butadiene and SO2, via classical cycloaddition chemistry, with hydrogenation

Thiochemicals: Mercapto Acids and Organosulfur (IV/VI) Compounds

of the sulfolene affording the sulfolane. Sulfolane is available as a crystalline anhydrous material, or with 3% added water as a freezing point depressant.

3105

Z defined as shown below (R and R0 ¼ substituted or unsubstituted aliphatic, aromatic or heterocyclic groups, and M ¼ metal, ammonium, phosphonium).

Z

Applications Both DMSO and sulfolane are extensively used in the chemical, pharmaceutical, polymer, and electronics industries as polar aprotic solvents, with unique properties such as a high dielectric constant, high polarity, and high miscibility with organic and aqueous materials. For example, sulfolane finds use in the refining industry for the separation of benzene, toluene, and xylene (BTX) fractions from paraffins. The use of DMSO as a reactant and as a solvent has been discussed in detail. Because of its highly polarized Sþ–O bond, DMSO is a strong solvator of water, inorganic salts, and most moderately to highly polar organic molecules. Because of its ready permeation through the skin and strong solvating properties, it is used as a carrier solvent for transdermal delivery of pharmaceuticals. As a reaction solvent, DMSO serves to activate NH, OH, and CH functional groups in a variety of reaction chemistries. Dimethyl sulfoxide oxidizes aromatic thiols to disulfides at room temperature.

CH3CH2SO3H CH3CH2CH2SO3H

HCl HOSO3H

2.8

FSO3H

CH3OSO3H

3.0

HO3SCH2SO3H p-O2NC6H5SO3H

1.5

204=1

140=20

172=0.1

245 (dec.)

174=1.3

163=1.3

16794-13-1

2386-47-2

5284-66-2

594-45-6

503-40-2

75-75-2

CAS Reg. No.

124–125 (1H2O) 83 (3H2O)

90 (dehy.)

106–107 38 (metastable) 106 (1H2O) 93 (3H2O)

62.1 (2H2O)

65–66 45–46 (1H2O) 43–44 (1.5H2O)

198 (dec.)

232.3 232.3

acid

L-()-Camphorsulfonic

16.1

15.9

152.2

CH3(CH2)4SO3H

Pentane-1-sulfonic acid

149=1

15.2

113

138.2 144.2

130=3

123=1

167=10

Boiling point ( C/torr)

7.5

CH3(CH2)3SO3H

124.2

CH3(CH2)2SO3H

Propane-1-sulfonic acid

17 4.3 (1H2O)

(CH3)3CSO3H

110.1

CH3CH2SO3H

Ethanesulfonic acid

96–100

Butane-1-sulfonic acid

176.2

HO3SCH2SO3H

Methanedisulfonic acid

19 11 (1H2O) 52 (3H2O)

Melting point ( C)

2-Methylpropane-2-sulfonic acid

96.10

MW

CH3SO3H

Formula

MSA

Name

Table 4A Physical properties of selected aliphatic and aromatic sulfonic acids

3106 Thiochemicals: Mercapto Acids and Organosulfur (IV/VI) Compounds

Thiochemicals: Mercapto Acids and Organosulfur (IV/VI) Compounds

3107

Table 4B Physical properties of selected sulfonic acids Name

Formula

MW

Fluorinated organic sulfonic acids Trifluoromethanesulfonic acid

CF3SO3H

150.1

Pentafluoroethanesulfonic acid Heptafluoropropane-1-sulfonic acid Nonafluorobutane-1-sulfonic acid Undecafluoropentane-1-sulfonic acid Tridecafluorohexane-1-sulfonic acid Heptadecafluorooctane-1-sulfonic acid

CF3CF2SO3H CF3(CF2)2SO3H CF3(CF2)3SO3H CF3(CF2)4SO3H CF3(CF2)5SO3H CF3(CF2)7SO3H

200.1 150.1 300.1 350.1 400.1 500.1

Aminosulfonic acids Sulfanilic acid Taurine

p-H3NþC6H4SO3 H3NþCH2CH2SO3

173.2 125.1

Methanesulfonic acid is primarily manufactured by the batch or continuous oxidation of the methyl mercaptan or dimethyl disulfide with chlorine in saturated aqueous hydrochloric acid.[3–5,8,9] This chemistry is also the basis for much of the worldwide production of MSC, with photochlorination of methane (vide infra) being the most significant commercial alternative. Other alkanesulfonyl chlorides and sulfonic acids have also been produced in lesser quantities by the Cl2=H2O oxidation. Reaction yields are typically 92–100%. The HCl by-product separates as vapor from the saturated reaction mixture. It is absorbed into water to afford a

Melting point ( C) 40 34 (1H2O)

Boiling point ( C=torr)

CAS Reg No.

162=760 217=760 (1H2O) 178=760 196=760 210–212=760 224–226=760 238–239=760 120=3

1493-13-6

288 317 (dec)

354-88-1 423-41-6 375-73-5 2706-91-4 355-46-4 1763-23-1 121-57-3 107-35-7

concentrated hydrochloric acid coproduct. RSH þ 3Cl2 þ 2H2 O

ð13Þ

1035 C

ƒƒƒƒƒƒ! RSO2 C1 þ 5HC1 aqueous HC1

>85 C

RSO2 C1 þ H2 O ƒƒƒƒƒƒ! RSO3 H þ HC1 aqueous HC1

ð14Þ

When the desired product is the alkanesulfonyl chloride RSO2Cl, reaction, temperatures are maintained at 10–35 C to inhibit hydrolysis of the poorly soluble sulfonyl chlorides. Extraction with clean

Table 5 Physical properties of selected sulfonyl halides and anhydrides Name

Melting point ( C)

Formula

MW

Methanesulfonyl anhydride

(CH3SO2)2O

174.2

Methanesulfonyl fluoride

CH3SO2F

98.10

Methanesulfonyl bromide

CH3SO2Br

159.0

MSC

CH3SO2Cl

114.6

Ethanesulfonyl chloride

CH3CH2SO2Cl

128.6

1-Propanesulfonyl chloride

CH3(CH2)2SO2Cl

142.6

44

Isopropanesulfonyl chloride

(CH3)2CHSO2Cl

142.6

1-Butanesulfonyl chloride

CH3(CH2)3SO2Cl

156.6

47

69–70

Boiling point ( C=torr)

CAS Reg. No.

125=4

7143-01-3

124=760

558-25-8 41138-92-5

33

161=730 46–47=9

124-63-0

51=4

594-44-5

50–51=4 75=13

10147-36-1

59–61=11

10147-37-2

81=9

2386-60-9

tert-Butanesulfonyl chloride

(CH3)3CSO2Cl

156.6

95–95.5

80=15

10490-22-9

Octanesulfonyl chloride

CH3(CH2)7SO2Cl

212.7

14.6

122=4

7795-95-1

Octanesulfonyl fluoride

CH3(CH2)7SO2F

196.3

Fluorinated organic sulfonyl halides and anhydrides Trifluoromethanesulfonyl anhydride Trifluoromethanesulfonyl chloride Trifluoromethanesulfonyl fluoride Nonafluorobutanesulfonyl fluoride Perfluorooctanesulfonyl fluoride

(CF3SO2)2O CF3SO2Cl CF3SO2F CF3(CF2)3SO2F CF3(CF2)7SO2F

282.1 168.5 152.1 302.1 502.1

40630-63-5

81–83=745 29–32.9=760 23=760 65=760 154–155=760

358-23-6 421-83-0 335-05-7 375-72-4 307-35-7

T

3108

Thiochemicals: Mercapto Acids and Organosulfur (IV/VI) Compounds

Table 6 Physical properties of selected sulfonamides Name Methanesulfonamide

Formula CH3SO2NH2

MW 95.1

Melting point ( C)

Boiling point ( C=torr)

92

CAS Reg No. 3144-09-0

Methanesulfonimide

(CH3SO2)NH

173.2

155

N-Methyl methanesulfonamide

CH3SO2NHCH3

109.2

3

5347-82-0

N-Phenyl methanesulfonamide

CH3SO2NHC6H5

171.2

93–97

1197-22-4

Ethanesulfonamide

CH3CH2SO2NH2

109.2

62

1520-70-3

Propane-1-sulfonamide

CH3(CH2)2SO2NH2

123.2

56–58

Propane-2-sulfonamide

(CH3)2CH2SO2NH2

123.2

67.5

Benzenesulfonamide

C6H5SO2NH2

157.2

150–152

N-Butyl benzenesulfonamide

C6H5SO2NHC4H9

213.3

153=15

105–106=0.1

1184-85-6

24243-71-8 98-10-2

314=760

3622-84-2

Trifluoromethanesulfonamide

CF3SO2NH2

149.1

117–119

421-85-2

N-Trifluoromethanesulfonyl trifluoromethanesulfonimide

(CF3SO2)2NH

281.1

46–57

82113-65-3

N-Phenyl trifluoromethanesulfonamide

CF3SO2NHC6H5

225.2

67

456-64-4

Nonafluorobutanesulfonamide

CF3(CF2)3SO2NH2

299.1

30334-69-1

Perfluorooctanesulfonamide

CF3(CF2)7SO2NH2

499.2

754-91-6

concentrated HCl removes any H2SO4 or RSO3H impurities, while residual HCl and water are removed by evaporation under vacuum. Alternately, the sulfonyl chloride can be vacuum distilled to separate the impurities. The alkanesulfonic acids are produced at elevated temperatures to ensure complete hydrolysis of the intermediate RSO2Cl, with residual HCl being stripped from the reaction product. Methanesulfonic acid and other alkanesulfonic acids are typically obtained as 60–70% aqueous solutions. Similarly, isolated alkanesulfonyl chlorides can be hydrolyzed to afford the aqueous or anhydrous sulfonic acids.[10] Anhydrous nonfluorinated alkanesulfonic acids can also be prepared by evaporative removal of water from the aqueous acids.[11] Nitric acid has also been shown to be an effective oxidant for converting thiols or disulfides to the corresponding sulfonic acids.[12,13] When this reaction is performed in the presence of HCl at low temperature, the sulfonyl chlorides can also be obtained. The use of nitric acid oxidant has traditionally resulted in higher levels of overoxidation, affording H2SO4 as an impurity. In electroplating applications, the major market for MSA, the specification for H2SO4 is typically less than 150 ppm and occasionally much lower. In one recently launched commercial MSA process involving nitric acid oxidant, the H2SO4 levels in the product were minimized by performing the reaction in two sequential reactors, the first operating at 80–120 C where the bulk of the conversion occurs, the second at 130–150 C. Residual HNO3 and NOx were removed by evaporative stripping. Distillation of the MSA then afforded the final product with low

impurities. In this process the HNO3 was regenerated from the nitric oxide by-product by reaction with air. Similarly, the use of atmospheric O2, catalytic HNO3= NOx, and Br=Br2cocatalyst to oxidize aliphatic thiols and disulfides to alkanesulfonic acids and alkanesulfonyl chlorides has also been described.[14] 1 5 RSSR þ HNO3 2 3 80150 C

ƒƒƒƒƒƒ! RSO3 H þ H2 O

5 1 NO þ H2 O 3 3

ð15Þ

Alkanesulfonic acids can be prepared from the thiols or disulfides and air using cocatalytic DMSO and HBr.[15] The use of aqueous DMSO in the absence of air, but in the presence of a halogen or hydrogen halide catalyst, readily converts most aliphatic, aromatic, and heterocyclic thiols or disulfides to the corresponding sulfonic acid.[16,17] In effect, DMSO oxidizes the hydrohalide to the molecular halogen, which then reacts with the organosulfur substrates. Water serves as a proton and oxygen source, and inhibits the Pummerer-type decomposition of the DMSO. Other oxidants for the conversion of alkanethiols and disulfide alkanesulfonic acids have been described. H2O2, in combination with catalysts, e.g., peroxycarbonates, alkanesulfonic acids, molybdates, tungstates, or HCl, was effective in producing alkanesulfonic acids.[18,19] At lower temperatures (25–35 C) the combination of H2O2 catalytic HCl afforded the alkanesulfonyl chlorides. Kinetic studies of the photooxidation with O2 in acetonitrile and other solvents have been reported and the relative stabilities of the various

Thiochemicals: Mercapto Acids and Organosulfur (IV/VI) Compounds

oxysulfur intermediates have been evaluated by ab initio methods.[20,21] Electrolytic oxidation of alkanethiols and disulfides has also been described.[22] RSH þ 3ðCH3 Þ2 SO 100130 H2 O

ƒƒƒƒƒƒƒƒƒ! RSO3 H þ 3ðCH3 Þ2 S catalyst HX or X2

2HX þ ðCH3 Þ2 SO ! X2 þ H2 O þ ðCH3 Þ2 S

ð16Þ

ð17Þ

3109

the Strecker reaction. It is currently used in the commercial production of disodium methanedisulfonic acid from CH2Cl2. Similarly, metal hydrogen sulfite also reacts with epoxides to afford 2-hydroxyethanesulfonates, the basis for the production of sodium isethionate (sodium 2-hydroxyethanesulfonate) used in the manufacture of surfactants. Free-radical addition of metal hydrogen sulfites to alkenes or alkynes is an alternate route to 2-alkyl- or 2-aryl-substituted ethanesulfonates or to geminal disulfonates.

Alkanesulfonic acids and alkanesulfonyl chlorides from aliphatic hydrocarbons Sulfoxidation of saturated hydrocarbons with SO3, or SO2 with O2, is an effective but nonselective method for producing alkanesulfonic acids, invariably producing a mixture of alkanesulfonic acid products.[23] Despite the poor selectivity, this procedure has been used extensively in the manufacture of long-chain alkanesulfonate salts as surfactants since the 1940s. Methane is a particularly nonreactive hydrocarbon and early efforts to sulfoxidate CH4 using mercury salt catalysts afforded mixtures of MSA, methanedisulfonic acid, and methyl esters.[24] More recent efforts have demonstrated highly selective sulfoxidations of methane with SO3, or air and SO2, to afford MSA, as well as a related sulfochlorination of methane with SO2Cl2 to make MSC.[25–28] Photochemical sulfochlorination of saturated hydrocarbons has been used in the production of long-chain alkanesulfonyl chlorides, again as precursors to the salts as surfactants. These processes involve free radical intermediates. Analogous to the sulfoxidation of long-chain alkanes, these processes also provide mixtures of alkanesulfonyl chloride and chloroalkane products.[29] Nonetheless, pure products can be obtained for short C1–C3 alkyl chain lengths. A highpressure gas phase process for the reaction of CH4 with Cl2 and excess SO2 is used in one commercial manufacturing route to MSC.[30] hv 32 C

CH4 þ Cl2 þ SO2 ƒƒƒƒƒ! CH3 SO2 Cl þ HC1 gas phase

ð18Þ

Alkanesulfonate salts from sulfite and alkyl halides, alkyl sulfate, olefins, or epoxides Alkyl halides, alkyl sulfonates, and alkyl sulfates undergo nucleophilic displacement by aqueous sulfite to afford sulfonic acid salts under very mild conditions.[3–5] This chemistry is traditionally referred to as

Sultones from olefins Olefins react with SO3 under free-radical conditions to afford cyclic sulfonate esters, i.e., sultones.[3–5] The initially formed products are believed to be the highly strained 1,2-adducts, which then rearrange to the 1,3adducts. This is the commercial route to propanesultone.

Aromatic sulfonic acids from aromatic hydrocarbons The traditional manufacturing processes for making aromatic sulfonic acids involve sulfonation of an aromatic hydrocarbon (or polymers) by falling film technology using vaporized SO3 in air. Alternate sulfonating agents include oleum (H2SO4 þ SO3), ClSO3H, SO3 in solvents, and SO2 with air. A complete description is beyond the scope of this entry, and for a more thorough overview interested readers should see Ref.[31]. The industrial preparation of PTSA illustrates one approach. Toluene is reacted with SO3 in liquid SO2 in a falling film reactor.[32] The final reaction mixture is treated with water to separate unreacted toluene and convert residual SO3 to H2SO4. The PTSA

T

3110

Thiochemicals: Mercapto Acids and Organosulfur (IV/VI) Compounds

product is then isolated by crystallization and drying.

Fluorinated alkanesulfonic acids Perfluoroalkanesulfonyl fluorides and of related materials are manufactured by the electrochemical fluorination of the corresponding alkanesulfonyl fluorides or cyclic sulfones.[3–5] This electrochemical synthesis results in replacement of all C–H bonds in the feedstock. The perfluorinated alkanesulfonyl fluoride products are neutralized to make the anhydrous salt, then acidified and distilled to afford the anhydrous sulfonic acid. Alkyl chain degradation in the electrochemical cell becomes more pronounced at longer chain lengths. Addition of short-chain divalent sulfur compounds (e.g., thiols, sulfides) to the cell inhibits buildup of tarry materials and loss of efficiency. ECF

C8 H17 SO2 F þ 17HF ! C8 F17 SO2 F þ 17H2

sulfonyl chloride. The sulfonamide products are very polar materials and separation from the by-product salts is invariably a key processing issue. The general approach is to choose a reaction solvent with good solubility for the sulfonamide and poor solubility for the by-product salt, thus permitting removal of the salt by filtration. Such solvents have included nitroalkanes, toluene, chlorobenzene, tetrahydrofuran, other ethers, and nitriles.[33] Running the reaction in water and recovery of the alkanesulfonamide by extraction have also been demonstrated.[34] Alternately, the separation of aqueous mixtures of alkanesulfonamides and ammonium chloride by electrodialysis is also feasible.[35] Alkanesulfonate esters are similarly prepared from the alkanesulfonyl chloride, the appropriate alcohol, and base.[36] Symmetrical N-sulfonyl sulfonamides, i.e., sulfonimides, can be prepared from the amine and two equivalents of sulfonyl chloride and base. The unsymmetrical N-sulfonyl sulfonamides and the related N-acyl sulfonamides can be obtained by the reaction of a sulfonamide or carboxamide with an acid halide.

2RSO2 X þ R0 NH2 þ 2 Base ! ðRSO2 ÞNR0 þ 2 Base  HX

ð26Þ

ð24Þ Aminoalkanesulfonates

Sulfonamides, sulfonimides, and sulfonate esters from sulfonyl halides Aliphatic and aromatic sulfonamides are prepared from the corresponding sulfonyl halide (RSO2X) and ammonia, 1 - or 2 -amines.[3–5] The reaction requires alkaline conditions and thus a stoichiometric amount of additional base to neutralize the by-product hydrohalide (see below Eq. (27)). This base may be a second molecule of the reactive amine, a 3 -amine, or an inorganic base. The reaction can be performed in organic or aqueous solvents, although excessive temperatures must be avoided in water to prevent hydrolysis of the

The 2-aminoethanesulfonates (i.e., taurines) are the prepared reaction of ammonia or 1 - or 2 -amines with metal 2-haloalkanesulfonate, vinylsulfonate, or hydroxyethanesulfonate salts under alkaline conditions.[37,38] The aminoethanesulfonate salts are then converted to the zwitterionic acids by cation exchange or acidification. 3-Aminopropanesulfonic acids are directly prepared by the reaction of propanesultone with the amine.

Thiochemicals: Mercapto Acids and Organosulfur (IV/VI) Compounds

Interconversion of Sulfonic Acid Derivatives A variety of procedures have been identified for the interconversion of sulfonyl derivatives. Because of their high acidity, sulfonic acids can readily be neutralized with inorganic bases or organic amines to form salts. Conversion of metal salts back to the acid form is often desirable in synthetic or electroplating applications. Direct acidification of alkali metal sulfonates with HCl affords the sulfonic acid with precipitation of the formed metal halide salt. An alternative approach is to convert the salt back to the acid by electrodialysis.[39] Recovery and recycle of MSA from spent metal plating baths have been described.[36] Sulfonic acids and alkali metal salts can be readily converted to the sulfonyl chlorides using a variety of reagents, including thionyl chloride, phosgene, phosphorous pentahalides, or phosphorous oxychloride. Sulfonyl bromides can be obtained from the corresponding chlorides by sequential reaction with alkaline Na2SO3 and Br2. Sulfonyl fluorides are obtained from sulfonyl chlorides by reaction with aqueous KF. Aromatic sulfonyl chlorides can be obtained from the sulfonamides by reaction with PCl5, but this is not applicable to the aliphatic analogs. Applications Sulfonic acids and their derivatives are used in innumerable industrial applications in chemical synthesis, electroplating of metals, surfactants, ion-exchange resins, and preparation of dyes, animal feeds, pesticides, and pharmaceuticals. Methane sulfonic acid is used as an electrolyte for electroplating of tin onto sheet steel, for plating tin and tin=lead alloy onto nickel or other base metal substrates in the manufacture of lead frames and bumpcontacts for microelectronic devices.[36] It can also be used for copper deposition during the manufacture of microprocessors. Other alkanesulfonic acids have also found use in electroplating applications. Disodium methanedisulfonate and other alkanedisulfonate salts are used in chrome plating.[40] As discussed previously, several processes for the recovery and recycle of alkanesulfonic acids from spent metal plating baths have been described. Because of their high acidity, sulfonic acids are extensively used as Bronsted acid catalysts for esterification,

3111

aromatic alkylation, and etherification. The primary sulfonic acids in these applications include MSA, PTSA, trifluoromethanesulfonic acid (triflic acid), and nonafluorobutanesulfonic acid (nonaflic acid), as well as the aromatic and fluorinated alkanesulfonic acid resins. In these applications, they compete against the traditional strong inorganic acids such as H2SO4, HCl, or HF. Triflic acid and MSA are both liquids and are easily handled in a manufacturing environment. This lends these materials to use as both acid catalyst and reaction solvent.[41] Methane sulfonic acid bound to poly(vinylpyridine) has been shown to be an effective esterification catalyst that can be readily recovered and recycled.[42] Because of its much higher cost, the commercial uses of triflic acid are generally limited to reactions where its higher acidity is a requirement, e.g., the potential replacement for HF or H2SO4 in paraffin alkylation in petroleum refining. p-Toluenesulfonic acid is a solid at ambient temperatures and poorly soluble in some organic matrices at low temperature. Thus, it is recyclable by filtration. In addition to its synthetic utility, MSA is employed as an amine-salt former in pharmaceutical final dosage forms. Other sulfonic acids are also used as pharmaceutical salt formers, but much less frequently. The major commercial application of aromatic sulfonic acid salts is anionic surfactants.[43] These are predominately linear long-chain alkyl benzene sulfonates and the naphthalene analogs, or a-olefinsulfonates. Short-chain alkylarenesulfonates are used in liquid detergent formulations as coupling agents, solubilizers, and hydrotropes where high concentrations of organic surfactants and inorganic compounds must be kept in aqueous solution. Fatty acid esters of sodium isethionate are mild surfactants unaffected by hard water, but with limited hydrolytic stability. Cocoyl isethionate is the principle ingredient in detergent bars for personal use. Aromatic sulfonic acid and perfluoroalkanesulfonic acids resins are widely used as ion-exchange resins in water treatment and multiple other industrial applications. In the form of membranes, they are routinely used in electrochemical cells, particularly in electroplating of metals and in battery applications. The lithium salts of trifluoromethanesulfonic acid and Ntrifluoromethanesulfonyl trifluoromethanesulfonamide are both employed as electrolytes in secondary battery applications. Taurine is an essential dietary nutrient in felines and is routinely added to packaged food for domestic cats.[3–5] Other aminoalkanesulfonic acids, e.g., N-2-hydroxyethylpiperazine-N 0 -2-ethanesulfonic acid (HEPES), 2-[Nmorpholino]ethanesulfonic acid (MES), or 3-(n-morpholino)propanesulfonic acid (MOPS), are commonly referred to as biological buffers or as ‘‘Good’s’’ buffers.

T

3112

These are intensively used in fermentation and protein separation.[37,38] Sulfonyl chlorides have extensive uses in organic synthesis in the preparation of sulfonamides and sulfonate esters. Methanesulfonyl chloride is a key raw material in the synthesis of critical components for photographic color developing formulations, as well as for herbicides and pharmaceuticals. Methanesulfonamide and N-methyl methanesulfonamide are used as chemical intermediates in the manufacture of agrochemical herbicides and fungicides.[44,45] Methyl methanesulfonamide is also used in the synthesis of the anticholesterol agent rosuvastatin.[46] N-Butyl benzenesulfonamide serves as a plasticizer for polyamide resins.

CONCLUSIONS Thiochemicals are among the important intermediates and end-user chemicals, varying from relatively simple molecules, e.g., H2S, to complex molecules, e.g., polysulfides. Their manufacturing routes vary and depend highly on their properties and quantity of interest. Their intermediate applications include the manufacturing of agricultural products, pharmaceuticals, detergents, lubricants, etc. Some thiochemicals have played important roles in solvents, polymer synthesis, electroplating, presulfiding of hydrotreating catalysts, coke prevention in steam cracking process, etc.

ACKNOWLEDGMENT The authors acknowledge the support and information provided by Arkema Inc.

REFERENCES 1. Fisch, M.H.; Bacalogulu, R.; Biesiada, K.; Brecker, L.R. Mechanism of organotin stabilization of poly(vinyl chloride). 1. The structure and equilibria of alkyltin alkyl mercaptopropionates and their compatibility with PVC. In Plastics: Plastics on My Mind, Proceedings Antec 1998 Brookfield, Society of Plastic Engineering; Vol. 3, 33291. 2. Fisch, M.H.; Bacalogulu, R.; Biesiada, K.; Brecker, L.R. Mechanism of organotin stabilization of poly(vinyl chloride). 2. Significance for PVC stabilization of structure and equilibria of alkyltin alkyl thioglycolates. In Plastics: Bridging the Millenia, Proceedings Antec 1999, Society of Plastic Engineering; Vol. 3, 33296.

Thiochemicals: Mercapto Acids and Organosulfur (IV/VI) Compounds

3. Patai, S., et al. The Chemistry of Sulfonic Acids, Esters, and Their Derivatives; Patai, S, Rappoport, Z., Eds.; John Wiley and Sons: New York, 1991. 4. Patai, S., et al. Supplement S–The Chemistry of Sulfur Containing Functional Groups; Patai, S., Rappoport, Z., Eds.; John Wiley and Sons: New York, 1993. 5. Patai, S., et al. The Chemistry of Sulphones and Sulphoxides; Patai, S., Rappoport, Z., Eds.; John Wiley and Sons: New York, 1988. 6. Wypych G., Ed.; Knovel Solvents: A Properties Database; ChemTec Publishing: Toronto, Ontario, 2000. 7. Flick, E.W. Ed.; Organic sulfur compounds. In Industrial Solvents Handbook, 5th Ed.; Noyes Data Corp.: Westwood, NJ, 1998; 222–237. 8. Guertin, R.M Method of Preparing Alkyl Sulfonyl Chloride. U.S. Patent 3,626,004, Dec 7, 1971. 9. Guertin, R.M Method of Preparing Alkane Sulfonic Acids. British Patent 1,350,328, Apr 18, 23, 1974. 10. Ollivier, J.; Lagaude, C.H.; Baptiste, H.; Larrouy, M Process for the Production of Alkanesulphonic AcidsU.S. Patent 4,859,373, Aug 22, 1989. 11. Comstock, P.; Keys, K.M Preparation of Anhydrous Alkanesulfonic Acid. U.S. Patent 4,938,846, Jul 3, 1990. 12. Eiermann, M.; Tragut, C.; Ebel, K Method of Producing Alkanesulfonic Acid. U.S. Patent 6,531,629, Mar 11, 2003. 13. Bechtel, M.; Thiel, J Improved Liquid Distributor for Distillation Columns Consisting of Glass. German Patent DE 10139587, Feb 20, 2003; Chem. Abstr. 138, 155467. 14. Chen, J Preparation of Alkane Sulfonic Acids and Sulfonyl Chlorides by Oxidation of Alkanethiols and Dialkyl Disulfides. U.S. Patent 6,124,497, , Sep 26, 2000. 15. Husain, A.; Wheaton, G.A Production of Alkanesulfonic Acids by Oxidation of Alkanethiols or Dialkyl Disulfides in the Presence of DMSO. European Patent 424,616, Feb 2, 1991; Chem. Abstr. 115, 8085. 16. Toland, W.G Bromide Ion Promoted Oxidation of Sulfide-Sulfur by Lower Dialkyl Sulfoxides. U.S. Patent 3,428,671, Feb 18, 1969. 17. Lowe, O.G Oxidation of Thiols and Disulfides to Sulfonic Acids. U.S. Patent 3,948,922, Apr 6, 1976. 18. Schreyer, G.; Geiger, F.; Hensel, J A Process for the Production of Alkyl Sulfonic Acids. U.S. Patent 4,239,696, Dec 16, 1980 (and cited references therein). 19. Husain, A.; Wheaton, G.A. Oxidation of Thiols, Disulfides, and Thiolsulfonates. U.S. Patent 4,956,494, Sep 11, 1990.

Thiochemicals: Mercapto Acids and Organosulfur (IV/VI) Compounds

20. Robert-Banchereau, E.; Lacombe, S.; Ollivier, J.; Micheau, J.C.; Lavabre, D. Kinetic modeling of the photooxidation of dimethyldisulfide in the liquid phase. Int. J. Chem. Kinet. 1997, 29 (11), 825–834. 21. Lacombe, S.; Loudet, M.; Dargelos, A.; RobertBanchereau, E. Oxysulfur compounds derived from dimethyl disulfide: an ab-initio study. J. Org. Chem. 1999, 63, 2281–2291. 22. Gardner, D.M.; Wheaton, G.A Preparation of Alkanesulfonyl Halides and Alkanesulfonic Acids. U.S. Patent 5,035,777, Jul 30, 1991 (and references therein). 23. Bost, H.W Sulfoxidation Process. U.S. Patent 3,413,337, Nov 26, 1968. 24. Snyder, J.C.; Grosse, A.V Reaction of Methane with Sulfur Trioxide. U.S. Patent 2,493,038, Jan 3, 1950. 25. Basickes, N.; Hogan, T.E.; Sen, A. Radicalinitiated functionalization of methane and ethane in fuming sulfuric acid. J. Am. Chem. Soc. 1996, 118, 13111. 26. Mukhopadhyay, S.; Bell, A.T. A High-yield approach to the sulfonation of methane to methanesulfonic acid initiated by H2O2 and a metal chloride. Angew. Chem. Int. Ed. 2003, 42, 2990– 2993 (and references cited therein). 27. Mukhopadhyay, S.; Zarella, M.; Bell, A.T.; Srinivas, R.V.; Smith, G.S. Synthesis of methanesulfonyl chloride (msc) from methane and sulfuryl chloride. Chem. Commun. 2004, 472–473. 28. Richards, A.K Anhydrous Conversion of Methane and Other Alkanes into Methanol and Other Derivatives Using Radical Pathways and Chain Reactions with Minimal Waste Products. World Patent Application 2004=041399A2, May 21, 2004. 29. Hertel, O.; Schlecht, H.; Schneider, R. Production of Substituted Alkanes. U.S. Patent 3,911,004, , Oct 7, 1975. 30. Ollivier, J.; Baptiste, H.; Laqaude, C.H.; Larrouy, M Proce´de´ et Appareil pour la Sulfochloration Photochimique d’Alcanes Gazeux. European Patent Application 194,931, Mar 14, 1985; Chem. Abstr. 106, 20289. 31. Lindner, O.; Rodefeld, L. Benzenesulfonic acids and their derivatives. In Ullmanns Encyclopedia of Industrial Chemistry; John Wiley and Sons: New York, 2002. 32. Wu, J.C.; Wang, B.H.; Zhang D.L. Song, G.F.; Yuan, J.T.; Liu, B.F. Production of p-toluenesulfonic acid by sulfonating toluene with gaseous

3113

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

44.

45.

46.

sulfur trioxide. J. Chem. Technol. Biotechnol. 2001, 76, 619–623. Smith, G.S.; Cordova, R.; Overgaard, T.H.; Budrick, M.T.; Brown, S.M Preparation of alkanesulfonamides with low residual ammonium impurities, U.S. Patent 5,599,983, Feb 4, 1997 (and references cited therein). Brown, S.M.; Muxworthy, J.P.; Gott, B.D Process for the Production of Sulfonamides. World Patent Application WO 98=25890, Nov 27, 1997 (and citations referenced therein). Gancet, C.; Lauranson, D.; Perie, F Desalination of aqueous sulphonamide solutions. U.S. Patent 6,036,830, Mar 14, 2000. Gernon, M.D.; Wu, M.; Buszta, T.; Janney, P. Environmental benefits of methanesulfonic acid: comparative properties and benefits. Green Chem. 1999, 127–140 (and references therein). Good, N.E.; Winget, G.D.; Winter, W. Hydrogen ion buffers for biological research. Biochemistry 1974, 5, 467–477. Good, N.E.; Izawa, S. Hydrogen ion buffers. In Photosynthesis; Methods Enzymol; San Pietro, A., Ed.; 1971; Vol. 24B, 53–68. Gavach, C.; Gancet, C.; Mirassou, A.; Perie, F Regeneration of Acids, Particularly Strong Organic Acids, Using Bipolar Membranes. U.S. Patent 5,993,629, Nov 30, 1999. Newby, K.R Protection of Lead-Containing Anodes During Chromium Electroplating. U.S. Patent 5,176,813, Jan 5, 1993. Kieczykowski, G.R Process for Preparing 4Amino-1-Hydroxybutylidene-1,1-Bisphosphonic Acid (ABP) or Salts Thereof. U.S. Patent 5,019,651, May 28, 1991. Gancet, C Preparation of Esters of Carboxylic Acids Directly From Carboxylic Acids and Alcohols Using a Catalyst System Comprising a Sulfonic Acid and a Polymer-Bound Tertiary Amine. European Patent 1,167,337, Jan 2, 2002. Lynn, J.L.; Bory, B.H. Surfactants. In Kirk-Othmer Encyclopedia of Chemical Technology; 1998 Electronic Release; Wiley–VCH: Weinheim, Germany, 1997. Schlegel, G Preparation of Pyrimidyl-Substituted Sulfonylureas. European Patent Application EP 560178, Sep 15, 1993; Chem. Abstr. 120, 8613. Madsen, H.B.; Klemmensen, H.K.-A N-Phenyl or N-Pyridyl Sulfamides and Methanesulfonamides. . U.S. Patent 4,148,901, Apr 10, 1979. Hirai, K.; Ishiba, T.; Koike, H.; Watanbe, M Pyrimidine Derivatives. U.S. Patent 5,260,440, Nov 9, 1993.

T

Tissue Engineering T Shang-Tian Yang Clayt Robinson Department of Chemical and Biomolecular Engineering, The Ohio State University, Columbus, Ohio, U.S.A.

INTRODUCTION Tissue engineering is a combination of biology and engineering for producing biological substitute structures that can reconstitute cellular or tissue function, which is lost, declining, or insufficient. Theorized for a long time, the field has evolved greatly within recent decades to meet the challenges of disease and a limited organ donor supply. For tissue substitutes to be successful in replacing lost function, they must accurately and reliably perform the desired function that is dependent on a multitude of variables. Parameters that must be carefully investigated include the cell source, the polymeric scaffold support, the tissue culture protocol, and the implantation procedure. Tissue engineering research covers a wide range of applications including many tissue substitutes, cell therapy, and diagnostic modeling. Although faced with as yet unmet challenges, tissue engineering stands to provide momentous advancements to health in the 21st century. First coined in 1987, tissue engineering, as a field, is defined by two main objectives. The first objective is to apply methods and principles of engineering and life sciences to understand the biological tissue construct.[1] In addition to deciphering the vast catalog of cellular functions, the interconnections of a cellular population, regulated by signaling pathways, and the interactions of a cellular composite with the noncellular surrounding environment must be understood. Since the threedimensional structure and organization of tissue components are integral in defining proper function of the tissue, this environmental arrangement must be mimicked to produce a tissue substitute that functions appropriately. A fundamental understanding of a tissue environment structure and how it enables or affects the tissue function is essential. The second objective is to apply established knowledge of the tissue construct toward developing biological substitutes to restore, maintain, or improve natural function to tissues or organs that are structurally or physiologically altered.[1] Included in this entry is a review of the development of tissue engineering, from theorized concepts and early experiments through advancements made in developing tissue substitutes in the recent past. A review of biological systems and components precedes Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120006824 Copyright # 2006 by Taylor & Francis. All rights reserved.

a thorough discussion of the components utilized in tissue engineering constructs. The entry concludes with tissue engineering applications and the challenges that remain for full realization of the field’s potential.

MOTIVATION Motivation for further development in the field of tissue engineering is to save the lives of patients suffering from organ failure or loss, who are waiting for donor organs to become available, and those suffering from debilitating diseases, such as Parkinson’s disease, in which essential cell or tissue function deteriorates or is lost over time. Further, people are victims to catastrophic events in which tissue repair or replacement is a grave concern with, commonly, no immediate remedy. As a result, the annual health care costs in the United States alone exceed $400 billion for patients with tissue loss or organ failure. The cost includes 8 million surgical procedures and 40–90 million hospital days annually.[2] In 2002, the waiting list for organ recipients exceeded 81,700 candidates, with over 38,600 people added to the list. However, only 12,800 organs were transplanted in 2002 owing to a severe shortage of donors.[3] With a limited organ supply, accumulating costs, and limited medical procedures available, there is a void to fill.

FROM FICTION TO REALITY There are several examples of tissue engineering well before the field was named in 1987. Stories that include the replacement of a person’s body parts with those from another person are found in literature centuries old. Possibly, the first tale of a tissue engineering procedure is found in the Bible, when Eve is given life from the rib of Adam. In 1818, Mary Shelley wrote Frankenstein in which the title character is given life through the compilation of body parts garnered from donor corpses. In the legend of St. Cosmas and St. Damien, about 200 A.D., the two physicians performed a procedure in which the gangrenous leg of a 3115

3116

man was successfully replaced with the leg of a recently deceased man.[4] More recently, Hollywood has produced films that portray ideas from tissue engineering. In 1991, the aptly titled Body Parts was released, in which a man loses his arm in an accident and receives a transplant from a recently executed inmate. Although successfully transplanted, incompatibility becomes an issue as the transplanted arm retains the murderous personalities of its donor. The collection of stories that involve tissue engineering concepts shows the promise and spectacular possibilities that the future could bring. However, rarely do these stories fully conceptualize or even mention the challenges involved in performing these acts in reality. Until the advancement of tissue engineering in the 1970s, replacements for bodily tissues were prostheses made of wood, ceramics, and plastics. Replaced body parts included arms, legs, eyes, ears, teeth, and noses. Experimentation with animals became more common for studying the growth of tissue. In the 1930s, the work done by Bisceglie became one of the earliest documented tissue engineering procedures.[2] Two important concepts were displayed including the transplantation of mouse tumor cells into the abdomen of a pig, while the cells were encased within a polymer membrane structure. First, the results showed that cells could survive within a foreign environment without rejection by the host immune system. Second, the procedure introduced the encapsulation approach. The semi-permeable polymer membrane system allowed for nutrient and waste fluxes into and out of the membrane system, respectively. Concurrently, the membrane selectively rejected the passage of immune system molecules and proteins, thus protecting the enclosed cells. In the work performed by Chick and coresearchers, islet cells were encapsulated within a semi-permeable membrane and transplanted into animal models to provide glucose level control as a cure for diabetes.[2] In the laboratory setting, the first tissue to be reconstituted was skin because of its relatively simple two-dimensional structure. During the late 1970s and 1980s, artificial skin was created using skin cells distributed within natural collagen or collagen– glycosaminoglycan composite support structures. The growth of tissue engineering in the 1990s and early 2000s is due to interdisciplinary advancements in the fields of engineering, genomics, proteomics, cell biology, and material science.[5] As understanding of the important relationship between tissue structure and function became fully realized, three-dimensional synthetic polymers were utilized to mimic the bodily in vivo environment as a support for cells to attach and grow on. The liver was the first tissue to be cultivated using these three-dimensional constructs owing to its relatively simple composition. Utilizing these newly

Tissue Engineering

developed tools, a multitude of tissues have been or are being studied and mimicked with tissue-engineered products for possible future usage in a wide array of applications that is discussed later in this chapter.

CELLULAR SYSTEMS BIOLOGY For a tissue substitute to function properly, many biological aspects of the tissue and component cells must be understood. Some aspects include the extracellular matrix, cell-specific gene expression and surface markers, cell growth parameters, population arrangement and behavior, and the immune system. Tissues and Organs An organ is a component of the human body system made of one or more tissue types and has specific jobs within the complex network. Tissue is a collection of similar cells and the surrounding supportive environment that together perform specific tasks. There are four basic categories of tissues: epithelium, which constitutes surfaces such as skin, connective tissue, muscle tissue, and nerve tissue.[4] Extracellular Matrix Tissue in vivo consists of cells that are engaged in an environment, called the extracellular matrix (ECM), that provides support to allow for proper cell function, cell–scaffold interactions, and tissue morphology. The ECM directly affects or controls cell shape, function, viability, and population structure. The ECM that supports the tissue structure and cells in vivo is a complex network of collagens, glycoproteins, such as fibronectin and laminin, hyaluronic acid, proteoglycans, glycosaminoglycans, and elastins to which cells adhere and interact. The three-dimensional ECM has a two-way interaction with the cells. The ECM surface properties and molecules provide cell-surface receptormediated signals to influence cellular spatial organization, migration, growth, differentiation, and death.[6] Important cell receptors that interact with the ECM include integrin and cadherin adhesion receptors.[7] The cells influence the ECM by remodeling the structure and secreting new ECM components.[8] Genes and Proteins The genes on chromosomes within each cell are the blueprints of the human body. The pattern in which the genes are expressed determines the cell type and behavior. Gene expression, which leads to protein

Tissue Engineering

production, is a highly controlled process dependent on signals originating from other cellular components within the same cell, externally from other cells within the immediate environment, or from distant locations, such as part of the endocrine system. External signals are transduced, or passed, with the aid of surface molecules that are receptors for the signal molecules. Other molecules and proteins are expressed to perform cellspecific functions within the cell, or externally following secretion. The presence of these surface receptors or expressed functional proteins is characteristic of the cell type and can be used as cell markers for identifying proper cell function.

3117

and mutated cancerous cells are targeted by the immune system, the resulting immune rejection mechanism kills, disrupts, or encloses the foreign object to prevent further harm to the body. Cells are screened by immune system antibody molecules or T-cells to verify whether the surface molecules are native or foreign. A key determinant of whether a cell is native or foreign is the major histocompatibility complex genes contained within the cell and expressed on the cell surface. Therefore, for cells to be compatible and to avoid immune rejection in a new host, the similarity of the major histocompatibility complex genes must be high. This is important in determining the success of a tissueengineered product within the host.

Cell Growth As cells grow, or proliferate, they proceed through a highly controlled process called the cell cycle. The cell cycle consists of four phases during which the DNA set is accurately replicated and subsequently divided into two complete sets partitioned into two new cells derived from dividing the original cell via mitosis. Regulatory proteins guard progression between phases to ensure cellular readiness and DNA integrity preservation. Errors in this regulatory process can lead to DNA mutations and uncontrolled growth, both characteristics of cancer. With specific signals present, cells exit the cell cycle for a maturation process called differentiation before re-entering the cell cycle. As cells differentiate, their function becomes more defined and limited. Morphology and Arrangement Cellular morphology and population arrangement within the tissue culture environment are dependent on the support surface properties. Cells that are highly proliferating appear round and smooth. As cells adhere to a surface, spreading occurs across the surface, and the cells flatten. As population of cells increases, interaction and rearrangement occur among the cells. In a two-dimensional environment, interactions are limited, but in a three-dimensional environment, cellular aggregates can form within the structure of the supporting material. Aggregation and interaction among cells in all directions mimic the body environment and allows for similar cellular function. Immune System The immune system defends the body from infection and illness using cellular and molecular components to detect and clear objects from the body that are not identified as normal. When objects such as invading viruses, micro-organisms, implanted materials,

TISSUE ENGINEERING CONSTRUCT COMPONENTS As summarized in Fig. 1, the components that must be customized based on the application include cell source, scaffold parameters, and cell culture procedures. Cell Sources The cells utilized for producing a tissue substitute come from multiple sources. When a specific cell type is needed to culture a certain tissue type, the availability of the cell type, the means of obtaining the cells, procedures for maintaining and multiplying the cell population in culture, and immune rejection upon transplantation must all be considered. Avoiding rejection is a huge challenge in the development of a tissue substitute, thus autologous cells obtained from the same person to whom the transplant will be given are ideal as the host body does not reject autologous cells. However, supply of autologous cells is frequently the problem. For example, in many situations, a large enough population of healthy cells is unavailable due to the extent of disease.[9] Further, the harvesting of cells from one location may cause long-term harm to the donor site. Once obtained, autologous cells are cultured, or grown, in a laboratory environment, or in vitro in the presence of a proper support structure and nutrient supply until a larger population is present. The autologous cells are then administered to the donor patient at the necessary site. When autologous cells are not appropriate or available, the cells can come from either another donor of the same species or a different species. The transplantation of cells between similar or dissimilar species is called allotransplantation or xenotransplantation, respectively. Although the supply of these types of grafts is plentiful, immune rejection is very common, so some additional strategy must be utilized to avoid rejection. Cells provided by family members, or

T

3118

Tissue Engineering

Fig. 1 Overview of the components utilized in tissue engineering constructs.

allogeneic cells, tend to be the most compatible source in order to avoid rejection due to similar gene sets, including, specifically, the histocompatibility complex genes. Xenografts provide an additional challenge since xenogeneic cells may contain components that are infectious when introduced into a human. This fact, in addition to ethical and moral issues regarding utilizing nonhuman parts in a human, has led to general unpopularity of this procedure.[9] A fourth cell source is stem cells. Stem cells are characterized by the ability to proliferate indefinitely and develop into different cell types, or pluripotent, depending on the stem cell origin and given the appropriate signals. Embryonic stem cells are present as an embryo first begins to develop and differentiate to form all components of the human body. These stem cells allow for generation of any tissue cell type, however, ethics and regulations limit their usage. An adult retains a limited supply of adult stem cells in the bone marrow and in tissues throughout the body. Most of these progenitor cells are partially differentiated into a lineage of cell types, but remain multipotent to develop into a more limited range of cell types. Examples are neural stem cells, hematopoietic stem cells, and mesenchymal stem cells. Owing to the limited supply of stem cells in the body, specialized techniques are necessary to find and separate them from the mixture. Stem cells may be isolated from an extracted tissue mass by digesting the ECM structure surrounding the cells, and then detecting the cells based on signature biomolecular expression profiles as a ‘‘fingerprint’’ for stem cells. Once identified, the stem cells can be separated from the total tissue population by passing the cells through a system that selectively removes the stem cells while

allowing the remainder of cells to exit the system separately. Selective retention technology systems extract stem cells from the mixed population by customizing the system to have a high affinity for expressed stem cell surface molecules or to secondary molecules previously bound selectively to the cell surface. Example stem cell surface markers are CD34 and CD45 for hematopoietic stem cells and Oct-4 and SSEA-3 for embryonic stem cells. Once harvested, stem cells can be cultured in specific biological, chemical, and physical stimuli to differentiate into the cell type of interest. Upon expansion in culture to a large enough population, the cells may be transplanted as therapy. However, strategies to avoid rejection are necessary unless the cells are autologous or cultured in such a way to disguise the fact that the cells are from a different source.[9] The difficulty lies in providing the correct composition, amounts, and timing of the stimuli to direct the differentiation to the desired cell type. A current approach is to direct the transformation of the stem cells in vitro until the cells are within one or two steps of the complete differentiation destination. The final transformation steps are accomplished in vivo, after transplantation to the site of the desired cell type where signals are provided to complete the differentiation.[10] One concern with stem cell usage is that transplanted cells derived from stem cells can be tumorigenic owing to undifferentiated stem cells present in the population that proliferate uncontrollably.[9] The use of stem cells, however, is promising because of the limited cell supply for many tissues. Cells that are not autogeneous must be able to avoid immune rejection. Somatic cell and nuclear transfer procedures could provide cells that function appropriately and

Tissue Engineering

yet will not be rejected.[9] The production of a universal donor cell source that can be used in any patient while avoiding immune rejection is a future goal. This probably involves altering the expression of or hiding the histocompatibility complex surface markers. Different types of stem cells are also being investigated as a source for multiple cell types if the differentiation can be precisely directed and controlled.

Scaffolds Studies performed in vitro with cells growing on a two-dimensional surface have observed isolated cell function performance, such as proliferation, glycolysis, respiration, and gene expression, by optimizing the media nutrient, hormone, and growth factor compositions. However, the proper regulation and control of these functions are dependent on cellular interactions

3119

present within a three-dimensional structure.[1] Therefore, scaffolds are essential for creating tissue substitutes that mimic in vivo function. Scaffolds can be foams, sponges, gels, membranes, or fibrous materials (Fig. 2). They are categorized as natural, synthetic, or a combination of both. Table 1 provides a list of scaffold materials and applications utilized in tissue engineering.[11–14] Natural biomaterials, such as collagen, are inherently equipped for cell interaction, but have the disadvantages of limited adaptability and customizable processing as well as relatively scarce availability compared to synthetic biomaterials.[15] The most commonly utilized synthetic biodegradable materials are poly(glycolic acid) (PGA), poly(lactic acid) (PLA), and poly(lactic co-glycolic acid) (PLGA), a blend of the former two polymers. Synthetic polymers can be used for the culture of many cell types, but it remains difficult to culture some cell types, such as nerve cells, on synthetic polymers.

Fig. 2 Five scaffold types: (A) sponge; (B) foam; (C) gel; (D) fibrous; and (E) membrane.

T

3120

Tissue Engineering

Table 1 Commonly utilized scaffold materials Polymer Natural Type I collagen Alginate Chitosan Fibrin Laminin Hyaluronic acid Polyhydroxyalkanoates (PHA) Isolated ECM from bone and small intestine Synthetic Poly(esters) Poly(glycolic acid) (PGA) Poly(lactic acid) (PLA) Poly(caprolactone) (PCL) Poly(lactic-co-glycolic) (PLGA) Poly(anhydride) Poly(hydroxybutyrate) Poly(vinyl alcohol) (PVA) Poly(ethylene glycol) (PEG) Poly(ethylene terephthalate) (PET) Expanded poly(tetrafluoroethylene) (e-PTFE) Poly(propylene fumarate)

Application Skin, bone, cartilage, tendon, nerve, kidney, cornea, vessels Cartilage, muscle, soft tissue Encapsulation, membranes Cartilage Epithelial tissues, islets Medical devices Skin, drug delivery, sutures Bone, blood vessels, ureters

References [7,11,12] [12] [11] [12] [7] [11] [11] [7]

Cartilage, bone, muscle, nerve, blood vessel, valves, bladder, liver, cardiac tissue, drug delivery, sutures

[11,12]

Bone, drug delivery Valves Cartilage, nerve Cartilage Cornea, blood vessels Cornea, blood vessels Bone, cardiovascular tissue

[12] [7] [12] [12] [13,14] [13,14] [12]

Advancement in three-dimensional polymer processing makes customization possible for polymer composition, mechanical strength, cell–surface attachment interactions, degradation rates, and high cell density.[15] Scaffold properties influence a plurality of cell culture aspects including proliferation, differentiation, adhesion, migration, gene expression, and function. Characteristics on three size scales influence these aspects. On the macroscopic scale, the scaffold is conformed to a specific shape and size to direct the formation of a three-dimensional structure. For example, the scaffold utilized for blood vessel regeneration would be tubular in shape in order to direct cell growth and tissue morphology accordingly. A three-dimensional matrix has a high surface area to volume ratio that allows for high-density cell populations and sufficient space for nutrient transfer. The mechanical strength of the scaffold may be an important consideration depending on whether the implanted tissue is subject to a large amount of stress, as for example, with cartilage. On the microscale, the porosity and pore structure regulate cell penetration, migration, interaction, and growth. Optimal porosities allow for penetration of cell seeding suspensions throughout the scaffold resulting in uniform distributions. The successful mixing and distribution of cells within a scaffold result in chondrocytes functioning properly by producing ECM molecules at high cellularity for enhanced cartilage strength

characteristics of the tissue.[16] The morphogenesis of the developing tissue is influenced by the allowable migration of cells. The pore size distribution relates to the migration ability because it determines the amount of space available. The porosity of the scaffold and the size of the pores affect the supply of nutrients and mediation of the waste concentrations via fluid and mass transfer mechanisms. Transfer considerations are increasingly important as high cell density cultures are obtained which limit the available space for fluid and nutrient transport. In fibrous scaffolds, the fiber diameter and affiliated surface curvature affect the spreading ability of attached cells. Spreading allows cells to increase proliferation and this is regulated by fiber dimensions. Additionally, the diameter affects the degree of cell–cell interactions allowable around the fiber which are necessary for proper tissue function. Patterning of the scaffold surface, such as grooves, directs cell adhesion as well as cell growth and function for certain cell types.[17] On the nanoscale, the surface chemistry of the scaffold must recreate the important cell–ECM properties of adhesion and control. Biocompatibility of the scaffold surface with cells is key for allowing adhesion and migration of cells. The amino acid sequence of arginine–glycine–aspartic acid (RGD) has been identified on fibronectin and other ECM glycoproteins as a key adhesion domain, and the design of synthetic scaffolds incorporating the peptide has been successful

Tissue Engineering

in improving adhesion and cytocompatibility.[18] The organization of RGD peptides on the scaffold surface affects adhesion as well, with a clustered arrangement optimal rather than randomly positioned RGD peptides. Scaffolds can be supplemented with binding or signal molecules bound to the polymer surface. For enhanced proliferation and differentiation, scaffolds can be designed to release growth factors efficiently. Neural cells were cultivated in rats on a scaffold equipped with degrading beads that released nerve growth factor in a controlled manner. If a scaffold is transplanted, the rate of biodegradability is important to ensure that the scaffold remains to support a transplant until a natural ECM replaces it. The biodegradation or resorption rate is a function of the scaffold composition, structure, and the mechanical load present at the site of transplantation.[7] The necessary rate at which the scaffold is degraded varies according to the tissue type. For example, slow degradation is allowable in bone tissue, whereas in other tissues chronic inflammation may occur if the rate is too low.[7] It is important that the degradation by-products are nontoxic to the body.

Cell Culture To develop a tissue in culture for use as a tissue substitute, the tissue cell density must be high (commonly 109 cells=ml) and uniform within the scaffold. In neomorphogenesis, cells are brought into contact with a porous scaffold and they form a structure together. The cell seeding process must be optimized to achieve uniformity of cell distribution within the scaffold, to maximize utilization of the cells, and to minimize seeding process time in order to avoid damage to the cells. To seed a scaffold, cells and scaffolds are incubated together to allow for adhesion to take place. Dynamic seeding protocols incorporate mixing or flow to distribute cells throughout the scaffold efficiently. When introduced, cells attach to the scaffold surfaces if the surface chemistry of the scaffold is compatible. Scaffolds can be pre-treated to alter the surface chemistry thus allowing improved compatibility with the cells. For example, increasing the concentration of hydrophilic compounds of the surface will improve cell adhesion. The medium utilized to provide complete nutrition for the growth of different cell types is based on a standard minimum essential nutrient composition. Most media consist of a sugar source, minerals, vitamins, and amino acids. Serum, such as fetal bovine serum, commonly supplements the medium to enhance cell growth. Growth factors and cytokines are utilized to accelerate cell growth through interaction with specific cell receptors. Differentiation inducers are added to direct the differentiation pathway of stem cells. Impor-

3121

tant parameters to control include the pH, pCO2, and pO2 of the media. There are several types of bioreactor designs utilized for cultivating new tissue growth, and four are shown schematically in Fig. 3. The environment within may be static or mixed using internally designed or externally applied mechanisms. Static environments rely on diffusion as the mass transfer mechanism for nutrient supply. Mixing within a culture vessel provides convective flow of oxygen and nutrients to the cells while removing waste from the surroundings. Mixinginduced shear stress levels are an important consideration since they may cause cell death. The simplest design is the plate or Petri dish. The spinner flask is larger in scale and has an internal agitation mechanism to provide a uniform nutrient concentration within the medium and the enclosed cell–scaffold construct. Perfused bioreactors are culture environments in which media are circulated within a closed system past the immobilized cell–scaffold components. The continuous flow allows for uniform nutrient supply with enhanced mass transfer. Long-term stability of the culture is attained using continuous perfusion reactors.[19] Hollow fiber bioreactors are specialized perfusion bioreactor designs in which a semipermeable membrane in a tubular configuration creates an interior and exterior region. Media can pass through the interior region of the hollow fiber axially, while the membrane allows nutrient and product transfer into and out of the extracapillary region, respectively, where cells are located growing on the exterior surface of the membrane.[20] As a novel bioreactor design, rotating-wall bioreactors spin on an axis, and the enclosed cell–scaffold constructs tumble within the rotating microgravity environment. Tissue culture in microgravity has been shown to improve cellular aggregation and produce highly differentiated tissue products.[21] Rotating-wall bioreactors avoid the high shear stress found within bioreactors that have agitation devices. As a result, altered gene expression favors improved aggregation resulting in aggregates up to ten times larger in diameter than those attained in conventional bioreactors. Additionally, necrosis of cells within the center of the aggregates due to mass transfer limitations is not seen. The stability of tissue constructs after removal from the microgravity environment remains to be proven in order to realize in vivo utilization.[21] Other stimuli may be incorporated into the culture environment to cultivate proper tissue function. Mechanical or electrical stimulation, provided at frequencies simulating in vivo conditions, have been shown to improve the resulting properties of the cultivated tissue.[19] For example, pulsatile conditions that simulate a beating heart are utilized in the culture of blood vessels resulting in improved strength and function relative to cultures lacking this stimulus.[2]

T

3122

Tissue Engineering

Fig. 3 Tissue engineering culture environment designs: (A) plate; (B) spinner flask; (C) perfusion; (D) hollow fiber; and (E) rotating wall bioreactor.

In cell culture, the essential nutrients must be present in the medium and be able to flow or diffuse to the cell membranes to allow for a viable culture. The bioreactor design needs to enable the cell population to expand to the cell density of the in vivo tissue; however, nutrient limitations deter this when cell growth decreases the space available within the scaffold for nutrient diffusion. Additionally, stagnant build-up of waste by-products increases acidity and harmful conditions. Therefore, the viability of a culture decreases within heavily populated regions of the scaffold. Supplying nutrients to these interior regions is a challenge, and currently, a tissue thickness of over 1 mm is not maintainable without cell death in the core of the tissue mass. In vivo, nutrient transfer within a tissue is achieved by a process called angiogenesis. As tissue mass increases, vasculature is established by promoting blood vessel growth within a starvation zone. A similar process of angiogenesis established within an in vitro culture would permit large-scale tissue growth. A tissue transplant can promote angiogenesis by providing angiogenic growth factors, such as vascular endothelial growth factor, along with the transplant that are

released over time to allow for high cell density tissue regeneration. Many tissues consist of more than one cell type. The fact that most major organs in the human body consist of more than one tissue and cell type adds to the complexity in recreating a functional organ replacement. The proper physiological function of these tissues depend on the interactions between these multiple cell types, so developing a tissue substitute with the ability to restore function should be a heterogeneous culture consisting of multiple cell types organized appropriately. Coculture of multiple cell types within one environment in order to accomplish this is very difficult. Currently, two different cell types are cultured together by growing each cell type in layers with membranes. The membranes allow for signal passage without direct cell–cell contact. The concept of ‘‘organ printing’’ may permit the generation of heterogeneous, vascularized, and three-dimensional organ constructs using a computer-controlled ‘‘printing’’ device that deposits multiple cell types, biomaterials, and other tissue components layer by layer to form an organized structure.[22]

Tissue Engineering

TRANSPLANTATION AND CRYOPRESERVATION When implanting a tissue substitute or therapy device, immune rejection by the host against the foreign implant is the primary concern unless a biocompatible scaffold with autologous cells is used. Several strategies exist to circumvent this challenge, depending on cell type, source, and desired function. Three specific strategies are immune system therapy before and after the transplantation procedure, gene modification in the cells prior to tissue development to allow immune system acceptance, and immunoisolation.[23] Traditionally, similar to organ transplants, the complete immune system is suppressed for a period of time following the transplant to increase the chances of the transplant eventually being accepted. There is a key 5–10 week period in which the successful integration of the transplant or graft rejection is determined.[23] However, during this period of immunosuppression, the patient is susceptible to other illnesses. A novel strategy involves building a tolerance in the host for a cell type prior to transplantation. By first performing a bone marrow transplant using donor hematopoietic cells, the host will circulate immune system components that match the future donor. Then, when the transplant occurs, the donor tissue is tolerated.[10] Immunoisolation strategies incorporate a semipermeable membrane enclosing the cell-based device, protecting the cells from immune recognition and yet allowing the transport of therapeutic cell-derived compounds to emanate from the device. When a tissue-engineered product is not to be used for transplantation upon its creation, cryopreservation allows for long-term storage until future application. Cryopreservation is commonly used for preparing cells for storage, but its utility for storage of tissues remains in development. Cryopreservation is a major focus for researchers, though, as it is a key component for improving the marketability of tissue-engineered products, allowing the development of an on-demand tissue supply, preservation of tissue genetic stability, and establishment of production quality control archives.[24] During the cryopreservation process, both the freezing and thawing procedures are equally important in regulating water displacement and replacement, respectively, while maintaining cellular and tissue integrity. Cryoprotectants and thermal processing protocols are utilized in this process. Commonly utilized cryoprotectants are dimethyl sulfoxide and glycerol, which remove intracellular water to avoid damaging ice crystal formation. Owing to the larger scale of tissues compared to cells, the challenges are greater, including the induction of chemical and thermal gradients within tissue, which must be resolved by utilizing optimized mass and heat transfer operations, respectively.

3123

Technologies to improve these processes and to monitor tissue parameters for performance control and modeling will further help to develop applicable cryopreservation protocols. APPLICATIONS Tissue engineering applications can incorporate the aforementioned components in various combinations to achieve specific goals. There are five general applications of tissue engineering: 1. The development of human tissues in vitro for future implantation into the body to replace lost tissue function. 2. In vivo tissue regeneration by transplantation of a seeded or unseeded scaffold to aid in regeneration at a deficient site in the body. 3. Development of an in vivo or extracorporeal device that supplements reduced tissue function. The in vivo device is encapsulated within a semipermeable membrane to allow for provision of a therapeutic molecule to the site while protecting the cells from the host immune system. An externally positioned device would provide deficient tissue function compounds through a tube directed to the body site while avoiding cellular contact with the immune system. 4. Establishment of an environment for expanding a cell population that is later extracted from the scaffold for implementation within the body in a cell-based or gene therapy application. 5. Development of a model to promote in vivo-like function of a population of cells in vitro for studying tissue development, pathology, pharmacology, and toxicology projects instead of using animal models. The future financial outlook for tissue engineering products have varied greatly with some predicting an $80 billion market in 2000.[2] However, owing to increasingly evident challenges involved in developing tissue replacements and a lack of realization of previous tissue engineering product success predictions, more recent market estimates have been around $15 billion annually.[5] Currently, over 20 different tissues have been researched for a tissue engineering application. A selection of these applications is shown in Table 2 and Fig. 4.[26–33] To successfully design a tissue substitute that mimics the normal in vivo counterpart, the cells must perform similar functions with the support of a biomaterial with appropriate biocompatibility, degradation, and strength characteristics and, subsequently, restore functionality of the tissue to the system.

T

Endothelial

Fibroblasts

Valvular disease

Disease, trauma

Diabetes

Renal failure

Injury

Disease, shortage

Acute liver failure

Heart valves

Cornea

Pancreas

Kidney

Tendons and ligaments

Red blood cells

Liver Porcine hepatocytes

Artificial with hemoglobin

Islets of Langerhans

Epithelial

Endothelial, fibroblast

Extracorporeal support device

Micro-encapsulation

Membrane

Polysulfone hollow fiber

Repair, replacement

Collagen

Support device, replacement

Encapsulated, restore

Alginate=poly(L-lysine)

Hollow fiber

Repair, replacement

Repair, replacement

Replacement

Detoxification activity, bridge-to-transplantation

Avoid removal of artificial cells from circulation

[20]

[33]

[32]

[31] Blood filtration, homeostasis maintenance Mechanical durability

[30]

[14]

[29]

[28]

[27]

Active glucose concentration control

Elasticity, refractive properties, transparency, curvature

Durability over mechanical devices

Flexible transplant with induced angiogenesis

Dopamine production, neural network re-established

[25,26]

Restored mechanical properties

Repair, replacement, regeneration Replacement, encapsulation

[13]

References

Thromboresistance

Goal

Replacement

Application

e-PTFE, PMMA, PVA

PGA

PGA, PLA, PCL

Lumpectomy, mastectomy

Breast

Smooth muscle cells (SMC), fibroblast, chondrocytes

Parkinson’s and Huntington’s diseases

Brain

PGA, PLA, calcium alginate Membrane

Periosteal, osteoblast, mesenchymal stem cells

Disease, radiation

Bone

PET, e-PTFE

Scaffold

Neurons

Endothelial

Cells

Vessel occlusion

Why

Blood vessels

Tissue

Table 2 Selected tissue engineering applications

3124 Tissue Engineering

Tissue Engineering

3125

T

Fig. 4 (A) In vitro modeling of colon cancer; (B) production of cord blood cells; (C) astrocyte culture for cell-based therapy of Parkinson’s disease; and (D) placenta model using trophoblast cells for transport.

Skin Skin was the first tissue to be produced as a tissue substitute. The relatively simple two-dimensional, bi-layered structure of skin primarily consists of keratinocytes, fibroblasts, and ECM components. Skin functions include roles as a protective barrier, fluid and heat regulator, and immune system reconnaissance for early warning of dangers. Common features are its flexibility, elasticity, and strength. The major application is to develop skin substitutes for use on burn victims, especially when skin from the same individual is not available for autogenous transplantation. The main role for the regeneration of skin is to re-establish the barrier function with the dermis and epidermis. Other functions can be developed in vivo via migration of other component cells from surrounding areas into the regenerated skin. Wound healing can be achieved by transplanting an unseeded scaffold at the wound site promoting migration of surrounding cells into the scaffold. Artificial skin substitutes were the first tissue engineering products to reach the market. One available dermal substitute on the market is DermagraftÕ, produced by Smith & Nephew (Florida, U.S.A.), which consists of fibroblasts, extracellular matrix, and a bioabsorbable scaffold. When applied, usually for healing diabetic foot ulcers, healthy cells surrounding the wound, including keratinocytes,

migrate into the scaffold to fully reconstitute healthy skin with natural barrier properties.[5]

Liver The liver was the first tissue engineered in three-dimensional scaffolding. Proper differentiated function of a tissue-engineered construct containing hepatocytes includes the production of albumin and the completion of urea and bilirubin metabolism. The cells used to seed the scaffold should be highly proliferative in order to develop a high cell density tissue. Owing to the size of the liver, nutrient diffusion limitations are a concern, and, therefore, vascularization of liver construct is necessary for blood to supply nutrients within the liver mass. When the engineered liver is transplanted to the host site, vascularization promoters can be included to initiate the blood vessel migration into the new tissue. Hepatocytes have also been cultured in hollow-fiber membranes to supply liver function from a device outside the body. The HepatAssist System (Circe Biomedical, Inc., Massachusetts, U.S.A.) is an extracorporeal device supporting liver function in patients waiting for transplantation.[20] The device utilizes porcine hepatocytes immobilized in the outer space of a hollow-fiber design to perform liver functions while plasma is circulated inside the hollow fiber.

3126

The polysulfone membrane allows for protein and toxin transport across the membrane while preventing the passage of foreign cells into the patient. Thus, hepatocytes in the device do not have to be compatible with the host immune system since there is no direct contact.

Bone Autogeneous bone grafts are difficult because there is a limited supply of cells, and donor sites are subject to post-graft morbidity and bone deformations. Tissueengineered bone structures consist of osteoblasts and osteocytes, which re-form in three-dimensional scaffolds to produce signature bone structures and features such as load-bearing ability and osteocalcin secretion for osteoblasts. The progenitor cells utilized in the osteogenesis can provide developmental signals such as bone morphogenetic proteins, which induce and promote bone growth. Periosteal cells, from the outer bone membrane periosteum, were seeded on calcium alginate hydrogel to allow for calcium extraction by the cells and to make customized bone-shaped molds for transplantation into bone defect sites.[25]

Gene Therapy Gene therapy is introducing DNA, which codes for an essential protein, into cells that are lacking productivity of the protein. Gene therapy is an alternative to supplying the deficient region with the protein itself or replacing the defunct tissue with a functional tissue substitute. If stem cells are utilized, the regenerating source of corrected, functional cells will eventually replace expired, deficient cells. DNA is carried and introduced into cells using a delivery system which is commonly viral or plasmid based.[34] Once the gene is expressed within the cell, persistence of the essential protein supply is key. Accompanying the huge potential of this process are important safety concerns with regards to the implementation of the DNA correctly within the chromosomes. Incorrect incorporation of the gene could alter important control genes or other essential genes. Once the successfully transduced cells are selected, the cell population is expanded in vitro. In Vitro Studies Tissue engineering can also be used to develop laboratory tools using human cells to perform pathological, developmental, pharmacological, and toxicological studies. Tissue cultured as an in vitro model can be used to study tissue or organ development processes including signaling and control mechanisms. Additionally,

Tissue Engineering

toxicology and carcinogen studies on tissue models provide insight without utilizing animal models.[35] Cancer models can be used to study tumor biology, control, and progression as well as use in cancer treatment studies.[36] Cell-Based Technologies Additional cell-based technologies are being developed by utilizing the knowledge of cells and their interaction with materials attained from tissue engineering research. Biosensors take advantage of the specificity of cell surface receptors for select target molecules and the high signal amplification for detecting low levels of chemical or biological agents. The brain consists of a neural network that processes a large number of signals. Neurons organized on a support in specific patterns could produce in vitro neural networks that pass signals similarly to the electronic structure on microchips.[37]

CURRENT CHALLENGES Although there has been some limited success in producing tissue substitutes, especially artificial skin, the overall tissue engineering aim of creating tissue that can replace, maintain, or improve deficient in vivo tissue is not an easy task. Multiple challenges remain in providing tissue products to meet demand. These include creating a greater supply of cells to avoid immune rejection, such as a universal donor cell supply. Advancements in biomaterial production and bioreactor design are necessary to provide customized support and environments to replicate the complex in vivo environment. When developing complex, threedimensional tissues, challenging issues include promoting angiogenesis to overcome tissue size limitations, the heterogeneous coculture of multiple cell types within one tissue, and the cryopreservation of the resulting tissue before transplantation.

CONCLUSIONS Through the multidisciplinary application of biological and medical knowledge and engineering skills, tissue engineering allows the development of tissues in the human body to restore, repair, or improve deficient biological components. Generally, utilizing cells, supportive scaffolds, and designed culture systems, tissue development processes face many challenges in reconstituting in vivo-like performance. Although many challenges remain, the field is a major research focus, and medical demand is strong, both ensuring that

Tissue Engineering

tissue engineering stands to provide momentous advancements to health in the 21st century.

REFERENCES 1. Ma, T. Fiber-Based Bioreactor Systems In Mammalian Cell Culture and Tissue Engineering Human Trophoblast Cells; Dissertation; The Ohio State University: Columbus, OH, 1999. 2. Langer, R. Tissue engineering. Mol. Ther. 2000, 1 (1), 12–15. 3. http: ==www.optn.org=latestData=step2.asp (accessed July 2003). 4. Schulthesis, D.; Bloom, D.A.; Wefer, J.; Jonas, U. Tissue engineering from Adam to the zygote: historical reflections. World J. Urol. 2000, 18, 84–90. 5. Flanagan, N. Engineering methods for tissue and cell repair. Genet. Eng. News 2003, 23 (8), 10. 6. Martins-Green, M. Dynamics of cell–ECM interactions. In Principles of Tissue Engineering, 2nd Ed.; Lanza, R.P., Langer, R., Vacanti, J., Eds.; Academic Press: New York, 2000; 33–55. 7. Griffith, L.G. Biomaterials. In WTEC Panel on Tissue Engineering Research: Final Report; McIntire, L.V., Ed.; Academic Press: New York, 2002; 9–22. 8. Hubbell, J.A. Matrix effects. In Principles of Tissue Engineering, 2nd Ed.; Lanza, R.P., Langer, R., Vacanti, J., Eds.; Academic Press: New York, 2000; 237–250. 9. Heath, C.A. Cells for tissue engineering. Trends Biotechnol. 2000, 18, 17–19. 10. Brower, V. Stem cell research and development advances in face of challenges. Genet. Eng. News 2003, 23 (9), 1. 11. Pachence, J.M.; Kohn, J. Biodegradable polymers. In Principles of Tissue Engineering, 2nd Ed.; Lanza, R.P., Langer, R., Vacanti, J., Eds.; Academic Press: New York, 2000; 263–277. 12. Elisseeff, J.H.; Langer, R.; Yamada, Y. Biomaterials for tissue engineering. In Tissue Engineering and Biodegradable Equivalents: Scientific and Clinical Applications; Lewandrowski, K.-U., Wise, D., Trantolo, D., Gresser, J.D., Yaszemski, M.J., Altobelli, D.E., Eds.; Marcel Dekker: New York, 2002; 1–24. 13. Xue, L.; Greisler, H.-P. Blood vessels. In Principles of Tissue Engineering, 2nd Ed.; Lanza, R.-P., Langer, R., Vacanti, J., Eds.; Academic Press: New York, 2000; 427–446. 14. Trinkaus-Randall, V. Cornea. In Principles of Tissue Engineering, 2nd Ed.; Lanza, R.-P., Langer, R., Vacanti, J., Eds.; Academic Press: New York, 2000; 471–491.

3127

15. Langer, R. Selected advances in drug delivery and tissue engineering. J. Control. Release 1999, 62, 7–11. 16. Vunjak-Novakovic, G.; Freed, L.; Biron, R.J.; Langer, R. Effects of mixing on the composition and morphology of tissue-engineered cartilage. Am. Inst. Chem. Eng. J. 1996, 42, 850–860. 17. Saltzman, W.M. Cell interactions with polymers. In Principles of Tissue Engineering, 2nd Ed.; Lanza, R.P., Langer, R., Vacanti, J., Eds.; Academic Press: New York, 2000; 221–235. 18. Griffith, L.G.; Naughton, G. Tissue engineering— current challenges and expanding opportunities. Science 2002, 295, 1009–1014. 19. Freed, L.E.; Vunjak-Novakovic, G. Tissue engineering bioreactors. In Principles of Tissue Engineering, 2nd Ed.; Lanza, R.P., Langer, R., Vacanti, J., Eds.; Academic Press: New York, 2000; 143–156. 20. Mullon, C.; Soloman, B.A. HepatAssist liver support system. In Principles of Tissue Engineering, 2nd Ed.; Lanza, R.P., Langer, R., Vacanti, J., Eds.; Academic Press: New York, 2000; 553–558. 21. Unsworth, B.R.; Lelkes, P.I. Tissue assembly in microgravity. In Principles of Tissue Engineering, 2nd Ed.; Lanza, R.P., Langer, R., Vacanti, J., Eds.; Academic Press: New York, 2000; 157–164. 22. Mironov, V.; Boland, T.; Trusk, T.; Forgacs, G.; Markwald, R.R. Organ printing: computer-aided jet-based 3D tissue engineering. Trends Biotechnol. 2003, 21 (4), 157–161. 23. Hardin-Young, J.; Teumer, J.; Ross, R.N.; Parenteau, N.L. Approaches to transplanting engineered cells and tissues. In Principles of Tissue Engineering, 2nd Ed.; Lanza, R.P., Langer, R., Vacanti, J., Eds.; Academic Press: New York, 2000; 281–291. 24. Karlsson, J.O.M.; Toner, M. Cryopreservation. In Principles of Tissue Engineering, 2nd Ed.; Lanza, R.P., Langer, R., Vacanti, J., Eds.; Academic Press: New York, 2000; 293–307. 25. Vacanti, C.A.; Bonassar, L.J.; Vacanti, J.P. Structural tissue engineering. In Principles of Tissue Engineering, 2nd Ed.; Lanza, R.P., Langer, R., Vacanti, J., Eds.; Academic Press: New York, 2000; 671–682. 26. Bruder, S.P.; Caplan, A.I. Bone regeneration through cellular engineering. In Principles of Tissue Engineering, 2nd Ed.; Lanza, R.P., Langer, R., Vacanti, J., Eds.; Academic Press: New York, 2000; 683–696. 27. Wahlberg, L.U. Brain implants. In Principles of Tissue Engineering, 2nd Ed.; Lanza, R.P., Langer, R., Vacanti, J., Eds.; Academic Press: New York, 2000; 773–783.

T

3128

28. Lee, K.Y.; Halberstadt, C.R.; Holder, W.D.; Mooney, D.J. Breast reconstruction. In Principles of Tissue Engineering, 2nd Ed.; Lanza, R.P., Langer, R., Vacanti, J., Eds.; Academic Press: New York, 2000; 409–423. 29. Love, J.W. Cardiac prostheses. In Principles of Tissue Engineering, 2nd Ed.; Lanza, R.P., Langer, R., Vacanti, J., Eds.; Academic Press: New York, 2000; 455–467. 30. Wang, T.G.; Lanza, R.P. Bioartificial pancreas. In Principles of Tissue Engineering, 2nd Ed.; Lanza, R.P., Langer, R., Vacanti, J., Eds.; Academic Press: New York, 2000; 495–507. 31. Humes, H.D. Renal replacement devices. In Principles of Tissue Engineering, 2nd Ed.; Lanza, R.P., Langer, R., Vacanti, J., Eds.; Academic Press: New York, 2000; 645–653. 32. Goulet, F.; Rancourt, D.; Cloutier, R.; Germain, L.; Poole, A.R.; Auger, F.A. Tendons and ligaments. In Principles of Tissue Engineering, 2nd Ed.; Lanza, R.P., Langer, R., Vacanti, J., Eds.; Academic Press: New York, 2000; 711–722.

Tissue Engineering

33. Chang, T.M.S. Red blood cell substitutes. In Principles of Tissue Engineering, 2nd Ed.; Lanza, R.P., Langer, R., Vacanti, J., Eds.; Academic Press: New York, 2000; 601–610. 34. Fradkin, L.G.; Ropp, J.D.; Warner, J.F. Genebased therapeutics. In Principles of Tissue Engineering, 2nd Ed.; Lanza, R.P., Langer, R., Vacanti, J., Eds.; Academic Press: New York, 2000; 385–405. 35. Li, A.P. Screening for human ADME=Tox drug properties in drug discovery. Drug Discov. Today 2001, 6 (7), 357–366. 36. Chung, L.W.K.; Zhau, H.E.; Wu, T.T. Development of human prostate cancer models for chemoprevention and experimental therapeutics studies. J. Cell. Biochem. Suppl. 1997, 28=29, 174–181; Supplement. 37. Mrksich, M. Cell-based technologies: nonmedical applications. In WTEC Panel on Tissue Engineering Research: Final Report; McIntire, L.V., Ed.; Academic Press: New York, 2002; 61–69.

Trace Elements T Ian D. Brindle Brock University, St. Catharines, Ontario, Canada

INTRODUCTION It is a truism that nothing in the universe is free from trace elements. As analytical techniques have improved with the development of ever more sophisticated measuring devices, the impact of trace and ultratrace concentrations of elements has been revealed much more clearly over the last two decades or so. We know from nutritional studies the importance of ultratrace concentrations of several elements including, for example, cobalt in cyanocobalamine (vitamin B12), where the daily requirement for this vitamin is only 1 mg=day for an adult. Some elements commonly occur in ores of other elements and so can become problems, even where they are present at ultratrace levels. These elements can have deleterious or beneficial effects upon the properties of the major element that is won from the ore. The impact of trace elements on properties of materials, even when they are present at ultratrace concentrations, can be striking, and understanding their influence on materials is important. Regulations, either environmental or prescriptive, will continue to have a major impact on our interest in trace elements.

ENVIRONMENTAL IMPACTS Treatment of many materials results in the liberation of the trace elements into the environment, which can have an impact on health. Coal is a particularly useful example of a major source of trace elements poured into the environment from coal combustion. Coal contains an alphabet soup of trace elements, including arsenic, mercury, uranium, selenium, and chromium. Pyrite is a ubiquitous mineral found in coal, but coal can also contain a variety of other mineral phases. West Virginia coal, for example, includes clay minerals such as kaolinite (35%) and illite (35%), quartz (18%), pyrite (7%), and calcite (3%).[1] A number of projects that utilize coal for power generation while minimizing the impact on the environment have been described. An excellent example is the SNOXTM (trademark owner Haldor Topsoe) demonstration project, which utilizes high-sulfur coal (2.8%).[2] The demonstration project of this technology, equally funded by the U.S. Department of Energy and participants at a total cost Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120007994 Copyright # 2006 by Taylor & Francis. All rights reserved.

of U.S.$ 31.5 million, was able to reduce NOx to nitrogen (>90% reduction) and to oxidize SO2 to SO3 (>95% transformation) and ultimately to make sulfuric acid for sale, and hence minimizing the amount of acid rain produced by the system. Mercury was not retained at all by the system and boron was only partially retained. Selenium and cadmium, normally problematic elements, were recovered in the processing of the spent gases. Particulate emissions were reduced by 99%. Mineral content of coal is variable and each new source must be evaluated closely for its trace element composition. For some applications, it may be a case where the raw material should not be processed. Thus, in Guizhou province of the People’s Republic of China, coal is mined from a deposit that is very high in arsenic.[3] Germanium is also relatively high in coal. The highest reported value of germanium found in coal ash was 1.1% in a sample from a particular seam of coal in Durham, U.K. Chalcophilic elements are usually associated with sulfur minerals in coal, and some success in removing these elements prior to burning the coal has been achieved by washing the coal and removing the heavier mineral components. Other fuels are also susceptible to contamination with trace elements at low, but significant, concentrations. A report prepared for the United States Environmental Protection Agency (USEPA) describes several sources of contamination from traces of mercury in oils and natural gas.[4] Elemental mercury is present in liquid petroleum oil or in natural gas, condensates at parts per million level, and must be removed from the product stream to prevent it from reacting with metallic components, particularly during the cryogenic treatments, where liquid mercury can be condensed from the gas phase. Mercury concentrations can vary considerably, and the EPA report describes one gas reservoir in Texas, where the concentration of mercury is sufficiently high to lead to the conclusion that the gas is in equilibrium with elemental mercury in the subsurface reservoir. In terms of processing of petroleum and natural gas, mercury is found in a variety of compartments and in a variety of forms or species including alkylmercury and inorganic mercury salts such as mercury halides and mercury sulfides (insoluble and largely found in suspension or in drilling wastes). Table 1, which presents predominant species in a variety of matrices, is adapted from the report. 3129

3130

Trace Elements

Table 1 Approximate distribution and abundance of mercury compounds in hydrocarbons Coal

Natural gas

T

D

D

D

(CH3)2Hg

a

T

T, (Sa)

T, (Sa)

HgCl2

a

N

S

S

Hg0

S

Gas condensate

Crude oil

HgS

D

N

Suspended

Suspended

HgO

Ta

N

N

N

a

N

Ta

Ta

CH3HgCl

Abundances, expressed as percentages of total Hg concentrations: D (dominant), greater than 50 percent of total; S (some), 10-50%; T (trace), less than 1%; N (none), rarely detected. a Inconclusive data. (From EPA-600=R-01-066.)

In view of the widely varying toxicities of these species and, in particular, the unexpectedly huge toxicity of dimethylmercury, for which we have only one reported human death (Professor Karen Wetterhahn of Dartmouth College, who died as a result of a single and very limited exposure to dimethylmercury in 1997), the significance of these trace concentrations of ‘‘species’’ must be considered when the health of workers who will be exposed to these compounds is evaluated. An element that is also found in natural gas represents the world’s best source for the element and that is helium which, in some U.S. and polishgas supplies, can reach 7% by weight. This valuable gas is collected by liquefaction of the natural gas, which leaves the remaining gas significantly enriched in helium. Helium originates from radioactive decay and, because it is a fugitive gas and can escape the earth’s atmosphere, it is not practical to recover helium from the air.

TRACE ELEMENTS IN METALS Trace concentrations of a variety of elements have huge effects upon metals and can, in many cases, determine the metals’ fitness for various purposes in manufacture and use or the traces may disqualify a particular form of the metal from use either because of the changes in physical or chemical properties or because it may affect the end use or disposal of the metal. Iron Iron is one of the most recycled metals and the recycling process can introduce unwanted elements into the alloys and can play havoc when they are ignored. Iron, cast into the engine blocks of internal combustion engines, is often made from gray iron or nodular iron. Nodular cast iron has a crystallinity and hardness that enables the casting to support the

continuous motion of pistons and piston rings, suitably lubricated, for more than 100,000 miles without significant wear. Reproducible production of nodular cast iron is clearly of great economic importance. One trace element that has been identified in guaranteeing the quality of nodular cast iron is lanthanum. Although not required as a pure additive, several rare earth elements confer advantages on the manufacturing process. Often, lanthanum is added as an ingredient in a rare earth mix of elements, which typically includes cerium at higher concentrations. Much of the control of microscopic structure of iron depends on the management of the carbon content of the metal. Carbon at levels less than 5% precipitates in a variety of forms and can also react with iron and other metals in the alloy to make carbides. Trace elements can significantly alter the form in which carbon occurs, and manipulation and control of trace elements are major determinants of quality standards for categorizing irons by the American Society for Testing Materials (ASTM). The structure of precipitated carbon determines the properties of the alloy, in particular its tensile strength. Lead at levels in iron that exceed 50 parts per million (ppm) results in the formation of Widmansta¨tten graphite, which confers a mossy or fuzzy appearance to graphite flakes in the iron. Without an appropriate modifier, Widmansta¨tten graphite reduces the tensile strength of iron to less than 15,000 pounds per square inch (psi). Addition of cerium, in a process usually called ‘‘inoculation,’’ reduces the effect of lead to more manageable levels. Nitrogen also affects the form of carbon found in iron. Normally, nitrogen equilibrium concentrations are less than 70 ppm. When the values exceed 150 ppm, the graphite form in gray iron is ‘‘fatter’’ than it would be in the presence of low concentrations of nitrogen. Titanium, present at trace concentrations, reverses this trend and can restore graphite flakes to normal, filamentous morphology. Boron-containing low-carbon steels have a number of advantages that are specifically because of the

Trace Elements

presence of boron. Subtle effects relate to boride concentration appearing at grain boundaries. In combination with niobium, boron confers several advantageous properties to the so-called bainitic steels. Thus boron low-carbon steels have improved hardness, strength, weldability, and corrosion resistance, which has made them the steel of choice in machinery manufacture, oil pipelines, and maritime applications (bridges, drilling rigs, and ships). Furthermore, as they can be rapidly cooled, these steels are relatively energy efficient materials. Trace concentrations of silicon and aluminum tend to decrease the strength and hardness of iron by increasing the ratio of ferrite to pearlite. Nickel, copper, and tin work in the contrary direction and, by increasing the amount of pearlite, increase the strength and hardness of iron. In some cases, not surprisingly, trace elements can confer benefits for some purposes and liabilities with others. The impact of trace elements on creep deformation and fracture was investigated in depth.[5] The conclusions were uncertain as to the effects of trace concentrations of several elements (P, As, Sn, etc.), but they determined that traces of titanium decrease embrittlement caused by trace elements. Other work suggested that arsenic and antimony reduce the ductility of steel and increase brittleness; antimony also improves the resistance of iron to corrosion.[6] Welz and Melcher[7] noted that traces of bismuth improve the machining properties of steel up to a point but, even at low levels, steels may break during cold working. Tin has a deleterious effect upon the hot working of steel. As we are better informed about the fate of materials in the environment by our greater knowledge of the cycling of elements in the environment, the present-day regulators are increasingly concerned about ‘‘cradle to grave’’ management of materials. One critical aspect of such regulations is the anticipation of issue around the disposal of wastes at the end of the life of materials. Thus, some iron and steel manufacturers are anticipating the fate of iron and steels when they re-enter the environment as they rust and disintegrate. One striking example will serve to illustrate this case. A study was commissioned by the U.S. Department of Energy to review the fate of ferritic steels with regard to shallow burial or recycling.[8] The anticipated problems relate to the activation, by neutron absorption, of various elements adventitiously or deliberately present in the steel. The radioactive isotopes, generated by neutron activation, limit the disposal or recycling options, and elements including silver, molybdenum, and niobium were identified as potentially problematic. Niobium is deliberately added to iron, as noted above, and because it is readily activated, it can render radioactivity to the iron. Other elements, including silver and gold, appear at trace

3131

concentrations and are likely carried into the irons and steels from the processing of the iron ores, which can be problematic when activated by the neutron flux from a nuclear reactor. Copper High purity copper is an essential component in the electronics industry, and a number of trace elements can decrease the conductivity of copper. As electrical resistance translates into energy losses as heat (I2R), manufacturers are anxious to ensure the lowest resistance of copper wiring in circuitry so that they can reduce both the energy and heat load in electrical and electronic equipment. Most trace impurities increase the resistivity of copper and so manufacturers look to develop the highest purity copper for applications in the electronics industry. Cadmium appears to have the least effect upon copper resistivity. Zinc The dry cell battery industry is a major consumer of zinc. The amount of current that can be withdrawn from a dry cell battery is limited because of polarization at the surface of the zinc, caused by the overpotential, which is in turn caused by hydrogen at the surface. This overpotential was traditionally reduced by incorporating an oxidizing agent (manganese dioxide) to react with the hydrogen. Incorporating small amounts of mercury in the battery extends battery life by limiting a phenomenon known as local action. Local action is caused by trace impurities in zinc, which set up independent galvanic cells that react and reduce the useful life of the battery. Incorporation of mercury in the zinc is proposed to separate the impurities from the zinc and thereby reduces the independent galvanic activity. As mercury has become increasingly regulated, indium has taken its place in dry cell batteries. Another concern for battery manufacturers is the concentration of the hydride-forming elements that appear to be responsible for significant losses in performance of dry cell batteries even at low concentrations. One of the products manufactured by the primary producers of zinc is a 30% m=v ZnSO4 solution. The detection limits desired by the manufacturers for hydrideforming elements in this concentrated solution are 10 mg=L for arsenic, antimony, bismuth, tin, selenium, and tellurium, and 2 mg=L for germanium.[9]

CATALYST POISONING BY TRACE ELEMENT Coal combustion remains the single largest source of energy used in the generation of electricity in the

T

3132

U.S.A. Energy production in 2000 was 1968 TW hr and coal represented 51% of the total. By 2010, energy production is anticipated to rise to around 4600 TW, with coal burning being responsible for about 2200 TW hr or 48% of the total previously noted.[10] Coal is a source of many trace elements; while mercury may be a problem for human health, other elements create problems in other areas. As the large amounts of coal are burned (around 10 million tons per year for some power stations), efforts are underway to improve the quality of emissions to fulfill a mandate to use ‘‘clean coal’’ technology. Selective catalytic reduction (SCR) is employed to reduce NOx emissions (a component of acid rain) into the environment from burning coal. Arsenic in the coal is usually oxidized to arsenic trioxide (As2O3), which has a significant vapor pressure in the flue gases and poisons SCR catalysts. Arsenic removal is possible when there are sufficient calcium oxide equivalents in the coal, which can react with the arsenic trioxide to give nonvolatile calcium arsenate [Ca3(AsO4)2]. The desirable presence of calcium-containing minerals in coal in this application suggests that minerals should be retained in coal to deal with arsenic. This retention, however, appears to run contrary to the desirability of removing minerals as a source of both arsenic and other toxic elements noted above. As coal washing is used to discriminate coal from noncoal, it may be difficult to separate calcium minerals that confer benefits from iron sulfide minerals that are problematic. Polyolefin production is enhanced by a new generation of metallocene cocatalysts that are used together with Ziegler–Natta catalysts. These metallocene catalysts are poisoned by the presence of oxygen in the feedstock. The presence of oxygen can also be problematic by causing some crosslinking, chain branching, and delay of induction of polymerization. Typical olefin feedstock has oxygen maintained at levels below 50 parts per billion (ppb). Reforming catalysts are poisoned by sulfur and halogens. The complexity of reactions involved in reforming means that catalysts can work to the advantage of some reactions over others. This advantage can work against the process for, for example, the production of hydrogen. The Boudouard reaction, for example, removes carbon from the reforming process. Some catalysts need to be poisoned to some degree. Sulfur, present in most natural gas, is an appropriate poison for some catalysts, and it is not totally removed from the process stream, as it poisons the catalyst for the Boudouard reaction without materially affecting the reforming reactions. Several important reforming reactions are presented in Scheme 1. Sulfur, chlorine, and phosphorus are classed as temporary poisons in reforming reactions, because the activity of the catalyst is recovered when the poison

Trace Elements

Scheme 1 Reforming reactions.

is removed from the gas stream. Permanent poisoning results from the presence of certain metal vapors including zinc, lead, and arsenic. Alkali and alkaline earths are also permanent poisons. The occurrence of permanent poisons is a rare event in reforming reactions. Sulfur often occurs at levels below 20 ppm, therefore sulfur removal may not be necessary, but the levels of sulfur should be monitored. Carbon monoxide is a common impurity in hydrogen that has been generated by a reforming process. Portable power sources, used by the military for energy generation in remote locations, use a methanol reforming reaction as shown in the following equation: CH3 OH þ H2 O ! CO2 þ 3H2

ð1Þ

Another significant reaction can also occur, which is shown in the following equation: CH3 OH ! CO þ 2H2

ð2Þ

This reaction, or the hydrocarbon reforming reactions noted above, can generate sufficient carbon monoxide to poison the catalysts used for hydrogenbased fuel cells that are used to generate electricity. Thus, carbon monoxide is a trace component of concern, and as the hydrogen economy is further developed and reformation reactions may become more important in the development of hydrogen fuels, technologies for the efficient removal of carbon monoxide must be developed. One of the most widely known examples of catalyst poisoning is taken from the automobile industry. Though tetra-ethyl lead has been removed from essentially all gasoline in North America, the ban on leaded gasoline is not worldwide, and leaded and unleaded gasoline is available in many countries. Catalytic converters, which contain precious metals like platinum, palladium, and rhodium, are used to both reduce NOx and oxidize CO and unburned hydrocarbons. Lead irreversibly destroys the catalytic ability of the converter. Concentrations of lead in leaded gasoline are nominally 150 mg=L.

Trace Elements

TRACE ELEMENTS IN SOLUTION Treatment of solutions, when they are used to crystallize compounds, is done in various ways to change the size and the nature of crystals formed. Sudden cooling usually results in the formation of small crystals, whereas crystals formed by slow cooling tend to be large. In addition, rapid precipitation often results in the incorporation of traces of contaminants in the crystals. Finely divided crystals have a large surface area and can be used advantageously as a way to remove impurities from solutions. This technique has been used for a variety of purposes, from the removal of trace elements from drinking water to the removal of radionuclides from solutions derived from spent nuclear materials. Magnesium hydroxide has also been used analytically as a method to preconcentrate elements. Using this technique, Brindle et al.[11] were able to preconcentrate germanium and determine its concentration in the range of 1 pg=ml. The impact that trace concentrations of elements in solutions can have on precipitation reactions is quite striking and can be problematic, or can confer economic benefits. Impurities at the levels of 10
3 to 10
4 M, which consist of relatively large organic molecules, have an inhibitory effect upon the growth of crystals but do not appear to fully inhibit their growth. Concentrations of inorganic complexes, even if they are low, can inhibit or ‘‘poison’’ the growth of crystals by adsorption at a surface step rather than their being adsorbed as a surface monolayer. Sears proposed such a mechanism for the inhibition of the growth of potassium bromide and lithium fluoride in the presence of FeF63
at concentrations in the range of 10
5 to 10
6 M.[12] In addition to affecting the crystal growth, these foreign ions can affect also the shape of crystals.[14] At higher concentrations, impurities can cause a variety of changes. If the impurity is isostructural with the major component, the morphology of the resulting crystal will be intermediate between the two forms. Where adsorption of ions onto a crystal surface takes place, small changes in the lattice position of ions on a facet of the crystal farthest from the center of symmetry have the largest effects on the surface energy, which in turn can modify the habit of the crystal. It appears to be a general principle that the adsorbed ions have a minor impact upon habit, and the ions that insert themselves into the crystal lattice have the greatest effect in altering crystal habit. The normal crystal habit of many commodity chemicals is incompatible with their use. Not all crystalline materials can flow readily. In applications where a smooth-flowing powder is required, the natural shapes of crystals may not lend themselves to being free-flowing, and ‘‘clumping’’ can occur, which can clog feed tubes and hoppers. Thus if one can alter

3133

the crystal habit to a form which allows a more free-flowing powder, time and product can be saved. The modification of habit has occupied process chemists for many years, and a number of high-volume chemicals are routinely treated with habit modifiers to improve the flow characteristics, especially of these commercial products. A crystal has several faces, and the nature of interactions with the surrounding solutions, including the impact of supersaturation, solvent, temperature, and the level of impurities in solution, varies widely according to the energies of the various faces. Because crystals expose a different chemical environment at each face, it should not be surprising that impurities will have a major impact on the crystal habit—all other conditions remaining the same. Face-specific interactions of impurities, even when they are present at low concentrations, can be considerable. As discussed in this entry, impurities tend to reduce the rate of growth at the site of adsorption, thus allowing other facets to grow at their expense. Manufacturers usually prefer granular or prismatic crystal habit, but other forms, including needle and flaky crystals, are sometimes desirable. There are thousands of papers in the literature that describe the modification of crystal habit, and this entry can only touch upon a few. Highly charged and=or complex ions appear to have the greatest effect upon crystal habit. Thus, Al3þ, Cr3þ, Fe3þ, and Fe(CN)64
are commonly used in a variety of applications to modify habit. A number of excellent monographs have been published.[13] This reference also includes a number of sources that provide further reviews that discuss the impact of trace concentrations upon crystallization. In Table 2, the range of initial morphologies, habit modifiers, and the crystal forms that can result from trace concentrations of several different trace modifiers are shown.

TRACE ELEMENTS IN WATER Drinking water is obtained from surface sources, including lakes and rivers, and from the subsurface aquifers that may be shallow or deep, ancient or relatively recent. Shallow aquifers are often recharged by rainwater at an aquifer recharge zone. Both surface and subsurface waters can be contaminated by toxic elements. Sometimes, surface waters can become contaminated ultimately by subsurface water and can create major environmental and potential health problems. An example of the contamination of surface water by toxic elements is the case of northern California’s problem with selenium in surface waters. This problem is acute in the area south of San Francisco and in the Kesterton National Wildlife Refuge. Groundwater flows through seleniferous formations and is used

T

3134

Trace Elements

Table 2 Habit modification by addition of trace ions or chemicals Commodity

Normal habit

Habit modifier

New habit

Ammonium alum

Octahedral

Borax

Cubes

NH4Cl

Dendrites

Cd2þ, Ni2þ

Cubes

NH4H2PO4

Needles

Al3þ, Fe3þ, Cr3þ

Tapered prisms

3þ

(NH4)2SO4

Prisms

Fe

Irregular crystals

MgSO47H2O

Needles

Borax

Prisms

AgNO3

Plates

Sodium oleate

Dendrites

Potassium alum

Octahedral

Borax

Cubes

3þ

KCN

Cubes

Fe

Dendrites

KCl

Cubes Cubes

Fe(CN)64
PbCl2

Dendrites Octahedral

K2SO4

Rhombic prisms

Fe3þ

Irregular needles

NaBr

Cubes

Fe(CN)64


Dendrites

3þ

NaCN

Cubes

Fe

Dendrites

NaCl

Cubes Cubes

Fe(CN)64
Na6P4O13

Dendrites Octahedral

NaClO3

Cubes

S2O62


Octahedral

[4]

(From Ref. .)

mostly for irrigation. The surplus flow eventually finds its way into the wetlands where it is highly toxic to migratory birds and other wildlife that depend on the wetland for food and protection. In a similar fashion, arsenic is found in groundwater in many countries including Argentina, Bangladesh, Malaysia, and Thailand, and also in some areas of the U.S.A. The levels of arsenic in the untreated groundwater in some parts of Bangladesh are sufficiently high to cause a number of health problems including keratosis and cancer. Regulations on the levels of arsenic in drinking water have been steadily modified and the acceptable levels have been lowered over the last three decades. Many municipal laboratories are currently incapable of measuring the levels of arsenic below 10 ppb, which is creating problems for regulatory agencies charged with protecting public health. Aluminum, in the form of alum, is used in the treatment of water to remove trace elements and particulate matter. The reaction involves the hydrolysis of the aluminum ion to form aluminum hydroxide, which is a flocculent precipitate with a very large specific surface area that is useful for the adsorption of ions and also for the collection of microscopic particles that are carried down by the precipitate. In the management of water treatment plants, the kinetics of the reactions are taken into consideration in designing plants so that there is sufficient time to allow all of the aluminum to hydrolyze and precipitate as the hydroxide. When there are upset conditions, aluminum can be carried into the water supply. As the public is concerned about Alzheimer’s disease and as there has

been some considerable discussion about the involvement of aluminum in the disease, operators of water treatment plants need to be mindful of this transfer reaction. Some water treatment plants use iron(III) chloride (ferric chloride) instead of aluminum. The hydrolysis reaction in this case tends to be less problematic and the precipitation forms more rapidly. Although water treatment tends to remove any traces of lead from water, lead still appears at the consumer’s tap and can be a cause for concern. There are several sources for this lead. In the early years of municipal water distribution, leader pipes, i.e., the pipes that deliver water from the street main to the house, were made from lead. With the development of flexible copper piping, leaders in new houses were constructed of this material. As the technologies used in water treatment have changed, the protective layers of hydroxides and hydroxycarbonates have been eroded. In recent years, water in some areas, such as Washington, DC, has become increasingly contaminated with unacceptable levels of lead. These municipalities are now faced with replacing lead leaders with copper—in part because of increasingly stringent standards for lead in drinking water. Other trace elements in drinking water become unacceptable because they undergo species transformation during the water treatment process. Both chromium and bromine as chromium(III) and bromide, respectively, are benign species and their presence in drinking water does not pose a hazard. As water is treated in the water treatment plant by strongly oxidizing reagents such as ozone, used to sterilize the water

Trace Elements

and render it germfree, chromium and bromine can be transformed into chromium(VI) (as chromate, CrO4
) and bromate (BrO3
), respectively. Both of these compounds are carcinogenic. Thus, drinking water standards for chromium and bromide are put into effect to control their likely transformation into the more toxic species by chemical treatment.

TRACE ELEMENTS IN FOOD PROCESSING Trace elements affect foods in a number of ways. Their effects also vary and can diminish or enhance the toxicity of trace concentrations of elements. As a result of the increased level of food processing that is done to increase the stability, shelf life, etc., many nutritionists feel that modern, highly processed foods are in fact missing many essential trace elements and that consumers in developed countries are showing deficiencies in a number of trace elements such as chromium and zinc that they would normally get through the husks, germs, and other parts of plants that are disposed off. In wines, traces of iron, which are picked up, perhaps, from processing and=or storage, or copper, which are picked up from mildew sprays, such as Bordeaux mixture, affect the oxidative stability of wines by acting as the redox shuttles as they transfer between oxidation states. Winemakers discovered that adding ferricyanide to wine, in a process known as ‘‘blue fining,’’ precipitates copper and iron and thereby reduces their concentrations below 1 ppm, which is considered to be acceptable. Critical control of ferricyanide addition is necessary, as cyanide is also a contaminant that must be measured. Where vineyards have replaced cherry and apple orchards, low concentrations of arsenic have started to appear; but they are present at very low concentrations in high quality wines. The arsenic appears from arsenical compounds such as lead and calcium arsenates that were used for many decades as pesticides on apples and cherry orchards. A bizarre manifestation of trace element interference with food processing was described in the early 1980s by Procter and Gamble food scientists. Their development of a lemon chocolate cake created very negative responses from the tasting panels that were set up to determine the acceptability of new formulae for foods.[15] In this case, the problem was iodine, which was present as iodide in the salt used in the formulation. The iodide was oxidized to iodine during the cooking process, which in turn reacted with one of the flavoring ingredients, cresol. This resulted in the formation of iodocresol, which has a very strong medicinal taste. This taste was responsible for stimulating the gag reflex in some members of the tasting panels. On the effect of iodide in various foods, UNICEF commissioned a report.[16]

3135

It is rapidly becoming clear that trace elements in foodstuffs acquire greater significance when they can undergo important metabolic transformations in which simple inorganic salts are converted by enzyme systems within the organism or the ingestion of previously transformed compounds that may appear in a biomagnification scenario. A number of elements are particularly noteworthy in this regard, such as mercury, arsenic, and tin. Mercury tends to occur at highest concentrations in fish, particularly the predatory fish at the top of the food chain. The toxicity of mercury compounds, as noted above, varies radically according to the chemical species. It is well known at this point that most of the mercury in fish is present as methylmercury. Bloom[17] reported that more than 90% of the mercury in fish muscle is present as methylmercury. The action of methylmercury is particularly troubling for women in the first trimester of pregnancy. Methylmercury, it is believed, coordinates with cysteine in the blood and this coordinated form can cross the blood–brain barrier, where it can result in a number of serious neurological problems and is implicated in mental retardation in babies. The mother typically exhibits no symptoms. The most infamous case of methylmercury poisoning amongst an adult population was in the Japanese community of Minamata. In the 1950s, methylmercury, a byproduct from the manufacture of acetaldehyde, was discharged into the bay at Minamata where it was rapidly incorporated into fish and shellfish. The effect on the population gave rise to the ‘‘Minamata disease,’’ which was characterized by tremors, hallucination, and death amongst the subsistence fishing families living around the bay. Similar but reduced effects were observed in northern Ontario amongst the aboriginal Cree tribes along the English– Wabigoon river systems, which were contaminated by phenylmercury from a local paper mill that had used phenylmercury salts as slimicides. Fish consumption guidelines are issued in many areas where mercury, and hence methylmercury, is a likely health problem. Recognizing the health benefits that derive from eating fish, consumption guidelines are often specific about the frequency of consumption recommended and particular warnings relating to women in the first trimester of pregnancy or women who intend to become pregnant. Dimethylmercury has not been reported as being present in fish, although there are reports of its presence in ultratrace concentrations in mangrove swamps and human breath. Arsenic is another element that occurs in a variety of seafoods; both vegetable and animal sources contain varying amounts of several organic arsenic compounds as well as traces of inorganic arsenic. Measuring the total arsenic concentration in crab or lobster gives an alarmingly high number, but arsenic is present in the

T

3136

Trace Elements

Table 3 Arsenic species commonly found in seafood Formula

Name

AsO43


Arsenate [As(V)]

AsO33


Arsenite [As(III)]

(CH3)AsO32


Methylarsenate [As(V)]

(CH3)AsO22
(CH3)2AsO22
(CH3)2AsO
(CH3)3AsþCH2CH2OH (CH3)3AsþCH2CO2


Methylarsenite [As(III)] Dimethylarsenate [As(V)] Dimethylarsenite [As(III)] Arsenocholine Arsenobetaine

form of arsenobetaine, a compound whose acute toxicity has been estimated to be similar to table salt. Arsenobetaine is a water-soluble compound and is usually excreted through the urine within 48 hr. Commonly found arsenic species are listed in Table 3. Arsenic also finds its way into the seaweed, which is eaten in a number of foods by the population in Japan, Wales, and Canada. Arsenosugars contain a dimethylarsenic moiety and these sugars also appear in creatures that graze on seaweed. Not all arsenosugars have been identified, but some of the species are illustrated in Scheme 2. Tin appears in seafood as butylated tin compounds and as methyltin compounds in leachate from plastics. Tributyltin was used for many years to control barnacles and other marine growths on the hulls of ships. These organisms attached to the ships increase the drag on the ships and make them less energy efficient. Tributyltin leaches from the paint into the surrounding waters and can reach levels of concern in confined areas, such as harbors. Filter feeding organisms, such as oysters and mussels, as well as detritus feeders accumulate butyltin compounds and these compounds

appear to have estrogenic effects. International food and health organizations have proposed standards between 0.25 and 1.6 mg=K of the body weight for the human population. Although there is not much clinical evidence, researchers suspect that organotins have an adverse effect upon the immune system. Methyltins appear as stabilizers in a wide variety of plastics. But there appears to be no strong evidence to suggest that they migrate readily into food and water that are in contact with plastics, which has been heat-stabilized with organotin compounds. There is evidence that inorganic tin is methylated in the environment.

CONCLUSIONS Trace elements have huge impacts in many aspects of processing. Trace elements’ effects upon crystallization, whether in alloys or in commodity chemicals, are of enormous significance in forming the final product. Our understanding of the effects of trace elements will develop further as the techniques for their determination improve. Concern about the form in which trace elements appear in food or water will lead to increased demands for techniques that will determine the concentrations of the traces of species of elements. In this discussion, it is clear that the toxicities of different chemical species in which elements can be found vary widely. Although we may feel that the water we drink and the food we eat are well characterized, we should also know that our knowledge of trace elements and their impact at these minuscule concentrations is relatively recent. To protect human and animal health adequately, this area of development will continue to gain importance.

Scheme 2 Arsenosugars found in seaweed and in animals that consume seaweed.

Trace Elements

The impact of trace elements in a variety of scenarios shows that we need to remain vigilant about their impact on materials that are processed and used in commerce. The anticipation of ‘‘cradle to grave’’ management of materials may provoke increasingly stringent regulations for the disposal of wastes. The effects of trace elements in foods will grow in importance not only from a nutritional perspective but also from a security perspective and the consequences of new regulations upon monitoring and regulation. Although trace elemental concentrations are important, speciation of elements will continue to grow in significance.

3137

9.

10. 11.

REFERENCES 1. http:==www.wvgs.wvnet.edu=www=datastat=te= Glossary.htm (accessed May 25, 2005). 2. http:==www.netl.doe.gov=cctc=summaries=snox= snoxtmdemo.html (accessed May 25, 2005). 3. Liu, J.; Zheng, B.; Aposhian, H.V.; Zhou, Y.; Chen, M.L.; Zhang, A.; Waalkes, M.P. Chronic arsenic poisoning from burning high arsenic coal in Guizhou, China. Environ. Health Perspect. 2002, 110 (2), 119–122. 4. National Risk Management Research Laboratory. Mercury in Petroleum and Natural Gas: Estimation of Emissions from Production, Processing, and Combustion, EPA=600=R-01=066; United States Environment Protection Agency Office of Air Quality Planning and Standards: Washington, DC, 2001; 28. 5. Larouk, Z.; Pilkington, R. Creep formation and fracture of a Cr=Mo=V bolting steel containing selected trace element additions. Metal. Mater. Trans. A 1999, 30A, 2049–2058. 6. Fleming, H.D.; Ide, R.G. Determination of volatile hydride-forming metals in steel by atomic absorption spectrometry. Anal. Chim. Acta 1976, 83, 67–82. 7. Welz, B.; Melcher, M. Determination of antimony, arsenic, bismuth, selenium, tellurium and tin in metallurgical samples using the hydride atomic absorption technique. 1. Analysis of low alloy steels. Spectrochim. Acta B 1981, 36, 439–462. 8. Klueh, R.L.; Cheng, E.T.; Grossbeck, M.L.; Bloom, E.E. Impurity Content of Reduced-Activation

12.

13. 14. 15.

16.

17.

Ferritic Steels and the Effect on the ReducedActivation Characteristics. U.S. Department of Energy Office of Fusion Energy Sciences Contract DE-AC05-96OR22464, 1998. Rigby, C; Brindle, I.D. Determination of arsenic, antimony, bismuth, germanium, tin, selenium, and tellurium in 30% zinc sulphate solution by hydride generation inductively coupled plasma atomic emission spectrometry. J. Anal. Atomic Spectrom. 1999, 14, 253–258. http:==www.nrcan.gc.ca=es=es=energypicture= chap6_e.cfm (accessed May 25, 2005). Brindle, I.D; Brindle, M.E.; Le, X.-C.; Chen, H. Preconcentration by coprecipitation.Part I. Rapid method for the determination of ultratrace amounts of germanium in natural waters by hydride generation—atomic emission spectrometry. J. Anal. Atomic Spectrom. 1991, 6, 129–132. Cahn, J.W.; Hilling, W.B.; Sears, G.W. The molecular mechanisms of crystallization. Acta Mettalurgica 1964, 12, 1421–1439. http:==www.lut.fi=
hhatakka=docit=impure.html (accessed May 26, 2005). Mullin, J.W. Crystallization, 3rd Ed.; Butterworth-Heinmann: London, 1993; 255. Sevenants, M.R.; Sanders, R.A. Anatomy of an of-flavor investigation: the ‘‘Medicinal’’ cake mix. Anal. Chem. 1985, 56, 293A–298A. Westnb, C.E.; Merx, R.J.H.M.; de Koning, F.L.H.A. Effect of Iodized Salt on the Color and Taste of Food; UNICEF: New York, PD=95=009, 1995. Bloom, N.S. On the chemical forms of mercury in edible fish and marine invertebrate tissues. Can. J. Fish Aquat. Sci. 1992, 49, 1010–1017.

BIBLIOGRAPHY Hans Lo¨ffler, Ed. Structure and Structure Development of Al–Zn Alloys; Academie Verlag: Berlin (VCH Publishers, Inc.: New York), 1995. Istva´n Pais, J.; Benton Jones, Jr. The Handbook of Trace Elements; St. Lucie Press: Boca Raton, FL, 1997. Jancic, S.J.; Grootscholten, P.A.M. Industrial Crystallization; Delft University Press: The Netherlands (D. Riedel Publishing Company: Dordrecht, The Netherlands), 1984.

T

Transmission Electron Microscopy for Materials Science T Rolf Erni Nigel D. Browning Department of Chemical Engineering and Materials Science, University of California Davis, Lawrence Berkeley National Laboratory, Berkeley, California, U.S.A.

INTRODUCTION

MICROSCOPE SETUP

Transmission electron microscopy (TEM) comprises a complete repertoire of imaging, diffraction, and analysis techniques. The combination of direct imaging as well as local structural and chemical analyses makes TEM a powerful tool to study materials down to the atomic level. TEM is used to characterize the microstructure of materials, the constitution of phases and nanoparticles, the local arrangement of atoms, and particularly to study crystal defects such as grain boundaries, dislocations, precipitates, and their impact on the physical properties of the solid. The increasing need to study materials on the atomic scale, which is primarily given by the scaling down of electronic devices, the optimization of catalysts, the development of nanostructures, and also by the goal to obtain a basic understanding of physical and mechanical properties of solids, can in many cases be met by selectively applying one or a combination of experimental TEM techniques. In a transmission electron microscope, a highly coherent electron beam passes through a thin sample. The electron beam interacts with the sample and is transferred to the specimen’s exit plane. The electron wave at the exit plane is magnified in order to form an image or alternatively a diffraction pattern of the sample. A brief depiction of a transmission electron microscope is provided first. A short section about electron scattering qualitatively describes what types of electron–atom interactions are relevant for TEM. The most common experimental techniques are then explained consecutively. Diffraction mode including nanodiffraction and convergent beam electron diffraction (CBED) are explained first. Direct imaging techniques, such as bright-field (BF) and dark-field (DF) imaging, as well as high-resolution transmission electron microscopy (HRTEM) are dealt with in the following section. The third part of the experimental techniques is about scanning transmission electron microscopy (STEM) and Z-contrast imaging. Finally, analytical methods such as energy-dispersive X-ray spectroscopy (EDS), electron energy-loss spectroscopy (EELS), and energy-filtered imaging are discussed.

Starting at the top of a microscope column (Fig. 1), two types of electron sources are common, thermionic sources, i.e., W and LaB6 cathodes, and field-emission guns (FEG). Compared to thermionic sources, fieldemission sources show higher brightness as well as coherence and a smaller energy spread,[1] which significantly increases the information limit of the microscope. Below the electron source, the emitted electrons are accelerated. The acceleration voltage U of a microscope defines the primary energy of the electrons and hence the wavelength of the electron radiation. Most microscopes used in materials science are operated between 100 and 400 kV corresponding to an electron wavelength l between 3.7 and 1.6 pm, which is given by

Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120030635 Copyright # 2006 by Taylor & Francis. All rights reserved.

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi   eU l ¼ 4p h 2me eU 1 þ 2me c2 2

ð1Þ

where h is Planck’s constant, e is the elementary charge, me is the rest mass of the electron and c is the speed of light in vacuum. The optical part of the microscope starts below the accelerator. The magnetic electron lenses can be divided into three main lens systems: condenser, objective, and projector lens system (Fig. 1). The condenser lens system, including C1 and C2 aperture, is used to form the illuminating electron beam in front of the objective lens. The objective lens, which focuses the electron beam to the specimen plane, consists of two parts, termed pole pieces. The sample is located between these pole pieces. The objective aperture, mainly used for BF and DF imaging, is in the back focal plane of the objective lens. The selected area aperture, used for selected area diffraction (SAD), is in the first image plane below the objective lens. A stack of projector lenses magnifies the electron wave below the sample. Depending on its setting, either an image or a diffraction pattern is formed on the fluorescence screen or the recording medium. Stateof-the-art microscopes are equipped with charge coupled device cameras;[2] however, imaging plates and more frequently films are also used. 3139

3140

Transmission Electron Microscopy for Materials Science

Fig. 1 Microscope setup. The electron beam is indicated as a green line. The projector lens forms either an image (solid line) or a diffraction pattern (dashed line) on the screen. (View this art in color at www.dekker.com.)

The resolution in TEM is limited by lens aberrations. In contrast to optical microscopy, where by serially ordering concave and convex lenses, aberrations can be compensated and hence the wavelength of the radiation is resolution limiting; in TEM lens aberrations cannot be compensated since concave electron lenses are not feasible.[3] The objective lens is the crucial part for image defining the microscope’s resolution. The quality of the objective lens is described by the constants of spherical CS(
0.5–3 mm) and chromatic aberration CC (
1–2 mm). Recently, microscopes equipped with complex correctors for the spherical aberration have become available.[4]

ELECTRON–ATOM INTERACTIONS Two types of electron–atom interactions have to be considered: elastic and inelastic interactions (Fig. 2). Generally, the interaction between electrons and matter is strong. Thin samples have, thus, to be prepared in order to make them electron transparent.[5] An electron-transparent foil has a thickness between 10 and
200 nm and for perforated samples, a wedge-shaped thickness profile is typical.

Elastic Interaction Elastic scattering is a result of the electrostatic interaction between the incident electrons and the atoms in the sample. Electrons elastically scattered on passing through the sample are used to form an image or a diffraction pattern. Elastic scattering of an electron by an atom is described by the elastic scattering factor fe. The scattering factor is a function of the scattering angle y, which in the Mott formula is written as:[6] fe ðyÞ ¼

gme e2 l2 ½Z  fx ðyÞ 128p5 h2 eo sin2 y

ð2Þ

where g is the relativistic factor, eo is the permittivity of free space, and Z is the atomic number of the element. The function fx(y), whose Fourier transform describes the distribution of the electrons surrounding the nucleus, corresponds to the X-ray scattering factor.[6] Eq. (2) can be interpreted as follows: the momentum of an incident electron is affected by the electrostatic potential of the nucleus, described by the term Z, and by the electrostatic potential of the electrons surrounding the nucleus, described by fx(y). Including the prefactor, each of these terms shows a different y-dependence. For small and

Transmission Electron Microscopy for Materials Science

3141

T

Fig. 2 Electron–atom interactions. (A) Elastic electron–electron interaction dominates the scattering intensity at low and medium scattering angles; (B) Rutherford scattering at the nucleus causes high-angle scattering; and (C) electrons can excite atom-bonded electrons from the ground state to higher unoccupied states or to the vacuum level, element specific X-rays are produced when the excited electron returns to the ground state. (View this art in color at www.dekker.com.)

medium scattering angles, scattering is dominated by electron–electron interactions (Fig. 2). However, since fx(y) rapidly decreases with increasing scattering angle, elastic scattering at the nucleus, i.e., Rutherford scattering, becomes important at large scattering angles (Fig. 2). With Eq. (2) and the first Born approximation, the elastic differential scattering cross-section of an atom is given by: dse ðyÞ ¼ jfe ðyÞj2 dO

ð3Þ

where O is the solid angle and se is the total elastic scattering cross-section. Compared to elastic X-ray  2  fx  1026 and neutron (b2  1028) scattering, the total elastic scattering cross-section for electron scattering is large (se  1019). The probability that an electron is scattered more than once has to be taken into account by considering dynamical diffraction.[6] Inelastic Interaction Electrons also interact inelastically with the sample. Incident electrons can excite plasmons and phonons, and atom-bonded electrons can be excited from the ground state to higher unoccupied electron states.[7] In the de-excitation process, i.e., when the excited electron returns to the unoccupied core state (Fig. 2),

element-specific X-rays are produced which are analyzed by EDS. Inelastic interactions with atom-bonded, inner-shell electrons cause element-specific electron energy losses. The collective excitation of outer-shell electrons gives rise to energy losses corresponding to the plasmon energy of the solid.[7] The electron energy distribution below the sample can be measured by EELS. Phonon scattering, referred to as quasi-elastic scattering, causes a very small relative energy change (
25 meV). The electron energy distribution is hardly affected by phonon scattering. However, the change of the electron momentum caused by phonon scattering is observable, particularly as thermal diffuse scattering (TDS) in diffraction pattern.[8] Radiation Damage Inelastic and elastic interactions between the incident electrons and the sample can cause radiation damage. By elastic interactions, electrons can transfer a certain amount of kinetic energy to a nucleus.[9] The maximum energy that can be transferred, called maximum recoil energy, depends on the primary electron energy and the atomic weight of the element. If the maximum recoil energy exceeds the minimum energy required for displacing an atom, radiation damage, i.e., the creation of point defects, likely occurs.[10] By inelastic

3142

Transmission Electron Microscopy for Materials Science

electron–electron and elastic electron–nucleus interactions, electrons deposit energy to the sample. As an additional damage mechanism, the sample can be heated and even be destroyed by the amount of energy transferred to the sample. ELECTRON DIFFRACTION In diffraction mode, the projector lens system is adjusted in order to image the electron wave located at the back focal plane of the objective lens. What is seen on the screen is the intensity of this electron wave, which for coherent elastic scattering is called an electron diffraction pattern. Diffraction Spots, Bragg’s Law, and the Reciprocal Lattice For crystalline samples under plane-wave illumination, a diffraction pattern is observed as a spot pattern. The individual spots depend on the crystal orientation, its structure factor, and obey Bragg’s law. Bragg’s law states that the difference between the scattered k and the incident k0 wave vector is equal to a vector g of the reciprocal lattice:[6] k
k0 ¼ g

ð4Þ

For elastic scattering, the absolute value of the scattered and the incident electron wave vector are equal, given by the reciprocal value of the wavelength, jkj ¼ jk0 j ¼ 1=l. The angle between scattered and incident wave vector is the scattering angle y and the vector q, given by q ¼ k
k0

ð5Þ

is the scattering vector. For elastic scattering in a crystal and according to Bragg’s law (Eq. 4), only scattering

vectors corresponding to reciprocal lattice vectors are allowed, q ¼ g. The reciprocal lattice is related to the crystal lattice; a general reciprocal lattice vector g can be written as the sum of primitive vectors, g ¼ hg1 þ kg2 þ lg3, where h, k and l are independent integers termed Miller indices.[11] The set of primitive reciprocal lattice vectors defines the reciprocal lattice obeying the condition gi aj ¼ dij, where aj is a primitive crystal lattice vector and dij is the Kronecker delta 6 j and dij ¼ 1 for i ¼ j). symbol (dij ¼ 0 for i ¼ Ewald Construction The general relation for elastic scattering in Eq. (5) can be visualized by using the concept of the Ewald sphere.[11] The incident wave vector points in beam direction to the origin of the reciprocal lattice, i.e., hkl ¼ f 0 0 0 g. For elastic scattering, the set of scattered wave vectors k form a sphere surrounding the source point of k0 (Fig. 3). The points of intersection of Ewald sphere and reciprocal lattice form the set of allowed scattering vectors obeying Bragg’s law, q ¼ g. These points are observed as diffraction spots. The orientation of the crystal, its reciprocal lattice and the size of the Ewald sphere define which spots appear in the diffraction pattern. Tilting a sample under an invariant illumination means to rotate the reciprocal lattice around the origin of the reciprocal lattice, whereas the Ewald sphere remains unaltered. Two points have to be considered when using the Ewald construction for electron diffraction in TEM. First, for the electron wavelength (2 pm) is much smaller than a typical lattice spacing (0.4 nm), the radius of the Ewald sphere is much larger than the spacing between nearby reciprocal lattice points, jkj & jgi j. For small and medium scattering angles, the curvature of the Ewald sphere is almost negligible. Second, the reciprocal lattice ‘‘points’’ of a thin foil are elongated perpendicularly to the foil plane and form rods.

Fig. 3 Ewald construction. The white half-circle indicates the Ewald sphere in two dimensions. The points of intersection between the reciprocal lattice rods and the Ewald sphere form the set of reciprocal lattice points (bright) which obey Bragg’s law and appear as diffraction spots in the diffraction pattern. Zero-, first- and second-order Laue zone are indicated. For electron diffraction in TEM, the ratio between the radius of the Ewald sphere and the reciprocal lattice unit is larger than visualized in the figure. (View this art in color at www.dekker. com.)

Transmission Electron Microscopy for Materials Science

This circumstance is called shape effect.[1] The rods themselves are modulated according to the interference function.[6] What is seen in an electron diffraction pattern are the points of intersection between Ewald sphere, which approaches a plane, and the reciprocal lattice rods (Fig. 3). For small and medium scattering angles, an electron diffraction pattern basically corresponds to one plane of reciprocal space normal to the incident wave vector. This plane containing the origin of the reciprocal lattice is called zero-order Laue zone (ZOLZ). At large scattering angles, the curvature of the Ewald sphere causes the appearance of contributions of higher-order Laue zones (HOLZ). Kikuchi Lines Besides diffraction spots, which are caused by coherent elastic scattering, a diffraction pattern also contains contributions of incoherently scattered electrons. Quasi-elastic phonon scattering for instance, which is incoherent scattering, causes TDS, which with increasing sample thickness becomes apparent as a diffuse background. Combined incoherent and coherent elastic scattering gives rise to faint lines. These Kikuchi lines appear in pairs consisting of a deficient and an excess line.[12] Particularly for thicker samples, they can even dominate the contrast features of an electron diffraction pattern. Similar to diffraction spots, Kikuchi lines reflect the symmetry of the crystal.

3143

Selected Area Diffraction Diffraction spots reveal the symmetry and the spacing of the reciprocal lattice. Since both of them are directly related to the crystal lattice, diffraction patterns recorded for different crystal orientations can be used to determine crystal symmetry and lattice parameters. In order to see sharp diffraction spots, the sample has to be illuminated by a plane wave. A plane-wave illumination warrants that only one incident wave vector k0 goes into the elastic scattering relation, k  k0 ¼ q. A plane-wave illumination, however, means that the entire sample is uniformly illuminated. The information contained in such a diffraction pattern is not localizable. For a local analysis of the crystal structure, i.e., for the study of individual grains or selected phases, a selected area aperture can be inserted. Only the sample area selected by the aperture contributes to the diffraction pattern (Fig. 1). A plane-wave illumination, i.e., the appearance of sharp diffraction spots, can thus be maintained (Fig. 4). This technique is called SAD.[13] The smallest area that can be selected is about 0.5 mm given by the smallest selected area aperture. Convergent Beam Illumination In case a diffraction pattern of a smaller area is required, which is frequently the case when nanoscale

Fig. 4 Selected area diffraction. SAD pattern of uniaxially aligned lamellar g=a2 titanium aluminide consisting of hexagonal a2-Ti3Al lamellae with D019 structure and tetragonal (slightly distorted cubic) g-TiAl lamellae with L10 structure. g-TiAl is present in two twin variants causing two sets of reflections (green, blue). The spots labeled in red are caused by a2-Ti3Al. Main reflections are labeled in bold font, superstructure reflections of the tetragonal phases are in normal font. The interfaces between the individual lamellae are fully coherent which causes the overlap of certain diffraction spots. Incident beam direction for the tetragonal phases is f1 1 0g, for the hexagonal phase f1 1 0g. The streaks in y-direction are caused by the lamellar structure of the material. (View this art in color at www.dekker.com.)

T

3144

materials are studied, a convergent beam is formed, which is focused to a small area. The area illuminated by a convergent beam is in the nanometer range. For a convergent beam, a continuous set of incident wave vectors k0, each forming an independent Ewald sphere, has to be considered. A convergent beam illumination thus results in the appearance of diffraction disks. Due to the appearance of HOLZ lines (Kikuchi lines of higher-order Laue zones), a convergent beam diffraction pattern contains information, which cannot be accessed by SAD. From CBED, lattice parameters, point and space groups, local lattice strains, and the sample thickness can precisely be measured.[14] For FEG transmission electron microscopes, it is possible to demagnify the electron beam to a subnanometer electron probe. The diameter of the electron probe determines the sample area contributing to the (nano-) diffraction pattern. Electron nanodiffraction (END) makes possible the recording of a diffraction pattern from areas smaller than 1 nm.[15] Therefore, END is the appropriate technique to study the structure of nanoparticles, as for instance nanotubes and quantum dots. Although a nanoprobe is a convergent beam, the convergence angle is usually smaller than in CBED. As will be shown, END is fundamental for STEM.

IMAGING Transmission electron microscopy images show a projection of the sample in beam direction. The three-dimensional information is projected to a twodimensional intensity map. Direct imaging techniques are commonly divided into three groups: 1) conventional imaging covering BF and DF imaging techniques; 2) high-resolution imaging; and 3) STEM, which will be dealt with in the next section. Electrons passing through a crystalline sample are elastically scattered according to Bragg’s law. At the focal plane of the objective lens and of any other lens in the projector lens system, the angular distribution of the diffracted beams is observed. This angular distribution is studied when doing electron diffraction. At the corresponding planes where these beams interfere, an image is formed. An image is thus formed by the interference of Bragg-diffracted beams, however, each diffracted beam contains part of the full image information and can be used to form an individual image. This is the fundamental idea behind TEM imaging.

Transmission Electron Microscopy for Materials Science

individual beams differs. Defects as for instance dislocations, precipitates, and grain boundaries may locally change the diffraction conditions. Owing to distortions, the Bragg condition may locally be breached and a certain diffracted beam will not be excited from a particular area. Forming an image using this beam would result in low intensity for the area the beam is not excited. This type of contrast is termed diffraction contrast. Additionally, due to the presence of phases consisting of different elements and=or thickness changes, the attenuation of the forward scattered beam (i.e., the f0 0 0g reflection) may locally vary. This position-dependent attenuation of the forward scattered beam gives rise to mass-thickness contrast. If an image is formed with the forward scattered beam only, areas containing heavy elements and thicker areas will show lower intensity. Diffraction and mass-thickness contrast are both caused by an intensity change of a diffracted beam over the field of view. Since the intensity, specifically the amplitude, of a beam causes these types of image contrast, diffraction and mass-thickness contrast are referred to as amplitude contrast.[1] Although this concrete explanation of the amplitude contrast is appropriate for a basic understanding, it does not account for all image features. It is based on the kinematical approach to explain electron diffraction. In the kinematical approach, an electron can be scattered once and once it is scattered it will not change its momentum. For very thin samples of light elements, this approximation is sometimes justifiable. Dynamical Diffraction Due to the strong Coulomb interaction between electrons and atoms, electron diffraction is generally treated by dynamical diffraction. In the dynamical approach,[6] an electron can be scattered more than once. As a consequence of multiple scattering, the forward scattered beam exchanges its intensity with the diffracted beams and each diffracted beam exchanges its intensity with any other diffracted beam. The intensity of a diffracted beam is thus not a smooth function of the specimen thickness; it is strongly modulated and the modulation period, which is termed extinction distance, depends on the elements in the sample as well as on the particular reflection. For heavy elements, the extinction distance is generally shorter, whereas for light elements, it is larger. Typical extinction distances are in the range of 10–
200 nm.[16]

Diffraction Contrast and Mass-Thickness Contrast

Conventional Imaging

Though each diffracted beam contains part of the full image information, the information contained in the

Conventional imaging is usually performed at low or medium magnification. Though the interference of all

Transmission Electron Microscopy for Materials Science

Bragg diffracted beams forms an image, it is often of low contrast. In order to make use of the amplitude contrast, a single diffracted beam can be selected to form an image. By inserting an objective aperture at the back focal plane of the objective lens (Fig. 1) and selecting one particular beam, an image dominated by amplitude contrast can be formed. This is done by switching to diffraction mode, centering the aperture on the reflection, and switching back to imaging mode. Using the forward scattered beam, which for thin samples has the strongest intensity, is called BF imaging. Forming an image with any other beam, the imaging technique is called DFimaging. Bright-Field and Dark-Field images are in a qualitative way complementary; what is bright in the BF image appears with low intensity in the DF image and vice versa. The complementary application of BF and DF imaging is frequently used to locate (ordered) phases, which show certain (super)structure reflections the bulk material does not show. By choosing a reflection characteristic for a certain phase, the phase that causes the reflection will appear with high intensity and can clearly be located (Fig. 5). Because the intensity of a diffracted beam is modulated as a function of the crystal thickness, using one diffracted beam to form an image results for a

3145

T

Fig. 6 Weak-beam DF imaging. Weak-beam DF images of: (A) lamellar titanium aluminide showing thickness contours and (B) a spiral dislocation in Al-3 at% Ag, the white speckling is caused by silver-rich Guinier-–Preston zones. (View this art in color at www.dekker.com.)

wedge-shaped sample in alternating bright and dark lines (Fig. 6A). The spacing of these thickness contours, depends on the angle of the wedge and the extinction distance of the selected beam. Weak-beam DF imaging is an imaging technique which is based on a special diffraction condition, i.e., a two-beam case; besides the forward scattered beam, only one Bragg-diffracted beam is excited which is used to form an image.[1] Because weak-beam DF imaging is highly sensitive to the local diffraction conditions, defects as for instance dislocations can be imaged with high contrast (Fig. 6B). High-Resolution Imaging

Fig. 5 Bright-field and dark-field imaging: (A) BF image of lamellar g=a2 titanium aluminide; (B) corresponding SAD pattern (see Fig. 4); (C) and (D) DF images of the reflections marked in (B). Each of these reflections is characteristic for one twin variant of tetragonal g-TiAl appearing with high intensity in the corresponding DF image. (View this art in color at www.dekker.com.)

High-resolution transmission electron microscopy can be understood as a general information-transfer process. The incident electron wave, which for HRTEM is ideally a plane wave with its wave vector parallel to a zone axis of the crystal, is diffracted by the crystal and transferred to the exit plane of the specimen. The electron wave at the exit plane contains the structure information of the illuminated specimen area in both the phase and the amplitude.. This exit-plane wave is transferred, however affected by the objective lens, to the recording device. To describe this information transfer in the microscope, it is advantageous to work in Fourier space with the spatial frequency of the electron wave as the relevant variable. For a crystal, the frequency spectrum of the exit-plane wave is dominated by a few discrete values, which are given by the most strongly excited Bloch states,[17] respectively, by the Bragg-diffracted beams. An ideal information transfer is described by a constant, frequency-independent transfer function with a value of one. A transmission electron microscope is a nonideal information channel, the individual spatial frequencies of the exit-plane wave are differently affected by the transfer and interfere. The complex

3146

Transmission Electron Microscopy for Materials Science

transfer process, which is a function of the defocus, lens aberrations, and the degree of coherency, is described by the transmission cross coefficient.[18,19] Both amplitude and phase of the exit-plane wave are transferred, however they are being mixed up. What is finally seen in a high-resolution micrograph (Fig. 7) is dominated by the phase of the exit-plane wave. Therefore, HRTEM is known as phase-contrast imaging. It is the imaginary part of the transmission cross coefficient, called transfer function, which is crucial for the imaging process (Fig. 8). Due to the highly coherent imaging process in HRTEM, images can be recorded over a certain defocus range; there is not a well-defined value in which the sample is in focus. However, because the ideal case of a plane wave illumination is usually not met, the convergence angle of the incident beam and additionally the energy spread of the electron source constrict the coherency. Owing to this partial coherence, the transfer function is damped at high spatial frequencies (Fig. 8). Optimizing the transfer function of a microscope means approaching the ideal case of a frequencyindependent transfer, i.e., a constant phase shift over a large spatial frequency range. Working at Scherzer condition,[20] which gives a criterion for an optimum defocus DfScherzer balancing the effect of spherical aberration CS against defocus Df: DfScherzer

rffiffiffiffiffiffiffiffiffiffiffiffi 4 CS l ¼  3

ð6Þ

results in a broad low-pass with almost constant (negative) value. The phase of the diffracted beams is similarly shifted up to a maximum spatial frequency where the transfer function becomes zero (Fig. 8). This first zero-crossing at Scherzer condition defines the

Fig. 7 High-resolution transmission electron microscopy. HRTEM micrograph of lamellar g=a2 titanium aluminide. From top to bottom, first twin variant of tetragonal g-TiAl, hexagonal a2-Ti3Al, second twin variant of g-TiAl and again a2-Ti3Al. Incident beam direction for the tetragonal phases is f1 1 0g, for the hexagonal phase f1 1 0g. (View this art in color at www.dekker.com.)

microscope’s point-resolution: qffiffiffiffiffiffiffiffiffiffi 4 rS ¼ 0:65 CS l3

ð7Þ

As a consequence of the higher coherence in FEG electron microscopes, information beyond the point resolution significantly contributes to the image. In addition to the point resolution, it is adequate to define an information limit. The information limit corresponds to the highest spatial frequency, which contributes to the coherent imaging process. For FEG electron microscopes, there is a pronounced gap between point resolution and information limit (Fig. 8). However, owing to the strongly oscillating behavior of the transfer function, the information beyond the point resolution is not directly interpretable and suffers from image delocalization. Image delocalization means that image details are displaced from their true locations in the specimen and blurred over a certain area. Image delocalization depends on the defocus and the spatial frequency. In addition to the Scherzer condition, there is another criterion, called Lichte defocus DfLichte, which minimizes the overall image delocalization up to a maximum spatial frequency kmax:[21]

3 DfLichte ¼  CS ðkmax lÞ2 4

ð8Þ

Although image delocalization is minimized when working at Lichte defocus, it is still present and has to be considered when analyzing high-resolution micrographs of FEG microscopes. Numerical methods, such as reconstruction of the exit-plane wave by analyzing focal series of HRTEM micrographs, are used to access the full information up to the information limit.[22] There are different criteria to optimize the transfer function in HRTEM; the two cited are: Scherzer defocus for a nearly frequency-independent phase shift and Lichte defocus for minimizing image delocalization. The highly coherent imaging process, particularly in the case of FEG microscopes, causing image delocalization and focus-dependent contrast reversal[1] complicates a direct image interpretation. What is seen in a HRTEM micrograph strongly depends on the transfer function, i.e., on the experimental conditions, and therefore, has to be considered thoroughly. SCANNING TRANSMISSION ELECTRON MICROSCOPY Imaging Process The incremental way an image is acquired in STEM is fundamentally different from the single-shot imaging

Transmission Electron Microscopy for Materials Science

3147

1.0

T

t(k)

0.5

0.0

-0.5

-1.0

0

1

2

3

4

5

6

7

8

9

10

k [nm-1]

process in normal transmission mode. In scanning mode, the electron beam is demagnified in the condenser lens system to a small, convergent electron probe ˚ ), which is scanned across the sample. At each (
1–2 A scan position, the electron probe is locally scattered on passing through the sample. The microscope is in diffraction mode; i.e., each scan position produces a nanodiffraction pattern of overlapping diffraction disks. On scanning the electron probe, the position of the diffraction pattern remains invariant. Scan detectors, which measure the electron current of a part of its diffraction pattern for each scan position, are either located below the retractable screen or can be inserted above it. Transmitted electrons are detected as a function of the electron-probe position and the scattering angle. The electron current measured for a single scan position reflects the integral intensity of a part of its diffraction pattern. This value is finally represented as the corresponding pixel intensity in the STEM image. Scan detectors are distinguished according to the scattering-angle in whose range electrons are being detected. Bright-field detectors measure the forward scattered beam, annular dark-field detectors measure the integral intensity of low-order diffracted beams, and high-angle annular dark-field (HAADF) detectors collect electrons scattered to high angles over a large scattering angle range (Fig. 9).

Z-Contrast Imaging According to the principle of reciprocity,[23] TEM and STEM mode are equivalent in a qualitative way. For instance, what is observed in a HRTEM image is similar to what is seen in a high-resolution BF STEM image. From the principle of reciprocity also follows that a large electron source in TEM mode is equivalent to the use of a large STEM detector that integrates the

Fig. 8 Transfer function. The microscope transfer function t(k) as a function of the spatial frequency k calculated for Scherzer conditions at 300 kV. Dashed line, microscope equipped with a thermionic source (LaB6) and CS ¼ 1.3 mm; full line, FEG microscope, CS ¼ 1.2 mm.

intensity over a large scattering angle range. A large, although fictitious electron source implies loss of (spatial) coherence and therefore, by using a large electron detector in STEM incoherently formed images can be recorded. This is realized when doing HAADF STEM (Fig. 9). The large HAADF detector area significantly reduces the coherence of the imaging process. Both TDS and the convergence of the electron beam reduce the coherence even further.[24] Contrast reversal and delocalization, caused by the coherent imaging process in (HR)TEM, are therefore absent when doing HAADF STEM. The incoherent nature of HAADF STEM allows for a direct, unambiguous image interpretation. Additionally, since the HAADF detector measures electrons scattered to high angles, Rutherford scattering (i.e., the Z-contribution in Eq. (2) dominates. The signal recorded approaches a Z2 behavior, where Z is the atomic number of an element. HAADF STEM, also referred to as Z-contrast imaging,[25] is thus a chemical-sensitive imaging technique (Fig. 10).

Spatial Resolution The spatial resolution in STEM mode is given by the size, i.e., the full width at half maximum (FWHM), of the electron probe, which depends on the electron wavelength, the convergence angle, the defocus, and the constant of spherical aberration CS of the objective lens. The optimum probe size can be set according to Scherzer incoherent conditions,[20] where the optimum values for defocus DfS,inc and semi-convergence angle aS,inc are given by:

DfS;inc

pffiffiffiffiffiffiffiffi ¼  CS l

and

aS;inc

sffiffiffiffiffiffi 4 4l ¼ : CS

ð9Þ

3148

Transmission Electron Microscopy for Materials Science

Fig. 9 Z-contrast imaging and EELS. The electron probe is scanned across the sample. For each scan position, the HAADF detector collects the highangle scattering intensity. The intensity of one scanposition is represented as the corresponding pixel intensity in the STEM image. The forward scattered beam is not affected by the detector and can be used for EELS. (View this art in color at www.dekker. com.)

The electron probe size (FWHM), specifically the spatial resolution, the becomes:

rSTEM

qffiffiffiffiffiffiffiffiffiffi 4 ¼ 0:43 CS l3

ð10Þ

For a well-aligned electron probe, which can be done by observing the electron Ronchigram of an amorphous specimen area,[26] atomic resolution STEM images are feasible for FEG microscopes (Fig. 10).

ANALYTICAL TECHNIQUES The most common analytical techniques used in (S)TEM are EDS and EELS. The combination of STEM and EDS and=or EELS can be used to analyze the specimen locally; when stopping the scan process, the electron probe can be positioned at the point of interest. EDS analyzes element characteristic X-rays caused by inelastic electron–atom interactions. It is mainly used to measure the composition of the sample. The spatial

Fig. 10 High-resolution Z-contrast imaging. Zcontrast image of a grain boundary in SrTiO3 (perovskite structure) recorded in f0 0 1g direction. One of the unit cells framed in the micrograph is illustrated on the left, the Sr columns (bright) are at the corner of the unit cell, in the center there is a TiO column. The pure oxygen columns, black in the model, are not observable in the Z-contrast image. The atomic number (Z) contrast is apparent; with increasing atomic number (Z) of the elements, the intensity increases. (View this art in color at www.dekker.com.)

Transmission Electron Microscopy for Materials Science

resolution of EDS carried out in scanning transmission mode is given by the excited material volume, which in projection is in the nanometer range. EDS lacks the possibility to detect light elements,[1] i.e., elements of atomic numbers smaller than about six. EELS measures the electron energy distribution below the sample. Two types of spectrometer are common, O-spectrometers and post-column energy filters.[7] O-spectrometers, located between objective lens and screen, are advantageous when analyzing angledependent inelastic scattering in diffraction pattern. As an independent add-on, post-column filters are mounted below the microscope column. Post-column filters use a magnetic prism, which deflects the electron trajectories as a function of their energy. The use of a post-column spectrometer is particularly advantageous when doing Z-contrast imaging in combination with EELS (Fig. 9). The forward scattered beam, which is not affected by the HAADF detector, can be analyzed in the spectrometer with a spatial resolution down to the atomic level.[27] A typical EEL spectrum consists of the zero-loss peak, caused by elastically and quasi-elastically scattered electrons, plasmon-loss peaks, a downwardsloping background and element-specific ionization edges superimposed on this background.[7] The characteristic ionization edges can be used for a chemical analysis with the advantage compared to EDS that light elements down to He can be detected.[7] Due to the better energy-resolution in EELS (1=2 h), the material has a lower bulk density, or absolute maximum feed capacity is needed, special elements with increased volume are useful. The self-wiping profile of the pushing flight (and sometimes the trailing flight) has been transformed into a square channel profile. This modification accomplishes two functions. First, it directs more force acting on the material to be in the downchannel direction. Second, it creates up to 40% additional free volume in the element. Powders that tend to fluidize, especially those such as silica which have a very low initial bulk density, are significantly more difficult to feed than pellets or flakes. The first step to successful feeding depends upon eliminating (or at least minimizing) fluidization, and controlling the separation of already entrained air, prior to material entering the extruder. In general, the vertical drop should be as short as possible. Also, direct the feed to the down-turning section of the screw. This is the apex region for the counter-rotating machine and the barrel wall for the co-rotating extruder. Ideally the screw configuration should be designed to allow air to travel down channel for removal through a vent section rather than be trapped and forced to flow countercurrent and back out the feed throat. However, in order to be effective, a plastification zone typically must be backed up by a restrictive element. This element then blocks the air from traveling downstream. In this case, special elements such as increased volume (undercut) or single flight (SF) elements play an important role in feed introduction. The choice of either increased volume or SF elements depends upon the amount and degree of air separation required. For a relatively low amount of air, the undercut element again provides a greater free volume and a more open path for air to flow backwards. However, as more air is forced back to the feed throat, the material fluidizes and the undercut elements lose their effectiveness. Under these circumstances the SF should be considered. Unlike the undercut element, which has greater free volume, the SF has approximately 15% lower free volume per unit length than a standard element. The SF elements do not allow air to flow easily in a countercurrent direction and, therefore, force it to flow past the restriction in the plastification section. These elements function in this manner as a result of the severe flow restrictions through the apex caused by the wide crest. These crests create a positive

T

3176

displacement flow greater than in any other co-rotating twin-screw element.

Downstream Feeding In many compounding operations it is necessary to split the feed streams. This may be required in order to: 1) achieve disperse phase size of an impact modifier; 2) retain aspect ratio of reinforcing fiber filler; or 3) obtain high level of loading for either low bulk density filler or incompatible low viscosity additives. The most efficient way to add solids in a downstream location is to use a twin-screw side feeder. The twin-screw side feeder is normally a co-rotating device with high OD=ID ratio (1.8). It has several advantages over single-screw side or top feeders. First, the twin-screw has better solids conveying characteristics, since a twin-screw device does not totally rely on drag flow. The twin-screw also has a wider longitudinal as well as a larger total cross-sectional discharge opening than a single-screw side feeder, and therefore provides lower pressure and more uniform feed introduction. Typically there is a vent above the side feed opening in the barrel. This permits entrained air to escape. For very low bulk density material containing significant air, the vent opening is moved one barrel upstream to avoid material particles getting entrained in the air as it exits the extruder. By having the air travel upstream before discharge, requires it to travel a path around the screw channel. This disrupts the air flow and causes the solid to disengage. Liquid additives are mostly added downstream of the plastification section because they tend to lubricate the pellets or cause powders to agglomerate in the feed throat. If a significant amount of liquid is to be incorporated, it can be added at several locations. The most effective method for low viscosity liquid incorporation is to inject into a fully filled distributive mixing section. This requires a pressure injection valve and positive displacement pump. For small amounts of compatible liquid, non-pressurized injection into a low degree of fill area of the screw configuration may also be acceptable.

Plastification Mechanisms Plastification of polymeric material requires energy to be transferred from an outside source into the material. In the twin-screw extruder, this energy transfer occurs through both mechanical and thermal mechanisms. However, as the extruder gets larger, the surface to volume ratio decreases significantly. Therefore, mechanical energy transfer is the dominant mechanism for plastification.

Twin-Screw Extrusion

Thermal Heat Transfer Normally, thermal energy is introduced through electric heaters surrounding the barrel or heat transfer medium that is pumped through barrel bores. As mentioned previously, smaller extruders can introduce a greater percentage of energy through heat transfer. Conversely, they can also loose a higher percentage of heat through heat transfer. If too much heat is lost in this manner, then on scale up to a production unit, the same percentage of heat cannot be removed. This will result in a higher discharge temperature. It is therefore very important that laboratory or development extruders run as close to adiabatic conditions as possible. Mechanical Heat Transfer In most cases, the majority of energy required for plastification comes from the mechanical input of the screw configuration. In counter-rotating intermeshing systems, material is compressed and deformed by reduction of screw pitch. In co-rotating systems, kneading blocks are the primary tool used to accomplish this task. The amount of mechanical energy introduced depends upon not only the number of kneading blocks, but also the configuration within the plastification zone, the screw rpm, and throughput rate. Specific mechanical energy (Sme), the energy introduced per pound or kilo of product, increases with rpm, especially when the plastification section is backed-up with a restrictive screw element. A second restrictive element further increases Sme. Mixing Mechanisms Mixing requirements for polymer compounding can be divided into two basic disciplines—dispersive and distributive. Dispersive mixing breaks down a particle into smaller units, while distributive mixing homogenizes the spatial relationship of the particles (whether dispersed or not). The basic building blocks for mixing in the co-rotating twin-screw are kneading blocks and special mixing bushings. These special elements include toothed mixing units and blister rings, both standard geometry and self-wiping. It is not surprising that the mixing element selection for distributive and dispersive blending processes is different. The distributive mixing profile uses narrow kneading blocks to maximize the number of flow divisions per machine length. In dispersive mixing, wide disc kneading blocks are used to increase shear stress applied to the material. These elements are typically backed up with a restrictive

Twin-Screw Extrusion

kneading block or screw bushing to increase the degree of fill as well as residence time. The number of restrictive elements that can be used is limited due to the resulting temperature buildup and potential material degradation. Therefore, in order to transmit increased stress to the material, a number of shorter mixing sections (1–2D long) are often more efficient than one long mixing section as this permits elastic materials to relax in the conveying sections between the mixing areas. For very low viscosity products, introduce sufficient energy for dispersive or distributive mixing by adjusting the feed sequence, such that only a small portion of the solvent or dilution oil is added in the upstream part of the twin-screw extruder and therefore high mechanical stress transfer can be achieved. The remainder of the diluent is introduced in the latter sections of the machine.

3177

end of the machine. For fully intermeshing counterrotating machines, no device for discharge assistance is required.

CONCLUSIONS There is a wide range of process applications for twinscrew extrusion. The two most prolific are profile extrusion and compounding. The former utilizes traditional counter-rotating geometry with either parallel or conical screws. The majority of compounding applications are run on co-rotating fully intermeshing extruders. The other twin-screw machine types, while not as widely used, each have specific areas where they are the preferred geometry. For a more detailed analysis of the operating mechanisms of the various geometries please see Refs.[29–33].

Devolitalization/Degassing In compounding lines, devolatilization typically involves removal of entrained air, moisture contained in the incoming feedstock, or byproducts from any reaction between recipe components. The amounts are generally less than 1% or 2% and typically less than 1%. Therefore, the staged vacuum set-ups and stripping techniques used in devolatilization processes where 10% to 60% volatile must be removed, are not necessary. However, even in compounding, devolatilization is a critical step, because while it is necessary to remove these volatile, it serves as a location for the potential introduction of contamination. The screw configuration in the vent area is typically comprised of screw bushing with a pitch of between 1 and 2D. This design permits sufficient residence time at a lower degree of fill to remove volatiles while maintaining the polymer within the bounds of the screw flight.

Discharge Material is discharged from most twin-screw compounding operations as pellets or specialized forms, such as sheets, tubes, ropes, or other more complicated profiles. In order to push material through these dies, the machine must generate the appropriate pressure. This can range from 200 psi or less for strand dies, and up to 5000 or so psi for sheet and other profile dies. When the discharge pressure exceeds 2000 psi, or precise product gauge control is critical, a singlescrew extruder or melt gear pump is typically used to generate the pressure in the co-rotating twin-screw. In large-scale polyolefin pelletizing operations, rates can be dramatically increased by integrating a gear pump into the system.[28] Also, fiber spinning and sheet extrusion operations typically use a gear pump at the

REFERENCES 1. Weinrich, C. Direct extrusion of wood flour= plastics composites on conical counterrotating twin screw extruders. In Wood-Plastic Conference, Conference Proceedings, Baltimore, MD, December 5–6, 2000; 207–215. 2. Dardenne, D.S. WoodtruderTM (Patents Pending) system for extrusion of wood fiber polymeric composites. In Wood-Plastic Conference, Conference Proceedings, Baltimore, MD, December 5–6, 2000; 187–188. 3. Jackson, S.M. Advanced process technology for manufacturing wood-polymer composites. In Wood-Plastic Conference, Conference Proceedings, Baltimore, MD, December 5–6, 2000; 21–28. 4. Wiedmann, W. Improving product quality through twin-screw extrusion and closed-loop quality control. In Food Extrusion Science and Technology; Kokini, J.L., Ho, C., Karwe, M.V., Eds.; Marcel Dekker: New York, 1992; 539–570. 5. Brauer, F. Continuous production of polyurethanes on a twin-screw extruder reactor. In PPS Summer Meeting Conference Abstracts, Amherst, MA, August 16–17, 1989; 7D. 6. Uberall, M.; Kapfer, K.; Brauer, F.; Trippler, R. Devolitalization operations for high volume polyolefin solutions. In Polyolefins IX, Conference Proceedings, Houston, TX, February 26–March 1, 1995; 323–338. 7. Andersen, P.G.; Kite-Powell, K.K. Method of Removing Liquids from Solids. US Patent 5,151,026, September 29, 1992. 8. Friedrich, R. Powder compression with corotating twin-screw extruders. J. Powder Bulk Solids Technol. 1980, 4 (4), 27–32.

T

3178

9. Beecher, E.D.; Carr, M.E.; Grillo, J.G. Starch compounding on co-rotating twin screw extruder: starch as an encapsulation medium for a controlled release application. SPE Antec Proceedings, Boston, MA, May 7–11,1995; 2037–2041. 10. Andersen, P.G. Twin screw extrusion guidelines for compounding nanocomposites. SPE Antec Proceedings, San Francisco, CA, May 5–8, 2002; 219–223. 11. Friedrich, R. Comparison between processing polymers and propelling charge powders on continuous co-rotating twin-screw mixers, 18th International Annual Confernece of the ICT, July 1–3, 1987. 12. Kapfer, K.; Schneider, H. Production of compounds with high filler or fiber loading on screw kneaders. SPE Antec Proceedings; Nashville, TN, May 4–8, 2003; CD-ROM. 13. Andersen, P.G.; Dickens, D. Selection criteria for concentrate and masterbatch compounding. AMI Proceedings Thermoplastic Concentrates 2002, New Orleans, LA, January 16–18, 2002. 14. Kapfer, K. Current developments in twin-screw design and its application in the preparation of polymer blends. SPE Antec Proceedings, Atlanta, GA, April 18–21, 1988; 96–101. 15. Laughner, M.; Parikh, D.; Walton, K. New developments in metallocene ethylene elastomers for automotive applications. In Polyolefins 2001, Proceedings, Houston, TX, February 25–28, 2001; 339–367. 16. Liberto, L.P. Ed. Powder Coating: The Complete Finishers Handbook; Powder Coating Institute: Alexandria, VA, 1999; p 12. 17. Charpentier, J.C. Compounding color toners. Proceedings Toners & Photoreceptors 2001, Santa Barbara, CA, June 10–13, 2001; Session 13. 18. Beecher, E. Techniques for short run continuous compounding of adhesives and sealants. Presented at ASC 2001 Fall Convention. New Orleans, LA, October 21–24, 2001. 19. Burbank, F.R.; Jackson, S.M. New twin-screw element design for elastomer compounding. SPE Antec Proceedings, New York, NY, May 2–6, 1999, 225–229. 20. Andersen, P.; Haering, E.; Kapfer, K. Understanding high rate and high speed compounding

Twin-Screw Extrusion

21.

22.

23.

24.

25.

26.

27.

28.

29.

30. 31.

32.

33.

on co-rotating twin-screw extruders. SPE Antec Proceedings, Toronto, Canada, April 27–May 2, 1997; CD-ROM. Kapfer, K.; Haering, E. Deeper screw flights offer new opportunities for co-rotating twin-screw extruders. SPE Antec Proceedings, San Francisco, CA, May 5–8, 2001; CD-ROM. White, J.L. Modular counter-rotating twin screw extruders: intermeshing and tangential. In Polymer Mixing: Technology and Engineering; White, J.L., Coran, A.Y., Moet, A., Eds.; Hanser: Munich, 2001; 167. Thiele, W.C. Counterrotating intermeshing twin-screw extruders. In Plastics Compounding: Equipment and Processing; Todd, D.B., Ed.; Hanser: Munich, 1998; 46–70. Brown, T. Extrusion of rigid PVC products. In Vinyl RETEC and Tutorial 0 96 Technical Papers; Cincinnati, OH, October 15–16, 1996; 57–67. Skidmore, R.H. Method of Separating an Insoluble Liquid from Polymer Composition. US Patent 3,742,093, June 26, 1973. Howland, C.; Erwin, L. Mixing in counter rotating tangential twin screw extruders. SPE Antec Proceedings, Chicago, IL, May 2–5, 1983; 113–116. Foster, R.W.; Lindt, J.T. Bubble free devolitalization in counter-rotating non-intermeshing twin screw extruders. Polym. Engng. Sci. 1990, 30 (7), 424–435. Schuler, E.W. Energy efficiency in high volume plasctics compounding operations. SPE Antec Proceedings, Chicago, IL, May 2–5, 1983, 924–927. Herrmann, H.; Jakopin, S. A comprehensive analysis of multi-screw extruder mechanisms. SPE Antec Proceedings, Montreal, Canada, April 25–28, 1977; 481–486. White, J.L. Twin Screw Extrusion; Hanser: Munich, 1990. White, J.L.; Coran, A.Y.; Moet, A., Eds.; Polymer Mixing: Technology and Engineering; Hanser: Munich, 2001. Manas-Zloczower, I.; Tadmor, Z., Eds.; Mixing and Compounding of Polymers: Theory and Practice; Hanser: Munich, 1993. Todd, D.B., Ed. Plastics Compounding: Equipment and Processing; Hanser: Munich, 1998.

Use of Lipases to Isolate Polyunsaturated and Oxygenated Fatty Acids and Form Value-Added Ester Products

U

Douglas G. Hayes Department Biosystems Engineering and Environmental Science, University of Tennessee, Knoxville, Tennessee, U.S.A.

INTRODUCTION Polyunsaturated and oxygenated fatty acids, obtained from triacylglycerols (TAG) of several different plant and animal species, are valuable materials feedstock for value-added products in a variety of industries: food, pharmaceutical, cosmetics, and paints and coatings. The acyl species, their chemical structure, and their most abundant sources are summarized in Table 1. In contrast to inexpensive C16 and C18 saturated and D9-unsaturated acyl groups, such as palmitic (16:0), stearic (18:0), oleic (18:1-9c), linoleic (18:2-9c, 12c), and a-linolenic acid (ALA; 18:3-9c, 12c,15c), recovered from the oil of soybean and other common sources, and C4–C16 saturates from palm oil and milk fat, polyunsaturated and oxygenated acids are derived from less common sources, and particularly for polyunsaturated fatty acid (PUFA), are typically present at only 20–40% purity. Although common C4–C18 acyl groups are readily isolated and modified by chemical means, PUFA and oxygenated fatty acids are susceptible to degradation at their double bonds and oxygenated groups, respectively. Biocatalytic means of isolation and modification provide mild operating conditions (low temperature and pressure, near-neutral pH, and the absence of toxic materials such as catalysts and harmful solvents), which will prevent degradation and promote an environmentally-friendly workplace.[1,2] The positional, substrate, and regioselectivity of lipases will lead to a narrower product distribution compared to chemical methods. The higher selling costs of PUFA and oxygenated fatty acid-derived products will help offset the high materials costs of enzymes, in contrast to common C16 and C18 acyl groups and their esters. This entry will briefly review the applications of PUFA and oxygenated fatty acids and the biocatalytic properties of lipases in a broad sense, discuss the lipase-mediated isolation of PUFA and oxygenated free fatty acids (FFA), and review lipase-catalyzed synthesis of PUFA and hydroxy FFA-enriched ester products and their applications. Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120030531 Copyright # 2006 by Taylor & Francis. All rights reserved.

APPLICATIONS OF POLYUNSATURATED AND OXYGENATED FATTY ACIDS D4–D6 PUFA such as docosahexaenoic (DHA), eicosapentaenoic (EPA), arichidonic (AA), and g-linolenic (GLA) and their derivatives are gaining popularity as food additives, or ‘‘nutraceuticals,’’ due to the many medical and nutritional benefits they present.[3] (DHA, EPA, and ALA are frequently grouped together in application-oriented discussions as ‘‘n-3’’ fatty acids, since each contains a double bond located three positions away from the terminal carbon atom on the acyl chain and provide similar medical benefits.) For instance, a literature search on patents involving DHA and AA from the year 1999 to the present resulted in 1014 and 878 hits, respectively. Of great benefit to world population suffering from an epidemic of obesity, a diet rich in PUFA is known to decrease blood TAG levels, blood pressure, and low density lipoprotein (or ‘‘bad’’) cholesterol. They also have many applications relating to brain and nervous system development in infants, treatment of arthritis and osteoporosis, and possess anti-inflammatory properties. In addition to the applications listed above, AA dietary supplements have been reported to increase muscle mass and reduce intravenous hemorrhaging in premature infants. AA, DHA, EPA, GLA, and ALA possess cis double bonds in positions whereby they can serve as oxidation substrates for the enzyme lipoxygenase to produce hydroperoxides, which are valuable precursors of prostaglandins, leukotrienes, and flavor ingredients: ketones, alcohols, and aldehydes, and are chiral synthons for combinatorial libraries employed in drug discovery. Although not technically a PUFA, petroselinic acid (18:1-6c) is listed in this category because it possesses a double bond positioned at C6, and shares many of the same nutritional and medical applications as DHA and EPA. Petroselinic acid is employed as a cosmetic ingredient for hair growth and skin rehydration products. Conjugated linoleic acids, CLAs, lipids that occur naturally in milk fat and beef tallow are becoming increasingly popular materials with applications similar to those

3179

3180

Use of Lipases to Isolate Polyunsaturated and Oxygenated Fatty Acids and Form Value-Added Ester Products

Table 1 Polyunsaturated and oxygenated free fatty acyl species FFA product

Chemical structure

Sources

Arachidonic (AA)

20:4-5c, 8c, 11c, 14c

Single-cell oils (25%)

Crepenynic

18:1-9c, 12-yne

Crepis alpine oil

Docosahexaenoic (DHA)

22:6-4c, 7c, 10c, 13c, 16c, 19c

Fish oils; single-cell oils (5–20%)

Eicosapenaenoic (EPA)

20:5-5c, 8c, 11c, 14c, 17c

Fish oils; single-cell oils (5–20%)

g-Linolenic (GLA)

18:3-6c, 9c, 12c

Blackcurrant (Ribes nigrum, 15–20%) borage (Borago officinalis, 20–25%), or evening primrose (Oenothera biennis, 8–14%) oil

Petroselinic

18:1-6c

Coriander (Coriandrum sativum) oil (15.3%)

Conjugated linoleic acids (CLAs)

18:2-9c, 11t; 18:2-10t, 12c

Milk fat; beef tallow; modification of ricinoleic acid; free radical isomerization of a-linolenic acid (equimolar mixture of two CLA isomers)

20:1-5c, 22:2-5c,13c

Meadowfoam (limnanthes alba) oil (83.5%)

20:3-5c, 11c, 14c, 20:4-5c, 11c, 14c, 17c

Biota orientalis oil (15.6%)

Dimorphecolic

S-18:2-10t, 12t,OH-9

Dimorphotheca pluvialis oil (61.8%)

Lesquerolic and auricolic

R-22:1-11c, OH-14, R-22:2-11c, 17c OH-14

Lesquerella fendleri oil (57%)

Ricinoleic

R-18:1-9c, OH-14

Castor (Ricinus comminus) oil (90%)

Vernolic

18:1-9c,-epoxy-12,13c

Vernonia galamensis (77–81%) or Euphorbia lagascae (60-65%) oil

listed above for DHA, with additional employment in the treatment of diabetes.[4] An equimolar mixture of the CLAs 18:2-9c, 11t and 18:2-10t, 12c are synthesized from free radical isomerization of ALA.[4] Long-chain PUFA from meadowfoam and Biota orientalis oils (Table 1) have applications as oleochemicals, surfactants, and cosmetics; B. orientalis PUFA may have activity in lipid metabolism. Crepenynic acid has received attention as a chemical feedstock, including for use in paints and coatings, due to its unique carbon–carbon triple bond. The second category, hydroxy- and epoxy-containing acyl groups (and their esters), have numerous applications as chemical feedstocks in lubricants, paints and coatings, food and cosmetics emulsifiers, nylon synthesis, laxatives, disinfectants, etc.[5,6] The utility of ricinoleic acid and its derivatives, for instance, the most commonly recognized member of this category, has existed for at least a century, demonstrated by US patents issued nearly 100 years ago, and 577 patents issued between the years 1990 and 2004.

LIPASES Lipases play the specific role of forming and hydrolyzing ester bonds involving long-chain carboxylic,

or fatty, acids (Fig. 1).[7] Lipase-catalyzed hydrolysis, or ‘‘lipolysis,’’ typically occurs at liquid–liquid or liquid–solid interfaces, while esterification and acyl exchange (reactions 2 and 3–5 of Fig. 1, respectively) are catalyzed by solid-phase lipase dispersed in lowwater, nonaqueous, media.[8] Lipases possess regio-, stereo-, and substrate selectivity to control the product distribution.[9] For instance, many lipases are strongly regioselective toward primary hydroxyl groups, allowing them to utilize the 1- and 3- position of glycerol as substrate but not the 2-position. Lipases which fit this description are categorized as ‘‘1,3-regioselective’’ while those that do not are ‘‘random.’’ The strong stereoselectivity of lipases has made them common tools in the separation of racemic acid or alcohol mixtures in the pharmaceutical industry. The fatty acyl substrate selectivity of lipases is based on chain length and the position, type (cis or trans), and number of double bonds. For instance, lipases discriminate against substrates with double bonds near the carbonyl terminus, such as the D4–D6 PUFA. Geotrichum candidum lipase has unique substrate selectivity, with a strong preference toward D9 acyl groups, such as oleic, linoleic, and a-linolenic acids. There are several lipases commercially available which possess high activity and thermo-, storage, and operational stability resulting from enhanced knowledge of the fundamental

Use of Lipases to Isolate Polyunsaturated and Oxygenated Fatty Acids and Form Value-Added Ester Products

O 1.

OR

O + H2O

+ ROH

HYDROLYSIS

OH

O OH

OR

O +

3.

R OH

+ ROH OR

O 4. +

O ACIDOLYSIS

+

OH

OR +

OH

O

O 5.

+ H2O

O

ALCOHOLYSIS

OR

OR

U

O

ESTERIFICATION

+ ROH

2.

3181

O OR

O

TRANSESTERIFICATION

OR

O

O

+

OR

OR

Fig. 1 Reactions catalyzed by lipases. (From Ref.[6].)

biochemistry of lipases, isolation of lipases from extremophiles, expression of lipase genes in recombinant hosts, directed evolution, and enhanced immobilization technology (e.g., NovozymeTM, Lipozyme RM IMTM, and Lipozyme TL IM, immobilized Candida antarctica, Rhizomucor miehei, and Thermomyces lauginosus lipases, respectively, from Novozymes, Inc., Franklinton, North Carolina, U.S.A.).[7]

the reaction medium or enzyme preparation. Its occurrence increases with the water content of the reaction medium and temperature. Although acyl migration is difficult to prevent, its extent is significantly lessened when the residence time of the reaction medium in the presence of the biocatalyst is minimized.

THE IMPORTANCE OF LIPID SEPARATIONS IN ENZYMATIC PROCESSES ACYL MIGRATION PLAGUES 1,3-POSITIONAL SELECTIVITY OF LIPASES 1,3-Selective lipases are particularly useful for lipolysis of TAG that possess acyl groups of interest only in the 1- and 3-glycerol positions and for forming ‘‘structured’’ TAG, both of which are discussed later. However, acyl migration, a non-enzymatic intramolecular acyl exchange within monoacyl- and diacyl-glycerol, or MAG and DAG, respectively, greatly reduces the anticipated yield and purity. Specifically, the following two reactions occur: I:

II:

2
MAG Ð 1
ð3
ÞMAG 1; 2
DAG Ð 1; 2
ð2; 3
ÞDAG

Reviewed elsewhere,[10] acyl migration is frequently catalyzed by electrostatically charged materials in

High yield of PUFA incorporation can only occur when the acyl donor is highly enriched in PUFA due to the relatively poor selectivity of all lipases toward D4–D6 FFA, with Pseudomonas sp. and C. antarctica lipase being the least discriminatory. Thus, when synthesizing structured or PUFA-enriched TAG from raw sources such as seed or fish oil or lard, one must employ a PUFA source that is highly pure in order to achieve a high yield. Thus, lipid separation steps are a very important component of the overall process. The most commonly employed lipid separation steps integrated with biocatalysis are listed in Table 2. Molecular distillation is perhaps the most common and rapid of the separation methods, but can degrade double bonds and involves a high energy cost. Saponification is the preferred choice for isolation of FFA from esters, but must be employed in a buffered solution when applied to hydroxy fatty acyl mixtures. Crystallization and precipitation can be slow

3182

Use of Lipases to Isolate Polyunsaturated and Oxygenated Fatty Acids and Form Value-Added Ester Products

Table 2 Separation methods that accompany lipase-catalyzed reactions Method

Selectivity

Application with enzyme reactions

Short-path molecular distillation

Separates molecules by their molecular weight

Fractionation of MAG, DAG, TAG, and fatty acid esters from lipolysate or reaction medium

Saponification

Removes FFA from esters

Downstream removal of FFA

Crystallization and precipitation

Fractionates based on melting point

In situ removal of long-chain and hydroxy FFA and polyol monoesters

Urea inclusion compounds

Fractionation of polyunsaturates from saturates

Downstream processing of lipase-catalyzed PUFA-enriched FFA

High performance and low pressure liquid chromatography (adsorption)

Separation of lipid classes, separation of lipids by molecular weight and degree of unsaturation

Impractical for most preparative or large-scale processes

Solvent extraction (‘‘countercurrent distribution’’)

Separation of neutral lipid classes and PUFA from saturated FFA

Readily scaled up, but poorly selective

processes that require high-energy costs due to the low temperatures and process time required; thus, they are impractical for large-scale processes. Exceptions to this rule may be the isolation of MAG, polyol monoesters, and long-chain saturated and hydroxy FFA and their derivatives, which have melting point temperatures slightly below ambient (10–20 C). A cold trap can be implemented into the bioreactor design for continuous isolation of MAG, thermodynamically driving the desired reaction in the forward direction. Chromatographic methods provide the greatest selectivity among the methods listed in Table 2; however, chromatography is very difficult to adapt to large-scale processing and is quite expensive. Solvent extraction, a technique that isolates FFA by their degree of saturation and=or the presence or absence of hydroxy groups, is poorly selective. It was employed frequently in the 1940s–1960s in the form of multistage, countercurrent, contacting equipment (known as ‘‘countercurrent distribution’’) before its replacement by chromatographic techniques for analytical and preparative scale applications. Urea inclusion compounds (UICs) have also been employed for over 50 years to separate lipids primarily by the degree of saturation.[11] UICs represent a process that has a high selectivity (greater than solvent extraction but less than molecular distillation or chromatography), can be readily scaled up, and involves much lower energy costs.[11] UIC-based fractionation is complementary to lipase-catalyzed selective esterification to purify PUFA.[11] Moreover, lipase-catalyzed esterification of an FFA mixture will remove common D9-unsaturated C18 FFA and UIC-based fractionation will remove saturated and monounsaturated FFA, resulting in a PUFA-enriched FFA product.[11]

ISOLATION OF FFA BY LIPOLYSIS Traditionally, FFA are isolated from a degummed and bleached seed oil by the Colgate–Emery process, which employs steam at high pressure and temperature (typically 250 C and 5 MPa) to cleave the TAG ester bonds. The high temperature of this process is known to degrade PUFA and oxygenated FFA. In addition, the Colgate–Emery process consumes a large amount of energy, about 800 kJ of energy per kg of oil. Although lipolysis may serve as a low-cost and environmentally safe alternative to the Colgate–Emery process for FFA isolation, its widespread use in the oleochemical industry is blocked by the high material costs of enzymes. Lipolysis will have a greater impact upon PUFA and oxygenated FFA isolation due to the susceptibility of double bonds, epoxy, and hydroxyl groups to degradative side-reactions and the higher selling costs of PUFA-related products. Partial isolation of PUFA with D4–D6 unsaturation occurs during lipolysis due to the inability of lipases to rapidly utilize TAG that contain PUFA groups.[12] Moreover, the PUFA are recovered from the MAG, DAG, and TAG molecules of the reaction mixture, while common acyl groups are preferentially released, then removed from the glycerides by saponification. (Alternatively, TAG are subjected to alcoholysis, reaction 3 of Fig. 1. The ester products are removed from the MAG=DAG=TAG by short-path molecular distillation. Since lipolysis and alcoholysis yield similar degrees of purification, selection between the two is based on economic considerations.) C. rugosa lipase has been employed in most instances for selective lipolysis. Due to the inability of many lipases to hydrolyze TAG with 2 or 3 D4–D6 PUFA groups, glyceride

Use of Lipases to Isolate Polyunsaturated and Oxygenated Fatty Acids and Form Value-Added Ester Products

3183

Saturated Salt Solution

U

aW ω

aW

Lipoly -sate ω aW aW Saturated Salt

ω Lipoly -sate

Solution

products contain PUFA at only moderately high concentrations (40–75%) and yield (60–90%). The purity and yield are not strongly affected by reaction medium type nor by addition of fresh lipase during the time course of reaction. Pseudomonas sp. lipases generally produced the highest degree of lipolysis (79% or greater), but exhibited a much lower degree of discrimination against PUFA acyl groups than most other lipases. 1,3-Selective lipolysis is employed to isolate the hydroxy FFA listed in Table 1 because, unlike ‘‘random’’ lipolysis, poly(hydroxy acid) byproduct formation cannot be catalyzed. The employment of 1,3-selective lipases is particularly beneficial for recovery of lesquerolic and dimorphecolic acids from lesquerella and dimorphotheca oils, respectively, because the hydroxyl acyl groups are located solely in the 1- and 3-glycerol positions within TAG. Lipolysis of lesquerella oil by Lipozyme RM IM occurred at 66.7% yield and produced lesquerolic and auricolic FFA at a combined purity of 80% among the FFA.[13] A greater extent of lipolysis resulted in a greater yield at the cost of lower purity due to the occurrence of acyl migration.[13] FFA enriched in hydroxyl, epoxy, and long-chain (>C20) cannot be isolated by a simple saponification procedure; the alkali extractant must be buffered at pH 10. Alternatively, hydroxy FFA-rich products are isolated by crystallization through use of cold trap placed on-line in a continuous-mode bioreactor.[14] When designing a bioreactor for lipolysis for preparative scale or larger, the use of immobilized lipases is essential to facilitate reusability and recovery of the biocatalyst from the reaction medium. The major issue to decide is the relative amounts of water and TAG, and the means of contacting the two poorly miscible liquids. The most common approach, to employ rapid stirring, resulting in either water-in-oil or oil-in-water

aW

Fig. 2 Bioreactor configurations for lipolysis that employ saturated salt solutions of high water activity to maintain water saturation. (View this art in color at www.dekker.com.)

emulsions, provides a high degree of interfacial area, but promotes poor dispersion of the immobilized biocatalyst and hinders rapid separation of liquid phases and the immobilized lipase. Microemulsions, nanometer-sized dispersions formed by the addition of a surfactant and alkane, provide excellent interfacial area and interfacial mass transport but are not recommended because of the occurrence of product inhibition and the very difficult separation of the water and oil phases due to the surfactant. Two approaches are recommended as alternatives to the employment of emulsions. The first is to operate with a watersaturated oil phase, with saturation maintained by the use of a compartmentally separated saturated salt solution (Fig. 2). The principle which drives water transport from the saturated salt solution to the lipophilic reaction solution is the thermodynamic requirement of equal water activity (i.e., chemical potential) for all phases. Hence, water consumed by the reaction will be replenished by the saturated salt solution. The water content of the lipolysate will increase during the course of reaction for a given water activity due to the formation of FFA, a surface-active agent. A salt which yields a high water activity, such as NaCl, should be employed. Two configurations are presented in Fig. 2, one where a saturated salt solution is circulated through the lipolysate through semipermeable tubing of hydrophobic material. The second configuration employs a common vapor head space between the saturated salt solution and the reservoir. The second alternative to emulsion-based systems is to employ membrane bioreactors, in which the membrane compartmentalizes the water and oil phases. Both flat membrane and hollow microfiber configurations have been employed.[15] Membrane bioreactors allow for continuous-mode operation and the continuous removal of glycerol, the latter of which will

3184

Use of Lipases to Isolate Polyunsaturated and Oxygenated Fatty Acids and Form Value-Added Ester Products

thermodynamically drive hydrolysis over its reverse reaction, esterification. However, the barrier to mass transport presented by the membrane and the low activity of lipases immobilized to the membrane surface lessen enthusiasm for this approach. A nitrogenrich, oxygen-poor reactor head space is recommended for all lipolytic reactions that produce PUFA or hydroxy FFA to reduce oxidative degradation.

ISOLATION OF PUFA VIA LIPASE-CATALYZED SELECTIVE ESTERIFICATION In contrast to lipolysis, the use of selective lipasecatalyzed esterification (reaction 2 of Fig. 1) in conjunction with a purification method is very successful for isolating PUFA.[16] Due to the discrimination most lipases possess against D4–D6 FFA (particularly from Rhizopus sp., G. candidum, Candida rugosa, and Chromobacterium viscosum), acyl groups such as the C18 D9s are rapidly esterified, while the D4–D6 PUFA are esterified much more slowly. As a first step, FFA are derived from a TAG source, for instance, by employing lipolysis. Next, selective esterification is operated in a stirred batch reactor containing the FFA source, an acyl acceptor (in stoichiometric excess), lipase, and perhaps a solvent. The acyl acceptor chain length (n-C12 reported optimal[17]), substrate ratio, temperature, water activity level, biocatalyst type and concentration, and extent of reaction are variables that should be optimized by the user in preliminary small-scale batch reactions.[18] Optimal conditions represent a position on the hyperbolic plot of purity versus yield that is dictated by economic factors. The time course of reaction is monitored so that the reaction can be stopped upon reaching the optimal extent of reaction by removal of the lipase via filtration or sedimentation. Saponification is then employed to isolate the PUFA-rich FFA from the reaction mixture. As an alternative to saponification, FFA, esters, and unreacted alcohol can be fractionated by short-path molecular distillation if long-chain acyl acceptors are used. An additional cycle of lipase-catalyzed selective esterification applied to the recovered PUFA-enriched FFA fraction will improve the purity of PUFA at the expense of reduced yield. For PUFA isolation from FFA mixtures derived from their natural sources (Table 1), typical purities and yields are 73–95% and 60–90%, respectively. To improve the PUFA purity, urea inclusion compounds (UICs) are frequently employed, as discussed above. Lipase-catalyzed alcoholysis (reaction 3 of Fig. 1) or hydrolysis of fatty acid methyl or ethyl ester mixtures is equally effective in discriminating against D4–D6 acyl groups. Assuming that the acyl acceptor substrate and acceptor group in the ester substrate in a selective alcoholysis reaction are significantly different

in chain length, D4–D6 polyunsaturated fatty acyl-rich esters can be isolated by short-path molecular distillation. Lipase-catalyzed esterification to isolate PUFA has been scaled up to the 1–10 kg level.[18] Lipase-selective esterification has recently helped fractionate PUFA species present in the same mixture. Fish oil-derived FFA was esterified with glycerol using Lipozyme IM in solvent-free media to fractionate DHA and EPA.[19] The unesterified FFA contained 78% DHA and only 3% EPA, with a 79% recovery of DHA; the esters contained the majority of the EPA and the other acyl groups, with a 91% recovery of EPA.[19] Lipases have also successfully fractionated the two most abundant species of CLAs: 18:2-9c,11t and 18:2-10t,12c from an equimolar mixture derived from acid-catalyzed isomerization of ALA.[20]

STRUCTURED AND PUFA-ENRICHED TAG Structured TAG are defined as TAG molecules that contain mixtures of short- (C1–C4), medium- (C6–C12), and long- (C14 or higher) chain acyl groups, with a given acyl group types frequently being confined to either the 1-(3-) or 2-position on the glycerol backbone.[21] (TAG rich in medium chain acyl groups are clinical pharmaceuticals for patients with lipid absorption or digestion disorders and high- and rapid-energy sources for athletes since medium chain acyl groups are readily metabolized via the portal vein and generally are not stored in adipose cells for long-term use.) Structured TAG have many applications as nutraceuticals. Commercial examples include SalatrimTM from Danisco Cultor (dietary agents consisting of TAG with at least one long-chain saturate, e.g., 18:0, and one short-chain acyl group), CapreninTM and CaprucinTM from Procter and Gamble, dietary agents containing C8, C10, and behenic (22:0), or erucic (22:1-13) acyl groups, respectively, CaptexTM 810-D from Abitec, Columbus, Ohio, U.S.A. (cosmetic agents containing TAG enriched in C8, C10, and 18:2 acyl groups), and StructolipidTM from Fresenius-Kabi AB, Sweden (parenteral nutrients formed by random interesterification of coconut and soybean oils). All of these products contain TAG with random distribution of the named acyl groups among the three acylglycerol positions. The employment of 1,3-selective lipases leads to structured TAG where specific acyl groups are confined to either the 1-(3-) or 2-acylglycerol position. One of the earliest examples is cocoa butter substitute formed by 1,3-selective lipase-catalyzed acidolysis (reaction 1 of Fig. 3) of palm oil midfraction by palmitic acid (resulting in the replacement of 1-, 3-dipalmityl, 2-oleyl TAG by 1-(3-) palmityl, 2-oleyl, 3- (1-) steryl TAG and 1-, 3-disteryl, 2-oleyl TAG, both of which are abundant in cocoa butter). The product

Use of Lipases to Isolate Polyunsaturated and Oxygenated Fatty Acids and Form Value-Added Ester Products

3185

U

(

)

Fig. 3 Reactions catalyzed by 1,3-selective lipases to synthesize structured TAG. (From Ref.[6].)

BoheninTM, consisting mostly of 1,3-behenyl, 2-oleyl TAG, is synthesized using lipases by Fuji Oil, Osaka, Japan, as a cocoa butter improver. A second product area is human milk fat substitute for use in formula for both prematurely born and full-term infants. Mammalian milk frequently contains TAG possessing a palmitic acyl group at the 2-position. The commercial product BetapolTM from Loders Croklaan, Wormerveer, Netherlands, a TAG product enriched in 1,3-dioleyl, 2-palmityl TAG, recently received the rating ‘‘generally regarded as safe (GRAS)’’ by the US Food and Drug Administration (FDA). BetapolTM, the first commercial product employing lipases to create structured TAG, improves the absorption of dietary fat and calcium by infants, and reduces constipation. Infant formula that contains PUFA-acyl groups at the 1- and 3-acylglycerol position is suggested to enhance brain and nervous system development.[22] Positioning of PUFA groups in the outer acylglycerol positions allows for proper digestion by infants, since only the 2-position acyl group is strongly absorbed in vivo.[22] This fact suggests that structured

TAG containing PUFA at the 2-position and mediumchain acyl groups at the 1- and 3-positions may be a useful nutraceutical, providing the benefits described above for SalatrimTM, CapreninTM, and CaprucinTM, with the added benefit of an essential FFA being adsorbed by the body. Such a structured TAG has been suggested as a nutrition source for patients with pancreatic deficiencies. However, a problem recently discovered for structured TAG prepared by 1,3-selective lipases is the product’s relatively low oxidative stability.[23] Synthesis of a structured TAG with PUFA confined to either the 1-(3-) or 2-position requires 1,3-selective lipases to interesterify a TAG using FFA or fatty acid esters (reaction 1), or to transesterify two different populations of TAG molecules (Fig. 3). Alternatively, a two-step approach can be employed, where a TAG is first subjected to 1,3-selective lipolysis, resulting in 2-MAG that is carefully isolated, then crystallized, to prevent acyl migration.[24] The second step is 1,3selective esterification of the 2-MAG.[24] Such an approach results in a 72–85% yield and >95%

3186

Use of Lipases to Isolate Polyunsaturated and Oxygenated Fatty Acids and Form Value-Added Ester Products

purity.[24] In addition, the acidolysis of a medium acyl chain-rich TAG by PUFA-rich FFA was enhanced by use of a membrane bioreactor in which the released medium-chain FFA were selectively removed by their permeation through the membrane.[25] To enrich a common TAG mixture in PUFA, either 1,3- selective or ‘‘random’’ lipases are employed to catalyze interesterification using PUFA-enriched FFA (acidolysis, reaction 4 of Fig. 1) or fatty acid methyl or ethyl ester (transesterification, reaction 5 of Fig. 1). PUFAenriched MAG=DAG=TAG, produced via selective lipolysis (discussed above), can replace the TAG substrate, resulting in a TAG highly enriched in PUFA. An alternative approach to produce PUFA-enriched TAG is direct esterification between glycerol and PUFA-rich acyl groups. When conducting interesterification for structured and PUFA-enriched TAG synthesis, the three main goals are high activity, low occurrence of hydrolysis (resulting in DAG and perhaps MAG), and, for structured lipid synthesis, low occurrence of acyl migration, because this reaction will lead to formation of TAG with undesired structure. One must optimize reaction parameters: residence time, reaction temperature, and fatty acyl donor-to-TAG substrate ratio, to arrive at a desired position between the two extremes of high percent interesterification and low degrees of acyl migration and hydrolysis.[26] An optimal water activity also exists: suboptimal water content reduces biocatalytic activity, while high water content leads to increased hydrolysis and acyl migration.[26] A continuous-mode packed bed bioreactor is recommended because it lowers residence time of reaction mixture in the presence of immobilized lipase, which leads to reduced acyl migration.[26]

PUFA AND HYDROXY ACYL-RICH MAG, DAG, AND POLYOL ESTERS MAG are well-known biodegradable and biocompatible emulsifiers in the foods, dairy, cosmetics (e.g., in toothpastes), and pharmaceutical industries.[27–29] 1-(3-) MAG are easily modified chemically to produce MAG sulfates, cosmetic surfactants which yield low irritability. As discussed above, 2-MAG are feedstock for the lipase-catalyzed synthesis of structured lipids. Approval of MAG as GRAS was recently granted by the FDA. DAG are cocoa butter additives that reduce the extent of crystallization or ‘‘blooming.’’ Isomerically pure DAG are possible feedstocks for the synthesis of glycerylphospholipids, glycolipids, and pharmaceuticals.[29] Current industrial methods to produce MAG and DAG (e.g., glycerolysis directed by a heterogeneous catalyst) involve high temperatures (180–220 C or 30–160 C if a high vacuum pressure

of 200–400 Pa is applied) that may produce byproducts that promote off-flavors, -odors, or -colors, and a broad product distribution containing various MAG, DAG, and TAG species.[27,29] Molecular distillation must be applied to remove the impurities, which increases operating costs and can result in further chemical degradation. Therefore, enzymatic preparation of MAG and DAG may gain further interest if energy costs continue to increase and will be of particular interest for products that contain degradationsusceptible PUFA and oxygenated acyl groups. PUFA-enriched MAG (and polyol monoesters, discussed below) are food emulsifiers that provide essential FFA nutrients. Hydroxy acyl-rich MAG are waxy materials employed in lipsticks. Lipase-catalyzed synthesis has been employed to synthesize FFA partial and poly-esters of glycols such as ethylene and propylene glycol, neopentanol, trimethylolpropane, and polyglycerol as a low-cost and low-temperature alternative to chemical processing for applications as biodegradable and biocompatible lubricants and emulsifiers in the food and cosmetics industry. Saccharide acyl acceptors (e.g., fructose, glycols, sucrose, xylose, maltose, and trehalose) have received the most attention of the polyols due to the narrow product distribution provided by regioselective lipases.[30] For instance, 1,3-selective lipase-catalyzed esterification of 1,2-propanediol results in only the 1-monoester. The substrate ratio, water activity, and reaction medium polarity can also be selected to control the relative proportion of mono- and di-esters. As a second example, lipases esterify only the primary OH groups of sucrose, at the 6, 60 and 10 positions, with the three hydroxyls listed in the order of preference by lipases. Chemical synthesis of sucrose esters requires temperatures above 100 C and results in a broad product distribution. Lipases catalyze MAG (or polyol ester) synthesis via esterification or glycerolysis (alcoholysis) in nonaqueous media (Fig. 4). The challenge to overcome when conducting lipase-catalyzed reactions is the poor miscibility of glycerol (or polyol) and the acyl donorrich lipophilic phase. The employment of polar solvents (e.g., tert-butanol, acetone, and, recently, room temperature ionic liquids) increases miscibility and the product distribution in favor of mono- rather than di- or poly-esters, but decreases the catalytic rates because of the partitioning away of essential water molecules from lipase to the solvent. Other approaches include the use of polyhydric alcohol complexation agents such as phenylboronic acid, the use of protective groups such as isopropylidene (which for glycerol covalently blocks access to two of its three hydroxyls, resulting in regioisomerically pure 1-(3-) MAG), water-in-oil microemulsions, and the suspension of silica gel saturated with glycerol in the reaction medium,

Use of Lipases to Isolate Polyunsaturated and Oxygenated Fatty Acids and Form Value-Added Ester Products

1.

3187

O

O HO

OH

O ESTERIFICATION

+ HO (or

O OR)

HO

+

H2O (or ROH)

(or TRANSESTERIFICATION)

HO

HO 1-(3-) MAG

2.

O O

O 2

O

O Acyl Migration

GLYCEROLYSIS

O

+ 2 O O

HO

HO

+

HO O

HO

HO

O

HO 1-(3-) MAG

2-MAG HO

Fig. 4 Reactions catalyzed by 1,3-selective lipases to synthesize monoacylglycerol (MAG). (From Ref.[6].)

all of which are impractical for large-scale manufacturing. Of note, glycosides (saccharide alkyl ethers) can be readily formed by chemical and enzymatic means (glucosidases) to solve the problem of poor substrate miscibility for saccharides. Penicillium camembertii lipase-catalyzed esterification will produce MAG and DAG, but not TAG due to its unique substrate selectivity.

POLY (HYDROXY FATTY ACID) ESTERS Random lipases are known to catalyze the formation of poly(hydroxy) fatty acids, primarily ricinoleic acid and its product via hydrogenation, 12-hydroxystearic acid; however, polymerization also occurs readily by chemical processing.[31] The value of lipase-catalyzed processing is the ability of 1,3-selective lipases to permit esterification of the free carboxyl moiety without cleavage of the ester bonds that join together the hydroxyacyl monomeric units. Esterification improved the physical properties of poly(hydroxy acids) as lubricant materials, evidenced by their reduced viscosity change with temperature, i.e., increased viscosity index.[31] Monoesters of polyglycerol and poly(ricinoleic acid) are well known emulsifiers for the food industry. Star polymers have been synthesized recently from

polyol and poly(ricinoleic acid), e.g., pentaerythritolpoly(ricinoleic acid) tetraester, using lipases in nonaqueous media. Such materials have very good lubricant properties and may have utility as drug delivery vehicles.[32]

STERYL ESTERS Lipases have also been employed to isolate tocopherols and sterols, important antioxidants from ‘‘deodorizer distillate,’’ a by-product formed during the deodorization step of seed oil purification, by hydrolyzing MAG and DAG in the ‘‘distillate,’’ and by esterifying sterols into sterol esters.[33] The latter serves as antioxidant in non-polar food products such as margarine (e.g., BenacolTM from the Raisio Group, Finland and ‘‘Take Control’’TM from Unilever, Englewood Cliffs, New Jersey, U.S.A.). PUFA-sterol esters would provide the benefits of essential fatty acid intake and antioxidant protection.

PEROXIDATED FFA Peroxy fatty acids, R-COOOH, formed from the esterification of FFA or alcoholysis of fatty acid methyl or

U

3188

Use of Lipases to Isolate Polyunsaturated and Oxygenated Fatty Acids and Form Value-Added Ester Products

ethyl esters with hydrogen peroxide, H2O2, are oxidants employed in the chemical industry for epoxidation of double bonds, conducting Baeyer-Villiger oxidation reactions, and hydroxylation of aromatic rings and amines. Traditionally, this reaction occurs via catalysis using a strong acid; however, for longchain and unsaturated FFA, harsh operating conditions (strong acids and high temperature) are required and by-products frequently occur. Thus, biocatalysis may be a safer alternative.[34] Lipase-catalyzed peroxidation of PUFA yields epoxy FFA, of potential use for paints, coatings, and disinfection agents (sporicides).[34] Epoxy FFA occur due to the isomerization, or ‘‘selfepoxidation’’ of peroxy FFA. For instance, peroxidation of ALA results in peroxy-ALA, which isomerizes, or ‘‘self-epoxidizes,’’ into a mixture of 9,10-epoxy, 12,13-epoxy, and 15,16-epoxy FFA. Lipase-catalyzed peroxidation typically results in a 70–90% yield of epoxy-FFA.[34] Hydroxy and peroxy acids are also formed from the chemical rearrangement of hydroperoxides, a class of lipid products formed via lipoxygenasecatalyzed oxidation of PUFA,[35] a topic that is beyond the scope of this entry. Although enzymes are very susceptible to denaturation by H2O2, NovozymeTM appears to be reasonably stable. To reduce denaturation, fed-batch addition of 30–60% aqueous H2O2 to the pure FFA or alkyl ester is employed. Typical solvents used for this reaction are toluene and dichloromethane.

CONCLUSIONS The employment of lipases to isolate polyunsaturated and oxygenated FFA and transform them into ester products will expand in the near future, especially for purifying PUFA and synthesizing structured TAG. Biocatalytic modification will become more competitive with chemical processing if energy costs continue to rise. Advances in lipase bioreactor design and process scale-up technology will be an area of focus in future research, including the implementation of reactive separations. Along the same lines, multistep processing schemes composed of biocatalytic and separation steps to convert raw lipid feedstock into products of interest such as structured TAG, have improved in their sophistication during the past five years, with the trend expected to continue, particularly with regard to the integration of reaction and separation steps. Bioreactor design will also be developed for multiple enzyme systems, for instance, the combination of random lipases and lipoxygenases to transform TAG into hydroperoxides. Another example may be a multistage bioreactor where random lipases hydrolyze PUFA-enriched TAG in an upstream stage, and 1,3-selective lipases catalyze the acidolysis of a

tripalmitin feed stream with the PUFA-enriched FFA transported from the upstream portion of the process in a downstream stage, resulting in an infant formula nutraceutical. Although acyl migration and hydrolysis side reactions cannot be prevented in manufacturing TAG, future bioreactor designs will facilitate the programming of operational parameters such as water activity, solvent system polarity, and temperature to improve yield and recovery. The ability to program the water activity and other parameters effectively will require improved capabilities in kinetic modeling and means of in situ control and rapid measurement of substrate and product concentration. The in situ measurement of water activity is being developed.[36] It is highly desired that a lipase be produced that will be less discriminatory against D4–D6 and long-chain saturated FFA to improve the rate and extent of PUFA acyl incorporation into ester products. Advancements in directed evolution, extremophile discovery and screening, and recombinant DNA technology will lead to more active and stable lipases that possess narrower substrate selectivity. In conclusion, although a mature field, lipasecatalyzed lipid modifications will require further improvement along the lines given in the preceding paragraphs for the further use of PUFA and hydroxy acyl material as chemical and biological intermediates. Although alternate technologies (lipid purification, chemical modification, transgenic plants,[37] and microbial bioconversions[38]) will improve, the selectivity of lipase-based processing will continue to make it a viable choice as unit operation in a lipid ‘‘biorefinery’’ as a substitute for petrochemical-based processes.

REFERENCES 1. Bornscheuer, U.T. Ed. Enzymes in Lipid Modification; Wiley-VCH: Weinheim, Germany, 2000. 2. Kuo, T.M.; Gardner, H.W. Lipid Biotechnology; Marcel-Dekker: New York, 2002. 3. Li, D.; Bode, O.; Drummond, H.; Sinclair, A.J. Omega-3 (n-3) fatty acids. In Lipids for Functional Foods and Nutraceuticals; Gunstone, F.D., Ed.; Oily Press Lipid Library: Bridgwater, UK, 2003; 225–262. 4. Pariza, M.W.; Park, Y.; Cook, M.E. The biologically active isomers of conjugated linoleic acid. Prog. Lipid Res. 2001, 40 (4), 283–298. 5. Schwitzer, M.K. Castor oil. Proceedings World Conference on Oleochemicals: Into the 21st Century; Applewhite, T.H., Ed.; AOCS Press: Champaign, IL, 1991; 111–118. 6. Hayes, D.G. Enzyme-catalyzed modification of oilseed materials to produce eco-friendly

Use of Lipases to Isolate Polyunsaturated and Oxygenated Fatty Acids and Form Value-Added Ester Products

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

19.

products. J. Am. Oil Chem. Soc. 2004, 81 (12), 1077–1103. Jaeger, K.E.; Dijkstra, B.W.; Reetz, M.T. Bacterial biocatalysts: molecular biology, three-dimensional structures, and biotechnological applications of lipases. Ann. Rev. Microbiol. 1999, 53 (1), 315–351. Halling, P. Enzymic conversions in organic and other low-water media. In Enzyme Catalysis in Organic Synthesis; 2nd Ed.; Drauz, K., Waldmann, H., Eds.; Wiley-VCH: Weinheim, Germany, 2002; 259–285. Villeneuve, P.; Foglia, T.A. Lipase specificities: potential application in lipid bioconversions. Inform 1997, 8 (6), 640–651. Fureby, A.M.; Virto, C.; Adlercreutz, P.; Mattiasson, B. Acyl group migrations in 2-monoolein. Biocatal. Biotransform. 1996, 14 (2), 89–111. Hayes, D.G. Purification of free fatty acids via urea inclusion compounds. In Handbook of Functional Lipids; Akoh, C.C., Ed.; CRC Press: Boca Raton, FL, 2005; 77–88. Shimada, Y.; Sugihara, A.; Tominaga, Y. Enzymatic enrichment of polyunsaturated fatty acids. In Lipid Biotechnology; Kuo, T.M., Gardner, H.W., Eds.; Marcel Dekker: New York, 2002; 493–515. Hayes, D.G.; Kleiman, R. 1,3-specific lipolysis of Lesquerella fendleri oil by immobilized and reverse micellar encapsulated lipases. J. Am. Oil Chem. Soc. 1993, 70 (11), 1121–1127. Derksen, J.T.P.; Krosse, A.M.; Tassignon, P.; Cuperus, F.P. Lipase-catalyzed production of functionalized fatty acids from Dimorphotheca pluvialis seed oil. Mededelingen van de Faculteit Landbouwwetenschappen, Universiteit Gent 1992, 57 (4a), 1741–1747. Malcata, F.X.; Reyes, H.R.; Garcia, H.S.; Hill, C.G., Jr.; Amundson, C.H. Immobilized lipase reactors for modification of fats and oils – a review. J. Am. Oil Chem. Soc. 1990, 67 (12), 890–910. Shimada, Y.; Sugihara, A.; Tominaga, Y. Enzymatic purification of polyunsaturated fatty acids. J. Biosci. Bioeng. 2001, 91 (6), 529–538. Shimada, Y.; Sugihara, A.; Nakano, H.; Kuramoto, T.; Nagao, T.; Gemba, M.; Tominaga, Y. Purification of docosahexaenoic acid by selective esterification of fatty acids from tuna oil with Rhizopus delemar lipase. J. Am. Oil Chem. Soc. 1997, 74 (2), 97–101. Shimada, Y. Application of lipase reactions to separation and purification of useful materials. Inform 2001, 12 (12), 1168–1174. Halldorsson, A.; Kristinsson, B.; Glynn, C.; Haraldsson, G.G. Separation of EPA and DHA

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

3189

in fish oil by lipase-catalyzed esterification with glycerol. J. Am. Oil Chem. Soc. 2003, 80 (9), 915–921. Nagao, T.; Yamauchi-Sato, Y.; Sugihara, A.; Iwata, T.; Nagao, K.; Yanagita, T.; Adachi, S.; Shimada, Y. Purification of conjugated linoleic acid isomers through a process including lipasecatalyzed selective esterification. Biosci. Biotechnol. Biochem. 2003, 67 (6), 1429–1433. Osborn, H.T.; Akoh, C.C. Structured lipids–novel fats with medical, nutraceutical, and food applications. Compreh. Rev. Food Sci. Food Safety 2002, 1 (3), 93–103. Yamane, T. Lipase-catalyzed synthesis of structured triacylglycerols containing polyunsaturated fatty acids: monitoring the reaction and increasing the yield. In Enzymes in Lipid Modification; Bornscheuer, U.T., Ed.; Wiley-VCH: Weinheim, Germany, 2000; 148–169. Timm-Heinrich, M.; Skall Nielsen, N.; Xu, X.; Jacobsen, C. Oxidative stability of structured lipids containing C18:0, C18:1, C18:2, C18:3 or CLA in sn2-position – as bulk lipids and in milk drinks. Innovat. Food Sci Emerg. Technol. 2004, 5 (2), 249–261. Schmid, U.; Bornscheuer, U.T.; Soumanou, M.M.; McNeill, G.P.; Schmid, R.D. Highly selective synthesis of 1,3-oleoyl-2-palmitoylglycerol by lipase catalysis. Biotechnol. Bioeng. 1999, 64 (6), 678–684. Xu, X.; Skands, A.; Jonsson, G.; Adler-Nissen, J. Production of structured lipids by lipasecatalysed interesterification in an ultrafiltration membrane reactor. Biotechnol. Lett. 2000, 22 (21), 1667–1671. Xu, X. Enzymatic production of structured lipids: process reactions and acyl migration. Inform 2000, 11 (10), 1121–1131. Bornscheuer, U.T. Lipase-catalyzed synthesis of monoacylglycerols. Enzyme Microb. Technol. 1995, 17 (7), 578–586. Boyle, E. Monoglycerides in food systems: current and future uses. Food Technol. 1997, 51 (8), 52–5456, 58–59. Diks, R.M.M.; Bosley, J.A. The exploitation of lipase selectivities for the production of acylglycerols. In Enzymes in Lipid Modification; Bornscheuer, UT, Ed.; Wiley-VCH: Weinheim, Germany, 2000; 3–22. Sarney, D.B.; Vulfson, E.N. Enzymatic synthesis of sugar fatty acid esters in solvent-free media. In Enzymes in Nonaqueous Solvents: Methods and Protocols (Methods in Biotechnology Vol. 15); Vulfson, E.N., Halling, P.J., Holland, H.L., Eds.; Humana Press: Totawa, 2001; 531–543.

U

3190

Use of Lipases to Isolate Polyunsaturated and Oxygenated Fatty Acids and Form Value-Added Ester Products

31. Hayes, D.G. The catalytic activity of lipases toward hydroxy acids (a review). J. Am. Oil Chem. Soc. 1996, 73 (5), 543–549. 32. Hayes, D.G. Lipase-catalyzed synthesis of polyhydric alcohol-poly(ricinoleic acid) ester star polymers. Polym. Prepr. (Am. Chem. Soc., Div. Polym. Chem.) 2005, 46 (1), 276–277. 33. Weber, N.; Weitkamp, P.; Mukherjee, K.D. Cholesterol-lowering food additives: lipasecatalysed preparation of phytosterol and phytostanol esters. Food Res. Intl. 2002, 35 (2=3), 177–181. 34. Ru¨sch gen. Klaas, M.; Warwel, S. Lipasecatalyzed peroxy fatty acid generation and lipid oxidation. In Enzymes in Lipid Modification; Bornscheuer, U.T., Ed.; Wiley-VCH: Weinheim, Germany, 2000; 116–127.

35. Iacazio, G.; Martini-Iacazio, D. Properties and applications of lipoxygenases. In Enzymes in Lipid Modification; Bornscheuer, U.T., Ed.; Wiley-VCH: Weinheim, Germany, 2000; 337–359. 36. Kang, I.J.; Rezac, M.E.; Pfromm, P.H. Membrane permeation based sensing for dissolved water in organic micro-aqueous media. J. Membr. Sci. 2004, 239 (2), 213–217. 37. Sayanova, O.V.; Napier, J.A. Eicosapentaenoic acid: biosynthetic routes and the potential for synthesis in transgenic plants. Phytochemistry 2004, 65 (2), 147–158. 38. Kuo, T.M.; Kaneshiro, T.; Hou, C.T. Microbiological conversions of fatty acids to value-added products. In Lipid Biotechnology; Kuo, T.M., Gardner, H.W., Eds.; Marcel Dekker: New York, 2002; 605–628.

Vapor–Liquid–Solid Synthesis of Nanowires V Brian A. Korgel Tobias Hanrath Forrest M. Davidson, III The University of Texas, Austin, Texas, U.S.A.

INTRODUCTION Fundamental aspects of vapor–liquid–solid (VLS) semiconductor nanowire growth are presented here. The synthesis of VLS semiconductor has been extended to different reaction media and pathways from the early chemical vapor deposition (CVD) approach, including solution–liquid–solid (SLS) and supercritical fluid–liquid–solid (SFLS), laser-catalyzed growth, and vapor–liquid–solid-epitaxy. The properties of nanowires grown by these VLS embodiments are compared. In this entry, VLS growth of nanowire heterostructures and oriented and hyperbranched arrays is examined. In addition, surface passivation and functionalization are assessed, and the importance of these techniques in the progress toward VLS produced nanowire devices is detailed.

BACKGROUND Semiconductor nanowires are ultrahigh aspect ratio (>1000) crystals that are micrometers in length and Pd  Rh > Ir. In recent years, the interest toward PM catalysts has grown from essentially academic to much more practical, as a result of efforts to develop on-board fuel processors for cars. Operational conditions envisaged for such fuel processors preclude the existing Cu–Zn–Al catalysts from being applicable, while PM catalysts are believed to be a viable candidate. Platinum-based catalysts with a great variety of promoters, modifiers, and supports, as well as their preparation conditions seem to be most widely explored in research papers and patent applications. Pt supported on ceria containing support materials have been widely explored for the WGS reaction.[35–38] In recent years, significant attention has also been given to gold-containing catalysts.[39,40] However, PM catalysts have not yet become an industrial reality, and different opinions regarding their viability have been expressed.[41,42] At present, these catalysts are still too expensive and their activity needs to be improved to assure good performance at low temperatures (200 C). However, PM catalysts offer important advantages such as insensitivity to frequent start– stops, thermal stability, and operational safety. Besides, significant cost savings and environmental benefits can result from PM recycling.

Reaction Kinetics for Precious Metal Catalysts Reaction orders and activation energies have been determined for Pt=ceria catalysts by several authors.[37,38,43] There is an agreement that the reaction order with respect to CO is approximately 0 at 200 C. Therefore high CO concentrations do not speed up the reaction for Pt-based catalysts at low temperature, as opposed to Cu-based catalysts which have approximately first-order kinetics. Activation energy estimates range from approximately 46 kJ=mol[37] to 80 kJ=mol.[34]

Water Gas Shift Reaction

PRACTICAL ASPECTS In industrial hydrogen and synthesis gas production, the WGS reaction is used to: 1. Adjust the CO=H2 ratio for the specific purpose of the synthesis gas (methanol production, Fischer-Tropsch synthesis) 2. Increase the hydrogen yield of the process (i.e., ammonia synthesis; refinery hydrogen production) 3. Reduce the CO concentration to a level amenable to clean-up by preferential oxidation catalysts (fuel cell applications). Water gas shift converters in industry are largely unchanged from their original design: adiabatic fixed bed reactors with particulate catalysts. There are only two kinds of these reactors in use in industry today: HTS and LTS. They differ in the operating temperatures and catalysts which they use. Two reactors are needed to convert the majority of CO, because of the equilibrium limitations of the process (see previous section). The inlet temperature of the HTS reactor is typically around 320 C. The outlet temperature rises to about 400–450 C because of the reaction exotherm. The gases are cooled to about 200 C before entering the LTS reactor, where the final 2–3% CO is partially converted to CO2 and H2.

HTS Reactors The HTS reactor converts the majority of CO (from about 12% for SR plants to 45% for partial oxidation and coal gasification plants) to CO2 and H2 and experiences the majority of the 100–400 C exotherm associated with CO conversion. This conversion can be done in one step (for SR gas) or two to three steps with addition of quench water or inter-cooling (for high CO content gas, i.e., partial oxidation or coal gasification) because of the associated exotherm of CO conversion.[8] Inlet HTS temperatures are typically kept in the range of approximately 320–360 C, which is the compromise determined by such criteria as the catalyst activity, equilibrium outlet CO concentration, catalyst lifetime, reactor materials cost, etc. In particular, higher operating temperatures lead to a more rapid decrease in the catalyst activity due to sintering of magnetite (Fe3O4) crystallites, which accelerates with temperature. Typically, the lifetime of the iron-chrome HTS catalyst amounts to 3–5 years and depends on how the catalyst has been handled in a given plant. Aging of the catalyst requires the corresponding gradual increase of the reactor inlet temperature to compensate for the activity decrease. This increase has to be carefully controlled, since unnecessarily high

3211

temperature spikes can shorten the catalyst life. Despite all precautions, however, the exit CO concentration eventually gets to a too high level, which necessitates a replacement of the aged catalyst by a fresh charge. The exit concentration depends on the inlet gas composition and temperature and usually lies between 1% and 4% (dry gas). The space velocity is kept between 500 hr
1 and 5000 hr
1. Poisoning of HTS Catalyst The HTS catalysts in modern plants are fairly resistant against poisoning. The most common poison almost inevitably present in the feedstocks is sulfur. The iron-chrome catalysts are less sensitive to sulfur compared to the other catalysts used in hydrogen generation processes, such as steam reforming and LTS catalysts, therefore the levels of sulfur-containing compounds in the feed gas that are tolerant for those catalysts do not affect the activity of the HTS catalyst. Moreover, the HTS catalysts, even if sulfided with Fe3O4 being converted into FeS (this may occur in the coal-based processes where high concentrations of sulfur-containing compounds are common), still retain the WGS activity at about one half that of Fe3O4.[7] For this reason the iron-chrome HTS catalyst is a good choice in processes where high concentrations of sulfur-containing compounds in the feed gas can be anticipated. It also serves as a sulfur guard bed for the more expensive Cu-Zn catalyst. Other known chemical poisons for the HTS catalyst are halides, although under normal operating conditions they are not present in the feed at an appreciable concentration. Decay in the catalytic activity was also observed with the feed gas which contained minor amounts of unsaturated hydrocarbons, oxygen, and nitric oxides.[7] Under the HTS conditions these compounds were converted into a heavy carbonaceous residue deposited on to the surface of the catalyst, blocking access of the reactants to the catalytic surface. Commercial HTS catalysts are mechanically quite strong. However, in an industrial process environment, the HTS catalyst can suffer from mechanical factors that deteriorate its performance, such as steam condensation leading to a gradual disintegration of the catalyst pellets, and deposition of foreign components (e.g. particulate matter from corrosion of the process equipment). Increase in the pressure drop across the catalyst bed resulting from these factors is yet another factor determining the catalyst lifetime. Discharge of HTS Catalyst The HTS catalyst deactivated beyond an acceptable level has to be discharged from the reactor and

W

3212

Water Gas Shift Reaction

replaced by a fresh charge. Due to a highly exothermic reoxidation of the magnetite phase 2Fe3 O4 þ

1 2

The catalyst bed temperature increases in the direction of gas flow due to the WGS reaction exotherm. Typical temperature gradients in the bed are about 20–30 C. The lifetime and state of activity of the catalyst is conveniently monitored by the temperature profile through the adiabatic bed. As the reaction front moves through the bed when the catalyst ages, so does the temperature rise from the reaction (Fig. 3).

O2 ! 3Fe2 O3

DH ¼
232:2 kJ=mol

which results in an adiabatic temperature rise of approximately 450 C, the reduced catalyst should not be exposed to air unless special measures are taken to avoid dangerous overheating. Thus, on the reactor cool down, the catalyst is usually kept under a reducing or inert atmosphere, then discharged under nitrogen and slowly oxidized by gradually admitting air into the catalyst storage vessel. Another procedure for safe catalyst discharge includes filling up the cooled reactor with water and removing the wet catalyst. Sulfided catalyst must be handled with exceptional precautions, since oxidation of iron sulfide is extremely exothermic:

Poisoning of LTS Catalysts The lifetime of the LTS catalysts in industrial processes is about 3–4 years. During this time, the catalyst slowly loses its activity due to inevitable sintering and poisoning of copper. To be able to compensate for deactivation and to operate the LTS stage for at least 3 years with required low outlet CO concentrations, the actual charge of LTS catalyst is always overestimated by a factor of about three or more, relative to the amount that would be needed if the activity did not decrease. Sulfur- and halide-containing compounds, the known poisons for copper-based LTS catalysts, gradually decrease the activity even if their concentrations in the feed are very low. The sensitivity to these substances for many years precluded copper-based catalysts from being used in industrial WGS processes. It is only with the massive transfer of hydrogen industry from coal gasification to steam reforming of natural gas and other hydrocarbons, which produces much cleaner gas, that copper-based LTS catalysts became a viable option. The effects of poisons on Cu-based LTS catalysts is described in detail in Refs.[7,8]. Another problem is the formation of methanol over the LTS catalyst. The methanol accumulates in the process condensate and in the gas entering the CO2 removal system.[44] Therefore, emissions from the plant can become an environmental concern. Catalysts doped with Cs have been developed to address this problem.[44,45] These catalysts reduce the production of methanol by nearly 90%.

6 FeS þ 13 12 O2 ! 2 Fe2 ðSO4 Þ3 þ Fe2 O3 DH ¼
5379 kJ=mol Therefore the sulfided catalyst initially needs to be steamed to convert the sulfide into magnetite.

Operation of LTS Catalyst Temperature dependence of the equilibrium CO concentrations and sintering of copper crystallites (the active phase) at elevated temperatures dictate that the LTS stage be operated at the lowest possible temperature. It is common practice to keep the inlet temperature at least at about 20 C above the dew point of the feed gas, which can be as high as 180–200 C. If water condenses on the catalyst, it can lead to breakage of particles and deactivation of the catalyst.

240

reaction zone (poisoning zone)

Temperature [˚C]

230 220

unused catalyst

210 200 inactive (poisoned) zone

190 180 0

20

40 60 Rel. reactor length [%]

80

100

Fig. 3 Typical temperature profile through an LTS bed.

Water Gas Shift Reaction

3213

Table 2 Catalyst requirement for different H2 generation applications Catalyst properties

Industrial H2 generation

Small scale H2 generation

Cost

Very important

Important

Size (volume)

Important

Very important

Sensitivity to water condensation

Accommodated by process

Insensitivity required

Sulfur tolerance

Desired

Desired

Attrition resistance

Important

Important

Reduction requirements

Accommodated by process

Use of process gas

Air sensitivity

Accommodated by process

Insensitivity required

Discharge of LTS Catalyst After several years of operation at the LTS stage, the CO level at the reactor exit begins to increase due to catalyst deactivation. This signals the need for the catalyst to be replaced and the plant shut down. Discharge of a spent catalyst from the reactor requires special precautions due to the strongly exothermic character of its reoxidation by air, which may generate an exotherm as high as 800–900 C, unless special measures are taken to moderate the temperature rise. Therefore, the common LTS catalyst discharge procedure includes purging the reactor with nitrogen, while cooling it down to below 50 C, followed by the discharge under nitrogen flow with immediate spraying of the catalyst with water to prevent rapid reoxidation. An alternate procedure for a safe catalyst discharge is analogous to the one employed for HTS catalysts and includes filling up the cooled reactor with water and removing the wet catalyst.

SIZE VERSUS COST—NEW CATALYSTS FOR ON-SITE H2 GENERATION AND FUEL CELL REFORMERS The industrial means of hydrogen production for methanol synthesis, ammonia synthesis, Fischer-Tropsch synthesis, and refinery hydrogen production have one thing in common: they are tightly controlled processes, run by trained plant operators in an industrial setting.[46] For the emerging applications in hydrogen production, such as fuel cells and small-scale hydrogen production, the operating conditions are different. Often, fuel processors and small hydrogen generators will be operated by inexperienced personnel.[5] Moreover, the typical ‘‘duty cycle’’ of operation for the catalyst is very different. For most industrial processes, this duty cycle consists of continuous steady-state operation with infrequent plant shutdowns. These planned shutdowns are tightly controlled by trained plant personnel and the catalyst is cautiously treated so that it retains its

Table 3 Limitation of WGS catalysts compared to new PM monoliths HTS catalyst (FeCr)

LTS catalyst (CuZn)

PM catalyst

Form

3–10 mm pellets

3–6 mm pellets

Monolith supported

Activity

Not active below 350 C

High activity starting below 200 C

Highly active above 250–300 C

Loss of activity above 260–280 C

Thermally stable to >400 C

Thermal stability

Thermally stable to 500 C 99.5% NaCl rejection) and moderate water flux. Brackish water membranes, on the other hand, are low-pressure membranes. The operating pressure of BW membranes is usually between 1.4 and 4.0 MPa. The inorganic salt rejection of BW membrane is approximately 95–99.5%, and the water flux is higher than that of SW membrane. When RO membranes have fixed charge groups (such as carboxylate anions), improved salt rejection can be obtained by means of Donnan exclusion. In principle, RO membrane is suitable for treatment of organic as well as aqueous solutions. However, problems with membrane stability under actual operation conditions make organic solvent applications very limited. Among others, spiral-wound and hollow-fiber modules are two commonly available RO membrane products on the market. Currently, spiral-wound modules have approximately 80% of the total RO membrane market. Thin film composite membrane

Fig. 2 Schematic diagram of a thin film composite membrane.

Water Reclamation

is usually sold in spiral-wound modules. Two flat membrane sheets are glued (sandwiched) together with a product spacer (permeate spacer) in between to form an envelope. The envelope is sealed on three sides with the fourth open side attached to a product tube. Many of these envelopes are then rolled around a product tube, with feed spacers between the facing membrane surfaces. Under pressure, the feed stream enters the module from one end and exits from the other end. The product water is collected from the center tube. Fig. 3 provides a cutaway view of a spiral-wound module. Reverse osmosis membrane is widely used in seawater and brackish water desalination processes. Compared to traditional distillation, there is no energy-intensive phase change involved in membrane processes. Therefore, desalination with RO membrane is more energy efficient. In addition to the traditional desalination processes, RO membranes have also found wide application in industrial and municipal wastewater treatment, in pure water production for the electronic and pharmaceutical industries, and in the food industries. Current RO membrane research is focused mainly on improving membrane performance and membrane resistance to fouling. Fouling is the deposition of components from the feed stream onto the membrane surface, which reduces membrane flux. Membrane fouling is a major obstacle for the efficient use of membrane technology. Membrane fouling can dramatically reduce process efficiency and, ultimately, also shorten membrane life. Methods to improve fouling resistance of membranes include chemical modifications of membranes and improvements in module design. The membrane surface can be modified to suit specific applications. For instance, making the membrane surface more hydrophilic will improve membrane biofouling characteristics; a smooth surface will reduce the tendency of particle deposition onto the membrane surface. Introducing surface charge is found to be helpful in certain types of applications. Because most colloidal particles, proteins, and cells are negatively charged in aqueous solution, one can expect that membrane with negative charges will be more fouling resistant.[5] Another area of continuing interest and need is to provide RO membranes having high flux at low pressure, while maintaining high solute rejection characteristics.[6,7] The major advantages of using such membranes are low energy consumption and low equipment cost. Compared to the first generation of asymmetric cellulose acetate RO membranes operated at 3 to 4 MPa, current low-pressure RO membranes can be run at approximately 1 MPa. Research efforts to further enhance the permeation rate of RO membranes and reduce energy requirements are ongoing.[8–10]

Water Reclamation

3219

W

Fig. 3 Cutaway view of a spiral-wound membrane module.

NANOFILTRATION MEMBRANES Nanofiltration (NF) membrane is a category of membrane between reverse osmosis and ultrafiltration (UF). An NF membrane can reject organic molecules with molecular weights (mol wt) in the range of 300 to 1000, a lower range that can be separated by UF. However, NF membranes typically have much higher flux than RO membranes at low pressures (0.5 MPa or lower), and they have lower rejection than RO membranes for neutral molecules below 150 mol wt and for monovalent inorganic salts.[11] For example, the NaCl rejection of most NF membranes is in the range of only 30–90%. Most NF membranes have a large concentration of negatively charged groups, such as carboxylate ions, covalently bonded to their surfaces. These surface anions give NF membranes a high rejection for salts with multivalent anions, such as sulfates, carbonates, and phosphates. Nanofiltration membranes are commonly used as water softening membranes, because they can very effectively remove most hard water components, i.e., carbonates and sulfates of calcium and magnesium. NF membranes designed for this purpose are usually rated in terms of their MgSO4 rejection, which is typically in the range of 95–99%. Other applications of NF membrane include organics removal from surface water, radium removal from ground water, sulfate removal from seawater, and food and pharmaceutical applications such as concentration of dilute solutions and desalting of cheese whey. Like RO membranes, many NF membranes are polyamide thin film composite membranes. These membranes can be prepared by interfacial reaction of piperazine with 1,3,5-benzenetricarbonyl trichloride and=or isophthaloyl dichloride, or by treating polyamide thin film composite RO membranes with compounds such as mineral acids, to increase their flux and lower their salt rejection.[2] A few ceramic NF membranes have also been developed. New methods

to produce NF membrane are a subject of ongoing research, as exemplified by two recently reported methods: filling the pores of microfiltration (MF) membrane with cross-linked poly(4-vinylpyridinium) salts,[12] and grafting charged groups onto the surface of UF membrane.[13]

ULTRAFILTRATION AND MICROFILTRATION MEMBRANES Ultrafiltration and microfiltration membranes are both porous membranes but with two different ranges of pore size. Microfiltration membranes are used to separate particles in the range of 0.02–10 mm. Ultrafiltration membranes are used to remove much smaller species, such as dissolved macromolecules and colloids with 1000–500,000 mol wt. Ultrafiltration membranes have pore diameters in the range of 1–100 nm and separate primarily by size exclusion. In this process, a liquid containing dissolved or suspended matter is driven through the membrane by an applied pressure difference across the membrane, and any species larger than the pore size of the membrane is rejected (prevented from passing through the membrane). The observed rejection R of a species is given by R ¼ 1
cp =cr where cp is the concentration of the species in the permeate liquid that passes through the membrane, and cr is the concentration of the same species in the retentate stream, which remains on the feed side of the membrane. The pore size of a UF membrane is described by its nominal molecular weight cutoff (MWCO), defined as the molecular weight of the smallest species for which the rejection is 90% or greater. Although the nominal

3220

MWCO is a useful measure of a membrane’s average pore size, it is only an approximate indication of how a UF membrane may perform in a given application. In reality, the rejection level for a solute molecule depends not only on its molecular weight, but also on its shape, its molecular orientation, and operating conditions such as feed pressure and solute concentration. To eliminate the effect of molecular orientation in MWCO measurement, spherical macromolecules such as globular proteins are recommended as test solutes. Although there is at present no generally accepted standardization of the procedure for MWCO measurement, the factors that affect the measurement have been reviewed in detail, and recommended operating conditions have been proposed.[14] Two phenomena that complicate UF processes are concentration polarization and fouling. During filtration, the solution upstream of the membrane becomes enriched in the retained molecules or particles. At high permeation flux rates, mixing and diffusion in the upstream liquid might not be rapid enough to maintain uniform concentration, resulting in a layer of highly concentrated solution next to the membrane. This effect is called concentration polarization, and it leads to lower flux and lower apparent rejection, because the effective solute concentration at the membrane surface is higher than that of the bulk solution. Fouling is the deposition of components from the feed stream onto the surface and in the pores of the membrane, which can cause total pore blockage or an effective reduction of pore diameter, and leads to lower flux and in some cases, increased rejection. Fouling can be partially reversed by membrane cleaning, but some fouling is also irreversible, eventually making membrane replacement necessary. Although some inorganic membranes are available, most UF membranes are made of polymers. The earliest UF membranes were made of cellulose acetate, but today, the most widely used polymers are polysulfone and polyethersulfone, which are preferred because of their higher resistance to extremes of pH and temperature. Other polymers used include polyvinylidenefluoride, polyacrylonitrile, and polyamides.[15] Most UF membranes are asymmetric, having a thin separating layer or ‘‘skin layer’’ with small pores on one side of the membrane, and a much thicker layer with larger pores below the membrane which provides structural support with minimum flow resistance. Asymmetric membranes are manufactured by wet phase inversion casting. In this process, a casting solution of a polymer in a water-miscible solvent is spread in a thin layer onto a flat surface and then immersed in water. The water causes extraction of solvent and precipitation of the polymer as a porous flat sheet. The skin layer is formed on the upper surface that was in direct contact with water, and the underlying

Water Reclamation

porous structure has progressively larger pores with increasing distance from the skin layer. The skin layer pore size and MWCO are determined by the casting solution formulation (polymer concentration, type of solvent, and additives) and by the phase inversion casting process conditions. Flat sheet UF membrane is usually manufactured with a polyester or polyolefin nonwoven fabric backing for added durability, and membrane sheets are rolled into spiral-wound membrane modules (see Fig. 3). Ultrafiltration membranes are also commonly manufactured as hollow membrane fibers and sold as modules containing bundles of fibers. Both spiral-wound and hollow fiber membrane modules are usually operated in crossflow filtration mode, wherein the feed liquid flows tangential to the surface of the membrane under pressure. The high tangential feed velocity helps to sweep particles away from the membrane surface and reduce concentration polarization and fouling. In crossflow filtration, only a portion of the feed passes through the membrane and the remaining liquid exits as a separate retentate stream, carrying rejected components away and out of the membrane module. Microfiltration membranes are similar to UF membranes but have larger pores. Microfiltration membranes are used to separate particles in the range of 0.02–10 mm from liquid or gas streams. Commercial MF membranes are made from a wide variety of materials including polymers, metals, and ceramics. A wide variety of membrane module designs are available including tubular, spiral wound, pleated sheet, hollow fiber, and flat sheet designs. Some modules are best suited for crossflow filtration,[16] and others are designed for dead-end filtration.[17] In dead-end filtration, the feed liquid flows normal to the surface of the membrane, and retained particles build up with time as a cake layer on the membrane surface or within the pores of the membrane.

LIQUID MEMBRANES AND FACILITATED TRANSPORT MEMBRANES Liquid membranes are ultra-thin films of water or an organic liquid that are stabilized by the presence of surfactant molecules and other additives. This film is placed between two miscible phases: one is a feed phase, and the other is a phase receiving the materials that are transferred from the feed phase through the membrane phase. This means that if the feed and the receiving phases are both aqueous, an organic liquid membrane will be needed, whereas if the feed and the receiving phases are organic, the liquid membrane will need to be aqueous. In commercial applications, liquid membranes are prepared via emulsion techniques as agglomerations of micrometer-sized droplets.

Water Reclamation

This emulsification technique aims to maximize membrane surface area. The internal phase of the droplets can be placed inside the pores of polymeric films, or inside hollow fibers.[18–21] The membrane phase can contain surfactants or no surfactants for easy coalescence. This aspect of the liquid membrane system will be discussed in more detail in the application section. The transfer of materials through liquid membranes is based on two facilitated transport mechanisms. In the type 1 facilitation, the solute reacts with a reagent dissolved in the receiving phase, or the internal phase in an emulsion liquid membrane system. This reaction should be selected so that it produces a nonpermeating compound. The purpose is to effectively reduce the concentration of the solute to zero in the receiving phase, thereby facilitating the continuous transfer of the materials. The type 2 facilitation uses a complexing agent for the transferred compound in the membrane phase. The complexing agent reacts with the transferred compound inside the membrane phase to form a complex, which permeates through the membrane phase to the receiving phase, or the internal phase in an emulsion liquid membrane system. The transfer through the membrane phase is based on the concentration gradient of the complex molecules, and the decomplexation at the interface between the membrane and the receiving phase is based on a counter-diffusion of ions. The counterdiffusing ions from the receiving phase are usually either hydrogen ions or hydroxide ions from the acid or base materials placed in the receiving phase (Fig. 4). The facilitated transport will stop when the amount of the acid or base material is consumed.[21–23] There have been several modified systems since the invention of liquid membranes, including a facilitated transport mechanism. One of them is to disperse the receiving solution in an organic membrane phase on one side of a porous hollow fiber.[24,25] Two plants to treat contaminated groundwater were built and operated based on this revised liquid membrane system. More discussion about this application is given below in the application section.

MEMBRANE PROCESSES FOR WATER RECLAMATION With world population increase and scarcity of water sources, the key to providing a sustainable development of water resources is to carry out water reclamation. The concept of water reclamation is to treat wastewater from different sources to meet different water quality requirements of various water applications economically. USEPA provided guidelines for nonpotable water reuse,[26] which stated various conventional technologies commonly used in treating municipal wastewater such as primary and secondary treatment, and specified the

3221

W

Fig. 4 Facilitated transport in a liquid membrane system with complexing agent in the membrane phase and acid in the receiving phase.

water quality level required for various applications such as irrigation, industrial cooling processes, and groundwater recharge. As for industrial wastewater reuse, the American Institute of Chemical Engineers Center for Waste Reduction Technologies published a book[27] to provide guidelines for a systematic approach, which was summarized by Zinkus.[28] With the advancement of membrane technologies and engineering in the integration of MF, UF, NF, and RO membrane systems, and possibly a liquid membrane system, wastewater reclamation can provide high quality water for underground water replenishment and direct household nonpotable use, and source water for ultra-pure water applications. Membrane technology provides several advantages over conventional treatment processes: 1. It is simpler to apply in treating a wide range of contaminants in wastewater effectively. 2. A membrane system occupies a smaller footprint. Efficient use of floor space is critical in meeting the future expansion needs of municipal wastewater treatment plants, which are typically located within densely populated areas. 3. Modular design of membrane treatment systems allows easier future expansion. 4. Less coagulation or flocculation chemicals are required in the treatment processes.

3222

Membrane technology will play a major role in water treatment and wastewater reclamation, as indicated in Desalination and Water Purification Technology Roadmap,[29] and will require continued research and development in: 1. reducing the operational cost of membrane systems; 2. enhancing membrane fouling resistance to reduce membrane cleaning frequency; and 3. providing better on-line monitoring of membrane module and system integrity to ensure a 100% safety barrier in rejecting bacteria and viruses. It should be noted that, in developing a membrane process for treating a specific type of wastewater efficiently and effectively, the Six Sigma quality management program was suggested to be incorporated in the development process.[30]

Municipal Wastewater Reclamation In the Orange County Sanitary District of California, the wastewater treatment plant (Fig. 5) treats approximately 200 million gallons per day (MGD) of municipal wastewater through preliminary treatment (bar screens and grit processing to separate solid waste from wastewater), primary treatment (a physical=chemical treatment process to settle small solids), and secondary treatment (an aerobic bio-treatment process consisting of trickling filter and activated sludge plants to treat dissolved matter with activated sludge). Effluent from secondary treatment is discharged into the ocean. Because of excessive exploitation of underground water in this area, however, a barrier to prevent seawater intrusion was necessary. In 1976, Water Factory 21 was built to provide highly treated (purified) water as this water barrier. This treatment plant reclaimed 15 MGD of the secondary treatment effluent from Orange Country Sanitary District’s wastewater plant by pretreatment (flocculation, recarbonation, and multi-media filtration), followed by division into two treatment processes: 6 MGD was purified by an RO membrane system, and 9 MGD was treated with activated carbon and disinfected with chlorine. This highly treated wastewater (with quality of potable water) was then blended with underground water before being injected into the ground as a water barrier for seawater intrusion prevention. To provide a sustainable water resource in Orange County, a 100,000 acre-foot per year Ground Water Replenishment System (GWRS) was proposed[31] on the basis of the success of the Water Factory 21 project. Because of limitations in land space available

Water Reclamation

to this treatment plant, a new pretreatment process using a membrane system (MF and UF system) was tried and approved by virtue of its superior performance, cost effectiveness, and smaller footprint than the conventional pretreatment process.[32] The first phase of GWRS, a 71,600 acre-foot wastewater reclamation project using a new integrated membrane process consisting of an MF system followed by an RO system, and last by a UV disinfection system, has been under construction since 2001. Membrane technology also shows its potential in replacing the secondary activated sludge treatment step. Although using a membrane system as a replacement for sedimentation in the activated sludge process was tried in the late 1960s, current membrane bioreactor (MBR) systems provide more robust engineering design to overcome membrane fouling.[33] Compared to the conventional activated sludge process, the MBR provides complete solids removal, a significant disinfection capacity, high treatment rate, and high efficiency of organic and nutrient removal. Therefore, the MBR strengthens the future role of membrane systems in wastewater recycling and reuse applications. Currently, more than 500 commercial MBR systems are in operation worldwide.

Industrial Wastewater Reclamation In addition to wastewater effluent treatment regulations as a driving force for industrial wastewater treatment, other factors such as water and wastewater management costs, operation costs, and recovery and recycle of processing chemicals play critical roles in determining industrial wastewater reclamation needs and treatment processes. With 16 million tons=day of water usage, the paper and pulp industry in the USA generates enormous wastewater pollution problems.[34] Characteristics of wastewater from pulp and paper are high BOD and COD, and high suspended solids content. The industrial ‘‘end of pipe’’ abatement approach is to treat all wastewater together, using primary treatment to settle the solids and to biological processes such as activated sludge to reduce BOD and COD content. Industry has tried membrane technology to reclaim wastewater, thereby reducing water consumption and wastewater pollution. Without significant pretreatment of wastewater, however, traditional hollow fiber, spiral-wound, or tubular membrane modules would be fouled within a very short time. Nuortila-Jokinen[35] used shear enhanced membrane modules to overcome the membrane fouling problem, thereby allowing membrane processes to treat and reuse pulp mill effluent effectively. Shear enhanced membrane modules used a vibration or spinning force to produce 150,000 s
1

Fig. 5 An overall block diagram of the Orange County municipal wastewater treatment process and Water Factory 21 process.

Water Reclamation 3223

W

3224

Water Reclamation

shear rate at the membrane surface, which was three to five times the rate attainable in crossflow systems.[36] This dispersal force prevented foulants from plugging up the membrane surface and resulted in a significantly higher permeation rate and longer operation time between cleanings. Wagner[37] described the application of membranes for wastewater treatment in a paper manufacturing plant. Wastewater generated during the process of paper formation was rather white because of the high content of suspended solids. This white water from the paper mill was reclaimed by the use of a plate and frame UF filter. Wagner also noted that as more water was recycled in a pulp or paper mill, more salt built up. Salt concentration had to be reduced in order to decrease the possibility of corrosion in machinery and pipelines. Nanofiltration or RO membranes showed capability in significantly reducing salt content. Using UF membrane, fresh water consumption of the pulp and paper industry could generally be reduced by 50%, while NF or RO membrane could provide a further 50% reduction. Two commercial size plants for groundwater treatment based on liquid membrane technology in general, and the supported liquid membrane using hollow fibers in particular, were built and operated in Baltimore, U.S.A.[20] Specifically, the purpose of the two plants is for hexavalent chromium cleanup. One plant went into commercial operation in March 1999 and the other approximately about a year later. The liquid membrane system in these two plants is able to reduce metal-ion concentration from 100–1000 ppm range to approximately 0.05 ppm and, meanwhile, produce a concentrated chromium solution, which is the spent strip solution, at approximately 20% Cr (VI). This concentration is suitable for sale for reuse. Other commercialized major applications for liquid membranes are the recovery of zinc from rayon plant

effluents,[38] the removal of phenol from industrial wastewater streams,[39] and the treatment of cyanidecontaining wastewater from gold leaching solutions.[40,41] The process of extraction of uranium from phosphoric acid was successfully developed in a pilot plant and is waiting for commercial opportunity.[42] Membrane technology also helped an aluminum can manufacturer (Fig. 6) treat processes wastewater to reduce ever-increasing fresh water and wastewater discharge fees.[43] Oily wastewater from the can forming process and other facilities processes was filtered by a UF unit. The UF permeate combined with can washing wastewater was further treated by an NF unit. Three RO units were used to purify the incoming municipal water and the NF permeates. Using such an integrated membrane system, this aluminum can manufacturer recovered approximately 88% of the process water.Cassano[44] reviewed the potential of using UF and NF in the treatment of aqueous solutions from leather processing industries. Water is used in almost every step during leather manufacturing processes. Through the application of various types of membranes, process water can be either recycled directly or used in other steps of leather manufacturing. Polysulfone spiral-wound UF modules with an MWCO of 20,000 could maintain the sulfide concentration during the unhairing steps, and NF element could concentrate the tanning solution for reuse instead of discharge into the wastewater treatment plant. Semiconductor wafer production also consumes a large quantity of fresh water. The Philips San Antonio facility was producing 150 mm wafers and planned to move into 200 mm wafer production in early 2000. One task it encountered was the need to increase its supply of high purity water. Its project team evaluated several options, including expanding the fresh water supply and recycling waste wafer rinse water. The team

Fig. 6 Integrated membrane system used in a can manufacturing plant.

Water Reclamation

concluded that reclaiming and recycling the process wastewater would meet both the project’s critical schedule milestones as well as its budget.[45] A 180 gpm double pass RO system was installed with weak acid cation resin as RO pretreatment and posttreatment. This system provided a maximum 87% recovery rate with 93% or better rejection of waste rinse water TOC.

CONCLUSIONS Water reclamation, the treatment of wastewater to meet the water quality standards of various applications economically, is becoming increasingly important in view of the increasing world population and scarcity of fresh water sources. The major technology used for water reclamation is membrane technology. This entry gives an overview of the major membrane types used for water reclamation: reverse osmosis, nanofiltration, ultrafiltration, microfiltration, and liquid membranes. Applications of these membranes in municipal and industrial wastewater reclamation have been described.

REFERENCES 1. Morgan, P.W. Interfacial polycondensation in unstirred systems. In Condensation Polymers: by Interfacial and Solution Methods; Interscience Publishers: New York, 1965; 19–64. 2. Petersen, R. Composite reverse osmosis and nanofiltration membranes. J. Membr. Sci. 1993, 83 (1), 81–150. 3. Mulder, M. Basic Principles of Membrane Technology; Kluwer Academic Publishers: Boston, 1991; 10–12. 4. Strathmann, H. Synthetic membranes and their preparation. In Handbook of Industrial Membrane Technology; Porter, M., Ed.; Noyes Publications: Park Ridge, NJ, 1990; 12–1445–49. 5. Chen, V.; Fane, A.G.; Fell, C.J.D. The use of anionic surfactants for reducing fouling of ultrafiltration membranes: their effects and optimization. J. Membr. Sci. 1992, 67 (2–3), 249–261. 6. Li, N.N.; Kuehne, M.A.; Petersen, R.J. High Flux Reverse Osmosis Membrane. US Patent 6,162,358, December 19, 2000, NL Chemical Technology, Inc. 7. Li, N.N.; Kuehne, M.A.; Petersen, R.J. A novel high flux membrane. ICOM ‘99: the 1999 International Congress on Membranes & Membrane Processes, Toronto, Canada, June 12–18, 1999; North American Membrane Society, 1999; Abstract Number 236.

3225

8. Kuehne, M.A.; Song, R.Q.; Li, N.N.; Petersen, R.J. Flux enhancement in TFC RO membranes. Environ. Prog. 2001, 20 (1), 23–26. 9. Kuehne, M.A.; Li, N.N.; Petersen, R.J. Flux enhancement in TFC membranes. ICOM ‘99: the 1999 International Congress on Membranes & Membrane Processes, Toronto, Canada, June 12–18, 1999; North American Membrane Society, 1999; Abstract Number 82. 10. Song, R.Q.; Ho, W.S.; Li, N.N.; Petersen, R.J. Polymeric TFC membrane formation studies. ICOM ‘99: the 1999 International Congress on Membranes & Membrane Processes, Toronto, Canada, June 12–18, 1999; North American Membrane Society, 1999; Abstract Number 469. 11. Eriksson, P. Nanofiltration extends the range of membrane filtration. Environ. Prog. 1988, 7 (1), 58–62. 12. Mika, A.M.; Childs, R.F.; Dickson, J.M. Ultralow pressure water softening: a new approach to membrane construction. Desalination 1999, 121 (2), 149–158. 13. Be´quet, S.; Remigy, J.-C.; Rouch, J.-C.; Espenan, J.-M.; Clifton, M.; Aptel, P. From ultrafiltration to nanofiltration hollow fiber membranes: a continuous UV-photografting process. Desalination 2002, 144 (1–3), 9–14. 14. Cheryan, M. Membrane properties. In Ultrafiltration and Microfiltration Handbook; Technomic Publishing Company: Lancaster, Pennsylvania, 1998; 89–110. 15. Kulkarni, S.S.; Funk, E.W.; Li, N.N. Ultrafiltration membranes. In Membrane Handbook; Ho, W.S., Sirkar, K.K., Eds.; Van Nostrand Reinhold: New York, 1992; 408–431. 16. Mir, S.; Michaels, S.L.; Goel, V; Kaiser, R. Crossflow microfiltration: applications, design, and cost. In Membrane Handbook; Ho, W.S., Sirkar, K.K., Eds.; Van Nostrand Reinhold: New York, 1992; 571–594. 17. Goel, V; Accomazzo, M.A.; DiLeo, A.J.; Meier, P.; Pitt, A.; Pluskal, M. Deadend microfiltration: applications, design, and cost. In Membrane Handbook; Ho, W.S., Sirkar, K.K., Eds.; Van Nostrand Reinhold: New York, 1992; 506–570. 18. Li, N.N. Separating Hydrocarbons with Liquid Membranes. US Patent 3,410,794, November 12, 1968. 19. Li, N.N.; Cahn, R.P.; Shrier, A.L. Removal of Organic Compounds by Liquid Membrane. US Patent 3,617,546, November 2, 1971. 20. Jacoby, M. Norman Li Wins Perkin Medal. Chem. Eng. News (March 6, 2000), 78 (10), 60–61. 21. Ho, W.S.; Li, N.N. Emulsion liquid membranes: definitions and theory. In Membrane

W

3226

22.

23.

24.

25.

26.

27.

28.

29.

30.

31. 32.

33.

34.

Water Reclamation

Handbook; Ho, W.S., Sirkar, K.K., Eds.; Van Nostrand Reinhold: New York, 1992; 597–655. Cahn, R.P.; Li, N.N. Commercial applications of emulsion liquid membranes. In Separation and Purification Technology; Li, N.N, Calo, J.M., Eds.; Marcel Dekker: New York, 1992; 195–212. Bartsch, R.A.; Charewicz, W.A.; Kang, S.I.; Walkowiak, W. Proton-coupled transport of alkali metal cations across liquid membranes by ionizable crown ethers. In Liquid Membranes: Theory and Applications; Noble, R.D., Way, J.D., Eds.; ACS Symp. Ser. No. 347; American Chemical Society: Washington, D.C., 1987; 86–97. Ho, W.S.; Poddar, T.K. New membrane technology for removal and recovery of chromium from waste waters. Environ. Prog. 2001, 20 (1), 44–52. Ho, W.S.; Wang, B.; Neumuller, T.E.; Roller, J. Supported liquid membranes for removal and recovery of metals from waste waters and process streams. Environ. Prog. 2001, 20 (2), 117–122. EPA. Guidelines for Water Reuse, EPA=625= R-92=004; Office of Compliance, US Environmental Protection Agency: Washington, Dc, 1992. Byers, W.; Doerr, W.; Krishnan, R.; Peters, D. How to Implement Industrial Water Reuse; American Institute of Chemical Engineers: New York, 1995. Zinkus, G.A.; Byers, W.D.; Doerr, W.W. Identify appropriate water reclamation technologies. Chem. Eng. Prog. 1998, 94 (5), 19–31. U.S. Bureau of Reclamation; Sandia National Laboratories. Desalination and Water Purification Technology Roadmap—A Report of the Executive Committee, Desalination & Water Purification Research & Development Program Report #95; 2003. Song, R.Q.; Li, J.C.; Kuehne, M.; Tsai, M.; Li, N. Development of an advanced membrane for water treatment. In Water Purification and Reuse. Potsdam, Germany, June 8–13, 2003; Engineering Conferences International: Brooklyn, New york, 2003. http:==www.gwrsystem.com (accessed April 2001). Lesile, G.L.; Dunivin, W.R.; Gabillet, P.; Conklin, S.R.; Mills, W.R.; Sudak, R.G. Pilot testing of microfiltration and ultrafiltration upstream of reverse osmosis during reclamation of municipal wastewater. Proceedings of American Desalting Association Biennial Conference, Monterey, California, August 1996. Stephenson, T.; Judd, S.; Jefferson, B.; Brindle, K. Membrane Bioreactors for Wastewater Treatment; IWA Publishing: London, England, 2000. EPA. Profile of the Pulp and Paper Industry, EPA=310-R-95-015; Office of Compliance, US Environmental Protection Agency: Washington, D.C., 1995.

35. Nuortila-Jokinen, J.; Kuparinen, A.; Nystro¨m, M. Tailoring an economical membrane process for internal purification in the paper industry. Desalination 1998, 119 (1–3), 11–19. 36. Culkin, B.; Plotkin, A.; Monroe, M. Solve membrane fouling problems with high-shear filtration. Chem. Eng. Prog. 1998, 94 (1), 29–33. 37. Wagner, J. Membrane technology in wood, pulp and paper industries. In Membrane Technology in Water and Wastewater Treatment; Hillis, P., Ed.; The Royal Society of Chemistry: Cambridge, UK, 2000; 233–240. 38. Draxler, J.; Marr, R.; Pro¨tsch, M. Commercialscale extraction of zinc by emulsion liquid membranes. In Separation Technology, Engineering Foundation Conference, Schloss Elmau, Germany, April 27–May 1, 1987; Li, N.N., Strathmann, H., Eds.; American Institute of Chemical Engineers: New York, 1988; 204–214. 39. Zhan, X.; Liu, J.; Fan, Q.; Lian, Q.; Zhang, X.; Lu, T. Industrial application of liquid Mmembrane separation for phenolic wastewater treatment. In Separation Technology, Engineering Foundation Conference, Schloss Elmau, Germany, April 27–May 1, 1987; Li, N.N., Strathmann, H., Eds.; American Institute of Chemical Engineers: New York, 1988; 190–203. 40. Treatment of Cyanide-Containing Waste Water from Gold Mine Operation by Liquid Membrane Technology. News release in Kexue Bao (Newspaper of Science), China, October 16, 1987. 41. Jin, M.; Zhang, Y. Study on extraction of gold and cyanide from alkaline cyanide solution by liquid membrane. Proceedings, The 1990 International Congress on Membranes and Membrane Processes, ICOM ’90, Chicago, August 20–24, 1990; American Institute of Chemical Engineers: New York, 1990; Vol. 1, 676–678. 42. Bock, J.; Klein, R.R.; Valint, P.L.; Ho, W.S. Liquid membrane extraction of uranium from wet process phosphoric acid. In Sulfuric=Phosphoric Acid Plant Operations; American Institute of Chemical Engineers: New York, 1982; 175–183. 43. Abolmaali, B.; Yassine, I.; Capone, P. Water recovery from an aluminum can manufacturing process using spiral wound membrane elements. In Membrane Technologies for Industrial and Municipal Wastewater Treatment and Reuse; Water Environment Federation: Alexandria, VA, 2000; 51–56. 44. Cassano, A.; Molinari, R.; Romano, M.; Drioli, E. Treatment of aqueous effluents of the leather industry by membrane processes. J. Membr. Sci. 2001, 181 (1), 111–126. 45. Weems, J.; Sohns, R.; Bell, C.; Pandya, K. Water reuse. Ultrapure Water 2003, 20 (4), 25–36.

Wide Band-Gap Electronics Materials W Mark A. Prelas Krishnendu Saha Nuclear Science and Engineering Institute, University of Missouri–Columbia, Columbia, Missouri, U.S.A.

INTRODUCTION A wide band-gap material is a semiconductor that has a band-gap approximately greater than 2 eV. Some of the most prominent wide band-gap materials are GaP (2.26 eV), 3C–SiC (2.36 eV), 6H–SiC (3.0 eV), 4H–SiC (3.23 eV), GaN (3.2 eV), diamond (5.46 eV), AlN (6.2 eV), and BN (6.1–6.4 eV). These materials have applications in high-efficiency optoelectronic devices such as blue and UV light emitting diodes (LEDs) and lasers, as well as high-power, hightemperature, and high-frequency electronic devices. Electronic devices formed in wide band-gap materials operate at high temperatures without suffering from intrinsic conduction effects because of the wide energy band-gap. Some wide band-gap materials emit and detect short-wavelength light, which has applications in blue LEDs and nearly solar blind UV photodetectors. Wide band-gap materials, in general, have a high breakdown voltage, which allows them to withstand a voltage gradient much greater than that of Si or GaAs without undergoing avalanche breakdown. This property makes wide band-gap materials excellent for applications in very high-voltage, high-power devices such as diodes, power transistors, power thyristors, and surge suppressors, as well as high-power microwave devices. Because of their excellent thermal properties, devices made of wide band-gap materials can be placed close together, allowing a high device packing density for integrated circuits. Packing density is dependent on molecular size of the element in the chip and an example of packing density is a few million transistors in an area of 4 mm2 with spacing between elements of approximately 1.5 mm. Diamond, for example, has the highest known thermal conductivity (five times larger than copper). Wide band-gap materials can operate at very high power levels and still dissipate even large amounts of excess heat. Many wide band-gap materials have a high electron drift velocity, which allows them to operate at high frequencies (RF and microwave). The electron drift velocity is the velocity of electrons in semiconductors under the influence of electric field. Wide band-gap materials have a high resistance to chemical attack, which allows them to be used in corrosive environments. Additionally, wide band-gap materials Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120025854 Copyright # 2006 by Taylor & Francis. All rights reserved.

are resistant to radiation, which makes them ideal for devices that require radiation hardening, such as components for satellites and spacecraft.

BACKGROUND Diamond Diamond is discussed in Diamond and Diamond-Like Film Applications. III–V Materials (Nitrides) Wide band-gap materials have always been a subject of interest for electronic applications.[1] Light emitting diodes with blue and green light emitting capability have been in the market and ultraviolet and blue laser diodes , high-speed transistors, and ultraviolet photodetector technology have been demonstrated. The III–V materials (nitrides) especially stand out. The properties like high electron mobility, high current carrying capability, high thermal capability, high temperature operation, and high breakdown field are important attributes of III–V materials. Electron mobility and hole mobility are the measure of scattering of electron or holes in semiconductors. The first report of synthesizing a small GaN crystal was made by Johnston and Parsons in 1932.[2] Grimmeiss and Koelmans in 1959 performed luminescence studies.[3] In 1969, Maruska and Tietjen succeeded in growing the first single-crystal GaN on a sapphire substrate by hydride vapor phase epitaxy (HVPE).[4] This work was a major breakthrough because of the difficulty of growing large bulk GaN single crystals and because the available methods relied on heteroepitaxial growth. Heteroepitaxial growth is a method in which a thin layer of single-crystal material is deposited on a single-crystal substrate, the chemical composition of the depositing material being different from that of the substrate. Additionally, it was found that GaN has a direct-transition band structure with band-gap energy of about 3.39 eV, at room temperature. This caused an acceleration in research on GaN as seen in Fig. 1, period A, which lists the number of worldwide 3227

3228

Wide Band-Gap Electronics Materials

Fig. 1 Number of worldwide publications on nitrides over the years. (From Ref.[1].)

publications related to GaN in any given year. By 1971, Dingle et al. demonstrated optically pumped UV stimulated emission from a GaN crystal at 2 K.[5] The demonstration used Zn as a deep acceptor. By changing doping of the Zn level, various colors can be produced and blue, green, yellow, or red can be emitted. The first blue LED using a metal–insulator– semiconductor structure was developed by Pankove et al. in 1971.[6] The basic studies of the physical properties of GaN crystals occurred in this period. Ejder reported energy dispersion of the refractive index of GaN in 1971.[7] Then, studies of luminescence including exciton recombination were reported by several researchers.[8,9] Through these studies some basic properties like positions of exciton ground state transition energies of GaN and the lowest band-gap of GaN were clarified. However, the crystalline quality of GaN in those days was not sufficient to characterize the intrinsic properties. Moreover, because of the high n-type background, p-type conduction was very difficult to obtain. N-Type is the lattice where there is an extra electron while p-type is where there is an one electron less in the lattice. Because of this limitation many researchers discontinued Group III nitride research and as a result, the number of publications gradually decreased during this period as seen in Fig. 1, period B. With the introduction of metalorganic vapor phase epitaxy, which is preferred compared to molecular beam epitaxy (MBE), dramatic improvement of crystalline quality of nitrides,

was obtained. This caused a revolutionary breakthrough in period C, which resulted in extensive research from period C to period D. AlN, GaN, InN, and their alloys are all wide bandgap semiconductors. They crystallize into both wurtzite and zincblende polytypes.[10] Wurtzite GaN, AlN, and InN have direct room temperature band-gaps of 3.4, 6.2, and 1.9 eV, respectively. In cubic form, GaN and InN have direct band-gaps while AlN has an indirect energy band-gap. Alloys of GaN with AlN and InN can form materials with a wide range of energy bandgaps. Group III nitrides span a continuous range of direct band-gap energies throughout much of the visible spectrum into the ultraviolet wavelength. Short-wavelength optoelectronic device application is one of the reasons for the increase in research papers after 1990. Group III nitride optoelectronic devices, like LEDs and lasers, can be fabricated for green, blue, and UV wavelengths. The LEDs have applications as elements for full color displays and also as elements for signal and illumination applications. Used as coherent sources they have applications in high-density optical read and write device. In the latter application, because of the fact that the diffraction limited optical storage density increases roughly quadratically as the probe laser wavelength is reduced, nitride based coherent sources at wavelengths down to UV are attracting attention. For optical storage applications these devices have shown storage and retrieval property of a large

Wide Band-Gap Electronics Materials

3229

number of images and large video files. Currently, optical storage devices use lasers made from InGaAlP heterostructures whose wavelengths are limited to about 550 nm. ZnSe alloys have recently been explored that operate in fringes of green and blue wavelength. These devices have a short lifetime probably owing to stacking faults, which is in the range of about 105=cm2. The potential short life of ZnSe because of mechanical instability is a significant concern. GaN and=or its alloys can overcome these problems because of longer lifetime and fill the green, blue, and UV regions of the spectrum. During the past several decades, lasers and LEDs have expanded in terms of both range of wavelengths and brightness. Introduction of a bright blue emitter paved the way for full color displays. The low power consumption of LEDs will allow full color display to extend battery life and reduce battery weight. Both n- and p-type doping is possible in SiC and it has an excellent power amplification device performance. As a result it is a direct competitor of the nitrides in this application. Nitrides form direct bandgap heterostructures allowing the placing of carriers at the interface and carrier confinement. In addition, nitrides can form good ohmic contacts, which are imperative for power devices. Many wide band-gap semiconductors are produced by epitaxial growth. A brief history of applications of these techniques is given below.[11] The first event of importance was the success of Al-ferov in 1969 in fabricating a continuous-wave laser diode that operated at room temperature by liquid phase epitaxy. For this, he received the Nobel Prize in physics in 2000 with two other scientists. Liquid phase epitaxy, which was proposed by Nelson, makes high-quality thin film semiconductors.[12] In III–V LPE, the compounds are grown from Group III metal solutions on a substrate. This produces highly stochiometric, thin, pure, and perfect films. The next important event in the history of III–V epitaxy is the invention of the high electron mobility transistor (HEMT) by Mimura et al., which was made by MBE.[13] They used the idea

of modulation doping that was originally proposed by Dingle et al.[14] The growth technology of high-quality and ultrathin III–V films of about 10 nm thickness was required for the fabrication of HEMT. At that time, such films could be grown only by MBE. Almost at the same time, the technology of metal organic chemical vapor deposition (MOCVD) was developed and used for growing high-quality thin films. Without MBE and MOCVD, neither HEMT nor high-performance laser diodes such as quantum well (QW) and multiquantum well (MQW) lasers could be made. Recent successes in growing high-quality GaN and related compounds were achieved by Amano et al. by using a low-temperature buffer layer on sapphire in MOCVD.[15] The improvement in crystal quality realized made it possible to get p-type GaN. The success of p-GaN was a real breakthrough for making blue to ultraviolet lasers and LEDs.

PROPERTIES OF WIDE BAND-GAP MATERIALS General Properties The properties like high electron mobility, high current carrying capability, high thermal capability, high temperature operation, and high breakdown field) of wide band-gap materials are superior to silicon. The Keyes figure of merit (KFM) takes into account the power density dissipation for closely packed integrated circuits. High thermal conductivity is an important element for the KFM. The KFM is based on Vsat (electron saturation velocity), st (thermal conductivity), and er (dielectric constant). The relative value of the KFM is related to the speed of the transistor in the material as given below: KFM ¼ st ðVsat =er Þ0:5 The higher the value of KFM, the higher will be the speed of the transistor (Table 1).

Table 1 Properties of some wide band-gap semiconductors Material Si

Band-gap (eV) 1.1

rt (300 K) (W/cm) 1.5

er 11.8

GaN

3.5

1.5

9.5

aSiC(6H)

3.0

5.0

10.0

bSiC(4H)

3.3

5.0

9.7

Diamond

5.47

20.0

5.5

BN

6.1

5.7

3.3

AlN

6.2–6.4

3.0

9.0

(From NSM Archive, Http:==www.ioffe.rssi.ru=SVA=NSM=Semicond=.)

Vsat (cm/sec) 7

KFM (W cm
1/2sec
1/2) 2

Ratio to silicon

1.0  10

13.8  10

1.0

2

1.76

2.0  107

5.12

7

70.7  102 2

5.8

2.7  107

444.0  102

32.2

7

54.8  102

4.0

7

2.5  10 2.5  10

3.1  107 3.0  10

24.3  10 80.3  10

174.7  102

12.7

W

3230

Wide Band-Gap Electronics Materials

The Johnson figure of merit (JFM) is used to compare materials for power microwave applications. Larger values indicate superior performance in microwave power applications. The higher the value, the higher will be the power capability. This figure of merit is the product of the breakdown voltage (Vb) and the electron saturation velocity (Vsat). JFM ¼ ðVb Vsat Þ=2p Wide band-gap materials have an excellent JFM, about two orders of magnitude higher than silicon, making them excellent candidates for high-power microwave electronics (Table 2). Overall, the properties of wide band-gap materials have many advantages over Si, GaAs, and related semiconductor materials (Table 3). Wide band-gap materials have a much higher breakdown voltage and a higher thermal conductivity than silicon and related semiconductors. In addition when you compare the electron mobility (Si: 1450 cm2=V=sec, GaAs: 8500 cm2= V=sec) and hole mobility (Si: 450 cm2=V=sec, GaAs: 400 cm2=V=sec) of standard semiconductor materials to that of diamond (electron mobility: 2200 cm2=V=sec, hole mobility: 2000 cm2=V=sec), there are some very distinct advantages that come to light. The large band-gap difference makes these materials suitable for high-temperature electronics, power electronics, and radiation-hardened electronics. The material properties of some individual wide band-gap materials are discussed below.

SPECIFIC MATERIAL PROPERTIES GaP Gallium phosphide is primarily used for the manufacture of red and green diodes. Its band structure is direct. Some properties of wide band-gap materials are given in Table 2. GaP has a number of potential shallow-level donors (n-type dopant). These include sulfur (substituted for

P-activation energy of 0.107 eV), selenium (substituted for P-activation energy of 0.105 eV), tellurium (substituted for P-activation energy of 0.093 eV), lithium (substituted for P-activation energy of 0.091 eV), germanium (substituted for Ga-activation energy of 0.204 eV), silicon (substituted for Ga-activation energy of 0.085 eV), tin (substituted for Ga-activation energy of 0.072 eV), and lithium (substituted for Ga-activation energy of 0.061 eV). Of these potential donors, typical commercial GaP wafers use sulfur (carrier concentration of 1.15
1017=cm3 and an electron mobility between 80 and 130 cm2=V=sec) or tellurium (carrier concentration of 1.7
1017=cm3 and an electron mobility between 100 and 140 cm2=V=sec). Wafer sizes are 50–80 mm. GaP also has a number of potential shallow-level acceptors (p-type dopant). These include germanium (substituted for P-activation energy of 0.265 eV), carbon (substituted for P-activation energy of 0.0543 eV), silicon (substituted for P-activation energy of 0.210 eV), beryllium (substituted for Ga-activation energy of 0.0566 eV), cadmium (substituted for Ga-activation energy of 0.1022 eV), magnesium (substituted for Gaactivation energy of 0.0599 eV), and zinc (substituted for Ga-activation energy of 0.0697 eV). Of these potential acceptors, typical commercial GaP wafers use zinc (carrier concentration of 1.2
1017=cm3 and a hole mobility between 50 and 80 cm2=V=sec). Wafer sizes are 50–80 mm. 3C–SiC 3C–SiC is the sole cubic polytype among the many SiC polytypes. It has the highest electron mobility of the SiC polytypes. Its band structure is indirect. Most of the commercial applications of SiC are with 4H–SiC and 6HSiC. 3C–SiC has a potential shallow-level donor (n-type dopant) with nitrogen (activation energy of 0.06–0.1 eV). 3C–SiC has several potential shallow-level acceptors (p-type dopant) with aluminum (activation energy of 0.26 eV), gallium (activation energy of 0.344 eV), and boron (activation energy 0.735 eV).

Table 2 Johnson figure of merit (JFM) of some wide band-gap semiconductors Material Si

Band-gap (eV) 1.1

Vb(V/cm) 5

3
10

7

Vsat(cm/sec) 7

1.0
10

7

JFM (V/sec)

Ratio to silicon

12

1.0

14

3
10

GaN

3.5

3
10

2.5
10

8.0
10

267

aSiC(6H)

3.0

3
107

2.0
107

6.0
1014

200

bSiC(4H)

3.3

7

3
10

7

2.5
10

14

8.0
10

267

Diamond

5.47

1
107–1
108

2.7
107

2.7
1014–27
1014

90–900

BN

6.1

1.2
107

3.1
107

3.72
1014

124

7

7

6.0
1014

200

AlN

6.2–6.4

2
10

3.0
10

445

Debye temperature (K)

1.12m0 0.22m0 0.79m0 0.14m0 3.8

Effective longitudinal electron mass (ml)

Effective transverse electron mass (mt)

Effective heavy hole mass (mh)

Effective light hole mass (mlp)

Electron affinity (eV) ˚) Lattice constant (A

5

Hardness Mohs scale 9.2

4.9

1.1

Thermal conductivity (W=cm= C)

50 2830

150 1457

Hole mobility (cm =V=sec)

Melting point ( C)

2

9.2

4.9

2830

40

380

3

1000

2

1.1

250

Breakdown voltage (MV=cm)

Electron mobility (cm2=V=sec)

3.0

2.26

2.26

104.2

Band-gap (eV)

102.8

A ¼ 3.0730, b ¼ 10.053

51

4.3596

0.42m0

Optical phonon energy (meV)

5.4505

0.20m0

0.68m0

9.11

Dielectric const (high frequency)

0.25m0

6.52

6.52

11.1 9.66

9.72

4.14

Dielectric constant (static)

3.21

1200

4 -P63mc C6v

Wurtzite

6H–SiC

Density (g=cm ) 3.166

1200

4.9  10

Number of atoms per cubic centimeter

3

Td2 -F43m

Td2 -F43m

Group of symmetry 22

Zinc blende (cubic)

3C–SiC

Zinc blende (cubic)

GaP

Crystal structure

Property

Table 3 Properties of some wide band-gap materials

9.2

4.9

2830

140

800

3

3.3

104.2

a ¼ 3.0730, b ¼ 10.053

0.42m0

0.29m0

6.52

9.66

1300

4 -P63mc C6v

Wurtzite

4H–SiC

1.3

2500

350

300

3

3.5

91.2

4

0.54

1977

80

3.37

10

20

4373

2000

2200

1–10

5.47

160

3.567

2.85

3273

14

300

1.2

6.2

99.2

A ¼ 3.11, c ¼ 4.98

0.6 a ¼ 3.189, c ¼ 5.186

3.53m0
0.070

3.53m0

0.4m0

4.84

9.14

3.255

1150

9.6  10

22

4 -P63mc C6v

Wurtzite

AlN

0.70m0

2.12m0

0.36m0

1.40m0

5.7

3.515

1860

1.7  10

23

O7h -Fd3m

Diamond

Diamond

4.1 a ¼ 4.75, c ¼ 2.92

3.75

8.75

5.642

4 -P63mc C6v

Wurtzite

ZnO

0.3m0

1.4m0

0.20m0

0.20m0

5.35

8.9

6.15

600

8.9  10

22

4 -P63mc C6v

Wurtzite

GaN

9.5

7.4

2973

500

200

2

6.2–6.4

130

3.6157

4.5

0.150m0

0.37m0

0.24m0

0.35m0

4.5

7.1

3.48

1700

Td2 -F43m

Zinc blende

BN

Wide Band-Gap Electronics Materials 3231

W

3232

6H–SiC 6H–SiC is an indirect band-gap material that is produced commercially. 6H–SiC along with 4H–SiC is used for short-wavelength optoelectronic, hightemperature, radiation-resistant, and high-power= high-frequency electronic devices. High-quality SiC is grown commercially by companies, such as CREE Research Inc., of Raleigh, NC, by bulk growth methods (4H and 6H structures) as well as by chemical vapor deposition (CVD). A p–n structure in SiC can be achieved by various methods. Some of these methods are sublimation epitaxy, container-free liquid phase epitaxy, CVD, ion implantation of aluminum, and boron diffusion. Many of the problems, which plagued SiC in the past, have been overcome. The micropipe defect density has been reduced to less than one per square centimeter. The crystal size is now being produced in 3 in. production wafers and large by CREE. 6H–SiC has potential shallow-level donors (n-type dopant) with nitrogen (activation energy of 0.085– 0.125 eV) and phosphorous (activation energy of 0.085 eV). 6H–SiC has several potential shallow-level acceptors (p-type dopant) with aluminum (activation energy of 0.239 eV), gallium (activation energy of 0.317 eV), beryllium (activation energy of 0.320 eV), and boron (activation energy of 0.31–0.38 eV).

4H–SiC 4H–SiC is an indirect band-gap material that is produced commercially. As stated previously, 4H–SiC along with 6H–SiC are used for short-wavelength optoelectronic, high-temperature, radiation-resistant, and high-power=high-frequency electronic devices. 4H–SiC has potential shallow-level donors (n-type dopant) with nitrogen (activation energy of 0.059– 0.102 eV), titanium (activation energy of 0.13 eV), and chromium (activation energy of 0.15–0.18 eV). 4H–SiC has several potential shallow-level acceptors (p-type dopant) with aluminum (activation energy of 0.19 eV), gallium (activation energy of 0.267 eV), and boron (activation energy of 0.647 eV).

GaN Gallium nitride is a direct band-gap material that is primarily used in green, blue, and UV LEDs and lasers, white-light sources, high-power RF and microwave sources, and high-temperature devices. Its band structure is direct. Some of the properties of gallium nitride are given in Table 3. GaN has been the extensively studied material for optoelectronic applications among all III–V nitrides. Blue LEDs and lasers made of GAN are

Wide Band-Gap Electronics Materials

obtained by overcoming constraints like doping and heteroepitaxial growth.[16,17] Gallium nitride (GaN) substrates are being grown by MOCVD, MBE, and HVPE. GaN is typically grown on sapphire (Al2O3), 6H–SiC, and ZnO. Most as-grown GaN (and InN) films exhibited high n-type conductivity because of native defects but p-type conductivity cannot be obtained.[16] p-Type GaN was achieved by doping with Mg, and GaN p–n homojunction and blue light emitting devices were demonstrated.[18,19] GaN LEDs are now being made commercially. The alloy of AlxGa1
xN has also been developed for blue to UV emitters.[18,19] However, only films with a small amount of Al (x  0.1 for p-type and x < 0.4) for n-type can be doped successfully.[18–20] GaN has a number of potential shallow-level donors (n-type dopant). An N vacancy is believed to be the dominant donor. The activation energy of the N vacancy is still under debate. Of these potential donors, typical commercial GaN wafers are naturally a donor because of the N vacancy. Wafer sizes are about 50 mm. Both magnesium and beryllium have been used to make p-type GaN; however, there is an unintentional incorporation of oxygen and silicon in the GaN growth process.[16,18,19]. p-Type materials are possible with beryllium or vanadium (in the Ga position with approximate activation energy of 0.236 eV), magnesium or vanadium (in the Ga position with approximate activation energy of 0.236 eV), and zinc or vanadium (in the Ga position with approximate activation energy of 0.232 eV). There are a number of deeplevel acceptors.

ZnO ZnO is a direct band-gap material that has gained interest for optoelectronic applications. It is an II–VI semiconductor, which gives it advantages in performance because II–VI semiconductors are not easily degraded through defects. ZnO is naturally an n-type material.[21] Its primary uses are for the polycrystalline form for piezoelectric transducers, varistors, phosphors, and transparent conducting films. Bulk crystals of ZnO have been fabricated by various methods including vapor phase growth, hydrothermal growth, and melt growth. Additionally, epitaxial growth of ZnO single-crystal films has been demonstrated by various techniques such as pulsed laser deposition, MBE, MOCVD, and HVPE. Acceptor levels in ZnO can be created with hydrogen and potentially sulfur and nitrogen.[22,23] A p–n junction in ZnO has been formed.[24] Various light emitting structures have been formed using ZnO including LEDs, and a UV nanowire laser.[25,26]

Wide Band-Gap Electronics Materials

Diamond Diamond is an indirect band-gap material that is used in grinding, polishing, cutting, wear resistance, tribology, acoustics, optics, thermal management, and electronic applications (see section on Diamond and Diamond-Like Materials). It has been used in electronic applications as a heat sink for large-scale integrated circuits. Recently, an 81 GHz high-frequency diamond device was made by Nippon Telegraph and Telephone Corp. (NTT).[27] This promises to make possible amplification e in the millimeter-wave band from 30 to 300 GHz. In addition, efforts by the Japanese government to initiate a joint research project with industry in fiscal year 2003 to develop diamond-based semiconductors indicates the progress that has been made in diamond electronics over the last decade.[28] Diamond, because of its superior properties is envisioned as an advanced chip technology that could one day replace silicon as the base for superfast, high-voltage semiconductors. There are two keys to diamond technology, the most challenging is making electronic grade high-quality synthetic diamond single crystals. NTT will focus its technology on a proven means of fabricating millimeter electronic single crystals. Other breakthroughs may lead to a source of large single crystalline diamond. A company called Genesis is producing large high-quality HPHT diamond crystals that may lead to a source of diamond wafers for the electronic industry. A company called Apollo is producing gem quality single-crystal diamond that may be economical for the diamond electronics business.[29] The second is in the formation of complex device structures. A fundamental device, a p–n junction, was fabricated in 1997 using a method called field enhanced diffusion with optical activation the details of which are given in Ref.[30]. Field enhanced diffusion with optical activation lends itself well to the fabrication of complex integrated circuits.[31] Donors for diamond are challenging. Lithium has been identified as a donor with an activation energy of 0.2 eV (searching did not reveal) using field enhanced diffusion with optical activation.[32] However, the lithium concentration was limited to about 5  1017 atoms=cm3. Other methods of introducing lithium into the diamond, such as in situ doping or ion implantation, did not produce n-type material owing to the defects introduced by the method. When dopant atoms are introduced into the semiconductor layers during its growth, most commonly during epitaxial growth of semiconductor layer, it is called in situ doping. Field enhanced diffusion with optical activation is a low-energy method that maintains the quality of the crystal and thus is able to limit the formation of defects. Another method of creating n-type conductivity has been reported. In this process, deuterating

3233

boron-doped diamond in a plasma at about 550 C converts the material from p-type to n-type with conductivities as high as 2 S=cm at room temperature.[33] Nitrogen is also a donor but it is deep level with activation energies of 1.7 and 4 eV, which may be due to substitutional impurity, simple aggregates, or platelets. Phosphorus is a donor with reported activation energies between 0.84 and 1.16 eV. Sulfur is a donor that has received a great deal of attention recently. Dr. Matt West demonstrated that diamond doped by the field enhanced diffusion with optical activation method has activation energy of 0.9 eV.[34] Boron is an acceptor in diamond with an activation energy of 0.37 eV. Natural type II b diamond is an example of a boron doped p-type diamond. There are many commercial sources of synthetic boron doped-diamond. Polycrystalline boron doped diamond films, made by CVD, have been made for electrode applications. High-pressure high-temperature diamonds with boron doping are also available.[35] AlN Aluminum nitride is direct band-gap material that has a very wide band-gap (6.2 eV). Also, it has a high thermal conductivity, high electrical resistivity, high acoustic velocity, high thermal stability, and high chemical resistance, and radiation stability. These properties make AlN suitable for UV optical devices, surface acoustic wave (SAW) devices, electrical insulators, or passive layers in microelectronics. Such a device can operate in a harsh environment with high temperatures and=or radiation. However, it is very difficult to dope AlN with impurities to make it to n- or p-type semiconductors. Also as grown AlN films do not show any n- or p-type characteristics. Although both n- and p-type doping of aluminum nitride have been reported as far back as 1967, almost no recent paper can be found that verifies these results.[36] Field enhanced diffusion with optical activation has been used for doping AlN and some promising results with magnesium as the p-type dopant and silicon as the n-type dopant.[37] Additional potential donors include C (Al), Ge (Al), and Se (N) while additional potential acceptors include Be (Al), C (N), Ca (Al), and Hg (Al). BN Despite the excellent properties of cubic boron nitride, it is a very difficult crystal to grow. Thus far the crystal size has been limited to micrometer-type size. Little is known about the ultimate potential of cubic boron nitride in electronics. Potential donors include Si with activation energy of 0.24 eV, C with activation energy of 0.28 eV, and sulfur with activation energy of

W

3234

0.41 eV.[38] A potential acceptor is Be with activation energy of 0.19 eV.[39]

APPLICATIONS There are many industries that utilize wide band-gap materials. The primary applications of wide band-gap materials are identified below.

Wide Band-Gap Electronics Materials

Insulators Because of the high breakdown voltage, wide band-gap material can and have been used as electrical insulators. There has been a great deal of interest in pressed nanophase SiC and diamond powders as an electrical insulator for capacitors.

Semiconductors

Grinding and polishing is one of the oldest applications for wide band-gap materials primarily owing to the property of hardness that some of these materials possess (e.g., diamond). SiC and cubic boron nitride, in addition to diamond, have found a commercial market in grinding and polishing, primarily for ferromagnetic materials with high carbon solubility.[39,40]

All of the wide band-gap materials have potential applications as semiconductors, ranging from hightemperature, high-power, and high-radiation resistance applications to high-speed integrated circuits to highfrequency devices. Recent research has shown that wide band-gap materials like ilmenite-hematite (IH) can be used for low-voltage varistor application for space research.[47] Semiconductor technology based on wide band-gap material composites has been widely used for MEMS devices.[48]

Tool and Die

Optoelectronic Devices

Because of the hardness of band-gap materials such as diamond, SiC, and boron nitride, wide band-gap materials have found applications in the tool and die industry.[40,41].Wide band-gap materials have been successfully used for coating field emission tips in SEM with results showing reduction in turn-on voltage by 100 V and more uniform emission during low-voltage operation.[42]

Wide band-gap semiconductors have been used for green, blue, and UV LEDs and lasers, and white-light sources.

Grinding and Polishing

Electronics One of the major markets for wide band-gap materials is in electronics. Specifically, they are suitable for and have been used for heat sinks (diamond), short wavelength optoelectronic devices (GaP, GaN, SiC), hightemperature electronics (SiC, GaN), radiation resistant devices, and high-power=high-frequency electronic devices (diamond, GaN, SiC).[40,43–45] Recent research showed that Mn-doped GaN can be used for spintronic applications.[46] Atomically flat technology developed by NASA for SiC and GaN WBG material can introduce a new dimension of application for WBG materials.

Heat Sinks Because of diamond’s exceptional heat conductivity (about five times that of copper), it has been used as a heat sink for large-scale integrated circuits.[40]

CONCLUSIONS As the wide band-gap electronic field matures, a wide variety of optoelectronic devices, high-temperature devices (for automotives, combustion chambers, smoke stacks, furnaces, etc.: 150 C to >600 C), radiation hardened devices (for satellites, nuclear power plants, etc.), high-power devices (high-voltage switching, power transistors, etc.), high-speed integrated circuits (LSI, CPU, etc.), and high-frequency devices (microwave amplification, communication, etc.) will be developed. Wide band-gap materials are positioned well to play an important role in the vibrant electronics industry over the next several decades.

REFERENCES 1. Isamu, A. Renaissance and progress in nitride semiconductors—my personal hisory of nitride research. Materials Research Society Symposium Proceedings, 2001; Vol. 639. 2. Johnson, W.C.; Parsons, J.B. Nitrogen compounds of gallium. I, II. J. Phys. Chem. 1932, 36, 2588.

Wide Band-Gap Electronics Materials

3. Grimmeiss, H.G.; Koelmans, H. Emission near the absorption edge and other emission effects of GaN. Zeitschrift fuer Naturforschung 1959, 14a, 264–271. 4. Maruska, H.P.; Tietjen, J.J. The preparation and properties of vapor-deposited single-crystal-line GaN. Appl. Phys. Lett. 1969, 15, 327. 5. Dingle, R.; Shaklee, K.L.; Leheny, R.F.; Zetterstrom, R.B. Stimulated emission and laser action in gallium nitride. Appl. Phys. Lett. 1971, 19, 5. 6. Pankove, J.I.; Miller, E.A.; Richman, D.; Berkeyheiser, J.E. Electroluminescence in GaN. J. Lumin. 1971, 4, 63. 7. Ejder, E. Refractive index of gallium nitride. Physica Status Solidi A: Applied Research 1971, 6 (2), 445–448. 8. Monemar, B. Fundamental energy gap of GaN from photoluminescence excitation spectra. Phys. Rev. 1974, B10, 676. 9. Jacob, G.; Boulou, M.; Bois, D. GaN electroluminescent devices: Preparation and studies. J. Lumin. 1978, 17, 263. 10. Mohammad, S.N.; Morkoc, H. Progress and prospects of Group III nitride semicondeuctors. Prog. Quant. Electr. 1996, 20 (5=6), 361–325. 11. Nishinaga, T.; Naritsuka, S. Epitaxial Growth of III–V Compounds. Department of Material Science and Engineering, Meijo University, Japan, 2003; 1–501. 12. Kressel, H.; Dunse, J.U.; Nelson, H.; Hawrylo, F.Z. Luminescence in silicon-doped gallium arsenide grown by liquid-phase epitaxy. J. Appl. Phys. 1968, 39 (4), 2006–2011. 13. Mimura, T.; Hiyamizu, S.; Fujii, T.; Nanbu, K. A new doped GaAs=n-AlxGa1-x as heterojunctions. Jpn. J. Appl. Phys. 1980, 19L, 225. 14. Dingle, R.; Stormer, H.L.; Gossard, A.C.; Wiegmann, W. Electron mobilities in modulation-doped semiconductor heterojunction superlattices. Appl. Phys. Lett. 1978, 33, 665. 15. Amano, H.; Sawaki, N.; Akasaki, I.; Toyoda, Y. Metalorganic vapor phase epitaxial growth of a high quality GaN film using an AlN buffer layer. Appl. Phys. Lett. 1986, 48, 353. 16. Strite, S.; Morkoc, H. GaN, AlN, and InN: a review. J. Vac. Sci. Technol. B 1992, 10 (4), 1237–1266. 17. Davis, R.F. III-V nitrides for electronic and optoelectronic applications. Proc. IEEE 1991, 79 (5), 702–712. 18. Akasaki, I.; Amano, H. Perspective of the UV= blue light emitting devices based on GaN and related compounds. Optoelectron. Devices Technol. 1992, 7 (1), 49–56. 19. Tanaka, T.; Watanabe, A.; Amano, H.; Kobayashi, Y.; Akasaki, I.; Yamazaki, S.; Koike, M. p-Type

3235

20.

21.

22.

23.

24.

25. 26.

27. 28. 29. 30.

31.

32.

33. 34.

35.

conduction in Mg-doped GaN and Al0.08Ga0.92N grown by metalorganic vapor phase epitaxy. Appl. Phys. Lett. 1994, 65 (5), 593–594. Yoshida, S.; Misawa, S.; Gonda, S. Properties of Al=x=Ga=1-x=N films prepared by reactive molecular beam epitaxy. J. Appl. Phys. 1982, 53 (10), 6844–6848. Jayaraj, M.K; Aldrin, A.; Manoj, R. Transparent conducting zinc oxide thin film prepared by off-axis rf magnetron sputtering. Bull. Mater. Sci. 2002, 25 (3), 227–230. Chris, G.; Van de Walle. Hydrogen as a cause of doping in zinc oxide. Phys. Rev. Lett. 2000, 85 (5), 1012–1015. Cruz-Va´zquez, C.; Rocha-Alonzo, F.; BurruelIbarra, S.E.; Inoue, M.; Bernal, R. Fabrication and characterization of sulfur doped zinc oxide thin films. Superficies y Vacı´o 13, Dec 2001; Sociedad Mexicana de Ciencia de Superficies y de Vacı´o, 2001; 88–91. Ohta, H.; Kawamura, K.; Orita, M.; Hirano, M.; Sarukura, N.; Hosono, H. Current injection emission from a transparent p-n junction composed of p-SrCu2O2=n-ZnO. Appl. Phys. Lett. 2000, 77 (4), 475–477. Look, D.C. Recent advances in ZnO materials and devices. Mater. Sci. Eng. 2001, B80, 383–387. Huang, M.H.; Mao, S.; Feick, H.; Yan, H.; Wu, Y.; Kind, H.; Weber, E.; Russo, R.; Yang, P. Roomtemperature ultraviolet nanowire nanolasers. Science 2001, 292, 1897–1899. Hara, Y. NTT verifies diamond semiconductor operation at 81 GHz. EE Times 2003, Aug 26. Miyake, K. Japanese government to launch diamond chip project. Info World 2002, Dec 27. http:==www.wired.com=wired=archive=11.09= diamond_pr.html. Popovici, G.; Melnikov, A.; Varichenko, V.V.; Sung, T.; Prelas, M.A.; Wilson, R.G.; Loyalka, S.K. Diamond ultraviolet photovoltaic cell obtained by lithium and boron doping. J. Appl. Phys. 1997, 81 (5), 2429–2431. Popovici, G.; Sung, T.; Prelas, M. Forced diffusion in diamond: a review. J. Chem. Vap. Deposition 1994, 3, 115–132. Sung, T. Doping Diamond by Forced Diffusion. Ph.D. thesis, University of Missouri, Columbia, May, 1996. High-conductivity n-type diamond demonstrated, photonics spectra. Presstime Bull. 2003, Aug. Matthew, W. Diffusion of Sulfur into Natural Diamond: Characterization and Applications in Radiation Detection. University of Missouri– Columbia, Dec 1999. Ekimov, E.A.; Sidorov, V.A.; Bauer, E.D.; Mel0 nik, N.N.; Curro, N.J.; Thompson, J.D.; Stishov, S.M.

W

3236

36.

37.

38. 39.

40.

41.

42.

43.

Wide Band-Gap Electronics Materials

Superconductivity in Diamond; Vereshchagin Institute for High Pressure Physics, Russian Academy of Sciences: Troitsk, Russia. Los Alamos National Laboratory, Condensed Matter, 2004; 1–13. Chu, T.L.; Ing, D.W.; Noreika, A.J. Epitaxial growth of aluminum nitride. Solid-State Electron. 1967, 10, 1023. Prelas, M.; Ghosh, T.; Tompson, R. Direct Conversion of Radioisotope Energy to Electricity. Final Report for DOE contract, DE FG07001D13927, August 2003. http:==www.ioffe.rssi.ru=SVA=NSM=Semicond= Semicond= BN=bandstr.html#Donors. Subramanian, K.; Shanbhag, V.R. Abrasive applications of diamond. In Handbook of Industrial Diamonds and Diamond Films; Marcel Dekker: 1998; 1023–1042. Prelas, M.; Popovici, G.; Bigelow, K. Handbook of Industrial Diamond and Diamond Films; Marcel Dekker: 1998; 1023–1042. Hay, R.A.; Galimberri, J.M. Cutting and wear applications. In Handbook of Industrial Diamonds and Diamond Films; Marcel Dekker: 1998; 1135– 1147. Graf, T.; Goennenwein, S.T.B.; Brandt, M.S. Prospects for carrier-mediated ferromagnetism in GaN. Walter Schottky Institut, Technische Universitaet Muenchen: Garching, Germany. Phys. Status Solidi B: Basic Res. 2003, 239 (2), 277–290. Dreifus, D.L.; Fox, B.A. Active devices. In Handbook of Industrial Diamonds and Diamond Films; Marcel Dekker: 1998; 1043–1072.

44. Eden, R.C. Applications of diamond in computers. In Handbook of Industrial Diamonds and Diamond Films; Marcel Dekker: 1998; 1073– 1102. 45. Brandes, G.R. Diamond vacuum electronics. In Handbook of Industrial Diamonds and Diamond Films; Marcel Dekker: 1998; 1103–1127. 46. Hajra, M.; Chubun, N.N.; Chakhovskoi, A.G.; Hunt, C.E.; Liu, K.; Murali, A.; Risbud, S.H.; Tyler, T.; Zhirnov, V. Field emission characterization of silicon tip arrays coated with GaN and diamond nanoparticle clusters. Electrical and Computer Engineering Department, University of California, Davis, CA. J. Vacuum Sci. Technol. B: Microelectron. Nanometer Struct.—Process. Meas. Phenom. 2003, 21 (1), 458–463. 47. Padmini, P.; Pulikkathara, M.; Wilkins, R.; Pandey, R.K. Neutron radiation effects on the nonlinear current–voltage characteristics of ilmenite-hematite ceramics. Department of Electrical and Computer Engineering, The University of Alabama, Tuscaloosa, AL. Appl. Phys. Lett. 2003, 82 (4), 586–588. 48. Luchinin, V.V.; Korlyakov, A.V.; Vasilev, A.A.; Jandjgava, G.I.; Prosorov, S.V.; Solomatin, A.K.; Sorokin, A.V.; Kucherkov, S.G.; Severov, L.A.; Ponomarev, V.K. SiC-AlN-compositionbased MEMS. Proceedings of SPIE—The International Society for Optical Engineering, 1999; Microtechnology Center, St. Petersburg State Electrotechnical University: St. Petersburg, Russia. Indo-Russian Workshop on Micromechanical Systems); 141–145.

Zeolite Membranes Z Yushan Yan Zijian Li Shuang Li Christopher Lew Department of Chemical and Environmental Engineering, University of California at Riverside, Riverside, California, U.S.A.

INTRODUCTION Zeolites and zeolite-type materials (hereafter referred to as zeolites for simplicity) are a class of crystalline oxides with uniform channels and cages ranging from 0.3 to 2 nm. Structurally, a zeolite consists of 3-dimensionally linked tetrahedra and each tetrahedron has one tetrahedrally coordinated atom (hereafter referred to as T-atom) (e.g., Si, Al, P, B, Ga, Ge, Zn, Be, etc.) at the center and four oxygen atoms at the corners. Each corner oxygen atom is shared by two neighboring tetrahedra. Depending on the T-atom, the zeolite framework can be charged or neutral. When the framework is charged, balancing ions are needed, and this is the origin of the ion exchange capability of zeolites. According to their structural symmetry and topology, zeolites are classified into different framework types. Each framework type is assigned a unique three-letter code (e.g., MFI) by the International Zeolite Association (IZA). These structure codes are generally from the zeolite names (e.g., MFI from Mobil Five) and do not follow any sort of established naming procedure. More than 150 framework types with numerous compositional variations have been verified and approved by the IZA. Zeolite materials are used commercially as shape= size selective catalysts in the petrochemical and petroleum refining industry, and as molecular sieving separation media for gases and hydrocarbons. For both applications, zeolites are used in powder composite form such as pellets and granules. In this entry, we focus on zeolite membranes. We define zeolite membranes as a continuous phase of zeolite-based materials (pure zeolite or composite) that separate two spaces. Zeolite membranes are generally uniform thin films attached to a porous or a nonporous substrate. They can also be self-standing without a substrate. Note that we have included zeolite films and layers on nonporous substrate in this entry because we believe many of the synthesis strategies and applications reported for those nonporous substrates are easily transferred to a porous substrate to prepare a zeolite membrane. Encyclopedia of Chemical Processing DOI: 10.1081/E-ECHP-120025023 Copyright # 2006 by Taylor & Francis. All rights reserved.

In this entry, we briefly discuss the types of the zeolite membranes and then focus on new applications that have been demonstrated recently. New developments are also included and analyzed, which are followed by some concluding remarks and future directions.

TYPES OF ZEOLITE MEMBRANES Polycrystalline Zeolite Membranes Polycrystalline zeolite membranes consist of intergrown zeolite crystals with no apparent cracks or pinholes[1] (Fig. 1A). These films are composed of only zeolite (i.e., there are no non-zeolite components such as amorphous silica or polymer). They are normally supported on a substrate although free-standing films have also been synthesized.[2] Membranes can be prepared on different substrates such as silicon wafer, quartz, porous alumina, carbon, glass, stainless steel (SS), gold, etc. Polycrystalline films are primarily prepared by hydrothermal synthesis methods including in situ crystallization,[1] seeded growth,[3] and vapor transport,[4] and have potential use in all of the applications discussed in this entry. Zeolite Matrix Composite Membrane A zeolite matrix composite membrane is defined as a membrane in which zeolite crystals are imbedded in a solid matrix that is either inorganic (e.g., silica and carbon)[5–8] or organic (e.g., polymer)[9,10] (Fig. 1B). Inorganic matrix membranes are often supported, while organic polymer matrices are normally freestanding. The selection between an inorganic and an organic phase as the imbedding matrix strongly depends on the intended application. For example, an inorganic matrix is ideal for applications where thermal stability is required,[5] whereas an organic polymer matrix is the better choice when flexibility is desired.[9,10] These composite films are usually 3237

3238

Zeolite Membranes

Fig. 1 Schematic of the three types of zeolite membranes: (A) a polycrystalline zeolite membrane; (B) a zeolite matrix composite membrane; and, (C) a zeolite crystal layer.

prepared by wash-coating, sol-gel processing, and polymer film casting. Zeolite Crystal Layer Another form of zeolite membranes is a zeolite crystal layer that consists of isolated crystals deposited on a solid substrate[11] (Fig. 1C). The substrate can be a variety of materials such as metal, ceramic, or silicon wafer. Crystal layers have to be supported. There has been exciting fundamental research carried out in this area, however, demonstrated applications have been limited to sensors. The organic linker approach appears very promising for the preparation of these types of membranes.[12]

APPLICATIONS OF ZEOLITE MEMBRANES Zeolite membranes have been demonstrated for many applications. Applications such as separation membranes, membrane reactors, adsorption, and catalysis have been covered in several reviews.[13–16] In this entry, we focus on new applications including sensors, low-dielectric constant (low-k) films, corrosion resistant coatings, hydrophilic coatings, heat pumps, and thermoelectrics.

To achieve selective adsorption and quick response, the zeolite film is required to be continuous (i.e., free of non-zeolitic pores) and thin, and have proper orientation. For facile mass transport, a monocrystal film is also preferred because there are no grain boundaries in the mass transport direction, which have proved to decrease diffusion in zeolite by orders of magnitude. Strong adhesion between the zeolite film and the sensor surface is also critical for the durability of the sensor. Recently, a b-oriented continuous pure-silica-zeolite (PSZ) MFI monolayer film was successfully synthesized on SS and silicon wafers using a very facile direct in situ crystallization method.[17–19] As-synthesized b-oriented films have good continuity and thin monocrystal layer thickness. The thin film thickness and preferred crystal orientation can reach a fast adsorption and desorption equilibrium, which can thus increase the stability of the sensor. In addition, the superior adhesion of this film to the substrate ensures high durability. All these features make this type of film a better candidate as sensor coating to achieve higher selectivity, sensitivity, stability, and durability.[11] The molecular sieving property of this continuous b-oriented PSZ MFI monocrystal film was demonstrated in a zeolite modified electrode (ZME) configuration using redox molecules of different sizes (Fig. 2).[20] Ru(NH3)63þ (diameter ˚ ) was able to traverse the film, while Co(phen)32þ
5.5 A ˚ ) was completely excluded. This film (diameter
13.0 A has excellent adhesion to the electrode and is stable in strong acidic conditions.

Sensors Chemical sensors are important for industrial process control and environmental monitoring. They also find critical use in medical and defense applications. Two of the critical performance parameters of a sensor are selectivity and sensitivity. When integrated with an appropriate sensor platform (e.g., quartz crystal microbalance or electrochemical quartz crystal microbalance), a zeolite film can improve the selectivity and sensitivity of the sensor due to its ability to selectively adsorb a component out of a mixture. A large number of publications have appeared in this area, and an in-depth review on zeolite films for chemical sensor applications up to 1996 was provided as a section in a recent review by Bein.[11] Here, we highlight a recent development of selective chemical sensor based on continuous oriented zeolite films.

Low-k Dielectrics Low-k (k for dielectric constant) dielectric materials have been identified as one of the most difficult challenges for interconnects in future generation integrated circuits (ICs). Many materials have been proposed, studied, or are under commercial development as potential candidates for low-k dielectrics. Among these materials, two major classes are dense organic polymers and porous inorganic based materials. It has been shown that some dense, organic polymers could easily have a k value between 2 and 3, but there are concerns about their low thermal stability and low heat conductivity. Also, due to their low mechanical strength, polymeric materials may have potential problems with the chemical and mechanical polishing (CMP) process.

Zeolite Membranes

Fig. 2 (A) SEM top view of b-oriented pure-silica-zeolite (PSZ) MFI monocrystal-thick film; (B) cross-sectional view of the sample in (A) after slight polishing; (C) schematic illustration of molecular sieving in b-oriented PSZ MFI monocrystal-thick film. A distance was intentionally kept between the MFI film and the electrode so that the sieving behavior can be illustrated more clearly. (From Ref.[20].)

Current dielectric material is dense silica that has a k value of 4. Porous silica has a lower k because of the incorporation of air. Sol-gel silica and mesoporous silica have been studied as possible low-k materials. However, these materials have drawbacks such as low mechanical strength and low heat conductivity. Sol-gel silica also has randomly occurring large pores that can cause electrical breakdown. High mechanical strength is needed for the new low-k materials to be compatible with the CMP process. Very recently, zeolites have been demonstrated to be a promising low-k material (Fig. 3). Zeolites have very uniform pores and their pore size falls into the micropore range (