Handbook of Membrane Reactors. Volume 1 Reactor Types and Industrial Applications [1 ed.] 978-0-85709-415-5

Citation preview

Handbook of membrane reactors

© Woodhead Publishing Limited, 2013

Related titles: Handbook of membrane reactors, Volume 1 (ISBN 978-0-85709-414-8) Advanced membrane science and technology for sustainable energy and environmental applications (ISBN 978-1-84569-969-7) Functional materials for sustainable energy applications (ISBN 978-0-85709-059-1) Details of these books and a complete list of titles from Woodhead Publishing can be obtained by: •

visiting our web site at www.woodheadpublishing.com

•

contacting Customer Services (e-mail: [email protected]; fax: +44 (0) 1223 832819; tel.: +44 (0) 1223 499140 ext. 130; address: Woodhead Publishing Limited, 80, High Street, Sawston, Cambridge CB22 3HJ, UK)

•

in North America, contacting our US office (e-mail: usmarketing@ woodheadpublishing.com; tel.: (215) 928 9112; address: Woodhead Publishing, 1518 Walnut Street, Suite 1100, Philadelphia, PA 19102-3406, USA)

If you would like e-versions of our content, please visit our online platform: www. woodheadpublishingonline.com. Please recommend it to your librarian so that everyone in your institution can benefit from the wealth of content on the site. We are always happy to receive suggestions for new books from potential editors. To enquire about contributing to our energy series, please send your name, contact address and details of the topic/s you are interested in to Sarah.Hughes@ woodheadpublishing.com. We look forward to hearing from you.

The Woodhead team responsible for publishing this book: Commissioning Editor: Sarah Hughes Publications Coordinator: Adam Davies Project Editor: Rachel Cox Editorial and Production Manager: Mary Campbell Production Editor: Mandy Kingsmill Project Manager: Vedhapriya Badrinarayanan, Newgen Knowledge Works Pvt Ltd Freelance Copyeditor: Dick Hill Proof reader: Suma George, Newgen Knowledge Works Pvt Ltd Cover designer: Terry Callanan

© Woodhead Publishing Limited, 2013

Woodhead Publishing Series in Energy: Number 56

Handbook of membrane reactors Volume 2: Reactor types and industrial applications Edited by Angelo Basile

Oxford

Cambridge

Philadelphia

New Delhi

© Woodhead Publishing Limited, 2013

Published by Woodhead Publishing Limited, 80 High Street, Sawston, Cambridge CB22 3HJ, UK www.woodheadpublishing.com www.woodheadpublishingonline.com Woodhead Publishing, 1518 Walnut Street, Suite 1100, Philadelphia, PA 19102-3406, USA Woodhead Publishing India Private Limited, G-2, Vardaan House, 7/28 Ansari Road, Daryaganj, New Delhi - 110002, India www.woodheadpublishingindia.com First published 2013, Woodhead Publishing Limited © Woodhead Publishing Limited, 2013. Note: the publisher has made every effort to ensure that permission for copyright material has been obtained by authors wishing to use such material. The authors and the publisher will be glad to hear from any copyright holder it has not been possible to contact. The authors have asserted their moral rights. This book contains information obtained from authentic and highly regarded sources. Reprinted material is quoted with permission, and sources are indicated. Reasonable efforts have been made to publish reliable data and information, but the authors and the publisher cannot assume responsibility for the validity of all materials. Neither the authors nor the publisher, nor anyone else associated with this publication, shall be liable for any loss, damage or liability directly or indirectly caused or alleged to be caused by this book. Neither this book nor any part may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, microfilming and recording, or by any information storage or retrieval system, without permission in writing from Woodhead Publishing Limited. The consent of Woodhead Publishing Limited does not extend to copying for general distribution, for promotion, for creating new works, or for resale. Specific permission must be obtained in writing from Woodhead Publishing Limited for such copying. Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation, without intent to infringe. British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library. Library of Congress Control Number: 2013930581 ISBN 978-0-85709-415-5 (print) ISBN 978-0-85709-734-7 (online) ISSN 2044–9364 Woodhead Publishing Series in Energy (print) ISSN 2044–9372 Woodhead Publishing Series in Energy (online) The publisher’s policy is to use permanent paper from mills that operate a sustainable forestry policy, and which has been manufactured from pulp which is processed using acid-free and elemental chlorine-free practices. Furthermore, the publisher ensures that the text paper and cover board used have met acceptable environmental accreditation standards. Typeset by Newgen Knowledge Works Pvt Ltd, India Printed by MPG Printgroup, UK

© Woodhead Publishing Limited, 2013

Contributor contact details

(* = main contact)

Chapter 2

Editor

Annarosa Gugliuzza* and Prof. Angelo Basile Institute on Membrane Technology ITM-CNR c/o University of Calabria Via P. Bucci 17/C 87036 Rende (Cs) Italy

Prof. Angelo Basile Institute on Membrane Technology ITM-CNR c/o University of Calabria Via P. Bucci 17/C 87036 Rende (Cs) Italy Email: [email protected]

Email: [email protected]. it; [email protected]; [email protected]

and

and

AST Engineering spa via Adolfo Ravà 30 00142 Rome Italy

Prof. Angelo Basile AST Engineering spa via Adolfo Ravà 30 00142 Rome Italy

Chapter 1 and Chapter 23 V. Calabrò Department of Engineering Modeling University of Calabria Via P. Bucci 39/C 87036 Rende (Cs) Italy Email: [email protected]

Chapter 3 G. Camera-Roda Dipartimento di Ingegneria chimica mineraria e delle Tecnologie ambientali University of Bologna Via Terracini 28 40131 Bologna Italy Email: giovanni.cameraroda@ unibo.it xv

© Woodhead Publishing Limited, 2013

xvi

Contributor contact details

Vincenzo Augugliaro, Vittorio Loddo and Leonardo Palmisano ‘Schiavello-Grillone’ Photocatalysis Group Dipartimento di Ingegneria Elettrica, Elettronica e delle Telecomunicazioni Università degli Studi di Palermo Viale delle Scienze 90128 Palermo Italy Email: vincenzo.augugliaro@ unipa.it; [email protected]; [email protected]

Chapter 4 Dr Antonio Comite, A. Bottino, G. Capannelli, C. Costa and R. Di Felice Department of Chemistry and Industrial Chemistry University of Genoa Via Dodecaneso 31 16146 Genoa Italy Email: [email protected]

Chapter 5 Prem Kumar Seelam*, Mika Huuhtanen and Riitta L. Keiski Mass and Heat Transfer Process Laboratory Department of Process and Environmental Engineering P.O. Box 4300 University of Oulu FI-90014 Finland

Chapter 6 Sylwia Mozia* and Antoni W. Morawski Institute of Chemical and Environment Engineering West Pomeranian University of Technology, Szczecin ul. Pułaskiego 10 70–322 Szczecin Poland Email: [email protected]; [email protected] Raffaele Molinari Department of Chemical Engineering and Materials University of Calabria Via P. Bucci 44/A 87036 Rende (CS) Italy Email: [email protected] Leonardo Palmisano and Vittorio Loddo ‘Schiavello-Grillone’ Photocatalysis Group Dipartimento di Ingegneria Elettrica, Elettronica e delle Telecomunicazioni Università degli Studi di Palermo Viale delle Scienze 90128 Palermo Italy Email: leonardo.palmisano@unipa. it; [email protected]

E-mail: [email protected]

© Woodhead Publishing Limited, 2013

Contributor contact details

xvii

Chapter 7

Chapter 9

Alfredo Cassano* and Prof. Angelo Basile Institute on Membrane Technology ITM-CNR c/o University of Calabria Via P. Bucci 17/C 87036 Rende (Cs) Italy

Pierre Millet Institut de Chimie Moléculaire et des Matériaux d’Orsay UMR 8182 Université Paris sud Centre d’Orsay Bâtiment 410 91405 Orsay Cedex France

Email: [email protected]; [email protected] and

Email: [email protected]

Chapter 10

Prof. Angelo Basile AST Engineering spa via Adolfo Ravà 30 00142 Rome Italy

Chapter 8 Xiaoyao Tan Department of Chemical Engineering Tianjin Polytechnic University Tianjin 300160 China Kang Li* Department of Chemical Engineering and Technology Imperial College London South Kensington London SW7 2AZ UK Email: [email protected]

Paolo Chiesa* and Matteo C. Romano Department of Energy Politecnico di Milano via Lambruschini, 4 20156 Milan Italy Email: [email protected] Thomas G. Kreutz Princeton Environmental Institute Princeton University 25 Guyot Hall Princeton, NJ 08544 New Jersey 08544 USA

Chapter 11 Marcello De Falco* Faculty of Engineering University Campus Bio-Medico of Rome via Alvaro del Portillo 21 00128 Rome Italy Email: [email protected]

© Woodhead Publishing Limited, 2013

xviii

Contributor contact details

G. Iaquaniello and E. Palo Tecnimont KT S.p.A. Viale Castello della Magliana 75 00148 Rome Italy Email: G.Iaquaniello@tecnimontkt. it; [email protected] B. Cucchiella Processi Innovativi s.r.l. Viale Castello della Magliana 75 00148 Rome Italy Email: Cucchiella.B@ processiinnovativi.it

1-5-20 Kaigan, Minato-ku Tokyo 105–8527 Japan Email: [email protected]

Chapter 13 M. De Falco Faculty of Engineering University Campus Bio-Medico of Rome Via Alvaro del Portillo 21 00128 Rome Italy

V. Palma and P. Ciambelli Dipartimento di Ingegneria Industriale University of Salerno via Ponte Don Melillo 84084 Fisciano (SA) Italy

G. Iaquaniello* and E. Palo Tecnimont KT S.p.A. Viale Castello della Magliana 75 00148 Rome Italy

Email: [email protected]; [email protected]

A. Salladini Processi Innovativi S.r.l. Viale Castello della Magliana 7500148 Rome Italy

Chapter 12 Mr Yoshinori Shirasaki* Fuel cell and hydrogen project section R & D Department The Japan Gas Association, Tokyo 1-15-12, Toranomon, Minato-ku Tokyo, 105-0001 Japan Email: [email protected] Dr Isamu Yasuda Technology Planning Department Tokyo Gas Co., Ltd

Email: G.Iaquaniello@tecnimontkt. it; [email protected]

Email: Salladini.a@ processiinnovativi.it

Chapter 14 Giampaolo Manzolini* Politecnico di Milano Dipartimento di Energia via Lambruschini 4 20156 Milano Italy Email: giampaolo.manzolini@ polimi.it

© Woodhead Publishing Limited, 2013

Contributor contact details Daniel Jansen Energy Research Centre of the Netherlands Westerduinweg 3 1755 LE Petten The Netherlands

Silvano Tosti* ENEA Unità Tecnica Fusione C.R. Frascati Via E. Fermi 45 00044 Frascati (RM) Italy

Email: [email protected]

Email: [email protected]

Chapter 15

Chapter 17

Prof. Michel Cassir*, Dr Armelle Ringuedé and Dr Virginie Lair Chimie ParisTech ENSCP UMR CNRS 7575 Laboratory of Electrochemistry, Chemistry of Interfaces and Modeling for Energy 11 rue Pierre et Marie Curie F-75231 Paris Cedex 05 France

Ayfer Veziroglu* Chief Researcher, International Association for Hydrogen Energy 5794 SW 40 Street # 303 Miami FL 33155 USA

Email: michel-cassir@ens. chimie-paristech.fr Dr Deborah Jones Institut Charles Gerhardt UMR CNRS 5253 Laboratoire Agrégats, Interace et Matériaux pour l’Energie Université de Montpellier 2 Montpellier Cedex 5 France

Chapter 16 Alfonso Pozio ENEA Unità Tecnica Fonti Rinnovabili C.R. Casaccia Via Anguillarese 301 00123 S. Maria di Galeria (RM) Italy

xix

Email: [email protected] Rosario Macario Instituto Superior Técnico Universidade Técnica de Lisboa Av. Rovisco Pais 1 1049–001 Lisboa Portugal Email: [email protected]

Chapter 18 Pierre Millet* Institut de Chimie Moléculaire et des Matériaux d’Orsay UMR 8182 Université Paris sud Centre d’Orsay Bâtiment 410 91405 Orsay Cedex France Email: [email protected]

Email: [email protected] © Woodhead Publishing Limited, 2013

xx

Contributor contact details

Prof. Angelo Basile Institute on Membrane Technology ITM-CNR c/o University of Calabria Via P. Bucci 17/C 87036 Rende (Cs) Italy Email: [email protected] and

Chapter 20 Faisal I. Hai* and Long D. Nghiem Strategic Water Infrastructure Laboratory School of Civil, Mining and Environmental Engineering The University of Wollongong, Wollongong NSW 2522 Australia Email: [email protected]; longn@ uow.edu.au

AST Engineering spa via Adolfo Ravà 30 00142 Rome Italy

Chapter 19 Alfredo Cassano*, Alberto Figoli and Francesco Galiano Institute on Membrane Technology, ITM-CNR c/o University of Calabria Via P. Bucci 17/C 87036 Rende (Cs) Italy Email: [email protected] Pietro Argurio and Raffaele Molinari Department of Chemical Engineering and Materials University of Calabria Via P. Bucci 44/A 87036 Rende (Cs) Italy Email: [email protected]; [email protected]

Oskar Modin Water Environment Technology section Department of Civil and Environmental Engineering Chalmers University of Technology Göteborg Sweden Email: [email protected]

Chapter 21 Raffaele Molinari* Department of Chemical Engineering and Materials University of Calabria Via P. Bucci 44/A 87036 Rende (Cs) Italy Email: [email protected] Leonardo Palmisano and Vittorio Loddo ‘Schiavello-Grillone’ Photocatalysis Group Dipartimento di Ingegneria Elettrica, Elettronica e delle Telecomunicazioni

© Woodhead Publishing Limited, 2013

Contributor contact details Università degli Studi di Palermo Viale delle Scienze 90128 Palermo Italy Email: leonardo.palmisano@unipa. it; [email protected] Sylwia Mozia and Antoni W. Morawski Institute of Chemical and Environment Engineering West Pomeranian University of Technology, Szczecin ul. Pułaskiego 10 70–322 Szczecin Poland Email: [email protected]; [email protected]

xxi

Sudip Chakraborty Department of Chemical Engineering and Materials University of Calabria Cubo-44a Via P. Bucci 17/C 87036 Rende (Cs) Italy Email: [email protected] Vincenzo Piemonte and Luisa Di Paola University Campus Bio-Medico of Rome Faculty of Engineering Via Alvaro del Portillo 00128 Rome Italy Email: [email protected]; [email protected]

Chapter 22 Tadashi Uragami* Department of Chemistry and Materials Engineering Kansai University 3-3-35 Yamate-cho, Suita Osaka 564–8680 Japan Email: [email protected]. ac.jp; [email protected]

Chapter 23 G. Iorio Department of Engineering Modeling University of Calabria Via P. Bucci 39/C 87036 Rende (Cs) Italy Email: [email protected]

© Woodhead Publishing Limited, 2013

Woodhead Publishing Series in Energy

1

Generating power at high efficiency: Combined cycle technology for sustainable energy production Eric Jeffs 2 Advanced separation techniques for nuclear fuel reprocessing and radioactive waste treatment Edited by Kenneth L. Nash and Gregg J. Lumetta 3 Bioalcohol production: Biochemical conversion of lignocellulosic biomass Edited by K. W. Waldron 4 Understanding and mitigating ageing in nuclear power plants: Materials and operational aspects of plant life management (PLiM) Edited by Philip G. Tipping 5 Advanced power plant materials, design and technology Edited by Dermot Roddy 6 Stand-alone and hybrid wind energy systems: Technology, energy storage and applications Edited by J. K. Kaldellis 7 Biodiesel science and technology: From soil to oil Jan C. J. Bart, Natale Palmeri and Stefano Cavallaro 8 Developments and innovation in carbon dioxide (CO2) capture and storage technology Volume 1: Carbon dioxide (CO2) capture, transport and industrial applications Edited by M. Mercedes Maroto-Valer 9 Geological repository systems for safe disposal of spent nuclear fuels and radioactive waste Edited by Joonhong Ahn and Michael J. Apted 10 Wind energy systems: Optimising design and construction for safe and reliable operation Edited by John D. Sørensen and Jens N. Sørensen

xxiii © Woodhead Publishing Limited, 2013

xxiv

Woodhead Publishing Series in Energy

11 Solid oxide fuel cell technology: Principles, performance and operations Kevin Huang and John Bannister Goodenough 12 Handbook of advanced radioactive waste conditioning technologies Edited by Michael I. Ojovan 13 Membranes for clean and renewable power applications Edited by Annarosa Gugliuzza and Angelo Basile 14 Materials for energy efficiency and thermal comfort in buildings Edited by Matthew R. Hall 15 Handbook of biofuels production: Processes and technologies Edited by Rafael Luque, Juan Campelo and James Clark 16 Developments and innovation in carbon dioxide (CO2) capture and storage technology Volume 2: Carbon dioxide (CO2) storage and utilisation Edited by M. Mercedes Maroto-Valer 17 Oxy-fuel combustion for power generation and carbon dioxide (CO2) capture Edited by Ligang Zheng 18 Small and micro combined heat and power (CHP) systems: Advanced design, performance, materials and applications Edited by Robert Beith 19 Advances in clean hydrocarbon fuel processing: Science and technology Edited by M. Rashid Khan 20 Modern gas turbine systems: High efficiency, low emission, fuel flexible power generation Edited by Peter Jansohn 21 Concentrating solar power technology: Principles, developments and applications Edited by Keith Lovegrove and Wes Stein 22 Nuclear corrosion science and engineering Edited by Damien Féron 23 Power plant life management and performance improvement Edited by John E. Oakey 24 Electrical drives for direct-drive renewable energy systems Edited by Markus Mueller and Henk Polinder 25 Advanced membrane science and technology for sustainable energy and environmental applications Edited by Angelo Basile and Suzana Pereira Nunes 26 Irradiation embrittlement of reactor pressure vessels (RPVs) in nuclear power plants Edited by Naoki Soneda

© Woodhead Publishing Limited, 2013

Woodhead Publishing Series in Energy

xxv

27 High temperature superconductors (HTS) for energy applications Edited by Ziad Melhem 28 Infrastructure and methodologies for the justification of nuclear power programmes Edited by Agustín Alonso 29 Waste to energy conversion technology Edited by Naomi B. Klinghoffer and Marco Castaldi 30 Polymer electrolyte membrane and direct methanol fuel cell technology Volume 1: Fundamentals and performance of low temperature fuel cells Edited by Christoph Hartnig and Christina Roth 31 Polymer electrolyte membrane and direct methanol fuel cell technology Volume 2: In situ characterization techniques for low temperature fuel cells Edited by Christoph Hartnig and Christina Roth 32 Combined cycle systems for near-zero emission power generation Edited by Ashok D. Rao 33 Modern earth buildings: Materials, engineering, construction and applications Edited by Matthew R. Hall, Rick Lindsay and Meror Krayenhoff 34 Metropolitan sustainability: Understanding and improving the urban environment Edited by Frank Zeman 35 Functional materials for sustainable energy applications Edited by John A. Kilner, Stephen J. Skinner, Stuart J. C. Irvine and Peter P. Edwards 36 Nuclear decommissioning: Planning, execution and international experience Edited by Michele Laraia 37 Nuclear fuel cycle science and engineering Edited by Ian Crossland 38 Electricity transmission, distribution and storage systems Edited by Ziad Melhem 39 Advances in biodiesel production: Processes and technologies Edited by Rafael Luque and Juan A. Melero 40 Biomass combustion science, technology and engineering Edited by Lasse Rosendahl 41 Ultra-supercritical coal power plant: Materials, technologies and optimisation Edited by Dongke Zhang

© Woodhead Publishing Limited, 2013

xxvi

Woodhead Publishing Series in Energy

42 Radionuclide behaviour in the natural environment: Science, impacts and lessons for the nuclear industry Edited by Christophe Poinssot and Horst Geckeis 43 Calcium and chemical looping technology for power generation and carbon dioxide (CO2) capture: Solid oxygen- and CO2-carriers P. Fennell and E. J. Anthony 44 Materials’ ageing and degradation in light water reactors: Mechanisms and management Edited by K. L. Murty 45 Structural alloys for power plants: Operational challenges and high-temperature materials Edited by Amir Shirzadi, Rob Wallach and Susan Jackson 46 Biolubricants: Science and technology Jan C. J. Bart, Emanuele Gucciardi and Stefano Cavallaro 47 Wind turbine blade design and materials: Improving reliability, cost and performance Edited by Povl Brøndsted and Rogier Nijssen 48 Radioactive waste management and contaminated site clean-up: Processes, technologies and international experience Edited by William E. Lee, Michael I. Ojovan, Carol M. Jantzen 49 Probabilistic safety assessment for optimum nuclear power plant life management (PLiM): Theory and application of reliability analysis methods for major power plant components Gennadij V. Arkadov, Alexander F. Getman and Andrei N. Rodionov 50 The coal handbook Volume 1: Towards cleaner production Edited by D. G. Osborne 51 The coal handbook Volume 2: Coal utilisation Edited by D. G. Osborne 52 The biogas handbook: Science, production and applications Edited by Arthur Wellinger and David Baxter 53 Advances in biorefineries: Biomass and waste supply chain exploitation Edited by K. W. Waldron 54 Geoscience of carbon dioxide (CO2) storage Edited by Jon Gluyas and Simon Mathias 55 Handbook of membrane reactors Volume 1: Fundamental materials science, design and optimisation Edited by Angelo Basile

© Woodhead Publishing Limited, 2013

Woodhead Publishing Series in Energy

xxvii

56 Handbook of membrane reactors Volume 2: Reactor types and industrial applications Edited by Angelo Basile 57 Alternative fuels and advanced vehicle technologies: Towards zero carbon transportation Edited by Richard Folkson 58 Handbook of microalgal bioprocess engineering Christopher Lan and Bei Wang 59 Fluidized-bed technologies for near-zero emission combustion and gasification Edited by Fabrizio Scala 60 Managing nuclear projects: A comprehensive management resource Edited by Jas Devgun 61 Handbook of process integration: Energy, water, waste and emissions management in processing and power industries Edited by Jiří Klemeš

© Woodhead Publishing Limited, 2013

Foreword

Membrane science and technology was transferred from the laboratory to industry several years ago, first for the process of uranium enrichment and soon after for water desalination using the reverse osmosis process. Pioneer researchers realized the potential of membrane technologies and a vast set of applications soon began to emerge from laboratories. Initially, membranes were considered for liquid and then for gas separation, and were then also integrated into membrane reactors, both biological and chemical. New processes are now being researched and are subsequently finding application in industry. These include micro-, ultra-, nano- and hyperfiltration, pervaporation, dialysis, electrodialysis and membrane contactors, and combined processes such as membrane distillation, membrane extraction, membrane adsorption, membrane crystallization, membrane reactors and electrochemical membrane reactors (namely fuel cells, photocatalytic membrane reactors and photoelectrochemical membrane reactors). Membranes themselves have evolved tremendously but polymer membranes – symmetric, asymmetric, composite, mixed matrix – have witnessed an even more extraordinary level of development. Alternative types of membranes made from other materials soon found their way into various membrane processes. Examples are carbon molecular sieve membranes, metal membranes, glass membranes, molecular imprinted membranes, mixed matrix membranes, smart membranes, catalytic membranes, ion exchange membranes and liquid membranes. The list is growing continuously as a result of researchers’ imagination and in response to society’s needs. This handbook focuses on the various different kinds of membrane reactors and their applications. It covers both chemical and biological membrane reactors, reviews various membrane materials and configurations and also discusses catalysts, characterization, thermodynamics and phenomenological modeling. It was carefully organized by my good friend and colleague Angelo Basile and targets research students, scientists and industry professionals. The book is a broad source of information on membrane reactors. Prof. Adelio Mendes Porto, Portugal xxix © Woodhead Publishing Limited, 2013

Preface

This handbook is dedicated in particular to those readers interested in emerging applications of membrane reactors in the field of energy and environment. The main motivation for this handbook is to give to the reader a panorama of the various aspects of research related to membrane reactors and their applications. The utilization of membrane reactor technology on a larger scale could constitute a relevant enhancement of conventional systems already in existence. For example, in the field of reforming processes, the main benefit of a membrane reactor is the selective removal of a compound such as hydrogen from the reaction side, which may allow the thermodynamic equilibrium restrictions of the conventional fixed bed reactors to be overcome. To this end, I invited an international team of highly expert scientists from the field of membrane science and technology to write about the state-of-the-art of the various kinds of membranes (polymeric, Pd- and non-Pd-based, carbon, zeolite, perovskite, composite, ceramic and so on) used in membrane reactors, modelling aspects related to all kinds of membrane reactors, the various applications of membrane reactors and, finally, economic aspects. Owing to the large amount of material available in the specialized literature, the handbook is composed of two volumes. It should also be mentioned that all the chapters are strictly interconnected; however, for practical use of the handbook, each volume is composed of different parts. In this second volume there are four parts. In Part I a selection of the types of membrane reactor is presented, together with chapters on the integration of membrane reactors with current industrial processes. To summarize, in Chapter 1 (Calabrò) membrane bioreactors are described from an engineering point of view, together with a straightforward description and simulation, with a simple mathematical approach, of the most important configurations and processes in which they are involved. Basic principles of bioconversion, bioreactors and biocatalysis with immobilized biocatalysts are also presented. For all the cited systems the most significant parameters are defined in order to estimate their performances. The best approaches for the preparation of xxix © Woodhead Publishing Limited, 2013

xxx

Preface

ideal interfaces for membrane contactor devices are illustrated in Chapter 2 (Gugliuzza and Basile), after considering the basics of this advanced technology. The potential of this type of membranes in versatile membrane operations such as membrane distillation, osmotic membrane distillation and membrane crystallization is discussed in depth in order to facilitate the adoption of these technologies by others. In comparison with other membrane reactors, pervaporation reactors present some very special characteristics, which are becoming important for new applications, such as the synthesis of fine chemicals and water detoxification. To this end, in Chapter 3 (Camera-Roda, Augugliaro, Loddo and Palmisano), the fundamentals of this integrated reaction-separation process are illustrated, the existing applications are reviewed, the key factors that affect the obtainable process intensification are analysed and, at the end, the opportunities for future developments are considered. Chapter 4 (Comite, Bottino, Capannelli, Costa and Di Felice) summarizes the features of the multi-phase catalytic membrane reactors applied to either gas–liquid or liquid–liquid systems. These multi-phase reactors are reviewed by examining some fundamental concepts involved in the application of a membrane both as an active interface contactor between two fluid phases and as the meeting place between reactants and the catalysts, resulting in an improved management of mass transfer phenomena coupled with heterogeneous reactions. An important aspect of membrane applications is the integration of membrane functionality into microfluidic devices. In this context, in Chapter 5 (Seelam, Huuhtanen and Keiski) microreactor and membrane microreactor concepts are introduced. The chapter provides fundamental knowledge and the state-of-the-art of membrane microreactors. In illustrating how, in these systems, three main functions are combined into only one unit, the authors also show that they can be used in equilibrium-limited reactions. The different membrane types, fabrication methods, applicability, materials and synergic effects are discussed in detail. Potential applications of membrane technology in microfluidic devices are in gas separations, removal of volatile organic compounds, pervaporation, emulsification, gas–liquid contactors and energy devices, especially for niche markets. Chapter 6 (Mozia, Morawski, Molinari, Palmisano and Loddo) reports the state-of-the-art of photocatalytic membrane reactors. These systems combine some advantages of both photocatalytic reaction and membrane separation. To this end, the authors describe the types of membranes used, and some aspects related to membrane operations, such as fouling, are discussed. The different methods of preparation of photocatalytic membranes as well as the most common configurations of photocatalytic membrane reactors are also shown. The distinction between photocatalytic membrane reactors utilizing pressure-driven membrane techniques (microfiltration,

© Woodhead Publishing Limited, 2013

Preface

xxxi

ultrafiltration and nanofiltration) and photocatalytic membrane reactors coupling photocatalysis with dialysis, pervaporation and membrane distillation are discussed in detail, especially in terms of the various mechanisms of mass transport in both types of membrane processes. The possibility of combining membrane operations with traditional separation systems offers significant advantages in terms of product quality, plant compactness, environmental impact, recovery of substances with high added value and energy consumption. In this context, selected applications related to wastewater treatment, agro-food productions, water desalination and others are reviewed in Chapter 7 (Cassano and Basile). Part II of the book deals with membrane reactors used in chemical and large-scale hydrogen production from fossil fuels. Considering again the inorganic dense membranes, Chapter 8 (Tan and Li) introduces the dense ceramic membrane reactors, which are able to lead to significantly improved yields, simplified production processes and reduced capital costs. This chapter describes the principles of various types of configuration (disc/flat-sheet, tubular and hollow fibre) of dense ceramic membrane reactors and also the fabrication of the membranes and membrane reactors. Concerning the aspect of commercializing of these systems, future work addressed at both material and engineering solutions for the fabrication of intact dense membrane with high performances in membrane reactors at a larger scale is also discussed. The purpose of Chapter 9 (Millet) is to provide an overview of chlor-alkali technology and associated cell separators. After a brief historical review of the chlor-alkali process, main reaction characteristics (thermodynamics, cell reactions and kinetics), main chlor-alkali technologies and main cell separators are described. Some improved electrolysis concepts are also presented. Chapter 10 (Chiesa, Romano and Kreutz) deals with the application of high temperature membranes for hydrogen and oxygen separation in large-scale energy systems, for electricity and hydrogen production. Issues related to CO2 removal are considered mainly because carbon capture is probably the most promising application of membranes in the energy sector. The chapter describes the different possibilities for applying membranes in such systems, elucidating the key principles that guide plant design and discussing the advantages resulting from membrane integration. Another aspect considered is related to the fact that although hydrogen-selective membrane technology is very promising and would have a strong impact on industrial processes, crucial issues related to manufacturing procedures and cost reduction have to be properly tackled. In Chapter 11, the authors (De Falco, Iaquaniello, Palo, Cucchiella, Palma and Ciambelli) identify a correct manufacturing strategy with the aim of improving the industrial competitiveness of selective membranes by lowering production costs. Moreover, the analysis of the water–gas shift reaction and, in particular its application

© Woodhead Publishing Limited, 2013

xxxii

Preface

downstream to a reforming step plant for reducing the CO content in the exhaust stream, is presented in order to assess the benefits of the membrane application in chemical reactors. An advanced large-scale Pd-based membrane reformer system with a nominal hydrogen production capacity of 40 Nm3/h developed at the Tokyo Gas Company (Japan) is presented in Chapter 12 (Shirasaki and Yasuda). The potential benefits of this commercial membrane reformer are illustrated in detail by the authors. Chapter 13 (De Falco, Iaquaniello, Salladini and Palo) describes the operating experience performed by the authors on the first pre-industrial natural gas steam reformer plant integrated with Pd-based membrane modules for hydrogen separation. Following the plant description, the plant performance after more than 1000 operating hours is reported in detail, in order to give the readers the real data on the selective membrane behaviour under industrial operating conditions. Chapter 14 (Manzolini and Jansen) discusses a general methodology for the economic assessment of large-scale plants for power and/or hydrogen production. The aim of economic assessment is the calculation of the levelized cost of electricity, cost of hydrogen production, and cost of CO2 avoided for the membrane reactor processes as well as reference cases. The fundamentals of the economic analysis are heat and mass balances for the considered plants and a detailed cost assessment of each component. Particular attention is given to membrane reactor equipment cost, the equipment being an innovative component, and the core of the systems. In Part III electrochemical devices and transport applications of membrane reactors are taken into consideration. Chapter 15 (Cassir, Jones, Ringuedé and Lair) concentrates on electrochemical devices for energy. This chapter focuses on fuel cells and electrolytic cells. It should be said that a fuel cell could also be seen as an electrochemical membrane reactor. Both electrochemical devices are part of an international effort at both fundamental and demonstration levels, and, in some specific cases, market entry has already begun. The principal features of these devices are outlined, focusing on the properties of the state-of-the-art membranes and on the present novelties in this area. In Chapter 16 (Pozio and Tosti), the theory of hydrogen evolution over Pd-based cathodes and its permeation through dense metal walls is introduced followed by a description of the Damköhler–Péclet analysis for these membrane reactors. Particular attention is given to new electrolyser prototypes using thin-wall Pd–Ag permeator tubes, and the efficient operation of cell efficiency and surface activation/deactivation phenomena. Future trends in development of materials for electrolyser cathodes are also presented. When combined with the right energy carrier, fuel cells have the highest efficiencies and lowest emissions of any vehicular power source.

© Woodhead Publishing Limited, 2013

Preface

xxxiii

Chapter 17 (Veziroglu and Macario) presents the relevant research and development work which has taken place on fuel cell vehicle technologies, focusing on economical and environmental aspects. Chapter 18 (Millet and Basile) provides an overview of palladium-alloy metallic membranes for application in the transport industry. Main membrane materials, manufacturing processes, basic principles of hydrogen sorption and permeation through palladium-based metallic membranes, membrane characterization and performances are discussed in depth in this chapter. Finally, some applications are presented and advantages and limitations of existing technologies are discussed. Next, Part IV deals with membrane reactors in environmental engineering, biotechnology and medicine. In particular, Chapter 19 (Cassano, Figoli, Galiano, Argurio and Molinari) provides a comprehensive overview of emerging techniques of particular interest, such as supported liquid membranes, pervaporation, membrane bioreactors and complexation-ultrafiltration developed as eco-friendly processes for treating wastewaters from different sources. The next chapter deals with the use of membrane reactors for enzyme-catalysed bioremediation processes showing that it is potentially beneficial in terms of improving the process economics by enabling enzyme reuse and enhancing overall efficiency and robustness. In fact, Chapter 20 (Hai, Nghiem and Modin) comprehensively reviews the laboratory scale studies on biocatalytic membrane reactors. In particular, it demonstrates the potential applications in wastewater treatment. Studies reported in the literature are only used as proof of concept. Issues that need to be addressed in order to achieve scale-up of such systems have also been discussed in this chapter. In Chapter 21 (Molinari, Palmisano, Loddo, Mozia and Morawski), the configurations and performances of photocatalytic membrane reactors assembled with membranes (also photocatalytic type) and/or suspended photocatalyst are described. Examples of application of photocatalytic membrane reactors in water/ wastewater treatment as well as in photocatalytic synthesis are shown. The outline on modelling of these systems is presented and a brief economical analysis and future trends are also discussed. Chapter 22 (Uragami, Chakraborty, Piemonte and Di Paola) presents a survey of membrane bioreactor applications in pharmaceutical, environmental and biomedical fields. The integration between the separation and the reaction units is demonstrated to highly improve the selectivity and productivity of bioreactors, especially when some factors (inhibition by products or reactants, contamination by xenobiotics) play a key role in the bioreactive systems. In Chapter 23 (Calabrò and Iorio), membrane bioreactors are described from an economical point of view. Most important rules and parameters have been preliminarily introduced. Some applications,

© Woodhead Publishing Limited, 2013

xxxiv

Preface

examples and case studies are also reported in order to have an idea of the impact of economics on plant design, project and control. I wish to take this opportunity to thank all the authors of the chapters for their expert contributions. A. Basile ITM-CNR, Italy

© Woodhead Publishing Limited, 2013

1 Engineering aspects of membrane bioreactors V. CALABR Ò, University of Calabria, Italy

DOI: 10.1533/9780857097347.1.3 Abstract: This chapter describes membrane bioreactors from an engineering point of view. A membrane bioreactor can be defined as a unit operation or a piece of chemical equipment that combines a biocatalyst-filled reaction chamber with a membrane system for the purposes of adding reactants or removing products from a reaction. The basic principles of bioconversion, bioreactors and biocatalysis are introduced, together with a description of the most important biocatalyst immobilization techniques. The mass transfer phenomena involved in membrane systems are discussed along with some representative configurations of membrane bioreactors, whose behaviour can be described using a simple mathematical approach. For all the aforementioned systems the most significant parameters have been defined to estimate the system performance. Key words: membrane bioreactors, biocatalysis, process modelling, biocatalyst kinetics.

1.1

Introduction

A bioreactor is a device within which biocatalysts, usually enzymes or living cells, carry out biochemical transformations. A bioreactor is frequently called a fermenter whether the transformation is carried out by living cells or in vivo cellular components, that is, enzymes. A membrane bioreactor can be defined as a unit operation or a piece of chemical equipment that combines a bioreactor with a membrane system. An enzyme membrane reactor is a membrane bioreactor in which the biocatalyst is an enzyme. In a membrane bioreactor, the membrane can be used for different tasks: • • • • •

separation, selective extraction of reactants, retention of the biocatalyst, distribution/dosing of a reactant and biocatalyst support (often combined with distribution of reactants). 3 © Woodhead Publishing Limited, 2013

4

Handbook of membrane reactors

Consequently, membrane bioreactors are an example of the combination of two unit operations in one step; for example, membrane filtration with the chemical reaction. In a typical membrane bioreactor, as well as acting as a support for the biocatalyst, the membrane can be a very effective separation system for undesirable reactions or products. The removal of a reaction product from the reaction environment can be easily achieved thanks to the membrane selective permeability, and this is of great advantage in thermodynamically unfavourable conditions, such as reversible reactions or product-inhibited enzyme reactions. A very interesting example of a membrane bioreactor is the combination of a membrane process, such as microfiltration or ultrafiltration (UF), with a suspended growth bioreactor. Such a set up is now widely used for municipal and industrial wastewater treatment, with some plants capable of treating waste from populations of up to 80 000 people (Judd, 2006). Over the last few decades membrane science and technology have offered a great contribution to the development of biotechnology and, more specifically, to the engineering of enzyme bioreactors. Membranes have been extensively used to support biocatalyst immobilization with the aim of creating workable membrane bioreactors (Atkinson, 1974; Belfort, 1989; Cheryan and Mehaia, 1986; Giorno et al., 2003; Iorio et al., 1994; Messing, 1975). Different membrane configurations and membrane bioreactors have been proposed in recent years for this purpose and have been widely discussed by several authors (Calabrò et al., 2008) and the optimization of novel immobilization techniques certainly improves biocatalyst behaviour, thus leading to the development of very effective bioreactors. Synthetic membranes, for example, have been well-assessed as supports for the immobilization of biological catalysts under milder conditions, compared to those that exist when biocatalysts are chemically bound to a membrane. A synthetic membrane, with a suitable molecular weight cut-off (MWCO), somehow artificially replicates the functions of a cell membrane ensuring the protection of a purified enzyme against contaminants and inhibitors. Biocatalysts are not always immobilized on membranes in bioreactors, though. As enzymes are macromolecules and often differ greatly in size from reactants they can be separated by size exclusion membrane filtration with ultra- or nano-filtration. This is used on an industrial scale in one type of enzyme membrane reactor for the production of enantiopure amino acids by kinetic racemic resolution of chemically derived racemic amino acids. The most prominent example is the production of L-methionine on a scale of 400 t/y (Liese et al., 2006). The advantage of this method over immobilization of the catalyst is that the enzymes are not altered in activity or selectivity as they remain solubilized. This principle can be applied to all macromolecular catalysts which can be separated from the other reactants by means of filtration. So far, only enzymes have been used to a significant extent.

© Woodhead Publishing Limited, 2013

Engineering aspects of membrane bioreactors

5

The aim of this chapter is to give a detailed overview of the characterization of biocatalysts and the development of membrane bioreactors, in particular, the main aspects of biocatalyst kinetics and their immobilization/ entrapment, either within the porous membrane structure, or on its surface. Transport models that can help to predict the behaviour of membrane bioreactors, as well as the most relevant theoretical models and operating parameters, are presented below. This data is then analysed in order to ascertain how to improve effectiveness and productivity of the membrane bioreactors. Some relevant fields of application are also discussed in order to show the potential of such systems.

1.1.1

Basic principles of bioconversion

Bioprocess plants which use microorganisms and/or enzymes, such as fermentation plants, have many characteristics similar to those of chemical plants. Therefore, an engineering approach to the design and operation of various plants which involve biological systems would be valuable, provided the differences in the physical properties of some materials are taken into account. Bioprocesses involve many reactions, both chemical and biochemical. In order to design a successful reactor, it is essential to understand how the composition of reactants and products and their utilization and production rates change under various conditions. To simulate behaviour in a bioreactor involving biochemical reactions, for example, the formation and disappearance as well as mass balance of specific components must be calculated. The study of the kinetics of the enzyme-catalysed biochemical reactions that are involved in the growth of microorganisms associated with bioreactors is essential in this respect (Katoh and Yoshida, 2009). Bioreactions can happen in both a homogeneous liquid phase as well as a heterogeneous phase, including gas, liquid and/or solid. An example of a reaction in the heterogeneous phase is the immobilization of enzymes and aerobic fermentation with oxygen supply. It is evident that for the engineering approach to succeed, background knowledge of the biological systems involved is required.

1.2

Biocatalysts and their immobilization

Enzymes or whole cells represent biocatalysts, that is, natural catalysts that initiate or modify the rate of a chemical reaction by reproducing the biological activity that they play in living organisms. There are some differences between enzymes or whole cells, mainly due to the fact that whole cells are living microorganisms (such as yeast or bacteria) whereas enzymes are proteins. As a consequence, reactions biocatalysed by enzymes do not suffer the disadvantage of simultaneous cell

© Woodhead Publishing Limited, 2013

6

Handbook of membrane reactors

growth. However, enzymes are more expensive, due to their production. Consequently, in order to prevent their damage and guarantee their reuse, it is very convenient to immobilize them on or in a support.

1.2.1

Enzymes as biocatalysts

Enzymes are globular proteins which catalyse the complex biological reactions of living organisms, a process also known as ‘metabolism’. They are usually confined within living cells and are often bound to cellular membranes. Enzymes catalyse specific reactions contributing to the energy balance of living cells and at the same time regulate their metabolic state. Although enzymes play a similar role in a specific reaction to that of synthetic catalysts, their catalytic action is extremely efficient and selective, well beyond the performance of synthetic catalysts. Enzyme engineering represents a new branch of biotechnology that involves the production of enzymes, separation and purification, as well as the characterization and design of enzyme reactors. The application of enzymes has increased enormously in the last few years, coupled with a series of related developments, such as: • enhancement of the techniques to induce microorganisms which produce selected enzymes; • improved enzyme purification techniques; • engineering of techniques to immobilize enzymes or whole cells on, or in, solid supports; • development of techniques for enzyme usage in continuous flow reactors. Enzymes produced by microorganisms such as fungi or bacteria have been used for years in batch fermentation plants for the production of pharmaceuticals, beverages, foods, etc. The availability of almost pure enzymes enables one to carry out specific reactions under mild conditions. Side-product formation can also be limited and the synthesis of chemically active compounds, which would otherwise require extremely long reaction times, can be performed with low yields. The use of enzymes as biocatalysts can therefore be of extreme interest for individual applications, mainly for the advantages it grants in terms of energy consumption, safety, pollution prevention and materials preservation. The most important fields of application of enzymes have been listed in Table 1.1. Enzyme activity can be measured in terms of the so-called turnover number, which represents the net number of substrate molecules reacted per catalytic activity site and per unit time. The turnover number, therefore,

© Woodhead Publishing Limited, 2013

Engineering aspects of membrane bioreactors

7

Table 1.1 Industrial applications of enzymes Application

Enzymes

Uses

Food processing

Amylases from fungi and plants

Production of sugars from starch, such as in making high-fructose corn syrup. In baking, to catalyse breakdown of starch in the flour to sugar. Yeast fermentation of sugar produces the carbon dioxide that raises the dough. Biscuit manufacturers use them to lower the protein level of flour.

Proteases Baby foods

Trypsin

To predigest baby foods.

Brewing industry

Enzymes from barley are released during the mashing stage of beer production Industrially produced barley enzymes Amylase, glucanases, proteases Betaglucanases and arabinoxylanases Amyloglucosidase and pullulanases Proteases

They degrade starch and proteins to produce simple sugar, amino acids and peptides that are used by yeast for fermentation.

Acetolactatedecarboxylase (ALDC)

Widely used in the brewing process to substitute for the natural enzymes found in barley. Split polysaccharides and proteins in the malt. Improve the wort and beer filtration characteristics. Low-calorie beer and adjustment of fermentability. Remove cloudiness produced during storage of beers. Increases fermentation efficiency by reducing diacetyl formation.

Fruit juices

Cellulases, pectinases

Clarify fruit juices.

Dairy industry

Rennin, derived from the stomachs of young ruminant animals Lipases

Manufacture of cheese, used to hydrolyze protein.

Lactases Meat tenderizers

Papain

Is implemented during the production of Roquefort cheese to enhance the ripening of the blue-mould cheese. Hydrolysis of lactose to glucose and galactose. To soften meat for cooking. (Continued)

© Woodhead Publishing Limited, 2013

8

Handbook of membrane reactors

Table 1.1 Continued Application

Enzymes

Uses

Starch industry

Amylases, amyloglucosidases and glucoamylases Glucose isomerase

Converts starch into glucose and various syrups.

Paper industry

Amylases, xylanases, cellulases and ligninases

Degrade starch to lower viscosity, aiding sizing and coating paper. Xylanases reduce bleach required for decolorizing; cellulases smooth fibres, enhance water drainage, and promote ink removal; lipases reduce pitch and ligninases remove lignin to soften paper.

Biofuel industry

Cellulases

Used to break down cellulose into sugars that can be fermented (see cellulosic ethanol). Use of lignin waste.

Ligninases Biological detergent

Primarily proteases, produced in an extracellular form from bacteria Amylases

Lipases Cellulases

Converts glucose into fructose in production of high-fructose syrups from starchy materials, with enhanced sweetening properties and lower calorific values than sucrose.

Used for presoak conditions and direct liquid applications helping with removal of protein stains from clothes.

Detergents for machine dish washing to remove resistant starch residues. Used to assist in the removal of fatty and oily stains. Used in biological fabric conditioners.

Contact lens cleaners

Proteases

To remove proteins on contact lens to prevent infections.

Rubber industry

Rubber industry

To generate oxygen from peroxide to convert latex into foam rubber.

Photographic industry

Protease (ficin)

Dissolve gelatin off scrap film, allowing recovery of its silver content.

Molecular biology

Restriction enzymes, DNA ligase and polymerases

Used to manipulate DNA in genetic engineering, important in pharmacology, agriculture and medicine. Molecular biology is also important in forensic science.

© Woodhead Publishing Limited, 2013

Engineering aspects of membrane bioreactors

9

allows for comparison of the activity of a particular enzyme with that of an inorganic catalyst operating under more severe conditions (Bailey and Ollis, 1986; Iorio et al., 1994). Operating conditions, such as pH and temperature, play a significant role since they strongly affect the catalytic activity of the enzyme. An optimal value of the above two parameters can be determined for each enzyme catalysing a specific reaction. Usually, if the reaction is carried out at an operating temperature lower than the optimal one, a reduced rate is observed; on the other hand, temperatures higher than the optimal value can cause enzyme deactivation. Similar considerations apply with reference to the variation in pH (Bernhard, 1968). During the course of a reaction, biocatalysts are subjected to a progressive decay of catalytic activity due to mechanical or thermal stresses that modify their native conformation and damage some catalytic sites. Enzyme deactivation rate can be estimated and expressed in terms of the process variables that determine catalytic sites damage.

1.2.2

Immobilization of biocatalysts

When an enzyme operates in its native form and is used in a stirred batch or a similar reactor, a number of drawbacks may be encountered in the homogeneous solution. The following disadvantages are noted (Chibata 1978; Zaborsky 1973): • high enzyme purification costs; • low productivity per reactor, per unit time; • difficult and expensive recovery and reuse of enzymes or cellular microorganisms; • product pollution; • difficulties in maintaining standard product quality. Most of the above problems can be overcome by using immobilized enzymes; in other words, enzymes which are confined in a well-defined region of space by means of a selective membrane, or immobilized by, for example, absorption or entrapment within the polymeric matrix of a membrane. Enzymes are prohibited, due to their molecular size, from diffusing out or permeating through the membrane, while substrates and products can readily permeate the membrane. The enzymes retain their catalytic properties and can be repeatedly and continuously used. Traditional immobilization techniques are summarized in Fig. 1.1. They include adsorption on a surface, covalent binding to an insoluble support, co-polymerization with a proteic carrier, encapsulation in a membrane shell and confinement in a gel. A comparison between some different techniques is also reported in Table 1.2. Hollow fibre membranes are commonly used for membrane reactors

© Woodhead Publishing Limited, 2013

10

Handbook of membrane reactors (a)

(b) E

E

E

E E

E

Enzyme

E Solid support

(c)

Polymeric matrix (d) E

Membrane

E E

E

E

E

1.1 Immobilized enzyme systems. (a) Enzyme non-covalently adsorbed to an insoluble particle; (b) enzyme covalently attached to an insoluble particle; (c) enzyme entrapped within an insoluble particle by a cross-linked polymer; (d) enzyme confined within a membrane. Table 1.2 Comparison of different enzyme immobilization techniques Characteristics

Adsorption

Covalent binding

Entrapment

Membrane confinement

Preparation Cost Binding force Enzyme leakage Applicability Running problems Matrix effects Large diffusional barriers Microbial protection

Simple Low Variable Yes Wide High Yes No

Difficult High Strong No Selective Low Yes No

Difficult Moderate Weak Yes Wide High Yes Yes

Simple High Strong No Very wide High No Yes

No

No

Yes

Yes

because of their high surface-to-volume ratio that allows for high biocatalyst density in a small reactor volume. Although diffusion is the primary mass transfer mechanism, the membrane bioreactors can also be operated with UF fluxes through the membrane wall. These fluxes are generally promoted by the application of a trans-membrane pressure (TMP) difference. If significant TMP values are applied, mass transfer rate can be enhanced by the addition of a convective component to the diffusive mass transfer mechanism. (Atkinson, 1974; Calabrò et al., 2002; Messing, 1975; Olson and

© Woodhead Publishing Limited, 2013

Engineering aspects of membrane bioreactors

11

Cooney, 1975). Several general advantages derive from the use of immobilized enzymes or microbial cells, such as: • • • • •

opportunity to design processes in a more rational way; cost cut-off in enzyme consumption; more compact plants; operating costs cut-off; high productivity per unit time per equipment, with small amounts of side-products.

The immobilization of microorganisms can also be performed in place of the immobilization of purified enzymes. This may be necessary, for example, when enzymes are intracellular, when an enzyme extracted from cells is unstable during and after immobilization, or when the microorganism does not contain any stable interfering enzymes (Giorno et al., 2003). The immobilization of whole cells is sometimes preferable when there is no limitation in terms of mass transport of both substrate and product(s) to the catalytic sites. In this case, the immobilization of whole cells has some significant advantages compared to enzyme immobilization. In particular, significant cost reduction can be achieved since enzyme extraction and purification are not necessary. Furthermore, the enzyme remains in its natural environment, therefore a high level of initial activity can be maintained for longer period. Another advantage offered by cell immobilization is the possibility to perform multi-step reactions, each of them catalysed by a different enzyme among those belonging to the cell enzyme heritage (Calabrò et al., 2008; Cheryan and Mehaia, 1986).

1.3

Membranes as enzyme supports and for downstream processing

Membranes represent good supports for enzyme immobilization because they enable the integration of biocatalysis and separation. Often, the available commercial membranes require modifications to make them suitable for enzyme immobilization. Different immobilization techniques can be used on such suitable membranes, depending on many factors mainly related to enzymes and membrane properties and bioprocess performances.

1.3.1 The role of membranes in biocatalyst immobilization Enzyme forms and types of carrier Immobilized enzymes can occur in different forms – particles, membranes, tubes and fibres, the most common of which is the particle form. The

© Woodhead Publishing Limited, 2013

12

Handbook of membrane reactors

advantages of having enzymes in this form are ease of handling and ease of application. Immobilized enzymes in membrane form are made either by moulding the enzymes into a membrane form, or by attaching the enzymes to membrane-type carriers. For the moulded membrane, the enzymes are usually enclosed within a semi-permeable polymer membrane utilizing the entrapment method (discussed in detail in the following subsection). The formation of enzyme fibres also utilizes the entrapment method for enzyme immobilization. Another form of enzyme immobilization is tube form. In order to produce enzyme tubes, polymer ‘carrier’ tubes are treated in a series of chemical reactions in which the enzyme is bound by diazo coupling. Examples of the polymers used for the carrier include nylon and polyacrylamide. In fact, both organic and inorganic solid supports can be used for enzyme immobilization. Some of the organic supports used are listed below (Jochems et al., 2011): • • • • • • • • •

carbon, proteins, polypeptides, polystyrenes, polysaccharides, polyamides, polyacrylates, maleic anhydride based copolymers and vinyl and allyl polymers.

The entrapment method for immobilizing enzymes Entrapment consists of the placement of an enzyme within the framework of a membrane or polymer matrix. In this way, the protein can be retained while allowing for penetration of the substrate. The process can be further categorized into lattice and microcapsule entrapment. This process requires the enzyme to be trapped, rather than be bound to the gel matrix or membrane as with covalent bonding and crosslinking (see subsection below), resulting in wide applicability. However, a loss in enzyme activity can be observed due to the relatively severe conditions required for the chemical polymerization reaction. Therefore, it is necessary to consider the most suitable conditions for the immobilization of different enzymes. For the lattice-type entrapment method a cross-linked water-insoluble polymer, such as polyvinyl alcohol or polyacrylamide, is used to trap the enzymes. A natural polymer, or starch, could also be used in this technique. For the microcapsule-type entrapment method, a semi-permeable polymer membrane is used to surround the enzymes. As with lattice-type entrapment, control of the conditions is very important as they can have a detrimental effect on the preparation of the enzyme microcapsules. There

© Woodhead Publishing Limited, 2013

Engineering aspects of membrane bioreactors

13

are several different methods for the microencapsulation of enzymes, and two such processes are described below. Interfacial polymerization is a method which involves the utilization of a water-immiscible organic solvent to emulsify an aqueous mixture containing the enzyme and a hydrophilic monomer. Additional amounts of the hydrophilic monomer are stirred into the organic solvent; at the interface between the aqueous and organic solvent phases within the emulsion, the monomers are polymerized. This results in the entrapment of the enzyme in the aqueous phase within a polymer membrane. Liquid drying is another method of microencapsulation, in which the enzyme is enclosed in a polymer membrane. First of all a polymer is dissolved into a water-immiscible organic solvent. For this process to work effectively the solvent must have a boiling point lower than the boiling point of water. An aqueous solution containing the enzyme is then dispersed into the organic phase to form a water-in-oil type emulsion, creating aqueous micro droplets. This emulsion is in turn distributed in an aqueous phase which contains surfactants and protective colloidal substances, such as gelatine. During this part of the process a secondary emulsion is created. The organic solvent is removed by warming in a vacuum, producing a polymer membrane to encapsulate the enzyme. Other methods of enzyme immobilization In other methods, the enzymes are not trapped, but rather they are absorbed into or bound to the gel matrix or membrane. Enzymes can be absorbed within symmetric macroporous membranes in order to establish high catalyst concentrations, cross-linked to prevent them from elution, or simply covalently or ionically bound either to symmetric or asymmetric membranes. In spite of short residence times, high conversions can be achieved in most kinds of enzyme membrane reactors. Concentration polarization phenomena can be used to form a gel-layer of enzyme proteins on a membrane, in either dynamic or static conditions. It is possible to establish more than one enzyme layer, without a coupling agent to carry out the immobilization. Due to high protein concentration on the membrane surface, enzyme stability can be improved compared with systems using enzymes homogeneously distributed in the reacting solution. To entrap an enzyme, a biocatalyst suspension can be forced through the unskinned surface of an asymmetric hollow fibre membrane so that biocatalysts, either enzymes or whole cells, although still suspended, are effectively immobilized within the macroporous spongy part of the membrane. In this way the enzymatic activity can be spread over a large surface, although substrates and products can only diffuse to and from the biocatalyst. In certain applications, though, the size of the biocatalyst is not suitable for entrapment into the supporting sponge layer of asymmetric membrane. In these

© Woodhead Publishing Limited, 2013

14

Handbook of membrane reactors

cases, the membrane acts mainly as a selective barrier; it defines a reactive zone in the vessel, usually the shell, accessible to substrates and products primarily by diffusive mass transfer protecting the catalyst from pollution or inhibition, possibly caused by other species in solution.

1.3.2

Downstream processing methods

In bioprocesses, the initial concentrations of target products are usually low, so separation and purification, so-called downstream processing, is required to obtain the final products. In the preliminary stage of downstream processing it is necessary to separate the cells from the fermentation broth. If it is necessary to obtain the intracellular products, the cells are ruptured first to solubilize the products. The next stage is to concentrate an amount of the target product using a range of different methods, such as UF, aqueous two-phase separation, salting-out and extraction. The filtration stage consists of separating the required particles from a suspension. This can be done in several ways. Positive pressure can be applied to the upstream side of the filtration medium, or a vacuum can be used to create the opposite effect on the downstream side. Alternatively, fluid can be forced through the filtration medium. For separation of relatively large precipitates and microorganisms, conventional filtration methods can be used. Particles larger than several microns in diameter can be treated using this method. However, for smaller particles centrifugation or microfiltration should be utilized in order to effectively separate them. Microfiltration has been used to separate cells and cell lysates (fragments of disrupted cells) from other soluble components. This was achieved by creating a membrane bioreactor in which the cells can operate, with a downstream separation process. Using this method of separation the following advantages can be gained: • A high retention rate of the cells (>99.9%) can be achieved. • The closed systems used in this process are free from aerosol formation. • The process does not depend on a difference in density between the cells and the media. • A filter aid is not required. Microfiltration membranes can have pore sizes ranging from 0.01 to 10 mm. The choice of pore size used is dependent on the desired clarity of the filtrate and the size of the cells/debris to be filtered. Cross-flow filtration (CFF; see Fig. 1.2) can provide a higher filtration flux than dead-end filtration (DEF, Fig. 1.3). This is because the liquid flows parallel to the membrane surface. Less of the retained species can collect on the membrane surface due to this liquid flow, which can sweep away small amounts of the

© Woodhead Publishing Limited, 2013

Engineering aspects of membrane bioreactors Cb (y )

C0

H

15

Cb Membrane

x Cw (y) y

δf (JAm), Cp

1.2 Scheme of CFF membrane systems in the separation of cells. In the scheme, C represents the cell concentration: Cb the bulk concentration, Cw the concentration at membrane wall, Cp the concentration in the permeate, C0 the inlet concentration. H is the reactor height, δ the cake thickness, J the permeate flux and Am the membrane area. C0

Cb H Membrane Cw

x

δf (JAm), Cp

1.3 Scheme of DEF membrane systems in the separation of cells. In the scheme, C represents the cell concentration: Cb the bulk concentration, Cw the concentration at membrane wall, Cp the concentration in the permeate, C0 the inlet concentration. H is the reactor height, δ the cake thickness, J the permeate flux and Am the membrane area.

retained species. The size of the layer of microparticles which builds up on the membrane surface is dependent on the balance between the quantity of the particles which are swept away by the cross-flow along the membrane, and the amount of particles transported by the bulk flow towards the membrane (Katoh, 2009).

1.4

Membrane bioreactor configurations

Various membrane bioreactor configurations have been described in the literature. Whatever the configuration, though, the main objective is to ensure the complete rejection of the enzyme in order to maintain the optimum level of efficiency inside the reacting volume. As described above, depending on the situation, enzyme molecules may be freely circulating on the retentate side, immobilized on the membrane surface itself, or inside the membranes

© Woodhead Publishing Limited, 2013

16

Handbook of membrane reactors

porous structure (Rios, 2004). Some of the different types of membrane bioreactors which have been developed are listed below: • enzyme membrane reactors (EMR) with biocatalyst in a homogeneous solution, • enzyme gel-layer membrane reactors, • membrane bioreactors with continuous biocatalyst recirculation, • membrane segregated enzyme reactors, • membrane bound enzymes in continuous flow reactors, • continuous stirred membrane bioreactors (CSMB) in which the biocatalyst is immobilized on the membrane surface and • whole cells or enzymes immobilized in capillary membrane reactors.

1.4.1

Bioreactors in which enzymes are not immobilized on or within a porous membrane support

Enzyme membrane reactors (EMR) with biocatalyst in a homogeneous solution In the EMR in Fig. 1.4, the enzyme acts as a biocatalyst in homogeneous solution, where diffusive resistances are minimized. It mainly consists of a continuously stirred tank, with a volume symbolically equal to V, equipped with a UF membrane and characterized by an exposed area Am. The membrane is chosen to ensure complete rejection of the enzyme. The tank is loaded with the required amount of enzyme. It is then continuously fed with the substrate solution, at a given concentration of substrate S0 and Q, S0

Eb, S, P

H

Membrane X

δf

Ew

(JAm), S, P

1.4 Schematic of an EMR of volume V, where the rejected enzyme profile is evidenced; EW is the enzyme concentration on the membrane wall, Eb is the bulk concentration.

© Woodhead Publishing Limited, 2013

Engineering aspects of membrane bioreactors

17

flow rate Q. The reaction takes place and the products, at concentration P, are continuously removed by permeation through the membrane. A uniform enzyme concentration E0 can be assumed before the filtration takes place. Continuous product withdrawal can be obtained, thus allowing for an increase of reaction yield in thermodynamically unfavourable systems; product inhibition is also minimized. Unfortunately, the rejected enzyme accumulates on the membrane surface, as a consequence of the concentration polarization phenomena that occurs in a thin layer close to membrane surface. Enzyme deactivation can also occur. The reaction rate and the conversion rate within the reactor decrease as a consequence of the reduced catalytic activity of the enzyme. With systems such as this it is simple to replace the inactivated enzyme with fresh biocatalyst. Enzyme gel-layer membrane reactors Enhancement of enzyme gel-layer membrane reactors can be obtained by means of co-gelation of the enzyme and of the high molecular weight inert compounds (such as polyalbumins) during UF. This should result in a more stable gel on the membrane surface. A further improvement of the mechanical stability of the enzyme gel can be achieved by means of a co-polymerization/gelation process, in which the enzyme is chemically bound to the high molecular weight compounds through bridge molecules, prior to the UF-gelation step. Membrane bioreactors with continuous biocatalyst recirculation The occurrence of concentration polarization phenomena typical of the first EMR configuration described in this section can be efficiently tackled using CFF in capillary or hollow fibre membrane modules, as shown in Fig. 1.5. Larger surface-to-volume ratio and higher compactness make such a configuration more suitable for large scale operations, as is required in most industrial applications. The enzyme is continuously recirculated to the reaction tank where reaction occurs at the greatest extent, with partial or total rejection. Permeate is continuously removed by filtration through the membrane: reaction products are, consequently, continuously removed. The selectivity of the membrane is chosen with respect to the various species which constitute the reacting solution; the relative size of the molecules and membrane pores are associated in this way. Due to the fact that a lot of enzymes have a molecular weight between 10 and 80 kDalton, UF membranes with a molecular cut-off between 1 and 100 kDalton are the most frequently used. Segregated enzyme reactors Segregated enzyme reactors are based on the method of confining the enzyme in a well-defined region of space. This technique can be applied to hollow fibre

© Woodhead Publishing Limited, 2013

18

Handbook of membrane reactors Retentate (F-JAm), SR, PR, ER

Q, S0

Eb, S, P

Membrane

F, S, P, Eb Permeate (JAm), SP, PP

1.5 Schematic of membrane bioreactor with continuous biocatalyst recirculation in the retentate.

membrane systems; enzyme segregation can be achieved both in the fibre lumen and in the reactor shell. When the biocatalyst is immobilized in the shell side, a high surface-to-volume ratio is attained. The enzyme can be easily replaced and control of the fluid dynamic conditions is straightforward. Enzymes can be also immobilized by binding them to a membrane by means of extremely active bridge molecules. The conversion of macromolecules to a lower weight species by enzymes favours an easy recovery of reaction products permeating through suitable supporting membranes. With these systems mentioned above the biocatalyst is stably bound to the matrix, so as to enhance enzyme stability. Nevertheless, some disadvantages characterize the actual behaviour of the chemical bound enzymes; namely, a significant activity loss, higher enzyme vulnerability (as they are much more exposed to contaminants) and a cost increase caused by the unfeasibility of recovering exhausted enzymatic preparations that, therefore, have to be discarded together with the support.

1.4.2

Bioreactors with immobilized enzymes on or within a porous membrane support

Membrane bound enzymes in continuous flow reactors The membrane supporting the biocatalyst can be fractionated into small pieces, and the biocatalytic particles used in different bioreactor

© Woodhead Publishing Limited, 2013

Engineering aspects of membrane bioreactors

19

Q, S0

Membrane with immobilized enzyme onto the surface

V, S, P

Q, S, P

1.6 Schematic of a CSMB in which biocatalyst is immobilized on the membrane surface.

configurations. There are two ways to operate a biochemical reactor with suspended solid catalyst particles or microbial cells: batchwise or continuous. These modes of operation can be further broken down into stirred batch, stirred semi-batch, continuous stirred and continuous plug flow reactors. In stirred batch, stirred semi-batch and continuous stirred modes, the contents of the tanks are completely mixed resulting in a composition consistent throughout. For batch reactors, the reactants are charged and then the products are recovered after a specified reaction time. For semi-batch reactors and fed-batch reactors, the reactants are constantly fed. In these processes the product, or products, are recovered batchwise. The reactants concentration changes over time, as does the concentration of the products (Katoh and Yoshida, 2009). Steady-state flow reactors, with a constant supply of reactants and continuous removal of products, can be operated as both a continuous stirred-tank bioreactor (CSTB) and as a plug flow bioreactor (PFB). It is possible to have different configurations of the membrane bioreactor where the biocatalyst is immobilized in the fractionated membrane support (Katoh and Yoshida, 2010). In Fig. 1.6 the scheme of a CSMB in which the biocatalyst is immobilized on the surface of the membrane beads is presented. The biocatalyst immobilized in the porous structure of a fractioned membrane can also be operated in CSMB. For example, two configurations are shown in Fig. 1.7: (a) for flat-sheet and (b) for spherical porous structures, respectively. Such structures could also be adopted for PFB, where a bed of membrane support with the immobilized biocatalyst could be utilized, in either a fixed or fluid configuration.

© Woodhead Publishing Limited, 2013

20

Handbook of membrane reactors (a)

Q, S0

Biocatalyst immobilized inside the membrane

V

Q, S, P

(b)

Q, S0

Biocatalyst immobilized inside the membrane

V

Q, S, P

1.7 (a) Scheme of a CSMB with biocatalyst immobilized inside a flat-sheet (slab) membrane. (b) Scheme of a CSMB with biocatalyst immobilized inside a spherical membrane.

Whole cells or enzymes immobilized in capillary membrane reactors Immobilization inside the membrane pores can occur both in the dense skin and in the spongy, porous layer (Calabrò et al., 2002; Curcio et al., 2006a) as reported in Fig. 1.8, where a hollow fibre membrane bioreactor is depicted. A membrane bioreactor with this configuration is made up of a collection of parallel, polymeric hollow fibres brought together in a cylindrical shell (Calabrò et al., 2001). This type of bioreactor is usually split into the shell side by the porous membrane wall and the luminal. The wall acts as a selective barrier for the transportation of the involved species through the membrane. Single pass (Fig. 1.8a) or recycle configurations (Fig. 1.8b) could also be realized. © Woodhead Publishing Limited, 2013

Engineering aspects of membrane bioreactors

21

(a) Substrate Product

Sample

Substrate Membrane bioreactor

Pump

E E E

E E

E

E

E

Spongy layer (Zone 3)

E

E

E

E

E

E

E E E

S: Substrate Dense layer P: Product (Zone 2) E: Enzyme V(t): Change of feed tank volume with time

E

E E

E

E E

E E E

E E

E

E E

S

E E E

E E E

E

E

E

E

E E

E

E E

E

E

E

E

E

E

P E

E

E E

E

E

E

E E

E

Membrane lumen (Zone 1)

1.8 Schematic of a continuous membrane bioreactor with enzyme immobilized in the membrane fibres: (a) single pass; (b) recycle of un-reacted substrate. (Continued)

1.5

Modelling and simulation: kinetics of enzyme reactions

The behaviour of membrane bioreactors is strongly affected by kinetics and mass transfer limitations (Satterfield and Sherwood, 1963). A detailed understanding of transport phenomena is required to optimize their performance. The basic concepts related to biocatalytic reactions, in terms of the kinetics and mass transport phenomena involved, have to be introduced in order to formulate detailed mass balance within the systems. Furthermore, in order to predict the performance of a membrane bioreactor, a detailed analysis of the effectiveness of the biocatalysed processes is necessary. Simulation of the behaviour of membrane bioreactors has been carried out with the aim of presenting an overview of this very relevant field of biocatalysis applications. © Woodhead Publishing Limited, 2013

22

Handbook of membrane reactors

(b)

Unreacted substrate (product)

V(t) Hollow fibres membrane bioreactor

Substrate (product)

Pump Sample P (S) Product (substrate)

E E

E

E E

E

E

Spongy layer (Zone 3)

E E

E

E

E

E

E

E E E

E E

E

E E

E E E

E E

E

E

E E

E E

E E

S (S) E

E E

E

E

E

E E

E

E E

E

E

E

E

E

Dense layer (Zone 2)

E

E

E

E E

E

E

E

E E

E

Membrane lumen (Zone 1)

1.8 Continued

By means of this analysis more detailed mathematical models could be formulated, applied and solved with little mathematical effort. The results can also be extended to preliminary analysis of more complex systems. The modelling and simulation can be related to dimensionless variables and parameters and might be generalized for a large number of applications. The mechanisms of enzyme catalytic actions are much more complex than those used to describe traditional catalysis, since enzyme configuration itself is strongly affected by the reaction environment and by the presence of specific substrates. Experimental investigations, as well as theoretical analyses, confirm the existence of a substrate−enzyme complex. The substrate binds to the enzyme active site, where reaction occurs and where the products are released. In order to promote this binding, the enzyme protein chain requires certain properties (Bernhard, 1968; Dixon and Webb, 1979; Lenhinger, 1975). An enzyme can enhance the reaction rate where it is capable of holding two substrates close to one another, known as the

© Woodhead Publishing Limited, 2013

Engineering aspects of membrane bioreactors

23

proximity effect. An advantageous approach angle (orientation effect) can also improve the reaction; moreover, for some enzymes the binding with the substrate can slightly change the enzyme shape to favour the reaction, known as induced fit. A great number of enzymatic reactions can be described by the well-known Michaelis–Menten equation rate, which was initially proposed by Henri, and then improved by Briggs and Haldane (Bailey and Ollis, 1986; Laidler and Bunting, 1973). The proposed kinetic rate Rk for the reaction of bioconversion: S

E

P

relative to a general reaction scheme: k

2 ⎯ 1⎯ →(ES) ⎯k⎯ S E← →P + E k −1

in which S represents the substrate, E the free enzyme and P the product, whereas (ES) is the substrate–enzyme complex, is the following: Rk =

RMAX S k2 E0 ⋅ S = KM S KM S

[1.1]

where Rk and RMAX are the actual and the maximum reaction rates, generally expressed as mol/(m3 s) or as g/(m3 s), S is the substrate concentration, KM is the Michaelis constant and has the same dimensions of S, that is, mol/m3 or g/m3, E0 is the total enzyme concentration and is expressed as the moles of enzyme/m3 or as the grams of enzyme/m3. The kinetic constant RMAX is equal to: RMAX

k2 ⋅ E0

[1.2]

where E0 is the total enzyme concentration and is expressed as the moles of enzyme/m3 or as the grams of enzyme/m3. It is evident that the constant k2, (mass of substrate/mass of enzyme)/ time, represents an inverse measure of reaction time. In k2 the dimensions of the substrate and enzyme concentration can be expressed as mass/volume or mol/volume. This uncertainty is dependent on the difficulty, for some applications, to know the exact molecular weight of the compounds involved; in such cases a mass balance has to be applied. The constant KM can be seen as a measure of the enzyme–substrate affinity; a high value of KM corresponds to a low affinity between enzyme and substrate, whereas a low value denotes high affinity.

© Woodhead Publishing Limited, 2013

24

Handbook of membrane reactors

Reaction rate (RK)

1

RMAX

0.8

0.6

RMAX/2 0.4

Slope = RMAX /KM

0.2

0 0

KM

0.4

0.8

1.2

1.6

2

Substrate concentration (S )

1.9 Reaction rate Rk as a function of substrate concentration S as expressed by the Michaelis–Menten kinetic equation.

An implicit relationship between substrate concentration and time can be determined analytically, as in a batch reactor, as follows: RMAX t = (S0

S) + K M ⋅ ln

S0 S

[1.3]

In Fig. 1.9 schematic representation of an enzyme reaction rate modelled by the Michaelis–Menten equation is shown; it can be observed that when Rk = RMAX, saturation condition is reached, so that a further increase in the amount of substrate does not correspond to an increase in the reaction rate. Other useful considerations may be drawn from comparisons of the KM values with the substrate concentration values S. When S > KM Equation [1.1] becomes: Rk

[1.5]

RMAX

a zero-order reaction rate applies corresponding to the asymptotic value of RMAX. When S equals KM the reaction rate becomes: Rk =

RMAX 2

[1.6]

Similar considerations apply to analyse more complicated kinetic mechanisms that are to be formulated when two or more substrates are involved in the reaction or when inhibition phenomena also have to be accounted for (Bailey and Ollis, 1986; Calabrò et al., 2009a; Laidler and Bunting, 1973; Lenhinger, 1975; Ricca et al., 2009a).

1.5.1

Inhibition of enzyme reactions

The enzyme reaction rate is often affected by the presence of various chemicals and ions. Enzyme inhibitors combine, either reversibly or irreversibly, with enzymes and cause a decrease in enzyme activity. Instead, effectors control enzyme reactions by combining with the regulatory site(s) of enzymes. There are several mechanisms of reversible inhibition and for the control of enzyme reactions (Katoh and Yoshida, 2009). Competitive inhibition An inhibitor competes with a substrate for the binding site of an enzyme. As an enzyme–inhibitor complex does not contribute to product formation, it decreases the rate of product formation. Many competitive inhibitors have steric structures similar to substrates, and are referred to as substrate analogues. Product inhibition is another example of such an inhibition mechanism of enzyme reactions, and arises due to structural similarity between the substrate and the product. The mechanism of competitive inhibition, in a uni-molecular irreversible reaction, could be described with the following kinetic rate equation of product formation: Rk =

RMAX S R S = MAX ap ⎛ I ⎞ S KM S KM ⋅ ⎜ 1 + ⎟ K I ⎠ ⎝

© Woodhead Publishing Limited, 2013

[1.7]

26

Handbook of membrane reactors

where S, E, I, and P designate the substrate, enzyme, inhibitor, and product, respectively and KI is the equilibrium constant of the inhibition reaction and is called the inhibitor constant. In comparison with Equation [1.1] for the reaction without inhibition, ap the apparent value of the Michaelis constant K M increases by (KM I)/KI, and hence the reaction rate decreases. At high substrate concentrations, the reaction rates approach the maximum reaction rate, because a large amount of the substrate decreases the effect of the inhibitor. Non-competitive inhibition In non-competitive inhibition, an inhibitor is considered to combine with both an enzyme and the enzyme–substrate complex. Thus, a further reaction is added to the competitive inhibition mechanism and, assuming that the equilibrium constants of the two inhibition reactions are equal in many cases, the following rate equation can be obtained by the Michaelis–Menten approach: Rk =

Rap ⋅ S RMAX RMAX S = = MAX ⎛ ) KM ⎞ ⎛ I ⎞ I ⎞ ( + ⎛ ⎜ 1 + S ⎟ ⋅ ⎜ 1 + K ⎟ ( S KM ) ⋅ ⎜ 1 + K ⎟ ⎝ ⎠ ⎝ I ⎠ I ⎠ ⎝

[1.8]

In comparison with Equation [1.1] for the reaction without inhibition, the ap apparent value of the maximum kinetic rate, RMAX decreases by I/KI, and hence the reaction rate decreases. Uncompetitive reaction For the case of uncompetitive inhibition, where an inhibitor can combine only with the enzyme–substrate complex, the rate equation is given as: Rk =

Rap S RMAX S = MAXapp ⎛ ⎞ S KM I S ⋅⎜1+ ⎟ + KM K I ⎠ ⎝

[1.9]

In comparison with Equation [1.1] for the reaction without inhibition, both ap the apparent values of the maximum kinetic rate, RMAX and Michaelis conap stant KM decrease by I/KI, with different effect on the reaction rate. The substrate inhibition, in which the reaction rate decreases at high concentrations of substrate, follows this mechanism.

© Woodhead Publishing Limited, 2013

Engineering aspects of membrane bioreactors

1.5.2

27

Enzyme deactivation

A further aspect that requires some comment is enzyme deactivation (Joly, 1965; Ricca et al., 2009b). A simple model can be used to approximate the deactivation rate of enzymes to a first-order reaction, in order to predict the time of the course of deactivation, as follows: Ea = e −kkd t E0

[1.10]

where Ea is the active enzyme (supposing it is subjected to an irreversible structural or chemical change leading to an inactive form), E0 is the initial enzyme concentration and kd is the deactivation constant (1/time). The Equation [1.1] is to be consequently modified, thus accounting for enzyme activity decay: Rk =

k2 (E0 ⋅ e k t ) S KM S

[1.11]

Figure 1.10 is a schematic representation of the effect of enzyme decay on the reaction rate is as the active enzyme concentration decreases, lower values of the maximum rate and a slope decay can be observed.

1.6

Transport phenomena and the effectiveness of immobilized biocatalysts

For the sake of simplicity, in such an overview of transport phenomena involved in membrane bioreactors, the analysis will be focused on only two of the most common cases of physical significance: • •

Biocatalyst immobilized onto the membrane surface and Biocatalyst immobilized within the membrane polymeric structure.

1.6.1

Biocatalyst immobilized onto the membrane surface

In this case, immobilization is carried out either by physical adsorption or by the establishment of a chemical bond between the enzyme and the membrane material, promoted by extremely active bridge molecules. The scheme of such immobilization is depicted in Fig. 1.11.

© Woodhead Publishing Limited, 2013

Handbook of membrane reactors 1

0.8

Reaction rate (RK)

28

0.6

0.4 Active enzyme concentration, Ea(t ) 0.2

0 0

0.4

0.8

1.2

1.6

Substrate concentration (S )

1.10 Reaction rate as a function of substrate concentration when an enzyme activity decay is observed, based on Equation [1.11].

Biocatalyst

Membrane

Sb S P Pb

U δf

1.11 Schematic of biocatalyst immobilized onto the membrane surface. Qualitative substrate and product concentration profiles are reported in order to show external mass transfer resistances in the film of thickness δ, when solution flows with velocity U.

© Woodhead Publishing Limited, 2013

2

Engineering aspects of membrane bioreactors

29

As far as the mass transfer is concerned, the enzyme reaction process can be modelled in terms of the film theory combined with the description of a surface reaction. In other words, it is assumed that substrate mass transfer resistance is concentrated in a thin film adjacent to the membrane surface where enzyme molecules have been entrapped. The substrate diffuses through the film and reacts on the membrane surface. Equation [1.1] can also be used in this case with the assumption of the Michaelis–Menten rate equation in which RMAX is defined as the amount of substrate that reacts per unit support volume and unit time: RMAX

k2 ⋅ Em,V

[1.12]

in which Em,V is the amount of enzyme immobilized on the support, in mass (or moles)/(m3 of support). A steady-state mass balance equation on the substrate can be written, assuming that all the substrate reaching the membrane surface reacts: ks a ⋅ ( Sb S ) =

RMAX S KM S

[1.13]

where S is the substrate concentration in the bulk and a is the specific surface (m−1): a=

Am Vm

[1.14]

defined as the ratio of the membrane surface Am and the membrane volume Vm; ks (m/s) is the mass transfer coefficient in the liquid film that might be estimated by means of the well-known semi-empirical correlations (Bird et al., 1960; Schlichting, 1960): jD =

k Sh 23 23 ⋅ ( Sc ) = s ⋅ ( Sc ) = f (Re) Colburn factor Re ⋅ Sc U

where, in this case: Sh = Sc =

lC ks Sherwood number Diff

ν Schmidt number Diff

(

(

) )

© Woodhead Publishing Limited, 2013

30

Handbook of membrane reactors Re =

lC U Reynolds number ν

(

)

in which lC is the process characteristic length (e.g., the diameter in a tubular system or the film thickness or the length of a flat-sheet module, etc.). Equation [1.13] can be rewritten in terms of the following dimensionless variables:

χ=

S dimensionless substrate concentration Sb

Da =

θ=

RMAX Damköhler number (dimensionless) (ks a) ⋅ Sb

KM saturation factor ( Sb

)

[1.15]

[1.16]

[1.17]

thus becoming: 1− χ χ = Da χ +θ

[1.18]

The physical meaning of the Damköhler number refers to a comparison of the relative significance between the maximum reaction rate RMAX and the maximum mass transfer rate (ks·a·S0). It is obvious that: Da > 1 corresponds to a diffusion-limited regime (i.e., extremely fast reaction). The dimensionless saturation factor represents a measure of enzyme saturation and allows for the approximation of the Michaelis–Menten rate, either to a first order (θ >> 1) or to zero order (θ > 1), where the reaction rate approaches the first order. The effect of mass transfer limitation on the overall process can also be expressed in terms of the effectiveness factor, relating the actual reaction rate to that observed if the catalyst surface were exposed to the same

© Woodhead Publishing Limited, 2013

Engineering aspects of membrane bioreactors

31

substrate concentration Sb present in the bulk solution (Curcio et al., 2000, 2006a), that is, without any diffusive resistance outside the enzyme molecule. =

Actual reaction rate Reaction rate in absence of diffusional resistances

[1.19]

In this case, the effectiveness factor η can be defined as:

η=

Rk (

Rk (

) ⋅ Am

=

) ⋅ Am

(RMAX ⋅ S) / (K M + S) (RMAX ⋅ Sb ) / (K M + Sb )

that, in dimensionless form and coupled with Equation [1.18] becomes:

η=

χ

(

θ)

χ θ

=

(

+θ )

Da

⋅( − χ )

[1.20]

It can be demonstrated that for a reaction controlled by kinetics (Da > 1), the reaction is to be considered so fast that substrate concentration immediately vanishes. The effectiveness factor, in this case, can be calculated by simplifying the equation as follows:

η=

1+ χ Da

[1.22]

that tends to zero as Da tends to infinity. The process rate RP is simply given by RP ks ⋅ a Sb or in dimensionless variables: RP 1 = RMAX Da

[1.23]

The process rate, RP, in this case is completely independent to the reaction rate parameters, RMAX and KM.

© Woodhead Publishing Limited, 2013

32

Handbook of membrane reactors

The region, where the process is reaction-rate limited, is not strictly affected by saturation factor θ and the effectiveness factor η approaches the unit value. In the diffusive control region, a linear dependence between η and Da can be observed in a log-log plot: log

l g ( + χ ) − log Da log

[1.24]

The saturation factor θ plays a significant role and tends to enlarge the region where kinetics controls the process. In the above conditions, the minimization of the external concentration gradients is required. By a proper choice of system fluid-dynamics, provided that the enzyme−membrane bonds are strong enough to avoid biocatalyst removal by shear.

1.6.2

Biocatalyst immobilized within the membrane polymeric structure

When the biocatalyst is immobilized on the internal surface of a porous support or entrapped in a porous matrix, the substrate has to diffuse in order to come into contact with the enzyme. In this case, the observed rate of substrate consumption requires the evaluation of the concentration profile within the matrix by a steady-state mass balance referred to the thin membrane section where the biocatalyst is actually immobilized. An effective diffusion coefficient, Deff, is to be introduced to properly consider the actual diffusion rate through the support. The substrate effective diffusion coefficient, Deff-S, might be evaluated, for example, as:

Deff-S

εp Kp DS0 ⋅ ⋅ τ p Kr

⎛ ⎞ r ≅ ⎜ 1 − substrate ⎟ Kr ⎝ rpore ⎠

Kp

4

where DS0 is the diffusivity of S in the bulk liquid, r the radius. The porosity εp is, generally, determined by proper experimental analysis. The tortuosity factor τp ranges from 1.4 to 7.0. The reaction rate of an immobilized enzyme exhibit a behaviour similar to that of enzyme in its native state, but with different kinetic parameters, assuming that the local intrinsic reaction kinetics of the immobilized biocatalyst can be expressed by the Michaelis–Menten equation, Equation [1.1] where: MAX

Eimm ⋅ ρ p ⋅ qE , imm

© Woodhead Publishing Limited, 2013

[1.25]

Engineering aspects of membrane bioreactors

33

Membrane Biocatalyst

z z+dz z

Sb

Concentration profiles

Pb

δ

δ

1.12 Schematic of a porous membrane of thickness 2δ, with enzyme immobilized in the pores. In the figure the thin layer of thickness dz is evidenced. Qualitative concentration profiles of both substrate S and product P are also reported.

where Eimm (molimmobilized enzyme/gsupport) is the amount of immobilized enzyme loaded on the support, qE,imm (molconverted substrate/(molimmobilized enzyme·s)) is the immobilized specific activity of immobilized enzyme, which corresponds to k2 in homogeneous conditions and ρp (gsupport/m3support) is the particle density. As a consequence, RMAX has the following units: (molconverted substrate/ (s·m3support)). A mass balance within the support is, therefore, necessary to account for the effects related to system geometry. Assuming that the enzyme is immobilized within a pellet, two geometrical configurations of support can be taken into account: a slab and a sphere, as shown in Figs 1.12 and 1.13. It is useful to calculate the effectiveness factor η that is a measure of the correct use of the biocatalyst. Effectiveness factor represents the ratio between the actual kinetic rate, when transport resistance actually occurs, and the kinetic rate that would have been observed if all the substrate were exposed to the enzyme, without any transport resistance (Aris, 1965; Bailey and Ollis, 1986; Perry and Geen, 1997; Satterfield and Sherwood, 1963). It is estimated as:

η=

RP obs Rk ( )

[1.26]

where RPobs is the actual (observed) reaction rate and Rk(Sb) is the reaction rate when the concentration S is equal to that in the bulk, Sb:

© Woodhead Publishing Limited, 2013

34

Handbook of membrane reactors Biocatalyst

r r+dr

R

1.13 Schematic of a porous support of radius R, with enzyme immobilized within the pores. A thin spherical shell of thickness dr is shown.

Rk ( Sb ) =

(E

p

KM

) ⋅ Sb

qE

[1.27]

Sb

Slab geometry (Fig. 1.12) – first-order reaction (S > 1) In this case, the equation rate [1.1] can be rewritten as: Rk =

(

E RMAX ⋅S = KM

p

qE

KM

)⋅S = K

cat

S

[1.28]

where the kinetic constants have been all grouped in the constant Kcat. The mass balance becomes: Defff

d2S S

dz2 d

= Kcat S

[1.29]

Introducing the dimensionless variables:

χ=

ζ=

φ

S dimensionless substrate concentration Sb z dimensionless z-coordinate δ

Kcat δ= Defff − S

RMAX δ2 ⋅ Thiele modulus analogue ( dimensionless) K M Defff − S [1.30]

© Woodhead Publishing Limited, 2013

Engineering aspects of membrane bioreactors

35

The dimensionless concentration profiles might be expressed as:

χ=

cosh(φ ς ) cosh(φ )

[1.31]

and the observed reaction rate might be calculated as: RP obs =

Ap Vp

Defff − S

dS dz d

z= δ

The dimensionless substrate concentration profile in a porous membrane, where the biocatalyst is entrapped, depends on the Thiele modulus φ. This parameter is widely adopted in heterogeneous catalyst, since it allows for the estimation of the penetration depth within the support and can also be used to identify the mechanism that controls the process rate. The Thiele modulus can be interpreted as a ratio between a diffusion time (δ2/Deff-S) and a kinetic time (KM/RMAX) or, equivalently, as a ratio between a characteristic kinetic rate VMAX/KM ΔS = Kcat ΔS and a merely diffusive transport mechanism, characterized by (Deff−S/δ) × (ΔS/δ). When φ > 1, the greater part of the enzyme is not effectively used since concentration gradients actually develop in a thin region close to the outer surface and tend to rapidly vanish. Diffusive transport controls the process rate since kinetics is much faster than the mass transfer by diffusion. This aspect is evidenced by the evaluation of effectiveness factor η, which is estimated using the definition (Equation [1.26]), introducing the dimensionless substrate profile χ (Equation [1.31]):

η=

1 Defff − S ⋅ K M d χ ⋅ ⋅ RMAX dζ δ2

ζ=1

=

1 φ i h (φ) φ 2 cosh ( φ )

the effectiveness η is calculated as:

η=

tanh ( φ )

[1.32]

φ

Thiele modulus only directly affects the process effectiveness factor when high values of Thiele module are chosen, so determining a

© Woodhead Publishing Limited, 2013

36

Handbook of membrane reactors

diffusion-controlled process. The effectiveness factor only approaches unity when low values of Thiele module occur; this corresponds to a good utilization of the enzyme. No influence is observed if the saturation factor changes, provided that the reaction rate can be approximated by a firstorder mechanism (θ >> 1). External and internal resistances When the substrate is first transported in a boundary layer surrounding the particle, before diffusing within the catalyst support where reaction occurs, external resistance needs to be considered (Calabrò et al., 2008; Truskey et al., 2004). An example is the case of a packed bed bioreactor, where fluiddynamics play a significant role in the optimization of system performances. In such a case the kinetic contribution has to be expressed in terms of overall effectiveness factor ηov. To estimate it, the mass balance, Equation [1.29], has to be solved by imposing the continuity of mass flux at the wall. For a flat-sheet support it corresponds to: −Defff −S

dS dz d

z =δ

= ks ⋅ ⎡⎣S (

) − Sb ⎤⎦

where ks is the mass transfer coefficient from the bulk solution to the support, which can be estimated by means of semi-empirical correlations (Bird et al., 1960; Perry and Geen, 1997; Satterfield and Sherwood, 1963; Schlichting, 1960). To solve the first-order differential equation in a flat geometry, another dimensionless parameter, that is, the Biot number, Bi, is introduced: Bi =

ks ⋅ δ Defff −S

Bi represents a comparison between a characteristic transport rate occurring outside the support and a characteristic diffusion rate occurring within the support where the biocatalyst has been immobilized. The overall effectiveness factor ηov becomes:

ηov =

tanh ( φ )

(

⎡ φ φ ⋅ ⎢1 + ⎢⎣

Bi

( φ )) ⎤ ⎥ ⎥⎦

© Woodhead Publishing Limited, 2013

[1.33]

Engineering aspects of membrane bioreactors

37

that might be rewritten as: 1 1 φ2 = + ηov η Bi

[1.34]

in which η is the effectiveness factor when no external resistance occurs. Spherical geometry (Fig. 1.13) – first-order reaction, (S 1, the transport mechanism follows Knudsen diffusion. 0.01 Tp. These temperatures reflect a significant reduction in the net driving force of the mass transfer. An important effect making the MD process different from traditional heat exchanges, the temperature polarisation and concentration polarisation occur in the membrane wall due to the transfer of both water vapour and latent heat. As previously stated, the heat and mass transfer across the membrane move from the hot feed stream to the cold permeate one. The temperature gradients cause a difference in temperature between the liquid−vapour interfaces and the bulk temperatures on both sides of the membrane. This effect, in membrane science called temperature polarisation, reduces the water vapour flux and in literature it is measured by the so-called temperature polarisation coefficient (τ), given by: τ = (Τfm − Τpm)/(Tf – Tp)

© Woodhead Publishing Limited, 2013

[2.15]

Membrane contactors

87

Heat flux Mass (vapour) flux δp

Tf

Permeate

Tfm Tpm

Tp

Feed Pfm

δf

Ppm Membrane

2.16 Temperature profile and temperature polarisation in MD.

where Tfm and Tpm are the temperatures at the hot and cold membrane surfaces, respectively, and Tf and Tp the temperatures in the feed and permeate bulk solutions, respectively (Fig. 2.16). The value of τ approaches unity in well-designed MD systems that are mass transfer limited, whereas it approaches zero in poorly designed MD systems which are limited by the transfer of heat through the two cited boundary layers. The concentration polarisation is quantified using the following coefficient (ζ): [2.16]

ζ = Cfm/Cf

where Cfm is the salt concentration at the hot membrane surface and Cf the bulk concentration in the feed bulk solution. These two parameters, τ and ζ must be taken into account when quantifying and explaining the corresponding reduction in the driving force. See, for example, details of the work published by Martinez-Diez and VasquezGonzales (1999) in which the solute concentration Cm1 at the feed solution/ membrane interface is expressed as: Cm1 = Cb1 exp(J/ρ Ks)

[2.17]

where Ks is the solute mass transfer coefficient, Cb1 the solute concentration in the bulk solution, J the mass flux and ρ the liquid density. In their experimental work, Ks was evaluated using the mass transfer analogy of the Graetz–Léveque equation: Sh = 1.86 (Re Sc de/L)0.33

© Woodhead Publishing Limited, 2013

[2.18]

88

Handbook of membrane reactors

Temperature polarisation means that the temperature near the membrane wall is lower than the main flow temperature, because water vapour absorbs the heat of the aqueous solution and is transferred across the membrane. Similarly, concentration polarisation means that the concentration near the membrane wall is higher than the main flow concentration, because of water vapour transfer across the membrane. These resultant concentration and temperature gradients across the membranes determine a reduction of the vapour pressure gradient, which in turn results in a reduction of the driving force of mass transfer, and so a reduction of the permeate flux. Sometimes this temperature polarisation is reflected in a decrease of the driving force by more than 50% (Susanto, 2011). It is important to observe that such a polarisation cannot be avoided, but only minimised. To proceed in this direction, the most important parameters influencing heat transfer must be identified: •

Heat transfer in the liquid phase is influenced by operating conditions, fluid property and hydrodynamic conditions. • The latent heat for vaporisation determines the amount of vapour to be transported from the feed to the permeate side: increasing heat for vaporisation increases permeate product. Nevertheless, the amount of latent heat depends on the extent of temperature polarisation and heat conduction (which is heat loss, thus should be minimised). This heat loss is influenced by membrane porosity and membrane thickness, and can be reduced by increasing the membrane thickness. However, such an increase results in a decrease of the resulting mass transfer. As such, this trade-off phenomenon could be solved by identifying and using an optimised thickness. • Another important parameter to be considered is related to the hydrophobic character of the membrane. The water vapour flow rate through the hydrophobic pores of the membrane is limited by the membrane intensity and by the LEP of water (Khayet et al., 2003; Wang et al., 2009), as previously explored in depth. In the case of RO, for example, concentration polarisation is a significant problem because the osmotic effect does not permit concentrations of interest to be obtained. On the other hand, the concentration polarisation in MD does not produce the same level of negative effect, and so starting from highly concentrated feeds, a very high quality of pure water can be obtained. This situation is reversed when temperature polarisation is considered. MD is a temperature-driven technique, and as such the phenomenon significantly influences the amount of water flux through the membrane. It is important to recognise that a reduction of the temperature profile at the membrane side decreases the effect of the temperature polarisation (Drioli et al., 1999).

© Woodhead Publishing Limited, 2013

Membrane contactors

89

The contribution of the thermal diffusion towards mass transfer was found almost negligible in MD (Banat and Simandl, 1994).

2.6

Applications of membrane distillation in membrane bioreactors

Ethanol production from biomass using membrane bioreactors (MBRs) is considered a plausible option for the production of alternative liquid fuels. It is therefore interesting to consider the application of the MD technique to ethanol productivity, coupled with a fermentation MBR. Gryta et al. (2000) combined batch fermentation with the removal of ethanol from the broth by means of the MD process. To separate volatile compounds from the feed (broth), formed as a result of fermentation, they used a porous capillary polypropylene membrane with the following characteristics: inner diameter (id) 1.8 mm and outer diameter (od) 2.6 mm, pore sizes with a nominal and maximum diameter of 0.22 and 0.6 mm, respectively, porosity about 73%, and effective membrane area 490 cm2. The distillate temperature was 293 K, while the fermentation was performed at 303 (conventional process) and 309 K (MBR). The bioreactor was connected through a pump with a module for MD. The best results correspond to the following factors: •

bioreactor combined with MD: efficiency of 0.40–0.51 gEtOH/gsugar production rate of 2.5–40 gEtOH/dm3·h

•

classical batch fermentation (i.e., without any removal of the products): efficiency of 0.35–0.45 gEtOH/gsugar production rate 0.8–2.0 gEtOH/dm3·h.

The ethanol flux obtained in MD varied in the range of 1–4 kgEtOH/m2 per day, and was dependent on the temperature and the feed composition. Following the conclusions of the authors, these results allow us ‘to perform the fermentation of concentrated sugar solutions, which normally lead to a significant increase of the ethanol concentration in the classical reactors’. More recently, in order to overcome the disadvantages of conventional fermentation systems, Lewandowicz et al. (2011) used a continuous fermentation process based on a bioreactor coupled with an MD unit. The MD technique for removing bioethanol from a reactor is able to maximise the volumetric productivity, and thus minimise production costs in the biofuel industry. Moreover, the combined system also facilitates a reduction in yeast stress, which commonly occurs in industrial fermentation. The experimental campaign was carried out using a module with 40 microporous hydrophobic

© Woodhead Publishing Limited, 2013

90

Handbook of membrane reactors

capillary membranes of PP, with i.d. 1.8 mm and o.d. 2.6 mm, pore size 0.2 µm, and effective membrane area 0.1 m2. The experimental results confirm that MD can be considered as a straightforward method for use in increasing ethanol production by decreasing glycerol synthesis levels, as well as for increasing the number and viability of yeast cells. According to the results of the authors, when compared to other membrane processes (for example UF and RO), MD is more difficult to apply on an industrial scale, due to some serious engineering problems that include module design and heat loss during processing, leading to uncertain economic costs. In a recent work, Phattaranawik et al. (2008) described a novel process using a membrane distillation bioreactor (MDBR) for the treatment of wastewater. This process used a combination of elements from both the MD process and the wastewater biological process, such as activated sludge, in which the MD module is submerged in an aerobic bioreactor. Water is removed from the bioreactor by a thermally driven process across the hydrophobic membrane in the MD module. The mixed liquor in the bioreactor is in direct contact with the submerged membrane on the feed side, and the permeate solution (clean water) is recirculated through the inner (lumen) side of the membrane. Product permeate is collected in a tank. The benefits of the process include the production of high quality water (TOC levels below 1 p.p.m. and negligible salts), independence of both organic and hydraulic residence times, fluxes in the range 2–5 L/m2·h and operation at atmospheric pressure. Furthermore, only a modest primary energy demand is required. Among the various potential applications, there is also the possibility of ‘one step’ wastewater reclamation. The authors also showed a qualitative economic comparison between two other membrane systems, including a hybrid of MBR with RO and MBR with UF, and MDBR. Their conclusion was that MDBR would produce enhanced capital and more efficient operating costs when compared with the other two hybrid membrane based processes.

2.7

Osmotic membrane distillation (OMD)

Osmotic distillation, also known as MD, osmotic evaporation, or direct OMD is a membrane contactor technique in which a microporous hydrophobic membrane separates two aqueous solutions of different solute concentrations (osmotic pressure). The result of the water transfer through the membrane (in the form of vapour) is the concentration of feed, and the dilution of the osmotic agent solution. Working at ambient temperature and pressure, the energy consumption in OMD is much lower than in RO. An important advantage of OMD, in comparison to MD, is that its lowest operating temperature avoids the degradation of heat sensitive components during the concentration of food and pharmaceutical products: colour, flavour

© Woodhead Publishing Limited, 2013

Membrane contactors

91

and other essences are retained and, in addition, taste is not altered (Kunz et al., 1996). In many cases, the concentration level obtained by evaporation is relatively high, but the use of the OMD technique avoids any thermal damage or loss of the solutes. The OMD process works in this way: a porous hydrophobic membrane (generally PTFE, PVDF, or PP), is in contact with two different non-wetting aqueous solutions; for example a juice solution and a concentrated salt solution (stripping solution or extractant). The different water activity of the two solutions corresponds to two different water vapour pressures. Due to the combination of hydrophobicity and the narrow pore size of the membrane, neither the first solution nor the extractant passes through the membrane pores. Only water vapour is allowed to diffuse, and the water vapour pressure difference between the two sides of the membrane constitutes the driving force of the process. As stripping solutions, inorganic salts such as sucrose, NaCl, CaCl2, MgCl2, MgSO4, K2HPO4, and KH2PO4 or organic solvents (e.g., glycerol and polyglycerol) are generally used. The choice of the salt solution reflects the boundary layer effect; at the same bulk pressure difference, different salts generate different fluxes (Kunz et al., 1996). Moreover, the stripping solution must have the following specific characteristics: high solubility in water, high superficial tension, low volatility and viscosity, and non-toxicity. Potassium salts of ortho- and pyro-phosphoric acid are generally preferred, for example, because they offer a low equivalent weight, high water solubility, steep positive temperature coefficients of solubility, and are safe for use in foods and pharmaceuticals. The OMD process provides the great advantage of working at both constant (and low) temperature and atmospheric pressure, with a consequent reduction in both thermal and mechanical damage. These gentle operating conditions are very important, especially in the concentration of heat sensitive materials in biomolecules, in natural colours, or in food processing (fruit juice production, for example).

2.7.1 Transport mechanisms To evaluate the vapour flux across the membrane, depending on the Knudsen number, two transport models are considered: •

Kn < 1. Poiseuille capillary model: the vapour molecules collide more frequently with each other than with the pore walls. The flux (Jp) is proportional to the square root of the pore radius: Jp = Kpois·r2·pav·(pf – pp)

© Woodhead Publishing Limited, 2013

[2.19]

92

Handbook of membrane reactors

where Kpois is the Poiseuille parameter (depending on temperature, vapour viscosity, membrane porosity, molar weight of the solvent and pore length), r the radius of the membrane pore, pav the mean vapour pressure in the membrane, and pf and pp the respective pressures at both sides (feed and permeate) of the membrane. •

Kn > 1. Knudsen diffusional model: the vapour molecules collide more frequently with the pore walls than with each other. The flux (JK) is proportional to the pore radius: JK = KK·r·pav·(pf – pp)

[2.20]

where KK is the Knudsen parameter (depending on temperature, membrane porosity, molar weight of the solvent and pore length). Both the Poiseuille and Knudsen models are generally valid in the absence of air in the pores of the membrane, and both show a large dependence of membrane flux on pore radius. In these three models the vapour flux is related to the total pressure difference between the two sides of the membrane and the membrane porosity. A third model is also used to describe the water vapour flux, and is related to the diffusion of vapour through the ends of the pores. In this model the interaction between molecules and wall are not considered, and the flux (JD) is generally given by: JD = KD·(pf – pp)

[2.21]

where KD is the diffusion parameter (depending on temperature, vapour diffusion coefficient, membrane porosity, molar fraction of air, molar weight of the solvent and pore length). All three models propose a linear relationship between the vapour flux and the vapour pressure difference, so Kunz et al. (1996) proposed a general expression for describing the mass flux. This is related through the overall mass transfer coefficient, which accounts for all three resistances (feed, membrane and stripping) of water vapour transport to the driving force. In OMD, flux is inversely proportional to the membrane thickness, so it must be as thin as possible (0.1–1.0 µm). Furthermore, in the case of OMD, a general mass transfer correlation like Equation [2.7], in which the Sherwood number is a function of the Reynolds and Schmidt numbers, can be written. In contrast to PV and RO, in OMD the vapour flux is not influenced by any physico-chemical interaction between the liquid in the feed and the membrane. It is only physical parameters of the membrane, such as pore size, porosity, tortuosity, thickness and waterproofness, which influence the flux.

© Woodhead Publishing Limited, 2013

Membrane contactors

93

2.7.2 Temperature and concentration polarisation effects Even if the bulk temperature of the two solutions is the same, OMD is not purely a mass transfer operation like direct osmosis, as heat transfer must be also taken into account (Thanedgunbaworn et al., 2009). In fact, the water vapour transport through the pores of the membrane implies water evaporation at the (dilute) feed solution side, and condensation at the extract (osmotic agent) solution side. The process of evaporation cools the feed side, whereas the condensation warms up the extract side. In other words, a temperature difference and, therefore, a temperature gradient through the membrane, are also created. As a consequence, during the water vapour transport a simultaneous mass and heat transfer occurs, which causes a difference in both the concentration and temperature near the two membrane surfaces in comparison to the two bulk streams. The coupled heat and mass transfer is very complicated and thus non-equilibrium studies based on irreversible thermodynamics are generally used to quantitatively analyse the simultaneous heat and mass transfer phenomena (Wang and Min, 2011), but this is beyond the scope of this chapter. These two concentration and temperature gradients significantly affect the performance of the OMD process, as both are directly reflected in a driving force reduction, and thus a reduction of the water vapour flux through the membrane (Ravindra Babu et al., 2008). The main disadvantage of the OMD technique is purely related to its low flux, which limits the full commercial application. It is therefore very important to understand the mechanism for the formation of concentration and temperature polarisation, in order to try to reduce their negative effects. During the mass transfer which occurs during an OMD process, two different boundary (polarisation) layers are created because of the increased solute concentration on the feed side of the membrane surface, and the decreased solute concentration on the opposite side. These two concentration polarisation layers have the effect of reducing the water vapour flux through the pores of the membrane by reducing the driving force. The overall concentration polarisation effect is the sum of these two boundary layers. Furthermore, there is also the driving force reduction instigated by the temperature polarisation effect. To quantify, during OMD operations, the temperature difference across the membrane is generally in the very low range of 0.3–1.5 K, whereas concentration polarisation produces a concentration change of about 0.5–3.2% near the membrane surface (compared to bulk concentration). Many studies are today focused on producing a better understanding of the influence of the concentration and temperature polarisation effects during the OMD process (Ravindra Babu et al., 2008). These authors found that the polarisation phenomenon depends on various parameters, such as the type of osmotic agent, the osmotic agent concentration, and the feed flow

© Woodhead Publishing Limited, 2013

94

Handbook of membrane reactors

rate. In particular, the driving force reduction due to concentration polarisation increases with an increase in the osmotic agent concentration. In contrast, the driving force reduction due to the temperature polarisation effect was, in their experimental conditions, relatively low. To give some data, the reduction in driving force due to the total polarisation effect increased from 52 to 357 Pa, with an increase in osmotic agent concentration. From these results it is evident that the polarisation phenomenon significantly affects the driving force reduction during the concentration of aqueous solutions, such as pineapple juice, by OMD processing. Furthermore, the same authors also showed that at any concentration used in their experiments, the driving force reduction due to the concentration polarisation effect was higher than that of the temperature polarisation effect. For example, at 8 mol/kg, CaCl2 corresponds to 225 Pa of reduction in driving force due to concentration polarisation, whereas the reduction due to temperature polarisation effect is only 75 Pa. The best result obtained by Ravindra Babu et al. (2008) in terms of pineapple juice concentration was up to 62° Brix, preserving the ascorbic acid content of the fruit. To give a comparison of OMD with, for example, MD, in the case of orange juice, Alves and Coelhoso (2006) found that the water flux was more than 50% lower in MD processing simply because of the thermal polarisation effect. Warczok et al. (2007) also found that the viscosity of the feed solution plays an important role. In the experimental conditions they considered, it was the key determining factor that influenced the water flux during the OMD process.

2.8

Membrane crystallisation

Membrane crystallisation is designed to work as an extension of MD (Curcio et al, 2001; König and Weckesser 2005). The basic concept is to use evaporative mass transfer through a microporous hydrophobic membrane in order to concentrate a solution above the supersaturation limits, and induce nucleation and growth of crystals (Drioli and Macedonio, 2010; Zhang et al., 2008). Crystal modification, shape, size and purity depend strongly on the level of local supersaturation, and the use of membranes can effectively limit and control the level of supersaturation during the process.

2.8.1

Approaches to achieve supersaturation in membrane crystallisation processes

Depending on the strategy selected, solvent evaporation and vapour transfer through the membrane pores can be induced by a partial pressure gradient,

© Woodhead Publishing Limited, 2013

Membrane contactors

95

or by instigating a difference of temperature between the two solutions (crystallising solution/stripping solution). In the former case, the osmotic process is induced to establish a difference of water activity between the two streams. The vapour is thus transferred from the dilute solution to the concentrated solution (stripping phase). In the second case, the crystallising solution is heated and the partially evaporated solvent is moved towards the permeate side, where it is condensed by cooling. In membrane crystallisers, the membrane acts as a heterogeneous phase to induce nucleation and growth of crystals by a change in the supersaturation profile in proximity to rough surfaces. This is in order to instigate concentration polarisation, preferential adsorption sites and the establishment of interfacial attractive/ repulsive interactions, enabling the thermodynamic and kinetic drive of the overall process. A high level of supersaturation could give rise to unstable modifications, as a formation of heterogeneous distributions of crystals with a low degree of purity. The control of the supersaturation through limitation of the mass transfer can overcome this drawback, and membranes can be interesting tools in combination with crystallisation technology, allowing precise, quick control of supersaturation levels. Similarly, a well-controlled increase in the level of supersaturation can result in a reduction of the induction nucleation time and, therefore, in a reduction of the crystallisation time. Numerous precipitant and additive types have been utilised to control turbidity and induction time periods, yielding various crystal forms (Zhang et al, 2006). Modulation of the crystal size has also been achieved by inducing massive nucleation through the establishment of attractive interaction between the membrane surface and crystallising solution. It has also been demonstrated that correctly functionalised PVDF membranes can work not only as simple physical barriers, but also as interactive interfaces, inducing protein agglomeration onto the membrane surface and reaching supersaturation conditions in very short time. After 3 h of incubation crystal nuclei were formed, and in less than 24 h regular and micro-sized protein crystals were achieved (Gugliuzza et al., 2006, 2009c). Amphiphilic molecules adsorbed on extensive hydrophobic domains enabled attractive Lifshitz–van der Waals and Lewis acid–base interfacial forces to direct massive nucleation at the membrane–solution interface (Aceto, 2006). As a result, the size of the crystals was modulated from a needle to a more compressed shape by changing the membrane affinity to the protein, for example Lysozime (Fig. 2.17). The decrease in the difference of the solubility parameters (Δδ), calculated from the difference of the respective solubility parameters (δ), was meant as an increase in the reciprocal affinity. A comparable trend was noted for crystal size profile, yielding a useful indication about the attractive choice of using rough hydrophobic surfaces, where binding functional

© Woodhead Publishing Limited, 2013

96

Handbook of membrane reactors 30 28 26 24 100 µm

22 20 0

50 100 150 Modifier loading (mg/cm2)

/w, (-)

Δδ (103 J1/2/m3/2)

200 µm

4 3.5 3 2.5 2 1.5 1 0.5 0 0 20 40 60 80 100 120140 Nucleating agent loading (mg/cm2)

2.17 Increasing affinity of the membrane to protein solution (Δδ), which expresses the reciprocal affinity (a) and formation of protein crystals with decreasing size (w) as a function of modifier loading.

sites are dispersed. The hydrophobic character of the membrane is the reason membrane pores must be prevented from flooding, while the functional sites induce quick agglomeration of crystallising molecules in proximity to the surface. In order to prevent blocking of the pores by crystal deposition, hydrodynamics, temperature and effectively assembled crystalliser plants must be carefully selected.

2.8.2

Controlling factors for crystallisation

Well-controlled crystallisation is the best approach for the preparation of material that is uniform in shape, size, structure and purity. However, the local gradient of supersaturation could be a limiting factor in respect to the uniformity of the product quality, as nucleation and crystal growth depend on the degree of supersaturation. If the combination of laminar flux and low shear stress is expected to produce crystals with good structural lattices (Drioli et al, 2006), it is important to note that the choice of the operating conditions can influence the nucleation rate, as described by the following equation: Rn

Kb mTξ ω e ΔC k

[2.22]

where Rn is the nucleation rate, mT is the concentration of crystals in the magma, ωe is the rotation rate, k is a numerical empirical coefficient and ΔC is the difference of concentration between the mother liquor and solubility value at equilibrium. The constant of the nucleation rate (Kb) ranges from 1 to 2.5, whereas ξ is close to unit if the collision between crystals and vessel prevail over the collisions between crystal and crystal.

© Woodhead Publishing Limited, 2013

Membrane contactors

97

Similarly, the dependence of the growth rate (G) on concentration and temperature can be described by the equation: G

K g ΔC g

[2.23]

where Kg is the growth rate constant and depends on temperature. Drioli et al. (2006) described the crystal growth as sequential steps, including ‘(1) the transport of the molecules from the mother liquor to the crystal surface, (2) adsorption and diffusion over the surface, (3) attachment and diffusion along a step, integration into the crystal at a kink site’. The rate-controlling step for the crystal growth is effectively the diffusion, while a gradient of concentration is generated from the bulk of the mother liquor and crystal surface. The diffusion of molecules through this boundary layer is described by the first and second Fick’s laws. However, the forced flow applied to membrane crystallisers increases convectively the diffusion of the molecules towards crystal growth, rapidly reaching supersaturation of the environment. However, the effective surface area of mass transfer in the membrane crystalliser is very large, and changes in the solvent evaporation rate can be controlled by modifying the difference of temperature between the crystallising solution and stripping solution, and by altering the concentration of the stripping solutions. Membrane crystallisation occupies an important role in the global vision of sustainable integrated membrane processes, offering interesting opportunities in the design, rationalisation, and optimisation of innovative new products. Through a focused combination of thermally driven membrane operations, such as MD and membrane crystallisation for example, it is possible to build up desalination plants, enabling the production of freshwater and the recovery crystals from natural salts, significantly reducing the costs related to polluting brines, and limiting the negative impact on ecosystems. Furthermore, membrane crystallisation also presents a viable and practical route to present molecules of pharmaceutical interest, such as proteins, in the form of perfect crystals. This is an important target for research, as the identification of protein–activity relationships can accelerate the design of new generation drugs, as well as assist in the development of advanced diagnostic tools and effective therapies for the treatment of human diseases.

2.9

Conclusions and future trends

MCs combine properties such as the modular concept of highly specific surface areas from membrane technology with the selectivity of

© Woodhead Publishing Limited, 2013

98

Handbook of membrane reactors

conventional separation processes like extraction or absorption. MCs are flexible, easy-to-integrate technologies which can be adapted to fit a variety of industrial applications, including water treatment, aromatics recovery, selective removal of heavy metals and gas treatment. Increasing demand for high product quality, reduced production costs, limited energy consumption and preservation and management of natural resources has in recent years increased interest in producing hybrids based on this type of technology. These needs are central to the global vision of process intensification, which means more compact equipment size, higher plant efficiency for productive cycles, minimisation of pollutant processes, improved natural resource management, exploitation of renewable energy and strengthening economy based on a high quality human environment. Despite the fact that membrane technology already fulfills the concept of process intensification in some areas, especially in water desalination, fruit juice concentration and the petrochemical industry, effective application of these technology types at an industrial level still needs additional effort addressed at developing and integrating new materials, new design concepts, economics and process control, scale-up and realistic assessment of the basic working parameters on real pilot plants.

2.10

References

Aceto M.C. (2006), Functionalization of hydrophobic microporous membranes: a suitable tool for the protein crystallization in contactor devices. Experimental thesis, ITM-CNR, Rende, Italy. Al-Asheh S., Banat F., Qtaishat M., Al-Khateeb M. (2006), Concentration of sucrose solutions via vacuum membrane distillation, Desalination, 195, 60–68. Alves M.D., Coelhoso I.M. (2006), Orange juice concentration by osmotic evaporation and membrane distillation: a comparative study, J. Food Eng., 57, 153–163. Baker R.W. (2004), Membrane technology and applications, 2nd ed. (R.W. Baker, Ed.), John Wiley & Sons, Chichester, West Sussex, England. Banat F.A., Simandl J. (1996), Removal of benzene traces from contaminated water by vacuum membrane distillation, Chem. Eng. Sci., 51, 1257–1265. Banat F.A., Simandl J. (1994), Theoretical and experimental study in membrane distillation, Desalination, 95, 39–52. Bodell B.R. (1963), Silicone rubber vapor diffusion in saline water distillation, US Patent Application. Boi C., Bandini S., Sarti G.C. (2005), Pollutants removal from wastewaters through membrane distillation, Desalination, 183, 383–394. Bottino A., Roda G.C., Capannelli G., Munari S.M. (1991), The formation of microporous polyvinylidene difluoride membrane by phase separation, J. Membrane Sci., 57, 1–20. Cardea S., Gugliuzza A., Schiavo Rappo E., Aceto M., Drioli E., Reverchon E. (2006), Generation of PEEK-WC membranes by supercritical fluids, Desalination, 200, 58–60.

© Woodhead Publishing Limited, 2013

Membrane contactors

99

Cardea S., Gugliuzza A., Sessa M., Aceto M.C., Drioli E., Reverchon E. (2009), Supercritical gel drying: a powerful tool for tailoring symmetric porous PVDF-HFP membranes, ACS Appl. Mater. Interfaces, 1(1), 171–180. Cardea S., Sessa M., Reverchon E. (2010) Supercritical phase inversion to form drug-loaded poly(vinylidene fluoride-co-hexafluoropropylene) membranes, Ind. Eng. Chem. Res., 49(6), 2783–2789. Charcosset C. (2009), A review of membrane processes and renewable energies from desalination, Desalination, 245, 214–231. Chen N., Hong L. (2002), Surface phase morphology and composition of the casting films of PVDF–PVP blend, Polymer, 43, 1429–1436. Cheng L.P. (1999), Effects of temperature on the formation of microporous PVDF membranes by precipitation from 1-octanol/DMF/PVDF and water/DMF/ PVDF systems, Macromolecules, 32, 6668–6674. Cheng L .P., Young T.H., Fang L., Gau, J. J. (1999), Formation of particulate microporous poly(vinylidene fluoride) membranes by isothermal immersion precipitation from the 1-octanol/dimethylformamide/poly(vinylidene fluoride) system, Polymer, 40(9), 2395–2403. Curcio E., Criscuoli A., Drioli E. (2001), Membrane Crystallizers, Ind. Eng. Chem. Res., 40(12), 2679–2684. Deshmukh S.P., Li K. (1998), Effect of ethanol composition in water coagulation bath on morphology of PVDF hollow fibre membranes, J. Membrane Sci., 150, 75–85. Ding Z., Liu L., Li Z., Ma R., Yang Z. (2006), Experimental study of ammonia removal from water by membrane distillation (MD): The comparison of three configurations, J. Membrane Sci., 286, 93–103. Drioli E., Criscuoli A., Curcio E. (2006), Membrane Contactors: Fundamentals, Applications and Potentialities, Membrane Science and Technology, Vol. 11, Elsevier: Amsterdam. Drioli E., Laganà F., Criscuoli A., Barbieri G. (1999), Integrated membrane operations in desalination processes, Desalination, 122, 141–145. Drioli E., Macedonio F. (2010), Membranes for water treatment, Peinemann, KV, Pereira Nunes, S (Eds), Wiley-CH Verlag GmbH & Co, Weinheim. El-Bourawi M.S., Khayet M., Ma R., Ding Z., Zhang X. (2007), Application of vacuum membrane distillation for ammonia removal. J. Membrane Sci., 301, 200–209. El-Zanati E., El-Khatib K.M. (2007), Integrated membrane-based desalination system, Desalination, 205, 15–25. Feng C., Shi B., Li G., Wu Y. (2004), Preparation and properties of microporous membrane from poly(vinylidene fluoride-co-tetrafluoroethylene) (F2.4) for membrane distillation, J. Membr. Sci., 237, 15. Findley M.E. (1967), Vaporization through porous membranes, Ind. Eng. Chem. Des. Dev., 6, 226–230. Gabelman A., Hwang S.-T. (1999), Hollow Fiber Membrane Contactors, J. Membr. Sci., 159, 61–106. Gaeta S.N. (2003), Membrane contactors in industrial applications, Proc. 1st Italy-Russia Workshop on membrane and membrane processes, Cetraro (I), 51–55. Garcıa-Payo M.C., Izquierdo-Gil M.A., Fernández-Pineda C. (2000), Air gap membrane distillation of aqueous alcohol solutions, J. Membrane Sci., 169, 61–80.

© Woodhead Publishing Limited, 2013

100

Handbook of membrane reactors

Gekas V., Hallström B. (1987), Mass transfer in the membrane concentration polarization layer under turbulent cross flow: I. Critical literature review and adaptation of existing Sherwood correlations to membrane operations, J. Membr. Sci., 30, 153. Gryta M., Tomaszewska M, Karakulski K. (2006), Wastewater treatment by membrane distillation, Desalination, 198, 67–73. Gryta M., Barancewicz M. (2010), Influence of morphology of PVDF capillary membranes on the performance of direct contact membrane distillation, J. Membrane Sci., 358(1–2), 158–167. Gryta M., Karakulski K., Morawski A. (2006), Separation of effluents from regeneration of a cation exchanger by membrane distillation. Desalination, 197, 50–62. Gryta M., Morawski A.W., Tomaszewska M. (2000), Ethanol production in membrane distillation bioreactor, Catal. Today, 56, 159–165. Gu M., Zhang J., Wang X., Tao H., Ge L. (2006), Formation of poly(vinylidenefluoride) (PVDF) membranes via thermally induced phase separation, Desalination, 192, 160. Gugliuzza A., Drioli E. (2007), PVDF and HYFLON AD membranes: Ideal interfaces for contactor applications, J. Membrane Sci., 300, 51–62. Gugliuzza A., Drioli E. (2009a), New performance of hydrophobic fluorinated porous membranes exhibiting particulate-like morphology, Desalination, 240, 14–20. Gugliuzza, A., Aceto, M. C., Drioli, E. (2009c), Interactive functional poly(vinylidene fluoride) membranes with modulated Lysozyme affinity: a promising class of new interfaces for contactor crystallizers, Polym. Int., 58 (12), 1452–1464. Gugliuzza A., Ricca F., Drioli E. (2006), Controlled pore size, thickness and surface free energy of super-hydrophobic PVDF® and Hyflon®AD membranes, Desalination, 200, 26–28. Gugliuzza A., Aceto M.C., Macedonio F., Drioli E. (2008), Water droplets as template for next generation self-assembled poly-(etheretherketone) with Cardo membranes, J. Phys. Chem. B, 112(34), 10483–10496. Gugliuzza A., Speranza V., Macedonio F., Drioli E. (2010), High-performance hydrophobic membranes for contactors and desalination technologies, Proc. of Advances in Science and Engineering for Brackish Water and Seawater Desalination (ECI), May 8–12, Cetraro, Italy, 122–124 Gugliuzza, A., Speranza, V., Trotta, F., Drioli, E. (2009b), Bio-inspired membranes with well-defined channels, Chem. Eng. Trans., 17, 1537–1542. He K., Hwang H.J., Moon I.S. (2011), Air gap membrane distillation on the different types of membrane, Korean J. Chem. Eng., 28(3), 770–777. Huang S., Wu G., Chen S. (2007), Preparation of microporous poly(vinylidene fluoride) membranes via phase inversion in supercritical CO2. J. Membrane Sci., 293, 100–110. Imhof A., Pine D.J. (1997), Ordered macroporous materials by emulsion templating, Nature, 389, 948–951. Iversen S.B., Bhatia V.K., Dam-Johansen K., Jonsson G. (1997), Characterization of microporous membranes for use in membrane contactors, J. Membrane Sci., 130, 205–217.

© Woodhead Publishing Limited, 2013

Membrane contactors

101

Khayet M., Godino M.P., Mengual J.I. (2003), Theoretical and experimental studies on desalination using the sweeping gas membrane distillation method, Desalination, 157, 297–305. König A., Weckesser D. (2005), Membrane based evaporation crystallization, 16th Int. Symp. on Industrial Crystallization ISIC 2005, September 11–14, 2005, Dresden, In: VDI-Berichte 1901.2, 1171–1176 Koonaphapdeelert S., Zhentao W., Li K. (2009), Carbon dioxide stripping in ceramic hollow fibre membrane contactors, Chem. Eng. Sci., 64, 1–8. Kosaraju P.B., Sirkar K.K. (2007), Novel solvent-resistant hydrophilic hollow fiber membranes for efficient membrane solvent back extraction, J. Membrane Sci., 288, 41–50. Koschilowski J., Wieghaus M., Rommel M. (2003), Solar thermal driven desalination plants based on membrane distillation, Desalination, 156, 295–304. Kotov N.A., Liu L., Wang S., Cumming C., Eghtedari M., Vargas G., Motamedi M., Nichols J., Cortiella J. (2004), Inverted colloidal crystals as three-dimensional cell scaffolds, Langmuir, 20(19), 7887–7892. Kresge C.T., Leonowicz M.E., Roth W. J., Vartuli J.C., Beck J.S. (1992), Ordered mesoporous molecular sieves synthesized by a liquid-crystal template mechanism, Nature, 359, 710–712. Kumar P.S., Hogendoorn J.A., Feron P.H.M., Vesteeg G.F. (2002), New absorption liquids for removal of CO2 from dilute gas streams using membrane contactors, Chem. Eng. Sci., 57, 1639–1651. Kunz W., Benhabiles A., Ben-Aim R. (1996), Osmotic evaporation through macroporous hydrophobic membranes: a survey of current research and applications, J. Membrane Sci., 121, 23–36. Lawson K.W., Lloyd D.R. (1997), Membrane distillation, J. Membrane Sci., 124, 1–25. Lewandowicz G., Białas W., Marczewski B., Szymanowska D. (2011), Application of membrane distillation for ethanol recovery during fuel ethanol production, J. Membrane Sci., 375, 212–219. Lin D.J., Chang H.H., Chen T.C., Lee Y.C., Cheng L.P. (2006), Formation of porous poly(vinylidenefluoride) membranes with symmetric or asymmetric morphology by immersion precipitation in the water/TEP/PVDF system, Eur. Polym. J., 42, 1581–1594. Lü L.Z., Yang Z.S., Wang Z.Y., Yang Y.C., Tian S.N. (2011), Preparation and performances of PVDF hydrophobic microporous membrane via immersion precipitation assisted with template, Journal of Tianjin Polytechnic University, 30(4), 6–10 Luo C., Huang W., Han Y. (2009), Formation of two kinds of hexagonally arranged structures in ABC triblock copolymer thin films induced by a strongly selective solvent vapor macromol, Rapid Commun., 30, 1917–1921. Macedonio F., Drioli E. (2008), Pressure-driven membrane operations and membrane distillation technology integration for water purification, Desalination, 223, 396–409. Martinez-Diez L., Florido-Diaz F. J. (2001), Desalination of brines by membrane distillation, Desalination, 137, 267–273.

© Woodhead Publishing Limited, 2013

102

Handbook of membrane reactors

Martinez-Diez L., Vasquez-Gonzales M.I. (1999), Temperature and concentration polarization in membrane distillation of aqueous salt solutions, J. Membrane Sci., 156, 265–273. Mathioulakis E., Belessiotis V., Delyannis E. (2007), Desalination by using alternative energy: review and state-of-the-art, Desalination, 203, 346–365. Meindersma G.W., Guijt C.M., de Haan A.B. (2006), Desalination and water recycling by air gap membrane distillation, Desalination, 18, 291–301. Nene S., Kaur S., Sumod K., Joshi B., Raghavarao K.S.M.S. (2002), Membrane distillation for the concentration of raw cane-sugar syrup and membrane clarified sugarcane juice, Desalination, 147, 157–160. Nymeijer D.C., Folkers B., Breebaart I., Mulder M.H.V., Wessling M. (2004), Selection of top layer materials for gas-liquid membrane contactors, J. of Appl. Polym. Sci., 92(1), 323–334. Peng P., Fane A.G., Li X. (2005), Desalination by membrane distillation adopting hydrophilic membrane, Desalination, 173, 45–54. Phattaranawik J., Fane A.G., Pasquier A.C.S., Bing W. (2008), A novel membrane bioreactor based on membrane distillation, Desalination, 223, 386–395. Qi Z., Cussler E.L. (1985), Microporous hollow fibres for gas absorption I. Mass transfer in the liquid, J. Membrane Sci., 23, 321–333. Ravindra Babu B., Rastogi N.K., Raghavarao K.S.M.S. (2008), Concentration and temperature polarization effects during osmotic membrane distillation, J. Membrane Sci., 322, 146–153. Reverchon E., Cardea S. (2006), PVDF−HFP membrane formation by supercritical CO2 processing: elucidation of formation mechanisms, Ind. Eng. Chem. Res., 45, 8939–8945. Speranza V., Trotta F., Drioli E., Gugliuzza A. (2010), High-definition polymeric membranes: construction of 3D lithographed channel arrays through controlling natural building blocks dynamics, ACS Appl. Mater. Interfaces, 2(2), 459–466. Srinivasarao M., Collings D., Philips A., Patel S. (2001), Three-dimensionally ordered array of air bubbles in a polymer film, Science, 292, 79–83. Stanojević M., Lazarevic B., Radic D. (2003), Review of membrane contactors designs and applications of different modules in industry, FME Transactions, 31, 91–98. Susanto H. (2011), Towards practical implementations of membrane distillation, Chem. Eng. Proc., 50, 139–150. Teoh M.M., Bonyadi S., Chung T.S. (2008), Investigation of different hollow fiber module designs for flux enhancement in the membrane distillation process, J. Membr. Sci., 311, 371–379. Thanedgunbaworn, R., Jiraratananon R., and Nguyen M.H. (2009) Vapour transport mechanism in osmotic distillation process, Int. J. Food Engineering, 5(5) art. 3, 1–19. Tomaszewska M. (1996), Preparation and properties of flat-sheet membranes from poly(vinyldene fluoride) for membrane distillation, Desalination, 104, 1–11. Vogelaar L., Wessling M. (2002), Fabrication of polymeric microsieves by phase separation micro moulding. In: Network Young Membrains, NYM, 5–7 July, 2002, Toulouse, France.

© Woodhead Publishing Limited, 2013

Membrane contactors

103

Wang L., Min J. (2011), Modelling and analyses of membrane osmotic distillation using non-equilibrium thermodynamics, J. Membrane Sci., 378, 462–470. Wang Z.-S., Gu Z.-L., Feng S.-Y., Li Y. (2009), Applications of membrane distillation technology in energy transformation process-basis and prospect, Chinese Sci. Bull. (Springer), 54, 2766–2780. Wang Z.Y., Li J.L., Kong X.S., Yang Z.S., Li C.L. (2011), Effect of octanol on wettability and permeability of PVDF porous membrane via dry-wet phase inversion, Tianjin Daxue Xuebao (Ziran Kexue yu Gongcheng Jishu Ban)/Journal of Tianjin University Science and Technology, 44(7), 628–632. Warczok J., Gierszewska M., Kujawski W., Guellet C. (2007), Application of osmotic membrane distillation for reconcentration of sugar solutions from osmotic dehydration, Sep. Purif. Technol., 57, 425–429. Weyl P.K. (1967), Recovery demineralized water from saline waters, United States Patent 3, 340,186. Winter D., Koschikowski J., Duever D. (2011), Spiral wound modules for membrane distillation: modelling, validation and module optimization, Proc. Int. Workshop on Membrane Distillation and Related Technologies, Ravello (Sa, Italy), Oct. 9–12, pp. 58–59. Yan S.-P., Fang M.-X., Zhang W.-F., Wang S.-Y., Xu Z.-K., Luo Z.-Y., Cen K.-F. (2007), Experimental study on the separation of CO2 from flue gas using hollow fibre membrane contactors without wetting, Fuel Proc. Techn., 88(5), 501–511. Young T.H, Cheng L.P., Lin D.J., Fane L., Chuang W.Y. (1999), Mechanisms of PVDF membrane formation by immersion-precipitation in soft (1-octanol) and harsh (water) nonsolvents, Polymer, 40, 5315–5323. Zhang X., El-Bourawi M.S., Wei K., Tao F., Ma R. (2006), Precipitants and additives for membrane crystallization of lysozyme, Biotechnol J., 1(11), 1302–1311. Zhang X., Zhang P., Wei K., Wang Y., Ma R. (2008), The study of continuous membrane crystallization on lysozyme, Desalination, 219, 101–117. Zhao Z.P., Ma F.W., Liu W.F., Liu D.-Z. (2008), Concentration of ginseng extracts aqueous solution by vacuum membrane distillation. 1. Effects of operating conditions, Desalination, 234, 152–157.

2.11 2.11.1 a b c C Cb1 Cf Cfm C m1 D

Appendix: nomenclature Notation numerical empirical coefficient numerical empirical coefficient numerical empirical coefficient concentration solute concentration in the bulk solution salt bulk concentration in the feed bulk solution salt concentration at the hot membrane surface solute concentration at the feed solution/membrane interface liquid diffusion coefficient of the component

© Woodhead Publishing Limited, 2013

104 de e f g G Gz H hf hmg hmv hp I id J JD Ji JK Jp Jw K Kb KD Kf Kg Kl KK Km km Kn kp Kp Kph1 Kph2 Kpois kr Ks L mT Nu od pav pf pb

Handbook of membrane reactors hydraulic diameter or effective diameter numerical empirical coefficient function of the geometry of the system numerical empirical coefficient growth rate of crystals Graetz number total heat transfer coefficient heat transfer coefficients in feed layer heat transfer through vapour and gas heat transfer coefficients for the liquid vaporisation (latent heat) heat transfer coefficients in permeate layer generic component inner diameter total mass flux water vapour flux due to diffusion of vapour through the pores total flux of the component i Knudsen flux Poiseuille flux total flux of the water vapour w overall mass transfer coefficient constant of nucleation rate diffusion parameter (Equation [2.21]) mass transfer coefficient of feed side growth rate constant mass transfer coefficient of the dense top layer Knudsen parameter mass transfer coefficient of the membrane membrane conductivity Knudsen number conductivity in permeate side mass transfer coefficient of permeate side mass transfer coefficient of the phase 1 mass transfer coefficient of the phase 2 Poiseuille parameter conductivity in retentate side solute mass transfer channel length concentration of crystals in the magma Nusselt number outer diameter mean vapour pressure in the membrane pressure at feed sides of the membrane breakthrough pressure © Woodhead Publishing Limited, 2013

Membrane contactors pp Pr Pv Q Qf Qm Qmc Qmg Qmv Qp R Re Rn Sc Sh T Tf Tfm Tp Tpm

pressure at permeate side of the membrane Prandtl number overall permeability of the water vapour total heat transfer heat transfer from feed side to membrane surface heat transfer across the membrane heat conduction through the membrane heat transfers through the vapour and gas filled pores heat transferred across the membrane due to liquid vaporisation heat transfer from the membrane to the permeate side pore radius of the membrane Reynolds number nucleation rate Schmidt number Sherwood number temperature temperature of the feed side membrane surface temperature feed side temperature of the permeate side membrane surface temperature permeate side

Greek symbols α γ δ δf δp Δ ε ζ θ ξ ρ τ ϕ ωe

2.11.2 AGMD DCMD G

105

separation factor surface free tension membrane thickness boundary layer feed side boundary layer permeate side variation porosity of the membrane concentration polarisation coefficient contact angle numerical parameter liquid density temperature polarisation coefficient packing fraction rotation rate

Abbreviations air gap membrane distillation direct contact membrane distillation gas

© Woodhead Publishing Limited, 2013

106 L LEP MC MD OMD PP PTFE PVDF SGMD VMD

Handbook of membrane reactors liquid liquid entry pressure membrane crystallisation membrane distillation osmotic membrane distillation polypropylene polytetrafluoroethylene polyvinylidenefluoride sweep gas membrane distillation vacuum membrane distillation

© Woodhead Publishing Limited, 2013

3 Pervaporation membrane reactors G. CAMERA-RODA, University of Bologna, Italy and V. AUGUGLIARO, V. LODDO and L. PALMISANO, University of Palermo, Italy

DOI: 10.1533/9780857097347.1.107 Abstract: Pervaporation is a peculiar membrane separation process which is currently being considered for integration with a variety of reactions in promising new applications. Indeed, pervaporation membrane reactors have some specific uses in sustainable chemistry, which is an area currently growing in importance. The fundamentals of this type of membrane reactor are presented in this chapter, along with the advantages and limitations of different processes. A number of applications are reviewed with particular attention given to potential future developments. Key words: pervaporation, membrane reactors, integrated process, process intensification, catalytic membranes.

3.1

Introduction

The utilization of pervaporation (PV) coupled with reactions was first studied in 1960 (Jennings and Binning, 1960) and a significant amount of research was undertaken in the years which followed. Recently, research on this topic has gained importance thanks to both the growing focus on energy saving and increasing interest in sustainable chemical processes. In these fields, pervaporation and pervaporation membrane reactors (PVRs) can offer some interesting advantages over other processes. This increase in research has also led to promising new types of applications for PVRs. Since the 8th International Symposium on Pervaporation, held in 1995 in Reno, Nevada, USA, there have been two international scientific conferences on PV, organized in 2010 and 2011 in Torun, Poland, which demonstrate a renewed interest in pervaporation. A notable number of new processes were presented, together with several studies on new or modified membrane materials with enhanced performance. For the membrane separation process to be integrated with a reaction in a ‘membrane reactor’ certain criteria have to be met. Straightforward 107 © Woodhead Publishing Limited, 2013

108

Handbook of membrane reactors

compatibility of the operative conditions is required, as well as of course the achievement of the specific separation capabilities for the task in hand. The final objective of the integration should be a ‘process intensification’ with a substantial enhancement of the yield of the conventional sequential connection of reaction and separation (Stankiewicz, 2003; Stankiewicz and Moulijn, 2000). It is illustrative that Lutze et al. (2010) utilized a pervaporation membrane reactor to exemplify the concept of process intensification. Taking into account the requirements above, pervaporation appears very well suited to some specific applications where it may outperform other membrane or conventional separation processes. In this chapter, after some general considerations about the opportunities and alternatives offered by PVRs, the principles of pervaporation are briefly presented to identify the essential characteristics that can be exploited in a PVR. Following this, the fundamentals which are the basis of the existing studies and applications are surveyed with special attention to recent developments. Finally, indications are given regarding expected future trends.

3.1.1

Pervaporation membrane reactor (PVR) as a special case of membrane reactors

Pervaporation is unique among the various membrane separation processes since it presents the following singular characteristics: • the membranes are non-porous; • a phase change accompanies the permeation (in practice only membrane distillation presents the same phenomenon); • the driving force is not given by a pressure difference (even if the permeate pressure is generally maintained under a vacuum), so that the pressure upstream the membrane is usually atmospheric; • selectivity for the separation of different compounds relies mainly on differences in their relevant solubilities and/or diffusivities in the membrane material; • PV covers a very large spectrum of separation processes. By the correct choice of the membrane material, it can be used to permeate preferentially either water from organics or organics from water, or organics from other organics. Other interesting characteristics of PV, some of which are shared with other membrane processes, are as follows: • •

mild operative conditions (low temperatures and pressures); modularity;

© Woodhead Publishing Limited, 2013

Pervaporation membrane reactors • • •

109

simplicity of control of the process; energy saving; and safety.

These distinguishing features make pervaporation suitable for a number of specific applications and make it a good candidate for green processes in sustainable chemistry.

3.2

The basic concepts of integrated pervaporation−reaction processes

In comparison with the simplest sequential connection at an equipment level of these two unit operations (reaction and pervaporation – usually, but not necessarily, in this order), the coupling of the two processes into a single unit even without a true interrelationship between them may create savings due to the smaller, cheaper plant required. However, a true integration is obtained when the coupling is accompanied also by a positive (synergistic) interaction of the two processes, which can be actuated through different mechanisms. When a real interrelationship or, equivalently, an integration (Schmidt-Traub and Górak, 2006) is established between a separation process and a reactive process, some of the following benefits can be obtained: • • • • • • • •

to retain the reagents into the reactor while the products are removed; to remove products that can hinder the desired reaction; to remove one or some products of a reversible reaction; to recover intermediates that could otherwise undergo further undesired reaction if maintained in the reactor; to maintain the catalyst in the reactor; to utilize the heat of reaction for the separation; to create a more intimate contact between reagents and catalyst (typical for catalytic membranes); to produce compounds which can be more easily removed by pervaporation with respect to the starting permeating species.

Most of the studies or applications of PVR are in two fields: esterification reactions, or more generally equilibrium reactions, and biological reactors. Other new applications which have recently emerged include treatment of polluted aqueous streams (water detoxification) and green chemistry for fine chemicals. It is worth briefly describing the pervaporation process in order to point out the characteristics that should be taken into account in the choice and the design of a pervaporation membrane reactor.

© Woodhead Publishing Limited, 2013

110

3.2.1

Handbook of membrane reactors

Principles of pervaporation

In pervaporation the components of a liquid solution permeate at different rates by dissolving and diffusing across a non-porous (‘dense’) membrane and evaporating downstream from the membrane. This mechanism explains the term pervaporation, which derives from the permeation and evaporation steps of the process. In this way, a feed stream is separated into two streams: the permeate and the retentate. The separation performance is determined by the flux and the selectivity, this latter being expressed as a separation factor or, alternatively, by an enrichment factor (Böddeker, 1990). The separation factor is similar to the relative volatility of species Ai with respect to species Aj and is defined as:

α PV,ij =

ci,PP / c j , P ci,FF / c j , F

=

ci,P ,P / c , F c j ,PP / c j , F

where c can be any convenient variable of composition (usually the concentration) and the subscripts P and F refer to the permeate (downstream side of the membrane) and to the feed (upstream side of the membrane), respectively. Alternatively, the enrichment factor, βi = ci,P/ci,F of Ai, can be used. Separation factors αPV,ij larger than 1 imply that the membrane permeates preferentially Ai with respect to Aj, and of course the higher the parameter the more effective the separation of Ai from Aj. The separation of the various chemical species is a consequence of their different solubility and diffusivity into the membrane. In fact, the value of the pervaporation selectivity (the separation factor) αPV,ij is usually different from that of the sorption selectivity αS,ij, which is defined as the ratio of the equilibrium solubility in the membrane of Ai in comparison with the one of Aj. Thus, a diffusion selectivity coefficient αD,ij can be formally defined as αD,ij = αPV,ij/αS,ij. In an aqueous solution of organics, water shows a diffusion selectivity coefficient higher than 1. This is an advantage for hydrophilic membranes which are used to preferentially permeate water, but a disadvantage for organophilic membranes, since the diffusivity counteracts the sorption selectivity. Usually the mathematical description of the process is expressed by a solution−diffusion model, which could take into account some non-linear phenomena for both the sorption and the diffusion steps. Anomalous behaviours are frequently observed and are typical for polymers, which very often constitute the material of the selective layer of the membrane. Swelling, which accompanies the sorption of chemicals into polymers, is just one of the most important non-linear phenomena. Due to the presence of the permeants in the membrane, swelling is not uniform and produces non-linear gradients of permeant concentration inside the membrane.

© Woodhead Publishing Limited, 2013

Pervaporation membrane reactors

111

Other aspects and phenomena which are often considered are concentration and temperature polarization, interactions between the permeating species, effect of fillers included in the membrane to induce positive modifications of the sorption and diffusion capabilities, interaction of the porous supporting layer, etc. The driving force for the permeation of a certain compound Ai is given by the difference of its chemical potential between the upstream and the downstream side of the membrane. This difference can be obtained by keeping the partial pressure of the permeating species Ai (pi = p × yi) in the permeate vapour low. Therefore, it is possible to operate at a low total pressure p (the more common choice) or alternatively to use an inert gas carrier to maintain a small value of the mole fraction yi. It is not common to work at a high feed pressure, since the influence of the feed pressure is usually minor, because the chemical potential of a compound in a liquid solution depends only very slightly on this parameter. In theory, a very high feed pressure is only really required if the permeate partial pressure approaches the saturation value. However, it is usually more economic and practical to keep the permeate pressure low or to increase the operative temperature. For these reasons, it is common practice to operate the feed at atmospheric pressure. In the case that a temperature gradient is created between the two sides of the membrane, ‘thermopervaporation’ occurs. Until now, some technical problems have limited the application of thermopervaporation, but researchers are now trying to overcome these difficulties since thermopervaporation promises increased energy efficiency (Sanchez Fernandez et al., 2010). It is worth noting that in pervaporation, contrary to multistage distillation, the only stream that needs to be vaporized (and then condensed or frozen) is the permeate, which usually constitutes a small fraction of the feed. Thus, in pervaporation the energy demand, which mainly concerns the latent heat of vaporization of the permeate plus the energy consumption of the vacuum pumps and of the refrigeration system to condense or freeze the permeate, is relatively low. As mentioned previously, the driving force depends on the permeate pressure. The flux of a given component decreases by increasing the permeate pressure and approaches zero at the partial pressure saturation. An increase of the permeate pressure may hinder or favour the selectivity, depending upon whether the membrane is selective towards the less volatile or more volatile component, respectively ( Böddeker, 1990; Néel, 1995). The flux is also affected by the temperature; an Arrhenius type law is very often a good representation of this dependence with an activation energy in the range of 17–63 kJ/mol (Huang and Rhim, 1991). As a rule of thumb, a twofold increase of the flux is often assumed for a 10°C temperature rise. On the contrary, the selectivity generally decreases as the temperature

© Woodhead Publishing Limited, 2013

112

Handbook of membrane reactors

decreases, but the variation in selectivity is not as important as the variation in flux. By taking into account this behaviour it can be concluded that the separation performance can be improved at higher temperatures. However, it is possible to operate also at relatively low temperatures, provided the permeate pressure is kept sufficiently low or the membrane area is sufficiently high to meet the required flow rate. In this way it is possible to avoid damage to temperature-sensitive compounds and to recover heat from low temperature sources. In principle, the mechanism of PV is similar to extraction, where the membrane material acts as a solid ‘solvent’ for the ‘extraction’ of the permeating species. This ‘solvent’ is continuously ‘recovered’ in situ by evaporating the permeate. The main difference is that contrary to conventional solvent extraction, mass transport properties (diffusivities) of the permeants in the membrane contribute to the performance of the process (flux and selectivity). As a consequence, the selectivity can be very different from the one of the simple vapour−liquid equilibrium and in many cases is so high that it equals that obtainable by several equilibrium stage units in a more conventional separation process, such as distillation (Baudot and Marin, 1997). For many systems which present an azeotrope, the equivalent number of distillation equilibrium stages approaches infinity. Consequently, the azeotrope breaking is currently the main application of PV. Furthermore with PV there is no need for an entrainer, as in extractive distillation, and an additional significant energy saving is achieved. Recent studies have shown that for some systems it is possible to obtain a ‘fractional condensation’ of the permeate (Brazinha and Crespo, 2009; Vane et al., 2004). In practice, fractional condensation allows an additional separation of the permeated compounds with respect to the separation achieved with the pervaporation membrane. For instance, it has been shown (Augugliaro et al., 2011; Böddeker et al., 1997; Brazinha et al., 2011) that the permeated vapours of vanillin can be selectively deposited as crystals at ambient temperature, thus obtaining a very effective separation from the other permeants, which under vacuum condense only at lower temperatures. PV is suitable to preferentially permeate the minor component of the feed solution, because: (i) in PV the flux is intrinsically low (the permeation takes place through a dense membrane), (ii) the energy consumption is proportional to the permeant flow rate and (iii) the separation factor is usually decreasing with increasing the feed concentration of the permeant. The practical lower limit of the feed concentration of the compounds to be removed is very small, typically ranging between 1 and 10 ppm. At even lower concentrations the separation is impractical, due to the decrease in the driving force and the increasing importance of concentration polarization.

© Woodhead Publishing Limited, 2013

Pervaporation membrane reactors

113

Depending on the membrane material used, it is possible to separate different types of compounds. Membranes are conventionally classified into two main categories: hydrophilic and organophilic (hydrophobic) membranes. The first type is used for the selective permeation of water with respect to organics, and the second one for the opposite operation. However, in some cases, for mixtures of organic compounds in aqueous or non-aqueous solutions, organophilic membranes are capable of providing satisfactory organics−organics separation (Smitha et al., 2004). New polymeric membranes, possibly with advanced or novel functions (Ulbricht, 2006), are still under development in order to enhance the separation performances and to reach the levels which are required for potential new applications, such as membrane reactors. Pervaporation, as a non-integrated process, is typically utilized for dehydration and for the recovery or removal of organics from aqueous solutions and sometimes also for the separation of organic mixtures (Néel, 1995). Also many ‘hybrid’ processes have been developed where PV is coupled with other processes, such as different membrane processes (e.g., reverse osmosis, or organophilic pervaporation coupled with hydrophilic pervaporation), distillation, reactive distillation and, of course, reaction. With these aspects in mind, PV appears particularly suitable to keep the concentration of a by-product low, or to continuously recover a product while it is formed. Note that these are the main objectives typically pursued in membrane reactors.

3.3

Classification of pervaporation membrane reactors

A pervaporation membrane reactor can be classified according to the different criteria listed below: 1. Type of coupling a. No effective interrelationship, just coexistence of the two processes in the same apparatus b. Real interrelationship between the two processes 2. Configuration of PVR a. Single equipment b. Separate equipments 3. Mode of operation a. Continuous b. Semi-continuous 4. Type of membrane utilized a. Hydrophilic b. Organophilic

© Woodhead Publishing Limited, 2013

114

Handbook of membrane reactors

5. Membrane material a. Organic membrane (usually polymeric) b. Inorganic membrane (usually zeolitic) c. Hybrid membrane (organic with dispersed inorganic powders or with an inorganic layer) 6. Target compound removed a. By-product b. Product The study of alternative solutions to interconnect the reaction and the pervaporation separation processes is useful as it discloses some of the implications of the integration. Figures 3.1–3.4 illustrate the possibilities of connecting reaction and separation according to different schemes. The choice of configuration is usually made on the basis of the requirements of the two processes and the expected advantages to be gained. The conventional sequence reactor−separator shown in scheme 1 represents the trivial case of no integration. If the two processes are really interrelated, their coexistence affects the rate or the yield of the reaction and/or separation. Both schemes 2 and 3 could be used in this circumstance. Scheme 3 offers the possibility to operate both processes in the same equipment, thus making a saving of space and investment costs in comparison to scheme 2. Catalytic membranes are just a particular case of scheme 3. However, scheme 2 may present the following advantages over scheme 3: • the ability to operate under different conditions in the two apparatuses (i.e., temperature, pressure and hydrodynamic regimes), so that these conditions can be optimized for each process; • the possibility to conduct decoupled maintenance; • the possibility to vary without limitations the ratio, δ, between the characteristic rate of permeation and the characteristic rate of reaction (Mohan and Govind, 1988). Different expressions for δ can be adopted, depending Retentate

Feed

Reaction unit

PV

Unit Permeate

3.1 Scheme 1 for the connection of reaction and separation.

© Woodhead Publishing Limited, 2013

Pervaporation membrane reactors

115

on the constitutive equations for the rate of reaction and for the rate of pervaporation. In the most common or simplest form it is conveniently expressed as δ ( / ρ) (k cs(,0− ) Vr ) where A is the membrane area, k the kinetic constant of the reaction, cs,0 the initial concentration or the feed concentration of the reagent, n the order of the reaction, Vr the  the permereactor volume, ρ the density of the liquid solution and m" ate mass flux. Note that the characteristic rate of membrane separation depends on the membrane area, whereas the characteristic rate of reaction is proportional to the volume of the reactor. So, by adopting scheme 2, these two factors can be changed independently in order to modify δ, possibly resorting to the typical modularity of the pervaporation process. On the contrary, if adopting scheme 3, the possible variation of δ is constrained within certain limits. These limitations are due to space restrictions inside the reactor−separation equipment, or, in a catalytic membrane, the constraints are imposed by the preparative procedures or by the physical and chemical resistance of the functionalized membranes as well as from the necessity to avoid a decrease of the catalytic activity and of the membrane permeability (Ozdemir et al., 2006). The possibility, offered by scheme 2, of modulating δ with no limitation is important, since in an integrated reaction-PV process the ratio δ should be optimized to maximize performance and/or economy of the integrated process. In some circumstances scheme 3 can guarantee more effective heat and mass transfer, in particular if a catalytic membrane is used, by promoting a more intimate contact of the reagents with the catalyst inside the membrane and a shorter path for mass transfer of the permeating products. The permeation of the products can be enhanced thanks to the local production of these chemical species directly into the membrane, thus bypassing their mass transfer from the bulk of the upstream solution to the membrane surface. This local production also avoids the dissolution of chemical species at the liquid solution−membrane interface and their diffusion into part of the membrane. Furthermore, it is argued that if the catalyst is put inside the membrane, it could act as a barrier against the permeation of the reactants since they are locally destroyed by the reaction. The final result would be an improvement of the permselectivity (Bagnell et al., 1993). Actually, it appears rather difficult to quantify this effect. The main perplexities about this mechanism derive from the observation that, in order to effectively utilize the catalyst inside the membrane, solubility of the reactants in the membrane needs to be relatively high, thus reducing the selectivity towards the products if the hypothesized barrier effect given by the catalyst is not very efficient. In order to obtain a significant barrier effect, a relatively high concentration of the catalyst in the membrane is necessary, but in this case the permeability or the resistance of the membrane could be negatively affected.

© Woodhead Publishing Limited, 2013

116

Handbook of membrane reactors

The experimental results obtained with catalytic pervaporation membranes for the well-explored case of esterification reactions are generally not as good as those observed in integrated reaction−pervaporation processes with non-catalytic membranes. For the reasons mentioned above, although pervaporation catalytic membranes are potentially interesting, they require additional research in order to better analyse these phenomena and optimize the immobilization techniques of the catalyst in the membranes. The effective integration of a reaction and a separation process requires that both processes work jointly on a solution with approximately the same composition. In this way the products can be recovered by pervaporation while they are being produced, avoiding any degree of ‘sequential’ separation. When the two processes take place in the same equipment, as in scheme 3, only mass transfer limitations from the bulk of the solution to the membrane surface can reduce the effectiveness of the integration and the methods of reducing such limitations (e.g., by enhancing mixing) are well known in chemical engineering. But, when the two processes take place in separate equipments, some additional conditions must be satisfied to get an ‘effective’ integration. If the process is semi-continuous (see scheme 2(i), without the optional streams) an effective integration can be attained if the variations of the composition in the passage of the solution through the reactor and the separator are very small (differential). Mathematically, this requirement is usually expressed referring to the ratios of the characteristic time of disappearance of the reactant in the reactor and in the separator with respect to the corresponding residence times. The following two conditions must be concurrently satisfied: Da > 1, where Da = k cs(,n0− ) × (Vr / V ) is the Damköhler number (the ratio of the residence time in the reactor to the  ′′ ) is the Péclet number characteristic time of the reaction), Pe = V ( A m (the ratio of the characteristic time of permeation of As in the PV unit to the residence time in the PV unit) and V represents the volumetric flow rate through the reaction and the membrane units. The simplest way to attain Da > 1 is to adopt a high flow rate of the recycle. Note that the parameters Da, Pe and δ are not independent since: δ = 1/(Da Pe). In a continuous process, the effectiveness of the integration still requires that the reactor is differential (for the definition of differential reactor see (n ) Levenspiel, 1999, p. 397), so that Da= × (Vr / V ) must be much less s, 0 than 1. There are two ways to fulfill this objective: a large flow rate, V , and/or a small reactor volume, Vr. For scheme 2(i), high values of V are obtained when the recycle ratio is close to 1. In fact if R → 1 the flow rate of the recycle, V , becomes infinitely larger than the flow rate of the fresh feed to the system. In this situation the relevance of backmixing is high and, as a consequence, regardless of whether the reactor is a plug flow reactor (PFR) or a continuous stirred tank reactor

© Woodhead Publishing Limited, 2013

Pervaporation membrane reactors

117

(CSTR), scheme 2(i) equals the performance that can be obtained with scheme 3(i) with the same reactor volume and the same membrane area. For scheme 2(ii) high integration effectiveness is reached if the volume, Vr, of each reactor is small in order to get again Da 70

Single equipment with non-catalytic membrane.

2

1

C and S

H

>99

Comparison between semi-continuous reactor and continuous PFR and CSTR.

3

1

S

H

H, HybSi® 1,1 Diethoxy membrane butane (Energy Research Center of the Netherlands) O, Pervap Methyl acetate 2011 (Sulzer Chemtech) O, Nafion Methyl acetate (Permapure and n-butyl Products Inc., acetate NJ, USA; home modified)

70 to >90

4

2

S

H

Ethyl lactate

98.6

5

2

S

H

Ethyl lactate and diethyl succinate

98.6

6

1

S

H

O, GFT pervap 1005 (Deutsche Carbone AG) O, GFT pervap 1005 (Deutsche Carbone AG) and T-1b (Texaco Research) O, PVA membrane (home prepared) + catalyst

Nafion membranes act also as catalytic membranes. A second mechanism of selective permeation of the products due to the presence of the catalyst inside the membrane is verified. The increase of conversion is directly proportional to the amount of removed water. The selectivity is low and significant amounts of reagents permeate. Homogeneous and heterogeneous esterification.

Isoamyl acetate 90

Flat membrane and spiral wound membrane.

Catalyic membrane. Investigation on the effect of the ratio membrane area/volume of reaction.

(Continued)

Table 3.1 Continued

© Woodhead Publishing Limited, 2013

Ref.a

Configurationb

Type of Mode memof operationc braned

Membrane materiale Product

Improvement of conversion or maximum conversion (%)

Notes

7

2

S

H

70 to >90

Experimental results compared with theoretical results from kinetic model.

8

1, 2

S

H

>95

9

2

S

H

Experimental and theoretical analysis on the effects of different parameters, e.g. the ratio membrane area/volume of reaction. Analysis of the effects of different parameters, e.g. the ratio membrane area/volume of reaction.

10

2

S

H

11

2

S

H

O, PVA membrane 1-propyl HS type (Carbone propionate Lorraine-GFT) and 2-propyl propionate O, PVA membrane 1-propyl HS type (Carbone propionate Lorraine-GFT) O, Pervap 2201 Ethyl lactate membrane (Sulzer Chemtech) O, GFT pervap Benzyl acetate 1005 (Deutsche Carbone AG) – –

12

1

S

H

O, Pervap 1000 Ethyl acetate (Sulzer) + catalyst

–

13

2

S

H

23 to >50

14

1

S

H

15

2

S

H

O, Pervap 1000 (Sulzer) H, PVA membrane with different zeolites as fillers (home prepared) O, Pervap 1000 and 1001 (Le Carbone Lorraine)

Ethyl oleate

–

99

–

Ethyl acetate

Ethy lacetate

Good agreement between the experimental data and the theoretical results from the model. Only mathematical model to investigate the effects of different parameters. Catalyic membrane. Investigation on the effect of the ratio of the membrane area to the volume of reaction.

Study on the effects of fillers on the pervaporation performances and consequently on the yield. 51 to >63.9

Only simulation of the integrated process.

© Woodhead Publishing Limited, 2013

16

1

S

O

O, PDMS (home prepared)

Ethyl acetate

Slight increase

17

2

C

H

Ethyl acetate

90

18

1

S

H

Methyl acetate and other esters

98

It is demonstrated that ‘the amount of removed water clearly correlates with the yield promotion’.

19

1

S

H

I, mordenite and zeolite A (home prepared) I, different types of zeolitic membranes (home prepared) –

Diethyl tartrate

–

20

2

S

H

I, sodalite (home prepared)

67 to >98; 62 to >92

21

1

S

H

O, PEI from General Electric-Ultem 1000 (home prepared)

Ethyl acetate and butyl acetate Ethyl oleate

22

1

S

O

O, PDMS from General Electric (home prepared)

Both concentration-based and activity-based rate constants of the reaction are experimentally obtained, showing that the rate of reaction can be predicted correctly also by concentration-based parameters. Simulations of the integrated process show that the ratio, A/V, of the membrane area to the reactor volume has an optimum. When A/V is too low the water removal is too low and when A/V is too high, too much alcohol is removed. The focus is on the properties of the inorganic membrane material. Membrane shows a very high selectivity and good stability. Initial composition with an excess of ethyl alcohol. Limited losses of ethyl alcohol in the permeate (99; 70 to >99

Isobutyl acetate 60 to >70

Utilization of an organophilic membrane, which removes selectively acetate, but also some water and reagents. Conversion in the integrated process is slightly enhanced. More resistant inorganic membranes are used to withstand the acid reaction medium.

(Continued)

Table 3.1 Continued

© Woodhead Publishing Limited, 2013

Ref.a

Configurationb

Type of Mode memof operationc braned

Membrane materiale Product

23

1

S

H, O

Isobutyl acetate 60 to >90

24

1

S

H

O, organophilic: PDMS (home prepared); hydrophilic: Nafion 117, Pervap 1201, Pervap 2216 O, PERVAP 1005 (GFT)

Ethyl acetate

–

25

2

S

H

Propylpropionate

–

26

–

–

–

O, PERVAP 2201D (Sulzer Chemtech) –

–

–

27

–

–

–

–

–

–

Improvement of conversion or maximum conversion (%)

Notes

Both organophilic and hydrophilic membranes are utilized. The effects of the temperature, of the excess of alcohol in the initial mixture and of the ratio of the membrane area to reactor volume are studied. The results obtainable with organophilic and hydrophilic membranes are compared. The performances of the membrane are evaluated experimentally and theoretically. The integrated process is simulated by a mathematical model, which shows that the results are strongly affected by the ratio of the membrane area to the mass of the mixture. –

Review of many PVR systems for esterification. The design of PVR systems for esterification reactions for various configurations is presented. The effects of many parameters, such as δ, Da and R, are taken into account. Review of many PVR systems, also with combination with other processes.

© Woodhead Publishing Limited, 2013

28

1

S

H

O, PVA on porous ceramic support (home prepared)

n-butyl acetate

75 to >92

29

1

S

H

O, PVA on support (home prepared)

n-butyl acetate

–

30

2

S

H

Ethyl lactate

66 to >80

31

2

S

H

H, hybrid tetraethoxysilane within chitosan membrane (home prepared) O, PVA (GFT)

Propylpropionate

–

32

1, 2

S

H

I, silica membranes Ethyl tartrate on γ-alumina (Pervatech BV)

–

33

1

S

H

O, PEI and POPMI (home prepared by phase inversion method)

>99

Ethyl oleate

Effects of various parameters: temperature, initial molar ratio of acetic acid to n-butanol, ratio of membrane area to the reacting mixture volume, catalyst content. A relatively simple mathematical model is introduced. Comparison of experimental and calculated values. Study on the effect of various parameters. The ratio of the rate of water removal by the membrane to that of water production is studied experimentally and theoretically. Characterization of the membrane by FT-IR, XRD, TGA and contact angle.

Several aspects of process design and analysis are systematically considered. A comprehensive model takes into account several phenomena. Comparison between experimental and calculated results. A realistic model, whose parameters have been obtained from independent reaction and pervaporation experiments, is used to design the PVR. The process is investigated experimentally and theoretically. It is shown that the presence of water inhibits the rate of the forward reaction at relatively high temperature. The effects of different parameters are studied.

(Continued)

Table 3.1 Continued

© Woodhead Publishing Limited, 2013

Ref.a

Configurationb

Type of Mode memof operationc braned

Membrane materiale Product

Improvement of conversion or maximum conversion (%)

Notes

34

1

C

H

>equil. at higher T

35

1

S

H

H, asymmetric Ethyl acetate γ-alumina porous tubular membrane (id of 0.0070 m, od of 0.010 m, MembraloxTM) coated by polyetherimide (UltemR-1000) (home prepared) I, silica membranes Ethyl lactate on γ-alumina (Pervatech BV)

36

1

S

H

I, zeolite catalytic coating on top of ceramic silica membrane

68 to >73

In the introduction a synthetic review of PVR applications is provided mainly for esterification reactions. The activities of the chemical species in the non-ideal mixtures are used to describe esterification reaction rates and the driving force for transmembrane permeation. The PVR follows a conventional PFR. The adsorption of water by CaSO4 at the permeate side enhances the conversion if the permeate pressure is relatively high. The emphasis is on the pervaporation process, in particular on the consequences of concentration and temperature polarization. The PVR is simulated by a mathematical model. The study is focused on the preparation of catalytic membrane. In the experiments the initial feed is a mixture of reactants and products at ‘equilibrium’. The loss of reactants and acetate in the permeate are very low thanks to the ‘dual-layer structure’, which gives better selectivity.

Butyl acetate

90

© Woodhead Publishing Limited, 2013

37

1

C

H

I, zeolite catalytic coating on top of ceramic pervaporation hollow fiber membranes (TNO-TPD, The Netherlands)

Butyl acetate

68 to >73

38

1

S

H

About 50

39

–

–

H

H, composite Butyl acetate ceramic/ poly(vinyl)alcohol membrane coated by catalytic Amberlyst 15 resin O, Pervap Isopropyl 2201 (Sulzer acetate Chemtech)

40

2

S

H

O, Pervap 2201 (Sulzer Chemtech)

>90

Isopropyl acetate

–

In addition to the same experimental results reported also by Peters et al., 2005a, simulations by a mathematical model are presented. The simulation shows that with the same catalyst loading the catalytic membrane reactor gives a higher conversion than the one obtainable with the catalyst simply dispersed in the bulk liquid. However the difference between the best performances is small. Investigation on the performance of catalytic membranes obtained by coating a composite ceramic/poly(vinyl)alcohol membrane with layers of catalytic Amberlyst 15 resin. The process appears to be limited by the low rate of permeation obtained with the catalytic membrane. Investigation on the kinetics of the esterification reaction in the presence of the ion-exchange resin Amberlyst 15. The Uniquac model is applied to evaluate the activities of the reacting species in the kinetic equation. The interactions in permeation between the components of the quaternary mixture composed by the reactants and the products are also studied. These preliminary kinetic data are used in part II of the article for the analysis of the integrated process. The PVR is experimentally and theoretically analysed. The effects of: (i) the initial reactant molar ratio; (ii) the ratio of the membrane area to the initial solution volume; (iii) the operating temperature; (iv) the catalyst concentration are studied.

(Continued)

Table 3.1 Continued

© Woodhead Publishing Limited, 2013

Ref.a

Configurationb

Type of Mode memof operationc braned

Membrane materiale Product

Improvement of conversion or maximum conversion (%)

Notes

41

1

S

H

O, PVA on polyethersulfone (home prepared)

Methyl oleate

98.7 to > 99.9

42

1

S

H

I, zeolite T (home prepared) on alumina support

Ethyl acetate

≈80 to > 99.5

43

2

S

H

–

Ethyl acetate

–

44

2

S

H

I, Hybsi ceramic membrane by Pervatech

Methyl esters

75 to >90

45

1

C

H

O, CMC-E membrane by CM-Celfa AG, Switzerland

Ethyl acetate

71 to >98.7

The performances with the home prepared membrane are a little better than those achievable with a commercial membrane by CM-Celfa. A simple model fits the experimental data satisfactorily. A detectable decline of the separation factor is observed as a consequence of the aggressive acid reacting solution. A simple model is used to evaluate the effect of the flux and of the selectivity on the efficiency of a semibatch PVR for esterification. The model is validated through a comparison with literature experimental data from Tanaka et al., 2001. The ceramic membrane withstands the acid reactive environment at relatively high temperatures. It is reported that the plant output is largely increased in comparison with that achievable with the conventional existing esterification process. A continuous process is studied with commercial membranes in a loop tube membrane reactor. Based on the data obtained in a pilot plant, an economic analysis demonstrates that significant savings can be obtained in operating and investment costs with the PVR in comparison with the conventional reaction distillation process.

© Woodhead Publishing Limited, 2013

46

–

S

H

–

Benzyl acetate

≈50 to >≈ 65

47

–

S

H

–

n-Butyl acetate

–

48

–

S

H

–

Ethyl acetate

–

49

1

S

H

n-Butyl acetate

>99

50

1

C

H

O, phosphatic PVA on PAN microfiltration support (home prepared) H, modified ceramic tubular membrane (from U.S. filter) with a γ-alumina layer coated by polyetherimide (UltemR-1000) (home prepared)

Ethyl acetate

≈74 to >≈65

The effect of some parameters is studied by a very simple model, which assumes the same uniform concentration in the reactor and in the PV unit. The model is validated by comparison with literature experimental data from Domingues et al., 1999. The effects of some parameters are studied by a very simple model, which assumes the same uniform concentration at any time in the reactor and in the PV unit. The model is validated by comparison with literature experimental data from Liu and Chen, 2002. The effects of some parameters are studied by a very simple model, which assumes the same uniform concentration at any time in the reactor and in the PV unit. The model is validated by comparison with literature experimental data from Tanaka et al., 2001. Evaluation of the membrane performances. Comparison between experimental and mathematical results.

The model considers the activities of the chemicals to evaluate the driving force of the permeation and the kinetics of the reaction. Satisfactory agreement is obtained between experimental data and model results. The effects of temperature and space time in the reactor are studied.

(Continued)

Table 3.1 Continued

© Woodhead Publishing Limited, 2013

Ref.a

Configurationb

Type of Mode memof operationc braned

Membrane materiale Product

Improvement of conversion or maximum conversion (%)

Notes

51

1

S

H

n-Butyl acetate

≈60 to >≈95

52

1

S

H

H, composite catalytic membrane with cross-linked PVA on porous ceramic plates –

n-Butyl acetate

>90

Composite membranes are produced and tested. The influences of the temperature, of the initial molar reactant ratio and of the catalyst concentration are experimentally evaluated, showing that the operating temperature is the most effective factor. The model of the process considers the chemical activities in the reaction kinetic equation. The model is validated by comparison with experimental data from Liu et al., 2002. The effects of the initial composition, the catalyst concentration and the ratio of the membrane area to the reactor volume are simulated.

a

1: Agirre et al., 2011; 2: Assabumrungrat et al., 2003; 3: Bagnell et al., 1993; 4: Benedict et al., 2003; 5: Benedict et al., 2006; 6: Castanheiro et al., 2006; 7: David et al., 1991a; 8: David et al., 1991b; 9: Delgado et al., 2010; 10: Dominguez et al., 1999; 11: Feng and Huang, 1996; 12: Figueiredo et al., 2008; 13: Figueiredo et al., 2010; 14: Gao et al., 1996; 15: Gonçalves et al., 2004; 16: Hasanoğlu et al., 2009; 17: de la Iglesia et al., 2007; 18: Inoue et al., 2007; 19: Keurentjes et al. 1994; 20: Khajavi et al., 2010; 21: Kita et al., 1988; 22: Korkmaz et al., 2009; 23: Korkmaz et al., 2011; 24: Krupiczka and Koszorz, 1999; 25: Lauterbach and Kreis, 2006; 26: Lim et al., 2002; 27: Lipnizki et al., 1999; 28: Liu and Chen, 2002; 29: Liu et al., 2001; 30: Ma et al., 2009; 31: Mitkowski et al., 2009; 32: Nemec and van Gemert, 2005; 33: Okamoto et al., 1993; 34: Park and Tsotsis, 2004; 35: Pereira et al., 2010; 36: Peters et al., 2005a; 37: Peters et al., 2005b; 38: Peters et al., 2007; 39: Sanz and Gmehling, 2006a; 40: Sanz and Gmehling, 2006b; 41: Sarkar et al., 2010; 42: Tanaka et al., 2001; 43: Tanna and Mayadevi, 2007; 44: Velterop, 2011; 45: Waldburger and Widmer, 1996; 46: Wasewar, 2007; 47: Wasewar et al., 2008; 48: Wasewar et al., 2010; 49: Xuheui and Lefu, 2001; 50: Zhu et al., 1996; 51: Zhu and Chen, 1998; 52: Zou et al., 2010. b

Configuration: 1, single equipment; 2, separate equipments.

c

Mode of operation: C, continuous; S, semi-continuous.

d e

Type of membrane: H, hydrophilic; O, organophilic.

Membrane material: O, organic; I, inorganic; H, hybrid.

Pervaporation membrane reactors

131

steps so that different amounts of catalyst are deposited. Nonetheless, it is observed that an increase in the thickness of the catalyst layer reduces to some extent the overall permeability of the two-layer structured membrane. The results show that ‘further adjustments of the catalytic layer thickness and an optimization of both the catalytic and membrane properties’ are needed. When a system equipped with catalytic membranes is compared with one where coupled reaction and permeation take place in separate zones, it remains to be demonstrated that the expected advantages deriving from the utilization of catalytic membranes (e.g., space savings, shorter diffusion path for the permeants, etc.) counterbalance the aforementioned intrinsic difficulties. In the design and analysis of PVRs for esterification reactions, many mathematical models have been proposed with different levels of complexity according to the number of phenomena and non-linear behaviours which are taken into account. In some cases, simple models were sufficient to adequately reproduce the essential features of the system. For instance Inoue et al. (2007) adopted a simple mathematical model, which was able to satisfactorily fit their experimental data. Mass balances were expressed as: dcester ″′ = n ester ,g dt −

dcacid dc dc = − alcohol = ester dt dt dt

dcwater ′″ = n water t ,g dt

J water

A Vr

with constitutive equations: ″′ n ester ,gg

rrev =

n w″′ater t g = rfor

k cester cwater ; K

rrev ;

rfo forr

k caalcohol lcohol cacid ;

J water = Pwater cwater

and initial conditions at t = 0: calcohol = calcohol,0; cacid = cacid,0; cester = cwater = 0. Among the simplifying assumptions of this model: the fixed value of the kinetic constant k, which actually depends on the ratio of the mass of the catalyst to the volume of the solution, constant permeability Pwater of the membrane and negligible permeation of reactants and ester.

© Woodhead Publishing Limited, 2013

132

Handbook of membrane reactors

Through a dimensional analysis it can be recognized that some dimensionless parameters determine the behaviour of the system. Of course the choice is not unique, but a convenient combination is: K (the equilibrium Vr ) . constant), M calcohol,0 cacid, acid id,0 and δ = Pwater A (kV These three parameters (or other equivalent dimensionless groups) must appear in whatever formulation of this type of problem (the esterification reaction in PVRs), possibly together with other parameters which take into account other aspects such as: additional phenomena (for example concentration polarization of the membrane), the presence of products in the initial mixture, the concentration of the catalyst and more complex constitutive equations. The dimensionless parameters have a more general validity than the individual dimensional parameters that appear grouped into them and characterize more univocally the behaviour of the system. The adoption of the parameter δ, the ratio of the characteristic rate of permeation to the characteristic rate of reaction, can be extended to any PVR and in general also to any membrane reactor. With this approach the comparison between different studies on PVRs is more direct and meaningful. On the other hand, the less acceptable, though often employed, dimensional parameter A/Vr is comprised in the definition of δ. PVRs equipped with hydrophilic membranes can be used for any liquid reaction where the water produced limits the equilibrium conversion or acts as an inhibiting agent. For example, PVRs have been studied for the dehydration reaction of butenediol to form tetrahydrofuran (Liu and Li, 2002), the synthesis of methylisobutylketone (Staudt-Bickel and Lichtenthaler, 1996), the Knoevenagel condensation reaction between benzaldehyde and ethyl cyanoacetate or ethyl acetoacetate or diethyl malonate (Zhang et al., 2004). Additional information can be found in reviews by Sanchez Marcano and Tsotsis (2002) and by Van der Bruggen (2010).

3.4.2

Pervaporation bioreactors (PVBRs)

In PVBRs the preferentially permeated compounds are organic products of the biological reaction, which often takes place in an aqueous solution. The main objective is to remove these compounds, since they have an inhibitory effect on the biological reaction. At the same time, a partial purification of the permeating product is achieved, which can be beneficial for further refinement. To this aim the separation factor of the product with respect to water should be high, thus reducing the energy wasted in evaporation of water downstream from the membrane. Good performance is also required for the organic−organic separation of the product from the organic substrate precursor, which must be retained in the biological reactor together with water. Therefore, the correct

© Woodhead Publishing Limited, 2013

Pervaporation membrane reactors

133

choice of membrane material is important as the feasibility of a given integrated process depends largely on the availability of a suitable membrane. Consequently a larger variety of organophilic membranes have been developed for the selective recovery of organics than for the removal of water. Much research on this topic is reported in the scientific literature and the obtained results are promising, but the process is not problem-free. For further information on this type of pervaporation reactor, see the following reviews that discuss the subject: Lipnizki et al. (2000); Sanchez Marcano and Tsotsis (2002); Vane (2005) and Wasewar and Pangarkar (2006). It is worthwhile mentioning some of the potential problems that are commonly encountered which can limit such applications, since this information can be useful when comparing with other alternative processes. The main bottlenecks of PVBRs are listed below: •

•

•

In broth fermentation the formation of CO2, which usually permeates with a high volumetric flow rate, makes vacuum pervaporation difficult and expensive. Bio-fouling on the membrane can reduce both the flux and selectivity. Microfiltration of the feed to the membrane unit can diminish this inconvenience. Alternatively, modules designed to establish hydrodynamic conditions that minimize the fouling can be utilized. However these solutions increase the cost and are not completely trouble-free. Relatively high temperatures are not compatible with the biological environment in the reactor. On the other hand the flux would be favoured by relatively high temperatures, thus reducing the membrane area and the capital costs. Advantages and disadvantages of increasing the temperature are widely discussed by Vane, 2005 (see Table 3.2). The conclusion is that the advantages of operating at higher temperatures prevail over the disadvantages; nonetheless, the operating temperature is ultimately a compromise between conflicting requirements.

3.4.3

Pervaporation photocatalytic reactors (PVPRs)

There are two major difficulties in the integration of a biological reaction with pervaporation: the difficulty of operating at a relatively high temperature to avoid the thermal death of the microorganisms or the deactivation of enzymes, and the (bio-)fouling effect of the fermentation broth on the membrane. The low temperatures limit the membrane flux, so that the capital cost for the large membrane area required increases. This is particularly detrimental for low volatile chemicals, such as many aromas, which would need

© Woodhead Publishing Limited, 2013

134

Handbook of membrane reactors

Table 3.2 Advantages and disadvantages of operating pervaporation at the maximum possible temperature Advantages

Disadvantages

Highest flux

Harmful to temperature-sensitive compounds and microorganisms Higher likelihood of precipitate formation Increased heater capital costs

Highest allowable permeate pressure (reduced vacuum requirement) Lowest permeate condensation costs (ability to operate at higher coolant temperatures) Yields system with the lowest required membrane area

Requires high temperature heat source Requires heat exchanger to recover heat from residual stream Material failure more likely Increased material property requirements Requires more insulation Reduced separation factor (dependent on compound and membrane properties)

Source: Adapted from Vane (2005).

relatively high temperatures to obtain acceptable fluxes (Böddeker et al., 1990, 1997). The problem could be solved by using separate equipments for the bioreaction and the pervaporation process, which should be kept at different optimized temperatures, and resorting to a heat-recovery system to minimize the energy losses, but even this solution has a cost. In the case of PVRs for esterification, or other equilibrium reactions which produce water as a by-product, the high concentration of some chemically aggressive reagents, which are typical for these reactions, can detrimentally affect the resistance of the membrane, reducing its life as a result. However, as a rule, membranes which can produce high separation performances and high chemical and thermal resistance are more costly. The feasibility of an integrated photocatalysis−pervaporation process can be preliminarily evaluated by taking into account some characteristics of photocatalysis. In the typical applications of photocatalysis, the reagents, which are termed ‘substrates’ in the common terminology of photocatalysis (Braslavsky et al., 2011), should be present in low concentrations in order to ensure the economy of the process. Therefore the system is not as chemically aggressive for the membrane, particularly if the pervaporation modules are kept in separate equipments. In this way, the membrane is not exposed to the radiation or to the photoactivated radicals, which are present only close to the surface of the photocatalyst, as long as it is illuminated, since the lifetime

© Woodhead Publishing Limited, 2013

Pervaporation membrane reactors

135

of hydroxyl radicals is very short, in the order of 10−7−10−10 s (see e.g., Ollis et al., 1991). The photocatalytic powders, which are used in slurry photocatalytic reactors (De Lasa et al., 2005), are not harmful to the PV membrane and the potential fouling does not alter the performance of the membrane (Camera-Roda et al., 2011a). Furthermore, photocatalysis is not a thermally activated reaction, so the effect of changes in temperature is minor and the operative temperature can be chosen to optimize pervaporation. With photocatalysis the main problems encountered in the previously examined applications of PVRs are avoided. It is also worth noting that: (i) the coupling of photocatalysis with the pervaporation process allows for a complementary exploitation of the solar spectrum. In fact, photocatalysis is normally only able to use its ultraviolet component, while the remaining thermal part of the spectrum can be utilized to heat the fluid and to evaporate the permeate; (ii) photocatalysis and pervaporation have common operative conditions: liquid solutions (often aqueous solutions), low concentrations of the reactants and consequently of the products, low temperature and atmospheric pressure; and (iii) once the type of light source is chosen, photocatalysis is a modular process like pervaporation. Therefore the integration of the two processes is straightforward and advantageous. Due to the fact that the photocatalyst is typically a non-toxic substance (titanium dioxide in most of the applications) and that no chemical additive is needed, alongside the low energy demand which can be satisfied by the solar radiation, it is evident that an integrated pervaporation−photocatalysis process can absolutely satisfy the requirements of sustainable chemistry. Photocatalytic membrane reactors have been analysed over the years (see e.g., Damodar and You, 2010; Grzechulska-Damszel et al., 2009; Molinari et al., 2002, 2008, 2009, 2010; Mozia et al., 2008, 2010), but only more recently has photocatalysis been coupled with pervaporation in an integrated process. The first study of a PVPR was carried out a few years ago by Camera-Roda and Santarelli (2007), and, recently studies have gone on to investigate the possible application of PVPR in the field of photocatalytic synthesis of fine chemicals. Till now, studies of PVPR have not been numerous; nonetheless, the potential applications appear to be of real interest and this area of research is expected to grow rapidly. Despite the low number of existing works on PVPRs, the investigations tend to cover the two main common applications of organophilic pervaporation i.e. the recovery of aroma compounds from process streams (Karlsson and Trägårdh, 1993; Pereira et al., 2006; Trifunović et al., 2006) and the removal of volatile organic compounds (VOC) from aqueous effluents (Konieczny et al., 2008; Lipnizki and Field, 2002; Peng et al., 2003; Urkiaga et al., 2002). In photocatalysis, a photon with appropriate wavelength (that has suitable energy) absorbed by a semiconductor may cause the passage of an electron

© Woodhead Publishing Limited, 2013

136

Handbook of membrane reactors

from the valence band to the conduction band with the concurrent formation of a hole in the valence band. The hole can recombine with the electron, losing the chance to activate a reaction, or it can produce a highly oxidizing agent, which, in the presence of liquid water, is principally the hydroxyl radical. Therefore, photocatalysis is classified as an advanced oxidation technology (AOT). The oxidizing strength of photocatalysis in aqueous solution allows the degradation of a very large spectrum of organic molecules, which can be degraded until complete mineralization is achieved. However, the mineralization process is not direct − it is reached through the formation of several intermediate compounds, except in a few cases with simple molecules such as formic acid. On the other hand, a low selectivity accompanies the oxidation capability in aqueous solution. The potential of photocatalysis in the destruction of a recalcitrant pollutant has been largely demonstrated (Ahmed et al., 2011; Blanco-Galvez et al., 2007; Chong et al., 2010; Gaya and Abdullah, 2008; Occulti et al., 2008; Ray, 2009; Thakur et al., 2010), but in recent years photocatalysis has also been successfully utilized for the synthesis of fine chemicals (Augugliaro et al., 2010, 2011; Hermann and Lacroix, 2010; Maurino et al., 2008; Palmisano et al., 2007; Shiraishi and Hirai, 2008), principally by partial oxidation of selected substrates. It is worth noting that pervaporation is also a candidate for the removal of pollutant VOCs from aqueous stream (Konieczny et al., 2008; Peng et al., 2003; Schofield et al., 1991; Urkiaga et al., 2002). Its efficiency in recovering organic compounds has been utilized, as mentioned previously, in the biological production of different organic compounds. Therefore the idea of integrating these two processes appears absolutely reasonable. Camera-Roda and Santarelli (2007) investigated the coupling of photocatalysis with pervaporation for the detoxification of an aqueous solution of 4-chlorophenol (4-CP). The presence of chlorinated organic compounds at low, but unacceptable, levels in water streams is a common problem that biological degradation cannot solve. This is principally because these pollutants can be highly refractory to a biological attack but also because the biological process would not be economically convenient for such low concentrations of pollutants. The scheme of the PVPR is shown in Fig. 3.5. Photocatalysis alone, or PV alone, are candidates for the treatment of these contaminated effluents, but the effectiveness of the process can be unsatisfactory. In this system photocatalysis is hindered by the formation of several intermediate compounds which slow down the rate of degradation of 4-CP (Camera-Roda and Santarelli, 2007; Camera-Roda et al., 2011a; Satuf et al., 2008; Theurich et al., 1996) since they are competitive with 4-CP for the utilization of the produced oxidation agents and for the adsorption

© Woodhead Publishing Limited, 2013

Pervaporation membrane reactors

137

UV lamp

Photocatalytic reactor

Liquid nitrogen traps Product stream

Pervaporation Tank

Membrane-cell

Incondensables to vacuum

Permeate

3.5 Scheme of a pervaporation photocatalytic reactor.

Initial rate of disappearance of 4-CP mg/(L h)

18 16 14 12 10 8 6 4 2 0 [4CP]=100 ppm

[4-CP]=100 ppm, [HQ]=50 ppm

[4-CP]=100 ppm, [BQ]=50 ppm

[4-CP]=100 ppm, [HQ]=50 ppm, [BQ]=50 ppm

3.6 Decrease of the rate of degradation of 4-chlorophenol (4-CP) in the presence of hydroquinone (HQ) and benzoquinone (BQ).

on the active sites of the catalyst (see Fig. 3.6). Also the removal of 4-CP by pervaporation is not very effective due to the low permeability and selectivity of the membranes towards 4-CP. The integration of photocatalysis and pervaporation shows positive effects on both the processes thus establishing a real synergy and a process intensification. Indeed, it is observed that the selectivity of the membrane towards some intermediates is high. So pervaporation removal takes advantage of the presence of photocatalysis which transforms the poorly permeable 4-CP into more permeable compounds. At the same time the rate of photocatalysis degradation of 4-CP is enhanced by PV which removes intermediate compounds that would hinder this reaction.

© Woodhead Publishing Limited, 2013

138

Handbook of membrane reactors

The degree of intensification is quantified by comparing the enhanced rate of disappearance of 4-CP in the integrated process with the sum of the rates that are obtained in PV alone and in photocatalysis alone. This sum represents an optimistic (conservative in the comparison) evaluation of the rate that would be achieved by operating sequentially with the two processes. An intensification factor, Ei, can be defined as

Ei =

Actual rate of disappearance of 4 4-CP in the integrated process Rate of disappearance of 4-CP C due to PC alone + rate of disappearance of organics due to PV alone

Various experiments, carried out with different photocatalytic reactors and commercial pervaporation membranes, by varying the relative weight of pervaporation with respect to photocatalysis, have demonstrated that the intensification factor depends solely on the parameter δ previously defined, as is apparent in Fig. 3.7. From the definition of Ei, it is obvious that, both when δ → 0 (only photocatalysis) and when δ → ∞ (only pervaporation), the intensification factor approaches unity. So a maximum is expected at an intermediate value of δ. The results in Fig. 3.7 demonstrate the existence of an optimal value of δ, which maximizes the intensification. Furthermore, it is worth noticing that the process is significantly enhanced even at relatively low pervaporation rate, since the optimal value of δ is significantly less than unity (δ ≈ 0.1). 2

Ei

1.75

1.5

1.25

1 0.01

0.1

1

10

δ

3.7 The intensification factor vs δ (adapted from Camera-Roda and Santarelli, 2007). The various symbols indicate experiments carried out with different photocatalytic reactors and different commercial pervaporation membranes.

© Woodhead Publishing Limited, 2013

Pervaporation membrane reactors

139

On the contrary, in most applications of PVRs, the yield of the integrated process monotonically increases with δ, and economic factors determine the optimal value of δ. PVPRs have also been applied to the photocatalytic synthesis of organic compounds. Camera-Roda et al. (2011a, 2011b) investigated the photocatalytic green synthesis of some aromatic aldehydes. It was observed that these products are intermediate compounds in a scheme of reactions in series. For instance, benzaldehyde and 4-methoxybenzaldehyde can be produced by partial oxidation of the corresponding primary alcohols in diluted aqueous solutions, but in the photocatalytic reactor they undergo further oxidation which destroys them. Organophilic pervaporation membranes (composite POMS/PEI membranes produced by GKSS (Forschungszentrum, Geesthacht)) were utilized to recover the aldehydes during their production, in this way preventing their degradation. Therefore, in this application the permeate represents the product stream. The membranes also retained the photocatalytic powders completely, which are usually preferred to support photocatalytic films thanks to their higher photocatalytic activity, but whose post-process separation is usually complex because of their nano-scale dimensions. The membranes were also scarcely permeable to the reagents, which for the most part remained within the reactive solution. In most cases, it has been observed that the separation factor between the aromatic aldehydes and the corresponding primary alcohol (Camera-Roda et al. 2011a; Lamer et al., 1996) is much higher than the one achievable between the ester and the alcohol, which is utilized as a reactant in an esterification reaction (Baudot and Marin, 1997). Fouling was not found to be a problem in this process and the performance of the membrane did not decay in all the time it was being utilized (more than 500 h). Interestingly, the rate of production was enhanced by the recovery of the aldehydes, since it is possible to limit the slowdown of the rate of reaction caused by the competitive utilization by the aldehydes of the photocatalyst active sites and the photo-generated oxidation agents (Camera-Roda et al., 2011a). In this system the process intensification is the result of multiple mechanisms acting simultaneously, such as: (i) the recovery of a product which could undergo an undesired reaction (as in esterification reactions where the undesired reaction is the reverse reaction), (ii) the removal of an inhibiting compound (as in PVBRs), (iii) the retention of the catalyst in the system, and (iv) the conservation of most of the unused reactants in the reacting solution. It was experimentally observed (Camera-Roda et al., 2011a) and confirmed by the mathematical model of a continuous process (Camera-Roda et al., 2011b) that the integration of the two processes brings an enhancement to the selectivity and the conversion for any residence time in the system. An example of the results of the mathematical model for the production of benzaldehyde in a continuous system is presented in Fig. 3.8.

© Woodhead Publishing Limited, 2013

140

Handbook of membrane reactors δ = 2.5, R = 0.95

0.4 0.35

Selectivity 0.3

Yield

PVR

0.25 0.2 0.15 Without PV 0.1 0.05 0 0

5

10

15

Residence time (h) (volume of reactor/feed volumetric flow rate)

3.8 Calculated values of selectivity and yield in a continuous system for the production of benzaldehyde with and without pervaporation.

It has also been demonstrated that the extent of the enhancement increases with the ratio δ of the characteristic time of reaction to the characteristic time of permeation, as long as the flow rate of the permeate, which increases with the membrane area, remains below the feed flow rate. The integrated process was also adopted for the production of vanillin (Augugliaro et al., 2011; Camera-Roda et al., 2011b), the most important aroma for industrial use, which is photocatalytically synthesized starting from an aqueous solution of ferulic acid or other aromatic precursors, and selectively and continuously recovered utilizing a PEBAX® 2533 membrane prepared by solvent casting. A significant purification of the product in the permeate (also thanks to the fractional condensation of the permeate with a deposition of part of the permeated vanillin as virtually pure crystals) and an enhancement of the yield were observed. In conclusion a PVPR applied to the photocatalytic synthesis of organic compounds displays the following interesting features: • increased yield and selectivity; • absence of the photocatalytic powders in the permeate product stream; • increased purity of the product; • low energy requirement; • green processing (mild conditions: low harmless temperatures and atmospheric pressure, reagent in aqueous solution without any chemical

© Woodhead Publishing Limited, 2013

Pervaporation membrane reactors

141

additive, possibility of using natural precursors, possibility of utilizing solar radiation to satisfy the energy requirements); • modularity − in fact both the outputs of the reaction unit and of the membrane separation unit can be varied disjointedly by adding or removing modules. It is worth noting that modular processes, which have higher scale-up factors in cost than non-modular processes, are particularly suitable for relatively small scale plants, such as the ones typical of the fine chemical industry; • possibility to choose between continuous or semi-continuous (batch) processing; and • very simple control so that specially qualified technicians are not required and the actions of operators are minimized.

3.5

Conclusions and future trends

The main enhancements of PVRs are likely to come through the development of better performing membranes, whether functionalized or not. Water selective membranes (hydrophilic) have already reached satisfactory separation performance levels, but their resistance requires improvement without affecting selectivity and permeation flux in order to withstand chemically and thermally aggressive reacting solutions, such as the ones that can be encountered in the esterification reactors. Organophilic membranes need to be improved in terms of their organic selectivity with respect to water, mainly in order to reduce the amount of water (whose specific latent heat of vaporization is particularly high), which is uselessly vaporized downstream from the membrane wasting energy. Also the selectivity of the products with respect to organic reactants must be augmented in order to reduce the loss of the reactants into the permeate. To this aim, fractional condensation is a technique worth investigating, since it might allow for the separation of the permeated reactants and their recycle to the reactor with limited additional cost. In some cases, for example vanillin, the permeated product may deposit as almost pure crystals, thus obtaining a direct separation from the other permeating compounds. Hybrid membranes are probably the most likely future solution, since they promise good performance and satisfactory stability at affordable production costs. Other reactive systems, which till now have not been investigated, could be taken into consideration in the future. In fact, despite the large number of systems where equilibrium limits are removed (see e.g., PVRs for esterification reactions), few works have studied the enhancement of the yield of intermediate products from reactions in series by their selective recovery.

© Woodhead Publishing Limited, 2013

142

Handbook of membrane reactors

It should be noted that reversible reactions can be considered a special case of reactions in series, where the reversed reaction is the consecutive reaction. This is most likely due to the presumed lack of membranes with sufficient selectivity to separate organic products from organic reactants. However for some systems (e.g., aromatic aldehydes from the corresponding primary alcohols) this separation is already effective. In the detoxification of water streams contaminated by VOC, reactions ‘designed’ to produce highly permeable compounds can assist the task of pervaporation in removing the pollutants. Future developments are also expected in thermopervaporation to obtain improved energy recovery and energy savings, as well as for fractional condensation, which is useful to obtain higher purification and better separation even with membranes which are not very selective. The coupling of PVR with other processes, such as distillation or reactive distillation, is a field which offers a high number of potential combinations, many of which are worth further investigation. Growth in the number of possible applications is expected; for example PVPR could open the way to a new class of green syntheses which is still largely unexplored.

3.6

References

Agirre I, Güemez M B, van Veen H M, Motelic A, Vente J F and Arias P L (2011), ‘Acetalization reaction of ethanol with butyraldehyde coupled with pervaporation. Semi-batch pervaporation studies and resistance of HybSi® membranes to catalyst impacts’, J Membr Sci, 371, 179–188. Ahmed S, Rasul M G, Brown R and Hashib M A (2011), ‘Influence of parameters on the heterogeneous photocatalytic degradation of pesticides and phenolic contaminants in wastewater: a short review’, J Environ Manag, 92, 311–330. Assabumrungrat S, Phongpatthanapanich J, Praserthdam P, Tagawa T and Goto S (2003), ‘Theoretical study on the synthesis of methyl acetate from methanol and acetic acid in pervaporation membrane reactors: effect of continuous-flow modes’, Chem Eng J, 95, 57–65. Augugliaro V, El Nazerb H A, Loddo V, Mele A, Palmisano G, Palmisano L and Yurdakal S (2010), ‘Partial photocatalytic oxidation of glycerol in TiO2 water suspensions’, Catal Today, 151, 21–28. Augugliaro V, Camera-Roda G, Loddo V, Palmisano G, Palmisano L, Parrino F and Puma M A (2011), ’Synthesis of vanillin in water by TiO2 photocatalysis’, Appl Catal B-Environ, 111–112, 555–561. Bagnell L, Cavell K, Hodges A M, Mau AW-H and Seen A J (1993), ‘The use of catalytically active pervaporation membranes in esterification reactions to simultaneously increase product yield, membrane permselectivity and flux’, J Membr Sci, 85, 291–299. Baudot A and Marin M (1997), ‘Pervaporation of aroma compounds: comparison of membrane performances with vapour-liquid equilibria and engineering aspects of process improvement’, Trans IChemE: part C, 75, 117–142.

© Woodhead Publishing Limited, 2013

Pervaporation membrane reactors

143

Benedict D J, Parulekar S J and Tsai S-P (2003), ‘Esterification of lactic acid and ethanol with/without pervaporation’, Ind Eng Chem Res, 42, 2282–2291. Benedict D J, Parulekar S J and Tsai S-P (2006), ‘Pervaporation-assisted esterification of lactic and succinic acids with downstream ester recovery’, J Membr Sci, 281, 435–445. Blanco-Galvez J, Fernández-Ibáñez P and Malato-Rodríguez S (2007), ‘Solar photocatalytic detoxification and disinfection of water: recent overview’, J Sol Energy Eng, 129, 4–15. Böddeker K W (1990), ‘Terminology in pervaporation’, J Membr Sci, 51, 259–272. Böddeker K W, Bengtson G and Bode E (1990), ‘Pervaporation of low volatility aromatics from water’, J Membr Sci, 53, 143–158. Böddeker K W, Gatfield I L, Jähnig J J and Schorm C (1997), ‘Pervaporation at the vapor pressure limit: vanillin’, J Membr Sci, 137, 155–158. Braslavsky S E, Braun A E, Cassano A E, Emeline A V, Litter M I, Palmisano L, Parmon V N and Serpone N (2011), ‘Glossary of terms used in photocatalysis and radiation catalysis (IUPAC Recommendations 2011)’, Pure Appl Chem, 83, 931–1014. Brazinha C and Crespo J G (2009), ‘Aroma recovery from hydro alcoholic solutions by organophilic pervaporation: modelling of fractionation by condensation’, J Membr Sci, 341, 109–121. Brazinha C, Barbosa D S and Crespo J G, (2011), ‘Sustainable recovery of pure natural vanillin from fermentation media in a single pervaporation step’, Green Chem, 13, 2197–2203. Camera-Roda G and Santarelli F (2007), ‘Intensification of water detoxification by integrating photocatalysis and pervaporation’, J Sol Energy Eng-T ASME, 129, 68–73. Camera-Roda G, Santarelli F, Augugliaro V, Loddo V, Palmisano G, Palmisano L and Yurkadal S (2011a), ‘Photocatalytic process intensification by coupling with pervaporation’, Catal Today, 161, 209–213. Camera-Roda G, Augugliaro V, Loddo V, Palmisano G and Palmisano L (2011b), Production of aldehydes by oxidation in aqueous medium with recovery of the product by means of pervaporation, patent pending, PCT/IB2011/052541. 2011. Camera-Roda G and Santarelli F (2012), ‘Design of a pervaporation photocatalytic reactor for process intensification’, Chem Eng Technol, 35, 1221–1228. Castanheiro J E, Ramos A M, Fonseca I M and Vital J (2006), ‘Esterification of acetic acid by isoamylic alcohol over catalytic membranes of poly(vinyl alcohol) containing sulfonic acid groups’, Appl Catal A-Gen, 311, 17–23. Chong M N, Jin B, Chow C W K and Saint C (2010), ‘Recent developments in photocatalytic water treatment technology: a review’, Water Res, 44, 2997–3027. Damodar R A and You S-J (2010), ‘Performance of an integrated membrane photocatalytic reactor for the removal of Reactive Black 5’, Sep Purif Technol, 71, 44–49. David M-O, Gref R, Nguyen T Q and Néel J (1991a), ‘Pervaporation-esterification coupling: part I. Basic kinetic model’, Trans Inst Chem Eng, 69, 335–340. David M-O, Nguyen T Q and Néel J (1991b), ‘Pervaporation-esterification coupling: part II. Modelling of the influence of different operating parameters’, Trans Inst Chem Eng, 69, 341–346.

© Woodhead Publishing Limited, 2013

144

Handbook of membrane reactors

de la Iglesia Ó, Mallada R, Menéndez M and Coronas J (2007), ‘Continuous zeolite membrane reactor for esterification of ethanol and acetic acid’, Chem Eng J, 131, 35–39. De Lasa H, Serrano B and Salaices M (2005), Photocatalytic Reaction Engineering, Springer Science+Business media, New York. Ch.2. Delgado P, Sanz M T, Beltrán S and Núñez L A (2010), ‘Ethyl lactate production via esterification of lactic acid with ethanol combined with pervaporation’, Chem Eng J, 165, 693–700. Domingues L, Recasens F and Larrayoz M A (1999), ‘Studies of a pervaporation reactor: kinetics and equilibrium shift in benzyl alcohol acetylation’, Chem Eng Sci, 54, 1461–1465. Feng X and Huang R Y M (1996), ‘Studies of a membrane reactor: esterification facilitated by pervaporation’, Chem Eng Sci, 51, 4673–4679. Figueiredo K C S, Salim V M M and Borges C P (2008), ‘Synthesis and characterization of a catalytic membrane for pervaporation-assisted esterification reactors’, Catal Today, 133–135, 809–814. Figueiredo K C S, Salim V M M and Borges C P (2010), ‘Ethyl oleate production by means of pervaporation-assisted esterification using heterogeneous catalysis’, Braz J Chem Eng, 27, 1–11. Gao Z, Yue Y and Li W (1996), ‘Application of zeolite-filled pervaporation membrane’, Zeolites, 16, 70–74. Gaya U I and Abdullah A H (2008), ‘Heterogeneous photocatalytic degradation of organic contaminants over titanium dioxide: a review of fundamentals, progress and problems’, J Photochem Photobiol C, 9, 1–12. Gonçalves F D, Borges L E P and Borges C P (2004), ‘Synthesis of ethyl acetate by coupling a heterogeneous catalytic system with a pervaporation unit’, Separ Sci Technol, 39, 1485–1500. Grzechulska-Damszel J, Tomaszewska M and Morawski A W (2009), ‘Integration of photocatalysis with membrane processes for purification of water contaminated with organic dyes’, Desalination, 241, 118–126. Hasanoğlu A, Salt Y, Keleşer S and Dinçer S (2009), ‘The esterification of acetic acid with ethanol in a pervaporation membrane reactor’, Desalination, 245, 662–669. Hermann J-M and Lacroix M (2010), ‘Environmental photocatalysis for green chemistry’, Kinet Catal, 51, 793–800. Huang R Y M and Rhim J W (1991), ‘Separation characteristics of pervaporation membrane separation processes’, in: Huang R Y M ed., Pervaporation Membrane Separation Processes, Elsevier, Amsterdam. Inoue T, Nagase T, Hasegawa Y, Kiyozumi Y, Sato K, Nishioka M, Hamakawa S and Mizukami F (2007), ‘Stoichiometric ester condensation reaction processes by pervaporative water removal via acid-tolerant zeolite membranes’, Ind Eng Chem Res, 46, 3743–3750. Jennings J F and Binning R C, (1960), Removal of water generated in organic chemical reactions. US Patent. 2956070. Karlsson H O E and Trägårdh G (1993), ‘Pervaporation of dilute organic-waters mixtures. A literature review on modelling studies and applications to aroma compound recovery’, J Membr Sci, 76, 121–146.

© Woodhead Publishing Limited, 2013

Pervaporation membrane reactors

145

Keurentjes J T F, Janssen G H R and Gorissen J J (1994), ‘The esterification of tartaric acid with ethanol: kinetics and shifting the equilibrium by means of pervaporation’, Chem Eng Sci, 49, 4681–4689. Khajavi S, Jansen J C and Kapteijn F (2010), ‘Application of a sodalite membrane reactor in esterification – Coupling reaction and separation’, Catal Today, 156, 132–139. Kita H, Sasaki S, Tanaka K, Okamoto K-I and Yamamoto M (1988), ‘Esterification of carboxylic acid with ethanol accompanied by pervaporation’, Chem Lett, 17, 2025–2028. Konieczny K, Bodzek M and Panek D (2008), ‘Removal of volatile compounds from the wastewaters by use of pervaporation’, Desalination, 223, 344–348. Korkmaz S, Salt Y, Hasanoğlu A, Ozkan S, Salt I and Dincer S (2009), ‘Pervaporation membrane reactor study for the esterification of acetic acid and isobutanol using polydimethylsiloxane membrane’, Appl Catal A-Gen, 366, 102–107. Korkmaz S, Salt Y and Dincer S (2011), ‘Esterification of acetic acid and isobutanol in a pervaporation membrane reactor using different membranes’, Ind Eng Chem Res, 50, 11657–11666. Krupiczka R and Koszorz Z (1999), ‘Activity-based model of the hybrid process of an esterification reaction coupled with pervaporation’, Sep and Purif Technol, 16, 55–59. Lamer T, Spinnler H E, Souchon I and Voilley A (1996), ‘Extraction of benzaldehyde from fermentation broth by pervaporation’, Process Biochem, 31, 533–542 Lauterbach S and Kreis P (2006), ‘Experimental and theoretical investigation of a pervaporation membrane reactor for a heterogeneously catalysed esterification’, Desalination, 199, 418–420. Levenspiel O (1999), Chemical Reaction Engineering, Third edition, John Wiley & Sons, New York. Lim S Y, Park B, Hunga F, Sahimi M and Tsotsis T T (2002), ‘Design issues of pervaporation membrane reactors for esterification’, Chem Eng Sci, 57, 4933–4946. Lipnizki F, Field R W and Ten P-K (1999), ‘Pervaporation-based hybrid process: a review of process design, applications and economics’, J Membr Sci, 153, 183–210. Lipnizki F, Hausmanns S, Laufenberg G, Field R and Kunz B (2000), ‘Use of pervaporation-bioreactor hybrid processes in biotechnology’, Chem Eng Technol, 23, 569–577. Lipnizki F and Field R W (2002), ‘Hydrophobic pervaporation for environmental applications: process optimization and integration’, Environ Prog, 21, 265–272. Liu Q, Zhang Z and Chen H (2001), ‘Study on the coupling of esterification with pervaporation’, J Membr Sci, 182, 173–181. Liu Q L and Chen H F (2002), ‘Modeling of esterification of acetic acid with n-butanol in the presence of Zr(SO4)2·4H2O coupled pervaporation’, J Membr Sci, 196, 171–178. Liu Q-L and Li Q-B (2002), ‘Membrane of PVA coated on porous catalytic ceramic disks supported H3PW12O40’, J Membr Sci, 202, 89–95. Lutze P, Ganib R and Woodley J M (2010), ‘Process intensification: A perspective on process synthesis’, Chem Eng Process, 49, 547–558.

© Woodhead Publishing Limited, 2013

146

Handbook of membrane reactors

Ma J, Zhang M, Lu L, Yin X, Chen J and Jiang Z (2009), ‘Intensifying esterification reaction between lactic acid and ethanol by pervaporation dehydration using chitosan–TEOS hybrid membranes’, Chem Eng J, 155, 800–809. Maurino V, Bedini A, Minella M, Rubertelli F, Pelizzetti E and Minero C (2008), ‘Glycerol transformation through photocatalysis: a possible route to value-added chemicals’, J Adv Oxid Technol, 11, 184–192. Mitkowski P T, Buchaly C, Kreis P, Jonsson G, Górak A and Gani R (2009), ‘Computer aided design, analysis and experimental investigation of membrane assisted batch reaction–separation systems’, Comput Chem Eng, 33, 551–574. Mohan K and Govind R (1988), ‘Effect of temperature on equilibrium shift in reactors with a permselective wall’, Ind Eng Chem Res, 27, 2064–2070. Molinari R, Palmisano L, Drioli E and Schiavello M (2002), ‘Studies on various reactor configurations for coupling photocatalysis and membrane processes in water purification’, J Membr Sci, 206, 399–415. Molinari R, Caruso A, Argurio P and Poerio T (2008), ’Degradation of the drugs Gemfibrozil and Tamoxifen in pressurized and de-pressurized membrane photoreactors using suspended polycrystalline TiO2 as catalyst’, J Membr Sci, 319, 54–63. Molinari R, Caruso A and Poerio T (2009), ’Direct benzene conversion to phenol in a hybrid photocatalytic membrane reactor’, Catal Today, 144, 81–86. Molinari R, Caruso A and Palmisano L (2010), ‘Photocatalytic processes in membrane reactors’, in: Drioli E and Giorno L eds., Comprehensive Membrane Science and Engineering, Elsevier. Book 3, Oxford,UK. ch. 3.07. Mozia S, Morawski A W, Toyoda M and Inagaki M (2008), ’Effectiveness of photodecomposition of an azo dye on a novel anatase-phase TiO2 and two commercial photocatalysts in a photocatalytic membrane reactor (PMR)’, Sep Purif Technol, 63, 386–391. Mozia S, Morawski A W, Toyoda M and Tsumura T (2010),’ Integration of photocatalysis and membrane distillation for removal of mono- and poly-azo dyes from water’, Desalination, 250, 666–672. Néel J (1995), ‘Pervaporation’, in: Noble R D and Stern S A eds., Membrane separations technology, Principles and Applications, Elsevier Science, Amsterdam,The Netherlands. Ch.5. Nemec D and van Gemert R (2005), ‘Performing esterification reactions by combining heterogeneous catalysis and pervaporation in a batch process’, Ind Eng Chem Res, 44, 9718–9726. Occulti F, Camera-Roda G, Berselli S and Fava F (2008), ‘Sustainable decontamination of an actual-site aged PCB-polluted soil through a biosurfactant-based washing followed by a photocatalytic treatment’, Biotechnol Bioeng, 99, 1525–34. Okamoto K-I, Yamamoto M, Otoshi Y, Semoto T, Yano M, Tanaka K and Kita H (1993), ‘ Pervaporation-aided esterification of oleic acid’, J Chem Eng Jpn, 26, 475–481. Ollis D F, Pelizzetti E and Serpone N (1991), ‘Destruction of water contaminants’, Environ Sci Technol, 25, 1523–1529. Ozdemir S S, Buonomenna M G and Drioli E (2006), ‘Catalytic polymeric membranes: Preparation and application’, Appl Catal A-Gen, 307, 167–183.

© Woodhead Publishing Limited, 2013

Pervaporation membrane reactors

147

Palmisano G, Augugliaro V, Pagliaro M and Palmisano L (2007), ‘Photocatalysis: a promising route for 21st century organic chemistry’, Chem Commun, 33, 3425–3437. Park B-G and Tsotsis T T (2004), ‘Models and experiments with pervaporation membrane reactors integrated with an adsorbent system’, Chem Eng Process, 43, 1171–1180. Peng M, Vane L M and Liu S X (2003), ‘Recent advances in VOCs removal from water by pervaporation’, J Hazard Mater, B98, 69–90. Pereira C C, Ribeiro C P, Nobrega R and Borges C P (2006), ‘Pervaporative recovery of volatile aroma compounds from fruit juices: review’, J Membr Sci, 274, 1–23. Pereira C S M, Silva V M T M, Pinho S P and Rodrigues A E (2010), ‘Batch and continuous studies for ethyl lactate synthesis in a pervaporation membrane reactor’, J Membr Sci, 361, 43–55. Peters T A, vander Tuin J, Houssin C, Vorstman M A G, Benes N E, Vroon Z A E P, Holmen A and Keurentjes J T F (2005a), ‘Preparation of zeolite-coated pervaporation membranes for the integration of reaction and separation’, Catal Today, 104, 288–295. Peters T A, Benes N E and Keurentjes J T F (2005b), ‘Zeolite-coated ceramic pervaporation membranes; pervaporation-esterification coupling and reactor evaluation’, Ind Eng Chem Res, 44, 9490–9496. Peters T A, Benes N E and Keurentjes J T F (2007), ‘Preparation of Amberlyst-coated pervaporation membranes and their application in the esterification of acetic acid and butanol’, Appl Catal A-Gen, 317, 113–119. Ray A K (2009), ‘Photocatalytic reactor configurations for water purification: experimentation and modeling’, Adv Chem Eng, 36, 145–184. Sanchez Fernand ez E, Geerdink P and Goetheer E L V (2010), ‘Thermo pervap: the next step in energy efficient pervaporation’, Desalination, 250, 1053–1055 Sanchez Marcano J G and Tsotsis T T (2002), Catalytic Membranes and Membrane Reactors, Wiley-VCH, Weinheim, Germany. Ch.3. Sanz M T and Gmehling J (2006a), ‘Esterification of acetic acid with isopropanol coupled with pervaporation. Part I: kinetics and pervaporation studies’, Chem Eng J, 123, 1–8. Sanz M T and Gmehling J (2006b), ‘Esterification of acetic acid with isopropanol coupled with pervaporation. Part II. Study of a pervaporation reactor’, Chem Eng J, 123, 9–14. Sarkar B, Sridhar S, Saravanan K and Kale V (2010), ‘Preparation of fatty acid methyl ester through temperature gradient driven pervaporation process’, Chem Eng J, 162, 609–615. Satuf M L, Brandi R J, Cassano A E and Alfano O M (2008), ‘Photocatalytic degradation of 4-chlorophenol: a kinetic study’, Appl Catal B-Environ, 82, 37–49. Schmidt-Traub H and Górak A (2006), ‘Introduction’, in: Schmidt-Traub H and Górak A eds., Integrated reaction and separation operations, Modelling and experimental validation, Springer-Verlag, Berlin, Heidelberg. Ch.1. Schofield W, McRay S, Ray R and Newbold D D (1991), ‘Opportunities for pervaporation in the water-treatment industry’, Proceedings of the fifth international conference on pervaporation processes in the chemical industry, Heidelberg, Germany, pp.409–420.

© Woodhead Publishing Limited, 2013

148

Handbook of membrane reactors

Shiraishi Y and Hirai T (2008), ‘Selective organic transformations on titanium oxide-based photocatalysts’, J Photochem Photobiol C, 9, 157–170. Smitha B, Suhanya D, Sridhar S and Ramakrishna M (2004), ‘Separation of organic–organic mixtures by pervaporation-a review’, J Membr Sci, 241, 1–21. Stankiewicz A and Moulijn J A (2000), ‘Process intensification: transforming chemical engineering’, Chem Eng Prog, 96, 22–34. Stankiewicz A (2003), ‘Reactive separations for process intensification: an industrial perspective’, Chem Eng Process, 42, 137–144. Staudt-Bickel C and Lichtenthaler R N (1996), ‘Integration of pervaporation for the removal of water in the production process of methylisobutylketone’, J Membr Sci, 111, 135–141. Tanaka K, Yoshikawa R, Ying C, Kita H and Okamoto K-I (2001), ‘Application of zeolite membranes to esterification reactions’, Catal Today, 67, 121–125. Tanna N P and Mayadeviy S (2007), ‘Analysis of a membrane reactor: influence of membrane characteristics and operating conditions’, Int J Chem React Eng, 5, article A5. Thakur R S, Chaudhary R and Singh C (2010), ‘Fundamentals and applications of the photocatalytic treatment for the removal of industrial organic pollutants and effects of operational parameters: A review’, J. Renewable Sustainable Energy, 2, 042701 (37 pages). DOI: 10.1063/1.3467511. Theurich J, Linder M and Bahnemann D W (1996), ‘Photocatalytic degradation of 4-chlorophenol in aerated aqueous titanium dioxide suspensions: a kinetic and mechanistic study’, Langmuir, 12, 6368–6376. Trifunović O, Lipnizki F and Trägårdh G (2006), ‘The influence of process parameters on aroma recovery by hydrophobic pervaporation’, Desalination, 189, 1–12. Ulbricht M (2006), ‘Advanced functional polymer membranes’, Polymer, 47, 2217–2262. Urkiaga A, Bolaño N and De Las Fuentes L (2002), ‘Removal of micropollutants in aqueous streams by organophilic pervaporation’, Desalination, 149, 55–60. Vander Bruggen B (2010), ‘Pervaporation membrane reactors’, in: Drioli E and Giorno L eds., Comprehensive Membrane Science and Engineering, Elsevier. Book 3, ch. 3.06. Vane L M (2005), ‘A review of pervaporation for product recovery from biomass fermentation processes’, J Chem Technol Biotechnol, 80, 603–629. Vane L M, Alvarez F R, Mairal A P and Baker R W (2004), ‘Separation of vapor-phase alcohol/water mixtures via fractional condensation using a pilot-scale dephlegmator: enhancement of the pervaporation process separation factor’, Ind Eng Chem Res, 43, 173–183. Velterop F (2011), ‘The potential of the HybSi ceramic membrane in process intensification’, Programme booklet of the International Scientific Conference on Pervaporation, Vapor Permeation and Membrane Distillation, Torun (Poland) 8–11 september 2011, 39–40. Waldburger R M and Widmer F (1996), ‘Membrane reactors in chemical production processes and the application to the pervaporation-assisted esterification’, Chem Eng Technol, 19, 117–126. Wasewar K L (2007), ‘Modeling of pervaporation reactor for benzyl alcohol acetylation’, Int J Chem React Eng, 5, article A6.

© Woodhead Publishing Limited, 2013

Pervaporation membrane reactors

149

Wasewar K L and Pangarkar V G (2006), ‘Intensification of recovery of ethanol from fermentation broth using pervaporation: economical evaluation’, Chem Biochem Eng Q, 20, 135–145. Wasewar K L, Patidar S and Agarwal V K (2008), ‘Pervaporation reactor for esterification of acetic acid with n-butanol: modeling and simulation’, Int J Chem React Eng, 6, article A93. Wasewar K L, Patidar S, Agarwal V K, Rathod A, Sonawane S S, Agarwal R V, Uslu H and Inci I (2010), ‘Performance study of pervaporation reactor (PVR) for esterification of acetic acid with ethanol’, Int J Chem React Eng, 8, article A 57. Xuehui L and Lefu W (2001), ‘Kinetic model for an esterification process coupled by pervaporation’, J Membr Sci, 186, 19–24. Zhang X, Lai E S M, Martin-Aranda R and Yeung K L (2004), ‘An investigation of Knoevenagel condensation reaction in microreactors using a new zeolite catalyst’, Appl Catal A-Gen, 261, 109–118. Zhu Y, Minet R G and Tsotsis T T (1996), ‘A continuous pervaporation membrane reactor for the study of esterification reactions using a composite polymeric/ ceramic membrane’, Chem Eng Sci, 51, 4103–4113. Zhu Y and Chen H (1998), ‘Pervaporation separation and pervaporation-esterification coupling using crosslinked PVA composite catalytic membranes on porous ceramic plate’, J Membr Sci, 138, 123–134. Zou Y, Tong Z, Liu K and Feng X (2010), ‘Modeling of esterification in a batch reactor coupled with pervaporation for production of n-butyl acetate’, Chinese J Catal, 31, 999–1005.

3.7

Appendix: nomenclature

3.7.1

Notation

A c Da Ei k K M  m" n n"’ ig p P Pe R t

membrane area concentration Damköhler number intensification factor kinetic constant of the reaction equilibrium constant ratio of the initial concentration of the alcohol to the initial concentration of the acid permeate mass flux through the membrane order of reaction molar rate of generation of Ai per unit volume pressure permeability Péclet number recycle ratio time

© Woodhead Publishing Limited, 2013

150

Handbook of membrane reactors volume

V V

volumetric flow rate mole fraction

y

Greek symbols separation factor enrichment factor ratio between the characteristic rate of permeation and the characteristic rate of reaction (dimensionless parameter) density

α β δ ρ Subscripts 0 acid alcohol D ester F for i j P PV r rev s S t water

3.7.2 AOT 4-CP CSTR PFR PV PVBR PVPR PVR VOC

initial or inlet conditions acid alcohol diffusivity ester feed forward reaction species Ai species Aj permeate pervaporation reactor reverse reaction substrate solubility total integrated system water

Abbreviations advanced oxidation technology 4-chlorophenol continuous stirred tank reactor plug flow reactor pervaporation pervaporation bioreactor pervaporation photocatalytic reactor pervaporation reactor volatile organic compound

© Woodhead Publishing Limited, 2013

Pervaporation membrane reactors Abbreviations for polymers PAN PDMS PEBA PEI POMS POPMI PVA

polyacrylonitrile polydimethylsiloxane polyether block amide polyetherimide polyoctylmethylsiloxane poly(4,4’-oxydiphenylene pyromellitimide) polyvinyl alcohol

© Woodhead Publishing Limited, 2013

151

4 Multi-phase catalytic membrane reactors A. COMITE , A. BOTTINO, G. CAPANNELLI , C. COSTA and R. DI FELICE , University of Genoa, Italy

DOI: 10.1533/9780857097347.1.152 Abstract: The present chapter deals with multi-phase catalytic membrane reactors where the reactants distributed in different fluid phases meet on catalytic sites located in the structure of a catalytic membrane. The aim of the chapter is to summarize the features of catalytic membrane reactors applied to gas−liquid and liquid−liquid systems in order to show the capabilities, advantages and limitations of this class of emerging multi-phase reactors. In multi-phase systems, the catalytic membrane is usually used as an interface between two fluid phases and allows strict control of the reactant mass transfer with enhancement of the overall heterogeneous kinetics. Moreover, the particular features of the catalyst structured as a membrane can improve the intrinsic kinetics and the catalyst’s effectiveness. This chapter, by discussing some relevant points of catalytic membranes applied to multi-phase systems, aims at being a simple starting point for researchers who would like to begin their investigation in this fascinating field. After an introduction devoted to a first general description of multi-phase reactor systems, the chapter illustrates several contact modalities between reactants distributed in two different phases and the catalytic membrane. The role of mass transfer phenomena and reaction kinetics in multi-phase catalytic membrane reactors is discussed on the basis of simple well-known chemical engineering principles and modelling examples. Typical materials for multi-phase membrane reactors are briefly listed. Finally, some typical applications (e.g., hydrogenation and oxidation reactions) are reported. Key words: catalytic membrane, three-phase catalytic membrane reactor, multi-phase reactions.

4.1

Introduction

Catalytic membrane reactors belong to the class of structured reactors. The catalytic membrane performs several functions (e.g., separation, reactant distribution, catalytic and/or interface roles) in order to enhance the process productivity (in terms of conversion, selectivity or yield of desired products) and/or the process safety. Due to their unique property 152 © Woodhead Publishing Limited, 2013

Multi-phase catalytic membrane reactors

153

of keeping two process volumes physically separated, membrane reactors offer flexibility in choosing the most suitable reactor configuration or contact modality for each specific reactive process. One of the most interesting applications of catalytic membrane reactors is in multi-phase systems (gas−liquid or liquid−liquid). The chapter will introduce multi-phase catalytic membrane reactors starting from general considerations on conventional multi-phase catalytic reactors. Several contact modalities between reactants in multi-phase catalytic membrane reactors will be illustrated. Then the role of mass transfer phenomena and reaction kinetics will be discussed on the basis of simple well-established chemical engineering principles. Literature on multi-phase catalytic membrane reactors concerning modelling, catalytic membrane materials and examples of application processes will be briefly reported. Finally, research suggestions and future trends will be drawn in order to assist investigation and the effort to bring multi-phase catalytic membrane reactors to a more mature application phase. The present chapter will not treat some relevant topics, including dense polymeric catalytic membranes when operating with a single fluid phase, biocatalytic membranes, non-catalytic membranes with an homogeneous catalyst in one of the two phases, and details on the preparation of catalytic membranes, since they are covered in other chapters of the handbook or in reviews which will be cited. For example, aspects of multi-phase enzyme membrane reactors can be found in Sisak et al. (2000). In a multi-phase catalytic reactor, the reacting species are dissolved in two different fluid phases (e.g., in a gas−liquid system or in liquid−liquid system) which are separated by a phase interface, and the catalyst is located in one of the fluid phases or in a dissolved form (as for example an homogeneous catalyst) or as a third heterogeneous phase (e.g., a solid phase). When a gas−liquid system is in the presence of a solid phase catalyst, the reactor is referred to as a three-phase reactor. Multi-phase reactors represent one of the most important classes of chemical reactors and they are widely used in many industrial sectors, as for example chemical, petrochemical, biotechnological, pharmaceutical and food processing industries (Barnett, 2006; Biardi and Baldi, 1999; Henkel, 2000; Nauman, 2008). Multi-phase reactors have typical industrial application in: • hydrocarbon hydrocracking; • coal liquefaction; • Fischer–Tropsch syntheses; • hydrotreating of mineral oil and hydrofinishing of lubricating oils; • hydrogenation processes, as for example the hydrogenation of adiponitrile to hexamethylenediamine, of α-methyl styrene to cumene, or of caprolactone to hexanediol;

© Woodhead Publishing Limited, 2013

154

Handbook of membrane reactors Liquid Gas or liquid A

A, B

Solid catalyst

P

Diffusion of the reactant A to the interface

Diffusion of reactants A and B to the liquid–solid phase interface

Repartition equilibrium of A between the two phases

Simultaneous diffusion and reaction of A and B A+B

P

4.1 Schematic representation of a typical multi-phase system where a gas or liquid phase is dispersed in a second liquid phase containing a solid porous catalyst. (P refers to products as schematized by the reaction: A + B → P.)

• • •

partial oxidation reactions, as for example for the production of oxalic acid; wet-air oxidations with heterogeneous catalysts; biological processes.

In a multi-phase catalytic reactor, the catalyst is usually a solid phase in contact with the liquid phase. Figure 4.1 shows a typical multi-phase catalytic system, where one fluid phase (gas or liquid) is dispersed in a liquid phase which contains porous catalyst particles. The reactants need to diffuse from their respective phases to the catalytic site where reaction products are formed and then they can diffuse back to one or both fluid phases. The overall reaction rate of the process will be affected by the inter-phase mass transfer rates near the gas−liquid and the liquid−solid interfaces, as well as by the intra-phase mass transfer rate competing with the intrinsic reaction rate inside the catalyst structure. In conventional industrial multi-phase reactors, the heterogeneous catalyst can be organized as a packed (or fixed) bed of catalyst particles (e.g., in trickle-bed reactors or in submerged up-flow reactors), as catalyst particles suspended or fluidized in one of the two phases (in the liquid phase of a three-phase reactor, as for example in a slurry-stirred reactor and a slurry-bubbling reactor) or finally as a structured catalyst (e.g., monolith and membrane reactors). Structured catalysts are regular solid structures which reduce randomness through a well-defined structure and shape at a reactor level. The selection of the most appropriate traditional multi-phase

© Woodhead Publishing Limited, 2013

Multi-phase catalytic membrane reactors

155

reactor type is a complex operation that has been summarized by Krishna and Sie (1994). Typically slurry reactors operate in batch or semi-batch mode, and at the end of each reaction cycle the finely powdered catalyst needs to be properly separated. Abrasion phenomena between the catalyst particles due to the mixing operation usually leads to the formation of a fraction of very fine catalyst particles which can easily undergo leaching phenomena. In conventional fixed-bed reactors the fluid phases move on a packed bed of catalyst particles. Since the two phases are not miscible and both need to be in contact with the catalyst, different operation modes are applied to improve the contact between the fluid phases and the catalysts. For example, in a trickle-bed reactor both fluid phases are introduced from the top in a vertically positioned reactor. Unfortunately poor distribution of the fluid phases gives rise to poor performance, local hot spots and sintering of the catalyst. In all the traditional multi-phase reactors, both fluid phases access the catalyst from the same direction with an evident strong and often uncontrollable influence of mass transfer phenomena. Recently, Centi and Perathoner (2003) have emphasized that a successful approach to multi-phase reactions requires a strict integration between both catalyst and reactor design at a nano-scale level (e.g., environment around the active site), at a micro-scale level (e.g., spatial distribution of the active site) and at a macro-scale level (e.g., structured reactors). At the turn of 1990s catalytic membrane reactors were proposed as a novel type of structured three-phase reactors which could improve the contact and the mass transfer of the reacting species on the catalyst. From that time many papers have been published on this specific topic. The simple concept was to use a thin catalytic membrane as a well-defined reacting interface between two fluid phases, in order to minimize the diffusion resistance and enhance the effectiveness of the catalyst (Fig. 4.2). The solid membrane interface between the gas and liquid phases is the meeting and reacting place for the reactants. Therefore the main advantage of a catalytic membrane reactor seems to be the possibility of decoupling mass transfer phenomena and kinetics at a reactor level. The main effort of researchers was initially oriented to exploring the possibilities and the characteristics of catalytic membrane reactors, especially in three-phase systems. Only a few papers are devoted to the application of catalytic membrane reactors to liquid−liquid systems. An excellent and comprehensive review of three-phase catalytic membrane reactors has been published by Dittmeyer et al. (2004). This exhaustive review covers several aspects of the application of catalytic membranes as three-phase reactors and critically discusses some examples in the literature.

© Woodhead Publishing Limited, 2013

156

Handbook of membrane reactors Catalytic membrane A+B

Gas or liquid A

P

Liquid B P

Diffusion of the Diffusion of the reactant A to the reactant B to the interface interface Repartition equilibrium of A, diffusion and reaction

4.2 Schematic representation of a catalytic membrane in a multi-phase system.

In particular, the role of the catalytic membrane in three-phase systems has been defined considering the following benefits: •

simpler processes when the membrane performs a separation function directly into the reactor; • a single unit can carry out the reaction and products separation processes; • an optimal contact between the phases and optimal distribution of reactants in the reactor; • the possibility of supplying a reactant to the catalytic site in the most active or selective form due to the role of the membrane. Therefore, a membrane is not only a simple selective separation barrier, but plays an active role due to the presence of the catalysts, the ability to improve the contact of reactants on the catalytic sites and, finally, the segregating effect of the membrane on the reactants in the different phases. To better understand the role played by a catalytic membrane reactor in a multi-phase system, it can also be fruitful to start from the concept of the

© Woodhead Publishing Limited, 2013

Multi-phase catalytic membrane reactors

157

membrane contactor as a means to control the mass transfer between two segregated phases and to subsequently widen it to the catalytic membrane reactor concept, considering the effect of the mass transfer in the membrane with a catalyst (Bottino et al., 2009; Di Felice et al., 2010). The aim of the chapter is to summarize the general features of catalytic membrane reactors applied to gas−liquid and liquid−liquid systems in order to show the capabilities, advantages and limitations of this emerging class of multi-phase reactors.

4.2

Contact modalities in multi-phase catalytic membrane reactors

A multi-phase catalytic membrane reactor favours the meeting of the reacting species on the catalyst, which is usually located in a thin reacting-zone layer and often acts as an interface between the two phases containing the reactants. In this last case, the membrane offers a geometrically well-defined interface between the two phases, and therefore the reactants can independently reach the catalytic sites located at the interface layer from two different membrane sides. Figure 4.3 compares from a general point of view the concentration profiles of reactants in a conventional three-phase reactor and in catalytic membrane reactors operating in various contact configurations. Similar considerations can be drawn for a liquid−liquid system. As can be observed, the main difference between conventional three-phase reactors and catalytic membrane reactors lies in the relative positions of the mass transfer resistances with respect to the catalytic phase. In a conventional porous catalyst the catalytic sites in the pores have only one way or path of access. The gaseous reactant will encounter the first two mass transfer resistances at the gas−liquid interface, where the solvation equilibrium of the species from one phase to the other will take place. The dissolved species will diffuse towards the surface of the catalytic pellet for quite a long path in the liquid phase and will meet an additional mass transfer resistance at the liquid−solid catalyst interface. It then needs to diffuse and react in the porous structure of the catalyst as well as the other reactant already present in the liquid phase. In the case of a traditional three-phase reactor (Fig. 4.3a), the concentration of at least one of the reacting species is limited by its solubility and diffusion in the other fluid phase with a long diffusion path and in some cases unknown interfacial area (e.g., bubbles with variable size depending on the type of the gas feeding and distribution device in slurry reactors, not uniform phase contact and distribution in trickle-bed reactors). In a porous membrane the pores present two distinct accesses corresponding to the two membrane sides (Figs 4.3b and c). When an asymmetric porous catalytic membrane constituted by a catalytic layer and an inert

© Woodhead Publishing Limited, 2013

Handbook of membrane reactors (b)

(a) Gas phase

(c)

Catalytic membrane Inert porous support

Gas phase

Solid catalytic phase

Liquid phase

A P

Liquid phase

A

Reactant from the gas phase

Inert porous support

B

MT

MT MT R

MT

A

(f)

MT

MT MT R

MT

Solid catalytic phase Liquid phase

Liquid phase

Gas phase A

Catalytic membrane Inert porous support

Liquid phase

Forced flow

B

P

Reactant from the gas phase

Gas phase

P

Dense catalytic phase

Liquid phase

A

MT R

Inert porous support

Reactant from the liquid phase

B Concentration

MT R

(e)

Dense catalytic phase Gas phase

Concentration

MT R

MT R

Concentration

(d)

MT MT MT Reactant from the liquid phase MT

Catalytic membrane

B

P Reactant from the gas phase

Solid catalytic phase

B Concentration

B Concentration

Solid catalytic phase

Liquid phase

Concentration

158

A

P MT R

Reactant from the liquid phase Reactant from the liquid phase

MT

MT

MT

Reactant from the liquid-phase

MT R

4.3 Contact modalities and concentration profiles in catalytic membrane reactors for three-phase systems. The concentration of reactants is represented on the y-axis and the spatial coordinate along the membrane cross-section is represented on the x-axis. Below the scheme of each case the sequence of the mass transfer (MT) resistances and of the reaction event (R) are reported. (a) Traditional slurry reactor; (b) supported thin porous catalytic layer with the liquid impregnating the porosity and the gas phase in contact with the catalytic layer; (c) supported thin porous catalytic layer with the liquid impregnating the porosity and the liquid phase in contact with the catalytic layer; (d) supported dense membrane which is perm-selective to the gas-phase reactant; (e) dense catalytic membrane perm-selective to both reactants in the gas and liquid phases; (f) forced flow of the liquid phase enriched with the gas-phase reactant through the thin catalytic membrane layer.

porous support interfaces two different phases, two main feeding configurations are feasible. The gas phase can be fed near the catalytic layer and the liquid phase along the porous support side (Fig. 4.3b) or vice versa (Fig. 4.3c). When the gas phase is fed directly on the catalyst-side layer and the porous catalytic membrane is wetted by the liquid phase, then it appears that the influence of overall mass transfer resistances can be minimized and the two main mass transfer phenomena from the fluid phases to the solid catalytic phase can occur independently and simultaneously. Moreover if the reaction takes place in a very thin porous layer (usually a few microns) an additional improvement of internal effectiveness can be easily achieved. Figures 4.3d and 4.3e consider the case of dense membranes as interfaces between the fluid phases. The case of Fig. 4.3d is limited to the situation

© Woodhead Publishing Limited, 2013

P

Multi-phase catalytic membrane reactors

159

where the dense layer is both catalytic and absolutely selective only to the reactant present in one of the phases. An example is given by palladiumbased membranes used for hydrogenation. In this case, the reaction takes place directly on the surface of contact between the liquid phase and the dense membrane. When the membrane is made of a precious and costly material, the dense layer should be the thinnest possible, because: • only the surface where both the reactants meet participates in the catalytic reaction; • the permeating flux is inversely proportional to the layer thickness. Figure 4.3e considers a dense catalytic layer which is permeable to both the reactants present in the segregated phases. This is the case of several catalytic polymeric membranes, either unsupported or supported on porous substrates. The reaction rate is governed by the relative rate of diffusion and reaction in the thickness of the catalytic membrane. In this case the solubility of the components in the membrane layer should also be taken into account. Figure 4.3f shows a way of contact of the reactants on the catalytic membrane based on the forced flow of both reactants through the catalytic layer directly from one membrane side. This configuration with respect to traditional reactors can offer an important improvement in the contact of reactants with the catalytic sites. Mass transfer from the gas phase to the liquid phase will occur in the same way as traditional reactors: the liquid phase needs to be previously saturated with the gas reactant. The features of this last feeding configuration have been reported by Reif and Dittmeyer (2003) for both the catalytic nitrite reduction and the dechlorination of chloroform. A particular case of the contact mode sketched in Fig. 4.3f is represented by the use of catalytic dense polymeric membranes working in cross-flow mode on the liquid feed side and in pervaporation mode through the membrane (Bengston et al., 2002). This particular class will be not discussed further, since Chapter 1 of Handbook of membrane reactors Volume 1: Fundamental materials science, design and optimisation is dedicated to polymeric membrane reactors..

4.3 4.3.1

Multi-phase membrane reactors: fundamental concepts, modelling and operations Porous membranes and interfaces between fluid phases

The nature of the membrane material with respect to each fluid phase (e.g., wettability properties), the pore size and its distribution, the pressure

© Woodhead Publishing Limited, 2013

160

Handbook of membrane reactors

of each fluid phase on both membrane sides will determine whether the pores of a catalytic porous membrane will be filled only by one fluid phase or by both of them. For sake of simplicity let us consider for a moment a symmetric porous membrane. A non-wetting fluid can displace the wetting fluid only when the pressure of the former can overcome the ‘breakthrough pressure’ or ‘capillary pressure’, Δp, defined by Laplace equation: Δ = Δp

2σ

ϑ

[4.1]

rp

where Δp is the pressure difference between the non-wetting fluid side and the other side of the pore, rp is the pore radius, σ is the surface tension (or the interfacial tension for liquid−liquid systems) and ϑ is the contact angle between the wetting fluid and the membrane pore. The previous equation reduces to Cantor’s equation when the contact angle is ϑ = 0° (the wetting liquid spontaneously penetrates the pores): Δ = Δp

2σ rp

[4.2]

The above equation shows that by application of a pressure on the non-wetting fluid side higher than the capillary pressure, pores will be filled by the incoming fluid phase. If we consider the typical example of a catalytic porous asymmetric membrane constituted by a thin catalytic layer supported by a macroporous substrate and a wetting liquid phase on the support side and a gas phase on the small pore catalytic side, the liquid will easily fill the pores of both the support and the porous catalytic layer. In order to move the gas−liquid interface from the support towards the catalytic porous layer, a pressure difference of the gas phase has to overcome the capillary pressure of the support. For the same reason, the position of the interface between the two fluid phases inside the porous catalytic layer will depend on the quality of the catalytic layer and on the strict control of the pressure difference between the two membrane compartments. In membrane contactors, usually the condition of gas-phase filled pore is preferred in order to reduce the overall mass transfer resistance across the membrane. In a catalytic membrane, both reactants in the two fluid phases need to reach the catalytic sites in the pore and therefore an ideal situation wherein the interface between the phases is very close to the catalytic sites is to be preferred in order to achieve the maximum reactant concentration in the reaction zone. This situation can be approximated by using a wetting liquid, a thin catalytic layer and fluid−fluid

© Woodhead Publishing Limited, 2013

Multi-phase catalytic membrane reactors

161

meniscus occurring in the catalytic pore located in a thin catalytic membrane layer. The pressure difference between the two membrane sides, namely the transmembrane pressure, determines the position of the gas−liquid interface along the membrane cross-section, since the phase−phase displacement will take place only in pores where the transmembrane pressure is greater than the breakthrough pressure. Therefore, in catalytic membrane reactors operating in contactor mode a strict control of the transmembrane pressure is very important. Iojoiu et al. (2005), using a catalytic ceramic membrane for the wet-air oxidation of formic acid, showed that when the reactor is fed by the liquid phase directly on the catalytic layer and the gas phase on the support side, the application of a gas transmembrane pressure greater than the capillary pressure improves the activity. Indeed, by application of an over-pressure the gas−liquid interface moves from the support side towards the catalytic zone, which is usually located in the top layer and sometimes in a fraction of the intermediate layers. Figure 4.4 shows calculated capillary pressures for the typical pore size of each layer in typical ceramic membranes used for three-phase reactions. Vospernik et al. (2003b) have measured the displaced water by the application of an increasing transmembrane pressure. The importance was pointed out of a proper transmembrane pressure application when gas is fed from the support side. It must be underlined that the presence of defects in the top and intermediate layers will set a critical pressure that, if overcome, will result in the formation of gas bubbles. Therefore, the quality of the top-layer membrane is an important issue in the development of suitable catalytic membranes.

4.3.2

Partition equilibrium between the fluid phases and external mass transfer

In a multi-phase system one of the reactants needs to dissolve in the phase surrounding the solid catalyst. The solubility of that reactant can be described in terms of a partition constant. For example, considering a gas−liquid system the partition law is well-known as Henry’s law: H

c/ p

[4.3]

Gas solubility in liquid phase is usually very low and therefore it can be a limiting factor influencing the overall reaction rate. Most reactions of interest for the application of three-phase catalytic membrane reactors are hydrogenation reactions using hydrogen and oxidation reactions with air,

© Woodhead Publishing Limited, 2013

162

Handbook of membrane reactors 2nd layer 10–30␮m Top layer 1–10␮m

Support 1.5–2mm

1st layer 15–50␮m

Capillary pressure (bar)

1000

100

10

1

0.1 1

10

100 1000 Pore size (nm)

10 000

100 000

4.4 Capillary pressure against the pore size. The characteristic thickness and pore size of each intermediate layer of typical catalytic ceramic membrane for three-phase reactions is reported.

pure oxygen or ozone. Table 4.1 reports Henry constants along with solubility at 1 bar in water for some gases typically used in three-phase reactions (a referenced freely accessible list of Henry constants in water can be found in Sander, 1999). Table 4.2 lists the Henry constant for hydrogen in various organic solvents used for chemical syntheses. As can be seen, the dissolved concentration of reacting gases typically used in three-phase catalytic systems is very low and can be a limiting factor, especially for fast reactions. For this reason, three-phase reactions are usually carried out under pressure in order to increase the gas solubility. Similar considerations can be drawn for a two-liquid phase system. In that case, the partition equilibrium of the reacting species between the two phases needs to be considered. The presence of the catalytic membrane as an interface between the fluid phases can improve their contact and provide a reliable interfacial area which is very close to the catalytic sites where the reaction occurs with a direct effect on the driving force for the mass transfer. External mass transfer phenomena can limit the overall reaction rate. For the evaluation and control of external mass transfer rates the same

© Woodhead Publishing Limited, 2013

Multi-phase catalytic membrane reactors

163

Table 4.1 Henry constant and solubility of some gases in water at 298 K

O2 H2 O3 CO

H × 103 (mol/L/atm)

C (mg/L)

1.3 0.8 12 0.9

41.5 1.6 576 26.6

Table 4.2 Henry constant for H2 in various organic solvents at 298 K Solvent

H × 103 (mol/L/atm)

Methanol Ethanol 1-propanol 1-butanol Methyl acetate Ethyl acetate Cyclohexane Cyclohexanone

9.4 12.4 13.7 15.7 14.0 18.8 20.2 12.2

approach as that used for the membrane contactor can be adopted. Typical correlations can be found for gas and liquid on hollow fibre membranes in the review of Gabelman and Hwang (1999) and for other geometries in Basmadjian (2004). Mass transfer coefficients can be assessed by using a correlation of the type: Sh ∝ Re α Scβ f (

)

[4.4]

where Sh is the Sherwood number, which compares the mass transfer rate in the interface with the diffusion rate in the bulk fluid phase: Sh =

kc A D

[4.5]

Re is the Reynolds number: Re =

Aνρ µ

[4.6]

© Woodhead Publishing Limited, 2013

164

Handbook of membrane reactors

and Sc is Schmidt number, Sc =

μ ρD

[4.7]

Liquid diffusivity is smaller than gas diffusivity by a factor of two orders of magnitude. Considering the typical size (a few mm) of a channel (internal side of a tubular membrane) or of the cavity on the other membrane side (between the membrane and the housing wall) the typical density and viscosity of solvents used in multi-phase reactions, high liquid phase flow rates have to be used in order to be in a turbulent regime (a high Reynolds number) which minimizes the effect of the external mass transfer limitations.

4.3.3

Mass transfer and reaction in the catalytic membrane

In a catalytic membrane, the catalytic layer is usually well-defined and very thin (e.g., from 1 up to 30 µm) and its behaviour can be described in analogy with the catalytic slab reported in several chemical reaction engineering textbooks. Material balances on the thin catalytic layer of a membrane lead to the definition of a Thiele modulus (Cini et al., 1991a). Simple considerations on the Thiele modulus and the effectiveness factor in a catalytic membrane reactor have been given by Bottino et al. (2009) and Di Felice et al. (2010). The Thiele modulus, φ, is a dimensionless number composed of the square root of the characteristic reaction rate (e.g., for an n-order reaction), r: C sn

[4.8]

divided by the characteristic diffusion rate in the pores, rdiff: ⎛D C ⎞ rdiff = ⎜ efff2 s ⎟ ⎝ δ ⎠

[4.9]

Therefore for an n-order reaction rate the following general expression for the Thiele modulus is:

φ

⎛ kC n −1 ⎞ δ⎜ s ⎟ ⎜ Defff ⎟ ⎝ ⎠

1

2

© Woodhead Publishing Limited, 2013

[4.10]

Multi-phase catalytic membrane reactors

165

the effectiveness factor, η, is defined in a general way as:

ηint =

actual reaction rate rate predicted fro f m intrinsic kinetics

[4.11]

This can be easily obtained for a flat geometry on the basis of the Thiele modulus as (Missen et al., 1999):

ηint =

tanh(φ ) φ

[4.12]

The thickness of the catalytic layer in a membrane reactor can be very low (e.g., in porous catalytic membranes usually 1–10 µm) compared to the pellet size of a traditional reactor (from 100 µm to few mm) and, as a consequence, depending on the specific reaction rate, the Thiele modulus can be low enough to achieve an intrinsic effectiveness of about 1, which corresponds to full and efficient catalyst utilization in the reactive process. Moreover, the distributed reactant feeds on the two sides of the catalytic layer improves the mass transfer of the reactants from the surface of the catalytic layer to the catalytic sites in the catalytic layer internal structure. The mass transfer through the non-catalytic support should also be taken into account, especially when, as occurs in most of the studied catalytic membranes, the support is quite thick (e.g., 1–2 mm). Effective diffusivity through a porous membrane can be simply evaluated by correcting the binary diffusivity by the support porosity and tortuosity: Defff

D

ε τ

[4.13]

A study aimed at determining mass transfer rates through a liquid-impregnated ceramic membrane contactor operating as a gas−liquid contactor at transmembrane pressure differences lower than the bubble-point pressure has been presented by Vospernik et al. (2003b). In this study the authors have experimentally obtained the diffusivity of some model compounds (p-nitrobenzoic acid, phenol and oxalic acid) and they have proved that the estimation of the diffusivity can be done using well-known correlations such as the Wilke–Chang equation (Perry and Green, 1997) corrected by the membrane porosity and tortuosity. An example of estimation of mass transfer coefficient for a gas-filled pore membrane contactor can be found in Bottino et al. (2008).

© Woodhead Publishing Limited, 2013

166

4.3.4

Handbook of membrane reactors

Overall reaction rate

The overall effectiveness for a catalytic membrane reactor operating in the mode described in Fig. 4.3b can be defined as

ηoverall =

ηint ηint = HkV VM ρM 1 + ηint Da 1 + ηint kg AM

[4.14]

where the Damköhler number expresses the relative rates of the intra-phase reaction to external mass transport. When the reaction is fast compared to the mass transfer, Da is very high, ηintDa ≫ 1 and ηoverall → 1/Da. Then two approaches are available to increase the productivity of the membrane reactor: to improve the mass transfer by acting on the fluid dynamic of the catalytic membrane module and/or to realize very compact modules with high interfacial surface area. The fluid dynamic regime is equally important, if we consider that in several experimental conditions reported in literature the laminar regime is prevalent in both the reacting phases. When the external mass transfer rate is fast compared to the reaction rate, then ηoverall → ηint. In addition, if the catalytic layer is very thin, then the reactor is not influenced by any mass transfer. These behaviours have been observed and reported in literature. Vospernik et al. (2004) showed the effect of the external mass transfer resistance by using a low and a higher liquid phase recirculation rate (Re in the range 660–4860) in the nitrite hydrogenation, while no effect was observed in the formic acid oxidation. Bottino et al. (2004) compared the behaviour of a catalytic porous membrane reactor and of a slurry reactor for the competitive hydrogenation−isomerization of methylenecyclohexane, and while the slurry reactor showed at higher temperature a lower activation energy, due to mass transfer limitations, the apparent activation energy for the catalytic membrane reactor was constant in the same temperature range.

4.3.5

Mass transport and reaction in dense membranes

When the catalytic membrane is dense (Figs 4.3d and f), then the two fluid phases are separated by a continuous barrier layer and therefore do not have a direct contact as can occur, for example, in the pores of a catalytic porous membrane. Then the two reacting species will need to dissolve and diffuse in the dense material of the membrane in order to be able to react

© Woodhead Publishing Limited, 2013

Multi-phase catalytic membrane reactors

167

together. The mechanism of dissolution and diffusion of species in a dense material is usually dependent on the material nature and on the specific permeating molecule. The solution−diffusion model approach is usually used for the description of mass transfer in dense membranes. Both reacting species can exhibit a permeation through the dense membrane or only one of them, as for example occurs in dense metal membranes (e.g., H2-palladium alloys, O2- silver). The equilibrium law which regulates the dissolution of the permeating species from a liquid phase inside the dense membrane material can be described as follows: c

S ⋅C

[4.15]

where c is the equilibrium solute concentration in the membrane material, C is the concentration in the liquid phase, and S is the solubility coefficient. In the case of a chemical compound in equilibrium between a gas and a solid phase, the previous equation can be expressed in terms of Henry’s law as: c

H H′ ⋅ p

[4.16]

where p is the partial pressure of the compound in the gas phase. Previous equations are usually valid for low solute concentrations. An introduction to mass transport in dense polymer membranes can be found in Paul (2010). In the specific case of a hydrogen−palladium system, the absorption mechanism involves the surface dissociation of hydrogen, and the concentration of hydrogen atoms in palladium can be related to hydrogen partial pressure by the Sievert’s equation: c

S ⋅ p1/ 2

[4.17]

Using a Fickian approach, the mass transfer rate through a dense metal membrane can be described by an equation of the type: J=

mDS m ( p1 (p δ

p2m )

[4.18]

where J represents the molar flux through the membrane, D is diffusivity of the permeating species in the material, δ is the thickness of the dense material, p1 and p2 are the partial pressures of the permeating species (e.g., H2) of both the surfaces of the membrane. The parameter m is related to the eventual dissociation mechanisms (e.g., m = 0.5, for dense thick palladium

© Woodhead Publishing Limited, 2013

168

Handbook of membrane reactors

membranes working in the Sievert regime), to the influence of other phenomena on the membrane surface or to the presence of defects in the membrane layer. When the membrane is permeable to only one of the reactants (e.g., hydrogen), the reaction will occur directly at the membrane surface on the side where the non-permeating reactant is present. When both the reactants are soluble and can diffuse in the dense catalytic membrane, two cases can be considered: 1. the catalyst is uniformly and continuously distributed in the catalytic layer. The approach will be similar to the one based on the Thiele modulus by properly arranging the material balances and replacing effective diffusivity in the pores with diffusivity in the dense membrane; 2. the catalyst is dispersed in the form of discrete particles: an additional diffusion path from the dense membrane surface to the catalytic particles, sorption phenomena and particle distribution in the dense membrane, need to be taken into account.

4.3.6

Modelling of multi-phase catalytic membrane reactors

Several authors have reported modelling of multi-phase membrane reactors and, in particular, of three-phase catalytic membrane reactors. Harold and Watson (1993) have considered the situation of a porous catalytic slab partially wetted by a liquid from one side and by a gas phase on the other side, and they have pointed out the complexity of the problem in presence of an exothermic reaction, capillary condensation and vaporization. Torres et al. (1994) studied nitrobenzene hydrogenation on an asymmetric catalytic porous tubular membrane where a thin catalytic layer (ca. 3 µm) is supported by a thick (ca 1.1 mm) macroporous inert tube and they developed a model using experimental kinetics. They showed that by using an inorganic membrane the most beneficial configuration to the overall reaction kinetics is when the gas phase flows directly on the catalytic layer while the liquid phase flows along the inert support. It should be noted that the support was impregnated by the liquid phase. Akyurtlu et al. (1988) carried out a theoretical investigation on multi-phase porous tubular catalytic membrane with the liquid phase on the external side and the gas phase in the membrane tube lumen. They showed the importance of the Thiele modulus on the reactor performance and that thin-walled catalyst tubes have larger effectiveness factors.

© Woodhead Publishing Limited, 2013

Multi-phase catalytic membrane reactors

169

Catalytic particle

Aqueous phase

Organic phase

Catalytic membrane thickness

4.5 Outline of a non-catalytic dense permeable matrix embedding discrete catalyst particles with a well-defined geometry (adapted from Yawalkar et al., 2001).

Vospernik et al. (2006) calculated the profile concentration along the porous catalytic layer by assuming the gas−liquid interface at the border between an intermediate porous catalytic layer and the coarse support (at about 50 µm from the top-layer surface). In their model they considered the complete depletion of oxygen in the reaction zone during the catalytic oxidation of formic acid. Their model did not take into account the catalyst distribution along the cross-section. They found the limiting influence of oxygen diffusion on the overall reaction rate. Yawalkar et al. (2001) has developed a model for a three-phase reactor based on the use of a dense polymeric composite membrane containing discrete cubic zeolite particles (Fig. 4.5) for the epoxidation reaction of alkene. Catalytic particles of the same size are assumed with a cubic shape and uniformly dispersed across the polymer membrane cross-section. Effects of various parameters, such as peroxide and alkene concentration in liquid phase, sorption coefficient of the membrane for peroxide and alkene, membrane–catalyst distribution coefficient for peroxide and alkene and catalyst loading, have been studied. The results have been discussed in terms of a peroxide efficiency defined as the ratio of flux of peroxide through the membrane utilized for alkene oxidation to the total flux of organic peroxide through the membrane. The paper aimed to show that, by using an organophilic dense membrane and the catalysts confined in the polymeric matrix, the oxidant concentration (in that reaction peroxides) can be controlled on the active site with an improvement of the peroxide efficiency and selectivity to desired products. Trusek-Holownia and Noworyta (2005) discussed a model for a two-liquid phase catalytic membrane enzymatic reactors. The organic phase contains

© Woodhead Publishing Limited, 2013

170

Handbook of membrane reactors

the reactant which through the membrane reaches the catalytic zone formed by an enzymatic gel layer immobilized on the membrane surface, while the water phase works as a hydrophilic product receiver. In particular, the authors have studied the influence of enzymatic gel layer parameters on the value of the substrate flux entering this layer and on the conversion degree, and they have found by Thiele module assessment the maximum limit value for the gel thickness for an efficient use of the enzyme activity. In other papers the authors, besides showing the preparation method of the catalytic membrane, have obtained experimentally kinetics parameter and an estimation of the catalytic gel layer thickness. Recently, Endre (2011) attempted a comprehensive approach to the modelling of catalytic membranes for multi-phase membrane reactors, showing the mutual effects of mass transfer and some typical kinetics laws.

4.3.7

Operations with multi-phase catalytic membrane reactors

In catalytic multi-phase membrane reactors the catalytic membrane usually plays the role of interface between the two fluid phases. As discussed in the previous subsections, especially in the case of porous catalytic membranes, a strict control of the position of the inter-phase interface in proximity to the catalytic layer is of paramount importance in order to minimize the diffusion resistances. Figure 4.6 reports a simplified outline of a three-phase experimental rig as can be found in several publications. A recycle loop for the liquid phase has been considered in order to achieve the desired conversion. Differential pressure transducer Compressor

Flow control valves

Gas A unreacted

Gas feed A

Liquid B unreacted products Liquid feed B

Pump

Recycle liquid tank

4.6 Typical co-current three-phase catalytic membrane reactor rig.

© Woodhead Publishing Limited, 2013

Multi-phase catalytic membrane reactors

171

The single step operation without recycle option would require the use of highly active catalytic membranes, catalytic membranes with a higher catalyst loading, or a larger catalytic membrane surface. In order to reduce the catalytic reactor size, membrane modules characterized by a high active surface to reactor volume ratio are desired. The position of the reaction front in the reaction zone can be realized by a proper choice of the feed configuration, the characteristics of the membrane materials, and proper regulation of the transmembrane pressure between the two compartments segregated by the membrane.

4.4

Materials and catalytic membranes for membrane reactors

A catalytic membrane can be realized by using polymeric membranes, inorganic membranes or composite organic−inorganic membranes. Each category of materials has some advantages and disadvantages which should be considered in catalytic membrane reactor design. Polymeric membranes are available in a very wide variety of materials which can be arranged in very compact module configurations with very high packing density factors (membrane surface area per unit volume of the module). The most compact configuration corresponds to the hollow fibre module. On the other hand, by using polymeric materials maximum operating temperatures range from 100°C to 250°C and also the polymer−solvent stability needs to be properly considered. Inorganic materials can work in a wider range of operating conditions, both in terms of pressure and temperature. Figure 4.7 shows some catalytic membranes and supports used in our group for studying multiphase membrane systems. Figure 4.7a shows an unsupported flat sheet poly(vinylidene fluoride) (PVDF) membrane loaded with Nickel Raney particles. Figure 4.7b shows a flat sheet PVDF membrane loaded with a Pt/ carbon catalyst supported by a polyester non-woven. Both the membranes have been designed to carry out in laboratory hydrogenation reactions in three-phase membrane reactors. The presence of a support improves the mechanical resistance of the membranes, which can therefore be much thinner than unsupported membranes. Polymeric flat sheet membranes can be arranged in plate and frame modules or in spiral wound modules. Tubular and hollow fibre (Fig. 4.7c and 4.7d) geometries are already used successfully in membrane contactor operations. A higher surface per membrane volume can be achieved using multi-channel configurations (Fig. 4.7e and 4.7f). Moreover, the membrane monoliths can be easily arranged in modules with higher packing density (available membrane surface per volume of the module). Inorganic membranes are available on the market in a variety of materials, including ceramics, metal, carbon, zeolites. The cost of inorganic membranes

© Woodhead Publishing Limited, 2013

172

Handbook of membrane reactors

4.7 Some types of membrane and catalytic membranes for three-phase reactions prepared in the Genoa membrane research group. Porous flat sheet catalytic membrane: (a) unsupported; (b) supported on a polymeric non-woven substrate. (c) Hollow fibres; (d) single-channel Pd membrane supported on alumina porous substrate; (e) three-channel catalytic zeolite membrane; (f) multi-channel porous alumina substrate.

is much higher than polymeric membranes. Since often operating conditions of several multi-phase reacting systems are quite mild, polymeric, inorganic or composite organic−inorganic materials can be practicably used. The choice of a suitable material for a membrane reactor is related to its stability in the chemical environment during the reaction and on the operating time, and to the possibility to load useful amounts of a stable catalyst. For example, in order to increase the catalyst loading and metal dispersion in a polymeric porous polyvinylidene fluoride membrane, Bottino et al. (2002) added polyvinylpyrrolidone to the membrane composition with the purpose of enhancing the hydrophilic character of the membrane and favouring a better interaction between catalyst precursors (present in an aqueous phase) and the pore surface. The success of a catalytic membrane is strongly related to a proper selection and development of both the catalyst and the membrane. Hall et al. (2002) prepared catalytic membranes by chemical modification with a silane of silica-impregnated polyethylene and poly(styrenedivinylbenzene) copolymer functionalized with quaternary ammonium groups in order to insert ion-exchangeable onium sites to catalyse the partial oxidation of benzyl alcohol (in the organic liquid phase) with hypochlorite (aqueous liquid phase). In this case the membrane materials needs to be chemical resistant to the oxidant environment in the aqueous phase. In a dense polymeric catalytic membrane the catalyst can be a thin layer on the membrane surface or distributed in the thickness of the polymeric matrix. An exhaustive review of methods for the preparation of catalytic polymeric membranes has been reported by Ozdemir et al. (2006). Vankelecom (2002) thoroughly reviewed the application of polymeric membranes in catalytic reactors.

© Woodhead Publishing Limited, 2013

Multi-phase catalytic membrane reactors

173

Nowadays, a wider variety of inorganic membranes are commercialized. A rough classification can be based on the type of the inorganic materials (e.g., carbon, metal, ceramic, glass, zeolite) and on the structure (e.g., porous or dense). The final catalytic membrane can be a composite of different inorganic or organic−inorganic materials. Microporous selective carbon layers can be obtained by pyrolysis of a polymer layer. Carbon molecular sieve membranes can be of interest for their application in multi-phase systems. Tennison et al. (2007) have reported the preparation of carbon layers supported on ceramic macroporous supports to be tested in the three-phase hydrogen peroxide synthesis reaction. These carbon coated ceramic membranes have been used for the deposition of dispersed metal catalytic particles (Abate et al., 2006b; Melada et al., 2005) or palladium-based dense layers (Abate et al., 2005) in order to study the hydrogen peroxide formation reaction. Another example of carbon-based materials, which can be included in the category of catalytic membranes for multi-phase reactions, is given by the catalytic gas diffusion electrodes for proton-exchange fuel cells. Catalytic gas diffusion electrodes for fuel cell applications are composed of a catalytic carbon microporous layer (e.g., platinum on carbon black) on a flexible carbon cloth or carbon non-woven. Of course they can be considered as catalytic membranes when the surface is homogeneous and without any crack or defect. Ong et al. (2008) developed gas diffusion layers for fuel cells with improved surface morphology (e.g., homogeneity, controlled porosity and low defectiveness) with respect to traditional ones. Porous metal membranes are commercially available in stainless steel and some other alloys (e.g., Inconel, Hastelloy) and they are characterized by a macroporous structure. On the other hand, porous ceramic membranes can be found commercially in various oxides and combination of oxides (e.g., Al2O3, TiO2, ZrO2, SiO2) and pore size families in the mesopore and macropore ranges (e.g., from 1 nm to several microns). Most of the literature studies on three-phase catalytic membrane reactors have been carried out by developing catalytic ceramic membranes. The deposition techniques for the preparation of catalytic ceramic membranes involve methods widely used for the preparation of traditional supported catalysts (Pinna, 1998), and methods specifically developed for the preparation of structured catalysts (Cybulski and Moulijn, 2006). Other methods to introduce a catalytic species on a porous support include the chemical vapour deposition and physical vapour deposition (Daub et al., 2001). The catalyst deposition method has a strong influence on the catalytic membrane reactor performance. Dense metal membranes exhibit an absolute permeability to specific species. Clear examples are given by palladium and palladium-alloy membranes, which are exclusively permeable to hydrogen, and by silver membranes,

© Woodhead Publishing Limited, 2013

174

Handbook of membrane reactors

which are permeable to oxygen. The use of palladium dense membrane for liquid phase hydrogenation was pioneered by Gryaznov and co-workers. An extensive description of metal dense membranes as structured catalytic reactors can be found in Gryaznov et al. (2006). The dense layer plays three main roles: (i) It is the interface between the two segregated fluid phases, (ii) it is selective to one of the reacting species or products, and (iii) it supplies the catalytic surface for the chemical reaction. Therefore the dense layer should be continuous and without defects. Palladium alloys have been studied, since pure palladium shows mechanical instability due to embrittlement caused by a transition from α-phase to β-phase in a hydrogen atmosphere. Metal dense membranes can be structured as metal foils or by depositing on porous metal supports very thin dense metal layers in order to obtain a high permeating flux and to save the amount of precious metal. Usually thin dense catalytic metal membranes are supported on porous ceramic or metal supports. Zeolite membranes indicate inorganic membranes with a selective/catalytic layer composed of a zeolite which is crystalline aluminosilicate with the feature of a high ordered porous structure with size comparable to molecular dimension. An example of the use of zeolites as a catalyst in a multi-phase membrane reactor can be found in Shukla and Kumar (2004) who have immobilized a lipase on a zeolite-clay composite membrane by using glutaraldehyde as a bifunctional ligand in order to carry out the hydrolysis of olive oil. An application of a zeolite-based membrane in a three-phase membrane reactor has been reported by Wu et al. (1998), where TS-1 zeolite crystallites were embedded in a polydimethylsiloxane (PDMS) membrane in order to catalyse the oxyfunctionalization of n-hexane (from a gas phase) with hydrogen peroxide (from a liquid phase). Although several studies on the preparation of catalytic membranes for multi-phase membrane reactors have been published, the choice of the proper membrane material to be combined with a suitable catalyst for the preparation of an effective catalytic membrane reactor is still not easy. For example, the selection of a porous material with a suitable wettability or non-wettabilty to one of the fluid phases can play a key role in determining the effectiveness of the catalytic membrane in relation with each particular configuration contact modes. Several investigations have reported that with a ceramic porous asymmetric catalytic membrane, the best feeding configuration is obtained by feeding the aqueous phase from the support side and the gas from the catalytic layer side, since the ceramic membrane is easily impregnated by the liquid and then the diffusion path of the gas phase can be shorter. On the other hand when the liquid is fed on the support side, the application of an over-pressure is necessary to move the phase interface near the catalytic active zone. Recently, Aran et al. (2011) have

© Woodhead Publishing Limited, 2013

Multi-phase catalytic membrane reactors

175

prepared very interesting ceramic catalytic membranes where the support has been hydrophobized in order to apply the gas phase (hydrogen) from the support side and the liquid phase from the catalytic layer side for nitrite conversion.

4.5

Typical reactions with three-phase membrane reactors

The three-phase membrane reactors have been mainly investigated for applications in both hydrogenation and partial oxidation reactions. Table 4.3 lists some typical gas−liquid hydrogenation reactions investigated in order to explore the features of three-phase catalytic membrane reactors. An example of the application of three-phase catalytic membrane reactors to the hydrogenation of sunflower seed oil can be found in Veldsink (2001), where it was shown that for this hydrogenation running under kinetically controlled conditions the interfacial transport resistances and intraparticle diffusion limitations did not have any effect. Unfortunately the catalyst underwent a serious deactivation process. Singh et al. (2009) have investigated the partial hydrogenation of soybean oil by using a metal/polymer composite asymmetric membrane with platinum catalyst deposited by magnetron sputtering on the ‘skin’ side of an integral asymmetric polymeric membrane. Hydrogen was supplied to the porous substructure side of the membrane while the oil flowed over the platinum-sputtered feed (skin) side of the membrane. The main consideration is about the thickness of the skinned layer, which was about 0.2 microns. Palladium acted therefore only as a source of active hydrogen. Cini et al. (1991b) proposed the use of a tubular Pd/Al2O3 mesoporous membrane for the hydrogenation of α-methylstyrene to cumene. A comparison between the tubular catalyst and a fully-wetted pellet revealed a rate increase by up to a factor of 20. From that study, several other theoretical (Torres et al., 1994) and experimental ones confirmed that a three-phase membrane reactor can improve the mass transfer rate of gas−liquid−solid systems. In particular, Bottino et al. (2004) explored the performance of different catalytic membranes in the hydrogenation–isomerization of methylenecyclohexane, in a temperature range between 288 and 343 K.The performance of the three-phase catalytic membrane reactor has been compared with that of a slurry reactor, resulting in a wider operating temperature range without mass transfer limitations. Typical oxidation reactions investigated by using a multi-phase catalytic membrane reactor are reported in Table 4.4 (gas−liquid systems) and in Table 4.5 (liquid−liquid systems).

© Woodhead Publishing Limited, 2013

Table 4.3 Examples of gas–liquid hydrogenation carried out with catalytic membrane reactors

© Woodhead Publishing Limited, 2013

Reaction

Catalytic membrane

Geometry

Contact mode

References

Hydrogenation of sunflower seed oil

Porous ceramic asymmetric membranes; Pd dispersed in the top layer

Liquid phase in the inner side, gas phase in the shell side

Veldsink (2001)

Hydrogenation of sunflower oil

Porous catalytic polymeric membranes; dense catalytic pervaporative membranes Pt and Pd dispersed particles

Tubular, single channel, catalyst in the inners side Flat

Bengtson et al. (2002)

Partial hydrogenation of soybean oil

Integral asymmetric polymeric membrane; Pt by sputtering on the skin layer Asymmetric ceramic membrane; Co–Pt dispersed in the top layer

Flat

Asymmetric ceramic membrane; Polymeric porous membrane; Pd, Pt or Ru dispersed in the top layer

Tubular single channel (ceramic) Flat (polymeric)

Flow-through mode (porous membranes or pervaporation mode (dense membranes)) Liquid on the catalyst layer, gas on the support side Gas on the support side and liquid on the catalyst side and vice versa Gas on the support side liquid on the catalyst side and vice versa

Hydrogenation of cinnamaldehyde

Hydrogenation– isomerization of methylenecyclohexane

Tubular single channel

Singh et al. (2009) Pan et al. (2000)

Bottino et al. (2002, 2004)

© Woodhead Publishing Limited, 2013

Nitrobenzene liquid-phase hydrogenation Various selective hydrogenations of organic compounds Hydrogenation of nitrate

Porous ceramic asymmetric membranes Pt dispersed in the top layer Porous ceramic asymmetric membranes Pt, Pd dispersed in the top layer Porous ceramic asymmetric membranes Pd–Cu, Pd–Sn dispersed in the top and intermediate layers

Tubular single channel

Hydrogenation of nitrate

Porous ceramic asymmetric membranes Pd/Cu dispersed in the top layer Porous ceramic asymmetric membranes; hydrophobized support Pd dispersed in the top and intermediate layers

Tubular single channel

Flow-through mode

Tubular single channel

Liquid on the catalyst layer side, gas on the support side

Hydrogenation of nitrite

Tubular single channel Tubular single channel

Gas on the support side liquid on the catalyst side and vice versa Gas on the support side liquid on the catalyst side and vice versa Liquid on the catalyst layer, gas on the support side

Peureux et al. (1995), Torres et al. (1994) Tilgner et al. (1998) Daub et al. (1999, 2001), Reif and Dittmeyer (2003), Vospernik et al. (2003b) Wehbe et al. (2010) Aran et al. (2011)

Table 4.4 Examples of gas–liquid oxidation reactions investigated by using catalytic membrane reactors

© Woodhead Publishing Limited, 2013

Reaction

Catalytic membrane

Geometry

Contact mode

Partial oxidation of organic compounds (formic acid) Formic acid wet-air oxidation

Asymmetric porous ceramic membranes; Pt dispersed in the top layer Asymmetric porous ceramic membranes; Pt dispersed in the top layer Asymmetric porous ceramic membranes; Pt dispersed in the top layer Asymmetric ceramic support; Pd, Pd–Ag dense layers; hydrophobic gas permeable polymer Porous asymmetric ceramic membrane with carbon coating; Pd, Pd–Ag, Pd–Pt dispersed; Pd and Pd/Ag dense layer Superacid catalytic membranes, Fenton catalyst, symmetric

Tubular single channel and multi-channel

Liquid phase on the catalytic Iojoiu et al. (2006, 2007) side, gas phase on the support side

Tubular single channel and multi-channel

Liquid inside, gas outside

Tubular single channel and multi-channel

Gas phase on the catalytic side, liquid phase on the support side

Tubular single channel

O2 saturated liquid on the hydrophobized Pd dense layer; H2 gas on the support side

Tubular single channel

O2 saturated liquid phase on Abate et al. (2005, 2006, the catalytic side, H2 gas 2006b), Melada et al. on the support side (2005, 2006)

Flat

The oxidant is in the liquid phase, alkanes are in the gas phase

Formic acid wet-air oxidation

Direct oxidation of hydrogen to hydrogen peroxide Synthesis of H2O2 from H2 and O2

Synthesis of liquid oxygenates from light alkanes (C1–C3)

References

Raeder et al. (2003), Iojoiu et al. (2005, 2005b), Vospernik et al. (2004, 2006) Miachon et al. (2003)

Choudhary et al. (2001)

Frusteri et al. (1999), Espro et al. (2001, 2003, 2006)

Multi-phase catalytic membrane reactors

179

Table 4.5 Examples of liquid–liquid reactions studied by using catalytic membrane reactors Reaction Selective oxidation of benzyl alcohol to benzaldehyde

Catalytic membrane

Geometry

Porous composite Flat inorganic–organic; quaternary ammonium catalyst The membrane is wettable by both liquid phases Oxyfunctionalization Zeolite particles Flat of n-hexane to in polymeric a mixture of membrane, hexanols and symmetric hexanones using H2O2 Hydrolysis of olive Lipase immobilized Flat oil on a zeolite membrane

Contact mode References –

Hall et al. (2002)

–

Wu et al. (1998)

Oil phase on Shukla the catalyst et al. layer, (2004) aqueous phase on the support side

An interesting class of polymer based three-phase catalytic reactors is based on the combination of catalytic layers and solid electrolyte polymer membranes as used in proton-exchange fuel cells. In particular, the direct methanol fuel cell is a special type of three-phase catalytic membrane reactor where the methanol oxidation reaction is carried out with modalities which enable the device to convert the chemical energy into the electrical energy. It is not in the scope of the present chapter to discuss protonexchange fuel cells; in any case, the use of proton-conducting membranes in three-phase system is an emerging field which needs further investigation, due to the possibility of exploiting hydrogen electrochemical pumping. Figure 4.8 shows a hydrogen pumping catalytic membrane reactor (An et al., 1998; Otsuka and Yamanaka, 1990) for hydrogenation reactions. A protonconducting dense membrane is used to transfer hydrogen in a very active form from the gas phase to the reaction side by the application of a voltage to two catalytic electrodes on the two sides of the dense membrane. Selective oxidation of n-paraffins has been carried out on catalytic ionomer membranes with Fe2+/H2O2 Fenton. The three-phase membrane reactors were constituted by a proton-conducting membrane (based on Nafion), fed on one side with a gaseous n-paraffins and, on the other side, with an aqueous solution of hydrogen peroxide and Fe2+ ions. The paraffin is activated by

© Woodhead Publishing Limited, 2013

180

Handbook of membrane reactors Substrate

Hydrogenated product

Cathode catalytic layer Cathode diffusion layer

e–

Proton conducting membrane Anode catalytic layer Anode diffusion layer

e–

H-donor compound

Dehydrogenated product

4.8 Representation of a hydrogen pumping three-phase catalytic membrane reactor.

the superacid Nafion-based membrane and then partially oxidized by the OH radicals generated during the Fenton process to products which were soluble in the liquid phase (e.g., Espro et al., 2006). Choudhary et al. (2001) reported the use of a thin Pd–Ag alloy membrane supported on a tubular membrane coated by an oxidized layer of Pd and by a hydrophobic polymeric layer for the non-hazardous and highly selective H2O2 production. Hydrogen is supplied from the external side of the porous support and the H atoms permeate through the membrane and react with molecular O2 in a liquid medium. The authors highlighted that the segregated reactant configuration had the advantage of improving the safety of the reactive process beside a better selectivity to desired products. Table 4.5 lists some references on the application of catalytic membrane reactors to liquid–liquid multi-phase systems. The number of studies of application of catalytic membrane reactors in liquid−liquid systems is still very limited and most of them consider the use of biocatalytic membranes. In Europe, the first project on the investigation of a catalytic membrane reactors in three-phase systems for fine chemistry applications dates back to 1991 (project reference: BREU0406, 2nd EU Framework Programme). In this project particular attention was devoted to membrane development. One example of realization of scale-up of a three-phase membrane reactors has been reported by Iojoiu et al. (2006) where laboratory catalytic membrane reactors (single- and multi-channel types) and demonstration pilot units using real wastewater have been compared. The work has been carried out in the 5th Framework Programme ‘Waste water treatment by catalytic oxidation contactor – Watercatox (EVK1-CT- 2000–00073)’, which was aimed at developing a wet-air oxidation process based on porous catalytic membrane contactors. Their technological efficiency was demonstrated by the results obtained using the pilot test unit on different industrial effluents from several origins. An important effort to investigate the engineering

© Woodhead Publishing Limited, 2013

Multi-phase catalytic membrane reactors

181

aspects of three-phase reactors has been carried out. During the project an industrial test unit with membrane area of 0.35 m2 has been designed and constructed. Caro (2010) reports the realization of a pilot-scale membrane contactor reactor for selective hydrogenation at Bayer technology Services GmbH. Therefore, due to the results being proved not only in academic laboratory experiments, but also by using small demonstration units, industry will probably become more and more interested.

4.6

Conclusion and future trends

Although catalytic membrane reactors have not yet been applied on an industrial scale, the attention of the scientific community on them remains unchanged. Concerning multi-phase catalytic membrane reactors, the research and development has already reached a certain grade of maturity with the involvement of some companies in proof-of-concept studies and realization of small demonstration units. In any case, a lot of work with clearer focus is still necessary and the technical challenges to be overcome are still numerous. Although an extensive literature on the preparation of catalytic membranes can be easily found, a further effort would be crucial for improving the catalyst−membrane pair (e.g., effect of the catalyst distribution) for each specific application. In particular, since the concept of the catalytic membrane contactor is based on a thin catalytic layer, the preparation procedures should be optimized in order to offer the highest catalytic surface (e.g., metal dispersion) at the high catalyst loadings. This becomes very important for the membrane supports with low−medium packing density as well as most of the inorganic membranes, although in recent years examples of hollow fibre ceramic membranes have emerged even at a commercial stage. Catalyst deposition on polymeric membranes needs additional effort in order to achieve stable anchoring of the active catalytic species. The number of studies devoted to the application of modules with higher membrane area per unit volume in multi-phase reactions remains still very limited, for both catalytic polymeric and inorganic membranes. To this day many commercial types of polymeric and inorganic membranes are available, with a wide variety of materials. Polymeric membranes seem to be manufactured with a much higher reproducibility than inorganic membranes. Inorganic membranes, especially on the emerging products, still show the presence of many defects and irregularities which make more difficult the assessment of the reactor performance in the application. Since the reactor acts as an interface between two phases, the thin catalytic layer should be without any defect which might shift

© Woodhead Publishing Limited, 2013

182

Handbook of membrane reactors

the reactant meeting interface to a non-reacting zone. The cost per unit surface area of inorganic membranes is typically 100 times more than the polymeric ones, but they, especially when the catalyst is well-anchored and chemically stable, should offer a much longer life span and regeneration possibilities. The opportunities and advantages offered by the application of catalytic membrane reactors to multi-phase systems need to be further explored. We hope that this introductory chapter on multi-phase catalytic membrane reactors will stimulate the reader to delve into this multidisciplinary topic.

4.7

References

Abate, S., Centi, G., Melada, S., Perathoner, S., Pinna, F., Strukul, G., 2005. Preparation, performances and reaction mechanism for the synthesis of H2O2 from H2 and O2 based on palladium membranes. Catalysis Today 104, 323–328. Abate, S., Melada, S., Centi, G., Perathoner, S., Pinna, F., Strukul, G., 2006. Performances of Pd-Me (Me = Ag, Pt) catalysts in the direct synthesis of H2O2 on catalytic membranes. Catalysis Today 117, 193–198. Abate, S., Perathoner, S., Genovese, C., Centi, G., 2006b. Performances, characteristics and stability of catalytic membranes based on a thin Pd film on a ceramic support for H2O2 direct synthesis. Desalination 200, 760–761. Akyurtlu, J.F.,Akyurtlu,A., Hamrin, C.E., 1988.A study on the performance of the catalytic porous-wall three-phase reactor. Chemical Engineering Communications 66, 169–187. An, W., Hong, J.K., Pintauro, P.N., 1998. Current efficiency for soybean oil hydrogenation in a solid polymer electrolyte reactor. Journal of Applied Electrochemistry 28, 947–954. Aran, H.C., Chinthaginjala, J.K., Groote, R., Roelofs, T., Lefferts, T., Wessling, M., Lammertink, R.G.H., 2011. Porous ceramic mesoreactors: a new approach for gas–liquid contacting in multiphase microreaction technology. Chemical Engineering Journal 169, 239–246. Barnett, S.M., 2006. Multiphase Reactors, in: Encyclopedia of Chemical Processing. Taylor & Francis, New York. Basmadjian, D., 2004. Mass transfer: principles and applications. CRC, Boca Raton. Bengtson, G., Scheel, H., Theis, J., Fritsch, D., 2002. Catalytic membrane reactor to simultaneously concentrate and react organics. Chemical Engineering Journal 85, 303–311. Biardi, G., Baldi, G., 1999. Three-phase catalytic reactors. Catalysis Today 52, 223–234. Bottino, A., Capannelli, G., Comite, A., Del Borghi, A., Di Felice, R., 2004. Catalytic ceramic membrane in a three-phase reactor for the competitive hydrogenation–isomerisation of methylenecyclohexane. Separation and Purification Technology 34, 239–245. Bottino, A., Capannelli, G., Comite, A., Di Felice, R., 2002. Polymeric and ceramic membranes in three-phase catalytic membrane reactors for the hydrogenation of methylenecyclohexane. Desalination 144, 411–416.

© Woodhead Publishing Limited, 2013

Multi-phase catalytic membrane reactors

183

Bottino, A., Capannelli, G., Comite, A., Di Felice, R., 2009. Three-phase membrane reactors and aspects of membrane contactors, in: Simulation of membrane reactors, (A. Basile and F. Gallucci, Eds). Nova Science Publishers, New York, pp. 311–340. Bottino, A., Capannelli, G., Comite, A., Di Felice, R., Firpo, R., 2008. CO2 removal from a gas stream by membrane contactor. Separation and Purification Technology 59, 85–90. Caro, J., 2010. Basic aspects in membrane reactors, in: Comprehensive membrane science and engineering. Elsevier, Amsterdam. Centi, G., Perathoner, S., 2003. Novel catalyst design for multiphase reactions. Catalysis Today 79–80, 3–13. Choudhary, V.R., Gaikwad, A.G., Sansare, S.D., 2001. Nonhazardous direct oxidation of hydrogen to hydrogen peroxide using a novel membrane catalyst. Angewandte Chemie International Edition 40, 1776–1779. Cini, P., Blaha, S.R., Harold, M.P., 1991a. Preparation and characterization of modified tubular ceramic membranes for use as catalyst supports. Journal of Membrane Science 55, 199–225. Cini, P., Harold, M.P., 1991b. Experimental study of the tubular multiphase catalyst. AIChE Journal 37, 997–1008. Cybulski, A., Moulijn, J.A., 2006. Structured catalysts and reactors. Taylor & Francis, Boca Raton. Daub, K., Emig, G., Chollier, M.-J., Callant, M., Dittmeyer, R., 1999. Studies on the use of catalytic membranes for reduction of nitrate in drinking water. Chemical Engineering Science 54, 1577–1582. Daub, K., Wunder, V.K., Dittmeyer, R., 2001. CVD preparation of catalytic membranes for reduction of nitrates in water. Catalysis Today 67, 257–272. Di Felice, R., Capannelli, G., Comite, A., 2010. Multiphase membrane reactors, in: Comprehensive Membrane Science and Engineering. Elsevier, Amsterdam. Dittmeyer, R., Svajda, K., Reif, M., 2004. A review of catalytic membrane layers for gas/liquid reactions. Topics in Catalysis 29, 1–2. Endre, N., 2011. Mass transfer through catalytic membrane reactor, in: Mass transfer in multiphase systems and its applications. InTech. Espro, C., Arena, F., Frusteri, F., Parmaliana, A., 2001. On the potential of the multifunctional three phase catalytic membrane reactor in the selective oxidation of light alkanes by Fe2+–H2O2 Fenton system. Catalysis Today 67, 247–256. Espro, C., Frusteri, F., Arena, F., Parmaliana, A., 2003. Innovative membrane-based catalytic process for environmentally friendly synthesis of oxygenates. Topics in Catalysis 22, 65–70. Espro, C., Arena, F., Frusteri, F., Tasselli, F., Regina, A., Drioli, E., Parmaliana, A., 2006. Selective oxidation of propane on Nafion/PEEK-WC catalytic membranes in a multifunctional reaction system. Catalysis Today 118, 253–258. Frusteri, F., Arena, F., Bellitto, S., Parmaliana, A., 1999. Partial oxidation of light paraffins on supported superacid catalytic membranes. Applied Catalysis A: General 180, 325–333. Gabelman, A., Hwang, S.-T., 1999. Hollow fiber membrane contactors. Journal of Membrane Science 159, 61–106. Gryaznov, V.M., Ermilova, M.M., Orekhova, N.V., Tereschenko, G.F., 2006. Reactors with metal and metal-containing membranes, in: Structured catalysts and reactors. Taylor & Francis, Boca Raton.

© Woodhead Publishing Limited, 2013

184

Handbook of membrane reactors

Hall, D.W., Grigoropoulou, G., Clark, J.H., Scott, K., Jachuck, R.J.J., 2002. The selective oxidation of benzyl alcohol using a novel catalytic membrane reactor. Green Chemistry 4, 459–460. Harold, M.P., Watson, P. C., 1993. Bimolecular exothermic reaction with vaporization in the half-wetted slab catalyst. Chemical Engineering Science 48, 981–1004. Henkel, K.-D., 2000. Reactor types and their industrial applications. Ullmann’s Encyclopedia of Industrial Chemistry. Iojoiu, E.E., Landrivon, E., Raeder, H., Torp, E.G., Miachon, S., Dalmon, J.A., 2006. The “‘Watercatox’” process: Wet air oxidation of industrial effluents in a catalytic membrane reactor. First report on contactor CMR up-scaling to pilot unit. Catalysis Today 118, 246–252. Iojoiu, E.E., Miachon, S., Landrivon, E., Walmsley, J.C., Ræder, H., Dalmon, J.A., 2007. Wet air oxidation in a catalytic membrane reactor: model and industrial wastewaters in single tubes and multichannel contactors. Applied Catalysis B: Environmental 69, 196–206. Iojoiu, E.E., Walmsley, J.C., Raeder, H., 2005. Catalytic membrane structure influence on the pressure effects in an interfacial contactor catalytic membrane reactor applied to wet air oxidation. Catalysis Today 104, 329–335. Krishna, R., Sie, S.T., 1994. Strategies for multiphase reactor selection. Chemical Engineering Science 49, 4029–4065. Melada, S., Pinna, F., Strukul, G., Perathoner, S., Centi, G., 2005. Palladium-modified catalytic membranes for the direct synthesis of H2O2: preparation and performance in aqueous solution. Journal of Catalysis 235, 241–248. Melada, S., Pinna, F., Strukul, G., Perathoner, S., Centi, G., 2006. Direct synthesis of H2O2 on monometallic and bimetallic catalytic membranes using methanol as reaction medium. Journal of Catalysis 237, 213–219. Miachon, S., Perez, V., Crehan, G., Torp, E., Ræder, H., Bredesen, R., Dalmon, J.A., 2003. Comparison of a contactor catalytic membrane reactor with a conventional reactor: example of wet air oxidation. Catalysis Today 82, 75–81. Missen, R.W., Mims, C.A., Saville, B.A., 1999. Introduction to chemical reaction engineering and kinetics. John Wiley & Sons, New York. Nauman, E.B., 2008. Chemical reactor design, optimization and scaleup. John Wiley & Sons, Hoboken, New Jersey. Ong, A.L., Bottino, A., Capannelli, G., Comite, A., 2008. Effect of preparative parameters on the characteristic of poly(vinylidene fluoride)-based microporous layer for proton exchange membrane fuel cells. Journal of Power Sources 183, 62–68. Otsuka, K., Yamanaka, I., 1990. One step synthesis of hydrogen peroxide through fuel cell reaction. Electrochimica Acta 35, 319–322. Ozdemir, S. S., Buonomenna, M. G., Drioli, E., 2006. Catalytic polymeric membranes: Preparation and application. Applied Catalysis A: General 307, 167–183. Pan, X.L., Liu, B.J., Xiong, G.X., Sheng, S.S., Liu, J., Yang, W.S., 2000. Exploration of cinnamaldehyde hydrogenation in Co–Pt/-Al2O3 catalytic membrane reactors. Catalysis Letters 66, 125–128. Paul, D.R., 2010. Fundamentals of transport phenomena in polymer membranes, in: Comprehensive membrane science and engineering. Elsevier, Oxford, pp. 75–90.

© Woodhead Publishing Limited, 2013

Multi-phase catalytic membrane reactors

185

Perry, R.H., Green, D.W., 1997. Perry’s chemical engineers’ handbook, seventh. ed. McGraw-Hill, New York. Peureux, J., Torres, M., Mozzanega, H., Giroir-Fendler, A., Dalmon, J.A., 1995. Nitrobenzene liquid-phase hydrogenation in a membrane reactor. Catalysis Today 25, 409–415. Pinna, F., 1998. Supported metal catalysts preparation. Catalysis Today 41, 129–137. Raeder, H., Bredesen, R., Crehan, G., Miachon, S., Dalmon, J.A., Pintar, A., Levec, J., Torp, E.G., 2003. A wet air oxidation process using a catalytic membrane contactor. Separation and Purification Technology 32, 349–355. Reif, M., Dittmeyer, R., 2003. Porous catalytically active ceramic membranes for gas– liquid reactions: a comparison between catalytic diffuser and forced through flow concept. Catalysis Today 82, 3–14. Sander, R., 1999. Compilation of Henry’s Law Constants for Inorganic and Organic Species of Potential Importance in Environmental Chemistry. URL: www.henrys-law.org (accessed 10 January 2013). Shukla, A., Kumar, A., 2004. Experimental studies and mass-transfer analysis of the hydrolysis of olive oil in a biphasic zeolite-membrane reactor using chemically immobilized lipase. Industrial and Engineering Chemistry Research 43, 2017–2029. Singh, D., Rezac, M.E., Pfromm, P.H., 2009. Partial hydrogenation of soybean oil with minimal trans fat production using a Pt-decorated polymeric membrane reactor. Journal of the American Oil Chemists’ Society 86, 93–101. Sisak, C., Nagy, E., Burfeind, E., Schugerl, K., 2000. Technical aspects of separation and simultaneous enzymatic reaction in multiphase enzyme membrane reactors. Bioprocess Engineering 23, 503–512. Tennison, S.R., Arnott, K., Richter, H., 2007. Carbon ceramic composite membranes for catalytic membrane reactor applications. Kinetics and Catalysis 48, 864–876. Tilgner, I.C., Lange, C., Schmidt, H.W., Maier, W.F., 1998. Three-phase hydrogenations with microporous catalytic membranes. Chemical Engineering and Technology 21, 565–570 Torres, M., Sanchez, J., Dalmon, J.A., Bernauer, B., Lieto, J., 1994. Modeling and simulation of a Three-phase catalytic membrane reactor for nitrobenzene hydrogenation. Industrial and Engineering Chemistry Research 33, 2421–2425. Trusek-Holownia, A., Noworyta, A., 2005. A catalytic membrane for hydrolysis reaction carried out in the two-liquid phase system—Process modelling. Journal of Membrane Science 259, 85–90. Vankelecom, I.F.J., 2002. Polymeric membranes in catalytic reactors. Chemical Reviews 102, 3779–3810. Veldsink, J.W., 2001. Selective hydrogenation of sunflower seed oil in a three-phase catalytic membrane reactor. Journal of the American Oil Chemists’ Society 78, 443–446. Vospernik, M., Pintar, A., Berčič, G., Levec, J., 2003a. Experimental verification of ceramic membrane potentials for supporting three-phase catalytic reactions. Journal of Membrane Science 223, 157–169. Vospernik, M., Pintar, A., Berčič, G., Levec, J., 2003b. Mass transfer studies in gas– liquid–solid membrane contactors. Catalysis Today 79–80, 169–179.

© Woodhead Publishing Limited, 2013

186

Handbook of membrane reactors

Vospernik, M., Pintar, A., Berčič, G., Levec, J., Walmsley, J., Ræder, H., Iojoiu, E.E., Miachon, S., Dalmon, J.A., 2004. Performance of catalytic membrane reactor in multiphase reactions. Chemical Engineering Science 59, 5363–5372. Vospernik, M., Pintar, A., Levec, J., 2006. Application of a catalytic membrane reactor to catalytic wet air oxidation of formic acid. Chemical Engineering and Processing: Process Intensification 45, 404–414. Wehbe, N., Guilhaume, N., Fiaty, K., Miachon, S., Dalmon, J.A., 2010. Hydrogenation of nitrates in water using mesoporous membranes operated in a flow-through catalytic contactor. Catalysis Today 156, 208–215. Wu, S., Bouchard, C., Kaliaguine, S., 1998. Zeolite containing catalytic membranes as interphase contactors. Research on Chemical Intermediates 24, 273–289. Yawalkar, A.A., Pangarkar, V.G., Baron, G.V., 2001. Alkene epoxidation with peroxide in a catalytic membrane reactor: a theoretical study. Journal of Membrane Science 182, 129–137.

4.8 4.8.1

Appendix: nomenclature Notation

AM C c Cs D Da Deff f (geometry) H, H′ k kc kg kl A m n J p P Δp r Re rp

membrane area (m2) molar concentration (mol m−3) molar concentration of the sorbate (mol m−3) concentration at the catalyst surface (mol m−3) diffusivity (m2 s−1) Damköhler number (−) effective diffusivity (m2 s−1) function of (geometry) Henry constant (mol m−3 Pa) kinetic constant (m s−1) local mass transfer coefficient (m s−1) mass transfer coefficient in the gas-phase boundary layer (m s−1) mass transfer coefficient in the liquid phase boundary layer (m s−1) characteristic length (m) exponent and coefficient in Equation [4.18] (−) reaction order (−) molar flux (mol s−1 m−2) partial pressure (Pa) pressure (Pa) breakthrough pressure (Pa) reaction rate per volume of catalyst (mol s−1 m3) Reynolds number (−) pore radius (m)

© Woodhead Publishing Limited, 2013

Multi-phase catalytic membrane reactors S Sh Sc u v VM

solubility coefficient (−) Sherwood number (−) Schmidt number (−) fluid velocity(m s−1) Velocity (m s−1) membrane volume (m3)

Greek symbols α β ε ϑ σ ηint ηoverall φ δ ρ τ µ

exponent for Reynolds number (−) exponent for Schmidt number (−) porosity (−) contact angle (−) surface tension (N m−1) internal effectiveness factor (−) overall effectiveness factor (−) Thiele modulus (−) catalytic layer thickness (m) density (kg m−3) tortuosity (−) kinematic viscosity (Pa s)

© Woodhead Publishing Limited, 2013

187

5 Microreactors and membrane microreactors: fabrication and applications P. K. SEELAM , M. HUUHTANEN and R. L. KEISKI , University of Oulu, Finland

DOI: 10.1533/9780857097347.1.188 Abstract: Process intensification (PI) is the future direction for the chemical and process industries and in this chapter, two key technologies to achieve this are discussed: microreactors and so-called membrane microreactors (MMRs). There is great potential to enhance the overall efficiency of microreactors by integrating them with membrane technologies to make MMRs and there are tremendous opportunities for the application of MMRs in many fields. This chapter reviews microreactor design, fabrication and applications as well as materials for micromembranes (MM). The integration of MMs with microreactors and the applications of the resulting MMRs are then discussed. Key words: membrane microreactor, microfluidic device, palladium micromembrane, zeolite, microfabrication.

5.1

Introduction

Process intensification (PI), which represents a new direction in modern chemical engineering, can be defined as a technological toolbox of process improvement tools (Gerven et al., 2009). It is a key area of study in green chemistry and engineering and an important part of strategies for more sustainable development. Engineers and scientists are currently working on green technologies to produce products with less or no environmental impact, and chemical reactor engineering has made great progress in this respect as a result of PI. Miniaturization is a fundamental PI concept and reducing the size of equipment has facilitated the conversion of raw materials to more useful products in a more energy efficient, safe and cost effective manner. (Anastas and Warner, 1998; Ramshaw, 1999). More detailed information on PI can be found in reports such as that by Moulijn et al. (2006) which describes the four main PI domains, namely the thermodynamic, spatial, temporal and functional domains (Moulijn et al. 2006; Van Gerven and Stankiewicz, 2009). Process-intensifying approaches have led to 188 © Woodhead Publishing Limited, 2013

Microreactors and membrane microreactors

189

the development of several methods and equipment set-ups, in which novel reactors as well as intensive mixing, heat transfer and mass transfer devices are employed (e.g., static mixers, monolithic catalysts, microreactors, rotating devices). Moreover, other methods such as the integration of a reaction with one or more unit operations (separation, heat exchange or phase transition) into so-called multifunctional reactors (reverse-flow reactor, reactive distillation, reactive adsorption, membrane reactors, catalytic membranes, reactive extrusion and fuel cells (FCs)) are considered (Moulijn et al., 2006; Rong et al., 2008; Sanders et al., 2011). This chapter considers microreactors and MMRs in particular, being two key technologies for PI. By definition, microreactor technology is the process miniaturization of chemical reactors in sub-micron or sub-millimetre (roughly 50 µm–2 mm) range dimensions, leading to an improvement in both the physical and chemical parameters of reaction engineering. Compared to macroreactor systems, microreactors have many benefits. As well as a high surface-tovolume ratio and high heat and mass transfer rates, microreactors are safe to operate, have low operation, maintenance and construction costs, short residence times and high energy and materials efficiency (Delsman et al., 2005; Gavriilidis et al., 2002; Hessel et al., 2005a). The linear growth in scientific studies on microreactor applications, particularly in the fields of FCs for stationary and portable small-scale power systems, hydrogen production, catalytic studies, fine chemical and organic synthesis, integrated energy systems, functional chemicals and highly exothermic reactions is shown in Fig. 5.1 (Gavriilidis et al., 2002; Gokhale et al., 2005; Holladay et al., 2004; Jensen et al., 2001; Klemm et al., 2007; Pattekar and Kothare, 2004; Watts and Haswell, 2005; Zhang et al., 2004). The introduction of membranes into microstructured reactors has not yet been studied extensively in some fields. However the potential benefits of such integration are wide-ranging. The synergistic effects of multifunctional reactors are the key concept in MMR technology. In a catalytic MMR, for example, both the reaction (catalyst effect) and separation (membrane effect) are combined in a single unit. This reduction in the number of unit processes and unit operations leads to yield enhancement, with more efficient heat and mass transfer achieved (Dittmeyer et al., 2001). Integrating a membrane into a microchannel or microstructured reactor system to create an MMR is a challenging task due to the dimensions of a microreactor, where the height, width and depth are in the range of a few micrometres to submillimetres. This involves multidisciplinary fields in order to effectively design, fabricate and test MMRs. The fact that MMR involves multidisciplinary fields and covers such a wide study area makes it impossible to cover all aspects of it in a single chapter. Most of the research on micromembranes for MMRs places emphasis on Pd-based membranes for hydrogen separation, purification and

© Woodhead Publishing Limited, 2013

190

Handbook of membrane reactors

700 Scopus Science Direct

650 600 550

No. of publications

500 450 400 350 300 250 200 150 100 50

11

10

20

09

20

08

20

07

20

06

20

05

20

04

20

03

20

02

20

01

20

00

20

99

20

98

19

97

19

96

19

95

19

94

19

19

19

93

0

Year

5.1 Number of published articles with the key word ‘microreactor’ in Scopus and Science Direct databases categorized by the year of publication (from 1993 to 2011) (accessed in September 2011).

production by dehydrogenation, steam reforming (SR) and water-gas-shift (WGS) reactions and zeolite MMs for Knoevenagel condensation reaction (KCR) and fine chemicals syntheses. Therefore this chapter is restricted to Pd-based and zeolite MM devices.

5.1.1

Membrane reactors

The term membrane means a permeable phase acting as a selective barrier and controlled by mass transport. A membrane can be porous or dense material, and separation takes place due to a difference in chemical potential gradients (Dittmeyer et al., 2001). There are two materials involved in an MR: a membrane and a catalyst. A membrane can have catalytic and separation functions by itself, or each material can function independently depending upon how the catalyst and membrane are incorporated in an MR. In a tubular MR, the catalyst bed is packed in the annulus or inside the tube, in which case the MR is termed a packed bed membrane reactor.

© Woodhead Publishing Limited, 2013

Microreactors and membrane microreactors

5.2

191

Microreactors

Microreactors are microfluidic devices with dimensions in the sub-millimetre range that is, the width−height−diameter−length are in the range of 10–1000 µm. Generally, microreactors or microstructured reactors are different from conventional reactors, and are manufactured via different fabrication methods using silicon or glass materials. Miniaturization of process and reactors has started in the late 1980s, and various emerging applications (e.g., micro-fuel cell processors) and technologies have been driving the trend of miniaturization over the last two decades (Mills et al., 2007). According to Holladay et al. (2004), on-site and on-demand production of hazardous chemicals is the main motivating factor. Due to environmental regulations and energy security issues, alternative techniques to produce products in a more sustainable way, by reducing, for example, the formation of waste by-products and energy consumption must be found. One such alternative is microreactor technology (MRT). The first attempt to translate this concept into practice was done by DuPont scientists at the beginning of 1987. They demonstrated a prototype microreactor which was fabricated using microelectromechanical systems (MEMS) techniques. This research continued until the 1990s, with studies containing experimental chemical reactions on a miniaturized scale (Mills et al., 2007). Microreactors can play a significant role in limiting the transportation and production of hazardous chemicals, reducing by-product formation, increasing atom efficiency and safety (e.g., phosgene synthesis), and other various factors reported by Mills et al. (2007), for example. The number of patents and scientific articles in the field of MRT has grown exponentially during the last decades (Ehrfeld et al., 2000; Hessel et al., 2008). In comparison to conventional reactor systems, microreactors are easier to scale up by numbering-up (external or internal numbering). Most microreactors are made from silicon wafer or Si bulk using traditional semiconductor microfabrication methods, whilst other materials such as ceramic, glass and stainless steel have also been used in their design. The design of microreactors made from these materials is based on the application type, thermal conductivity, mechanical, electrical and electronic properties of each material.

5.2.1

Advantages and disadvantages of microreactors

To build a miniaturized process unit with an integrated microreactor system can only be considered beneficial if there is no technical or economic advantage in using a conventional system instead. It is therefore necessary to identify those reactions and processes which can be run successfully and beneficially in a microreactor compared to a conventional reactor. There

© Woodhead Publishing Limited, 2013

192

Handbook of membrane reactors

Table 5.1 Advantages and disadvantages of microreactors (Ehrfeld 2000a; Mills et al., 2007) Advantages

Disadvantages

High surface-to-volume ratio Improved heat and mass transfer rates

Clogging or fouling Challenges in numbering-up (scale-up) Catalyst deactivation Malfunctioning of distributors Reliability for long time on stream Leaks between the channels Mixing efficiency Cost issues

Compactness Short and narrow RTD Enhanced safety Mitigation of runaway reactions Faster system response Faster research results and process development Light weight

Short residence times require fast reactions

Better process control High yield and selectivity Increased conversion Quick start and shutdown Easier scale-up Distributed production (on-site)

are several advantages, as well some disadvantages, in using microreactors as summarized in Table 5.1. Regulations concerning the safety, health and environmental issues in the process industry, as seen in the chemical process industry for example, have recently become increasingly rigorous. Thus, the potential application of MRTs has begun to receive greater attention, as the reactions in microstructured or microchannel reactors can be run in safe conditions. A number of examples can be found of MRT applications reported as process improvements in industries such as fine chemicals, organic synthesis, pharmaceuticals and hydrogen production (Hessel et al., 2005a; Jensen 1999; Jensen et al., 2001). Furthermore, the potential for portable automotive fuel processing or on-site hydrogen production (facilitated by compact, lightweight technology) are exciting applications, and good examples of efficient utilization of the key advantages of MRT (Holladay et al., 2004).

5.2.2

Flow phenomena in microreactors

The flow phenomena in microfluidic devices such as microchannels have been studied extensively during the last few decades (Ehrfeld et al., 2000; Hessel et al. 2005b; Papautsky et al., 2001). Fluid flow in microstructured or microchannel reactors is quite different from that in conventional macro reactor systems, due to the smaller hydraulic diameter of the

© Woodhead Publishing Limited, 2013

Microreactors and membrane microreactors

193

channels. Generally, flow phenomena are analysed using the Navier-Stokes equations (convection−diffusion equations). However, these equations are not always capable of describing the flow phenomena in microreactors. In macro-scale conventional reactors, some parameters are neglected (e.g., molecular, interfacial tension, wall friction, roughness of surface, viscosity of the fluid and rarefactions effects) due to the large dimensions. These parameters have a significant role in microreactor design for gaseous flows. The molecular effect on the momentum transfer in directions other than the streamwise direction can increase significantly when the length of the flow channels is reduced and the continuum assumption becomes invalid (Alfadhel and Kothare, 2005; Ratchananusorn, 2007). When the length scale of the flow domain is reduced, the surface phenomena become more important, and surface effects such as wetting and spreading, for example, become dominant, whereas on the macro-scale these phenomena can be ignored. Pfahler et al. (1991) have reported that the variation in fluid properties such as viscosity can occur due to temperature variation in microfluidic transport on a micro-scale, which invalidates assumptions of constant properties (Coleman and Colin, 1999; Pfahler et al., 1991). For gaseous flows, the local statistical distribution of Maxwell–Boltzmann is assumed for the velocity of the particles in the co-moving fluid, but this assumption may breakdown when the gases flow in microchannels at high temperature or low pressures. The flow in a microreactor is predominantly laminar due to the small hydraulic diameter of the channels, making the Reynolds number (Re) very small. The diffusion paths for heat and mass transfer are also very small, thus making microreactors ideal candidates for heat and mass transfer limited reactions. In conventional macro-scale systems with large diameter flow channels, the flow is mainly dominated by the influence of gravitational force, whereas in micro-scale systems with small diameter flow channels, the flow is dominated by the wall friction force, molecular effects and viscosity (Kolb et al., 2004). For gas, the standard continuum flow regime and its deviation is described by the Knudsen number (Kn) (Equation [5.1]): Kn =

λ L

[5.1]

λ=

K BT

[5.2]

2 2π pd dm

where Kn is the ratio of two length scales, λ is the mean free path of gas molecules (Equation [5.2]), L is the characteristic length scale of the flow domain, KB is the Boltzmann constant, T is temperature, p is pressure and

© Woodhead Publishing Limited, 2013

194

Handbook of membrane reactors

dm is the molecular diameter. When Kn > 1, it is likely that a gas molecule collides with the channel wall, rather than with another molecule. The flow behaviour may change dramatically if Kn exceeds 1. This may happen when gas flows through narrow channel are considered, and also when the temperature is high and/or pressure is low. Four different flow regimes can be distinguished based on Kn: continuum flow with no-slip boundary conditions (Kn ≤ 10−2); continuum flow with slip boundary conditions (10−2 < Kn ≤ 10−1); transitions flow (10−1 < Kn ≤ 10) and free molecular flow (Kn > 10) (Hessel et al., 2005b). In microreactors, the friction factor is not independent of wall surface roughness. Moreover, molecular interaction with the walls increases relative to intermolecular interactions when compared to macro-scale flows. In macro-scale systems, two boundary conditions will be applied, that is, a no-slip-flow in which the fluid next to the wall exhibits the velocity of the fluid normally being zero in the most common conditions, and a slip flow in which the velocity of the fluid next to the wall is not zero, and is affected by the wall friction effects and shear stress at the wall. In the case of the slip-flow conditions, a significant reduction in the friction pressure drop and thus reducing the power consumption required to feed the fluid into the microchannel reactor. For most cases in microreactors, the Kn = 0.1 continuum flow with slip boundary conditions is applied. In addition, the pressure drop inside the microreactor is minimal in comparison to that of macro-scale systems (Hessel et al., 2005b). The two main non-dimensional parameters used to characterize the fluid flow are the Re and the Darcy friction factor (f). The Re depends on four quantities: the diameter of the flow, viscosity, density and average liner velocity of the fluid (Equation [5.3]). Re =

ρ f U m Dh µ

[5.3]

where ρf is the density of the fluid (kg/m3), Um is the mean fluid velocity (m/s), Dh is the hydraulic diameter (m) and µ is the fluid viscosity (Pa·s). Dh can be calculated by Equation [5.4]: Dh =

2WH W H

[5.4]

where W is the width of the channel (m) and H is the height of the channel (m). When the hydraulic diameter decreases, the pressure difference increases by an order of two. Simultaneously, the Re decreases and the friction factor

© Woodhead Publishing Limited, 2013

Microreactors and membrane microreactors

195

increases due to the pressure drop. The Darcy friction factor (f) is calculated using Equation [5.5] and the pressure drop, which is the function of the friction factor, is described by Equation [5.6]. The roughness of the microchannel walls plays a role in the friction factor, as well as in the transition to the turbulent flow regime. If the wall roughness in a microchannel structure increases, the friction factor increases in line with the pressure drop. The friction coefficient (Cf) which determines the relation between the Darcy friction factor and the Re is presented in Equation [5.7]. f =

2 Δp pD p Dh

[5.5]

2 ρL LU m

Δ = Δp

2 fLρU m 2 Dh

[5.6]

Cf = fRe

[5.7]

For a single-phase flow, in a rectangular channel for example, the flow phenomenon is assumed to be laminar layered and fully developed due to small hydraulic diameter (Dh), resulting in low Re values. A transition or even a turbulent flow regime may occur in the corrugated flow channel, resulting in larger Re values. Moreover, the rarefaction effects can also occur at normal pressures in a microchannel, resulting in a deviation from the continuum flow behaviour. With regards to slip boundary conditions, a three-dimensional Navier-Stokes equation is relevant and commonly applied for the flow in microchannels. A single-phase flow in a micro-scale reactor is similar to that in a macro-scale reactor, but the interactions between the fluid and the surface properties of the wall on the rarefaction effect for gas and liquid flow should be taken into consideration whilst determining the boundary conditions (Hessel, 2005b). The no-slip boundary condition is not valid for all cases in micro-scale flows, as these boundary conditions can result in a slip flow which can cause a reduction in the pressure drop. A two-phase flow in a microreactor is much more difficult to analyse than a single-phase flow. A two-phase flow, gas−liquid, for example, exhibits many flow regimes and parameters which will affect the flow and patterns in a microchannel (Waelchli and Rohr, 2006). The flow velocities of the dominant fluids, flow regimes (bubbly, slug, stratified, wavy and annular flow regimes) and contacting principles of the two phases are the two most significant factors (Doku et al., 2005; Serizawa et al., 2002). The ratio of flow velocities of each phase is critical and determines the flow domains. Generally, two-phase flows in a microchannel reactor are based on a gas−liquid flow, with the two phases fed separately with different types of mixing or contacting principles

© Woodhead Publishing Limited, 2013

196

Handbook of membrane reactors

in a continuous contact phase. The regimes and transitions of flow depend on many factors, such as the geometry of the microchannels, flow velocities and the mixing or contacting principle. In the case of round tubes, slug, annular, bubbly and churn flow regimes can be observed, whereas for squared tubes irregular flow behaviour is found (Gokhale et al., 2005). As observed by Coleman and Colin (1999) and Serizawa et al. (2002), the most common flow regimes, including stratified, intermittent, annular, dispersed and bubbly, occur when air (dispersed phase) and water (continuous phase) are fed through a rectangular microchannel reactor. In both gas−liquid and liquid−liquid flows, the interfacial mass transfer between the two phases is high when the flow pattern consists of alternating plugs or bubbles (Kashid et al., 2007; Serizawa et al., 2002). Understanding flow regimes and the transitions plays an important role in designing microreactors. The manifold structures for flow distribution of compounds/streams inside microchannels should be considered thoroughly in order to ensure uniform flow, and efficient heat and mass transfer are achieved. For a liquid phase flow in a microreactor, mixing is the critical design issue due to the small Re values, and it has been found that mixing occurs due to diffusion. A concise and detailed review of characterization in the single or multiphase flow, and mixing phenomena in microchannels is reported by Aubin et al. (2010) and Doku et al. (2005).

5.3 5.3.1

Microreactor design and fabrication methods Microreactor design

The design of microreactors requires extensive knowledge with regard to material choice, fabrication methods, kinetics, transport rates, catalyst coating and loading, location of sensors and intrinsic conductivity (Gokhale et al., 2005). Computational fluid dynamics (CFD) is one possible and good method for the design and optimization of microreactor parameters (Kashid et al., 2007). Recent advances in CFD modelling and simulations provide relatively precise knowledge on flow, temperature and pressure distribution, without the need to perform any experiments. Various types and shapes of microreactors exist for single and multiphase flow systems in the field of MRT. The selection of an appropriate mixer and/ or reactor type and shape for a certain process depends on the characteristics of the reaction. For heterogeneous gas phase reactions, a microchannel device (a thin wall coated catalyst) with an integrated heat exchanger, sensors and temperature controllers can be selected. In the case of endothermic reactions, a combustion channel capable of supplying energy integrated with the reaction channels is the most optimal design, for a micro-fuel cell processor, for example. Catalytic wall and packed bed microreactors are

© Woodhead Publishing Limited, 2013

Microreactors and membrane microreactors

197

appropriate choices for single-phase reactions, whereas capillary and falling film microreactors, as well as microcontactors and the aforementioned reactor types, are used for multiphase reactions (Kiwi-Minsker and Renken, 2005). From a design point of view, it is important to understand how to introduce two separate flows into one microchannel. In addition, the relative velocities of the flows have a significant influence on the resulting pattern of the multiphase flow. Another important aspect is how to introduce the catalysts’ active phase for a heterogeneous reaction where the solid catalyst is coated on the wall and/or placed as a packed bed inside a reactor. Even though the packed bed reactors are easier to fabricate than catalytic wall microreactors (CWM), CWMs are still favoured in most cases due to lower pressure drop and as they exhibit higher heat transfer rates (Kin et al., 2006). Besides choosing the proper type of a reactor, the geometry and appropriate microreactor structure are also important decisions. Multi-channel CWMs are most commonly used and have numerous advantages over conventional reactors, overcoming potential limitations related to volumetric flow rates and numbering-up for microchannel units. The dimensions of a microreactor and its channels have to be determined based on the throughput, and optimal dimensions should maximize the most important characteristics and parameters. The residence time distribution in a microchannel with a laminar flow profile is strongly dependent on the diffusion coefficients of the species, and also on the channel dimensions (Kolb et al., 2004). According to the study conducted by Tonomura et al. (2004a) an optimal design for a plate-fin microreactor typically contains parallel microchannels with inlet and outlet manifolds. The two main design parameters are thermal and fluid design. In thermal design, the main objective function is unformalization of fluid temperature, with the optimization of variables such as the microchannel shape, flow rate, coolant temperature and constraints (maximum pressure drop, reaction temperature and yield/selectivity). In fluid design, minimization of the total residence time distribution (RTD), the manifold shape and the number of microchannels are the optimization variables. Moreover, throughput, RTD and maximum pressure drop are constraints which must be considered. In the thermo−fluid design approach, by changing the channel width of the longitudinal position, a uniform temperature distribution with no hot spots has been achieved during exothermic reactions. Thus, optimal thermo−fluid design is the main goal for the microreactor system (Tonomura et al., 2004a, 2004b). McMullen and Jensen (2010) have made a review of the automation and construction materials used in integrated microreactors, highlighting various materials and fabrication methods. Generally, five types of materials were used: ceramic, glass, plastic, silicon and stainless steel (SS). The most

© Woodhead Publishing Limited, 2013

198

Handbook of membrane reactors

common fabrication techniques are stereolithography (for ceramic), photolithography (for glass and silicon), soft lithography (for plastic), powder and injection moulding (for ceramic and plastic), wet and dry etching (for glass and silicon), laser and electric discharge machining (for ceramic), ultra machining (for glass), electro plating (for SS), micromachining (for SS) and embossing (for plastic). Each material offers specific advantages and disadvantages, and selecting appropriate materials for a microreactor depends on the reaction characteristics, along with the chemical properties and compatibility of the reactants or reagents. For high temperature reactions and separations, for example, ceramic is a good candidate with a low heat loss and chemical resistance. However, the costs are high. For high pressures and temperatures with superior heat conductivity and high aspect-ratio design, silicon and SS microreactors are the most effective choices. Plastic is preferable for fast and inexpensive development, but it is incompatible with organic solvents and not suitable for high temperatures and pressures. Most microreactors formed by silicon wafers are patterned to form microstructures or channels with heaters, sensors and catalytic reaction zones.

5.3.2

Microfabrication methods for microreactor devices

Microfabrication (MF) methods are widely used in microchemical systems design, especially in the case of MMRs where the reaction and separation are performed in a single unit. The use of MF methods for micro-devices has increased very rapidly in the design of MEMS, MSs, electronic circuits, microelectronics, semiconductors and energy systems (Jensen, 1999, 2001; Qin et al., 1998). MF techniques create new opportunities for chemical reaction engineering, and are used to build compact, efficient microreactor systems. The use of MF methods offers many advantages, including a reduction in the consumption of expensive reagents, fluidic components with small dead volumes, improved separation resulting from higher surface-to-volume ratios, integration of sensors, actuators and parallel screening, replication for multiple unit production and compactness. MF techniques used in microreactor components design and manufacturing have the potential to replace conventional energy production devices and macro-scale reactors. One such potential application is on-site hydrogen production using a micro-scale reformer unit, integrated with a micro-fuel processor (a PC-card sized microchemical system with integrated microfluidic, sensor, controller and reaction components) (Jensen, 1999). Moreover, miniaturization offers improved heat and mass transfer rates, and enables the design of more compact and efficient reaction and separation units.

© Woodhead Publishing Limited, 2013

Microreactors and membrane microreactors

199

By applying MF techniques, the reaction heat can be controlled by varying the thickness and thermal conductivity of the wall. Geometrical parameters and construction materials are the key selection criteria in designing the microreactor systems, whilst for microsystems which perform chemical reactions, separations, analyses, and sensing devices, the channels, cavities, valves and electrodes all need to be designed and selected properly. In order to create MSs or micro units capable of performing the various required operations, it is important to understand which materials and methods are the most economic, reliable, accurate, stable and efficient (Ehrfeld et al., 2000). Generally, rigid materials such as Si are preferred, and many microfluidic devices are thus built on silicon substrates, along with other substrates such as crystalline and amorphous Si, glass, metal, plastic and polymers (Qin et al., 1998). Single-crystal Si substrates are used in many microsystems due to the shapes and patterns which can be reproduced with high precision by bulk and surface micromachining techniques. Moreover, Si/SiO2 is chemically and thermally stable and also extensively used in the electronics industry (e.g.in integrated circuits). Microreactors can be fabricated using high-volume and low-cost techniques, but the final price of a microreactor depends on many factors, including design parameters, materials and fabrication costs. Some of the MF techniques most widely applied in the creation of microstructured reactors are as follows (Ehrfeld et al., 2000; Gavriilidis et al., 2002; Hessel et al., 2005c; Qin et al., 1998): 1. Microlithographic techniques (photolithography, soft lithography, stereolithography, microcontact printing, moulding organic polymers, etc.); 2. LIGA – combination of deep lithography, electroforming, moulding, micromachining with laser radiation; 3. micromilling; 4. laser ablation; 5. micro moulding; 6. wet and dry chemical etching (on Silicon, glass materials); 7. electro-discharge micromachining (EDM); 8. other advanced techniques like turning, sawing, punching, embossing, drilling, laser micromachining, etc.; and 9. bonding techniques – gaskets, welding, sintering, electron-beam welding, diffusion bonding, soldering and laser welding. Fabrication of MMs has been studied by many authors, who have investigated Pd and zeolite based membranes on Si substrates. Generally, MMRs are fabricated using silicon (Si) wafer substrates for micro-fuel

© Woodhead Publishing Limited, 2013

200

Handbook of membrane reactors (a)

Si wafer (b)

(c)

(d)

(e)

5.2 Microfabrication procedure of a thin Pd membrane by Zhang et al. (a) Preparation of a Pd thin micromembrane on a Si wafer by dc sputtering. (b) Preparation of negative resist (65 μm thick). (c) Deposition of about 50 μm Ni layer by electroplating. (d) Removal of the negative resist. (e) Removal of Si wafer by wet etching in KOH. (f, g) SEM images of the prepared Pd micromembrane (thickness 2.5 μm) (Zhang et al., 2006) (Copyright permission 2006 Elsevier). (Continued)

processors, and function as integral components of miniature devices. Ye et al. (2005) have studied oxidized porous silicon (PS) supported thin Pd membranes. They used the MF technique to produce better adhesion between the support and the Pd with a smooth surface of PS. The microfabrication procedure for thin Pd membranes on Si substrates is presented in Fig. 5.2.

© Woodhead Publishing Limited, 2013

Microreactors and membrane microreactors

201

(f)

Pd Ni metal

5kV

⫻500

50 µm

(g)

Ni metal Pd membrane

2.5 µm

5kV

⫻2, 200

10 µm

5.2 Continued

5.4

Micromembranes

In the following section, methods for the fabrication and deposition of Pd-based and zeolite MMs are discussed, as well as applications in (de) hydrogenation, SR, WGS, partial oxidation (POx) reactions and fine chemical synthesis. The research on Pd-based MMRs for hydrogen separation, purification and production (by dehydrogenation, SR and WGS reactions) has been selected as a case study, as significant research and, therefore, much information can be found in the literature on this field.

© Woodhead Publishing Limited, 2013

202

5.4.1

Handbook of membrane reactors

Pd-based micromembranes

Conventional Pd-based membranes are categorized into self-supported Pd-based membranes, and supported composite membrane with a thin Pd layer deposited on porous supports. Self-supported Pd-based membranes (SSP) are available commercially, are capable of providing adequate mechanical strength and can easily be integrated into a reactor set-up. SSPs are normally relatively thick (50 µm or thicker) and exhibit low hydrogen fluxes with high hydrogen perm-selectivity (almost infinite with respect to other gases). However, the cost of Pd has increased exponentially during recent years, and as the flux is inversely proportional to the thickness, it makes SSPs an expensive choice. In order to reduce the thickness (i.e., higher fluxes) and to have better mechanical and thermal stability, thin Pd films are deposited on various supports such as porous stainless steel (PSS), Al2O3, ceramic and ZrO2. In Seelam et al. (2012), a 20 µm thick Pd thin layer was deposited by the ELP method on a PSS supported membrane module. The prepared membrane was investigated in SR reactions, and was concluded to exhibit good selectivity and a high hydrogen flux in comparison to a dense 50 µm thick Pd–Ag layer. By using a sub-micron thickness Pd-based membrane, the cost is not only reduced but the hydrogen flux is also enhanced. However, it is difficult to prepare a defect-free hydrogen separating membrane including both a Pd film and porous support. The commercially available porous supports are not defect or pinhole free materials, and also have surface imperfections such as non-uniform pore sizes, which make the metal films unable to completely cover up the support, leading to membrane defects and cracks. Furthermore, the supports may have a thick form which has a considerable higher mass transfer resistance, thus negatively affecting the separation flux (Tong, 2004). Thin film coverage of the pores of the support may also be insufficient, and thus the walls of the pore may not be covered completely by the metal film. In the studies by Bryden and Ying, (1995) a thin Pd-based film with a nanostructured membrane was deposited on porous Vycor glass disks. Leakage was found to occur due to small pinholes related to surface defects in the Vycor glass support. When exposed to hydrogen, the membrane can also cause cracks due to the expansion and contraction of Pd. In order to suppress a phase transition from α to β in the Pd metal lattice and grain-growth, alloying of Pd with other metals such as Ag, Ru, Cu, Rh, Ni is used (Bryden and Ying, 1995). The support structure is very important in avoiding pinholes or defects in the membranes that could potentially lead to gas leakages. Thus, the pore size of the top layer support must be reduced. In order to deposit an appropriate thin Pd film, avoiding surface imperfections of the support and film coverage, more advanced methods should be employed to produce durable, cost effective membranes for hydrogen separation and

© Woodhead Publishing Limited, 2013

Microreactors and membrane microreactors

203

purification. Thin SSP membranes cannot withstand high pressures, and there is a clear need for strong Pd-supported membranes with good adhesion and gas-tightness, which can be improved by a reduction in pore size and surface roughness of the support (Ye et al., 2005). Most Pd-based membranes operate at high temperatures (HT) ranging from 300°C to 500°C. In addition, the hydrogen embrittlement factor, lattice strains and thermal stability are major problems. Furthermore, high temperature Pd-based MMs offer potential for application in miniaturized micro-FCs, WGS, SR and hydrogenation reactions. These membranes can be permselective to hydrogen, and in addition, occasionally to CO2 when operated at high pressures, providing even greater advantages for further processes. As these developments reveal, in the 21st century advanced MMs are developed with high permeation efficiency using micro- and nano-engineering techniques. In the study conducted by Kim (2008), HT MMs using Si wafers and SS supports were prepared, and the permeation properties for hydrogen and CO2 were reported. The permeation property of the MMs depends strongly on the geometry, morphology, preparation methods of the support and uniformity of film deposition. Chen and Gobina (2010) reported that a support with high hydrogen affinity is much more efficient, due to the resistance to hydrogen-induced failures or cracks. Generally, thin membranes exhibit higher hydrogen flux (2–10 mol/ m2s) but a selectivity decline. Thus, perm-selectivity and permeability are trade-off parameters. Tong et al. (2003, 2004, 2005a) have used microfabrication techniques to produce thin Pd–Ag MMs on substrates such as silicon wafer and MPSS. The performance of Pd-based MMs manufactured with a thickness less than 20 µm Pd and deposited using different fabrication techniques, are summarized in Table 5.2. Furthermore, palladium MMs with very thin features can be applied in microfluidic devices, and some applications are already suggested by the referred authors. Zhang et al. (2006) have also reported on thin Pd MMs with thicknesses of 2.5 µm and 0.7 mm, produced on an Si wafer for hydrogen separation applications, via MF technology and using a sputtering method to deposit the Pd film. Permeation tests were conducted using pure hydrogen and hydrogen/ CO2 mixtures at a temperature range of 200–400°C. The hydrogen permeability for this thin 2.5 µm Pd membrane was 50–60% of that achieved with a 0.7 mm membrane. Figure 5.2a–e presents the fabrication procedure for the Pd thin membranes with 50 µm Ni layer as a support, and Fig. 5.2f, g presents scanning electron micrograph images of thin Pd 2.5 µm and 50 µm Ni supports before the permeation test was done. During the SR and WGS reactions and separation processes, the inhibiting effects of CO2 and steam on the hydrogen flux for thin Pd–Ag membranes (1 µm and 0.5 µm thickness) are stronger than for thicker membranes (>10 µm) (Gielens et al., 2006). Moreover, CO2 and CO are reported to

© Woodhead Publishing Limited, 2013

Table 5.2 Performance of microfabricated Pd-based micromembranes deposited by various techniques

© Woodhead Publishing Limited, 2013

Membrane

Thickness Hydrogen (µm) flux (mol/m2·s)

Separation factor

Operating temperature (oC)

Deposition Remarks method

Pd–Ag/ceramic

0.35

0.015

250

SD

Δp = 1 atm; n = 1

Pd/MPSS Pd/SS

6 10

0.3 0.089

H2/N2 = 5.7, H2/He = 2.2 H2/He, Ar = ∞ H2/N2 > 1000

500 480

MD/ELP ELP/O

Pd–Ag Pd/ceramic

1.5–2 11.4

0.07 0.71

H2/N2 = 24 H2/N2 = 650

500 550

SP ELP

Δp = 1 atm Δp = 0.987 atm; n = 0.5 n=1 n = 0.6

Pd75%Ag25%γ-Al2O3

0.16–0.52

0.01

300

SD

n > 0.5

Pd/porous α-alumina Pd76%Ag24%/ polymeric Pd and Pd–Ag/γ-αAl2O3 Pd–Ag/ceramic Pd77%–Ag23%/Si wafer Pd–Fe5%

3–5

>0.1

H2/N2 = up to 116, H2/He = 3845 H2/N2 = 1000

300–500

MOCVD

pH2

0.25–1

0.002

H2/CO2 > 100

25

SD

n=1

0.1–1.5

0.01–0.02

H2/N2 = 30–200

25–300

0.7–1.1 1

0.05 0.5

H2/N2 = 4–80 H2/N2 > 550

350 450

MOCVD Δp = 1 atm and MS MS Δp = 1–4.93 atm CS Δp = 0.297 atm

18

10*

H2/He = 30

200

PE

Δp = 1 atm

Pd/cordierite Pd

8–16 0.2

0.001–0.005 H2/He = 40–360 350 5.2 100

ELP SD

Δp = 0.1–0.5 atm; n = 0.5 Δp = 0.56 atm; n = 0.5

1 atm n

References

Jayaraman and Lin, 1995 Goltsov, 1977 Li et al., 1998 Li et al., 1993 Collins and Way, 1993 McCool et al., 1999

05

Yan et al., 1994 Athayde et al., 1994 Xomeitakis and Lin, 1997 O’Brien et al., 2001 Tong et al., 2003 Bryden and Ying, 2002 Kim et al., 2009 Karnik et al., 2003

© Woodhead Publishing Limited, 2013

Pd–Ag/SiN wafer

0.5

4

H2/N2 > 1500

450

SD

pH2 = 0 82 atm

Tong et al., 2005a

Pd–Ag/Si wafer Pd–Ni/Si wafer

0.2 2.5

3–4 0.8

H2/Ar = 1000 –

350 200–400

EB CS

Δp = 0.96 atm pH2 0 2 atm n = 0.73

Wilhite et al., 2004 Zhang et al., 2006

Pd–Ag/Si microsieve

1

1

H2 > 1500

400–450

Δp = 0.8 atm; n =1

Keurentjes et al., 2004

Pd–Ag/Si3N4

0.75

0.02–0.95

H2/He > 1500

350–450

Single cannon SD CS

Pd

0.34

0.112

H2/N2 = 46, H2/ He = 10

250

SD

Δp = 1.1 atm; pH2 = 2 atm

Pd–Ag/Si3N4

1

3.6

450

CS

Pd77%–Ag23%/Si

2.2

8.4

H2/He = 1500–2000 H2/N2 = 1400

400

MS

pH2 = 25.66 atm; n = 0.5

Pd–Ag

5.5

0.35

400

ELP

Δp = 3.95 atm

Pd/ZrO2 Pd/PSS Pd/ceramic/PSS Pd–Ag/PSS Pd/Al2O3 HF† Pd/Al2O3 HF‡ Pd/Al2O3 HF§ Pd/Si wafer Pd–Ag23% Pd/porous alumina

10 6 5 16 5 2.5 1.5 6 0.2 1

0.2 – 0.78 0.19 0.135 0.198 0.2387 0.302 2.21

350–500 500 450 450 450 450 450 500 400 300

ELP ELP ELP ELP ELP ELP ELP ELP SD CVD

Δp = 1 atm; n = 0.5 Δp = 0.5 atm; n = 1 Δp = 3.35 atm; n = 0.5 n = 0.6 Δp = 1 atm; n = 1 Δp = 1 atm; n = 1 Δp = 1 atm; n = 1 Δp = 1 atm; n = 1 n = 0.97

H2/N2 = upto 4500 He/Ar = 3.1 H2/N2 = 450 – H2/N2 = 380 H2/N2 = 340 H2/N2 = 1400 H2/N2 = 3115 H2/N2, Ar = ∞ H2/He > 1000 H2/He > 104

pH2 = 0.2 atm

pH2 = 0.82 atm

Gielens et al., 2002 Ye et al., 2005 Gielens et al., 2002 Peters et al., 2008 Hou and Hughes, 2003 Wang et al., 2004 Su et al., 2005 Li et al., 2007 Yepes et al., 2006 Sun et al., 2006 Sun et al., 2006 Sun et al., 2006 Tong et al., 2005b Wilhite et al., 2006 Yamamoto et al., 2006 (Continued)

Table 5.2 Continued

© Woodhead Publishing Limited, 2013

Membrane

Thickness Hydrogen (µm) flux (mol/ m2·s)

Pd75%–Ag25%/SiO2 Pd–CeO2 /MPSS Pd/HF

1 13 11

1.1 0.275 0.136

Pd–Ag/PG Pd–Ag/MPSS Pd/PSS

21.6 4 5

0.067 0.28 0.155

Pd/Al2O3

7–15

Pd90%–Cu10%/Al2O3

Separation factor

H2/N2 = ∞ H2/N2 = 1000

Operating temperature (oC)

Deposition Remarks method

References

450 500 430

SD/PECVD Δp = 0.18 atm; n = 0.73 CMS/ELP Δp = 0.2 atm ELP Δp = 1.1 atm

400 500 400

ELP IM/ELP ELP

Δp = 0.199 atm Δp = 1.1 atm Δp = 1.1 atm; n = 0.6–0.7

McLeod et al., 2009 Tong et al., 2005c Tong and Matsumura, 2006a Uemiya et al., 1988 Tong et al., 2006b Dittmeyer et al., 2001 Dittmeyer et al., 2001 Roa and Way, 2003

400

ELP

Δp = 1.1 atm; n = 0.7

3.5

H2/N2 = ∞ H2/N2 = ∞ H2/N2 = 100–200 0.086–0.134 H2/ N2=100–1000 0.056 H2/N2 ≤ 7000

350

ELP/O

Pd60%–Cu40%/Al2O3 Pd/α–Al2O3 HF†

1.5 1.1

0.499 0.4

Pd/α–Al2O3 HF‡ Pd/α–Al2O3 HF§ Pd/silicon wafer/ PDMS Pd–Ag23% Pd/PSS

2.6 0.6 4 1 4.4

350 370

ELP/O ELP

0.16 0.25 –

H2/N2 = 93 H2/N2 = 3000–8000 H2/N2 = 500 H2/N2 = 50 –

Δp = 1.7–2.4 atm; n = 0.7–1 Δp =1.7 atm; n = 0.5 Δp = 3.95 atm

370 370 200

ELP ELP SD

Δp = 3.95 atm Δp = 3.95 atm –

Roa and Way, 2003 Nair and Harold, 2007 Nair et al., 2007 Nair et al., 2007 Cui et al., 2000

0.05¶ 0.92

– H2/He = 1124

298 500

SD ELP

Δp = 1.1 atm n = 0.5

Naddaf et al., 2009 Chi et al., 2010

Pd/α–Al2O3

4

0.55

H2/N2 = 6600

500

ELP

pH2

Pd77% Ag23%/SOI

0.2

46

H2/N2 > 1000

350

EB

pH2

Pd–Ag23%

1.9–3.8

18.3

H2/N2 = 2900

400

MS

pH2

Pd–Ag/Si wafer

1

3.6

H2/He = 1500–2000

450

Single cannon SD

pH2

4 93 atm n = 0.5

Israni et al., 2009

Deshpande et al., 2010 25 66 atm n = 0.631 Peters et al., 2011 0 1 2 atm n

0 82 atm n = 1

05

Gielens et al., 2004

© Woodhead Publishing Limited, 2013

* Standard cm3 min−1. Pd encapsulated membrane. ‡ Pd nanopore. § Aged Pd nanopore. ¶ Pa m3 s−1. †

n = Pressure exponential factor; Δp = pressure differential; pH2 = hydrogen partial pressure (retentate); SD = sputter deposition; EP = electroless plating; CVD = chemical vapour deposition; SP = spray pyrolysis; CS = co-sputtering; MS = magnetron sputtering; PSS = porous stainless steel; MOCVD = metal-organic chemical vapour deposition; PE = pulsed electrodeposition; EB = electron-beam deposition; CMS/ ELP = combined method of physical sputtering and electroless plating; O = osmotic pressure method; MPSS = macroporous stainless steel; PG = porous glass; HF = hollow fibre; MD/ELP = multidimensional plating mechanism; IM/ELP = improved method of electroless plating; CMS/ELP = combined method of physical sputtering and electroless plating; RT = room temperature; DSD = dual sputtering deposition; SOI = silicon-on-insulator. In the first column % refers to wt%.

208

Handbook of membrane reactors

chemisorb on the surface of the Pd MM, which may have a negative impact on the hydrogen flux. Su et al. (2005) have reported a very high hydrogen permeability value in comparison to various other studies, using an approximate 2.5 µm thickness of Pd-supported on PSS graded with a SiO2 layer, deposited by a sol–gel technique. A new type of ultra-thin, hydrothermally stable, molecular-sieve supported hydrogen-selective membrane was fabricated as part of a project by the National Research Council, reported under DOE, and operated at high temperatures and pressures. The main goal of the project was to scale up economically, providing an easy way to manufacturing and fabricate hydrogen-selective membranes with high flux and selectivity. Not only was this achieved, but a hydrogen permeance higher than 10−7mol/m2·s·Pa with H2/CO2 selectivity greater than 100 at temperatures in the range of 500– 700°C, and pressure of 20 bar was also obtained. (Plasynski et al., 2008) Franz et al. (2000) have reported a novel Pd-based MMR with controlled selective permeation and/or increased hydrogen flux, using MF techniques and sequential steps to fabricate Pd MM as presented in Fig. 5.3a. Moreover, in-situ palladium and its alloy based MM were prepared as part of microfluidic devices, as shown in Fig. 5.3b (Gielens et al., 2004).

5.4.2

Zeolite micromembranes

Zeolites are crystalline aluminosilicates characterized by a structure comprising a three-dimensional pore system and regular framework formed by linked TO4 tetrahedral (T = Si, Al) with different morphological and physico-chemical properties. Due to their impressive selectivity and uniform pore structure, they have very efficient molecular sieving properties, and are able to separate molecules based on size and shape. Zeolite powders, films and membranes are widely used in catalysis, adsorption and separation applications (McLeary et al., 2006; Pina et al., 2011). Zeolites are cheap and widely available due to their abundance in both natural and synthetic forms. The application of zeolites in the membrane field is growing very fast, and has been the subject of increased research focus during the last few decades (McLeary et al., 2006). In this section, zeolite micromembranes for MMRs will be discussed, and synthesis methods and applications in gas separation and fine chemical synthesis will be introduced. Incorporating zeolite into microreactors presents a range of challenges, including difficulties in the production of homogeneous layers, sufficient coatings, and adhesion and reproducibility. A zeolite itself may act as either a catalyst or a membrane separating layer in microreactors. A zeolite film can be directly grown on microchannels, and during the synthesis the thickness and crystal orientation can be regulated. Many

© Woodhead Publishing Limited, 2013

Microreactors and membrane microreactors (a) Starting material: Si wafer with 0.25 µm of oxide and 0.3 µm low pressure chemical vapour deposition (LPCVD) nitride

Backside KOH etch, to form channel/membrane structure

Pattern backside (dry nitride etch followed by BOE)

Blanket deposition of Pd (200 nm) with a thin Ti (10 nm) adhesion layer.

Pattern perforations on frontside (dry nitride etch)

Opening of Pd membrane using BOE

Heater patterning and metallization (Pt/Ti)

Pt

209

Packaging

Front view

SiNx SiO2 Si Pd Al Side view

5.3 (a) Microfabrication steps for a palladium membrane microreactor (Jensen et al., 2001; Franz et al., 2000) (Copyright permission 2001 Elsevier). (LPCVD in the figure refers to low pressure chemical vapor deposition.) (b) Procedure to manufacture a microsieve-supported Pd–Ag membrane micro system using MF techniques (Gielens et al. 2004) (Copyright permission 2004 Elsevier). (Continued) © Woodhead Publishing Limited, 2013

210 (b)

Handbook of membrane reactors Si wafer double-side polished Deposition of 0.3 µm SiO2 and 0.7 µm SiN

Selective removal of SiO2 and SiN with dry etching

KOH etching to create apertures

Open windows on back-side, dry etching SiN

5 µm

SixNy SiO2 Si

Remove Si completely by KOH, etch stop at SiO2

Co-sputtering of Pd/Ag membrane layer

Release of membrane by removal of SiO2 with BHF

5 µm

SixNy SiO2 Pd/Ag

5.3 Continued

factors may influence the zeolite coatings inside a microreactor, including gel composition, time, support, crystal orientation, temperature, synthesis procedure and Si/Al ratio. These factors can be manipulated to produce the desired characteristics of the selected reaction and separation. Since 2004, many articles on preparation of zeolite MMs have been published, on such areas as MFI or Sil-1 zeolite etched on the Si substrate for gas separation applications, and MMRs for KCR and fine chemical synthesis (Coronas and Santamaria, 2004; Kwan et al., 2010; Wan et al., 2001; Yeung et al., 2005). Coronas and Santamaria (2004) have reported on the use of zeolite films and interfaces in micro-scale and portable applications, including the removal of volatile organic compounds from indoor air, recovery of catalysts in homogeneous reactions, zeolitic microreactors and microseparators, for example. Moreover, zeolite coated microreactors and microseparators exhibit high surface-to-volume ratio, and are capable of high productivity as a result of the good contact between reactants and catalyst wall.

© Woodhead Publishing Limited, 2013

Microreactors and membrane microreactors

211

In the work by Chau et al. (2003) a zeolite layer was grown uniformly on the Si wafer using a variety of fabrication steps. Firstly, a zeolite−silicon composite was fabricated via hydrothermal synthesis. This was then micro-patterned with a resistant coating, using a photolithography etching process (by UV radiation), after which it was exposed to buffered oxide etch (BOE) and photoresist stripping. In an alternative approach, zeolite nanocrystals were grown directly on silicon, with Teflon tape used prior to the growth. Leung and Yeung (2004) have investigated three different types of freestanding microfabricated zeolite MMs (Sil-1 and ZSM-5 with two different Si/Al ratios, namely 40 and 60). In their investigations, the micromembranes were prepared on silicon substrate, and SEM images of this are presented in Fig. 5.4. These zeolite MMs were subjected to gas permeation tests for hydrogen, He, CH4, CO2 and mixtures of these gases. Gas permeation characteristics are dependent on the morphology, Si/Al ratio and synthesis conditions of the zeolite. A similar study was conducted by Kwan et al. (2010) in which ZSM-5 and Sil-1 MMs (Si/Al ratio 14 to ∞) were fabricated on silicon substrate, and gas permeation tests were conducted for pure gas and gas mixtures. Synthesis parameters such as composition, thickness, orientation, crystal grain size, intergrowth and morphology were adjusted to obtain a homogeneous MS. Surface diffusion is the dominant mechanism at low temperature, and the effect of Al of ZSM-5 MM on gas permeation for single gas components (i.e., H2, He, N2, Ar and methane) is greater than the Knudsen diffusion. When the Si/Al ratio in ZSM-5 increases, the deviation from the Knudsen diffusion simultaneously increases. For propaneN2 separation, the ZSM-5 MM is able to separate the gases based on the kinetic diameters of the molecules, producing the molecular sieving effect. The ZSM-5 based MMs exhibit excellent permeance and perm-selectivity values for single, binary and ternary gas permeation and separation. The zeolite MMs are less expensive than the Pd-based membranes and offer strong competition in both hydrogen production and separation capabilities (especially in miniature devices such as microreformer or micro-fuel cell processors). Wan et al. (2001) prepared zeolites (ZSM-5, Sil-1 and TS-1) as catalysts, MMs and structural materials for micro-devices using 3–16 µm layer thickness and film orientation. Micromachining techniques were employed for the device architecture, and four different fabrication methods were compared: firstly, zeolite powder coatings on Si; secondly, with or without seeding via hydrothermal synthesis for uniform film growth on the Si; thirdly, with confined seeding and; fourthly, zeolite-silicon composite fabrication using a photolithographic etching process (by UV radiation, masking, patterning by photo resist).

© Woodhead Publishing Limited, 2013

212

Handbook of membrane reactors

(a)

(b)

100 µm

250 µm

260 µm (c)

(d)

(e)

(f)

5.4 SEM images of (a) zeolite micromembrane units, (b) higher magnification image of a single unit, (c) cross-sectional view of support structure after zeolite growth before etching, (d) zeolite support layer deposited inside the micro-cavity which was etched on silicon, (e) a freestanding zeolite micromembrane layer after etching and (f) layer formed as a result of excessive etching (Adapted from Leung and Yeung, 2004) (Copyright permission 2004 Elsevier).

These kinds of MMs are useful for MMRs and/or incorporation into microfluidic devices with thin-films. The potential applications of such zeolite MMs include separations in pharmaceutical and fine chemical synthesis, as well as in lab-on-chip devices, sensors, zeolite micro-FCs and adsorption screening tools (Pina et al., 2011).

© Woodhead Publishing Limited, 2013

Microreactors and membrane microreactors

5.5 5.5.1

213

Catalyst coating techniques and hydrogen production in microreactors Catalyst coating techniques inside the microreactors

Introducing a catalyst into a microchannel system is an important step in catalyst preparation for heterogeneously catalysed reactions. There are many methods for the incorporation of catalytic materials inside a microstructured reactor. Wash coating, packed bed catalyst filaments or structured monoliths, coatings of commercial catalysts, or catalysts as a part of microreactor fabrication (as in the direct formation of zeolite crystal on a metallic structure) may all be used (Cybulski and Moulijn, 2006). Kolb et al. (2004) reported various coating methods including spin coating, dip coating and drop coating for the incorporation of porous support materials into microchannels, and later introduced Pt by sputtering and wet impregnation. As previously suggested, wall coated catalysts are more efficient than packed bed microstructured catalysts due to issues related to pressure drop. There are two methods generally used: the material-independent method (coating technologies such as wash coating, spray coating and dip coating) or the material-dependent method (e.g., anodic oxidation of aluminium) (Hessel et al., 2005c). Microchannel plates washcoated with porous Al2O3, for example, are available commercially and metals or metal oxides can be introduced to these by impregnation. Alternatively, the commercial catalyst can be coated using a MS. Before coating the ready-made catalyst, the MSs are pre-treated to ensure strong adhesion between the wall and the catalyst. Conventional impregnation methods are employed to prepare a supported catalyst, and the procedure for washcoating by impregnation is the common procedure for catalytic MS preparation. The process for MS wash coating begins with cleaning of the MS in an ultrasonic bath, before thermal pre-treatment, positioning and masking, channel filling with suspension or slurry (e.g., γ-alumina, water and binder), removal of excess suspension, drying, calcination, pre-treatment of porous wash-coats, impregnation (the incorporation of metal and metal oxides for example), and finally drying and calcination. The washcoating of commercial catalyst powders can be carried out using a similar procedure, excluding the pre-treatment of porous wash-coats. Similarly, catalyst coatings can be used to produce MSs via thin film gas phase deposition, with chemical (up to 10 µm thickness) and physical vapour deposition methods (< 1 µm) typically used to deposit metals and metal oxide thin-films. Other methods such as electrophoretic deposition of alumina, ZnO, CeO2 and ZrO2 coating layers inside the microreactors are also used (Cybulski and Moulijn, 2006; Hessel et al., 2005c).

© Woodhead Publishing Limited, 2013

214

Handbook of membrane reactors

Klemm et al. (2007) have reported on microstructured reactors for heterogeneously catalysed reactions (gas or liquid phase reactions) using wall catalysts. In wall coated microreactors, the mechanical stability and the adhesion between the catalyst and the wall are the most important considerations. It is important to avoid blistering of the wall catalyst during operations, and prevent wall degradation through shear stresses, bending and impact loads. Furthermore, whilst some reactions require a catalyst layer of greater thickness (> 100 µm) to achieve a desired level of yield and selectivity, the thickness of the wall catalyst must be kept uniform to prevent negative effects on flow distribution. Thus, optimum porosity can also greatly influence the catalyst mass per unit area; otherwise it can lead to cracks and thus limit the mass transfer. Based on all of this knowledge, new catalyst materials are being developed and tested in microstructured reactors to produce high levels of activity and selectivity with better stability.

5.5.2

Hydrogen production in microreactors

Hydrogen production in microreactors is studied extensively using a range of different fuels and processing technologies. Most of the studies focus on catalytic SR, as this technique offers high efficiency, commercial experience and a high hydrogen/CO ratio (Holladay et al., 2009). Generally, SR reactions are endothermic, fast and equilibrium-limited. They are based on a fuel (a hydrocarbon or alcohol) reacting with steam to give hydrogen and CO2 as major products, whilst forming CO, CH4 and coke in addition. Two key points to note are that fast heat and mass transfer rates are needed to drive the endothermic SR reaction, and that SR reactions are fast with short residence times. These two points, in combination with other motivating factors, make microreactor utilization in SR reactions beneficial. Around 70–80% of hydrogen is produced by SR from natural gas, using Ni as the catalyst in conventional macro-scale reformer systems (Pattekar and Kothare, 2004). This process is highly energy intensive and is not environmentally friendly, due to the high air emissions produced. Moreover, hydrogen is now not only used in the production of ammonia, fertilizers and in oil refineries, but also as fuel for fuel cell devices. Hydrogen can be used as a fuel in all types of FCs, especially proton exchange membrane fuel cell (PEMFC) and alkaline fuel cell (AFC) energy systems which can be used in stationary and portable electronic applications. For the production of hydrogen on-site and on-demand, it can be a challenging task to build a compact, highly efficient system, but, as previously discussed, the compactness and light weight of microreactors makes them very promising for use as on-board fuel processing micro-devices. Exploring the use of biofuels like bio-ethanol, bio-methanol, bio-butanol and bio-

© Woodhead Publishing Limited, 2013

Microreactors and membrane microreactors

215

glycerol as raw materials for hydrogen production could be advantageous when operating at low temperatures with closed carbon cycles (Seelam et al., 2012). Studies of hydrogen production via SR of alcohols and hydrocarbon based fuels in microreactors are summarized in Table 5.2. These reactions can also be beneficial in MMRs. In microreactors, hydrogen is produced via SR, POx, WGS reaction, autothermal reforming (ATR) and other processing technologies. Delsman et al. (2005) studied the comparison between the conventional fixed-bed system and MRT for portable hydrogen production, with a complete methanol steam reformer system designed with a reformerburner (RB), which coupled endo- and exo-thermic reactions and a POx reactor with a heat exchanger for power output of 100 W and 5 kW. It was revealed that, for both levels of power, structured microreactors perform better than conventional reactors in the same conditions.

5.6

An overview of membrane microreactors

Integration of sub-micron thick membranes into devices with dimensions in the sub-millimetre range has the potential to produce highly efficient results, for example in high-purity hydrogen production. The MM can act as a permselective barrier, or be used to facilitate the addition of specific molecules to enhance microreactor performance. Moreover, the high pressures that are critical for conventional membrane systems can be operated safely in MMRs. The concept of lab-on-chip has created a range of new opportunities and has already been successfully applied to many fields, including catalysis, analytical chemistry and integration of sensors and micro-separation units. Whilst the integration of membrane functionality into microchemical systems (micro-devices) offers great benefits, designing this kind of micro unit can be challenging. In MMRs, three main units are integrated: the membrane separation, reaction zone and reactor channel units. In the previous section, the advantages, disadvantages and role in hydrogen production of microreactors have already been discussed. Table 5.3 summarizes microreactors for hydrogen production, which consist of a reformate stream of gases with other by-products, such as CO, CO2 and CH4. This stream cannot, however, be fed directly to a PEMFC as even very small amounts of CO (i.e., > 10 ppm) can damage the anode electrode of PEMFC, and the inhibiting effects of by-products lower the overall performance of the FC system. Microstructured membrane systems can be elegant, efficient alternatives to conventional catalytic CO clean up or WGS reaction systems. In order to produce on-site streams of pure or COx free hydrogen for a micro-fuel cell processor, a thin Pd-based membrane is introduced after the micro-reforming section, or the membrane is attached to the reforming catalyst in order to selectively separate hydrogen. Similar catalytic MM devices for high-purity hydrogen generation were designed by Wilhite et al.

© Woodhead Publishing Limited, 2013

Table 5.3 Hydrogen production in microreactors with various dimensions and operating temperatures

© Woodhead Publishing Limited, 2013

Microreactor (its material and shape)

Dimensions

Reaction

Catalyst-coated

T (oC)

Reference

Autocatalytic Microchannel, micro-combustion PCFSF supported Stainless steel Tree shaped serpentine flow Si wafer Rectangular microchannel

D = 3 mm w = 700 µm, D = 1.5 mm, tw = 0.3 mm, tc = 10–100 µm, L = 20 mm L = 10–20 mm w = 1 mm, h = 20 mm, L = 50 mm w = 2 mm, h = 1 mm, L = 110 mm

MeSR MSR

15% NiO/Al2O3 Cu/ZnO/Al2O3 and Pt

600–840 230

Levent et al., 2003 Tadbir and Akbari, 2011

MSR ESR MSR

Cu/Zn/Al/Zr Cobalt talc CuO/ZnO/Al2O3

240–360 350 230

Zhou et al., 2009 Domínguez et al., 2011 Chen et al., 2011

ESR

Rh/CeO2

350–650

Gorke et al., 2009

ESR

Co/ZnO

200–500

Casanovas et al., 2008

ESR

CO3O4

400

Casanovas et al., 2009

ESR

Rh/CeO2/Al2O3

400–600

Peela et al., 2011

MeATR

Ni/Al2O3

600

Akbari et al., 2011

MSR

Cu/ZnO

100–300

Kim and Kwon, 2006

ESR

Ni/Al2O3

500–700

Wang et al., 2008

w = 200 µm, h = 200 µm, L = 80 mm Stainless steel w = 700 µm, d = 350 µm, tw = 750 µm, L = 78 mm Silicon micromonoliths w = 0.9 mm, D = 3–4 µm, d = 350 µm, L = 210 µm Micromachined and w = 500 µm, d = 400 µm, L = 60 welded mm Rectangular microchannel w = 340 µm, h =340 µm, L = 8.5 mm Glass rectangular w = 0.5 mm, h = 1 mm, L = 2 cm microchannel Ceramic rectangle w = 0.3 mm, h = 0.4 mm, tc = 0.3 µm, L = 20 mm microchannel

w = width; d = depth of the channel; D = diameter of the microchannel; L = length of the microchannel; h = height; tc = catalyst thickness; tw = wall thickness; MeSR = methane steam reforming; MSR = methanol steam reforming; ESR = ethanol steam reforming; MeATR = methane autothermal reforming.

Microreactors and membrane microreactors

217

(2006) for portable power applications. The introduction of membrane phenomena in the sub-millimetre ranges, using mass transport controlled by a pressure-driven process in channels, flow and reaction domain, is also of interest. For example, hydrogen production in a tubular membrane reactor module was studied extensively using thick Pd-based membranes (Basile et al., 2011; Seelam et al., 2012) and reported in many publications. However, there have not yet been many studies focussed on membranes in microreactors for hydrogen production.

5.6.1

Methods and approaches

Jong et al. (2006) have reported different approaches and methods for integration of the membrane into microfluidic devices. These approaches are summarized as follows: (a) Direct incorporation of commercial membranes: achieved by clamping or gluing flat commercial membranes, followed by functionalisation. The membrane is introduced during micro-stereo lithography (ML) and HF membranes, rather than flat, and are used between capillaries. Sealing of the membrane with the micro-device is the main problem, particularly in the use of inorganic substrates such as glass or silicon with polymeric membranes. Furthermore glue can fill the pores of the membrane − a problem that can be solved by use of the ML technique. This approach is simple to apply to processes, and a wide choice of membrane materials and morphologies exist Jong et al. (2006). (b) Membrane preparation as part of the chip fabrication process: for many applications, this is the most appropriate method. Moreover, many authors have worked on this approach, developing methods to introduce membrane into microfluidic devices for hydrogen separation and purification using SR, WGS and KCR reactions (Chau et al., 2003; Deshpande et al., 2010; Karnik et al., 2003; Tong et al., 2003, 2005a; Wilhite et al., 2006; Zhang et al., 2006). Microfabrication techniques such as etching for well-defined pores, growing zeolite crystals on Si substrate, SiO2, porous Si, alumina or molecular-sieve materials, thin metallic film deposition on a SiO2 wafer, preparing polymeric membranes by casting and creating pores by ion track technology are just some of the approaches developed for this kind of method Wan et al. (2001). The adhesive and mechanical strength between the support structure (e.g., Silicon) and the membrane (e.g., thin Pd metal) during preparation and operation are the most important elements of this method. (c) In-situ preparation of membranes: this method is based on the fabrication of micro membranes in-situ with microfluidic channels or devices.

© Woodhead Publishing Limited, 2013

218

Handbook of membrane reactors

One example of this method is the production of liquid membranes inside microfluidic channels, where membranes are formed between the pillars by laser induced phase separation of the acrylate monomer. This method is very complex and can only be applied to limited materials. (d) The final method is using the membrane itself as chip material: this simple method involves fabrication of a zeolite membrane into microchannels for the MMR devices, and is very promising as it offers many advantages over the other methods. No additional materials are required in this method, for example. Jong et al. (2006) studied the PDMS materials due to their high gas permeability, and exploited this characteristic in microfluidic devices. This kind of method is applied to polymeric chips, hydrogel based chips and also fabrication of completely porous chips capable of utilizing membrane as chip materials. Some key features must be considered in selecting the most desirable membrane for microfluidic devices. The chemical, mechanical, thermal and surface properties of membrane materials, as well as their compatibility with the chip, and reaction to fouling are important, as are the selectivity, porosity, morphology and geometry (flat, HF, tubular etc.) of the membrane type. Feasibility of the fabrication method is another key feature and selection of the method and the approach, from the four previously discussed approaches, is the main application-dependent choice. Finally, the availability and functionality of operating conditions such as pressure, temperature, chemicals and reagents must be considered. In combination, these features are very important for the successful design of MMR devices (Jong et al., 2006; McMullen et al., 2010; Jensen et al., 2001).

5.6.2

MMR applications

Currently listed reactions performed in MMRs are reported in Table 5.4. Most of the studies are devoted to hydrogen separation, using hydrogenextractor and hydrogen-distributor MMR devices. A microfabricated Pd–Ag MMR integrated device, using reforming catalysts to produce highpurity hydrogen for portable micro-fuel cell applications, has been studied by Wilhite et al. (2006) (Fig. 5.5). In this study, hydrogen generation on the catalyst-coated wall layer and purification by Pd–Ag MM were integrated into one compact unit, improving the overall thermal and reactor performance. This was achieved by MF of a microfluidic device, followed by the deposition of a hydrogen permselective and 0.2 µm thick Pd–Ag MM on the microchannels, with a final coating by a catalyst suspension (prepared by co-precipitation method, milling, suspension in MeOH and aluminium oxide binder addition). The coating solution was added uniformly drop wise on the membrane surface using a syringe. In this case, the membrane and

© Woodhead Publishing Limited, 2013

Table 5.4 Reactions performed in membrane microreactors Membrane microreactor

Fabrication methods

Pd/PS

© Woodhead Publishing Limited, 2013

Etching, BOE, DRIE, SD Pd–Ag23%/Si oxide Evaporation, sputtering, spin-casting, or electrodeposition Pd–MMR – ZSM-5/NaA/PSS

NaY/NaX/NaA/ PSS Pd–Ag multichannelled

Pd/silicon

Fabricated microchannel on porous stainless steel plate Electro-discharge micromachining Structured catalytic filament bed membrane microreactor Sputtering, wet etching, patterning, plasma etch

Thickness Reaction (µm)

Temperature (°C)

Catalyst

Membrane function

References

0.34

250

–

H2 distributor

Ye et al., 2005

0.2

1-butene hydrogenation POM

400

LaNi0.95Co0.05O3

H2 separation

Wilhite et al., 2006

0.2

MeSR

887

Ni/MgAl2O4

H2 separation

30

KCR

100

ZSM-5

Water removal

Goto et al., 2003 Yeung et al., 2005

6

KCR

100

Water removal

–

PDH

550

Cs-exchanged NaA 0.5 wt%Pt/1%Sn/ Si–Al

0.2

WGS

200

Cu/Zn/Al2O3

H2 permeation

H2 permeation

Yeung et al., 2005 Wolfrath et al., 2001

Karnik et al., 2003 (Continued)

Table 5.4 Continued

© Woodhead Publishing Limited, 2013

Membrane microreactor

Fabrication methods

Thickness Reaction (µm)

Temperature (°C)

Catalyst

Membrane function

References

Pd/silicon wafer/ PDMS

4

Dehydrogenation of cyclohexane

200

Pt

H2 permeation

Cui et al., 2000

Pd/alumina

Wet and dry micromachining, PDMS moulding, bonding, etching, sputtered –

1

300

–

H2 permeation

Dense polymeric

–

–

–

–

EO separation

ZSM-5/PSS

ZSM-5 membrane microchannel on PSS plate Fabrication of porous metallic HFs, CNF on PSS, PDMS coating, ELP

25

Dehydrogenation of cyclohexane Ethylene epoxidation Oxidation of aniline

70

Titanium silicalite-1

Water removal

Yamamoto et al., 2006 Schiewe et al., 2001 Wan et al., 2005

20

Catalytic reduction of nitrite (G-L)

RT

Pd

H2 distributor

Aran et al., 2011

6

WGS

200–500

30%CuO/CeO2

H2 separation

70

PDH

550

Pt/Sn/alumina

H2 extractor

Rahman et al., 2011 Kiwi-Minsker et al., 2002

Dense polymer PDMS/CNFs/ PSS Pd/Al2O3 HF Pd–Ag/ filamentous catalyst

Filaments structured catalytic packing

Microreactors and membrane microreactors (a)

Ar

CH3OH + O2

221

Ar, H2

H2 + CO + CO2 + CH3OH

Pt resistive heaters, 200 nm

Pd–Ag permselective film, 200 nm

Catalyst washcoat

Silicon nitride, 300 nm

Silicon substrate, 0.65 mm

Silicon oxide, 250 nm

(b)

(c)

5.5 Catalytic micromembrane device for high-purity hydrogen generation: (a) micro-device unit with side-view (left) and cross section (right), (b) completed micro-device prior to assembly and (c) catalyst washcoat of microchannel with LaNi0.95Co0.05O3/Al2O3 (Wilhite et al., 2006) (Copyright permission 2006 John Wiley and Sons).

the catalyst are the active phases in which reaction and separation happen simultaneously and instantly. Goto et al. (2003) have simulated the MMR for fuel cell applications using a methane feed, and have gone on to compare the efficiencies of the three FC system configurations; one with Pd–MMR followed by PEMFC, solid-oxide fuel cell and the proton conducting solidoxide fuel cell (SOFC). The simulation results show that Pd–MMR is the most effective system for power generation in comparison to the other two systems. In Takahashi et al. (2005) a suspended MEMS based micro-fuel reformer was designed and manufactured, and the performance of the reformer evaluated. In this study, in-situ chemical vapour deposition (CVD) of the alumina catalyst bed on a membrane was used as the preparation method for better mechanical and thermal isolation of the reaction zone on the membrane. Most of the microfabricated Pd-based MMs are much more efficient than the conventional thicker or large scale devices, as reported by many authors.

© Woodhead Publishing Limited, 2013

222

Handbook of membrane reactors

Water

Set-points from external supervisory predictive controller

Adaptive PI parameters

V1 (sp) I1

PI control V1

Adaptive PI parameters

V2 (sp)

Si substrate

PI control I2

V2

Non-permeate from lower microchannel T2

T1

Permeate hydrogen from upper microchannel

Heaters and temp. sensors Methanol Mixer/vaporizer

Catalytic reformer Palladium membrane microreactor Heaters and sensors on an insulator Pd

Resistor 1 µm 500 µm

Cu Cu

Pyrex 1000 µm

5.6 A microreactor system with integrated Pd membrane with heaters and sensors (Karnik et al., 2003) (Copyright permission 2003 IEEE). (In the figure, V1 (sp) and V2 (sp) refer to voltage regulation signal set-point 1 and 2, from and to the PI controllers. T1 and T2 refer to temperature sensors to the heaters and I1 and I2 refer to current to the resistive heaters from the PI controllers.)

In the first experiment of its kind, a Pd-based MMR for hydrogen production via SR of methanol was designed, fabricated and tested in a study conducted by Karnik et al. (2003). The MMR consists of four main components: a mixer/vaporizer for methanol and water, a catalytic steam micro-reformer, a WGS reactor with Cu and integrated Pd MM, and integrated resistors and sensors (shown in Fig. 5.6). This Pd–MMR is utilized to produce pure or CO-free hydrogen for a micro-fuel cell processor. The complete MMR was built on a silicon substrate using MEMS MF techniques. The Pd–MMR device consists of an integrated WGS reactor and a Pd MM (hydrogen gas separator) in one compact unit (as shown in Fig. 5.6) which is constructed from a Cu−SOG−Al−Pd layer (spin-of-glass (SOG), Cu as a catalyst). Cu, SOG and aluminium provide structural support to the MM Pd. In another study, a MEMS based thin Pd–MMR with 0.34 µm Pd thickness was fabricated on an oxidized PS support. The MMR system was characterized by permeation experiments with hydrogen, N2 and He and also carried out the 1-butene hydrogenation in the temperature range of around 200–250oC. The

© Woodhead Publishing Limited, 2013

Microreactors and membrane microreactors Liquid

223

Liquid

Gas

Porous wall Dense polymer layer

Gas Liquid

Gas

5.7 Scheme for membrane microreactors for multiphase reactions, that is, gas−liquid reactions (left), a PSS supported dense PDMS gaspermeable membrane with CNFs as a catalyst support (right). (Aran et al., 2011) (Copyright permission 2011 Elsevier).

hydrogen flux was reported to be 0.112 mol m−2 s−1, with a partial pressure difference of 110 kPa (Ye et al., 2006). A new hybrid catalytic MMR has been designed and fabricated for gas−liquid−solid (G−L−S) reactions (Aran et al., 2011). In this MMR, carbon nanofibres (CNF) are grown as catalyst supports on the PSS substrate, and Pd catalyst are immobilized onto the support. The outer part is coated with an encapsulated hydrogen gas-permeable PDMS membrane, as shown in Fig. 5.7. This complete catalytic MMR system is used to study the G−L−S reaction, such as nitrite (NO−2 ions) reduction in water using a PSS/Pd–CNF/ PDMS reactor. Due to intrinsic reductive properties of CNFs on PSS and catalytic activity of Pd–CNF on PSS, the porous catalytic MMR is highly active and thus very promising in relation to multiphase reactions in microreactors. A new integrated ceramic membrane micro-network constructed from 8 µm thick defect-free Pd films was introduced into the alumina coated channels (Kim et al., 2009). This micromembrane system consists of three channels: SR, WSG and permeate sides. As proposed, this thermally integrated MMR system can be utilized as a portable reformer to produce highpurity hydrogen for PEMFC. The use of miniature zeolite membranes inside a microchannel reactor allows the removal of the water by-product from the aniline oxidation reaction mixture, thus prolonging the life of the TS-1 catalyst. Zeolite MMRs were studied extensively by the Yeung’s research group on KCR (Lau et al., 2003, 2007; Wan et al., 2001; Yeung et al., 2005) and, as KCR is an equilibrium or thermodynamically limited reaction, the results were particularly interesting. In this case free-standing ZSM-5 MMs fabricated on the Si chip were used in the KCR reaction, along with a zeolite (NaX) catalyst. In KCR, the water by-product is selectively removed from the condensation reaction by the MM, thus a supra-equilibrium conversion was achieved with high product purity, shifting the chemical equilibrium towards the desired product side.

© Woodhead Publishing Limited, 2013

224

Handbook of membrane reactors

100%

Product yield

80% Equilibrium conversion

60% 40% 20% 0% 0

1

2

3

4

5

Residence time (h)

5.8 Product yield as a function of residence time for a fixed-bed reactor (triangles), a multi-channel microreactor (circles) and a multi-channel membrane microreactor (squares) (Lai et al., 2003) (Copyright permission 2006 Royal Society of Chemistry).

As shown in Fig. 5.8 a higher product yield was achieved using a multichannel MMR in comparison to FBR and a microreactor (Lai et al., 2003). Lau et al. (2007) have reported on a zeolite MMR for KCR, offering two different design approaches. In the first approach, a thin Cs-exchanged NaX catalyst powder was coated on the wall of the EDM-fabricated microchannel (thirty-five straight channels with 300 µm wide, 600 µm deep and 25 mm long). The NaA membrane was grown on the back side on the microchannel plate by pre-seeding with 150 nm NaA zeolite nanocrystals and seeds were attached using mercapto-3-propyltrimethoxysilane, before being assembled on the PSS microchannel plates. In the second design approach, a hybrid NaA membrane with a Cs-NaX catalyst film was deposited on the microchannel plate, and NaA nanocrystal membrane was grown by seeding the microchannels with a thin layer. The procedure was repeated three times until a membrane thickness of approximately 6 µm was achieved, and the faujasite X zeolite layer was then deposited on top of the membrane, before the wall was finally coated with a Cs-NaX catalyst. The KCR between benzaldehyde and ethyl cyanoacetate, ethyl acetoacetate and diethyl malonate were conducted in the zeolite MMR using NaX as the membrane and Cs−NaA as the catalyst. Zeolite MMR was found to function more efficiently than a microreactor without a membrane, due to the fact that water formed during the condensation reaction was continuously removed by the membrane, thus shifting the equilibrium towards the desired products (Lau et al., 2003, 2007).

© Woodhead Publishing Limited, 2013

Microreactors and membrane microreactors

5.7

225

Conclusions and future trends

By selective removal of a product or addition of a reactant through a membrane, chemical equilibrium can be controlled. Integration of the reaction and separation in one unit offers great advantages, including improved efficiency over conventional systems, and MMR devices are multifunctional, novel approach to achieve such integration in reaction engineering. In order to enhance the high surface-to-volume ratios and high heat and mass transfer rates in a microreactor, it is beneficial to include functional membranes capable of helping the reactor to achieve maximum efficiency. The high degree of parallelization in micro-devices makes them ideal candidates for high throughput screening and testing devices for membrane based processes and, as such, membrane technology is foreseen to have a bright future in micro and/or chemical process technology. Enhancing this view, membranes are already being successfully exploited and applied in many areas, including analytical chemistry and gas separation, and MF techniques already exist in the semiconductor and MEMS based industries. It therefore seems greatly advantageous to build membrane based micro-devices, with decreasing costs and existing commercial experience. MF technology will be critical to the reproduction of membrane micro-devices, as well as being crucial in enabling precise control of pore sizes, depositing thin membrane separating layers and manufacturing integrated systems for application in, for example, portable electronic applications. However, much research is still needed into the application of functioning membranes inside microfluidic devices. Much research has been conducted on microstructured devices for reactions, analytical devices, catalyst screening and organic synthesis. However, methods for depositing a very thin membrane layer into the MSs have not yet been widely discussed in the scientific literature and only a few areas, such as palladium-based and zeolite MMs for microfluidic devices, have been explored. In the future, new membrane and catalyst materials will be developed and tested in microreactors, and the effect of miniaturization, use of membranes and optimization will be thoroughly investigated. It can be forecast that the application of membrane technology in microfluidic devices will become increasingly important in gas separations, volatile organic compound removal, pervaporation, emulsification, gas−liquid contactors and energy devices, especially for niche markets. Another important aspect for investigation is the catalyst, which plays an important role in catalytic MMRs. In this case, the membrane functions only to separate, and does not participate in the chemical reaction. A thorough understanding of the process for introducing an appropriate quantity of a catalyst into the MSs will be crucial in ensuring sufficient mechanical strength, adhesion of the catalyst to the walls and prevention of ageing

© Woodhead Publishing Limited, 2013

226

Handbook of membrane reactors

phenomena. Therefore, the synergic effects of a catalyst and a membrane should be studied carefully in order to assess their robustness, selectivity and activity before they are introduced into micro-devices. Moreover, understanding the manner in which catalysts affect the membrane function will be very important. Dense Pd-based membrane reactors have been widely studied in the conventional tubular and fixed-bed reactors with thick membranes. However, only the permeation characteristics have been studied using microfabricated micro-devices, meaning research is needed into the reaction and separation characteristics in MMRs. Zeolites are very promising materials, with vast potential for use as membrane, catalyst or substrate materials for microreactors in MMR applications.

5.8

References

Akbari M H, Ardakani A H S and Tadbir M A (2011), ‘A microreactor modeling, analysis and optimization for methane autothermal reforming in fuel cell applications’, Chem Eng J, 166, 1116–1125. Alfadhel K A and Kothare M V (2005), ‘Microfluidic modeling and simulation of flow in membrane microreactors’, Chem Eng Sci, 60, 2911–2926. Anastas P T and Warner J C (1998), ‘Green chemistry: Theory and practice’, New York, Oxford University Press, 30. Aran H C, Benito S P, Luiten-Olieman M W J, Er S, Wessling M, Lefferts L, Benes N E and Lammertink R G H (2011), ‘Carbon nanofibers in catalytic membrane microreactors’, J Membrane Sci, 381, 244–250. Athayde A L, Baker R W and Nguyen P (1994), ‘Metal composite membranes for hydrogen separation’, J Membrane Sci, 94, 299–311. Aubin J, Ferrando M and Jiricny V (2010), ‘Current methods for characterising mixing and flow in microchannels’, Chem Eng Sci, 65, 2065–2093. Bryden K J and Ying J Y (1995), ‘Nanostructured palladium membrane synthesis by Magnetron sputtering’, Mater Sci Eng, A204, 140–145. Bryden K J and Ying J Y (2002), ‘Nanostructured palladium–iron membranes for hydrogen separation and membrane hydrogenation reactions’, J Membrane Sci, 203, 29–42. Basile A, Pinacci P, Iulianelli A, Broglia M, Drago F, Liguori S, Longo T and Calabrò V (2011), ‘Ethanol steam reforming reaction in a porous stainless steel supported palladium membrane reactor’, Int J Hydrogen Energy, 36, 2029–2037. Casanovas A, Domínguez M, Ledesma C, Lopez E, Llorca J (2009), ‘Catalytic walls and micro-devices for generating hydrogen by low temperature steam reforming of ethanol’, Catal Today, 143, 32–37. Casanovas A, Saint-Gerons M, Griffon F and Llorca J (2008), ‘Autothermal generation of hydrogen from ethanol in a microreactor’, Int J Hydrogen Energy, 33, 1827–1833. Chau J L H, Leung A Y L and Yeung K L (2003), ‘Zeolite micromembranes’, Lab Chip, 3, 53–55. Chen C and Gobina E (2010), ‘Ultra-thin palladium technologies enable future commercial deployment of PEM fuel cell systems’, Membrane Technol., 3, 6–13. Chen Y, Zhang C, Wu R and Shi M (2011), ‘Methanol steam reforming in with constructal tree-shaped network’, J Power Sources, 196, 6366–6373.

© Woodhead Publishing Limited, 2013

Microreactors and membrane microreactors

227

Chi Y H, Yen P S, Jeng M S, Ko S T and Lee T C (2010), ‘ Preparation of thin Pd membrane on porous stainless steel tubes modified by a two-step method’, Int J Hydrogen Energ, 35, 6303–6310. Coleman J W and Garimella S (1999), ‘Characterization of two-phase flow patterns in small diameter round and rectangular tubes’, Int J Heat Mass Transfer, 42, 2869–2881. Colin R (1999), ‘Process Intensification and Green Chemistry’, Green Chem, 1, G15-G17. Collins J P and Way J D (1993), ‘Preparation and characterization of a composite palladium-ceramic membrane’, Ind Eng Chem, 32, 3006–3013. Coronas J and Santamaria J (2004), ‘The use of zeolite films in small-scale and micro-scale applications’, Chem Eng Sci, 59, 4879–4885. Cui T, Fang J, Zheng A, Jones F and Reppond A (2000), ‘Fabrication of microreactors for dehydrogenation of cyclohexane to benzene’, Sensor Actuator B Chem, 71, 228–231. Cybulski A and Moulijn J A (2006), ‘Structured catalysts and reactors’, 2nd ed., CRC Press. Delsman E R, Laarhoven B J P F, De Croon M H J M, Kramer G J and Schouten J C (2005), ‘Comparison between conventional fixed-bed and microreactor technology for a portable hydrogen production case’, Chem Eng Res Des, 83, 1063–1075. Deshpande K, Meldon J H, Schmidt M A and Jensen K F (2010), ‘SOI-supported microdevice for hydrogen purification using palladium–silver membranes’, J Microelectromech S, 19, 402–409. Dittmeyer R, Höllein V and Daub K (2001), ‘Membrane reactors for hydrogenation and dehydrogenation processes based on supported palladium’, J Mol Catal, 173, 135–184. Doku G N, Verboom W, Reinhoudt D N and van den Berg V (2005), ‘On-microchip multiphase chemistry—a review of microreactor design principles and reagent contacting modes’, Tetrahedron, 61, 2733–2742. Domínguez M, Cristiano G, López E and Llorca J (2011), ‘Ethanol steam reforming over cobalt talc in a plate microreactor’, Chem Eng J, 176–177, 280–285. Ehrfeld W, Hessel V and Löwe H (2000), Microreactors – New technology for modern chemistry, Wiley-VCH, Weinheim. Franz A, Jensen K F and Schmidt M A (2000), ‘Palladium based micromembranes for hydrogen separation and hydrogenation/dehydrogenation reactions’, Ed. Ehrfeld W, Microreaction technology: industrial prospects, 267–276. Gavriilidis A, Angeli P, Cao E, Yeong K K and Wan Y S S (2002), ‘Technology and applications of microengineered reactors’, Trans IchemE, part A, 80. Gielens F C, Knibbeler R J J, Duysinx P F J, Tong H D, Vorstman M A G and Keurentjes J T F (2006), ‘Influence of steam and carbon dioxide on the hydrogen flux through thin Pd/Ag and Pd membranes’, J Membrane Sci, 279, 176–185. Gielens F C, Tong H D, van Rijn C J M, Vorstman M A G and Keurentjes J T F (2002), ‘High-flux palladium-silver alloy membranes fabricated by microsystem technology’, Desalination, 147, 417–423. Gielens F C, Tong H D, van Rijn C J M, Vorstman M A G and Keurentjes J T F (2004), ‘Microstystem technology for high-flux hydrogen separation membranes’, J Membrane Sci, 243, 203–213. Gokhale S V, Tayal R K, Jayaraman V K and Kulkarni B D (2005), ‘Microchannel reactors: Applications and use in process development’, Int J Chem React Eng, 3, R2.

© Woodhead Publishing Limited, 2013

228

Handbook of membrane reactors

Goltsov V A (1977), ‘Hydrogen in metals, Questions of atomic sciences and engineering’, Ser.: Atomic-Hydrogen Energetics, 1 (2), 65. Gorke O, Pfeifer P and Schubert K (2009), ‘Kinetic study of ethanol reforming in a microreactor’, Appl Catal A Gen, 360, 232–241. Goto S, Tagawa T, Assabumrungrat S and Praserthdam P (2003), ‘Simulation of membrane microreactor for fuel cell with methane feed’, Catal Today, 82, 223–232. Hessel V, Knobloch C and Löwe H (2008), ‘Review on patents in microreactor and micro process engineering’, Recent Patents on Chemical Engineering, 1, 1–16. Hessel V, Hardt S and Löwe H (2005a), Chemical micro process engineering: Fundamentals, modelling and reactions, Wiley-VCH Verlag GmbH, 1–106 Hessel V, Hardt S and Löwe H (2005b), Chemical micro process engineering: Fundamentals, modelling and reactions, Wiley-VCH Verlag GmbH, 125–147. Hessel V, Löwe H, Müller A and Kolb G (2005c), Chemical micro process engineering processing and plants, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, 386–399. Holladay J D, Hu J, King D L and Wang Y (2009), ‘An overview of hydrogen production technologies’, Catal Today, 139, 244−260. Holladay J D, Wang Y and Jones E (2004), ‘Review of developments in portable hydrogen production using microreactor technology’, Chem Rev, 104, 4767–4790. Hou K and Hughes R (2003), ‘Preparation of thin and highly stable Pd/Ag composite membranes and simulative analysis of transfer resistance for hydrogen separation’, J Membrane Sci, 214, 43–55. Israni S H, Nair B K R and Harold M P (2009), ‘Hydrogen generation and purification in a composite Pd hollow fiber membrane reactor: Experiments and modeling’, Catal Today, 139, 299–311. Jayaraman V and Lin Y S (1995), ‘Synthesis and hydrogen permeation properties of ultrathin, palladium-silver alloy membranes’, J Membrane Sci, 104, 251–262. Jensen K F (1999), ‘Microchemical systems: Status, challenges, and opportunities’, 45, 2051–2054. Jensen K F (2001), ‘Microreaction engineering is small better?’, Chem Eng Sci, 56, 293–303. Jong J D, Lammertink R G H and Wessling M (2006), ‘Membranes and microfluidics: a review’, Lab Chip, 6, 1125–1139. Karnik S V, Hatalis M K and Kothare M V (2003), ‘Towards a palladium micro-membrane for the water gas shift reaction: Microfabrication approach and hydrogen purification results’, J Microelectromech S, 12, 93–100. Keurentjes J T F, Gielens F C, Tong H D, Rijn C J M and Vorstman M A G (2004), ‘High-flux palladium membranes based on microsystem technology’, Ind Eng Chem Res, 43, 4768–4772. Kim T and Kwon S (2006), ‘Catalyst preparation for fabrication of a MEMS fuel reformer’, Chem Eng J, 123, 93–102. Kim T W (2008), Studies of transport phenomena in hydrotalcite membranes and their use in direct methanol fuel cells, PhD thesis, Faculty of the Graduate School University of Southern California. Kim D, Kellogg A, Livaich E and Wilhite B A (2009), ‘Towards an integrated ceramic micro-membrane network: Electroless-plated, palladium membranes in cordierite supports’, J Membrane Sci, 340, 109–116. Kiwi-Minsker L and Renken A (2005), ‘Microstructured reactors for catalytic reactions’, Catal Today, 110, 2–14.

© Woodhead Publishing Limited, 2013

Microreactors and membrane microreactors

229

Kiwi-Minsker L, Wolfrath O and Renken A (2002), ‘Membrane reactor microstructured by filamentous catalyst’, Chem Eng Sci, 57, 4947–4953. Klemm E, Döring H, Geisselmann A and Schirrmeister S (2007), ‘Microstructured reactors in heterogeneous catalysis’, Chem Eng Technol, 30, 1615–1621. Kolb G and Hessel V (2004), ‘Micro-structured reactors for gas phase reactions’, Chem Eng J, 98, 1–38. Kwan S M, Leung A Y L and Yeung K L (2010), ‘Gas permeation and separation in ZSM-5 micromembranes’, Sep Purif Technol, 73, 44–50. Lai S M, Martin-Aranda R and Yeung K L (2003), ‘Knoevenagel condensation reaction in a membrane microreactor’, Chem Commun, 218–219. Lau W N, Yeung K L, Zhang X F and Martin-Arand R (2007), ‘Zeolite membrane microreactors and their performance’, Stud Surf Sci Catal, 170, 1460–1465. Levent M, Gunn D J and El-Bousiffi M A (2003) ‘Production of hydrogen-rich gases from steam reforming of methane in an automatic catalytic microreactor’, Int J Hydrogen Energy, 28, 945–959. Leung Y L A and Yeung K L (2004), ‘Microfabricated ZSM-5 zeolite micromembranes’, Chem Eng Sci, 59, 4809–4817. Li A, Grace J R and Lim C J (2007), ‘Preparation of thin Pd-based composite membrane on planar metallic substrate Part II. Preparation of membranes by electroless plating and characterization’, J Membrane Sci, 306, 159–165. Li A, Liang W and Hughes R (1998), ‘Characterization and permeation of palladium/stainless steel composite membrane’, J Membrane Sci, 149, 259–268. Li Z Y, Maeda H, Kusakabe K, Marooka S, Anzai H and Akiyama S (1993), ‘Preparation of palladium-silver alloy membranes for hydrogen separation by the spray pyrolysis method’, J Membrane Sci, 78, 247–254. McCool B, Xomeritakis G and Lin Y S (1999), ‘Composition control and hydrogen permeation characteristics of sputter deposited palladium-silver membranes’, J Membrane Sci, 161, 67–76. McLeary E E, Jansen J C and Kapteijn F (2006), ‘Zeolite based films, membranes and membrane reactors: Progress and prospects’, Micropor Mesopor Mat, 90, 198–220. McLeod L S, Degertekin F L and Fedorov A G (2009), ‘Determination of the rate-limiting mechanism for permeation of hydrogen through microfabricated palladium–silver alloy membranes’, J Membrane Sci, 341, 225–232. McMullen J P and Jensen K F (2010), ‘Integrated microreactors for reaction automation: New approaches to reaction development’, Annu Rev Anal Chem, 3, 19–42. Mills P L, Quiram D J and Ryley J F (2007), ‘Microreactor technology and process miniaturization for catalytic reactions − A perspective on recent developments and emerging technologies’, Chem Eng Sci, 62, 6992−7010. Moulijn J A, Stankiewicz A, Grievink J and Gorak A (2006), ‘Process intensification and process system engineering: a friendly symbiosis’, Computer Aided Chem Eng, 21, 29–37. Naddaf A, Bart H J, Krätz L, Detemple P, Schmitt S, Hessel V and Faqir N (2009), ‘Microfabricated hydrogen sensitive membranes’, Chem Eng Technol, 32,103–113. Nair B K R and Harold M P (2007), ‘Pd encapsulated and nanopore hollow fiber membranes: Synthesis and permeation studies’, J Membrane Sci, 290, 182–195.

© Woodhead Publishing Limited, 2013

230

Handbook of membrane reactors

O’Brien J, Hughes R, and Hisek J (2001), ‘Pd-Ag membranes on porous alumina substrates by unbalanced magnetron sputtering’, Surf Coat Tech, 142, 253–259. Papautsky I, Ameel T and Frazier A B (2001), ‘A review of laminar single-phase flow in microchannels’, Proceeding of ASME International Mechanical Engineering Conference and Exposition, New York, Nov. 21–26. Pattekar A V and Kothare M V (2004), ‘A microreactor for hydrogen production in micro fuel cell applications’, J Microelectromech S, 13, 7–18. Peela N R, Mubayi A and Kunzru D (2011), ‘Steam reforming of ethanol over Rh/ CeO2/Al2O3 catalysts in a microchannel reactor’, Chem Eng J, 167, 578–587. Peters T A, Stange M, Klette H and Bredesen R (2008), ‘High pressure performance of thin Pd–23%Ag/stainless steel composite membranes in water gas shift gas mixtures; influence of dilution, mass transfer and surface effects on the hydrogen flux’, J Membrane Sci, 316, 119–127. Peters T A, Stange M and Bredesen R (2011), ‘On the high pressure performance of thin supported Pd–23%Ag membranes—Evidence of ultrahigh hydrogen flux after air treatment’, J Membrane Sci, 378, 28–34. Pfahler J, Harley J, Bau H and Zemel J N (1991), ‘Gas and liquid flow in small channels’, Micromechanical Sensors, Actuators and Systems, ASME Winter Annual Meeting, Atlanta, GA, 32, 49–59. Pina M P, Mallada R, Arruebo M, Urbiztondo M, Navascués N, de la Iglesia O and Santamaria J (2011), ‘Zeolite films and membranes: Emerging applications’, Micropor Mesopor Mat, 144, 19–27. Plasynski S, Lang D and Tsapatsis M (2008), http://www.netl.doe.gov/publications/ factsheets/project/Proj306.pdf (accessed on Jan 2012). Rahman M A, García-García F R, Irfan Hatim M D, Kingsbury B F K and Li K (2011), ‘Development of a catalytic hollow fibre membrane micro-reactor for high purity H2 production, J Membrane Sci, 368,116–123. Ratchananusorn W (2007), ‘Flow phenomena, heat and mass transfer in microchannel reactors’, MSc. Thesis, Faculty of Technology, The Degree Program of Chemical Technology, Lappeenranta University of Technology, Finland. Roa F and Way J D (2003), ‘Influence of alloy composition and membrane fabrication on the pressure dependence of the hydrogen flux of palladium–copper membranes’, Ind Eng Chem Res, 42, 5827–5835. Rong B G, Kolehmainen E and Turunen L (2008), ‘Methodology of conceptual process synthesis for process intensification’, Computer Aided Chem Eng, 25, 283–288. Sanders J P M, Clark J H, Harmsen G J, Heeres H J, Heijnen J J, Kersten S R A, van Swaaij W P M and Moulijn J A (2011), ‘Process intensification in the future production of base chemicals from biomass’, Chem Eng Process: Process Intensification, 51, 117–136. Schiewe B, Staudt-Bickel C, Vuin A and Wegner G (2001), ‘Membrane-based gas separation of ethylene/ethylene oxide mixtures for product enrichment in microreactor technology’, ChemPhysChem, 2, 211–218. Seelam P K, Liguori S, Iulianelli A, Pinacci P, Drago F, Calabrò V, Huuhtanen M, Keiski R, Piemonte V, Tosti S, De Falco M and Basile A (2012), ‘Hydrogen production from bio-ethanol steam reforming reaction in a Pd/PSS membrane reactor’, Catal Today, 193, 42–48.

© Woodhead Publishing Limited, 2013

Microreactors and membrane microreactors

231

Su C, Jin T, Kuraoka K, Matsumura Y and Yazawa T (2005), ‘Thin palladium film supported on SiO2-modified porous stainless steel for a high-hydrogen-flux membrane’, Ind Eng Chem Res, 44, 3053–3058. Sun G B, Hidajat K and Kawi S (2006), ‘Ultra-thin Pd membrane on α-Al2O3 hollow fiber by electroless plating: High permeance and selectivity’, J Membrane Sci, 284, 110–119. Tadbir M A and Akbari M H (2011), ‘Methanol steam reforming in a planar wash coated microreactor integrated with a micro-combustor’, Int J Hydrogen Energy, 36, 12822–12832. Takahashi T, Tanaka S, Chang K S and Esashi M (2005), ‘In-situ chemical vapor deposition of alumina catalyst bed on a suspended membrane microreactor’, Solid-State Sensors, Actuators and Microsystems, TRANSDUCERS ‘05. The 13th International Conference, 1, 899–903. Tong H D, Berenschot J W E, De Boer M J, Gardeniers J G E, Wensink H, Jansen H V, Nijdam W, Elwenspoek M C, Gielens F C and van Rijn C J M (2003), ‘Microfabrication of palladium–silver alloy membranes for hydrogen separation’, J Microelectromech S, 12, 622–629. Tong H D (2004), ‘Microfabricated palladium-based membranes for hydrogen separation’, PhD thesis, Transducers Science and Technology of the MESA+ Research Institute at the University of Twente, Enschede, Netherlands. Tong H D, Gielens F C, Gardeniers J G E, Jansen H V, Berenschot J W, de Boer M J, de Boer J H, van Rijn C J M and Elwenspoek M C (2005a), ‘Microsieve supporting palladium-silver alloy membrane and application to hydrogen separation’, J Microelectromech S, 14(1), 113–124. Tong J, Suda H, Haraya K and Matsumura Y (2005b), ‘A novel method for the preparation of thin dense Pd membrane on macroporous stainless steel tube filter’, J Membrane Sci, 260, 10–18. Tong J, Matsumura Y, Suda H and Haraya K (2005c), ‘Thin and dense Pd/CeO2/ MPSS composite membrane for hydrogen separation and steam reforming of methane’, Sep Purif Technol, 46, 1–10. Tong J and Matsumura Y (2006a), ‘Pure hydrogen production by methane steam reforming with hydrogen-permeable membrane reactor’, Catal Today, 111, 147–152. Tong J, Su L, Kashima Y, Shirai R, Suda H and Matsumura Y (2006b), ‘Simultaneously depositing Pd–Ag thin membrane on asymmetric porous stainless steel tube and application to produce hydrogen from steam reforming of methane’, Ind Eng Chem Res, 45, 648–655. Tonomura O, Noda M, Kano M and Hasebe S (2004a), ‘Optimal design approach for microreactors with uniform residence time distribution’, The 10th Asian Pacific Confederation of Chemical Engineering (APCChE), 4B-07, Kitakyushu, Japan, Oct. 17–21. Tonomura O, Noda M, Kano M and Hasebe S (2004b), ‘Thermo-fluid design approach to microreactors with uniform temperature and residence time distribution’, AIChE Annual Meeting 2004, 430d. Uemiya S, Kude Y, Asugino K, Sato N, Matsuda T and Kikuchi E (1988), ‘A palladium/porous glass composite membrane for hydrogen separation’, Chem Lett, 17, 1687–1690. Van Gerven T and Stankiewicz A (2009), ‘Structure, energy, synergy, time – The fundamentals of process intensification’, Ind Eng Chem Res, 48, 2465–2474.

© Woodhead Publishing Limited, 2013

232

Handbook of membrane reactors

Waelchli S and Rohr PRV (2006), ‘Two-phase flow characteristics in gas-liquid microreactors’, Int J Multiphas Flow, 32, 791–806. Watts P and Haswell S J (2005), ‘The application of micro reactors for organic synthesis’, Chem Soc Rev, 34, 235–246. Wan Y S S, Chau J L H, Gavriilidis A and Yeung K L (2001), ‘Design and fabrication of zeolite-based microreactors and membrane microseparators’, Micropor Mesopor Mater, 42, 157–175. Wang D, Tong J, Xu H and Matsumura Y (2004), ‘Preparation of palladium membrane over porous stainless steel tube modified with zirconium oxide’, Catal Today, 93–95, 689–693. Wang J, Liu G, Xiong Y, Huang X, Guo Y and Tian Y (2008), ‘Fabrication of ceramic microcomponents and microreactor for the steam reforming of ethanol’, Microsystem Technol, 14, 1245–1249. Wilhite B A, Schmidt M A and Jensen K F (2004), ‘Palladium-based micromembranes for hydrogen separation: Device performance and chemical stability’, Ind Eng Chem Res, 43, 7083–7091. Wilhite B A, Weiss S E, Ying J Y, Schmidt M A and Jensen K F (2006), ‘High-purity hydrogen generation in a microfabricated 23 wt% Ag–Pd membrane device integrated with 8:1 LaNi0.95Co0.05O3/Al2O3 catalyst’, Adv Mater, 18, 1701–1704. Wolfrath O, Kiwi-Minsker L and Renken A (2001), ‘Novel Membrane Reactor with Filamentous Catalytic Bed for Propane Dehydrogenation’, Ind Eng Chem Res, 40, 5234–5239. Xomeitakis G and Lin Y S (1996), ‘Fabrication of thin metallic membranes by MOCVD and sputtering’, J Membrane Sci, 133, 217–230. Yamamoto S, Hanaoka T, Hamakawa S, Sato K and Mizukami F (2006), ‘Application of a microchannel to catalytic dehydrogenation of cyclohexane on Pd membrane’, Catal Today, 118, 2–6. Yan S, Maeda H, Kusakabe K and Morooka S (1994), ‘Thin palladium membrane formed in support pores by metal-organic chemical vapor deposition method and application to hydrogen separation’, Ind Eng Chem Res, 33, 616–622. Ye S Y, Tanaka S, Esashi M, Hamakawa S, Hanaoka T and Mizukami F (2005), ‘Thin palladium membrane microreactor with porous silicon support and their application in hydrogenation reaction’, Solid-State Sensors, Actuators and Microsystems, 2005. Digest of Technical Papers. Transducers ‘05. The 13th International Conference, 2, 2078–2082. Ye S Y, Tanaka S, Esashi M, Hamakawa S, Hanaoka T and Mizukami F (2006), ‘MEMS-based thin palladium membrane microreactors’, Proc SPIE, 6032, 603207. Yepes D, Cornaglia L M, Irusta S and Lombardo E A (2006), ‘Different oxides used as diffusion barriers in composite hydrogen permeable membranes’, J Membrane Sci, 274, 92. Yeung K L, Zhang X, Lau W N and Martin-Aranda R (2005), ‘Experiments and modeling of membrane microreactors’, Catal Today, 110, 26–37. Zhang X, Stefanick S and Villani F J (2004), ‘Application of Microreactor technology in Process Development’, Org Proc Res Dev, 8, 455–460.

© Woodhead Publishing Limited, 2013

Microreactors and membrane microreactors

233

Zhang Y, Gwak J, Murakoshi Y, Ikehara T, Maeda R and Nishimura C (2006), ‘Hydrogen permeation characteristics of thin Pd membrane prepared by microfabrication technology’, J Membrane Sci, 277, 203–209. Zhou W, Tang Y, Pan M, Wei X, Chen H and Xiang J (2009), ‘A performance study of methanol steam reforming microreactor with porous copper fibre sintered felt as catalyst support for fuel cells’, Int J Hydrogen Energy, 4, 9745–9753.

5.9 5.9.1 Cf d dm D Dh H KB Kn L n p Re T tc tw Um w

Appendix: nomenclature Notation friction coefficient depth of the microchannel molecular diameter diameter of the microchannel hydraulic diameter height of the channel Boltzmann constant Knudsen number length scale pressure exponential factor pressure Reynolds number temperature catalyst thickness wall thickness mean fluid velocity width of the channel

Greek symbols Λ µ ρf

5.9.2 AFC ATR BHF BOE

mean free path fluid viscosity density of the fluid

Abbreviations alkaline fuel cell autothermal reforming buffered hydrofluoric etch buffered oxide etch

© Woodhead Publishing Limited, 2013

234

Handbook of membrane reactors

CMS/ELP CNF CS CVD CWM DCH DOE DRIE DSD EB EDM ELP ESR FBR FC G–L–S HF HT IM/ELP KCR LIGA MD/ELP MeATR MeOH MEMS MeSR MF MOCVD MM MMR MPSS MS MSP MSR MR MRT OPM PCFSF PDH PDMS PE PECVD

combined method of physical sputtering and electroless plating carbon nanofibre co-sputtering chemical vapour deposition catalytic wall microreactor dehydrogenation of cyclohexane US Department of Energy deep reactive ion etching process dual sputtering deposition electron-beam deposition electro-discharge micro machining electroless plating ethanol steam reforming fixed-bed reactor fuel cell gas–liquid–solid hollow fibre high temperature improved method of electroless plating Knoevenagel condensation reaction lithography-electroforming-moulding-micromachining multidimensional plating mechanism methane autothermal reforming methanol microelectromechanical systems methane steam reforming microfabrication metal-organic chemical vapour deposition micromembrane membrane microreactor macroporous stainless steel microstructure magnetron sputtering methanol steam reforming membrane reactor microreactor technology osmotic pressure method porous copper fibre sintered felt propane dehydrogenation polydimethylsiloxane pulsed electrodeposition plasma-enhanced chemical vapour deposition

© Woodhead Publishing Limited, 2013

Microreactors and membrane microreactors PEMFC PG PI POx PS PSS PVD RT RTD SD SEM Si SOFC SOG SP SS SSP SR UV WGS

proton exchange membrane fuel cell porous glass process intensification partial oxidation porous silicon porous stainless steel physical vapour deposition method room temperature residence time distribution sputter deposition scanning electron microscope silicon solid-oxide fuel cell spin-of-glass spray pyrolysis stainless steel self-supported Pd-based membranes steam reforming ultra violet water-gas-shift

© Woodhead Publishing Limited, 2013

235

6 Photocatalytic membrane reactors: fundamentals, membrane materials and operational issues S. MOZIA and A.W. MORAWSKI, West Pomeranian University of Technology, Szczecin, Poland, R. MOLINARI, University of Calabria, Italy, L. PALMISANO and V. LODDO, University of Palermo, Italy

DOI: 10.1533/9780857097347.1.236 Abstract: This chapter reports the properties of semiconductor materials used in heterogeneous photocatalysis together with a comparison of heterogeneous photocatalytic systems and a brief description of the types of membranes that can be used. Some aspects of membrane operations, such as fouling, separation of a photocatalyst and effectiveness of photodegradation on permeate quality are discussed. Key words: heterogeneous photocatalysis, membranes, photocatalytic membrane reactors.

6.1

Introduction

A promising method for both the abatement of organic and inorganic pollutants (Fujishima et al., 1999; Hoffmann et al., 1995; Schiavello, 1997) and for syntheses (Palmisano et al., 2007a, 2007b; Yurdakal et al., 2008, 2009) is heterogeneous photocatalysis. This is based on particular properties of semiconductor materials that can give rise to redox reactions when they are irradiated with light of suitable energy. Numerous investigations on photocatalytic oxidation have been focused on reduction of the amounts of organic pollutants in wastewater. The semiconductor catalyst is generally used as powder suspended in a liquid medium. The inconvenience of this approach at large scale is the catalyst-recovering step from the solution at the end of operation. The solid−liquid separation is an extremely important issue for the development of the photocatalytic technology; indeed, the best possible recovery of particles must be ensured, in order to prevent their wash out and consequent decrease of their amount in the reactor system. 236 © Woodhead Publishing Limited, 2013

Photocatalytic membrane reactors

237

A promising approach to overcoming these problems is the combined application of photocatalysis and membrane processes. Photocatalytic membrane reactors (PMRs) are useful for the catalyst separation and for the control of photo-oxidation products and/or by-products. The membrane may also ensure continuous operation in systems where the reaction of interest and the separation of product(s) can occur in a single step. The membrane process can be carried out without chemical additives and involves low energy costs. Different systems combining photocatalysis with pressure driven membrane techniques, such as nanofiltration (NF) and ultrafiltration (UF) for the degradation of organic pollutants are described in the pertinent literature (Molinari et al., 2000, 2002a, 2002b; Sopajaree et al., 1999a, 1999b). One of the main drawbacks of these systems is membrane fouling. Coupling of photocatalysis and membrane distillation (MD) could avoid this problem, and an almost complete retention of total organic carbon (TOC) content is reported (Mozia and Morawski, 2006; Mozia et al., 2005, 2007). Self cleaning properties can be generated on a membrane by coating its surface with TiO2 particles (Kim et al., 2003; Madaeni and Ghaemi, 2007). Dialysis membranes were successfully coupled with a photoreactor in order to mineralize organic compounds contained in artificial turbid waters (Azrague et al., 2007). The pervaporation operation integrated with heterogeneous photocatalysis has been demonstrated as a very promising method to improve the detoxification efficiency of water streams containing organic pollutants at low concentration (Camera-Roda and Santarelli, 2007). This chapter introduces some properties and definitions of semiconductor materials used in heterogeneous photocatalysis. The comparison of heterogeneous photocatalytic systems and a brief description of the types of membranes that can be used is also reported. Some aspects of membrane operations such as fouling, separation of a photocatalyst and effectiveness of photodegradation on permeate quality are discussed.

6.2

Physico-chemical and photocatalytic properties of semiconductor materials

The valence bond theory is useful in explaining the structure and the geometry of molecules, but it is not suitable to explain the properties of semiconductor materials. The energy-band model for electrons can be applied to all crystalline solids and allows identifying a conductor, an insulator or a semiconductor material. Indeed, the properties of a solid are determined by the difference of energy between the different bands and the distribution of the electrons contained within.

© Woodhead Publishing Limited, 2013

238

Handbook of membrane reactors

When n atoms with their empty and filled atomic orbitals are assembled in a crystal lattice, molecular orbitals are formed. The energy of these molecular orbitals is so similar as to generate continuous energy bands. The width of the various bands and the separation between them depends on the inter-nuclear equilibrium distance between adjacent atoms. If the energetic levels of isolated atoms are not so different, the progressive enlargement of the bands may lead to their overlapping by decreasing the inter-nuclear distance. The most external energetic band full of electrons is called the valence band (VB). If the VB is partially filled or it is full and overlapped with the band of higher energy, electrons can move allowing conduction (conductors), as in the case of metals that have relatively few valence electrons which occupy the lowest levels of the most external band. On the contrary, the VB is completely filled in the case of ionic or covalent solids but it is separated by a high energy gap from the subsequent empty band. In this situation no electrons can move even if high electric fields are applied and the solid is an insulator. If the forbidden energy gap is not so high, some electrons could pass in the energetic empty band by means of thermal excitation, and the material behaves as a weak conductor, that is, as a semiconductor. The empty band, which allows the movement of the electrons, is called conduction band (CB). The energy difference between the lowest CB edge and the highest VB edge is called band gap (Eg). A material is generally considered a semiconductor when Eg ≤ 3 eV, whereas it is considered a wide band gap semiconductor when its band gap value ranges between 3 and 4 eV. Figure 6.1 shows the position of the energy bands of different types of materials. A semiconductor is called direct band gap semiconductor if the energy of the top of the VB lies below the minimum energy of the CB without a change in electron momentum, whereas it is called indirect band gap semiconductor if the minimum energy in the CB is shifted by a difference in momentum (see Fig. 6.2).

(a)

(b)

(c)

(d)

6.1 Schematic representation of the energetic bands for: conductors (a) and (b); insulators (c); semiconductors (d).

© Woodhead Publishing Limited, 2013

Photocatalytic membrane reactors

Energy

(b)

Energy

(a)

239

Band gap

Band gap

Difference in momentum

Momentum

Momentum

6.2 Energy vs momentum for (a) direct band gap semiconductor and (b) indirect band gap semiconductor.

The electron momentum, p , is defined as p

= λ

[6.1]

k

where ħ is the reduced Plank constant (h/2π), λ is the wavelength and k is the wave number. The probability f(E) that an energetic level of a solid is occupied by electrons can be determined by the Fermi-Dirac distribution function (Dekker, 1957). It applies to fermions (particles with half-integer spin, including electrons, photons, neutrons, which must obey the Pauli exclusion principle) and states that a given allowed level of energy E is function of temperature (T) and of the Fermi level, EF0 , according to the following equation: f (E ) =

(

1

1 + exp e p (E (E EF0 ) kT

)

[6.2]

in which k is the Boltzmann constant. The level EF0 represents the probability of 50% of finding an electron in it. For intrinsic semiconductors and for insulating materials, EF0 value falls inside the energetic gap and its value depends on the effective mass of electrons present at the end of the CB (m*e), on the effective mass of electrons at the beginning of the VB (m*h), and on the amplitude of the band gap (Eg) according to the following equation: EF0

m *h 1 3 Eg + kT ln 2 4 m *e

© Woodhead Publishing Limited, 2013

[6.3]

240

Handbook of membrane reactors f(E ) T2 >T1 T1 > 0 K

1

T= 0K 0.5

0 EF0

E

6.3 Fermi-Dirac distribution for several temperatures.

The value of EF0 is equivalent to the electrochemical potential of the electron, that is, it can be considered as the work necessary to transport an electron from infinite distance to the semiconductor. Figure 6.3 shows that the Fermi-Dirac distribution is a step function at T = 0 K. It has the values of 1 and 0 for energies below and above the Fermi level, respectively. As the temperature increases above T = 0 K, the distribution of electrons in a material changes. At T > 0 K the probability that energy levels above F0 are occupied, and similarly energy levels below EF0 are empty, is not zero. Moreover, the probability that an energy level above EF0 is occupied increases with temperature (distribution sigmoidal in shape) because some electrons begin to be thermally excited to energy levels above the chemical potential, µ ( EF0 ). In order to understand the meaning of the chemical potential some considerations are presented in the following. Some types of impurities and imperfections may drastically affect the electric properties of a semiconductor. In fact the conductivity of a semiconductor can be significantly increased by adding foreign atoms in the lattice (doping) that make available electrons in the CB and holes in the VB. For example, silicon has a crystal structure similar to that of diamond (see Fig. 6.4) and each silicon atom forms four covalent bonds with the four nearest atoms, corresponding to a chemical valence equal to four. The addition of atoms, for instance arsenic, phosphorous or antimony, having one valence electron more, will lead to an excess of positive charge (Fig. 6.5a), due to the transfer of an electron from the foreign atom to the CB (donor doping). If the foreign atom, for instance boron, gallium or indium, has one valence electron less, it can accept one electron from the VB (acceptor doping) (Fig. 6.5b). In the first case an energetic level close to the CB is introduced;

© Woodhead Publishing Limited, 2013

Photocatalytic membrane reactors

241

6.4 Crystal structure of silicon.

(a)

Si

Si

Si +

Si e–

Si

As

Si

Excess positive charge

Si

Excess electron from arsenic atom

(b)

Si

Si

Si –

Si

Excess negative charge

B

h+ Si

Si

Si

Positive hole formed by the removal of the electron from this bond

6.5 (a) With arsenic impurity an electron is available for conduction; (b) with boron impurity a positive hole is available.

© Woodhead Publishing Limited, 2013

242

Handbook of membrane reactors

EC E (eV) E F EV

EC E (eV) E F EV

EC E (eV) E F EV (a)

(b)

(c)

6.6 Energetic levels of a semiconductor: (a) intrinsic semiconductor; (b) n-type semiconductor; (c) p-type semiconductor.

consequently electrons can pass more easily in it. In this case the solid is called an n-type semiconductor and the Fermi level will be close to the CB (Fig. 6.6b). In the second case an energetic level close to the VB is formed, in which electrons can be promoted with the formation of holes. The semiconductor is of p-type and its Fermi level will be close to the VB (Fig. 6.6c). The notion of energetic levels of electrons in solids can be extended to the case of an electrolytic solution containing a redox system (Gerischer, 1970). The occupied electronic levels correspond to the energetic states of the reduced species whereas the unoccupied ones correspond to the energetic states of the oxidized species. The Fermi level of the redox couple, EF,redox, corresponds to the electrochemical potential of electrons in the redox system and is equivalent to the reduction potential, V0. In order to correlate the energetic levels of a semiconductor to those of a redox couple in an electrolyte, two different scales can be used. The first is expressed in eV, the other one in V (Fig. 6.7a). The difference between the two scales is due to the fact that in solid state physics zero is the level of the electron in vacuum, whereas in electrochemistry the reference is the potential of the normal hydrogen electrode (NHE). The correlation between the two scales can be calculated from the value of potential of NHE which is equal to −4.5 eV when it is referred to that of the electron in vacuum (Lohmann, 1967). If a semiconductor is placed in contact with a solution containing a redox couple, the equilibrium is reached when the Fermi levels of both phases become equal. This occurs by means of an electron exchange from solid and electrolyte which leads to the generation of a charge inside the semiconductor. This charge is distributed in a spatial charge region near to the surface in which the values of hole and electron concentrations differ also considerably from those inside the semiconductor. In Fig. 6.7, the energetic levels of an n-type semiconductor and a redox electrolyte before the contact are drawn. In particular, as the energy of the Fermi level is higher than that of the electrolyte, equilibrium is reached by electron transfer from the semiconductor to the solution. The electric field produced by this transfer is represented by the upward band bending (Fig. 6.7b). Due to the presence of the field, holes generated in excess in the spatial charge region move toward the semiconductor surface, whereas electrons in excess migrate from the surface to the bulk of the solid. Figure 6.8 shows the contact between a redox electrolyte and a p-type semiconductor. In this case, transfer of electrons

© Woodhead Publishing Limited, 2013

Photocatalytic membrane reactors

243

E (eV) Semiconductor 0 (H+/H2) –4.5

Electrolyte

EC EF

EF, redox

0.0 (H+/H2) V0, redox

EV

E (V) (a) EC

EF, redox

EF EV (b)

6.7 Formation of a junction between an n-type semiconductor and a solution: (a) before the contact; (b) at equilibrium.

E (eV) Semiconductor 0

Electrolyte

EC

(H+/H2) –4.5

EF, redox EF EV

0.0 (H+/H2) V0, redox

E (V) (a)

EC EF, redox

EF EV

(b)

6.8 Formation of a junction between a p-type semiconductor and a solution: (a) before the contact; (b) at equilibrium.

© Woodhead Publishing Limited, 2013

244

EC

Handbook of membrane reactors EC

ΔV

VFB

EF EF, redox

EF

EF, redox EV

EV

E (V )

E (V )

(a)

(b)

6.9 Scheme of the energetic levels at the semiconductor–electrolyte interface for an n-type semiconductor: (a) at equilibrium; (b) flat band potential.

EC EC

EF, redox

EF

EF, redox

EF EV

EV (a)

(b)

6.10 Generation of electron–hole pair due to irradiation of: (a) n-type semiconductor; (b) p-type semiconductor. (h is Planck constant and ν is frequency.)

occurs from the electrolyte to the semiconductor, and the band bending is downward. If the potential of the electrode changes due to an anodic or cathodic polarization, a shift of the Fermi level of the semiconductor with respect to that of the solution occurs with an opposite curvature of the bands (Fig. 6.9). For a particular value of the electrode potential, the excess charge disappears and the bands become flat from the bulk to the surface of the semiconductor. The corresponding potential is called flat band potential, VFB, and the determination of this potential allows calculating the values of the energy of the conduction and the valence bands (Pleskov and Gurevich, 1986). When a semiconductor is irradiated by a radiation of suitable energy equal or higher to that of the band gap, Eg, electrons can be promoted from the VB to the CB. Figure 6.10 shows the scheme of formation of an electron−hole pair due to the absorption of a photon by the semiconductor. The existence of an electric field in the spatial charge region allows the separation of the photogenerated pairs. In the case of n-type semiconductors, electrons migrate toward the bulk whereas holes move to the surface (Fig. 6.10a). In the case of p-semiconductors, holes move towards the interior of the semiconductor and electrons toward the surface (Fig. 6.10b).

© Woodhead Publishing Limited, 2013

Photocatalytic membrane reactors

Ox2

Ox2

CB

CB

VB

VB

CB

Red1

Red1

(a)

245

Ox2

VB

CB

VB

Ox2

Red1

Red1

(b)

(c)

(d)

E (V)

6.11 Relative positions between the valence (VB) and the conducting (CB) bands and the energies of two redox couples with different values (cases a–d). OX2 and Red1 represent the oxidized and the reduced species, respectively, of two different redox couples (1 and 2). Only for (d) are both reduction and oxidation reactions thermodynamically allowed.

Photoproduced holes and electrons, during their migrations toward opposite directions, can (i) recombine, dissipating their energy as electromagnetic radiation (photons emission) or more simply as heat or (ii) can react with electron−acceptor or electron−donor species present at the semiconductor−electrolyte interface reducing or oxidizing them, respectively. The energy of the CB edge, EC, corresponds to the potential of the photogenerated electrons, whereas the energy of the VB edge, EV, corresponds to the potential of the holes. If the value of EC is more negative than the potential of a species present in solution, electrons reaching the interface are able to reduce the oxidized form of the redox couple. Conversely, if the potential of EV is more positive than that of the redox couple, photoproduced holes can oxidize its reduced form (Fig. 6.11). Knowledge of the relative edge positions of the bands and of the energetic levels of the redox couples is essential to establish if thermodynamic allows the occurrence of oxidation and/or reduction of the species in solution. Figure 6.12 reports the values of band gap and the positions of the VB and CB edges for various semiconductors. The photocatalytic properties of a semiconductor depend on the position of the energetic levels, on the mobility and mean lifetime of the photogenerated electrons and holes, on the light absorption coefficient and on the nature of the interface. Moreover, the photoactivity depends on the methods of preparation of the powders which allows varying many physico-chemical properties of the semiconductor as the crystalline structure, the surface area and the distribution of the particle size. In a photocatalytic system the behavior of each single particle of semiconductor is similar to that of a photo-electrochemical cell constituted by

© Woodhead Publishing Limited, 2013

246

Handbook of membrane reactors E (eV) −2.0 ZnS −1.0

CdS SrTiO3

CdTe CdSe

Si

TiO2

0.0 3.6

3.0

3.2

2.4

1.4

1.7

1.1

WO3 Fe2O3

MoS2 1.75

1.0 2.8

2.0

2.3

3.0

6.12 Positions of the band edges for some semiconductors in contact with aqueous electrolyte at pH = 0. Red1

Ox1 −

CB

hν

VB

+ Red2

Ox2

6.13 Scheme of the photocatalytic process occurring on an illuminated particle in contact with a redox system. The oxidizing agent Ox1 is oxygen and the reducing one is an organic substrate.

a semiconductor electrode in contact with an electrode of an inert metal (Bard, 1979). In a photo-electrochemical cell an oxidation or reduction reaction may occur on the semiconductor electrode, whereas in a semiconductor particle immersed in an electrolyte solution both reactions occur simultaneously by hole transfer from the VB and by electron transfer from the CB (see Fig. 6.13). An advantage to use semiconductor powder suspensions lays in the fact that each particle acts as a small photocell, and in 100 mg of powder, consisting for example of 0.1 mm diameter particles, more than 1011 independent particles are present.

© Woodhead Publishing Limited, 2013

Photocatalytic membrane reactors

247

It can be taken into account that generally the powders have great surface areas (1−100 m2/g) which favors the occurrence of charge transfer reactions at the interface between the semiconductor surface and the redox electrolyte. The separation of the electron−hole pairs increases by increasing the thickness of the spatial charge region, which depends on the doping of the semiconductor. When the volume of the particles decreases, the effect on the charge separation becomes a minimum, as sometimes the particle size is smaller than the thickness of the spatial charge region. The absorption of radiation of suitable wavelength by the semiconductor allows the transformation of light energy into chemical energy, and this phenomenon represents a fundamental step in heterogeneous photocatalysis. In particular, when aqueous suspensions of semiconductor powders are irradiated, at the solid−liquid interface a great variety of photoinduced chemical reactions, able to degrade many organic and inorganic molecules, can occur by means of formation of very reactive radical species which are generated in the presence of O2 and H2O. The following scheme shows the events which can occur at the semiconductor−water interface when, for instance, TiO2 is used as the photocatalyst: TiO2

h

OH − + O2 •

+ ( VB)

(

TiO2 e(−CB) + h(+VB)

)

→ • OH

[6.5]

e(−CB) → • O2−

[6.6]

O2− + H + → • HO2

[6.7]

2 • HO 2 → O2 + H 2 O2 H 2 O2 + • O2−

OH

[6.4]

[6.8] •

OH O2

[6.9]

The semiconductor most frequently used is undoubtedly titanium dioxide, produced in large amounts as a low cost pigment. The photocatalytic activity of anatase, rutile and brookite polymorphic modifications of TiO2 is affected by several factors, such as the crystalline structure, the surface area, the particle size distribution and the density of surface hydroxyl groups. Although the positions of valence and conduction bands of both anatase and rutile are positive enough to allow the oxidation of many organic molecules, anatase

© Woodhead Publishing Limited, 2013

248

Handbook of membrane reactors

phase shows a higher activity as the level of its CB is more favorable for the electron transfer needed for the complementary reduction reaction. The poor efficiency of rutile is due mainly to the high recombination rate of electron−hole pairs and to its low ability to photoadsorb oxygen (Sclafani et al., 1990). It can be taken into account, however, that the most used TiO2 photocatalysts, as for example Degussa P25 (ca. 80% anatase, 20% rutile), contain a mixture of both the crystalline modifications. The high value of band gap for anatase (Eg = 3.2 eV) and rutile (Eg = 3.0 eV) phases allows utilization only of radiation with wavelength lower than 400 nm ca., which represents ca. 5% of solar light (Augugliaro et al., 2010; Fox and Dulay, 1993; Mills and Le Hunte, 1997; Ollis et al., 1989; Schiavello, 1997; Turchi and Ollis, 1990).

6.3

Heterogeneous photoreactors and photocatalytic systems

The ideal reactors mostly studied in chemical engineering are: the batch or semi-batch reactor, the plug flow reactor (PFR) and the continuous flow stirred tank reactor (CSTR). A batch reactor has no input or output of mass, and the stirring of the reactor avoids temperature or concentration gradient in the reaction medium. The reaction rate is uniform and can be considered everywhere equal to the average value. A PFR can be visualized as a tubular reactor for which three conditions must be satisfied: (i) the axial velocity profile is flat; (ii) there is complete mixing across the tube, so that all the reaction variables are a function of the axial dimension of the reactor (named z); and (iii) there is no mixing in the axial direction. PFRs have spatial variations in concentration and temperature. Such systems are called distributed, and analysis of their steady state performance requires the solution of differential equations. The CSTR is a flow reactor, in which the contents are mechanically agitated. If the mixing is adequate, the entering feed will be quickly dispersed through the vessel, and the composition and temperature at any point will approximate the average composition and temperature. Perfect mixers have no spatial distribution of compositions and temperatures. Such systems are called lumped. The steady state performance of lumped systems is determined by algebraic equations rather than differential ones. The utilization of radiation and solid catalysts in a synergic effect for performing chemical transformations of species present in liquid or gaseous mixtures makes heterogeneous photocatalytic reactors more complex than heterogeneous catalytic reactors. The main components of a photocatalytic system are the photoreactor and the radiation source (Braun et al., 1991). For thermal catalytic processes, the reactor configuration is chosen on the basis of: (i) the mode of operation;

© Woodhead Publishing Limited, 2013

Photocatalytic membrane reactors

249

(ii) the phases present in the reactor; (iii) the flow characteristics; (iv) the needs of heat exchange; and (v) the composition and the operative conditions of the reacting mixture. In addition to previous parameters, the choice of the heterogeneous photoreactors is related to their geometry and materials in order to guarantee the penetration of radiation all over the reacting mixture, having in mind that the absorbed photon energy should be equal or higher than the band gap of the used photocatalyst. In the case of stirred photoreactors the presence of the photocatalyst, usually a powdered micro- or nano-crystalline semiconductor, affects the depth of radiation penetration in a complex way. For heterogeneous photocatalytic reactions the contact among reactants, photons and catalyst must be maximized. Mixing and flow characteristics of the photoreactor may greatly enhance these contacts. For liquid−solid and gas−solid systems, continuously stirred tank photoreactors and fluidized bed photoreactors are the most suitable. If a fixed bed photoreactor is used, the irradiation aliquot of catalyst is limited to a thin layer and a large reactor volume is required. In order to enhance the reactor performance, devices which use the photocatalyst deposited as thin film on various kinds of materials represent one of the best configurations. For instance, using technologies created in the development of ultra-high temperature resistant ceramics, photocatalytic fibers with large surface area and good fiber strength have been developed (Ishikawa, 2004). The performance of a photoreactor is strongly dependent on the irradiation source. Six main types of radiation sources may be used: (a) arc lamps; (b) fluorescent lamps; (c) incandescent lamps; (d) lasers; (e) LEDs and (f) solar radiation. In arc lamps, the emission is obtained by the activation of a gas by collision with accelerated electrons generated by an electric discharge between two electrodes, typically tungsten-made. The type of lamp is often denoted by the gas contained in the bulb: including neon, argon, xenon, krypton, sodium, metal halide and mercury. In particular, for mercury lamps, the following classification, based on the Hg pressure, is done: 1. Low pressure. The lamp contains Hg vapor at a pressure of ca. 0.1 Pa at 298 K and it emits mainly at 253.7 and 184.9 nm. 2. Medium pressure. This lamp contains Hg vapor at a pressure from 100 to several hundred kPa. It emits mostly from 310 nm to 1000 nm with most intense lines at 313, 366, 436, 576 and 578 nm. 3. High pressure. This lamp contains Hg vapor at a pressure equal or higher than 10 MPa and it emits in a continuous background between ca. 200 and 1000 nm with broad lines. 4. Hg–Xe. This lamp is used to simulate the solar radiation as it is an intense ultraviolet (UV), visible and near-IR radiation source. The lamp contains

© Woodhead Publishing Limited, 2013

250

Handbook of membrane reactors

a mixture of Hg and Xe vapors at different concentrations under high pressure. The presence of Hg vapor increases the emission in the UV region. The above lamps are usually of cylindrical shape, with the arc length increasing as the pressure decreases and power increases. The power ranges from few watts to ca. 60 000 W. Generally medium and high pressure mercury lamps need to be cooled by circulating liquids in their thimbles. Fluorescent lamps are filled with gas containing a mixture of low pressure mercury vapor and argon (or xenon), more rarely argon−neon, sometimes even krypton. The inner surface of the lamp is coated with a fluorescent (often slightly phosphorescent) coating made of varying blends of metallic and rare-earth phosphor salts. The cathode is generally made of coiled tungsten, which is coated with a mixture of strontium, barium and calcium oxides that have a relatively low thermo-ionic emission temperature. When the light is turned on, the cathode is heated enough to emit electrons. These electrons collide with gas atoms which are ionized to form a plasma by a process of impact ionization. As a result the conductivity of the ionized gas rises, allowing higher currents to flow through the lamp. The mercury, which exists at a stable vapor pressure equilibrium point of about one part per thousand inside the tube (with the noble gas pressure typically being about 0.3% of standard atmospheric pressure), is then likewise ionized, causing it to emit light in the UV region of the spectrum, predominantly at wavelengths of 253.7 nm and 185 nm. The efficiency of fluorescent lighting is due to the fact that low pressure mercury discharges emit about 65% of their total light at the 254 nm line (also about 10–20% of the light emitted in UV is at the 185 nm line). The UV light is absorbed by the bulb fluorescent coating, which re-radiates the energy at lower frequencies (longer wavelengths: two intense lines of 440 nm and 546 nm wavelengths appear on commercial fluorescent tubes) to emit visible light. The blend of phosphors controls the color of the light and, along with the bulb glass, prevents harmful UV light escape. The actinic fluorescent tubes have emission in the near-UV region (ca. 360 nm) due to their particular fluorescent coating. The emission of incandescent lamps is obtained by heating at very high temperature suitable filaments of various substances by current circulation. One of the most used incandescent lamps is the halogen lamp, wherein a tungsten filament is sealed into a small envelope filled with a halogen gas such as iodine or bromine. A tungsten-halogen lamp creates an equilibrium reaction in which the tungsten that evaporates when giving off light is preferentially re-deposited at the hot-spots, preventing the early failure of the lamp. This also allows halogen lamps to be run at higher temperatures, which would cause unacceptably short lamp lifetimes in ordinary incandescent lamps, allowing for higher luminous efficacy, apparent brightness and

© Woodhead Publishing Limited, 2013

Photocatalytic membrane reactors

251

whiter color temperature. Because the lamp must be very hot to create this reaction, the halogen lamp envelope must be made of hard glass or fused quartz, instead of ordinary soft glass, which would soften and flow too much at these temperatures. In applications that require UV radiation, the lamp envelope is made out of undoped quartz. Thus, the lamp becomes a source of UV-B radiation. A typical halogen lamp is designed to run for about 2000 hours, twice as long as a typical incandescent lamp. A laser (light amplification by stimulated emission of radiation) is a device that emits light through a specific mechanism. This is a combined quantum-mechanical and thermodynamic process. A typical laser emits light in a narrow and well-defined beam and with a well-defined wavelength (or color). This is in contrast to a light source such as the incandescent light bulb, which emits in almost all directions and over a wide spectrum of wavelengths. All these properties are summarized in the term ‘coherence’. A laser consists of a gain medium inside an optical cavity, in order to supply energy to the gain medium. The gain medium is a material (gas, liquid or solid) with appropriate optical properties. In its simplest form, a cavity consists of two mirrors arranged such that light bounces back and forth, each time passing through the gain medium. Typically, one of the two mirrors, the output coupler, is partially transparent. All light that is emitted by the laser passes through this output coupler. Light of a specific wavelength that passes through the gain medium is amplified (increases in intensity); the surrounding mirrors ensure that most of the light makes many passes through the gain medium. Part of the light that is between the mirrors (i.e., it is in the cavity) passes through the partially transparent mirror and appears as a beam of light. The process of supplying the energy required for the amplification is called pumping and the energy is typically supplied as an electrical current or as light at a different wavelength. In the latter case, the light source can be a flash lamp or another laser. Most practical lasers contain additional elements that affect properties such as the wavelength of the emitted light and the shape of the beam. LEDs are special p-n junction diodes formed by a thin layer of doped semiconductor material. They are based on the semiconductor optical properties to produce photons by the recombination of electron−hole pairs. Electrons and holes are injected in a recombination zone through two parts of the diode doped in different ways, n-type impurities for electrons and p-type for holes. Figure 6.14 shows the working scheme. UV light is achievable by using GaN as the active layer, with emission in the near-UV region with wavelengths around 350–370 nm. Also nitrides containing aluminum, such as AlGaN and AlGaInN, can be used to achieve shorter wavelengths, whereas by using aluminum nitride, wavelengths

© Woodhead Publishing Limited, 2013

252

Handbook of membrane reactors Light emission

Hole

p-type

Electron

n-type +

_

EC E (eV)

EC EF

EF EV

EV

6.14 Working scheme and energetic levels in an LED.

Epoxy lens-case Wire bond Reflective cavity Semiconductor die

Lead frame

Flat spot − Cathode

+ Anode

6.15 Main parts of an LED.

down to 210 nm can be obtained. Figure 6.15 shows the main parts of an LED. In addition to the artificial radiation light sources above described, sunlight can be used to illuminate the photocatalysts. The earth receives about 1.7 × 1014 kW of solar radiation (1.5 × 1018 kWh per year). Extraterrestrial radiation has an intensity of 1367 W·m−2 in a wavelength range between 200 and 5000 nm, which is reduced to 280–4000 nm when it reaches the ground, due to absorption phenomena by atmospheric compounds such as ozone, oxygen, carbon dioxide, aerosols, water vapor, clouds, etc.

© Woodhead Publishing Limited, 2013

1.8

1000

1.6

900 800

1.4

700

1.2

600 1.0 500 0.8

400

0.6

300

0.4

200

0.2

100

0 300

400

500 600 700 800 Wavelength (nm)

253

Integrated irradiance (W m–2)

Irradiance per wavelength (W m–2 nm–1)

Photocatalytic membrane reactors

0 900 1000

6.16 ASTM global irradiance standard solar spectrum (air mass (AM) 1.5) up to wavelength of 1000 nm, incident on a plane tilted 38° facing south, normalized to 1000 W m−2 for the whole spectrum.

The solar radiation that reaches the planet without being absorbed or scattered is called direct-beam radiation, whereas radiation that reaches the ground but has been dispersed before is called diffuse radiation; the sum of both is called global radiation. Figure 6.16 shows the standard solar spectrum at sea level, where the spectral irradiance data are for the sun at a solar zenith angle of 48.19°. This zenith angle corresponds to an air mass (AM) of 1.5, which is the ratio of the direct-beam solar irradiance path length through the atmosphere at a solar zenith angle of 48.19° to the path length when the sun is in the zenith. The radiation effectively reaching the ground level changes strongly due to several factors, such as date, time, geographic latitude, atmospheric conditions (aerosols, humidity, etc.) or cloudiness. To judge the feasibility and profitability of solar applications at a specific site, studies have to be performed to estimate or to measure the amount of radiation actually available at the specific site along the year. It is worth noting that the comparison of the performance of the various heterogeneous photocatalytic systems is an important issue, and the irradiation conditions have to be taken into account. Unfortunately the comparison is not always easy, due to the different experimental conditions used in the different laboratories. The spectrum of the lamps, their power and the distance from the photoreactor, are parameters that strongly influence the reaction rate and the distribution of the products formed. Sunlight is

© Woodhead Publishing Limited, 2013

254

Handbook of membrane reactors

an inexpensive source of light, but some drawbacks should be faced, as for instance the fact that the irradiation is not constant during the day and to carry out systematic studies and comparison among different photocatalysts is quite difficult. An open question is how to perform the experiments in laboratories in order to correctly compare the photoreactivity results. One way is to work by using equal amounts of photocatalyst along with the other experimental conditions, that is, type of irradiation source, initial concentration of starting reagents, pH in an aqueous ambient, etc. Another way is to carry out the experiments by using different amounts of different photocatalysts, but these quantities should be sufficient to absorb completely the impinging photons. In the latter case, the experiments would be carried out with equal amounts of absorbed photons, and a cylindrical photoreactor geometry would be suitable to measure the transmitted light in the presence of different amounts of powdered photocatalysts. The minimum amount of photocatalyst sufficient to absorb all the photons will be that for which the transmitted light is negligible. Another point to be considered is the value of specific surface area of the photocatalysts. Many authors express the photocatalytic parameters in terms of ‘per area’ and denote them as area-related parameters. Notably in photocatalysis, the area-related parameters do not characterize the catalyst as they do in heterogeneous dark catalysis. In fact, the properties of a photocatalyst are not directly proportional to its area, because the light may not equally reach every part of a particle. In the most frequent cases the surface area accessed by the light cannot be determined, and thus it is recommended to use the total surface area in terms of the BET(N2) area. This means that generally a lower limit of the quantity under examination is obtained, because the surface determined by N2 adsorption will be larger than that reached by the radiation.

6.4

Materials and design of photocatalytic membranes

PMRs with photocatalytic membranes are devices in which photodegradation of contaminants occurs on the external surface and within the pores of a membrane, while reactants are permeating in a one-pass flow. Therefore, the element which has to be irradiated is the membrane itself. The photocatalytic membrane in a PMR acts as a support for the photocatalyst and might act as a barrier for the molecules present in the solution (initial compounds and products or by-products of their decomposition). Due to the photocatalytic degradation of organic contaminants an improvement of permeate quality is observed. Modification of a membrane with photocatalyst particles

© Woodhead Publishing Limited, 2013

Photocatalytic membrane reactors Gas phase reactions

255

Liquid phase reactions APPLICATION

Photocatalytic membranes

Inorganic

Polymeric

Photocatalyst supported on a membrane surface

Photocatalyst entrapped in a membrane structure

MEMBRANE MATERIAL

Direct grafting of TiO2 nanotubes

Dip-coating

Hydrothermal synthesis – filtration

Phase inversion Dip-coating

Physical deposition

Bi-axial stretching PREPARATION METHOD

Photochemical method

6.17 Division of photocatalytic membranes.

leads also to fouling mitigation due to an increase of membrane hydrophilicity (Alaoui et al., 2009; Bae and Tak, 2005; Ciston et al., 2009; Damodar et al., 2009; Madaeni and Ghaemi, 2007; Mansourpanah et al., 2009; Yang et al., 2007). However, fixation of a photocatalyst often results in a loss of photoactivity compared to the suspended systems. Moreover, since the light source must be positioned near the membrane, the latter should be resistant to UV irradiation and oxidation by hydroxyl radicals or other reactive oxygen species (ROS). In general, photocatalytic membranes could be divided into polymeric and inorganic ones. The photocatalyst could be supported on a membrane surface or entrapped in a membrane structure. The method of photocatalyst incorporation in the membrane structure depends mainly on the membrane material. In Fig. 6.17 a simplified division of photocatalytic membranes with reference to the application, membrane material and preparation method is shown.

© Woodhead Publishing Limited, 2013

256

Handbook of membrane reactors

6.5

Polymeric membranes

The techniques of preparation of polymeric photocatalytic membranes can be divided into: I. Methods in which a photocatalyst is being supported on a membrane surface. II. Methods in which a photocatalyst is being entrapped within a membrane structure. The membranes with a photocatalyst supported on a membrane surface can be prepared according to the following techniques: (a) dip coating (Bae and Tak, 2005; Kim et al., 2003; Kim and Van der Bruggen, 2010; Kwak et al., 2001; Madaeni and Ghaemi, 2007; Mansourpanah et al., 2009; Vankelecom, 2002), (b) photochemical process based on photografting and photopolymerization (Barni et al., 1995; Bellobono et al., 1992, 1994, 2005, 2006a, 2006b, 2008; Moroni et al., 1999), (c) physical deposition of TiO2 layer: deposition of the photocatalyst layer is performed by ultrafiltration of TiO2 suspension through a polymer membrane (Bai et al., 2010; Molinari et al., 2002a, 2004). Supporting a TiO2 layer on a membrane surface is usually performed by the dip coating method (a). In this method a dry porous substrate is dipped into a suspension containing photocatalyst particles and subsequently withdrawn from it. During the dipping the porous surface is wetted by the dispersion liquid. The capillary suction caused by the porous substrate drives photocatalyst particles to concentrate at the substrate/suspension boundary, and a wet and more or less dense cake of well-defined thickness is formed if the particles cannot enter into the pores. In this capillary filtration, the driving force behind the fluid flow is the capillary suction pressure of the substrate, as in the slip casting process (Gu and Meng, 1999). Figure 6.18 shows the membrane formation process on a porous substrate by the capillary suction pressure. During preparation of a photocatalytic membrane by the dip coating method a membrane is dipped in a TiO2 suspension in water (Kim et al., 2003; Kim and Van der Bruggen, 2010; Kwak et al., 2001; Madaeni and Ghaemi, 2007), alcohol, for example, 2-propanol (Vankelecom, 2002) or other liquids. The membrane after dipping could be additionally pressurized with a compressed gas (Bae and Tak, 2005). The photocatalyst particles are self-assembled on a polymeric membrane due to a coordination of the functional groups present on the membrane surface (e.g., carbonyl or

© Woodhead Publishing Limited, 2013

Photocatalytic membrane reactors Suspension of a photocatalyst in water

257

Concentration of photocatalyst particles at the suspension/membrane boundary

Filtration cake

Membrane

Water in membrane pores

6.18 Formation of a photocatalyst active layer on a porous membrane support by the capillary suction pressure during the dip coating process. (Adapted from Gu and Meng, 1999.)

sulfone), or by a hydrogen bond between the functional group of the membrane and surface group of TiO2. In Fig. 6.19a schematic diagram showing the self-assembly of TiO2 nanoparticles on the membrane surface is presented (Kim et al., 2003; Kim and Van der Bruggen, 2010). Figure 6.20 shows SEM microphotographs of (a) virgin reverse osmosis (RO) membrane and (b) the membrane dip coated with TiO2 particles. In the case of photocatalytic membranes prepared by the dip coating technique, there is a risk of detachment of TiO2 particles during long-term operations or high shear stress introduced by the filtration velocity (Bae and Tak, 2005; Mansourpanah et al., 2009). As a result, the membrane might be losing its properties after some time. In order to improve the strength of attachment of TiO2 particles, a modification based on the induction of –OH groups on the membrane surface was proposed (Mansourpanah et al., 2009). The –OH functionalized membrane made of polyethersulfone and polyimide (PES/PI) was obtained by immersion of the membrane in an aqueous solution of diethanolamine. The occurred reaction is the nucleophilic and electrophilic addition. After modification the membranes were immersed in TiO2 suspension. The self-assembly between TiO2 nanoparticles and functional groups of the modified polymers is presented in Fig. 6.21. The modification of the membrane by diethanolamine resulted in an increase of the amount of –OH groups on the membrane surface and in a decrease in space hindrance for attracting TiO2 particles. As a consequence of this, more TiO2 nanoparticles were bound, which led to an increase of membrane surface roughness and hydrophilicity. Moreover, the filtration experiments revealed that the quantity of TiO2 nanoparticles on the –OH functionalized membrane surface was less decreased compared to the unmodified membrane (Mansourpanah et al., 2009).

© Woodhead Publishing Limited, 2013

258

Handbook of membrane reactors (a)

O Ti4+ C R

O

(b) R

Ti4+

HO

O C OH

(c)

O S

O Ti4+

O

O Ti4+

(d)

O O

S

OH

O

Ti4+

HO

O Ti4+

6.19 Self-assembly of TiO2 nanoparticles on a membrane surface: (a) bidentate coordination of carboxylate to Ti4+; (b) H-bond between a carbonyl group and a surface hydroxyl group; (c) coordination of sulfone group and ether bond to Ti4+; (d) H-bond between sulfone group and ether bond and surface hydroxyl group of TiO2. (Adapted from Kim et al., 2003; Kim and Van der Bruggen, 2010.)

During the dip coating of the membranes, the nanoparticles could adsorb not only on the membrane surface but also on the walls of membrane pores. Therefore, usually a decrease of the pure water flux compared to the unmodified membrane is observed. The magnitude of the flux deterioration

© Woodhead Publishing Limited, 2013

Photocatalytic membrane reactors

259

(a)

(b)

6.20 SEM images of surface of (a) virgin and (b) TiO2-dip-coated TFC-SR (thin film composite selective rejection) composite RO membrane (Fluid Systems Company). Membrane material: PVA/polyaryl sulfone/ polyester. (Reprinted from Madaeni and Ghaemi, 2007 with permission from Elsevier.)

depends on both the material and the structure of the neat membrane. However, the membrane performance during processing of real water/ wastewater is usually significantly improved by deposition of TiO2 particles (Bae and Tak, 2005; Kim et al., 2003). The dip coating technique could be also coupled with the cross linking method. Such a technique of preparation of photocatalytic membranes was proposed by Liu et al. (2009). Polyethylene terephthalate (PET) filter cloth was dip coated with a suspension containing TiO2 obtained by the sol−gel method, FeSO4 and activated carbon fibers (ACF). After the process, the resulting composite membrane was chemically cross-linked with polyvinyl alcohol (PVA) and glutaraldehyde (GA). The immobilized membrane was treated with NaBH4 to accelerate the reduction reaction. The composite membranes were found to be stable and durable over a long period of time.

© Woodhead Publishing Limited, 2013

260

Handbook of membrane reactors O

O

O

C

C

C

N

*

N C

C

O

O

*

DEA OH

OH HO O

O

O

C

C

C

N

N NH

*

OH

HN C

C

O

O

*

TiO2

Ti4+ Ti4+

OH

OH

Ti4+

Ti4+

HO

O

O

C

C

C

OH N

N *

O

HN

HN C

C

O

O

*

OH Ti4+

Ti4+

6.21 Mechanism of the polyimide (PI) membrane modification by diethanolamine (DEA) and the self-assembly of TiO2 nanoparticles on the DEA-modified surface of the PI membrane. (Adapted from Mansourpanah et al., 2009.)

Photocatalytic membranes could be also prepared by the photochemical method (b). During the process the controlled amounts of appropriate monomers and prepolymers, containing the semiconductor to be immobilized, with or without addition of a suitable photosensitizer (usually an

© Woodhead Publishing Limited, 2013

Photocatalytic membrane reactors

261

organometallic derivative of Fe, Co and V, for example, triethylvanadate(V), oxo-(diquinolyloxo)vanadic(V) acid, µ-peroxo-bis[N,N’-ethylene-bis(salicyl ideneiminato)cobalt(III), etc.), are photografted onto a support (e.g., perforated polyester network, polypropylene non-woven tissue, cellulose acetate membrane, etc.). The final porosity of the photosynthesized membranes can be regulated by controlling the rheological and photochemical parameters during a membrane manufacture. The method was invented in the 1990s by Bellobono and co-workers (Bellobono et al., 1992). They prepared a series of photocatalytic membranes from which the best performance was exhibited by the patented Photoperm BIT/313 photocatalytic membrane containing 30 wt% of TiO2 Degussa P25. Another method of preparation of photocatalytic membranes is physical deposition of a TiO2 layer on a membrane surface (c). The method is based on filtration of TiO2 suspension through a polymer membrane (Molinari et al., 2002a, 2004). In this case a filtration cake built of photocatalyst particles, which plays a role of the so-called ‘dynamic membrane’, is formed on a polymer membrane surface. Depending on the process parameters, such as the amount of TiO2 in suspension, filtration pressure, etc., different amounts of photocatalyst particles could be deposited on the membrane. The results obtained for flat sheet polyacrylonitrile (PAN) membrane modified with TiO2 P25 (Degussa) revealed that the optimal TiO2 layer density was about 2.04 mg TiO2/cm2. For lower values, the cake layer showed poor mechanical stability and the photocatalyst particles detached very easily from the membrane, while for higher values the cake layer presented peaks of accumulated particles. Immobilization of TiO2 by ultrafiltration of its aqueous suspension is a very simple technique. However, in the case of the membranes modified with nanoparticles such as TiO2 P25 a significant decrease of the pure water flux, compared to the neat membrane, was observed. Moreover, the pure water flux through the membrane with immobilized TiO2 showed a linear trend changing very little with the applied transmembrane pressure and the amount of TiO2. This is due to the fact that the photocatalyst layer behaves as a filtration cake with a very dense structure (Bai et al., 2010; Molinari et al., 2002a). However, it was found that application of TiO2 photocatalyst with a nano-thorn structure for preparation of the photocatalytic membrane resulted in quite different membrane behavior. The pure water flux was proportional to the transmembrane pressure, as shown in Fig. 6.22. The reason was that the surface of the TiO2 nano-thorn membrane was hierarchically porous, which is favorable for water to pass through, whereas the surface of the membrane modified with TiO2 P25 was dense, which would block water from passing through (Bai et al., 2010). The second group of methods of preparation of polymeric photocatalytic membranes includes techniques by which membranes with a

© Woodhead Publishing Limited, 2013

262

Handbook of membrane reactors

Nano-thorn TiO2 modified membrane Permeate flux

TiO2 P25 modified membrane

Transmembrane pressure

6.22 The influence of TiO2 morphology on permeate flux through a photocatalytic polymer membrane prepared by the dip coating technique. (Adapted from Bai et al., 2010.)

photocatalyst entrapped within their structure are obtained. The membranes with entrapped photocatalyst can be divided into: (a) membranes prepared by the phase inversion technique from the casting solution containing photocatalyst particles (Alaoui et al., 2009; Artale et al., 2001; Bae and Tak, 2005; Damodar et al., 2009; Kleine et al., 2002; Li et al., 2009; Molinari et al., 2004), (b) membranes prepared by bi-axial stretching of polymer extrusion containing catalytic (TiO2) filler (Morris et al., 2004). TiO2-entrapped membranes obtained by the phase inversion technique (a) were prepared from different polymers, such as polysulfone (PSU) (Bae and Tak, 2005), polyethersulfone (PES) (Li et al., 2009), poly(vinylidene fluoride) (PVDF) (Alaoui et al., 2009; Bae and Tak, 2005; Damodar et al., 2009), polyacrylonitrile (PAN) (Bae and Tak, 2005; Kleine et al., 2002; Phonthammachai et al., 2006), cellulose acetate (CA) (Artale et al., 2001; Molinari et al., 2004) and others. Both wet and dry phase inversion methods are used for the membranes’ preparation. In the wet method the casting solution, consisting of polymer, a proper solvent and TiO2 nanoparticles, is cast with a casting knife onto a glass plate, from which the membrane is removed after gelation. Alternatively, the solution can be cast on a support (e.g., polyester non-woven fabric). After casting the membrane is immersed in a coagulation bath with or without solvent evaporation before the immersion. In the dry method a casted membrane is left in air for solvent evaporation until the gelation of the film is completed. Figure 6.23 shows SEM microphotographs of surface of PVDF membranes obtained by the phase

© Woodhead Publishing Limited, 2013

Photocatalytic membrane reactors

263

(a)

Mag = 5.00 KX

2µm

EHT = 4.00 kV Signal A = SE2 Date: 5 Dec 2007 WD = 12 mm Photo No. = 6867

2µm

EHT = 4.00 kV Signal A = SE2 Date: 5 Dec 2007 WD = 12 mm Photo No. = 6846

(b)

Mag = 5.00 KX

6.23 SEM images of membrane surfaces: (a) PVDF membrane; (b) photocatalytic TiO2/PVDF membrane (TiO2/PVDF ratio of 0.5 wt). (Reprinted from Alaoui et al., 2009 with permission from Elsevier.)

inversion technique without modification (a) and modified with TiO2 nanoparticles (b). The amount of TiO2 added to the casting solution must be chosen carefully. However, since the prediction of membrane properties, on the basis of the amount of TiO2 incorporated in its structure, is very difficult, the TiO2 loading is usually adjusted experimentally. From one side, addition of TiO2 might increase membrane porosity. This is due to the fact that TiO2 has higher affinity to water compared to polymer and therefore during the

© Woodhead Publishing Limited, 2013

264

Handbook of membrane reactors Table 6.1 Contact angles of PSU/TiO2 membranes TiO2 content (wt%)

Contact angle (°)

0 1 2 3 5

85 52 41 44 47

Source: Yang et al. (2007).

coagulation step of the phase inversion process the penetration velocity of water into the membrane may increase with TiO2 content. As a result, more porous membranes are formed. However, from the other side, the presence of TiO2 might lead to a decrease of membrane porosity, due to plugging of some membrane pores by the TiO2 particles (Damodar et al., 2009). At too high an amount of a photocatalyst, the membrane performance and stability of the casting solution can be poorer, since nanoparticles plug membrane pores and hinder the interaction between polymer and solvent molecules. In the case of the membranes made of polysulfone the optimum ratio of TiO2/ PSU in the casting solution was found to be 0.3 (Bae and Tak, 2005). In the case of PVDF membranes the optimum polymer/TiO2 ratio amounted to 0.5 (Alaoui et al., 2009). The porosity of the membranes increased up to this value and decreased at higher TiO2 amounts. The increase in porosity with the TiO2/PVDF ratio was attributed to (i) an anatase-promoted increase in the number of the nuclei of polymer-lean phase, which ultimately become the membrane pores; and (ii) to the presence of inter-granular spaces in the embedded anatase aggregates. The decrease in porosity in the case of a TiO2/PVDF ratio higher than 0.5 was explained by insufficient polymer material to form the polymer scaffold of the porous membrane, leading to a system collapse, that is, a decrease in the porosity and the pore size of the whole structure (Alaoui et al., 2009). The amount of TiO2 entrapped in a membrane influences not only membrane porosity but also its hydrophilicity. Contact angle, which represents the wettability of a membrane surface in the case of the membranes with entrapped TiO2, is in general lower compared to the unmodified membranes. Table 6.1 presents contact angles of polysulfone membranes prepared by the phase inversion method and modified with TiO2. The influence of membranes’ modification with TiO2 on their permeability is a complex issue. The permeability strongly depends on the membrane porosity and pore structure, which was discussed above. In most cases when the amount of the entrapped TiO2 is low, an improvement of pure water flux compared to the flux measured for neat membranes is observed. At higher TiO2 concentration the pure water flux of TiO2-entrapped membranes

© Woodhead Publishing Limited, 2013

Photocatalytic membrane reactors

265

Permeate flux J (L/m2h)

550 500 450 400 350 300 250 200 0

1

2 3 TiO2 (wt%)

4

5

6.24 Pure water flux (J) of neat PSU and PSU/TiO2 membranes obtained by phase inversion technique. (Adapted from Yang et al., 2007.)

usually decreases slightly compared to membranes prepared under the same conditions but without addition of a photocatalyst (Yang et al., 2007). This might result from plugging of membrane pores by nanoparticles during the phase inversion (Bae and Tak, 2005). Figure 6.24 shows an example of pure water fluxes of neat and TiO2-entrapped PSU membranes. The method of preparation of polymeric photocatalytic membranes based on bi-axial stretching of a polymer extrusion containing TiO2 as a catalytic filler (b) was applied for production of PTFE membranes (Morris et al., 2004). The obtained membranes contained 2 wt% of TiO2 distributed uniformly within the membrane. The SEM photographs revealed that the structure of the membrane resembled a net in which PTFE strands were knotted at multiple regions. The photocatalyst particles were primarily accumulated in the vicinity of these polymer knots. From the overview presented above it can be found that there are numerous methods of preparation of polymeric photocatalytic membranes. However, it should be stressed that in case of polymer membranes there is always a danger of destruction of the membrane structure by UV light or hydroxyl radicals. This risk is associated with the reactor configuration. Application of a photocatalytic membrane requires irradiation of the membrane itself in order to perform the photodecomposition of pollutants. The lowest UV resistance is exhibited by membranes prepared from polyethersulfone (PES) and polysulfone (PSU) (Chin et al., 2006; Molinari et al., 2000). This can be attributed to the fact that PES and PSU contain sulfone groups which are highly sensitive to UV light. Other membranes exhibiting low UV resistance are those made of polypropylene (PP), polyacrylonitrile

© Woodhead Publishing Limited, 2013

266

Handbook of membrane reactors

(PAN) and CA. In the case of these membranes, irradiation with UV light leads to a breakage of the chemical bonds of the methyl group (–CH–). The least effect of the UV/oxidative environment on the membrane stability was observed in case of polytetrafluoroethylene (PTFE) and poly(vinylidene fluoride) (PVDF) membranes. However, the UV resistance of the membranes should not be generalized from polymer materials, but based on the results obtained for the actual polymer formulations used in practice (e.g., taking into account different additives). This is because membranes made of the same polymer but manufactured by different companies differ in stability under UV light (Chin et al., 2006).

6.6

Inorganic membranes

In contrast to the polymeric membranes, the inorganic membranes are resistant to UV light and oxidation by hydroxyl radicals. There are several methods of preparation of these membranes, including: (a) dip coating (Alem et al., 2009; Choi et al., 2006a, 2006b; Ciston et al., 2008; Ding et al., 2006; Djafer et al., 2010; Ma et al., 2009a, 2009b; Naszályi et al., 2008; Tsuru et al., 2001, 2003, 2006; Wang et al., 2007, 2008; Zhang et al., 2006a, 2006b), (b) hydrothermal synthesis–filtration–calcination method based on filtration of TiO2 nanowire suspension through a glass filter followed by calcination (Zhang et al., 2008a) or hot press process (Zhang et al., 2009), (c) direct grafting of TiO2 nanotubes in the channels of a ceramic membrane (Zhang et al., 2008b). In the dip coating method (a) the photoactive layer is deposited on a porous support, which is usually made of α-alumina (Al2O3), although supports with layered structure, such as the alumina–titania–zirconia (the so-called ATZ) (Ciston et al., 2008) and hydroxyapatite-alumina (HAP/Al2O3) (Ma et al., 2009a) are also applied. The photoactive layers are mostly formed from TiO2, although other photocatalysts, for example, ZnO (Naszályi et al., 2008) are also used. The photocatalytic inorganic membranes can be also prepared from TiO2. In this case, an anatase photoactive layer is deposited on a support prepared from rutile (Wang et al., 2007, 2008). A scheme showing the procedure of preparation of photocatalytic membranes by the dip coating method is presented in Fig. 6.25. The dip coating of the support can be performed by immersing it in a suspension of a photocatalyst in pure water or in water containing additives (e.g., dioctyl sulfosuccinate surfactant (Ciston et al., 2008)). A commercially available TiO2 basic hydrosol or a TiO2 colloidal sol obtained by the sol–gel method from a TiO2 precursor can also be applied. Sometimes binders, such

© Woodhead Publishing Limited, 2013

Photocatalytic membrane reactors Aqueous solution of TiO2 nanoparticles (with or without additives)

267

Ceramic support (e.g. Al2O3 membrane)

Dip coating

Drying

Sintering

Inorganic membrane

6.25 A scheme of preparation procedure of ceramic photocatalytic membranes by the dip coating method.

as hydroxyethyl cellulose (Djafer et al., 2010) or hydroxypropyl cellulose (HPC) and PVA (Alem et al., 2009), are added to the TiO2 sol before the dip coating process. Using PVA as a single additive causes flocculation of the titania sol, therefore to prevent this phenomenon a combination of PVC with HPC is recommended. Addition of such modifiers allows adjustment of the sol viscosity and lowers the sol surface tension. Moreover, in the presence of the binders the strength of the unfired membrane layer increases, which helps in formation of a crack-free membrane. An improvement of porosity of the photoactive layer might be obtained by application of a TiO2 sol containing surfactant as a pore forming agent (Choi et al., 2006a). The presence of templates such as surfactants in sol–gel chemistry plays a crucial role in creating the porous structure of the TiO2 inorganic network, which reduces the hydraulic resistance of TiO2 membranes (Choi et al., 2006b). In order to obtain asymmetric photocatalytic membranes with an intermediate layer between the support and the active top layer, the TiO2 particles with different sizes are used (Fig. 6.26). Such asymmetric membranes are prepared by coating of MF alumina membrane with a mixture of anatase particles (200 nm) and anatase sol solution followed by coating with TiO2 colloidal sol obtained by the sol–gel process (Tsuru et al., 2006). After dip coating the membranes are dried and then sintered at a defined temperature. Both heating rate and calcination temperature significantly influence the membrane structure. The heating rate should be slow enough to prevent cracking of the membrane. Similarly, membrane cracking could be avoided by selection of a proper temperature. Usually, the calcination is performed at 450°C. In general, a temperature below the temperature of phase transition of anatase to rutile should be selected. This is because the

© Woodhead Publishing Limited, 2013

268

Handbook of membrane reactors (a)

Top layer Intermediate layer

Support

(b)

Surface Top layer Intermediate layer

Support 10 kV ×10,000 1 µm

6.26 (a) Scheme of a ceramic asymmetric membrane. (Adapted from Kim and Van der Bruggen, 2010.). (b) SEM image of cross-section of a TiO2 membrane. (Reprinted from Tsuru et al., 2006 with permission from Elsevier.)

stresses generated during the phase transition can cause severe cracking of the membrane. Moreover, anatase exhibits significantly higher photoactivity than rutile, which is a very important factor in the case of photocatalytic membranes. When a ceramic support is dipped into a suspension of photocatalyst particles, the particles are packed on the support surface and are grown into filter cake. It is well known that the permeability decreases with the increase of the thickness of a membrane or, in case of asymmetric membranes, the thickness of a membrane skin. Moreover, in the case of asymmetric photocatalytic membranes, too thick a photoactive filtration layer might be cracked very easily during drying or sintering. On the other hand, if the layer is too thin an incomplete and defective membrane might be formed. Therefore, it is very important to control the thickness of the TiO2 layer. This might be performed for example, by adjusting the dipping time. It was found that, in order to obtain a defect-free titania layer on the top of the support, the dipping time should be at least 30 s (Ding et al., 2006). It was also reported that the optimum thickness at which a crackfree layer is formed is 1 µm (Alem et al., 2009). The influence of dipping time on the photoactive layer thickness is shown in Fig. 6.27. It can be clearly seen that the layer thickness increases linearly with the square root of the dipping time.

© Woodhead Publishing Limited, 2013

269

Thickness

Photocatalytic membrane reactors

(Dipping time)1/2

6.27 The linear relationship between membrane thickness and square root of dipping time. (Adapted from Ding et al., 2006.)

The defects in a membrane structure can be avoided by application of the repeated coating procedure. In general, the more cycles of repeated coating are performed, the smoother the membrane surface could be obtained. It was reported that at least 3 (Choi et al., 2006b) to 6 (Ma et al., 2009b) coating cycles are necessary to fabricate a skin layer with good integrity and smoothness. During the repeated coating procedure the photocatalyst layer formed from first several coating cycles plays the role of an intermediate layer on which the final smooth active layer is created. The amount of a photocatalyst supported on the membrane surface increases linearly with the number of coating cycles, as shown in Fig. 6.28a. However, the dependence of the active layer thickness on the number of cycles is not linear (Fig. 6.28b) because of incorporation of particles forming subsequent layers in the pores of the intermediate layer, as was explained above (Ma et al., 2009b). In order to improve the properties of inorganic photocatalytic membranes prepared by the dip coating technique, in terms of increasing their permeability and reduction of the number of intermediate layers, a modification of the method was proposed (Bosc et al., 2006). This modification was based on improving the membrane porosity by generation of three levels of pores: macropores, mesopores and micropores. In the first step of the preparation process, the stable complex organic–inorganic hybrid suspensions are prepared by mixing a polystyrene latex aqueous suspension, a titania hydrosol and a non-ionic triblock copolymer. These suspensions are then deposited by the dip coating method on alumina supports. During drying of such obtained membranes solvent evaporation induce the formation of spherical micelles by self-assembly of the amphiphilic molecules. Subsequent calcination at 410°C led to formation of spherical macropores and mesopores,

© Woodhead Publishing Limited, 2013

270

Handbook of membrane reactors

TiO2 mass per membrane unit area

(a)

Repeated coating cycles

Thickness of TiO2 layer

(b)

Repeated coating cycles

6.28 The influence of number of coating cycles on (a) mass of TiO2 coated and (b) thickness of TiO2 layer on a ceramic photocatalytic membrane. (Adapted from Ma et al., 2009b.)

which are generated inside the layers by the thermal removal of the polystyrene particles and of the micelles, respectively. The remaining inorganic network exhibits an additional interconnected microporosity resulting from the aggregation of the anatase nanoparticles. An advantage of this proposed method is that the organic–inorganic hybrid suspensions can be deposited directly on a macroporous support, whereas a conventional microporous layer requires an additional intermediate mesoporous layer, which might negatively affect the membrane permeability. Although preparation of inorganic photocatalytic membranes by dip coating is the most popular method of membrane fabrication, several other techniques are also used. In the hydrothermal synthesis – filtration method (b) a suspension of TiO2 nanowires is prepared by the hydrothermal method (Zhang et al., 2008a). Subsequently, a low amount of surfactant is added and such obtained suspension is vacuum-filtered through a glass fiber filter (0.45 µm). During filtration a TiO2 nanowire membrane is formed, owing to the accumulation of TiO2 nanowires on the surface of the glass filter. Residual surfactant left in the membrane is washed away with distilled water. This

© Woodhead Publishing Limited, 2013

Photocatalytic membrane reactors

271

prepared TiO2 nanowire membrane is dried at room temperature. After drying the glass filter is removed. The free-standing TiO2 nanowire membrane is then calcined at a defined temperature (e.g., 700°C). The filtration method of membrane preparation has several advantages, mainly (Zhang et al., 2008a): (1) a homogeneous distribution of TiO2 nanowires can be obtained by filtration; (2) the membrane thickness can be easily controlled by simply adjusting the nanowire concentration and volume of the suspension filtered; (3) the TiO2 nanowire membrane before calcination exhibits high flexibility with no observed change in its shape after repeated flexure, which enables the membrane to be formed into various shapes; (4) calcination above 300°C ensures the membrane retains its desired shape. In order to obtain an asymmetric TiO2 nanowire membrane with hierarchical structure, two types of TiO2 nanowires with different diameters are applied. A scheme explaining the method of preparation of the TiO2 nanowire membrane is shown in Fig. 6.29. In this method a two step filtration is performed. In the first filtration step a supporting layer composed of nanowires having larger diameters is formed. In the second step a separation layer built of nanowires with smaller diameters is placed on the supporting layer. Unlike the previously described method, the membrane after filtration is not calcined but hot-pressed. The supporting layer of the membrane provides mechanical strength, while the functional layer built of smaller nanowires is responsible for the membrane permeability and separation properties (Zhang et al., 2009). Another type of photocatalytic ceramic membranes is TiO2 nanotube membranes prepared by direct grafting of TiO2 nanotubes in the channels of alumina MF membrane (c) (Zhang et al., 2008b). In the process the alumina support is immersed in a supersaturated TiF4 solution for a defined time. A hydrolysis process occurs, leading to formation of TiO2: TiF4 → Ti(OH)4−x → TiO2

[6.10]

Thin crystalline films of TiO2 are grafted through the heterogeneous nucleation in the channels of the alumina membrane, thus forming nanotubes (Fig. 6.30). Since the free-standing TiO2 nanotube membrane is friable it is not recommended to remove the TiO2 nanotubes from the alumina support. The properties of TiO2 nanotube membranes depend on several factors, including solution pH and deposition time. At pH less than 2 the deposition rate is slow and the TiO2 coating is tightly packed, whereas at pH > 2.5 the deposition rate is fast and the TiO2 coating is loose. The deposition time influences the inner diameter of TiO2 nanotubes (Fig. 6.30). A long deposition time reduces the inner diameter of nanotubes by increasing their wall thickness. The inner diameter of nanotubes after 2 h grafting is in the range

© Woodhead Publishing Limited, 2013

272

Handbook of membrane reactors STEP 1 Suspension of nanowires with large diameters

STEP 2 Suspension of nanowires with small diameters

STEP 3

Hot press

Active layer

Photocatalytic membrane

Support

6.29 Preparation of TiO2 nanowire membrane with hierarchical structure (Zhang et al., 2009). Inner diameters Channel

Nanotube

Nanotube

Alumina template before grafting

Short grafting time

Long grafting time

6.30 The influence of grafting time on properties of TiO2 nanotube membranes. With increasing grafting time the walls of TiO2 nanotubes become thicker resulting in the reduction of the inner diameter of the nanotubes. (Adapted from Zhang et al., 2008b.)

© Woodhead Publishing Limited, 2013

Photocatalytic membrane reactors

273

of 50–70 nm. Pure water flux through the TiO2 nanotube membranes is proportional to the grafting time, which is associated with decreasing pore size with elongation of the time (Zhang et al., 2008b).

6.7

Photocatalytic membrane reactors with suspended photocatalyst

The most common configurations of PMRs with suspended photocatalyst are those utilizing pressure driven membrane processes, such as microfiltration (MF), ultrafiltration (UF) and NF. In other types of PMRs, photocatalysis is combined with dialysis, pervaporation (PV) or direct contact membrane distillation (DCMD, MD) (Mozia, 2010). In PMRs with a photocatalyst in suspension, the primary role of a membrane is separation and recovery of photocatalyst particles from the treated solution. Moreover, when the process is used for the degradation of organic pollutants, the membrane should be able to reject the compounds and products/by-products of their decomposition, while if photocatalysis is applied in a synthesis, the role of the membrane is often the separation of the product(s) from the reaction environment (Molinari, 2010). The possibility of separation of organics and products of their degradation depends on the properties of the membrane used and the membrane process applied. The products and by-products of photodegradation of organic contaminants are, in general, low molecular weight compounds. Therefore, in the case of pressure driven membrane processes, NF and RO only might be considered as able to separate the photodecomposition products. In the case of membrane techniques in which the mechanism of separation is other than the sieve one, other properties of reactants, for example volatility, should be taken into consideration. The photocatalytic reaction in PMRs with suspended photocatalyst might be conducted in (a) a feed tank, (b) a membrane module or (c) an additional reservoir (photoreactor) located between the feed tank and the membrane module. In some cases the reaction is conducted in both the membrane module and the feed tank. The above mentioned configurations are the most popular ones; however, one might find modifications of these solutions. In the PMRs with suspended photocatalyst, the light source must be positioned above or inside the element of the membrane installation in which the photocatalytic degradation is carried out. Schematic diagrams of the most common configurations are presented in Fig. 6.31. There are numerous configurations of PMRs, including pressurized and de-pressurized (submerged) systems working in either batch or continuous modes. A detailed description of these configurations is presented in Chapter 21. The present chapter is devoted to the key issues relating to the PMRs with suspended photocatalyst, including: (i) membrane fouling, (ii) separation of photocatalyst particles and (iii) permeate quality.

© Woodhead Publishing Limited, 2013

274

Handbook of membrane reactors (a)

Light source Retentate

Feed

Feed tank (photocatalyst in suspension)

Membrane module Membrane Permeate

(b)

Light source Retentate

Feed

Feed tank (photocatalyst in suspension)

Membrane module Membrane Permeate

(c)

Light source Retentate

Feed Membrane module

Membrane Permeate Feed tank (photocatalyst in suspension)

Additional reservoir (photoreactor)

6.31 The most common configurations of PMRs with suspended photocatalysts: (a) photocatalytic reaction conducted in a membrane module; (b) photocatalytic reaction conducted in a feed tank; (c) photocatalytic reaction conducted in an additional reservoir (photoreactor) located between the feed tank and membrane module.

© Woodhead Publishing Limited, 2013

Photocatalytic membrane reactors

6.7.1

275

Membrane fouling

The major problems occurring in case of PMRs utilizing pressure driven membrane techniques, such as MF, UF and NF, are permeate flux decline and membrane fouling caused by photocatalyst particles. On the other hand, the deterioration of the permeate flux in the presence of TiO2 photocatalyst has not been observed in case of PMRs coupling photocatalysis with dialysis, PV and MD. The reason is the mass transport mechanism in both types of membrane processes. The application of pressure difference as a driving force leads to enhanced transport of photocatalyst particles towards the membrane surface and compression of the TiO2 filtration cake, which eventually results in permeate flux decline. With techniques such as PV, MD or dialysis, mass transport can be achieved without the application of a pressure difference as a driving force. As a result, the main factor responsible for membrane fouling is excluded and, therefore, the deterioration of the permeate flux can be avoided. The severity of membrane fouling caused by photocatalyst particles in case of pressure driven techniques depends on several factors. A brief summary of the influence of the process conditions on the permeate flux deterioration is presented in Table 6.2. The influence of photocatalyst loading on membrane fouling is a complex issue. In general, an increase of photocatalyst concentration in feed solution leads to a more severe decline of permeate flux (Molinari et al., 2006; Shon et al., 2008; Sopajaree et al., 1999a, 1999b; Xue et al., 2008). This is associated with an increase of the cake thickness, which results in higher resistance. During experiments conducted with application of a UF membrane (MWCO 30 000 Da), the permeate flux decreased from 99.1 to 39.8 dm3/m2h with the increase of TiO2 concentration from 0.1 to 1.0 wt% (Xue et al., 2008). However, it was also observed (Xi and Geissen, 2001) that the interdependence of photocatalyst concentration and permeate flux was not straightforward (Fig. 6.32). The decrease of the flux with increasing photocatalyst loading took place only in the case of very low and very high TiO2 concentrations, which was especially noticeable when the pH of the suspension was close to the point of zero charge (pHpzc) of TiO2. In the medium range of TiO2 loadings, very little influence of the presence of the photocatalyst on permeate flux was observed. Such behavior was attributed to the fact that TiO2 particles in a cake layer have a wide size and shape distribution. The boundary layer near the membrane surface contains larger particles only and is highly porous, while the upper layer formed later consists of finer particles and exhibits lower porosity. At relatively low TiO2 concentrations, the boundary layer near the membrane surface is so thin that the finer particles from the upper layer can enter into the pores of the porous lower

© Woodhead Publishing Limited, 2013

276

Handbook of membrane reactors

Table 6.2 The influence of operating variables on permeate flux and membrane fouling in PMRs with suspended photocatalyst utilizing pressure driven membrane techniques Variable

Influence

Photocatalyst concentration Photocatalyst type

More severe decrease of the permeate flux at higher photocatalyst concentrations Photocatalyst particles with larger size exhibit less fouling tendency

Feed solution composition: - NOM Strong interactions between photocatalyst particles and NOM lead to a severe membrane fouling; however, in the presence of UV light NOM is decomposed and flux could be restored - pH Less severe membrane fouling at pH close to the isoelectric point of TiO2 (pH ≈ 6.8) Hydraulic conditions: - Cross-flow velocity At high shear rates the building up of the polarization/gel layer is hindered - Transmembrane pressure High pressure contributes to the deposition of TiO2 cake on the membrane surface especially at low values of cross-flow velocities Operating mode: - Pressurized Significant fouling caused by TiO2 particles Less prone to fouling caused by TiO2 particles - De-pressurized than the pressurized systems (submerged membrane) Operating conditions in membrane modules: - Aeration In submerged systems could improve the process performance; however, at high aeration rates the bubble clouds might attenuate the UV light transmission in the photoreactor - Intermittent permeation In submerged systems reduces accumulation of the photocatalyst particles on the membrane; however, leads to a decrease in water production - Gas backflushing In pressurized systems prevents fouling caused by TiO2 particles Source: Mozia (2010).

one, resulting in a higher cake layer resistance and lower permeate flux (Fig. 6.32). On the other hand, above a certain TiO2 concentration the total cake layer thickness increases and reaches its crucial effect on the cake resistance and the permeate flux decreases (Xi and Geissen, 2001). The main factors affecting the severity of membrane fouling by TiO2 particles were found to be (Table 6.2): • •

composition of the feed solution (e.g., pH, ionic strength, NOM), hydraulic conditions.

© Woodhead Publishing Limited, 2013

Photocatalytic membrane reactors

277

1500

Permeate flux J (dm3/m2h)

Pure water flux pH ≈ pHpzc (pH 6.9) 1000 pH < pHpzc (pH 5.0) 500

0

J Low TiO2 concentrations 0

0.1

J

J

0.2

Medium TiO2 concentrations 0.3

0.4

High TiO2 concentrations 0.5

TiO2 concentraion (wt%)

6.32 The influence of TiO2 concentration in feed on MF permeate flux (based on of Xi and Geissen, 2001).

The TiO2 surface is positively charged at pH < pHpzc, whereas it is negatively charged at pH > pHpzc. This means that, for the zeta potential different from 0 mV, forces of repulsion exist between the particles which hinder aggregation of the particles. However, at a pH near to the point of zero charge (pHpzc ≈ 6.8 for TiO2 P25, Degussa) the photocatalyst particles tend to aggregate, which is beneficial from the permeate flux point of view. This is because the photocatalyst cake layer composed of larger aggregates exhibits higher porosity that with smaller particles, causing less resistance to the permeate flux. The influence of pH on the cake structure explains well the difference between the permeate flux obtained during the MF processes conducted at pH 5.0 and pH 6.9, which can be observed in Fig. 6.32. It was also found that the presence of electrolytes, such as CaCl2, has beneficial effect on the flux (Xi and Geissen, 2001; Xue et al., 2008). During MF of TiO2 suspension it was observed that the steady state flux was four times higher in the process conducted with addition of CaCl2 compared to the one in the absence of the electrolyte. This phenomenon can be explained by the flocculation of TiO2 particles and formation of a cake layer of higher porosity in the presence of salts. The presence of NOM, mainly humic and fulvic acids, in the feed solution was found to be another factor contributing to membrane fouling in PMRs. TiO2 particles or humic acids molecules can be deposited or sorbed on the membrane surface during permeation when they exist as single components (Fig. 6.33a and 6.33b). However, when they are mixed together, humic

© Woodhead Publishing Limited, 2013

278

Handbook of membrane reactors (a) TiO2 only Porous TiO2 cake layer Membrane skin

(b) Humic acids (HA) only

HA gel layer Membrane skin Adsorption of HA onto the pore walls

(c) Mixture of HA and TiO2 without UV HA/TiO2 dense cake layer Membrane skin

6.33 A schematic diagram of possible mechanism of the formation of cake/gel layers on the membrane surface during UF of: (a) TiO2 suspension in water; (b) humic acids (HA) solution and (c) mixture of TiO2 and HA in water (adapted from Lee et al., 2001).

acids can be sorbed onto the TiO2 particles and, additionally, can occupy the vacancies between the TiO2 particles (Fig. 6.33c). Therefore, the cake layer composed of humic acids and TiO2 particles should be denser than that of the TiO2-only cake layer. As a result, an increased resistance of the cake layer, associated with more severe decrease of the permeate flux, is observed (Lee et al., 2001). However, it should be stressed that the application of UV light leads to decomposition of humic substances in the presence of TiO2, and, eventually, the permeate flux can be almost completely recovered (Fig. 6.34) (Bai et al., 2009; Choo et al., 2008a). By taking into account the above considerations, it could be concluded that the influence of TiO2 on permeate flux in PMRs is dual. First, a deterioration of the flux due to the presence of photocatalyst particles in the feed solution might take place. Second, during treatment of water polluted with NOM the application of TiO2 together with UV irradiation might improve the process performance due to removal of high molecular organic compounds which are forming the gel layer on a membrane surface. Other factors influencing the permeate flux in PMRs utilizing pressure driven membrane techniques are those referring to hydraulic conditions in the membrane module (Table 6.2). The most important are: cross-flow

© Woodhead Publishing Limited, 2013

Photocatalytic membrane reactors

279

Permeate flux J

Pure water (i) TiO2 suspension in water (ii) HA + TiO2 + UV (v) HA solution (iii)

HA + TiO2 + (no UV) (iv)

Time

6.34 Schematic diagram showing permeate flux behavior during UF of: pure water, TiO2 suspension in water, HA solution, suspension of TiO2 in humic acid solution without UV irradiation, and suspension of TiO2 in humic acid solution with UV irradiation (based on Lee et al., 2001 and Bai et al., 2009).

velocity of the feed and transmembrane pressure. An increase of the cross-flow velocity might result in minimization of the negative influence of the presence of TiO2 on permeate flux, because at high shear rates the building up of the polarization/gel layer is hindered (Sopajaree et al., 1999a, 1999b; Xue et al., 2008). At low values of cross-flow velocities the transmembrane pressure was found to be an important factor contributing to the deposition of TiO2 cake on a membrane surface. Under such conditions the thickness of the filtration cake increases, resulting in an increased filtration resistance. On the other hand, in the high cross-flow velocity regime, the gel formation was inhibited and transmembrane pressure played a minor role. With unfavorable combination of low cross-flow velocity and high pressure, about 50% of the initial TiO2 amount could be accumulated at the membrane surface (Sopajaree et al., 1999b). The severity of membrane fouling by photocatalyst particles depends also on the operation mode (Table 6.2). The submerged membrane systems are less prone to fouling because of TiO2 particles than the pressurized ones (Molinari et al., 2008), although the fouling might not be totally avoided. The filtration efficiency in the submerged PMRs might be improved by increasing aeration or the application of intermittent permeation. Aeration could reduce the membrane fouling and help to keep the photocatalyst particles well suspended in the solution. However, increasing aeration would result in higher energy consumption and could also attenuate UV transmission in the photoreactor, as shown in Fig. 6.35 (Chin et al., 2007a, 2007b). Therefore, in order to maintain the high permeate flux at low aeration rate, the intermittent permeation might be applied (Chin et al., 2007a, 2007b). When suction is stopped and no permeate is collected, there is a

© Woodhead Publishing Limited, 2013

280

Handbook of membrane reactors

UV light intensity (mW/cm2)

1.4 1.2 1 0.8 0.6 0.4 0.2 0 0

1

2 3 Aeration rate (dm3/min)

4

5

6.35 The influence of aeration rate on UV light transmission in a photoreactor (based on Chin et al., 2007b).

Air bubbles Capillary membranes

Permeate flow direction

TiO2 particles

Detachment of TiO2 particles

Aerator Suction off (no permeate)

6.36 The idea of intermittent permeation in a submerged membrane module. After the suction pump is turned off, the collection of permeate is stopped and removal of photocatalyst particles from membrane surface by air bubbles starts.

period for the aeration to exert shear on the membrane surface to facilitate the detachment of TiO2 particles (Fig. 6.36). This prevents accumulation of the particles on the membrane. However, the intermittent permeation results in a decrease in water production. From a practical point of view, the application of intermittent

© Woodhead Publishing Limited, 2013

Photocatalytic membrane reactors

281

permeation could be beneficial if the intermittence frequency (IF), defined as the ratio of the off-time to on-time, is low enough. The water production rates with intermittence frequency (IF) of 0.4 and 0.1 were found to be approximately 70% and 90% with respect to the continuous production rate (Chin et al., 2007a, 2007b).

6.7.2

Separation of a photocatalyst

The main role of a membrane in the PMRs is separation and recovery of a photocatalyst from the treated solution. The highest risk of leakage of photocatalyst particles to permeate exists with pressure driven techniques such as MF and UF. This is because the pores of the MF and UF membranes are very often larger than the diameters of photocatalyst particles, which additionally exhibit a wide particle size distribution. Due to large membrane pores and application of pressure as a driving force, the photocatalyst could possibly pass through the membrane. However, it was experimentally proved that the pressure driven membrane processes such as MF, UF and NF are very efficient in retention of TiO2 particles. A high effectiveness of TiO2 separation by MF/UF membranes was attributed to the formation of a dense cake layer on the membrane surface (so-called ‘dynamic membrane’), which improved the rejection of smaller particles (Choo et al., 2008b). Efficient separation of TiO2 by MF membranes was also explained by the change of the photocatalyst properties after dosing into the reactor. When staying in the reactor the fine particles aggregate to form larger agglomerates, which are rejected by the membrane (Meng et al., 2005). The membrane techniques, such as dialysis, PV and MD, were also found to be efficient in retention of photocatalyst particles. This is associated mainly with the mass transport mechanism in these processes, which precludes permeation of photocatalyst particles through the membranes used.

6.7.3

Effectiveness of photodegradation and permeate quality

PMRs with suspended photocatalyst have been applied for removal of dyes, pharmaceuticals, humic and fulvic acids, as well as for treatment of real dyeing wastewater or surface waters (Mozia, 2010). The effectiveness of photodegradation and the permeate quality in PMRs strongly depends on the membrane properties and membrane process used, as well as on the process parameters applied. Apart from the factors associated with the photocatalysis, such as for example, light wavelength and intensity, photocatalyst type and loading or solution composition, some additional parameters, associated

© Woodhead Publishing Limited, 2013

282

Handbook of membrane reactors

Table 6.3 Operating variables influencing the permeate quality in different types of PMRs Operating variable

Description

Effectiveness of Related to the factors affecting the photocatalytic photocatalytic process: reactor design, light wavelength and degradation of pollutants intensity, photocatalyst type and loading, pH, oxygen content, initial concentration of pollutants, the presence of ions, etc. In general, the higher the effectiveness of photodegradation, the higher the permeate quality Hydraulic retention time Elongation of the residence time enhances the (residence time) degradation efficiency of organic pollutants and leads to an improvement of permeate quality. Therefore, the lower the permeate flux, the better the permeate quality Operational mode: - Batch vs continuous flow Batch operation was more efficient than the continuous flow one. However, the continuous system is more promising than the batch one due to a possible potential industrial application of the former configuration - Intermittent vs continuous The effect of intermittent permeation on the degradation efficiency and product quality was found to be negligible Type of membrane For the pressure driven processes the quality of process and separation permeate increases with decreasing pore size characteristics of the of the membranes (MF < UF < NF) membrane MD: high quality of permeate since only volatile components could pass through a membrane PV: permeate contains organics (initial compounds and by-products of photocatalytic reaction); retentate is free from organics; however, it requires additional treatment due to the presence of photocatalyst particles Dialysis: both dialysate and retentate need to be further treated Source: Mozia (2010).

with the membrane process itself, should be taken into consideration. They include mainly (Table 6.3): • residence time (hydraulic retention time), which is associated with the permeate flux, • operational mode (e.g., batch vs continuous flow, dead end vs cross-flow, intermittent vs continuous operation, etc.), • separation characteristics of a membrane.

© Woodhead Publishing Limited, 2013

Photocatalytic membrane reactors

Remaining TOC (mg/dm3)

7

100

Permeate flux (dm3/m2h) 60

283

40

6 5 4 3 2 1 0 1.95

3.25

4.87

Residence time (h)

6.37 The influence of residence time on TOC removal in a PMR utilizing MF. (Adapted from Chin et al., 2007b.)

Long residence time enables good degradation efficiency of organic contaminants. In case of the PMRs utilizing membrane techniques which are characterized by high permeate fluxes, such as MF or UF, the residence times are relatively short. Therefore, it is very important to control the hydraulic retention time, especially in the case of these processes. However, monitoring of this parameter in other types of PMRs is also important. The influence of residence time on composition of MF permeate in terms of remaining TOC concentration is shown in Fig. 6.37. Another parameter affecting the treatment efficiency in PMRs is the operational mode (Table 6.3). The PMRs can work either in batch or continuous modes. Sometimes, semi-batch systems are also applied. In batch mode, the feed tank is filled with the treated solution at the beginning of the process and no refilling of the tank before the end of the operation is done. The permeate is collected in a permeate tank and the retentate (concentrate) is recycled back to the feed tank. At the end of the batch process, a small volume of concentrate remains in the feed tank. Then the system is drained, the membranes can be cleaned if necessary, and the tank is refilled with a new batch. A modification of the batch mode is the semi-batch system. In this case the feed tank is refilled with fresh feed solution as the permeate is removed, in order to keep a constant volume in the feed tank. After a defined time, or once a predetermined concentration or flux is reached, the supply of fresh feed is stopped and the remaining solution in the tank is concentrated, as in batch mode. In continuous mode the feed solution is continuously supplied to the feed

© Woodhead Publishing Limited, 2013

284

Handbook of membrane reactors

tank, while permeate and retentate are drained from the system. The rate of feed supply to the feed tank is dependent on the permeate flux and the final concentration desired. In general, continuous flow PMRs are less efficient in photodegradation of organics than batch ones (Li and Zhao, 1999; Molinari et al., 2002a, 2002b). This is associated mainly with shorter residence time of contaminants in the case of the former systems. Batch mode is preferred in applications in which the treated volumes of feed solutions are relatively small. However, when there are large volumes of feed to be treated, the continuous flow PMRs are more promising, even though the effectiveness of purification compared to the batch systems is lower (Molinari et al., 2002a). In the case of PMRs with submerged membranes, the treatment can be conducted either in continuous or in intermittent suction mode. Intermittent permeation could improve permeate flux and reduce membrane fouling, as was described in Section 6.7.1. However, no significant difference between the quality of permeate obtained during continuous and intermittent operations was observed. Similarly, no significant influence of the intermittent frequency (IF) on the efficiency of organics removal in the PMRs was found (Chin et al., 2007a; Ryu et al., 2005). The PMR permeate quality depends also on the membrane properties and membrane process applied. MF and UF membranes are efficient in the separation of photocatalyst particles, but are not able to reject low molecular weight compounds. Therefore, products and by-products of photodegradation towards which the membrane does not exhibit separation properties can freely permeate through the membrane. However, it should be mentioned that the quality of permeate obtained in PMRs is higher than in the case of the MF/UF processes conducted without photocatalysis, because of photodegradation of contaminants in the feed solution. The quality of permeate in PMRs utilizing NF is higher than in systems coupling photo-oxidation with MF or UF. Nonetheless, even in case of NF some of the photodegradation products or the initial compounds having small molecules can be transported through the membranes (Molinari et al., 2004). When NF is considered, it must be remembered, however, that the separation mechanism differs from that in MF and UF. In MF/UF processes, the separation is based on the sieve effect, whereas in the case of NF the solution−diffusion mechanism and the Donnan effect must be taken into account. Due to the Donnan effect repulsive or attractive interactions between the substrate molecules and the membrane surface may occur if the charges are of the same or of different signs, respectively. Under this condition, repulsive interactions increase rejection values, whereas attractive ones decrease them (Molinari et al., 2010). From the second group of membrane techniques applied in PMRs, MD is the one which assures high permeate quality. During MD the retention

© Woodhead Publishing Limited, 2013

Photocatalytic membrane reactors

285

coefficient of non-volatile compounds is (theoretically) equal to 100%. Therefore, only water vapor, some dissolved gases (e.g., CO2) and volatile compounds (initial contaminants and products or by-products of their degradation) can pass through the membrane (Mozia et al., 2009, 2010). An important advantage of the application of MD in PMRs is long residence time of contaminants because of low permeate flux, which assures high photodegradation efficiency. The main application of all the above described PMRs, that is, utilizing MF, UF, NF, RO and MD, is treatment of water/wastewater. The two other types of the hybrid systems, that is, utilizing PV or dialysis, have more potential for being applied in other processes. In case of the PMR utilizing PV, the product of the process could be feed solution free from organic contaminants (Camera-Roda and Santarelli, 2007) or permeate containing desired volatile compounds being products of a photocatalytic reaction (Camera-Roda et al., 2011). The composition of permeate in the case of this PMR depends on both the course of the photocatalytic reaction and the separation factor of the pervaporation membrane used. Since the products of photodegradation, which could slow down the rate of the photocatalytic reaction, are continuously removed from the reaction environment, an enhancement of the process is observed. It was also reported that operating at high values of a coefficient R, defined as the ratio of the characteristic rate of permeation and the characteristic rate of the photocatalytic oxidation, could result in an improvement of the recovery of the desired product. The ratio R could be increased by increasing the area of the membrane (Camera-Roda and Santarelli, 2007; Camera-Roda et al., 2011). In the case of a PMR utilizing dialysis, both dialysate and retentate can be regarded as products of the process, although both streams need further treatment (compare with Section 21.5). An important factor affecting the process performance is selection of process parameters assuring equal rates of the mass transfer through a membrane and the photodegradation of contaminants. This is because the degradation is performed in the dialysate compartment to which the organics are transported by diffusion from the feed compartment (Azrague et al., 2006).

6.8

Conclusions and future trends

Heterogeneous photocatalysis is a powerful tool for the abatement of organic and inorganic pollutants as well as for the synthesis of useful compounds. For successful application of photocatalysis in purification or production processes, a deep understanding of its mechanistic aspects and knowledge on parameters influencing reaction pathways and efficiency is necessary.

© Woodhead Publishing Limited, 2013

286

Handbook of membrane reactors

One of the key issues in the application of photocatalysis is separation and recovery of photocatalyst from the reaction media. A promising solution for this is coupling of photocatalysis with membrane technology. PMRs are systems in which the advantages of both photocatalytic reaction and membrane separation are combined. Due to the application of membrane processes, not only recovery and reuse of the photocatalyst but also elongation of the residence time of the substrates in the reactor, as well as selective separation of the reaction products, is possible. Both photocatalytic membranes and photocatalyst in suspension can be applied in PMRs. In the former case, the photocatalytic reaction occurs on the external surface and within the pores of a membrane, while reactants permeate in a one-pass flow. Therefore, the element which has to be irradiated is the membrane itself. This means that the membranes should be prepared from materials resistant to UV irradiation and oxidation by hydroxyl radicals or other ROS. Since the stability of polymer membranes under such conditions is limited, more promising membranes for PMRs are ceramic ones. Different methods of preparation of photocatalytic membranes have been developed. In the case of polymer membranes they include (i) methods in which a photocatalyst is supported on a membrane surface and (ii) methods in which a photocatalyst is being entrapped within a membrane structure. In the case of ceramic membranes the photocatalyst can be supported on a membrane, entrapped in it, or the membrane can be made from the photocatalytic material only. The stability and photoactivity of the photocatalytic membranes depend not only on the membrane material (polymer or ceramic) but also on parameters applied during photocatalyst immobilization. More common configurations of PMRs are systems with suspended photocatalyst. In these PMRs various membrane techniques are utilized: MF, UF, NF, dialysis, PV, or MD. The photocatalytic reaction in PMRs with suspended photocatalyst might be conducted in (a) a feed tank, (b) a membrane module or (c) an additional reservoir (photoreactor) which is located between the feed tank and the membrane module. Thus, the danger of polymer membrane destruction by UV light could be avoided by a proper selection of reactor configuration. The major problems occurring in the case of PMRs utilizing pressure driven membrane techniques, such as MF, UF and NF, are permeate flux decline and membrane fouling caused by photocatalyst particles. On the other hand, the deterioration of the permeate flux in the presence of TiO2 is not observed in the case of PMRs coupling photocatalysis with dialysis, PV and MD. These differences result from the various mechanisms of mass transport in both types of membrane processes.

© Woodhead Publishing Limited, 2013

Photocatalytic membrane reactors

287

The severity of membrane fouling in PMRs with suspended photocatalyst depends on several factors, including photocatalyst type and loading, feed solution composition, hydraulic conditions or operating mode. The extent of flux deterioration might be controlled by proper selection of process parameters. The effectiveness of photodegradation and the permeate quality in PMRs strongly depend on the membrane properties and membrane process used, as well as on the process parameters applied. Except from the factors associated with the photocatalysis, such as e.g. light wavelength and intensity, photocatalyst type and loading or solution composition, some additional parameters, associated with the membrane process are also important. They include mainly: residence time, operational mode and separation characteristics of a membrane. This overview of the present state of the art in the area of PMRs has revealed that further studies are still necessary to improve the properties of photocatalytic membranes, as well as to overcome problems associated with membrane fouling and permeate quality in PMRs with suspended photocatalyst. Nonetheless, PMRs can be considered as promising systems in water/ wastewater purification as well as in the synthesis of useful compounds.

6.9

References

Alaoui O T, Nguyen Q T, Mbareck Ch and Rhlalou T (2009), ‘Elaboration and study of poly(vinylidene fluoride)–anatase TiO2 composite membranes in photocatalytic degradation of dyes’, Appl Catal A-Gen, 358, 13–20. Alem A, Sarpoolaky H and Keshmiri M (2009), ‘Titania ultrafiltration membrane: Preparation, characterization and photocatalytic activity’, J Eur Ceram Soc, 29, 629–635. Artale M A, Augugliaro V, Drioli E, Golemme G, Grande C, Loddo V, Molinari R, Palmisano L and Schiavello M (2001), ‘Preparation and characterisation of membranes with entrapped TiO2 and preliminary photocatalytic tests’, Annali di Chimica, 91, 127–136. Augugliaro V, Loddo V, Pagliaro M, Palmisano G and Palmisano L (2010), Clean by light irradiation. Practical applications of supported TiO2, R.S.C. Publishing, Cambridge. Azrague K, Aimar P, Benoit-Marquié F and Maurette M T (2007), ‘A new combination of a membrane and a photocatalytic reactor for the depollution of turbid water’, Appl Catal B-Environ, 72, 197–205. Bae T-H and Tak T-M (2005), ‘Effect of TiO2 nanoparticles on fouling mitigation of ultrafiltration membranes for activated sludge filtration’, J Membrane Sci, 249, 1–8. Bai H, Zhang X, Pan J, Sun D D and Shao J (2009), ‘Combination of nano TiO2 photocatalytic oxidation with microfiltration for NOM removal’, Water Sci Technol: Water Supply, 9(1), 31–37.

© Woodhead Publishing Limited, 2013

288

Handbook of membrane reactors

Bai H, Liu Z and Sun D D (2010), ‘Hierarchically multifunctional TiO2 nano-thorn membrane for water purification’, Chem Commun, 46, 6542–6544. Bard A J (1979), ‘Photoelectrochemistry and heterogeneous photo-catalysis at semiconductors’, J Photochem, 10, 59–75. Barni B, Cavicchioli A, Riva E, Zanoni L, Bignoli F and Bellobono I R (1995), ‘Laboratory-scale photodegradation of phenol in aqueous solution by photocatalytic membranes immobilizing titanium dioxide’, Chemosphere, 30, 1847–1860. Bellobono I R, Bonardi M, Castellano L, Selli E and Righetto L (1992), ‘Degradation of some chloro-aliphatic water contaminants by photocatalytic membranes immobilizing titanium dioxide’, J Photoch Photobio A, 67, 109–115. Bellobono I R, Carrara A, Barni B and Gazzotti A (1994), ‘Laboratory- and pilot-plant-scale photodegradation of chloroaliphatics in aqueous solution by photocatalytic membranes immobilizing titanium dioxide’, J Photoch Photobio A, 84, 83–90. Bellobono I R, Morazzoni F, Bianchi R, Mangone E S, Stanescu R, Costache C and Tozzi P M (2005), ‘Laboratory-scale photomineralisation of n-alkanes in aqueous solution, by photocatalytic membranes immobilising titanium dioxide’, Int J Photoenergy, 7, 79–85. Bellobono I R, Stanescu R, Costache C, Canevali C, Morazzoni F, Scotti R, Bianchi R, Mangone E S, de Martini G and Tozzi P M (2006a), ‘Laboratory-scale photomineralization of n-alkanes in gaseous phase by photocatalytic membranes immobilizing titanium dioxide’, Int J Photoenergy, 2006, 1–8, Article ID 73167. Bellobono I R, de Martini G, Tozzi P M, Canevali C, Morazzoni F, Scotti R and Bianchi R (2006b), ‘Modelling of quantum yields in photocatalytic membrane reactors immobilising titanium dioxide’, Int J Photoenergy, 2006, 1–8, Article ID 26870. Bellobono I R, Rossi M, Testino A, Morazzoni F, Bianchi R, de Martini G, Tozzi P M, Stanescu R, Costache C, Bobirica L, Bonardi M L and Groppi F (2008), ‘Influence of irradiance, flow rate, reactor geometry, and photopromoter concentration in mineralization kinetics of methane in air and in aqueous solutions by photocatalytic membranes immobilizing titanium dioxide’, Int J Photoenergy, 2008, 1–14, Article ID 283741. Bosc F, Lacroix-Desmazes P and Ayral A (2006), ‘TiO2 anatase-based membranes with hierarchical porosity and photocatalytic properties’, J Colloid Interf Sci, 304, 545–548. Braun A M, Maurette M T and Oliveros E (1991), Photochemical technology, John Wiley & Sons, New York. Camera-Roda G and Santarelli F (2007), ‘Intensification of water detoxification by integrating photocatalysis and pervaporation’, J Sol Energ – T ASME, 129, 68–73. Camera-Roda G, Santarelli F, Augugliaro V, Loddo V, Palmisano G, Palmisano L and Yurdakal S (2011), ‘Photocatalytic process intensification by coupling with pervaporation’, Catal Today, 161, 209–213. Chin S S, Chiang K and Fane A G (2006), ‘The stability of polymeric membranes in a TiO2 photocatalysis process’, J Membrane Sci, 275, 202–211.

© Woodhead Publishing Limited, 2013

Photocatalytic membrane reactors

289

Chin S S, Lim T M, Chiang K and Fane A G (2007a), ‘Hybrid low-pressure submerged membrane photoreactor for the removal of bisphenol A’, Desalination, 202, 253–261. Chin S S, Lim T M, Chiang K and Fane A G (2007b), ‘Factors affecting the performance of a low-pressure submerged membrane photocatalytic reactor’, Chem Eng J, 130, 53–63. Choi H, Stathatos E and Dionysiou D D (2006a), ‘Sol–gel preparation of mesoporous photocatalytic TiO2 films and TiO2/Al2O3 composite membranes for environmental applications’, Appl Catal B-Environ, 63, 60–67. Choi H, Sofranko A C and Dionysiou D D (2006b), ‘Nanocrystalline TiO2 photocatalytic membranes with a hierarchical mesoporous multilayer structure: synthesis, characterization, and multifunction’, Adv Funct Mater, 16, 1067–1074. Choo K-H, Ran T and Kim M-J (2008a), ‘Use of a photocatalytic membrane reactor for the removal of natural organic matter in water: Effect of photoinduced desorption and ferrihydrite adsorption’, J Membrane Sci, 322, 368–374. Choo K-H, Chang D-I, Park K-W and Kim M-H (2008b), ‘Use of an integrated photocatalysis/hollow fiber microfiltration system for the removal of trichloroethylene in water’, J Hazard Mater, 152, 183–190. Ciston S, Lueptow R M and Gray K A (2008), ‘Bacterial attachment on reactive ceramic ultrafiltration membranes’, J Membrane Sci, 320, 101–107. Ciston S, Lueptow R M and Gray K A (2009), ‘Controlling biofilm growth using reactive ceramic ultrafiltration membranes’, J Membrane Sci, 342, 263–268. Damodar R A, You S-J and Chou H-H (2009), ‘Study the self cleaning, antibacterial and photocatalytic properties of TiO2 entrapped PVDF membranes’, J Hazard Mater, 172, 1321–1328. Dekker A J (1957), Solid State Physics, Prentice-Hall, Englewood Cliffs, New Jersey. Ding X, Fan Y and Xu N (2006), ‘A new route for the fabrication of TiO2 ultrafiltration membranes with suspension derived from a wet chemical synthesis’, J Membrane Sci, 270, 179–186. Djafer L, Ayral A and Ouagued A (2010), ‘Robust synthesis and performance of a titania-based ultrafiltration membrane with photocatalytic properties’, Sep Purif Technol, 75, 198–203. Fox M A and Dulay M T (1993), ‘Heterogeneous photocatalysis’, Chem Rev, 93 341–357. Fujishima A, Hashimoto K and Watanabe T (1999), TiO2 Photocatalysis: Fundamentals and Applications, Tokyo, Bkc. Gerischer H (1970), ‘Semiconductor Electrochemistry’ in Eyring H, Henderson D and Host W, Physical Chemistry, An Advanced Treatise, Vol. IX, Academic Press, New York. Gu Y and Meng G (1999), ‘A model for ceramic membrane formation by dip-coating’, J Eur Ceram Soc, 19, 1961–1966. Hoffmann M R, Martin T S, Choi W and Bahnemann D W (1995), ‘Environmental applications of semiconductor photocatalysis’, Chem Rev, 95, 69–96. Ishikawa T (2004), ‘Photocatalytic fiber with gradient surface structure produced from a polycarbosilane and its applications’, Int J Appl Ceram Tec, 1, 49–55.

© Woodhead Publishing Limited, 2013

290

Handbook of membrane reactors

Kim S H, Kwak S-Y, Sohn B-H and Park T H (2003), ‘Design of TiO2 nanoparticle self-assembled aromatic polyamide thin-film-composite (TFC) membrane as an approach to solve biofouling problem’, J Membrane Sci, 211, 157–165. Kim J and Van der Bruggen B (2010), ‘The use of nanoparticles in polymeric and ceramic membrane structures: Review of manufacturing procedures and performance improvement for water treatment’, Environ Pollut, 158, 2335–2349. Kleine J, Peinemann K-V, Schuster C and Warnecke H-J (2002), ‘Multifunctional system for treatment of wastewaters from adhesive-producing industries: separation of solids and oxidation of dissolved pollutants using doted microfiltration membranes’, Chem Eng Sci, 57, 1661–1664. Kwak S-Y, Kim S H and Kim S S (2001), ‘Hybrid organic/inorganic reverse osmosis (RO) membrane for bactericidal anti-fouling. 1. Preparation and characterization of TiO2 nanoparticle self-assembled aromatic polyamide thin-film-composite (TFC) membrane’, Environ Sci Technol, 35, 2388–2394. Lee S-A, Choo K-H, Lee Ch-H, Lee H-I, Hyeon T, Choi W and Kwon H-H (2001), ‘Use of ultrafiltration membranes for the separation of TiO2 photocatalysts in drinking water treatment’, Ind Eng Chem Res, 40, 1712–1719. Li X Z and Zhao Y G (1999), ‘Advanced treatment of dyeing wastewater for reuse’, Water Sci Technol, 39, 249–255. Li J-F, Xu Z-L, Yang H, Yu L-Y and Liu M (2009), ‘Effect of TiO2 nanoparticles on the surface morphology and performance of microporous PES membrane’, Appl Surf Sci, 255, 4725–4732. Liu L, Chen F and Yang F (2009), ‘Stable photocatalytic activity of immobilized Fe0/ TiO2/ACF on composite membrane in degradation of 2,4-dichlorophenol’, Sep Purif Technol, 70, 173–178. Lohmann F (1967), ‘Fermi-Niveau und Flachbandpotential von Molekülkristallen aromatischer Kohlenwasserstoffe’, Z Naturforsch, A22, 843–844. Ma N, Fan X, Quan X and Zhang Y (2009a), ‘Ag–TiO2/HAP/Al2O3 bioceramic composite membrane: Fabrication, characterization and bactericidal activity’, J Membrane Sci, 336, 109–117. Ma N, Quan X, Zhang Y, Chen S and Zhao H (2009b), ‘Integration of separation and photocatalysis using an inorganic membrane modified with Si-doped TiO2 for water purification’, J Membrane Sci, 335, 58–67. Madaeni S S and Ghaemi N (2007), ‘Characterization of self-cleaning RO membranes coated with TiO2 particles under UV irradiation’, J Membrane Sci, 303, 221–233. Mansourpanah Y, Madaeni S S, Rahimpour A, Farhadian A and Taheri A H (2009), ‘Formation of appropriate sites on nanofiltration membrane surface for binding TiO2 photo-catalyst: Performance, characterization and fouling-resistant capability’, J Membrane Sci, 330, 297–306. Meng Y, Huang X, Yang Q, Qian Y, Kubota N and Fukunaga S (2005), ‘Treatment of polluted river water with a photocatalytic slurry reactor using low-pressure mercury lamps coupled with a membrane’, Desalination, 181, 121–133. Mills A and Le Hunte S (1997), ‘An overview of semiconductor photocatalysis’, J Photoch Photobio A, 108, 1–35. Molinari R, Mungari M, Drioli E, Di Paola A, Loddo V, Palmisano L and Schiavello M (2000), ‘Study on a photocatalytic membrane reactor for water purification’, Catal Today, 55, 71–78.

© Woodhead Publishing Limited, 2013

Photocatalytic membrane reactors

291

Molinari R, Palmisano L, Drioli E and Schiavello M (2002a), ‘Studies on various reactor configurations for coupling photocatalysis and membrane processes in water purification’, J Membrane Sci, 206, 399–415. Molinari R, Borgese M, Drioli E, Palmisano L and Schiavello M (2002b), ‘Hybrid processes coupling photocatalysis and membranes for degradation of organic pollutants in water’, Catal Today, 75, 77–85. Molinari R, Pirillo F, Falco M, Loddo V and Palmisano L (2004), ‘Photocatalytic degradation of dyes by using a membrane reactor’, Chem Eng Process, 43, 1103–1114. Molinari R, Pirillo F, Loddo V and Palmisano L (2006), ‘Heterogeneous photocatalytic degradation of pharmaceuticals in water by using polycrystalline TiO2 and a nanofiltration membrane reactor’, Catal Today, 118, 205–213. Molinari R, Caruso S, Argurio P and Poerio T (2008), ‘Degradation of the drugs Gemfibrozil and Tamoxifen in pressurized and de-pressurized membrane photoreactors using suspended polycrystalline TiO2 as catalyst’, J Membrane Sci, 319, 54–63. Molinari R, Caruso A and Palmisano L (2010), ‘Photocatalytic processes in membrane reactors’, in Drioli E and Giorno L, Comprehensive membrane science and engineering, vol. 3, Oxford, Academic Press, Elsevier B V Amsterdam, Netherlands, 165–193. Moroni A, Bellobono I R and Gawlik B M (1999), ‘Elementary steps of reaction pathway in the pilot plant photomineralisation of s-triazines on to photocatalytic membranes immobilising titanium dioxide and promoting photocatalysts’ in Froment G F and Waugh K C, Reaction Kinetics and the Development of Catalytic Processes, Elsevier Science B.V., 385–392. Morris R E, Krikanova E and Shadman F (2004), ‘Photocatalytic membrane for removal of organic contaminants during ultra-purification of water’, Clean Techn Environ Policy, 6, 96–104. Mozia S, Tomaszewska M and Morawski A W (2005), ‘A new photocatalytic membrane reactor (PMR) for removal of azo-dye Acid Red 18 from water’, Appl Catal B-Environ, 59, 131–137. Mozia S and Morawski A W (2006), ‘Hybridization of photocatalysis and membrane distillation for purification of wastewater’, Catal Today, 118, 181–188. Mozia S, Tomaszewska M and Morawski A W (2007), ‘Photocatalytic membrane reactor (PMR) coupling photocatalysis and membrane distillation—Effectiveness of removal of three azo dyes from water’, Catal Today, 129, 3–8. Mozia S, Morawski A W, Toyoda M and Tsumura T (2009), ‘Effect of process parameters on photodegradation of Acid Yellow 36 in a hybrid photocatalysis–membrane distillation system’, Chem Eng J, 150, 152–159. Mozia S (2010), ‘Photocatalytic membrane reactors (PMRs) in water and wastewater treatment. A review’, Sep Purif Technol, 73, 71–91. Mozia S, Morawski A W, Toyoda M and Tsumura T (2010), ‘Integration of photocatalysis and membrane distillation for removal of mono- and poly-azo dyes from water’, Desalination, 150, 666–672. Naszályi L, Bosc F, El Mansouri A, van der Lee A, Cot D, Hórvölgyi Z and Ayral A (2008), ‘Sol–gel-derived mesoporous SiO2/ZnO active coating and development of multifunctional ceramic membranes’, Sep Purif Technol, 59, 304–309. Ollis D F, Pelizzetti E and Serpone N (1989), ‘Heterogeneous photocatalysis in the environment: application to water purification’ in Serpone N and Pelizzetti E

© Woodhead Publishing Limited, 2013

292

Handbook of membrane reactors

Photocatalysis: Fundamentals and Applications, Wiley Interscience, New York, 603–637. Palmisano G, Addamo M, Augugliaro V, Caronna T, Di Paola A, García-López E, Loddo V, Marcì G, Palmisano L and Schiavello M (2007a), ‘Selectivity of hydroxyl radical in the partial oxidation of aromatic compounds in heterogeneous photocatalysis’, Catal Today, 122, 118–127. Palmisano G, Yurdakal S, Augugliaro V, Loddo V and Palmisano L (2007b), ‘Photocatalytic selective oxidation of 4-methoxybenzyl alcohol to aldehyde in aqueous suspension of home-prepared TiO2 catalyst’, Adv Synth Catal, 349, 964–970. Phonthammachai N, Gulari E, Jamieson A M and Wongkasemjit S (2006), ‘Photocatalytic membrane of a novel high surface area TiO2 synthesized from titanium triisopropanolamine precursor’, Appl Organometal Chem, 20, 499–504. Pleskov Y V and Gurevich Y Y (1986), Semiconductor Photoelectrochemistry, Consultants Bureau, New York. 101. Ryu J, Choi W and Choo K-H (2005), ‘A pilot-scale photocatalyst-membrane hybrid reactor: performance and characterization’, Water Sci Technol, 51, 491–497. Schiavello M (1997), Heterogeneous photocatalysis, Vol. 3, Wiley Series in Photoscience and Photoengineering, Chichester, John Wiley & Sons. Shon H K, Phuntsho S and Vigneswaran S (2008), ‘Effect of photocatalysis on the membrane hybrid system for wastewater treatment’, Desalination, 225, 235–248. Sclafani A, Palmisano L and Schiavello M (1990), ‘Influence of the preparation methods of TiO2 on the photocatalytic degradation of phenol in aqueous dispersion’, J Phys Chem, 94, 829–832. Sopajaree K, Qasim S A, Basak S and Rajeshwar K (1999a), ‘An integrated flow reactor-membrane filtration system for heterogeneous photocatalysis. Part I: Experiments and modelling of a batch-recirculated photoreactor’, J Appl Electrochem, 29, 533–539. Sopajaree K, Qasim S A, Basak S and Rajeshwar K (1999b), ‘An integrated flow reactor-membrane filtration system for heterogeneous photocatalysis. Part II: Experiments on the ultrafiltration unit and combined operation’, J Appl Electrochem, 29, 1111–1118. Tsuru T, Toyosada T, Yoshioka T and Asaeda M (2001), ‘Photocatalytic reactions in a filtration system through porous titanium dioxide membranes’, J Chem Eng Japan, 34, 844–847. Tsuru T, Kan-no T, Yoshioka T and Asaeda M (2003), ‘A photocatalytic membrane reactor for gas-phase reactions using porous titanium oxide membranes’, Catal Today, 82, 41–48. Tsuru T, Kan-no T, Yoshioka T and Asaeda M (2006), ‘A photocatalytic membrane reactor for VOC decomposition using Pt-modified titanium oxide porous membranes’, J Membrane Sci, 280, 156–162. Turchi C and Ollis D F (1990), ‘Photocatalytic degradation of organic water contaminants: Mechanisms involving hydroxyl radical attack’, J Catal, 122, 178–192. Vankelecom I F J (2002), ‘Polymeric membranes in catalytic reactors’, Chem Rev, 102, 3779–3810. Wang Y H, Liu X Q and Meng G Y (2007), ‘Preparation of asymmetric pure titania ceramic membranes with dual functions’, Mat Sci Eng A, 445–446, 611–619.

© Woodhead Publishing Limited, 2013

Photocatalytic membrane reactors

293

Wang Y H, Liu X Q and Meng G Y (2008), ‘Preparation and properties of supported 100% titania ceramic membranes’, Mater Res Bull, 43, 1480–1491. Xi W and Geissen S-U (2001), ‘Separation of titanium dioxide from photocatalytically treated water by cross-flow microfiltration’, Water Res, 35, 1256–1262. Xue X-d, Fu J-f, Zhu W-f and Guo X-ch (2008), ‘Separation of ultrafine TiO2 from aqueous suspension and its reuse using cross-flow ultrafiltration (CFU)’, Desalination, 225, 29–40. Yang Y, Zhang H, Wang P, Zheng Q and Li J (2007), ‘The influence of nano-sized TiO2 fillers on the morphologies and properties of PSF UF membrane’, J Membrane Sci, 288, 231–238. Yurdakal S, Palmisano G, Loddo V, Augugliaro V and Palmisano L (2008), ‘Nanostructured rutile TiO2 for selective photocatalytic oxidation of aromatic alcohols to aldehydes in water’, J Am Chem Soc, 130, 1568–1569. Yurdakal S, Palmisano G, Loddo V, Alagöz O, Augugliaro V and Palmisano L (2009), ‘Selective photocatalytic oxidation of 4-substituted aromatic alcohols in water with rutile TiO2 prepared at room temperature’, Green Chem, 11, 510–516. Zhang H, Quan X, Chen S, Zhao H and Zhao Y (2006a), ‘Fabrication of photocatalytic membrane and evaluation its efficiency in removal of organic pollutants from water’, Sep Purif Technol, 50, 147–155. Zhang H, Quan X, Chen S and Zhao H (2006b), ‘Fabrication and characterization of silica/titania nanotubes composite membrane with photocatalytic capability’, Environ Sci Technol, 40, 6104–6109. Zhang X, Dua A J, Lee P, Sun D D and Leckie J O (2008a), ‘TiO2 nanowire membrane for concurrent filtration and photocatalytic oxidation of humic acid in water’, J Membrane Sci, 313, 44–51. Zhang X, Dua A J, Lee P, Sun D D and Leckie J O (2008b), ‘Grafted multifunctional titanium dioxide nanotube membrane: Separation and photodegradation of aquatic pollutant’, Appl Catal B-Environ, 84, 262–267. Zhang X, Zhang T, Ng J and Sun D D (2009), ‘High-performance multifunctional TiO2 nanowire ultrafiltration membrane with a hierarchical layer structure for water treatment’, Adv Funct Mater, 19, 3731–3736.

6.10

Appendix: nomenclature

6.10.1

Notation

e(− E

electron in CB energy

)

EC

energy of CB edge

EF0 EF,redox

Fermi level Fermi level of redox couple

Eg EV

band gap energy (eV) energy of VB edge

h(+

H

)

hole in VB Planck constant (6.62606896 × 10−34 J·s)

© Woodhead Publishing Limited, 2013

294

Handbook of membrane reactors

ħ k

reduced Planck constant (h/2π) Boltzmann constant (1.380650 × 10−23 J/K)

k m*e m*h

wave number effective mass of electrons at the end of CB effective mass of electrons at the beginning of VB

p pHpzc

electron momentum point of zero charge

V0 VFB

reduction potential flat band potential

Greek symbols µ

chemical potential

6.10.2

Abbreviations

ACF AM ATZ BET CA CB CSTR DCMD GA HAP HPC IF IR LED MD MF MWCO NF NHE NOM PAN PES PET PFR PI PMR

activated carbon fibers air mass alumina–titania–zirconia Brunauer-Emmett-Teller cellulose acetate conduction band continuous flow stirred tank reactor direct contact membrane distillation glutaraldehyde hydroxyapatite hydroxypropyl cellulose intermittence frequency infrared light-emitting diode membrane distillation microfiltration molecular weight cut off nanofiltration normal hydrogen electrode natural organic matter polyacrylonitrile polyethersulfone polyethylene terephthalate plug flow reactor polyimide photocatalytic membrane reactor

© Woodhead Publishing Limited, 2013

Photocatalytic membrane reactors PP PSU PTFE PV PVA PVDF RO ROS SEM TOC UF UV VB

polypropylene polysulfone polytetrafluoroethylene pervaporation polyvinyl alcohol poly(vinylidene fluoride) reverse osmosis reactive oxygen species scanning electron microscopy total organic carbon ultrafiltration ultraviolet valence band

© Woodhead Publishing Limited, 2013

295

7 Integrating different membrane operations and combining membranes with conventional separation techniques in industrial processes A. CASSANO and A. BASILE, ITM-CNR, Italy

DOI: 10.1533/9780857097347.1.296 Abstract: Various membrane operations are available today for a wide spectrum of industrial applications. Microfiltration (MF), ultrafiltration (UF), nanofiltration (NF), reverse osmosis (RO), electrodialysis and gas separation are well-known membrane unit operations with high potentiality in molecular separations, clarifications, fractionations and concentrations both in liquid and gas phases. In this chapter the possibility of integrating different membrane unit operations in the same industrial cycle or in combination with conventional separation systems is analysed and discussed. Many original solutions in water desalination, agro-food productions and wastewater treatments are reviewed highlighting the advantages achievable in terms of product quality, compactness, rationalization and optimization of productive cycles, reduction of environmental impact and energy saving. Moreover, there is a potential application of polymeric membranes for integrated gasification combined cycle (IGCC) power plants, some aspects of the integration of a membrane reactor with a fuel cell, the possibility to integrate a membrane reformer into a solar system, and the potential application of membrane integrated systems in the fusion reactor fuel cycle, which are attracting many scientists and so will also be introduced and discussed in this chapter. Key words: water desalination, wastewater treatment, integrated membrane systems, agro-food production, gasification, fuel cell, solar membrane reformer.

7.1

Introduction

Membrane unit operations such as microfiltration (MF), ultrafiltration (UF), nanofiltration (NF), reverse osmosis (RO), electrodialysis (ED) and gas separation (GS) are well-established applications today at industrial level offering interesting opportunities in the rationalization and optimization of productive cycles. 296 © Woodhead Publishing Limited, 2013

Integrating different membrane operations

297

In addition, the possibility of integrating different membrane operations in the same process, or in combination with traditional separation units, offers significant advantages in terms of product quality, plant compactness, environmental impact, recovery of high added value substances and energy consumption (Drioli and Romano, 2001; Drioli and Fontananova, 2004, 2009). In fact, in order to take advantage of the benefits of each technology, in many industrial membrane applications, membranes are today being used in conjunction with other conventional separation techniques (Baker, 2004). For this reason, it is very important to study and optimize the ways in which membrane operations and traditional separation units can be combined. For example, membranes can be integrated into power plants in a variety of ways, but they are most effective when applied to concentrated, high pressure streams due to the greater driving forces for separation. The use of both conventional and membrane technologies gives the potential for eliminating the drawbacks of the single process. An excellent example in this direction is the application of polymeric membrane into IGCC power plants (Section 7.5). Moreover, the possibility to apply integrated systems in which all the steps of the productive cycle are based on molecular membrane separations can be considered a valid approach for sustainable industrial growth within the process intensification strategy. The aim of this strategy is to introduce into the productive cycles new technologies characterized by low hindrance volume, advanced levels of automation capacity, modularity, remote control, and reduced energy consumption or waste production (Stankiewicz and Moulijn, 2000). Various chapters on membrane reactors (MR) consider different aspects of the integration of membranes with other conventional systems: pervaporation, zeolite, bioreactors, fuel cells, wastewater treatment, systems for electrical energy, and so on. However, among the various possible examples not cited in these chapters, in the following, due to the lack of space, only seven, but very interesting, case studies are taken in consideration.

7.2

Water desalination

Desalination of saline water (sea and brackish waters) is a well-established means of water supply in many countries. Basically, desalination processes in this area can be divided into two groups: (1) phase-change/thermal, and (2) membrane-based separation processes. Phase-change processes include multi-stage flash, multiple effect boiling, vapour compression, freezing, humidification/dehumidification and solar stills. RO, ED and membrane distillation (MD) are typical membrane separation processes (Charcosset, 2009). RO is a pressure-driven separation technique based on the use of semi-permeable membranes with high permeability for water and low

© Woodhead Publishing Limited, 2013

298

Handbook of membrane reactors

Table 7.1 Key operational data of thermal and membrane-based desalination systems Operational data

MSF

RO

ED

Thermal energy consumption (kWh m−3) Electrical energy (kWh m−3) Typical salt content of raw water (g L−1) Product water quality (ppm TDS)

12 35 30–100 99%; SCO > 98% JO2 = 0.3 ml/cm2.min

>1000 h

Balachandran et al., 1997

Ce0.8Sm0.2O2−δ/ La0.8Sr0.2CrO3−δ dual phase Sm0.4Ba0.6Co0.2Fe0.8O3-δ

Tube packed with CaSrTi-cat.

950

X = 17%; SH2

La0.4Ba0.6Fe0.8Zn0.2O3-δ La0.8Sr0.2Fe0.7Ga0.3O3-δ SrCo0.4Fe0.5Zr0.1O3-α La0.6Sr0.4Co0.2Fe0.8O3-δ La2NiO4 Ba0.5Sr0.5Co0.8Fe0.2O3-δ YSZ-SrCo0.4Fe0.6O3-δ La0.5Sr0.5Fe0.8Ga0.2O3-δ on α-Al2O3 Ba0.5Sr0.5Co0.8Fe0.2O3

–

Tian et al., 2010

X = 90%; SCO = 98%

–

Ikeguchi et al., 2005

X→100%; SCO > 95%

500 h

Gong and Hong, 2011

900

X = 74%; SH2

142 h

Delbos et al., 2010

950

Sco > 90%

–

Hoang and Chan, 2006

825–885

X > 96%; SCO > 97%

3–7 h

Jin et al., 2000

900 875

X = 89%; SCO = 96%; H2/CO = 1.5 X = 94%; SCO > 95%; JO2 8.0 ml/cm2.min

– 500 h

Zhu et al., 2003 Wang et al., 2003a

X = 64%; SCO ~ 100%

220 h

Gu et al., 2002

X = 97%; SCO = 100%; H2/CO = 1.76

–

Ritchie et al., 2001

X = 98.5%; SCO = 93% JO2 115 ml/cm2.min

500 h

Shao et al., 2001

Disk packed with Rh/ 900 MgO Disk packed with Ni-cat. 900 Tube coated with La0.8Sr0.2Fe0.7Ni0.3O3-δ cat. Disk packed with NiO/ Al2O3 Tube packed with Ni/γ-Al2O3 Tube Tube packed with LiLaNiO/γ-Al2O3

75%

750–850 Disk, packed with NiO/ Al2O3 Tube, Ru-cat. packed on 850 shell 875 Disk packed with LiLaNiO/γ-Al2O3 cat.

50%

Ca0.8Sr0.2Ti0.7Fe0.3O3-α YBa2Cu3O7-x SrFe0.7Al0.3O3-δ BaCo0.4Fe0.4Zr0.2O3-δ

Disc coated with Ni/ Ca0.8Sr0.2Ti0.9Fe0.1O3-α cat. Disc with or without Ni/ ZrO2 cat. Disk packed with SrFe0.7Al0.3O3-δ cat. Disk with catalyst

© Woodhead Publishing Limited, 2013

900

X = 13.7%, SCO = 98%

–

Hamakawa et al., 2000

875

X = 100%; SCO = 95%

5h

Hu et al., 2006

950

X = 65%; SCO = 48%

–

Kharton et al., 2005

850

X = 98%; Sco = 100% JO2 5.6 ml/cm2.min

2200 h

Tong et al., 2002

925

X = 96%; SCO = 97%; H2/CO ≈ 2

–

Wang et al., 2006a

950

X > 98%; SCO > 98%; H2/CO ≈ 2 JO2 4.3 ml/cm2.min

1100 h

Zhu et al., 2008

X

550 h

Zhang et al., 2011

X > 98%; SCO > 98%

500 h

Zhu et al., 2010

SCO = 90%; H2/CO ≈ 2

–

Wu et al., 2010

X

400 h

Luo et al., 2010

1000 h

Li et al., 2010

–

Zhang et al., 2010

Sm0.15Ce0.85O1.925/ Sm0.6Sr0.4Fe0.7Al0.3O3-δ

Hollow fiber packed with Ni-cat. on shell Disk packed with LiLaNiO/γ-Al2O3

BaCo0.7Fe0.2Nb0.1O3-δ

Disk packed with Ni-cat. 875

Ba(Co,Fe,Zr)O3-δ

JO2 Ce0.85Sm0.15O1.925/ Sm0.6Sr0.4FeO3-δ dual phase La0.8Sr0.2MnO3-δ−YSZ BaCo0.7Fe0.2Ta0.1O3−δ

Disk packed with LiLaNiO/γ-Al2O3

950

Hollow fiber with 950 NiO-YSZ as cat. layer Disk packed with Ni-cat 900

JO2 BaCe0.1Co0.4Fe0.5O3-δ 3%Al2O3 doped SrCo0.8Fe0.2O3

Disk packed with LiLaNiO/γ-Al2O3 cat.

950

Tube packed with Ni-cat. on shell

900

X JO2

92%; SH2

90% 2

15 ml / cm .min

99%; % SH2

94% 2

16.2mll / cm .min 99%; SCO

93%

9.5 mll / cm2.min

X = 99%; SCO > 93%

356

Handbook of membrane reactors CO2, CO, H2O CH4/He

CH4(g)

C2H6(g)

•CH3 +h•

CH4(s)

h•

VO••

C2H4(s)

C2H6(s)

eⴕ

×

Product gas (C2H4, C2H6, CO, CO2)

+OO

×

OO

MIEC membrane

O2

Air 1 2

× O2 + VO•• ↔ OO + 2h•

8.3 The mechanism of OCM reactions in the MIEC membrane reactor (‘g’ refers to gas phase and ‘s’ to surface phase.)

the reactor, and methane loss due to back-permeation can be prevented. Furthermore, the oxide membranes deliver oxygen to the reaction compartment in a dissociated or ionized form, hence the formation of COx from by-reactions due to the presence of gaseous oxygen can be suppressed leading to increased C2 selectivity. However, the low oxygen concentration also slows down the reaction rate. Consequently, a long contact time is required for a high methane conversion, but this may again leads to lowering the C2 selectivity. Therefore, so far the C2 yields obtained in the membrane reactors have not been able to exceed 35% (Bhatia et al., 2009). Figure 8.3 demonstrates the OCM process in MIEC membrane reactors. The membrane itself has intrinsically catalytic activity to the OCM reaction. On the membrane surface exposed to air, the following exchange reaction takes place at a high temperature: 1 O2 2

Vo

k k

Oox + 2 h•

[8.8]

where kf and kr represent the forward and the reverse reaction constants, respectively. The lattice oxygen (O0x ) and electron holes (h•) are transported through the membrane to the other side of the membrane (surface exposed to methane) and react with methane adsorbed to form methyl radicals. Local charge neutrality is maintained by joint diffusion of oxygen vacancies and electrons on the surface exposed to air: CH 4 ( ) ⇔ CH 4 ( ) CH 4 ( ) h• +

1 x OO 2

[8.9] CH

1 1 H O + VOii 2 2

© Woodhead Publishing Limited, 2013

[8.10]

Applications of dense ceramic membrane reactors

357

The methyl radicals are coupled in gas phase to form C2 product. In the meantime, the methyl radicals also react with gaseous oxygen to form carbon oxide: 2 • CH 3 → C 2 H 6

[8.11]

•CH 3 +

[8.12]

x O2 → CO x 2

Since the detailed mechanism of methyl radical oxidation is unknown, an undefined stoichiometric coefficient, x is introduced here. The further oxidation of ethane follows the same mechanism as that for methane oxidation as: C 2 H6 ( ) ⇔ C 2 H6 ( ) C 2 H 6 ( ) h• + •C 2 H 5 +

1 x OO 2

[8.13] C H

1 1 H O + VOii 2 2

x O2 → CO x 2

[8.14]

[8.15]

The C2 yield is very sensitive to the membrane characteristics, as well as the reaction conditions and the reactor dimensions (Wang and Lin, 1995). In order to achieve high C2 yields the oxygen permeation flux, the methane flow rate and the intrinsic reaction rate must match well with each other. Insufficient oxygen supply leads to poor conversion, but a high oxygen flux may result in low selectivity, because of the complete oxidation reactions, especially at high temperature and pressure (Haag et al., 2007). This implies that the oxygen permeability of the membrane has to match the catalytic activation of the membrane surface (Olivier et al, 2009). For a given composite membrane, the oxygen flux can be readily improved by decreasing membrane thickness or by improving the surface exchange kinetics. Therefore, the selection of a membrane material with good intrinsic catalytic properties, or the modification of these high oxygen-permeable ceramic membrane surfaces with an appropriate OCM catalyst such as lead oxides and alkali compounds, has become the most critical step in the development of dense membrane reactors for OCM. Some experimental results are summarized in Table 8.2. Among various developed membranes, Bi1.5Y0.3Sm0.2O3-δ exhibits not only high oxygen permeability and catalytic activity but also high chemical and mechanical stability under OCM conditions. The C2 yield in the Bi1.5Y0.3Sm0.2O3-δ membrane reactor has reached 35% (Akin and Lin, 2002a).

© Woodhead Publishing Limited, 2013

Table 8.2 MIEC membrane reactors for methane oxidative coupling

© Woodhead Publishing Limited, 2013

Membrane

Reactor configuration

T(ºC)

Main results

Ba0.5Sr0.5Co0.8Fe0.2O3-δ

Tube, no cat. Packed with La-Sr/CaO cat. Tube, no cat.

800–900

SC2

62% YC2 = 13 −15 15%; SC2

900

YC2

35 35% SC2

54%

Akin and Lin, 2002a

18 8% SC2

65%

Olivier et al., 2009

Bi1.5Y0.3Sm0.2O3-δ

References 54 58%

Wang et al., 2005

950

YC2

Ba0.5Sr0.5Co0.8Fe0.2O3-δ

Disk coated with La-Sr/ CaO cat. Disk

850

XCH4 > 40 – 70%; SC2

BaCe0.8Gd0.2O3-δ

Tube

778

YC2

16 1 6 5%; SC2

La0.6Sr0.4Co0.8Fe0.2O3-δ

Disk, no cat.

800–900

YC2

1 3% ; SC2 → 7 %

ten Elshof et al., 1995

Y-doped Bi2O3

Disk

750–950

YC2

16 1 6 14% 14 SC2

20 – 90%

Zeng and Lin, 2000

La0.8Sr0.2Co0.6Fe0.4O3-δ

Disk

850

YC2

10 18 8% SC2

70 – 90%

Zeng et al., 1998

La0.8Sr0.2CoO3

Disk

800–850

YC2

1 12 2 14 SC2 = 40 – 56%

La0.6Sr0.4Co0.2Fe0.8O3

Hollow fiber, packed with SrTi0.9Li0.1O3 cat.

–

YC2

Ba0.5Sr0.5Co0.8Fe0.2O3-δ

21% SC2

40 – 70%

62.5%

71.9%

Shao et al., 2001 Lu et al., 2000a, 2000b

Lin and Zheng, 1996 Tan et al., 2007

Applications of dense ceramic membrane reactors Methane in

359

Product gas (C2H4, C2H6, CO, CO2)

Ammeter Volt meter

Air out YSZ tube LSM electrode

Pt wire

LaAlO electrode

Air in

8.4 Schematic diagram of the tubular YSZ based SOFC membrane reactor for OCM.

In the SOFC-type membrane reactor made of pure ionic conducting membranes such as yttria-stabilized zirconia (YSZ, 8%Y2O3-ZrO2), electrical power can be co-generated with the OCM reaction. Figure 8.4 illustrates schematically a tubular YSZ based SOFC-type membrane reactor for OCM reaction (Tagawa et al., 1999). A YSZ tube with one dead end is used as the electrolyte. La0.85Sr0.15MnO3 (LSM) powder is pounded and mixed with glycerol, pasted into thin film on the outside of the YSZ tube, and heated to an elevated temperature to form the cathode. La1.8Al0.2O3 prepared on the inside of the YSZ tube by a mist pyrolysis method is used as the OCM catalyst as well as the anode. Pt wire is connected to platinum mesh placed on both electrodes to serve as the current collector. Oxygen ions are transferred from the cathode through the membrane to the anode side and react with CH4 to yield C2 products: Cathode side reaction : O2 O2 Anode side reaction : 2CH 4 + 2O

O22− C 2 H 4 + 2 H 2 O + 4e −

© Woodhead Publishing Limited, 2013

[8.16] [8.17]

360

Handbook of membrane reactors

The theoretical electromotive force is given by: E=−

ΔG nF

[8.18]

where ΔG is Gibbs free energy, n is the number of electrons and F is the Faraday constant (96.2 kJ/mol.V). The oxygen permeation rate is determined by the electrical current I. Fo2 =

I 4F

[8.19]

The anode catalyst plays a key role in the C2 selectivity. For example, when silver was used as electrodes and 1 wt% Sr/La2O3-Bi2O3 as catalyst, an electric current of 20–40 mA with C2 selectivity of 90–94% and C2 yields of 0.2–1% was obtained at 730°C (Guo et al., 1999). When using La1.8Al0.2O3 as anode catalyst, the electric current and C2 yields could reach 180 mA and up to 4%, respectively (Tagawa et al., 1999). However, all the membrane reactors tested in practice so far have not shown very high C2 yields. This was attributed to the low oxygen permeation flux, which did not match the catalytic activation of methane on the membrane surface. In fact, if an external power source is applied to form an EOP membrane reactor, the catalytic activity and C2 selectivity of the metal and metal oxide catalysts can be altered dramatically and reversibly due to supplying more active oxygen species (Eng and Stoukides, 1991), leading to much higher C2 yields. In general, although the SOFC-type MR requires an operational temperature approximately 200 K higher than the others, the electricity simultaneously generated as a by-product still makes it attractive.

8.2.3

Oxidative dehydrogenation of alkanes (ethane and propane)

The selective oxidation of alkanes such as ethane and propane to corresponding olefins is an important catalytic process: C 2 H6 + C 3 H8 +

1 O2 2

C2 H4

H 2 O, H 0298 = −105 kJ mol

1 O2 2

C 3 H6

H2O

© Woodhead Publishing Limited, 2013

1

[8.20]

[8.21]

Applications of dense ceramic membrane reactors

361

Compared with the steam cracking technology which is currently used extensively for ethylene and propene production in industry, the oxidative dehydrogenation process is more energy- and cost-effective, due to the exothermic nature of the reactions. Furthermore, the presence of oxygen may limit the bad coke formation. However, the product yields with current catalysts in conventional co-feed reactors are too low for commercial application, as attempts to replace the existing technology would require a yield of at least 70%. Since the products (olefins) can be deeply oxidized into COx more easily than the raw materials, it is hard to achieve a high selectivity. In order to increase the olefins’ selectivity, control of the contact mode between the reactants is necessary. This can be easily realized by using dense oxygen-permeable MRs. The process of oxidative dehydrogenation of ethane or propane in MIEC reactors can be seen in Fig. 8.3 without consideration of the coupling reactions. On the oxygen-rich side, gaseous O2 is first adsorbed on the membrane surface, reduced to O2− and then transported through the bulk of the membrane to the reaction side surface. On the reaction side, ethane is oxidized by surface O2−. As the surface oxygen is depleted, the bulk O2 diffuses from the oxygen-rich side to fill in the oxygen vacancies. The anode reaction may be written as: C 2 H 6 ( ) + O2

C2 H4

H 2 O + 2e −

[8.22]

or x C 2 H 6 ( ) OO

h•

C H

H O VOii

[8.23]

Obviously, the contact mode of the reactants (ethane and oxygen) with each other, and with the catalytically active surface, is quite different from that of conventional reactors and can be controlled at very high levels (Bhatia et al., 2009). In addition, dense membrane reactors also demonstrate many other advantages, such as in situ air separation, possible increase in the yield and selectivity across the thermodynamic limitations, high energy efficiency and operational safety (Wang et al., 2002). Some results for the oxidative dehydrogenation of ethane/propane in dense ceramic MRs are summarized in Table 8.3. It is noteworthy that the performance of the membrane reactor can be changed by application of surface catalysts. For example, with V/MgO micron grains or Pd nano cluster to modify the membrane surface respectively, the ethylene yield in the BSCF membrane reactor could reach 75% at 1040–1050 K (Rebeilleau-Dassonneville et al., 2005). However, the Ni cluster deposition led to a decreased ethane conversion compared to the bare membrane without changing the ethylene selectivity. In addition,

© Woodhead Publishing Limited, 2013

Table 8.3 Oxidative dehydrogenation of ethane/propane in the dense ceramic membrane reactors © Woodhead Publishing Limited, 2013

Reaction

Membrane

Configuration

Cat.

T (oC)

Main results

References

C2H6→C2H4 C2H6→C2H4

BSCF BYS

/ /

650 875

S = 90% Y = 56%; S = 80%

Wang et al., 2002 Akin and Lin, 2002b

C2H6→C2H4 C2H6→C2H4

BCFZ BSCF

Tube Tube with a dead end Hollow fiber Disk

800 770

S = 64%; X = 63% Y = 75%; S > 92%

C3H8→C3H6

BSCF

/ V/MgO, coated on surface /

750

S = 44.2%; X = 23–27%

Wang et al., 2006b Rebeilleau-Dassonneville et al., 2005 Wang et al., 2003b

Tube with a dead end

BSCF = Ba0.5Sr0.5Co0.8Fe0.2O3-δ; BCFZ = BaCoxFeyZrzO3-δ (x + y + z = 1); BYS = Bi1.5Y0.3Sm0.2O3.

Applications of dense ceramic membrane reactors

363

the contact time between the reactant and the membrane has significant effect on selectivity. For example, the ethene selectivity in hollow fiber MR is lower than that in the disk-shaped MR, despite a higher conversion being achieved (Wang et al., 2006b).

8.2.4

Decomposition of H2O, NOx and CO2

Hydrogen has received increasing attention as an energy source and/or as an energy transfer medium. However, it does not exist in nature and has to be produced from hydrogen-containing compounds. There is a particular interest in using water as the hydrogen source because it is clean and abundant. At a high temperature water can be dissociated into oxygen and hydrogen: H 2 O( ) ⇔ H 2

1 O2 2

[8.24]

Generally, very low concentrations of hydrogen and oxygen can be generated even at a very high temperature (e.g., 0.1% and 0.042% for hydrogen and oxygen, respectively, at 1600°C), because of the small equilibrium constant. However, if the equilibrium is shifted toward dissociation, by removing either oxygen or hydrogen, significant amounts of hydrogen or oxygen can be produced at moderate temperatures. This can be achieved by using an MIEC membrane without the need for electric power or electrical circuitry, as shown in Fig. 8.5. The driving force for oxygen permeation may be achieved by using an inert gas, or a reducing gas such as methane, as the sweeping gas. The hydrogen production rate highly depends on the rate at which oxygen is removed from the water-dissociation zone. This requires the membrane to possess high electron and oxygen ion conductivities and good surface exchange properties. Any methods to promote the oxygen permeation rate, such as decreasing the membrane thickness or increasing the active membrane surface area or by applying a water-dissociation catalyst to the membrane surface, may be able to improve the hydrogen production rate (Balachandran et al., 2004, 2007; Jiang et al., 2010; Wang et al., 2011). Table 8.4 summarizes recent work on the vapor dissociation in MIEC reactors. The maximum hydrogen production rate has reached 7.44 µmol cm−2 s−1. Nitrogen oxides (NOx, i.e., NO, NO2, N2O) are considered as major air pollutants responsible for photochemical smog, acid rain, ozone depletion and climate change. The conventional route to eliminate NOx pollution is to catalytically reduce NOx into N2 using NH3, urea, H2, CO and hydrocar-

© Woodhead Publishing Limited, 2013

364

Handbook of membrane reactors ×

H 2O + VO•• ↔ OO + 2h• + H 2

H2

or H 2O + 2e⬘↔ 12 O 2− + H2 H2O

eⴕ

O2–

VO••

h•

MIEC membrane Sweep gas +O2

Sweep gas O2 ×

OO + 2h • ↔ 2−

or O ↔

1 2

1 2

O2 + VO••

O2 + 2e ⴕ

8.5 Hydrogen production by water-dissociation with the help of MIEC ceramic membrane.

bons as reducing agents. The decomposition of N2O is a kinetically limited reaction and inhibited by the product molecule oxygen: N 2 O( ) ⇔ N 2

1 O2 2

[8.25]

Most perovskite catalysts cannot tolerate the co-existence of O2 because the adsorbed oxygen blocks the catalytically active sites for N2O decomposition. Using an MIEC membrane reactor, the inhibitor oxygen can be removed as oxygen ion (O2−) through the membrane: N 2 O( ) ⇔ N 2

O* → N 2

1 O2 2

[8.26]

Accordingly, the total decomposition of N2O can be achieved in the membrane reactor (Jiang et al., 2010). To increase the driving force for oxygen transport in the membrane, methane or ethane can be fed to the permeate side to consume the permeated oxygen. Figure 8.6 shows the principle of the MIEC membrane reactor for N2O decomposition combined with the partial oxidation of hydrocarbons. It should be mentioned that the decomposition of water for hydrogen production can also be enhanced by coupling with the partial oxidation of hydrocarbons. Recently, more and more attention has been focused on CO2 capture and sequestration because of global climate change. One potential route

© Woodhead Publishing Limited, 2013

Table 8.4 Dense ceramic membrane reactors for the decomposition reactions Reaction © Woodhead Publishing Limited, 2013

H2O→H2

Membrane

H2O→H2

Gd-doped CeO2–40%Ni SrFeCo0.5Ox

H2O→H2

GDC-GSTA

H2O→H2

BCFZ

N2O→N2 C2H6→ C2H4 CO2→CO CH4→syngas

BCFZ SCFA

Configuration

Cat.

T (oC)

Main results

−2 2

References

Disk (0.13 mm)

—

900

rH2 = 4 46 μm mol cm s

Disk (0.09 mm)

—

900

rH2 > 7 44 μm mol cm−22s 1

Disk (25 µm coated support) Hollow fiber (0.17 mm) Hollow fiber (0.17 mm) Tube

—

900

rH2 = 7 μmol m cm−22s 1

—

950

rH2 = 2 31μm mol cm−22s 1

Ni/Al2O3

875

XN2O

100%; XN2O = 91%; SC2H4 = 80%

Ni/Al2O3

900

XCO2

12 12.4% 4%; XCH4

SCO=93%; H2/CO=1.8

1

86%

Balachandran et al., 2004 Balachandran et al., 2007 Wang et al., 2011 Jiang et al., 2010 Jiang et al., 2010 Zhang et al., 2009

GDC = Gd0.2Ce0.8O1.9-δ; GSTA = Gd0.08Sr0.88Ti0.95Al0.05O3-δ; BCFZ = BaCoxFeyZr1−x−yO3−δ; SCFA = 3%Al2O3 doped SrCo0.8Fe0.2O3−δ.

366

Handbook of membrane reactors N 2O ↔ O * + N 2 N2O

O * +2e⬘↔ O 2−

N2

O*

eⴕ

CH4 (C2H6)

MIEC membrane

O2–

O 2− ↔ 12 O2 + 2e⬘ CH 4 + 12 O2 ↔ CO + 2H 2 C2 H 6 +

1

2

CO, H2 (C2H4, H2O)

O2 ↔ C2 H 4 + H 2 O

8.6 N2O decomposition in the MIEC membrane reactor enhanced by coupling with the partial oxidation of hydrocarbons.

CH4 (s) + O 2− → CO + 2H 2 + 2e⬘

CH4 CO, H2

O2–

eⴕ

VO••

h•

MIEC membrane

CO

CO2

CO2 + 2e⬘ ↔ CO + O 2−

8.7 Decomposition of CO2 coupled with POM in the MIEC membrane reactor.

for the consumption of CO2 is the thermal decomposition of carbon dioxide to CO and O2: 2CO2

2CO + O2 ,

H 0298 = 552 kJ mol

1

[8.27]

However, this is a highly endothermic reaction occurring only at high temperature and cannot be carried out in conventional fixed-bed reactors. Using a dense ceramic membrane reactor, the CO2 decomposition reaction can be coupled with POM to syngas, as shown in Fig. 8.7. In this process, the decomposition reaction takes place on one side of the membrane while the released oxygen from the decomposition permeates through the oxygenpermeable membrane to the other side and reacts with CH4 to yield syngas

© Woodhead Publishing Limited, 2013

Applications of dense ceramic membrane reactors

367

over the catalyst. The CO2 actually serves as the oxygen source for the POM reaction. The oxygen permeation rate through the membrane plays a key role in CO2 conversion (Zhang et al., 2009).

8.3

Hydrogen permeable membrane reactors

The hydrogen permeable membrane reactor is made of the proton or mixed proton-hole conducting oxides, hence it is also called the proton conducting membrane reactor (PCMR). In PCMRs, hydrogen is supplied to the hydrogenation reactions, or extracted from the dehydrogenation reactions, through the membrane in the form of protons, driving the reaction to the product side. With the oxide membrane as an electrolyte to supply or remove ions from the catalyst surface, catalytic activity and selectivity can be enhanced significantly. Compared to metal membranes, such as Pd or Pd–Ag alloys which have been studied extensively for dehydrogenation reactions, the dense ceramic membrane has higher stability, and thus is more attractive for use in high temperature and harsh environments. A variety of hydrogenation and dehydrogenation reactions have been conducted in PCMRs, which are summarized in the following. More details can be obtained from a recent review article (Kokkofitis et al., 2007a).

8.3.1

Dehydrogenation reaction

Methane coupling to C2 products Using oxygen separation membranes for methane coupling may have two problems: (1) deep oxidation of C2 products by gaseous oxygen, which may lead to lowering the C2 selectivity and yield; and (2) water vapor formation in the product stream, which may result in complexity of the reactor system and harm to the catalytic activity of membrane and catalyst. As an alternative to the oxidative process, methane coupling can also be achieved through a dehydrogenation process, where gaseous oxygen can be avoided: 2CH 4

C2

6

H2 ,

2CH 4

C2

4

2H 2 ,

0 1000 0 1000

77.3 kJ mol

1

9 .9 kJ mol 92

[8.28a] 1

[8.28b]

However, the thermodynamics of these reactions are unfavorable and display equilibrium-limited conversion. By removing the hydrogen from the coupling side, the equilibrium of the methane coupling may be driven towards the C2 product side to a complete conversion of the reactant, resulting in

© Woodhead Publishing Limited, 2013

368

Handbook of membrane reactors

significant improvement of C2 yield. This can be achieved by using a PCMR with the principle illustrated schematically in Fig. 8.8. On the anode side, methane is adsorbed and catalytically converted into methyl radicals, which enter into the gas phase for the coupling reaction. The protons released from the methane coupling reaction are electrochemically driven to the cathode side of the membrane, where they combine with electrons into hydrogen. In the presence of oxygen, the protons are oxidized to water. The electrodes are attached to the membrane surfaces to collect/distribute electrons. Three operational modes can be applied to PCMRs, depending on the primary goal as well as the conducting property of the membrane. If the goal is the production of electricity, the membrane should be of a pure ionic conductor so that chemical energy can be converted directly into electrical energy (so called fuel cell mode, Fig. 8.8b). If, on the other hand, the primary goal is the production of C2 hydrocarbons and/or hydrogen, an external power source can be used to impose a current through the cell (so called pumping mode, Fig. 8.8a). In this case, the hydrogen separation rate can be controlled by the modulation of applied current. For given methane feed rate, a high yield requires a high proton permeation rate or equivalently, a high current density. But the imposed current density usually cannot exceed an upper limit due to the thick membranes used (Chiang et al., 1995), so the C2 yields are usually very low, which severely limits the practicality of the reactors to enhance methane coupling. The proton transference number of the membrane oxides may vary and, depending primarily on temperature and the partial pressures of hydrogen and oxygen, the membranes may exhibit almost pure protonic, mixed protonic−electronic and even mixed oxygen ion−protonic−electronic conductivity. Hence, mixed conducting membranes, either with or without electrodes, have been also utilized in an effort to achieve acceptably high yields to C2 products. When the membrane is a mixed proton−electron conductor, the hydrogen can be extracted in the reaction system by the self-discharge phenomenon. In this case, the external electric source and the electrode materials as well as current collectors are unnecessary for transporting protons across the solid electrolyte, and the construction of the reactor is much simpler and cheaper (permeation mode, Fig. 8.8c). However, the methane activation and dimerization rate may be many times lower than that passing the direct current through the cell (Chiang et al., 1993; Hamakawa et al., 1993). Catalysts are usually applied to activate methane dissociation and hydrogen combination. Such catalysts may be of Pt, Ag or other specially designed oxides. It is worth noting that several mixed conducting perovskites (e.g., strontium cerates) are very efficient methane coupling catalysts and have been used as such in regular catalytic reactors. Studies on the methane

© Woodhead Publishing Limited, 2013

Applications of dense ceramic membrane reactors

369

0 ⇔ h• + eⴕ C2 H 6 CH 4 + h• ⇔ •CH 3 + H •

(a)

Anode eⴕ PC membrane

H• eⴕ

Cathode

2H • + 2e⬘ ⇔ H 2 C2 H 6 0 ⇔ h• + e⬘ CH 4 + h• ⇔ • CH3 + H •

(b) eⴕ

Anode

H•

PC membrane Cathode

eⴕ

O2 1 2

O2 + 2H • ⇔ H 2O + 2h•

h• + e⬘↔ 0 C2 H 6 CH 4 + h• ⇔ •CH 3 + H •

(c)

Anode

h•

H•

e⬘

MIEC membrane Cathode

O2 1 2

O2 + 2H • ⇔ H 2O + 2h•

8.8 Principles of PCMR for methane coupling: (a) pumping mode; (b) fuel cell mode; (c) permeation mode.

coupling in PCMRs are summarized in Table 8.5. As can be seen, most of the reported C2 yields are very low (less than 2%). The reasons for this include the low catalytic activity of the membrane surface and catalyst, the large membrane thickness and methane pyrolysis in the absence of oxygen: CH 4 ⇔ C

H2

© Woodhead Publishing Limited, 2013

[8.29]

© Woodhead Publishing Limited, 2013

Configuration

Cat./ Operation mode electrode

T (ºC)

Disk (1 mm) Disk (1.5 mm)

Hollow fiber

BCM SCYb

SCYb

–

– Ag

Ag Ag Ag

Permeation

Permeation Pumping

Pumping Pumping Permeation

Pt

Disk (30 µm on porous support)

BCY Fuel cell

Fuel cell

2e ′; 2H• 1 2O2 2e9↔ H2O

700

700

Main results

100%

13 13 4%; SC2 = 21%

X = 34%; S = 96%; P = 174 mW cm−2 X = 36.7%; S = 90.5%; P = 216 mW cm−2

–

YC2

8.9 µmol min–1 cm–2 YC2 55 55 SC2 = 64%

SC2

0.23–0.52 µmol min–1 cm–2 – 0.23–1.33 µmol min–1 cm–2

2e9↔ H2O

650–750

950

950 750

900 750 900

SCYb = SrCe0.95Yb0.05O3-δ; BCM = BaCe0.95Mn0.05O3-δ; BCY = BaCe0.85Y0.15O3-δ.

Pt

Disk (0.5 mm)

BCY

Ethane to ethylene: C2H6 ↔ C2H4 + 2H•

Propane to propylene: C2H6 ↔ C2H4 + 2H H• e9 SCYb Disk (1.5 mm) Pt, Pd/Ag Pumping

Disk Disk Disk (1mm)

SCYb SCYb SCYb

Methane coupling to C2 hydrocarbons: CH4 ↔ •CH3 + H• + e′ ; 2H• 1 2O2

Membrane

Table 8.5 PCMRs for dehydrogenation reactions

Fu et al., 2010

Shi et al., 2008

Karagiannakis et al., 2005, 2006

Liu et al., 2006b

White et al., 1998 Langguth et al., 1997

Hamakawa et al., 1993 Chiang et al., 1993 Hamakawa et al., 1994

References

Applications of dense ceramic membrane reactors

371

Conversion of alkanes into alkenes The catalytic dehydrogenation of propane is a potential method for the production of propylene, a key chemical in the polymerization and organic synthesis industries. C 3 H8 ⇔ C 3 H6 + H 2 ,

H 0298 = 124.2 kJ mol

1

[8.30]

This reaction requires high temperatures (500–700°C) and low pressures (0.3–1 atm) to operate. At such high temperatures, however, the side-reaction of C3H8 thermal decomposition takes place: C 3 H 8 ⇒ C 2 H 4 + CH 4

[8.31]

With a proton conducting electrochemical reactor to withdraw the hydrogen from the reactor, the limited conversion of the catalytic dehydrogenation of propane dictated by the thermodynamic equilibrium can be shifted, leading to promotion of propylene selectivity and yield. The driving force is the difference in electrochemical potential of protons at the two sides of the membrane, which is generated by using an external power source (e.g., a galvanostat or a potentiostat) (Fig. 8.8a). In addition, high purity hydrogen can be obtained as a product at the cathode. Polycrystalline Pt is superior to palladium as the anode catalyst to yield higher rates of propane decomposition (Karagiannakis et al., 2005). When propane steam mixture is introduced, instead of pure propane in the reactor, up to 90% of the produced hydrogen could be electrochemically separated (Karagiannakis et al., 2006). Similarly, ethane can also be converted into ethylene in a PCMR with enhanced ethylene yields (Shi et al., 2008). Electrical power can be co-generated with ethylene product in fuel cell operation mode (Fig. 8.8b). The performance of the fuel cell MR can be improved by reducing the membrane thickness, because of the increased permeation property (Fu et al., 2010). Some results from the literature on propane/ethane dehydrogenation in PCMRs have been summarized in Table 8.5.

8.3.2

Hydrogen production

The water gas shift (WGS) reaction is a key technology in the hydrogen purification processes of syngas obtained by steam reforming or partial oxidation of hydrocarbons: CO + H 2 O ⇔ CO2 + H 2 ,

H 0298 = −41.1 kJ mol −1

© Woodhead Publishing Limited, 2013

[8.32]

372

Handbook of membrane reactors

PCMRs can be used to selectively remove hydrogen from the reacting mixture and, at the same time, shift the reaction equilibrium to the product side, obtaining higher conversions. SrCe0.95Yb0.05O3-δ (SCYb) is a well-known proton conductor extensively applied as electrolyte membrane, but it is relatively intolerant to CO2 (Matsumoto et al., 2002). SrZr0.95Y0.05O3-δ electrolyte demonstrates high tolerance to CO2 exposure (500 h) but low hydrogen permeability (Kokkofitis et al., 2007b). In order to separate the hydrogen with higher efficiency, it has to use thin electrolyte membranes and an improved anode, but the applied voltage should be as low as possible. SrCe0.9Eu0.1O3-δ exhibits mixed proton−electron conduction, thus it can be used for hydrogen separation without the need for any external circuits (Li et al., 2009). Another effective way to produce hydrogen is by the electrolysis of water if electric power is available, preferably from renewable or sustainable energies, such as solar and wind powers, bio-energy, etc. (Matsumoto et al., 2002). The hydrogen can be electrochemically pumped out through a proton conducting membrane by using an external current. The current efficiency increases with increasing partial pressure of water vapor and temperature, and decreases with increasing current density (Kobayashi et al., 2001). Studies on hydrogen production in the PCMRs are summarized in Table 8.6.

8.3.3

Synthesis of ammonia

The dominant process for ammonia synthesis (Haber process) was developed at the beginning of the twentieth century and involves the reaction of gaseous nitrogen and hydrogen on an Fe-based catalyst at high pressures (150–300 bar): N2

3H 2 ←

Fe cat Fe-cat

→ 2 NH 3

[8.33]

The conversion to ammonia is limited by thermodynamics. Since the gas volume decreases with reaction, very high pressures must be used in order to push equilibrium to the ammonia side. Although the reaction is exothermic, and therefore conversion increases with decreasing temperature, the reaction temperature must be high so as to achieve industrially acceptable reaction rates. Due to the application of more active catalysts, most modern synthesis loops nowadays are operated at 100–300 bar and 450–500°C, with equilibrium conversion in the order of 10–15%. In a PCMR ammonia synthesis can be achieved at atmospheric pressure and the conversion may be attained higher than the equilibrium value of the reaction. Figure 8.9 illustrates schematically the principle of ammonia synthesis in a typical PCMR. Pd–Ag alloy is usually served as the working

© Woodhead Publishing Limited, 2013

Table 8.6 PCMRs for hydrogen production © Woodhead Publishing Limited, 2013

Membrane Configuration

Electrode/Cat.

WGS process: CO + H2O ↔ CO2 + 2H• + 2e9 ; 2H•

Operation mode

T (ºC)

Main results

References

SCYb

Disk (0.5 mm)

Pt

2 ↔ H2 Pumping

800

–

Matsumoto et al., 2002

SZY

Disk

Pd

Pumping

600–750

rH2 = 0.8 3.2 mol s−1

Kokkofitis et al., 2007b

SCEu

Tube (23 µm on porous support)

Ni-SrCeO3

Permeation

900

H• Steam electrolysis: H2O ↔ 2H SCYb SZYb

Disk (0.5 mm) Tube with one closed end

CO

46 6%; % YH2

32%

Li et al., 2009

O2 ; 2H• + 2 ′ ↔ H2 Pt Pt cermet

Pumping Pumping

800 460

– –

SCYb = SrCe0.95Yb0.05O3-δ; SZY = SrZr0.95Y0.05O3-δ; SCEu = SrCe0.9Eu0.1O3-δ; SZYb = SrZr0.9Yb0.1O3-δ.

Matsumoto et al., 2002 Kobayashi et al., 2001

374

Handbook of membrane reactors N2

N 2 + 6H • + 6 e⬘

NH 3

H•

eⴕ

H 2 ⇔ 2H • + 2e⬘

Anode (Pd–Ag) Sealant PC membrane

Porous support

eⴕ Cathode (Pd–Ag)

H2

8.9 Principle of ammonia synthesis in a typical PCMR.

electrode. Gaseous H2 passing over the anode of the cell reactor is converted to protons: 3

2

→ 6H 6 H • + 6e −

[8.34]

The protons (H•) are then transported through the solid electrolyte to the cathode where the following half-cell reaction takes place: N2

6H•

66e − → 2 NH 3

[8.35]

Since the hydrogen is supplied electrochemically to the catalyst surface in the form of protons, the reaction may follow a different mechanism, or at least the rate-determining step is different from that in the typical catalytic system where the rate-determining step is the dissociative adsorption of gaseous molecular nitrogen onto the catalyst’s active sites. As a result, NH3 production rates in the membrane reactor may exceed those in normal reactors by at least three orders of magnitude, and the conversion of H+ into NH3 may be as high as 78% (Marnellos et al., 2000). Upon ‘pumping’ H+ to the catalyst surface, the reaction rate could increase by as much as 1300%, while the catalytic activity would decrease and eventually be completely lost on ‘pumping’ H+ away from the catalyst. Some studies on ammonia synthesis in PCMRs are summarized in Table 8.7. Generally, the reaction rate

© Woodhead Publishing Limited, 2013

Table 8.7 Ammonia synthesis in the PCMRs

© Woodhead Publishing Limited, 2013

Membrane

Configuration

Electrode/cat.

T (ºC)

Main results

References

SCYb

Tube with one closed end

Pd

570–750

rNH3 increased at least three orders Marnellos et al., 2000

LCZ

Disk (0.8 mm)

Pd–Ag

520

of magnitude rNH3 = 2.0 0 10−99mol s 1cm−2

Xie et al., 2004

BCNb

Disk (0.8 mm)

Pd–Ag

620

rNH3 = 2.16 6 10−99mol s 1cm−2

Li et al., 2005

CM

Disk (0.8 mm)

Pd–Ag

400–800

rNH3 = 7.2 2− 8 2 × 10 9mol s 1cm−2

Liu et al., 2006

LSGM

Disk

Pd–Ag

550

rNH3 = 2.37 3 10−99mol s 1cm−2

Zhang et al., 2007

BCY

Disk on NiO-BCY Pd–Ag/BSCF support

530

r = 4.1 × 10−9 mol s−1 cm−2

Wang et al., 2010

SCYb = SrCe0.95Yb0.05O3-δ; LCZ = La1.9Ca0.1Zr2O6.95; BCNb = Ba3(Ca1.18Nb1.82)O9-δ; CM = Ce0.8M0.2O2-δ (M = La, Y, Gd, Sm); LSGM = La0.9Sr0.1Ga0.8Mg0.2O3−α; BCY = BaCe0.85Y0.15O3-δ; BSCF= Ba0.5Sr0.5Co0.8Fe0.2O3−α.

376

Handbook of membrane reactors NOx, He or N2O, He

H2O, N2, O2, He H • or H ad

eⴕ

H•

H•

H•

H•

Cathode (Pd)

H•

H•

eⴕ

Proton conductor SrCe0.95Yb0.05O3-δ Anode (Ag)

H2O, He

H2O, O2, He

8.10 Schematic principle of the reduction of nitrogen oxides in a PCMR.

is limited by the rate of proton transport through the membrane. Therefore, any methods to improve the proton transport rate, such as decreasing the membrane thickness and promoting the surface exchange kinetics, are able to increase the ammonia yields.

8.3.4

Decomposition of NOx

In addition to the oxygen separation membranes, the proton conducting membranes can also be applied to reduce NOx emission by combining heterogeneous catalysis and solid state electrochemistry. The solid electrolytes in MRs serve to electrochemically control chemisorptive bonds and enhance catalytic activity. Figure 8.10 shows the schematic diagram of a steam electrolysis cell constructed with a proton conductor for reducing NO. Steam is electrolyzed at the anode. It shows the produced H+ is electrochemically pumped to the cathode and reacts with NOx to produce N2 and H2O: Anode reaction : H 2 O ↔ 2 H•

1 O2 + 2e − 2

Cathode reaction : NO + 2H H• + e →

1 N2 + H2O 2

[8.36]

[8.37]

Strontium cerates or strontium zirconates proton conductor can be used as the electrolyte membrane, with the results of nitrogen oxide reduction summarized in Table 8.8. When Pt/Ba/Al2O3 or Pt/Sr/Al2O3 is used as working electrodes, it is possible to reduce the NOx even in the presence of excess O2 (Kobayashi et al., 2000, 2002). The reduction of NO proceeds through the electrochemical reduction of NO absorbed into Sr/Al2O3 but not through the chemical reduction of NO by H2 gas.

© Woodhead Publishing Limited, 2013

Applications of dense ceramic membrane reactors

377

Table 8.8 PCMRs for the reduction of nitrogen oxides T(ºC) Main results

Membrane Configuration

Electrode/ cat.

SZYb

Pt/Sr-Al2O3 350

SCYb SCYb

Tube with one closed end Tube with one closed end –

Pt/Ba/ Al2O3 Ag/Pd

450

50% removal, 20% efficiency Removal→100%

–

SN2

100%

References

Kobayashi et al., 2000 Kobayashi et al., 2002 Kalimeri et al., 2010

SZYb = SrZr0.9Yb0.1O3-δ; SCYb = SrCe0.95Yb0.05O3-δ.

8.4

Conclusions and future trends

Dense ceramic membrane reactors have been applied in a variety of oxidation and dehydrogenation reactions, aiming at improving the reactant conversion and product yields. The membranes may function either as reactant distributors or as product extractors. Catalysts can be readily incorporated with the membrane by packing the reactor or by coating the membrane surface. In most cases, the reaction rate is limited by the oxygen/hydrogen permeation rate in the ceramic membranes but not by the reaction kinetics. However, dense ceramic MRs have a long way to go before they can find any practical applications. High permeability and sufficient stability are always required for dense ceramic MRs in practical applications. However, most currently developed membranes exhibit limited permeation rate to match the high activity of catalysts. Furthermore, either the oxygen or the hydrogen separation membranes are not sufficiently stable under a wide range of oxygen/hydrogen partial pressures and especially in highly reducing atmosphere. Therefore, developing novel ceramic membranes with high permeability and long-term stability is the main challenge for successful application. As well, the further efforts should also include the preparation of dense ceramic membranes with ultrathin separation layer and large specific membrane area (ratio of effective membrane area to the reactor volume), the improvement of the catalytic properties of the membranes (which may be achieved through loading an appropriate catalyst on the membranes by various methods) and the assembly of membrane reactors and systems.

8.5

Acknowledgements

The authors gratefully acknowledge the research funding provided by the National Natural Science Foundation of China (NNSFC, No. 20976098, No. 21176187) and the Royal Academy of Engineering Research Exchanges with China and India Scheme.

© Woodhead Publishing Limited, 2013

378

Handbook of membrane reactors

8.6

References

Akin F T and Lin Y S (2002a), ‘Oxidative coupling of methane in dense ceramic membrane reactor with high yields’, AIChE J, 48(10), 2298–2306. Akin F T and Lin Y S (2002b), ‘Selective oxidation of ethane to ethylene in a dense tubular membrane reactor’, J Membrane Sci, 209, 457–467. Andersen A G, Hayakawa T, Suzuki K, Shimizu M and Takehira K (1994), ‘Electrochemical methane conversion over SrFeO3-δ perovskite on an yttrium stabilized zirconia membrane’, Catal Lett, 27, 221–233. Balachandran U, Dusek J T, Maiya P S, Ma B, Mieville R L, Kleefisch M S and Udovich C A (1997), ‘Ceramic membrane reactor for converting methane to syngas’, Catal Today, 36, 265–272. Balachandran U, Lee T H, Wang S and Dorris S E (2004), ‘Use of mixed conducting membranes to produce hydrogen by water dissociation’, Int J Hydrogen Energy, 29, 291–296. Balachandran U, Lee T H and Dorris S E (2007), ‘Hydrogen production by water dissociation using mixed conducting dense ceramic membranes’, Int J Hydrogen Energy, 32, 451–456. Bhatia S, Thien C Y and Mohamed A R (2009), ‘Oxidative coupling of methane (OCM) in a catalytic membrane reactor and comparison of its performance with other catalytic reactors’, Chem Eng J, 148, 525–532. Chiang P H Eng D and Stoukides M (1993), ‘Electrocatalytic nonoxidative dimerization of methane over Ag electrodes’, Solid State Ionics, 61, 99–103. Chiang P H, Eng D, Panagiotis T and Stoukides M (1995), ‘Ion transport and polarization studies in a proton conducting solid electrolyte cell’, Solid State Ionics, 77, 305–310. Czuprat O, Caro J, Kondratenko V A and Kondratenko E V (2011), ‘Dehydrogenation of propane with selective hydrogen combustion: a mechanistic study by transient analysis of products’, Catal Comm, 11, 1211–1214. Delbos C, Lebain G, Richet N and Bertail C (2010), ‘Performances of tubular mixed conducting La0.8Sr0.2Fe0.7Ga0.3O3-δ membrane reactor for under pressure methane conversion to syngas’, Catal Today, 156, 146–152. Eng D and Stoukides M (1991), ‘Catalytic and electrochemical oxidation of methane on platinum’, J Catal, 130, 306–309. Fu X Z, Luo J-L, Sanger A R, Xu Z-R and Chuang K T (2010), ‘Fabrication of bi-layered proton conducting membrane for hydrocarbon solid oxide fuel cell reactors’, Electrochi Acta, 55, 1145–1149. Gong Z and Hong L (2011), ‘Integration of air separation and partial oxidation of methane in the La0.4Ba0.6Fe0.8Zn0.2O3-δ membrane reactor’, J Membrane Sci, 380, 81–86. Gu X, Jin W, Chen C, Xu N, Shi J and Ma Y H (2002), ‘YSZ-SrCo0.4Fe0.6O3-δ membranes for the partial oxidation of methane to syngas’, AIChE J, 48, 2051–2060. Guo X M, Hidajat K and Ching C B (1999), ‘Simulationof a solid oxide fuel cell for oxidative coupling of mehtane’, Catal Today, 50, 109–116. Haag S, van Veen A C and Mirodatos C (2007), ‘Influence of oxygen supply rates on performances of catalytic membrane reactors: application to the oxidative coupling of methane’, Catal Today, 127, 157–164. Hamakawa S, Hibino T and Iwahara H (1993), ‘Electrochemical methane coupling using protonic conductors’, J Electrochem Soc, 140, 459–462.

© Woodhead Publishing Limited, 2013

Applications of dense ceramic membrane reactors

379

Hamakawa S, Hibino T and Iwahara H (1994), ‘Electrochemical methane coupling in a proton-hole mixed conductor and its application to a membrane reactor’, J Electrochem Soc, 141, 1720–1726. Hamakawa S, Hayakawa T, Suzuki K, Murata K, Takehira K, Yoshino S, Nakamura J and Uchijima T (2000), ‘Methane conversion into synthesis gas using an electrochemical membrane reactor’, Solid State Ionics, 136/137, 761–766. Hoang D L and Chan S H (2006), ‘Effect of reactor dimensions on the performance of an O2 pump integrated partial oxidation reformer—a modelling approach’, Int J Hydrogen Energy, 31, 1–12. Hu J, Xing T, Jia Q, Hao H, Yang D, Guo Y and Hu X (2006), ‘Methane partial oxidation to syngas in YBa2Cu3O7-x membrane reactor’, Applied Catalysis A: General, 306, 29–33. Ikeguchi M, Mimura T, Sekine Y, Kikuchi E and Matsukata M (2005), ‘Reaction and oxygen permeation studies in Sm0.4Ba0.6Co0.2Fe0.8O3-δ membrane reactor for partial oxidation of methane to syngas’, Appl Catal A-Gen, 290, 212–220. Ishihara T and Takita Y (2000), ‘Partial oxidation of methane into syngas with oxygen permeating ceramic membrane reactors’, Catal Surveys Japan, 4, 125–133. Ishihara T, Tsuruta Y, Todaka T, Nishiguchi H and Takita Y (2002), ‘Fe doped LaGaO3 perovskite oxide as an oxygen separating membrane for CH4 partial oxidation’, Solid State Ionics, 152/153, 709–714. Jiang H, Wang H, Liang F, Werth S, Schirrmeister S, Schiestel T and Caro J (2010), ‘Improved water dissociation and nitrous oxide decomposition by in situ oxygen removal in perovskite catalytic membrane reactor’, Catal Today, 156, 187–190. Jiang Q, Faraji S, Slade D A and Stagg-Williams S M (2010), ‘A review of mixed ionic and electronic conducting ceramic membranes as oxygen sources for high-temperature reactors’, Membr Sci Technol, 14, 235–273. Jin W, Li S, Huang P, Xu N, Shi J and Lin Y S (2000), ‘Tubular lanthanum cobaltite perovskite-type membrane reactors for partial oxidation of methane to syngas’, J Membrane Sci, 166, 13–22. Kalimeri K K, Athanasiou C I and Marnellos G E (2010), ‘Electro-reduction of nitrogen oxides using steam electrolysis in a proton conducting solid electrolyte membrane reactor (H+-SEMR)’, Solid State Ionics, 181, 223–229. Karagiannakis G, Kokkofitis C, Zisekas S and Stoukides M (2005), ‘Catalytic and electrocatalytic production of H2 from propane decomposition over Pt and Pd in a proton-conducting membrane-reactor’, Catal Today, 104, 219–224. Karagiannakis G, Zisekas S, Kokkofitis C and Stoukides M (2006), ‘Effect of H2O presence on the propane decomposition reaction over Pd in a proton conducting membrane reactor’, Appl Catal A-Gen, 301, 265–271. Kharton V V, Yaremchenko A A, Valente A A, Sobyanin V A, Belyaev V D, Semin G L, Veniaminov S A, Tsipis E V, Shaula A L, Frade J R and Rocha J (2005), ‘Methane oxidation over Fe-, Co-, Ni- and V-containing mixed conductors’, Solid State Ionics, 176, 781–791. Kobayashi T, Abe K, Ukyo Y and Iwahara H (2000), ‘Reduction of nitrogen oxide by steam electrolysis cell using a protonic conductor SrZr0.9Yb0.1O3-α and the catalyst Sr/Al2O3’, Solid State Ionics, 134, 241–247. Kobayashi T, Abe K, Ukyo Y and Matsumoto H (2001), Study on current efficiency of steam electrolysis using a partial protonic conductor SrZr0.9Yb0.1O3−α’, Solid State Ionics, 138, 243–251.

© Woodhead Publishing Limited, 2013

380

Handbook of membrane reactors

Kobayashi T, Abe K, Ukyo Y and Iwahara H (2002), ‘Performance of electrolysis cells with proton and oxide-ion conducting electrolyte for reducing nitrogen oxide’, Solid State Ionics, 154/155, 699–705. Kokkofitis C, Ouzounidou M, Skodra A and Stoukides M (2007a), ‘High temperature proton conductors: Applications in catalytic processes’, Solid State Ionics, 178, 507–513. Kokkofitis C, Ouzounidou M, Skodra A and Stoukides M (2007b), ‘Catalytic and electrocatalytic production of H2 from the water gas shift reaction over Pd in a high temperature proton-conducting cell-reactor’, Solid State Ionics, 178, 475–480. Langguth J, Dittmeyer R, Hofmann H and Tomandl G (1997), ‘Studies on oxidative coupling of methane using high-temperature proton-conducting membranes’, Appl Catal, 158, 287–305. Li J, Yoon H, Oh T-K and Wachsman E D (2009), ‘High temperature SrCe0.9Eu0.1O3-δ proton conducting membrane reactor for H2 production using the water-gas shift reaction’, Applied Catalysis B: Environmental, 92, 234–239. Li Q, Zhu X, He Y and Yang W (2010), ‘Partial oxidation of methane in BaCe0.1Co0.4Fe0.5O3-δ membrane reactor’, Catalysis Today, 149, 185–190. Li S Z, Liu R Q, Xie Y H, Feng S and Wang J D (2005), ‘A novel method for preparation of doped Ba3(Ca1.18Nb1.82)O9-δ: Application to ammonia synthesis at atmospheric pressure’, Solid State Ionics, 176, 1063–1066. Lin Y S and Zheng Y (1996), ‘Catalytic properties of oxygen semipermeable perovskite-type ceramic membrane materials for oxidative coupling of methane’, J Catal, 164, 220–231. Liu R Q, Xie Y H, Wang J D, Li S Z and Wang B H (2006), ‘Synthesis of ammonia at atmospheric pressure with Ce0.8M0.2O2-δ (M=La, Y, Gd, Sm) and their proton conduction at intermediate temperature’, Solid State Ionics, 177, 73–76. Liu S, Tan X, Li K and Hughes R (2001), ‘Methane coupling using catalytic membrane reactors’, Catal Rev, 43, 147–198. Liu Y, Tan X and Li K (2006a), ‘Mixed conducting ceramics for catalytic membrane processing’, Catal Rev, 48, 145–198. Liu Y, Tan X and Li K (2006b), ‘Non-oxidative methane coupling in the SrCe0.95Yb0.05O3-α (SCYb) hollow fibre membrane reactor’, Ind Eng Chem Res, 45, 3782–3790. Lu Y P, Dixon A G, Moser W R, Ma Y H and Balachandran U (2000a), ‘Oxidative coupling of methane using oxygen-permeable dense membrane reactors’, Catal Today, 56, 297–305. Lu Y P, Dixon A G, Moser W R, Ma Y H and Balachandran U (2000b), ‘Oxygen permeable dense membrane reactor for the oxidative coupling of methane’, J Membrane Sci, 170, 27–34. Luo H, Wei Y, Jiang H, Yuan W, Lv Y, Caro J and Wang H (2010), ‘Performance of a ceramic membrane reactor with high oxygen flux Ta-containing perovskite for the partial oxidation of methane to syngas’, J Membrane Sci, 350, 154–160. Marnellos G, Zisekas S and Stoukides M (2000), ‘Synthesis of ammonia at atmospheric pressure with the use of solid state proton conductors’, J Catal, 193, 80–87. Matsumoto H, Okubo M, Hamajima S, Katahira K and Iwahara H (2002), ‘Extraction and production of hydrogen using high-temperature proton conductor’, Solid State Ionics, 152/153, 715–720.

© Woodhead Publishing Limited, 2013

Applications of dense ceramic membrane reactors

381

Olivier L, Haag S, Mirodatos C and van Veen A C (2009), ‘Oxidative coupling of methane using catalyst modified dense perovskite membrane reactors’, Catal Today, 142, 34–41. Ramachandra A M, Lu Y, Ma Y H, Moser W R and Dixon A G (1996), ‘Oxidative coupling of methane in porous Vycor membrane reactors’, J Membrane Sci, 116, 253–264. Rebeilleau-Dassonneville M, Rosini S, van Veen A C, Farrusseng D and Mirodato C (2005), ‘Oxidative activation of ethane on catalytic modified dense ionic oxygen conducting membranes’, Catal Today, 104, 131–137. Ritchie J T, Richardson J T and Luss D (2001), ‘Ceramic membrane reactor for synthesis gas production’, AIChE J, 47, 2092–2101. Shao Z, Dong H, Xiong G, Cong Y and Yang W (2001), ‘Performance of a mixed-conducting ceramic membrane reactor with high oxygen permeability for methane conversion’, J Membrane Sci, 183, 181–192. Shi Z, Luo J-L, Wang S, Sanger A R and Chuang K T (2008), ‘Protonic membrane for fuel cell for co-generation of power and ethylene’, J Power Sources, 176, 122–127. Tagawa T, Moe K K, Ito M and Goto S (1999), ‘Fuel cell type reactor for chemicals-energy co-generation’, Chem Eng Sci, 54, 1553–1557. Tan X and Li K (2009), ‘Design of mixed conducting ceramic membranes/reactors for the partial oxidation of methane (POM) to syngas’, AIChE J, 55(10), 2675–2685. Tan X, Pang Z, Gu Z and Liu S (2007), ‘Catalytic perovskite hollow fibre membrane reactors for methane oxidative coupling’, J Membrane Sci, 302, 109–114. ten Elshof J E, Bouwmeester H J M and Verweij H (1995), ‘Oxidative coupling of methane in a mixed-conducting perovskite membrane reactor’, Appl Catal, A130, 195–212. Tian T, Wang W, Zhan M and Chen C (2010), ‘Catalytic partial oxidation of methane over SrTiO3 with oxygen-permeable membrane reactor’, Catalysis Comm, 11, 624–628. Tong J, Yang W, Cai R, Zhu B and Lin L (2002), ‘Novel and ideal zirconium-based dense membrane reactors for partial oxidation of methane to syngas’, Catal Lett, 78, 129–137. Wang H, Cong Y and Yang W S (2002), ‘High selectivity of oxidative dehydrogenation of ethane to ethylene in an oxygen permeable membrane reactor’, Chem Commun, 14, 1468–1469. Wang H, Cong Y and Yang W (2003a), ‘Investigation on the partial oxidation of methane to syngas in a tubular Ba0.5Sr0.5Co0.8Fe0.2O3−δ membrane reactor’, Catal Today, 82, 157–166. Wang H, Cong Y, Zhu X and Yang W (2003b), ‘Oxidative dehydrogenation of propane in a dense tubular membrane reactor’. React Kinet Cat Lett, 79(2), 351–356. Wang H, Cong Y and Yang W (2005), ‘Oxidative coupling of methane in Ba0.5Sr0.5Co0.8Fe0.2O3-δ tubular membrane reactors’, Catal Today, 104, 160–167. Wang H, Tablet C, Schiestel T, Werth S and Caro J (2006a), ‘Partial oxidation of methane to syngas in a perovskite hollow fiber membrane reactor’, Catal Commun, 7, 907–912. Wang H, Tablet C, Schiestel T and Caro J (2006b), ‘Hollow fiber membrane reactors for the oxidative activation of ethane’, Catal Today, 118, 98–103.

© Woodhead Publishing Limited, 2013

382

Handbook of membrane reactors

Wang H, Gopalan S and Pal U B (2011), ‘Hydrogen generation and separation using Gd0.2Ce0.8O1.9-δ-Gd0.08Sr0.88Ti0.95Al0.05O3-δ mixed ionic and electronic conducting membranes’, Electroch Acta, 56, 6989–6996. Wang W and Lin Y S (1995), ‘Analysis of oxidative coupling of methane in dense oxide membrane reactors’, J Membrane Sci, 103, 219–233. Wang W B, Cao X B, Gao W J, Zhang F, Wang H T and Ma G (2010), ‘Ammonia synthesis at atmospheric pressure using a reactor with thin solid electrolyte BaCe0.85Y0.15O3−δ membrane’, J Membrane Sci, 360, 397–403. White J H, Schwartz M and Sammells A F (1998), Solid state proton and electron mediating membrane and use in catalytic membrane reactors. US Patent 5821185. Wu Z, Wang B and Li K (2010), ‘A novel dual-layer ceramic hollow fibre membrane reactor for methane conversion’, J Membrane Sci, 352, 63–70. Xie Y H, Wang J D, Liu R Q and Su X T (2004), ‘Preparation of La1.9Ca0.1Zr2O6.95 with pyrochlore structure and its application in synthesis of ammonia at atmospheric pressure’, Solid State Ionics, 168, 117–121. Zeng Y and Lin Y S (2000), ‘Oxygen permeation and oxidative coupling of methane in yttria doped bismuth oxide membrane reactor’, J Catal, 193, 58–64. Zeng Y, Lin Y S and Swartz S L (1998), ‘Perovskite-type ceramic membrane: synthesis, oxygen permeation and membrane reactor performance for oxidative coupling of methane’, J Membrane Sci, 150, 87–98. Zhang C, Jin W, Yang C and Xu N (2009), ‘Decomposition of CO2 coupled with POM in a thin tubular oxygen-permeable membrane reactor’, Catal Today, 148, 298–302. Zhang F, Yang Q, Pan B, Xu R, Wang H and Ma G (2007), ‘Proton conduction in La0.9Sr0.1Ga0.8Mg0.2O3−α ceramic prepared via microemulsion method and its application in ammonia synthesis at atmospheric pressure’, Materials Lett, 61, 4144–4148. Zhang P, Chang X,Wu Z, Jin W and Xu N (2005), ‘Effect of the packing amount of catalysts on the partial oxidation of methane reaction in a dense oxygen-permeable membrane reactor’, Ind Eng Chem Res, 44, 1954–1959. Zhang Y, Liu J, Ding W and Lu X (2011), ‘Performance of an oxygen-permeable membrane reactor for partial oxidation of methane in coke oven gas to syngas’, Fuel, 90, 324–330. Zhu D C, Xu X Y, Feng S J, Liu W and Chen C S (2003), ‘La2NiO4 tubular membrane reactor for conversion of methane to syngas’, Catalysis Today, 8, 151–156. Zhu X, Li Q, Cong Y and Yang W (2008), ‘Syngas generation in a membrane reactor with a highly stable ceramic composite membrane’, Catal Commun, 10, 309–312. Zhu X, Li Q, He Y, Cong Y and Yang W (2010), ‘Oxygen permeation and partial oxidation of methane in dual-phase membrane reactors’, J Membr Sci, 360, 454–460.

© Woodhead Publishing Limited, 2013

Applications of dense ceramic membrane reactors

8.7

Appendix: nomenclature

8.7.1

Notation

•CH3 e′ E F FO 2

methyl radicals electron theoretical electromotive force, V Faraday constant, 96.2 kJ/mol.V oxygen permeation rate, mol/s

ΔG h•

Gibbs free energy, kJ/mol electron hole

ΔH 0298

reaction heat at standard state, kJ/mol

I N

electrical current, A electron number

x OO

lattice oxygen

VO••

oxygen vacancy

© Woodhead Publishing Limited, 2013

383

9 Chlor-alkali technology: fundamentals, processes and materials for diaphragms and membranes P. MILLET, University of Paris (11), France

DOI: 10.1533/9780857097347.2.384 Abstract: Ion-conducting materials are used as cell separators in electrolysis cells for the double purpose of carrying electric charges between electrodes and separating the products formed at each electrode. The purpose of this chapter is to provide an overview of chlor-alkali technology and associated cell separators. After a brief historical review of the chlor-alkali process, the main reaction characteristics (thermodynamics, cell reactions and kinetics) are detailed in Section 9.1. Main chlor-alkali technologies are described in Section 9.2. Main cell separators are described in Section 9.3 (diaphragm materials) and in Section 9.4 (membrane materials). Some improved electrolysis concepts are described in Section 9.5. Key words: chlorine, soda, electrolysis, electrolysers.

9.1

Introduction

Chlorine and sodium hydroxide (caustic soda) are among the top ten chemicals produced in the world. They are involved in the manufacturing of a wide variety of products (e.g., pharmaceuticals, detergents, deodorants, disinfectants, herbicides, pesticides and plastics). In 2010, the world production of chlorine amounted to approximately 50 ktons/year. The synthesis of chlorine by the electrolysis of brine is attributed to Cruikshank in 1800, but it took almost 90 years before the electrolytic method was used successfully on a commercial scale. Although the laws that govern the electrolysis of aqueous solutions were formulated by Faraday in 1833, electrochemical processes could not develop through the nineteenth century because of a lack of electrical generation capacity. The development of the dynamo (around 1865) allowed Edison, Siemens, Varley, Wheatstone and others to invent generators of electricity with large capacity and high efficiency. This in turn opened the way to industrial electrolysis processes, 384 © Woodhead Publishing Limited, 2013

Chlor-alkali technology

385

for example, water electrolysis, aluminium electrosynthesis and electrolytic production of chlorine and caustic soda. In the early stages of development, the main challenge during the electrolysis of NaCl solutions was that of achieving continuous separation of chlorine generated at the anode and sodium hydroxide produced at the cathode. While it was easy to keep the chlorine and hydrogen gases in U-shaped tubes, the sodium hydroxide formed at the cathode reacted with chlorine to form sodium hypochlorite. The British scientist Charles Watt is credited with having devised the concept of a current-permeable separator (1851), which allowed the electric current to pass but kept the anode and cathode products separated. At that time, the separator was made of a fibrous and porous inorganic fabric impregnated with liquid electrolyte and for this reason was called a diaphragm. An alternative method of producing chlorine and caustic soda involves the use of mercury cells (Castner−Kellner process), whose cathode reactions differ from those occurring in diaphragm cells. The first patent dealing with the mercury cell was British patent number 4349, issued in 1882 to Nolf, but the mercury-cell process for producing chlorine and caustic soda was developed by Hamilton Y. Castner and Karl Kellner in the USA (1892). This kind of cell was in continuous operation from 1897 to 1960. In parallel to this, Brichaux and W.Wilsing of Solvay built the first ‘long’ cell in Germany (1898). In the late 1960s, concern over environmental problems associated with mercury-cell plants initiated a revival of interest in the membrane-cell technology. As a spin-off of the space program, DuPont developed a perfluorinated ion-exchange membrane while performing fuel cell research. This membrane trademarked as ‘Nafion®’ was found to have good ion-exchange properties as well as high resistance to the harsh environment of the chlor-alkali cell and was therefore increasingly used at industrial scale. Alternatively, the electrolysis of hydrochloric acid solutions is also used to produce chlorine. Some historical details mentioned in this chapter are taken from a detailed review of the chlor-alkali industry.1

9.1.1 Thermodynamics In a hydrochloric acid electrolysis cell, electricity is used to produce chlorine at the anode and reduce protons into hydrogen at the cathode according to the following electrode reactions: Anode: Cl− → 1/2 Cl2(g) + 1e−

[9.1]

Cathode: H3O+(aq) + 1e− → 1/2 H2(g) + H2O

[9.2]

Full reaction: H3O+(aq) + Cl−(aq) → 1/2 Cl2(g) + 1/2 H2(g)

[9.3]

© Woodhead Publishing Limited, 2013

386

Handbook of membrane reactors

At equilibrium, the amount of electricity (nFE) required to split one mole of hydrochloric acid is equal to the change of Gibbs free energy ΔGd of the hydrochloric acid dissociation Reaction [9.3]: ΔGd – nFE = 0

where ΔGd > 0

[9.4]

•

n = 1 (number of electrons exchanged during the electrochemical splitting of HCl). • F = ≈ 96 485 C.mol−1 (Faraday). • E = thermodynamic voltage (Volt) associated with Reaction [9.3]. • ΔGd = free energy change in J.mol−1 associated with Reaction [9.3]. ΔGd is a function of both operating temperature and pressure, and thus : ΔG Gd (T P )

ΔHd (T , P P)) − T ΔSd (T , P ) > 0

[9.5]

ΔHd(T,P) and ΔSd (T,P) are respectively the enthalpy change (J.mol−1) and entropy change (J.mol−1.K−1) associated with Reaction [9.3]. To split one mole of HCl, ΔGd (J.mol−1) of electricity and TΔSd (J.mol−1) of heat are required. The thermodynamic electrolysis voltage E in Volt is defined as: E(T , P ) =

ΔGd (T , P ) nF

[9.6]

The thermo-neutral voltage V in Volts is defined as: V (T , P ) =

ΔH (T , P ) nF

[9.7]

In standard conditions of temperature and pressure (T° = 298 K, P° = 1 atm), HCl is dissolved in liquid water, H2 and Cl2 are gaseous. The enthalpies and entropies of formation of reactants and products appearing in Reaction [9.3] at 298 K are compiled in Table 9.1. From the data of Table 9.1, the standard free energy, enthalpy and entropy changes for Reaction [9.3] are: ΔHd° (

1

ΔSd° (

)

57 J mol −1 .K −1 ) = +243.57

ΔG Gd° (

)

1

V

ΔH Δ Hd° (

) / 2F ≈ 1.74 V

E

ΔG Δ Gd° (

) / 2F ≈ 1.36 V

© Woodhead Publishing Limited, 2013

Chlor-alkali technology

387

Table 9.1 Enthalpies and entropies of formation at 298 K

(

)

(

)

Compound

State

° Δf H298 kJmoll−1 kJ

° S298 JK K −1 mol 1

H2 Cl2 O2 H3O+ Cl− H 2O HCl

g g g aq aq l g

0 0 0 0 −167.54 −285.25 −92.35

130.74 223.09 205.25 0 55.13 70.12 186.79

g = gas phase; l = liquid phase; aq = in aqueous solution with unit activity.

A voltage term T. ΔSd° / ( F ) = .38 V must be added to the thermodynamic voltage E° to provide the heat required by Reaction [9.3]. In a brine electrolysis cell, electricity is used to produce chlorine at the anode (the reaction is pH independent) and reduce water into hydrogen and caustic soda (the reaction is pH dependent) at the cathode according to the following electrode reactions: Anode: 2Cl−(aq) → Cl2(g) + 2e−

[9.8]

Cathode: 2H2O(aq) + 2e− → H2(g) + 2OH−(aq)

[9.9]

Full reaction: 2NaCl(aq) + 2H2O(aq) → Cl2(g) + H2(g) + 2NaOH(aq)

[9.10]

Therefore, the cell voltage of Reactions [9.3] and [9.10] is significantly different (see next section). It should also be noted that the thermodynamic cell voltage E° is a function of operating temperature. According to Mussini and Longhi, the higher the operating temperature, the lower the thermodynamic voltage:2 E(T , P )

9.1.2

.484867 3, 958492 10 4 T − 2.750639 × 10 −6 T 2

[9.11]

Effect of pH on cell voltage

From the data of Table 9.1, it can be seen that the enthalpy and entropy changes for the water decomposition reaction (H2O → H2 + 1/2 O2) are ΔHd° ( 2 ) = 285.840 kJ.mol −1 and

© Woodhead Publishing Limited, 2013

388

Handbook of membrane reactors

ΔSd° ( 2 ) = 163.15 J.mol −1 .K −1 . Therefore, the standard free energy change is ΔG Gd° ( ) ΔHd° ( 2 ) − T .ΔSd° (H 2 O) = .22 kJ.mol −1 . The thermodynamic voltage required to decompose water into hydroGd° ( 2 ) / F ≈ 1.23 V , a value close gen and oxygen is therefore E to the voltage required to evolve chlorine from NaCl (1.36 V). However, the electric potential of the anode and the cathode varies markedly with the pH of the electrolyte. Let us consider the water splitting reaction in acidic media: Anode (+): H2O(liq) → 1/2 O2(g) + 2H+ + 2e−

[9.12]

Cathode (−): 2H+ + 2e− → H2(g)

[9.13]

Full reaction: H2O(liq) → H2(g) + 1/2 O2(g)

[9.14]

Let E+ and E− be the potential of anode and cathode, respectively. From the Nernst equation, it follows:

E+

° EH + 2 O / O2

( )() ( )

2 aH fO1/22 RPG T Ln nF aH 2O

[9.15]

At 298 K, when the pressure of oxygen is one bar (ideal gas), E+ ≈ 1.23 − 0.06pH.

E

° EH

2 /H

+

+

a2 + RPG T Ln H ≈ −0 06 pH nF fH 2

[9.16]

At 298 K, when the pressure of hydrogen is one bar (ideal gas), E− ≈ −0.06pH. Therefore, under the same conditions, the cell voltage Ecell = E+ − E− = 1.23V. Now let us consider the water splitting reaction in alkaline media : Anode (+): 2OH− → H2O + 1/2 O2(g) + 2e−

[9.17]

Cathode (−): 2H2O + 2e− → H2(g) + 2OH−

[9.18]

Full reaction: H2O(liq) → H2(g) + 1/2 O2(g)

[9.19]

© Woodhead Publishing Limited, 2013

Chlor-alkali technology

389

From the Nernst equation, it follows: 0 E + = EH + 2O / H2

2 aH RPG T 2O Ln 2 nF fH 2 aHO −

[9.20]

At 298 K, when the pressure of oxygen is one bar (ideal gas), E+ ≈ 1.23 + pKe − 0.06pH. 0 E = EH + 2O / H2

2 aH RPG T 2O Ln 2 nF fH 2 aHO −

[9.21]

At 298 K, when the pressure of hydrogen is one bar (ideal gas), E− ≈ pKe − 0.06pH. Therefore, under the same operating conditions, the cell voltage Ecell = E+ − E− = 1.23V (Fig. 9.1). The thermodynamic voltage required to split water molecules into hydrogen and oxygen is the same, whatever the pH. The only difference is that the potential of each electrode is shifted along the potential axis, as a function of electrolyte pH. The consequences are mostly to do with electrode material stability. In brine electrolysis, the pH in the anodic compartment (NaCl) is close to 7 and the oxygen-evolution reaction takes place at a potential (1.23 – 0.06 × 7 = +0.81 V), much lower than the potential of the chlorine evolution reaction (+1.36 V). Therefore, only pure oxygen should evolve. However, by using appropriate anode materials (such as carbon), chlorine gas evolution Acidic water electrolysis (pH = 0)

E/V

Alkaline water electrolysis (pH = 14)

1.23 NHE –0.06 pH ΔE = 1.23 V +0.390 V/NHE 0 NHE ΔE = 1.23 V –0.06 pH

–0.840 V/NHE

9.1 Electrode potential versus pH for the water splitting reaction. (NHE refers to normal hydrogen electrode.)

© Woodhead Publishing Limited, 2013

390

Handbook of membrane reactors

is favoured due to the poor kinetics of the oxygen-evolution reaction (OER) on carbon. Decomposition of hydrochloric acid into chlorine and hydrogen can also be used and requires less energy than the electrolysis of brine but does not produce caustic soda.

9.1.3

Electrocatalysis and cell kinetics

The free energy changes (ΔGd) of Reactions [9.3] and [9.10] correspond to the minimum energy required to transform one mole of reactant. However, there is no practical interest in performing electrolysis close to equilibrium conditions. On the contrary, it is necessary to significantly increase operating current densities in order to reduce investment costs. In conventional electrolysers, current densities ranging from 0.1 A.cm−2 up to several A.cm−2 are commonly achieved. To reach such values, it is necessary to overcome different kinds of current-dependent irreversibilities, such as charge transfer overvoltages (η in Volt) at electrode−electrolyte interfaces and ohmic drop across the electrolyte due to its ionic resistance (Re). Since voltage losses increase with operating current density, an economic optimum has to be found between energy (more efficient at low current density) and investment (less expensive at high current density) costs. To a certain extent, anodic (ηa) and cathodic (ηc) charge transfer overvoltages can be minimised by choosing appropriate electrocatalysts. In the early 1900s, the anode catalyst used for chlorine generation was either platinum or magnetite. However, because of the high cost of platinum and because of strong current density limitations (≈ 40 mA.cm−2) with the magnetite, R&D has been actively seeking for alternative materials. Graphite became the predominantly used anode material until the mid-1970s, although periodic replacement of anodes was necessary (because of corrosion, increasing interpolar distances yielded increasing energy consumption). This situation motivated the search for dimensionally stable anode structures. The chlor-alkali industry was revolutionised by H.B. Beer.3 His patented work was based on the idea of using thermal decomposition techniques to coat titanium substrates with mixed crystals of valve-metal oxides and platinum group metal oxides (with a valve-metal oxide content being more than 50%). Gradually, most technology suppliers started to use these anodes, which were given the registered trademark DSA® (for dimensionally stable anodes4). A schematic plot of electrode potential versus current density is provided in Figs 9.2 and 9.3 for the hydrochloric acid and brine electrolysis reactions, respectively.

9.1.4

Cell efficiency

The efficiency ε of an electrolysis cell relates the theoretical amount of energy Wt required to split one mole of reactant to the real amount of

© Woodhead Publishing Limited, 2013

Chlor-alkali technology

391

Potential (V) 2 Cl– → Cl2 + 2 e–

ηan

+1.36

E ° = 1.36 V

2 H+ + 2 e– → H2

0

ηcat

0.1

Current density (A.cm–2)

1.0

9.2 Electrode potentials for HCl electrolysis (pH = 0). Ohmic drop in electrolyte not included. Potential (V) 2 Cl– → Cl2 + 2 e–

ηan

+1.36

E ° = 2.2 V 0

H2O + 2 e– → H2 + 2 OH–

–0.84 ηcat

0.1

Current density (A.cm–2)

1.0

9.3 Electrode potentials for brine electrolysis (pH = 14). Ohmic drop in electrolyte not included.

energy Wr required by the process. Because of the above-mentioned irreversibilities, Wr > Wt. The cell efficiency can be defined as:

ε=

Wt Wr

[9.22]

where: •

Wr = (Ucell. I. t), Ucell is the cell voltage during operation (Volt), I is current (A) and t is the time in seconds

© Woodhead Publishing Limited, 2013

392

Handbook of membrane reactors

• Wt can be defined from the thermodynamic voltage E: Wt,ΔG = (E. I. t) • Wt can also be defined from the thermo-neutral voltage V: Wt,ΔH = (V. I. t) Therefore, two different definitions can be used to express the efficiency of the electrolysis cell. Since E and V are both functions of operating temperature (T) and operating pressure (P), and since Ucell is also a function of the operating current density j, the two different cell efficiencies can also be expressed as a function of T, P, j:

ε ΔG (T P j )

E(T , P ) U cell (T , P, j )

ε ΔH (T (T , P, j ) =

V (T , P ) U cell (T , P, j )

[9.23]

At low current densities, cell efficiencies close to 100% can be obtained. The cell efficiency decreases as the operating current density increases.

9.1.5

Parasite reactions

The primary products of brine electrolysis are chlorine, hydrogen and sodium hydroxide. Gaseous products are easily collected on top of the cells. However, hydroxyl ions formed at the cathode must be prevented from reaching the anode compartment where two parasite reactions could potentially take place: Cl2(g) + 2OH−(aq) → ClO−(aq) + Cl−(aq) + H2O

[9.24]

At the operating potential of the anode, the subsequent oxidation of hypochlorite ions (ClO−) can also take place: 6 ClO− (aq ) + 3H 2 O

2ClO3 (aq ) 4Cl − (aq ) + 6H 6 + + 3 / 2 O2

6e − [9.25]

As a result, caustic soda could be polluted by ClO−3 ions and Cl2 by O2. From a technology viewpoint, it is therefore necessary to prevent the migration of OH− to the anodic compartment by using a diaphragm or a membrane, or by choosing a cathodic reaction which does not form OH− during the first reaction step, as discussed in the next section.

9.2

Main electrolysis technologies

The electrochemical reactions taking place in a brine or in a HCl electrolysis cell are non-spontaneous transformations. On the other hand, the reverse reactions are spontaneous. It is therefore necessary to prevent

© Woodhead Publishing Limited, 2013

Chlor-alkali technology (a)

(b)

: Membrane

393

(c)

: Current collector

: Supported catalyst

9.4 Two-dimensional schematic diagrams of: (a) a gap-cell; (b) a zero-gap cell; (c) an SPE cell.

the spontaneous recombination of reaction products. Liquid electrolytes are not efficient separators, because chemicals can be transported rapidly from one electrode to the other by diffusion and/or by convection. Solids are more efficient and in practical cases, a charge-conducting separator is inserted between the anode and the cathode. Basically, three different cell concepts are used in the industry (Fig. 9.4). The first concept is called the ‘gap-cell’ (Fig. 9.4a). This is the most conventional configuration. Two planar electrodes (in fact, a cheap electron-conducting substrate surface covered by a thin layer of electrocatalyst) are placed face-to-face in an electrolyte, and a separator is inserted in the liquid electrolyte to prevent recombination of products. The separator can be an inert and porous material (a diaphragm) impregnated with the liquid electrolyte or a ion-conducting polymer (a membrane). Since transport of electric charges in an electrolyte follows Ohm’s law, the larger the distance between the two electrodes, the larger the ohmic losses and the lower the cell efficiency. Therefore, cell design is of uppermost significance in reducing energy consumption. When the electrolysis leads to the formation of gaseous products (e.g., H2 and Cl2 during brine electrolysis), the maximum current density is limited to a few hundred mA.cm−2, because non-conducting gaseous films tend to form over the electrode surfaces. Therefore, the concept of the ‘gap-cell’ pictured in Fig. 9.4a is not appropriate for gas-evolving electrodes. The second concept is called the ‘zero-gap cell’ (Fig. 9.4b). In that case, porous electrodes are pressed onto the separator, to reduce as much as possible the distance between anode and cathode (and corresponding ohmic losses). Products are released through the pores at the backside of the electrodes. This is an interesting concept when gases are produced, from at least one electrode. The third concept is called the ‘SPE cell’ (Fig. 9.4c). The term SPE stands for Solid Polymer Electrolyte. In such cells, ions which convey electric

© Woodhead Publishing Limited, 2013

394

Handbook of membrane reactors

charges from one electrode to the other are immobilised inside the membrane, which acts as a solid electrolyte. There is no liquid electrolyte in circulation in the electrolyser. This concept was first proposed for application in H2/O2 fuel cells in the 1950s, at the dawn of the US space programme, to solve the problems associated with operation in a low gravity environment and those due to the corrosion of metals in acidic media. The idea was used later for the development of water electrolysers. In the literature, it is referred to as SPE water electrolysis or PEM (proton exchange membrane or polymer electrolyte membrane) water electrolysis. In a PEM water electrolysis cell, electrodes are deposited at the surface of the membrane, and porous current collectors are pressed on each side of the cell. There is no liquid electrolyte in circulation. Only de-ionised water is circulated in the anodic chamber to feed the electrochemical reaction. A similar concept is used in high temperature fuel cells and electrolysis cells, using oxide-ion conducting membranes. However, SPE cells cannot be used in brine electrolysis because a ion-conducting aqueous solution of NaCl has to be supplied to the cell. The ‘zero-gap cell’ configuration is preferred. As discussed in the Introduction section, three main types of electrolytic cells have been used for the large scale production of chlorine and caustic soda: mercury, diaphragm and membrane cells. The main difference in these technologies lies in the manner by which the chlorine gas and the sodium hydroxide are prevented from mixing with each other to ensure generation of pure product. Alternatively, the electrolysis of hydrochloric acid solutions is also used to produce chlorine. Individual electrolysis cells can be electrically wired in parallel (monopolar electrolysers) or in series (bipolar electrolysers).

9.2.1

Brine electrolysis with diaphragm cells

A diaphragm is a porous barrier between the anode and cathode compartments intended to allow sodium ion flow through to balance the charge of the hydroxide ion formed and complete the flow of electricity in the circuit, also allowing a build-up of sodium hydroxide in the cathode compartment. The first commercial cell used for chlorine production is usually credited to the Griesheim Company in Germany (1888). This type of non-percolating diaphragm cell was predominantly used for Cl2 production in the early 1900s. It was using a porous cement diaphragm invented by Brauer in 1886. The cell was operated batch-wise with saturated potassium chloride solutions at 80–90°C and low current density (10–20 mA.cm−2) for several days until a concentration of 7% KOH was obtained. The cell voltage was close to 4 V and the current efficiency ≈ 70–80%. According to the literature, the first diaphragm cell developed in Great Britain was the Hargreaves–Bird cell, operated in 1890 by the United Alkali

© Woodhead Publishing Limited, 2013

Chlor-alkali technology

395

Company. Each cell consisted of a rectangular iron box lined with cement. The box was divided into three parts of two separate asbestos sheet diaphragms. Six carbon anodes were placed in the anode compartment, and the cathode was a copper gauze attached to the diaphragms. This basic approach of anchoring the diaphragm onto the cathode is still used in modern diaphragm cells. The anode compartment was filled with saturated brine, and during electrolysis chlorine evolved at the anode. The sodium ions, along with sodium chloride and water, percolated through the diaphragm into the cathode compartment. Back-migration of hydroxyl ions was suppressed by injecting CO2 and steam into the cathode compartment to form sodium carbonate. The major contribution of this cell was its vertical diaphragm configuration, which is the basis of modern cells. Twelve cells were run in series at 20 mA.cm−2, at 4 to 4.5 V, when 60% of the salt was converted to sodium carbonate. The schematic diagram of more recent diaphragm cells is pictured in Fig. 9.5. The anodes are formed of hollow expended titanium braids covered with a surface layer of RuO2. The cathodes are formed of steel braid or iron mesh, which are not affected by the hydroxide solution. The diaphragm is made of synthetic plastic or asbestos surrounding the cathode. Typical operating conditions and performances of a diaphragm cell are as follows: • • •

Operating pressure = atmospheric pressure. Operating temperature ≈ 85°C. In monopolar technology, the maximum current density ≈ 275 mA.cm−2. • The production of one ton of chlorine requires ≈ 1.7 t of NaCl, ≈ 3000 kWh and ≈ 2.5–3 tons of water vapour.

In the ‘gap-cell’ configuration (Fig. 9.4), the formation of gas bubbles at both anode and cathode lead to high apparent resistances. The problem is alleviated in ‘zero-gap’ configurations.

9.2.2

Brine electrolysis with mercury cells

In a mercury cell (Fig. 9.6), chlorine directly evolves at the anode (a DSA). But at the cathode, the formation of hydroxide ions is prevented by using liquid mercury. Thus, the hydrogen evolution reaction (HER) is replaced by the reduction of sodium ions and the formation of sodium−mercury metallic amalgam: Na+(aq) + xHg + 1e− → NaHgx

[9.26]

The potential of formation of a 0.2% Na–Hg amalgam is E = –1.78 V for 5 M NaCl. On mercury, the potential for the HER is very close

© Woodhead Publishing Limited, 2013

396

Handbook of membrane reactors Saturated brine (~ 300 g/l) H2

H2 Cl2

Cl2

Depleted brine + NaOH

Depleted brine + NaOH

9.5 Schematic diagram of a diaphragm cell for brine electrolysis.

(≈ −1.7 V). However, the formation of the amalgam is predominant. The HER is inhibited because the formation of hydrogen at the mercury−electrolyte interface tends to significantly increase the local pH, thus reducing the H2O/ H2 thermodynamic voltage. The Na–Hg amalgam entering the electrolysis cell has an Na-content of approximately 0.01%. On the exhaust side, the Na-content is close to 0.2% (at higher concentrations, hydrogen is evolved and the amalgam is increasingly viscous). The amalgam is then washed with de-ionised water in the decomposer along with a ballast of graphite beads. As a result, hydrogen is evolved, caustic soda is formed and the sodium content in the amalgam decreases, according to: NaHgx + 2H2O → 2NaOH + xHg + H2

[9.27]

The flow of liquid water in the decomposer is adjusted to obtain a 50 wt% NaOH solution and mercury is pumped back to the electrolysis cell. Typical operating conditions and performances of a mercury cell are as follows: • •

Operating pressure = atmospheric pressure. Operating temperature ≈ 85°C.

© Woodhead Publishing Limited, 2013

Chlor-alkali technology

397

Cl2

Saturated brine (~ 300 g/l)

Depleted brine (~ 250 g/l)

Hg–Na amalgam H2

Q

NaOH (~ 50 wt.%)

Deionised water

Hg

9.6 Schematic diagram of a mercury cell for brine electrolysis. (Q = heat flow (J/s).)

• In monopolar technology, the current density ≈ 1–1.5 A.cm−2 and the cell voltage ≈ 4.0 V. • The production of one ton of chlorine requires ≈ 1.7 ton of NaCl, ≈ 3500 kWh of electricity and no water vapour. The two main advantages of this process are that it produces very pure soda with no salt, and it avoids using asbestos. The disadvantage, however, is that mercury is released into the environment, sometimes at a concentration close to 100–200 g/ton of NaOH produced. This is due to unavoidable mechanical transfers of mercury to the solution. Metallic mercury is extremely insoluble in water but there are some bacteria in rivers or oceans that can convert it into dimethyl mercury, which can be taken up by organisms and passed along the food chain. The concentration increases along the way to the human endconsumer who eats fish and can receive doses of mercury which can cause brain damages. In the late 1960s, rising concerns over environmental issues led to severe emission standards (a threshold value of 1 g of mercury released per ton of NaOH produced has been set). However, this value is difficult to meet and as a result, fewer mercury cells are being built.

© Woodhead Publishing Limited, 2013

398

Handbook of membrane reactors

H2

Cl2 Na+

Depleted brine (~ 200 g/l)

Water

H2O

Saturated brine (~ 300 g/l)

Cl–

Cl– traces OH– traces

2Cl–(aq) → Cl2(s) + 2 e–

NaOH (~ 32 wt.%) OH–

2H2O(aq) + 2 e– → H2(g) + 2 OH–(aq)

9.7 Schematic diagram of a ‘membrane’ brine electrolysis cell.

9.2.3

Brine electrolysis with membrane cells

The main advantages of diaphragm and mercury brine electrolysis technologies, respectively a low cell voltage and an NaCl-free alkali, are combined in membrane technology. Actually, the membrane cell is a modification of the diaphragm cell in which the diaphragm is replaced with a permselective ion-exchange membrane that inhibits the passage of negative chloride ions but allows positive sodium ions to move through freely (Fig. 9.7). The membrane cell is either a ‘gap-cell’ or a ‘zero-gap’ cell (Fig. 9.4). A cationic ion-exchange membrane is placed between the anode and the cathode. The transport of current across the membrane is ensured selectively by hydrated sodium ions Na+. In most cases, the separator is a bi-layer membrane made of perfluorocarboxylic (PFCA) and perfluoro-sulfonic (PFSA) acid-based films, to prevent the transport of caustic soda from the cathodic to the anodic compartment. From a historical perspective, Diamond and Hooker Chemicals in the USA have demonstrated an interest in ion-exchange membranes since the mid-1940s. S. G. Osborne et al. of Hooker Chemical Corporation hold the original patent for the first membrane cell in 19525 and intensive R&D was carried out until the late 1950s. The introduction of Nafion® membrane on the market renewed interest in membrane-cell technology and initial problems associated with membrane cells were steadily being alleviated during the early 1970s. In addition, the development of metal anodes negated problems associated with graphite anode erosion. In 1975, Asahi Chemical started a pilot plant operation with ion-exchange membrane

© Woodhead Publishing Limited, 2013

Chlor-alkali technology

399

cells, using DuPont’s Nafion®. Since then, several membrane-cell technologies were developed in Japan, as a pollution-free chlor-alkali process, and Japan was the first major chlorine producing country to convert entirely to membrane-cell technology. Many chlor-alkali producers and technology licensing companies have refined and optimised membrane cell designs to realise low energy consumption and long life. The new cell designs incorporated the ‘zero-gap’ concept (Fig. 9.4), which eliminated the electrolyte gap between the electrodes and the membrane. In the membrane process, anodic and cathodic circuits are separated. It is therefore possible to use directly solid salt (sodium chloride) in place of saturated brine, which is more easily stored and handled. To prevent the pollution of the membrane by incorporation of di-valent metallic cations (mostly by calcium and magnesium cationic species) the NaCl-saturated brine has to be purified by addition of NaOH + Na2CO3. Thus, Ca and Mg concentrations are reduced down to the 1–5 ppm range. The solution is then filtrated using ion-exchange resins to further reduce the Ca–Mg concentration to about 0.02 ppm. It is also necessary to reduce the concentration of miscellaneous constituents such as Al, Fe, SiO2, iodine, sulfates, chlorates, Ba and Sr. In such aggressive media, the lifetime of the membranes is reported to be approximately three years. Compared to the diaphragm and mercury cells, the main advantages of the membrane processor are: (i) high purity chlorine; (ii) low energy consumption; and (iii) high flexibility. On the less positive side, (i) a high purity brine is required, because ion-exchange membranes are very sensitive to the presence of trace amounts of poisonous chemicals; and (ii) membranes are expensive. A typical chlorine production plant using membrane cells is pictured in Fig. 9.8. Electrolysers are operating at atmospheric pressure and ≈ 85°C. The main electrochemical characteristics of brine electrolysis cells using membranes are: (i) operating current density: 300–500 mA.cm−2; (ii) cell voltage: 3.0–3.6 V; (iii) NaOH concentration: 33–35 wt%; (iv) energy consumption: 2600–2800 kWh/ton Cl2 at 500 mA.cm−2; (v) efficiency: 50%; and (vi) steam consumption for concentrating NaOH to 50%: 180 kWh/ton Cl2. The production of one ton of chlorine requires ≈ 1.7 tons of NaCl and less than 1 ton of water vapour.

9.2.4

Electrolysis of aqueous hydrochloric acid

In the industry, gaseous HCl is often released as a by-product. It is not always possible to find a use for the acid, and discharge to waste treatment is usually not a satisfactory solution. However, like brine, aqueous solutions

© Woodhead Publishing Limited, 2013

400

Handbook of membrane reactors

9.8 Chlorine-alkali cell room with BL-2.7 membrane cells (courtesy of Uhde GmbH).

of HCl can be used to produce chlorine by electrolysis (Reaction [9.3]). The annual world production is close to 2 ktons/year (approximately 4% of the total chlorine world production). The recovery of chlorine from hydrochloric acid offers several industrial advantages: • • • •

no costs related to hydrochloric acid disposal; business is made independent from Cl2 and HCl prices; reduced investment costs compared to the installation of a traditional chlor-alkali plant of the same capacity; the process is interesting for regions with high Cl2 demand or excess caustic soda.

The schematic diagram of an HCL electrolysis unit is pictured in Fig. 9.9. The aqueous solution of HCl is formed by dissolution of gaseous HCl in deionised water. The HCl concentration in the circulating electrolyte is closely monitored, and in stationary conditions of operation the difference of HCl concentration between feed and exhaust solutions remains small (within 5 wt%). At pH = 0 (Fig. 9.10), the thermodynamic potential of the chlorine-evolving anode is close to +1.36 V whereas the thermodynamic voltage of oxygen evolution from water in acidic media is approximately +1.23 V. The difference is not very large but the kinetics of the OER are slow and require elevated operation potentials of ca. +1.6 V to proceed at a significant rate. Therefore, pure chlorine is evolved at the surface of the DSA.

© Woodhead Publishing Limited, 2013

Chlor-alkali technology Cl2 H2O

H2

HCl 22 wt. %

HCl 22 wt. %

D

Q

Q Gaseous HCl

17 wt.

17 wt.

HCl 30 wt. %

S

Q

401

S

S

9.9 Schematic diagram of a hydrochloric acid electrolysis cell. Q = heat exchangers. S = storage. D = diaphragm.

Potential (V) H2O → ½O2 + 2 H+ + 2 e–

ηan +1.36 +1.23

2 Cl– → Cl2 + 2 e–

E ° = 1.36 V 2 H+ + 2 e– → H2

0

ηcat

0.1

Current density (A.cm–2)

1.0

9.10 HCl electrode potentials at pH = 0. Ohmic drop in electrolyte not included.

9.2.5

Advantages and limitations of each technology

In 2005, the distribution of the production between the different processes was approximately: diaphragm cell (45%); mercury cell (20%); membrane cell (35%). Membrane-cell technologies were developed for licensing by

© Woodhead Publishing Limited, 2013

402

Handbook of membrane reactors

Table 9.2 Comparison of main advantages and limitations of the three main brine electrolysis processes Process

Advantages

Limitations

Diaphragm

• Good energy efficiency • High purity chlorine • Uses brine

Mercury

• Obtaining concentrated soda with no salt • No asbestos • No water vapour consumption • High current density • No asbestos and no mercury • High purity caustic soda with no salt ( 10 mS.cm−1); (ii) a poor electronic conductivity; (iii) a good chemical and mechanical stability; (iv) a high thermal conductivity (λ > 0.1 J.s−1.m−1.K−1); (v) a limited permeability to hydrogen and oxygen; and (vi) a good permselectivity. Proton conductivity, affinity to

© Woodhead Publishing Limited, 2013

Chlor-alkali technology CF2

CF2 + SO3

F2C

CF2

O

SO2

O +

FSO2CF2COF

2 F3CFC

FSO2CF2CF2OCFCF2OCFCFO CF3

CF2

405

FSO2CF2COF

F–

Na2CO3 Δ

FSO2CF2CF2OCFCF2OCF

CF2

CF3

CF3

9.11 Synthesis process for Nafion® membrane co-monomer.

–[(CF2

CF2)m

CF

CF2]n

O CF2 CF

CF3

z

O m = 5–13.5 n = ca 1000

CF2

z = 1, 2, 3,...

CF2 SO3H

9.12 The chemical composition of Nafion®.

water, hydration stability at high temperatures, electro-osmotic drag, and mechanical, thermal and oxidative stability − all these key properties are directly determined by the chemical microstructure of the polymer. Among the earliest concepts that were set forth regarding microstructure of PSA are those of Gierke et al. and Yeager and Steck.15 From a historical perspective, the cluster-network model of Gierke et al.16 has been used for many years for interpreting the properties of Nafion® membranes (especially ion and water transport and ion permselectivity). In this model (Fig. 9.13), it is assumed that there are ≈ 40 Å-in-diameter clusters of sulfonate-ended perfluoroalkyl ether groups that are organised as inverted micelles and arranged on a lattice. These micelles are connected by pores or channels that are ≈ 10 Å in size. These -SO3– coated channels were invoked to account for inter-cluster transfer of cations and ion conductivity. In the model of Yeager and Steck, clusters do not have a strict geometry and their geometrical distribution has a lower degree of order. Most

© Woodhead Publishing Limited, 2013

406

Handbook of membrane reactors SO3–

SO3– SO3–

SO3– SO3– SO3–

SO3–

SO3– SO3–

SO3–

SO3–

SO3–

SO3–

SO3–

SO3– SO3–

SO3–

SO3–

SO3–

SO3–

9.13 Gierke model of nanoclusters in perfluoro-sulfonic acid materials.

importantly, intermediate interphase domains are found between hydrophobic and hydrophilic regions. PFSA polymers can be considered as a homogeneous and isotropic two-phase medium, a mixture of hydrophobic regions concentrating fluoro-carbon backbones and hydrophilic regions containing water, where proton conductivity takes place.

9.4.2

Ionic conductivity

The role of a membrane in an electrolysis cell is to separate reaction products and to convey electric charges (ions) between electrodes. Therefore, ionic conductivity is a key physical property of membrane materials. Charge transport ions in PFSA polymers are usually H+ (in PEM fuel cells, and in water electrolysis and HCl cells), or Na+ (brine electrolysis). These monovalent charge carriers jump from one sulfonate group to another, the driving force of the motion being the electric field set across the membrane during operation. It is known from the PEM fuel cell literature17 that the ionic resistance of Nafion® membranes depends on the hydrogen partial pressure. In particular, it has been reported that a surface proton transfer mechanism along pore walls predominates at low gas pressure values and that the membrane resistivity decreases when H2 partial pressure is lowered. However, this effect has been observed only for low H2 partial pressures 10 kPa ) and ionic conductivity of Nafion® is mainly due to both ion ( PH 2 migration in the aqueous phase and proton tunnelling between adjacent sulfonate groups in narrow pores. In conventional electrolysers, the operating current density ranges from a few hundred mA.cm−2 to 1 A.cm−2. A maximum ohmic drop of 100 mV

© Woodhead Publishing Limited, 2013

Chlor-alkali technology

407

at 1 A.cm−2 is a good target for water electrolysers. Therefore, the surface resistance should be less than 0.1 Ω.cm2. Membranes have a typical thickness of 200 µm (this is the case for commercial Nafion® products). At 80°C, the resistivity of 1100 EW Nafion® in H+ form is ρ ≈ 5.0 Ω.cm and the corresponding conductivity is σ ≈ 0.2 S.cm−1.18 The surface resistance of the membrane is therefore 0.1 Ω.cm2. However, the ionic mobility of H+ and Na+ is quite different and the resistivity of Nafion® membranes in Na+ form is approximately seven times higher.

9.4.3

Permselectivity

When the membrane is used in a SPE cell (no liquid electrolyte), there is only one kind of charge carrier: cations in cation-exchange materials (e.g., proton in PEM water electrolysis cells) and anions in anion-exchange materials. When the membrane is used in a ‘gap-cell’ or in a ‘zero-gap cell’ (Fig. 9.4) with a liquid electrolyte, both cations and anions participate in the transport of electric charges, according to their charge and individual mobility. Donnan exclusion prevails only at low electrolyte concentrations. Therefore, anions (Cl− and/or OH− in the case of brine electrolysis) can penetrate cationic ion-exchangers and can contribute to charge transport. Ionpairing (Na+-Cl− or Na+-OH−) can explain the phenomenon. In spite of this, the permselectivity of a membrane is much larger than the permselectivity of a porous diaphragm impregnated with liquid electrolyte, and brine electrolysis using a membrane cell is used to produce high purity caustic soda with no salt (2.7 is generally required to avert carbon deposition on the surface of the catalyst (and its consequent deactivation). The reaction is usually carried out at 20–30 bar, and is limited by equilibrium considerations mentioned above as well as structural limits related to creep stress in the high temperature alloy tubes. The products of steam reforming are CO and H2 (Equation [10.1]), along with carbon dioxide (CO2) which is produced by the slightly exothermic water-gas shift (WGS) reaction: CO + H2O ↔ CO2 + H2

[10.2]

(ΔH0 = −41.1 kJ/kmol at 25°C) that also occurs freely on the Ni catalyst. Because of the relatively high reaction temperatures and activities of the catalysts, both reactions (Equations [10.1] and [10.2]) operate near equilibrium in the reactor operating conditions. •

In autothermal reforming (ATR), heat for the endothermic reforming reaction is provided internally by partial combustion of the feed stream. Air can be used as oxidant when a significant amount of nitrogen is acceptable in the outlet syngas, for example, in ammonia synthesis. In hydrogen production, however, downstream nitrogen separation is not economical, and thus high purity oxygen is preferred as oxidant. ATR employs an adiabatic reactor composed of (in series) a burner, a conical combustion/mixing zone, and an Ni-based catalyst bed where

© Woodhead Publishing Limited, 2013

Membranes in systems for electric energy and hydrogen production 419 chemical equilibrium is attained. ATR permits higher temperatures and pressures than FTR because the FTR reactor wall must permit the efficient transfer of large amounts of heat. In industrial operation the syngas outlet temperature is typically 900–1100°C, with pressures as high as 70 bar. The steam-to-carbon mole ratio is generally 0.6–1.5 with an oxygen-to-carbon mole ratio of 0.55–0.65. Catalytic partial oxidation (CPO) is a novel technology based on the same principles as ATR, except that all reactions take place catalytically, without upstream combustion; after careful mixing, the reactants are sent directly on to the catalyst. CPO catalysts are usually based on noble metals (Pt, Pd, Rh, Ir)1 whose high reactivity permits very short contact times (0.1–10 ms) and thus high space velocities, which enables the design of compact reactors. Apart from considerations of hydrogen supply pressure, the choice of reforming technology is governed by process economics. At small scales, FTR is the technology of choice. In O2-blown ATR, the cost of cryogenic air distillation can constitute up to 40% of the total investment.2 Since air separation units exhibit sharp economies of scale up to 90 000 Nm3/h, O2-blown ATR is considered competitive for hydrogen output greater than 250 000 Nm3/h.3 Typical plant flowsheets for FTR and ATR are shown in Figs 10.1 and 10.2, respectively. At the reactor exit, syngas is usually cooled down by evaporating – but not superheating – steam. This constraint is applied in order to reduce the risk of metal dusting, a carburization-based corrosion process that rapidly destroys steel alloys that are in contact with gas streams rich in carbon compounds, especially between 400°C and 800°C. Fast cooling of the process gas stream by steam generation allows the temperature of the metal surface to be kept below this critical range, thus obviating the need for special coatings and surface treatments designed to mitigate metal dusting. Syngas produced by steam reforming contains a substantial amount of CO that can be converted to H2 by operating the WGS reaction (Equation [10.2]) at a temperature lower than that at the reformer exit. Various catalysts are used in order to promote the WGS reaction in different temperature ranges. Between 330°C and 510°C, the catalyst is typically chromium-stabilized iron oxide promoted by small amounts of copper (1–2% by weight), while more active copper- and zinc-based catalysts are preferred in the 180–330°C temperature range. The arrangement of the WGS reactors is governed by the desired degree of CO conversion and the steam-to-carbon ratio in the charge. In some cases, it is convenient to split the overall conversion into separate (usually two) adiabatic reactors with intermediate cooling in order to combine fast kinetics

© Woodhead Publishing Limited, 2013

420

Handbook of membrane reactors Adiabatic pre-reformer Fired tubular reformer

Desulfurization

Natural gas Pressure swing PSA Hydrogen adsorber off-gas fuel output

Combustion air

Steam export

HT WGS Condensate

Make-up boiler feedwater (BFW)

10.1 Layout of a hydrogen production plant from natural gas based on a fired tubular reformer.

and high temperature heat recovery in the first (hot) stage while insuring a high degree of CO conversion by means of a second (cold) stage. After WGS, the syngas is cooled to ambient temperature for hydrogen recovery, generally performed with pressure swing adsorption (PSA), a purification technique based on the principle of selective concentration of certain gaseous species at the surface of microporous solid sorbents such as zeolites and activated carbon. Sorbents used for hydrogen purification selectively adsorb species other than hydrogen and helium, with an impurity loading proportional to partial pressure of contaminants. PSA operates at constant temperature, ‘swinging’ between two pressure levels, adsorbing impurities at high pressure and releasing them at low pressure. Since the operating cycle is composed of at least two phases (contaminant absorption and regeneration), a minimum of two beds in parallel is needed to ensure continuous operation. In industrial practice, PSA plants employ a higher number of beds (typically 4 to 12) in order to reduce the consumption of high pressure, high purity hydrogen during bed repressurization. PSA

© Woodhead Publishing Limited, 2013

Membranes in systems for electric energy and hydrogen production 421 Combustion air Waste N2

O2

Air separation unit

Adiabatic pre-reformer Air

Autothermal reformer Natural gas

Desulfurization

Steam export HT WGS

Make-up BFW

Condensate

PSA off-gas fuel Pressure swing adsorber

Hydrogen output

10.2 Layout of a hydrogen production plant based on autothermal reforming of natural gas.

produces hydrogen at a pressure slightly lower than the feed stream (the pressure drop is typically < 1 bar), with a purity higher than 99.9% and with a H2 recovery efficiency in the range 75–92% depending on the composition of the feed stream, equipment design and operating parameters. The contaminant or ‘purge’ stream (including the residual H2) exits the PSA at nearly ambient pressure. The purge gas is generally combusted to recover its (not insignificant) heating value. Natural-gas-to-hydrogen conversion efficiency, defined as the ratio of the lower heating values of the hydrogen output and the natural gas input, is in the range 70–80% for both the FTR and ATR configurations shown in Figs 10.1 and 10.2. Coal gasification Gasification is the preferred method of producing syngas from heavy fossil feedstocks such as coal, petroleum coke (petcoke) and refinery residues. Gasification is usually an adiabatic process in which feedstock is partially

© Woodhead Publishing Limited, 2013

422

Handbook of membrane reactors

oxidized with a substoichiometric amount of oxygen (termed the ‘blast’) as well as steam (to moderate the reaction temperature). Gasifiers can be conceptually classified into three categories, depending on how the gaseous and solid reactants are contacted:4 •

In moving-bed (sometimes called fixed bed) gasifiers, the coal moves slowly downward by gravity, usually flowing countercurrent to the gas produced in the bed. As the coal descends, it is first preheated and then pyrolyzed, releasing reactive, volatile gaseous hydrocarbons. The remaining char falls into the gasification zone at the bottom where it reacts with oxidant and steam at slagging temperatures (>800–900°C). This countercurrent arrangement leads to low oxidizer consumption and low syngas outlet temperatures (450–550°C); however, not all products of pyrolysis are completely cracked, but instead become entrained in the output syngas stream. When cooled, these tar compounds condense into a complex, often toxic liquor which must be removed and treated, or recycled back to the bed. This is a significant complication – one of the critical points of this technology – along with the difficulty of using coals that tend to produce fines which impede the countercurrent passage of syngas. • In fluidized bed gasifiers, the feedstock is uniformly distributed in a bed of inert particles (or the ash itself, in the case of coal) that is fluidized by injecting the blast and steam with carefully controlled velocities at locations near the bottom of the reactor. The operating temperature must be kept below the softening point of the ash (~800–900°C) because its agglomeration inhibits proper bed fluidization. These low temperatures significantly limit the reaction kinetics and thus constrain fluidized beds to the gasification of reactive feedstocks such as low rank coals and biomass rather than hard coal. The uniform distribution of material within the reactor results in the inadvertent removal of partially reacted fuel along with the ash during bleeding, generally limiting the extent of carbon conversion to 73.8 bar) and dehydrated ( 50 nm), while metallic membranes can be categorized into supported and unsupported. Generally, inorganic membranes are stable between 200°C and 800°C and in some cases can operate at elevated temperatures (ceramic membranes over 1000°C).

© Woodhead Publishing Limited, 2013

Palladium-based membranes for hydrogen separation

459

Membranes can be subdivided into four categories according to their geometry: tubular, hollow fiber, spiral wound and flat sheet (Basile et al., 2011). Tubular membranes are used most frequently, even though they require relatively high volumes and involve high costs. Gas separation can be basically attributed to four mechanisms: (i) Knudsen diffusion, (ii) molecular sieving, (iii) surface diffusion and (iv) solution−diffusion (Adhikari and Fernando, 2006; Kluiters, 2004; Koros and Fleming, 1993). Separation based on Knudsen diffusion occurs when the pore diameter of the effective barrier layer is smaller than the mean free path of the gas being separated, while separations based on molecular sieving operate on a size-exclusion principle. Surface diffusion can occur in parallel with Knudsen diffusion: gas molecules are adsorbed on the pore walls of the membrane and migrate along the surface. The solution−diffusion separations are based on both the solubility and mobility of one species in a solid effective barrier. A gas molecule is adsorbed on one side of the membrane, dissolves in the membrane material, diffuses through the membrane and desorbs on the other side of the membrane. If diffusion through the membrane takes place in the form of ions and electrons or as atoms (e.g., for hydrogen transport through dense metal), the molecule must split up after adsorption and recombine after diffusing through the membrane. The most commonly investigated performance characteristics of gas separation membranes are flux, permeability coefficient/permeance and selectivity. The flux is the amount (mass or moles) of gas that permeates through the membrane per unit time and unit surface area; the permeability coefficient is the quantitative expression of a specific measure of gas moving through a membrane; and the selectivity is the separating ability of a given membrane (Ockwig and Nenoff, 2007). Criteria for selecting membranes are complex and dependent upon the application. The main parameters affecting the choice are productivity, permeation rate, separation selectivity, membrane life time, mechanical and chemical integrity at the relevant operating conditions and, particularly, the cost. The relative importance of each of the above parameters varies according to the application. However, selectivity and permeation rate are clearly the most critical properties of a membrane. The higher the selectivity, the more efficient the process, the lower the driving force required to achieve a given separation and thus lower the operating costs of the separation system. The higher the flux, the smaller the membrane area required and thus, the lower the capital cost of the system (Lu et al., 2007). Hydrogen selective membranes are among the most interesting and promising applications: for these purposes, dense metallic membranes are the dominant technology thanks to their high selectivity and good permeability. Although the hydrogen permeability of metals such as niobium,

© Woodhead Publishing Limited, 2013

460

Handbook of membrane reactors

vanadium and tantalum is higher than that of palladium at temperatures between 0°C and 700°C (Basile et al., 2011; Gallucci et al., 2007b), these metals have greater resistance to hydrogen transport than does palladium. Dense palladium membranes are therefore preferred for use in this application. The molecular transport of hydrogen in palladium membranes occurs through a solution−diffusion mechanism. Depending on temperature, pressure, gas mixture composition and thickness of the membrane, each of the key steps of this mechanism may control hydrogen permeation through the dense film. As a result, the hydrogen permeating flux can be expressed by means of the following equation: JH2

H2

/δ

(p

n H 2 rett

n − pH 2

perm

)

[11.1]

where n (variable in the range 0.5–1) is the dependence factor of the hydrogen flux to the hydrogen partial pressure, J H the hydrogen flux permeat2 ing through the membrane, Pe the hydrogen permeability, δ the membrane n n thickness, pH 2 , ret and pH 2 perm the hydrogen partial pressure in the retentate (high pressure) and permeate (low pressure) sides, respectively. When the pressure is relatively low and the Pd layer is relatively thick (>10 micron), the diffusion of atomic hydrogen through the membrane is assumed to be the rate-limiting step and the factor n is equal to 0.5 (Sieverts– Fick law). The commercialization of pure palladium membranes is still limited by several factors. Above all, pure Pd membranes undergo embrittlement when exposed to hydrogen at a temperature lower than 300°C and pressure below 2 MPa. This phenomenon may produce pinholes on the membrane, which will negatively affect its hydrogen permselectivity (Lemier and Weissmuller, 2006). Furthermore, pure palladium membranes are deactivated by carbon compounds at temperatures higher than 450°C (Amandusson et al., 2000; Gallucci et al., 2007a; Li et al., 2000) and can be irreversibly poisoned by sulfur compounds (Castro et al., 2002; Kulprathipanja et al., 2005). The final major limiting factor is their high cost. All these disadvantages may be overcome through the use of Pd-alloy, rather than pure Pd, membranes, obtained by alloying the Pd with Ni, Cu and especially Ag (Li et al., 1993). The optimal silver concentration seems to be in the region of 23 wt%.

11.2

Membrane preparation techniques

A suitable hydrogen membrane must have the following features: (i) high selectivity for hydrogen; (ii) high permeability, in order to operate with high flows and limited surfaces, and (iii) good chemical and structural stability, in order to avoid deterioration under exertion.

© Woodhead Publishing Limited, 2013

Palladium-based membranes for hydrogen separation

461

The most promising configurations for a hydrogen selective membrane are thus: 1. ceramic support + thin hydrogen selective layer, 2. metallic support + metallic interdiffusion barrier + thin hydrogen selective layer. It must be noted that the interdiffusion barrier plays a unique role in preventing the interdiffusion of the Pd-alloy in the steel support. This barrier must have good chemical stability and reduced thickness (of only a few microns) in order to allow the gas to cross. One of the main objectives in creating an interdiffusion barrier is to obtain a layer that is extremely adherent, dense, homogeneous and with a consistent thickness. To date, several types of barrier materials have been tested in various studies: TiN (Shu et al., 1996), Al2O3 (Yepes et al., 2006), YSZ (Zhang et al., 2009). A number of methods can be used in the production of Pd-based membranes, depending on factors such as the nature of the metal used in the selective layer, the manufacturing facilities, the required thickness, surface area, shape, purity and so on. Nevertheless, no single method can produce a membrane that is able to combine all these factors effectively. The choice of production method is therefore based on a compromise between these factors (Shu et al., 1991). Any thin-film deposition process involves three main steps: (i) production of the appropriate atomic, molecular or ionic species; (ii) transport of these species to the substrate through a medium; and (iii) condensation on the substrate, either directly or via a chemical and/or electrochemical reaction, to form a solid deposit. The most commonly employed processes for film deposition can be categorized as variants of a basic physical or chemical process.

11.2.1

Physical processes

The physical process, physical vapor deposition (PVD), can be divided into two types: (i) thermal evaporation and (ii) sputtering. The thermal evaporation process involves the evaporation of source materials in a vacuum chamber below 10−4 Pa, followed by the condensation of the evaporated particles on a substrate. Thermal evaporation processes can further be divided into the following sub-categories (Iaquaniello et al., 2011): (a) Vacuum deposition: resistive heating is most commonly used for the deposition of thin films. The source materials are evaporated by a resistively heated filament or boat, generally made of refractory metals such

© Woodhead Publishing Limited, 2013

462

Handbook of membrane reactors

as W, Mo or Ta, with or without ceramic coatings. Crucibles of quartz, graphite, alumina, beryllia, boron-nitride or zirconia are used with indirect heating. The refractory metals are evaporated by electron-beam deposition, since simple resistive heating cannot evaporate high melting point materials. (b) Pulsed laser deposition (PLD): this process is an improved thermal process used for the deposition of alloys and/or compounds with a controlled chemical composition. In laser deposition, a high-power pulsed laser is irradiated onto the target of source materials through a quartz window. A quartz lens is used to increase the energy density of the laser power on the target source. Atoms that are ablated or evaporated from the surface are collected on nearby sample surfaces to form thin films. The sputtering process is characterized by the fact that when a solid surface is bombarded with energetic particles, such as accelerated ions, surface atoms of the solid are scattered backward due to collisions between the surface atoms and the energetic particles. This phenomenon is called back-sputtering, or simply sputtering. Transmission sputtering, on the other hand, occurs when a thin foil is bombarded with energetic particles, leading to some of the scattered atoms being transmitted through the foil (Iaquaniello et al., 2011). PVD sputtering has several advantages, including: (i) synthesis of ultrathin films with minimal impurity; (ii) easily controllable process parameters; (iii) flexibility for synthesizing alloys; and (iv) the ability to generate nanostructured films. The last two points are very important in the preparation of membranes for hydrogen separation, because the fabrication of the membrane alloys helps in overcoming the problem of hydrogen embrittlement, while the nanostructured films may have unique size-dependent properties, such as high hydrogen permeation.

11.2.2

Chemical processes

Chemical processes can be classified as: (i) thermal chemical vapor deposition (CVD) and (ii) plating. The CVD process involves the vaporization of a volatile compound of the substance to be deposited, after which the vapor is thermally decomposed into atoms or molecules, and/or reacts with other gases, vapors or liquids at the substrate surface to yield non-volatile reaction products on the substrate. Most CVD processes operate at relatively high temperature (near 1000°C) at pressures of between a few hundred Pa to above the atmospheric pressure of the reactants. Several CVD processes have been put

© Woodhead Publishing Limited, 2013

Palladium-based membranes for hydrogen separation

463

forward as means of increasing the efficiency of the chemical reaction at lower substrate temperatures. Plasma-assisted chemical vapor deposition (PACVD) is one of the modifications made to the conventional CVD process. In the PACVD system, electric power is supplied to the reactor to generate the plasma. The ions in the plasma show slightly higher energy than the neutral gas molecules at room temperature. Typically, the temperature of the ions in plasma is around 500 K. Plating can be further categorized as follows: (a) Electroplating consists of the deposition of a metallic coating on an electrically conducting surface which acts as the cathode in an electrolytic cell, whose solution contains ions of the metal to be deposited. This is a relatively complex technology that involves a large number of steps (Almeida, 2001). (b) Electroless plating is a non-galvanic plating method that involves several simultaneous chemical reactions in an aqueous solution occurring without the use of external electric current. The chemical reactions occur when hydrogen is released by a reducing agent, normally sodium hypophosphite, and is oxidized to produce a negative change on the surface of the substrate. This autocatalytic deposition method enables the metal coating of non-conductive textile materials, which can be used for precision work in conventional manufacturing. Unlike electroplating, the absence of any electric field contributes to a uniform plating thickness. Under properly controlled conditions all of the above-mentioned methods produce good quality thin layers; however, electroless plating has the advantage of easy scale-up and the flexibility to coat the metal film on supports of different geometries. Its principal disadvantage is the difficulty in controlling the composition of the alloy.

11.3 11.3.1

Membrane cost analysis Current Pd-based membranes market

Only a small number of companies are able to supply Pd-based membranes or membrane ‘modules’; the market is still restricted to laboratory scale membranes or modules for small pilot units. Below is a short overview of some membrane providers. The Energy research Centre of the Netherlands (ECN) produces and offers a line of hydrogen separation modules (Hysep) on a pre-commercial basis for evaluation purposes. The technology is based on palladium membranes, which are capable of separating high purity hydrogen from a gas mixture.

© Woodhead Publishing Limited, 2013

464

Handbook of membrane reactors

An essential element of the Hysep® technology is the use of thin-film palladium composite membranes to allow reliable hydrogen separation at low cost. The supported palladium layer in the Hysep® module has a thickness as low as 3–9 µm, a substantial improvement over current commercially available palladium membranes, which are based on self-supporting metal foils with a thickness of 20–100 microns. MRT is a Vancouver-based private company interested in hydrogen purifiers to provide high purity hydrogen and to recover hydrogen from mixed gas streams. MRT produces membranes either as rolled foils or as deposited thin films (8–15 microns). In addition, patented bonding techniques have been developed to permanently attach membranes to support modules with a perfect, hydrogen-tight seal. For membranes thinner than 15 microns, MRT uses a proprietary coating technique. Prototype membranes as thin as 8 microns, tested by MRT, have been produced and show excellent performance and longevity. An important Japanese company (JC) is developing a gas separation membrane able to efficiently recover hydrogen, by forming a film on a porous ceramic substrate using a palladium alloy known for its selective permeation of hydrogen. The key is to simultaneously achieve cost effectiveness and high hydrogen selectivity by making expensive palladium membranes thinner. A three-step procedure is used to produce the membranes: first, Pd is deposited onto the Al2O3 support by the electroless plating technique; then, Ag is layered on by electroplating using the Pd layer as the electrode; and finally, the layered Pd–Ag membrane is heat-treated to obtain the Pd–Ag alloy membrane. The resulting membranes are tubular with an external diameter of about 1.0 cm, an effective length of about 9.0 cm and a Pd–Ag coating deposited on the external surface with a selective layer of about 28.3 cm2. In recent years, the three companies mentioned above have developed three membrane modules (Fig. 11.1a–c) employed in the pilot plant developed by Tecnimont KT in Chieti Scalo (Italy) within the framework of the Italian FISR project ‘Pure hydrogen from natural gas through total conversion reforming obtained by integrating chemical reactions and membrane separation’. The aim of the project was to assess the industrial feasibility of the integration of a steam reformer reactor with a membrane for hydrogen separation. The plant was successfully running continuously for up to 1000 h, with no catalyst or membrane deterioration (De Falco et al., 2011). Finally, SINTEF has developed a technique for the manufacture of Pd-based hydrogen separation membranes based on a two-step process, allowing a reduction in membrane thickness. First, a defect-free Pd-alloy thin film is prepared by magnetron sputtering onto a silicon wafer. In the second step the film is removed from the wafer. These films may subsequently be either used self-supported or integrated with various supports of different pore size, geometry and size. This allows, for example, the preparation of

© Woodhead Publishing Limited, 2013

Palladium-based membranes for hydrogen separation

465

(a)

A = 0.40 (b)

A = 0.60 (c)

A = 0.13

11.1 The three membrane modules developed by (a) ECN, (b) MRT and (c) JC, with relevant selective layer permeation surface (A in m2) employed for the Chieti Scalo pilot plant.

very thin (approximately 2–3 microns) high-flux membranes supported on macroporous substrates which can operate at high pressures.

11.3.2

Membrane manufacturing strategy

In order to lower production costs, the correct membrane manufacturing strategy must be adopted. This involves two main aspects: 1. the manufacturing process itself, which will give the business a distinct advantage in the marketplace (e.g., through the use of unique technology); 2. manufacturing-associated activities and infrastructure design, such as controls, procedures, choice of subcontractors, and so on. Based on our current understanding and experience in the field of composite membranes, one way of simplifying the manufacturing process is to

© Woodhead Publishing Limited, 2013

466

Handbook of membrane reactors

separate the preparation of the Pd-based selective layer from its integration onto the support. This has been demonstrated by Bredesen and Klette (2000), and by SINTEF, as shown in Section 11.3.1. Advancing this concept, the manufacturing process should consist of a batch process with three independent steps. The choice of a batch process is a logical one, because it provides similar items on a repeat basis, usually in larger volume. The batch procedure divides the manufacturing task into a series of appropriate operations, which will combine to make the required product. The main steps in this process are easily described: firstly, the Pd-alloy selective layer is prepared; secondly, the support is prepared; and finally, the membrane module is assembled and tested. One way to approach the fabrication of the membrane and support is to produce the two products on distinct lines, and then to test them for quality control at the end of the process prior to integration. The selection of the correct manufacturing process and necessary hardware is not simple. The choice of process is related to the features of the hardware, and the tangible ways in which the products are manufactured, but it also involves more than this. The associated structures, controls, procedures and other systems within manufacturing are equally necessary for successful, competitive manufacturing performance. In our membrane manufacturing strategy, quality controls related to the thin films, for instance, but also to the support and the membrane overall, constitute an essential aspect of the manufacturing task. Creating a quality function in the organization to supervise such operational controls is a crucial issue to be developed and properly managed. Failure to develop a proper infrastructure for the complex process of membrane production may result in target costs becoming impossible to reach.

11.3.3

Experience curve for membrane production

In order to forecast production costs for thin Pd-based membranes, it is important to introduce the concept of ‘economics of learning’ in understanding the behavior of all added costs of membranes as the cumulative production volume increases. This idea of economics of learning, or law of experience, may be expressed as follows: cn = c1n−a

[11.2]

where c1 is the cost of the unit production (square meter of membrane for instance), cn is the cost of the nth unit of production, n is the cumulative volume of production and a is the elasticity of cost with regard to output. An experience curve can then be constructed by using the data available, which are limited, for Pd-based or ceramic membranes, to a minimal surface

© Woodhead Publishing Limited, 2013

Palladium-based membranes for hydrogen separation

467

Table 11.1 Cost per m2 of membrane module versus cumulated production Cumulated production, m2

€ cost per m2

1000 10 000 100 000 1 000 000 10 000 000

8900 5000 2800 1600 900

area (less than 1 m2). Another issue associated with drawing an experience curve is that cost and production data must be related to a ‘standard product’; this is impossible in this case, as no standard has yet emerged in membrane technology. It is a fact, however, that costs decline systematically with increase in cumulative output. The assumptions made are: c1 = 50 000 € and a = 0.25, where the c1 value is derived from Tecnimont KT’s recent experience in building pilot units, while the ‘a’ factor is assumed to be an average value typically between 20% and 30%. On the basis of this data, it is possible to predict the cost/m2 of membrane module as a function of the cumulative value of production, expressed in terms of m2, as reported in Table 11.1. Obviously, the experience curve is characterized by a progressively declining gradient. The size of the experience effect is measured by the proportion by which costs are reduced with subsequent doublings of aggregate production. For Pd-based membranes, a cost of 8900 €/m2 is predicted for a cumulative production of 1000 m2, while the cost would be reduced to 900 €/m2 if the cumulative production is increased to 10 000 000 m2. The key question now is how to achieve cumulative production of millions of square meters of Pd-based membranes within the next few years. R&D efforts in membrane cost reduction, the reliability of thin-film fabrication methods, and the demonstration of the benefits of selective membrane application in chemical processes can be expected to hasten the introduction of this new technology onto the market.

11.4 11.4.1

Membrane application case study: water gas shift (WGS) reactor Process description

The WGSR [11.3] was discovered more than two centuries ago, by the Italian physicist Felice Fontana in 1780 (Burns et al., 2008). CO + H2O ↔ CO2 + H2

© Woodhead Publishing Limited, 2013

[11.3]

468

Handbook of membrane reactors

Although the first report was published in 1888 by Mond and Langer (1888), the technical importance of the WGSR was not recognized until the development of the Haber process. Currently, the WGSR is used in various chemical processes, such as hydrogen and ammonia production, Fischer−Tropsch and methanol synthesis. It is also considered to be an important process for the removal of CO in small-scale future power generation, based on fuel cells for both mobile and stationary applications. The WGSR is not affected by pressure and is moderately exothermic ΔH 0298 K 41 1 kJ mol 1 ; therefore, its equilibrium constant decreases with temperature, favoring higher CO conversions at lower temperatures. The addition of greater stoichiometric quantities of steam leads to improvements in conversion. Under adiabatic conditions, the heat of the reaction increases the process temperature; consequently, conversion in a single catalyst bed is thermodynamically limited. Significant improvements can be achieved with a double catalyst bed operation, with the second bed operating at the lowest possible inlet temperature. The high-temperature shift (HTS) reactor usually operates at temperatures of 350–420°C with an iron−chromium oxide based catalyst, and the low-temperature shift (LTS) reactor operates at temperatures of 180–340°C with a Cu–ZnO/Al2O3 catalyst (Navarro et al., 2007). With this reactor architecture, the exit concentration of CO could be as low as 0.1–0.3%. The mechanism of the catalyzed shift reaction for both copper- and iron-based catalysts remains controversial. Two types of mechanism have been proposed: adsorptive and regenerative. In the former, the reactants adsorb on the catalyst surface, where they react to form surface intermediates such as formates, followed by decomposition to products and desorption from the surface. In the regenerative mechanism, on the other hand, the surface undergoes successive oxidation and reduction cycles by water and carbon monoxide, respectively to form the corresponding hydrogen and carbon dioxide products of the WGS reaction. Some major disadvantages associated with the WGSR can also be identified. The HTS catalyst has low activity at lower temperatures and the process is thermodynamically limited at high temperatures. The low temperature Cu–ZnO catalyst is sensitive to air exposure, promotes temperature excursions and requires lengthy preconditioning for intermittent operation. Significant efforts have therefore been made to improve the performance of iron- (Lei et al., 2005a, 2005b, 2006; Maroño et al., 2009; Martos et al., 2009; Natesakhawat et al., 2006; Rhodes et al., 2002) and Cu–ZnO- (Guo et al., 2009; Nishida et al., 2008; Tang et al., 2008) based catalysts by optimizing the preparation and formulation of the catalyst. Because of its unique redox properties, ceria was tested as a CuO-based catalyst support (Djinović et al., 2008a, 2008b, 2009a, 2009b; Li et al., 2000). Other recent studies have also focused on the use of noble metals such as Pt (González et al., 2010; Jacobs

(

)

© Woodhead Publishing Limited, 2013

Palladium-based membranes for hydrogen separation

469

et al., 2006; Kim et al., 2009; Panagiotopoulou et al., 2007; Wheeler et al., 2004) and Au (Bond, 2009). One potential application of the WGSR is the integrated gasification combined cycle (IGCC) process (Ciambelli et al., 2011), in which the gasification of coal produces synthesis gas and the WGSR converts the CO to produce additional H2. Despite coal being an available raw material with relatively stable cost for H2 production as an alternative to gaseous and liquid hydrocarbons, the IGCC process presents additional challenges due to the lower pressure and lower H2 content of the clean gas. The first generation of IGCC plants was developed in the 1990s. Although these are reliable and have demonstrated environmental benefits, further improvements are needed to simplify the process, increase efficiency and reduce costs, in order to advance the commercial outlook of the IGCC scheme. Despite the existence of several commercial entrained flow gasification systems for the production of fuel gas or syngas, the process has yet to be demonstrated on the commercial scale as part of an integrated plant for the production of H2 with the collection and storage of species of environmental concern such as CO2. These systems, including the gasifiers developed by Shell, Texaco, Destec and Prenflo, all share certain features, namely the utilization of pulverized coal, and common operating conditions of 20 to 70 bar and 1500°C with very high fuel heating rates. However, the systems also differ from each other in the way in which the fuel is introduced, the concentration of steam and the methods employed for heat recovery.

11.4.2

Membrane assisted WGS reaction

Reactor configurations – open and closed architecture The typical and most straightforward configuration for a membrane reactor is composed of two concentric tubes, where the catalyst is packed in the annular zone while the inner tube is the membrane itself (closed architecture) as shown in Fig. 11.2. A sweeping gas is fed through the inner tube, co-currently or counter-currently, in order to carry out the permeated hydrogen. Membrane integration can of course also be achieved by assembling many smaller Reactants Sweep gas + H2

Products

Catalyst H2

H2

H2

H2

Reactants

Sweep gas Products

Catalyst

11.2 Membrane integration in closed architecture.

© Woodhead Publishing Limited, 2013

470

Handbook of membrane reactors Sweep gas + H2

Reactants

Sweep gas

Reactor

Membrane module

Retentate

Sweep gas

Reactor

Sweep gas + H2

11.3 Membrane integration in open architecture.

tubes, thereby increasing the volume of the specific membrane surface on the reactor and consequently the global permeated hydrogen flow. In an alternative configuration, the selective membrane is placed outside the reactor in units located downstream (open architecture, Fig. 11.3). In this case, after the membrane separation module another reaction unit is required, in which the enhancement in hydrocarbon conversion may be observed. The feedstock is sent to a first reactor where it is partially converted into the products; then one of the products is recovered through a selective membrane separation module, while the retentate is sent to the next step or recycled to the first module. By means of a heat recovery system, the operating temperature can be reduced before the membrane unit, in order to ensure a thermal level suitable for the correct operation of the membrane unit, and then increased again before the second reactor up to the values required to support the reactions. These reaction−separation steps can be repeated until the desired natural gas conversion is achieved. The two configurations both have several benefits and drawbacks. Globally, at the same operating conditions, the closed architecture is more compact, shows ease of scalability and avoids catalyst waste; however, its major drawback is in the technological difficulties involved in designing and maintaining the reactor. In contrast, the open architecture allows the decoupling of reaction and separation unit operating conditions, meaning that the temperatures in the two blocks can be optimized independently; further, the mechanical design of the membrane module is simpler and its maintenance is easier. One of the main advantages of the open architecture is the possibility of revamping

© Woodhead Publishing Limited, 2013

Palladium-based membranes for hydrogen separation

471

the already existing plants. On the contrary, the main drawbacks are the loss of compactness, the larger membrane surface and the higher cost. Literature review A growing interest in membrane assisted WGSR is evident from the substantial volume of the relevant studies that address one of the two most popular types of hydrogen permselective membranes: those based on palladium (Barbieri et al., 2008; Basile et al., 1995, 1996, 2001; Bi et al., 2009; Brunetti et al., 2009a, 2009b; Criscuoli et al., 2000; Iyoha et al., 2007a, 2007b) and those based on silica (Battersby et al., 2008, 2009; Brunetti et al., 2007a; Giessler et al., 2003). The references provided here are only the most recent and most representative studies; in addition, a very recent extensive review by Babita et al. (2011) summarizes the status, challenges and opportunities of WGS membrane reactors. Basile et al. (1995) studied the WGSR in a palladium membrane reactor and demonstrated the importance of the membrane preparation method in obtaining high-quality membrane materials. Magnetron sputtering, PVD and co-condensation techniques were used to realize submicron palladium membranes. The best membrane increased CO conversion up to 99.89% at about 330°C and stable performance was reported for more than two months (Basile et al., 1996). Iyoha et al. (2007a) assessed the performance of a Pd and Pd80 wt%Cu membrane reactor at 900°C with a 241 kPa trans-membrane pressure differential intended to be positioned downstream of a coal gasifier. No catalyst was used, as it was expected that the membrane tubes would sufficiently catalyze and further enhance the fast rate of the WGSR at this temperature. There was a significant increase in CO conversion from the equilibrium value of 54% to 93% with the Pd membrane, while conversion of 66% was obtained with the Pd80 wt%Cu membrane. The markedly lower conversion with the Pd80 wt%Cu membrane was attributed to its lower H2 permeance. However, after about eight days on-stream, pinhole formation was confirmed by SEM-EDS for both membranes. In follow-up WGSR studies (Iyoha et al., 2007b), these membranes were applied to the simulated coal gasification syngas feeds. CO conversion of 99.7% was achieved at 900°C in a counter-current, Pd multi-tube membrane reactor operated at a 2 s residence time. The conversion of CO was considerably higher than the approximately 32% equilibrium conversion allowed in a conventional reactor. As previously determined, the Pd80 wt%Cu membrane tubes gave a significantly lower CO conversion of only 68%. Exposure of both membranes to syngas mixtures containing H2S at levels below the threshold required for the formation of thermodynamically stable sulfides (H2S/H2 < ~0.0011) did not affect the mechanical strength of the membranes, but caused a

© Woodhead Publishing Limited, 2013

472

Handbook of membrane reactors

steep drop in CO conversion due to deactivation of the catalyst surface. Any further increase in H2S concentration destroyed the membranes within minutes. Recently, Barbieri et al. (2008) proposed an innovative configuration of a membrane reactor consisting of a conventional fixed bed reactor followed by a membrane reactor loaded with the same catalyst. This configuration was considered to be particularly suitable for reactions characterized by slow kinetics, such as the WGSR, as it stimulates more effective membrane utilization by optimizing the driving force and minimizing the H2 back-permeation to the reaction side. The WGSR was tested between 280°C and 320°C using a commercial CuO/CeO2-based catalyst and Pd–Ag membrane without sweeping for an equimolecular H2O/CO stream at pressure between 2 and 6 atm and gas hourly space velocity (GHSV) between 2000 and 10 000 h−1. A significant reduction in the required reaction volume to reach CO conversion similar to that of the conventional reactor was achieved, which should result in a decreased cost of operation. Bi et al. (2009) employed a noble metal based catalyst Pt/Ce0.6Zr0.4O2 and Pd membrane reactor to study the WGSR using feeds obtained by autothermal reforming of natural gas. CO conversion remained above thermodynamic equilibrium up to feed space velocities of 9100 l kg−1 h−1 at 350°C, Ptotal = 1.2 MPa and S/C = 3, but H2 recovery decreased from 84.8% at space velocity of 4050 l kg−1 h−1 to 48.7% at the highest space velocity. This rapid decline of separation performance was attributed to slow H2 diffusion through the catalyst bed, suggesting that external mass flow resistance has a significant impact on the H2 permeation rate in such membrane reactors. Near-complete conversion of CO was obtained by Kikuchi et al. (1989) and by Uemiya et al. (1991) in the WGSR carried out at 400°C in a double tubular Pd membrane reactor with a commercial iron–chromium oxide as the catalyst. Despite all these studies, a recent economic feasibility study of membrane assisted WGSR, conducted to assess its advantages over conventional technology, showed that the concept was not feasible for Pd membrane reactors (Criscuoli et al., 2001) and instead supported the use of ceramic membranes (Bracht et al., 1996). Microporous silica hydrogen permselective membranes have been extensively studied as a potentially more practical alternative to Pd membranes. Cutting edge research into silica membranes has shown that they have good hydrogen flux and separation, as well as adequate thermal stability. However, the hydrothermal stability of a silica hydrogen permselective membrane is a key factor in determining its suitability for a commercial application of membrane assisted processes. A silica membrane, prepared by the soaking−rolling procedure on a porous stainless steel disk, was employed by Brunetti et al. (2007) for the

© Woodhead Publishing Limited, 2013

Palladium-based membranes for hydrogen separation

473

WGSR for temperatures of 220–290°C and pressures of up to 600 kPa using a commercial CuO/CeO2-based catalyst. The CO conversion always exceeded the conversions that could be obtained in a conventional reactor at temperatures higher than 250°C. The CO conversion difference between the conventional reactor and the membrane reactor increased with temperature and was more pronounced at lower reaction pressures. It was determined that the best operating conditions for the membrane reactor were 280°C and 400 kPa. The use of a silica membrane supported on a molecular sieve in a low-temperature WGSR at 280°C was reported by Giessler et al. (2003). Although almost complete conversion of CO was reported, the low H2/N2 separation of the membrane allowed CO to cross the membrane into the sweep stream and for the sweep gas to enter the product stream, artificially boosting the conversion. Battersby et al. (2008) investigated the WGSR in a silica membrane at temperatures between 150°C and 250°C. The H2/CO separation increased from 5 to 15 with temperature, but at conversions below 40%, the H2 driving force for permeation was lower than that for CO, dictating the minimum conversion for which the membrane reactor could be effective. In a follow-up study (Battersby et al., 2008b) a cobalt-silica membrane was used to test the WGSR up to 300°C. The initial H2/CO separation increased with temperature, resulting in up to 95% purity of the permeated H2. A 7% increase in conversion above that obtained in a conventional reactor was achieved at 300°C and the membrane showed a reasonable on-stream stability for about 200 h. However, the membrane had no effect on conversion below 200°C, indicating that the application of a membrane reactor in a kinetically controlled range of the WGS reaction is perhaps not the best choice of conditions to demonstrate a paradigm offered by membrane reactors, as the equilibrium conversion is already high but the reaction rate is low. The hydrothermal stability of the membrane was most likely an issue at higher temperatures and dictated the choice of these less representative conditions.

11.4.3

Modeling

The modeling and simulation of the membrane assisted WGS reaction in traditional packed-bed reactors has received some attention in the literature (Boutikos and Nikolakis et al., 2010; Gosiewski et al., 2010; Piemonte et al., 2010). The modeling of the WGS reaction assisted by membranes was carried out using both closed and open architecture. In the case of the closed architecture, the model was developed for the ‘tube in tube’ configuration. The

© Woodhead Publishing Limited, 2013

474

Handbook of membrane reactors

reaction zone is packed with the catalyst inside the inner tube; this tube is made of membrane material separating the products of the WGS reaction into the outer tube (shell). On the shell side, an inert sweep gas can be introduced either co-currently or counter-currently with respect to the direction of flow in the tube. In our modeling, a co-current configuration was adopted. The mathematical model has been developed under the following assumptions: • • • • • • • • •

Monodimensional model Steady state conditions Plug flow behavior both in reaction and permeation zone (no axial mixing) Constant catalyst effectiveness factor all along the reactor length Infinite hydrogen selectivity of the membrane Ideal gas behavior All the impurities are considered as a single inert component and as such, do not take part in the reaction Traditional reactor kinetics are also considered valid for the membrane reactor The momentum balance equations were not addressed, since constant values of pressure along axis in the tube and shell were assumed for preliminary studies.

The material balances in the reaction zone are described by the following Equations [11.4–11.8]: dFC CO = − η ⋅ ( − ε) ⋅ (rCO ) π (ri2,i ) dz dFH2O dz dFC CO2 dz dFH 2 dz

[11.4]

= − η ⋅ (1 − ε) ⋅ (rCO ) π (ri2,i )

[11.5]

= η ⋅ (1 − ε ) ⋅ (rCO ) π (ri2,i )

[11.6]

perm p = η ⋅ (1 ε ) (rCO ) π (ri2,i ) J H 2 π re,i 2

dFiinert =0 dz

[11.7]

[11.8]

© Woodhead Publishing Limited, 2013

Palladium-based membranes for hydrogen separation dT reaz = dz

⎡r π ri2, i ⋅ η ⋅ ( − ⎢( CO ) ⋅⎢ ⎢ Fi ccp pi (T ) ⎣ 1

n

∑

) ⋅ ( − Δ H reaz (T ) ) + −U 2 (T ) ⋅ 2 ⋅ π ⋅ re,i ⋅ (

−

( ( ) − h (T

+ J H 2 ⋅ 2 ⋅ ⋅ re i ⋅ h

H2

2

475

reaz membrane

))

)⎤⎥ ⎥ ⎥ ⎦

i=1

[11.9] Fi is the component molar flowrate (i = H2O, CO, CO2, H2, Inert); η is the catalyst effectiveness factor; ε is the catalyst bed porosity (assumed to be equal to 0.5); ri,i is the shell tube radius; re,i is the membrane tube radius; (rCO) is the rate of WGS reaction; T reaz is the reaction temperature (K); n  reaz is is the number of chemical species; cpi is specific heat [J/mol K]; ΔH ̂ ΔH of reaction at temperature T [J/mol]; and U2 is a global heat transport coefficient from catalytic zone to permeation one [J/m2 ·s·K]. Reaction rates for CO are given by the Langmuir−Hinshelwood kinetics equation (Criscuoli et al., 2000): ( CO )

⎛ pCO2 pH 2 ⎞ k ⋅ KCO K H 2O ⋅ ⎜ pCO pH 2O − ⎟ Keq ⎝ ⎠

[11.10] −2

(1 + KCO pCO + K H 2O ⋅ pH 2O + KCO2 ⋅ pCO2 ) ⋅ ρcat Adsorption constants are: KCO

KH2O

KCO2

10

⎛ 3064 6 74 ⎞ − ⎜ ⎟ 1.987 ⋅T 1.987 ⎠ ⎝ e

5

10

5

10

5

⎛ −6216 12.77 ⎞ + ⎜ ⎟ 1.987 ⋅T 1.987 ⎠ e⎝ ⎛ 12542 18.45 ⎞ − ⎜ ⎟ 1.987 ⋅T 1.987 ⎠ e⎝

The equilibrium constant is: Keq

⎛ 4577.8 ⎞ ⎜ T − 4 33 ⎟ ⎠

e⎝

The kinetics rate constant is: ⎛ −29364 40.32 ⎞

k

1 ⎜⎝ 1.987 ⋅T − 1.987 ⎟⎠ e 60

© Woodhead Publishing Limited, 2013

476

Handbook of membrane reactors

In the permeation zone the material balances are the following: dFHperm 2

dz dFssweep dz

=+

(

BH 05 ⋅ pH 2 R δ

)

05 pH ⋅ 2 ⋅ π ⋅ re,i 2 ,P

[11.11]

=0

[11.12]

⎡ dT perm ⎢ =⎢ dz ⎢ Fsweepc p sweep T ⎣ ⎡ ⎢⎣U 2 (T ) ⋅ 2 ⋅ π ⋅ re,i ⋅ ⎡K ⎤ in ⎢ ⎥ ⎣m⎦

(

(

⎤ ⎥ ⎥⋅ perm FH c p H perm T ⎥ ⎦

1

)

)+ J

−

(

)

perm m ⋅ 2 ⋅ π ⋅ re,i ⎛ hH 2 T perm − hH 2 H2 ⎝

(

)⎞ ⎤

membrane ⎠ ⎥ m ⎦

[11.13] perm is the flux of hydrogen across the membrane and it is described by JH 2 Fick–Sieverts’ law: perm JH = 2

(

BH 0 5 pH 2 R δ

p0 52

p

)

[11.14]

BH is the membrane permeability, the value of which depends on the membrane type and the temperature BH =

⎛ ⎞ 8.686 × 10 −6 10300 ⋅ exp ⎜ − 3600 ⎝ 8 31 ⋅ Tmedium ⎟⎠

⎡ mol ⎤ ⎢ m s ⋅ Pa 0 5 ⎥ ⎣ ⎦

In both reactor configurations, a WGS catalyst with a diameter of 3.5 m and a height of 3.5 m was employed, while the composition of the inlet to the WGSR was CO = 9.4%, H2O = 32.7%, CO2 = 5.3%, H2 = 49.6%, CH4 = 3.0%, which is typical for a steam reforming reactor fed with natural gas. The WGSR inlet temperature was set to 350°C, while the pressure was assumed to be constant and equal to 25.6 bar. In the study of an open architecture, two WGSRs with a membrane module separating them were employed. In both cases, the membrane permeation surface was assumed to be the same. However, the temperature profiles were different: in the closed architecture, the temperature along the membrane

© Woodhead Publishing Limited, 2013

Palladium-based membranes for hydrogen separation

477

module was assumed to increase due to the exothermicity of WGS reaction, whereas in the open architecture, the temperature was assumed to be constant, and equal to the temperature outlet of the first reactor. The results of mathematical modeling developed for closed architecture with a catalyst effectiveness factor of 0.6 clearly show the effect of the presence of the membrane on CO conversion profile along the catalyst bed. The most important effect is that the reactor, in this case, can overcome the thermodynamic limitations and the maximum CO conversion is higher than that obtained without the membrane. In addition, it is worth noting that the equilibrium CO conversion can be obtained by using a lower catalyst volume, or in other words at a shorter reactor length. The effect of the catalyst effectiveness factor on the membrane assisted WGSR in closed architecture is shown in Fig. 11.4, in terms of CO conversion (Fig. 11.4a) and reactor temperature profile (Fig. 11.4b).

(a)

80

CO conversion (%)

70

Equilibrium

60 50 η = 0.3

40 30

η = 0.6

20 η = 0.9

10

(b)

0 420

Temperature (°C)

410

Equilibrium

400 390 380

η = 0.3

370

η = 0.6

360 η = 0.9

350 340 0.0

0.2

0.4 0.6 Reactor length (adim)

0.8

1.0

11.4 Effect of catalyst effectiveness factor on WGS membrane reactor performance in closed architecture: (a) CO conversion and (b) temperature.

© Woodhead Publishing Limited, 2013

478

Handbook of membrane reactors

The effectiveness factor plays a very interesting role in determining membrane reactor performance: essentially, in the absence of extra- and intra-particle diffusion, the behavior of the CO conversion profile changes in the same way as observed in a standard catalytic reactor. In terms of the temperature profile, the main drawback caused by the presence of greater diffusional limitations is the lower heat production rate, corresponding to a slower temperature increase along the catalytic bed. When the effectiveness factor is closer to 1, the system is in the ‘kinetic regime’, and the effect of membrane addition may be even greater, leading to increased CO conversion, and correspondingly a higher temperature at the catalytic bed outlet. It should be noted that in the case of closed architecture the catalyst and the membrane tube can be arranged in such a way that the hydrogen flux can be centrifugal (the catalyst in the inner tube) or centripetal (the catalyst in the outer tube). Clearly, such an arrangement (i) would require a specific catalyst support such as an open cell foam, allowing a radial flow; and (ii) would influence the temperature profile established in the catalyst bed, thus potentially leading to an improvement in catalyst performance (Palma et al., 2009). The results of mathematical modeling in open architecture are given in Fig. 11.5 in terms of (a) CO conversion and (b) temperature profile along the catalytic bed. The data in Fig. 11.5 clearly demonstrates the beneficial effect of the membrane: the final CO conversion value is even higher than before, reaching a value of about 90%. This result is likely to be due to the effect of temperature, for which maximum values are assumed here, with an isothermal profile. The assumptions made on the basis of this model will bear a close resemblance to the real situation, where the physical separation of the reactor module by the membrane module allows easy control of the heat losses, thereby achieving a fairly constant temperature. If we consider the potential industrial application of this technology, although there is an obvious advantage for the closed architecture, mainly due to the lower cost and higher space saving, we propose that the intrinsic higher flexibility of open architecture offers greater advantages in revamping existing plants; in the short and medium terms, this is a more realistic approach to the application of membrane reactors (Iaquaniello et al., 2011a) Finally, when a membrane reactor was employed in open architecture configuration along with a catalyst effectiveness factor of 0.6, the hydrogen recovery factor was evaluated as a function of the membrane permeation surface. The results are shown in Fig. 11.6.

11.4.4

Economic perspectives

The successful introduction of membrane WGSRs depends on the improvement of membrane technology. Clearly, the absence of well-developed

© Woodhead Publishing Limited, 2013

Palladium-based membranes for hydrogen separation

479

(a) 100 90 CO conversion (%)

80

Second reactor

Equilibrium

70 60

First reactor

50 η = 0.3

40 30

η = 0.6

20 10

η = 0.9

0 (b) 470

Temperature (°C)

450 430 410 η = 0.3 390 η = 0.6 370 η = 0.9 350 0.0

0.2

0.4 0.6 Reactor length (adim)

0.8

1.0

11.5 Effect of catalyst effectiveness factor on WGS membrane reactor performance in open architecture: (a) CO conversion and (b) temperature. 100 90 H2 recovery (%)

80 70 60 50 40 30 20 10 0 0

20

40 60 80 Membrane surface (m2)

100

11.6 Effect of membrane surface on hydrogen recovery factor.

© Woodhead Publishing Limited, 2013

480

Handbook of membrane reactors

membrane reactor technology and of a reliable estimate of membrane costs, leads to difficulties in developing a detailed economic analysis of membrane assisted WGSRs. Further, as reported by Babita et al. (2011), the main techno-economic factors to be taken into account in an overall economic analysis of these systems should include the reactor configuration, the performance of both the catalyst and the membrane, and the purity of hydrogen required. In the first instance, apart from the quantitative contribution of the reactor configuration, a preliminary economic analysis was carried out taking into account only the cost of the catalyst and membrane, under the following assumptions: • Catalyst cost: 10 000 €/m3 • Membrane cost: 4000 €/m2 The two configurations were compared by calculating the costs relative to the specific hydrogen production, with a CO inlet flowrate of 1000 kmol/h. The catalyst volume was taken to be equal to 33.7 m3 for the closed architecture, and twice that for open architecture, whereas the membrane surface was the same for both configuration and equal to 38.5 m2. However, it must be observed that, in the case of open architecture, based on the performance reported in the Fig. 11.5, only the first 20% of the catalytic bed in the second reactor was considered for the purposes of economic evaluation, since it appears that this length is sufficient to reach the maximum CO conversion. On the basis of these assumptions, the costs obtained in the case of closed architecture were 705 €/kmol/h H2 produced, whereas in the case of open architecture, the costs are slightly at 680 €/kmol/h H2 produced, with a gain of about 5%.

11.5

Conclusions and future trends

Selective membrane application in chemical processes represents one of the most interesting scientific and technological topics to have emerged in recent years, in the context of the tendency towards the intensification of industrial processes and the improvement of process efficiency. Although reactor performance in the laboratory and assessments by mathematical model simulations have demonstrated the excellent potential of membrane integration in chemical processes, the performance of membrane reactors remains heavily dependent on the behavior of the selective membrane in terms of permeability, selectivity and stability. Some technological challenges remain, such as the optimization of the fabrication method, the durability of the membrane in a contaminating environment and the

© Woodhead Publishing Limited, 2013

Palladium-based membranes for hydrogen separation

481

configuration adopted for the integration of the membrane in the reactor. These challenges all need to be addressed before hydrogen selective membranes, and in particular Pd-based membranes, will become enough reliable and cost competitive for industrial applications, making the commercialization of hydrogen selective membranes and of membrane reactors a reality. However, we are confident that in the next 5–10 years this technology will definitely emerge.

11.6

References

Adhikari S, Fernando S (2006), ‘Hydrogen separation techniques’, Ind. Eng. Chem. Res., 45, 875–881. Alexander S R, Winnick J (1994), ‘Removal of hydrogen sulfide from natural gas through an electrochemical membrane separator’, AIChE J., 40, 613–620. Almeida E (2001), ‘Surface treatments and coatings for metals. A general overview. 2. Coatings: Application processes, environmental conditions during painting and drying, and new tendencies’, Ind. Eng. Chem. Res., 40, 15–20. Amandusson H, Ekedahl L-G, Dannetun H (2000), ‘The effect of CO and O2 on hydrogen permeation through a palladium membrane’, Appl. Surf. Sci., 153, 259–267. Babita K, Sridhar S, Raghavan K V (2011), ‘Membrane reactors for fuel cell quality hydrogen through WGSR – Review of their status, challenges and opportunities’, Int. J. Hydrogen Energy, 36, 6671–6688. Barbieri G, Brunetti A, Granato T, Bernardo P, Drioli E (2005), ‘Engineering evaluations of a catalytic membrane reactor for the water gas shift reaction’, Ind. Eng. Chem. Res., 44, 7676–7683. Barbieri G, Brunetti A, Tricoli G, Drioli E (2008), ‘An innovative configuration of a Pd-based membrane reactor for the production of pure hydrogen. Experimental analysis of water gas shift’, J. Power Sources, 182, 160–167. Basile A, Chiappetta G, Tosti S, Violante V (2001), ‘Experimental and simulation of both Pd and Pd/Ag for a water gas shift membrane reactor’, Sep. Purif. Technol., 25, 549–571. Basile A, Criscuoli A, Santella F, Drioli E (1996), ‘Membrane reactor for water gas shift reaction’, Gas. Sep. Purif., 10, 243–254. Basile A, Drioli E, Santella F, Violante V, Capannelli G, Vitulli G (1995), ‘A study on catalytic membrane reactors for water gas shift reaction’, Gas. Sep. Purif., 10, 53–61. Basile A, Iulianelli A, Longo T, Liguori S, De Falco M (2011), ‘Pd-based selective membrane state of the art’, in Membrane reactors for hydrogen production processes, M. De Falco, L. Marrelli, G. Iaquaniello (Eds.) ISBN 978-0-85729-150-9, Springer. Basile A, Paturzo L, Vazzana A (2003), ‘Membrane reactor for the production of hydrogen and higher hydrocarbons from methane over Ru/Al2O3 catalyst’, Chem. Eng. J., 93, 31–39. Battersby S, Duke M C, Liu S, Rudolph V, Diniz da Costa J C (2008), ‘Metal doped silica membrane reactor: Operational effects of reaction and permeation for the water gas shift reaction’, J. Membrane Sci., 316, 46–52.

© Woodhead Publishing Limited, 2013

482

Handbook of membrane reactors

Battersby S, Smart S, Ladewig B, Liu S, Duke M C, Rudolph V, Diniz da Costa J C (2009), ‘Hydrothermal stability of cobalt silica membranes in a water gas shift membrane reactor’, Sep. Purif. Technol., 66, 299–305. Bi Y, Xu H, Li W, Goldbach A (2009), ‘Water-gas shift reaction in a Pd membrane reactor over Pt/Ce0.6Zr0.4O2 catalyst’, Int. J. Hydrogen Energy, 34, 2965–2971. Bond G (2009), ‘Mechanisms of the gold-catalysed water-gas shift’, Gold Bull., 42, 337–342. Boutikos P, Nikolakis V (2010), ‘A simulation study of the effect of operating and design parameters on the performance of a water gas shift membrane reactor’, J. Membrane Sci., 350, 378–386. Bracht M, Alderliesten P T, Kloster R, Pruschek R, Haupt G, Xue E, Ross J R H, Koukou M K, Papayannakos N (1996), ‘Water gas shift membrane reactor for CO2 control in IGCC systems: techno-economic feasibility study’, Energy Convers., 38, S159–164. Bredesen R, Klette H (2000), ‘Method of manufacturing thin metal membranes’, US Patent 6,086,729. Brunetti A, Barbieri G, Drioli E (2009a), ‘Upgrading of a syngas mixture for pure hydrogen production in a Pd-Ag membrane reactor’, Chem. Eng. Sci., 64, 3448–3454. Brunetti A, Barbieri G, Drioli E (2009b), ‘Pd-based membrane reactor for syngas upgrading’, Energy Fuels, 23, 5073–5076. Brunetti A, Barbieri G, Drioli E, Granato T, Lee K-H (2007a), ‘A porous stainless steel supported silica membrane for WGS reaction in a catalytic membrane reactor’, Chem. Eng. Sci., 62, 5621–5626. Brunetti A, Barbieri G, Drioli E, Lee K-H, Sea B, Lee D-W (2007b), ‘WGS reaction in a membrane reactor using a porous stainless steel supported silica membrane’, Chem. Eng. Process., 46, 119–126. Burns D T, Piccardi G, Sabbatini L (2008), ‘Some people and places important in the history of analytical chemistry in Italy’, Microchim. Acta, 160, 57–87. Castro F J, Meyer G, Zampieri G (2002), ‘Effects of sulfur poisoning on hydrogen desorption from palladium’, J. Alloys Compd., 330, 612–616. Chen Y, Wang Y, Xu H, Xiong G (2008), ‘Efficient production of hydrogen from natural gas steam reforming in palladium membrane reactor’, Appl. Catal. B: Env., 80, 283–294. Ciambelli P, Palma V, Palo E, Galuszka J, Giddings T, Iaquaniello G (2011), ‘Technical and economical evaluation of WGSR’, in Membrane reactors for hydrogen production processes, M. De Falco, L. Marrelli, G. Iaquaniello (Eds.) ISBN 978-085729-150-9, Springer. Criscuoli A, Basile A, Drioli E (2000), ‘An analysis of the performance of membrane reactors for the water-gas shift reaction using gas feed mixtures’, Catal. Today, 56, 53–64. Criscuoli A, Basile A, Drioli E, Loiacono O (2001), ‘An economic feasibility study for water gas shift membrane reactor’, J. Membrane Sci., 181, 21–27. De Falco M, Iaquaniello G, Salladini A (2011), ‘Experimental tests on steam reforming of natural gas in a reformer and membrane modules (RMM) plant’, J. Membrane Sci., 368, 264–274. Djinović P, Batista J, Pintar A (2008a), ‘Calcination temperature and CuO loading dependence on CuO-CeO2 catalyst activity for water-gas shift reaction’, Appl. Catal. A, 347, 23–33.

© Woodhead Publishing Limited, 2013

Palladium-based membranes for hydrogen separation

483

Djinović P, Batista J, Levec J, Pintar A (2009a), ‘Comparison of water–gas shift reaction activity and long-term stability of nanostructured CuO-CeO2 catalysts prepared by hard template and co-precipitation methods’, Appl. Catal. A, 364, 156–165. Djinović P, Batista J, Pintar A (2009b), ‘WGS reaction over nanostructured CuO– CeO2 catalysts prepared by hard template method: Characterization, activity and deactivation’, Catal. Today, 147S, S191–S197. Djinović P, Levec J, Pintar A (2008b), ‘Effect of structural and acidity/basicity changes of CuO–CeO2 catalysts on their activity for water–gas shift reaction’, Catal. Today, 138, 222–227. Ferreira-Aparicio P, Rodríguez-Ramos I, Guerrero-Ruiz A (2002), ‘On the applicability of membrane technology to the catalysed dry reforming of methane’, Appl. Catal. A, 237, 239–252. Gallucci F, Chiaravalloti F, Tosti S, Drioli, E, Basile A (2007a), ‘The effect of mixture gas on the hydrogen permeation through a palladium membrane: experimental studies and theoretical approach’, Int. J. Hydrogen Energy, 32, 1837–1845. Gallucci F, De Falco M, Tosti S, Marrelli L, Basile A (2007b), ‘The effect of the hydrogen flux pressure and temperature dependence factors on the membrane reactor performances’, Int. J. Hydrogen Energy, 32, 4052–4058. Galuszka J, Giddings T, Iaquaniello G (2011), ‘Membrane assisted WGSR – Experimental study and reactor modeling’, Chem. Eng. J., http://dx.doi. org/10.1016/j.cej.2011.05.035 Giessler S, Jordan K, Diniz da Costa JC, Lu GQ(M) (2003), ‘Performance of hydrophobic and hydrophilic silica membrane reactors for the water gas shift reaction’, Sep. Purif. Technol., 33, 255–264. González I D, Navarro R M, Wen W, Marinkovic N, Rodriguéz J A, Rosa F, Fierro J L G (2010), ‘A comparative study of the water gas shift reaction over platinum catalysts supported on CeO2, TiO2 and Ce-modified TiO2’, Catal. Today, 149, 372–379. Gosiewski K, Warmuzinski K, Tanczyk M (2010), ‘Mathematical simulation of WGS membrane reactor for gas from coal gasification’, Catal. Today, 156, 229–236. Guo P, Chen L, Yang Q, Qiao M, Li H, Xu H, Fan K (2009), ‘Cu/ZnO/Al2O3 water–gas shift catalysts for practical fuel cell applications: the performance in shut-down/ start-up operation’, Int. J. Hydrogen Energy, 34, 2361–2368. Guo X, Hidajat K, Ching C (1997), ‘Oxidative coupling of methane in a solid oxide membrane reactor’, Ind. Eng. Chem. Res., 36, 3576–3582. Iaquaniello G, Borruto A, Lollobattista E, Narducci G, Katsir D (2011), ‘Hydrogen palladium selective membranes: and economic perspective’; in Membrane reactors for hydrogen production processes, M. De Falco, L. Marrelli, G. Iaquaniello (Eds.) ISBN 978-0-85729-150-9, Springer. Iaquaniello G, Cucchiella B, Antonetti E (2011a), ‘Membrane upstream water gas shift reaction’, European patent application No. 11170211.4 Itoh N, Kaneko Y, Igarashi A (2002), ‘Efficient hydrogen production via methanol steam reforming by preventing back-permeation of hydrogen in a palladium membrane reactor’, Ind. Eng. Chem. Res., 41, 4702–4706. Iyoha O, Enick R, Killmeyer R, Howard B, Morreale B, Ciocco M (2007a), ‘Wall-catalyzed water-gas shift reaction in multi-tubular Pd and 80 wt%Pd-20 wt%Cu membrane reactors at 1173 K’, J. Membrane Sci., 298, 14–23.

© Woodhead Publishing Limited, 2013

484

Handbook of membrane reactors

Iyoha O, Enick R, Killmeyer R, Howard B, Ciocco M, Morreale B (2007b), ‘H2 production from simulated coal syngas containing H2S in multi-tubular Pd and 80 wt% Pd-20 wt% Cu membrane reactors at 1173 K’, J. Membrane Sci., 306, 103–115. Jacobs G, Ricote S, Davis B H (2006), ‘Low temperature water-gas shift: Type and loading of metal impacts decomposition and hydrogen exchange rates of pseudo-stabilized formate over metal/ceria catalysts’, Appl. Catal. A, 302, 14–21. Jorgensen S, Nielsen P E H, Lehrmann P (1995), ‘Steam reforming of methane in membrane reactor’, Catal. Today, 25, 303–307. Julbe A, Farrusseng D, Guizard C (2001), ‘Porous ceramic membranes for catalytic reactors–overview and new ideas’, J. Membrane Sci., 181, 3–20. Kikuchi E, Nemoto Y, Kajiwara M, Uemiya S, Kojima T (2000), ‘Steam reforming of methane in membrane reactors: comparison of electroless-plating and CVD membranes and catalyst packing modes’, Catal. Today, 56, 75–81. Kikuchi E, Uemiya S, Sato N, Inoue H, Ando H, Matsuda T (1989), ‘Membrane reactor using microporous glass supported thin film of palladium. Application to the water gas shift reaction’, Chem. Lett., 18, 489. Kim Y T, Park E D, Lee H C, Lee D, Lee K H (2009), ‘Water-gas shift reaction over supported Pt-CeOx catalysts’, Appl. Catal. B, 90, 45–54. Kluiters S C A (2004), ‘Intermediate report EU project MIGREYD NNE5–2001– 670, ECNC – 04–102’. Koros W J, Fleming G K (1993), ‘Membrane-based gas separation’, J. Membrane Sci., 83, 1–80. Kulprathipanja A, Alptekin G, Falconer J, Way J D (2005), ‘Pd and Pd-Cu membranes: inhibition of H2 permeation by H2S’, J. Membrane Sci., 254, 49–62. Langguth J, Dittmeyer R, Hofmann H, Tomandl G (1997), ‘Studies on oxidative coupling of methane using high-temperature proton-conducting membranes’, Appl. Catal. A: General, 158, 287–305. Lei Y, Cant N W, Trimm D L (2005a), ‘Activity patterns for the water gas shift reaction over supported precious metal catalysts’, Catal. Lett., 103, 33–36. Lei Y, Cant N W, Trimm D L (2005b), ‘Kinetics of the water–gas shift reaction over a rhodium-promoted iron–chromium oxide catalyst’, Chem. Eng. J., 114, 81–85. Lei Y, Cant N W, Trimm D L (2006), ‘The origin of rhodium promotion of Fe3O4– Cr2O3 catalysts for the high-temperature water–gas shift reaction’, J. Catal., 239, 227–236. Lemier C, Weissmuller J (2006), ‘Grain boundary segregation, stress and stretch: Effects on hydrogen absorption in nanocrystalline palladium’, J. Acta Mater., 55, 1241–1254. Li A, Liang W, Hughes R (2000), ‘The effect of carbon monoxide and steam on the hydrogen permeability of a Pd/stainless steel membrane’, J. Membrane Sci., 165, 135–141. Li Y, Fu Q, Flytzani-Stephanopoulos M (2000), ‘Low-temperature water-gas shift reaction over Cu- and Ni-loaded cerium oxide catalysts’, Appl. Catal. B, 27, 179–191. Lin Y M, Liu S L, Chuang C H, Chu Y T (2003), ‘Effect of incipient removal of hydrogen through palladium membrane on the conversion of methane steam reforming. Experimental and modeling’, Catal. Today, 82, 127–139.

© Woodhead Publishing Limited, 2013

Palladium-based membranes for hydrogen separation

485

Li Z Y, Maeda H, Kusakabe K, Morroka S, Anzai H, Akiyama S (1993), ‘Preparation of palladium-silver alloy membranes for hydrogen separation by the spray pyrolysis method’, J. Membrane Sci., 78, 247–254. Lu G Q, Diniz da Costa J C, Duke M, Giessler S, Socolow R, Williams R H, Kreutz T (2007), ‘Inorganic membranes for hydrogen production and purification: a critical review and perspective’, J. Colloid Interface Sc., 314, 589–603. Maroño M, Ruiz E, Sánchez JM, Martos C, Dufour J, Ruiz A (2009), ‘Performance of Fe–Cr based WGS catalysts prepared by co-precipitation and oxi-precipitation methods’, Int. J. Hydrogen Energy, 34, 8921–8928. Martos C, Dufour J, Ruiz A (2009), ‘Synthesis of Fe3O4-based catalysts for the hightemperature water gas shift reaction’, Int. J. Hydrogen Energy, 34, 4475–4481. Mendes D, Mendes A, Madeira L M, Iulianelli A, Sousa J M, Basile A (2010), ‘The water-gas shift reaction: from conventional catalytic systems to Pd-based membrane reactors-a review’, Asia-Pac. J. Chem. Eng., 5, 111–137. Michaels A S (1968), ‘New separation techniques for CPI’, Chem. Eng. Prog., 64, 31–43. Mond L, Langer C (1888) Improvements in obtaining hydrogen. British Patent 12608. Natesakhawat S, Wang X, Zhang L, Ozkan US (2006), ‘Development of chromiumfree iron-based catalysts for high-temperature water-gas shift reaction’, J. Mol. Catal. A: Chem., 260, 82–94. Navarro R M, Peña M A, Fierro J L G (2007), ‘Hydrogen production reactions from carbon feedstocks: Fossil Fuels and Biomass’, Chem. Rev., 107, 3952–3991. Nishida K, Atake I, Li D, Shishido T, Oumi Y, Sano T, Takeira K (2008), ‘Effects of noble metal-doping on Cu/ZnO/Al2O3 catalysts for water–gas shift reaction. Catalyst preparation by adopting “memory effect” of hydrotalcite’, Appl. Catal. A, 337, 48–57. Nozaki T, Fujimoto K (1994), ‘Oxide ion transport for selective oxidative coupling of methane with new membrane reactor’, AIChe J., 40, 870–877. Ockwig N W, Nenoff T M (2007), ‘Membranes for hydrogen separation’, Chem. Rev., 107, 4078–4110. Palma V, Palo E, Ciambelli P (2009), ‘Structured catalytic substrates with radial configurations for the intensification of the WGS stage in H2 production’, Catal. Today, 147S, S107–S112. Panagiotopoulou P, Papavasiliou J, Avgouropoulos G, Ioannides T, Kondarides D I (2007), ‘Water–gas shift activity of doped Pt/CeO2 catalysts’, Chem. Eng. J., 134, 16–22. Patil C S, Annaland M V S, Kuipers J A M (2007), ‘Fluidised bed membrane reactor for ul-trapure hydrogen production via methane steam reforming: Experimental demonstration and model validation’, Chem. Eng. Sci., 62, 2989–3007. Piemonte V, De Falco M, Favetta B, Basile A (2010), ‘Counter-current membrane reactor for WGS process: Membrane design’, Int. J. Hydrogen Energy, 35, 12609–12617. Ramachandra A M, Lu Y, Ma Y H, Moser W R, Dixon A G (1996), ‘Oxidative coupling of methane in porous Vycor membrane reactors’, J. Membr. Sci., 116, 253–264. Ravanchi M T, Kaghazchi T, Kargari A (2009), ‘Application of membrane separation processes in petrochemical industry: a review’, Desalination, 235, 199–244.

© Woodhead Publishing Limited, 2013

486

Handbook of membrane reactors

Rhodes C, Williams BP, King F, Hutchings GJ (2002), ‘Promotion of Fe3O4/Cr2O3 high temperature water gas shift catalyst’, Catal. Commun., 3, 381–384. Saracco G, Neomagus H W J P, Versteeg G F, van Swaaij, W P M (1999), ‘High-temperature membrane reactors: potential and problems’, Chem. Eng. Sci., 54, 1997–2017. Shu J, Adnot A, Grandjean B P A, Kaliaguine S (1996), ‘Structurally stable composite Pd-Ag alloy membranes: an introduction of a diffusion barrier’, Thin Solid Films, 286, 72–79. Shu J, Grandjean B P A, Kaliaguine S (1995), ‘Asymmetric Pd-Ag/stainless steel catalytic membranes for methane steam reforming’, Catal. Today, 25, 327–332. Shu J, Grandjean B P A, Van Neste A, Kaliaguine S (1991), ‘Catalytic palladium-based membrane reactors: a review’, Can. J. Chem. Eng., 69, 1036–1060. Tang X-J, Fei J-H, Hou Z-Y, Lou H, Zheng X-M (2008), ‘Copper-zinc oxide and manganese promoted copper-zinc oxide as highly active catalysts for water-gas shift reaction’, React. Kinet. Catal. Lett., 94, 3–9. ten Elshof J E, Bouwmeester H J M, Verweij H (1995), ‘Oxidative coupling of methane in a mixed-conducting perovskite membrane reactor’, Appl. Catal. A: General, 130, 195–212. Tong J, Matsumura Y (2005), ‘Effect of catalytic activity on methane steam reforming in hydrogen-permeable membrane reactor’, Appl. Catal. A: Gen., 286, 226–231. Tonkovich A L Y, Zilka J L, Jimenez D M, Roberts G L, Cox J L (1996), ‘Experimental investigations of inorganic membrane reactors: a distributed feed approach for partial oxidation reactions’, Chem. Eng. Sci., 51, 789–806. Uemiya S, Sato N, Ando H, Matsuda T, Kikuchi E (1991), ‘Steam reforming of methane in a hydrogen-permeable membrane reactor’, Appl. Catal., 67, 223–230. Wheeler C, Jhalani A, Klein E J, Tummala S, Schmidt L D (2004), ‘The water-gas-shift reaction at short contact times’, J. Catal., 223, 191–199. Xu S J, Thomson W J (1997), ‘Perovskite-type oxide membranes for the oxidative coupling of methane’, AIChE J., 43, 2731–2739. Yepes D, Cornaglia L M, Irusta S, Lombardo E A (2006), ‘Different oxides used as diffusion barriers in composite hydrogen permeable membranes’, J. Membrane Sci., 274, 92–101. Zeng Y, Lin Y S (2000), ‘Oxygen permeation and oxidative coupling of methane in yttria doped bismuth oxide membrane reactor’, J. Catal., 193, 58–64. Zeng Y, Lin Y S, Swartz S L (1998), ‘Perovskite-type ceramic membrane: Synthesis, oxygen permeation and membrane reactor performance for oxidative coupling of methane’, J. Membrane Sci., 150, 87–98. Zhang K, Gao H, Rui Z, Liu P, Yongdan L, Lin Y S (2009), ‘High temperature stability of palladium membranes on porous metal supports with different intermediate layers’, Ind. Eng. Chem. Res., 48, 1880–1886.

© Woodhead Publishing Limited, 2013

12 Membrane reactor for hydrogen production from natural gas at the Tokyo Gas Company: a case study Y. SHIRASAKI , The Japan Gas Association, Japan and I. YASUDA , Tokyo Gas Co., Ltd, Japan

DOI: 10.1533/9780857097347.2.487 Abstract: An advanced large-scale Pd-based membrane reformer system with a nominal hydrogen production capacity of 40 Nm3/h has been developed. The system demonstrates the potential advantages of the membrane reformer: the simplicity of the system configuration (high purity (99.999% level) hydrogen is produced in a single step from natural gas), its compactness and its high energy efficiency. Other benefits of the membrane reformer include the fact that the CO2 in the off-gas can easily be liquefied and captured, due to the high CO2 concentration of 90%. The results of preliminary operational tests of the membrane reformer show that CO2 emissions decreased by over 50% with only a minor energy loss. Key words: membrane reformer, hydrogen production, steam methane reforming, palladium, CO2 capture.

12.1

Introduction

The expectation is that hydrogen energy will play a major role in a future low carbon society, contributing significantly to global environmental conservation and reducing dependence on fossil fuel resources. However, hydrogen is a secondary energy, and for the time being we will be dependent on hydrogen from fossil fuels, a process which produces CO2 emissions. A pragmatic approach to the future hydrogen energy society is required − accepting that hydrogen must be produced from fossil fuels during a transition period, but acknowledging also that the CO2 emissions created during this process must be reduced. In order to reduce CO2 emissions, the total energy efficiency of the path, from exploitation of fossil fuel to hydrogen production, must be improved markedly. In particular, significant improvement of the efficiency of conventional steam reforming for hydrogen production is 487 © Woodhead Publishing Limited, 2013

488

Handbook of membrane reactors

required. Furthermore, to realize further mass reductions in CO2 emissions, the capture and storage of by-product CO2 in hydrogen production need to be examined. A membrane reformer equipped with palladium membrane modules for in situ hydrogen separation is a compact, simple and highly efficient hydrogen production system, and an improvement in these respects on the conventional steam methane reformer. In addition, CO2 in the off-gas of a membrane reformer can be easily separated and captured by direct liquefaction, owing to the high concentration of CO2. Tokyo Gas Co., Ltd. (TGC) has developed a 40 Nm3/h-class membrane reformer system with the world’s highest efficiency (a value of 81.4%). The company has demonstrated the use of the hydrogen produced to refuel fuel cell vehicles (FCV), together with CO2 capture at the hydrogen station. An advanced hydrogen separation membrane module consisting of a palladium alloy membrane on a structured porous catalyst, which can be used to produce a membrane reformer that is more compact and less expensive, has also been developed. This chapter introduces the development of these two membrane reformer technologies.

12.1.1

Hydrogen production from natural gas

Natural gas steam reforming is widely used in industrial markets for hydrogen and synthesis gas production. The conventional hydrogen production system is composed of a steam methane reformer, a shift converter and a hydrogen purification system based on pressure swing adsorption (SMR-PSA). The steam reforming reaction (Equations [12.1] and [12.2]) and the water gas shift (WGS) reaction (Equation [12.3]) convert natural gas into a mixture of hydrogen, carbon monoxide, carbon dioxide and steam. The reforming reaction is reversible and largely endothermic. High temperatures of 700–800°C are usually preferred for producing a hydrogen-rich gas in conventional reformers (Rostrup-Nielsen, 1984). SMR is usually carried out using Ni supported on alumina as catalyst with external heating. C n H m + nH H 2 O → nCO + (n + m / 2)H 2

(

CH 4 + H 2 O ⇔ CO + 3H 2 CO + H 2 O ⇔ CO2 + H 2

(

(

H 0298 < 0

)

[12.1]

)

[12.2]

H 0298 = −206.2 kJ/mol H 0298 = 41.2 kJ/mol

© Woodhead Publishing Limited, 2013

)

[12.3]

Membrane reactor for hydrogen production from natural gas

489

Heat

Membrane tube (Pd-based alloy) Catalyst H2 Natural gas (CH4)

CO Hydrogen

CH4 H2

H2

H2O

Steam

CH4

CO2

Carbon dioxide CO2

Reactor tube

12.1 Conceptual diagram of membrane reformer.

12.1.2

Principle and features of membrane reformers

Palladium or its alloys are the most practical membrane materials, due to their high hydrogen permeability and stability at high temperatures. The membrane reformer is composed of a steam reformer equipped with palladium-based alloy modules in its catalyst bed, and can perform steam reforming reaction and hydrogen separation processes concurrently with no help from shift converter and PSA, as shown in Fig. 12.1. The new concept of simultaneous generation and separation of hydrogen means that membrane reformer system can be configured more compactly and can provide higher efficiency than conventional steam reformers. The simultaneous process of hydrogen generation and separation frees the reactions from the limitation of chemical equilibrium and thus can reduce the reaction temperature from the conventional 700–800°C to 500–550°C. This means that expensive heat-resistant metals need not be used for structural components and long-term durability increases as a result of the lower operation temperature. In addition to the above-mentioned features, another advantage of the membrane reformer system is the high concentration of CO2 in the off-gas, which enables easy capture of CO2 by direct liquefaction.

12.1.3

History of membrane reformer development

The discovery of hydrogen selective diffusion through a palladium membrane by Sir Thomas Graham in 1866 and its subsequent development, a palladium−hydrogen system, have been extensively investigated (Lewis, 1967). A self-sustained palladium dense metal membrane tube with thickness

© Woodhead Publishing Limited, 2013

490

Handbook of membrane reactors

of about 100 µm was commercially available early on as a purifier for the production of ultrapure hydrogen. Uemiya (1991) developed composite membrane tubes consisting of palladium thin film on porous supports using electroless plating. As a result, a membrane tube with sufficient hydrogen flux in hydrogen separation from a mixed gas such as steam methane reforming gas with low hydrogen partial pressure was able to be produced. This result triggered the development of the membrane reformer. TGC envisioned a membrane reformer as an excellent hydrogen production system that can offer high efficiency yet is a simple and compact system. TGC started research and development of the membrane reformer in 1992 in collaboration with Mitsubishi Heavy Industries (MHI). TGC and MHI developed a 4 Nm3/h-class membrane reformer and experimentally verified the effectiveness of the concept, demonstrating the operation of the membrane reformer in connection with a polymer electrolyte fuel cell (PEFC) stack to generate 5 kW DC electricity. This was the world’s first PEFC to operate with pure hydrogen produced from natural gas using the membrane reformer (Kuroda, 1996). Then the technology was further developed to scale up the membrane reformer to 15 Nm3/h-class. The operational tests proved it to be capable of producing 99.99% pure hydrogen at a rate of 15 Nm3/h and stable operation for about 1500 h with 40 start-up and shut-down cycles (Fujimoto, 2001); therefore the technical feasibility of the membrane reformer was proved and established. From fiscal year (FY) 2000 to FY 2004, as a member of The Japan Gas Association, TGC joined a Japanese national project, which was supported by The New Energy and Industrial Technology Development Organization (NEDO). In the project, TGC successfully developed a 40 Nm3/h-class membrane reformer system and demonstrated its high efficiency of 76% (higher heating value (HHV)) with 99.99% or higher purity hydrogen. The product hydrogen was supplied to FCVs (Yasuda, 2007). The impurities in the produced hydrogen increased gradually after the operational test of the first membrane reformer for 3000 h with 60 start-and-stop cycles, which indentified improvement of durability of palladium membrane as the most important issue for commercializing the membrane reformer technology. Meanwhile, a high CO2 concentration of 90% in the off-gas of the membrane reformer was also observed, which suggested that it would be possible to capture CO2 from distributed hydrogen production systems (Shirasaki, 2009). From FY 2005 to FY 2007, the second membrane reformer was developed with the aim of improving energy efficiency and establishing system engineering technologies under a NEDO 3-year program. The target was to develop a membrane reformer system that could produce 99.99% or higher purity hydrogen from natural gas at a rate of 40 Nm3/h with hydrogen

© Woodhead Publishing Limited, 2013

Membrane reactor for hydrogen production from natural gas

491

production energy efficiency of over 80%. TGC had been developing an advanced membrane reformer system for decades and has achieved the world’s highest efficiency in hydrogen production from natural gas. In addition, CO2 separation and a capture-from-hydrogen production system by direct liquefaction was demonstrated utilizing another feature of membrane reformer technology: the membrane reformer can produce pure hydrogen at high efficiency, while at the same time the highly concentrated CO2 in the process off-gas flows out (Kurokawa, 2011). At present, TGC has been promoting research and development focusing on membrane durability and reliability improvement as well as cost reduction, which are critical issues for the commercialization of membrane reformer technology. In parallel with the development of the membrane reformer system, a new concept membrane module, which has a palladium alloy membrane coated on the porous support tube with catalytic activity has been developed (Nishii, 2009). This membrane module is expected to provide a more compact reactor because the reactor does not require a separate catalyst. It is also expected that this module can be manufactured at low cost by applying the industrially-established mass production process used to make oxygen sensors for combustion control in vehicles with internal combustion engines.

12.2 12.2.1

Performance of the 40 Nm3/h-class membrane reformer Hydrogen separation membrane module

The membrane module has a plate-type structure 40 mmW × 460 mmL × 8 mmT in size. Figure 12.2 illustrates the configuration of the membrane module, and Fig. 12.3 shows a view of the membrane modules. The membrane modules consist of palladium−rare earth alloy thin film with thickness of less than 20 µm and a porous structural support. The hydrogen permeability of the membrane is several times higher than that of the widely used conventional palladium−silver alloy membrane (Sakamoto, 1992).

Palladium alloy membrane Hydrogen

Stuctural support Palladium alloy membrane

12.2 Configuration of membrane module.

© Woodhead Publishing Limited, 2013

492

Handbook of membrane reactors

12.3 Photograph of 20 membrane modules.

12.2.2

Reactor tube configuration

Figure 12.4 shows the structure of the reactor tube, which is the basic building unit of the reformer. The reactor tube contains two membrane modules and catalyst. The catalyst has been changed from Ni/Al2O3 to a precious metal catalyst to allow operation at lower steam-to-carbon ratios for increasing the energy efficiency in the second membrane reformer. The reactor tube has two forms of catalysts. The one is in pellet form in the primary catalyst bed, and the other is in a specially designed monolithic corrugated form and placed close to the membrane surface without causing mechanical damage to the membrane surface as a result of friction between the catalyst and membrane. The mixture of natural gas and steam is supplied to the primary reforming catalyst bed from the nozzle N-1. The process gas flow turns at the end of the primary catalyst bed, and is reformed while flowing through the membrane part, and hydrogen is selectively separated into the inside of the membrane modules. The product high-purity hydrogen comes out from the nozzle N-3, while the off-gas, the reformed gas after hydrogen extraction, flows out of the catalyst bed through the nozzle N-2. The unit reactor tube is 86 mmW × 615 mmL × 25 mmT.

12.2.3

Membrane reformer configuration

The 40 Nm3/h-class membrane reformer has a structure in which multi-reactor tubes are packed in the rectangular vessel. TGC made a few modifications to the first membrane reformer, which attained an efficiency level of 76.2%. The second membrane reformer increased the number of membrane modules from 224 to 256, in order to operate the system for higher methane

© Woodhead Publishing Limited, 2013

Membrane reactor for hydrogen production from natural gas Catalyst

Membrane module

N-3 N-1

N-3 N-2 Catalyst

493

Hydrogen Off-gas Natural gas, steam Off-gas Hydrogen

Membrane module Hydrogen

12.4 Configuration of reactor tube.

12.5 External view of 40 Nm3/h-class membrane reformer system.

conversion and reduce natural gas input. Figure 12.5 shows an external view of the packaged reformer system. The packaged system is very compact, 3.56 m wide, 2.56 m deep and 2.3 m high. The volume and footprint of the advanced membrane reformer system are only one-third and one-half respectively of the conventional SMR-PSA system with comparable hydrogen production capacity. The structure of the 40 Nm3/h-class membrane reformer is illustrated in Fig. 12.6. The 40 Nm3/h-class membrane reformer, including heat insulation, is 1200 mm wide, 750 mm deep and 1350 mm high. The reactor contains 128 reactor tubes.

12.2.4

System design of membrane reformer

Figure 12.7 shows a system flow diagram of the 40 Nm3/h-class membrane reformer system. The reactor is started up by combusting natural gas, and

© Woodhead Publishing Limited, 2013

494

Handbook of membrane reactors

Membrane module Reactor tube

Reactor unit Header

Natural gas + steam Hydrogen Off-gas Jacket Reactor tube Membrane module Primary catalyst bed

Exhaust gas

Distributor Furnace Burner

12.6 Structure of the 40 Nm3/h-class membrane reformer.

the mixture of natural gas and steam is introduced into the reactor after the temperature has reached a given operating temperature, which is controlled by the feed rate of the burner fuel. The pressure of natural gas is increased by a compressor up to 0.95 MPa after removal of sulfur odorants by a desulfurization agent, the natural gas is mixed with steam generated by a boiler, and then the gas mixture is added to the catalyst bed. The fuel for the burners is switched from natural gas to the off-gas after hydrogen production starts. The reactor is connected to a vacuum suction and booster unit and the pressure of product hydrogen can be increased to 0.74 MPa.

© Woodhead Publishing Limited, 2013

Membrane reactor for hydrogen production from natural gas

495

Natural gas Desulfurizer Compressor

Off-gas

Boiler

Burner Catalyst bed

Membrane reformer

Exhaust gas

Steam

Exhaust gas Membrane module

Drain

Cooling water

Hydrogen Hydrogen

Air Water

Cooling water

Blower Water treatment unit

12.7 Schematic flow diagram of 40 Nm3/h-class membrane reformer system. (In the figure, represents a connection port.)

An energy efficiency rating of 76% had already been achieved in the production of hydrogen with the first 40 Nm3/h-class membrane reformer system; the second system was designed to improve on this. For this purpose, the 40 Nm3/h-class membrane reformer was operated at a higher methane conversion rate and reduced natural gas input, steam-to-carbon ratio, auxiliary power consumption and heat losses. These improvements were expected to increase the efficiency up to 80% on the system design basis.

12.2.5

Demonstration of hydrogen supply to fuel cell vehicles

The 40 Nm3/h-class membrane reformer system was installed and tested at Senju hydrogen station in Tokyo, which was built and operated under the Japan Hydrogen & Fuel Cell Demonstration Project (JHFC). The system was installed at the right-back corner of the station as shown in Fig. 12.8. The volume of the packaged membrane reformer system is only a third of the 50 Nm3/h-class conventional SMR-PSA system which was also installed at Senju hydrogen station. Under the usual operation of the membrane reformer system, hydrogen generated from the system was combusted and exhausted. After the first membrane reformer system was connected to the hydrogen station facility, hydrogen generated from the system was supplied to two FCVs (Daimler F-CELL and Toyota FCHV leased by Tokyo Gas) in 2004.

© Woodhead Publishing Limited, 2013

496

Handbook of membrane reactors

12.8 Senju Hydrogen Station and fuel cell passenger cars leased from Toyota (right, FCHV) and DaimlerChrysler (left, F-Cell). Both cars are used for our company’s daily business and occasional events and exhibitions.

12.2.6

Performance of 40 Nm3/h-class membrane reformer

The system was tested at 495–540°C at a process gas pressure of 0.9 MPa and a product hydrogen pressure of 0.02–0.04 MPa, and S/C (steam-to-carbon ratio) of 2.8–3.6. Hydrogen production rate and efficiency, conversion rate of natural gas, and chemical composition of the off-gas were measured and analyzed by changing the feed rate of natural gas. The chemical composition of the off-gas was analyzed with a thermal conductivity detector (TCD) gas chromatograph. Natural gas conversion and hydrogen production efficiency were obtained by the following equations: Conversion (%) =

CCO + CCO2 CCO + CCO2 + CCH 4

× 100

[12.4]

where CCO,CCO2, and CCH4 are the concentrations of components (%) in the reformed gas. Efficiency (%) =

F (H 2 ) Q(H (H 2 ) × 100 F((NG G) (NG) + W (AUW)

[12.5]

where F(H2) is the flow rate of product hydrogen, F(NG) is the flow rate of natural gas, Q(H2) is the gross heating value of hydrogen, Q(NG) is

© Woodhead Publishing Limited, 2013

100

60

80

50

60

40

Eff. (1st MRF) Eff. (2nd MRF) Conv. (1st MRF) Conv. (2nd MRF)

30

40 H2 prod. rate (1st MRF) H2 prod. rate (2nd MRF)

20

0

20

Temperature: 495–540°C S/C: 2.8–3.6 Reaction side pressure: 0.9 MPa Permeation side pressure: 0.02–0.04 MPa

2

4

6

8

10

12

497

Hydrogen production rate (Nm3/h)

Efficiency (%, HHV), conversion (%)

Membrane reactor for hydrogen production from natural gas

10 14

Natural gas feed rate (Nm3/h)

12.9 Variations of efficiency, conversion and hydrogen production rate of 40 Nm3/h-class membrane reformer as a function of natural gas feed rate.

the gross heating value of natural gas and W(AUW) is auxiliary power consumption. Figure 12.9 shows the natural gas feed rate dependence of conversion, hydrogen production rate and efficiency of both the first and second 40 Nm3/h-class membrane reformer systems. In the first membrane reformer system (MRF), when the natural gas feed rate was 11.6 Nm3/h, the hydrogen production rate was 40.1 Nm3/h, the hydrogen production efficiency was 76.2% and the conversion was 78.7%. In the second membrane reformer system (2nd MRF) at a natural gas feed rate of 11.18 Nm3/h, the hydrogen production rate was 40.5 Nm3/h, and the hydrogen production efficiency and the conversion were 81.4% and 85.0% respectively, which were 5% and 7% higher respectively than those of the first membrane reformer. These results almost corresponded to the designed performance. Although additional natural gas was used for heating at lower natural gas feed rates, the hydrogen production efficiency remained at high levels of over 80%. The purity of the product hydrogen was higher than 99.999% (5-nines, 5N).

12.2.7

CO2 separation and capture

A membrane reformer is able to produce pure hydrogen at high efficiency, 10% higher than the conventional hydrogen production system based on

© Woodhead Publishing Limited, 2013

498

Handbook of membrane reactors

Table 12.1 Comparison of gas composition of membrane reformer off-gas, natural gas combustion exhaust and SMR-PSA off-gas

H2 CO CO2 CH4 N2 O2 Total

MRF off-gas 28% load

MRF off-gas 100% load

Natural gas combustion gas*

SMR-PSA off-gas†

6.0 0.9 89.7 3.4 0 0 100

15.0 1.7 65.2 18.1 0 0 100

0 0 12.2 0 87.8 0 100

51.5 2.2 44.1 2.2 0 0 100

*

Calculated value at air ratio of 1. Calculated value at hydrogen recovery of 70%.

†

the SMR-PSA system, therefore CO2 emissions are decreased. In addition, the membrane reformer has another advantage: that CO2 in the off-gas is concentrated to as high as 90% due to accelerating the steam reforming reaction to CO2 production side. Because of the enhanced reaction by removal of hydrogen from the reformed gas, the concentration of CO2 in the off-gas increased significantly. Table 12.1 compares the compositions of the off-gas of the first MRF at a natural gas feed rate of 11.6 Nm3/h (100% load) and 3.2 Nm3/h (28% partial load) with the stoichiometric combustion exhaust of natural gas and the offgas from a conventional SMR-PSA system. When the natural gas feed rate was 3.2 Nm3/h, CO2 concentration was as high as 90%. Although the stoichiometric combustion exhaust of natural gas contains 12% CO2, the CO2 concentration in actual thermal power plants is only about 3%, which makes it difficult to capture CO2. The off-gas of an SMR-PSA contains significantly higher CO2 of 44%, but a costly CO2 purification system is required prior to CO2 separation by liquefaction. However, in the membrane reformer, CO2 in the off-gas can easily be liquefied and captured without CO2 purification system because of the high CO2 concentration of up to 90% in the off-gas. The CO2 emissions and the possible reduction in its rate from a distributed highly efficient hydrogen production system based on a membrane reformer was estimated from the actual operation test results and data from the first 40 Nm3/h-class MRF. The CO2 emissions and possible reduction rate in the 40 Nm3/h-class MRF were calculated by assuming that only CO2 in the off-gas was captured by liquefaction process. The material balance of the MRF with CO2 capture process at a rated load is shown by flow diagrams in Fig. 12.10.

© Woodhead Publishing Limited, 2013

Membrane reactor for hydrogen production from natural gas CO2 released into air: 11.5 kg/h

499

CO2 emission from electricity: 4.8 kg/h

Off-gas: 5.7 Nm3/h

(0 Nm3/h)

Captured CO2: 21.0 kg/h

NG: 13.8 Nm3/h (11.6 Nm3/h)

MRF

CO2 capture system (2.2 Nm3/h) Hydrogen: 40.1 Nm3/h H2O: 36.0 kg/h

Boiler

Electric power:10.6 kW

Electric power: 3.5 kW

12.10 Flow diagram of 40 Nm3/h-class membrane reformer system with CO2 capture system.

The CO2 capture system is composed of a compressor, a chiller and a gas−liquid separation unit. In the industrial CO2 liquefaction process, CO2 gas is compressed up to 3 MPa and cooled down to −25ºC. In this study, it was designed that the off-gas from the MRF should be compressed to 7.0 MPa and cooled to room temperature to liquefy and separate CO2 in the gas−liquid separation unit. The concentration of CO2 in the off-gas was 65% and the off-gas flow rate was 16.4 Nm3/h at the rated hydrogen production of 40 Nm3/h. The total amount of liquefied CO2 capture was estimated as 21 kg/h and the CO2 reduction rate was calculated to be 55%. The CO2 emissions per unit of natural gas of 2.36 kg-CO2/Nm3 (Tokyo Gas website) and electric power consumption of 0.332 kg-CO2/ kWh based on Tokyo electric power generation mix in 2008 (TEPCO website) were used in this study. To prove the technical viability of CO2 liquefaction and separation from the off-gas, a CO2 capture apparatus was designed and assembled. The appearance of the membrane reformer system equipped with CO2 capture apparatus is shown in Fig. 12.11. The experimental apparatus was composed of the water removal equipment, a gas compressor, a chiller, gas-liquid separator and liquefied CO2 tank. In the preliminary operation test in connection with the 40 Nm3/h-class membrane reformer, it was demonstrated that over 90% of CO2 in the off-gas can be captured. The total CO2 emission in hydrogen production was decreased by 50% with only 3% energy loss. This experimental result suggests that the membrane reformer has the potential to reduce its CO2 emission to half with a minor energy loss by applying the CO2 capture system.

© Woodhead Publishing Limited, 2013

500

Handbook of membrane reactors

12.11 Appearance of membrane reformer system equipped with CO2 capture apparatus.

12.3 12.3.1

Advanced hydrogen separation module with membrane on catalyst Concept of a new type of membrane on catalyst (MOC) module

A membrane reformer offers the highest efficiency for hydrogen production from fossil fuels. As has been described in Sections 12.1. and 12.2, TGC has developed and operated a 40 Nm3/h-class MRF and demonstrated its high efficiency (HHV) of 81%. However, cost reduction in manufacturing hydrogen separation modules and development of a more compact reformer are required for commercialization, since the rare metal palladium is expensive. Downsizing membrane modules will also decrease the start-up time, which is important for practical application. TGC and NGK Spark Plug Co., Ltd have been developing a new type of hydrogen separation module, which we call a MOC module. In the MOC concept, the porous support itself has the function of reforming catalyst in addition to the role of membrane-support. The integrated structure of support and catalyst makes the membrane reformer more compact since separate catalysts placed around the membrane modules in the conventional membrane reformers can be eliminated. With relative ease, a thinner palladium film on the porous support can be made and the manufacturing process is appropriate to a mass production system, which will lead to significant reduction of module production cost.

© Woodhead Publishing Limited, 2013

Membrane reactor for hydrogen production from natural gas (a)

501

(b)

Natural gas with steam

Product hydrogen

Catalyst

Natural gas with steam

Membrane

Support

Product hydrogen

Structured catalyst

12.12 Structure of the conventional module and catalyst (a) and the MOC module (b).

12.3.2

Configuration of MOC module

The structure of the MOC is schematically shown in Fig. 12.12 as compared with the conventional membrane module. The conventional module consists of reforming catalyst and hydrogen separation membrane on a porous support as shown in Fig. 12.12a. Hydrogen produced from reforming of methane and steam in the catalyst layer permeates the membrane and passes through the porous support. In contrast, the MOC module consists of a hydrogen separation membrane on a porous support which has catalytic activity and requires no separate catalyst layers, as shown in Fig. 12.12b. Feedstock gases, that is, methane and steam, are reformed and converted to hydrogen, carbon monoxide and carbon dioxide in the catalytic porous support and only hydrogen permeates the membrane to produce very pure product hydrogen. The MOC module has four favorable characteristics. Firstly, it is more compact than the conventional module because the porous support has catalytic ability and requires no separate catalysis layers. Secondly, it may be durable because no friction of membrane with catalyst will occur. Thirdly, it can offer higher hydrogen permeability because plating technology can be applied to prepare a thin membrane on the porous support. Lastly, manufacturing costs may be significantly reduced because well-established mass production technology can be applied to produce the porous supports and because the amount of expensive rare metals, that is, palladium and silver, can be reduced by plating technology of thin membrane.

12.3.3

Fabrication of MOC module

The typical manufacturing method of the MOC is as follows. The fabrication process of the MOC module is roughly divided into three steps:

© Woodhead Publishing Limited, 2013

502

Handbook of membrane reactors

12.13 The MOC module (size: 9 mm od × 7 mm id × 300 mm long).

production of porous support, coating of barrier layer and plating of palladium alloy membrane. NiO and 8YSZ (8 mol% Y2O3-ZrO2) powders were blended at the weight ratio of 60:40 and formed into a tube by an extrusion molding method or a rubber pressing method. The porous support was produced by sintering the tube in air at 1400°C. A slurry of 8YSZ was dipcoated on the outer surface of the support and fired to form a barrier layer. The barrier layer is 20–60 µm in thickness and prevents interdiffusion of palladium and silver in the membrane and Ni in the catalytic support. The hydrogen permeation membrane was formed on the support coated with barrier layer by electroless plating of palladium and silver and subsequent heat treatment for alloying them. The thickness of the membrane was 7–20 µm. The module was heat-treated in hydrogen atmosphere at 600°C for 3 h to reduce NiO in the support into Ni. The size of the MOC module was 10 mm outer diameter and 100–300 mm length. Figure 12.13 is a photograph of MOC module of 300 mm in length. The MOC module has a membrane area of about 93 cm2.

12.3.4

Effectiveness of barrier layer in MOC module

To investigate whether the membrane stability was influenced by the existence of the interlayer between the Ni–YSZ porous support catalyst and the palladium membrane, the membrane-support assemblies with and without the interlayer were annealed at 600ºC in hydrogen atmosphere. Figure 12.14a, b are photographs of cross-section SEM images of the membranesupport assembly without the interlayer, before and after annealing. After 1000 h annealing, it was found that palladium membrane disappeared in the membrane-support assembly without the interlayer, which implies that palladium had disappeared by the interaction between the palladium and the constituent of Ni–YSZ as a result of the contact of the membrane-support directly. Figure 12.15a and b are photographs of cross-section SEM images of the membrane-support assembly with the interlayer before and after annealing in hydrogen atmosphere. It was verified that the membranesupport assembly with interlayer still preserved the structural soundness of palladium membrane after 2000 h.

© Woodhead Publishing Limited, 2013

Membrane reactor for hydrogen production from natural gas

503

(b)

(a)

Pd film

Porous support

12.14 Cross-section SEM images of the membrane–support assembly without the interlayer after the hydrogen exposure test at 600ºC for 1000 h. (a) Cross-sectional view before annealing (×2000) and (b) cross-sectional view after annealing (×2000).

(a)

(b) Pd film

Pd film Barrier layer

Barrier layer

Porous support

Porous support

12.15 Cross-section SEM images of the membrane–support assembly with the interlayer after the hydrogen exposure test at 600ºC for 2000 h. (a) Cross-sectional view before annealing (×2000) and (b) cross-sectional view after annealing (×2000).

12.3.5

Performance of MOC module

Reforming tests were carried out with a single tube closed at one end of the MOC module. The size of the MOC module used in the tests was 10 mm in outer diameter, 300 mm in length, 93 cm2 in membrane area and 7.3 µm in palladium−silver alloy membrane thickness. The MOC module was set in a cylindrical reactor tube and heated to a given temperature by electric furnace. The fuel gases were introduced into the inside of MOC module (upstream side) through the fuel-feeding tube, and hydrogen permeated through the membrane to the outside of the MOC module (downstream side). The MOC module was tested at 0.9 MPa of upstream pressure and the pressure of the downstream was atmospheric pressure or under ambient pressure at 500 ~ 600°C. The natural gas at the flow rate of 280 Nml/min (3 Nml/min/cm2) was introduced into the unit at an S/C of 3.0.

© Woodhead Publishing Limited, 2013

Handbook of membrane reactors (a)

100

12

Hydrogen flux (Nml/min/cm2)

10

80

8 60 6 40 4 Hydrogen flux Hydrogen recovery Conversion Equilibrium conversion

2 0 490

500

510

520

530

540

550

20

Conversion (%), hydrogen recovery (%)

504

0 560

Temperature (°C) 100

12 10

80

8 60 6 40 4 Hydrogen flux Hydrogen recovery Conversion Equilibrium conversion

2 0 490

500

510

520

530

540

550

20

Conversion (%), hydrogen recovery (%)

Hydrogen flux (Nml/min/cm2)

(b)

0 560

Temperature (°C)

12.16 Temperature dependence of the reforming performance (experimental conditions: reaction side pressure/permeation side pressure: (a) 0.9 MPa/0.1 MPa and (b) 0.9 MPa/0.04 MPa; natural gas input, 3.0 Nml/min/cm2; S/C, 3.0).

Figure 12.16a and b shows the test results of temperature dependence of reforming performance at the hydrogen permeation side pressure of 0.1 MPa and 0.04 MPa. The reforming performance is plotted in terms of the product hydrogen flux, the hydrogen recovery and the conversion. The conversion is much

© Woodhead Publishing Limited, 2013

Membrane reactor for hydrogen production from natural gas

505

higher than the equilibrium conversion shown in the dashed curves. This means that the membrane reformer functioned successfully in producing hydrogen without limitation of chemical equilibrium. In the best case where the permeate side pressure was 0.04 MPa and the temperature was 546.5ºC, the product hydrogen flux was as high as 11.3 Nml/min/cm2 and the conversion was as high as 87.8%. This result shows that the MOC module has a potential similar to that of a conventional membrane reformer which is composed of a membrane module and separated fixed bed catalyst.

12.4

Conclusions and future trends

Hydrogen, which does not generate CO2 at the time of use as a fuel, will be a major energy option and one making an important contribution to the realization of a low carbon society. Recently there has been extensive development and demonstration globally of fuel cell systems for residential use, FCVs and hydrogen refueling stations. The commercial sale of a 1 kW-class residential fuel cell system, which is named ENE-FARM as the unified brand name for residential fuel cell systems, was started by some gas utility companies in May 2009 in Japan. As for FCVs, which are expected to contribute to the reduction of CO2 emission in the transportation sector, several auto manufacturers in the Europe, USA and Asia have announced that they are going to launch commercial models of FCVs in 2015. In order for these applications of hydrogen to become widespread, hydrogen supply systems must be established, together with a desire for environmental friendliness, energy security and economic efficiency. Hydrogen is a secondary energy, which is obtained from primary energy sources through some processing. Hydrogen from renewable resources will essentially go a long way toward solving global warming; however, the supply stability and volume of renewable energy is not enough for practical use under the present circumstances and it is believed that the use of renewable energy is unlikely to be realized in the near future. Therefore, use of hydrogen produced from fossil fuels is necessary. This, in combination with CO2 capture and storage technologies, will lead to substantial reductions in CO2 emissions and will be a pragmatic approach to a low carbon society. Hydrogen production systems based on natural gas steam reforming is a potential hydrogen supply infrastructure options, and to put it into wide use, high efficiency of hydrogen production and easy CO2 capture are required in the near future. A membrane reformer with a rated hydrogen production capacity of 40 Nm3/h has been developed under the NEDO program and the operation test proved the potential advantages of the membrane reformer of simple system configuration, compactness and energy efficiency as high as 80%.

© Woodhead Publishing Limited, 2013

506

Handbook of membrane reactors

Membrane reformer technology is believed to give the highest efficiency in producing hydrogen from natural gas among the various competing technologies. The technical feasibility of CO2 capture from the MRF by direct liquefaction was also shown. As a result of preliminary operation tests of CO2 capture, it was demonstrated that CO2 emissions can be reduced by 50–60% with only minor energy penalties. The distributed on-site highly efficient hydrogen production system based on the membrane reformer with CO2 capture is one of the promising technologies that can contribute to realizing a low carbon society and sustainable growth for energy supply while using fossil fuels. To commercialize the membrane reformer technology, the durability and reliability must be improved and the cost needs to be reduced significantly. Presently, TGC continues the long-term operation test of the MRF for verification of durability and reliability, and development of an advanced membrane module which will be a key technology for cost reduction.

12.5

Acknowledgments

Development of a 40 Nm3/h-class membrane reformer was conducted with subsidy from The New Energy and Technology Development Organization (NEDO). Financial support and technical assistance are greatly acknowledged. The authors would like to thank the coworkers of Mitsubishi Heavy Industries, Ltd for collaboration in development of the 40 Nm3/hclass membrane reformer and of NGK Spark Plug Co., Ltd for collaboration in developing advanced hydrogen separation module, which we call MOC (Membrane On Catalyst) module. The authors also acknowledge Dr Hisataka Yakabe and Dr Hideto Kurokawa of Tokyo Gas Co., Ltd for their helpful suggestions and supports to compile this report.

12.6

References

Fujimoto Y (2001), ‘Development of hydrogen production with membrane reactor’, Mitsubishi Juko Giho, 38(5), 246–249. Kuroda K (1996), ‘Study on performance of hydrogen production from city gas equipped with palladium membranes’, Mitsubishi Juko Giho, 33(5), 346–349. Kurokawa H (2011), ‘Energy-efficient distributed carbon capture in hydrogen production from natural gas’, Energy Procedia, 4, 674–680. Lewis F A (1967), The palladium hydrogen system, London, Academic Press. Nishii T (2009), ‘Reforming performance of hydrogen production module based on membrane on catalyst’, Proceedings of 9th International Conference on Catalysis in Membrane Reactors, Lyon, France. Rostup-Nielsen J R (1984), Catalysis, Berlin, Springe-Verlag, Vol. 5, p. 1.

© Woodhead Publishing Limited, 2013

Membrane reactor for hydrogen production from natural gas

507

Sakamoto Y (1992), ‘Permeability and diffusivity of hydrogen in palladium-rich Pd-Y(Gd)-Ag ternary alloys’, J Alloys Compounds, 185, 191–205. Shirasaki Y (2009), ‘Development of membrane reformer system for highly efficient hydrogen production from natural gas’, Int J Hydrogen Energy, 36, 4482–4487. TEPCO website, TEPCO Group Sustainability Report 2010, Available from: http:// www.tepco.co.jp/en/challenge/environ/pdf-1/10report-e.pdf (Accessed 1 August 2011). Tokyo Gas website, Tokyo Gas Group CSR Report 2010, Available from: http:// www.tokyo-gas.co.jp/csr/report_e/environment/06_09.html (Accessed 1 August 2011). Uemiya S (1991), ‘Hydrogen permeable palladium–silver alloy membrane supported on porous ceramics’, J Membrane Sci, 56 (3), 315–325. Yasuda I (2007), ‘Development and demonstration of membrane reformer system for highly-efficient hydrogen production from natural gas’, Materials Science Forum Vols. 539–543, Switzerland, Trans Tech Publications, 1403–1408. Available from: http://www.scientific.net.

© Woodhead Publishing Limited, 2013

13 Integrating membranes into industrial chemical processes: a case study of steam reforming with membranes for hydrogen separation M. DE FALCO, University Campus Bio-Medico of Rome, Italy, G. IAQUANIELLO, Tecnimont KT S.p.A., Italy, A. SALLADINI, Processi Innovativi S.r.l., Italy and E. PALO, Tecnimont KT S.p.A., Italy

DOI: 10.1533/9780857097347.2.508 Abstract: This chapter focuses on the operating experience and the results obtained from experiments carried out at a natural gas steam reformer plant integrated with Pd- and Pd–Ag-based membrane modules for hydrogen separation. The plant is based in Chieti Scalo, Italy, and was developed within the framework of an Italian research project entitled ‘Pure hydrogen from natural gas reforming up to total conversion obtained by integrating chemical reaction and membrane separation’. The first section of the chapter deals with the advantages relevant to the integration of membranes in an equilibrium reaction environment, paying particular attention to steam reforming reactions. Two different configurations are described and analyzed. The second section deals with the description of the reformer and membrane module plant in Chieti Scalo. The process scheme is presented, along with descriptions of the main sections of the plant, including reforming reactors, membrane modules and the control system. The third section reports the experimental results obtained. Attention is focused on the performance of membranes and their integration into the steam reforming plant, as well as on the performance of the catalysts used. Experimental data are discussed in order to characterize the permeability of the membranes employed and to define their long-term stability. Key words: methane steam reforming, Pd/Pd–Ag membrane, hydrogen production, industrial operation.

13.1

Integration of selective membranes in industrial plants

The application of selective membranes in chemical processes represents one of the most interesting scientific and technological topics of the last 508 © Woodhead Publishing Limited, 2013

Integrating membranes into industrial chemical processes

509

ten years, as indicated by the increasing number of publications on the subject (Basile, 2008). The integration of a membrane in an environment in which reactions are equilibrium-limited allows the conversion of specific reactants under milder conditions. The result has, consequently, increased overall efficiency and process intensification. In recent years, research activities have shown the potential of membrane technology applied to different equilibrium reactions, mainly with the aim of hydrogen production, such as steam reforming (Botero et al., 2008; Carrara et al., 2011; Lin and Rei, 2001; Matsumura and Tong, 2008; Shirasaki et al., 2009), partial oxidation and autothermal reforming (Basile et al., 2001; Chang et al., 2010; Simakov and Sheintuch, 2009), water gas shift (Criscuoli et al., 2000; Pinacci et al., 2010) and dry reforming (Gallucci et al., 2008; Paturzo et al., 2003). Other applications concern hydrogen purification for use in fuel cells (Brunetti et al., 2008; Swesi et al., 2007) and for use in an Integrated Gasification Gas Combined Cycle (IGCC) with CO2 capture (Amelio et al., 2007; Ku et al., 2011). Due to its thermodynamic characteristics, the methane steam reforming reaction appears to be one of the best candidates for membrane application. The reaction is strongly endothermic and equilibrium-limited, and high temperatures of up to 850–900°C are necessary to reach a significant feed conversion. As the process requires a high heat input level, the catalyst-filled tubes are placed in the radiant firebox section of the furnace where a fraction of the natural gas feed has to be burned as fuel. The latter causes a reduction of global process efficiency, together with increased greenhouse gas emissions. The integration of Pd-based membranes able to selectively separate hydrogen from the reaction environment allows for an increase in hydrogen yield at a lower temperature, which can have significant effects on global process efficiency. The lowering of the reaction temperature from 850–900°C to 450–650°C means that a low-grade heat, with a lower energy than that of hot flue gas from burners, can be used. Exhaust from gas turbines or clean heating fluids not derived from combustion may be used, thereby reducing or eliminating CO2 emissions (De Falco et al., 2008, 2011b). Moreover, the lower temperature allows the heat supplied to be better exploited and permits the use of cheaper alloy steel for the reformer tubes due to the lower tube metal temperature. A selective membrane device can be integrated into a reaction environment in two different ways: directly inside the reaction environment, or assembled in a series of reaction units according to an open architecture. In the following sections, these two configurations are described and evaluated with a steam methane reforming reaction used as an example to elucidate benefits and drawbacks.

© Woodhead Publishing Limited, 2013

510

Handbook of membrane reactors

13.1.1

Integrated membrane reactor

A membrane reactor (MR) is a device in which a selective membrane is assembled directly inside the reaction environment, allowing reaction and separation within the same unit. Although more compact, MR architecture presents some technical problems related to membrane thermal stability as well as to catalyst and membrane maintenance. When reaction and separation take place in the same environment, the rate-limiting step is represented by the maximum allowable membrane temperature that, despite many attempts to develop thermally stable thin membranes, is limited to 500°C (Okazaki et al., 2011). A compromised temperature of about 550°C is usually chosen for MR architecture, so as to not penalize excessively the endothermic steam reforming reaction. However, higher selective layers are necessary to assure long term membrane stability at temperatures over 450°C. The increased thickness is accompanied by an increased membrane area, which is necessary to guarantee the same hydrogen separation capacity, since membrane permeance is inversely proportional to its thickness. The palladium used in the membrane is expensive; therefore, the increased area and thickness required may bring about an important additional cost. Chen et al. (2008) cite an economic evaluation stating that when the thickness of metallic palladium or palladium alloy films is higher than 20 microns, the process is no longer cost effective in comparison with other processes. An MR scheme is reported in Fig. 13.1. The simplest configuration is composed of two concentric tubes. The catalyst is packed in the annular zone, while the inner tube is the membrane itself. A sweeping gas may be fed through the inner tube, co-currently or counter-currently, in order to increase the driving force of hydrogen separation. Two different zones can be recognized: the reaction zone at higher pressure, which is the

CH4 + H2O

Filled catalyst bed

CH4 + H2O

Membrane

Retentate CO, CO2, CH4, H2O, (H2) Permeate H2 (+ sweep steam) Retentate CO, CO2, CH4, H2O, (H2)

13.1 Membrane reactor scheme applied to steam reforming reaction.

© Woodhead Publishing Limited, 2013

Integrating membranes into industrial chemical processes

511

annular section where the catalyst is packed, and the permeation zone at lower pressure, where the sweeping gas is fed. Although not often used on an industrial scale, the advantages of using a steam reforming reactor equipped with a hydrogen selective membrane over a conventional fixedbed reactor for the methane steam reforming reaction have been widely recognized and investigated. Conversions in the range of 80–95% were obtained by Shirasaki et al. (2009) during steam reforming of natural gas in a compact MR equipped with Pd–Ag membranes working at temperatures of 495–540°C, a pressure of 0.9 MPa, and a steam-to-carbon ratio of 3–3.2. Methane conversion of up to 98.8% was also obtained in a MR equipped with an ultra-thin permeable Pd-based membrane under mild working conditions (Chen et al., 2008).

13.1.2

Reformer and membrane module

In a reformer and membrane module (RMM) configuration, a hydrogen selective membrane is assembled downstream of the reaction units in a series of reaction and separation modules. By decoupling reaction and separation, it is possible to optimize reforming temperature independently from membrane constrains and to adopt milder operating conditions at the membrane stage. The latter not only increases membrane durability but also allows for the use of thinner membranes, achieving higher hydrogen separation efficiency. Although less compact than MR, since it requires two devices to carry out reaction and separation, RMM appears more feasible from an engineering point of view. RMM reformer tubes may be designed traditionally, while in MR architecture they need to be enlarged in order to accommodate a sufficient membrane area (Barba et al., 2008; Li et al., 2008), with a consequent increase in heat transfer area and reformer costs. Together with a simplification of the mechanical design of both devices, the open architecture allows for a simpler membrane module maintenance and catalyst replacement, which makes this configuration more suitable for industrial scale applications. An RMM scheme based on the two stages of reaction and separation is shown in Fig. 13.2. A mixture of natural gas and steam is sent to the first reactor, where it is partially converted into the products. The stream produced is cooled prior to being routed to the membrane module, where hydrogen is recovered and retentate is sent onto the next step. It is possible to replicate the reaction−separation steps until the desired natural gas conversion is achieved. In the following section, results obtained from a RMM Tecnimont KT industrial pilot plant, based on two stage reaction and separation, will be reported and discussed.

© Woodhead Publishing Limited, 2013

512

Handbook of membrane reactors

CH4 + H2O

I Reaction stage

II Reaction stage Heater Retentate

Cooler

CO, CO2, CH4, H2O, (H2)

Cooler

Permeate I Separation stage

II Separation stage

H2 (+ Sweep steam)

13.2 RMM scheme based on two stages.

13.2

Reformer and membrane module Tecnimont KT plant

The RMM Tecnimont KT plant was built within the framework of the Italian FISR research project entitled ‘Pure hydrogen from natural gas reforming up to total conversion obtained by integrating chemical reaction and membrane separation’ and financed by the Italian Ministry of Research and University (MIUR). As the title suggests, the aim of the project was to test the integration of hydrogen selective membranes within a steam reforming scheme according to an open architecture. The effectiveness of membrane integration, the performance of structured catalysts working at low temperatures, and the long term stability of a Pd-based membrane in a real steam reforming environment were all areas of research at the plant. The building of the plant was concluded at the end of 2009 and experimental tests began in April 2010. With regard to the catalysts used, noble metals based on activated and stabilized Al2O3, supported on SiC foam were selected. Under low reaction temperature in fact, noble metal based show a higher activity than traditional nickel-based catalysts. The open cross-flow typical of foam has the advantage, in a process where the rate-limiting factor is the heat transfer from the flue gas to the reforming tube wall, of increasing the effectiveness of the heat transfer through convection and thereby reducing the tube metal temperature. The catalyst is the SR-10, produced by BASF and composed of Rh–Pt. It is shaped into cylindrical elements with a diameter of 60 mm and a length of 150 mm.

© Woodhead Publishing Limited, 2013

Integrating membranes into industrial chemical processes

513

With regard to the membranes used, three different modules of dissimilar shape, using different kinds of supports and with selective layers of different thicknesses, were selected. In particular, two membranes were installed in parallel on the first reforming stage while a third membrane was installed downstream for the second reforming stage. Due to its modular design, the plant is quite flexible, and allows different kinds of catalyst and membrane modules to be tested.

13.2.1

Plant description

The plant process scheme is illustrated in Fig. 13.3a, together with a bird’s eye view of the constructed industrial plant, which covers an area of 1000 m2 inside the Scientific and Technological Park of Chieti Scalo (Fig. 13.3b). The plant is based around two-step RMMs designed for a capacity of 20 Nm3/h. The radiant boxes of both reformer stages contain a catalyst tube with a heated length of about 3 m where the common convective section receives hot flue gas from the radiant chambers and contains three coils to preheat feeds and superheat process steam. Exhaust flue gases are evacuated into the atmosphere through a vertical stack placed above a convective section. The design of the radiant chamber is quite conventional; it differs only by the heated length of the reformer tube, the tube metallurgy, such as stainless steel instead of the exotic and quite expensive material HP25/35 chromium/ nickel alloy, and the contained catalyst. Natural gas is derived from the town gas line and is available at battery limits at a pressure of 12 barg. It is introduced through the pressure regulator and flow controller to the feed desulfurization (DS) reactor, where sulfur compounds are removed up to 0.1 ppm. The DS feed is then mixed with steam in a ratio ranging from 3 to 5, preheated in the convection section and fed to the first reformer tube. The heat for the reaction in both reforming steps is provided by two independent hot gas generators, allowing the required reforming temperatures to be set. Flue gas temperatures range from a minimum of 450°C to a maximum of 900°C, allowing reforming temperatures of up to 750°C to be reached. Reformate streams are cooled prior to being sent to the membrane modules through a dedicated air cooler equipped with a variable speed motor. In this way, the inlet membrane temperature may be controlled and kept to the required design value. Membrane modules are protected from excessive temperatures by the use of a pressure relief regulator installed on the income lines and venting to the flare. On the permeate side, steam may be added as a sweeping stream to increase the driving force for separation by lowering the partial pressure of hydrogen. A retentate, and a mixture of hydrogen plus

© Woodhead Publishing Limited, 2013

514

Handbook of membrane reactors

(a)

Steam

Stack

∝

Connection section

Natural gas ∝ ∝ DS

N2

∝

1st reformer Flue gas step

Flare ∝

2nd reformer Air step

Air

∝

∝ ∝

Retentate Air

∝ Air

2nd/3rd membrane module

∝

∝

∝ Sweep steam

∝

∝

Pure hydrogen

∝

1st membrane module

∝ Water condensate

(b)

13.3 (a) Schematic of the process and (b) bird’s-eye view of the industrial test plant.

sweeping steam are produced from the membrane modules. Retentate from the first membrane stage is recycled to the second reformer while retentate from the second membrane is sent to the flare. Permeate streams from both modules are mixed and sent to the final cooling and condensate separation. The pressure of both the shell and permeate sides is controlled using

© Woodhead Publishing Limited, 2013

Integrating membranes into industrial chemical processes

515

back-pressure regulators. All the vent points are connected to the main vent system and routed to the flare. The latter receives the produced hydrogen, the retentate from the second membrane stage and the exhaust from the control valves together with a continuous pilot fuel gas stream. Utilities required for the running of the plant include nitrogen, instrument air, cooling water, steam and boiler feed water production. Nitrogen for plant start-up and purging is stored in liquid state and vaporized at a pressure of 18 barg. Gaseous nitrogen in cylinders is also available at battery limits to be used in case of a fault with the liquid nitrogen package. A dedicated compressor, equipped with a membrane dryer for moisture removal, produces dry instrument air while a package comprising two reverse osmosis stages, followed by a finishing step of ion exchange resins, produces boiler feed water. A hot oil boiler produces saturated steam up to a pressure of 27 barg to be fed to the superheated section of the convective. Cooling water to be used for sweep steam condensation is produced from an air cooling tower working in a closed loop. Sample connections, together with pressure, temperature and flow measurement points, are located at the inlet and outlet of the reformer and membrane modules to measure the performance of the RMM. A multipoint thermocouple is installed inside the first reformer tube in order to monitor the axial temperature profile along the heated catalyst length while two glass peepholes allow the reformer tube metal temperature to be measured by an infrared pyrometer. The control room is located in a safety area with a bird’s eye view of the plant area.

13.2.2

Selective membrane modules

During the last decade, a significant effort has been made to produce thin or low thickness Pd-based membranes supported on different porous substrates able to realize high hydrogen fluxes together with high hydrogen selectivity. From among the few membrane provider/developers able to produce membranes with surface dimensions higher than laboratory scale, three of them – the Energy Research Centre of the Netherlands (ECN), an ‘undisclosed’ company from Japan and Membrane Reactor Technologies Ltd (MRT) from Canada – were selected and included in the RMM research project. A fourth company, Acktar from Israel, was also involved; however, its membrane will be tested later. Characteristics of the three installed membrane modules are reported here. Figure 13.4a shows the ECN module equipped with 13 tubular membranes achieving a total area of 0.4 m2. A Pd membrane layer with a thickness of around 2.5 microns is applied by electroless plating onto a support manufactured from commercial alumina tubes with the addition of two alumina

© Woodhead Publishing Limited, 2013

516

Handbook of membrane reactors (c)

(a)

(b)

13.4 Membrane details: (a) ECN module; (b) MRT module; (c) Japanese module.

layers for improved surface smoothness. Membrane tubes with an effective length of 69 cm and an outer diameter of 1.4 cm are housed in a 5 inch shell. Outlet connections are made with flexible metal hoses and Swagelok ends in order to absorb mechanical stresses due to thermal expansion of surrounding piping and keep membrane module below allowable forces. The module may be used with or without sweeping steam. Figure 13.4b shows the MRT module equipped with five membrane elements consisting of two double-sided planar membrane panels welded in series. Each panel has an active palladium−silver alloy membrane area of 0.03 m2 achieving a total installed area of 0.6 m2 with a selective layer of 25 microns. These modules incorporate MRT’s patented membrane sealing techniques to provide a leak free seal between the 25 micron silver−palladium foil and the stainless steel module substrate. The membrane elements are housed in a rectangular core which, along with the inlet distributor, promotes uniform reformate flow across the membrane modules. The entire system is housed in a 6 inch shell along with the arrangement for a sweeping gas steam. Figure 13.4c shows the Japanese module, equipped with three tubular membranes having an external diameter of 3.0 cm, an effective length of 45 cm and a Pd–Ag layer thickness of about 2.5 micron. They are produced in a three-step procedure. First, Pd is deposited onto the Al2O3 support by electroless plating. Ag is then layered on by electroplating, using the Pd layer as the electrode. Finally, the layered Pd–Ag membrane is heat-treated to obtain the Pd–Ag alloy membrane. The module achieves a total membrane area of about 0.13 m2. The three elements are housed in a 6 inch shell without a sweeping gas stream arrangement.

© Woodhead Publishing Limited, 2013

Integrating membranes into industrial chemical processes

517

13.5 ECN, MRT and Japanese modules installed on RMM Tecnimont KT plant.

Figure 13.5 shows all three modules installed in the RMM Tecnimont KT plant: the first two from the left are the ECN and MRT modules respectively, working in parallel or one at a time downstream to the first reformate. The third is the Japanese module working on the second reformate.

13.2.3

Control system and design of testing phase

The main features of the control system of the plant are reported here. A dedicated programmable logic controller (PLC) handles both regulation loops and an alarms−interlocks system. If any process condition exceeds the design values, a trip system is activated and an interlock sequence restores a safe state. The control panel operator monitors alarm indicators and may take appropriate action when needed to avoid a plant shut-down. Figures 13.6a and 13.6b show two main control graphic pages related to the steam reformer and membrane separation sections, respectively. Other graphic interfaces allow the operation of utilities and the state of the permissive/interlock system to be supervised. A dedicated control page also allows the main operating parameters to be monitored. The latter is particularly useful during plant start-up because the membrane heating rate has

© Woodhead Publishing Limited, 2013

(a)

(b)

13.6 (a) Steam reforming section control system; (b) membrane separation section control system.

© Woodhead Publishing Limited, 2013

Integrating membranes into industrial chemical processes

519

to be kept below the specified design level to avoid thermal stresses. The ECN and the Japanese modules may be heated at a rate of up to 2.5°C/min, while the MRT module may be heated at a rate of up to 5°C/min. Heating is performed with an increasing nitrogen flow up to dew point temperature, then a mixture of nitrogen and steam is used to increase the heat capacity of the heating stream. To prevent membrane damage due to hydrogen swelling and embrittlement, membranes should not be exposed to a hydrogen environment where the surface temperatures are below 350°C for the MRT module and 300°C for the other two modules. This means that natural gas should not be introduced into the reformer until the membrane temperatures are below these values. The flow of the natural gas, steam and produced hydrogen are controlled through a regulation loop. Temperatures and pressures are measured at the inlet and outlet of both the reformer and membrane stages, while a differential pressure sensor monitors the pressure drop across the desulfurizer reactor as a measure of its absorbent capacity. A variety of syngas compositions, ready to be fed to the membrane modules, can be produced by acting on the steam/carbon ratio and on the reformer outlet temperature through the variation of the firing of two hot gas generators. This makes it possible to study the role played by partial pressures of the main components in separation performance. The effect of temperature on membrane permeance at fixed reformer working conditions may be studied by varying the temperature set point of two air coolers placed at the inlet of the membrane modules. To evaluate RMM performance regarding methane conversion and hydrogen recovery, it is necessary to measure the content of the output stream from the reformer and membrane modules. The composition of the reformed gas and retentate streams was detected by an ABB analyzer. CH4, CO and CO2 concentrations were measured by the online non-dispersive infrared (NDIR) multiple analyzer, ABB URAS14. H2 was analyzed using the thermal conductivity detector ABB Caldos 17. A Perkin Elmer Gas Chromatographer unit (CLARUS 500) was used to analyze the composition of the permeate streams.

13.3

Reformer and membrane module plant behavior

Two sets of experimental tests (Set I and Set II) were carried out at the RMM plant spanning about 1000 h during 2010 and 200 h during 2011, respectively. Weekly as well as daily continuous tests were performed with the system undergoing frequent heating-cooling cycles and adopting a wide variety of operating conditions. In particular reforming outlet temperatures were investigated from 500°C to 680°C, steam to carbon ratio from 3.8 to

© Woodhead Publishing Limited, 2013

520

Handbook of membrane reactors 74% Hydrogen 72% 15%

70% 68%

10% 66% 5% 64%

0% 540

560 580 600 Reformer outlet temperature (°C)

Hydrogen dry molar fraction (%)

Methane dry molar fraction (%)

20%

Methane

62% 620

13.7 Methane and hydrogen content at the outlet of the reforming reactor as a function of reformer outlet temperature.

4.8 and gas hourly space velocity (GHSV) from 4300 to 7700 h−1, where the GHSV value is defined as the ratio between the total gaseous flow rate fed to the reactor (referred to 0°C and 1 atm) and the total catalytic bed volume. Catalytic activity tests carried out by changing the GHSV value in the range 4300–6900 h−1 showed a decrease in methane conversion from 48.7% to 46.5%, since the steam reforming reactions are affected by low contact times (De Falco et al., 2011a). Through an increase of outlet reformer temperature from 550°C to 615°C, methane content decreases quite linearly from 18% to 8.5% whereas hydrogen content increases from 64% to 70.5%, as indicated in Fig. 13.7. The latter appears to reach a plateau at higher temperature since, although steam reforming reaction is favored, hydrogen derived from carbon monoxide through shift reaction is penalized. However, no catalyst deactivation occurred, as evidenced by stable methane content over the entire test period of Set I (Fig. 13.8). Regarding membrane behavior, all three membrane modules showed good stability under the real environment of the reformed gas mixture. Over 600 testing hours were collected at the ECN module. Set I was carried out with weekly tests; the plant was started up on Monday and was kept running continually throughout the week. As reported in Fig. 13.9a, permeance was subjected to a little decay after about 250 h. This slight drop in performance seems to coincide with the test carried out at membrane temperatures lower than 380°C and with a higher content in CO. These operating conditions may have favored a reversible adsorption of carbon monoxide on the membrane

© Woodhead Publishing Limited, 2013

Integrating membranes into industrial chemical processes

Methane dry molar fraction (%)

35 30

GHSV = 4290 h−1

GHSV = 5145 h−1

−1

GHSV = 6870 h−1

GHSV = 6000 h

521

25 20 15 10 5 0 0

200

400 600 Time test (h)

800

1000

13.8 Methane content at the outlet of the reforming reactor as a function of testing time and GHSV.

surface with a consequent inhibition effect (Scura et al., 2008); however, this hypothesis is under more extensive study by the authors. Set II was carried out with a daily test: every morning the plant was started up and kept running until evening. Although membranes were subjected to more stress, due to the daily start-up, frequent heating cycles under nitrogen could have kept the membrane surface cleaner, resulting in the higher permeance achieved than that of the last tests of Set I. Due to the better heat conservation of the membrane and the more accurate temperature measurement adopted during Set II, permeance data showed a better correlation coefficient on Arrhenius plot (Fig. 13.9b) as well as lower activation energy, as indicated by a lower slope. This behavior may also suggest that the membrane was not influenced by any relevant inhibition or polarization effect influencing membrane permeance dependency on temperature. Tests of the ECN module as part of Set I showed an activation energy much higher than literature values. This could be explained by the coexistence of phenomena promoting membrane inhibition, resulting in a fictitious higher activation energy. A total of about 330 testing hours were collected on the MRT module. Set I and Set II were carried out in the same way as on the ECN module with weekly and daily tests, respectively. Although the testing time was lower than that for the ECN module, no sign of performance decay was observed during Set I. As shown in Fig. 13.10a the performance of the MRT module showed a temporary decline only during the first test of Set II. Starting from a value about five times lower than that of previous test (Set I), membrane performance improved progressively before reaching a steady state value similar to that of Set I. If initial non-steady state points are excluded from

© Woodhead Publishing Limited, 2013

522

Handbook of membrane reactors (a) 0.30

0.25

420 400

0.20

380

0.15

360 0.10

Temperature (°C)

Permeance (kmol m−2h−1kPa−0.5)

440

340 0.05

320

Permeance I

Temperature

II

0.00 0

100

200

300 400 Time (h)

500

600

300 700

In(permeance) (kmol m−2h−1kPa−0.5)

(b) −1.4 Set I-2010

−1.6

Set II-2011

−1.8 −2.0 −2.2 −2.4 −2.6 −2.8 −3.0 1.40E-03

1.45E-03

1.50E-03 T

1.55E-03

1.60E-03

(K−1)

13.9 (a) Permeance of ECN module as a function of temperature and testing time; (b) Arrhenius plot for ECN module.

correlation, the dependency of membrane permeance on temperature on Arrhenius plot appears to be the same as Set I, as indicated by the same trend of correlating line. Although the MRT module showed a higher permeability than the ECN module, it achieved a lower level of permeance due to a membrane thickness an order of magnitude higher. Membrane permeance values were calculated by assuming the Arrhenius law for the dependence of permeability from temperature and the Sievert–Fick’s law

© Woodhead Publishing Limited, 2013

Integrating membranes into industrial chemical processes 440

0.10

420

0.08

400 380

0.06

360 0.04 340 0.02

Permeance I

Temperature

0.00 0

50

100 150

II

200 250 300 350 Time (h)

Temperature (°C)

Permeance (kmol h–1 m–2 kPa–0.5)

(a) 0.12

523

320 300

In(permeance) (kmol h−1 m−2 kPa−0.5)

(b) −2.0 −2.5

Set I-2010 Set II-2011 Set III-2011 fouled

−3.0 −3.5 −4.0 −4.5 −5.0 1.40E-03 1.45E-03 1.50E-03 1.55E-03 1.60E-03 T (K−1)

13.10 (a) Permeance of MRT module as a function of temperature and testing time; (b) Arrhenius plot for MRT module.

for hydrogen flux expression. Results obtained from correlation during Set I are reported in Table 13.1. The integration of the ECN and MRT modules into steam reforming architecture allows for an improved hydrogen yield due to further conversion in the second reforming stage. Figure 13.11 shows methane content at the outlet of two reforming reactors and of the intermediate membrane module. To balance the lower amount of hydrogen at the outlet of the membrane module, molar fractions of other components in the mixture increases. The methane molar fraction increases from about 10% to 30%, depending on the hydrogen recovery factor; it is further reduced in the second reformer reactor. Figure 13.12 shows enhancement of methane conversion due to

© Woodhead Publishing Limited, 2013

Handbook of membrane reactors Table 13.1 Permeability membranes characterization results Membrane module

A (kmol h−1m−1kPa−0.5)

Ea (kJ/mol)

R2

ECN membrane MRT membrane Japanese membrane

1.72E-1 5.75E-4 9.31E-2

77.0 35.3 80.4

0.71 0.61 0.84

Source: De Falco et al. (2011a). Ea, activation energy; R2, regression coefficient.

Methane dry molar fraction (%)

30 25 20 15 10 First reformer outlet 5

Membrane outlet Second reformer outlet

0 150

350

550 Time (h)

750

950

13.11 Methane content at the outlet of first and second reforming reactors and membrane modules.

75 Methane conversion (%)

524

70 65

Without membrane With ECN module With MRT module

60 55 50 45 40 590

600

610 620 630 640 650 Reformer temperature (°C)

660

13.12 Methane conversion of RMM architecture with ECN and MRT modules.

© Woodhead Publishing Limited, 2013

Integrating membranes into industrial chemical processes

525

integration of the ECN and MRT modules. The RMM architecture applied to the ECN module performed a feed conversion 10–12% higher than that of equilibrium values. An overall feed conversion of 57.3% was achieved at 610°C, about 26% higher than what can be achieved in a conventional reformer at the same temperature. Due to the lower hydrogen recovery factor, which characterizes the MRT module, enhancement in feed conversion is lower than that for the ECN module of about 3%. Working downstream from the second reforming reactor, the Japanese module does not help to increase global feed conversion.

13.4

Conclusions and future trends

The RMM architecture with a rated hydrogen production capacity of 20 Nm3/h was successfully tested. Results have demonstrated the potentialities of the integration of a Pd-based membrane in steam reforming plants according to an open architecture. However, to move towards an industrialization of membrane technologies it is necessary to fully characterize and improve membrane performance with relation to long term stability. Results obtained at the RMM Tecnimont KT plant could suggest a preferred procedure for plant management to reduce phenomena responsible for the loss of performance such as inhibition or polarization effects. Moreover, R&D efforts are aimed at designing reliable thin film fabrication methods directly in collaboration with membrane developers. In order to increase performance in terms of hydrogen recovery, it is necessary to optimize the preparation procedure by improving membrane permeability and reducing membrane thickness. The latter however still represents an issue in membrane manufacture. Future works will cover experimental test campaigns aimed at increasing testing time to completely understand long term behavior of Pd-based membranes, as well as to test different membrane modules in a realistic steam reforming environment. RMM architecture may be further optimized by implementing a suitable number of reaction−separation stages as well as by increasing membrane area. Making use of these experimental results, combined with larger membrane surfaces and more modules in series (De Falco et al., 2011b), the RMM architecture may achieve almost a complete feed conversion.

13.5

References

Amelio M, Morrone P, Gallucci F and Basile A (2007), ‘Integrated gasification gas combined cycle plant with membrane reactors: Technological and economical analysis’, Energy Convers Manage, 48, 2680–2693. Barba D, Giacobbe F, De Cesaris A, Farace A, Iaquaniello G, Pipino A (2008), ‘Membrane reforming in converting natural gas to hydrogen (part one)’, Int J Hydrogen Energy, 33, 3700–3709.

© Woodhead Publishing Limited, 2013

526

Handbook of membrane reactors

Basile A (2008), ‘Hydrogen production using Pd-based membrane reactors for fuel cells’, Top Catal, 51, 107–1224. Basile A, Paturzo L and Laganà F (2001), ‘The partial oxidation of methane to syngas in a palladium membrane reactor: simulation and experimental studies’, Catal Today, 67, 55–64. Botero M, Boyd T, Gulamhusein A, Comyn N, Lim J, Grace J R, Shirasaki Y and Yasuda I (2008), ‘Pure hydrogen generation in a fluidized-bed membrane reactor: Experimental findings’, Chem Eng Sci, 63, 2752–2762. Brunetti A, Barbieri G and Drioli E (2008), ‘A PEMFC and H2 membrane purification integrated plant’, Chem Eng Process, 47, 1081–1089. Carrara A, Perdichizzi A and Barigozzi G (2011), ‘Pd-Ag dense membrane application to improve the energetic efficiency of a hydrogen production industrial plant’, Int J Hydrogen Energy, 36, 5311–5320. Chang H-F, Pai W-J, Chen Y-J and Lin W-H (2010), ‘Autothermal reforming of methane for producing high-purity hydrogen in a Pd/Ag membrane reactor’, Int J Hydrogen Energy, 35, 12986–12992. Chen Y, Wang Y, Xu H and Xiong G (2008), ‘Hydrogen production capacity of membrane reformer for methane steam reforming near practical working conditions’, J Membr Sci, 322, 453–459. Criscuoli A, Basile A and Drioli E (2000), ’An analysis of the performance of membrane reactors for the water–gas shift reaction using gas feed mixtures’, Catal Today, 56, 53–64. De Falco M, Barba D, Cosenza S, Iaquaniello G and Marrelli L (2008), ‘Reformer and membrane modules plant powered by nuclear reactor or by solar heated molten salts: assessment of the design variables and production cost evaluation’, Int J Hydrogen energy, 33, 5326–5334. De Falco M, Iaquaniello G and Salladini A (2011a), ‘Experimental tests on steam reforming of natural gas in a reformer and membrane modules (RMM) plant’, J Membr Sci, 368, 264–274. De Falco M, Iaquaniello G and Salladini A (2011b), ‘Reformer and membrane modules (RMM) for methane conversion: experimental assessment and perspectives of said innovative architecture’, ChemSusChem, 4, 1157–1165. Gallucci F, Tosti S and Basile A (2008), ‘Pd-Ag tubular membrane reactors for methane dry reforming: A reactive method for CO2 consumption and H2 production’, J Membr Sci, 317, 96–105. Ku A Y, Kulkarni P, Shisler R and Wei W (2011), ‘Membrane performance requirements for carbon dioxide capture using hydrogen-selective membranes in integrated gasification combined cycle (IGCC) power plants’, J Membr Sci, 367, 233–239. Li A, Lim C J and Grace J R (2008), ‘Staged-separation membrane reactor for steam methane reforming’, Chem Eng J, 138, 452–459. Lin Y-M and Rei M-H (2001), ‘Separation of hydrogen from the gas mixture out of catalytic reformer by using supported palladium membrane’, Sep Purif Technol, 25, 87–25. Matsumura Y and Tong J (2008), ‘Methane steam reforming in hydrogen-permeable membrane reactor for pure hydrogen production’, Top Catal, 51, 123–132. Okazaki K, Ikeda T, Pacheco Tanaka D A, Sato K, Suzuki T M and Mizukami F (2011), ‘An investigation of thermal stability of thin palladium-silver alloy

© Woodhead Publishing Limited, 2013

Integrating membranes into industrial chemical processes

527

membranes for high temperature hydrogen separation’, J Membr Sci, 366, 212–219. Paturzo L, Gallucci F, Basile A, Vitulli G and Pertici P (2003), ‘An Ru-based catalytic membrane reactor for dry reforming of methane: its catalytic performance compared with tubular packed bed reactors’, Catal Today, 82, 57–65. Pinacci P, Broglia M, Valli C, Capannelli G and Comite A (2010), ‘Evaluation of the water gas shift reaction in a palladium membrane reactor’, Catal Today, 156, 165–172. Scura F, Barbieri G, De Luca G and Drioli E (2008), ‘The influence of the CO inhibition effect on the estimation of the H2 purification unit surface’, Int J Hydrogen Energy, 33, 4183–4192. Shirasaki Y, Tsuneki T, Ota Y, Yasuda I, Tachibana S, Nakajima H and Kobayashi K (2009), ‘Development of membrane reformer system for highly efficient hydrogen production from natural gas’, Int J Hydrogen Energy, 34, 4482–4487. Simakov D S A and Sheintuch M (2009), ‘Demonstration of a scaled-down autothermal membrane methane reformer for hydrogen generation’, Int J Hydrogen Energy, 34, 8866–8876. Swesi Y, Ronze D, Pitault I, Dittmeyer R and Heurtaux F (2007), ‘Purification process for chemical storage of hydrogen for fuel cell vehicles applications’, Int J Hydrogen Energy, 32, 5059–5066.

© Woodhead Publishing Limited, 2013

14 Economic analysis of systems for electrical energy and hydrogen production: fundamentals and application to two membrane reactor processes G. MANZOLINI , Politecnico di Milano, Italy and D. JANSEN, Energy Research Centre of the Netherlands, The Netherlands

DOI: 10.1533/9780857097347.2.528 Abstract: The aim of this chapter is to introduce a general methodology for the economic assessment of large scale plants for power and/or hydrogen production. The entire methodology is based on the definition and assessment of economic performance characteristics as (i) levelized cost of electricity (LCOE), (ii) cost of hydrogen production, and (iii) cost of CO2 avoided. The necessary inputs to determining these parameters are the thermodynamic performance of plants, as discussed in previous chapters, and the calculation of construction and operating costs. Costs assessment is discussed in detail here, with particular attention to innovative components such as membrane reactors. Finally, the methodology is tested in two particular applications of membrane reactors: power production, and combined hydrogen and power production, both including CO2 capture. Key words: economic analysis, cost of electricity, cost of CO2 avoided, cost of hydrogen production, membrane reactor cost assessment.

14.1

Introduction

This chapter presents a methodology to assess the economics of membrane reactors when applied to power and hydrogen production. From the many types and applications of membrane reactors discussed in previous chapters, this chapter will investigate only hydrogen selective membranes for power and/ or hydrogen production with CO2 capture; this type of membrane is considered the most promising application and with the potentially highest impact. Discussion will focus on natural gas derived syngas applications, as state-of-the-art hydrogen membranes cannot be applied to sour syngas derived from heavier fuels, including coal or oil, due to the effects of sulfur 528 © Woodhead Publishing Limited, 2013

Economic analysis of systems

529

poisoning and the presence of several other contaminants. However, the proposed methodology is general, and can be extended also to other membrane applications. In order to emphasize the potential of membrane reactors, reference cases for power and hydrogen production using state-of-the-art syngas separation technology will be also presented. As a starting point, it must be outlined that in the literature there are several methodologies for the economic assessment of power plants: proposed approaches vary from country to country and on author’s perspective (i.e., company employee or researcher). Examples of methodologies include those employed by the US Department of Energy National Energy Technology Laboratory, DOE-NETL (DOE-NETL, 2007) and the International Energy Agency, IEA (IEA, 2000, 2004). In this chapter, the methodology presented is that proposed by the European Benchmark Task Force (EBTF, 2011; Franco et al., 2010). This methodology is similar to that used by the IEA as far as the calculation of the cost of electricity is concerned. It is selected because it resulted from the contribution and agreement of several authors belonging to universities, research centers and companies. For this reason, it is considered an independent and neutral approach. This chapter is composed of four sections: in the first section, the main economic parameters are introduced. The second and third sections consider the assessment of investment cost and variable cost; in particular, a methodology to assess membrane reactor costs and their impact on variable cost is proposed. The last section discusses examples of proposed procedure applications for the reference cases (i.e., systems with more conventional separation technologies) and cases where membrane reactors are applied.

14.2

Calculation of the cost of electricity, hydrogen production and CO2 avoided

As anticipated in the Introduction, the application of membrane reactors for power and/or hydrogen production is now discussed. The advantages of membrane reactors in these applications include simultaneous fuel conversion and hydrogen separation, leading to thermodynamic advantages and potential equipment savings compared to conventional technologies. The first step before starting the economic assessment is the definition of the reference year in order to have consistent cost figures. For example in this work, 2010 was selected as reference year. The levelized cost of electricity (LCOE), cost of hydrogen production, and cost of CO2 avoided are considered as the main economic performance characteristics of the applications investigated.

© Woodhead Publishing Limited, 2013

530

Handbook of membrane reactors

The calculation method of these three parameters is now described in detail: The LCOE is different from the price cost, which includes profits, and is calculated by setting to zero the net present value (NPV) of the power plant (similar also to IEA approach (Finkenrath, 2011; IEA, 2004)). The procedure consists of varying the kWhel price to balance the cost of the power plant over the whole life time. The analytical formulation is now presented: NPV = =

∑

LT

∑

0

CASHFLOW

( + )k ( + )k LT REVENUE CC E S ( + )k + ( k CT ( + )k k =1 ( )

k = CT

k=

∑

REVENUES = (LCOE ⋅

ggrid

− OP Pco costs sts )

[14.1] )k

0

[14.2]

where i is the discount rate, CT is the construction time, LT is the plant operating lifetime and j is the inflation. Revenues are the incomes for net electricity sold to the grid (Elgrid) minus the operating costs (OPcosts) to run the power plant. Resulting actualized cash flow for a power plant is presented in Fig. 14.1, showing that from the beginning of plant construction to year zero, the CASHFLOW is negative because of construction costs (CC), and from year 1 to plant life time CASHFLOW becomes positive because of the revenues for electricity sale minus operating costs. The CASHFLOW trend depends on operating hours increase from year 1 to year 3, and then actualization with discount rate. The LCOE takes into account total capital cost of the plant and operating costs as fixed (i.e., labor) and variable (i.e., fuel and consumables), but it does not take into account any CO2 emission price (some studies include it (OECD, 2011)). However, because of uncertainties in several assumptions, a sensitivity analysis of main parameters is suggested, in order to outline the impact of each figure on the final results. Examples of main economic assumptions for LCOE calculations are reported in Table 14.1. The assumed operating hours per annum are typical for early 2000, where large scale fossil fuel power plants worked predominantly at base load, while hydro-power and small scale plants worked at peak-load. Recently, because of renewable energy diffusion, this value has significantly decreased.

© Woodhead Publishing Limited, 2013

Economic analysis of systems CT

531

LT

40 Relative variation (%)

20 0 −20 −40 Cash flow −60 NPV −80 −100

14.1 Characteristic actualized cash flow and NPV for power plants.

Table 14.1 Main economic assumptions Discount rate (%) First year operating hours (h) Second year operating hours (h) Rest of lifetime operating hours (h) Combined cycle operating lifetime (years) Pulverized coal operating lifetime (years) Construction time natural gas plants (years) Construction time coal plants (years)

8 3500 5700 7500 25 40 3 4

However, it is difficult to predict what the future will be and if power plant with CO2 capture will be assimilated to green-energy sources or not. Plants with CO2 capture have limited emissions and reduced impact on the environment; for this reason, they are closer to renewable energy rather than conventional fossil fuel power stations. However, at the present time, no plant has ever been built and there is no directive and/or law on this matter. Membrane reactors are usually integrated in the power plant for reducing CO2 emissions and, eventually, producing pure hydrogen. In both cases, investment costs and energy penalties are higher than for conventional cases that do not implement any separation technology, leading to higher electricity cost. Sometimes, LCOE for plant with CO2 capture does not take into account CO2 transportation and storage costs, because they are outside the plant boundaries. However, a rigorous analysis should include them, but also recognize that these costs are very location and case specific. This paper has assumed these costs are in the range of 1−4 $/tco2 for transport (variation depends on distance between power plant and storage site) and storage, which accounts for 6−13 $/tco2 (GCCSI, 2011).

© Woodhead Publishing Limited, 2013

532

Handbook of membrane reactors

In the case of CO2 capture, penalties related to a reduction of CO2 emissions are expressed by the cost of CO2 avoided, that represent the economic penalties to avoid emission of one kg of CO2. The cost of CO2 avoided is defined as: Cost of CO2 avoided =

(LCOE)CO2 cap − (LCOE)ref (CO2 kWh 1 )ref

(CO2 kWh W −1 )CO2 cap

[14.3]

where CO2 kWh−1 are the specific emissions, ref is the same type of power plant without carbon capture and CO2cap is the plant with carbon capture. The cost of CO2 avoided parameter from an economic point of view is the CO2 emission price that equalizes the LCOE of a plant without capture to the plant with capture. In the case of a hydrogen production plant, the cost of the hydrogen production is calculated setting the NPV of the power plant to zero, varying the cost of hydrogen produced similarly to the LCOE calculation previously presented. In the case of simultaneous electricity and hydrogen production, it is not straightforward to assess the economic values of the two outcomes. Two options are available: the hydrogen is assumed as revenue, with a price equal to a reference plant dedicated solely to hydrogen production, and the LCOE is calculated to balance investment costs as previously discussed; or the second option, which consists of assuming the electricity sold to the grid as revenue evaluated with the reference plant, and calculating the cost of hydrogen to achieve a zero NPV. In this chapter, the latter option will be assumed. It must be outlined that in the case of simultaneous hydrogen and electricity production with CO2 capture, the reference plant must include CO2 capture, too.

14.3

Calculation of construction and operating costs

The LCOE is the production price of electricity, which must balance the CC of the power plant as well as operating costs during the plant lifetime, taking into account the required return and time value of money. This section describes a possible methodology to assess both construction and operating costs. As for the calculation of LCOE, the procedure adopted is that proposed by EBTF (EBTF, 2011), outlining that several other methodologies have been proposed in literature.

14.3.1

Construction costs

Construction costs (CC) are indicated as total plant costs and are the sum of the engineering and procurement costs (EPC) and contingency and owner’s costs (C&O).

© Woodhead Publishing Limited, 2013

Economic analysis of systems

533

EPC can be calculated with two different approaches, usually named Bottom-Up (BUA) and Top-Down (TDA). The adoption of one methodology compared to the other depends on the quality of the source data available and will be appropriate for the stage of the project. The two different approaches will be clear after a brief description of each. 1. The BUA consist of breaking down the power plant into basic components or equipment and adding installation and indirect costs. A more detailed description, as described by EBTF, is reported here: (a) Component costs/equipment costs – Estimation of capital costs for each main basic equipment module, by a step-count exponential costing method, using the dominant or a combination of parameters derived from mass and energy balance computations, combined with cost data obtained from equipment suppliers and/or other available data. The total equipment cost (TEC) is the sum of all module costs in the plant. (b) Installation costs – The basic module costs are supplemented by estimations of additional expenses to integrate the individual modules into the entire plant, such as costs for piping/valves, civil works, instrumentation, electrical installations, insulation, painting, steel structures, erections and outside battery limits (OSBL). A rigorous methodology usually implies the adoption of dedicated coefficients for each item and plant component. However, this approach would require many arbitrary assumptions and could give misleading results when applied to different, unconventional power plants. For simplicity, a constant coefficient for each installation cost item can be adopted. (c) Total direct plant cost (TDPC) – The direct cost is the sum of the module/equipment costs and the installation costs. (d) Indirect costs – The indirect expenditures are fixed at 14% of the TDPC for all the technology options and include the costs of the yard improvement, service facilities and engineering costs as well as the building and sundries. A breakdown of possible indirect costs value is given in Table 14.2. (e) EPC – The EPC is the sum of TDPC and indirect costs. 2. The Top-Down Approach (TDA) is based on direct estimation of EPC from the equipment supplier assessment of the entire power plant. The direct estimation from an equipment supplier is usually possible only when plant design is finalized; it is not indicated during preliminary studies. When total plant cost (TPC) is presented, the level of learning and development of the technology being assessed should always be outlined. Usually, it

© Woodhead Publishing Limited, 2013

534

Handbook of membrane reactors Table 14.2 Typical breakdown of indirect costs Percentage of TDPC Yard improvement Service facilities Engineering/consultancy cost Building Miscellaneous

1.5 2 4.5 4 2

Table 14.3 Total capital requirement calculation template BUA

TDA

Plant component XXXXXX YYYYYY ZZZZZZ

A B C

– – –

TEC

A+B+C+…

–

Direct costs as percentage of the TEC Piping/valves, civil works, instrumentation, steel-structure, erection, etc.

XX%

XX% TEC

–

Total installation costs (TIC)*

80%

80% TEC

–

TEC + TIC

–

TDPC Indirect costs (IC) EPC Contingencies and owner’s costs (C&OC)

14% 15%

14% TDPC – TDCP + IC EPC 15% EPC

TPC

EPC + C&OC

*Total installation costs depends on type of plant and its complexity; 80% is a correct figure for power plants with CO2 capture or coal based plant as Integrated Gasification Combined Cycles; for Natural Gas Combined Cycles, this coefficient decreases to 68%.

is referred to as first-of-a-kind (FOAK), an innovative system which involves great uncertainty. Experience gained from the construction and operation of these plants will generate learning with cost reductions (nth-of-a-kind, NOAK). In this work, a diffuse technology for all plants components is assumed. A summary of TPC calculation with the two different methodologies is summarized in Table 14.3. Higher costs for FOAK technology are usually taken into account as contingencies, assuming a value in the range of 30–50%, which is three to five times higher than the NOAK figure. A rule of thumb to select the more correct methodology is the following: TDA is indicated for cases where no significant modifications to the power

© Woodhead Publishing Limited, 2013

Economic analysis of systems

535

section are required leading to constant EPC in this part (e.g., end-of-pipe technologies such as post-combustion CO2 capture, where only limited modification to the steam turbine are required), while BUA is suggested when the power plant lay-out is significantly modified from the base case (e.g., precombustion decarbonization technologies, or hydrogen production plant) requiring a detailed list of equipment costs such as that summarized in Table 14.4. For each component/subsystem a scaling parameter is selected and the actual erected cost C is derived from the cost C0 of a reference component of size S0 by the relationship: C = n C0[S/(n S0)]f

[14.4]

where S is the actual size and f is the scale factor. The coefficient n refers to the number of components for the base case. When the innovative components are not available on market, but will need to be integrated in the power plant (e.g., membrane reactors), a methodology similar to BUA must be adopted to perform cost assessment of the innovative component. Here follows an example of membrane reactor cost assessment.. The design of a membrane reactor, where hydrogen separation and water gas shift reaction occur, is assumed to be based upon a shell-and-tube heat exchanger, and for this reason the starting point is the price of a commercial heat exchanger made from stainless steel (e.g., AISI 316L) that keeps a high tensile strength even at high temperature (400–500°C is a typical operating condition) and oxidative conditions as in the feed stream. It must be outlined that the driving force for hydrogen permeation is its partial pressure difference from feed to permeate side, so total pressure in the shell is in the range of 4–7 MPa. The cost of the reactor shell is then calculated by applying correction factors for materials, pressure and additional pipe manifolding. The size of the reactor is selected in order to have 2000 m2 of membrane surface area (membrane tube length is expected to be in the range of 3–5 m, limited by support manufacturing). If a larger membrane surface area is required, several reactors can be placed in parallel, with advantages from a reliability point of view, but increasing piping connections and indirect costs. Additional costs for membranes, sweep tubes and catalyst are then added, to provide the final figure for membrane module costs. As far as membrane is concerned, considering that no membranes for hydrogen separation are produced, an estimation based on the manufacturing process, production time and materials is made, leading to a cost of €1500 m2 assuming a thin layer of Pd–Ag (3–5 µm) on porous support (Dijkstra et al., 2011; Manzolini et al., 2006a, 2006b, 2011; Pex et al., 2004). Other studies (Dolan et al., 2010) indicate a higher price but for a ten times thicker Pd–Ag layer. A summary of the methodology adopted is shown in Table 14.5.

© Woodhead Publishing Limited, 2013

Table 14.4 Typical equipment costs for main components

© Woodhead Publishing Limited, 2013

Plant component

Scaling parameter

Reference erected cost C0 (M€)

Reference size (S0)

Scale factor (f)

n

GT, generator and auxiliaries (DOE-NETL, 2007; Gas Turbine World, 2010) HRSG, ducting and stack (DOE-NETL, 2007; Gas Turbine World, 2010) Steam turbine, generator and auxiliaries (DOE-NETL, 2007; Gas Turbine World, 2010) Cooling water system and BOP (DOE-NETL, 2007) Amine scrubbing CO2 separation system (CAESAR, 2010) CO2 compressor and condenser (CAESAR, 2010) Desulfurization process (CACHET, 2008) Auto-thermal reformer (Manzolini et al., 2013) Fired tubular reformer (Manzolini et al., 2006a, 2006b) Waste heat boiler Water gas shift: high and low temperature shift (Martelli et al., 2008) High temperature shift (Manzolini et al., 2013) Air compressor/air blower (CACHET, 2008)

GTNet Power (MW)

49.4

272.12

0.3

2

U*S (MW/K)

32.6

12.9

0.67

2

STGross Power (MW)

33.7

200.0

0.67

1

Q_rejected (MW) CO2 captured (kg/s)

49.6 29.0

470.0 38.4

0.67 0.8a

1 2

Compressor power (MW) Thermal input LHV (MW) Thermal input LHV (MW) Thermal input LHV (MW)

8.9 0.66 37.29 37.66

13.0 413.82 1665.9 360.75

0.67 0.67 0.67 0.75

1 2 1 3

Heat transferred (MW) Thermal power input LHV

1.23 7.44

493.0 815.2

1.0 0.67

2 2

Thermal power input LHV Compressor power (MW)

2.56 14.77

827.6 47.61

0.67 0.67

2 1

a

The adoption of two GT and two HRSG requires two separated carbon capture systems, one for each HRSG.

Economic analysis of systems

537

Table 14.5 Equipment cost calculation for membrane reactor (single module of 2000 m2 membrane surface area) Reactor shell Catalyst Insulation Hydrogen membrane Sweep tubes Bought out cost Manifolding, valves, etc. Contingency costs Equipment cost

C0 C1 C2 C3 C4 C5 C6 C7 C8

– 17200 (€/m3) 1000 (€/m2) 1500 (€/m2) 150 (€/m2) C0 + C1 + C2+ C3+ C4 C0 × 100% (C5 + C6) 20% C 5 + C6 + C7

1.67 M€ 0.07 M€ 0.07 M€ 3.00 M€ 0.20 M€ 5.01 M€ 5.01 M€ 2.00 M€ 12.02 M€

Membrane reformer and oxygen membrane reactor costs assessment are similar to the one just presented, taking into account different working temperature and pressures. These aspects have a significant impact on reactors costs, because a higher working temperature of 200–300°C can require nickel-based material instead of a more conventional stainless steel reactor with a significant increase in costs. Moreover, the design of the reactor becomes more complex because heat must be supplied to support the reaction. An example of this solution was discussed in Manzolini et al. (2006a, 2006b, 2011).

14.3.2

Operating costs

Operating costs are daily expenses incurred to run a business, and are usually divided into fixed and variable costs. Fixed costs are constant and do not depend directly on the quantity of production, as variable costs do. Fixed costs in a power plant include personnel, administration and insurance, and part of the maintenance costs. While personnel costs depend on country and number of employees (figures proposed in Table 14.6 are an average among EU countries, but can change significantly if other nations such as China or Australia are considered), insurance and maintenance costs are usually assumed to be a percentage of the TPC. Within the variable costs category, the most significant figures are fuel, cooling and process water and consumables such as catalyst, membranes, solvent, etc. Catalyst and membranes are included among consumables because they have a lifetime in the range of three to five years, thus several replacements are required during plant lifetime. A template of operating costs is summarized in Table 14.6. As for CCs, considering uncertainties and price fluctuation, a sensitivity analysis is suggested. For example, a fuel cost variation should range from −50 to +50%.

© Woodhead Publishing Limited, 2013

538

Handbook of membrane reactors

Table 14.6 O&M and consumable costs Cost item

Units

Value

Fuel Natural gas costs

€/GJLHV

6

PAI (Personnel, administration and insurance) Labor costs for reference cases w/o CO2 capture Labor costs for plants with CO2 capture Insurance

M€ M€ % of TPC

6 9 2

Maintenance costs

% of TPC

2.5

Consumables Evaporative tower blow-off Cooling water make-up costs HRSG water blow-off Process water costs MEA make-up, MEA make-up costs Catalyst lifetime Catalyst costs Membrane lifetime Membrane substitution

% of evaporated water €/m3 % of steam produced €/m3 kgMEA/tCO2 €/tMEA years €/m3 years €/m2

100 0.35 1 1 1.5 1042 5 17200 5 3000

Sources: EBTF (2011) and Manzolini et al. (2013).

14.4

Procedure application

This section presents an application of the methodology discussed above with some preliminary results about membrane reactor for power and/or hydrogen production with CO2 capture. When a comparison has to be carried out, the first step consists of defining base cases to benchmark the proposed solutions. Given membrane reactors are of interest for power production with reduced CO2 emissions, the reference cases considered will include CO2 capture. Natural gas feed-stock will be investigated because of known membrane challenges applying to sour syngas derived from heavier fuels. Natural Gas Combined Cycle (NGCC) will be chosen as the reference plant. The state-of-the-art of CO2 capture in natural gas application is post-combustion amine scrubbing, so this technology will be considered as the reference case. However, the proposed methodology is general and can be applied also to other plants as coal based. Detailed energy balances of these two cases can be found in Franco et al. (2010), GHG IA (2009), Manzolini et al. (2011) and NETL (2010). The membrane cases will be based on the same gas turbine in terms of size and specific costs as well. In terms of the hydrogen production plant, the state-of-the-art consists of steam reforming which covers more than 90% of the hydrogen production worldwide.

© Woodhead Publishing Limited, 2013

Economic analysis of systems

14.4.1

539

Reference cases

As reference NGCC, a large scale ‘F class’ 50 Hz gas turbine (GT) with a net power output in the range of 260–300 MW is examined. NGCC lay-out is based on two GTs, each equipped with a heat recovery steam generator (HRSG) and a single steam turbine. This type of arrangement (2 + 1) is quite popular among utilities, adding operational flexibility as required by a competitive electricity market. Of the BUA and the TDA options, the former is selected in order to apply the same methodology to reference cases and innovative membrane applications; in this way, the results will be consistent. Specific equipment costs for the NGCC reference plant (see Table 14.4) are taken as a typical F class turbine described in the Gas Turbine World Handbook (Gas Turbine World, 2010). For the CO2 capture process, there are no commercial costs available because no full-scale plant has ever been built. For this reason, the equipment costs of the amine scrubbing section are determined from a BUA, as reported for the membrane case. As already stated, the assumed costs are estimated for a NOAK plant. The results are summarized in Table 14.7. The BUA allows allocation of costs for each component; for example, in the NGCC case, the GT accounts for 42% of total costs, while the other components account for about 15% each. In the capture case, the share changes because of the additional CO2 capture and compression section. Calculated specific investment costs for the reference case with CO2 capture is about 50% higher than for the case without capture, because of additional TPC and lower efficiency, accounting for 30% and 20%, respectively. GT cost is consistent across the two cases, as it is not affected by the CO2 capture, while the steam turbine power output reduces, due to the large steam extraction for solvent regeneration, leading to decrease in steam turbine equipment costs. In this case a dedicated design for the steam turbine in the CO2 capture application is considered. The resulting LCOE for the NGCC without CO2 capture is €54.10 /MWh. As shown in Table 14.8, most of the LCOE depends on fuel costs, while the sum of fixed and variable costs accounts for less than 10%. In the CO2 capture case, the LCOE increases by €15 /MWh as a consequence of the higher investment costs and lower efficiency, which leads to higher specific fuel consumption. The calculated cost of CO2 avoided is 47.5€/tCO2 aligned with costs presented in other studies (Finkenrath, 2011; GHG IA, 2011; NETL, 2010). An economic assessment of the reference cases for hydrogen production is now presented, and are summarized in Table 14.9 and Table 14.10. A detailed discussion of heat and mass balances can be found in Consonni (2005) and Forster Wheeler (1996). The hydrogen cost for the reference case is about 11€/GJLHV or 1.4€/kgH2, assuming an electricity price to the

© Woodhead Publishing Limited, 2013

540

Handbook of membrane reactors Table 14.7 TPCs for power production reference cases NGCC

NGCC with CO2 capture

Performances Net power output (MW) Thermal power inputLHV (MW) Net electric efficiency (%LHV) Net electric efficiency (%HHV) CO2 avoided (%)

829.9 1422.6 58.34 52.70 –

709.9 1422.6 49.90 45.08 89.7

Component cost (M€) GT HRSG Steam turbine Cooling tower, feedwater CO2 capture section CO2 compressor BOP

98.8 45.7 43.2 49.4 – – 0.4

98.8 44.8 35.1 54.9 56.7 14.4 2.8

TEC (M€) TIC (M€) TDPC (M€) Indirect cost (M€) EPC (M€) C&OC (M€) TPC (M€)

237.5 161.5 399.0 55.9 454.9 68.2 523.1

307.4 217.8 525.2 73.5 598.7 89.8 688.5

Net power output (MW) Specific costs (€/kW)

829.9 630.4

709.9 969.9

Sources: EBTF (2011) and Manzolini et al. (2013).

Table 14.8 LCOE and cost of CO2 avoided for reference cases

Investment cost (€/MWh) Fixed costs (€/MWh) Variable costs (€/MWh) Fuel costs (€/MWh) LCOE (€/MWh) Cost of CO2 avoided (€/tCO2)

NGCC

NGCC with CO2 capture

9.55 3.85 0.62 40.11 54.10 N/A

15.59 5.24 1.38 46.89 69.10 47.5

grid equal to €54.10 /MWh for the no-capture case. Considering the limited amount of power produced, this assumption has a small impact on final costs. The calculated cost of H2 production is consistent with other studies in literature (Hua et al., 2011; IEA Energy Technology Essential, 2011; NETL, 2011), deeming the higher NG price, whose effect for most of the cost are 75% and 60% for no-capture and with capture case, respectively.

© Woodhead Publishing Limited, 2013

Economic analysis of systems

541

Table 14.9 Energy balances and TPCs for hydrogen production cases H2 plant

H2 plant with CO2 capture

Performances Net power output (MW) Hydrogen produced (MWLHV) Thermal power input (MWLHV) Hydrogen production efficiency (%LHV)

41.28 904.91 1195.18 75.71

1.17 933.04 1195.18 78.07

Component cost (M€) Reforming + water gas shift section Steam turbine CO2 separation and compression H2 compressor Heat exchangers BOP

142.2 15.5 – 14.5 31.5 20.0

142.2 12.2 157.2 9.7 38.5 19.0

TEC (M€) Total installation costs (M€) TDPCs (M€) Indirect cost (M€) Engineering, procurement and construction (M€) Contingencies and owner’s cost (M€) TPC (M€)

223.7 169.1 392.7 55.0 447.7

378.8 292.8 671.6 94.0 765.6

67.2 514.9

114.8 880.4

Specific costs (€/kWH2)

569.2

944.0

Source: Consonni and Viganò (2005).

Table 14.10 H2 production costs and cost of CO2 avoided for reference cases H2 plant Investment cost (€/MWh) Fixed costs (€/MWh) Variable costs (€/MWh) Fuel costs (€/MWh) Electricity to the grid (€/MWh) H2 production costs (€/MWh) H2 production costs (€/kg) Cost of CO2 avoided (€/tCO2)

H2 plant with CO2 capture

8.6 3.5 0.3 30.9 2.5 40.9 1.36 N/A

14.1 5.3 0.4 30.0 0.1 49.8 1.66 45.1

Looking at investment costs (see Table 14.9), reforming and water gas shift section accounts for about 60% in the case without CO2 capture, while in the CO2-capture case, the share decreases significantly because of the high cost of the CO2 separation and compression section. The cost of hydrogen production with CO2-capture increases by 0.3€/kgH2, or ~25%, and the cost of CO2 avoided is similar to the power production case and equal to 45.1€/tCO2. In this case, the electricity price was assumed equal to €69.10 /MWh.

© Woodhead Publishing Limited, 2013

542

14.4.2

Handbook of membrane reactors

Membrane reactor case studies

This section presents the application of the proposed methodology for economic assessment with two different membrane cases taken from the literature (Atsonios et al., 2012; Manzolini, 2009): the first is about power production with CO2-capture and the second coproduction of hydrogen and electricity. For both cases, a membrane reactor integrated with water gas shift reaction is considered: Pd membranes applied to reforming reactor and oxygen transport are considered less mature technologies. The results are indicative of the potential of this technology in CO2 capture and hydrogen-cost reduction. The membrane surface area has been determined using a 1-D simplified model assuming a permeance of 7.21 × 10−5 mol/s/m2/Pa0.5 at 450°C (Pex et al., 2004). Being an application of the methodology without any further purpose, the adoption of a simplified 1-D model is considered satisfactory. CO2 capture Integration of membrane reactors for electricity production has been widely investigated during the last ten years. One of the most recent projects is CACHET-II (CACHET-II website, 2011), which is an FP7 project financed by the EU. In this case, membrane reactors are applied downstream to an oxygen-blown reforming section. The proposed lay-out taken from Atsonios et al. (2012) is shown in Fig. 14.2. This solution allows keeping the highest hydrogen partial pressure inside the membrane reactor, reducing surface area with economic benefits. The membrane reactor requires a surface area of 15 400 m2 at 400°C; this temperature was selected in order to guarantee a high reliability and permeability. Net electric efficiency calculated for membrane reactor case is 49.7%, which is close to the post-combustion CO2 capture reference efficiency, but it avoids 100% of CO2 stream rather than the 90% in the reference case. Looking at TPC for membrane case (Table 14.11), results show that with the assumed permeance and membrane equipment cost, the membrane reactor cost is about €100 M, which corresponds to 20% of the TEC. An improvement of membrane performance and/or a cost reduction are crucial to reduce final investment costs. Moreover, in order to limit the membrane surface area, an oxy-blown auto-thermal reformer was adopted, adding €40 M for the air separation unit. The resulting specific investment cost is higher than post-combustion capture via amine scrubbing (see Table 14.7). The calculated LCOE (see Table 14.12) is higher than the reference cases previously presented. The more significant difference is in the variable cost, which increases by 125% due to the cost of membrane replacement. The fuel costs are quite similar because of the similar efficiency. The calculated

© Woodhead Publishing Limited, 2013

Economic analysis of systems

543

To CO2 storage

Saturator

Reforming

NG Oxygen

ASU

Air Membrane

HTS Nitrogen

Air

14.2 Possible integration of membrane reactors for power production with CO2 capture.

Table 14.11 TPCs for membrane case for power production Performances Membrane surface area (m2) Net power output (MW) Thermal power inputLHV (MW) Net electric efficiency (%LHV) Net electric efficiency (%HHV) CO2 avoided (%)

15400 886.4 1784.8 49.66 44.86 100.0

Component cost (M€) Reforming and water gas shift section Membrane reactors Oxygen production and compression GT Heat recovery steam cycle Cooling tower, feedwater CO2 compressor Balance of plant

37.0 92.6 39.6 106.3 93.0 55.1 4.7 30.1

TEC (M€) TPC (M€)

458.8 905.0

Net power output (MW) Specific costs (€/kW)

1040.7 1174.1

© Woodhead Publishing Limited, 2013

544

Handbook of membrane reactors Table 14.12 LCOE and cost of CO2 avoided for membrane reactor case

Levelized cost of electricity ( /MWh)

Investment cost (€/MWh) Fixed costs (€/MWh) Variable costs (€/MWh) Fuel costs (€/MWh) LCOE (€/MWh) Cost of CO2 avoided (€/tCO2)

80 75

16.8 6.1 2.5 47.1 72.5 52.4

Membrane case (Table 14.12) Reactor cost +50% −50%

70

Ref. NGCC with CO2 capture

65 60

NGCC 55 50 0

10

20

30

40

50

60

70

Carbon tax ( /tCO2)

14.3 Sensitivity analysis on LCOE varying carbon tax and membrane reactor cost.

cost of CO2 avoided is also higher, which means that higher CO2 avoidance cannot balance cost penalties. A sensitivity analysis on the LCOE varying membrane reactor cost (±50%) and carbon tax is presented in Fig. 14.3. Membrane reactor costs can be reduced either by increasing membrane permeance and/or reducing membrane specific costs. The results show that reactor costs have a significant impact on LCOE and cost of CO2 avoided (that is the intersection between the line of the NGCC case and membrane cases) which vary from 50 to 70€/tCO2. Hydrogen production Membrane reactors can also be applied to electricity and hydrogen coproduction, as in the solution presented in Fig. 14.4 (Manzolini and Viganò, 2009). In this case, two reactors in series are integrated, wherein the first separates

© Woodhead Publishing Limited, 2013

Economic analysis of systems Process steam

545

Stack H3

1st H2

2nd

Prereformer

H6 H7

De-hydratation

NG saturator

CO2@ 150bar

HRSC Deareator

WGS HSMR

H4 H5

CO2 compr.

WGS HSMR

HT WGSR

Hydrogenation and sulfur absorption

H20

K.O. drum H19

H18

K.O. drum

ATR H17

H8 O2 IC compr.

ASU

NG compr.

H16

H9

Vent

N2 N2 IC compr.

Steam turbine

H11

H2 IC compr.

H12 H13 H14 EVA SH ECO

H10

Air

NG

H2 @ 60bar

N2 saturator

Air

Gas turbine

14.4 Example of plant for electricity and hydrogen coproduction with membrane reactors (Manzolini and Viganò, 2009). Reproduced with permission from Elsevier.

pure hydrogen, while in the second the permeated hydrogen is diluted with nitrogen for the GT combustor. The adoption of two reactors in series allows separating pure hydrogen when the partial pressure at the feed side is higher, thus reducing membrane surface area (balanced by hydrogen compressor work), or keeping a higher hydrogen back-pressure with penalties on membrane surface area, but reducing auxiliaries consumptions. Meanwhile, the adoption of nitrogen as the sweep gas in the second reactor allows keeping a high driving force inside the membrane, pushing the hydrogen conversion and permeation as well as diluting the fuel before the combustor (i.e., combustor cannot accept pure hydrogen as fuel). Other lay-outs, such as for example the adoption of a single reactor with steam as the sweep gas to produce hydrogen both for the pipelines and the GT, might be considered (this solution is detrimental from thermodynamic point of view, but can reduce investment costs). However, other options will not be discussed, since the aim of this chapter is the discussion of a general methodology for economic assessment, and not the presentation of an optimal solution. The effect of the amount of hydrogen produced is investigated in order to evaluate its impact on power plant performance and hydrogen production

© Woodhead Publishing Limited, 2013

546

Handbook of membrane reactors

Table 14.13 TPCs for hydrogen production cases with membrane reactor HRF for membrane reactor 45%

50%

55%

Performances Membrane surface area (m2) Hydrogen produced (MW) Net power output (MW) Thermal power input (MW) Hydrogen production efficiency (%) Net electric efficiency (%) CO2 avoided (%)

8593 378.9 298.3 1110.7 34.09 26.86 100

9583 462.8 299.9 1219.6 37.95 24.59 100

10976 571.6 303.6 1367.4 41.80 22.20 100

Component cost (M€) Reforming and water gas shift section Membrane reactors Oxygen production GT Heat recovery steam cycle CO2 separation and compressor H2 compressor Heat exchangers Balance of plant

24.8 51.7 18.5 44.8 60.4 3.1 7.8 44.8 45.8

26.6 57.6 19.8 44.9 59.9 3.3 9.7 48.8 47.0

29.0 66.0 21.5 45.0 60.2 3.5 11.8 54.6 48.4

TEC (M€) TPC (M€)

301.6 627.8

317.5 658.3

340.0 701.4

Specific costs (€/kWH2)

1658.0

1422.4

1227.2

costs. In particular, three different hydrogen recovery factors for the first membrane reactor are considered, hence pure hydrogen production, and results are shown in Table 14.13. The hydrogen recovery factor (HRF) indicates the amount of hydrogen permeated divided by the total amount of hydrogen in the feed side. Increasing the HRF leads to higher hydrogen production (from 380 to 570 MW), as well as membrane surface area (from 8600 to 11 000 m2): the former is predominant, since specific costs decrease significantly, as shown in Table 14.13, from 45% to 55% of HRF. For all HRF, this solution has a lower amount of hydrogen produced than reference case with methane steam reforming. As for the electricity production case, membrane reactors allow capture of the entire CO2 stream, leading to 100% CO2 avoidance, which is 10% higher than reference case. The share of membrane reactor costs over TPC is in the range of 17% to 19%, increasing with HRF as membrane surface area (8600 m2 to 10976 m2 from 45% to 55%). The lower membrane surface area, compared to only the electricity production (15 400 m2, see Table 14.12),

© Woodhead Publishing Limited, 2013

Economic analysis of systems

547

Table 14.14 Cost of hydrogen production and CO2 avoided for hydrogen production cases with membrane reactors HRF

Investment cost (€/MWh) Fixed costs (€/MWh) Variable costs (€/MWh) Fuel costs (€/MWh) Revenue for electricity to the grid (€/MWh) H2 production costs (€/MWh) H2 production costs (€/kg) Cost of CO2 avoided (€/tCO2)

45%

50%

55%

32.8 4.6 1.3 68.6 −54.4

26.8 4.7 1.4 61.7 −44.8

21.8 4.8 1.5 56.0 −36.7

52.9 1.76 45.0

49.8 1.66 33.5

47.4 1.58 24.4

arises because the adoption of two reactors allows a better optimization, and the total pressure in the feed side is higher, reducing necessary membrane surface area. For the calculation of the cost of hydrogen produced, the price of electricity sold to the grid is assumed equal to €69.1 /MWh (reference plant for power production with CO2 capture), and its impact is significant because of the relevant net power output. Hydrogen production cost reduces with the amount of hydrogen produced in the power plant, because the electricity sold to the grid cannot balance the lower hydrogen outcome (see Table 14.14). The hydrogen production costs are lower than the reference case, as also is the cost of CO2 avoided.

14.5

Conclusions

Economic assessment when applied to innovative technologies or plants is one of the most difficult jobs because of the cost uncertainties for component scale-up, industrialization of the product and integration issues, as well as price variation with time (i.e., fuel price). This chapter has presented a methodology for the economic assessment of plant for electricity and/or hydrogen production. The aim was not to determine the exact cost of electricity (LCOE), cost of CO2 avoided or hydrogen production, but to provide a clear and transparent method: it can be applied as presented or by varying some assumptions for a different location, new input on technology development, or even period. Compared to other models present in literature, the main difference arises from the transparent methodology and assumptions behind the model: all cases investigated with this proposed methodology will be consistent and comparable. The proposed methodology was applied to a power plant with CO2 capture and a plant with combined power−hydrogen production.

© Woodhead Publishing Limited, 2013

548

Handbook of membrane reactors

14.6

Acknowledgments

The authors also acknowledge all published sources that are referred to.

14.7

References

Atsonios K, Panopoulos KD, Doukelis A, Koumanakos A, Kakaras Em, 2012. Exergy analysis of a hydrogen fired combined cycle with natural gas reformingand membrane assisted shift reactors for CO2 capture. Energy Conversion and Management 60 196–203 CACHET, 2008. D1.4.11: Technical and economic performance of the process concept incorporating the advanced technology ‘Sorption Enhanced Water Gas Shift’. CACHET-II website, 2011. http://www.cachet2.eu/. visited June 2011. CAESAR, 2010. D4.7 Test cases and preliminary benchmarking results from the three projects. Consonni S, Viganò F, 2005. Decarbonized hydrogen and electricity from natural gas. International Journal of Hydrogen Energy, 30, 701–718. Dijkstra JW, Pieterse JAZ, Li H, Boon J, van Delft YC, Raju G, Peppink G, vander Brink RW, Jansen D, 2011. Development of membrane reactor technologies for power production with pre-combustion CO2 capture. Energy Procedia, 4, 715–722. DOE-NETL 2007. Cost and performance baselines for fossil energy plants, Volume 1. 2007–1281. Dolan MD, Donelson R, Dave NC, 2010. Performance and economics of a Pd-based planar WGS membrane reactor for coal gasification. International Journal of Hydrogen Energy, 35, 10994–11003. EBTF, European Benchmark Task Force, 2011. European best practice guide for assessment of CO2 capture technologies. Vol. http://www.energia.polimi.it/ news/D%204_9%20best%20practice%20guide.pdf, visited June 2011 Finkenrath M, 2011. Cost and perforamnce of carbon dioxide capture for power generation. International Energy Agency. Forster Wheeler, 1996. Decarbonisation of fossil fuels. IEA Greenhous Gas R&D programme report PH2/2. Franco F, Anantharaman R, Bolland O, Booth N, van Dorst E, Ekstrom C, Sanchez E, Macchi E, Manzolini G, Prins M, Pfeffer A, Rezvani S, Robinson L, Zahra AM, 2010 Common framework and test cases for transparent and comparable techno-economic evaluations of CO2 capture technologies – the work of the European Benchmarking Task Force, Proceedings of GHGT-10, Amsterdam (NL), Sept. 2010. Gas Turbine World Handbook, 2010, Southport 06890 USA: Pequot Publishing Inc. GCCSI, Global CCS Institute, 2011. http://www.globalccsinstitute.com/sites/default/ files/eco-assess-ccs-tech-2010–4b.pdf. Accessed June 2011. GHG IA (Greenhouse Gas Implementing Agreement), 2009. Biomass CCS study. 2009. Report Number 2009–9. Hua TQ, Ahluwalia RK, Peng JK, Kromer M, Lasher S, Me Kenney K, Law K, Sinha J, 2011. Technical assessment of compressed hydrogen storage tank systems for automotive applications. International Journal of Hydrogen Energy, 36, 3037–3049.

© Woodhead Publishing Limited, 2013

Economic analysis of systems

549

IEA (International Energy Agency), 2004. Improvement in power generation with post-combustion capture of CO2. IEA, November 2004. PH4/33. IEA, 2000. Leading options for the capture of CO2 emissions at power stations. IEA, Februray 2000. PH3/14. IEA Energy Technology Essential, 2011. http://www.iea.org/techno/essentials5.pdf. visited on June 2011. Manzolini G, Macchi E, Dijkstra JW, Jansen D, 2006a. Electricity production with low CO2-emissions through precombustion decarbonisation using membrane reactor: Proceedings of 8th GHGT International conference. Vols. Trondheim (No), June 2006. Manzolini G, Dijkstra JW, Macchi E, Jansen D, 2006b. Technical economic evaluation of a system for electricity production with CO2 capture using membrane reformer with permeate side combustion. Proceedings of ASME Turbo Expo, May 2006, Barcelona (Spain). Manzolini G, Macchi E, Binotti M, Gazzani M, 2011. Integration of SEWGS for carbon capture in Natural Gas Combined Cycle. Part B: Reference case comparison. International Journal of Greenhouse Gas Control, 5(2), 214–225. Manzolini G, Macchi E, Gazzani M, 2013. CO2 capture in natural gas combined cycle with SEWGS. Part B: Economic assessment. International Journal of Greenhouse Gas Control, 12, 502–509. Manzolini G, Viganò F, 2009. Co-production hydrogen and electricity from an auto-thermal reforming of natural gas by means of Pd-Ag. Energy Procedia, 1, 319–326. doi:10.1016/j.egypro.2009.01.044. Martelli E, Kreutz T, Consonni S, 2008. Comparison of coal IGCC with and without CO2 capture and storage: Shell gasification with standard vs. partial water quench. Elsevier, 2008. Vol. 1 Energy Procedia. NETL, 2010. Life Cycle Analysis: Natural Gas combined cycle (NGCC) power plant. DOE/NETL-403–110509. www.netl.doe.gov/energy-analyses/pubs/SCPC_ LCA_final.zip, accessed June 2011. NETL, 2011. http://www.netl.doe.gov/technologies/hydrogen_clean_fuels/systems_ studies.html. visited June 2011. Pex P, van Delft Y, Correia L, 2004, Palladium alloy membranes for energy efficient membrane reactors. Proceedings of the 8th international conference on Inorganic membranes. Cincinnati, July 18–22 : s.n., Vols. pp. 524–527. OECD (Organization for Economic co-operation and development), 2010. Projected cost of generating electricity. Paris, France : s.n.

14.8

Appendix: nomenclature

14.8.1

Abbreviations

BUA C&OC CT EPC EBTF FOAK GT

bottom-up approach contingencies and owner’s cost construction time engineering and procurement cost European Benchmarking Task Force first-of-a-kind gas turbine

© Woodhead Publishing Limited, 2013

550 HRF HRSG LCOE LT NGCC NPV NOAK TDA TEC TIC

Handbook of membrane reactors hydrogen recovery factor heat recovery steam generator levelized cost of electricity life time natural gas combined cycle net present value nth-of-a-kind top-down approach total equipment cost total installation costs

© Woodhead Publishing Limited, 2013

15 Electrochemical devices for energy: fuel cells and electrolytic cells M. CASSIR , Chimie ParisTech ENSCP, France, D. JONES, Université de Montpellier, France and A. RINGUEDÉ and V. LAIR , Chimie ParisTech ENSCP, France

DOI: 10.1533/9780857097347.3.553 Abstract: This chapter is dedicated to some significant applications of membranes in the field of energy, focusing on fuel cells and electrolytic cells. Both electrochemical devices are part of an international effort at both fundamental and demonstration levels and, in some specific cases, market entry has already begun. Membranes can be considered as separators between cathodes and anodes. As fuel cells are extremely varied, with working temperatures between 80°C and 900°C, and electrolytes from liquid to solid passing by molten salts, they are of particular interest for the research and development of new membranes. The situation is quite similar to the case of electrolysers dedicated to water electrolysis. The principal features of these devices will be outlined, with emphasis on the properties of the state-of-the-art membranes and on the present innovations in this area. Key words: membranes, fuel cells, electrochemical devices, electrolyser, water electrolysis, energy.

15.1

Introduction

The search for new energy resources and devices for energy production is becoming one of the major challenges of this century. Fossil fuels should be progressively replaced by renewable energy sources, such as solar (thermal and photovoltaic), wind, geothermal, hydrothermal, biomass, wastes, etc. Nevertheless, for a long intermediate period it will be necessary to rely on fossil fuels. The probable reserves of oil (1.62 × 1014 L) should last around 50 years, according to the predicted consumption. In the case of natural gas (1.42 × 1014 m3) and carbon (9.11 × 1011 tons), their lifetimes would be 70–100 and 200–250 years, respectively. It will be compulsory to use these fossil resources more efficiently to decrease their greenhouse effects and increase their duration. Fuel cells are ideal electrochemical devices, either 553 © Woodhead Publishing Limited, 2013

554

Handbook of membrane reactors

to efficiently transform fossil fuels (after their conversion to hydrogen) or renewable sources into electrical energy. These systems are also well adapted to use hydrogen as a fuel. Hydrogen, which can be considered as an ‘energy vector’ is surely one of the main issues in the future energy bouquet. Production of hydrogen can be realised in several ways, among which are reforming, electrolysis, photosynthesis, bacterial activity, etc. Electrolysis, whose operation is the reverse of that of a fuel cell system, is also an electrochemical process that will play an important role in the energy arena. In general, electrochemical devices, such as fuel cells, batteries and electrolysers are energy convertors. In the case of fuel cells and batteries, electrical energy and heat are produced from chemical reactions. In electrolysis, external energy is converted into electrochemical reactions. In all these systems, two electrodes are required, separated by a membrane permitting a pure ionic conduction between the electrodes, which are, in general, pure electronic conductors (or mixed ionic−electronic conductors in some specific cases). These three elements constitute a single cell. The ion-conducting membrane allows the avoidance of short-circuits between the electrodes and ensures the electrical operation of the cell. It is the heart of the cell and, in a sense, the electrolyte has a similar function in fuel cells, batteries or even electrolysers. In all these processes of energy conversion, two parameters are fundamental − the tensions between both electrodes, and the current delivered or produced, which is proportional to the mass flow according to Faraday’s law (N = q/nF; with N: the number of moles of species transformed, q: the total electric charge passed through the electrical circuit; n: the number of electrons exchanged and F = 96 485 C mol−1 is the Faraday constant). The loss of yield in these processes corresponds mainly to heat production, which can be exploited in some specific cases. In this chapter, we will not mention the field of batteries, which is becoming of growing importance and would require a specific treatment. We will focus mainly on fuels cells, which are varied and promising electrochemical devices. In the case of a fuel cell, the fuel (hydrogen, natural gas, biomass, alcohols, etc.) can be fed continuously into the system, and lifetimes of several years can be achieved, contrary to batteries which need to be replaced or recharged. Fuel cells allow converting chemical (electrochemical) energy into electrical energy, without the need for mechanical energy as in the case of turbines or gas engines; therefore, the lack of vibration decreases the noise (58 dB at 10 m) dramatically. Theoretically, electrical yields of 80–90% can be obtained, but in practice these yields vary from 45% to 60% (30% for gas engines). Moreover, fuel cells can be used as cogenerators of electricity and heat, with global yields higher than 80%. Due to their electrical

© Woodhead Publishing Limited, 2013

Electrochemical devices for energy

555

efficiency, fuel cells are more respectful of the environment, and lower carbon dioxide emissions are produced. Also, nitrogen oxide emissions are lower than the norm. Coupling fuel cells with other technologies, such as gas turbines, photovoltaic and nuclear, will be mentioned briefly: a broader view would require a whole chapter. Water electrolysis, which is the reverse of fuel cells, exploiting electrical energy to produce hydrogen and oxygen, will also be developed: it is a key technology in what would be a ‘hydrogen economy’. The fact that water electrolysis can nowadays be realised at high temperature makes it even more attractive for future applications. All the electrochemical devices that will be introduced in this chapter are constituted by a central membrane, the electrolyte, and they involve an electrochemical circuit. The role of fuel cells will be detailed because, under this name, different systems are involved with varied features and scientific/ technical aspects, for example, according to the temperature and the electrolyte, different kinds of electrochemistry can be seen: solid-state, molten salt, ionic liquids and more common aqueous solutions. Furthermore, fuel cells have reached a state of maturity and are excellent examples for understanding the behaviour of membranes in electrochemical devices. As electrolysis is constituted of similar elements to fuel cells, we will be much more synthetic with respect to this thematic. Our main objective through this chapter is to describe all the basic elements of these electrochemical devices focusing, in particular, on the role and characteristics of the electrolytic membrane. We will not limit our presentation to state-of-the art materials and systems, but will also focus on new materials and new concepts, to show the enormous potentialities of these systems together with new applications. For example, we will introduce new membranes for low-temperature fuel cells (polymer electrolyte membrane fuel cell (PEMFC)) working at higher temperatures, as well as the principle of microbial fuel cells which is attracting many researchers. We will show recent investigations to lower the temperature of high-temperature solid oxide fuel cells (SOFCs) and even to shift from oxide ion conduction to proton conduction (proton conductor fuel cells (PCFCs)). We will also describe new concepts in high-temperature fuel cells, such as the direct carbon fuel cells (DCFCs) which seem to be realising the old dream of electrochemists, to oxidise carbon electrochemically in a clean way, or new composite electrolytes combining the properties of molten salts and those of solid oxides. Membranes can be also seen in other environmental perspectives, for example, carbon capture and storage or separation and concentration of CO2. The field is too huge to embrace all the advances and possibilities, but we aim at giving an up-to-date panorama of this fertile field of research, development and commercialisation, through examples. Whatever the difficulties

© Woodhead Publishing Limited, 2013

556

Handbook of membrane reactors

in developing these environmentally friendly devices, they will have a major role in the new energy opportunities, due to their flexibility and the enormous potential of hydrogen energy.

15.2

Principles and features of fuel cells

The history of fuel cells (Grove and Phil, 1842; Justi and Winsel, 1962; Bacon, 1969; Tilak et al., 1981; Kordesch and Simader, 1996) is of particular interest, because their concept was discovered more than 170 years ago, but their development fluctuated between periods of high interest and others of complete silence. From simple curiosity deriving from batteries and electrolysis, which were already known, fuel cells attracted much attention during the carbon industrial revolution at the end of the nineteenth and the beginning of the twentieth century. However, the lack of serious knowledge of electrochemical kinetics stopped their evolution until the mid twentieth century, when the kinetics formalism was established. Rapidly, fuel cells became important through the space adventure since the 1960s. But it was only at the end of the twentieth century and the beginning of the present one that fuel cells began to achieve their maturity and progress to reaching the market. The first fuel cell was discovered in 1839 by William Robert Grove (1811– 1896), a lawyer in the United Kingdom, who succeeded in generating an electrical current by reversing water electrolysis. He used a hydrogen/oxygen cell in diluted sulfuric acid with platinised platinum as electrodes. In the same period, Christian Schönbein (1799–1868) in Switzerland made a similar discovery. In 1855, A.C. Becquerel (1788−1878) did similar experiments using platinum and carbon electrodes in molten nitrates. However, it was not until 1899 that Ludwig Mond and his co-worker Charles Langer engineered the first practical fuel cell, using coal gas as fuel and air as oxidant. The electrolyte was made of clay impregnated by dilute sulfuric acid. They gave the name ‘fuel cell’ to this electrochemical device. The theoretical base of fuel cells and the role of interfaces were given by Friedrich Wilhem Ostwald in 1893. In 1897, Walther Hermann Nernst (1864–1941) developed for the first time solid electrolyte membranes based on zirconia, which would be exploited in an SOFC system more than 70 years later. Francis Thomas Bacon (1904–1992) developed the first successful system using H2, O2, an alkaline electrolyte, and replaced platinum electrodes with Ni. The catalytic activity of this common metal was enhanced by increasing the temperature and pressure. In 1959, working for the Marshall Aerospace in the United Kingdom, he optimised his system and obtained a power of 5 kW. This key finding was the basis of the space programmes of the National Aeronautics and Space Administration (NASA)

© Woodhead Publishing Limited, 2013

Electrochemical devices for energy

557

in the early 1960s and since then fuel cells have supplied electricity and pure water in space missions. In the late 1960s, the first electric vehicles using liquid or gaseous hydrogen were developed. In the 1970s, a broader interest in fuel cells attracted industries (electricity, gas and even oil companies, car industries, etc.). Systems from a few kW to several MW were built in many countries, but the commercial maturity and real market of fuel cells is just beginning to grow in this new century, in three directions: small mobile applications, transport and stationary power applications. It should be added that combining fuel cells with other engines, such as gas turbines, etc. is also a very promising area of application.

15.2.1

Basic principles of the oxygen–hydrogen fuel cell

A good introduction to the principle of a fuel cell is given by the example of the low-temperature cell with an acidic electrolyte and working on hydrogen and oxygen (Vetter, 1967; Tanase et al., 1987; Lamy and Léger, 1994; Stevens et al., 2000). The following electrochemical reactions occur at the: Anode: H2 → 2 H+ + 2 e− Ean =

0 Ean

2 aH + RT + Ln nF aH 2

Cathode: 0.5 O2 + 2 H+ + 2 e− → H2O Ecat

[15.1]

0 Eaan n +

[15.2]

05 aO a2 + RT Ln 2 H nF aH 2O

The global chemical reaction is: H2(g) + 0.5 O2(g) ↔ H2O(l) + energy

[15.3]

The general scheme is presented in Fig. 15.1. At the standard state, T = 25°C, the free enthalpy of the previous reaction, related to the transformation of the chemical energy into electrical energy, is: ΔrG = –n.F.emf (−237 kJ/mol H2)

[15.4]

emf = Ecat – Ean = 1.23 V

[15.5]

© Woodhead Publishing Limited, 2013

558

Handbook of membrane reactors

2e−

Cathode

H2O

H+

Membrane

Anode

−

+ ½O2

H2

15.1 General scheme of an oxygen–hydrogen fuel cell with a proton membrane electrolyte.

15.2.2

Different families of fuel cells

According to their operating temperature and the nature of the electrolyte membrane, fuel cells can be classified into two large families: Low-temperature fuel cells: • • • • •

PEMFCs, comprising a proton exchange membrane, working between 60°C and 120°C depending on the pressure used. Direct methanol fuel cells (DMFCs), which are PEMFCs fed by methanol directly oxidised at the anode. Microbial fuel cells (MFCs), in general work like a PEMFC, but bacteria are responsible for the oxidation reaction. Alkaline fuel cells (AFCs) with a hydroxide conducting electrolyte, working between 70°C and 120°C. Phosphoric acid fuel cells (PAFCs), with an electrolyte containing concentrated phosphoric acid, working between 150°C and 210°C. High-temperature fuel cells:

• •

Molten carbonate fuel cells (MCFCs), with a molten carbonate eutectic as electrolyte, working between 600°C and 650°C. SOFCs, with an oxide conducting membrane.

15.2.3

Kinetics of the electrochemical reactions

It is not our purpose here to describe in detail the general kinetic equations but to give a rapid overview of essential aspects related to fuel cells. When the cell produces an electrical current with intensity I ≠ 0, it is out of equilibrium. The current is, in general, normalised with respect to the area A of the electrode and expressed as current density: j = I/A (in A cm2).

© Woodhead Publishing Limited, 2013

Electrochemical devices for energy

559

The cell potential E(j) = Ecat(j) – Ean(j) (difference between cathode and anode potentials for a current density j) is lower than emf, because electrochemical reaction rates and transport rates are finite. Activation overpotentials ηan and ηcat are necessary to produce anode and cathode reactions, respectively. In order to decrease η, good electrocatalysts are required, which is a major aspect in improving the efficiency of fuel cells. Anode and cathode overpotentials are related to the reaction rates: v=

dN I dq 1 = = . dt nF dt nF

[15.6]

where v is given in mol cm2 s−1, dN is the elementary variation of reacting moles during the time dt, and dq is the associated amount of electricity. At the anode, where the oxidation of the fuel occurs, the overpotential is ηan = Ean(j) – Ean > 0 and the corresponding anode current density is: jan = nFvan

[15.7]

At the cathode, where the oxidant reduction occurs, the overpotential is ηcat = Ecat(j) – Ecat < 0 and the corresponding current density is: jcat = nFvcat

[15.8]

The global current density is: J = nF(van−vcat)

[15.9]

The cell tension can be expressed as: Ei(j) = emf – j(Rct + Rel) – |ηan| – |ηcat|

[15.10]

where, Rel is the cell resistance (mainly due to the electrolyte) and Rct is the resistance due to charge transfer. In the case of a redox system out of equilibrium, the current density can be expressed as a function of the charge transfer overpotential ηct, according to the Butler−Volmer equation: j

(1 − )nF F ηct −α nF nF ηct ⎤ ⎡ j0 ⎢exp − exp RT RT ⎥⎦ ⎣

[15.11]

j0 is the exchange current density at equilibrium and α is the charge transfer coefficient.

© Woodhead Publishing Limited, 2013

560

Handbook of membrane reactors

In the case of a fuel cell, the situation is more complex because several phenomena occur with a relative importance depending on the fuel cell and experimental conditions: ohmic drop due to the electrolyte and charge transfer resistances, anode and cathode overpotentials due to charge transfer and/or mass transfer (essentially limited by diffusion). Therefore, anode overpotential is the sum of charge transfer and diffusion transfer (ηd) at the anode: ηan = ηct + ηd. In the case of low-temperature fuel cells (PEMFCs, AFCs and PAFCs), some simplified expressions can be obtained. • The oxygen reduction reaction is considered as very slow; in this case j0 → 0 and ηct → ∞ and the Butler−Volmer equation becomes: j0 exp

j

−α cat nF F ηct RT

[15.12]

and ⎛ −j⎞ ⎛ RT ⎞ ηct = − ⎜ Ln ⎜ ⎟ ⎝ α cat nF ⎟⎠ ⎝ j0 ⎠

[15.13]

with αcat the cathode charge transfer coefficient. • The hydrogen oxidation reaction can be considered as sufficiently rapid and in this case j0 → ∞ and ηct → 0. The Butler−Volmer equation can be linearised: j

j0

nF ηct RT

[15.14]

⎛ RT ⎞ ηct = ⎜ j F 0 ⎟⎠ ⎝ nFj

[15.15]

• If mass transfer rate is the limiting phenomenon (e.g., when the charge transfer is rapid), the corresponding diffusion overpotential becomes:

ηd =

RT Ln nF

⎛ j ⎞ 1− ⎜ ⎟ ⎝ jlim cat ⎠ ⎛ j ⎞ 1− ⎜ ⎟ ⎝ jliman ⎠

© Woodhead Publishing Limited, 2013

[15.16]

Electrochemical devices for energy

561

Voltage, V 1.23V

Activation polarisation

Theoretical potential

Ohmic loss

Concentration polarisation

Pseudo linear shape

Current density, A/cm2

15.2 Current density–potential characteristic (or polarisation curve) of a fuel cell.

• Therefore, the general equation describing fairly the current densitypotential characteristic (or polarisation curve), for such low-temperature fuel cells, is:

E( j )

f

j(Rct

j ⎛ 1− ⎜ ⎛ RT ⎞ ⎛ −i ⎞ ⎛ RT ⎞ jlim.cat Rel ) − ⎜ Ln ⎜ ⎟ − ⎜ ⎟ Ln ⎜ ⎝ α cat nF ⎟⎠ ⎝ j0 ⎠ ⎝ nF ⎠ ⎜ 1− j ⎜⎝ jlim.an

⎞ ⎟ ⎟ ⎟ ⎟⎠ [15.17]

A schematic representation of this polarisation curve is given in Fig. 15.2. The first part of the curve corresponds to activation loss, mainly the kinetics of oxygen reduction (with j0 = 10−6−10−8 A cm−2) involved; the second part is linear and due to ohmic loss, mainly the electrolyte resistance; the third part is due to mass transfer or diffusion loss: when the value of j becomes close to jlimcat or jliman, E(j) tends to zero. The optimal operating point is located in the linear part of the curve. These current-density-potential curves are very important for any type of fuel cell, because they summarise the influence of all the important parameters on the performance of a cell. Even though the equation is more complex in the case of high-temperature fuel cells, the general features of the current density vs potential characteristics are similar.

15.2.4

Fuel cell efficiency

When comparing a fuel cell to a thermal engine for the production of electrical energy, it should be noted that the conversion yield is significantly

© Woodhead Publishing Limited, 2013

562

Handbook of membrane reactors

higher in a fuel cell. In effect, the maximum theoretical efficiency of a thermal engine is given by the Carnot’s theorem:

ε thermal =

T Wr = 1− c (− H ) Th

[15.18]

ΔH is the enthalpy of the reaction; Wr is the mechanical work produced; Tc and Th define the absolute temperatures corresponding to a cold and a hot reservoir, respectively. The efficiency εthermal is the ratio of the work done by the engine to the heat drawn out of the hot reservoir. This efficiency is lower than 40% for gas turbines and lower than 30% for internal combustion engines. In the case of a fuel cell operating under reversible conditions, at constant pressure and temperature, the efficiency is:

ε rev. =

We ΔG = ( − H ) ΔH

[15.19]

with We, the electrical work. Knowing that: ΔG = ΔH − TΔS (ΔS is the isothermal variation of entropy, TΔS is the reversible isothermal heat exchanged, which represents the fraction of enthalpy that cannot be transformed into electrical energy):

ε rev. = 1 −

T S ΔH

[15.20]

However, the practical efficiency value is lower than the reversible efficiency, because of two factors: •

Faradaic efficiency, which is the ratio between the real current produced, with nexp, electrons exchanged, and the maximum current (related to n electrons exchanged in the global cell reaction) corresponding to the cell reaction:

εF = •

I I max

=

nexp n

≤1

[15.21]

Potential efficiency, which is the ratio of the cell potential under current to the thermodynamic potential (in equilibrium conditions)

εE =

E( j ) emf

≤1

[15.22]

© Woodhead Publishing Limited, 2013

Electrochemical devices for energy

563

Therefore, the practical efficiency is the product of the three mentioned efficiencies:

ε cell

ε rev. ⋅ ε F ε E

[15.23]

In practice, the electrical efficiency is between 40% and 60% depending on the fuel cell, the fuel and the operating conditions.

15.3

15.3.1

Low-temperature fuel cells: proton exchange membrane fuel cells (PEMFCs) and direct methanol fuels (DMFCs) Principle of the ion exchange membrane

This technology was created in the beginning of the 1960s by Thomas Grubb and Leonard Niedrach from General Electric in the US. The first important application was the NASA Gemini Project, but problems of contamination and oxygen leakage limited its growth. It is really since 1987 that Ballard Power System developed the Nafion membrane, which was the start of the modern emergence of PEMFCs in three fields: transport, stationary and mobile. This type of fuel cell has the highest power density and the lowest working temperature, but has the disadvantage of using expensive electrocatalysts such as platinum. The principle of PEMFCs has already been described in Section 15.2.1 when the fuel is pure hydrogen. The DMFC has the same kind of membrane as the PEMFC, but the fuel is methanol, directly oxidised at the anode side, the reduction of oxygen being similar:

15.3.2

Anode: CH3OH + H2O → CO2 + 6H+ + 6e−

[15.24]

Cathode: 1.5O2 + 6H+ + 6e− → 3H2O

[15.25]

State of the art of the components

We will focus mainly on the membrane of such fuel cells because they are at the heart of intense research activity and many applications are in progress.

15.3.3

Electrodes and catalysts

The role of PEMFC gas-diffusion electrodes is to support the electrocatalyst, generally Pt, to allow protons moving from hydrogen oxidation catalytic

© Woodhead Publishing Limited, 2013

564

Handbook of membrane reactors

sites to oxygen reduction sites and to evacuate electrons (from anode catalytic sites and redistribute them in cathode catalytic sites). The backing of both anode and cathode electrodes is a porous carbon cloth with a hydrophobic coating which includes carbon powder recovered by Pt nanometric particles, particles of polytetrafluoroethylene (PTFE) and liquid polymer electrolyte (see next section on membranes). These electrodes are heatpressed on both sides of the electrolytic membrane. The thickness of the membrane electrode assembly (MEA) is less than 1 mm. The MEA is integrated between two interconnecting plates. Platinum content varies from 0.01 to 0.5 mg cm−2. The main challenge at the level of electrodes is to minimise the Pt content and to distribute it more efficiently.

15.3.4

Membranes: state of the art and new trends

The membrane is the heart of the MEA. The principal roles of the membrane are both to separate the anode and cathode compartments, and to ensure ionic conductivity. It consists of a proton-conducting polymer structure with functional fixed groups, generally carboxylic -CO2 or sulfonate -SO3, and mobile protons when the membrane is humidified. The past decade has seen significant advances in the development of new polymer membrane materials for low-temperature fuel cells, including for direct methanol (ethanol) fuel cells and high-temperature PEMFCs (Tanase et al., 1987; Rozière and Jones, 2003; Hickner et al., 2004; Lafitte and Jannasch, 2007; Gubler et al., 2008; Li et al., 2009). Progress has been made both with perfluorosulfonic acid membrane materials, with regard to new polymer compositions and membrane processing technologies, and with non-fluorinated membrane systems that have reached, in some cases, a good level of technological maturity. For PEMFCs, particular efforts have been dedicated to satisfying the demanding set of properties required for automotive use, and the extended durability requirements of power stationary applications. This section will highlight the current state-of-the-art in this field, and point out the research directions for new membrane materials. Perfluorosulfonic acid (PFSA) membranes continue to be the industry standard for low-temperature PEMFCs due to their excellent proton conductivity, mechanical and chemical stability that is difficult to surpass. The Nafion® membrane produced by DuPont has been the most studied (Mauritz and Moore, 2004; Grot, 2008). Nafion® membranes are coded according to the polymer equivalent weight (EW) (first two digits), the membrane thickness (in mil, 1/1000 inch, corresponding to 25 µm) – third, or third and fourth digits); thus Nafion® N117 is polymer EW 1100, 7 mil thickness. In parallel with these developments, advances have been made on related perfluorinated ionomers that differ from the Nafion®-type polymer

© Woodhead Publishing Limited, 2013

Electrochemical devices for energy

565

by the absence of a pendant –CF3 group in the side chain, and from each other by the length of the perfluorovinylether side chain (Jones and Rozière, 2009). These compositional and structural differences impart specific properties. For example, the short-side chain (SSC) type Aquivion® ionomer developed by Solvay-Solexis (Arcella et al., 2003) has a significantly greater heat of fusion than the long-side chain (LSC) Nafion® type membranes, at a given polymer EW, enabling semi-crystalline character to be retained even at low EW. Aquivion® membranes have been operated at temperatures up to 110–120°C in stacks developed in the FP6 Autobrane ‘automotive fuel cell membranes’ project (Aricò et al., 2010), and they represent a viable option for automotive and small power stationary applications, although their long-term durability at higher temperatures must be increased further. The hydrophobicity of the perfluorinated polymer segments, and the hydrophilicity of sulfonic acid groups of PFSA ionomers lead to self-organisation into hydrophobic and hydrophilic domains with distinct phase-separated morphology. In the presence of water, the hydrophilic regions swell and provide channels for proton transport. Various models describing this microstructure/nanostructure have been proposed (Mauritz and Moore, 2004); however, the enduring conceptual basis is a cluster-network model of hydrated ionic aggregates of diameter ≈4 nm first proposed some 30 years ago by Gierke on the basis of X-ray scattering data (Gierke et al., 1981). The transformation of data from scattering and diffraction techniques into structural models is delicate and not always unambiguous. Microscopic investigations, simulation and modelling (Eikerling et al., 2008) are increasingly useful in providing direct visualisation of domain sizes and membrane morphologies, and in better understanding the effects of polymer structure, (Liu et al., 2010) molecular weight and EW (Wu et al., 2009), side chain length and hydration level on membrane morphology. Proton conductivity in PFSA membranes depends upon the polymer EW (number of charge carriers), the hydration number (λ, number of water molecules/sulfonic acid group), polymer structure, membrane morphology and temperature, all of which affect proton mobility. The conductivity of Nafion® NR-211 at 30°C, 50°C and 80°C, over a range of relative humidity values, is shown in Fig. 15.3. (Peron et al., 2010). The conductivity of the new generations of low EW PFSA membranes is a factor 2–3 higher, even at temperatures above 100°C and at lower relative humidity. Recent developments have tended towards the use of low EW PFSA polymer membranes of thickness only 25–30 µm. The difficulty lies in limiting the high water uptake that is accompanied by macroscopic swelling of the membrane, increased plasticity and softening, which generally occurs in membranes having the high charge carrier concentration required for high proton conductivity. Such effects are pronounced, even in state-of-the-art perfluorosulfonic acid membranes, and a current challenge lies in the development of means to

© Woodhead Publishing Limited, 2013

566

Handbook of membrane reactors

Conductivity (S cm−1)

1.E+00

1.E−01

1.E−02

1.E−03 0

20

40

60

80

100

Relative humidity (%)

15.3 In-plane conductivity of NR-211 membranes at 30°C (diamonds), 50°C (squares) and 80°C (triangles) (Peron et al., 2010).

ensure the mechanical stability of thin, low EW PFSA membranes. Nafion® XL is a thin, reinforced, dispersion cast membrane with improved tensile strength. An approach followed at 3 M is to modify the PFSA side chains such that they carry more than one acid site; since the crystallinity and morphological properties arise essentially from the ratio of non-substituted TFE to functionalised TFE of the backbone polymer repeat unit, multi-acid side chain ionomer membranes have the potential to demonstrate the mechanical properties of a higher EW polymer (characteristic of a single acid site per side chain), and the proton-conducting properties of a lower EW material (conferred by the presence of multi-acid sites per side chain) (Yandrasits, 2011). Other strategies are also being developed that limit membrane swelling, as chemical cross-linking. New perfluoropolymers with cross-linkable side chains have also been developed by copolymerisation of novel multifunctional monomers (Wainright et al., 2003). The presence of an inorganic phase is also effective in enhancing interaction between components, in limiting dimensional change and in improving fuel cell performance under high temperature and low relative humidity conditions (Tchicaya-Bouckary et al., 2002; Alberti and Casciola, 2003; Baglio et al., 2005; Jones and Rozière, 2008). Various approaches to macrocomposite dimensionally stabilised PFSA membranes have been developed and integrated into benchmark MEAs. For example, Gore-Select membranes (W. L. Gore and Associates) comprise a Teflon-like component providing mechanical strength and dimensional

© Woodhead Publishing Limited, 2013

Electrochemical devices for energy

567

stability, and a PFSA component for proton conductivity (Bahar et al., 1999). Highly porous reinforcements for PFSA have also been produced by electrospinning of fluorinated or hydrocarbon polymers (Lee et al., 2010; Subianto et al., 2012). Despite the very high chemical stability generally associated with perfluorinated polymers, the past ten years have witnessed not only rapid agreement that chemical degradation of PFSA membranes does indeed occur in the fuel cell environment (Peron et al., 2008), but also implementation of mitigation strategies. This chemical degradation is caused primarily by the attack of peroxide species on polymer end groups, being mainly -COOH, COF and CF2H. These act as points of lower chemical and thermal stability, and furthermore non-perfluorinated end groups are preferential sites for attack by peroxide radicals formed under fuel cell operating conditions (Curtin et al., 2004). Different approaches are under evaluation, including chemical stabilisation by conversion of the above end groups by fluorination or by the addition inside the membrane structure of an oxide (Coms et al., 2008; Trogadas et al., 2011a) or metal nanoparticle (Trogadas et al., 2011b) scavenger to minimise the presence of free radical species; chemically stabilised versions are available of the commercial PFSAs. The design and development of a broad range of polymers and membranes as alternatives to conventional and advanced PFSA types has progressed immensely over the past 10–15 years. These polymers generally have less demanding preparation routes, and are potentially less costly. Moreover, and most importantly, they can extend the range of fuel cell operation conditions, for example to significantly higher temperatures. New materials have primarily explored sulfonated aromatic or heterocyclic polymers, in particular polysulfones (Kim et al., 2004), poly(aryl ether ketone)s (Bauer et al., 2000; Liu et al., 2010), polybenzimidazoles (Jones and Rozière, 2001) and polyimides (Marestin et al., 2008), in which the ionic groups are covalently bound either directly to the polymer backbone or via a spacer. Many routes have been developed to sulfonic acid functionalised polyaromatics, including direct sulfonation of a polymer backbone, grafting of a sulfonated functional group on to a polymer main chain, graft polymerisation followed by sulfonation of the graft component and copolymerisation of functionalised monomers to random, alternating or block copolymers. A degree of enhancement of phase-separated morphology can be achieved by spatial separation of the hydrophilic sulfonic acid groups from the hydrophobic polymer main chain by locating the sulfonic acid groups on side chains grafted onto the polymer backbone and in multiblock copolymers of controlled segment length. Such observations lead to ‘design rules’ for proton electrolyte membranes that are only recently starting to take convincing form. Ionomers prepared by direct sulfonation generally show random substitution along the polymer chain, although for polymers with different

© Woodhead Publishing Limited, 2013

568

Handbook of membrane reactors

types of non-equivalent substitution site, kinetic or thermodynamic activation of a particular site has also been reported, and different sulfonating agents may activate different sites to sulfonation (Ng et al., 2010). In general, however, development of a polymer with a defined sulfonation pattern is not possible and, despite the increased interest in more organised polymer systems, statistical copolymers are still mainly used by industry in PEMFC applications. Increasing the degree of phase separation, and greater phase continuity, require more sophisticated approaches to tailoring than can be attained by simple polymer modification. Desirable polymer architectures include multiblock copolymers of controlled segment length, as well as grafted or branched sulfonated hydrocarbon copolymers, the properties of which convincingly demonstrate the impact of polymer nanostructure on membrane properties and have enabled identification of structure−morphology−property relationships (Gross et al., 2009; Peckham and Holdcroft, 2011). This is nicely illustrated by the results collected in Fig. 15.4 (Park et al., 2011), which correlates proton conductivity at ca. 25°C with water uptake for various sulfonated hydrocarbon-based membranes, and which shows that sulfonated hydrocarbon multiblock proton exchange membranes have comparatively the highest level of proton conductivity (> 0.1–0.3 S cm−1) and are attractive candidates at both low and high relative humidity. All the above materials have relied on the presence of water as proton carrier. Although practicable at lower temperatures, water management at temperatures higher than 100°C has a significant impact on the viability of the MEAs and on overall operation of the system. Over the last 10–15 years, new concepts have evolved involving alternative proton carriers that mark a move towards reducing the need for high levels of hydration (Schuster et al., 2005). The role played by acid-doped basic polymers such as polybenzimidazole (PBI) is increasingly being recognised for its importance in membrane electrode assemblies for PEM fuel cells operating in the medium and high temperature ranges, above the temperature at which PFSA membranes can be used (Seel et al., 2009). Acid-imbibed PBI is prepared by immersion of PBI membranes in acid solutions (Wainright et al., 1994). This ‘conventional’ method of acid doping through immersion of membranes in acid baths can be generally applied to other acids, other basic polymers and members of the PBI family, and cross-linking improves membrane mechanical stability (Li et al., 2007). Alternatively, H3PO4-PBI membranes are elaborated through a process whereby a polybenzimidazole polymer is prepared by a polycondensation reaction from its monomer components in polyphosphoric acid (PPA). After casting of the PBI-PPA solution, controlled hydrolysis of PPA to phosphoric acid leads directly to PBI membranes containing high quantities of H3PO4 (Xiao et al., 2005). This is the PPA process of BASF that provides the Celtec-P membranes and MEAs. The conductivity properties

© Woodhead Publishing Limited, 2013

Electrochemical devices for energy

569

Proton conductivity (S cm−1)

1

0.1

0.01

1E–3

0

50

100

150

200

Water uptake (wt%)

15.4 Comparison of proton conductivity of various types of sulfonated hydrocarbon proton exchange membranes at ca. 25°C as a function of the water uptake (wt%). Open stars (Nafion®), black stars (reference sulfonated hydrocarbon PEMs), open circles (sulfonated hydrocarbon proton exchange membranes incorporating functional groups such as nitrile), up-triangles (sulfonated hydrocarbon multiblock membranes), inverted triangles (grafted and branched sulfonated hydrocarbon proton exchange membranes), squares (sulfonated hydrocarbon PEMs with high IEC), and circles with a dot in the centre (sulfonated hydrocarbon PEMs with highly sulfonated monomers) (Park et al., 2011).

of H3PO4-PBI membranes produced by the PPA process differ substantially from those prepared by imbibing PBI membranes with phosphoric acid, which suggests that the PPA process imparts a polymer membrane structure capable of providing a proton transport mechanism that is more efficient than that of the imbibed membrane. Despite the high temperatures of operation at which H3PO4-PBI type MEAs operate, they show exceptional stability and durability, including withstanding stop−start and load cycling events. A number of degradation mechanisms have been proposed for PBI-based MEAs, such as phosphoric acid loss from the membrane, faster catalyst dissolution in hot acid medium, Pt catalyst sintering, thermal stress on fuel cell parts, thermal degradation of the catalyst support and carbon support corrosion (Schmidt, 2009). In the vast majority of cases, phosphoric acid has been used as acid dopant for PBI or other basic polymers. Early research, however, also considered doping with other acids, including sulfuric and perchloric acids (Xing and Savadogo, 1999). More recently, and with the particular objective of avoiding leaching of the acidic component from the membrane in the presence of a liquid feed or, polymerised organic derivatives of phosphoric acid have

© Woodhead Publishing Limited, 2013

570

Handbook of membrane reactors

been considered as the electrolyte component of high-temperature membranes. The less acidic and less hydrophilic character of phosphonic acid than of sulfonic acid functions allow for reduced water uptake (Brunklaus et al., 2009). This clearly facilitates operation temperatures beyond the boiling point of water, even in a non-pressurised fuel cell, and validates phosphonic acid functionalised membranes as a viable option for ‘medium’ temperature (120–130°C) – as well as higher temperatures – and low RH operation. The conductivity of polybenzimidazoles grafted with PVPA increases with temperature up to 160°C in an open conductivity cell (i.e., low RH conditions), giving a conductivity of c. 28 mS/cm at 160°C (Sukumar et al., 2007), and PVPA-grafted polysulfones have a proton conductivity of 5 mS/cm at 120°C and nominally dry conditions (Parvole and Jannasch, 2008). In Celtec-V, developed by BASF, PVPA is immobilised in a PBI matrix by polymer network interpenetration and cross-linking. Celtec-V has been investigated as a membrane for direct methanol (Gubler and Scherrer, 2007) and hydrogen-air fuel cells, and has recently been shown to give remarkable performance and several thousand hours durability at high temperature, with performance increasing over the range of temperature 60–190°C, as well as with pressure (Nedellec et al.,2010).

15.3.5

Application fields

The PEMFC system has seen such an important development in the last ten years that it would be impossible to describe in few lines the extent of its application. The transport area is surely the most challenging. After fluctuating periods, serious advances have been made on the cost and efficiency levels. Daimler is announcing a well-planned market entry for a fuel cell vehicle in 2014. US and Japanese developers are also ready. Already there are several buses fleet in many European, American and Asian cities. Light portable applications have their niche markets and numerous PEMFC units are already in use for residential applications, most particularly in Japan. In Canada, the first commercial benefits from PEMFC systems have been registered in the three last years. Moreover, the intense research activity on membranes and the opportunities offered by higher operating temperatures and relatively low humidity open an important field of development.

15.4 15.4.1

Other types of low-temperature fuel cell Microbial fuel cells (MFCs)

This new and important field would require a whole chapter and we only intend here to describe some specific features. An MFC system is close to a PEMFC,

© Woodhead Publishing Limited, 2013

Electrochemical devices for energy

e−

571

e−

Glucose

MED

CO2

H2O

O2 H+

Anode

e−

e−

H+ MED

Bacterium

H+

Membrane

Cathode

15.5 Scheme of an MFC (MED: electrochemical mediator allowing electron transfer to the anode).

but instead of platinum it uses bacteria as the catalyst for oxidising mostly an organic substrate such as glucose, etc. and converting chemical energy into electrical energy (Min and Logan, 2004; Du et al., 2007; Liu et al., 2008; Zhou et al., 2011). The classical system consists of anode (carbon) and cathode (carbon with Pt electrocatalyst) compartments separated by a proton exchange membrane, such as the Nafion® type, as shown schematically in Fig. 15.5. On the anode side, the substrate is transformed by the microbial process under anaerobic conditions producing electrons and protons. Electrons are transferred to the anode through electrochemical mediators (MED) and flow to the cathode. Protons are transferred to the cathode side through the membrane. As in a PEMFC, water is produced in the cathode side. The interest of such a fuel cell is in generating electricity by consuming waste water, pollutants, etc. Even though it has been known since 1911 that bacteria can generate electrical current, it was only in the 1980s that the use of redox mediators, promoting the electron mobility from bacteria to the anode, increased the output power of such devices significantly. Moreover, coating biofilms

© Woodhead Publishing Limited, 2013

572

Handbook of membrane reactors

containing the bacteria at the anode surface also opened new development opportunities. In the last five years important progress has been achieved and many applications are in view, among which the most relevant is the biological treatment of waste water. Nevertheless, the application of such systems is much larger, for example, elimination of nitrates in contaminated soils, breaking down varied organic substrates, biosensors and sensors for biological oxygen. The application field of MFCs is still limited because of their low power density, but the rapid progress in electrode materials, membranes and electrochemical mediators offers an optimistic horizon.

15.4.2

Alkaline fuel cells (AFCs)

AFCs can work at rather low temperatures of 80–100°C. They are made of hydroxide ion (OH−) conducting membrane. The reactions are as follows: Anode : 2H 2 +4OH − → 4H 2 O Cathode: O2 + 4e 4 − + 2H 2 O

44e −

[15.26]

4OH −

[15.27]

In the 1960s, AFCs were ‘rediscovered’. They had been used by NASA (Gemini program) and aboard military submarines. The AFC could compete with the PEMFC, at least in some applications, that is, stationary power production. The activation losses are lower in alkaline medium, and catalysts such as silver or manganese could be used instead of platinum. The main drawback of current AFCs containing a liquid electrolyte (aqueous KOH solution) is their sensitivity to carbon dioxide pollution. Consequently, performance is drastically reduced. One attractive solution could be the replacement of the KOH solution by an anion conducting polymer electrolyte. To ensure anionic conduction properties, the polymer backbone must incorporate the cationic functions. In such electrolytes, the cation is a quaternary ammonium that exhibits a higher thermal and chemical stability than other quaternary cations such as quaternary phosphonium or tertiary sulfonium groups. In alkaline medium, quaternary ammoniums can be very sensitive to Hoffman degradation. This leads to an elimination of tertiary amine from the neighbouring carbon, accompanied by the formation of an olefin. Only a few membranes were evaluated for use as solid polymer electrolytes for AFCs: chitosan-based electrolyte composite membrane, polysulfone matrix, polyether and chloride polymers networks based on polyvinylbenzyl chloride or polyepichlorhydrin homopolymer and polyoxyethylene-co-polyepichlorydrin copolymer. More

© Woodhead Publishing Limited, 2013

Electrochemical devices for energy

573

recently, an anion conducting membrane based on a polyepichlorhydrin copolymer using allyl glycidyl ether as cross-linking agent has been considered. This copolymer is an aliphatic polyether, chemically stable in alkaline medium. It has an elastomer character with a low glass transition temperature (−25°C) that permits stable mechanical properties up to 100°C (Stoica et al., 2007).

15.4.3

Phosphoric acid fuel cells (PAFCs)

PAFC systems have seen an impressive development during the 1990s and the beginning of this century, reaching a quasi-commercial level. Hundreds of 100 to 200 kW systems have been disseminated all over the world. The largest fuel cell system ever developed (11 MW in Japan) is precisely a PAFC. Lifetimes close to 40 000 h, with yields of 35–40%, have been reached. The principal developers were ONSI and Fuji. Even though there still has been some production in the last years, the general growth is declining. However, it is interesting at least to give a brief overview of this type of fuel cell. The usual operating temperature is between 150–210°C under 4 to 6 bar. The electrode reactions are similar to those in a PEMFC. Concentrated phosphoric acid is the electrolyte supported by a silicon carbide matrix. It is a low volatility medium, good ionic conductor and allows a good oxygen solubilisation. Nevertheless, it solidifies at 42°C, which obliges maintaining a minimum of heat when the system is stopped. Both anode and cathode are made of carbon. In fact, carbon grains with Pt catalyst are supported in a spongy porous structure with the help of polytetrafluoroethylene. Generally, the fuel is natural gas, transformed by reforming into hydrogen and carbon monoxide. The problems of PAFCs are their low electrical yields and the level of electrolyte solidification and leakage.

15.5 15.5.1

High-temperature fuel cells: solid oxide and proton conductor fuel cells Solid oxide fuel cells (SOFCs)

The operating temperature of SOFCs is higher than all the other fuel cells described. Initially optimised for a working temperature of 960°C, the current goal is to greatly decrease it in order to develop more compact architectures, allowing the use of metallic interconnect materials instead of fragile ceramics, and limiting the ageing of such high-temperature ceramic cells. Moreover, ceramic interconnect components are more expensive to manufacture, and it is more difficult to machine them with channels and complicated shapes.

© Woodhead Publishing Limited, 2013

574

Handbook of membrane reactors Cathode Electrolyte Anode (a)

(b)

15.6 (a) Scheme of an electrolyte-supported single SOFC. (b) Scheme of an electrode-supported single SOFC.

The reactions at both electrodes are the following: Anode: 2H 2

2O O2

2H 2 O + 4e −

Cathode: O2 + 4e 4 − → 2O2 −

[15.28] [15.29]

The electrolyte membrane is an oxide ion conductive ceramic, whose thickness depends on the cell design. One may distinguish electrolyte-supported cell from electrode-supported cell (Fig. 15.6). In the first case, anode and cathode are deposited onto both faces of the electrolyte membrane. As a direct consequence, the membrane must be mechanically strong, and a minimal thickness of 100 µm is required. In the case of the electrode-supported cell, the anode is actually the mechanical support of the electrolyte first, and next the cathode on the top. Thus, the electrolyte thickness can be greatly reduced, down to 8 µm for classical SOFC devices. More recently, with the development of micro-SOFC, it can reach 100 nm to 1 µm. Brief state of the art of the electrodes The usual anode is a cermet (composite material made of a ceramic and a metal). Porous Ni–YSZ (yttria-stabilised zirconia) is the state-of-the-art electrode, presenting electronic and ionic conductivities in order to increase the number of reaction sites, called triple phase boundaries. It corresponds to the area where O2−, e− and H2 are all present for the time required for the oxidation reaction to occur. No single phase has been found to completely fit all the requirements for an anode: thermal and chemical compatibilities with the electrolyte, mixed ionic and electronic conductivity, high electrocatalytic activity and stability in reductive atmosphere. Concerning the cathode materials, porous La1−xSrxMnO3−δ was used at the highest temperature, presenting mainly electronic conductivity even if a partial ionic conductivity was observed around 1000°C. Due to the need to decrease the operating temperature of SOFCs, a lot of work has been done in order to increase the cathode performance. LSM is still used when associated to YSZ in a composite, in order to increase the ionic conductivity of

© Woodhead Publishing Limited, 2013

Electrochemical devices for energy

575

the material and the thermal compatibility with the electrolyte. However, a new strategy has been adopted since middle of 1990s by developing new oxide phases with higher intrinsic ionic conductivity, allowing a larger number of oxygen reduction sites. The most promising are probably materials of Ln2NiO4+δ family (Ln = La, Pr, Nd, etc.), suggested few years ago by Grenier’s group (Mauvy et al., 2003; Lalanne et al., 2008). Criteria for a SOFC electrolyte membrane An oxide that may be considered as a good electrolyte candidate should have the following properties: (a) almost pure ionic conductivity σ ionic transport number tion > 0.98, higher than 10−2 S.cm−1; (b) chemical stability in oxidising and reducing atmospheres, thus in a large range of oxygen partial pressure (10−20 atm to 1 atm); (c) chemical stability towards both cathode, anode and interconnect materials; (d) thermal compatibility towards both cathode and anode, in the whole range of temperature; (e) mechanical strength; (f) low cost and high availability. Zirconia-based materials The state-of-the-art SOFC electrolyte is YSZ. It consists of zirconium oxide doped with yttrium oxide, crystallising in a fluorite structure. Y-doping allows firstly stabilising the cubic structure in the whole operating temperature range, from room temperature up to sintering temperature (and more!), and second, to present high pure ionic conductivity in a wide range of oxygen partial pressures (10−21 to 1 atm, even at 1000°C). Ionic conductivity results from the formation of oxide ion vacancies, as shown in the Kröger−Vink notation: 2 ZrO2

Y2 O3 ⎯

..

′ ⎯ → 2Y YZr + VO

x 33O OO

[15.30]

Up to now, only YSZ has fulfilled all the requirements for a SOFC electrolyte. Ceria-based materials Despite a higher conductivity than zirconia-based materials, ceria-based compounds could not be considered for SOFCs working at a temperature higher than 550°C. Indeed, the main drawback of ceria is its reducibility under hydrogen, introducing a non-negligible electronic conductivity through the membrane, and as a direct consequence a drastic reduction of performance. However, because of the economic interest of decreasing the operating temperature, ceria-based materials are more and more attractive,

© Woodhead Publishing Limited, 2013

576

Handbook of membrane reactors

as electrolyte materials but also as thin interfacial layers. In this later case, the ceria-based layer acts as a diffusion barrier (in order to prevent cation diffusion between electrolyte and cathode materials, provoking the negative formation of secondary and insulating phases), sticking layer or in promoting electrochemical activity at the anode/electrolyte interface. Ceria, when un-doped, is highly resistive, while an appropriate doping by rare earth elements (Gd, Y, Sm, etc.) promotes the ionic conductivity by the formation of oxide ions vacancies, as for zirconia-based compounds. Even co-doping has been considered more recently. Lanthanum gallate-based materials The LaGaO3 family has been suggested (Ishihara et al., 1994) as a potential base compound for intermediate temperature SOFCs (IT-SOFCs), operating close to 700°C. This compound crystallises in the perovskite structure, even if substituted on A- or/and B-sites, which considerably improves the oxide ion conductivity. Nevertheless, these materials suffer from Ga volatilisation at high temperature, mainly during sintering step, and a non-sufficient stability under very low oxygen partial pressure. Other materials Apatites Apatite-type A10−x(MO4)6O2±d (M = Si, Ge) phases, where A corresponds to rare- and alkaline-earth cations, present high oxygen ionic conductivity. The apatite crystal can be described as a ‘hybrid’ structure consisting of covalent SiO4 tetrahedra and ionic-like La/O channels (Fig. 15.7). The oxygen transport in Re10−xSi6O26+d (Re = La–Yb; x = 0–0.67) increases with increasing Re3+ radius, whereas the activation energy decreases. Germanate-based apatites possess higher conductivity compared to silicates, but suffer from Ge volatility, transformation into La2GeO5, and high activation energies for ionic transport. The silicate oxides with apatite-type structure and composition La10−x(SiO4)6O2±δ have been suggested as promising electrolytes for SOFCs, exhibiting high ionic conductivity compared to YSZ in the intermediate temperature range (Fig. 15.8), combined with moderate thermal expansion coefficients and low electronic conductivity (Nakayama et al., 1995, 1999; Arikawa et al., 2000; Abram et al., 2001; Sansom et al., 2001; León-Reina et al., 2004; Slater et al., 2004). Numerous works, including theoretical atomistic simulations (Islam et al., 2003; Tolchard et al., 2003; Kendrick et al., 2008) and neutron powder diffraction (León-Reina et al., 2006, 2007), show that the ionic conduction occurs mainly via interstitial oxide migration both parallel and perpendicular to the channels and it increases with the oxygen concentration in the lattice.

© Woodhead Publishing Limited, 2013

Electrochemical devices for energy

577

Ca12+

b PO43−

c

a

CaII2+

F−

15.7 View of hexagonal apatite structure of Ca10(PO4)6F2.

Silicate apatites present several limitations for their potential use as SOFC electrolyte. One of the main drawbacks of apatite-type materials resides in the difficulty in preparing dense ceramic materials, which are needed for SOFC operation. On the other hand, degradation of the ionic transport properties with time in reducing atmospheres at high temperatures has been reported for these materials, which was ascribed to silica migration and volatilisation in the ceramic surface (Yaremchenko et al., 2004). The apatite structure allows a large number of cation substitutions. Among them, the partial substitution of Si4+ by Al3+ seems to enhance the ionic conductivity and partially suppress silicon volatilisation (Shaula et al., 2005, 2006). Despite the number of reports on the structural and transport properties of many apatite electrolytes, their potential use in SOFC devices has not been widely studied (Brisse et al., 2006; Yoshioka et al., 2008; Bonhomme et al., 2009). In this sense, Tsipis et al. (2007) and Yaremchenko et al. (2009) reported the electrochemical behaviour of different cathode materials in contact with silicate apatite electrolytes and they found that silicon migration towards the surface layer blocks the electrochemical reaction zones increasing the electrode polarisation. In addition, high area-specific resistances were found (Brisse et al., 2006) with NiO-La9SrSi6O26.5 cermets. However, a complete single fuel cell with apatite-type electrolyte, using different electrode materials, has not been reported before. Hence, further studies are still needed for a better characterisation of these potential electrolyte materials for SOFCs on several issues, such as the chemical reactivity and electrochemical performance between apatite electrolytes and different

© Woodhead Publishing Limited, 2013

578

Handbook of membrane reactors T (°C) −1

800 700

600

500

400

300

Log (σ/S cm−1)

−2

−3

−4 La10 Si5.5 Al0.6 O26.75 Zr0.84 Y0.16 O1.92 −5

La0.8 Sr0.2 Ga0.6 Mg0.2 O2.8 Ce0.8 Gd0.2 O1.9

−6

1.0

1.2

1.6

1.4 3/T

10

1.8

2.0

(K−1)

15.8 Silicate apatite conductivity compared to that of: YSZ, GDC (gadolinium-doped ceria) and LSGM (lanthanum- and strontium-doped magnesium gallate) (Marero-Lopez et al., 2010).

electrode materials. Ceria-based compounds have been introduced in the best case as a protective buffer layer in order to improve electrochemical performances, preventing from cation diffusion. LaMOx family Lacorre et al. (2000) and Goutenoire et al. (2000) have evidenced fast oxide ion conduction in La2Mo2O9, larger at high temperature than that of YSZ, the most commonly used electrolyte for SOFCs. However, this binary oxide exhibits a reversible α→β structural phase transition around 580°C (monoclinic → cubic ≡ order → disorder), with a 0.5% abrupt volume cell variation, detrimental to any electrochemical applications. It can cause a drastic drop in the ionic conductivity below 580°C and electrolyte breakdown due to repeated cycling between the high- and low-temperature polymorphs. Therefore, suppressing this structural transition may be a good

© Woodhead Publishing Limited, 2013

Electrochemical devices for energy

579

way to improve the conducting properties of this material family. As a consequence, a lot of work has been devoted to the stabilisation of the high conducting β-form at room temperature, through cationic site substitution on La and/or Mo sites (LAMOX family) (Goutenoire et al., 2001; Collado et al., 2002; Khadasheva et al., 2002; Wang et al., 2002; Georges et al., 2003a, 2003b; Hayward and Redfern, 2004; Marozau et al., 2004; Subasri et al., 2004; Tealdi et al., 2004; Marrero-Lopez et al., 2005). It seems that some of these substitutions may stabilise the β-cubic structure down to room temperature. The additives failed, however, to increase the ionic conductivity of the La2Mo2O9 to any significant extent. Moreover, as shown by VegaCastillo et al. (2010), even if La2Mo2O9 remains stable above a PO2 of 10–7 Pa, it decomposes to a mixture of reduced and crystallised molybdates for an oxygen partial pressure below this value. This is not favourable for the use of molybdenum-compounds such as the LAMOX family’s materials as electrolytes in SOFCs.

15.5.2

Proton conductor fuel cells (PCFCs)

PCFCs are made of a ceramic electrolyte, presenting protonic conductivity and working at high temperature. This technology is more recent, and under study in view of increasing the conversion efficiencies compared to PEMFCs, and at the same time decreasing the operating temperature in order to increase the lifetime of the all-ceramic fuel cell. PCFCs may exhibit more advantages than SOFCs in many respects, such as low activation energy (Sammells et al., 1992), high energy efficiency (Demin et al., 2002), and water produced on cathode side which will not dilute the fuel concentration on the anode side (Yang et al., 2008). Many perovskite-type (ABO3) oxides show high proton conductivity in a reducing atmosphere. The most common proton-conducting oxide is BaCeO3, which has been doped with various oxides, including those of samarium (Ranran et al., 2006), neodymium (Schober, 2003; Kobayashi et al., 2005; Gorbova et al., 2006; Su et al., 2006) and ytterbium (Taherparvar et al., 2003), for use in SOFCs. The conductivity of neodymium–yttria co-doped barium cerate is particularly high. Nevertheless, cerate samples easily degrade as they react to form insulating barium carbonate (BaCO3) and cerium oxide (CeO2), when exposed to atmosphere containing CO2 and H2O in a working environment (Gopalan and Virkar, 1993; Schober, 2003). Compared with doped cerates, doped zirconates have better chemical stability but lower conductivity. The stability of doped-barium cerates could be improved by the introduction of Zr at B site in order to achieve high proton conductivity and sufficient chemical stability in fuel cell operation conditions.

© Woodhead Publishing Limited, 2013

580

Handbook of membrane reactors

However, it is quite difficult to prepare such materials as high density pellets. To improve the sintering activity, Babilo and Haile (2005) and Tao and Irvine (2006) introduced Zn into Y- and Zr- doped BaCeO3. Results show that it is an effective way to obtain a stable proton-conducting electrolyte BaCe0.5Zr0.3Y0.16Zn0.04O3−δ (BCZYZ) by co-doping of Zr and Zn, which could be sintered densely at lower temperature (150°C less) than the materials without Zn. Moreover, BCZYZ can be easily fabricated as a 20 µm thin membrane for electrolyte application (Ding et al., 2010). Other proton-conducting oxides for potential use in SOFCs include BaSc0.5Zr0.5O3, (La,Pr)0.9Ba1.1GaO3.95 and Nd0.9Ba1.1GaO3.95 (Kendrick et al., 2005).

15.6 15.6.1

High-temperature fuel cells: molten carbonate fuel cells (MCFCs) and new concepts Molten carbonate fuel cells (MCFCs)

Concept of a molten salt electrolyte membrane Molten carbonates are non-toxic salts forming eutectics with low melting points (about 488 and 501°C for Li2CO3-K2CO3 and Li2CO3-Na2CO3 melt, respectively) (Broers, 1958; Sangster and Pelton, 1987). Their high conductivity (from 1.2 to 2.5 S cm−1 at 700°C) and chemical/electrochemical properties makes them suitable for electrochemical applications, such as the MCFCs (Maru, 1984). At operating temperatures between 600°C and 700°C the kinetics of fuel oxidation (H2, CO, natural gas, gasified coal) and oxidant reaction (air), are sufficiently high to allow the use of electrode materials cheaper than noble metals used in low-temperature fuel cells. They also permit cogenerating electricity and heat with global yields higher than 80%. The first MCFC was constructed in 1921, but Broers in the late 1950s was the first to initiate its real development using a ternary eutectic Li2CO3-Na2CO3-K2CO3 with MgO matrix (Appleby and Nicholson, 1980). In 1965, the replacement of MgO by LiAlO2 as a new electrolyte support increased significantly MCFC lifetime up to 12 000 h (Lu and Selman, 1992). Since the early 1970s, where lithiated nickel oxide cathodes and chromium-doped nickel anodes are used, this technology has been progressing constantly. Power density is now reaching 200 mW cm−2 and lifetime about 35 000 h for 200–300 kW systems. More than 90 demonstration devices have been tested in the United States, Asia and Europe. A system developed by Fuel Cell Energy (USA), the DFC3000 (2.8 MW), is shown in (Fig. 15.9).

© Woodhead Publishing Limited, 2013

Electrochemical devices for energy

581

15.9 Photo of the DFC3000 (2.8 MW) MCFC system developed by Fuel Cell Energy.

Principle and scheme of an MCFC The fuel directly introduced in the anode compartment is H2 or a mixture of H2 + CO, resulting from natural gas conversion by reforming or thermal cracking. The oxidant is constituted by a mixture of air and CO2. Schematically, the electrochemical reactions occurring at the electrodes are the following: At the cathode:

1 O2 2

CO2

2e

CO23 −

[15.31]

This global reaction always proceeds, but a detailed analysis has shown that − reduced oxygen species, O2− 2 and O2 are involved in the reduction process and the mechanism is complex (Nishina et al., 1990; Cassir et al., 1993; Weever et al., 1995; Cassir et al., 1997). At the anode: H 2 + CO23 − → H 2 O + CO2 + 22e −

[15.32]

The kinetics of this reaction, even if controversial, always involves adsorption and desorption of hydrogen and is considered as more rapid than the reduction one (Flood and Forland, 1947; Trémillon, 1997).

© Woodhead Publishing Limited, 2013

582

Handbook of membrane reactors

The global cell reaction is the following: H2

1 O2 + CO2 (cathode) → H 2 O + CO2 (anode) 2

[15.33]

The CO2 formed at the anode is recycled and consumed at the cathode. The emf of the cell reaction is: emf

emf e f0 +

RT P(H 2 ) (O2 )1/ 2 RT P(CO2 )cathode ln + ln 2F P(H 2 O) 2F P(CO2 )anode

[15.34]

with emf0 the standard value of the cell reaction (all activities are 1) and P(i) the partial pressure of i gas. Chemical properties of molten carbonates: oxoacidity concept The self-ionisation constant of a molten carbonate is characterised by the equilibrium: M 2 CO3 (l) ⇔ M 2 O(s)+CO2 (g)

[15.35]

where (l), (s) and (g) refer to the liquid, solid and gas phase, respectively. Given that ionic dissociation is very important in molten media, this equilibrium can be expressed in terms of ionic species: CO23 ⇔ O2 + CO2 In this kind of melt, containing an electron pair donor such as O2−, it is possible to develop the concept of oxoacidity. In this case, O2− can be associated to an oxoacid CO2 to form an oxobase CO2− 3 (Yamada and Uchida, 1994; Tomczyk et al., 1995). The ‘acidity’ of the melt can be stabilised by fixing either the carbon dioxide pressure or the oxide ion concentration. The extent of the accessible acidity range, defining the limits fixed by the saturation of the molten salt by M2O (the most basic medium) and the molten salt under a carbon dioxide pressure of one atmosphere (the most acidic medium), can be characterised by the following apparent constant, Kd, derived from the self-ionisation constant of the molten carbonate, K*: pKd* = − log Kd* , with Kd* a( a( 2 ) P( K * a( 2 2) 3) Brief state of the art of MCFC materials Only a few basic ideas will be given on the state-of-the-art electrode materials in MCFC devices, mainly with the purpose of understanding the compulsory requirements for the electrolyte membrane and the way to optimise it.

© Woodhead Publishing Limited, 2013

Electrochemical devices for energy

583

The state-of-the-art cathode has consisted of porous nickel since the 1970s. When metallic nickel is in contact with the carbonate melt under an oxidising atmosphere (air/CO2), a NiO layer, more inert than the metal, appears. NiO, formed in situ at high temperature, is a p-type semi-conductor containing crystal defects in its lattice. In the cathode MCFC environment, alkali species can be incorporated into the NiO lattice, which creates positive holes. As the conductivity of NiO is strongly dependent on the defects in the crystal, the presence of these incorporated species increases the cathode conductivity (Fukui et al., 1998; Janowitz et al., 1999). LixNi1−xO cathode presents a relatively high solubility in the electrolyte, which can provoke the formation of metallic nickel and short-circuiting between the anode and the cathode. Alternative electrode materials should be more stable in the carbonate medium and allow obtaining electrical performance as well as LixNi1−xO. In order to solve this problem, different materials can be possible substitutes: (a) Single oxides: LiFeO2, Li2MnO3, La0.8Sr0.2CoO3 and LiCoO2 (Carewsca et al., 1997; Fukui et al., 1998; Janowitz et al., 1999; Belhomme et al., 2001; Mohamedi et al., 2001); (b) NiO-based materials containing other oxides less soluble and with performance close to that of LixNi1-xO, such as alloys Ni-M (M = Nb, Al, Ti, Co) (Wijayasinghe et al., 2004; Ringuedé et al., 2006) and mixed oxides LiFeO2-LiCoO2-NiO (Daza et al., 2000; Fukui et al., 2000), NiO-LiCoO2 (Kuk et al., 2001) or NiO-CeO2 (Durairajan et al., 2002); (c) Ni state-of-the-art cathode by thin coatings of oxides, mostly LiCoO2 or cobalt-based oxides, more corrosion-resistant and presenting interesting electrocatalytic properties (Yuh et al., 1995; Mendoza et al., 2003; Mendoza et al., 2004; Pauporté et al., 2005; Escudero et al., 2006; Lair et al., 2008; Paoletti et al., 2009; Kim et al., 2011). Since the early 1970s, porous nickel has been the current MCFC anode used (Wendt et al., 1993; Frangini and Masci, 2004); nevertheless, its stability, due to mechanical deformation under compressive load, is a major problem. Thus, to avoid mechanical stress and cracking during operation, the anode should be strengthened. Its structure is commonly stabilised via the addition of about 2–10% of chromium; however, the LiCrO2 surface layer formed under anode conditions decreases the wettability of molten carbonates and modifies the anode surface (Wendt et al., 1993; Moon and Lee, 2003; Frangini and Masci, 2004). For this reason, the amount of Cr was reduced, but the creep problem increased and other solutions have been analysed. One of the most promising solutions is the addition of aluminium to nickel, permitting a good protection of the anode by improving its mechanical resistance (Cassir and Belhomme, 1999; Perez et al., 2002; Moon and Lee, 2003; Yoshiba et al., 2004).

© Woodhead Publishing Limited, 2013

584

Handbook of membrane reactors

State of the art and improvement of MCFC electrolyte matrix membrane The membrane separating the anode and the cathode is constituted by a carbonate eutectic, commonly Li2CO3/K2CO3 (62/38 mol%) and Li2CO3/ Na2CO3 (52/48 mol%) (Glugla and Decarlo, 1982; Morita et al., 2002; Scaccia, 2005), supported by a porous ceramic body constituted with thin, isolating and chemically inert particles insoluble in the electrolyte, γ-LiAlO2, the most stable crystalline structure of lithium aluminate, is impregnated by the carbonate eutectic to form a paste structure. The resulting matrix contains about 45 wt% of carbonate melt and 55 wt% of lithium aluminate and is manufactured as a thin plate by tape casting (Durairajan et al., 2002). The corrosiveness of the electrolyte is one of the most serious drawbacks of the MCFC system. It may act at the level of the interconnectors or the electrodes, but the corrosion and the dissolution of the cathode is the most critical aspect. Therefore, it is necessary to optimise the electrolyte composition, by adding other species in the carbonate eutectic in order to control its oxoacidic properties and decrease the NiO (Morita et al., 2002). An oxobasic medium favours in principle cathode resistance to corrosion and decreases its solubility, which can be reached by adding a few percent of alkali earth, Mg, Ca, Sr and Ba (Perez et al., 2002; Frangini and Masci, 2004; Mendoza et al., 2004; Yoshiba et al., 2004). But most of these additives have the inconvenience of decreasing the conductivity of the salt and the performance of the cell (Selman and Maru, 1981). Contrary to the addition of strontium, which is negative with respect to cell performance, it appears that the best compromise is the use of a ternary salt Li/Na 52 mol% with CaCO3 9 mol% and BaCO3 9 mol% (Tanimoto et al., 2003). Fe, La and W have also been added to Li-K carbonates, because they inhibit NiO dissolution, but may provoke cation segregation in long-term operation (Mitsushima et al., 2002; Matsuzawa et al., 2005). It has also been found that rare earth additions in molten carbonates have a beneficial effect with respect to nickel solubility; the more soluble this oxide is in the carbonate melt, the lower is the nickel solubility; therefore, La2O3 is the best rare earth additive (Ota et al. 2006). New carbonate mixtures, such as Li2CO3-Cs2CO3 mixtures, have also been tested in single cells, showing lower overpotentials and higher cell potentials than Li-K or Li-Na carbonate eutectics (Yuh et al., 1992). In effect, oxygen solubility is higher in such melts, which favours the reduction kinetics, especially at low temperature; however, these melts are more acidic, which favours NiO dissolution (Paetsch et al., 1993). Surface tension of molten carbonates is another significant parameter, which controls the distribution of the electrolyte between the electrodes, the reactant gas solubility in the electrolyte and the electrochemical reactions and, therefore, the performance. Surface tension of carbonate mixtures can

© Woodhead Publishing Limited, 2013

Electrochemical devices for energy

585

be estimated as a function of the composition and temperature by the theory of corresponding states (Kim et al., 2009), which allows prediction of the effect of additives on the properties of the carbonate melt. An important issue to improve the stability and the performance of MCFCs is to reduce the resistance of the lithium aluminate matrix, which is responsible for 70% of the ohmic resistance of the cell. A first solution is to decrease its thickness (around 0.3 mm) (Yuh et al., 1992). As γ-LiAlO2 suffers stresses during thermal cycling, it is nowadays mixed with thicker particles of α-LiAlO2 and fibres of α-LiAlO2, (proportions: 55%-30%-15%) (Paetsch et al., 1993). LiAlO2 matrix can be reinforced by adding Al and Li2CO3 particles, which increases its mechanical strength significantly (Kim et al., 2009).

15.6.2

New concepts of high-temperature fuel cells

Direct carbon fuel cell Carbon is a fossil fuel known to be polluting, but it cannot be neglected in the future for two reasons: it has a high energy density, and there are important resources of carbon, for at least 200 years, spread all over the world. However, it is necessary to find more efficient devices for carbon conversion. Different electrolytes, such as solid oxide or molten carbonates, are being considered for new devices such as the direct carbon fuel cell, which consists in oxidising electrochemically a carbon fuel (pyrolysis of biofuels, farming products, etc.). In this case the energy produced per unit volume is four times higher than with methane (Zecevic et al., 2004). An attractive solution is the hybrid direct carbon fuel cell, HDCFC, combining the properties of solid oxides with those of molten carbonates (Irvine et al., 2006; Pointon et al., 2006; Nabae and Pointon, 2008). It is an SOFC system with a reservoir filled with carbonates at the anode side, allowing easier transfer of carbon fuel. The molten phase favours the kinetics of the global oxidation process, because it is an excellent fuel carrier and electrochemical mediator. This concept is very promising because it allows a classical SOFC cathode to be protected from the corrosive effects of molten carbonates, and is an efficient way to oxidise carbon fuel directly through molten carbonates (Nabae and Pointon, 2008). Composite solid oxide/molten carbonate electrolyte membrane Recently, combining ceramics with molten salts has been of growing interest for innovative high-temperature fuel cell applications. In the last ten years, doped-ceria oxides mixed with molten salts, such as chlorides, fluorides, carbonates and sulphates (Zhu and Mat, 2006; Di et al., 2010; Lapa et al., 2010;

© Woodhead Publishing Limited, 2013

586

Handbook of membrane reactors

Li and Sun, 2010; Raza et al., 2010; Xia et al., 2010; Zhang et al., 2010; Zhu et al., 2010; Benamira et al., 2011) have been investigated. This phase being molten or partially molten at intermediate temperature (> 500°C for alkali molten carbonates) would create an interfacial conduction pathway. Oxide ions ensure the conductivity in the oxide phase, the conductivity being attributable mostly to carbonates in the carbonate phase. These composites form highly disordered interfacial regions (‘superionic highways’, ‘percolating conducting paths’) between the oxide phase and the carbonate phase (Zhu et al., 2010). Some achievements have already been realised. Li and Sun (2010) have developed a so-called composite (oxide/molten carbonates) NANOSOFC (nano-solid oxide fuel cell), obtaining a maximum output power density of 140 mW cm−2, stable for 200 h (Li and Sun, 2010). Di et al. (2010) studying the same kind of composites, obtained a maximum power density of 590 mW cm−2 at 600°C (Di et al., 2010; Raza et al., 2010) showed that the use of Na2CO3 improved the performance, reaching 1.15 W cm−2 at 500°C. Carbon capture and storage (CCS) An MCFC system can act as a CO2 separator and concentrator, given its capability of transporting CO2 from the cathode to the anode stream while producing electricity from fuels at the anode side. CO2 can be extracted from the flue gas of a combined cycle power plant while generating electricity, avoiding loss in plant efficiency and consequent increase in primary energy consumption. The benefit of using an MCFC seems to be proven but the effectiveness still has to be evaluated against other alternatives for CO2 emission mitigation: (a) pre-combustion (gasification + CCS); (b) oxy-fuel combustion (low volumes, high cost); (c) post-combustion (amine, ammonia, electrochemical separators, etc.). MCFCs can reach 90% CO2 separation efficiency while producing electricity at high generation efficiency. The potential and the market are therefore enormous, but the capacity for producing MCFCs for this application is as yet inadequate for large power plants (100 Mton CO2/y). Smaller power plants (15 MtonCO2/y) can find more immediate implementation. Different systems are under investigation and seem to have interesting features (Suguira et al., 2003; Wade and Lackner, 2007).

15.7

Economic aspects of fuel cell development

Nowadays, the most promising fuel cell for transportation is the PEMFC technology, due to its efficiency, compactness, its temperature range and its lifetime. Nevertheless, stacks are still expensive, essentially because of the cost of the membrane (about 150 € for a 0.3 × 0.3 m2) (available from:

© Woodhead Publishing Limited, 2013

Electrochemical devices for energy

587

http://www.nafionstore.de [accessed 20 December 2011]) and the presence of noble materials to catalyst the reactions. It is quite difficult to estimate the economic loss or gain with utilisation of fuel cell energy because mass production is not established yet and there are many different models to calculate the cost. Moreover, life-cycle cost depends also on the fuel cell size, its cost and the hydrogen cost. Anyway, different studies can be found in literature. For instance, a study concluded that the fuel economy of hydrogen fuel cell vehicles (H2-FCVs) can be 2.5–3 times the fuel economy of a conventional internal combustion engine (ICE) using gasoline (Ahluwalia et al., 2004) if the power density per cell is 0.6 or 0.7 V. These results can be lower or higher depending on assumptions, the kind of car and the way of doing the calculations, but one can say the economy of H2-FCVs compared to ICE can be in the range 2–4 times. The near future will probably be more favourable for fuel cell hybrid vehicles (FCHVs) because depending on driving situation (highways or stop-and-go urban use), the energy consumption can be different. From an economic viewpoint, it should be noted that most of the existing studies only compare the best efficiency obtained with a fuel cell stack to the ICE. Then it seems that hybridisation can improve the fuel economy up to 27% on a highway (Ahluwalia et al., 2005). More generally, the economy is closer to 3–5% depending on the car used (Zheng et al., 2011). It seems that when the cost of the fuel cell is higher than 300 €/kW, hybridisation is preferable, whereas if the cost is lower, the use of the pure fuel cell is economically advantageous (Jeong and Oh, 2002).

15.8

Principles, features and applications of electrolysis cells

Water electrolysis is a process older than the fuel cell invention. The first experiment was realised by the British chemist William Nicholson (1753– 1815) and Sir Anthony Carlisle (1768–1840). Regarding Alessandro Volta’s publication on the voltaic pile, they observed, at the very beginning of the nineteenth century, that placing the two conductive materials in water lead to the formation of hydrogen at the negatively charged electrode (the so-called cathode) and oxygen at the positively charged electrode (the so-called anode) (De Levie, 1999; Trasatti, 1999). Originally, the electrolysis cells used to have monopolar electrodes, so that each anode was connected to the positive potential of the electric generator and each cathode was connected to the negative potential of the electric generator. That is to say the electrolysis cells used to work in parallel. Subsequently, electrolysis cells were developed that used bipolar plates, where one side plays the role of the anode and the other of the cathode. Then the electrolysis cells are

© Woodhead Publishing Limited, 2013

588

Handbook of membrane reactors

working in series. This later cell assembly allows to reduce of the ohmic loss, is more compact and presents a higher current density, but it is less resistant to corrosion because the bipolar plates are in contact either with a reducing atmosphere or an oxidative one. Nevertheless, industrial cells mostly work in series. The first electrolysis units were developed at the beginning of the nineteenth century, using an acid electrolyte (mainly phosphoric acid). Due to corrosion problems with the acid electrolyte, at the beginning of the 1950s the use of solid electrolyte appears for fuel cell applications and air space conquest. Then, the idea of such a polymer electrolyte membrane was used for the electrolysis cell in the 1970s. Nevertheless, no large electrolysers are actually available, because the technology is expensive, so it is mainly used in spacecraft or submarines to produce oxygen. Steam electrolysis (at very high temperature), using an oxygen ion-conducting ceramic electrolyte should present higher efficiency in the future but this technology is still under investigation at laboratory scale. At the beginning of the past century, a few hundred electrolysers were in operation. Up to now, many electrolysis cells have been sold to produce a small amount of hydrogen from water electrolysis, but most commercial water electrolysers use an alkaline electrolyte even if other techniques are still being studied.

15.8.1

Basic principles of an electrolysis cell

Direct decomposition of water requires high energy according to thermodynamic data associated with the overall reaction: H2O = H2 + ½ O2

[15.36]

where the enthalpy of water dissociation is ΔrH = +285 kJ/mol at 298 K. This dissociation is not spontaneous and requires an external energy production. This energy can be provided by electric power which provokes the following reactions: Cathode reduction: 2H+(aq) + 2e− = H2(g)

[15.37]

Anode oxidation: 2H2O (l) = O2(g) + 4H+(aq) + 4e−

[15.38]

As these reactions are not rapid and the electrolyte membrane has a specific resistance, the energy required is always higher than the energy predicted thermodynamically. It is therefore important to optimise the electrolyte and the electrode catalysts in order to minimise this additional energy.

© Woodhead Publishing Limited, 2013

Electrochemical devices for energy

15.8.2

589

Different families of electrolysis cells

According to their temperature and the nature of the electrolyte membrane, electrolysis cells can be classified in to two families, as for fuel cells (see Section 15.2.2). Here only the main viable processes are presented. Low-temperature electrolysis cell: • Alkaline electrolysis cell: the electrolyte is a conductive hydroxide salt (mainly KOH) and the process works around 100°C. • Proton exchange membrane electrolysis cell (PEMEC): the electrolyte is a conductive proton exchange membrane (mainly Nafion®) and the process works around 100°C. High-temperature electrolysis cell: •

Solid oxide electrolysis cell (SOEC): the electrolyte is a solid oxide conductive membrane and the process works around 800°C.

15.8.3

Low-temperature electrolysis cells

Two main processes are considered to produce hydrogen from water electrolysis at quite low temperature (about 100°C). These processes are the most advanced ones and, up to now, one is already commercialised: the alkaline electrolysis cell. The second one, PEMEC, seems to be very promising, especially for coupling with renewable energy. Both systems presented below are based on the fuel cell technology associated and described in Section 15.3. Alkaline electrolysis cell (AEC) The most conventional electrolysers use 25–35 wt% KOH solutions as electrolyte. The reactions occurring at the anode and the cathode are: At the cathode: 2H2O + 2e− = H2 + 2HO−

[15.39]

At the anode: 2HO− = ½ O2 + H2O + 2e−

[15.40]

Then, hydroxide ions move from the anode to the cathode. As the electrolyte is liquid, a diaphragm is required to separate the produced gases and to avoid the recombination of H2 and O2. Most commercial industrial processes have used porous and thick diaphragms made out of asbestos. But asbestos is toxic and carcinogenic and has a long-term instability at temperatures higher than 80°C (Gupta, 2009). Then, polymeric separators have been developed,

© Woodhead Publishing Limited, 2013

590

Handbook of membrane reactors

such as thermoplastic polymers and polysulfone fibre (Vandenborre et al., 1980). Later, cermet diaphragms were developed. They were called solid polymer electrolytes (SPE) and demonstrated quite high efficiency, even if the research was more focused on electrodes than on the electrolyte (Wendt and Hofmann, 1985). More recently, diaphragms called bipolar membranes (BM) have been developed. They consist of an interfacial layer separating an anion exchange layer, allowing the migration of HO−, and a cation exchange membrane, allowing the migration of H+ (Hung et al., 2012). Globally, for hydrogen production from water electrolysis, the membrane has to be of low ohmic resistance, gastight, and mechanically and chemically stable. This kind of electrolyser is already in the market because it is much cheaper than other processes. Nevertheless, CO2 can absorb in the membrane and carbonation phenomenon occurs and limits the performances. Moreover, this kind of system can work under pressure and small units can work from 3 to 30 bar. Proton exchange membrane electrolysis cells (PEMECs) Most conventional electrolysers use a Nafion® membrane (Ito et al., 2011). The proton-conducting polymeric film is immersed in distilled water. Under a sufficiently high potential, protons migrate through the membrane and are reduced at the cathode. The general reactions are: At the cathode: 2H+ + 2e− = H2

[15.41]

At the anode: H2O = ½ O2 + 2H+ + 2e−

[15.42]

Acid electrolysis is different from alkaline electrolysis because the electrolyte is always a solid membrane. Especially the membrane is a proton exchange membrane. This is a big advantage, because the systems are then more compact, and corrosion problems are reduced. Nevertheless, this polymeric membrane is quite expensive and noble metal (Pt for instance) electrocatalysts are required. As for any electrolyser, the membrane has to be mechanically stable to eventually support pressure. It has to be chemically stable regarding reducing and oxidising chemicals (O2 and H2) and gastight to avoid gas recombination. The membrane has to be a good protonic conductor and a poor electronic one. Moreover, it has to resist temperature (up to 100–120°C) and pressure (up to 30 bar or more). The most used membrane is Nafion®. As in PEMFCs, hydration of the membrane is very important to get the best performance; indeed, if it is too low, the membrane gets ‘dry’, proton migration is reduced and it is rapidly damaged. If the membrane is too wet, the electrodes can get ‘drowned’ and electrochemical reactions are partially blocked.

© Woodhead Publishing Limited, 2013

Electrochemical devices for energy

591

2e−

H2

+ O2−

Anode

Cathode

− H2O

Membrane

½O2

15.10 Scheme of a solid oxide electrolyser single cell.

PEMEC technology is already used to produce oxygen for life (e.g., in submarines or spacecraft) but it is still expensive. A new, challenging field is the combination of PEMFCs with renewable energies (solar or wind power), because it better resists current fluctuation than alkaline processes.

15.8.4

High-temperature electrolysis cells

This technology is based on the reverse of a SOFC as depicted in Fig. 15.10. The research is quite new and the same membranes used for SOFCs are used for the production of hydrogen from steam electrolysis. Most of the time, the YSZ state-of-the-art electrolyte is used. The electrode reactions are the following: At the cathode: H2O + 2e− = H2 + O2−

[15.43]

At the anode: O2− = ½ O2 + 2e−

[15.44]

In a SOEC, the membrane is an oxygen ion-conducting electrolyte. It is a solid ceramic which allows migration of oxide ions. The most common electrolyte used is YSZ, because of its high oxide ion conductivity and its good mechanical strength. As for SOFC applications, the membrane has to be dense and chemically stable. Moreover, the best conversion efficiency is obtained when the electrolyte has a high ionic conductivity. To minimise ohmic loss, the membrane has to be as thin as possible (Ni et al., 2008). Other doping elements can be used in the membrane to improve ionic conductivity (Arachi et al., 1999), such as Gd2O3, Sc2O3, etc. Even other kinds of electrolytic membrane can be used, as described in SOFCs (see Section 15.5). This technology is not yet mature and research effort is still going on, but lately one may notice significant achievements. However, SOEC is a promising device because high temperatures can theoretically be more efficient

© Woodhead Publishing Limited, 2013

592

Handbook of membrane reactors

and the costs of electrocatalysts are lower. Moreover, at 800°C, the ratio of electric power compared to heat energy is lower than at low temperature (about 20% less), which is an economic advantage.

15.8.5

Coupling with other technologies

All the technologies of water electrolysis will probably be coupled with other technologies in order to reduce costs. Indeed, it is important to find cheap energy sources to provide heat or electrical energy required to run electrolysers, nuclear energy, wind or solar energies (Gupta, 2009). The first idea is to couple high-temperature electrolysers with nuclear power plants. In this case, the excess of heat can be used to bring part of the global energy required to produce H2. It is an interesting process which uses waste heat to produce massively large-scale hydrogen or even syngas (O’Brien et al., 2010; Stoots et al., 2010). In the future, intermittent energies will be more and more present in many countries. The problem with solar or wind energy is that one cannot have energy production 24 h a day, 7 days a week, due to night or lack of wind. Then the idea is to couple these intermittent energies with water electrolysis power plants. Part of the energy can be devoted to the alimentation of electrolysers in order to produce H2 which can be used when solar cells or wind turbine cannot work. For instance, PEMEC can be coupled with solar energy as a house H2 supplier or as a mobile unit to provide H2 anywhere (Clarke et al., 2010; Gibson and Kelly, 2010; Obara, 2010). Furthermore, wind energy is also promising for coupling renewable energy and the electrolysis of water. The role of the electrolysers is then to absorb the fluctuating electricity generated by the wind turbine, by producing H2 which can be stored (Honnery and Moriarti, 2009; Green et al., 2011). In this perspective, Norway has already run an autonomous wind/hydrogen energy demonstrator, using PEMEC of 10 kW, and they observe that the system can give 2–3 days of full energy autonomy for 10 houses (Ulleberg et al., 2010).

15.9

Conclusions and future trends

It can be concluded from this chapter that membranes have an essential role in electrochemical devices such as fuel cells and water electrolysis. Moreover, considering the variety of systems and their specific electrolytic membranes, a huge field emerges, embracing research, from testing to fundamentals, demonstration programmes and, in some cases, a real market entry. Commercialisation of electrical fuel cell vehicles in a couple of years, as announced by Daimler, should favour the development of PEMFCs. Whereas, the use of MCFCs and SOFC for co-generation and varied

© Woodhead Publishing Limited, 2013

Electrochemical devices for energy

593

stationary applications will respond to the low-carbon energy demand. From the membrane science viewpoint it is a highly valuable opportunity, because it creates an input for developing new functional and stable membranes. In our opinion, new membranes for PEMFCs resistant to higher temperatures beyond 100°C and a better dispersion of lower-cost catalysts in the interface between electrodes and electrolyte offer a fruitful investigation fields. More efficient membranes, or even hybrid membranes, for a new-generation of MCFC and SOFC devices, is also offering a valuable area of investigation and application. In most fuel cells and electrolysers, there are important breakthroughs mostly related to membrane development.

15.10

References

Abram E J, Sinclair D C and West A R (2001), ‘A novel enhancement of ionic conductivity in the cation-deficient apatite La9.33(SiO4)6O2’, J Mater Chem, 11, 1978–1979. Ahluwalia R K, Wang X and Rousseau A (2005), ‘Fuel economy of hybrid fuel-cell vehicles’, J Power Sources, 152, 233–244. Ahluwalia R K, Wang X, Rousseau A and Kumar R (2004), ‘Fuel economy of hydrogen fuel cell vehicles’, J Power Sources, 130, 192–201 Alberti G and Casciola M (2003), ‘Composite membranes for medium-temperature PEM fuel cells’, Annu Rev Mater Res, 33, 129–154. Appleby A J and Nicholson S (1980), ‘Reduction of oxygen in lithium-potassium carbonate melt’, J Electroanal Chem, 112, 71–76. Arachi Y, Sakai H, Yamamoto O, Takeda Y and Imanishai N (1999), ‘Electrical conductivity of the ZrO2–Ln2O3 (Ln=lanthanides) system’, Solid State Ionics, 121, 133–139. Arcella V, Ghielmi A and Tommasi G (2003), ‘High Performance Perfluoropolymer Films and Membranes’, Ann New York Acad Sci, 984, 226–244. Aricò A S, Di Blasi A, Brunaccini G, Sergi F, Dispenza G, Andaloro L, Ferraro M, Antonucci V, Asher P, Buche S, Fongalland D, Hards G A, Sharman J D B, Bayer A, Heinz G, Zandonà N, Zuber R, Gebert M, Corasaniti M, Ghielmi A and Jones D J (2010), ‘High Temperature Operation of a Solid Polymer Electrolyte Fuel Cell Stack Based on a New Ionomer Membrane’, Fuel Cells, 10, 1013–1023. Arikawa H, Nishiguchi H, Ishihara T and Takita Y (2000), ‘Oxide ion conductivity in Sr-doped La10Ge6O27 apatite oxide’, Solid State Ionics, 136–137, 31–37. Babilo P and Haile S M (2005), ‘Enhanced sintering of yttrium-doped barium zirconate by addition of ZnO’, J Am Cer Soc, 88, 2362–2368. Bacon F T (1969), ‘Fuel cells, past, present and future’, Electrochim Acta, 14, 569–585. Bahar B, Cavalca C, Cleghorn S, Kolde J, Lane D, Murthy M and Rusch G (1999), ‘Effective selection and use of advanced membrane electrode power assemblies’, J New Mater Electrochem Syst, 2, 179–182. Baglio V A. Arico A S, DiBlasi A, Antonucci V, Antonucci P L, Licoccia S, Traversa S E and Serraino Fiory F (2005), ‘Nafion-TiO2 composite DMFC membranes:

© Woodhead Publishing Limited, 2013

594

Handbook of membrane reactors

physico-chemical properties of the filler vs electrochemical performance’, Electrochim Acta 50, 1241–1246. Bauer B, Jones D J, Rozière J, Tchicaya L, Alberti G, Casciola M, Massinelli L, Peraio A, Besse S and Rammuni E (2000), ‘Electrochemical characterisation of sulfonated polyetherketone membranes’, J New Mater Electrochem Syst, 3, 93–98. Belhomme C, Gourba E, Cassir M and Tessier C (2001), ‘Chemical and electrochemical behaviour of NiO in the cathodic conditions used in molten carbonate fuel cells MCFC’, J Electroanal Chem, 503, 69–77. Benamira M, Ringuedé A, Albin V, Vannier R-N, Hildebrandt L, Lagergren C and Cassir M (2011), ‘Gadolinia-doped ceria mixed with alkali carbonates for SOFC applications: I- a thermal, structural and morphological insight’, J Power Sources, 196, 5546–5554. Bonhomme C, Beaudet-Savignat S, Chartier T, Geffroy P-M and Sauvet A-L (2009), ‘Evaluation of the La0.75Sr0.25Mn0.8Co0.2O3-δ system as cathode material for ITSOFCs with La9Sr1Si6O26.5 apatite as electrolyte’, J Eur Ceram Soc, 29, 1781–1788. Bradley J, Irvine J T S, Lakeman J B, ‘Electricity producing fuel cell or battery device’, Brit Patent Appl., 2004, GB0426879.3 WO2006061639-A2, 2006. Brisse A, Sauvet A-L, Barthet C, Beaudet-Savignat S and Fouletier J (2006), ‘Electrochemical characterizations of Ni/doped lanthanum silicates cermets deposited by spin coating’, Fuel Cells, 6, 59–63. Broers G H J (1958), High Temperature Galvanic Fuel Cells, Ph.D. Thesis, University of Amsterdam, the Netherlands. Brunklaus G, Schauff S, Markova D, Klapper M, Müllen K and Spiess H-S (2009), ‘Proton mobilities in phosphonic acid based PEM probed by H solid state NMR spectroscopy’, J Phys Chem B, 113, 6674–6681. Carewsca M, Scaccia S, Croce F, Arumugam S, Yang Y and Greenbaum S (1997), ‘Electrical conductivity and 6,7Li NMR studies of Li1+yCoO2’, Solid State Ionics, 93, 227–237. Cassir M, Moutiers G and Devynck J (1993), ‘Stability and characterization of oxygen species in alkali molten carbonate: a thermodynamic and electrochemical approach’, J Electrochem Soc, 140, 3114–3122. Cassir M, Malinowska B, Peelen W, Hemmes K and de Wit J H W (1997), ‘Identification and electrochemical characterization of in situ produced reduced oxygen species in molten Li-K carbonate under P(CO2) = 1 atm and P(CO2) = 0.1 atm + P(O2) = 0.9 atm. Influence of oxide, peroxide and superoxide additions’, J Electroanal Chem, 433, 195–205. Cassir M and Belhomme C (1999), ‘Technological applications of molten salts: the case of the molten carbonate fuel cell’, Plasmas Ions, 2, 3–15. Clarke R E, Giddey S and Badwal S P S (2010), ‘Stand-alone PEM water electrolysis system for fail safe operation with a renewable energy source’, Int J Hydrogen Energy, 35, 928–935. Collado J A, Aranda M A G, Cabeza A, Olivera-Pastor P and Bruque S (2002), ‘Synthesis, structures, and thermal expansion of the La2W2-xMoxO9 series’, J Solid State Chem, 167, 80–85. Coms F D, Liu H and Owejan J E (2008), ‘Mitigation of perfluorosulfonic acid membrane chemical degradation using cerium and manganese ions’, ECS Trans, 16 1735–1747.

© Woodhead Publishing Limited, 2013

Electrochemical devices for energy

595

Curtin D E, Lousenberg R D, Henry T J, Tangeman P C and Tisack M E (2004), ‘Advanced materials for improved PEMFC performance and life’, J Power Sources, 131, 41–48. Daza L, Rangel C M, Cassais M T, Martinez M J, Alonso J A (2000), ‘Modified nickel oxides as cathode materials for MCFC’, J Power Sources, 86, 329–333. De Levie R (1999), ‘The electrolysis of water’, J Electroanal Chem, 476, 92–93. Demin A K, Tsiakaras P E, Sobyanin V A and Yu Hramova S (2002), ‘Thermodynamic analysis of a methane fed SOFC system based on a protonic conductor’, Solid State Ionics, 152–153, 555–560. Di J, Chen M, Wang C, Zheng J, Fan L, Zhu B (2010), ‘Samarium doped ceria–(Li/ Na)2CO3 composite electrolyte and its electrochemical properties in low temperature solid oxide fuel cell’, J Power Sources, 195, 4695–4699. Ding H, Xue X, Liu X and Meng G (2010), ‘High performance protonic ceramic membrane fuel cells (PCMFCs) with Sm0.5Sr0.5CoO3−δ perovskite cathode’, J Alloys Compd, 494, 233–235. Du Z, Li H and Gu T (2007), ‘A state of the art review on microbial fuel cells: a promising technology for wasterwater treatment and bioenergy’, Biotechnol Adv, 25, 464–489. Durairajan A, Colon-Mercado H, Haran B, White R and Popov B (2002), ‘Electrochemical characterization of cobalt-encapsulated nickel as cathodes for MCFC’, J Power Sources, 104, 157–168. Eikerling M, Kornyshev A and Spohr E (2008), ‘Proton conducting polymer electrolyte membrane: water and structure in charge’, Adv Polym Sci, 215, 15–54. Escudero M J, Mendoza L, Cassir M, Gonzalez T and Daza L (2006), ‘Studies on the behaviour of porous nickel coated by cobalt oxide in molten carbonate’, J Power Sources, 160, 775–781. Flood H and Forland T (1947), ‘The acidic and basic properties of oxides’, Acta Chem Scand, 1, 592–604. Frangini S and Masci A (2004), ‘Intermetallic FeAl based coatings deposited by the electrospark technique: corrosion behavior in molten (LiqK) carbonate’, Surf Coat Technol, 184, 31–39. Fukui T, Okawa H and Tsunooka T (1998), ‘Solubility and deposition of LiCoO2 in a molten carbonate’, J Power Sources, 71, 239–243. Fukui T, Ohara S, Okawa H, Hotta T and Naito M (2000), ‘Properties of NiO cathode coated with lithiated Co and Ni solid solution oxide for MCFCs’, J Power Sources, 86, 340–346. Georges S, Goutenoire F, Laligant Y and Lacorre P (2003a), ‘Reducibility of fast oxide-ion conductors La2-xRxMo2-yWyO9 (R = Nd, Gd)’, J Mater Chem, 13, 2317–2321. Georges S, Goutenoire G, Altorfer F, Sheptyakov D, Fauth F, Suard E and Lacorre P (2003b), ‘Thermal, structural and transport properties of the fast oxide-ion conductors La2-xRxMo2O9 (R=Nd, Gd, Y)’, Solid State Ionics, 161, 231–241. Gibson T L and Kelly N A (2010), ‘Predicting efficiency of solar powered hydrogen generation using photovoltaic-electrolysis devices’, Int J Hydrogen Energy, 35, 900–911. Gierke T D, Munn G E and Wilson F C (1981), ‘The morphology in Nafion perfluorinated membrane products, as determined by wide- and small-angle X-ray studies’, J Polym Sci, Part B: Polym Phys, 19, 1687–1704.

© Woodhead Publishing Limited, 2013

596

Handbook of membrane reactors

Glugla P G and Decarlo V J (1982), ‘The specific conductance of molten carbonate fuel cell tiles’, J Electrochem Soc, 129, 1745–1747. Gopalan S and Virkar A V (1993), ‘Thermodynamic stabilities of SrCeO3 and BaCeO3 using a molten salt method and galvanic cells’, J Electrochem Soc, 140, 1060–1065. Gorbova E, Zhuravlev B V, Demin A K, Song S Q and Tsiakaras P E (2006), ‘Charge transfer properties of BaCe0.88Nd0.12O3−δ co-ionic electrolyte’, J Power Sources, 157, 720–723. Goutenoire F, Isnard O, Retoux R and Lacorre P (2000), ‘Crystal structure of La2Mo2O9, a new fast oxide-ion conductor’, Chem Mater, 12, 2575–2580. Goutenoire F, Isnard O, Suard E, Bohnke O, Laligant Y, Retoux R and Lacorre P (2001), ‘Structural and transport characteristics of the LAMOX family of fast oxide-ion conductors, based on lanthanum molybdenum oxide La2Mo2O9’, J Mater Chem, 11, 119–124. Green R, Hu H and Vasilakos N (2011), ‘Turning the wind into hydrogen: The long-run impact on electricity prices and generating capacity’, Int J Hydrogen Energy, 39, 3992–3998. Gross M, Maier G, Fuller T, MacKinnon S and Gittleman C S (2009), ‘Design rules for the improvement of the performance of hydrocarbon-based membranes for proton exchange membrane fuel cells (PEMFC)’, in Handbook of Fuel Cells, Vol. 5 (W. Vielstich, H. Yokokawa, and H. A. Gasteiger, eds.), Wiley. Grot W (2008), Fluorinated Ionomers, Elsevier Inc. Grove W R (1842), ‘On a gaseous voltaic battery’, The London and Edinburgh Philosophical Magazine and Journal of Science, 21, 417–420. Gubler L, Kramer D, Belack J, Unsel O, Schmidt T J and Scherer G G (2008), ‘A proton conducting polymer membrane as solid electrolyte – function and required properties’, Adv Polym Sci, 215, 1–14. Gubler L and Scherrer G G (2007), ‘Celtec-V, a polybenzimidazole-based membrane for the direct methanol fuel cell’, J Electrochem Soc, 154, B981–B987. Gupta R B (2009), Hydrogen Fuel: Production, transport, Storage, Boca Raton, CRC Press. Hayward S A and Redfern S A T (2004), ‘Thermodynamic nature of and spontaneous strain below the cubic-monoclinic phase transition in La2Mo2O9’, J Phys Condens Matter, 16, 3571–3583. Hickner H, Ghassemi H, Kim Y S, Einsla B R and McGrath J E (2004), ‘Alternative polymer systems for proton exchange membranes (PEMs)’, Chem Rev, 104, 4587–4611. Honnery D and Moriarty P (2009), ‘Estimating global hydrogen production from wind’, Int J Hydrogen Energy, 34, 727–736. Hung C-Y, Li S-D, Wang C-C and Chen C-Y (2012), ‘Influences of a bipolar membrane and an ultrasonic field on alkaline water electrolysis’, J Membrane Sci, 389, 197–204. Ishihara T, Matsuda H and Takita Y (1994), ‘Doped LaGaO3 perovskite type oxide as a new oxide ionic conductor’, J Am Chem Soc, 116, 3801–3803. Islam M S, Tolchard J R and Slater P R (2003), ‘An apatite for fast oxide ion conduction’, Chem Commun, 13, 1486–1487. Ito H, Maeda T, Nakano A and Takenada H (2011), ‘Properties of Nafion membranes under PEM water electrolysis conditions’, Int J Hydrogen Energy, 36, 10527–10540.

© Woodhead Publishing Limited, 2013

Electrochemical devices for energy

597

Janowitz K, Kah M and Wendt H (1999), ‘Molten carbonate fuel cell research. Part I. Comparing cathodic oxygen reduction in lithium:potassium and lithium:sodium carbonate melts’, Electrochim Acta, 45, 1025–1037. Jeong K S and Oh B S (2002), ‘Fuel economy and life-cycle cost analysis of a fuel cell hybrid vehicle’, J Power Sources, 105, 58–65. Jones D J and Rozière J (2001), ‘Recent advances in the functionalization of polybenzimidazole and polyetherketone for fuel cell applications’, J Membr Sci, 185, 41–58. Jones D J and Rozière J (2008), ‘Advances in the development of inorganic/organic membranes for fuel cell applications’, Adv Polym Sci, 215, 219–264. Jones D J and Rozière J (2009), ‘Ambient temperature proton electrolyte membrane fuel cells: current trends’ in Encyclopedia of Electrochemical Power Sources, Vol. 2 (C. D. J. Garche, P. Moseley, B. Scrosati, Z. Ogumi, D. Rand, ed.), Amsterdam, Elsevier. Justi E W and Winsel A W (1962), Cold combustion Fuel Cells, Wiesbaden, Steiner. Kendrick E, Islam M S and Slater P R (2008), ‘Atomic-scale mechanistic features of oxide ion conduction in apatite-type germinates’, Chem Commun, 6, 715–717. Kendrick E, Islam M S and Slater P (2005), Electrochemical Society Proceedings, 2005–07, SOFC IX, 2005, pp. 1142–1148. Khadasheva Z S, Venskovskii N U, Safronenko M G, Mosunov A V, Politova E D and Yu Stefanovich S (2002), ‘Synthesis and properties of La2(Mo1-xMx)2O9 (M = Nb, Ta) ionic conductors’, Inorg Mater, 38, 1168–1171. Kim J-E, Patil K Y, Han J, Yoon S-P, Nam S-W, Lim T-H, Hong S-A, Kim H and Lim H-C (2009), ‘Using aluminum and Li2CO3 particles to reinforce the a-LiAlO2 matrix for molten carbonate fuel cells’, Int J of Hydrogen Energy, 34, 9227–9232. Kim Y S, Hickner M A, Dong L, Pivovar B S and McGrath J E (2004), ‘Sulfonated poly(arylene ether sulfone) copolymer proton exchange membranes: composition and morphology effects on the methanol permeability’, J Membr Sci, 243, 317–326. Kim Y-S, Yi C-W, Choi H S and Kim K (2011), ‘Modification of Ni-based cathode material for molten carbonate fuel cells using Co3O4’, J Power Sources, 196, 1886–1893. Kobayashi T, Watanabe H, Hibino M and Yao T (2005), ‘Electrical conductivity of Ba1−xLaxSc1−yZryO3−δ with defective perovskite structure’, Solid State Ionics, 176, 2439–2443. Kordesch K V and Simader G R (1996), Fuel Cells and Their Applications, Germany, VCH. Kuk S T, Song Y S, Suh S I, Kim J Y and Kim K (2001), ‘The formation of LiCoO2 on a NiO cathode for a molten carbonate fuel cell using electroplating’, J Mater Chem, 11, 630–635. Lacorre P, Goutenoire F, Bohnké O, Retoux R and Laligant Y (2000), ‘Designing fast oxide-ion conductors based on La2Mo2O9’, Nature, 404, 856–858. Lafitte B and Jannasch P (2007), ‘On the prospects for phosphonated polymers as proton-exchange fuel cell membranes’, Adv Fuel Cells, 1, 119–185. Lair V, Ringuedé A, Albin V and Cassir M (2008), ‘Characterization in MCFC conditions of high-temperature potentiostatically deposited Co-based thin films on NiO cathode’, Ionics, 14, 555–561. Lalanne C, Prosperi G, Bassat J M, Mauvy F, Fourcade S, Stevens Ph, Zahid M, Van Herle J and Grenier J C (2008), ‘Neodymium-deficient nickelate oxide

© Woodhead Publishing Limited, 2013

598

Handbook of membrane reactors

Nd1.95NiO4+δ as cathode material for anode-supported intermediate temperature solid oxide fuel cells’, J Power Sources, 185, 1218–1224. Lamy C and Léger J M (1994), Les piles à combustible: application au véhicule électrique, Phys IV, 4, C1–253–281. Lapa C M, Figueiredo F M L, de Souza D P F, Song L and Marques F M B (2010), ‘Synthesis and characterization of composite electrolytes based on samaria-doped ceria and Na/Li carbonates’, Int J Hydrogen Energy, 35, 2953–2957. Lee H Y, Hwang H K, Park S S, Choi S W and Shul Y G (2010), ‘Nafion impregnated electrospun polyethersulfone membrane for PEMFC’, Membr J, 20, 40–46. León-Reina L, Losilla E R, Martínez-Lara M, Bruque S and Aranda M A G (2004), ‘Interstitial oxygen conduction in lanthanum oxy-apatite electrolytes’, J Mater Chem, 14, 1142–1149. León-Reina L, Porras-Vázquez J M, Losilla E R and Aranda M A G (2006), ‘Interstitial oxide positions in oxygen-excess oxy-apatites’, Solid State Ionics, 177, 1307–1315. León-Reina L, Porras-Vázquez J M, Losilla E R, Sheptyakov D V, Llobet A and Aranda M A G (2007), ‘Low temperature crystal structures of apatite oxygen-conductors containing interstitial oxygen’, Dalton Trans, 20, 2058–2064. Li Q, Pan C, Jensen J O, Noye P and Bjerrum N J (2007), ‘Cross-linked polybenzimidazole membranes for fuel cells’, Chem Mater, 19, 350–352. Li Q F, Jensen J O, Savinell R F and Bjerrum N J (2009), ‘High temperature proton exchange membranes based on polybenzimidazoles for fuel cells’, Prog Polymer Sci, 34, 449–477. Li S and Sun J (2010), ‘Electrochemical performances of NANOCOFC in MCFC environments’, Int J Hydrogen Energy, 35, 2980–2985. Liu B, Hu W, Robertson G P, Kim Y S, Jiang Z and Guiver M D (2010), ‘Sulphonated Biphenylated Poly(aryl ether ketone)s for Fuel Cell Applications’, Fuel Cells, 10, 45–53. Liu H, Cheng S A, Huang L P and Logan B E (2008), ‘Scale-Lip of membrane-free single-chamber microbial fuel cells’, J Power Sources, 179, 274–279. Liu J, Suraweera N, Keffer D J, Cui S and Paddison S J (2010), ‘On the relationship between polymer electrolyte structure and hydrated morphology of perfluorosulfonic acid membranes’, J Phys Chem C, 114, 11279–11292. Lu S H and Selman J R (1992), ‘Electrode kinetics of oxygen reduction in molten carbonate’, J Electroanal Chem, 333, 257–271. Marrero-Lopez D, Canales-Vazquez J, Ruiz-Morales J C, Irvine J T S and Nunez P (2005), ‘Electrical conductivity and redox stability of La2Mo2−xWxO9 materials’, Electrochim Acta, 50, 4385–4395 Marrero-Lopez D, Martin-Sedeno M C, Pena-Martinez J, Morales J C, Nunez P, Aranda M A G and Ramos-Barradao J R (2010), ‘Evaluation of apatite silicates as solid oxide fuel cell electrolytes’, J Power Sources, 195, 2496–2506. Marestin C, Gebel G, Diat O and Mercier R (2008), ‘Sulfonated polyimides’, Adv Polym Sci, 216, 185–258. Marozau I P, Marrero-Lopez D, Shaula A L, Kharton V V, Tsipis E V, Nunez P and Frade J R (2004), ‘Ionic and electronic transport in stabilized beta-La2Mo2O9 electrolytes’, Electrochim Acta, 49, 3517–3524. Maru H C (1984), Molten Salt Techniques, vol. 2, New York, Plenum Press.

© Woodhead Publishing Limited, 2013

Electrochemical devices for energy

599

Matsuzawa K, Mizusaki T, Mitsushima S, Kamiya N and Ota K-I (2005), ‘The effect of La oxide additive on the solubility of NiO in molten carbonates’, J Power Sources, 140, 258–263. Mauritz K A and Moore R B (2004), ‘State of understanding of Nafion’, Chem Rev, 104, 4535–4585. Mauvy F, Bassat J M, Boehm E, Manaud J P, Dordor P and Grenier J C (2003), ‘Oxygen electrode reaction on Nd2NiO4+δ cathode materials : impedance spectroscopy study’, Solid State Ionics, 158, 17–28. Mendoza L, Albin V, Cassir M and Galtayries A (2003), ‘Electrochemical deposition of Co3O4 thin layers in order to protect the nickel-based molten carbonate fuel cell’, J Electroanal Chem, 548, 95–107. Mendoza L, Baddour-Hadjean R, Cassir M and Pereira-Ramos J P (2004), ‘Raman evidence of the formation of LT-LiCoO2 thin layers on NiO in molten carbonate at 650 °C’, Appl Surf Sci, 225, 356–361. Min B and Logan B E (2004), ‘Continuous electricity generation from domestic wastewater and organic substrates in a flat plate microbial fuel cell’, Environ Sci Technol, 38, 5809–5814. Mitsushima S, Matsuzawa K, Kamiya N and Ota K-I (2002), ‘Improvement of MCFC cathode stability by additives’, Electrochim Acta, 47, 3823–3830. Mohamedi M, Hisamitsu Y, Kihara K, Kudo T, Itoh T and Ushida I (2001), ‘Ni– Al alloy as alternative cathode for molten carbonate fuel cells’, J Alloys and Compd, 315, 224–233. Moon Y and Lee D (2003), ‘Corrosion resistance of 316L stainless steel with surface layer of Ni2Al3 or NiAl in molten carbonates’, J Power Sources, 115, 1–11. Morita H, Komoda M, Mugikura Y, Izaki Y, Watanabe T, Masuda Y and Matsuyama T (2002), ‘Performance analysis of molten carbonate fuel cell using a Li/Na electrolyte’, J Power Sources, 112, 509–518. Nabae Y and Pointon K D (2008), ‘Electrochemical oxidation of solid carbon in hybrid DCFC with solid oxide and molten carbonate binary electrolyte’, Energy Env Sci, 1, 148–155. Nakayama S, Kageyama T, Aono H and Sadaoka Y (1995), ‘Ionic conductivity of lanthanoid silicates, Ln10(SiO4)6O3 (Ln=La, Nd, Sm, Gd, Dy, Y, Ho, Er and Yb)’, J Mater Chem, 5, 1801–1805. Nakayama S, Sakamoto M, Higuchi M, Kodaira K, Sato M, Kakita S, Suzuki T and Itoh K (1999),’Oxide ionic conductivity of apatite type Nd9.33(SiO4)6O2 single crystal’, J Eur Ceram Soc, 19, 507–510. Nedellec Y, Dupont M, Kundler I, Belack J, Rozière J and Jones D J (2010), Progress in Materials for Medium and High Temperature Fuel Cells, La Grande Motte, France. Ng F, Jones D J, Rozière J, Jeske M and Schuster M J (2010), ‘Novel sulfonated poly(arylene ether benzimidazole) Cardo proton conducting membranes for PEMFC’, J Membrane Sci, 362, 184–191. Ni M, Leung M K H and Leung D Y C (2008), ‘Technological development of hydrogen production by solid oxide electrolyzer cell (SOEC)’, Int J Hydrogen Energy, 33, 2337–2354. Nishina T, Takahashi M and Uchida I (1990) ‘Gas electrode reactions in molten carbonate media’, J Electrochem Soc, 137, 1112–1121.

© Woodhead Publishing Limited, 2013

600

Handbook of membrane reactors

O’Brien J E, McKellar M G, Harvego E A and Stoots C M (2010), ‘High-temperature electrolysis for large-scale hydrogen and syngas production from nuclear energy – summary of system simulation and economic analyses’, Int J Hydrogen Energy, 35, 4808–4819. Obara S (2010), ‘Study of a water electrolysis system using a compact solar cell module with a plant shoot configuration’, Int J Hydrogen Energy, 35, 26–36. Ota K-I, Matsuda Y, Matsuzawa K, Mitsushima S and Kamiya K (2006) , ‘Effect of rare earth oxides for improvement of MCFC’, J Power Sources, 160, 811–815. Paetsch L M, Doyon J and Farooque M (1993), Proceedings of the Third International Symposium of Carbonate Fuel Cell Technology, Pennington, The Electrochemical Society. Paoletti C, Carewska M, Lo Presti R and Mc Phail S (2009), ‘Performance analysis of new cathode materials for molten carbonate fuel cells’, J Power Sources, 193, 292–297. Park C-H, Lee C-H, Guiver M D and Lee Y-M (2011), ‘Sulfonated hydrocarbon membranes for medium-temperature and low-humidity proton exchange membrane fuel cells (PEMFCs)’, Prog Polym Sci, 36, 1443–1498. Parvole J and Jannasch P (2008), ‘Poly(arylene ether sulfones) with phosphonic acid and bis phosphonic acid on short alkyl side chains for PEM’, J Mater Chem, 18, 5547–5556. Patrakeev M V and Marques F M B (2004), ‘Ionic and electronic conductivity of La9.83-xPrxSi4.5Fe1.5O26 +/-δ apatites’, Solid State Ionics, 171, 51–59. Pauporté T, Mendoza L, Cassir M, Galtayries A and Chivot J (2005), ‘Direct low temperature deposition of crystallized Co3O4 films by potentiostatic electrolysis’, J Electrochem Soc, 152, C49–C53. Peckham T J and Holdcroft S (2011), ‘Structure-morphology-property relationships of non-perfluorinated proton conducting membranes’, Adv Mater, 22, 4667–4690. Perez F J, Duday D, Hierro M P, Gomez C, Agüero A, Garcia M C, Muela R, Sanchez Pascual A and Martinez L (2002), ‘Hot corrosion study of coated separator plates of molten carbonate fuel cells by slurry aluminides’, Surf Coat Technol, 161, 293–301. Peron J, Nedellec Y, Jones D J and Rozière J (2008), ‘The effect of dissolution, migration and precipitation of platinum in Nafion-based membrane electrode assemblies during fuel cell operation at high potential’, J Power Sources, 185, 1209–1217 Peron J, Mani A, Zhao X, Edwards D, Adachi M, Soboleva T, Shi Z, Xie Z, Navessin T and Holdcroft S (2010), ‘Properties of Nafion® NR-211 membranes for PEMFCs’, J Membrane Sci, 356, 44–51. Pointon K D, Lakerman B, Irvine J T S, Bradley J and Jain S (2006), ‘The development of a carbon–air semi fuel cell’, J Power Sources, 162, 765–772. Ranran P, Yan W, Lizhai Y and Zongqiang M (2006), ‘Electrochemical properties of intermediate-temperature SOFCs based on proton conducting Sm-doped BaCeO3 electrolyte thin film’, Solid State Ionics, 177, 389–393. Raza R, Wang X, Ma Y, Liu X and Zhu B (2010), ‘Improved ceria–carbonate composite electrolytes’, Int J Hydrogen Energy, 35, 2684–2688. Ringuedé A, Wijayasinghe A, Albin V, Lagergren C, Cassir M and Bergman B (2006), ‘Solubility and Electrochemical studies of LiFeO2-LiCoO2-NiO materials for the MCFC cathode application’, J Power Sources, 160, 789–795.

© Woodhead Publishing Limited, 2013

Electrochemical devices for energy

601

Rozière J and Jones D J (2003), ‘Non-fluorinated polymer materials for proton exchange membrane fuel cells’, Annu Rev Mater Res, 33, 503–555. Sammells A F, Cook R L, White J H, Osborne J J and MacDuff R C (1992), ‘Rational selection of advanced solid electrolytes for intermediate temperature fuel cells’, Solid State Ionics, 52, 111–123. Sangster M and Pelton A D (1987), Special Report to the Phase Equilibria Program,Westerville, Ohio, American Ceramic Society. Sansom J E H, Richings D and Slater P R (2001), ‘A powder neutron diffraction study of the oxide-ion-conducting apatite-type phases, La9.33Si6O26 and La8Sr2Si6O26’, Solid State Ionics, 139, 205–210. Scaccia S (2005), ‘Investigation on NiO solubility in binary and ternary molten alkalimetal carbonates containing additives’, J Mol Liq, 116, 67–71. Schmidt T J (2009), Polymer Electrolyte Fuel Cell Durability (F. N. Büchi, M. Inaba, and T. J. Schmidt, eds.), New York, Springer Science. Schober T (2003), ‘Applications of oxidic high-temperature proton conductors’, Solid State Ionics, 162–163, 277–281. Schuster M, Rager T, Noda A, Kreuer K D and Maier J (2005), ‘About the choice of the protogenic group in PEM separator materials for intermediate temperature, low humidity operation: A critical comparison of sulfonic acid, phosphonic acid and imidazole functionalized model compounds’, Fuel Cells, 5, 355–365. Seel D C, Benicewicz B C, Xiao L and Schmidt T J (2009), Handbook of Fuel Cells, Vol. 5 (W. Vielstich, H. Yokokawa, and H. A. Gasteiger, eds.), New York, John Wiley & Sons. Selman J R and Maru H C (1981), Advances in Molten Salt Chemistry, vol. 4, New York, Plenum Press. Shaula A L, Kharton V V and Marques F M B (2005), ‘Oxygen ionic and electronic transport in apatite-type La10-x(Si,Al)6O’, J Solid State Chem, 178, 2050–2061. Shaula A L, Kharton V V and Marques F M B (2006), ‘Ionic and electronic conductivities, stability and thermal expansion of La10-x(Si,Al)6O solid electrolytes’, Solid State Ionics, 177, 1725–1728. Slater P R, Sansom J E H and Tolchard J R (2004), ‘Development of apatite-type oxide ion conductors’, Chem Rec, 4, 373–384. Stevens P, Novel-Cattin F, Hammou A, Lamy C and Cassir M (2000), Piles à combustible, Paris, Techniques de l’Ingénieur, traité de génie électrique. Stoica D, Ogier L, Akrour L, Alloin F and Fauvarque J-F (2007), ‘Anionic membrane based on polyepichlorhydrin matrix for alkaline fuel cell: synthesis, physical and electrochemical properties’, Electrochimica Acta, 53, 1596–1603. Stoots C M, O’Brien J E, Condie K G and Hartvigsen J (2010), ‘High-temperature electrolysis for large-scale hydrogen production from nuclear energy – Experimental investigations’, Int J Hydrogen Energy, 35, 4861–4870. Su X-T, Yan Q-Z, Ma Y-H, Zhang W-F and Ge C-C (2006), ‘Effect of co-dopant addition on the properties of yttrium and neodymium doped barium cerate electrolyte’, Solid State Ionics, 177, 1041–1045. Subasri R, Matusch D, Nafe H and Aldinger F (2004), ‘Synthesis and characterization of (La1-x Mx)2Mo2O9-δ ; M=Ca2+, Sr2+ or Ba2+’, J Eur Ceram Soc, 24, 129–137. Subianto S, Cavaliere S, Jones D J and Rozière J (2013), ‘Effect of side chain length on the electrospinning of perfluorosulfonic acid ionomers’, Polymer, 51, 118–128.

© Woodhead Publishing Limited, 2013

602

Handbook of membrane reactors

Suguira K, Takei K, Tanimoto K and Miyazaki Y (2003), ‘The carbon dioxide concentrator by using MCFC’, J Power Sources, 118, 218–227. Sukumar P R, Wu W, Markova D, Ünsel Ö, Klapper M and Müllen K (2007), ‘Functionalized poly(benzimidazole)s as membrane materials for fuel cells’, Macromol Chem Phys, 208, 2258–2267. Taherparvar H, Kilner J A, Baker R T and Sahibzada M (2003), ‘Effect of humidification at anode and cathode in proton-conducting SOFCs’, Solid State Ionics, 162–163, 297–303. Tanase S, Miyazaki Y, Yanagida M, Tanimoto K and Kodama T (1987), Progress in Batteries and Solar Cells, 6, 195–200, Cleveland, Ohio, JEC Press, Inc. Tanimoto K, Kojima T, Yanagida M, Nomura K and Miyazaki Y (2003), ‘Optimization of the electrolyte composition in a (Li0.52 Na0.48)2−2x AExCO3 (AE = Ca and Ba) molten carbonate fuel cell’, J Power Sources, 131, 256–260. Tao S W and Irvine J T S (2006), ‘A stable, easily sintered proton-conducting oxide electrolyte for moderate-temperature fuel cells and electrolyzers’, Adv Mater, 18, 1581–1584. Tchicaya-Bouckary L, Jones D J and Rozière J (2002), ‘Hybrid polyaryletherketone membranes for fuel cell applications’, Fuel Cells, 2, 40–45. Tealdi C, Chiodelli G, Malavasi L and Flor G (2004), ‘Effect of alkaline-doping on the properties of La2Mo2O9 fast oxygen ion conductor’, J Mater Chem, 14, 3553–3557. Tilak B V, Yeo R S and Srinivasan S (1981), Electrochemical Energy Conversion and Storage, in “Comprehensive Treatise of Electrochemistry”, J. O’M. Bockris, B.E. Conway, E. Yeager and R.E. White (Eds), vol.3, New York, Plenum Press. Tolchard J R, Islam M S and Slater P R (2003), ‘Defect chemistry and oxygen ion migration in the apatite-type materials La9.33Si6O26 and La8Sr2Si6O26’, J Mater Chem, 13, 1956–1961. Tomczyk P, Sato H, Yamada K, Nishina T and Uchida I (1995), ‘Oxide electrodes in molten carbonates. Part 2. Electrochemical behaviour of cobalt in molten Li+K and Na+K carbonate eutectics’, J Electroanal Chem, 391, 133–139. Trasatti S (1999), ‘Water electrolysis: who first?’, J Electroanal Chem, 476, 90–91. Trémillon B (1997), Reactions in solution, an applied analytical approach, New York, John Wiley & Sons. Trogadas P, Parrondo J, Mijangos F and Ramani V (2011a), ‘Degradation mitigation in PEM fuel cells using metal nanoparticle additives’, J Mater Chem, 21, 19381–19388. Trogadas P, Parrondo J, Mijangos F and Ramani V (2011b), ‘Platinum supported on CeO2 effectively scavenges free radicals within the electrolyte of an operating fuel cell’, Chem Commun, 47, 11549–11551. Tsipis E V, Kharton V V and Frade J R (2007), ‘Electrochemical behavior of mixed-conducting oxide cathodes in contact with apatite-type La10Si5AlO26.5 electrolyte’, Electrochim Acta, 52, 4428–4435. Ulleberg Ø, Nakken T and Eté A (2010), ‘The wind/hydrogen demonstration system at Utsira in Norway: Evaluation of system performance using operational data and updated hydrogen energy system modeling tools’, Int J Hydrogen Energy, 35, 1841–1852.

© Woodhead Publishing Limited, 2013

Electrochemical devices for energy

603

Vandenborre H, Leysen R and Baetsle L H (1980), ‘Alkaline inorganic-membrane-electrolyte (IME) water electrolysis’, Int J Hydrogen Energy, 5, 165–171. Vega-Castillo J, Mogni L, Corbel G, Lacorre P and Caneiro A (2010), ‘On the thermodynamic stability of La2Mo2O oxide-ion conductor’, Int J Hydrogen Energy, 5, 5890–5894. Vetter K J (1967), Electrochemical Kinetics: Theoretical and Experimental Aspects, New York, Academic Press. Wade J L and Lackner K S (2007), ‘Transport model for a high temperature, mixed conducting CO2 separation membrane’, Solid State Ionics, 178, 1530–1540. Wainright J S, Litt M H and Savinell R F (1994), ‘Acid doped PBI, a new polymer electrolyte’, Proc Electrochem Soc, 94, 255–264. Wainright J S, Wang J-T, Savinell R F, Litt M, Moaddel H and Rogers C (2003), Handbook of Fuel Cells, 3, 436–446. Wang X P, Fang Q F, Li Z S, Zhang G G and Yi Z G (2002), ‘Dielectric relaxation studies of Bi-doping effects on the oxygen-ion diffusion in La2-xBixMo2O9 oxide-ion conductors’, Appl Phys Lett, 81, 3434–3436. Weever R, Hemmes K and de Wit J H W (1995), ‘The mechanism of hydrogen oxidation at gold and nickel flag electrodes in molten Li/K carbonate’, J Electrochem Soc, 142, 389–397. Wendt H, Böhme O, Leidich F U and Brenscheidt T (1993), Proc. of the 3rd Int. Symp. on Carbonate Fuel Cell Technology, The Electrochemical Society Proceedings Series 93–3, Massachussets, The Electrochem. Soc. Inc. Wendt H and Hofmann H (1985), ‘Cermet diaphragms and integrated electrode-diaphragm units for advanced alkaline water electrolysis’, Int J Hydrogen Energy, 10, 375–381. Wijayasinghe A, Bergman B and Lagergren C (2004), ‘A study on LiCoO2-rich cathode materials for the MCFC based on the LiCoO2–LiFeO2–NiO ternary system’, Electrochim Acta, 49, 4709–4717. Wu D, Paddison S J and Elliott J A (2009), ‘Effect of Molecular Weight on Hydrated Morphologies of the Short-Side-Chain Perfluorosulfonic Acid Membrane’, Macromolecules, 42, 3358–3367. Xia C, Li Y, Tian Y, Liu Q, Wang Z, Jia L, Zhao Y and Li Y (2010), ‘Intermediate temperature fuel cell with a doped ceria–carbonate composite electrolyte’, J Power Sources, 195, 3149–3154. Xiao L, Zhang H, Scanlon E, Ramanathan L S, Choe E -W, Rogers D, Apple T and Benicewicz B C (2005), ‘High-temperature polybenzimidazole fuel cell membranes via a Sol-Gel process’, Chem Mater, 17, 5328–5333. Xing B and Savadogo O (1999), ‘The effect of acid doping on the conductivity of polybenzimidazole (PBI)’, J New Mater Electrochem Syst, 2, 95–101. Yamada K and Uchida I (1994), ‘In-situ formation process of LiCoO2 in the molten lithium-potassium carbonate eutectic at 923 K’, Chem Lett, 2, 299–302. Yandrasits M (2011), Advances in New Materials for Proton Electrolyte Fuel Cell Systems, CA, Asilomar. Yang L, Zuo C, Wang S, Cheng Z and Liu M (2008), ‘A novel composite cathode for low-temperature SOFCs based on oxide proton conductors’, Adv Mater, 20, 3280–3283.

© Woodhead Publishing Limited, 2013

604

Handbook of membrane reactors

Yaremchenko A A, Kharton V V, Bannikov D O, Znosak D V, Frade J R and Cherepanov V A (2009), ‘Performance of perovskite-related oxide cathodes in contact with lanthanum silicate electrolyte’, Solid State Ionics, 180, 878–885. Yaremchenko A A, Shaula A L, Kharton V V, Waerenborgh J C, Rojas D P, Patrakeev M V and Marques F M B (2004), ‘Ionic and electronic conductivity of La9.83-xPrxSi4.5Fe1.5O26 +/-δ apatites’, Solid State Ionics, 171, 51–59. Yoshiba F, Morita H, Yoshikawa M, Mugikura Y, Izaki Y, Watanabe T, Komoda M, Masuda M and Zaima N (2004), ‘Improvement of electricity generating performance and life expectancy of MCFC stack by applying Li/Na carbonate electrolyte. Test results and analysis of 0.44m2/10 kW- and 1.03m2/10 kW-class stack’, J Power Sources, 128, 152–164. Yoshioka H, Nojiri Y and Tanase S (2008), ‘Ionic conductivity and fuel cell properties of apatite-type lanthanum silicates doped with Mg and containing excess oxide ions’, Solid State Ionics, 179, 2165–2169. Yuh C, Farooque M and Johnsen R (1992), Proceedings of the Fourth Annual Fuel Cells Contractors Review Meeting, U.S. DOE/METC. Yuh C, Johnsen R, Farooque M and Maru H (1995), ‘Status of carbonate fuel cell materials’, J Power Sources, 56, 1–10. Zecevic S, Patton E M and Parhami P (2004), ‘Carbon–air fuel without a reforming process’, Carbon, 42, 1983–1993. Zhang L, Lan R, Petit C T G and Tao S (2010), ‘Durability study of an intermediate temperature fuel cell based on an oxide–carbonate composite electrolyte’, Int J Hydrogen Energy, 35, 6934–6940. Zheng C H, Oh C E, Park Y I and Cha S W (2011), Int J Hydrogen Energy, doi: 10.1016/j.ijhydrogene.2011.09.147 Zhou M, Chi M, Lio J, He H and Jin T (2011), ‘An overview of electrode materials in microbial fuel cells’, J Power Sources, 196, 4427–4435. Zhu B and Mat M D (2006), ‘Studies on dual phase ceria-based composites in Electrochemistry’, Int J Electrochem Sci, 1, 383–402. Zhu B, Li S and Mellander B (2010), ‘Theoretical approach on ceria-based two-phase electrolytes for low temperature (300–600 °C) solid oxide fuel cells’, Electrochem Commun, 10, 302–305.

15.11 Appendix: nomenclature 15.11.1 aq g l s Δr G Δr H Ea Ean Ecat

Notation aqueous phase gas phase liquid phase solid phase free enthalpy of reaction (kJ.mol−1) enthalpy of reaction (kJ.mol−1) activation energy (eV) potential at the anode at equilibrium (V) potential at the cathode at equilibrium (V)

© Woodhead Publishing Limited, 2013

Electrochemical devices for energy potential at the anode under j (V) potential at cathode under j (V) electromotive force (V) current density (A.cm−2) current density at the anode (A.cm−2) current density at the cathode (A.cm−2) exchange current density at equilibrium (A.cm−2) anode overpotential (V) cathode overpotential (V) charge transfer coefficient conductivity (S.cm−1) efficiency of a reversible fuel cell Faradaic efficiency of a fuel cell potential efficiency of a reversible fuel cell electrolyte resistance (Ω) resistance due to charge transfer (Ω) temperature (K, °C) tension of a half-cell (V) reaction rate (mol.cm2.s−1)

Ean(j) Ecat(j) emf j jan jcat j0 ηan ηcat α σ εrev εF εE Rel Rct T U v

15.11.2

Abbreviations

AEC AFC BCZYZ BM CCS DCFC DMFC EW FCHV H2-FCV HDCFC ICE IT-SOFC LAMOX LSC LSM MCFC MEA MFC NANOSOFC

alkaline electrolysis cell alkaline fuel cell BaCe0.5Zr0.3Y0.16Zn0.04O3-δ bipolar membrane carbon capture and storage direct carbon fuel cell direct membrane fuel cell equivalent weight fuel cell hybrid vehicle hydrogen fuel cell vehicle hybrid direct fuel cell internal combustion engine intermediate temperature solid oxide fuel cell La2Mo2O9 family long-side chain lanthanum strontium manganite molten carbonate fuel cell membrane electrode assembly microbial fuel cell nano-SOFC

© Woodhead Publishing Limited, 2013

605

606

Handbook of membrane reactors

NASA PAFC PBI PCFC PEFC PEMEC PEMFC PFSA PPA PTFE PVPA RH ScSZ SDC SOEC SOFC SPE SSC TEC TPB YSZ

National Aeronautics and Space Administration phosphoric acid fuel cell polybenzimidazole proton conductor fuel cell proton electrolyte fuel cell proton exchange membrane electrolysis cell proton exchange membrane fuel cell perfluorosulfonic acid polyphosphoric acid polytetrafluoroethylene poly(vinylphosphonic acid) relative humidity scandia-stabilised zirconia samarium-doped ceria solid oxide electrolysis cell solid oxide fuel cell solid polymer electrolyte short-side chain thermal expansion coefficient triple phase boundary yttria-stabilised zirconia

© Woodhead Publishing Limited, 2013

16 Palladium-based hollow cathode electrolysers for hydrogen production A. POZIO and S. TOSTI , ENEA, Italy

DOI: 10.1533/9780857097347.3.607 Abstract: Ultra-pure hydrogen can be obtained through the use of water electrolysers with dense metal cathodes. In this chapter, the theory of hydrogen evolution over Pd-based cathodes and its permeation through dense metal walls is introduced, and the Damköhler−Péclet (DaPe) analysis for these membrane reactors is described. Optimization of the electrolysis process is considered, based on the relationship of DaPe number with the cell overpotential and the apparent activation energy of permeation. Literature studies on hollow cathode water electrolysers are discussed, as are new prototypes using thin-wall Pd–Ag permeator tubes, and the efficient operation of cell efficiency and surface activation/ deactivation phenomena. The main applications of cells with Pd-based hollow cathodes are the production of ultra-pure hydrogen at high pressure for small scale laboratory electrolysers and the treatment of tritiated water. The results of tests on prototypes are examined in detail, and attention is given to future trends in material development. Key words: palladium membranes, thin-wall Pd–Ag tubes, hydrogen permeation, Damköhler−Péclet analysis, alkaline electrolysers.

16.1

Introduction

Hydrogen is widely considered to be a vector for clean energy. In fact, hydrogen is not naturally available, though it can be obtained from renewable sources. If hydrogen is ‘burnt’ to produce energy then water is the only by-product, and so no greenhouse gases are emitted.1–5 Most hydrogen is currently produced by reforming methane, which produces significant quantities of CO2.6–8 Recent studies have considered the use of biomass feedstocks such as bio-ethanol for producing hydrogen via reforming. In these processes, the release of CO2 into the environment can be greatly reduced, and the entire cycle could be considered at least CO2-neutral.9–11 Due to its high cost, only small amounts of hydrogen are produced via electrolysis of water, the subject of this chapter. However, electrolysis can be useful

607 © Woodhead Publishing Limited, 2013

608

Handbook of membrane reactors

when a very high purity of hydrogen is required (e.g., for laboratory application or chemicals) or if large amounts of low cost, renewable electricity are available.12 Electrolysis of aqueous alkaline hydroxide solutions has historically been one of the most popular methods for obtaining hydrogen and oxygen from water. Notable recent studies concerned with the improvement of alkaline electrolysis processes have focused on the use of advanced anode materials in order to reduce electrode polarization, and thereby to improve the overall electrical efficiency of the electrolysers.13–15 However, for several applications, the hydrogen produced via water electrolysis requires purification. It has been understood for many years that electrolytic hydrogen can diffuse through the walls of the tubular palladium cathode against which it is generated. This hitherto neglected possibility of simultaneous generation and purification has been utilized in the design of some compact electrolytic cells, which can yield ultra-pure hydrogen suitable for laboratory or other small scale applications. The first experiment on the electrolytic transmission of hydrogen through palladium was reported by Wahlin.16,17 Hydrogen pressures of 48 bars were built up via electrolysis, and it was shown that the diffused hydrogen was very catalytically active. Later investigations have shown that pure palladium is not an ideal cathode material for electrolytic cells, due to its unsatisfactory mechanical properties and limited diffusion rate for hydrogen. Better results were obtained with the 25% silver−palladium alloy, which expanded less than pure Pd when fully charged with hydrogen. Using this cathode, Darling generated small quantities of ultra-pure hydrogen by electrolytic diffusion in a small prototype electrolyser equipped with 76 µm thick cathodes.18 The author claimed hydrogen transmission efficiency of the order of 95% in dilute sulfuric acid electrolyte at 80°C with current densities up to 0.3 A cm−2. This acid electrolyser was capable of producing 2 L min−1 of hydrogen containing less than one part per million of impurities. The hydrogen permeation efficiency did not decrease with increasing membrane thickness, though it naturally took longer for the hydrogen to permeate a thicker membrane. In addition, electrolysis in acid media required an expensive iridium based anode. At the same time, Clifford et al. proposed a gravity-independent water vapour alkaline electrolysis cell based on a Pd–Ag alloy cathode, and demonstrated its feasibility.19 In this investigation a Pd–Ag cathode with a wall thickness of 125 µm was used in 50% NaOH electrolyte at 145°C. The cell operated at 2.0 V and at about 81 mA cm−2 with a 97–100% hydrogen transmission, but only for a limited operating period. Further applications have been studied by Lovelock et al. for producing ultra-pure hydrogen at high pressure20 and by Bellanger for reducing tritiated water.21 A more detailed description of these studies is provided

© Woodhead Publishing Limited, 2013

Pd-based hollow cathode electrolysers for hydrogen production

609

in Section 16.4. The beneficial aspects of low temperature permeation of high purity hydrogen are significant. A recent study explored the possibility of using a thin Pd 75%−Ag 25% alloy tubular cathode of wall thickness 50 µm and a nickel anode directly in a KOH electrolyte at 70–80°C as a source of high purity hydrogen,22 in order to realize the laboratory scale prototype alkaline electrolyser described in Section 16.3.2. By using a dense Pd-based permeator tube as the hollow cathode of a water electrolyser, it is possible to recover a pure hydrogen stream permeated through the membrane. Analysis of this electrolytic process must take into account both the mass transfer of hydrogen through a dense palladium alloy wall of the hollow cathode and the electrolytic reactions.

16.2

Theory

16.2.1

Hydrogen permeation through Pd–Ag alloys

Hydrogen can permeate selectively dense metal membranes, behaviour that permits the separation of hydrogen from gas mixtures. The mass transfer mechanism consists of several steps: dissociation of hydrogen molecules into atoms, interaction of hydrogen atoms with the metal surface and their adsorption, diffusion of hydrogen into the metal lattice, and desorption of hydrogen atoms from the other metal surface and their recombination into molecules.23–25 The overall transport process through the metal wall is called permeation and is ruled by the expression: J

P

pua

pda

[16.1]

δ

where J is the hydrogen permeation flux (mol m−2 s−1), P is the hydrogen permeability coefficient (mol m−1 s−1 Pa−n), δ is the membrane thickness (m), and pu and pd are the hydrogen partial pressures at the upstream and downstream sides (Pa), respectively. The hydrogen permeation flux is inversely proportional to the membrane wall thickness, while the dependence on the hydrogen partial pressures is determined by the exponent a, which varies in the range 0.5–1. When the diffusion of hydrogen into the metal is the controlling mass transfer step, the pressure exponent is 0.5, and the hydrogen permeation flux is given by the following formula:

J

P

pu

pd

δ

© Woodhead Publishing Limited, 2013

[16.2]

610

Handbook of membrane reactors

The effect of temperature is considered by introducing the dependence of the permeability coefficient from the temperature according to an Arrhenius-kind expression: P

Po e

− Ea

[16.3]

RT

where Po is the permeability pre-exponential coefficient (mol m−1 s−1 Pa−0.5), Ea the apparent activation energy (J mol−1), R the gas constant (8.31 J mol−1 K−1) and T the absolute temperature (K). The Richardson’s expression is obtained by combining Equations [16.2] and [16.3]: J

16.2.2

Po e

− Ea

RT

pu

pd

[16.4]

δ

Hydrogen production through a dense Pd–Ag hollow cathode

In order to analyse hydrogen production through a Pd–Ag cathode, it is necessary to introduce some basic principles of alkaline water electrolysis. The half reactions occurring on the cathode and anode of alkaline electrolysers,26 respectively, can be written as: 2OH 2H 2 O

1

2 2e

2 O2

+ H 2 O + 22e −

E°

0 401 V

[16.5]

H 2 + 2OH

E° = −0.827 V

[16.6]

The overall chemical reaction of water electrolysis can be written as: H2O → H2 + ½ O2

E° = +1.23 V

[16.7]

The hydrogen evolution reaction (HER) mechanism on a metallic cathode (M) in alkaline media is widely accepted to involve the formation of adsorbed hydrogen (Volmer mechanism):27 H2O + e− + M → M-Hads + OH−

[16.8]

which is followed by either electrochemical desorption (Heyrovsky mechanism): H2O + e− + M-Hads → H2↑ + OH−

© Woodhead Publishing Limited, 2013

[16.9]

Pd-based hollow cathode electrolysers for hydrogen production

611

or chemical desorption (Volmer mechanism): M-Hads + M-Hads → H2↑ + 2 M

[16.10]

where the subscript ‘ads’ represents the adsorbed status. In addition, the dissolution of absorbed hydrogen in the metal lattice on the side immersed in the electrolyte occurs on a Pd–Ag cathode:28 Ag/Pd-H2ads ↔ Ag/Pd-H2diss

[16.11]

followed by hydrogen diffusion through the cathode wall and its chemical desorption similar to Equation [16.9] on the other side: Ag/Pd-Hads + Ag/Pd-Hads → H2↑ + 2 Ag/Pd

[16.12]

So, the HER on a Pd–Ag electrode involves several stages: (a) (b) (c) (d)

generation of H2 by electrolysis of H2O adsorption of H2 on the Pd–Ag alloy cathode membrane H2 diffusion through the Pd–Ag desorption of H2 from the other side of the palladium−silver cathode.

The first stages (a, b, c) are the most critical, since desorption (d) occurs freely on the exit face. In order to maximize H2 permeation through the metal wall, with respect to the total amount of hydrogen produced by electrolysis, several parameters must be considered. Experimentally, Reaction [16.11] is found to be a faster process than diffusion in the metal,27,29 and is therefore considered to be in equilibrium in the overall permeation process. So, the flux of H entering a metal lattice is proposed to be proportional to the coverage (θ) on the surface: CO = (kd/k-d) θ = K θ

[16.13]

where CO is the hydrogen solubility, kd and k-d are the kinetics direct and inverse constants, and K is the equilibrium constant of Reaction [16.11]. The current density (i) of steady state permeation, therefore, is directly proportional to θ. Further, the coverage is strictly dependent on the overpotential and the mechanism of HER. The overpotential of hydrogen η is generally measured by the Tafel equation:

η = 2.3

RT i log αF io

© Woodhead Publishing Limited, 2013

[16.14]

612

Handbook of membrane reactors

where α is the transfer coefficient, F the Faraday constant, R the gas constant, i the current density (A m−2) and T the absolute temperature. In this equation, io, the exchange current density of the reaction, which can be analogized as the rate constant of reaction, is a function of the nature of the electrode (cathode) material. The overpotential of hydrogen production results in an extra energy barrier in the hydrogen formation process. The overpotential on the cathode is directly related to the formation of hydrogen in the vicinity of the electrode. The formation of hydrogen is intrinsically determined by the bond between hydrogen and the electrode surface, which can be expressed by the heat of activation. The path of the HER mechanism taking place on Pd was characterized by a fast discharge step (8) and a slow recombination (9 or 10).27 On the opposite, the rate of hydrogen evolution at the Ag cathodes in alkaline solutions is controlled by a slow discharge from water molecules.30 The behaviour of a Pd–Ag (75:25) alloy tends to approach a coupled discharge−combination mechanism.27,31,32 For the HER, for example, on Pd–Ag, it was observed32 that the exchange current density shows a maximum value, and the overpotentials a minimum value, for alloy compositions containing about 60% Pd (atomic % on the surface), see Fig. 16.1. Another advantage of alloying is that the mechanical strength can be higher than for pure palladium; the addition of silver increases the tensile strength and hardness of palladium by up to 20–40%.33 In general, a palladium membrane becomes brittle after a certain number of cycles of transformation between α ↔ β-phases, due to the accompanied lattice expansion.

−log io (A cm−2 at ER)

8 7 6 5 4 3 0

10

20

30 40 50 60 70 Alloy composition (at% Pd)

80

90

100

16.1 Exchange current density versus the composition of the Pd–Ag alloy in 1 N H2SO4–1 M Na2SO4 solution (pH 0.7; T = 25°C). Reprinted from Surface Technology, 15, Bélanger A., Vijh A K., The hydrogen evolution reaction on Ag–Pd alloys: influence of electronic properties on electroactivity, pp. 59–78, Copyright (1982), with permission from Elsevier.

© Woodhead Publishing Limited, 2013

Pd-based hollow cathode electrolysers for hydrogen production

613

In palladium–silver alloys, larger amounts of hydrogen can be solubilised because the metal lattice has already been expanded by the silver atoms, and thus it is less brittle than the pure Pd lattice. Based on these considerations, the ideal silver content of the Pd cathode is between 20 wt% and 30 wt%.34 In order to obtain high flow rates of pure hydrogen, the maximum permeability of the Pd–Ag cathode must be realized. The hydrogen permeability P through the lattice of the Pd–Ag tubular electrode follows Sieverts’ law.21,23 The H2 permeating flow rate QF (mol s−1) can be derived from Equation [16.2]: QF

PA P

pu

pd

δ

[16.15]

In this case pu and pd are respectively the upstream (external cathode tube-electrolyte side) and downstream (cathode tube lumen side) pressure and A is the membrane area (m2). The value of Q tends to zero when pd is equal to pu. In this case, the exit pressure pd is the atmospheric pressure, and the input pressure depends on the overpotential η. The possible pressure that can be developed inside the cathode, when the hydrogen is electrochemically generated, depends upon the mechanism of HER. For a coupled discharge−combination mechanism, the fugacity of H2 (Pa) can be expressed by the following equation:21,27 f = 101 5 exp

− ηF 2 RT

[16.16]

The H2 fugacity can be related to the sensible pressure by the well-known thermodynamic equation: p

RT (

f

f0 ) =

d ∫ v dp

[16.17]

p0

where v is the molar volume (m3 mol−1). The fugacity of hydrogen rises faster than its sensible pressure. For an instance, the pressure corresponding to a fugacity of 1010 atm is only 104 atm.27,35 For low fugacity values the upstream pressure can be approximated by the relation: pu

K ′ exp

− ηF 2 RT

where K′ is a numerical constant.

© Woodhead Publishing Limited, 2013

[16.18]

614

Handbook of membrane reactors

As shown, the cathode charging method introduces a very high ‘fictitious’ hydrogen input pressure, which depends on the overpotential. The electrolyte concentration of alkaline hydroxide also affects the permeation when the system is not limited by the other parameters. The rate of the discharge reaction can be derived from the following kinetic equation: i

F OH exp FkC

( − )ηF RT

[16.19]

where COH is the molar concentration (mol L−1) of alkaline hydroxide and k is a constant. According to Faraday’s law, the number of moles of the electrolysed species (H2 or O2), N, is given by: N=

Q nF

[16.20]

where Q is the total electrical charge (coulomb) transferred during the reaction (n = 2 for both Reactions [16.5] and [16.6]). The rate of electrolysis r (cm3 min−1) can be expressed as: r=

dN dt

[16.21]

From Equation [16.20] it results that dN 1 dQ = dt nF dt and dQ/dt can be noted as Faradic current i(A).36 Then r=

dN i = dt nF

The surface area over which the reaction takes place needs to be taken into account. The rate of the electrolysis reaction can be expressed as: r=

i nF

[16.22]

The permeation yield is an important parameter defining the tubular membrane’s capacity to separate pure hydrogen. It is defined as the ratio of the hydrogen permeated through the tube cathode to the overall hydrogen

© Woodhead Publishing Limited, 2013

Pd-based hollow cathode electrolysers for hydrogen production

615

produced by electrolysis. The permeation yield can be obtained simply by measuring, over the same time, the hydrogen volume passed through the Pd–Ag cathode as: Yield =

V (H 2 permeated) V (H 2 total)

[16.23]

This ratio is equivalent to the inverse of the Damköhler−Péclet number usually proposed for the preliminary design and optimization study of membrane reactors.37–40

16.2.3

Damköhler−Péclet analysis

In membrane reactors, the ‘DaPe’ refers to a dimensionless parameter representing the ratio of the reaction to the permeation rate. In a well-designed membrane reactor, DaPe should be close to 1, since under ideal conditions all the reaction product permeates through the membrane. When the DaPe is greater than 1 the reaction rate overcomes the permeation rate: this is the case, for instance, when optimizing the membrane reactor requires an increase in permeation area or a reduction in catalyst volume, and vice versa when the DaPe is less than 1. The water electrolyser described in the previous section functions like a membrane reactor − producing hydrogen over the Pd-based tubular cathode through Reaction [16.6]. The retentate stream consists of non-permeated hydrogen plus oxygen, while the pure hydrogen separated and collected in the membrane lumen is the permeate stream. A Damköhler−Péclet analysis could therefore provide a useful preliminary study of the behaviour of this type of electrolyser, by varying the main operating parameters. The DaPe number is the ratio of the overall hydrogen produced by electrolysis (as given by Equation [16.22]) to the hydrogen permeated (as given by Equation [16.15]). By ignoring the hydrogen pressure in the cathode lumen (i.e., pd 0, DaPe decreases as the temperature increases, because the permeability coefficient increases more than the term pu, and vice versa. The apparent activation energy Ea is positive for most metals, including Pd, Ni, and Fe, but negative for refractory metals such as Nb, V and Ta. This can be recognized in Fig. 16.2, where the hydrogen permeability of certain metals is plotted against the inverse of the absolute temperature.41 In metals where Ea < 0, from Equation [16.25] it is evident that DaPe always increases with the temperature. In fact, the overpotential is negative, too. For metals with Ea > 0, the DaPe number decreases with the temperature when operating at low overpotential 4Ea > |ηF|, as in the example reported below. © Woodhead Publishing Limited, 2013

Pd-based hollow cathode electrolysers for hydrogen production

617

100 –100 mV –200 mV

DaPe

–300 mV 10

1

0

20

40

60

80

100

T (°C)

16.3 DaPe number versus the temperature for three different values of the overpotential.

The above Expression [16.25] permits carrying out a parametric analysis, which is useful in optimizing the water electrolyser design. A Pd–Ag tube with a wall thickness of 50 µm can be used as an example cathode for studying the effect of the operating temperature (in the range 20–90°C) under different values of the overpotential (−100, −200 and −300 mV) when a current density of 200 mA cm−2 is discharged over the tubular cathode. The pre-exponential coefficient Pio = 1.23 × 10−5 mol m−1 s−1 Pa−0.5 and the apparent activation energy Ea = 24044.92 J mol−1 have been used to give the hydrogen permeability.42 The results of the DaPe analysis for the above example are reported in the Fig. 16.3. Because the example only includes DaPe values greater than 1, the hydrogen permeated is much less than that produced. The DaPe decreased when the overpotential increased, that is, a higher overpotential involves higher pu, which increases the permeation driving force.

16.3

Water electrolysers using thin-wall Pd–Ag tubes

When selecting materials for preparing Pd−Ag cathodes, one can begin by considering those used in conventional H2 gas permeation processes. The use of thin-wall Pd–Ag permeator tubes as cathodes in alkaline water electrolysis cells is described in the following section.43

16.3.1 Thin wall Pd–Ag tubes In the past, dense selective Pd–Ag thin-wall tubes have been produced via cold rolling and diffusion welding procedures.44,45 According to this technique, commercial Pd–Ag foils have been cold rolled in order to reduce

© Woodhead Publishing Limited, 2013

618

Handbook of membrane reactors

their thickness down to 50–60 µm, this metal working increases the Brinell hardness from about 100 HB (annealed) to 190 HB (worked). Several thermal treatments can be carried out during cold rolling in order to anneal the Pd-based metal foils at 800–1000°C per 1–2 h under controlled atmosphere (inert gas or vacuum). The limbs of the annealed thin Pd–Ag foil are then welded to form the permeator tube. Several joining techniques have been considered: • •

•

brazing risks by the introduction of metal impurities that could poison the Pd−Ag alloy and reduce hydrogen permeability. joining by arc welding produces thermal stressed zones, where hydrogen uploading could cause defects, cracks or holes to form, along with the loss of membrane selectivity. therefore, in order to overcome these drawbacks, the metal limbs of the rolled and annealed foils are typically joined by diffusion welding.

This technique consists of pressing the metal parts during heat treatment carried out at high temperatures, usually at 50–75% of the metal’s melting point. The procedure uses a thermo-mechanical press consisting of two stainless steel plates tightened by screws made of Invar, a nickel−steel alloy with a very low coefficient of thermal expansion. The thin Pd–Ag foil is wrapped around an alumina bar and its limbs are kept close by the thermomechanical press shown in Fig. 16.4. The heat treatment is carried out under vacuum or inert gas, and the temperature increases so that all materials (Pd–Ag foil, steel plates, etc.) expand except for the Invar screws, which apply pressure to the metal parts being joined. As shown in detail in Fig. 16.5, the resulting Pd–Ag thin-wall membrane tube is joined to two more rigid steel tube ends (100 µm wall thickness), in order to give the permeator the necessary stiffness for a tight connection to the hydrogen recovery system.

16.3.2

Electrolytic cell

The thin-wall tube cathode described above has been used in an electrolytic cell, as illustrated in Fig. 16.6.43 Part of the hydrogen obtained via water electrolysis permeates the dense tubular membrane cathode; this is known as the permeate stream, and the hydrogen produced is ultra-pure. Residual amounts of hydrogen of about 99.5% purity can be recovered over the alkaline solution, known as the retentate stream. The cell was a 400 cm3 glass vessel container (height 13 cm, diameter 6.1 cm) equipped with a magnetic stirrer. The anode of the cell was a nickel 99.9% (Leico Industries) 25 µm thick foil (200 cm2) rolled up symmetrically around the cathode, which consisted of a thin-wall tube of Pd–Ag

© Woodhead Publishing Limited, 2013

Pd-based hollow cathode electrolysers for hydrogen production

619

Pd–Ag membrane Alumina bar

INVAR screw and nuts

Stainless steel plates

16.4 Thermo-mechanical press used to perform the diffusion welding of thin Pd–Ag tubes. Reprinted from ‘Membranes for Membrane Reactors: Preparation, Optimization and Selection’, Basile and Gallucci Ed., Ch. 4, S. Tosti, Metallic membranes prepared by cold rolling and diffusion welding, pp. 155–166, Copyright (2011), with permission from Wiley.

(25 wt%) about 6 cm long, 1 cm of external diameter and 50 µm of wall thickness. This cathode is shown in Fig. 16.7. It consists of the Pd–Ag thinwall membrane tube (a), a closed steel end (b), and an open steel end (c and d) that collects the permeated hydrogen. Both steel ends (the close and

© Woodhead Publishing Limited, 2013

620

Handbook of membrane reactors

Brazed joint

Diffusion welded seam Stainless steel tube end Brazed joint Pd–Ag thin wall membrane

Stainless steel tube end

16.5 Details of a thin-wall Pd–Ag permeator tube. Reprinted from ‘Membranes for Membrane Reactors: Preparation, Optimization and Selection’, Basile and Gallucci Ed., Ch. 4, S. Tosti, Metallic membranes prepared by cold rolling and diffusion welding, pp. 155–166, Copyright (2011), with permission from Wiley.

the open one) are covered in Teflon in order to avoid hydrogen discharge over the steel wall. The cell was plunged into a thermostatic bath (Haake) furnished with a thermocouple in order to record temperature during experiments. Electrolysis was carried out in the range 25–80°C with a KOH 0.02–5 M solution as the electrolyte. All the electrolysis tests were performed using a potentiostat (Autolab Pgstat30) connected to a current booster (Autolab BSTR20A), which was essential for working with electric currents up to 20 A and software control (GPES 4.7, Ecochemie). Two distinct hydrogen streams were generated: an upstream from the outer cathode tube−electrolyte side, which yielded H2 polluted with O2, and a downstream from the cathode tube lumen side, which produced high purity H2. A calibrated mass flow meter (MKS 1179A) was used to assess the hydrogen-permeated stream. The upstream hydrogen was measured during the same test in order to check the electrolysis efficiency. The pattern of current density vs time (Fig. 16.8) for the electrolysis cell at 2.6 V indicated a sharp reduction, similar to a kind of filling trap, until a steady state was achieved at about 100 mA cm−2 where the reaction was diffusion controlled. Figure 16.8 also presents the total and permeated hydrogen flow rates. The overall hydrogen flow rate was determined by means of Equation [16.22] and the pattern was clearly close to that of the current density. The H2 permeated flow rate showed a similar, repeatable three-stage trend. At

© Woodhead Publishing Limited, 2013

Pd-based hollow cathode electrolysers for hydrogen production

621

H2 ultrapure

+

H2 purity 99.5%

−

+ O2

O2

O2

H2

H2

Ni anode

Electrolyte

Pd–Ag tube cathode Porous septum

16.6 Scheme of an alkaline water electrolysis cell. Reprinted from International Journal of Hydrogen Energy, 36, A. Pozio, M. De Francesco, Z. Jovanovic, S. Tosti, Pd-Ag hydrogen diffusion cathode for alkaline water electrolysers, pp. 5211–5217, Copyright (2011), with permission from Elsevier.

the time t0, electrolysis started to generate H2 (Reaction [16.8]), which was adsorbed into the metal Pd–Ag (Reaction [16.7]). At the time t1, the permeated H2 began to flow (Reaction [16.10]) into the membrane lumen. At t > t1, the Pd–Ag continued to adsorb the H atoms into the lattice. Both the H/metal ratio and the quantity of permeated H2 increased until the time t2, when they attained their maximum values. At t > t2, the Pd–Ag lattice was saturated with H and the permeating flow was constant. The lapse of about 700–800 s matches the time-lag, that is, the transitory regime matched the stationary regime. Figure 16.9 shows the corresponding permeation yield for a Pd–Ag cathode. The yield defined by Equation [16.23] increased sharply in the first 1000 s, approaching a maximum of about 55%, and then levelled out. In this state

© Woodhead Publishing Limited, 2013

622

Handbook of membrane reactors

(d)

(c)

(b)

(a)

(b)

16.7 The permeator tube used as a cathode: (a) thin-wall Pd–Ag membrane, (b) Teflon covering entirely the closed steel end (below) and partially the open steel end (above), (c) the open steel end without Teflon, (d) steel tube connection to the hydrogen storage.

50

Hydrogen (cc min-1)

t0

250

40

200 30 150

H2 total

20

100

t2

t1 10

H2 permeate 0 0

600

1200

1800

2400

3000

50

Current density (mA cm-2)

300

0 3600

Time (s)

16.8 Total (- -) and permeated (−) hydrogen flow rate and current density (…..) vs. time on Pd–Ag cathode for electrolysis at 2.6 V and 70°C in KOH 1 M. Reprinted from International Journal of Hydrogen Energy, 36, A. Pozio, M. De Francesco, Z. Jovanovic, S. Tosti, Pd–Ag hydrogen diffusion cathode for alkaline water electrolysers, pp. 5211–5217, Copyright (2011), with permission from Elsevier.

© Woodhead Publishing Limited, 2013

Yield (%)

Pd-based hollow cathode electrolysers for hydrogen production

623

100 90 80 70 60 50 40 30 20 10 0 0

600

1200

1800 Time (s)

2400

3000

3600

Theoretical yield (%)

16.9 Permeation yield on Pd–Ag cathode at 70°C and 2.6 V in KOH 1 M. Reprinted from International Journal of Hydrogen Energy, 36, A. Pozio, M. De Francesco, Z. Jovanovic, S. Tosti, Pd-Ag hydrogen diffusion cathode for alkaline water electrolysers, Pages No. 5211–5217, Copyright (2011), with permission from Elsevier. 18 17 16 15 14 13 12 11 10 9 8

y = 1.0728x R 2 = 0.9886

8

9

10

11

12

13

14

15

16

Measured yield (%)

16.10 Theoretical versus measured permeation yield (small circles) and linear regression (line) on Pd–Ag cathode at 70°C and 2.6 V in KOH 1 M for a long-term electrolysis test. Reprinted from International Journal of Hydrogen Energy, 36, A. Pozio, M. De Francesco, Z. Jovanovic, S. Tosti, Pd–Ag hydrogen diffusion cathode for alkaline water electrolysers, pp. 5211–5217, Copyright (2011), with permission from Elsevier.

the Pd–Ag cathode could generate high purity permeated H2 at a flow rate of 10.7 mL min−1 and a balance of 8.8 mL min−1. At 99.5%, the purity of H2 obtained via this process is equivalent to that produced by an alkaline electrolyser. Surface processing and annealing of the Pd–Ag membrane before use could allow for increased permeability in our operative environment.21,46 In order to check the electrolysis efficiency, flow rates of non-permeated and permeated hydrogen streams were investigated during a long electrolysis

© Woodhead Publishing Limited, 2013

624

Handbook of membrane reactors

−1.80

E/V versus NHE

−1.60 −1.40 −1.20 −1.00 −0.80 −0.60 −0.40 −0.20 0.00 0.000

0.020

0.040

0.060 0.080 0.100 Current density (A cm−2)

0.120

0.140

0.160

16.11 Electrode potential versus current density for Pd–Ag (-○-), Ni (-Δ-) and Ni/Pd–Ag (-□-) at 75°C in KOH 1 M. (NHE – normal hydrogen electrode.)

test. Figure 16.10 shows the theoretical permeation yield, estimated by means of Equations [16.22] and [16.23], and the total hydrogen volume. Electrolysis efficiency is always lower than 100% and for this reason the true yield is higher than the theoretical yield. However, because the pattern is linear, the real value can be extrapolated correctly by multiplying Equation [16.23] for a constant value of 1.07.

16.3.3

Metal coated Pd–Ag tubes

In order to decrease the electrolysis cell voltage, the Pd–Ag cathode could be activated by electrodeposition of a third metal, such as Ni or Pt. Figure 16.11 compares the curves of the electrode potential vs the current density for a Pd–Ag cathode, a nickel and a Ni covered Pd–Ag cathode. It is evident that the HER is influenced by the metal surface, showing that it is possible to reduce the overpotential by choosing an appropriate metal coating for the cathode.

16.4

Applications of Pd–Ag membrane cathodes

Following earlier studies into the use of Pd–Ag cathodes, which are described in the Introduction,16–19 in 1970 Lovelock et al. proposed an electrolytic cell with a hollow Pd–Ag alloy cathode (125 mm thickness) as a source of high pressure, ultra-pure hydrogen for gas chromatographic applications.20 The cell temperature was in the range 160–250°C, and a mixture of KOH 67.5%, LiOH 10% and water 22.5% was used as the electrolyte. Production of hydrogen by permeation through the cathode tubing was limited by a

© Woodhead Publishing Limited, 2013

Pd-based hollow cathode electrolysers for hydrogen production

625

maximum current density of 0.1 A cm−2, in order to attain a highly efficient operation. In comparison to acidic electrolytes, electrolysis in alkaline media allowed the use of a less expensive nickel anode, but required a higher operating temperature. More recently, Bellanger and Rameau21 proposed a method of reducing tritiated water with an electrolytic process using Pd-based cathode membranes. These studies have been of great interest in the nuclear field, since the tritiated water is produced as liquid waste by existing nuclear facilities, and will be released in large quantities by future fusion reactors.47 In the first study, Bellanger and Rameau described a prototype electrolyser based on the principle of a gas diffusion Pd–Ag electrode incorporating a tritium charging cathode, which produces very pure hydrogen isotope gases.21 They used Pd–Ag tubular cathodes with wall thicknesses in the range 50–250 µm, and mainly studied the effect of the overpotential, the temperature, the electrolyte concentration and the isotopic fractionation. They found that such devices operate a non-negligible tritium enrichment of the liquid phase (by about a factor of 6), while the permeated gas stream contains more hydrogen (protium). A current density of 700 mA m−2 and temperature of 80°C were selected in order to reduce the isotopic effect. In a later investigation into the enrichment of tritiated water at low concentration, Bellanger considered the use of a composite membrane consisting of a thin Pd layer coated over a Nafion support as described in Fig. 16.12.48 The use of a thin Pd membrane increased the hydrogen permeation and made operation at low temperatures possible. The best operating conditions were shown to be 75 mA m−2 at a temperature of 20°C with a Pd thickness of 2 µm. Higher temperatures were also investigated by carrying out electrolysis with melted sodium hydroxide. The aim of the prototype electrolyser, which is represented in Fig. 16.13, was to recover tritium from highly tritiated water in fusion reactors.34 The high temperature permits an increased rate of tritium diffusion and the use of thicker membranes, which are more reliable for nuclear applications. For this prototype electrolyser, a current density of 230 mA m−2 and a temperature of 150°C were adopted as the best operating conditions for a Pd–Ag membrane 86 µm thickness. Studies are currently being carried out by Pozio et al. into developing an electrolysis cell using a thin-wall Pd–Ag tube as the cathode.43,46,49 These papers describe experimental work that involved the direct-current electrolysis of potassium hydroxide solutions at medium temperatures (up to 80°C) and atmospheric pressure. In Section 16.3, electrolytic cells with a palladium−silver alloy cathode have been described in detail. The objective of these studies was to obtain the best possible yield of hydrogen recovered on the inside surface of the Pd–Ag cathode in alkaline electrolysis. Several electrolysis parameters were studied in order to establish a steady

© Woodhead Publishing Limited, 2013

626

Handbook of membrane reactors 11

12

1 9

7

10

4 3H

3

2O

Circuit

3

3H

2

2

Circuit

8

+

−

6 5 3

16.12 Schematic of the tritiated water prototype electrolyser: (1) tank for storage of tritiated water; (2) electrolyser; (3) loop for cycling the tritiated water; (4) Pd cathode membrane; (5) ionic solid polymer membrane made of Nafion; (6) anode; (7) pump; (8) hydrogen compartment; (9) storage container; (10) sampling and instrumentation; (11) ionization chamber; (12) release in atmosphere. Reprinted from Fusion Engineering and Design, 82, G. Bellanger, Enrichment of slightly concentrated tritiated water by electrolysis using a palladium cathode coated on ionic solid polymer membrane—Design and results, pp. 395–405, Copyright (2007), with permission from Elsevier.

configuration. A 55% yield of permeating hydrogen was easily reached at a low temperature (70°C). The hydrogen permeability was found to be higher than that obtained with a conventional permeation system, in which hydrogen gas was fed directly into the lumens of Pd–Ag membrane tubes at high temperatures 200–400°C. Preliminary planning for a scale up of the experimental cell to a commercial laboratory hydrogen generator has been based on these results. Table 16.1 summarises the characteristics of an alkaline electrolyser with a Pd–Ag cathode composed of a permeator tube with a wall thickness of 50 µm. In order to verify long-term material stability, the effect of discontinuous electrolysis on permeation performance and methods for Pd–Ag electrode reactivation were also analysed.46,49 Structural changes to the Pd–Ag alloy, induced by electrolytic hydrogen adsorption, have been studied in detail

© Woodhead Publishing Limited, 2013

Pd-based hollow cathode electrolysers for hydrogen production

627

7

12 1

8 10 11 13

5

6

3

14 9 4

2

16.13 Schematic of the tritiated water prototype electrolyser: (1) 3H2O feed tank; (2) electrolyser; (3) tritiated water level probe; (4) hollow thimble-shaped Pd–25%Ag membrane; (5) argon circuit; (6) upstream condenser; (7) catalyst reactor; (8) downstream condenser; (9) zeolite container for 3H2O storage and re-injection; (10) sampling point for mass spectrometry; (11) container for tritium gas storage after diffusion, (12) container for tritide storage; (13) instrumentation for diffusion calculations; (14) mass spectrometry; (T) valve.

and were correlated with hydrogen uploading into the metal (i.e., the H/ Pd atomic ratio). By performing electrolysis for a set amount of time, the authors managed to control the H/Pd ratio over the whole range of α and β-phase stability, as defined by the phase diagram.46 The influence of surface oxidation and hydrogen uploading into the Pd–Ag membrane on hydrogen permeation has also been examined. It has been verified that hydrogen uploading into the Pd–Ag lattice increases hydrogen permeation, while surface oxides formed during and after electrolysis cycles were the main factors involved in decreasing the hydrogen permeation.49 Silver oxides were particularly identified as permeation-blocking species. Several treatments were also proposed that aimed to reduce electrode deactivation, including wet chemistry treatments applicable after electrolysis and the in situ electrode reactivation by ultrasonic pulses.49

© Woodhead Publishing Limited, 2013

628

Handbook of membrane reactors Table 16.1 Characteristics of an alkaline electrolyser with a Pd–Ag permeating cathode Temperature Molarity KOH Pd–Ag cathode geometric area Nickel anode geometric area Non-permeated low purity H2 Permeated high purity H2 Cell voltage Current density Power efficiency Permeation yield Power

70–80°C 1M 500 cm2 1000 cm2 203 mL min−1 248 mL min−1 2.6 V 103 mA cm−2 47% 55% 130 W

Source: Reprinted from International Journal of Hydrogen Energy, 36, A. Pozio, M. De Francesco, Z. Jovanovic, S. Tosti, Pd-Ag hydrogen diffusion cathode for alkaline water electrolysers, pp. 5211–5217, Copyright (2011), with permission from Elsevier.

16.5

Conclusions and future trends

Water electrolysis cells that use hollow metal cathodes offer the advantage of increasing the amount of ultra-pure hydrogen produced as a proportion of the total. Particular attention has been paid to the use of thin-wall Pd–Ag permeator tubes and the design, manufacture and testing of prototype cells. The principles of gas permeation and water electrolysis have been introduced in order to model the behaviour of these electrolysers. The Damköhler−Péclet analysis has been also used in order to describe the influence of the main parameters, such as the effects of overpotential and the permeation activation energy on the permeation yield of the electrolytic cell. The main applications are concerned with the production of ultra-pure hydrogen for laboratory and small scale electrolysers and the processing of tritiated water. Recent studies into alkaline electrolysis cells using thinwall Pd–Ag tubes have demonstrated the applicability of these technologies for commercial hydrogen electrolysers. Other tests have verified the use of these hollow cathode cells for recovering tritium from tritiated water in the fuel cycle of the next fusion reactors.

16.6 1. 2.

References

Janssen L (2007), Hydrogen fuel cells for cars and buses, J Appl Electrochem, 37, 1383–7. Marban G, Valdes-Solis T (2007), Towards the hydrogen economy, Int J Hydrogen Energy, 32, 1625–37.

© Woodhead Publishing Limited, 2013

Pd-based hollow cathode electrolysers for hydrogen production 3. 4. 5. 6.

7.

8. 9. 10.

11. 12.

13.

14.

15.

16. 17. 18. 19. 20.

21. 22.

629

Balat M (2008), Potential importance of hydrogen as a future solution to environmental and transportation problems, Int J Hydrogen Energy, 33, 4013–29. Stojic D, Marceta M, Sovilj S, Miljanic S (2003), Hydrogen generation from water electrolysis – possibilities of energy saving, J Power Sources, 118, 315–9. Armor J (1999), The multiple roles for catalysis in the production of H2, Appl Catal A, 176, 159–76. Levent M, Gunn D, El-Bousiffi M (2003), Production of hydrogen-rich gases from steam reforming of methane in an automatic catalytic microreactor, Int J Hydrogen Energy, 28, 945–59. Xu J, Yeung C, Ni J, Meunier F, Acerbi N, Fowles M, Tsang S-C (2008), Methane steam reforming for hydrogen production using low water-ratios without carbon formation over ceria coated Ni catalysts, Appl Catal A, 345, 119–27. de Levie R (1999), The electrolysis of water, J Electroanal Chem, 476, 92–3. Barbir F (2005), PEM electrolysis for production of hydrogen from renewable energy sources, Solar Energy, 78, 661–9. Liu M, Yu B, Xu J, Chen J (2008), Thermodynamic analysis of the efficiency of high-temperature steam electrolysis system for hydrogen production, J Power Sources, 177, 493–9. Bilgen E (2004), Domestic hydrogen production using renewable energy, Solar Energy, 77, 47–55. Shakya B, Aye L, Musgrave P (2004), Technical feasibility and financial analysis of hybrid wind-photovoltaic system with hydrogen storage for Cooma, Int J Hydrogen Energy, 30, 9–20. Dabo P, Menard H, Brossard L (1997), Electrochemical characterization of graphite composite coated electrodes for hydrogen evolution reaction, Int J Hydrogen Energy, 22, 763–70. Krstajic N, Grgur B, Mladenovic N, Vojnovic M, Jaksic M (1997), The determination of kinetics parameters of the hydrogen evolution on Ti-Ni alloys by AC impedance, Electrochim Acta, 42, 323–30. Marshall A, Sunde S, Tsypkin M, Tunold R (2007), Performance of a PEM water electrolysis cell using IrxRuyTazO2 electrocatalysts for the oxygen evolution electrode, Int J Hydrogen Energy, 32, 2320–4. Wahlin HBJ (1951), The transmission of hydrogen through metals, Appl Phys, 22, 1503. Wahlin HBJ, Naumann VO (1953), The transmission of hydrogen through palladium by electrolysis, J Appl Phys, 24, 42–4. Darling AS (1963), Thermal and electrolytic palladium alloy diffusion cells, Platinum Met Rev, 7(41), 126–9. Clifford JE, Kolic E, Faust CL (1963), Equipment for life support in aerospace, Battelle Memorial Insitute Technical Report, Contract N°AF33(657)10988. Lovelock JE, Simmonds PG, Shoemake GR (1970), The palladium generator-separator a combined source and sink for hydrogen in closed circuit gas chromatography, Anal Chem, 42(9), 969–73. Bellanger G, Rameau JJ (1999), Tritium recovery from tritiated water by electrolysis, Fusion Technol, 36, 296–308. Pozio A, Tosti S , Bettinali L, Borelli R, De Francesco M, Lecci D, Marini F (2009), Elettrolizzatore alcalino con catodo tubolare in Pd-Ag per la produzione di idrogeno ultrapuro, Italian Patent RM2009U000200.

© Woodhead Publishing Limited, 2013

630

Handbook of membrane reactors

23. Shu J, Grandjean BPA, Van Neste A, Kalaguine S (1991), Catalytic palladium-based membrane reactors: a review, Can J Chem Eng, 69, 1036–60. 24. Paglieri SN, Way JD (2002), Innovations in palladium membrane research, Sep Pur Methods, 31(1), 1–169. 25. Basile A, Gallucci F, Tosti S (2008), Synthesis, characterization, and applications of palladium membranes, Membrane Science and Technology, 13, 255–323. 26. Zeng K, Zhang D (2010), Recent progress in alkaline water electrolysis for hydrogen production and applications, Prog Energ Combust, 36, 307–26. 27. Subramanyan PK (1981), Electrochemical aspects of hydrogen in metal, in Comprehensive Treatise of Electrochemistry Vol. 4, ed. Plenum Press, New York, 429. 28. Makrides AC (1964), Absorption of hydrogen by silver-palladium alloys, J Phys Chem, 68(8), 2160–69. 29. De Luccia JJ (1988), Electrochemical aspects of hydrogen in metals, in Hydrogen embrittlement: prevention and control, ASTM STP962, Raymond L., Ed., American Society for Testing and Materials, pp 17–34. 30. Ammar IA, Awad SA (1956), Hydrogen overpotential on silver in sodium hydroxide solutions, J Phys Chem, 60, 1290–3. 31. Bellanger G (1995), Embrittlement of Pd and Pd-Ag alloy cathode membranes by tritium, Fusion Technol, 27, 36–45. 32. Bélanger A, Vijh AK (1982), The hydrogen evolution reaction on Ag-Pd alloys: influence of electronic properties on electroactivity, Surf Technol, 5, 59–78. 33. Davis JR, Allen P, Lampman SR, Zorc TB, Henry SD, Daquila JL, Ronke AW (1990), ASM Handbook, Formerly Tenth Edition, Metals Handbook, Vol. 2, ‘Properties and selection: Nonferrous alloys and special-purpose materials’. ASM International, United States of America, 716. 34. Bellanger G (2009), Optimization for the tritium isotope separation factor and permeation by selecting temperature and thickness of the diffusion Pd–Ag alloy cathode, Fusion Eng Des, 84, 2197. 35. Oriani RA (1978), Hydrogen embrittlement of steels, Ann Rev Mater Sci, 8, 327–57. 36. Ganley JC (2009), High temperature and pressure alkaline electrolysis, Int J Hydrogen Energy, 34(3), 3604–11. 37. Bernstein LA, Lund CFR (1993), Membrane reactors for catalytic series and series–parallel reactions, J Membrane Sci, 77, 155–64. 38. Battersby S, Teixeira PW, Beltramini J, Duke MC, Rudolph V, Diniz da Costa JC (2006), An analysis of the Peclet and Damkohler numbers for dehydrogenation reactions using molecular sieve silica (MSS) membrane reactors, Catal Today, 116, 12–7. 39. Tosti S, Borelli R, Borgognoni F, Favuzza P, Rizzello C, Tarquini P (2008), Study of a dense metal membrane reactor for hydrogen separation from hydroiodic acid decomposition, Int J Hydrogen Energy, 33, 5106–14. 40. Tosti S, Basile A, Borelli R, Borgognoni F, Castelli S, Fabbricino M, Gallucci F, Licusati C (2009), Ethanol steam reforming kinetics of a Pd–Ag membrane reactor, Int J Hydrogen Energy, 34, 4747–54. 41. Tosti S (2003), Supported and laminated Pd-based metallic membranes, Int J Hydrogen Energy, 28, 1455–64. 42. Tosti S, Borgognoni F, Santucci A (2010), Electrical resistivity, strain and permeability of Pd-Ag membrane tubes, Int J Hydrogen Energy, 35, 7796–802.

© Woodhead Publishing Limited, 2013

Pd-based hollow cathode electrolysers for hydrogen production

631

43. Pozio A, De Francesco M, Jovanovic Z, Tosti S (2011), Pd-Ag hydrogen diffusion cathode for alkaline water electrolysers, Int J Hydrogen Energy, 36, 5211–17. 44. Tosti S, Bettinali L (2004), Diffusion Bonding of Pd-Ag Membranes, J Mater Sci, 39, 3041–46. 45. Tosti S (2011), ‘Metallic membranes prepared by cold rolling and diffusion welding’ in Membranes for Membrane Reactors: Preparation, Optimization and Selection, Basile A and Gallucci F (Ed.), Wiley, Ch. 4, 155–167 46. Jovanovic Z, De Francesco M, Tosti S, Pozio A (2011), Influence of surface activation on the hydrogen permeation properties of PdAg cathode membrane, Int J Hydrogen Energy, 36, 15364–71. 47. Glugla M, Antipenkov A, Beloglazov S, Caldwell-Nichols C, Cristescu IR, Cristescu I, Day C, Doerr L, Girard JP, Tada E (2007), The ITER tritium systems, Fusion Eng Des, 82, 472–87. 48. Bellanger G (2007), Enrichment of slightly concentrated tritiated water by electrolysis using a palladium cathode coated on ionic solid polymer membrane— Design and results, Fusion Eng Des, 82, 395–405. 49. Jovanovic Z, De Francesco M, Tosti S, Pozio A (2011), Structural modification of PdAg alloy induced by electrolytic hydrogen adsorption, Int J Hydrogen Energy, 36, 7728–36.

16.7

Appendix: nomenclature

16.7.1 a A CO COH DaPe dQ/dt Ea F f HER i J k K K′ kd and k-d n NHE P

Notation pressure exponent of Equation [16.1] membrane area (m2) hydrogen solubility molar concentration (mol L−1 ) of alkaline hydroxide Damköhler−Péclet number Faradic current (A) apparent activation energy (J mol−1) Faraday constant fugacity of H2 (Pa) hydrogen evolution reaction current density (A m−2) hydrogen permeation flux (mol m−2 s−1) equilibrium constant of Reaction [16.13] constant of Equation [16.19] numerical constant of Equation [16.18] kinetics direct and inverse constants of Equation [16.13] electrical charge exchanged during the Reactions [16.5] and [16.6] normal hydrogen electrode hydrogen permeability coefficient (mol m−1 s−1 Pa−n)

© Woodhead Publishing Limited, 2013

632 pd Po pu Q QF r R T

Handbook of membrane reactors hydrogen partial pressures at downstream side (Pa) permeability pre-exponential coefficient (mol m−1 s−1 Pa−0.5) hydrogen partial pressures at upstream side (Pa) total electrical charge transferred during the reaction (Coulomb) H2 permeating flow rate (mol s−1) rate of electrolysis (cm3 min−1) gas constant (8.31 J mol−1 K−1) absolute temperature (K)

Greek symbols α δ θ

transfer coefficient membrane wall thickness (m) coverage on the surface

© Woodhead Publishing Limited, 2013

17 Fuel cell vehicles (FCVs): state-of-the-art with economic and environmental concerns A. VEZIROGLU, International Association for Hydrogen Energy, USA and R. MACÁRIO, Universidade Técnica de Lisboa, Portugal

DOI: 10.1533/9780857097347.3.633 Abstract: Hydrogen fueled fuel cell vehicles (FCVs) will play a major role as a part of the change towards a hydrogen based-energy system. When combined with the right source of energy, fuel cells (FCs) have the highest potential efficiencies and lowest potential emissions of any vehicular power source. As a result, extensive work into the development of hydrogen fueled FCVs is taking place. This chapter highlights the relevant research and development work which has taken place on FCV technology, with a focus on economic and environmental concerns. Key words: fuel cell vehicles (FCVs), hydrogen fuel cell vehicles, hydrogen fueled vehicles.

17.1

Introduction

Concerns about the finite nature of fossil fuel resources and the global climate change due to the combustion of fossil fuels have sparked the scientists and engineers of the world into seeking a clean and sustainable energy source for our ever increasing energy demands (Granovskii et al., 2006b; Corbo et al., 2009). Hydrogen has been called the optimal replacement for fossil fuels, particularly in the transportation sector which represents the majority of petroleum consumption world-wide. The properties of hydrogen (H2) make it a unique fuel and give it certain advantages, as well as disadvantages, over conventional fuels. Hydrogen can be used for automotive applications via a blended mix of hydrogen and hydrocarbons in internal combustion engines (ICE), or used in an FC stack onboard light duty vehicles. The latter option, that is, vehicular applications of FCs, is the focus of this chapter. There has been much research into fuel cell electric vehicles (FCVs or FCEVs) in the recent past. Within the last few years, research has been 633 © Woodhead Publishing Limited, 2013

634

Handbook of membrane reactors

published regarding a variety of FC types and their properties. It should be noted that on a macro scale FCVs are still in the research and development phase. As such, the existing literature on FCs covers areas such as specific FC mechanisms and phenomena, comparative analyses between FCs and other power sources, environmental impacts of FCVs, economics which justify or discredit FCVs, and even papers concerning their effect on human health (Al-Baghdadi, 2009a). For vehicular applications, the polymer electrolyte fuel cell (PEFC), also known as the proton exchange membrane fuel cell (PEMFC), seems best suited (Boettner and Moran, 2004; Corbo et al., 2009). There continues to be research into optimal purification methods of FC grade hydrogen (Swesi et al., 2007), optimal operating points and automatic controls (Dalvi and Guay, 2009), FC cold-start ability in low temperature conditions (Schießwohla et al., 2009), and many other operating characteristics. Recently published simulation tools will assist researchers in the future as they continue to bring FC technology to maturity (Moore et al., 2005; Al-Baghdadi, 2009b). Also, component degradation and durability is anticipated to be a critical issue for the practical use of FCs (Wu et al., 2010b). The United States Department of Energy (DoE) has specified long-term targets for vehicular PEFC development. By the year 2015, FCs for these applications are expected to be 60% efficient, and cost US$30/kW. Furthermore, vehicular FC stacks are expected to have a nominal lifetime of at least 5000 h, which is equivalent to 150 000 miles at 30 miles per hour. These targets are echoed within the international community and represent concrete milestones for vehicular PEFC development (Moore et al., 2006). Some researchers are taking existing data on available FC stacks and comparing them to each other to determine which configurations are optimal. This work acts as feedback for fundamental research efforts, and steers future studies in the most promising directions (Boettner et al., 2004; Moore et al., 2006; Thomas, 2009a). There is enough fundamental information about FCs for larger analyses to be carried out. An environmental analysis of the impact of FCVs is a popular topic for research, and has appeared frequently in the recent literature. A common interest of many of these studies is comparing emissions from the entire hydrogen supply chain infrastructure to those of an analogous fossil fuel infrastructure (Feng et al., 2004; Colella, 2005; Granovskii et al., 2006a). Other environmental research has begun to look more deeply into changes in both total and urban emissions (Huo et al., 2009). In the US, some research is focused on the possibility of using coal as transportation fuel in response to the growing desire for energy security (Jaramillo et al., 2009). An environmental analysis can be coupled with an economic analysis to obtain a realistic indication of the viability of an FCV market. FC technology is currently being considered by marketing experts to determine the

© Woodhead Publishing Limited, 2013

Fuel cell vehicles

635

best strategies for marketing and growing an FCV economy (Williams and Kurani, 2006; Ajanovic, 2008). Some think that niche roles such as PEFC auxiliary power units (APU) could provide short- and medium-term growth (Contestabile, 2010), while others are beginning to investigate a possible symbiotic relationship between FCVs and BEVs (Offer et al., 2010). The purpose of this chapter is to provide a general overview of the current research on FCs for vehicular applications. It is not intended to be allinclusive. Rather, it will serve as a starting point for future research, and to gain perspective in the field of FCs. It should be noted that FC technology is still experiencing significant research and development. The next decade will most likely see some dramatic changes to the general tone of research into these quintessential components of the hydrogen economy.

17.2

Technical aspects in the development of fuel cell vehicles (FCVs)

FCs are a technology which is still seeing significant development. The best FC configuration has yet to be determined, and it will likely be different for various combinations of operating conditions, working loads and desired sizes. The four major subsystems of any hydrogen FC system are the FC stack, air supply, hydrogen supply, and water and thermal management. An accepted method of studying these elements of an FCV is through a dynamic simulation tool such as FCVSim. This program places an emphasis on FCVs, uses logical forward-looking causal structures, incorporates dynamic aspects, utilizes modular topography, and supports hardware-inthe-loop and rapid prototyping. The program can be extended to work with direct hydrogen (DH) in the DH-FCVSim extension (Moore et al., 2005). A large portion of the current work in FCs is devoted to PEFC, sometimes called PEMFC, as they are the most widely suitable FC technology for vehicular applications. Compact methanol reformers are an attractive option for generating hydrogen on board light duty vehicles that are fitted with PEFC. The catalytic burner in the methanol reformers provides energy for the reforming reaction by burning gases into carbon dioxide and water. Resulting gases have to be kept below EZEV (Equivalent Zero Emission Vehicle) standards using an appropriate burner design. Compact methanol reformers can be utilized to generate hydrogen on board a small duty vehicle with emissions of 10 to 100 times less than those of an ICE. When the complete fuel cycle is considered, the emissions from a PEFC are on the order of 100 to 1000 times less than those of an ICE of the same size (Emonts et al., 1998). One recent study examines the role of reactant feeding, humidification, and cooling systems for two versions of a hybridized energy supply in a PEFC (Corbo et al., 2009). The specific process by which

© Woodhead Publishing Limited, 2013

636

Handbook of membrane reactors

a FC degrades in vehicular applications over time is a new and expanding field. Table 17.1 highlights some of the recent studies into PEFC degradation. A comprehensive study of FC degradation can be found in (Borup et al., 2007). In a recent study, potential hydrogen production methods have been covered. Hydrogen can be reformed from fossil fuels, produced via water electrolysis, or it can be extracted from biomass via gasification. FCVs may either obtain hydrogen from a fueling station or produce it onboard. If FCVs are to refuel with hydrogen at a fueling station, the hydrogen must either be produced locally or transported from a central production facility. The likely sources of hydrogen for the present century and their market shares versus time have been illustrated diagrammatically in Fig. 17.1 (Thomas, 2009a). The most likely sources of hydrogen, in order of timely implementation, begin with distributed hydrogen by reforming natural gas locally at the fueling station; followed by reforming biofuels such as cellulosic ethanol locally at the fueling station; central production by biomass gasification; coal integrated gasification combined cycle (IGCC) with carbon capture and storage (CCS); and eventually electrolysis by zero-carbon electricity from nuclear and/or renewables. When comparing the costs related to future fuel options, it is important to include the cost of the fuel, vehicle price and maintenance cost. Figure 17.2 shows life-cycle fuel costs for different H2 fuel chains relative to ICEV, as well as those of coal-methanol, NG-methanol and petroleum-gasoline. It is clearly visible that the electricity−H2 fuel conversion is the most expensive, while all other paths result in some savings. The coal-methanol, NG-methanol, and petroleum-gasoline fuels are easier to transport, which reduces the fuel costs, resulting in savings in life-cycle fuel costs. It is also interesting to note that there is not much difference between large-scale (Coal−H2 and NG−H2) and small-scale (NGsmall−H2, Coal-Methanol−H2 and NG-Methanol−H2) fuel paths (Wang et al., 2005). In the area of onboard hydrogen production, the purification process for hydrogen in FCVs has been considered. Purification methods may include fueling of the vehicle with cycloalkane dehydrogenation in the vehicle, discharge of aromatics from the vehicle, and regeneration in a hydrogenation plant. Swesi et al. analyzed the MTH (methylcyclohexane-toluene-hydroge n) cycle due to its hydrogen storage capacity of 6.1 wt% and good reactivity in dehydrogenation (Swesi et al., 2007). There are two main separation techniques to extract hydrogen: membranes and adsorption. Studies indicate that separation of hydrogen through zeolite membranes is ineffective for FCV applications, since the toluene content in the permeate is too high (>2000 ppm). When toluene was present at high concentrations, the diffusion of hydrogen was hindered due to a strong adsorption of toluene in the membranes. Palladium membranes are

© Woodhead Publishing Limited, 2013

Table 17.1 Performance degradation research focuses for PEM fuel cells PEMFC component

Degradation effect

Reference

Entire FC Entire FC Entire FC

Trade-off between efficiency and degradation performance Bus city driving cycles effect on voltage degradation Review of literature on effects and potential mitigation of various degradation modes Difference between reversible and irreversible voltage degradation under open-circuit conditions Sub-zero operation effect on ice formation Driving cycle dynamic loading Catalyst decay and membrane failure under close to open-circuit conditions Freeze/thaw cycles Cathode flooding, membrane drying, and anode catalyst poisoning by CO Excess air bleeding Cell reversal during operation with fuel starvation Air−air start-up, platinum crystallite precipitation Structural changes in PEM and catalyst layers due to platinum oxidation or catalyst contamination under open-circuit conditions On/off cyclic operation under different humid conditions Effect of hygro-thermal cycle on membrane stresses Water uptake effect on cyclic stress and dimensional change, hydrogen crossover Increasing hydrogen gas crossover, comparing with Nafion 112 membranes Imide function hydrolysis inducing polymer chain scissions, comparison with Nafion membranes Pt catalyst ripening, electrocatalyst loss or re-distribution, carbon corrosion, electrolyte and interfacial degradation

Gemmen and Johnson, 2006 Lu et al., 2007 Wu et al., 2008

Entire FC © Woodhead Publishing Limited, 2013

Entire FC Entire FC Entire FC Entire FC Cathode, membrane and anode MEA, anode catalyst MEA MEA MEA MEA FC membranes Nafion NR111 membrane Nafion 212 membrane Sulfonated polyimide membranes Platinum catalysts, carbon-support, Nafion ionomer

Kundu et al., 2008 Alink et al., 2008 Lin et al., 2009 Wu et al., 2010b Luo et al., 2010 Rubio et al., 2010 Inaba et al., 2008 Taniguchi et al., 2004 Ettingshausen et al., 2009 Zhang et al., 2010 Seo et al., 2010 Kusoglu et al., 2006 Tang et al., 2007 Fernandes and Ticianelli, 2009 Meyer et al., 2006 Zhang et al., 2009

(Continued)

Table 17.1 Continued PEMFC component

Degradation effect

Reference

Catalysts

Fuel and oxidant starvation effects on catalyst and carbon-support degradation Catalyst treatment with acid effect on decreasing oxygen reduction reaction High temperature operation effect on carbon corrosion, platinum dissolution, and sintering

Yousfi-Steiner et al., 2009

Pt/C/MnO2 hybrid catalysts Pt/C and PtCo/C catalysts

Trogadas and Ramani, 2007 Aricò et al., 2008

© Woodhead Publishing Limited, 2013

Pt/C catalysts

Degradation due to Cl−, F−,

Pt/C catalysts

Increasing particle size of Pt/C catalyst due to dissolution mechanism, oxygen electroreduction at cathode catalyst Surface area loss due to carbon corrosion and increasing platinum particle size Platinum dissolution and deposition on cathode, Pt diffusion in MEA, hydrogen permeation Potential static holding conditions and potential step conditions effect on platinum dissolution and carbon corrosion Toluene-induced cathode degradation CO and CO2 poisoning Degradation effect on oxygen diffusion polarizations

Wang et al., 2009

Elevated temperature and flow rate effect on mechanical stress and material loss Sealing decomposition effect on catalysts Exposure time effect on de-crosslinking and chain scission

Wu et al., 2010a

Platinum catalyst Platinum catalyst Platinum catalyst Platinum catalyst Platinum catalyst Electrode porous catalyst layer and gas diffusion layer Gas diffusion layer Sealing material Gasket silicone rubber

2− 4

, or NO3−

Matsuoka et al., 2008

Cai et al., 2006 Bi and Fuller, 2008 Shao et al., 2008 Li et al., 2008 Yan et al., 2009 Aoki et al., 2010

Schulze et al., 2004 Tan et al., 2007

Fuel cell vehicles

639

100% Central electrolysis (renewable and nuclear)

90% 80% 70%

Coal IGCC + CCS

60% 50% 40% 30%

Biomass gasification

Natural gas at fueling station

20% 10%

Biofuels (ethanol) at fueling station 0% 2010

2020

2030

2040

2050 2060 Year

2070

2080

2090

2100

17.1 Likely sources of hydrogen over this century (Thomas, 2009a).

150%

126%

100%

69%

70%

83%

87% 68%

83%

50% 0% −50% −100%

−21% −43%

−32%

NGPetroleum- Coal-H2 Coalmethanol methanol gasoline

NG-H2 Electricity- NG(small)- CoalPetroleumNGH2 H2 methanol- methanol- gasolineH2 H2 H2

17.2 Life-cycle fuel costs relative to that of ICEV (Wang et al., 2005).

more promising (Swesi et al., 2007). Ultimately, it is more likely that future vehicles will refuel with hydrogen to avoid onboard purification. Onboard hydrogen storage is one of the paramount hurdles that FCVs are trying to overcome to become competitive with the current fleet of ICE vehicles. Storage options include metal hydrides, carbon nanotubes, compressed gas and liquid hydrogen. Currently, all of these options are both heavier and larger than their gasoline tank counterparts, but they are being further developed to achieve that goal as illustrated in Fig. 17.3 (Mori and Hirose, 2009).

© Woodhead Publishing Limited, 2013

640

Handbook of membrane reactors Hydrogen-absorbing alloy

Heavy

Advanced hydrogen-absorbing alloy

Tank weight (kg)

Carbon nanomaterial

Organic hydride

35 MPa

25 MPa

70 MPa

Light

ICE gasoline Small

Liquid hydrogen

High-pressure hydrogen

Tank volume (L)

Large

17.3 Hydrogen storage technologies and targets (Mori and Hirose, 2009).

The durability of PEFCs in vehicular applications has been the subject of recent research, which is a good indicator of progress toward FCVs. Computational fluid dynamics (CFD) models of FCs now exist, allowing the study of failure mechanisms in order to generate much more accurate life prediction models. There are a number of commercially available CFD programs that support PEFC research, including Fluent, CFX-5, STAR-CD, and FEMLAB (Guvelioglu and Stevgner, 2005). The best CFD programs, however, are built inhouse by researchers looking into specific aspects of FC operation. Three-dimensional, multiphase, non-isothermal CFD programs can account for all the major transport phenomena in a PEFC, including convective and diffusive heat and mass transfer, electrode kinetics, transport and phase change mechanisms of water, and potential fields. This allows investigation into the displacement, deformation, and stresses inside the whole FC as they develop during operation due to changes in temperature and relative humidity. A recent study found that non-uniform distribution of stresses caused by temperature gradients induce localized bending stresses, contributing to delamination between the membrane and gas diffusion layers. These stresses also contribute to delamination between gas diffusion layers and the flow field channels, particularly on the cathode side. These findings help explain cracks and pinholes that develop in FC components during regular operation, and will certainly help guide FC development in the future (Al-Baghdadi, 2009a). Table 17.2 lists the PEFC research work that has recently been done utilizing CFD programs.

© Woodhead Publishing Limited, 2013

Table 17.2 PEFC CFD programs and recent research implementation Model description

Research focus and general results

Reference

Non-isothermal, 3D

Solves for electric and ionic potentials in electrode and membrane, resolves local activation overpotential distribution, and predicts local current density distribution Results: Can predict maximum current densities and underlying causes (ohmic losses, concentration losses, asymmetry parameter, etc.) Solves for local current density distribution, wetting behavior of gas diffusion layers, and conditions that may lead to pore plugging Results: This model can effectively identify parameters for wetting behavior of the gas diffusion layers, it can also identify conditions that may lead to the onset of pore plugging Solves for displacement, deformation, and stresses inside the whole cell during cell operations due to changes in temperature and relative humidity Results: Temperature gradients create non-uniform stress distributions that induce bending stresses, causing delamination between membrane and gas diffusion layers, and gas diffusion layers and channels on cathode side Solves for species profiles, temperature distribution, potential distribution, and local current density distribution in airflow-channel and air-breathing FCs Results: Air-breathing designs achieve higher power densities, have a better gas replenishment rate at catalyst sites, and have a more uniform local current density distribution Solves for local activation over potentials and accurate local current density distribution Results: Varied, study analyzed multiple operating conditions for electrochemical and transport phenomena, and study identified various limiting steps and components under different operating conditions

Sivertsen and Djilali, 2005

Non-isothermal, 3D multiphase © Woodhead Publishing Limited, 2013

Non-isothermal, 3D multiphase

Non-isothermal, 3D multiphase

Non-isothermal, 3D single phase

Al-Baghdadi et al., 2007b

Al-Baghdadi 2009b

Al-Baghdadi 2009a

Al-Baghdadi et al., 2007a (Continued)

Table 17.2 Continued

© Woodhead Publishing Limited, 2013

Model description

Research focus and general results

Reference

Non-isothermal, 3D single phase

Solves for current density distribution across catalyst layer, anode and cathode activation overpotentials, oxygen transport limitations, and ohmic loss distributions Results: There are non-uniform distributions of current density across catalyst layer, differences in anode and cathode activation overpotentials, oxygen transport limitations, and ohmic losses distributions Solves for species profiles, temperature distribution, potential distribution, and local current density distribution in tubular shaped PEFCs Results: Varied, study analyzed multiple operating conditions for electrochemical and transport phenomena, and study identifies various limiting steps and components under different operating conditions Used as a direct problem solver to work with simplified conjugate-gradient method optimizer to solve for optimal gas channel width fraction, gas channel height, and thickness of gas diffusion layer Results: This model can be used as a direct problem solver in optimizing geometric parameters of PEFCs given a set of base case conditions, always leading to a unique final solution Solves for effects of channel geometry and water management Results: High current density operations require smaller width channels and bipolar plate shoulders, higher porosity electrodes result from increasing electrode area under bipolar plate shoulder, relative humidity in anode gas stream is more important for FC performance than relative humidity in cathode gas stream

Baca et al., 2008

Non-isothermal, 3D single phase

Isothermal, 3D single phase

Isothermal, 2D single phase

Al-Baghdadi et al., 2008

Cheng et al., 2007

Guvelioglu and Stenger, 2005

Fuel cell vehicles

643

Another study focusing on long-term durability of six-cell PEFCs found two different causes for cell degradation. During a 1600-hour test, PEFC cell voltage decreased at an average rate of 0.128 mV/h under close to open-circuit conditions. However, the first 800 h had a much slower degradation rate caused by the gradual coarsening of the platinum catalyst, while the second 800-hour period had a dramatic degradation rate caused by catastrophic failure of the membrane. Understanding these changes in failure mode is critical in enhancing the durability of PEFC (Wu et al., 2010b). Logistical thinking has led some researchers to look at the operating conditions in which an FC must work if it were utilized in a passenger vehicle. During the winter months, vehicular FCs need to start up in the same amount of time as the present day vehicles. The United States DoE proclaimed that an FCV should be able to start up from −20°C within 30 s using less than 5 MJ of additional energy. Looking more closely at the cold-start scenario, it has been found that reducing the start-up time requires minimizing the freezing of process water in the catalyst layer of the membrane electrode assembly (MEA). The best way to do this was found to be a strategized shut-down, including a 30-minute purge with dry gases (Schießwohla et al., 2009). Other researchers have looked into start-up/shut-down procedures for PEFCs and their effects on performance and durability. These are summarized in Table 17.3. There has been other logistical work on optimizing hybrid FCV operation during driving. One such study investigated the issue of oxygen starvation during transients in power demand. Oxygen starvation can lead to ‘burn-through’ effects on the membrane surface, which result in permanent damage. The potential solution, albeit costly, was found to be placing one ultra-capacitor at the load to buffer the FC during load changes, and another ultra-capacitor at the compressor to improve phase characteristics of the system. In an FC, ultra-capacitor and battery-hybrid system, the FC is the permanent energy source, the ultra-capacitor handles the need for energy bursts, and the battery acts as an intermediate energy supply and storage for reclaimed energy from braking (Ayad et al., 2011). Simulations showed that a controller could find optimum operating points for this hybrid system without requiring previous knowledge of the system dynamics (Dalvi and Guay, 2009). For such a hybrid FCV, its designed storage system should be able to provide the demanded maximum power at any point without completely discharging the battery. Hybrid vehicles with a high primary energy source (e.g., IC or FCEV) mainly use the onboard battery for energy recovery from braking and for additional power during rapid acceleration (Pede et al., 2004). Pulse discharge power (P) and total available energy (E) are two important parameters for vehicle storage systems. Figure 17.4 presents P/E plots for several electric storage systems. As shown in this figure,

© Woodhead Publishing Limited, 2013

Table 17.3 Recent research in start-up/shut-down procedures for PEFCs Research area Effect studied and general results

Reference

Normal start

Semelsberger and Borup, 2005 Shan and Choe, 2006

Normal start © Woodhead Publishing Limited, 2013

Normal start

Normal start

Normal start

Normal start

Normal shut-down

Gasoline, methanol, ethanol, dimethyl ether and methane effects on hydrogen production Results: modeled overall efficiencies were 37% for gasoline, 38.3% for methanol, 34.5% for ethanol, 38.5% for dimethyl ether, and 33.2% for methane Endplate effects on temperature profile Results: an asymmetric temperature profile develops due to greater heat generation on cathode side; membrane swelling phenomena, caused by continuous water content variation, increases electrical and thermal resistance; latent water heat produced at catalysts can be stored in the stack; the non-uniform temperature distribution can be minimized by coupling coolant for central cells with the end cells Hydrophobic treatment (HT) and micro-porous layer (MPL) in addition to gas diffusion layer (GDL) effect on water balance Results: HT without MPL increases liquid water accumulation at electrode, limiting oxygen transport to catalyst and lowering cell voltage, also decreases water at GDL; HT with MPL addition suppresses water accumulation at electrode, increasing current; increasing air permeability of GDL increases current, also improving start-up performance Liquid water, temperature, GDL thickness and porosity Results: liquid water increases time for current density to reach steady state; temperature does not have significant effect on current density; increasing porosity decreases mass transport time scale; increasing GDL thickness delays influence of liquid water Cathode, anode, and membrane potentials during start-up and shut-down Results: hydrogen/air boundary at anode creates voltage between membrane inlet and outlet and voltage at interface of cathode and membrane outlet, causing carbon corrosion Internal currents during open-circuit conditions Results: internal currents are caused mostly by capacitive effects; carbon oxidation occurs simultaneously and has negligible contribution to internal currents Close/open state of outlets and application of dummy load effect on degradation of MEA Results: using a thin electrolyte membrane, outlets should be closed to limit degradation during on/off operation; using a thick electrolyte membrane, the dummy load should be applied to limit degradation

Nakajima et al., 2007

Mishra and Wu, 2009

Shen et al., 2009

Maranzana et al., 2009 Kim et al., 2008

Cold start

Cold start

Cold start © Woodhead Publishing Limited, 2013

Cold start

Cold start

Cold start

Cold start

Start current density dependence on membrane humidity, operation voltage, and gas flows Results: start-up below 0°C depends on membrane humidity and operation voltage; current decay depends on constant gas flows of reactant gases; ice formation does cause degradation effects in the porous structures that leads to performance loss Buildup of ice in cathode catalyst and electrode structure, operations near short-circuit conditions Results: near short-circuit conditions improves start-up below –20°C by maximizing hydrogen utilization, producing waste heat absorbed by stack, and delaying loss of electrochemical surface area to ice formation; bipolar plates should be made from metal instead of graphite Residual water effects on performance, electrode electrochemical characteristics, and cell components Results: during start-up from –5°C, residual water did not alter the electrochemical active surface area or charge resistance at low current density; less water was stored in the catalyst layer than in the cell Energy requirement based on one-dimensional thermal model Results: an optimum range exists for current density given a stack design for rapid cold start-up; thermal isolation of the stack reduces start-up time; end plate thickness has no effect beyond a certain threshold; of internal/external heating options, flow of heated coolant above 0°C is the most effective way to achieve rapid start-up Operations under constant current and constant cell voltage conditions Results: water vapor concentration in cathode gas channel affects ice formation in cathode catalyst layer; the membrane plays important role in start-up by absorbing product water and becoming hydrated Ice formation and inner-cell temperature increase dependence on water vapor concentration in cathode gas channel, initial water content in membrane, current density, and start-up temperature Results: ice precipitation can be delayed by decreasing interfacial water vapor concentration at GDL and gas channel surface on cathode side; start-up performances improves by decreasing operation current density, decreasing initial water content in membrane, and increasing start-up cell temperature Water freezing phenomena at interface between GDL and MEA Results: ice formation at the GDL and MEA interface causes air gas stoppage, causing a drop in cell performance

Oszcipok et al., 2005

Ahluwalia and Wang, 2006

Hou et al., 2007

Khandelwal et al., 2007

Meng, 2008a

Meng, 2008b

Ishikawa et al., 2008

(Continued)

Table 17.3 Continued Research area Effect studied and general results

Reference

Cold start

Sun et al., 2008

Cold start © Woodhead Publishing Limited, 2013

Cold start

Cold start

Cold start

Cold start

Cold start

Develop procedure to assist start-up: react hydrogen and oxygen in the FC flow channel to heat it up Results: at temperatures below −20°C, a catalytic hydrogen reaction in FC flow channel is effective and safe way to heat up the FC, hydrogen concentration must be less than 20 vol%; gas flow rate, gas concentration, and active area are the key interdependent factors in this process Initial water in membrane, operating voltage, cell temperature, current Results: ice formation in cathode layer pores and in active reaction sites increases electrical resistance and decreases performance; performance reduces less than 1% per cold start-up Cell voltage, initial water content and distribution, anode inlet relative humidity, heat transfer coefficients, cell temperatures Results: heating-up time can be reduced by decreasing cell voltage; effective purge is critical; humidification of the supplied hydrogen has negligible effect; surrounding heat transfer coefficients significantly affect heating-up time Shut-down strategy importance on freezing of process water on catalyst layer of MEA Results: the degree of dryness in the stack significantly influences cold start-up ability, increasing dryness improves performance; the optimal shut-down strategy allows start-up from −6°C without any performance loss, lower temperatures will see temporary performance loss Adding hydrophilic nano-oxide (SiO2) to catalyst layer of cathode to increase water storing capacity Results: cold start process is strongly related to cathode water storage capacity; SiO2 slightly decreases cell performance under normal operating conditions but drastically improves cold start (−10°C) running time before cell voltage drops to zero; SiO2 does not accelerate cell degradation compared with cells without SiO2 layer Product water: absorbed in ionomer in catalyst layer, taken away as vapor in gas flow, and frozen into ice in catalyst layer pores Results: increasing membrane thickness increases water capacity but decreases water absorption process, increasing ionomer volume fraction increases ionomer water capacity and enhances membrane water absorption; cell start-up is better under potentiostatic condition than galvanostatic condition Ionomer content in catalyst layer in galvanostatic cold start Results: start-up from −30°C improves significantly with higher ionomer content in catalyst layer due to increased oxygen permeation of ice formation in catalyst layer

Pinton et al., 2009 Jiao and Li, 2009

Schiebwohl et al., 2009

Miao et al., 2010

Jiao and Li, 2010

Hiramitsu et al., 2010

Fuel cell vehicles

647

Specific energy (W/kg)

1000

Li-ion

100 Secondary cells 10

DEV,P/E = 4

de, l mo Dua 27 = P/E

NIMH Pb

Power assist, P/E = 83 1 Ultracapacitors 0.1 Alu-elco 0.01 10

100

1000

10,000

Specific power (W/kg)

17.4 Ragone plot and P/E ratio for several electric storage systems (Pede et al., 2004). (DEV – direct electric vehicle; P/E – pulse discharge power/total available energy; NiMH – nickel metal hydride battery; Aluelco – aluminum electrolytic capacitor.)

while Li-ion, NiMH and Pb batteries provide large amounts of energy, they cannot provide the discharge power that is available from a capacitor. Therefore, a battery system combining both secondary cells and capacitors can better satisfy a large spectrum of power needs in FC hybrid vehicles (Pede et al., 2004). Power management strategies determine when the vehicle power will come from the battery or the FC in an FCV application. These strategies affect the fuel consumption and battery life cycle, yet presently there are no standard measures to evaluate the relative performances of different strategies. Power management system determines the power sources on board an FCV based on the battery state of charge (SOC) and the power demand. If the battery is allowed to deplete almost completely, the power management optimizations result in least number of cycles extending the battery life; however, this increases fuel consumption. In the opposite case, fuel consumption can be optimized, while reducing the battery life (Fadel and Zhou, 2011). There are literally hundreds, if not thousands, of specific research interests into the physical operations of hydrogen FCs for passenger vehicles. It is a topic that is expected to grow until 2015–2020, when the first commercial versions of FC cars are expected to be introduced to the light duty vehicle market.

17.3

Environmental impacts of FCVs

While the total cost of FCVs might still be higher than fossil fueled vehicles, the environmental impacts of FCVs are very small compared to fossil

© Woodhead Publishing Limited, 2013

648

Handbook of membrane reactors

fueled vehicles. One detailed paper studied the change in emissions if FCVs come to dominate the US market. It was assumed that fossil fueled on-road vehicles (FFOV) would be replaced with hydrogen FCVs. Emissions were analyzed after production of hydrogen via decentralized steam reforming of natural gas, decentralized electrolysis powered by wind power, and centralized coal gasification. Conservative assumptions were made to strengthen the credibility of results, which were compared against a 1999 vehicle fleet base case (Colella et al., 2005). The reductions in emissions are the true advantage of FCVs over fossilbased technologies. In nearly every case, net quantities of nitrogen oxides (NOx), volatile organic compounds (VOCs), particulate matter (PM2.5 and PM2.5–10), ammonia (NH3), and carbon monoxide (CO) would decrease significantly. The conversion to either electric vehicles or to hydrogen vehicles, with hydrogen derived from natural gas, wind or coal, would reduce the global warming impact of greenhouse gases (GHG) by 6, 14, 23 and 1%, respectively. Remarkably, even for an inefficient hydrogen supply chain, where the FCVs are fueled by natural gas, no carbon is sequestered, and there is a 1% methane leak from feedstock, the scenario still achieves a reduction of 14% in CO2 equivalent GHGs (Colella et al., 2005). Furthermore, transition to hydrogen transportation will improve the local air quality by reducing the emissions of HC, CO, NOx, and PM up to 85%. When the full life cycle of the vehicle fleet is considered, a reduction of 44% in CO2 emissions will be attained. However, when full cradle-to-grave analysis is considered, the reduction in total CO2 is 20%. The emissions are highly dependent on the hydrogen production pathway, and electrolysis from the electric grid is the worst in terms of energy efficiency (Baptista et al., 2010). Greenhouse gas pollution is one of the primary concerns with new vehicle technology. The Intergovernmental Panel on Climate Change has suggested that 60–80% cuts in 1990 light duty vehicle emissions should be required in order to achieve the necessary CO2 reductions. Identifying which future technology platform can achieve this goal is not a simple matter. Projecting forward to 2100, emissions scenarios have a wide range of possibilities, as shown in Fig. 17.5 (Thomas, 2009a). Among the seven scenarios considered, FCEVs is considered the best option, followed by H2 ICE, hybrid electric vehicles (HEV), and BEV options. China is becoming more and more interested in transitioning their vehicle fleet to FCVs as they look forward into this century. In 1999, China imported 23% of its oil demand, largely to support their growing private vehicle fleet. By 2030, if the number of vehicles per 1000 people reaches 100, then there will be an additional demand of 130 million metric tons of oil per year using today’s standards. This equates to more than 50% imported oil needs, creating a serious energy security issue. Furthermore, in downtown Shanghai, fossil fueled vehicles account for 86% of total CO emissions, 96%

© Woodhead Publishing Limited, 2013

Greenhouse gas pollution (light duty vehicles only) (billion tonnes CO2-equivalent/year)

Fuel cell vehicles 100% Gasoline vehicles

2.5

Base case: gasoline hybrid scenario

2.0

Gasoline plug-in hybrid scenario

1.5

1.0

649

1990 GHG

Ethanol plug-in hybrid scenario Battery EV scenario

0.5 GHG goal: 80% below 1990 pollution

H2 ICE HEV scenario

– Fuel cell electric 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100vehicle scenario Year

17.5 Primary model output showing the greenhouse gas pollution (Thomas, 2009a).

of VOC emissions, and 56% of NOx emissions. Converting the private vehicle fleet to FCVs would greatly improve the local air quality in Shanghai (Huang and Xu, 2006). Following a well-to-wheels (WTW) assessment of hydrogen FCVs in Shanghai for ten different supply pathways (Table 17.4), six conclusions were reached. First, all hydrogen supply pathways could reduce emissions by at least 20% compared to petroleum use. Second, all but two hydrogen pathways (#7 and #8) significantly reduce WTW emissions in urban areas. Third, natural gas based pathways have the best energy efficiency (30–58%), electrolysis pathways have the worst (15–21%), and four of ten supply chains have higher energy efficiencies than supply chains using coal. Fourth, changes in WTW greenhouse gas emissions follow WTW energy use almost exactly. Fifth, all pathways achieve significant reductions in CO and VOCs. Other emissions, NOx, PM10 and SO2, can be reduced through some supply chains but not others. Lastly, it was found that the WTW assessment was necessary to adequately evaluate fuel/vehicle systems (Huang and Xu, 2006). China’s concerns about energy security are rightly justified since, internationally, we continue to rely on oil as our primary energy source for transportation. Looking forward to 2100, the demand for oil will greatly surpass the supply should there ever be political unrest in the OPEC nations. Our choice of future vehicle platform will weigh heavily on energy security concerns (Mori and Hirose, 2009). This, of course, favors hydrogen, which can be produced using energy and all primary energy sources. Electric, hybrid electric and FCVs are all important factors for a healthier environment in urban areas (Maggetto and Van Mierlo, 2001). Let us

© Woodhead Publishing Limited, 2013

650

Handbook of membrane reactors

Table 17.4 Supply pathways analyzed by Huang and Zhang (2006) Reference 1. China Statistic Bureau, 2011 2. Huang et al., 2001 3. Spath and Mann, 2001 4. Wang, 1999a 5. Wang, 1999b 6. Wang and Huang, 1999 7. General Motors Corporation et al., 2001a 8. General Motors Corporation et al., 2001b 9. General Motors Corporation et al., 2001c 10. Wang, 2002

Feedstock

Fuel

Petroleum Natural gas

Gasoline GH2 central plant

Natural gas Natural gas Natural gas Petroleum based naptha Petroleum based naptha Coal

GH2 refueling station LH2 central plant LH2 refueling station GH2 central plant LH2 central plant GH2 central plant

Coal

LH2 central plant

Electricity with Shanghai generation mix Electricity with Shanghai generation mix

GH2 refueling station GH2/LH2 refueling station

Source: Huang and Xu (2006).

expand on the idea of total vs urban emissions for a moment. While total emissions are critical for global climate change, urban emissions are a subset of total emissions and have a large impact on human health in cities. The cost of urban emissions can be separated from total emissions and quantified. Current US urban air pollution costs, estimated by various researchers and organizations, are shown in Table 17.5 (Thomas, 2009a). The total cost of urban emissions throughout this century will depend greatly on our choice of future vehicle platform. Figure 17.6 presents such costs for the United States for seven scenarios (Thomas, 2009a). FCVs result in the cleanest option, followed by BEVs, H2 ICE vehicles, and HEVs in that order. The desire to reduce urban emissions has been the driver of many ‘new’ fuels, such as corn or switch-grass based ethanol. One study found that using E85 corn-based ethanol in flexible-fuel vehicles increases total emissions, but reduces urban emissions by up to 30%, because the main emissions are related to farming equipment, fertilizer manufacture and ethanol plants, all of which are typically located in rural areas. HEVs can reduce both total and urban emissions due to higher fuel efficiency. BEVs may increase total PM emissions by 35–325%, but they reduce urban PM emissions by over 40%. FCVs increase both total and urban PM emissions. These results point to the use of BEVs in cities, where the vehicles typically have shorter driving ranges, and ethanol FCVs in suburban and rural areas, where they require longer driving ranges and the emissions have fewer adverse effects on human health (Huo et al., 2009).

© Woodhead Publishing Limited, 2013

© Woodhead Publishing Limited, 2013

Table 17.5 Urban air pollution costs ($/metric tonne) Pollutant

Delucchi average (Lipman and Delucchi, 2003)

Litman (Litman, 2002)

EU AEA (average of four estimates)

EU (Holland and Watkins, 2002)

ANL damage cost

ANL control cost

Average air pollution costs

VOC CO NOx PM-10 PM-2.5 SO2

1086 76 17 129 138 257 165 019 69 094

17 706 534 18 934 6565

2722

3412

3940

11 714

6825 22 750

7860 10 599

16 195 4420 17 319 6005

8450

4733

11 581

7510 1677 13 297 36 835 118 552 21 873

Source: Thomas (2009a).

72 085 15 506

652

Handbook of membrane reactors 100% Gasoline ICVs

US urban air pollution costs ($Billions/year)

70 Base case: gasoline HEV scenario

60

Gasoline PHEV scenario

50 40

Ethanol PHEV scenario

30

H2 ICE HEV scenario

20 10

Battery EV scenario

PM cost from brake and tire wear

2000

2020

2040

Fuel cell electric vehicle scenario

2060

2080

2100

Year

17.6 The US costs of urban air pollution (Thomas, 2009a). (ICV – internal combustion vehicle.)

In Beijing, a life-cycle assessment was performed for 11 supply streams to analyze energy efficiencies and emissions reductions. This study found the most efficient supply chain to be coal gasification and pipeline transport, with a total energy efficiency of 30%. Based on the criteria of global warming, human toxicity, photochemical oxidation, acidification and eutrophication, environmentally the best option was to produce hydrogen via steam reformation of natural gas and pipeline transport, the best overall option was coal gasification with cylinder tank truck delivery when considering the energy, environment and economy in Beijing (Feng et al., 2004). A similar life-cycle assessment of hydrogen FCVs was performed in Canada, again seeking the optimal supply chain. From an environmental standpoint, the best option was found to be wind power production of hydrogen via electrolysis, followed by utilization in a PEFC vehicle (Granovskii et al., 2006b). Another study, focused on types of vehicles, found that an electric car with the capability of onboard electricity generation would be a worthy future investment since it could be nearly environmentally benign (Granovskii et al., 2006a). In the United States, some research is rightly devoted to using coal for transportation, due to the large, indigenous supply of this fossil fuel (approximately 250 years supply at current consumption). Coal can be used to create liquid hydrocarbon fuels, hydrogen or electricity to power ICE, FC or

© Woodhead Publishing Limited, 2013

Fuel cell vehicles

653

BE vehicles. Results of one study found out that coal-to-liquid fuels and coal-to-hydrogen will most likely increase emissions, while coal-to-electricity combined with carbon capture and sequestration could cut emissions in half using short range (60 km) plug-in hybrid electric vehicles (PHEV) for some of the vehicle fleet demand. In reality, this study proves that coal for transportation could be argued for increased energy security (Jaramillo et al., 2009). However, coal-based electricity with carbon sequestration costs as much as, or more than, wind power does today. The cost of photovoltaic electricity is steadily falling, as well. The quality of hydrogen used for transportation is important, as impurities could poison FCs. Therefore, it is necessary to define the guidelines for hydrogen quality, which would of course have an effect on the cost of hydrogen. When produced from natural gas steam reforming, the cost of hydrogen is sensitive to the cost of natural gas. It must also be considered that the cost of quality testing will reflect on the price of hydrogen. Therefore, it is important to find test instruments that are inexpensive yet efficient (Papadias et al., 2009).

17.4

Economic analysis of FCVs

Cost effective, investor friendly economics for FCVs have yet to be demonstrated. Conventional vehicles have had the great advantage of over a century of time to mature to the current status of the market, where consumers expect a vehicle that is reliable, durable, has a long range, high acceleration and good power characteristics. FCVs are still in the research and development phase, so they are not as advanced as fossil technologies. In the Beijing case study, the optimal supply chain involved onboard methanol reforming, although this was not competitive with gasoline powered systems (Feng et al., 2004). In an Austrian case study, FCVs do not look attractive until at least 2030, assuming very favorable key parameters for developing the hydrogen infrastructure (Ajanovic, 2008). A case study in California showed that, even though people have positive attitudes towards hydrogen transportation, a large percentage of them are not willing to pay more for hydrogen transportation (Martin et al., 2009). A reasonable scenario for vehicle introduction results in FCV market penetration in the coming decade, followed by a fast increase until about 2040, resulting in rapid market share control (Thomas, 2009a). This scenario is presented in Fig. 17.7. Before we continue probing the possible FCEV scenarios of this century, it is worthwhile to understand the cost of continuing to use oil as our primary source of transportation energy. In addition to the urban costs of oil use shown in Table 17.5, there are military and economic costs of oil dependence as well. Such costs for the United States are presented in Tables 17.6 and 17.7 (Thomas, 2009a). Table 17.8 presents a summary of the military

© Woodhead Publishing Limited, 2013

654

Handbook of membrane reactors

100%

Percentage of new car sales

90%

Fuel cell electric vehicle (FCEV)

80% 70% 60% 50%

Ethanol PHEVs

40% 30% 20% 10%

Gasoline HEVs

Gasoline ICVs

0% 2005 2015 2025 2035 2045 2055 2065 2075 2085 2095 Year

17.7 Estimated fractions of light duty vehicle sales versus time (Thomas, 2009a). Table 17.6 Estimates of the annual military costs of securing petroleum (US$ billions) Source

Low

High

Klare (Klare, 2004) Copulos, National Defense Council Foundation (Copulos, 2003, 2007) Kimbrell, International Center for Technology Assessment (Kimbrell, 1998) Danks, National Priorities Project (Danks et al., 2008) Average Per barrel military costs based on total oil consumption Per barrel military cost based on imported oil

132 49

150 138

48

113

100

210

82 $11.7/bbl

153 $21.8/bbl

$17.1/bbl

$31.9/bbl

Source: Thomas (2009a).

and economic costs (i.e., societal costs) for the United States. These costs are huge and are the compeling reasons for search for indigenous and clean energy sources. The mounting external costs indicate that the longer we wait to transit away from oil, the more expensive it becomes. Total societal costs for this century, including greenhouse gas pollution, urban air pollution, and economic and military costs of continuing to import oil, for six different scenarios are

© Woodhead Publishing Limited, 2013

Fuel cell vehicles

655

Table 17.7 Estimates of the economic costs of oil dependence (US$ billions) Cost items

Low

High

Transfer of wealth Loss of production capacity Disruption losses Totals Per barrel economic cost based on total oil consumption Per barrel economic cost based on imported oil

100 10 50 160 $22.8/bbl

150 50 170 370 $52.7/bbl

$33.4/bbl

$77.1/bbl

Source: Thomas (2009a).

Table 17.8 Summary of estimated societal costs of US petroleum dependence Cost items

Low

High

Average

Average annual military oil supply protection costs ($US billions/yr) Average annual economic costs of oil dependence ($US billions/yr) Total annual costs of oil dependence ($US billions/yr) Per barrel oil dependence cost based on total oil consumption Per barrel economic cost based on imported oil

82

153

118

160

370

265

242

523

383

$34.5/bbl

$74.5/bbl

$55/bbl

$50.5/bbl

$109/bbl

$80/bbl

Source: Thomas (2009a).

shown in Fig. 17.8 (Thomas, 2009a). Note that these costs are additional to the price consumers pay for their vehicles and refueling. Putting off the transition away from fossil fuels will no doubt increase these costs (Thomas, 2009a). It is clear that FCVs and BEVs are the two best options. According to a life-cycle assessment comparison between FCVs and gasoline vehicles, PEFC efficiency, when using hydrogen produced from steam methane reforming, needed to be 25–30% higher than that of a gasoline power source to be competitive. It would be better for the environment to produce hydrogen from wind power and electrolysis; however, this method was found to depend strongly on the ratio of costs of hydrogen and natural gas. When this ratio is 2:1, production of hydrogen from natural gas is about five times cheaper than that from wind (Granovskii et al., 2006b). Another study compared the economic viability of conventional, hybrid, electric, and hydrogen FC vehicles to determine which would be the cheapest. It was found that economic efficiency of electric cars depends substantially on the source of electricity. If the electricity comes from renewable sources,

© Woodhead Publishing Limited, 2013

656

Handbook of membrane reactors 100% Gasoline ICEVs

Total societal costs ($Billion/year)

600

Base case: gasoline hybrid scenario

500 400 300

Gasoline plug-in hybrid scenario

200

Ethanol plug-in hybrid scenario

100

Battery electric vehicle scenario

2000

2020

2040 2060 Year

2080

Fuel cell electric

2100 vehicle scenario

17.8 Estimates of the total societal costs versus time (Thomas, 2009a).

the electric car is advantageous to the hybrid. If the electricity comes from fossil fuels, the electric car can only be competitive with electricity generation onboard. Electricity efficiency of a gas turbine on the order of 50–60% may also make the electric car advantageous (Granovskii et al., 2006a). One seemingly overlooked option in recent research is the idea of a FC hybrid electric vehicle (FCHEV). This essentially combines the FCV with the BEV. All three platforms utilize an electric drive train, and appear to be the contenders for the vehicle fleet after about 2030. Using this 2030 scenario, one study found out that powertrain life-cycle costs for an FCEV range from $7360 to $22 580, for a BEV range from $6460 to $11 420, and for an FCHEV range from $4310 to $12 540. Also, vehicles in 2030 would be relatively insensitive to electricity costs but quite sensitive to hydrogen costs. The principal advantage of the FCHEV is that it can overcome the short driving range of BEVs using an FC range extender. Also, refueling a hydrogen tank takes minutes, whereas recharging a battery takes hours. Capital cost reduction must continue to be a key target for all three platforms, and recycling of platinum and lithium should be of key concern. It is most important to realize that BEVs and FCEVs are not necessarily antagonistic, either/or options, but both technologies should continue to be supported and pursued (Offer et al., 2010). If the price of energy generated from FC goes below 400 US$/kW, pure FCVs are more profitable than FCHEVs. It is beneficial to couple the FC with a battery in a hybrid system, as the additional power stored in the battery will allow for a smaller FC, as well as allowing the FC run at a stable

© Woodhead Publishing Limited, 2013

Fuel cell vehicles

657

more efficient condition, as the battery can supply the additional power for short bursts of acceleration. The battery also acts as storage for power captured by the regenerative braking. However, the battery has to be large enough to be beneficial as there are also charge−discharge losses, reducing the efficiency of the overall system (Jeong and Oh, 2002). Hydrogen production weighs heavily in the consideration between FCVs and BEVs. At the production scale necessary to produce hydrogen to supply the vehicle fleet, the most economically attractive renewable energy source is wind power, which could contribute about 70% of the total required energy in the US at a cost 40% lower than that of solar photovoltaics. Moreover, Class 4 wind resources (increasing class means increasing average wind speeds) may be utilized more than Class 5 or Class 6 resources because of their proximity to population centers and consequently lower transmission costs. Producing hydrogen via electrolysis, and assuming an electricity price of 4–8 cents/kWh, the hydrogen cost would be $2.75 to $4.50 per gallon of gasoline equivalent. One of the inefficient supply chains for hydrogen is production of liquid hydrogen fuel (which consumes 30% of the heating value), shipping it to distribution centers, and using it to fuel an ICE (Colella et al., 2005). In California, some researchers are considering a novel marketing approach to the FCV. This was premised on the idea that new consumer values must drive hydrogen FCV adoption. This solution is part of a larger idea called Mobile Energy (ME) innovation. This notion accepts the fact that FCVs will not be superior to today’s vehicles on dimensions conventionally valued by consumers, and hence product value must flow from other sources. Hydrogen fueled vehicles have some unique advantages over conventional vehicles that should be emphasized in their marketing. One opportunity for FCVs comes from their ability to produce clean electrical power for uses other than propulsion (Williams and Kurani, 2006). This ME could be used ‘on the go’, ‘in need’, or ‘for a profit’ (Williams and Kurani, 2007). ME is consistent with the slow convergence of transportation and other energy systems. The studies into ME considered household market segments, and found that only about 4 million out of 34 million California residents would most likely be able to adopt ME-enabled FCVs. There does appear to be a trade-off relationship between ME-power and driving range, as well as similar give-and-take situations within the supply framework. However, as questions arise over BEVs, market forces may well be opening the door for ME innovation in the FCV sector (Williams and Kurani, 2006, 2007). One niche role for ME may be in the use of PEFC APUs onboard long-haul trucks. These trucks are idling overnight, but still demand auxiliary power beyond what is produced by the engine. The US has recently passed an anti-idling regulation to limit pollution caused when idle. As a

© Woodhead Publishing Limited, 2013

658

Handbook of membrane reactors

result, there is a window of opportunity for PEFCs. If PEFCs can meet US DoE 2015 targets in terms of efficiency (40%), specific cost (99.999%)

H2

Biofuel + water

Heat

Air

683

Retentate

PEM fuel cell

Wel

Catalytic burner Heat

18.2 Schematic diagram of a steam reformer of biofuel. A composite metallic membrane is used to extract hydrogen and feed a H2/O2 PEM fuel cell.

In particular, the on-board storage of hydrogen for automotive applications remains a critical issue. Pressurized vessels (up to 350 or 700 bars) are potentially dangerous, liquid hydrogen is energy-consuming and metal hydrides suffer from inappropriate gravimetric energy densities (Schlapbach and Züttel, 2001). In this context, there is a need for alternative technologies.

18.1.2

On-board biofuels steam reforming

Steam reforming is a petrochemical process used for the transformation of hydrocarbons and naphtha with water vapor (Sutterfield, 1991). Steam reforming of fossil or biological fuels can be used for the production of hydrogen for application in the automotive industry. In particular, the possibility of using biofuels (such as methanol or ethanol obtained from cultures or vegetal wastes) as a liquid (easy to store) source of CO-free hydrogen to feed polymer electrolyte membrane (PEM) fuel cells has been the subject of intensive research over the past decades. A key step in the process is the extraction of purified hydrogen from the reaction chamber to fuel the H2/O2 PEM fuel cell (Fig. 18.2). For this purpose, composite metallic membranes (CMM) have been developed (Abdollahi et al., 2012). Such membranes contain a thin active layer of palladium alloy supported on an adequate porous structure. High purity (99.99999%) hydrogen is thus produced in accordance with gas-purity requirements for operation in a PEM fuel cell. Electric power can be used directly for motoricity (an average power of 70 kWel is required for a conventional car) or for on-board power generation with auxiliary power units of several kWel (with potential application in trucks or airplanes).

© Woodhead Publishing Limited, 2013

684

Handbook of membrane reactors

The steam reforming reactions of oxygenated hydrocarbons leading to the formation of either CO2 or CO are: CnHmOp + (2n – p) H2O → n CO2 + (2n + m/2 – p) H2 CnHmOp + (n – p) H2O → n CO + (n + m/2 – p) H2 The steam reforming reactions of methanol and ethanol are therefore: CH3OH + H2O → 2H2 + CO2 C2H5OH + 3H2O → 6H2 + 2CO2 The main thermodynamic properties of the methanol and ethanol steam reforming reactions are plotted in Fig. 18.3 for comparison. Both are endothermic. However, they are spontaneous, due to large entropy changes associated with the dissociation of the alcohols. Spontaneous methanol dissociation is obtained at a lower temperature (ΔG < 0 at ≈ 325 K) compared to ethanol (ΔG < 0 at ≈ 475 K), which requires practical dissociation temperatures of 600 K. It is therefore possible to incorporate a palladium permeation membrane into a methanol steam reformer (in such cases, both processes are operating at the same temperature) whereas for ethanol, reforming and extraction of hydrogen are usually performed in two separate 500 400

T⌬S

Energy (KJ.mol–1)

300 ⌬H

200 T⌬ S

100 ⌬K 0 –100 ⌬G

–200 –300 300

400

500

600 700 Temperature (K)

800

900

18.3 Plots of ΔH(T ), ΔG(T ) and T. ΔS(T ) for the steam reforming of methanol (circles) and ethanol (up triangles).

© Woodhead Publishing Limited, 2013

1000

Design and engineering of metallic membranes

685

steps. The nominal gas composition of the ethanol reformate mixture is H2 (40%), H2O (37%), CO2 (9%), CO (8%) and CH4 (6%). Compared to direct combustion, on-board steam reforming of biofuels can potentially offer several advantages. First, the emission of soot, sulfur, nitrogen oxides (NOx) and aromatic compounds is reduced. Second, the biofuel-to-hydrogen yield is high (70–80%) − this is due to the fact that the extraction of hydrogen is used to shift the steam reforming reaction towards the dissociation of the biofuel. In this energy cycle, the carbon dioxide produced by the steam reactor is released into the atmosphere and recycled through biomass. As a result, approximately 60–80% of greenhouse gas emissions can be prevented compared to the direct combustion of fossil fuels. Third, the energy efficiency of the fuel cell (50%) and the high efficiency of the electricity-to-work process are larger than the Carnot efficiency of the combustion. Fourth, it is possible to design very compact reactors.

18.1.3

Challenges

The use of steam reformers for transport application in the automotive industry is not a new concept. Several technologies have been developed and tested, but most have been abandoned due to the resulting complexity and problems of integration. It should be outlined that the concept faces several challenges, from both the material and system viewpoints, due to the specificity of the application and market requirements. From the engineering viewpoint, several constraints are imposed by the application: (i) a high compactness is required and (ii) heat management to meet short starting-up requirements is required. From the material viewpoint, the most significant requirements for the metallic membrane are: (i) cost (ultra thin and defect-free membranes are required); (ii) ability to sustain operating temperature conditions (from –20°C up to +500°C, thermal cycling); (iii) appropriate permeation kinetics (to meet acceleration requirements); and (iv) resistance to corrosion (tolerance to miscellaneous corrosion products issued from the steam reforming of biofuels). So far there is still a need for thin supported membranes that can operate efficiently over an extended temperature range and can sustain a large number of thermal cycles, compatible with the application (5000 h for the automotive industry).

18.2

Membrane materials, manufacturing and reactor design

For any application, the choice of the metal used in the active layer is critical. This choice directly determines the cost, the level of hydrogen purity, the rate of permeation and the operating conditions. According to the literature, group V metals (tantalum, vanadium, niobium) (Buxbaum and

© Woodhead Publishing Limited, 2013

686

Handbook of membrane reactors

Kinney, 1993; Peachey et al., 1996) and palladium (Darling, 1958; Lewis, 1967) are the most commonly used metals for such applications. Although palladium gives the best performance in terms of kinetics, palladium-based membranes are prone to surface corrosion when in contact with several pollutants, such as H2S, CO, CO2 and water vapor (Foley et al., 1993; Ali et al., 1994; Collins et al., 1996). Another problem comes from the elevated (300°C) critical temperature of the Pd–H system. As can be seen from Fig. 18.4, hydrogen dissolves into palladium to form a solid solution (α-PdH) at low hydrogen content on a restricted composition range (from zero up to ≈ PdH0.02 at 298 K). At higher hydrogen content, a substoichiometric (≈ PdH0.6 at 298 K) hydride phase (β-PdH) precipitates and a pressure plateau is observed on the isotherms over a large composition range (from the saturated solid solution up to the composition of the hydride phase). At even larger hydrogen content, a solid solution of hydrogen in the hydride phase is observed. As the temperature increases, the width of the two-phase domain decreases continuously up to the critical value. The three main Pd phases (Pd, α-PdH and β-PdH) have the same face-centered cubic (fcc) structure. At 298 K, the palladium lattice expands continuously from 3.889 Å (Pd) to 3.895 Å (α-PdH) to 4.025 Å (β-PdH) as the hydrogen content increases. The same behavior is observed at higher temperatures. Therefore, the precipitation of the hydride β-PdH produces a significant expansion of the metal lattice, which in turn can lead to the formation of different kinds of defects that can damage the mechanical property of the membrane, especially when thin films are used. This is why palladium membranes are not appropriate for operating at temperatures of less than 300°C. One solution to avoid these problems is to use palladium alloys. For example, palladium alloyed with silver (Hunter, 1956) improves mechanical stability during sorption/desorption cycles (Hunter, 1960) and also contributes to significantly reducing the critical temperature from 300°C down to 57°C for Pd77Ag23. A large variety of binary alloys have been investigated. Palladium−silver (Jayaraman and Lin, 1995), palladium−gold (McKinley, 1967), palladium−copper (Nam and Lee, 2001), palladium−yttrium (Hughes and Harris, 1978; Juda et al., 2000), palladium−ruthenium (Gryaznov et al., 1974), palladium−tantalum (Buxbaum and Kinney, 1996; Peachey et al., 1996) and alloys of other metals (Kajiwara et al., 1999) offer several advantages. They are chemically more stable and the critical temperature is significantly reduced compared to pure palladium. Among these different materials, let us consider in detail Pd–Ag and Pd–Cu alloys.

18.2.1

Palladium−silver alloys

Palladium and silver are miscible in all proportions to form an interstitial solid solution with a cubic face-centered structure. The addition of silver

© Woodhead Publishing Limited, 2013

Design and engineering of metallic membranes

687

28 26

310ºC

24 22 20

290ºC

Pressure (atm)

18 16

270ºC

14 12

250ºC

10 8 6

230ºC

4

200ºC

2

160ºC

0 0.0

0.1

0.2

0.3

0.4

0.5

0.6

H/Pd

18.4 Phase diagram of the Pd−H systems.

is used to dilate the palladium mesh. At 25°C, the mesh parameter varies from 3.889 Å (pure Pd) to 3.990 Å (5% silver) and 3.907 Å (10% silver). Palladium−silver isotherms measured at 50°C for different silver contents are plotted in Fig. 18.5 (Wicke and Brodowsky, 1978). As the silver content increases, the hydrogen solubility decreases and the horizontal two-phase pressure plateau becomes more and more sloppy. The phase diagrams of the Pd–H, Pd95Ag5−H and Pd90Ag10−H systems are plotted in Fig. 18.6 (Uemiya et al., 1991). As the silver content increases, the limit of solubility of hydrogen in the solid solution domain increases and the critical temperature decreases. Values of hydrogen diffusion coefficient and hydrogen solubility in the Pd–Ag alloy are plotted in Fig. 18.7 as a function of the silver content. The addition of silver increases the mesh parameter and reduces the compactness of the unit cell. As a result, hydrogen solubility increases because H−H mean distance increases. On the other hand, the hydrogen diffusion coefficient decreases because the activation energy to jump from one site to the other increases, due to the longer mean distance. Even though the hydrogen coefficient decreases when the silver content increases, the higher solubility is favorable and largely contributes to good permeation properties. Maximum permeability is reached for a silver content between 20 and 25 wt%. The value is approximately twice that

© Woodhead Publishing Limited, 2013

Handbook of membrane reactors 10000

(e)

(d) (b)

(a)

(c) (f)

Pressure (mbar)

1000

100

10

1

0.1

0.01 0.0

0.1

0.3

0.2

0.4

0.5

0.6

0.7

H/M

18.5 Palladium−silver isotherms measured at 50°C for different silver contents: (a) 40 wt% Ag; (b) 30 wt% Ag; (c) 20 wt% Ag; (d) 10 wt% Ag; (e) 5 wt% Ag; (f) pure Pd.

350 Tc

300

250 Temperature (ºC)

688

α

α+β

β

200

150

100

50

0 0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

H/M

18.6 Phase diagrams of (●) pure Pd; (■) Pd95Ag5; (▲) Pd90Ag10. (Tc refers to the critical temperature.)

© Woodhead Publishing Limited, 2013

689

3.5e–5

50

3.0e–5

40

2.5e–5

30

2.0e–5

20

1.5e–5

10

H solubility (cm3 · cm–3 · atm–0.5)

DH (cm2 · s–1)

Design and engineering of metallic membranes

0

1.0e–5 0

10

20 wt % Ag

30

40

18.7 Plot of hydrogen diffusion coefficient and hydrogen solubility in the Pd–Ag alloy as a function of the silver content.

of pure palladium, but decreases when the silver content is further increased (Fig. 18.8) (Knapton, 1977). According to more accurate measurements, the optimum is obtained for Pd77Ag23, a composition commercially available (Shu et al., 1994).

18.2.2

Palladium−copper alloys

The use of palladium−copper metallic membranes (in particular the alloy with a copper content of 40 wt%) for hydrogen purification by permeation is motivated by cost considerations, but also by the need for materials less sensitive to gas corrosion (Scholtus and Hall, 1984; Lide, 1999). PdCu membranes are used in purification units coupled with advanced water gas shift (WGS) membrane reactors (Shu et al., 1991). This kind of reactor is used to increase the conversion efficiency of fuel into hydrogen. But the permeation membrane is in direct contact with the gaseous components of the reactor and must sustain this corrosive environment containing (H2S, CO). Experimental phase diagrams of the PdCu−H system are plotted in Fig. 18.9 (Taylor, 1934; Subramanian and Laughlin, 1991). There is a significant discrepancy between experimental data from one author to the other, especially in the low temperature range (T < 450°C), and therefore, phase boundaries are not accurately known. Li et al. (2008) provided a detailed analysis of the PdCu phase diagram. For T < 600°C and copper contents ranging from 37 to 58 wt%, the structure of the alloy is ordered B2 (CsCl

© Woodhead Publishing Limited, 2013

690

Handbook of membrane reactors 3.0 Ag

Permeability (cm3 cm–2 s–1)

2.5

Ce

2.0 Cu

Au

1.5 1.0 0.5

Ru Ni

0.0

0

10

20

30 40 % Allied element

50

60

18.8 Hydrogen permeability as a function of the content of alloying element at T = 350°C. wt % Cu 70

60

50

40

30

20

700 cfc

60 wt % Pd

Temperature (°C)

600

500 B2 + cfc

400

B2 + cfc

B2

300

200 30

40

50

60

70

80

wt % Pd

18.9 Phase diagram of PdCu: (—) (according to Subramanian and Laughlin, 1991) and (- - -) (according to Taylor, 1934).

type). The system contains two phases (cfc + B2) for copper contents < 37 wt% and > 58 wt% Cu). At T > Tc, a disoriented cfc structure is obtained. For a copper content of 47 wt% (Pd content of 53 wt%), the system is single phased (B2 for T < 600°C and cfc for T > 600°C). At low temperature

© Woodhead Publishing Limited, 2013

Design and engineering of metallic membranes (a)

10000

(b)

(c)

(d)

(e)

1000 Pressure (mbar)

691

100

10

1 0.0

0.1

0.2

0.3

0.4 H/Pd

0.5

0.6

0.7

0.8

18.10 Isotherm of the Pd−Cu−H system for different copper contents at 30°C: (a) 20 wt% Cu; (b) 13 wt% Cu; (c) 9.5 wt% Cu; (d) 6 wt% Cu; (e) pure Pd.

(T < 450°C), the alloy composition 40 wt% Cu (60 wt% Pd) is either in the B2 domain or in the cfc domain. This is a problem because the permeability to hydrogen of the two phases differs significantly. According to Flanagan et al. (2003) the permeability of hydrogen across B2 is higher than the permeability across cfc. Therefore, for such copper content, there is a need to obtain the B2 phase instead of the cfc one. In the literature, the Pd−Cu−H system has been extensively studied for low copper contents (3–20 wt%) in the temperature range of 30–90°C (Burch and Buss, 1974, 1975). At such compositions, the PdCu alloy offers a cfc structure at any temperature. When copper is added to Pd–H, the miscibility gap of Pd–H extends to the ternary system (Fig. 18.10) (Flanagan et al., 2003). Hydrogen solubility and the critical temperature both decrease rapidly when the copper content increases. For a copper content close to 20 wt%, Tc ≈ 30°C, and even lower for a copper content of 40 wt%. This is of particular interest for application in the automotive industry to cope with the problems of cold starting. At 350°C, the PdCu alloy with a copper content of 40 wt% (ordered B2 phase) offers the best permeability to hydrogen (Roa et al., 2002). The values of the hydrogen diffusion coefficients for the different phases of the PdCu system are plotted in Fig. 18.11 (Piper, 1966; Völkl and Alefeld, 1978; Opalka et al., 2007). In the B2 phase, DH is 1000 times larger than in the cfc phase. Also the plot of hydrogen permeance as a function of operating temperature shows a sharp discontinuity at 600°C (Fig. 18.12) (Howard et al., 2004). The discontinuity is caused by the phase

© Woodhead Publishing Limited, 2013

692

Handbook of membrane reactors 10–6

400ºC

Predicted Pd0.5Cu0.5 (bcc)

10–7 Diffusion coefficient (m2 · s–1)

100ºC 25ºC Völkl (1978) Pd0.47Cu0.53 (bcc)

10–8 10–9 10–10 10–11 10–12

Piper (1966) Pd0.475Cu0.0.525 (bcc)

Völkl (1978) Pd0.47Cu0.53 (fcc)

10–13 10–14 10–15 10–16 0.001

0.002

0.003

0.004

0.005

0.006

0.007

0.008

T 1 (K–1)

18.11 Comparison of hydrogen diffusion coefficient values in PdCu (40 wt% Cu) from Piper (1966) and Völkl and Alefeld (1978).

transformation B2 → cfc. According to Völkl and Alefeld (1978) (Fig. 18.11), the activation energy for diffusion in 40 wt% PdCu is 0.035 eV (ten times lower than in the cfc phase) whereas the pre-exponential factor D0 is ten times larger in the cfc phase than in the B2 phase. In addition, the structure of the cfc phase is less compact (a = 0.375 nm) than the structure of the B2 phase (a = 0.297 nm). The mean distance between interstitial sites is lower in the B2 phase than in the cfc phase. As a result, the activation energy for diffusion is lower in the B2 phase and the permeability of the B2 phase is larger. According to the phase diagram, the B2 phase is the most stable one. The higher value of the hydrogen diffusion coefficient is obtained for a copper content of 40 wt% (Piper, 1966): DH = (3.2 ± 0.2) × 10–5 cm2 s−1. This composition is used in practical applications. However, it is close to the limit of stability of the B2 phase. In addition, the hydrogen permeability in B2 PdCu is very sensitive to composition. A change of the Pd content by 1.8 wt% can lead to a 50% reduction of the permeability. The fast decrease of permeability on the copper-rich content is attributed to a fast decrease of hydrogen solubility and hydrogen diffusion (Kamakoti et al., 2005). The fast decrease of permeability on the Pd-rich content is attributed to the phase transformation ordered (B2)/disordered (cfc), which leads to an increase of the less permeable cfc phase (Yuan et al., 2007). For this composition, there is a risk that the B2 phase will not remain stable under thermal cycling up to temperatures larger than 450°C (the minimum temperature used in WGS

© Woodhead Publishing Limited, 2013

Design and engineering of metallic membranes

693

Permeance (mol.m–2 .s–1 .Pa–0.5)

1e–4

B2

cfc

1e–5

1e–6 300

400

500

600

700

800

Temperature (ºC)

18.12 Comparison of the hydrogen permeance of cfc and B2 structures.

reactors). Unfortunately, for temperatures above 450°C, or for larger Pd contents, the less permeable cfc phase is more stable.

18.2.3

Membrane manufacturing

Thick Pd-based membranes (δ > 10 µm) are used for the production of hydrogen of high purity for different applications (laboratory applications, micro-electronics). Thin Pd-based membranes (δ < 5 µm) obtained by new thin layer deposition technologies can potentially be used in a large number of applications, such as fuel desulfurization in the petrochemical industry, production of hydrogen from fossil fuels and subsequent CO2 sequestration. Also, as discussed in the introduction of this chapter, metallic membrane reactors can be used for extracting hydrogen from different gas mixtures for on-board generation of purified hydrogen for transport applications. By coupling the membrane with a steam reforming reactor, it is possible to increase the efficiency of the steam reforming reaction and to simply significantly the design of the reactor. Significant progress has been made over the last decade in the manufacturing of thin membranes. The thickness of the thinnest laminated membranes which are commercially available is usually limited to ≈ 15 µm. Thinner membranes (1 µm < δ < 10 µm) can be obtained by sputtering palladium or its alloys over adequate substrates (physical vapor

© Woodhead Publishing Limited, 2013

694

Handbook of membrane reactors

18.13 SEM micrograph of a Pd–Ag membrane deposited on top of a microsieve.

deposition (PVD), chemical vapor deposition (CVD) magnetron sputtering, roll to toll processes) or using either chemical (electroless) or electrochemical plating processes. Micron and submicron thick membranes require a supporting substrate. Various types of engineered porous supports (ceramics or stainless steel) have been used for that purpose. As can be seen from Fig. 18.13, perforated supports have been fabricated through micromachining and manufacturing techniques borrowed from the integrated circuit industry (Wilwhite et al., 2004). Ceramic porous substrates obtained by the sintering of appropriate powders can also be used (Fig. 18.14). Metallic supports are less brittle but dilatation effects are more significant. As a result, the long-term behavior of metal-supported coatings during thermal cycling is less satisfactory. Two different approaches are usually followed to coat the active layer onto the substrate. The first option is to deposit the active layer directly onto the substrate. To avoid metallic interdiffusion problems at the temperature of operation that may pollute the active layer, ultrathin interdiffusion barriers (e.g., sintered TiO2 particles) are plated between the active layer and porous substrate. Another option is to produce self-supported ultrathin Pd-alloy films by PVD (membranes are formed on different smooth substrates, such as silicon wafers). As a result, mechanically self-supporting resistant thin membranes are obtained (Fig. 18.15). In the second approach, the film is removed from the wafer and transferred to appropriate disks (Lanning and Arps, 2005) or to porous stainless steel supports (Peters

© Woodhead Publishing Limited, 2013

Design and engineering of metallic membranes

695

18.14 SEM micrograph of a ceramic-supported thick (5 µm) Pd film obtained by PVD.

18.15 1 µm thick free-standing PdCu membranes (source: EC FP6 Cachet project).

et al., 2009). This allows the preparation of very thin (≈ 2 µm) membranes supported on macro-porous substrates.

18.2.4

Reactor design

Two major cell designs are commonly used for applications: planar or cylinder. The photograph of a micrometer thick palladium−silver deposited

© Woodhead Publishing Limited, 2013

696

Handbook of membrane reactors

18.16 Pd thin film membrane supported on tubular ceramic substrates.

18.17 Bundle of 24 Pd77Ag23 coated membrane tube.

at the surface of a planar microsieve is pictured in Fig. 18.13. Micrometer thick palladium films, deposited at the surface of porous stainless steel substrates of various diameters and lengths by electroless plating, are shown in Fig. 18.16. There is no practical limitation in terms of diameter and length. Tubes can be associated in parallel to increase the production rate. A bundle of 24 Pd77Ag23 coated membrane tubes is shown in Fig. 18.17. Alternatively, palladium-alloy coatings of given composition can be obtained by first

© Woodhead Publishing Limited, 2013

Design and engineering of metallic membranes H2

Water Reactor

697

Fuel cell

Reformate

Fuel Membrane Retentate

Fuel

Fuel cell

H2

Water

Membrane

Reactor

Retentate

18.18 Schematic diagrams of vaporeforming membrane reactors: the vaporeforming reactor and the gas-separation units are separated (top) and integrated (bottom).

coating the substrate with a foil of the alloying element and then a second foil of palladium, and then by applying an appropriate thermal treatment that will favor metallic interdiffusion and lead to the formation of an active layer of appropriate chemical composition.

18.2.5 Technology developments Two main configurations can be used to design vaporeforming membrane reactors (Fig. 18.18): (i) the membrane is separated from the reactor or (ii) the membrane is coupled with the reactor (steam reformer of WGS). Steam reforming of ethanol requires higher temperatures compared to methanol (e.g., Gallucci et al., 2007) and in such cases the former configuration is more appropriate. The advantage of the latter design is its compactness and the high fuel-to-hydrogen conversion rate, due to the displacement of the vaporeforming reaction. It can also be used for the WGS reaction. It operates at reduced temperatures and is therefore appropriate for the steam reforming of methanol or light hydrocarbons. However, the integration of the membrane into the reactor is more complicated to achieve, and some specific thermo-mechanical problems still have to be solved. Photographs of a methanol steam reformer operating at 400°C (delivery rate up to 250 NLH2 h−1) and of an integrated ethanol reformer operating at 600°C (delivery rate up to 1 Nm3 H2 h−1) are provided in Figs 18.19 and 18.20, respectively.

© Woodhead Publishing Limited, 2013

698

Handbook of membrane reactors

18.19 Photograph of a methanol steam reformer using four 20 µm thick PdCu membranes operating at 400°C and delivering 250 L H2 h−1 (source: CETH).

18.3 18.3.1

Hydrogen permeation mechanism and solubility Permeation mechanism

Permeation of atomic hydrogen across metallic membranes from surface 1 to surface 2 is a multi-step process (Martin et al., 1996) where ‘ad’ refers to surface-adsorption and ‘ab’ to bulk-absorption: H2(g) ↔ H2surface1 (step 1) H2surface1 ↔ 2 Hadsurface1 (step 2) 2 Hadsurface1 ↔ 2 Habsub-surface1 (step 3) 2 Habsub-surface1 ↔ 2 Habsub-surface2 (step 4) 2 Habsub-surface2 ↔ 2 Hadsurface2 (step 5) 2 Hadsurface1 ↔ H2surface1 (step 6) H2surface1 ↔ H2(g) (step 7) •

Step 1: physisorption of molecular hydrogen at the surface of the palladium foil (Weast, 1975): The physisorption step (physical adsorption)

© Woodhead Publishing Limited, 2013

Design and engineering of metallic membranes

699

18.20 Photograph of an ethanol steam reformer delivering 1 Nm3 H2 h−1 operating at 600°C (source: CETH).

is due to a long-distance attractive interaction (van der Waals). The adsorption energy (≈ 10–20 kJ mol−1) is not sufficient to split the H−H bound because the energy of dissociation of H2 (H2 → 2H) is Ediss = 436 kJ.mol−1. However, the interaction between hydrogen and the metal surface significantly modifies the electronic structure of H2 molecules, which are firmly trapped at the interface in a deep potential well (80–120 kJ.mol−1). • Step 2: surface dissociation of physisorbed hydrogen molecules into H ad-atoms: The activation energy Eact must be sufficient to overcome the energy required to form a hydrogen−metal bound. This activation energy is due to the repulsion between the electrons of the hydrogen molecule and surface electrons of the metal. Therefore, the activation energy depends on the chemical composition at the surface of the metal.

© Woodhead Publishing Limited, 2013

700

Handbook of membrane reactors

Once the activation barrier is crossed, hydrogen ad-atoms are adsorbed (Echem ≈ 50 kJ mol−1) and they share their electrons with surface metal atoms. • Step 3: transfer of H ad-atoms to sub-surface regions: Chemisorbed hydrogen ad-atoms can then enter the metal lattice. For transition metals, some d-electrons contribute to the metal bond (Bond, 1962). The involvement of d-electrons contributes to the reduction of the energy barrier (Litovchenko and Efremov, 1999). Palladium has a strong d-character and this is why it can easily split hydrogen molecules. • Step 4: diffusion-controlled transport of atomic hydrogen to bulk regions: Finally, sub-surface hydrogen ad-atoms can diffuse from the surface to bulk regions. At low hydrogen content, a solid solution forms. At higher hydrogen contents, a hydride phase forms (two-phase domains), and step (4) must be modified because H transport takes place over a layer (the growing surface β-PdH during absorption and α-PdH during desorption) of variable thickness and an additional and irreversible step related to the phase transformation process must be added. • Steps 5 to 7 are the same as steps 1 to 3 in the opposite direction: The permeation mechanism can be simplified by considering only the series connection of three main steps: (i) the surface resistance on the upstream side of the membrane due to the surface dissociation of molecular hydrogen and leading to the formation of sub-surface H ad-atoms; (ii) the bulk impedance associated with the transport of hydrogen ad-atoms; and (iii) the surface resistance on the downstream side of the membrane due to the surface recombination of hydrogen ad-atoms into molecular hydrogen and subsequent transport of gas away from the interface.

18.3.2

Hydrogen solubility

The solubility of molecular hydrogen H2 in some materials (e.g., some polymers) is significantly large. This is a non-dissociative dissolution. Membranes of such materials can be used for purification purposes. This is the non-dissociated molecule H2 that diffuses across the membrane with a diffusion coefficient DH 2 . The gradient of H2 chemical potential (pressure) set across the membrane acts as the driving force for the transport. Since it is usually difficult to find a material offering selective gas solubility, ‘molecular permeation’ as it is called lacks selectivity and the concentration level of impurities in the gas released after permeation remains significantly high. There is a linear relationship between hydrogen partial pressure in the feed gas PH 2 and hydrogen concentration DH 2 in the

© Woodhead Publishing Limited, 2013

Design and engineering of metallic membranes

701

membrane, given (at least for the low hydrogen concentration domain) by Henry’s law: CH 2

H H 2 (T )p pH 2

[18.1]

H H 2 (T ) expressed in mol m−3 Pa−1 is Henry’s constant. This is mostly a function of operating temperature and chemical composition of host material. Equation [18.1] is usually verified by ideal systems, at low concentration, when there is no interaction between hydrogen and the host matrix. Hydrogen dissolution in metals (group V metals, palladium, Pd-alloys) is a dissociative process leading to the formation of surface hydrogen adatoms that diffuses across the membrane with a diffusion coefficient DH. Assuming that hydrogen behaves as an ideal gas, its chemical potential is:

µ H 2 = µ DH 2 + RT ln pH 2

[18.2]

In the metal, at infinite dilution, the chemical potential of atomic hydrogen is:

µ H = µ DH + RT ln CH

[18.3]

At equilibrium:

µ H2

2µ H

[18.4]

Combination of Equations [18.1–18.4] yields the so-called Sieverts’ relationship: CH

Ks (T ) pH 2

[18.5]

where ⎛ µ oH − 2µ oH ⎞ K s (T ) = exp ⎜ 2 ⎟ ⎜⎝ 2 RT ⎟⎠

[18.6]

KS, expressed in mol m−3 Pa−1/2, is Sieverts’ constant and in Pa is the partial pressure of hydrogen in the gas phase. Equation [18.5] is an ideal limiting case that pertains to diluted solid solutions formed at low pressures with defect-free metallic hosts. As the hydrogen content increases, H−H interactions come into play that may introduce deviations.

© Woodhead Publishing Limited, 2013

702

Handbook of membrane reactors

Both Equations [18.1] and [18.5] describe ideal cases. In many experimental situations, sorption isotherms are non-linear, showing a negative behavior compared to Henry’s law. Some cases can be described using a dual sorption theory (Paul and Koros, 1976) in which the global sorption capacity is the result of two different contributions, one that follows Henry’s law and a second that follows Langmuir behavior: CH 2

H 2 ( ) pH 2 +

′ b pH CH 2

[18.7]

1 + b pH 2

′ where CH is the maximum void sorption concentration and b an affinity constant.

18.4

Permeation kinetics

Three different cases of rate-determining step are encountered (Fig. 18.21): (a) the surface step is rate determining (the concentration gradient across the membrane is zero but the rate is infinite because the diffusion coefficient is infinite; in the case considered here, surface resistances on both membrane sides are equal); (b) the bulk diffusion step is rate determining (interfaces are at equilibrium with the gas phase, whatever the gas permeation flow); and (c) this is the most common intermediate case, where both surface and bulk rate contributions are playing a role.

18.4.1

Surface step is rate determining

A first limit rate expression is obtained when surface rate contributions are rate determining. The concentration profile is plotted in Fig. 18.22. Although the use of Henry’s law (Equation [18.1]) as boundary conditions is usually limited to the case of molecular diffusion (polymer membranes), it can also be used to describe permeation across metallic membranes with surface rate-determining step (rds). In such cases, the dissociative physisorption step of H2 into H is assumed to be fast and at equilibrium. Steps (3) and (5) of the sorption mechanism are rds and the relationship between surface hydrogen ad-atoms and pressure is given by Equation [18.8]: CH

C H2

H H 2(T )pH 2

[18.8]

(

)

where H H 2 (T ) expressed in mol m−3 Pa−1 is Henry’s constant. Surface kinetics on face 1 is: J1H k1 C1 C x 0 where k1 in m s−1 is the surface rate constant. Surface kinetics on face 2 is: J 2H k2 C x* l C2 where k2 in m s−1 is the surface rate constant. The concentration gradient

© Woodhead Publishing Limited, 2013

(

)

Design and engineering of metallic membranes CH

CH

CH

(a)

(b)

0

1

x

703

0

(c) 1

x

0

1

x

18.21 Rate-determining steps: (a) surface; (b) bulk; (c) mix control.

CH P1

C1

C*

C*

C2 0

P2 x

1

18.22 Schematic diagram of the concentration profile in a surface-controlled permeation process.

across the membrane is close to zero (the hydrogen diffusion coefficient is large to insure a fast transport of hydrogen atoms from one side to the other). The hydrogen flux by diffusion across the membrane is therefore: JH

(

2 J H 2 = k1 C1 − C x*

0

)

(

) = (k C

k C x*

l

d C2

2 H H 2 (T ) p2

C

− k2C2 ) − C * (k1 − k2 )

Since: C1

2

(T ) p1

then: JH

2 J H 2 = 2 H H 2 (T ) ( k1 p1 k2 p2 ) C * (k1 k2 )

© Woodhead Publishing Limited, 2013

704

Handbook of membrane reactors

when the kinetics on both sides are the same (this is usually the case, since this is the same material unless during permeation the upstream side of the membrane is corroded due to exposure to poisonous species), then k1 = k2 and: JH

2 J H 2 = 2k1

setting Φ H = 2 k1 finally obtain: JH

H 2 (T )

H 2 (T )

( p1

p2 )

[18.9]

= membrane permeance in mol m−2 s−1 Pa−1, we

2 J H 2 = Φ H ( p1

p2 )

[18.10]

From the measurement of the permeance, it is in principle possible to determine the value of the surface rate constant when the hydrogen solubility is known. This permeance is independent of membrane thickness since only surface steps are rds.

18.4.2

Bulk diffusion step is rate determining

A second limit rate expression is obtained when surface rate contributions are negligible (Fig. 18.23). This is usually the case (i) at elevated temperature (T > 200°C); (ii) with relatively thick membranes (several tens of microns); and (iii) when permeation takes place under a small difference of pressure between the upstream and downstream sides of the membrane. In such cases, it is assumed that the concentration of atomic hydrogen in the CH P1

C1

P2

C2 0

1

x

18.23 Schematic diagram of the concentration profile in a bulk diffusion-controlled permeation process.

© Woodhead Publishing Limited, 2013

Design and engineering of metallic membranes

705

sub-surface region of the gas−metal interface is in rapid equilibrium with the partial pressure of hydrogen in the gas phase and that there is neither a concentration gradient in the gas phase (at least when pure or strongly concentrated hydrogen is used) nor kinetic contributions of the surface step. From a general viewpoint, transport of a hetero-element in a given matrix is governed by the so-called Gorsky’s effect (Sakamoto et al., 1991). The flow of the hetero-element induces an elastic deformation of the host matrix, in particular when the size of the diffusing species is large compared to the size of diffusion sites. In the case of hydrogen permeation, the operating temperature is usually significantly larger than the critical temperature of the metal−hydrogen system. Hydrogen dissolution leads to the formation of a solid solution (not a hydride phase) and the deformation of the host matrix remains limited. Thus, Fick’s laws of diffusion are sufficiently accurate to describe hydrogen transport. According to Fick’s first law of diffusion, the flux of atomic hydrogen JH inside a homogeneous and isotropic media is a function of the gradient of chemical potential, which in turn is a function of the gradient of concentration: d(DH CH )

JH

[18.11]

JH is only a function of the atomic hydrogen diffusion coefficient and of the local concentration gradient. When atomic hydrogen diffusion takes place across a membrane of planar geometry, the transient hydrogen concentration CH(x,t) is a function of one space coordinate perpendicular to the surface (x axis) and time (Fig. 18.23). Assuming that the hydrogen diffusion coefficient DH is independent of atomic hydrogen concentration (an assumption acceptable over at least a limited range of hydrogen concentration), the flux of atomic hydrogen JH (mol H s−1 cm−2) is given by: JH

DH

dC H dx

[18.12]

In stationary conditions of flux (a constant difference of pressure is set across the membrane of thickness δ), the flux of molecular hydrogen is constant and given by: JH2 =

JH C C2 = DH 1 2 δ

[18.13]

where: •

C1 and C2 (mol m−3) are the hydrogen concentration on the respectively upstream and downstream sides of the membrane;

© Woodhead Publishing Limited, 2013

706

Handbook of membrane reactors

• DH (m2 s−1) is the diffusion coefficient of atomic hydrogen in the metal; • δ (m) is the thickness of the metallic membrane. In order to obtain a quantitative relationship between the gradient of pressure set across the membrane and the flux of molecular hydrogen, it is necessary to relate the hydrogen pressure in the gas phase and the hydrogen concentration in the metal. Combining Sieverts’ equation (which pertains to equilibrium) and Fick’s first law of diffusion (which pertains to the transport of atomic hydrogen by diffusion), the following relationship, known as Richardson’s equation, is obtained (Lewis, 1967; Buxbaum and Marker, 1993): JH2 =

J H DH KS (T) 1 2 = ( p1 2 δ

p21 2 )

[18.14]

The product Pe H = DH KS , in mol m−1 s−1 Pa−1/2, is called the hydrogen permeability, and Pe H = 2 PeH 2. Therefore, the hydrogen flux can be rewritten as: JH2 =

J H PeH 1 2 = ( p1 2 δ

p21 2 )

[18.15]

In Equation [18.15], the ratio PeH/δ represents the hydrogen permeance ΦH, in mol m−2 s−1 Pa−1/2, which takes into account the thickness of the membrane, and so it can be used for the direct comparison of the performances of membranes having different thicknesses: ΦH =

DH KS PeH = δ δ

[18.16]

Therefore, the hydrogen flux is: JH2 =

18.4.3

JH = Φ H ( p11 2 2

p21 2 )

[18.17]

Differentiation of surface and bulk rate-determining step

From the above considerations, a plot of hydrogen flow rate as a function of the difference of pressure or the difference of pressure roots should be used to assess the presence of either surface or bulk rate-determining steps.

© Woodhead Publishing Limited, 2013

Design and engineering of metallic membranes

JH

707

Q l

2

n P1n – P2n (Pa)

18.24 Plot of the hydrogen flow vs the difference of pressure. n = 1 for surface rds and n = 0.5 for bulk rds.

However, experimentally many other factors (such as the existence of support, pine-holes and cracks, surface of grain boundary contamination, etc.) can also play a role. Such diverse contributions are usually accounted for by considering the more general Equation [18.18]: JH2 =

JH = Φ H ( p1n 2

p2n )

[18.18]

where n = 1 for surface rds and n = 0.5 for bulk diffusion rds. Intermediate values of n are supposed to account for a mix rate control, or for the presence of other factors listed above. In Equation [18.18], the slope of the n hydrogen flux J H 2 versus ( 1n 2 ) is equal to the membrane permeability (Fig. 18.24). Based on these considerations, it is usually considered that, for membranes a few microns thick or more, diffusion-controlled transport of hydrogen is rate determining (Ward and Doa, 1999). With submicron membrane thicknesses, it is generally considered that surface steps are rate determining. However, the use of Equation [18.18] to make a distinction between surface and bulk rate limitations is not always a simple task. This is not totally surprising, due to the number of underlying hypotheses. Many contradictory studies have been reported in the literature on the subject, and the minimum membrane thickness for which the rds shifts from surface to bulk differ from one author to the other. According to some authors (Kishimoto et al., 1986), the hydrogen flow across membrane films thicker than 100 nm follows Equation [18.17]. On the contrary, other authors (Jayaraman and Lin, 1995) show that membrane thicknesses must be larger than 350 nm to

© Woodhead Publishing Limited, 2013

708

Handbook of membrane reactors

obey Equation [18.17]. It has also been observed that the presence of macroscopic defaults and even pinholes in the membrane could lead to a rate law that follows Equation [18.17] (in such cases, this is logical because there is no surface dissociation/recombination of H2). When palladium alloys are used, surface segregation of the doping element sometimes leads to a decrease in the concentration of surface active sites for adsorption and as a result, the kinetics of surface dissociation decreases in spite of low thicknesses. In such cases, surface steps are rate determining and the hydrogen flux follows Equation [18.10]. Finally, it can happen that the experimental plots of the hydrogen permeation flows as a function of the difference of pressure or as a function of the difference pressure roots are both linear (at least on limited temperature ranges), and it is therefore difficult to reach clear conclusions in such cases. And when the permeation rate is under the mix control of surface and bulk steps, under transient conditions of flow, the situation becomes even worse. A technique that can provide a separate measure of surface resistances and a measure of hydrogen diffusion coefficient is therefore required (see next section).

18.5 18.5.1

Membrane characterization and performances Experimental characterization of metallic membranes

Basically, two different kinds of experiments can be carried out to analyze the kinetics of hydrogen absorbing metallic membranes (Fig. 18.25): (i) permeation experiment (the membrane is positioned between two reaction chambers) and (ii) sorption experiments (the membrane is not positioned between two reaction chambers but in a dead-end reactor). Boundary conditions differ but the same information can be obtained from the two kinds of experiment. During a permeation experiment, the sum of the absorption resistance (on the feed side) and the desorption resistance (on the permeate side) is measured. During an absorption experiment, the sum of the absorption-surface resistances of each side is measured. During a desorption experiment, the sum of the desorption-surface resistance of each surface is measured. There is no possibility to separately measure one surface resistance at a time. An example of the experimental set-up used for the measurements is shown in Fig. 18.26. Basically, this set-up is an adaptation of the original Sievert, volumetric technique (Fig. 18.27). It can be used to record experimental isotherms (thermodynamics) and collect kinetic data during permeation or sorption (absorption/desorption) experiments, by sequentially transferring known amounts of hydrogen from (to) the calibrated reference volume chamber (Ch1) to (from) the reaction chamber (Ch2). A reservoir

© Woodhead Publishing Limited, 2013

Design and engineering of metallic membranes Desorption

Permeation

Feed

709

Permeate

Absorption

18.25 Comparison of permeation (left) and sorption (right) processes.

18.26 Photograph of the experimental set-up used to measure gas-phase impedances.

© Woodhead Publishing Limited, 2013

710

Handbook of membrane reactors P2 T2 NV T2⬘

PV1

Ch2 MV2

P1

P3 TR

PV0

MV3

M

P0

T1 Ch1 Ch3 T3 Ch0

18.27 Schematic diagram of the experimental set-up of Fig. 18.26 used to measure gas-phase impedances. (T – temperature; MV – manual valve; NV – needle valve; P – pressure; Ch – volume chamber.)

(Ch0) is used as hydrogen source (sink) and the set-up can be gas-purged down to secondary vacuum levels (1 × 10−6 mbar) using a turbo-molecular pumping station. Gas transfer experiments are fully automated using pneumatic valves. The reaction chamber that comes apart from the set-up can be introduced in a glove box for the purpose of loading the sample (into Ch2) in an inert atmosphere. Pressure transmitters of different pressure range can be used to increase the accuracy of pressure measurements. Their choice depends on the hydride studied and on the pressure plateau value at the temperature of investigation. The driving force to gas transfer experiments is the initial pressure difference Δp° = p1° – p2°, set between Ch1 and Ch2: Δp° > 0 for absorption and Δp° < 0 for desorption. A needle valve, with a metering tip connected to a micrometer handle, is used to slow down the rate of gas transfer and maintain the temperature at constant values. During each experiment, transient pressure signals p1(t) and p2(t) in both the reference and the reaction chambers are sampled for real-time or post calculation of the pneumato-chemical impedance. The time-dependent hydrogen mass flow between Ch1 and Ch2 can be directly measured using an on-line mass flow meter (not represented in Fig. 18.27). Alternatively, it can be calculated using the needle valve which, after calibration, acts as a mass flow meter, or more simply, by derivation of the transient p1(t), assuming that the law of perfect gas prevails.

© Woodhead Publishing Limited, 2013

Design and engineering of metallic membranes

711

0.5 ⌬P = 3 bar 0.4 JH (mol.s–1 .m–2)

⌬P = 2.5 bar ⌬P = 2 bar

0.3

⌬P = 1.5 bar 0.2 ⌬P = 1 bar 0.1 ⌬P = 0.5 bar 0.0 100

150

200

250 300 Temperature (ºC)

350

400

450

18.28 H2 flow rate as a function of temperature for different pressure gradients.

18.5.2 Time-domain analysis of the kinetics of permeation The conventional way of analyzing the kinetics is to measure the permeation rate in stationary conditions of flow (sorption/desorption experiments are used for analysis in transient modes due to the limited volume of the membrane). Typical permeation curves measured on a B2 PdCu (40 wt% Cu) membranes are plotted in Fig. 18.28. The hydrogen flow increases steadily up to 300°C and then a plateau forms. This is related to the phase transformation of the B2 phase into the cfc phase which, in the presence of hydrogen, occurs at much lower temperatures than without hydrogen. Permeation curves for the two rate limiting cases (n = 1 for surface rds and n = 0.5 for bulk rds) are plotted in Figs 18.29 and 18.30, respectively. The linearity criterion is clearly respected for the second plot, suggesting that bulk diffusion is rds. From the slopes of the different curves of Fig. 18.30, it is possible to calculate the hydrogen diffusion coefficient when the hydrogen solubility is known (Equation [18.17]). Results are plotted in Fig. 18.31. The hysteresis (different DH values between the two temperature scans) is due to the fact that the measurements were made close to the two-phase domain.

© Woodhead Publishing Limited, 2013

Handbook of membrane reactors 0.5 395ºC 300ºC

JH (mol.s–1 .m–2)

0.4

250ºC 0.3 190ºC 0.2 150ºC 0.1

0.0 0

1

2 p1 – p2 (bar)

3

4

18.29 Plot of hydrogen flow as a function of the difference of pressure set across the metallic membrane.

0.5

395ºC 300ºC

0.4 JH (mol.s–1 .m–2)

712

250ºC 0.3 190ºC 0.2 150ºC 0.1

0.0 0

100

200

300

p11/2 – p21/2 (Pa0.5)

18.30 Plot of hydrogen flow as a function of the difference of the square root of pressure set across the metallic membrane.

© Woodhead Publishing Limited, 2013

400

Design and engineering of metallic membranes

713

1.2e–4 Up 1.0e–4 Down

DH (cm2 s–1)

8.0e–5 6.0e–5 4.0e–5 2.0e–5 0.0 0

50

100

150 200 Temperature (ºC)

250

300

350

18.31 Plot of the hydrogen diffusion coefficient in PdCu (40 wt% Cu) as a function of temperature.

18.5.3

Frequency-domain analysis of the kinetics of permeation

The experimental set-up of Figs 18.26 and 18.27 can also be used to analyze the kinetics in the frequency domain (sorption experiments are preferred although permeation experiments can also be performed). This so-called ‘pneumato-chemical impedance spectroscopy’ is used to separately measure surface and bulk membrane impedances and to obtain quantitative values for the surface resistance and the bulk hydrogen diffusion coefficient. Details about the technique and mathematical models used to determine rate parameters can be found in Millet and Ngameni (2011). To illustrate the interest of the technique, impedance diagrams measured on a non-recrystallized Pd77Ag23 sample are plotted in Fig. 18.32. The diameter of the high frequency semicircle is related to the surface resistance of the sorption process. The diameter significantly decreases when the temperature increases, indicating that the surface step becomes increasingly fast. The impedance is reduced by a factor of almost ten when the temperature is raised from 60°C to 250°C. After recrystallization the impedance of the sample increases significantly (Fig. 18.33). The same trend has been observed over the entire temperature range of investigation (60–250°C). By fitting experimental impedance diagrams with model equations, it is possible to obtain accurate measurements of the surface resistance RS and of the hydrogen diffusion coefficient DH (Millet et al., 2011). Data are plotted in

© Woodhead Publishing Limited, 2013

714

Handbook of membrane reactors 4e+10 60ºC

–Im (Pa.mol–1 .s)

3e+10

2e+10 73ºC

1e+10

80ºC

210ºC

150ºC

0 0

1e+10

2e+10

3e+10

4e+10

Re (Pa.mol–1 .s)

18.32 Experimental impedance diagrams measured on a 125 µm thick Pd–Ag (23 wt% Ag) at H/M ≈ 0.3 and different temperatures.

Fig. 18.34. The hydrogen diffusion coefficient rises exponentially with temperature (Arrhenius behavior). Before recrystallization, the following value is obtained: ln DH ( ) = −2703 ×(( / T )

.09

[18.19]

From Equation [18.19], an activation energy EA = +23 ± 2 kJ mol−1 and a pre-exponential factor D0 = (2.3 ± 0.2) × 10−3 cm2 s−1 can be determined for this material in these operating conditions. These values are consistent with measurements made with other techniques (Völkl and Alefeld, 1978; Wicke and Brodowsky, 1978). After recrystallization, the following value is obtained: ln DH

2285 (1 / T ) − 7.73

[18.20]

From Equation [18.20], an energy of activation EA ≈ 19 ± 2 kJ mol−1 and a pre-exponential factor D0 = 4 × 10−4 cm2 s−1 are obtained.

© Woodhead Publishing Limited, 2013

Design and engineering of metallic membranes

715

5e+10

4e+10

–Im (Pa.mol–1 .s)

(b) 3e+10

2e+10

1e+10 (a) 0 0

1e+10

2e+10 Re

3e+10

(Pa.mol–1

4e+10

5e+10

.s)

18.33 Impedance diagrams measured during a H2 sorption experiment on a 125 µm thick Pd77Ag23 membrane (a) before and (b) after recrystallization using the same driving force ∆P = 150 mbar at T = 150°C.

The surface resistance recrystallization: d ⎡⎣ log(

S )⎤ ⎦

dT

decreases

with

temperature.

Before

= − 7.2 × 10 −33 P Pa moll 1s K −1

After recrystallization: d ⎡⎣ log(RS )⎤⎦ dT

= −9.1 × 10 −3 P Pa moll 1sK

1

From the literature, it is known that hydrogen transport takes place preferably at grain boundaries. The diminution of the activation energy (from 23 ± 2 kJ mol−1 before, to 19 ± 2 kJ mol−1 after recrystallization) may be related to the decrease of the free enthalpy of the system during recrystallization.

© Woodhead Publishing Limited, 2013

716

Handbook of membrane reactors

1e+11

1.4e–5 (a)

(b)

1.0e–5

(a)

8.0e–6 1e+10 6.0e–6

DH (cm2 .s–1)

Rs (Pa.mol–1 .s)

1.2e–5

4.0e–6 2.0e–6

(b)

0.0

1e+9 50

100

150

200

250

Temperature (°C)

18.34 Plot of the surface resistance and of the hydrogen diffusion coefficient as a function of operating temperature measured on 125 µm thick Pd77Ag23 membranes at H/M ≈ 0.3: (a) before and (b) after recrystallization.

The reduction of grain boundary areas could also contribute to reducing the kinetics of surface dissociation of hydrogen molecules. In addition, silver is prone to surface segregation at grain boundaries during the recrystallization process, leading to an increase of surface resistance. These results clearly point out that surface and bulk microstructure of the membrane play significant roles in the kinetics of hydrogen permeation. Both surface and bulk steps are impacted by the thermal treatment and operating conditions. For a better understanding of surface and bulk contribution to the permeation process, it is necessary to analyze in detail the role played by surface structure and grain size and distribution. It should be noted finally that many experimental results obtained at the laboratory-scale have been obtained using pure hydrogen. Of course, the objective is to extract hydrogen from miscellaneous gas mixtures. As the hydrogen concentration in the feed gas tends to decreases, mass-transport limitations of hydrogen up to the membrane surface become more and more significant. To a certain extent, the problem can be simplified by promoting turbulent conditions of flow in the vicinity of the membrane surface, but there is a limit beyond which it is not possible to go. In such cases, significant mass-transport impedance is observed. Such impedance can be put into evidence and measured using pneumato-chemical impedance spectroscopy. The measuring technique of course does not offer a solution to reduce that impedance but can be used

© Woodhead Publishing Limited, 2013

Design and engineering of metallic membranes

717

Temperature (ºC) 560

496

Permeance (mol.m–2 .s–1 .Pa–1)

2.5e–5

2.0e–5

440

394

352

315

283

1 µm (Nam, 1999)

0.5 µm (Jan, 1999)

1.5e–5

1.0e–5 2 µm (Nam, 2000) 1 µm (Zhao, 1999)

5.0e–6 2 µm (Yan, 1994)

4 µm (Itoh, 2000)

0.0 1.1

1.2

1.3

1.4

1.5

1000 ×

1.6

1.7

3 µm (Abdollahi, 2011) 1.8

1.9

T–1 (K–1)

18.35 State-of-art membrane permeance values.

to measure its value under different operating conditions and help the engineer in his attempt to design efficient reactors and optimize mass-transport conditions.

18.5.4

Membrane performances

Some reference values of membrane permeance for thin Pd–Ag membranes (surface rds) are plotted in Fig. 18.35. As expected, the thinner the membrane and higher the operating temperature, the higher the hydrogen permeance is, although some inversions are observed due to the membrane microstructure.

18.6 18.6.1

Customized membranes for application in the automotive industry Requirements and constraints

A 60 g H2 h−1 is required to feed a 1 kWe fuel cell. A fuel cell of 70 kWel is required to power a conventional vehicle. Typical process requirements for application in the automotive industry are compiled in Table 18.1.

© Woodhead Publishing Limited, 2013

718

Handbook of membrane reactors

Table 18.1 Main characteristics of the purification unit for application in vehicular transport Operating temperature (°C) Input pressure (bar) Delivery pressure (bar) H recovery (%); H2 recovered/H2 in the fuel Hydrogen purity (%) Max CO level (ppm) Permeance coefficient (mol m−1 s−1 Pa−1/2 at 700°C) Membrane surface (m2/kWel) Thickness of porous substrate (mm) Thickness of active (Pd) film (µm) Maximum amount of Pd (g/kWel) Compactness of gas separator (l/ kWel) Nominal delivery rate (min); time required for the separator to produce H2 at nominal rate Time for response (sec); response time of the separator when the load change from 10% to 90% Lifetime (h) Thermal cycling (number of cycles)

18.6.2

750 15 2 > 90 > 99.999 < 10 < 10–8 < 0.015