ASM Handbook, Volume 24: Additive Manufacturing Processes 1627082883, 9781627082884

Citation preview

W

ASM Handbook

Volume 24 Additive Manufacturing Processes Prepared under the direction of the ASM International Handbook Committee

Volume Editors David L. Bourell, University of Texas at Austin William Frazier, Pilgrim Consulting LLC Howard Kuhn, University of Pittsburgh Mohsen Seifi, ASTM International

Division Editors Allison M. Beese, Pennsylvania State University David L. Bourell, University of Texas at Austin Howard Kuhn, University of Pittsburgh Ming Leu, Missouri University of Science and Technology Eric MacDonald, Youngstown State University

ASM International Staff Victoria Burt, Content Developer Steve Lampman, Content Developer Robert Meyer, Content Developer Amy Nolan, Content Developer Susan Sellers, Content Developer Madrid Tramble, Manager of Production Vince Katona, Production Coordinator Jennifer Kelly, Production Coordinator Karen Marken, Senior Managing Editor Scott D. Henry, Senior Content Engineer Editorial Assistance Ed Kubel Elizabeth Marquard

ASM InternationalW Materials Park, Ohio 44073-0002 www.asminternational.org

Copyright # 2020 by ASM InternationalW All rights reserved No part of this book may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the written permission of the copyright owner. First printing, August 2020 Second printing, September 2020

This Volume is a collective effort involving hundreds of technical specialists. It brings together a wealth of information from worldwide sources to help scientists, engineers, and technicians solve current and long-range problems. Great care is taken in the compilation and production of this Volume, but it should be made clear that NO WARRANTIES, EXPRESS OR IMPLIED, INCLUDING, WITHOUT LIMITATION, WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, ARE GIVEN IN CONNECTION WITH THIS PUBLICATION. Although this information is believed to be accurate by ASM, ASM cannot guarantee that favorable results will be obtained from the use of this publication alone. This publication is intended for use by persons having technical skill, at their sole discretion and risk. Since the conditions of product or material use are outside of ASM’s control, ASM assumes no liability or obligation in connection with any use of this information. No claim of any kind, whether as to products or information in this publication, and whether or not based on negligence, shall be greater in amount than the purchase price of this product or publication in respect of which damages are claimed. THE REMEDY HEREBY PROVIDED SHALL BE THE EXCLUSIVE AND SOLE REMEDY OF BUYER, AND IN NO EVENT SHALL EITHER PARTY BE LIABLE FOR SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES WHETHER OR NOT CAUSED BY OR RESULTING FROM THE NEGLIGENCE OF SUCH PARTY. As with any material, evaluation of the material under end-use conditions prior to specification is essential. Therefore, specific testing under actual conditions is recommended. Nothing contained in this Volume shall be construed as a grant of any right of manufacture, sale, use, or reproduction, in connection with any method, process, apparatus, product, composition, or system, whether or not covered by letters patent, copyright, or trademark, and nothing contained in this Volume shall be construed as a defense against any alleged infringement of letters patent, copyright, or trademark, or as a defense against liability for such infringement. Comments, criticisms, and suggestions are invited, and should be forwarded to ASM International. Library of Congress Cataloging-in-Publication Data ASM International ASM Handbook Includes bibliographical references and indexes Contents: v.1. Properties and selection—irons, steels, and high-performance alloys—v.2. Properties and selection—nonferrous alloys and special-purpose materials—[etc.]—v.24. Additive manufacturing processes 1. Metals—Handbooks, manuals, etc. 2. Metal-work—Handbooks, manuals, etc. I. ASM International. Handbook Committee. II. Metals Handbook. TA459.M43 1990 620.1’6 90-115 SAN: 204-7586 ISBN-13: 978-1-62708-288-4 (print) ISBN-13: 978-1-62708-289-1 (pdf) ISBN-13: 978-1-62708-290-7 (electronic)

ASM InternationalW Materials Park, OH 44073-0002 www.asminternational.org Printed in the United States of America

Foreword Whenever a new volume is considered for addition to the ASM Handbook series, great care and thought are taken to ensure the subject matter meets a demand from the market and that the leading technical experts are at the helm to develop the content. ASM International’s latest effort, ASM Handbook, Volume 24, Additive Manufacturing Processes, hits both marks. The list of contributors reads like a who’s who of the additive manufacturing world, and together they have created a comprehensive compendium of reference information on the growing topic of additive manufacturing. This handbook is a practical resource covering the processes used to additively manufacture polymers, ceramics, and metals, including direct-write methods. ASM International is grateful for the work and dedication of volunteer editors, authors, and reviewers who devoted their time and expertise to develop a reference publication of the highest technical and editorial caliber. A special note of thanks is offered to the volume and division editors who put forth extraordinary efforts to keep this project focused and completed on schedule.

Dr. Zi-Kui Liu, FASM President ASM International Ron Aderhold Acting Managing Director ASM International

v

Contributors Magnus Ahlfors Quintus Technologies

Corson Cramer Oak Ridge National Laboratory

Heinrich Kestler Plansee SE

Arulselvan Arumugham Akhilan University of Louisville

Carl Dekker MET-L-FLO Inc.

Vince Anewenter Milwaukee School of Engineering

E.R. Denlinger Autodesk Inc.

Samyeon Kim Singapore University of Technology and Design

Sundar V. Atre University of Louisville

Phill Dickens University of Nottingham

John Barnes The Barnes Group Advisors

Amy Elliott Oak Ridge National Laboratory

Saurabh Basu Pennsylvania State University

Ravi K. Enneti Global Tungsten and Powders Corp.

Joseph J. Beaman University of Texas at Austin

David Espalin The University of Texas at El Paso

Allison Beese Pennsylvania State University

David Fletcher Cooksongold

Lindsey B. Bezek Virginia Polytechnic Institute and State University

Diana Gamzina SLAC National Accelerator Laboratory

David L. Bourell University of Texas at Austin

Jerard V. Gordon Carnegie Mellon University

Carelyn E. Campbell National Institute of Standards and Technology

Robert J. Griffiths Virginia Tech

Prem Chahal Fraunhofer Center for Coatings and Diamond Technologies Kristin M. Charipar U.S. Naval Research Laboratory

Edward Kinzel University of Notre Dame M.M. Kirka Oak Ridge National Laboratory Howard Kuhn University of Pittsburgh David K. Leigh EOS North America Ming C. Leu Missouri University of Science and Technology Wenbin Li Missouri University of Science and Technology Xiangjia Li University of Southern California Guangyi Ma Dalian University of Technology

Gautam Gupta University of Louisville John Halloran University of Michigan

Eric MacDonald Youngstown State University John Martin Youngstown State University

Adam Hehr Fabrisonic LLC

Richard P. Martukanitz University of Virginia and Commonwealth Center for Advanced Manufacturing

Neil Hopkinson XAAR3D

Eric Maynard Jenike & Johanson

Timothy Horn North Carolina State University

Brian McTiernan Powdered Metals Consulting LLC

Wayne Hung Texas A&M University

Nicholas Meisel Pennsylvania State University

Harish Irrinki University of Louisville

Frank Cooper Birmingham City University School of Jewellery

P. Michaleris Autodesk Inc.

Jay Keist Pennsylvania State University

Amir Mostafaei Carnegie Mellon University

Jose Coronel The University of Texas at El Paso

Dominic Kelly The University of Texas at El Paso

Peeyush Nandwana Oak Ridge National Laboratory

Chase Cox MELD Manufacturing Corp.

Shawn Kelly Oerlikon AM

Abdalla R. Nassar Pennsylvania State University

Yong Chen University of Southern California Zhangwei Chen Shenzhen University Kenneth Church nScrypt, Inc. Brett P. Conner Youngstown State University

vii

Subrata D. Nath University of Louisville

Anthony D. Rollett Carnegie Mellon University

Andrew Triantaphyllou Manufacturing Technology Centre

Fangyong Niu Dalian University of Technology

David Rosen Georgia Institute of Technology

A. Cagri Ulusoy Michigan State University

Virginia Osterman Solar Atmospheres Inc.

Carolyn Carradero Santiago Youngstown State University

Ryan Wicker The University of Texas at El Paso

Todd Palmer Pennsylvania State University

Corey Shemelya The University of Massachusetts Lowell

Paolo Parenti Politecnico di Milano

James Shipley Quintus Technologies AB

Christopher B. Williams Virginia Polytechnic Institute and State University

Svenja Pestotnik Fraunhofer Center for Coatings and Diamond Technologies

S.L. Sing Nanyang Technological University

Christian M. Petrie Oak Ridge National Laboratory Brian Pettinger Jenike & Johanson Andrew Pinkerton Lancaster University Alberto Pique U.S. Naval Research Laboratory David A. Prawel Colorado State University Prahalada Rao University of Nebraska-Lincoln Edward W. Reutzel Pennsylvania State University Richard Ricker National Institute of Standards and Technology Aljoscha Roch Michigan State University

Don Smith Baxter Healthcare Corp. (Retired) Zackary Snow Pennsylvania State University Xuan Song University of Iowa Niyanth Sridharan Oak Ridge National Laboratory Thomas L. Starr University of Louisville Lukas Stepien Fraunhofer-Institut fur Werkstoff und Strahltechnik Mark R. Stoudt National Institute of Standards and Technology

Terry Wohlers Wohlers Associates Inc. Dongjiang Wu Dalian University of Technology Ziheng Wu Carnegie Mellon University Shuai Yan Dalian University of Technology Li Yang University of Louisville Srujana Rao Yarasi Carnegie Mellon University Evren Yasa Eskisehir Osmangazi University W.Y. Yeong Nanyang Technological University Hang Yu Virginia Tech

Joseph Strauss HJE Company, Inc.

Fan Zhang National Institute of Standards and Technology

Juan L. Trasorras Global Tungsten and Powders Corp.

Shanshan Zhang University of Louisville

viii

Officers and Trustees of ASM International (2019–2020) Zi-Kui Liu President Pennsylvania State University Diana Essock Vice President Metamark, Inc. David U. Furrer Immediate Past President Pratt & Whitney Ron Aderhold Acting Managing Director Prem K. Aurora Aurora Engineering Co.

Diana Lados Worcester Polytechnic Institute Toni Marechaux U.S. Department of Defense Thomas M. Moore Wavix Inc. Jason Sebastian Questek Innovations, LLC Larry Somrack NSL Analytical Services, Inc. Judith A. Todd Pennsylvania State University

Priti Wanjara National Research Council Canada Ji-Cheng Zhao University of Maryland Student Board Members Kimberly Gliebe Case Western Reserve University Ashwin Kumar Vanderbilt University Nisrit Pandey Carnegie Mellon University

Members of the ASM Handbook Committee (2019–2020) Craig J. Schroeder, Chair Element Scott M. Olig, Vice Chair U.S. Naval Research Lab Alan P. Druschitz, Immediate Past Chair Virginia Tech Sabit Ali AVIVA Metals Kevin R. Anderson Mercury Marine Scott Beckwith BTG Composites Inc. Lichun (Leigh) Chen Superior Essex Narendra B. Dahotre University of North Texas

John Harkness Retired Martin Jones Ford Motor Company Brad Lindner Element Materials Technology Dana Medlin EAG Laboratories, Inc. Roger Narayan UNC-NCSU Dept of Biomed Eng Valery Rudnev Inductoheat Incorporated Muthukumarasamy Sadayappan Natural Resources Canada Satyam Suraj Sahay John Deere Technology Center India

Jeffery S. Smith Material Processing Technology LLC John M. Tartaglia Element Materials Technology Wixom Inc. George E. Totten G.E. Totten & Associates LLC George Vander Voort, Immediate Past Chair Vander Voort Consulting L.L.C. Christopher Viney University of California –MERCED Junsheng Wang Beijing Institute of Technology Valerie L. Wiesner NASA Glenn Research Center Dehua Yang Ebatco

Chairs of the ASM Handbook Committee J.F. Harper (1923–1926) (Member 1923–1926) W.J. Merten (1927–1930) (Member 1923–1933) L.B. Case (1931–1933) (Member 1927–1933) C.H. Herty, Jr. (1934–1936) (Member 1930–1936) J.P. Gill (1937) (Member 1934–1937) R.L. Dowdell (1938–1939) (Member 1935–1939) G.V. Luerssen (1943–1947) (Member 1942–1947) J.B. Johnson (1948–1951) (Member 1944–1951) E.O. Dixon (1952–1954) (Member 1947–1955) N.E. Promisel (1955–1961) (Member 1954–1963) R.W.E. Leiter (1962–1963) (Member 1955–1958, 1960–1964) D.J. Wright (1964–1965) (Member 1959–1967)

J.D. Graham (1966–1968) (Member 1961–1970) W.A. Stadtler (1969–1972) (Member 1962–1972) G.J. Shubat (1973–1975) (Member 1966–1975) R. Ward (1976–1978) (Member 1972–1978) G.N. Maniar (1979–1980) (Member 1974–1980) M.G.H. Wells (1981) (Member 1976–1981) J.L. McCall (1982) (Member 1977–1982) L.J. Korb (1983) (Member 1978–1983) T.D. Cooper (1984–1986) (Member 1981–1986) D.D. Huffman (1986–1990) (Member 1982–1991) D.L. Olson (1990–1992) (Member 1982–1992) R.J. Austin (1992–1994) (Member 1984–1985)

ix

W.L. Mankins (1994–1997) (Member 1989–1998) M.M. Gauthier (1997–1998) (Member 1990–2000) C.V. Darragh (1999–2002) (Member 1989–2002) Henry E. Fairman (2002–2004) (Member 1993–2006) Jeffrey A. Hawk (2004–2006) (Member 1997–2008) Larry D. Hanke (2006–2008) (Member 1994–2012) Kent L. Johnson (2008–2010) (Member 1999–2014) Craig D. Clauser (2010–2012) (Member 2005–2016) Joseph W. Newkirk (2012–2014) (Member 2005–) George Vander Voort (2014–2016) (Member 1997–) Alan P. Druschitz (2016–2019) (Member 2009–) Craig Schroeder (2019–present) (Member 2016–)

Policy on Units of Measure

By a resolution of its Board of Trustees, ASM International has adopted the practice of publishing data in both metric and customary U.S. units of measure. In preparing this Handbook, the editors have attempted to present data in metric units based primarily on Syste`me International d’Unite´s (SI), with secondary mention of the corresponding values in customary U.S. units. The decision to use SI as the primary system of units was based on the aforementioned resolution of the Board of Trustees and the widespread use of metric units throughout the world. For the most part, numerical engineering data in the text and in tables are presented in SI-based units with the customary U.S. equivalents in parentheses (text) or adjoining columns (tables). For example, pressure, stress, and strength are shown both in SI units, which are pascals (Pa) with a suitable prefix, and in customary U.S. units, which are pounds per square inch (psi). To save space, large values of psi have been converted to kips per square inch (ksi), where 1 ksi = 1000 psi. The metric tonne (kg  103) has sometimes been shown in megagrams (Mg). Some strictly scientific data are presented in SI units only. To clarify some illustrations, only one set of units is presented on artwork. References in the accompanying text to data in the illustrations are presented in both SI-based and customary U.S. units. On graphs and charts, grids corresponding to SI-based units usually appear along the left and bottom edges. Where appropriate, corresponding customary U.S. units appear along the top and right edges. Data pertaining to a specification published by a specification-writing group may be given in only the units used in that specification or in dual units, depending on the nature of the data. For example, the typical yield strength of steel sheet made to a specification written in customary U.S.

units would be presented in dual units, but the sheet thickness specified in that specification might be presented only in inches. Data obtained according to standardized test methods for which the standard recommends a particular system of units are presented in the units of that system. Wherever feasible, equivalent units are also presented. Some statistical data may also be presented in only the original units used in the analysis. Conversions and rounding have been done in accordance with IEEE/ASTM SI-10, with attention given to the number of significant digits in the original data. For example, an annealing temperature of 1570  F contains three significant digits. In this case, the equivalent temperature would be given as 855  C; the exact conversion to 854.44  C would not be appropriate. For an invariant physical phenomenon that occurs at a precise temperature (such as the melting of pure silver), it would be appropriate to report the temperature as 961.93  C or 1763.5  F. In some instances (especially in tables and data compilations), temperature values in  C and  F are alternatives rather than conversions. The policy of units of measure in this Handbook contains several exceptions to strict conformance to IEEE/ASTM SI-10; in each instance, the exception has been made in an effort to improve the clarity of the Handbook. The most notable exception is the use of g/cm3 rather than kg/m3 as the unit of measure for density (mass per unit volume). SI practice requires that only one virgule (diagonal) appear in units formed by combination of several basic units. Therefore, all of the units preceding the virgule are in the numerator and all units following the virgule are in the denominator of the expression; no parentheses are required to prevent ambiguity.

x

Preface For almost 100 years the ASM Handbook has captured the ongoing growth of applied knowledge covering the complete array of industrial materials and manufacturing processes. Within this archive, a major theme is the progressive development of materials having ever greater strength and heat resistance, primarily to meet performance objectives of transportation and defense applications. Unfortunately, the characteristics of such materials present challenges to the principal shaping processes of forming and material removal in reaching the complex shapes required. This, in turn, led to a parallel trend toward net shape to reduce or eliminate the negative aspects of material removal.

boundaries of the part effectively reduces its density, which is of great value in any transportation related application. The same approach can be used to spatially modify the localized density, strength, and thermal properties of a part, enabling functionally gradient materials to accommodate different needs in different locations of a part or component. In an advanced form, AM enables spatial variation of properties by building the parts with different materials point-to-point, or by varying process parameters to accomplish different microstructures within the same part. This wide latitude in shape, structure, and compositional control has injected a spirit of excitement in the materials, design, and manufacturing communities. Materials science and engineering has a new field in which to apply the basic concepts of materials structure through advanced tools for material characterization. Likewise, designers now wander into a new world of possibilities opened by the seemingly limitless geometric flexibility of AM. Manufacturers can now consider a new array of development and production processes with potentially more efficient materials use, reduced time to market, and greater performance.

In more recent times, however, a completely different approach to shape making has evolved in which basic materials in liquid or particulate form, as well as filament or sheet form, are assembled point by point or layer by layer into the objective shape. Using the pinpoint accuracy of a laser, miniscule droplets from a printhead, or extrusion of material through a narrow nozzle, a variety of clever mechanisms have been devised to carry out these high-resolution building processes, now collectively known as additive manufacturing (AM), and frequently referred to as 3D printing. The current and rapidly expanding importance of AM merits its capture within this, ASM Handbook, Volume 24, Additive Manufacturing Processes. As a starting point, the first division presents an overview of the subject as well as deep insights into its historical development, authored by some of the key participants in that history as they trace the evolution of AM from its pre-computer roots to early commercialization of (largely) rapid prototyping machines, to modern serious tools for production of parts from all material classes.

This Volume of the ASM Handbook series seeks to promote the excitement of AM by providing the latest knowledge in materials, processes, and applications. Following the history and introductory division, the complete suite of materials and processes for polymers and ceramics are detailed in the next two divisions. The fourth division describes the metal AM processes, but begins with in-depth description of the production and characterization of metal powders; such information has an outsized effect on success or failure of metal AM processes. The fifth division describes AM processing of a wide variety of materials, illustrating differences in characteristics of metal alloys produced by AM processes in contrast to conventional processes. The final division covers direct-write processes, taking advantage of AM processes to combine materials and devices for multifunctional engineering applications. Additional volumes are planned covering design and applications for additive manufacturing.

A primary result of these newfound processes is the capability to produce shapes of greater complexity and with more refined geometric detail than can be obtained by conventional processes covered in previous ASM Handbook volumes. In fact, such capabilities enable designers and manufacturers to think beyond net shape and toward optimum shape – the placement of material only where it is needed to carry out the required transmission of stress, temperature, or electromagnetic fields. In addition, thermo-fluid management systems, such as heat exchangers and molding tools, can incorporate non-round and nonstraight internal channels for enhanced efficiency. A further advantage of AM’s geometric flexibility is the combination of multiple parts into one component, eliminating assembly operations as well as individual part tooling and inventory. One highly publicized example involves the integration of some 20 parts into one fuel injection nozzle for aircraft turbine engines. To illustrate the advanced industrial development of AM, this component has been in mass production for more than a year at this writing.

We wish to acknowledge the immense efforts by the article authors and division editors to bring this volume together. Considerable time is required to complete these assignments which, unfortunately, come at a time when the talents of the authors are in high demand within this rapidly expanding and dynamic industry as it evolves continuously to new levels of achievement. Howard Kuhn, FASM David L. Bourell, FASM William Frazier, FASM Mohsen Seifi

Another exciting opportunity afforded by AM is modification of a material’s properties. The introduction of engineered porosity (i.e., printing material around void spaces) and lattice structures within the

vi

g

Contents Material Jetting of Polymers Christopher B. Williams and Lindsey B. Bezek, Virginia Polytechnic Institute and State University . . . . . . . . . . . . . . Process Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Process Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Part Design and Processing Considerations . . . . . . . . . . . . . Applications. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Future of Polymer Material Jetting . . . . . . . . . . . . . . . . Modeling for Polymer Additive Manufacturing Processes Neil Hopkinson, XAAR3D David Rosen, Georgia Institute of Technology . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Material Extrusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Polymer Powder-Bed Sintering/Fusion . . . . . . . . . . . . . . . . Vat Photopolymerization . . . . . . . . . . . . . . . . . . . . . . . . . . Material Jetting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Additive Manufacturing Overview David Bourell, University of Texas at Austin . . . . . . . . . . . . . . . 1 Introduction to Additive Manufacturing David Bourell, University of Texas at Austin Terry Wohlers, Wohlers Associates Inc.. . . . . . . . . . . . . . . . . . . 3 Vat Photopolymerization . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Material Jetting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Powder Bed Fusion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Directed Energy Deposition . . . . . . . . . . . . . . . . . . . . . . . . . 7 Material Extrusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Binder Jetting. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 Sheet Lamination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 History of Additive Manufacturing David L. Bourell and Joseph J. Beaman, University of Texas at Austin Terry Wohlers, Wohlers Associates Inc. . . . . . . . . . 11 Additive Manufacturing Terminology . . . . . . . . . . . . . . . . . 11 Historical Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 Additive Manufacturing Prehistory (~1860–1965). . . . . . . . . 11 Additive Manufacturing Precursors (1968–1984) . . . . . . . . . 13 Modern Additive Manufacturing (~1981–Late 2000s) . . . . . . 14 Growth of Additive Manufacturing since 2010. . . . . . . . . . . 17 Standards Development . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Design and Manufacturing Implications of Additive Manufacturing David Rosen, Georgia Institute of Technology Samyeon Kim, Singapore University of Technology and Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 Characteristics of Additive Manufacturing Processes . . . . . . 19 Purposes of Additive Manufacturing . . . . . . . . . . . . . . . . . . 20 Design Implications of Additive Manufacturing . . . . . . . . . . 21 Manufacturing Implications of Additive Manufacturing. . . . . 26

69 69 69 70 73 75

Ceramic Additive Manufacturing Processes Ming Leu, Missouri University of Science and Technology . . . . 79 Vat-Photopolymerization-Based Ceramic Manufacturing Xiangjia Li and Yong Chen, University of Southern California . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 Vat Photopolymerization . . . . . . . . . . . . . . . . . . . . . . . . . . 82 Vat-Photopolymerization-Based Ceramic Fabrication . . . . . . 83 Postprocessing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86 Property Identification . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88 Summary and Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . 92 Material Extrusion Based Ceramic Additive Manufacturing Wenbin Li and Ming C. Leu, Missouri University of Science and Technology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98 Post-Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 Mechanical Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 Future Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 Innovation Opportunities . . . . . . . . . . . . . . . . . . . . . . . . . . 107 Material Jetting of Ceramics Brett P. Conner, Youngstown State University . . . . . . . . . . . . 112 Ink Properties and Delivery . . . . . . . . . . . . . . . . . . . . . . . . 112 Applications. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114 Binder Jetting of Ceramics Li Yang, University of Louisville . . . . . . . . . . . . . . . . . . . . . 118 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118 Processability Considerations for Binder Jetting Additive Manufacturing of Ceramics . . . . . . . . . . . . . . . . . . . . . . 118 Binder Jetting Additive Manufacturing Technology Potential. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128 Directed-Energy Deposition for Ceramic Additive Manufacturing Fangyong Niu, Shuai Yan, Guangyi Ma, and Dongjiang Wu, Dalian University of Technology. . . . . . . . . . . . . . . . . . . . 131 Directed-Energy Deposition Equipment for Ceramic Additive Manufacturing . . . . . . . . . . . . . . . . . . . . . . . . . 133 Directed-Energy Deposition Materials for Ceramic Additive Manufacturing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134

Polymer Additive Manufacturing Processes David L. Bourell, University of Texas at Austin . . . . . . . . . . . . 31 Vat Polymerization Don Smith, Baxter Healthcare Corp. (Retired) . . . . . . . . . . . . . Feedstocks for the Process . . . . . . . . . . . . . . . . . . . . . . . . . Safety Issues with Feedstock Handling . . . . . . . . . . . . . . . . Manufacturing Issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . Postprocessing/Finishing . . . . . . . . . . . . . . . . . . . . . . . . . . Part Properties and Common Defects . . . . . . . . . . . . . . . . . Special Topics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Material Extrusion Additive Manufacturing Systems David A. Prawel, Colorado State University . . . . . . . . . . . . . . Melt Extrusion 3D Printing . . . . . . . . . . . . . . . . . . . . . . . . Viscous Extrusion 3D Printing . . . . . . . . . . . . . . . . . . . . . . Powder Bed Fusion of Polymers David K. Leigh, EOS North America David Bourell, University of Texas at Austin . . . . . . . . . . . . . . Thermal Issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Safety Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . Manufacturing Issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . Postprocessing and Finishing . . . . . . . . . . . . . . . . . . . . . . . Common Defects and Part Properties . . . . . . . . . . . . . . . . . Case Studies in Polymer Powder Bed Fusion . . . . . . . . . . . .

58 58 59 61 63 64 66

33 34 35 36 37 38 38 40 40 50 52 53 54 54 55 55 56

xi

Directed-Energy Deposition Process . . . . . . . . . . . . . . . . . . Microstructure and Properties of Typical Directed-Energy Deposition Ceramics . . . . . . . . . . . . . . . . . . . . . . . . . . . Powder Bed Fusion S.L. Sing and W.Y. Yeong, Nanyang Technological University . . . Powder Bed Fusion for Ceramics . . . . . . . . . . . . . . . . . . . . Ceramics Processed by Powder Bed Fusion . . . . . . . . . . . . . Challenges and Potential in Powder Bed Fusion of Ceramics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Applications of Powder Bed Fusion Ceramics . . . . . . . . . . . Conclusion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Binder Jetting and Sintering in Additive Manufacturing Amy Elliott, Corson Cramer, Peeyush Nandwana, Oak Ridge National Laboratory . . . . . . . . . . . . . . . . . . . . . . . . . . . . History of Binder Jetting . . . . . . . . . . . . . . . . . . . . . . . . . . Overview of Binder Jetting Process . . . . . . . . . . . . . . . . . . Advantages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Current Applications in Metals. . . . . . . . . . . . . . . . . . . . . . Opportunities and Challenges . . . . . . . . . . . . . . . . . . . . . . . Materials Exploration . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ultrasonic Additive Manufacturing Niyanth Sridharan and Christian M. Petrie, Oak Ridge National Laboratory . . . . . . . . . . . . . . . . . . . . . . . . . . . . Process Fundamentals and Process Parameters . . . . . . . . . . . Metallurgical Aspects in Ultrasonic Additive Manufacturing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Part Qualification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sensor Embedding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Future Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Deformation Processes in Additive Manufacturing Robert J. Griffiths and Hang Yu, Virginia Tech Chase Cox, MELD Manufacturing Corporation . . . . . . . . . . . Benefit of Deformation in Additive Manufacturing . . . . . . . . Ultrasonic Additive Manufacturing . . . . . . . . . . . . . . . . . . . Cold Spray. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Friction Stir Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . Modeling Thermomechanical Effects on Additive Manufacturing E.R. Denlinger and P. Michaleris, Autodesk Inc. . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Formation of Residual Stress and Distortion . . . . . . . . . . . . Mathematical Formulations . . . . . . . . . . . . . . . . . . . . . . . . Directed-Energy Deposition . . . . . . . . . . . . . . . . . . . . . . . . Laser Powder-Bed Fusion . . . . . . . . . . . . . . . . . . . . . . . . . Defects in Metal Additive Manufacturing Processes M.C. Brennan, J.S. Keist, and T.A. Palmer, The Pennsylvania State University. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Types of Defects in Fusion-Based Processes . . . . . . . . . . . . Types of Defects in Solid-State/Sintering Processes . . . . . . . Defect-Mitigation Strategies. . . . . . . . . . . . . . . . . . . . . . . . Current Knowledge Gap . . . . . . . . . . . . . . . . . . . . . . . . . . In-line Process Monitoring of Powder-Bed Fusion and Directed-Energy Deposition Processes Thomas L. Starr, University of Louisville . . . . . . . . . . . . . . . Laser or E-Beam Powder-Bed Fusion . . . . . . . . . . . . . . . . . Directed-Energy Deposition . . . . . . . . . . . . . . . . . . . . . . . . Postprocessing of Additively Manufactured Metal Parts Wayne Hung, Texas A&M University . . . . . . . . . . . . . . . . . . Powder Removal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Powder Recycling and Conditioning . . . . . . . . . . . . . . . . . . Part Removal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Part Enhancement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Heat treating . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Deburring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hot Isostatic Pressing for Metal Additive Manufacturing Magnus Ahlfors, Quintus Technologies . . . . . . . . . . . . . . . . . History of Hot Isostatic Pressing . . . . . . . . . . . . . . . . . . . . Hot Isostatic Pressing Process . . . . . . . . . . . . . . . . . . . . . . Fundamentals of Densification . . . . . . . . . . . . . . . . . . . . . . Hot Isostatic Pressing Equipment . . . . . . . . . . . . . . . . . . . . The Hot Isostatic Pressing Cycle . . . . . . . . . . . . . . . . . . . . Hot Isostatic Pressing for Metal Additive Manufacturing. . . . Vacuum Heat Treating Additively Manufactured Parts Virginia Osterman, Solar Atmospheres Inc. . . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Vacuum Heat Treating Processes for Additively Manufactured Parts . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sintering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

135 136 152 152 153 154 159 160

Metal Additive Manufacturing Processes Allison Beese, Pennsylvania State University, and Howard Kuhn, University of Pittsburgh . . . . . . . . . . . . . . . . . . . . . . . . . . 165 Metal Powder Production and Powder Size and Shape Distributions Zackary Snow, The Pennsylvania State University . . . . . . . . . 167 General Methods of Metal Powder Production . . . . . . . . . . . 167 Particle Sizing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167 Particle Shape . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170 Powder Characterization for Metal Additive Manufacturing Ziheng Wu, Srujana Rao Yarasi, Amir Mostafaei, and Anthony D. Rollett, Carnegie Mellon University . . . . . . . . . . . . . . . 172 Powder Flowability in Metal Additive Manufacturing . . . . . . 172 Powder Porosity in Additive Manufacturing. . . . . . . . . . . . . 175 Safety in Handling of Metal Powders Eric Maynard and Brian H. Pittenger, Jenike & Johanson, Inc.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180 Safety Standards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180 Metal Powder Use and Handling . . . . . . . . . . . . . . . . . . . . 181 Metal Powder Hazards . . . . . . . . . . . . . . . . . . . . . . . . . . . 185 Dust Hazards Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 191 Hazard Prevention and Protection with Metal Powders . . . . . 192 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198 Energy Sources for Fusion Additive Manufacturing Processes Abdalla R. Nassar and Edward W. Reutzel, The Pennsylvania State University Applied Research Laboratory . . . . . . . . . . 200 Arcs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200 Laser. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202 Electron Beam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205 Utility of Sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206 Alternative Sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206 Laser Powder Bed Fusion Harish Irrinki, Subrata D. Nath, Arulselvan Arumugham Akilan, Gautam Gupta, and Sundar V. Atre, University of Louisville Magnus Ahlfors, Quintus Technologies . . . . . . . . . . . . . . . 209 Effects of Powder Attributes and Processing Conditions on Laser Powder Bed Fusion . . . . . . . . . . . . . . . . . . . . . . . 210 Microstructures. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213 Corrosion Properties of Laser Powder Bed Fusion Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217 Directed-Energy Deposition Processes Richard P. Martukanitz, University of Virginia and Commonwealth Center for Advanced Manufacturing . . . . . . 220 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220 Process Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220 Components of Directed-Energy Deposition Systems . . . . . . 220 Directed-Energy Deposition Process . . . . . . . . . . . . . . . . . . 225 Materials for Directed-Energy Deposition . . . . . . . . . . . . . . 228 Postprocessing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232 Properties of Metallic Materials Produced by Using Directed-Energy Deposition . . . . . . . . . . . . . . . . . 232 Unique Materials for Directed-Energy Deposition . . . . . . . . 234 Applications for Directed-Energy Deposition Processes. . . . . 234

xii

239 239 239 241 243 244 244 247 247 249 255 256 257 261 261 262 263 263 265 265 265 266 268 270 277 277 281 281 283 287 287 295 298 298 299 300 300 310 312 316 316 316 316 317 318 318 324 324 324 326

Laser Melting and Electron Beam Processing of Precious Metals. . . . . . . . . . . . . . . . . . . . . . . . . . . . . Establishment and Growth of Laser Melting of Precious Metals. . . . . . . . . . . . . . . . . . . . . . . . . . . . . Business and Market Incentive for Laser-Melted Precious Metal Jewelry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Binder Jet Printing of Precious Metals . . . . . . . . . . . . . . Material Extrusion of Precious Metals. . . . . . . . . . . . . . . Material Jetting of Precious Metals. . . . . . . . . . . . . . . . . Photopolymerization with Precious Metals. . . . . . . . . . . . Future of Additive Manufacturing in Precious Metals . . . .

Vacuum Heat Treatment of Metals . . . . . . . . . . . . . . . . . . . 326 Evaporation of Metals in Vacuum Furnaces. . . . . . . . . . . . . 328 Additive Manufacturing of Metals Howard Kuhn, University of Pittsburgh. . . . . . . . . . . . . . . . . 331 Additive Manufacturing of Titanium Alloys. . . . . . . . . . . . . . . . 333 Additive Manufacturing Processing for Titanium Alloys . . . . 333 Titanium Alloys for Additive Manufacturing . . . . . . . . . . . . 334 Additive Manufacturing of Nickel-Base Superalloys M.M. Kirka, Oak Ridge National Laboratory . . . . . . . . . . . . . 339 Nickel-Base Superalloys for Additive Manufacturing . . . . . . 339 Binder Jet Additive Manufacturing . . . . . . . . . . . . . . . . . . . 340 Nickel-Base Superalloy Fusion Additive Manufacturing . . . . 340 Powder Recyclability . . . . . . . . . . . . . . . . . . . . . . . . . . . . 343 Additive Manufacturing of Steels and Stainless Steels Carelyn E. Campbell, Mark R. Stoudt, and Fan Zhang, National Institute of Standards and Technology . . . . . . . . . 346 Classification of Steels . . . . . . . . . . . . . . . . . . . . . . . . . . . 346 Additive Manufacturing Processes for Steels . . . . . . . . . . . . 346 Powder Feedstock Characteristics . . . . . . . . . . . . . . . . . . . . 348 Solidification Microstructure and Microstructure Evolution during Build . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 348 Postbuild Thermal Processing . . . . . . . . . . . . . . . . . . . . . . 352 Material Properties of Additively Manufactured Steels . . . . . 353 Summary and Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . 359 Additive Manufacturing of Tool Steels Peeyush Nandwana, Oak Ridge National Laboratory . . . . . . . 366 Laser Powder Bed Fusion . . . . . . . . . . . . . . . . . . . . . . . . . 367 Electron Beam Powder Bed Fusion. . . . . . . . . . . . . . . . . . . 370 Directed Energy Deposition . . . . . . . . . . . . . . . . . . . . . . . . 370 Binder Jet Additive Manufacturing . . . . . . . . . . . . . . . . . . . 371 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 372 Additive Manufacturing of Cobalt Alloys Amir Mostafaei, Jerard V. Gordon, and Anthony D. Rollett, Carnegie Mellon University . . . . . . . . . . . . . . . . . . . . . . . 374 Additive Manufacturing of Cobalt Alloys . . . . . . . . . . . . . . 374 Process-Microstructure-Properties Correlation in Additively Manufactured Cobalt Alloys . . . . . . . . . . . . . . . . . . . . . . . 374 Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 378 Additive Manufacturing of Tungsten, Molybdenum, and Cemented Carbides Ravi K. Enneti and Juan L. Trasorras, Global Tungsten and Powders Corp. Heinrich Kestler, Plansee SE . . . . . . . . . . . . . . . . . . . . . . . . 380 Selective Laser Melting of Molybdenum and Tungsten . . . . . 380 Additive Manufacturing of Cemented Carbides . . . . . . . . . . 383 Additive Manufacturing of Copper and Copper Alloys Timothy J. Horn, North Carolina State University Diana Gamzina, SLAC National Accelerator Laboratory. . . . . 388 Copper and Copper Alloys for Additive Manufacturing . . . . 388 Additive Manufacturing Processing of Copper and Copper Alloys. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 395 Applications of Copper and Copper Alloys in Additive Manufacturing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 403 Copper Feedstock Considerations . . . . . . . . . . . . . . . . . . . . 405 Properties of Copper and Copper Alloys Produced by Additive Manufacturing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 411 Additive Manufacturing of Precious Metals Joseph Strauss, HJE Company Inc.. . . . . . . . . . . . . . . . . . . . 419 Indirect Additive Manufacturing in the Casting of Precious Metals. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 419 Direct Additive Manufacturing Methods of Precious Metals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 420

. . 422 . . 424 . . . . . .

. . . . . .

425 425 425 426 426 426

Direct Write Processes Eric MacDonald, Youngstown State University . . . . . . . . . . . 429 Microdispensing Processes Kenneth Church, nScrypt, Inc. . . . . . . . . . . . . . . . . . . . . . . . Microdispensing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Material Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pen Tips . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Time Pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Positive Displacement . . . . . . . . . . . . . . . . . . . . . . . . . . . . Auger . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Progressive Cavity Pump . . . . . . . . . . . . . . . . . . . . . . . . . . Valving . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Gap. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Microelectronic Packaging . . . . . . . . . . . . . . . . . . . . . . . . . Printing to Pads . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Printing Patterns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Printed Antennas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Printed Circuits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Textile Printing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Flexible Printing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Structural Printing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Microdispensing Systems. . . . . . . . . . . . . . . . . . . . . . . . . . Aerosol Jetting for Multifunctional Additive Manufacturing Svenja Pestotnik and Prem Chahal, Fraunhofer Center for Coatings and Diamond Technologies Lukas Stepien, Fraunhofer Institute for Material and Beam Technology A. Cagri Ulusoy and Aljoscha Roch, Michigan State University Carolyn Carradero Santiago and Eric MacDonald, Youngstown State University. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Aerosol Jet Printing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Literature Review of Applications of Aerosol Jetting . . . . . . Comparison with Ink Jetting . . . . . . . . . . . . . . . . . . . . . . . Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Laser-Induced Forward Transfer Processes in Additive Manufacturing Alberto Pique´ and Kristin M. Charipar, U.S. Naval Research Laboratory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Origins and Background . . . . . . . . . . . . . . . . . . . . . . . . . . Other Laser Transfer Techniques . . . . . . . . . . . . . . . . . . . . Fundamentals of Laser-Induced Forward Transfer . . . . . . . . Laser-Induced Forward Transfer of Solid Materials . . . . . . . Laser-Induced Forward Transfer of Liquid Materials . . . . . . Laser-Induced Forward Transfer with a Laser-Absorbing Layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Laser-Induced Forward Transfer of Intact Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Laser-Induced Forward Transfer for Microadditive Manufacture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Industrial Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

xiii

431 431 431 432 432 432 432 433 433 433 434 434 434 435 435 435 435 436 436

437 437 438 438 442

446 446 446 446 448 448 451 451 452 453 453

Ultrasonic and Thermal Metal Embedding for Polymer Additive Manufacturing Carolyn Carradero Santiago and Eric MacDonald, Youngstown State University Jose Coronel, Dominic Kelly, Ryan Wicker and David Espalin, The University of Texas at El Paso . . . . . . . . . . . . . . . . . . 456 Wire Embedding in Additive Manufacturing . . . . . . . . . . . . 457

Applications of Wire Embedding in Additive Manufacturing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 458 Conclusion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 460 Reference Information. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 463 Index. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 465

xiv

Additive Manufacturing Overview Division Editor: David L. Bourell, University of Texas at Austin

Introduction to Additive Manufacturing Vat Photopolymerization . . . . . . . . . . . Material Jetting . . . . . . . . . . . . . . . . . Powder Bed Fusion. . . . . . . . . . . . . . . Directed Energy Deposition . . . . . . . . . Material Extrusion . . . . . . . . . . . . . . . Binder Jetting. . . . . . . . . . . . . . . . . . . Sheet Lamination . . . . . . . . . . . . . . . .

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

1 5 6 6 7 8 9 9

History of Additive Manufacturing . . . . . . . . . . . . . . . . . . . . . 11 Additive Manufacturing Terminology . . . . . . . . . . . . . . . . . . . 11 Historical Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

Additive Manufacturing Prehistory (~1860–1965). . . . Additive Manufacturing Precursors (1968–1984) . . . . Modern Additive Manufacturing (~1981–Late 2000s) . Growth of Additive Manufacturing since 2010. . . . . . Standards Development . . . . . . . . . . . . . . . . . . . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

11 13 14 17 17

Design and Manufacturing Implications of Additive Manufacturing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Characteristics of Additive Manufacturing Processes . . . . . . . . Purposes of Additive Manufacturing . . . . . . . . . . . . . . . . . . . . Design Implications of Additive Manufacturing . . . . . . . . . . . . Manufacturing Implications of Additive Manufacturing. . . . . . .

19 19 20 21 26

Copyright # 2020 ASM InternationalW All rights reserved www.asminternational.org

ASM Handbook, Volume 24, Additive Manufacturing Processes D. Bourell, H. Kuhn, W. Frazier, M. Seifi, editors DOI 10.31399/asm.hb.v24.a0006555

Introduction to Additive Manufacturing David Bourell, University of Texas at Austin Terry Wohlers, Wohlers Associates Inc.

ADDITIVE MANUFACTURING (AM), popularly known as 3D printing, is a collection of manufacturing processes, each of which builds a part additively based on a digital solid model. The solid model-to-AM interface and material deposition are entirely computer controlled. According to the official terminology standard for AM, ISO-ASTM 52900 (Ref 1), AM is defined as the “process of joining materials to make parts from 3D model data, usually layer upon layer, as opposed to subtractive manufacturing and formative manufacturing methodologies.” The standard further categorizes current commercially viable AM processes into seven categories, based largely on the broad binding mechanism for part creation. They are vat photopolymerization (vpp), material jetting (MJT), powder bed fusion (PBF), directed energy deposition (DED), material extrusion (MEX), binder jetting (BJT), and sheet lamination (SHL). From a purely economic perspective, the traditional application space for AM has been for low-production runs of parts with complex shapes and geometric features. Examples include prototypes; tooling; jigs/fixtures; models for aerospace, automotive, biomedical, and other industrial sectors; mass customization; jewelry; and artwork. For low-cost AM systems such as some material extrusion and binder jetting equipment, part cost can be much lower than that of conventional manufacturing for small to medium quantities. At the other extreme, part production with stringent service requirements using PBF or DED can be expensive due to high feedstock cost and the large capital cost of the machine, supporting equipment, and postprocessing labor. Also, AM is advantageous economically where reduced part count and elimination of assembly requirements result in significant cost savings. This is shown in Fig. 1, where a 16-part assembly is replaced by a single AM part. Performance-based applications of AM are notable. Certain parts have sufficiently complex shapes and geometric features where no other manufacturing routes are available. A common example is internal flow fields in parts, including conformal cooling channels

(a)

Fig. 1

(b)

Example of a multiple-part assembly (a) reduced to a single part (b) through additive manufacturing. Courtesy of 3D Systems

in molds and dies, as well as gas/aerosol flow channels for optimally mixing and distributing fuel and oxidants. Figure 2 shows a conformal cooling channel implemented in an injection mold. The cooling time for injection-molded parts was reduced by 55%, enabling the manufacturer to improve molding cycle time by almost 25%. AM is also an enabling technology for topology optimization, an approach to design in which predictably loaded parts are analyzed using solid modeling/stress analysis software. This ensures that most or all elements of the part carry loads with an objective of reducing material to a minimum. Topology-optimized parts often have a biomimetic look that can be difficult or impossible to manufacture using conventional methods (Fig. 3). Another feature of AM is the ability to easily create designed lattice/truss and cellular structures in a part. Lightweighting applications for the biomedical industry involve hard-tissue scaffolding, for which engineered porosity facilitates adhesion to living material (Fig. 4). Some AM applications are driven by materials issues, particularly for metal part production. Because the active build volume for AM is relatively small, parts tend to be chemically homogeneous without macrosegregation, as in powder methods. Certain AM processes, such as DED and material

Fig. 2

Conformal cooling channel in an injection molding tool. Courtesy of Renishaw

extrusion, facilitate creation of functionally graded compositions (Fig. 5). Here, various compositions of material are deposited during the build in specific locations. Gradation of porosity is also effected by AM. For AM processes involving fusion (PBF, DED, and material jetting of thermoplastics and metals),

4 / Additive Manufacturing Overview

Fig. 3

Nacelle hinge bracket for Airbus A320 (a) and topology-optimized design produced by additive manufacturing (b). Courtesy of EADS and Altair (Ref 2,3)

Fig. 5

Fig. 4

Cellular bone scaffold for dental implantation. Courtesy of Nanyang Technological University

the energy input affects the heating and cooling rate locally, which can influence the microstructure. For example, it is possible to control the texture in a nickel-based alloy processed using electron beam PBF by altering the solidification kinetics via processing parameter adjustment (Fig. 6). Increasingly, AM is being employed in part repair. In preparing to fixture a part for computer numerical control (CNC) machining, DED is useful for building up the part, especially for expensive, high-end parts such as aerospace engine components. Beyond the economies and performance of parts, AM technology also impacts the process and supply chain. Often, prototype part production was delayed until late in the product

development cycle due to cost and time factors. With AM, it becomes feasible to manufacture much earlier in the process, and in many cases, create iterative functional prototypes in lieu of or in addition to form-and-feel prototypes. Having a prototype part early in the process facilitates final design. The notion here is exemplified by the adage, “If a picture is worth a thousand words, a physical part is worth a thousand pictures.” The potential for distributed manufacturing enabled by AM also positively impacts the supply chain because customers are given more vendors and sources for subassemblies and products. Another impact on the supply chain is the cost to transport parts. With a distributed manufacturing capability enabled by AM, it is

(a) Functionally graded additive manufacturing cross section (Ref 4) and (b) a multimaterial photopolymer part (Stratasys J750 printer)

possible to print the part at or near its final service location, resulting in savings in transportation costs. Corporate access to multiple certified manufacturing facilities and service providers positively affects the availability of parts and reduces risk associated with unforeseen changes in part provision from a single supplier. A societal impact of AM is the transition partially from centralized manufacturing to distributed manufacturing, especially for lowcost fabricators. This occurred with computing in the late 1970s and 1980s, when microprocessors spawned the commercialization of desktop computing. Additive manufacturing effectively enables the same concept for manufacturing, in which the final user becomes equipped to manufacture with a reduction in assistance from other companies. The impact of distributed manufacturing is not yet fully realized, but its strength is demonstrated anecdotally almost every day in stories spanning the breadth of society where AM makes a difference.

Introduction to Additive Manufacturing / 5 Additive manufacturing processes share several broad, common characteristics. Typically, the build rate of ~0.5 to 5 cm3/min is slow compared with conventional casting, molding, or forming. Exceptions are weld-head deposition approaches, large-scale material extrusion, and certain other AM approaches (e.g. binder jet) that rely on binding an entire layer simultaneously rather than using a point-to-point or raster mechanism such as a laser or nozzle, respectively. Surface finish is variable, with finishes on the order of 1 mm attainable for vat photopolymerization to tens to hundreds of micrometers for certain other AM processes. Most AM processes require support structures for all but the simplest shapes. Exceptions are sheet lamination, binder jetting, and polymeric PBF. Support structures are defined by ISO/ ASTM 52900 as a “structure separate from the part geometry that is created to provide a base and anchor for the part during the building process.” Removal of the support material and finishing the part, grouped with other activity and termed “postprocessing,” may

Fig. 6

constitute a significant portion of the time, effort, and cost of manufacture.

Vat Photopolymerization The surface of a photosensitive liquid thermoset polymer is exposed to a prescribed wavelength of “light,” which chemically initiates the cross-linking reaction. This results in the formation of a solid in the liquid where the material is exposed to the light. Wavelengths often lie in the ultraviolet range. A typical schematic is shown in Fig. 7. In this embodiment, the photopolymer resides in a vat, and a low-power (milliwatt) laser scans the surface of a platform, inducing cross-linking and creating a solid layer. A recoater mechanism delivers a thin layer of liquid photopolymer to the surface, and the process is repeated. Because the laser scanning is computer controlled, the layers vary in shape, resulting in the ultimate creation of a fully three-dimensional part (Fig. 8).

Crystallographic texture control in additive manufacturing (Ref 5)

The liquid polymer resin is typically an epoxy or acrylate mixed with a photoinitiator. Epoxies are more common because they are generally stronger and do not shrink as much as acrylates on cross-linking. Common photoinitiators are benzoin, acetophenone, benzyl ketal, and cyclohexyl phenyl ketone. When exposed to a specified wavelength, the photoinitiator reacts with the liquid monomer to form free radicals, thus initiating cross-linking. Termination of the polymeric long chains occurs by one of several mechanisms detailed in the article “Vat Photopolymerization” in this Volume. It is possible to mix additives, typically ceramic or metallic particulate, into the liquid photopolymer. In these cases, the refractive index of the particles should match that of the liquid photopolymer to prevent undue dispersion of the light. Several variants of the vat photopolymerization process have been developed and commercialized. One builds the part upside down from the bottom of the vat. The light is transmitted through a window at the bottom of the vat onto the platform. Historically, an obstacle to this approach has been adhesion of the cross-linked polymer to the window. This has been mitigated in one instance by using oxygen-sensitive photopolymers and oxygen-permeable windows. The oxygen inhibits the cross-linking reaction at the surface of the window, which creates a thin layer of inactive liquid on the window surface. The advantage is continuous printing. Another variation is the use of a digital micromirror array, such as digital light processing technology from Texas Instruments, to project the entire layer simultaneously instead of using laser beam scanning. This results in a substantial increase in the build speed. Parts produced by vat photopolymerization have excellent feature detail and surface finish

Laser

Fig. 7

Schematic of vat photopolymerization

Fig. 8

Stereolithography wind tunnel part designed for Lotus Formula 1. Courtesy of 3D Systems (Ref 6)

6 / Additive Manufacturing Overview compared with those produced by other AM approaches, on the order of 1 mm for supportstructure-free, non-bottom-facing surfaces. They may be postprocessed by blasting, sanding, polishing, and coating to produce transparency. Mechanical properties are consistent with the cured thermoset resin matrix, which is relatively low in strength and toughness and has low thermal deflection temperatures, compared with common thermoplastics. For these reasons, vat photopolymerization is very suitable for creation of quality prototypes and nonstructural parts. According to the Wohlers Report 2019 (Ref 7), approximately one-third of all feedstock currently used in AM is thermosetting photopolymers for vat photopolymerization.

a material-jetted polymer part. Limited work, primarily research, has explored the use of molten metal droplet AM. Early study of the physics of metal droplet formation in an AM context was performed in the early 1990s by Orme (Ref 8). Material jetting presents several unique considerations. For polymers, the viscosity of the photopolymer must be sufficiently low to enable formation of a fine droplet stream. Droplet size and momentum define the degree of splat on the surface of the part, and these parameters must be controlled. The kinetics of the cross-linking reaction must also be sufficiently rapid to enable efficient, rapid deposition.

Powder Bed Fusion Material Jetting One embodiment of material jetting uses a photopolymer akin to those used in vat photopolymerization. It employs an inkjetting approach to deposit the material in specific locations on a build platform or part that is bathed in the cross-linking electromagnetic radiation (Fig. 9). In broad terms, material jetting uses a localized material presence within indiscriminate cross-linking radiation, in contrast to vat photopolymerization, for which a localized energy source cross-links polymer within an indiscriminate material resin pool. It has been demonstrated that material jetting is possible using metallic feedstock. With multihead inkjets, multimaterial systems can be created, leveraging future optimization technology. A second embodiment involves melting a polymer and atomizing it to create a fine particle stream for jetting. Feedstocks include polypropylene (PP), high-density polyethylene, polystyrene (conventional and high impact), poly(methyl methacrylate), polycarbonate, acrylonitrile butadiene styrene (ABS), and environmentally degradable polymers. Figure 10 shows

Essential to PBF is the localized fusing of particulate in a bed using an energy source, typically a laser, electron beam, or light source (Fig. 11). Powder is spread over a build area in a thin layer on the order of 30–100 mm.

Fig. 10

Powder spreading is accomplished by a moving blade “recoater” or a counterrotating cylinder. The energy source selectively focuses energy on the surface, resulting in localized melting and fusion of the powder, both to adjacent particles in the layer and to the previous layer. Figure 12 shows a PBF part made of nylon. Currently, three common energy sources are lasers, electron beams, and indiscriminate electromagnetic energy. Lasers are typically ~100 W neodymium-doped yttrium aluminum garnet (Nd:YAG) lasers for polymer PBF. For metals, fiber lasers provide improved coupling and are 200–500 W. Electron beams use a 5–10 kW power source. Another approach such as multijet fusion from HP or rapid laser sintering deposits a fine, particulate coupling agent, such as fine carbon powder, by inkjetting after the spreading of each layer of powder. When exposed to light of proper wavelength and energy, the coupling agent is heated to an extent that adjacent powder particles melt and fuse. The process is faster than point-source PBF because

Material jetting part. Courtesy of Loughborough University

Laser

Fig. 9

Schematic of material jetting

Fig. 11

Schematic of powder bed fusion

Introduction to Additive Manufacturing / 7 coupling agent deposition and entire surface binding is more rapid than raster scanning of a point energy source. Polymeric parts are suitable for production using laser and light sources. Metal parts are produced using electron and laser beam approaches. Virtually any pulverized material can be used in PBF when a suitable transient or permanent binder is mixed with the primary feedstock. Postprocessing may be used in the case of metal and ceramic parts to either convert or burn off the binder, and can be followed by optional conventional sintering or infiltration to densify the part. When residual stresses are controlled, as is the case for laser PBF of polymers, the loose, unfused powder supports the part, eliminating the need for support structures. For metal PBF, the part cake does not provide sufficient rigidity to prevent distortion from the heat, so parts are attached to a build plate using support structures, sometimes referred to as anchors. Polymer feedstocks for PBF are generally semicrystalline thermoplastics, including polyamide 11 (PA11), PA12, PP, polyether ether ketone, and polyaryletherketone. For laserbased PBF, the feedstock is preheated to a temperature near but less than the melting point. Once the material is melted by the laser, the bed temperature should remain above the crystallization temperature to minimize residual stresses and distortion. Therefore, a large temperature difference between the melting

and crystallization temperatures is desirable. For PA, this temperature difference is ~20  C (~36  F). Cooling prior to part removal generally requires the same amount of time it takes to build the parts. For example, if building the parts takes 10 h, the parts in the powder bed should cool for 10 h. If the parts are removed too quickly, they will distort and possibly oxidize. Feedstock particle size is on the order of 50–80 mm. Metallic feedstock for PBF is typically weldable and castable and includes aluminum alloys, cobalt-chromium alloys, nickel alloys, gold, silver, stainless steel, tool steel, and titanium alloys. Typical power sources for fusion are lasers and electron beams. Powder particle size is in the range of 20 to 40 mm for laser PBF and 45 to 100 mm for electron-beam PBF.

the energy beam using separate nozzles/feeders. Since the part is fully exposed during the metal additive process, DED is well suited to creating large-volume parts as well as repairing or adding features to existing parts. For electron beam and welding heat sources, the coupling and thermal efficiency can be high. Representative parts are shown in Fig. 14.

Energy source

Powder spray

Directed Energy Deposition Directed energy deposition feeds material into an energy beam, typically a laser, electron beam, or plasma arc weld head (Fig. 13). Feedstock is typically in the form of metal powder or wire. Compared with other AM process categories, DED offers unique features. First, it is possible to deposit large amounts of material, particularly with wire-fed processes, up to 200 to 400 in3/min (508 to 1016 cm3/min). Directed energy deposition is suited for multiple-material deposition, because the various feedstocks can be fed into

Fig. 13

Fig. 14 Fig. 12

Powder bed fusion part, nylon. Courtesy of 3D Systems

Schematic of directed energy deposition

Representative directed energy deposition parts. (a) Pure copper septagon structure 175 mm in diameter and 200 mm tall with 1 mm wall thickness. (b) Repairing a titanium turbine compressor vane. Courtesy of Optomec, Inc.

8 / Additive Manufacturing Overview One issue related to DED processes is that nearly all parts require postmachining, and it can be significant, time consuming, and costly. Also, DED limits the geometric complexity of a part. It is extremely difficult to use DED to produce internal channels, cavities, or other features that are typically easy, even trivial, for PBF. Lasers are usually CO2, Nd:YAG, fiber, disk, or diode. Laser power varies between 400 and 4000 W, with spot sizes ranging widely between about 50 mm and 25 mm. The plasma arc approach uses either gas tungsten arc, gas metal arc, or plasma-transferred arc welding heads. Heat sources range in power from less than 1 kW to 60 kW or more. Working distance from the power source is large for laser and electron beam approaches, and usually on the order of 300 mm (11.8 in.). For welding approaches, the working distance is about 25 mm (0.98 in.). Atmosphere control is maintained using either an inert build chamber, common for laser-based processes; vacuum in the case of electron beam processes; and a localized shielding gas for arc welding approaches. For DED with powder feedstock, the powder size and shape are dictated by the feeder specifications. The size is in the range of 5 to 150 mm. The capture efficiency in the energy beam is typically 40 to 80%. Powder mass flow rates are typically 1 to 50 g/min. Wirefeed systems can deposit material on the order of 300 g/min. Hybrid AM generally refers to a combination of AM, particularly DED, and other processes, such as CNC milling, that alternates after 1 to 20 layers of the build. Most often, the secondary process is subtractive (i.e., material removal) and involves machining of some sort during the build for the purpose of creating a precision surface finish and tight tolerances, compared with AM only.

road-path pattern is typically varied between layers to improve structural integrity. Characteristic of the road-path pattern is inter-road-path porosity, which can be as high as 25%. This porosity may be controlled to some extent by adjusting the rate and amount of fill deposited. Material extrusion nozzle size is largely dictated by the desired surface characteristics as well as the force needed to extrude the feedstock. These considerations conflict, as the surface is improved by small diameter nozzles while the force required for extrusion increases with decreasing nozzle diameter. The amount of material deposited, and thus the speed of the process, increases as the size of the nozzle diameter increases. Typical nozzle diameters for polymer material extrusion are in the range of 0.2 to 1.0 mm. Material extrusion, due in part to the simplicity of the process, includes the lowest-cost AM machines, on the order of 200 to 1500 USD. As such, by far the most fabricators in service are from this group. According to the Wohlers Report 2019, sales of material extrusion machines were more than 590,000 units annually worldwide. Feedstock for material extrusion typically is from the class of amorphous thermoplastics, most popularly polylactic acid and ABS. Amorphous polymers are suitable for forming slurries with a rather continuous range of viscosity, as opposed to semicrystalline polymers, which have a sharp transition from solid to melt characteristics. This feature of amorphous polymers makes them generally more suitable for material extrusion. Low melt viscosity is desirable from the perspective of the force required to extrude the feedstock. This is particularly an issue when neat polymer is loaded with fillers or particulate. As in metal and

ceramic injection molding, the amount of added nonpolymeric material is limited to less than about 60% by volume for material flow considerations. Due largely to residual porosity, parts made using material extrusion are typically not used in structural applications. Ongoing developments are underway to improve structural integrity of parts, motivated by the relatively low cost of parts produced by material extrusion compared with some other AM methods. They include development of materials such as polyetherimide, control of porosity through process parameter optimization, and structural design using the porous structure, such as use of Weibull statistical methods. Concrete printing is an example of large-scale material extrusion and has been used for demonstration building construction as well as transportationrelated concrete structures, such as bridges. Large-scale material extrusion is a high-volume rate deposition technique usually based on a gantry system, as shown in Fig. 16.

Heat

Fig. 15

Schematic of material extrusion

Material Extrusion Material extrusion is accomplished by forcing feedstock through a nozzle that moves relative to a build platform. The typical approach involves melting a polymer or polymer matrix, but methods for extruding slurries are also available. The latter approach is most popular for creating large concrete structures and is popular for printing food such as icing, cookie dough, and blended vegetables. Optimal feedstocks are those that are shear thinning. Shear thinning results in feedstock low in viscosity during extrusion, but maintains a firm, high viscosity after placement, which minimizes distortion and sagging. Robocasting is an early example of shear-thinning material extrusion. A schematic diagram of material extrusion is shown in Fig. 15. The material is deposited crudely as a cylindrical shape with a circular or oval cross section, the axis of which as-deposited is termed the road path. The

Fig. 16

Schematic of large-area material extrusion (contour crafting). Courtesy of Behrokh Khoshnvis, University of Southern California

Introduction to Additive Manufacturing / 9 Material extrusion of parts of general shape requires the use of support structures. They are typically generated automatically by the system or other special software. Material extrusion is amenable to multimaterial part construction, enabled generally by multiple extrusion heads or by multiple feedstock feeds into a single extrusion head, which serves a mixing/blending function. A common multimaterial embodiment involves use of a different material from the part material to build the support structure. This facilitates support structure removal after the build is complete, for example, by using a dissolvable support material.

Binder Jetting Binder jetting is a powder bed process in which an adhesive is sprayed selectively onto the surface of the powder bed (Fig. 17). Historically, inkjet printing technology has been used with low-viscosity adhesive inks. Traditional binder jetting for prototypes and figures is amenable to the creation of color parts by employing standard color inkjet technology using multiple print heads. This facilitates the use of binder jetting for visual prototypes and nonstructural parts requiring aesthetic qualities. The quality of the part is dependent on the interaction mechanism of the adhesive with the powder bed. The goal is for the droplet to wet and bind powder particles rapidly and locally without splatting or otherwise dispersing on impact. The deposition rate, adhesive droplet size, droplet velocity, and binding mechanism all impact part quality. As long as the powder feedstock is wet by the adhesive, almost any pulverized or atomized material (e.g., carbides, oxides, and any metal) may be used. Immediately after the build, parts are “green” or loosely bound and are brought to full density via infiltration or postprocess sintering. For metallic and ceramic powder beds, postprocessing for structural part creation is similar to metal or ceramic injection molding. The advantage is that much higher metal/ceramic loading is possible compared with flowing a polymer/feedstock mixture into a mold. The surface quality of as-built binder jet parts is defined largely by the powder bed particle size and the inkjet droplet size. The density of as-built binder jet parts is largely defined by the powder bed density during the build, which is on the order of 65% or less, so the amount of infiltrant used or the amount of shrinkage needed to reach full density is largely dictated by this powder bed density. A unique property of binder jetting is that because the powder bed can completely support parts as they print, parts can be nested over and around each other in the build box. This ability to pack large batches of parts into a single build box makes for a very highproductivity printing scenario. Further, because postprocessing steps such as depowdering and

Fig. 17

Schematic of binder jetting

Cutter Sheet

Fig. 18

Schematic of sheet lamination

sintering/infiltration can also be completed in large batches, binder jetting is well suited for large-volume production. Recent developments in binder jetting have focused on densifying steels and nickel alloys through printing and sintering of fine powder feedstocks. Fine powders on the order of 10 mm have significant drive for sintering and can reach upwards of 99% density using solid state sintering. This combined with the high productivity and material flexibility of the binder jetting process cycle has spurred recent market growth around the technology for metal induction molding replacement manufacturing for the automotive industry.

Sheet Lamination Sheet lamination AM processes involve binding and shaping of sheet feedstock. The earliest sheet lamination processes used continuous roll paper, which was adhered to the previous layer and then cut to create the shape. Cutting sources include mechanical cutters such as knife blades, lasers, and milling machine tools. Most processes involve placement of the sheet, followed by cutting (see Fig. 18), although it is possible to cut the sheet before stacking. Each approach offers advantages. Stack-and-cut approaches are simpler to implement and do not require precise registration of each new layer relative to

those already processed. Specifically designed support structures are not needed for stack-andcut processes because the processed sheet material serves to support what is above it. Depending on the features of the part being produced, specially designed supports may be needed to complete a cut-and-stack part. For stack-and-cut technologies, the part at the end of the build is completely encased in a block of bound feedstock, which must be removed, usually by hand. This poses limits on the degree of fine detail obtained using this approach. In each layer during the cutting process, the cutter not only differentiates the part from the rest of the sheet but also cuts a grid array in the unused region of the sheet to facilitate part removal. Material waste is usually inherent to sheet lamination processes. Sheet feedstock not used in the part typically cannot be reused in the process and is discarded. A principal advantage of sheet lamination, like binder jetting, is that the build process takes place at room temperature. This facilitates use of dissimilar metals with large differences in melting point and insertion of polymeric devices and sensors into metal parts. Feedstock material for commercial sheet lamination is either paper or metal foil. For the former, sheets are bound with an adhesive. The primary feedstock constraints are creating the sheet morphology and ensuring that the adhesive is effective in binding layers. Paperbased sheet lamination parts may be colored

10 / Additive Manufacturing Overview during the build by inkjet printing part surfaces in each layer during the build. Such color parts are useful for visual models and prototypes. For the process, ultrasonic AM, metal foil is used, and sheets are bound by solid-state ultrasonic welding using a rolling sonotrode. Sheets are machined using a CNC milling process that is a part of the system. Common feedstocks include aluminum, copper, stainless steel, and titanium. REFERENCES 1. “Additive Manufacturing—General Principles— Terminology,” ISO/ASTM 52900:2015, ASTM, 2015 2. M. Tomlin and J. Meyer, “Topology Optimization of an Additive Layer Manufactured (ALM) Aerospace Part,” paper presented at the 7th Altair CAE Technology Conference, Altair Engineering, 2011 3. N. Gardan and A. Schneider, Topological Optimization of Internal Patterns and Support in Additive Manufacturing, Journal of Manufacturing Systems, Vol 37 (No. 1), Oct 2015, p 417–425

4. V.E. Beal, P. Erasenthiran, N. Hopkinson, P. Dickens, and C.H. Ahrens, Fabrication of X-Graded H13 and Cu Powder Mix Using High Power Pulsed Nd:YAG Laser, Solid Freeform Fabrication Symposium Proc., University of Texas at Austin, 2004, p 187–197 5. R.R. Dehoff, M.M. Kirka, W.J. Sames, H. Bilheux, A.S. Tremsin, L.E. Lowe, and S.S. Babu, Site Specific Control of Crystallographic Grain Orientation through Electron Beam Additive Manufacturing, Materials Science and Technology, Vol 31 (No. 8), 2015, p 931–938, DOI: 10.1179/17432847 14Y.0000000734 6. Lotus F1 Team and 3D Systems Move Together towards 3D Printed Race-Ready Mass Production of Parts, 3D Systems, https://www.3dsystems.com/learning-center/ case-studies/lotus-f1-team-and-3d-systemsmove-together-towards-race-ready-mass 7. “Wohlers Report 2019,” Wohlers Associates, Ft. Collins, CO, 2019 8. M.E. Orme and E.P. Muntz, Method for Droplet Stream Manufacturing, U.S. Patent 5,171,360, 1992

SUGGESTED REFERENCES  I. Gibson, D.W. Rosen, and B. Stucker,

 



 



Additive Manufacturing Technologies: Rapid Prototyping to Direct Digital Manufacturing, Springer, 2010 J.P. Kruth, Material Incress Manufacturing by Rapid Prototyping Techniques, CIRP Annals, Vol 40 (No. 2), 1991, p 603–614 J.J. Beaman, J.W. Barlow, D.L. Bourell, R.H. Crawford, H.L. Marcus, and K.P. McAlea, Solid Freeform Fabrication: A New Direction in Manufacturing, Kluwer, 1997 J.P. Kruth, M.C. Leu, and T. Nakagawa, Progress in Additive Manufacturing and Rapid Prototyping, CIRP Annals, Vol 47 (No. 2), 1998, p 525–540 “The Wohlers Report,” Wohlers Associates, Ft. Collins, CO, published annually D. Bourell, J.P. Kruth, M. Leu, G. Levy, D. Rosen, A. Beese, and A. Clare, Materials for Additive Manufacturing,” CIRP Annals, Vol 2, (No. 66), Aug 2017, p 657–680 Solid Freeform Fabrication Symposium Proc., University of Texas at Austin, 1990–present, http://sffsymposium.engr.utexas.edu/archive

ASM Handbook, Volume 24, Additive Manufacturing Processes D. Bourell, W. Frazier, H. Kuhn, M. Seifi, editors DOI 10.31399/asm.hb.v24.a0006548

Copyright # 2020 ASM InternationalW All rights reserved www.asminternational.org

History of Additive Manufacturing David L. Bourell and Joseph J. Beaman, University of Texas at Austin Terry Wohlers, Wohlers Associates Inc.

Additive Manufacturing Terminology The group of technologies today (2020) termed additive manufacturing (AM) began commercialization in 1988. The original term was rapid prototyping, indicative of the original use in accelerating the modeling and prototyping of new designs, mostly in the automotive industry. Also in 1988, J. Beaman at the University of Texas at Austin, appreciating that the potential application of the technology was broader than just prototyping, coined the term solid freeform fabrication. These two terms perpetuated until the early 2000s, when a large number of terms came into use, some with greater popularity than others. Examples from the period include additive fabrication, additive processes, additive techniques, layer manufacturing, freeform fabrication, rapid tooling, additive layer manufacturing, rapid manufacturing, direct digital manufacturing, additive manufacturing, and three-dimensional (3-D) printing. The term 3D printing was clearly applied to modern AM processes in the 1980s, according to multiple independent sources. (Information dissemination in the 1980s was much different from today. By way of example, the worldwide web was not released to the general public until 1991, and in the 1980s, typical background literature was obtained by manually searching local library shelves and by networking at conferences.) The earliest known use of the term 3-D printing as applied to AM was in an article by Wohlers in 1988 that reviewed the then-new stereolithography machine recently commercialized by Hull (Ref 1). The concept of printing an object was in use at least as early as 1984, when Hull filed his famous stereolithography patent, which said, “‘Stereolithography’ is a method and apparatus for making solid objects by successively ‘printing’ thin layers of a curable material. . .” (Ref 2). The term 3D printing was popularized by Emanuel Sachs from the Massachusetts Institute of Technology (MIT), and was used in 1989 in a patent on a binder jetting process that he co-invented (Ref 3). Sachs chose the

name 3D printing from a list of names that he and a student created, despite the fact that many of those interviewed for their opinions did not react positively to it. “But printing is 2D!” was a typical response. In addition to its descriptive appeal, Sachs’ affinity for the name stemmed from the fact that his father was a publisher and had taken him to tour commercial printing presses. The name caught on. By the early 2000s, media outlets around the world were using 3D printing as a common vernacular to refer to AM. By the end of the 2000s, the term was entrenched for multiple uses: reference to the MIT binder jetting process, to AM in general, and, specifically, to low-cost AM machines. This is still the case today. The term additive manufacturing was formally adopted on January 14, 2009, at the charter meeting of an ASTM International technical committee meeting in West Conshohocken, PA. Near the end of the meeting and leading to a motion to form a technical committee for standards development, there was a detailed discussion of what to call the technology. The term 3D printing was discussed but abandoned, due to its association with the MIT binder jetting process, among other reasons. A motion was made and adopted to form the ASTM International Committee on Additive Manufacturing Technologies, which resulted in the formation of ASTM Committee F42. Within two years, the International Organization for Standardization (ISO) formed a comparable committee, ISO/TC 261 on Additive Manufacturing, which further advanced adoption of the term. One of the most persistent aspects of AM is the standard triangle language (STL) file format used to transfer computer-aided design (CAD) model data to AM systems. This file format was developed by Dave Albert of the Albert Consulting Group for 3D Systems in or around 1987, preliminarily for the commercialization of the first modern AM machine, the 3D Systems SLA-1. While the limitations and shortcomings of the format have been widely discussed, and alternative formats have been proposed, STL remains the de facto standard for AM.

Historical Overview The history of AM may be split into three segments. The earliest, called AM prehistory, is characterized by additive part creation without the use of a computer. These approaches involved hand lay-up and date back to at least 1860. A second segment of AM process development occurred in the period from ~1968 to 1984. Called AM precursors, the AM machines here relied on the use of a dedicated computer. In this period, distributed computational capacity was growing and available. However, programming the computer was difficult and required a significant skill set. This skill set was known to the inventors of the processes in this period, but none of the inventions were successfully commercialized. This was in part due to the fact that an AM fabricator owner must have computer knowledge to use the machine, and few in society or technical fields had this knowledge or the interest in gaining it. Paramount was the requirement for a user/consumer to describe their AM part digitally by using design software such as CAD software. The landscape of distributed computing changed remarkably in 1984 with Apple Computer’s release of the Macintosh computer. With its graphical user interface facilitated by a mouse, learning how to access the capabilities of a computer moved from learning a cryptic text-based screen and commands to “point and click” and “plug and play.” Over a few years, societal knowledge of the use of and access to distributed computing exploded. This development created a foundation for commercialization of AM technologies, moving into the third segment of AM historical development, modern AM.

Additive Manufacturing Prehistory (~1860–1965) This period of AM development is characterized by AM without the use of a computer. The oldest of the three main areas of AM development is photosculpture. Figure 1 shows the photosculpture studio of Franc¸ois Wille`me

12 / Additive Manufacturing Overview in Paris (Ref 4). A subject was positioned on a platform. Twenty-four cameras, each behind a small window, simultaneously took a photo of the subject from all sides. Two such small windows are shown in Fig. 1(a). Wille`me then used a pantograph apparatus to create 24 sectors of the sculpture by tracing the silhouettes of the photos. The pantograph linkage mechanisms shown in Fig. 1(b) are part of this technology. The assembled sculpture was then hand-finished by artisans. A photographic technique was invented by Baese (Ref 6) in which multiple cameras situated around a subject simultaneously took images. Photo plates consisted of a photosensitive gel that expanded upon exposure when treated with water, reproducing the surface contour of the subject. Monteath (Ref 7) used silhouettes and photoexpanding gelatin plates to create bas-reliefs of subjects. Morioka took multiple photographs of a subject illuminated by a banded light pattern of parallel light and dark lines (Ref 8). The approach allowed layered sections to be fabricated, which then were stacked to create an actual object representing the subject. Munz invented a vat polymerization process in the mid-1950s (Ref 9). The schematic of the device is shown in Fig. 2. The surface of a photosensitive resin based on silver halide or bichromated gelatin was exposed to light, which resulted in forming a solid, with the help of stopping/fixing agents. A piston was lowered to allow infill of more photosensitive resin, followed by re-exposure by the light source. The second area of AM prehistory involves sheet lamination approaches using a hand cutand-stack method. Blanther patented a method for creating 3-D paper contour maps from aerial topographical maps (Fig. 3, Ref 10). A thin wax sheet was placed over the map, and lines of constant elevation were cut into the sheet. The pieces were separated and stacked before repeating the process with different elevations. The end product was a set of molds, which, after smoothing and backing, was used as a die set for pressing the paper sheets into 3-D contour maps. The general concept was advanced by Perera, who used cardboard sheets (Ref 11), and Concordet (Ref 12), Zang (Ref 13), and Gaskin (Ref 14), who used glass or clear plastic sheets. The third area of AM prehistory involved the use of weld deposition. The earliest known example of this is a patent by Baker that was issued in 1925 (Ref 15). Referred to as a class of ornamental welding, a weld head was used as a deposition tool and moved by hand to create a part, in similar fashion to modern-day polymer “doodlers.” Figure 4 shows several illustrations from the Baker patent. A series of patents in the 1960s to 1980s dealt with weld deposition on a rotating mandrel. These included Garver (Ref 16), White (Ref 17), Ujiie (Ref 18), Brandi and Luckow (Ref 19), Gale and Fair (Ref 20), and Brown, Breinan, and Kear (Ref 21). All include the buildup of metal around a rotating cylinder, as shown in Fig. 5.

Fig. 1

Fig. 2

(a) Wille`me’s photo apparatus (Admiral Farragut, seated). (b) View of Wille`me’s studio in Paris, ~1865. Source: Ref 5

Munz’ photosensitive resin printer. The build platform (5) is lowered during the build. The light source (4) travels through optics (16) and into the vat of feedstock A, exposing the resin at level D. The mountain image at the top of the sketch is a radar-based recording device that sends an image to the light source (4). The part is shown by dotted lines in the build chamber and is built on the platform (5). Source: Ref 9

History of Additive Manufacturing / 13

Additive Manufacturing Precursors (1968–1984) Modern computers may date to the work of Turing in the mid-1930s. Early portable, that is, distributed, computers appeared in the mid-1970s. It is against this framework that precursor AM processes appeared. The inventors knew how to operate a computer and how to use it to drive a process. However, the operation of a computer was largely unknown to the general public. Therefore, it was not feasible for a computer-based machine to enter the market, because there essentially was no customer base. Common to the precursor AM processes were:    

A computer was used to drive the process. None were successfully commercialized. Most were forgotten. A number were rediscovered and successfully commercialized 10 to 20 years later.

Fig. 3

Perhaps the earliest known use of a computer was in a laser crossed-beam approach proposed by Swainson in 1968 (Ref 22). Figure 6 shows a nontactile laser arrangement used to capture the geometric image. It is processed by a computer (item 66), and information is sent to a process chamber, where crossed lasers reproduce the object by solidifying material from liquid at the point of intersection of the two process lasers. A number of mechanisms were proposed for solidification, including use of photosensitive gelatins and “photoreversible photochromic material.” The use of a computer to control weld metal deposition was described by Ciraud in 1972 in a German patent application (Ref 23). As shown in Fig. 7, powder (item 9) is fed through a dispenser (item 8), where it is deposited into one or more energy beams (items 7 and 7a). The part (item 1) is formed in a layerwise fashion. In 1984, Masters filed a patent that described one or more computer-controlled

Blanther’s image showing a cut-and-stack approach for creation of a die set from a topographical map, 1892. Source: Ref 10

Fig. 4

Fig. 5

“energy beams” that were used to create a freeformed object based on a computer solid model (Ref 24). Bronowski in 1985 patented a moveable weld head freeform approach (Ref 25). As shown in Fig. 8, a computer solid image of a part is created (item 11). It is processed through a computer (item 3), which sends build information to both a robotic weld head (item 11) as well as weld dams or “shoes” (items 14 and 15), which control the molten deposition at the part edges. Matsubara (Ref 26) proposed a technique whereby photopolymer resin sheets containing refractory particles were selectively exposed to light. The unhardened portions were dissolved away, leaving a part that eventually could serve as a casting mold. Several inventors machined sheets of metal using a milling cutter followed by stacking and fixing the sheets. These included DiMatteo (Ref 27), Nakagawa et al. (Ref 28), and Kunieda and Nakagawa (Ref 29). In 1979, Housholder filed a U.S. patent on what today (2019) would be described as powder-bed fusion. His intent was to create sand casting molds (Ref 30). One embodiment used a scanning laser beam and a powder bed, complete with laser scanners and recoating blade (Fig. 9). A computer was used to control the process. An early sand part produced by Housholder is shown in Fig. 10.

Illustration of weld metal buildup around a rotating mandrel. Source: Ref 19

Fig. 6

Several objects made by using weld deposition. Source: Ref 15

First use of a modern computer to form a freeformed object. Source: Ref 22

14 / Additive Manufacturing Overview

Fig. 7

Ciraud’s powder-based metal buildup, a precursor to directed-energy deposition. Source: Ref 23

In 1981, Kodama published a paper on the use of “modern” photosensitive polymers to create freeformed parts (Ref 33). This is an early reference to vat polymerization. He proposed three process arrangements (Fig. 11). All employ an elevator mechanism in a vat, upon which the part is constructed. The use of a mask with a top-down energy delivery to the vat is similar to modern digital micromirror array approaches (Fig. 11a). The mask bottom-up approach (Fig. 11b) is similar to a current approach in which an oxygenpermeable membrane prevents cross linking of liquid photopolymer on the bottom window. Most modern vat polymerization processes use the general arrangement shown in Fig. 11(c), in which a point energy source provides radiation for cross linking. An early Kodama part is shown in Fig. 12. In 1982, Herbert developed a benchtop apparatus for vat polymerization (Ref 34). His setup, shown in Fig. 13, is based on a Hewlett Packard 9815A computer, introduced in 1975 and with a minicartridge storage capability of 94 kB. A gantry plotter was used to move a laser beam over the surface of a small vat of photopolymer. The technology was never commercialized or patented. An early part produced using this apparatus is shown in Fig. 14. Material jetting was proposed by Masters (Ref 24). Metals, ceramics, and thermoplastics were covered. A computer-controlled interface translated the solid model into build information sent to one to three material-dispensing nozzles. Termed ballistic particle manufacturing, the process was eventually commercialized but later failed (Ref 35).

Modern Additive Manufacturing (~1981–Late 2000s) A collection by Wohlers and Gornet detailing the historical development of modern AM is

Fig. 8

Schematic of computer-controlled weld deposition process. Source: Ref 25

found in the Wohlers Report (Ref 35). Much of what follows traces to this source. Critical to the successful commercialization of AM was distributed computing. Each AM machine needs a dedicated computation capability to form the part digitally, transfer the 3D computer model to a machine, and control the build. The user must be comfortable using the computer to effect these activities. Distributed computing was well established by the early 1980s. In fact, IBM, for example, was quite popular, with $4 billion in personal computer sales in 1984. That same year, Apple introduced the Macintosh, the first mass-marketed personal computer with a graphical user interface and mouse. This enabled “point and click” technology and “plug and play” simplicity. IBM followed the next year with its Windows system, replacing the unintuitive DOS operating system. Over a few short years, the personal computer became ubiquitous in society. This familiarity enabled the commercialization of computer-based manufacturing equipment, and, over the course of 10 to 15 years, an explosion of AM platforms were invented, developed, and commercialized. What follows is the development of modern AM to the late 2000s. The meteoritic growth of the field from the late 2000s is too extensive to document here. The first modern AM company formed was Helisys, founded by Feygin in 1985

(Ref 31). His process was a stack-and-cut sheet lamination process termed laminated object manufacturing (Ref 36). The first machine shipment was not until 1991, and the company closed in 2000. The Denken venture in Japan was established in 1985, but its first stereolithography machine, the SLP-3000, was not introduced before 1993. Hull and Freed formed 3D Systems in 1986. The earliest commercial modern AM fabricator, the SLA-1, was introduced to a small handful of customers in 1987 as a beta test system. The first commercial sale of a production SLA system occurred in 1988. Hull enlisted the services of Dave Albert of the Albert Consulting Group to develop the now common STL file format for handling CAD solid models. The STL files could be sliced digitally, resulting in a stack of horizontal cross sections for fabrication. The following details the early developments in modern AM, arranged by current ISO/ASTM 52900 classification (Ref 37) and in order of introduction.

Vat Polymerization After 3D Systems commercialized stereolithography in the United States in 1987, Japan’s NTT Data CMET and Sony/D-MEC commercialized versions of stereolithography in 1988 and 1989, respectively. NTT Data CMET (now

History of Additive Manufacturing / 15

Fig. 9

Fig. 10

Early part produced by Housholder (powder-bed fusion), ~1979. Source: Ref 31, 32

Fig. 11

Kodama’s 1981arrangements for vat polymerization. (a) Mask top-down arrangement. (b) Mask bottom-up arrangement. (c) Gantry arrangement with a moving energy beam. Reprinted from Ref 33 with the permission of AIP Publishing

Fig. 13

Herbert’s vat polymerization setup, 1982. Source: Ref 34

Housholder powder-bed fusion. Source: Ref 30

Fig. 12

Early part produced by Kodama (vat polymerization), ~1981. Source: Ref 31–33

a part of Teijin Seiki, a subsidiary of Nabtesco) called its system solid object ultraviolet plotter, while Sony/D-MEC (now D-MEC) called its product the solid creation system. Sony stopped manufacturing stereolithography (SL) systems for D-MEC in 2007. In 1988, Asahi Denka Kogyo introduced the first epoxy resin for the CMET stereolithography machine. The following year, Japan Synthetic Rubber (now JSR

Corp.) and DSM Desotech began to offer resins for the Sony/D-MEC machines. In 1990, Electro Optical Systems (EOS) of Germany sold its first Stereos stereolithography system. The same year, Quadrax introduced the Mark 1000 SL system, which used visible-light resin. The following year, Imperial Chemical Industries introduced a visible-light resin product for use with the Mark 1000; the company stopped selling its resin approximately one year later when Quadrax dissolved due to a legal conflict with 3D Systems. The Soliform stereolithography system from Teijin Seiki became available in 1992. The Soliform technology was originally developed

by DuPont under the Somos name and was subsequently licensed to Teijin Seiki for exclusive distribution rights in parts of East Asia. Also in 1992, Allied Signal introduced vinylether Exactomer resin products for SL. In 1993, 3D Systems and Ciba commercialized their first epoxy resin product. At approximately the same time, the QuickCast build style was introduced. QuickCast is a method of producing investment casting patterns that are mostly hollow, making it possible to burn them out without fracturing the ceramic shell. While technically not a vat polymerization process, the Israeli company Cubital Ltd.

16 / Additive Manufacturing Overview developed and commercialized solid ground curing in 1986, with the first shipments in 1991. Photopolymer was spread using a dispenser and was cured using a mask/ultraviolet lamp arrangement. The mask was generated using electrostatic toner on a glass plate. A second material, composed of wax, was dispensed and used to fill voids, serving as a support material. It was later removed by heating. In late 2001, Envisiontec of Germany showed its Perfactory machine. The technology used acrylate photopolymer and digital light processing technology from Texas Instruments to harden an entire layer at once.

Material Extrusion S. and L. Crump founded Stratasys in 1988, based on an invention filed by S. Crump in 1989 (Ref 38). The first shipment of a fused-deposition modeling (material extrusion) machine was in 1991. In 2000, Stratasys introduced Prodigy, a machine that produces parts in acrylonitrilebutadiene-styrene plastic using the company’s fused-deposition modeling technology. In 2012, Stratasys and Objet (Israel) merged.

Powder-Bed Fusion

Feygin founded Helisys in 1985, and laminated object manufacturing (LOM), a stack-and-cut process, was commercialized in 1991 (Ref 36). In 1994, Kira Corp. commercialized Japan’s first nonstereolithography system. Called Solid Center, it used a standard laser printer engine, toner, and an x-y plotter and knife to produce woodlike models by paper lamination. Kira referred to Solid Center as the first plain-paper 3D printer. In 1996, Schroff Development initiated sales of its semiautomated paper lamination system for under $10,000. Sales ceased in 2002. Helisys stopped producing LOM machines in 2000, after selling more than 375 systems worldwide over a period of 9 years. Solidica (now Fabrisonic) sold and installed a new version of its ultrasonic foil consolidation system, called Formation, in 2004. In 2007, MCOR Technologies of Ireland introduced its Matrix system, which uses a mechanical cutting blade in a sheet paper stack-and-cut method. By integrating with a color inkjet printing technology, it was possible to create low-cost color parts.

Deckard invented selective laser sintering in 1986 (Ref 39). With Beaman, the company DTM Corporation was founded in 1987, with the first selective laser sintering machine shipped in 1992. Figure 15 shows the first part produced using powder-bed fusion. The German company EOS commercialized the EOSINT machine based on laser sintering technology in 1994. A distinction between the Texas patent and the EOS patent was the powder-spreading mechanism. The former used a counterrotating cylinder, while the latter employed a recoating spreader bar. The earliest known materials patent for modern AM was for multiple powder feedstock for powder-bed fusion, filed by Bourell et al. in 1989 (Ref 40). A year later, Bourell produced the first direct-metal part using a modern AM machine (Fig. 16). The part was a copper/ solder elemental blend built on a low-power precommercial laser sintering machine (Ref 41). Arcam, a Swedish company, was founded in 1997. Its principal product was an electron-beam-based powder-bed fusion machine based on development in collaboration with Chalmers University of Technology in Gothenburg (Ref 42). Fockele & Schwarze of Germany introduced its steel-powder-based selective laser melting system in 1999,

Fig. 14

Fig. 15

Sheet Lamination

Early part produced by Herbert (vat polymerization), ~1982. Source: Ref 31, 32, 34

developed in cooperation with the Fraunhofer Institute for Laser Technology. This was the harbinger of a series of high-power directmetal machines for laser-based powder-bed fusion. In 2001, 3D Systems acquired DTM Corp., the first commercial powder-bed fusion company. EOS introduced its EOSINT M 270 direct-metal laser sintering machine in 2003. This system was one of the first to use a high-power fiber laser. The same year, Trumpf introduced its TrumaForm LF and TrumaForm DMD 505 machines. The LF machine uses a 250 W laser and fiber optic cable to direct light onto a metallic powder bed. The DMD 505 included a five-axis motion system, making it one of the earliest hybrid AM machines. As early as 2004, Matsuura in Japan had a demonstration hybrid AM machine based on powderbed fusion and subtractive manufacturing. This formed the basis for their LUMEX line of AM systems. Concept Laser of Germany introduced the M1 cusing laser melting machine in 2004. MCP Tooling Technologies (which became MTT Technologies Group) introduced the SLM ReaLizer 100 selective laser melting machine in 2005. MTT transitioned into the company Realizer, which is now a part of DMG Mori.

Binder Jetting Soligen, the first company based on the MIT binder jetting process (Ref 3), was founded in 1991 and shipped its first parts in 1993. Their direct shell production casting used an inkjet mechanism to deposit liquid binder onto ceramic powder to form shells for use in the investment casting process. Soligen ceased operations in 2006. In 1996, Z Corp. launched its Z402 3D printer, primarily for concept modeling. Based on MIT’s inkjet printing technology, the Z402 produced models using starch- and plaster-based powder

Deckard’s first powder-bed fusion polymer part (acrylonitrile-butadiene-styrene), 1986. Side dimension is ~50 mm (2 in.).

History of Additive Manufacturing / 17 2100 from BPM Technology was sold commercially in 1996. The ballistic particle manufacturing process deposited wax materials by using an inkjet print head. The company ceased operations in 1997. In 2000, Objet Geometries of Israel announced Quadra, a 3D inkjet printer that deposited and hardened photopolymer by using 1536 nozzles and an ultraviolet light source. In late 2003, 3D Systems began to sell and ship its InVision 3D printer, a machine that jets and hardens photopolymer.

Growth of Additive Manufacturing since 2010 Fig. 16

First fully metallic part made using a modern additive manufacturing fabricator. Source: Ref 41

materials and a water-based liquid binder. In 2000, Z Corp. introduced its Z402C machine, the world’s first commercially available multicolor 3D printer. In 1999, Extrude Hone’s AM business (now ExOne) installed its first ProMetal RTS-300 machine for building metal parts at Motorola. This indirect process uses an inkjet binder and metallic powder bed to create a part that must be postprocessed to remove the binder and sinter the part. Specific Surface and Therics were also early licensees of the MIT binder jetting process. In 2016, the Therics drug-printing process resurfaced when Aprecia commercialized the first 3D-printed pills to receive Food and Drug Administration clearance.

Directed-Energy Deposition Respecting directed-energy deposition, an early patent by Rabinovich (Ref 43) described a wire-fed single laser beam system for processing metals and ceramics. The laser-engineered net shaping technology was developed by Keicher and others at Sandia National Laboratories in the mid-1990s (Ref 44). The technology was spun off with the launch of Optomec, which shipped its first machine in 1998. AeroMet was founded in 1997 as a subsidiary of MTS Systems Corp. The company developed a process called laser AM, which used a high-power laser and powdered titanium alloys. Until it shut down in December 2005, AeroMet manufactured parts for the aerospace industry as a service provider.

Material Jetting In 1994, ModelMaker from Solidscape (then called Sanders Prototype) became available. ModelMaker deposited wax materials by using an inkjet print head. 3D Systems sold its first material jetting 3D printer (Actua2100) in 1996, using a technology that deposits wax material layer by layer with an inkjet printing mechanism. Personal Modeler

Over the past decade, 3D printing has increasingly become a household name. A number of factors contributed to this transition. According to Ref 45, the tremendous growth of AM since 2010 resulted from several interrelated events:  The development of low-cost units, led by

 

  

Fab@Home, an open-source 3D printer developed at Cornell University. The technology initially was adopted by hobbyists and do-it-yourselfers, then by the general public. The expiration of patents, which spurred competition in the marketplace and a broadening of the market, heralding the Makerbot era The development of free or inexpensive open-source hardware and software. Unexperienced designers were able to begin making virtual 3D parts using such programs as Google Draft. The availability of purchasable 3D parts or downloadable 3D models at 3D clip art sites such as 3D Warehouse and shape ways The ability to take and convert 2D photos into 3D printable files through iPhone apps such as Autodesk’s Catch 123D The formation of the National Additive Manufacturing Innovation Institute by the U.S. government in 2012 (now America Makes)

Between 2010 and 2012, AM transitioned from a relatively obscure collection of processes to a serious tool for a wide range of production and tooling applications. Adoption of the technology for series manufacturing across many industries is still in its early days. Even so, companies in aerospace and medicine have made investments of tens, and even hundreds, of millions of dollars into the development and use of AM for manufacturing. Major corporations that produce footwear, eyewear, sporting goods, and niche consumer products are in the process of rolling out programs in which AM is used to manufacture products. Some are taking advantage of AM to produce custom or semicustom products in a way that is not affordable or practical when using conventional methods of design and manufacturing. This has led to new markets and entirely new businesses and business models. Many of the largest producers of chemicals and

materials have also entered the AM market, some in substantial ways.

Standards Development In May 2008, the Society of Manufacturing Engineers’ Rapid Technologies and Additive Manufacturing Steering Committee made standards a priority. The group charged Professor Brent Stucker, then at the University of Louisville, with exploring possible standards development organizations with which to partner. He identified ASTM International and made initial contact with the organization in July 2008. A planning meeting of a small group of AM individuals was held on November 3, 2008, at ASTM International headquarters, at which meeting it was decided to convene a larger group for the purpose of establishing an ASTM technical committee on AM. This organizational meeting was held at ASTM International on January 13 and 14, 2009. More than 59 people attended. At this meeting, the group consensus was to name both the technical committee and the technology “additive manufacturing.” Within the ISO community, Technical Committee TC 261 on Additive Manufacturing was formed in 2011. The same year, a Partner Standards Developing Organization agreement was signed between ASTM International and ISO, providing a means for joint development of standards. The agreement created a mechanism by which a single standard may be jointly developed, cobranded, and marketed by both ASTM International and ISO. The earliest joint standard was ISO/ASTM 52900 on terminology for AM, issued in 2015 (Ref 37) and replacing the older ASTM F2792 and ISO/ CD 17296-1 terminology standards. In March 2016, America Makes, also known as the National Additive Manufacturing Innovation Institute, and the American National Standards Institute launched the Additive Manufacturing Standardization Collaborative (AMSC). The AMSC was established to provide a common space for the coordination of standards development among any interested standards development organization. The organization’s “Standardization Roadmap for Additive Manufacturing” (Ref 46) identified existing and developing standards, assessed gaps, and made recommendations for priority areas in which a perceived need was found for additional standardization. Initially, ten standards development organizations were participants. REFERENCES 1. T. Wohlers, 3D Printing from CADD Model to Prototype, Comput. Graph. World, May 1988, p 52–54 2. C.W. Hull, Apparatus for Production of Three-Dimensional Objects by Stereolithography, U.S. Patent 4,575,330, filed 1984, issued 1986

18 / Additive Manufacturing Overview 3. E.M. Sachs et al., Three-Dimensional Printing Techniques, U.S. Patent 5,204,055, filed 1989, issued 1993 4. F. Wille`me, Photo-Sculpture, U.S. Patent 43,822, issued 1864 5. M. Bogart, In Art the End Don’t Always Justify Means, Smithsonian, 1979, p 104–110 6. C. Baese, Photographic Process for the Reproduction of Plastic Items, U.S. Patent 774,549, filed 1902, issued 1904 7. F.H. Monteath, Photomechanical Process for Producing Bas-Reliefs, U.S. Patent 1,516,199, filed 1922, issued 1924 8. I. Morioka, Process for Manufacturing a Relief by the Aid of Photography, U.S. Patent 2,015,457, filed 1933, issued 1935 9. O.J. Munz, Photo-Glyph Recording, U.S. Patent 2,775,758, filed 1951, issued 1956 10. J.E. Blanther, Manufacture of Contour Relief Maps, U.S. Patent 473,901, filed 1890, issued 1892 11. B.V. Perera, Process of Making Relief Maps, U.S. Patent 2,189,592, filed 1937, issued 1940 12. N. Concordet, Three-Dimensional Exhibit, U.S. Patent 2,556,798, filed 1948, issued 1951 13. E.E. Zang, Vitavue Relief Model Technique, U.S. Patent 3,137,080, filed 1962, issued 1964 14. T.A. Gaskin, Earth Science Teaching Device, U.S. Patent 3,751,827, filed 1971, issued 1973 15. R. Baker, Method of Making Decorative Articles, U.S. Patent 1,533,300, filed 1920, issued 1925 16. F.W. Garver, Method of Producing Metal Rollers, U.S. Patent 3,007,231, filed 1960, issued 1961 17. W.D. White, Pressure Roller and Method of Manufacture, U.S. Patent 3,156,968, filed 1962, issued 1964 18. A. Ujiie, Method of Constructing Substantially Circular Cross-Section Vessel by Welding, U.S. Patent 3,665,143, filed 1970, issued 1972 19. H.T. Brandi and H. Luckow, Method of Making Large Structural One-Piece Parts of Metal, Particularly One-Piece Shafts, U. S. Patent 3,985,995, filed 1974, issued 1976 20. P.L. Gale and J.E. Fair, Method of Making Aluminum Piston with Reinforced Piston Ring Groove, U.S. Patent 4,125,926, filed 1977, issued 1978

21. C.O. Brown, E.M. Breinan, and B.H. Kear, Method for Fabricating Articles by Sequential Layer Deposition, U.S. Patent 4,323,756, filed 1979, issued 1982 22. W.K. Swainson, Method, Medium and Apparatus for Producing Three-Dimensional Figure Product, U.S. Patent 4,041,476, filed 1968, issued 1977 23. P.A. Ciraud, Verfahren und Vorrichtung zur Herstellung Beliebiger Gegenstaende aus Beliebigem Schmelzbarem Material (Process and Device for the Manufacture of Any Objects Desired from Any Meltable Material), German Patent Application DE 22 63 777 A1, 1972 24. W.E. Masters, Computer Automated Manufacturing Process and System, U.S. Patent 4,665,492, filed 1984, issued 1987 25. H. Bronowski, Device for Building up a Workpiece by Deposit Welding, U.S. Patent 4,621,762, filed 1985, issued 1986 26. K. Matsubara, Molding Method of Casting Using Photocurable Substance, Japanese Kokai Patent Application, Sho 51 (1976)10813, 1974 27. P.L. DiMatteo, Method of Generating and Constructing Three-Dimensional Bodies, U.S. Patent 3,932,923, filed 1974, issued 1976 28. T. Nakagawa et al., Press Technique, 1979, p 93–101 29. M. Kunieda and T. Nakagawa, Manufacturing of Laminated Deep Drawing Dies by Laser Beam Cutting, Proc. of ICTP, Vol 1, 1984, p 520 30. R.F. Housholder, Molding Process, U.S. Patent 4,247,508, filed 1979, issued 1981 31. J.J. Beaman, J.W. Barlow, D.L. Bourell, R.H. Crawford, H.L. Marcus, and K.P. McAlea, Solid Freeform Fabrication: A New Direction in Manufacturing, Kluwer, Boston, MA, 1997 32. J.J. Beaman, Solid Freeform Fabrication: An Historical Perspective, Proceedings of the Solid Freeform Fabrication Symposium, University of Texas Mechanical Engineering Department, Aug 6–8, 2001, p 584–595 33. H. Kodama, Automatic Method for Fabricating a Three-Dimensional Plastic Model with Photo Hardening Polymer, Rev. Sci. Instrum., 1981, p 1770–1773 34. A.J. Herbert, Solid Object Generation, J. Appl. Photo. Eng., Vol 8 (No. 4), Aug 1982, p 185–188

35. T. Wohlers and T. Gornet, History of Additive Manufacturing, Wohlers Report, Wohlers Associates, Ft. Collins, CO, 2016 36. M. Feygin, Apparatus and Method for Forming an Integral Object from Laminations, U.S. Patent 4,752,352, filed 1987, issued 1988 37. “Additive Manufacturing—General Principles—Terminology,” ISO/ASTM 52900, International Organization for Standardization and ASTM International, 2015 38. S. Crump, Apparatus and Method for Creating Three-Dimensional Objects, U.S. Patent 5,121,329, filed 1989, issued 1992 39. C.R. Deckard, Method and Apparatus for Producing Parts by Selective Sintering, U.S. Patent 4,863,538, filed 1986, issued 1989 40. D.L. Bourell, H.L. Marcus, J.W. Barlow, J.J. Beaman, and C.R. Deckard, Multiple Material Systems for Selective Beam Sintering, U.S. Patent 4,944,817, filed 1989, issued 1990 41. J.A. Manriquez Frayre and D.L. Bourell, Selective Laser Sintering of Binary Metallic Powder, Proceedings of the Solid Freeform Fabrication Symposium, University of Texas Mechanical Engineering Department, Aug 6–8, 1990, p 99–106 42. R. Larson, Method and Device for Producing Three-Dimensional Bodies, U.S. Patent 5,786,562, filed 1994, issued 1998 43. J.E. Rabinovich, Rapid Prototyping System, U.S. Patent 5,578,227, filed 1993, issued 1996 44. D.M. Keicher, J.E. Smugeresky, J.A. Romero, M.L. Griffith, and L.D. Harwell, Using the Laser Engineered Net Shaping (LENS) Process to Produce Complex Components from a CAD Solid Model, Lasers as Tools for Manufacturing I, Vol 2993, SPIE Conference (San Jose, CA), 1997, https://doi.org/10.1117/12.270018 45. D. Bourell, Perspectives in Additive Manufacturing, Annu. Rev. Mater. Res., Vol 46, 2016, p 1–18, DOI: 10.1146/annurevmatsci-070115-031606 46. “America Makes and ANSI Overview,” America Makes and ANSI Additive Manufacturing Standardization Collaborative (AMSC), https://www.ansi.org/standards_ activities/standards_boards_panels/amsc/ America-Makes-and-ANSI-AMSC-Overview

ASM Handbook, Volume 24, Additive Manufacturing Processes D. Bourell, W. Frazier, H. Kuhn, M. Seifi, editors DOI 10.31399/asm.hb.v24.a0006560

Copyright # 2020 ASM InternationalW All rights reserved www.asminternational.org

Design and Manufacturing Implications of Additive Manufacturing David Rosen, Georgia Institute of Technology Samyeon Kim, Singapore University of Technology and Design

ADDITIVE MANUFACTURING (AM) processes fabricate parts in a layer-by-layer manner in which material is added and processed repeatedly. This layer-by-layer process has important implications on part characteristics as well as on design opportunities and on manufacturing practices, supply chains, and even business models. This article introduces these implications and describes how they relate to the fundamental nature of AM processes, starting with the characteristics of AM processes and the parts they fabricate.

Characteristics of Additive Manufacturing Processes The unique features of AM processes have important implications on part characteristics and design opportunities. Complex or unique custom geometries can be fabricated readily. Complex material compositions are also possible that can lead to optimized mechanical property distributions. Additionally, functional devices with kinematic joints and embedded components can be fabricated in the AM machine, enabling highly integrated, efficient device designs.

Additive Manufacturing Process and Material Characteristics Seven classes of AM processes are recognized in an international standard (Ref 1). Insights can be gained by grouping the process classes according to how material is added and processed to fabricate a layer of a part. More specifically, some processes pattern material, others pattern energy, and one patterns both material and energy. Material extrusion (MEX), material jetting (MJ), binder jetting (BJ), and some sheet lamination (SLam) processes deposit material in a patterned manner. That is, MEX deposits a filament of material to fabricate a part layer, while MJ and BJ processes deposit patterns of droplets. Energy is

provided either through the deposition head or in a blanket manner over the entire layer, for example, when a lamp is used to cure the droplets of photopolymer resin that were just deposited. In contrast, powder-bed fusion (PBF) and vat photopolymerization (VPP) processes fabricate layers by projecting energy patterns onto a bed or vat of material. In this case, the shape of the cured layer is defined by the energy pattern. Directed-energy deposition fabricates layers by patterning both energy and material simultaneously. These process classes highlight some important similarities and differences among processes. The first characteristic to highlight is the insight that the material is constructed as the part is constructed. This is a critically important characteristic that differentiates AM from other manufacturing processes. The consequence is that material and part construction becomes coupled; it is impossible to separately control the formation of the material from the formation of the part. Consider the machining of a metal part from stock. The alloy composition and microstructure were controlled in the foundry, while the machining process affects only the surface and subsurface regions of the part. In injection molding, the chemistry and composition of the polymer are fixed in the feedstock pellets. In contrast, in PBF processes, alloy composition and part microstructure depend on the thermal history at each point in the powder bed. In MEX and polymer PBF processes, polymer parts have porosity that depends on powder size, process conditions, and/or scan paths. Other part characteristics that are almost universal include part dimensions being slightly larger than as-designed and rough surface finishes. The first of these has been called a material-safe characteristic. The idea is related to the surface finish characteristic: Many AM parts require finishing operations, such as sanding or machining, to yield smooth surfaces. As a consequence, preprocessing software ensures that part surfaces have excess

material that could be finished off without having part dimensions become too small. Due to the layer-by-layer and digital, or discretized, nature of AM processes, part surfaces are often rougher than surfaces produced by other manufacturing processes. In some cases, layer “stair steps” are evident on slanted or curved surfaces, particularly for MEX, SLam, and up-facing surfaces from VPP processes. In other cases, surface roughness originates from the material feedstock, particularly for powders used in PBF and BJ processes.

Unique Characteristics Due to layer-by-layer fabrication, AM processes have some unique characteristics (Chapter 17 of Ref 2). More specifically, they can exhibit complexity along several different dimensions:  Shape complexity: Parts can have virtually

any shape. Related to this is the capability to fabricate very complex shapes on one hand, and unique, customized shapes on the other. Furthermore, the shape complexity can become hierarchical across several size scales.  Material complexity: Material can be processed differently in different regions of the part, leading to different properties in these different regions. In some processes, multiple materials can be deposited and processed, leading to potentially complex material compositions and property distributions.  Functional complexity: Working devices can be fabricated in situ, without assembly, inside AM machines. This includes working mechanisms as well as parts with embedded devices. A little more explanation of functional complexity is warranted. In most processes, it is possible to fabricate working mechanisms directly in the AM machine, without the need for postfabrication assembly (Ref 3). In the PBF, BJ, and VPP processes, kinematic joints

20 / Additive Manufacturing Overview can be fabricated easily and will be operational if powder or liquid resin is removed from joint regions. Joints can also be fabricated in the MJ and MEX processes if the support structure can be removed. A corollary to the characteristic on concurrent material and part construction is that the interior of a part is always accessible during fabrication. One method for leveraging this characteristic is to insert external components and devices into parts as they are being fabricated (Ref 4). As an example, sensors can be embedded in parts in which high stresses are expected during usage. Or, actuators such as electric motors can be embedded into a mechanism during fabrication. In both cases, a high level of integration and efficiency can be achieved. See the section “Design Implications of Additive Manufacturing” in this article for further discussion.

Additive Manufacturing Machine and Technology Characteristics The reputation of industry-grade AM machines is that they are expensive, use expensive materials, and have small build volumes. Furthermore, the machines tend to be closed systems that do not integrate with other manufacturing hardware or software systems from other vendors. During the first 15 to 20 years of the AM industry (late 1980s through early 2000s), machines and technologies were developed by small companies, with patents that protected their technologies and with limited research and development budgets. Machines and materials were expensive due to the significant research and development required and the relatively small markets they served. At that time, the primary use of AM was to fabricate prototypes, hence the original name of the industry as “rapid prototyping.” To this day, AM machines and materials for production remain expensive. However, a quickly growing segment of the industry now provides much less expensive machines and materials, many of which fabricate good-quality parts (even if they are not quite as good as their expensive cousins). Conceptually, across the spectrum of AM processes, part quality and cost have a direct relationship. That is, if one wants a very highquality part, they can expect it to be expensive, regardless of the material or process. Conversely, if one is willing to accept lower quality, such as some porosity, limited small-feature definition, or limited accuracy, then its cost can be dramatically lower. However, even lowerquality parts can be much more expensive than mass-produced parts using conventional manufacturing processes. This is due to the layer-bylayer, sequential fabrication process inherent in AM. A part may be injection molded in seconds but require many hours in an AM machine. Fortunately, the reputation of AM machines and technology is changing. In recent years,

technologies have improved, patents have expired, and large companies have invested in the AM industry. Related technologies, such as lasers, have also improved tremendously. These trends led to lower machine costs, higher-quality machines, and many more applications that have become enabled. The emergence of low-cost, consumer-grade three-dimensional (3-D) printers selling for under $2000 has challenged the industry. On the positive side, millions of people now have access to AM technology, enabling the democratization of the technology and allowing people to explore innovations and new applications. The capability of installing lowcost manufacturing technology where it is needed has profound implications on what a factory could look like. Implications for distributed manufacturing opportunities are explored later in this article. This section closes with some observations of machine characteristics and their trends. As with any manufacturing machine, each AM machine has a limited build volume; that is, the size of parts that can be fabricated is limited. Many AM machines have build volumes that are less than 250 by 250 by 250 mm (10 by 10 by 10 in.) in size, with many lower-end machines less than one-eighth of that (100 to 125 mm, or 4 to 5 in., on a side). The largest commercially available machines can fabricate parts that are approximately 1 m (3 ft) in length. Reasons for this are historical (most mechanical part prototyping needs were for parts that can be held in one’s hand), economic (machine cost scales with build volume), and practical (1 m3, or 35 ft3, metal powder beds and resin vats have masses of hundreds of kilograms, which causes handling challenges). Process and machine innovations have led to AM machines with much larger build volumes. Gantry-style machine architectures have emerged for directed-energy deposition (DED) and MEX processes. Using shield gas instead of a closed chamber, some DED machines have build volumes several meters in length. By mounting an extruder on a gantry system, some MEX machines have similar build volumes. Other MEX and DED machines use robot arms to expand their build volumes. In the construction industry, MEX-based machines have used gantry or rail-mounted robots to have build volumes many meters in size.

Purposes of Additive Manufacturing Historically, AM was a rapid prototyping (RP) technology; that is, it was used to fabricate prototype parts directly from computeraided design (CAD) models without the need for hand work, hard tooling, process planning, and other delays. Additive manufacturing had a tremendous impact through RP, with many cases documenting significant savings in product development time and cost. In the early

1990s, AM parts were primarily visual prototypes that were used to communicate intended 3-D shapes. For parts with complex geometries, these physical prototypes were far better at communicating design intent than twodimensional (2-D) drawings or even 3-D CAD models. Soon afterward, the role of form-fit prototypes was filled. That is, prototypes were assembled, possibly with other AM parts or with existing conventional manufacturing (CM) parts, so that designers could check if parts fit together as intended and if the overall shapes were correct. After assembling the prototype product, many designers then tested the functionality of the product. During this phase of RP, many related applications of AM parts were investigated. Parts were used to test assembly processes and tooling by running the AM parts down assembly lines in place of production parts that were not yet available. In time, if parts were of high enough quality, AM parts were assembled into products that were shipped to customers. When the CM production parts became available, a field upgrade was performed to replace the AM part with the production part. These examples illustrate the significant impact that early AM parts had on product development, in particular, the impact on time to market. Another application served by AM parts was as prototype molds and dies. The term rapid tooling was used to indicate the application of RP technology to the development of tooling for conventional processes. This rapid tooling was used to fabricate a small number of prototype parts, with the advantage being the prototypes were fabricated by the intended production manufacturing process. In some cases, small production lots were molded using the rapid tools, at least until the prototyping tooling was available; this situation is known as bridge tooling. Even today (2019), considerable interest remains in the use of AM-fabricated tools for injection molding. Of even greater interest, however, is the use of AM to fabricate jigs and fixtures for assembly and tools for composites manufacturing, machining, stamping, and other manufacturing processes. In many cases, these jigs, fixtures, and tools are used for production purposes. Customized medical devices were one area of application for AM for both production manufacturing as well as production tooling. Hearing aid shells are almost exclusively produced by AM processes. Many dental restorations have AM-fabricated components, such as metal copings for crowns and bridges. Wax patterns for these dental restorations are also fabricated by using MJ; the copings are then investment cast. Molds for clear orthodontic aligners (e.g., Align Technologies) enable the production of millions of aligners per year. These applications require custom, patient-specific shapes. The economics of AM processes are such that small, uniquely shaped parts can be fabricated inexpensively.

Design and Manufacturing Implications of Additive Manufacturing / 21 Another broad application area is the production of aerospace components. Until recently, the applications were for low-productionvolume polymer components for military aircraft, which were installed during aircraft upgrades. More recently, some production metal components are being manufactured for aircraft engines. The fuel nozzles for General Electric (GE) LEAP engines are a well-known example, with over 30,000 produced per year (Ref 5). Applications are being pursued for metal airframe components fabricated by either PBF or DED processes. More than 500,000 AM machines were sold worldwide in 2017, including more than 14,000 industry-grade machines (Ref 6). The rest were low-cost desktop, or consumer-grade, machines. The breadth of applications for which these machines are employed is difficult to ascertain. Needless to say, AM machines are being used for many, many different applications, including toys, visualization aids, education items, prototypes, production, and other uses. The range of industries covered spans cinema (miniature sets, stop-action movies), architectural (models), engineering, health and medicine, and even food (cake decoration to food printing). Many trade magazines publish articles on these and other applications.

Design Implications of Additive Manufacturing

Design for Additive Manufacturing: Opportunities versus Constraints Design for additive manufacturing (DFAM) encourages designers to explore new concepts or develop new designs by using the unique capabilities of AM. The main objective of DFAM is to maximize product performance through the synthesis of shapes, sizes, hierarchical structures, and material compositions, subject to the capabilities of AM processes (Chapter 17.4 in Ref 2). In adopting AM, redesigning a component or system is necessary to take advantage of its benefits and opportunities, which vary depending on the extent of the redesign—from redesign of an existing single component to redesign of a whole system (Fig. 1). The number of benefits that can be derived from AM is based on the level of commitment and is related to the concepts of reversible substitution and nonreversible substitution, as shown in Fig. 1. As the commitment level progresses (with increasing benefit), there is a design threshold before which there is the possibility of reversible substitution and after which there is no going back, due to irreversible substitution. In the earliest commitment stage, long before the design threshold, an existing CM part is replaced like-for-like with an AM part, which may save on cost and tooling but does not

Benefit

Design for manufacture and assembly (DFM) has been developed to tailor designs to alleviate manufacturing constraints imposed by conventional manufacturing methods. The layer-based and additive nature of AM processes allow many more design opportunities and freedom than conventional manufacturing.

In this regard, when designers are adapting to AM, a shift in thinking is required to change from DFM to design for additive manufacturing (DFAM). Design for additive manufacturing takes into account the unique capabilities of AM and considers the manufacturability for intended AM processes, for example, minimum wall thickness. Accordingly, the next section describes unique AM capabilities.

offer the other benefits of AM (Ref 7). This is the concept of reversible substitution, in that the AM-printed part can be reversibly replaced by an existing CM part. The form-fit-function of the component is not changed, or the form is changed in a minor way for manufacturability for AM. In the next stage of commitment, also before the design threshold, a functionally improved AM part also could offer some cost benefits but also perhaps some weight benefits, thus a functional improvement (Ref 7). The component shape can be drastically redesigned to improve functionality, provided that the envelope and interfaces of the component remain unchanged. Needless to say, design under the concept of reversible substitution requires less design commitment. As the commitment increases further and the design threshold is crossed, a redesigned part or assembly that is irreversibly redesigned for AM can bring more benefits, such as fewer production tools and less waste (Ref 7). This is the concept of nonreversible substitution, in that the redesigned component cannot be replaced by an existing CM part design. Further still in the commitment, multifunctional design can occur, in which AM structural components could have embedded systems, resulting in weight and cost reductions. At the highest commitment level is system-level design for AM, in which a complete change in design philosophies would result in the realization of all of the benefits from AM. This level of commitment requires that AM be considered from the beginning in future cleansheet designs (Ref 7). Form-fit-function of an existing part, subsystem, or system would be open to change to take advantage of the unique capabilities of AM. Accordingly, it requires high commitment and in-depth knowledge about the AM potential of the existing design

System-level design for AM e.g., functionally integrated product

Multifunctional design e.g., radiator built into chassis

Functionally improved component

nottingham.ac.ut/research/groups/3dprg/experience-ourgroup.aspx

Redesigned part e.g., component with different packaging space or with graded material properties

Existing component built using AM

MTC

Commitment

Reversible substitution

Fig. 1

No going back

Concept of reversible and nonreversible substitutions. AM, additive manufacturing. Courtesy of the Manufacturing Technology Centre (MTC), Coventry, U.K.

22 / Additive Manufacturing Overview and AM capabilities to create a novel and customized design that may have complex geometries and multifunctions. The significant change of the form-fit-function would lead to novel design, which CM cannot fabricate, and changes in production planning and the supply chain that resist reversible substitution by CM designs. Table 1 provides the concept of reversible and nonreversible substitutions. The following are examples of design benefits and opportunities to support these concepts. Customized Parts with Complex Geometry Additive manufacturing can fabricate parts with complex geometry without tooling, such as molds and cutting tools, so that customized and complex geometries can be achieved readily. The complex design can be realized to yield unprecedented structural properties by optimizing part design. A part can be optimized to achieve the desired functions if the function of a part is defined mathematically (Ref 8). The complex design can be structured by using topology optimization, shape optimization, internal features such as conformal cooling channels in the injection molding tool, and cellular structures, for example, foam, lattice, and honeycomb structures. Depending on design applications, these complex geometries can lead to performance improvements: weight reduction, custom material properties and functionality, uniform temperature distribution, active cooling, energy absorption, and vibration control. The design freedom of AM can be used more extensively for product design. The degree of design complexity when using conventional manufacturing significantly influences manufacturing cost and time. However, manufacturing cost and time are relatively independent of the level of customization or geometric complexity of a part. Part Consolidation Additive manufacturing allows replacement of multiple parts that were manufactured by conventional manufacturing with a single part or a smaller number of parts. This redesign practice is known as part consolidation. It results in part count reduction, assembly interface integration, working mechanism

fabrication without assembly, and multiple functions that may also be achieved by multiple materials and complex shapes. These advantages reduce assembly time, repair time, shop floor complexity, replacement part inventory, qualification cost, and the number of required tooling (Ref 8). As an example, GE Aviation needed 900 separate components for a commercial helicopter engine, requiring some 10 to15 suppliers, various tolerances, and different types of fasteners, such as bolts, nuts, and welds, for assembly operations. However, through part consolidation achieved by using AM, GE Aviation consolidated 900 components to 16, which were 40% lighter and 60% less expensive (Ref 9). To consolidate parts that exploit AM benefits successfully, there are two critical factors to be considered: part candidate identification and feasibility check according to available AM processes. First, parts that are good candidates for AM tend to have complex and customized geometries, low production volumes, or special combinations of mechanical properties (Ref 8). Additionally, AM potential parts for consolidation may face manufacturing and assembly difficulties during conventional manufacturing processes, for example, the GE LEAP engine fuel nozzle. Secondly, a design feasibility check is required to decide the boundary for selecting part candidates. The feasibility check is performed after deciding a specific AM process based on rules (Ref 10) such as:  Material availability: Check the availability

of engineering materials for consolidating parts.  Material variance: Choose parts with the same material or with materials that are available on an AM machine. 100% density

 Size limitation: Consider the size of the

consolidated parts, which should be less than the build chamber size.  Maintenance frequency difference: Choose parts with similar or reduced maintenance frequency.  Related function: Choose parts that have similar or related functions, a module consisting of a collection of parts, or modules that support primary functions.  High assembly difficulty between parts: Choose parts or assembly interfaces that have high assembly difficulty, which is related to assembly time, cost, and quality. Functionally Graded Material As mentioned previously, material complexity can be achieved by varying the material organization within a component to achieve an intended function (Ref 11). The aim is to derive heterogeneous characteristics in different regions in the part. The best example of material complexity is functionally graded materials (FGMs). The FGM parts attain their multifunctional status from densification of material or gradation of materials, or both, as shown in Fig. 2 (Ref 12). Densification of material means providing different mechanical properties in a part by controlling the material density locally. In terms of densification of material, a single material from powder-bed fusion for metal can be used to fabricate functionally graded parts. Gradation of material means control of the transitional gradation of two or more materials. Accordingly, multiple-material designation and a number of possible combinations provide the functionally graded characteristics that design engineers want in a part. Design for Constraints Although the value of AM is a high degree of design freedom, the design freedom is 50% density

10% density

Material densification (single material)

Table 1 Summary of reversible and nonreversible substitution

Parameter

Reversible substitution

Nonreversible substitution

Design for manufacturing

Design for function

Form of an existing Fixed or design partially fixed Fit of an existing Fixed design Function of an Fixed existing design Redesign level Part level

Free to redesign

Material A

Material A + material B

Material gradation (multimaterial)

Free to redesign Free to redesign Part, subsystem, and system level

Fig. 2

Continuously graded microstructure of functionally graded materials

Material B

Design and Manufacturing Implications of Additive Manufacturing / 23 limited by AM manufacturing constraints that often lead to problems in many AM processes, such as overhangs, abrupt thickness transitions, trapped volumes, layering, and cleanliness issues (Ref 2, 8). In this regard, the designfor-constraints methodology has been studied to discover manufacturing constraints from AM through experimental and benchmark studies (Ref 13). These design constraints can be embodied by design rules that are function independent and result in direct constraints in relation to part design, process, and material properties (Ref 14, 15). The design rules provide acceptable design parameter ranges of specific design features, joints, or mechanisms when printing these features by a specific AM process (Ref 16, 17). Most design rules focus on constraints during the manufacturing process rather than on potential constraints for postprocessing, such as support structure accessibility and inspection.

Additive versus Conventional Manufacturing Processes Conventional manufacturing is restricted by the need for tooling, while AM, which fabricates parts without tooling, provides design freedom in this regard (Ref 18). However, compared to CM, current AM processes have drawbacks such as slow manufacturing speed, limited available materials, limited size of final parts, and lower surface quality. The selection of suitable manufacturing processes between CM and AM can be challenging. Accordingly, the general selection process in Fig. 3 is introduced to guide users in deciding why and when AM should be used. The first step is to translate design requirements into process requirements by identifying design objectives, functions, and constraints. Design objectives must be decided to achieve goals such as weight reduction and performance enhancement. They exert constraints on the selection of processes. Two types of constraints would be considered: technical constraints and quality constraints. Because both constraint types are closely linked to manufacturing processes, a deeper understanding of both CM and AM is required to support the selection of appropriate manufacturing processes. Examples of information for this step are:  Objectives:

a. Minimize cost, weight, material usage, and so on b. Maximize performance  Constraints: a. Technical constraints: material type and its availability, technology limitations b. Quality constraints: achieve desired requirements, such as accuracy, surface finish, and mechanical properties, from manufacturing processes

Fig. 3

General additive manufacturing versus conventional manufacturing selection process

The second step is to check material availability, which is critical for process selection. Selecting material is a way to find the best match between design requirements and material properties. The AM processes for a given material can significantly influence the material properties of final parts and their consistency. Therefore, mechanical properties of AM materials, such as tensile strength, yield strength, elongation at yield/break, toughness, and hardness, should be considered while selecting a material. In addition to mechanical and material properties, the compatibility of the material with the part usage environment can be important. The resistance of the material to chemicals, humidity, corrosion, and wear should be considered if relevant to the usage environment. For medical devices, resistance to bodily fluids and tissues is important, because biocompatibility can be critical. After materials selection, material availability for manufacturing processes should be considered to match process attributes and design requirements. This is critical to adopt AM because materials have limited availability. If a material for a part is not available for AM processes, CM should be applied to fabricate the part. The third step is to decide process alternatives after materials selection. After AM process alternatives have been determined for a specific part design, several factors for process selection should be considered to screen unsuitable AM processes. There are four process selection factors:

 Factor related to technical constraints:

a. No optimized process for a given material: Theoretically, metal materials such as SUS316L or 1.2709 (maraging steel) can be printed by all BJ and PBF processes, especially selective laser melting and electron beam melting (EBM). However, when there are no identified process parameters to print a defect-free part for a given material, the process should not be used to fabricate parts with that material (Ref 19). For example, EBM does not support SUS316L or 1.2709 but can fabricate parts with TiAl, nickel alloy 718, Ti-6Al-4V, and CoCr.  Factors related to quality constraints: b. Surface quality: Surface quality is closely related to the staircase effect because of layer-by-layer AM fabrication. Vertical or horizontal surfaces may have acceptable surface roughness, but slanted surfaces, curved surfaces, and holes and other shapes may have unacceptably high roughness or have illformed shapes. Machining or other postprocessing operations may be required to achieve the desired surface quality. c. Accuracy: Accuracy is also linked to surface roughness, which can be dealt with through postprocessing. Machine capability influences accuracy. Part design and the build chamber environment can cause significant dimensional change to occur because of distortion

24 / Additive Manufacturing Overview during manufacturing and heat treatment. Accordingly, selecting processes that are stable to provide accuracy or minimum tolerance would be preferred. A trade-off between accuracy and lead time may be required. To fabricate parts with high feature resolution, a minimal layer thickness is required, which makes the production time longer. d. Postprocessing: When selecting AM processes or comparing AM to CM, postprocessing plays a significant role. Postprocessing for AM includes support structure removal, machining, coating, and heat treatment, to name a few. Types of postprocessing and corresponding costs and efforts should be considered when selecting processes. For example, when selecting among metals, their machinability should be considered if good surface finish is required on some surfaces. This is particularly relevant for superalloys because many are difficult to machine due to poor thermal conductivity, inclusions, or deformation mechanics. Additionally, postprocessing difficulty should be considered for parts with complex geometry. Support structure removal in lattice structures or internal cavities can cause problems. Similarly, accessibility for machining to achieve good surface finish can be difficult for these geometries. The fourth step, which may be carried out concurrently with the third step, is to check AM potential for part design and design feasibility based on technical constraints and economic aspects. Checks based on technical constraints are for manufacturability analysis because of the different capabilities of AM processes. Feasibility checks based on economic aspects check economic risks and benefits that are raised by adopting AM for design applications. Specific concerns for design feasibility are:

consolidation, complex geometry, multifunctional design, and material complexity. For example, AM supports the manufacture of a hydraulic manifold with curved fluid channels that can minimize pressure drop, which is a performance enhancement at the part level. At the same time, the AM-printed hydraulic manifold may allow the capacity of the pump supplying the fluid to be downsized, which is a performance enhancement at the system level. Therefore, design complexity is a good indicator of the suitability of AM.  Build time and cost: Build time and cost are critical factors in the decision to adopt AM for applications. They can be estimated by calculating material cost, energy consumption, layer thickness (resolution), part volume (or part size) (Ref 20), and the number of parts per batch during manufacture. Postprocessing is often required, for example, to remove support structure and improve surface quality. In this regard, postprocessing time and cost should also be considered when selecting AM processes. Further discussion is offered in the section “In the AM Machine” in this article.  Production volume and rate: Financial analysis would be required to analyze cost and benefit for adopting AM in terms of production volume and rate. Generally, AM is more beneficial if small production volumes are required, while CM would be more beneficial for mass production of identical parts. To identify costs and benefits for AM, production volume, rate, and DFAM should all be considered together, as shown in Table 2. The last step is to select a suitable process. As described in Table 3, from previous steps a selected material and the AM process alternatives are decided by considering various factors. In this step, decision-making is performed by multiple stakeholders, who may have different concerns, constraints, changeable priorities, and uncertainties, which affect the process selection. All of these considerations must be

taken into account when developing the proper set of selection attributes. For example, the selection attributes can be mechanical properties, such as ultimate tensile strength, Rockwell hardness, or density, as well as geometric complexity, build time, and cost. After that, the relative importance of each attribute must be decided according to the preferences of the stakeholders. When the process selection is changed by the replacement of material, attributes, or process alternatives, the relative importance of each attribute may be changed accordingly. Finally, the suitable AM process can be selected after identifying the trade-offs among the competing alternatives.

Design for Additive Manufacturing Methods Although the term design for additive manufacturing has been used extensively, only a few studies have focused on developing DFAM methods (Ref 21). Design for additive manufacturing is defined as the “synthesis of shapes, sizes, geometric meso-structures, and material compositions and microstructures to best utilize manufacturing process capabilities to achieve desired performance and other lifecycle objectives” (Ref 2, 22, 23). From the definition, it is realized that the primary concern of DFAM methods is the investigation of AM design potentials and design optimization opportunities. This new concept of DFAM methods is classified into three levels of abstractions from system level, to part design level, to processspecific design level. The DFAM methods that introduce relationships between design and AM and its impact on designers are classified into system-level design methods that are comprehensive and systematic. Reference 24 proposed a global DFAM method to take into account product functionality and manufacturing constraints for AM simultaneously at the early design stage. This is because

 Part size: The build chamber is a region

within the AM system where the parts are fabricated (Ref 1). It limits the size and volume of part designs. If the part does not fit in the build chamber, then the part cannot be printed by AM or should be divided into multiple parts.  Minimum feature size: The minimum feature size depends on machine capability. If the minimum feature size of a part is less than the printable minimum feature size of an AM machine, it may cause manufacturing failure or oversized features, depending on the AM process.  Design complexity and design requirements: Complex part designs may be generated by part consolidation or design optimization to achieve the desired performance. A part with a complex design requires the unique capabilities of AM, such as part

Table 2 Suitable manufacturing process according to production volume and part design Production volume

Small volume Large volume

Identical design

Complex design

AM more beneficial CM more beneficial

AM more beneficial Depends on design complexity(a)

Note: AM, additive manufacturing; CM, conventional manufacturing. (a) If a complex part design requires unique AM capability, e.g., a complex design, then AM can be the suitable manufacturing process for mass customization. If the part design is achievable by CM, then CM may be the suitable manufacturing process.

Table 3

Selection of a suitable additive manufacturing (AM) process

Given A selected and available material; AM process alternatives for the material Identify Set of attributes:  Aforementioned factors in previous steps (design objectives, functions, and constraints); process alternatives (can be considered as attributes for decision-making)  New attributes from different stakeholders can be considered. Rate Decide relative importance of attributes Rank Rank AM process alternatives Adapted from Ref 2

Design and Manufacturing Implications of Additive Manufacturing / 25 functional optimization and manufacturing processes such as part orientation and manufacturing paths are the keys to determining the appropriate part design in the global DFAM method. Specifically, part orientation determines the functional surfaces of a part that are critical for accuracy and surface finish requirements and influences mechanical properties and overall manufacturing time and cost. Manufacturing path optimization involves optimizing a specific tool path for achieving geometry, mechanical properties, and target manufacturing time and cost. Reference 8 provides an overall DFAM process that indicates various design considerations for AM decisions. Detailed design considerations and warnings to designers related to manufacturing constraints are described in the standard. Rosen (Ref 22) provides an overall DFAM method based on process-structure-propertybehavior mappings. In the mappings, the overall DFAM method represents that functional requirements are mapped to properties and geometry that satisfy these requirements. Reference 25 analyzed conventional DFAM methods and proposed a newly structured DFAM method to fully exploit AM potential by integrating existing methods and tools such as quality function deployment. Reference 26 proposed an AM-enabled design method, which is a design process for functional integration. The design process highlights two main steps: design analysis at the functional level and structure optimization to achieve better performance. At the part design level of abstraction, DFAM methods that focus on design for functionality are considered. Design for functionality indicates that if the functions and design objective of a part are defined mathematically, the part can be optimized to achieve these functions and objectives (Ref 27). Design for functionality aims to maximize performance of designs or determine the trade-off between performance and weight. Unlike the system-level design methods, the design methods for functionality focus on generating optimal structures according to design objective rather than modeling holistic design processes (Ref 26). Three main types of optimization problems have been explored for DFAM (Ref 2, 27):

mechanical, thermal, optical, and biological properties accompanied by relatively light weight (Ref 21), for example, light weight with strong mechanical properties in aerospace engineering, good energy absorption, good thermal and acoustic insulation in mechanical engineering, and osseointegration in the biomedical field (Ref 22). At the process-specific design level of abstraction, DFAM focuses on design rules and design guidelines for different AM processes. Design guidelines focus on design principles to encourage designers to design parts that take advantage of AM (Ref 26, 28), which is not mandatory but recommended. In contrast, design rules deal with manufacturability, related to specific AM process limitations, and provide suitable ranges of design parameters, such as overhang angle, thin-wall thickness, and gap heights, to achieve best quality. Design rules have been derived from benchmark and experimental studies for different AM processes. The results of these studies provide specific design rules to alleviate manufacturing constraints for AM in the design stage for successful fabrication (Ref 13, 16, 29).

Examples Complex Geometry A satellite part with lattice structures from Thales Alenia Space, shown in Fig. 4, is one of the best examples to elaborate complex geometry that cannot be manufactured by subtractive manufacturing. The part size is 134 by 28 by 500 mm (5.3 by 1.1 by 20 in.), and its mass is 1.7 kg (3.8 lb). Heat-exchanger designs like this can exhibit better heatexchange efficiencies in smaller sizes and reduced masses than designs fabricated by conventional manufacturing processes. More generally, the benefits of lattice structures and other complex geometries are parts that are lighter and have better mechanical properties, per unit mass, than conventionally designed and manufactured parts. The original GE jet engine bracket in Fig. 5(a) was designed to support the weight of the engine during handling. It may be used periodically but is always attached in the engine. The original bracket weighed 2033 g (4.5 lb). To significantly reduce the weight of the engine, GE and GrabCAD held a global 3D printing challenge (Ref 31). The redesigned bracket

Fig. 4

Metal printed satellite component with lattice structure from Thales Alenia Space. Source: Ref 30. Reprinted from https://www.metal-am.com/large-scale-metal-lattice-structure-complex-satellite-part/, copyright 2016 Thales Alenia Space, by permission from Thales Alenia Space

Fig. 5

GE jet engine bracket. (a) Conventional bracket. Source: Ref 31. (b) Redesigned bracket. Source: Ref 32

 Size optimization: in which values of dimen-

sions are determined

 Shape optimization: in which shapes of part

surfaces are changed

 Topology optimization: in which distribu-

tions of material are explored Topology optimization, as a DFAM approach, has received much attention recently due to its tendency to create geometries that are often too complex for conventional manufacturing processes to fabricate readily. Similarly, cellular materials such as honeycombs, lattices, and foams can be used to optimize structure. The key advantage of cellular structures is that they can produce specific

26 / Additive Manufacturing Overview

Part Consolidation In the example of the hydraulic manifold in Table 4, the AM-redesigned manifold becomes lighter and occupies less volume. Furthermore, internal paths for fluids are optimized, so pressure drop has reduced. This part is a good example to illustrate the reversibility concept introduced in Fig. 1. If only the form of the original part is redesigned, then the substitution is reversible. If the form-fit-function of the original part is redesigned, for example, changing location of inlet, outlet, and assembly interface of the part, then the substitution is nonreversible. This is because these design improvements may influence the supply chain and the assembly process.

Manufacturing Implications of Additive Manufacturing The layer-by-layer nature of AM processes has important implications for AM machines, product fabrication processes, factory layouts, and even supply chains. These implications can lead to new business models as well. These topics are explored in this section.

Implications in the Additive Manufacturing Machine In the section “Characteristics of AM Processes” in this article, the characteristics of

AM parts and machines were described, which have important implications for the use of AM machines and the development of future machines. Part fabrication is a sequential process. Parts are fabricated layer by layer and, in some cases, layer fabrication occurs scan by scan. As such, part fabrication can take a long time and be significantly more costly than conventional mass production processes. A rough order-of-magnitude example illustrates the consequences. Consider a machine that requires 10 s to fabricate a layer, and each layer is 50 mm thick. For a part that is 10 cm (4.0 in.) tall, the build will require 2000 layers and 20,000 s (5.55 h). If the parts are polymer, the AM machine may cost $10/h to operate, including material cost, which translates to a part cost of $55.50. If 100 parts can be fabricated simultaneously by arraying them in the build volume, the part cost reduces to $0.55, which is more reasonable. However, this may be an order of magnitude more than a comparable injection-molded part. If the AM process is scan based (e.g., MEX or uses a laser), layer fabrication times can be 10 to 100 longer, which translates directly to increased cost. Note that metal parts cost at least 10 more than comparable polymer parts using typical PBF machines. On the positive side, the AM process can fabricate just about any cross-sectional shape, enabling high geometric complexity, as stated earlier. Additionally, no hard tooling is required, meaning that tools do not need to be designed, fabricated, retrieved from a warehouse, installed, maintained, and so on. To a first approximation, AM part cost is

Table 4 Examples of hydraulic manifold before and after redesign for using additive manufacturing Parameter

Redesign level Redesign purpose

Material Mass, kg

Original block manifold

First-iteration geometry

Second-iteration geometry

Single part

Single part

Scope: reduce material mass and improve functionality, within the same space

Extract flow paths of original design Reduce volume Use desirable material (stainless steel 316L) Optimize the flow paths using computational fluid dynamics analysis Design permanent support structure Aluminum alloy 316L stainless steel 12.3 ( 52%) 16.3 ( 36%)(higher density than Al) 4650 ( 52%) 2040 ( 79%) Optimized flow paths Weight reduction (max 52%) Significantly smaller volume (79%) Compatibility with existing design

Aluminum alloy 25.6

Volume, cm3 9600 ... Benefit Adapted from Ref 33

independent of production volume. Stated differently, tooling costs need not be amortized over the production volume of parts, as in molding, casting, stamping, and other conventional manufacturing processes. Machining often requires process planning, setup, and tooling; associated costs are amortized over the production volume. To a first approximation, the relative part costs are as illustrated in Fig. 6. The trade-off points, where the curves cross, vary greatly depending on part size and complexity. However, the trends indicate why small, customized medical parts and low-production-volume parts are fabricated by AM. Novel part consolidation strategies that yield multifunctional parts and greatly reduce assembly costs can push the trade-off points far to the right, as in the case of the GE fuel nozzles. Additional insights can be gained by further considering the strategy taken to add and process material. Layer fabrication times are inherently slow when using a scanning strategy, whether it is to deposit material (e.g., MEX) or pattern energy (scanning for PBF, VPP, and DED). These are referred to as onedimensional (1D) process strategies. If scanning is parallelized (i.e., parallel 1D) or replaced entirely by 2D processing methods, fabrication times can be reduced tremendously. In MJ processes, up to thousands of nozzles can be arrayed to deposit material simultaneously. In digital micromirror device (DMD)-based VPP processes, entire part cross sections can be fabricated at one time. With these processes, layer fabrication time can be very short, sometimes only a few seconds. Sheet lamination processes are inherently 2D because an entire sheet is deposited to comprise a layer; cutting out the part cross section is a scanning process but needs to be conducted only for the periphery, which is typically much shorter than scanning a cross-section interior. In recent years, the AM industry has seen an increasing number of process innovations that significantly increase fabrication speed. The adoption of DMDs, instead of scanning lasers, in VPP processes has been mentioned. The introduction of inverted builds and the continuous liquid interface production technology

Single part Cost per parts

that won first place in the challenge is shown in Fig. 5(b). Its mass was reduced approximately 84%, to just 372 g (0.8 lb) (Ref 32).

Additive

Subtractive Forming

Number of parts

Fig. 6

Cost comparison according to the number of parts. Source: Ref 34

Design and Manufacturing Implications of Additive Manufacturing / 27 has virtually eliminated recoating times in printers from Carbon. The invention of novel binding and binding-prevention inks has enabled 2D PBF processes to emerge, such as the multijet fusion technology by HP. The adoption of powder metallurgy materials in BJ processes enabled the emerging low-cost metal AM technologies from Desktop Metal, Markforged, HP, and potentially others. This trend is expected to continue. Due to their simultaneous deposition of both material and energy, DED processes have fabrication freedoms that other processes do not. The DED processes can be used to fabricate parts but are frequently used for part repair or for fabricating features on conventionally manufactured parts. Because parts need not be grown from a powder bed or resin vat, metal can be added directly onto an existing part. Worn metal parts and molds are routinely repaired in this manner. Similarly, features such as bosses or ribs can be added on to cast, forged, or machined parts, enabling each manufacturing process to be used for the shapes and situations to which it is best suited. Some MEX processes are emerging that use robot arms to carry the extrusion head. These developments should enable optimized deposition strategies for large, complex parts. Additionally, they should enable feature fabrication on conventionally manufactured polymer and polymer composite parts. The final topic to be discussed is that of machine controls. Most AM machines operate in an open-loop manner, meaning that machines do not use sensors for closed-loop control. The exception is many DED machines, which have melt-pool sensors to control deposited bead quality. As such, AM machines lack the capability to monitor material and part quality or to adjust their process as parts are fabricated. Monitoring the effects on material of lasers or electron beams as they scan at 10 to 50 m/s (33 to 164 ft/s) is challenging. Nonetheless, researchers and companies are developing sensing and control strategies that will impact AM machines in the near future. Such technologies are expected to greatly increase part fabrication repeatability and reliability.

Implications on the Factory Floor As the focus expands from the AM machine to the work cell and factory floor, other implications of AM emerge. Two main types of implications are highlighted in this section: new physical part workflows and opportunities for integration. As discussed in the section “Unique Characteristics” in this article, components and devices can be inserted into AM parts as they are fabricated. This enables a product assembly strategy that is sometimes called monument assembly, in which parts are conveyed to a central location for assembly into the product, in contrast to typical assembly lines in which work in process travels along conveyors and parts are delivered to different assembly points. One can imagine

putting robot arms into the build chamber of an AM machine or having a robot feed parts into an AM machine. Physical part workflows could be reconfigured into a hubs-and-spokes arrangement, in which the hubs are AM machines or work cells consisting of AM machines and robots, while the spokes are conveyors for parts to be assembled. In this manner, functional products with moving parts and embedded sensors, actuators, or other high-value components can be manufactured directly in an AM machine. The monument assembly strategy and huband-spokes work cell configuration can result in a highly flexible, agile manufacturing capability. The AM machines are digitally driven. If the part-delivery system is also digitally controlled, changes can be made easily to product configurations, enabling the factory floor to respond quickly to changes in customer demands. It may be possible to achieve simplified factory floor operations with significant increases in product variety with this strategy. The second topic relates to integration of equipment across the factory floor. At present, most AM machines are stand-alone pieces of equipment, whose only connection to the outside world is an ethernet plug. Increasingly, however, AM machines are being designed to integrate with other manufacturing equipment, because the reality is that AM machines are often part of a process chain that is required to deliver high-quality parts for production applications. Most parts require postprocessing, which may include powder removal (PBF and BJ), cleaning (VPP, DED), support structure removal (metal PBF, VPP, MEX, MJ), finish machining, painting, and so on. In the early 2000s, Align Technology developed specialized equipment to automate the cleaning, postcure, and support-removal operations for their aligner molds, along with specialized conveyors. They collaborated with 3D Systems to develop modified VPP machines that could interoperate with their automated systems to fabricate millions of aligner molds per year. Increasingly, companies would like such automated systems, customized to their products, without the considerable effort to develop specialized equipment. The manufacturing sector is working toward using AM equipment that physically integrates with conventional manufacturing equipment and work cells. Also, AM equipment is increasingly integrated electronically using industry standard protocols, such as MTConnect, and with enterprise software systems, such as manufacturing execution systems. This trend is expected to continue for many years as more companies seek to use AM for production purposes and to improve the efficiency and effectiveness of their factory operations.

New Manufacturing Models Additive manufacturing also has implications at the enterprise and supply chain levels. A number of factors combine to enable several

models for distributed manufacturing. Taken to an extreme, some online networks, such as 3dhubs.com and makexyz.com, have emerged that enable individuals who purchase low-cost printers to build parts for (typically local) people who want parts fabricated. The point is that AM technologies fabricate designs that are delivered digitally, can operate automatically, do not require tooling, and can be economical for short-run or custom manufacturing applications. That is, the old rules of trying to achieve economies of scale for mass production no longer apply with AM. People and companies appear to want products that meet their needs exactly or are customized to their preferences or even body shapes. Furthermore, they prefer immediate delivery. Mass production in centralized factories fails each of these desires. Local production capabilities that enable customer co-design can provide customized products with quick turnaround. Kiosks on the beach that fabricate sunglasses with custom-designed frames are one example. It is now technically feasible for audiologists to offer same-day hearing aids. With recently developed scanning technology, they can scan a patient’s ear and design a custom hearing aid shell. The shell can be printed in minutes with a small AM machine and the final hearing aid produced by assembling the electronics into the shell for delivery to the patient within an hour or two. Car dealerships could fabricate customized add-on components before delivery to the customer. Or, they could fabricate parts as needed for repair, instead of stocking spare part inventories. More generally, the concept of digital inventory or digital warehouses becomes feasible. A typical home supply store (e.g., Home Depot or Lowe’s) stocks many thousands of different parts and products. Imagine if they were equipped with AM machines that could fabricate parts on demand, instead of having to carry such large inventories. They would stock powders, resins, and other feedstock materials instead. Additionally, they could offer much greater product variety and customization opportunities. Following this line of thinking, it can be seen that different types of supply chains will emerge. The home supply store would not need to be serviced by thousands of different companies; rather, they could have only a few suppliers of their feedstock materials. A more near-term supply chain disruption is the use of local service bureaus to supply parts. Many service bureaus are now capable of supporting production manufacturing of AM parts for the aerospace or medical industries. Some operate on a centralized model but take orders delivered electronically from anywhere in the world, because it is easy to email a CAD file. Others have regional facilities to reduce shipping time and cost. One example is Fast Radius, which has several U.S.-based facilities and an Asia-based facility. One of their U.S. facilities is co-located with the United Parcel

28 / Additive Manufacturing Overview Service (UPS) hub in Louisville, Kentucky, so they can use UPS logistics capabilities to minimize delivery delays. Additive manufacturing technologies are enabling significant changes and new opportunities at manufacturing companies. New types of products with new customization capabilities are enabled. Companies have new opportunities for “getting close to the customer,” with a range of distributed facilities options to be explored. All of this will facilitate the emergence of new supply chains and even business models, product skews, and product ranges. In addition to the technical aspects of DFAM and AM, the opportunities afforded by AM will require new approaches to educating engineers as well as training AM-qualified technicians on how to gain the most advantage from AM.

10.

11.

12.

13. REFERENCES 1. “Standard Terminology for Additive Manufacturing—General Principles—Terminology,” ISO/ASTM 52900, International Organization for Standardization, ASTM International, 2015 2. I. Gibson, D. Rosen, and B. Stucker, Additive Manufacturing Technologies: 3D Printing, Rapid Prototyping, and Direct Digital Manufacturing, 2nd ed., Springer, 2015 3. C. Mavroidis, K.J. DeLaurentis, J. Won, and M. Alam, Fabrication of Non-Assembly Mechanisms and Robotic Systems Using Rapid Prototyping, J. Mech. Des., Vol 123 (No. 4), 2000, p 516–524, doi:10.1115/ 1.1415034 4. A. Kataria and D.W. Rosen, Building around Inserts: Methods for Fabricating Complex Devices in Stereolithography, Rapid Prototyp. J., Vol 7 (No. 5), 2001, p 253–262, doi:10.1108/13552540110410459 5. “New Manufacturing Milestone: 30,000 Additive Fuel Nozzles,” GE Additive, Oct 4, 2018, https://www.ge.com/additive/ blog/new-manufacturing-milestone-30000additive-fuel-nozzles 6. T. Wohlers, “Wohlers Report 2018: 3D Printing and Additive Manufacturing State of the Industry Annual Worldwide Progress Report,” 2018 7. “Additive Manufacturing: Applications in Aerospace,” INSIGHT_08, Sept 2018, The Aerospace Technology Institute (ATI), Cranfield, U.K., https://www.ati. org.uk/media/sogpibd0/insight08-additivemanufacturing.pdf 8. “Additive Manufacturing—Design—Requirements, Guidelines and Recommendations,” ISO/ASTM 52910, International Organization for Standardization, ASTM International, 2018 9. T. Kellner, “An Epiphany of Disruption: GE Additive Chief Explains How 3D Printing Will Upend Manufacturing,” GE Reports, Nov 13, 2017, https://www.ge.com/ reports/epiphany-disruption-ge-additive-

14.

15.

16.

17.

18.

19.

20.

chief-explains-3d-printing-will-upend-manu facturing/ S. Yang and Y.F. Zhao, Additive Manufacturing-Enabled Part Count Reduction: A Lifecycle Perspective, J. Mech. Des., Vol 140 (No. 3), 2018, p 031702– 031712, doi:10.1115/1.4038922 G.H. Loh, E. Pei, D. Harrison, and M.D. Monzo´n, An Overview of Functionally Graded Additive Manufacturing, Add. Manuf., Vol 23, 2018, p 34–44, https:// doi.org/10.1016/j.addma.2018.06.023 E. Pei, G.H. Loh, D. Harrison, H.D.A. Almeida, M.D. Monzo´n Verona, and R. Paz, A Study of 4D Printing and Functionally Graded Additive Manufacturing, Assem. Autom., Vol 37 (No. 2), 2017, p 147–153, doi:10.1108/AA-01-2017-012 G.A.O. Adam and D. Zimmer, Design for Additive Manufacturing—Element Transitions and Aggregated Structures, CIRP J. Manuf. Sci. Technol., Vol 7 (No. 1), 2014, p 20–28, https://doi.org/10.1016/j. cirpj.2013.10.001 H. Jee and P. Witherell, A Method for Modularity in Design Rules for Additive Manufacturing, Rapid Prototyp. J., Vol 23 (No. 6), 2017, p 1107–1118, doi:10.1108/ RPJ-02-2016-0016 S. Kim, D.W. Rosen, P. Witherell, and H. Ko, “A Design for Additive Manufacturing Ontology to Support Manufacturability Analysis,” paper presented at the ASME 2018 International Design Engineering Technical Conferences and Computers and Information in Engineering Conference (Quebec, Canada), 2018; Vol 2A: 44th Design Automation Conference (Paper DETC2018-85848), http://dx.doi.org/10.1115/ DETC2018-85848 J. Kranz, D. Herzog, and C. Emmelmann, Design Guidelines for Laser Additive Manufacturing of Lightweight Structures in TiAl6V4, J. Laser Appl., Vol 27, 2015, p S14001, doi:10.2351/1.4885235 C.C. Seepersad, T. Govett, K. Kim, M. Lundin, and D. Pinero, “A Designer’s Guide for Dimensioning and Tolerancing SLS Parts,” paper presented at the 23rd Annual International Solid Freeform Fabrication Symposium—An Additive Manufacturing Conference, Aug 6–8, 2012 (Austin, TX) R. Hague, S. Mansour, and N. Saleh, Material and Design Considerations for Rapid Manufacturing, Int. J. Product. Res., Vol 42 (No. 22), 2004, p 4691–4708, doi:10.1080/00207840410001733940 P.K. Gokuldoss, S. Kolla, and J. Eckert, Additive Manufacturing Processes: Selective Laser Melting, Electron Beam Melting and Binder Jetting—Selection Guidelines, Materials, Vol 10 (No. 6), 2017, p 672, doi:10.3390/ma10060672 M. Baumers, C. Tuck, R. Wildman, I. Ashcroft, E. Rosamond, and R. Hague, “Combined Build-Time, Energy Consumption

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

and Cost Estimation for Direct Metal Laser Sintering,” paper presented at the 23rd Annual International Solid Freeform Fabrication Symposium—An Additive Manufacturing Conference, Aug 6–8, 2012 (Austin, TX) M.K. Thompson, G. Moroni, T. Vaneker, G. Fadel, R.I. Campbell, I. Gibson, et al., Design for Additive Manufacturing: Trends, Opportunities, Considerations, and Constraints, CIRP Ann., Vol 65 (No. 2), 2016, p 737–760, https://doi.org/10.1016/j.cirp. 2016.05.004 D.W. Rosen, Computer-Aided Design for Additive Manufacturing of Cellular Structures, Comput.-Aided Des. Appl., Vol 4 (No. 5), 2007, p 585–594, doi:10.1080/ 16864360.2007.10738493 D.W. Rosen, “Design for Additive Manufacturing: A Method to Explore Unexplored Regions of the Design Space,” paper presented at the 18th Annual International Solid Freeform Fabrication Symposium—An Additive Manufacturing Conference, Aug 6–8, 2007 (Austin, TX) R. Ponche, J.Y. Hascoet, O. Kerbrat, and P. Mognol, A New Global Approach to Design for Additive Manufacturing, Virtual Physical Prototyp., Vol 7 (No. 2), 2012, p 93–105, doi:10.1080/17452759. 2012.679499 M. Kumke, H. Watschke, and T. Vietor, A New Methodological Framework for Design for Additive Manufacturing, Virtual Physical Prototyp., Vol 11 (No. 1), 2016, p 3–19, doi:10.1080/17452759.2016. 1139377 S. Yang and Y.F. Zhao, Additive Manufacturing-Enabled Design Theory and Methodology: A Critical Review, Int. J. Adv. Manuf. Technol., Vol 80 (No. 1), 2015, p 327–342, doi:10.1007/s00170015-6994-5 D.W. Rosen, Research Supporting Principles for Design for Additive Manufacturing, Virtual Physical Prototyp., Vol 9 (No. 4), 2014, p 225–232, doi:10.1080/ 17452759.2014.951530 E. Atzeni, L. Iuliano, P. Minetola, and A. Salmi, Redesign and Cost Estimation of Rapid Manufactured Plastic Parts, Rapid Prototyp. J., Vol 16 (No. 5), 2010, p 308–317, doi:10.1108/13552541011065 704 I. Gibson, G. Goenka, R. Narasimhan, and N. Bhat, “Design Rules for Additive Manufacture,” paper presented at the 21st Annual International Solid Freeform Fabrication Symposium—An Additive Manufacturing Conference, Aug 9–11, 2010 (Austin, TX) “Large Scale Metal Lattice Structure for Complex Satellite Part,” Metal AM, Jan 6, 2017, https://www.metal-am.com/largescale-metal-lattice-structure-complexsatellite-part/ “GE Jet Engine Bracket Challenge,” GE and GrabCAD, 2013, https://grabcad.

Design and Manufacturing Implications of Additive Manufacturing / 29 com/challenges/ge-jet-engine-bracketchallenge 32. T. Kellner, “Jet Engine Bracket from Indonesia Wins 3D Printing Challenge,” GE Reports, 2013, https://www.ge.com/

reports/post/77131235083/jet-engine-bracketfrom-indonesia-wins-3d-printing/ 33. “Hydraulic Block Manifold Redesign for Additive Manufacturing,” Renishaw, https:// www.renishaw.com/en/hydraulic-block-

manifold-redesign-for-additive-manufacturing– 38949 34. A.B. Varotsis, “3D Printing vs. CNC Machining,” 3D Hubs, https://www.3dhubs.com/kno wledge-base/3d-printing-vs-cnc-machining

Polymer Additive Manufacturing Processes Division Editor: David L. Bourell, University of Texas at Austin

Vat Polymerization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Feedstocks for the Process . . . . . . . . . . . . . . . . . . . . . . . . . . . Safety Issues with Feedstock Handling . . . . . . . . . . . . . . . . . . Manufacturing Issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Postprocessing/Finishing . . . . . . . . . . . . . . . . . . . . . . . . . . . . Part Properties and Common Defects . . . . . . . . . . . . . . . . . . . Special Topics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

33 34 35 36 37 38 38

Material Extrusion Additive Manufacturing Systems . . . . . . . . 40 Melt Extrusion 3D Printing . . . . . . . . . . . . . . . . . . . . . . . . . . 40 Viscous Extrusion 3D Printing . . . . . . . . . . . . . . . . . . . . . . . . 50 Powder Bed Fusion of Polymers Thermal Issues . . . . . . . . . . . . Safety Considerations . . . . . . . Manufacturing Issues . . . . . . . Postprocessing and Finishing . .

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

52 53 54 54 55

Common Defects and Part Properties . . . . . . . . . . . . . . . . . . . 55 Case Studies in Polymer Powder Bed Fusion . . . . . . . . . . . . . . 56 Material Jetting of Polymers . . . . . . . . . . . . . . . . . . . . . . . . . . Process Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Process Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Part Design and Processing Considerations . . . . . . . . . . . . . . . Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Future of Polymer Material Jetting . . . . . . . . . . . . . . . . . .

58 58 59 61 63 64 66

Modeling for Polymer Additive Manufacturing Processes. . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Material Extrusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Polymer Powder-Bed Sintering/Fusion . . . . . . . . . . . . . . . . . . Vat Photopolymerization . . . . . . . . . . . . . . . . . . . . . . . . . . . . Material Jetting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

69 69 69 70 73 75

Copyright # 2020 ASM InternationalW All rights reserved www.asminternational.org

ASM Handbook, Volume 24, Additive Manufacturing Processes D. Bourell, W. Frazier, H. Kuhn, M. Seifi, editors DOI 10.31399/asm.hb.v24.a0006553

Vat Polymerization Don Smith, Baxter Healthcare Corp. (Retired)

VAT POLYMERIZATION (VP) is an additive manufacturing (AM), or three-dimensional (3D) printing, process in which 3D objects are produced by hardening a liquid polymer into the desired shape. The vat is the liquid resin that is hardened by way of light energy. The light energy polymerizes the resin into the shape of an object by printing two-dimensional layers, much like the pages of a book. User input is a 3D computer-aided design file that is sliced into crosssectional data called a Z-slice. (The terms Z-slice and layer are used interchangeably.) Each Zslice is printed individually on the VP machine, using either a directed beam of light or a projected image. As each layer is produced, the light energy also penetrates into previous layers, adhering the layers together. This “gluing” effect bonds each successive layer to the object, creating a cohesive part. Polymerization in the X, Yplane is likewise interconnected, creating a very solid part. Vat polymerization parts are almost fully dense and can be polished for clarity. In the 1980s, VP parts were clear amber in color. By the 2000s, the clarity of parts had improved, and white or black parts became available. By this writing (2019), part properties have improved and now rival thermoplastic molded resins. Vat polymerization resins offer a wide variety of final part properties that make VP one of the most versatile of the AM processes. Carbon 3D and EnvisionTEC introduced resin with elastomeric properties in 2017, and 3D Systems, Inc. in 2019. One benefit over other AM processes is that this versatility can be experienced with little investment. Resin can be changed by switching out the vat and material cartridge and cleaning the elevator. Resins are available that can fit many secondary processes, such as finishing and painting, casting, and lowpressure molding. With the introduction of new materials and improvements in material properties, VP offers a good alternative for AM for low-volume production. Users are able to use the same process/material for prototype testing as in production. Production methods, process controls, and part properties rival other AM production methods. Production and material quality controls are being developed that will mainstream VP as a compliant group of processes.

As with other AM processes, throughput is lower in VP than in other automated processes, such as computer-aided manufacturing and injection molding. However, some companies are using several machines at once to create a workflow that increases part production. Also, if material is not changed between builds, a stereolithography apparatus machine can be restarted within minutes after a previous build is completed. Vat polymerization is a group of two main processes: stereolithography apparatus (SLA) and digital light processing (DLP). Both processes use a liquid plastic polymer that changes state to a solid by means of light energy. As light energy is absorbed by the liquid resin, the resin turns to a solid form. The action is analogous to film emulsion, where light focused on photographic film changes to an image. The SLA uses light and heat from a focused laser to accomplish the transformation, whereas DLP uses a light to initiate a chemical reaction.

Fig. 1

Figures 1 and 2 show the functional aspects of the SLA and DLP processes, respectively. The SLA builds a part on a platform submerged in the resin vat. As each layer is printed, the platform drops farther into the vat and deposits uncured resin on the top surface of the part, which is then printed for the next layer. The liquid polymer is solidified by means of an ultraviolet (UV) beam of light. The UV light travels through a series of lenses and moving mirrors focused on the resin surface. Beam diameter can be focused from 75 mm (0.003 in.) for fine detail and 750 mm (0.030 in.) for speed. Printing continues until the part is complete. Polymerization is approximately 90% at this stage. The part is removed from the platform and washed with a solvent or detergent to clean off uncured resin. After the part dries, it is then put into a UV postcuring station to fully cure the part. The vat dimensions can be as large as 1500 by 750 by 550 mm (59 by 30 by 22 in.).

Schematic of stereolithography apparatus (SLA) process. Source: Ref 1. Used by permission from Formlabs

34 / Polymer Additive Manufacturing Processes

Fig. 2

Schematic of digital light processing (DLP) process. Source: Ref 2

The SLA can be used to build a mix of very large and smaller parts. Very small parts require high-definition settings and should be built on one platform with other small parts. Parts built with thick layer settings build faster than high-definition layers of 0.05 mm (0.002 in.). The recoating process takes anywhere from 20 s to 1 min, and if the number of layers doubles because of thinner layers, recoating time doubles. The DLP process uses light to project an image onto the bottom of a sheet of glass. The light energy hardens the liquid resin in the same way that the SLA works. The vat in this case is a pool of liquid resin deposited onto a sheet of glass. This pool is several millimeters thick. The initial support layers are cured onto an elevator platform, and, as subsequent layers are printed, the part travels up and away from the glass. As the part moves away from the glass, liquid resin fills the space between the part and glass. The process is repeated until the part is complete. Postprocessing consists of removing the part from the elevator platform and baking the part or rinsing the part of resin. The light source is tuned for the resin and can be anything between a laser, incandescent light, or liquid crystal display and projector technology to project the image onto the glass screen. Digital light processing is faster than the SLA because there is no drawing but rather a projected image of light. Digital light processing has a size restriction smaller than the SLA because of the recoating and imaging process. The larger the cross-sectional area, the longer it takes for resin to reach the center of the area as the part moves upward. Also, as the part cross section becomes larger, the light source must move farther from the glass to maintain accuracy, and light energy must be more intense. For DLP, postprocessing consists of washing the parts and curing with heat and/or light for final mechanical properties.

Digital light processing has several variants for projecting light and recoating resin. Original machines had an incandescent bulb and a proprietary projector to project one image per layer. This has the benefit of time over the process of “drawing” a layer, because all layers and platforms take the same time to print. Other sources of light energy are laser and light-emitting diode projectors or directedlaser printing, as in traditional stereolithography projector technology. The variation in light source has led to an increase in resin mechanical properties. For recoating, there are also several processes. While the SLA uses a recoater blade and dipping to deposit a new layer of resin, DLP uses either a rocking motion or membrane technology to recoat the part with new resin. Stereolithography was patented in the 1980s by Chuck Hall. During the same time period, engineers in Japan and France were doing development work on the SLA but failed to carry the work to a finished process or to patent it. Chuck Hall founded 3D Systems, Inc., which built the first commercially available SLA machine in 1987. In the late 1980s, Cubital Ltd. in Tel Aviv, Israel, produced a liquid polymer process/ machine that used UV light energy, projected through a xerographic image printed on glass, to solidify a 0.10 mm (0.004 in.) layer of resin. In the process, named solid ground curing, UV light passed through a negative xerographic image of the layer to polymerize the resin. Unpolymerized resin was vacuumed away, and the vacuumed space was backfilled with hot wax. The wax was cooled and the entire layer was milled with a large-diameter fly cutter. After milling the cured resin and wax layer to 0.10 mm (0.004 in.) thickness, fresh liquid resin was applied and a new xerographic image was prepared for the next layer. When complete, the block was washed with hot water to clear away wax, revealing the fully cured parts. This process is no longer available but illustrates the variety of AM production methods. Materials for early machines were very brittle; although the VP process was intriguing, low-value material properties gave stereolithography a bad reputation as a robust prototyping/ manufacturing process. Material development was occurring at Somos and 3D Systems; within several years, less brittle and faster resins became available. Several user groups and university laboratories helped to promote AM by experimentation and providing feedback on material and machine technology through member forums and wish lists.

Feedstocks for the Process Material for all of the VP processes is a liquid polymer that changes to a solid when exposed to light energy. The material is stored in light-blocking plastic containers, usually

dark brown to black. Shelf life is usually 1 year. The liquid photopolymer is clear, milky, or dark in color, with the color being similar to the color of a fully cured part. Raw material and uncured parts should be handled carefully, with safety precautions for the user. The raw material is a specific composition, customized by the manufacturer for specific properties and machine capabilities. Raw material should not be cross contaminated, and care should be taken not to intermix material containers. For the same reason, containers should not be reused unless a check is in place to ensure that the same material goes into the reused container. Materials should be stored at room temperature or cooler in a closed cabinet to prevent unwanted temperature fluctuation and exposure to light. Too cool of an environment for material storage will lengthen the time needed before the material is ready for use in the VP machine. Too warm of a storage environment will shorten the shelf life and may start polymerization. Use first-in, first-out rotation of material. Consult with the manufacturer and review the safety data sheet for resin-specific data. If unsure of the quality of the material, a viscosity test can be used to help determine if the material has started polymerization. When transferring material to the VP machine, the liquid material is either poured or pumped into the vat as raw material stock for part production. In some machines, material is slightly heated above 25  C (75  F), so that it is at a constant machine-controlled temperature for photoinitiation and cross linking. The vat portion of the machine is enclosed to maintain temperature, filter light and avoid contamination. Materials for VP are photopolymers and are either epoxy or acrylic based, although some newer photopolymers are elastomeric with a base of urethane. Vat polymerization resins are thermoset plastics that are polymerized by molecular binders. Once polymerized, the finished form is set and cannot revert to a moldable form. The photopolymer is made up of binder, monomer, and photoinitiator molecules. Photopolymers for VP are designed to initiate at a specific wavelength of light energy, depending on the process, light source, and intensity. As light is absorbed by the polymer, it changes state into a solid form. The photoinitiator starts the chemical reaction when exposed to light energy, and the binder and monomer combine into more complex molecules to form a solid. The reaction can occur with heat also, but at a slower rate. Cured materials have a wide range of properties. Table 1 shows a sample of commonly available materials and their published properties. During the building process, the material is partially cured after exposure to light, describing the layer of geometry. As the next layer is produced, the light penetrates into previous layers to promote bonding of the layers. This bonding of the current layer to previous layers penetrates the part by one or two layers and is

Vat Polymerization / 35 Table 1 Sample of several materials and their physical properties

Material

Accura25(a) Accura Xtreme White 200(a) Accura Clear View(a) Accura PEAK(a) Clear FLGPCL04(b) FLTOTL05(b) FLDUCL02(b) Carbon Resin RPU 70(c) Carbon Resin EPX 82(d) Carbon Resin FPU 50(e) ABS Flex White(f) E-Clear Series(g) E-Poxy(h)

Tensile modulus (ASTM D 638)

Flexural modulus (ASTM D 790)

MPa

ksi

MPa

ksi

Elongation (ASTM D 638), %

1380–1660 2350–2550 1980–2310 4180–4790 2800 2700 1260 1900 ± 200 2800 860 ± 110 1772 2150–3250 4400–4630

200–240 340–370 287–335 606–695 406 392 183 275 ± 29 406 125 ± 16 257 312–471 638–671

1590–1660 2300–2630 2270–2640 4220–4790 2200 1600 820 1800 ± 300 3000 831 ± 36 2543 1200–1500 ...

230–240 334–381 329–383 612–695 319 232 119 261 ± 44 435 120 ± 5 369 174–218 ...

13–20 7–20 3–15 1.3–2.5 6 24 49 100 ± 20 5.9 280 ± 15 6.6 2–4 3.0–3.7

Notched impact strength at 23  C (73  F) (ASTM D 256) J/m

ft-lb/in.

19–24 55–66 40–58 21.3–27.3 25 38 109 22 ± 1 44 40 ± 5 ... 30 ...

0.36–0.45 1.03–1.24 0.75–1.09 0.40–0.51 0.47 0.71 2.04 0.41 ± 0.019 0.82 0.75 ± 0.09 ... 0.56 ...

Heat-deflection temperature at 455 kPa (66 psi) (ASTM D 648) 

C

58–63 47 51 153 73.1 48.5 43.3 70 104(a) 78 ... 75 at 1.82MPa 92 at 1.82 MPa



F

136–145 117 124 307 163.6 119.3 109.9 158 219(a) 172 ... 167 at 264 psi 198 at 264 psi

(a) Source: Ref 3. (b) Source: Ref 4. (c) Source: Ref 5. (d) Source: Ref 6. (e) Source: Ref 7. (f) Source: Ref 8. (g) Source: Ref 9. (h) Source: Ref 10

one of the key factors in the success of VP. Figures 3 and 4 show the theory of laser printing. The theory of adhesion also applies to DLP projection. The materials that are available for a specific machine or process are limited by the manufacturer for the capabilities of their technology. Materials selection for a particular machine is limited to materials compatible to that process. Manufacturers design and test materials to run successfully on their machines. Changing materials for VP machines is a simple process but must be controlled to avoid contamination. Material properties and samples should be examined carefully before selecting a process. For the SLA process, the height of the largest part (in the Z-dimension) plus the elevator height must fit the vat available. When selecting a process, the Z-height of parts should be taken into consideration. The vat must be full prior to running the process, and there must be enough resin in reserve to keep the vat full. The vat of material can be changed to another material/vat in a matter of hours. The complete vat is changed, and the elevator and recoater system must be cleaned. Materials selection for an individual machine is limited to the number of vats the user is willing to purchase. For the DLP process, the amount of material in the reservoir is much less, and the travel of the elevator determines the maximum Z-height. The pumping system and build reservoir contact the liquid resin. To switch to another resin, these must be cleaned of the old resin. New resin is added to the system, and building can continue. In either case, the liquid resin must remain uncontaminated by other resins and solvents, so that it cures properly and exhibits the desired mechanical properties.

Safety Issues with Feedstock Handling Vat polymerization resins are a liquid plastic thermoset consisting of a photoinitiator,

z y x z y x

Fig. 3

Fig. 4 “Bullet” of light penetrating and curing stereolithography apparatus resin. Source: Ref 11

monomer, and binder. Most of the VP materials in the raw state are considered sensitizers and have a National Fire Protection Association health safety rating of “2” (Ref 12–15). Sensitizers can cause an allergic reaction after repeated low-dose exposure or a single highdose exposure. Sensitization is usually from repeated low-dose exposure. Skin sensitization manifests as a dermatitis reaction, and, if contracted, the person should consult a physician: “Sensitization dermatitis is the result of an allergic reaction to a given substance. Direct skin contact is necessary to cause sensitization. Individuals may become sensitized to a substance after a troublefree period of exposure. There are many factors which affect a person’s susceptibility, including existing skin diseases, personal habits, and individual sensitivity. Once a person is sensitized, even a minute exposure may trigger a severe dermatitis reaction that may spread over the body. Sensitization is permanent, so a sensitized individual should be removed from potential contact with the sensitizer” (Ref 16). A rash could develop in the exposed area. Once sensitized, the person is always sensitive, and lower doses can cause a dermatitis reaction. The best protection against sensitization is protective clothing and clean work practices.

Overcure of the resin, which adheres layers together. Source: Ref 11

The resin manufacturer or distributor and safety coordinator should be consulted to decide on the proper personal protective equipment (PPE) and help in developing a workflow. Ask questions and discuss alternatives. Use the material safety data sheet (MSDS) as a resource, and ensure that all warnings are understood. Trade articles and brochures provided by resin manufacturers also should be reviewed. Some suggestions for PPE include:  Chemical-resistant gloves for the hands.

Use gloves that fit well and are not too loose, so that fine work can be accomplished when wearing them. Do not leave the work area or go to “clean” benches while wearing a contaminated glove. Spreading resin to other work surfaces or parts is very easy and can happen without user intent. Remove both gloves when leaving the “wet” work area, and put on a new pair when returning to work.  Eye protection. Uncured SLA resins can cause serious irritation to the eyes. Minimum eyewear should be safety glasses with side shields. Face splashes can happen during cleaning. A face shield will offer extra protection for the face and eyes. Put on and take off safety glasses with clean hands or gloves. Do not wear contact lenses. Vapors can get underneath contacts and harm the eye.  Sleeve protection for arms. Sleeve protection or cuffs offer a defense against

36 / Polymer Additive Manufacturing Processes splashes and uncured parts when reaching and working during the cleaning process. Long-sleeved lab coats offer protection, but there is usually a gap at the wrist between the glove and sleeve. Sleeve cuffs help cover this gap.  Lab coat, long pants, and closed-toe shoes are a good defense for exposed skin and clothing against splashes, reaching, and drops of resin or uncured parts. Splashes and resin can soak into clothing and lay against the skin for hours. Use shoe covers if needed. Leather can absorb resin and cause irritation over time.  It is also important to use safe handling practices while removing parts from the machine and loading platforms. For example, open a latch/door with a clean gloved hand, and keep a finger or two clean so that the door can be closed again upon leaving the machine with a platform.  Think ahead so that the production area will not become contaminated with resins. Develop a workflow that promotes safety. Change gloves as needed. Do not skimp on gloves but rather think ahead and plan a process that will keep the user and area as clean as possible. Use “clean” tools when working on fully cured parts, and use an extra set of needed tools for the “wet” area, where parts are cleaned before curing. Some of the VP processes require cleaning with a solvent such as isopropanol, acetone, or tripropylene glycol methyl ether (TPM). Safe work practices, engineering controls, and PPE are different for resins than solvents or detergents. Consult the MSDS and the manufacturers for all chemicals used in the process, and follow the safest guidelines for the worst case. Identify incompatible chemicals and ensure they are kept separate in the work area. Some examples of engineering controls that may be needed include:    

Ventilation (hood) to remove vapor fumes Eye wash station Flammable cabinet for volatile liquids “Wet” work area, where uncured resin can be cleaned and cured. This is necessary even for the processes that produce fully cured parts. Elevator platforms, production components, and parts will have uncured resin, and the user will need a specific place to work with wet parts.  A sink where resin is cleaned should be separate from a hand-washing sink.  A work area to glue and assemble parts  Sanding table with a side or down draft to minimize inhalation of dust from finishing A key safety consideration for VP processes is laser/light safety for the eye. Liquid resins cure at a specific wavelength and can be glued outside of the machine with UV-curable glue or epoxy resins. Work stations should be available for this work, and eye protection for UV-

curable glue is specific. Read and follow safety precautions for tinted eyewear. Not all tinted eyewear is safe for the intense UV light used in gluing parts. Laser light for VP is highenergy light and can damage the eye in a matter of seconds. Lasers should not be adjusted or repaired without proper, specific training. Most adjustments will void the warranty, so consult the machine manufacturer to perform laser maintenance. Likewise, eye protection should be used when working on the light projector or lasers on a DLP system. Keep food and drink away from the production areas. Ingestion of liquid resin and cleaner is different than inhalation and can be toxic. Food and drink in the proximity of any chemical work area is dangerous. Walls or negative air flow should be constructed to keep the desk area isolated from the work area for machines and postprocessing. Keep air moving away from the office area and toward the cleaning work benches. Postprocessing chemicals should be exhausted, so the flow of air should be from the desks, to machines, to the cleaning stations, and out. Figure 5 shows a schematic of a proper room layout for air flow in a production area.

Manufacturing Issues As with any manufacturing process, AM has characteristics that are unique to the process and will dictate certain constraints to part orientation and design. Each technology is different, but for VP the main characteristics for consideration are support structures, recoating, resin viscosity and color, and laser power. Design factors influence part production and, when controlled, produce a better part. Design constraints for VP are wall thickness, hollow or bowl-shaped cross sections, feature mass, and feature orientation during the build process. While almost any part can be built successfully, designing for the process and reviewing the design with the manufacturing group will increase the chances of success.

Fig. 5

Support structures, as shown in Fig. 6, keep the first layer of a structure from floating away during recoating and support the first several layers during the recoating process. The sweeper arm, dipping, and elevator movement between layers put vertical forces on the layer. Because the first layer(s) is very thin, it would bulge or bow without supports. With the action of the recoating mechanism causing movement of the resin, the layer would float out of position. Supports are generated for any surface facing the elevator mechanism (downward-facing surface for SLA; upward-facing surface in DLP) no matter how small, but support-generation software usually creates too many supports. Some key tips for VP support structures include:  Large, flat, downward/upward-facing areas

need support. As the recoating process and elevator move the part, sagging and movement of the first few layers would occur, causing an irregular part surface or bad build. This applies to all downward/upward features on the part, not just the first layers. The manufacturer’s default support spacing should be used and modified as the user gains experience. Curved-bottom surfaces will sometimes be supported as one feature by the support-generating software. The sliced geometry should be viewed to make sure the first layers are supported correctly.  Overhanging geometry must be supported, and the support generator will add these. However, some overhanging geometry is so small that it does not need a support. A small overhang (1 to 2 mm, or 0.04 to 0.08 in.) can possibly be supported by the part without the need for additional supports.  Downward-facing surfaces that are not the first layer, are smaller than the support spacing, and are supported by the part do not need a support. A few examples of this include a downward-facing counterbore for a screw, embedded lettering, or a side-facing hole.

Schematic diagram of air flow for a vat polymerization work area

Vat Polymerization / 37

Fig. 6

Support structures for a stereolithography apparatus. Parts built by vat polymerization are supported by pinpoint support structures. Source: Ref 17

 Hollow cross sections (trapped volume) still

need a support structure on the elevatorfacing surface(s). The resin will not support these first layers on the top of the trapped volume during the recoating process. They are similar to the initial layers of the part. For a closed volume when supports cannot be removed from the part, a geometric pattern feature can be substituted for the support structure. Thickening the support will also make the support strong enough to mimic a feature.  Hollow cross sections must be drained of resin in the cleaning process, so a draining mechanism, such as a hole, will need to be designed or added. A slot or hole can be added and a plug designed to close off the cleaned trapped volume as a secondary process. Leave approximately 0.15 mm (0.006 in.) clearance for the plug assembly process. The design and manufacturing group should work together to produce a good part(s).  Tall supports require close review to ensure the supports will stay in place during recoating. The recoating process creates lateral and vertical forces that can displace supports, causing subsequent layers of supports and geometry to not be attached. The unattached support and part features will float away or cause the elevator mechanism to jam and stop the build process. Add thickness to tall supports, or create a shaped support rather than a line, to withstand recoating forces. Recoating of the parts with new resin on each layer is a part of the VP process. As each layer is finished, the part dips, moves, or a recoating action takes place so that a new uncured resin layer is in place to be cured by the light source. Too little resin during recoating causes voids in the part and possible crashed builds, while too much resin causes the part to swell and can also cause a crash. With larger cross sections and closed volumes, the user must pay close attention to recoating, resin viscosity, and dipping. Thick resin viscosity will increase force on the part and increase recoating time. If the resin is changed or becomes too viscous, the resin manufacturer should be consulted to maintain quality building of supports and parts and proper operation of the machine. When resin is changed, machine parameters must be reviewed.

The resin must be kept clean and free of debris. In a vat of SLA resin, uncured supports, small particles, and partially cured resin build up over time. Stray light or loose particles cause the photoinitiator and binder to continue to react with the monomer and solidify the resin into a gel. The vat must be maintained by periodically cleaning the resin with a strainer to prevent the whole vat from becoming too viscose. A screen with an open mesh is a good tool for cleaning the vat. The DLP machines that have a smaller vat also must be kept clean, so that resin contamination does not hinder the recoating process. The VP resins are sensitive to light, so all resin must be kept in closed containers and machine doors kept closed. Ultraviolet shielding on room light fixtures helps to reduce ambient light from polymerizing resin in the machine and on parts left for rinsing. If parts are left in the cleaning area without rinsing, the liquid resin will polymerize with light exposure and time. Finished parts should be cleaned as soon after building as possible. Resin color also affects the build. Light does not penetrate colored resin as easily as clear resin. A more solid color requires changed settings for higher laser power and/ or slower draw speed to build. Laser/light intensity must be maintained for each resin in accordance with the machine manufacturer and resin parameters. Mirrors and lens clarity can change over time from debris and chemical vapor. These must be cleaned by trained technicians. Light intensity will also change over time and should be checked regularly during a periodic preventive maintenance.

Postprocessing/Finishing Vat polymerization parts are cured to approximately 90 to 95% when finished building. Postprocessing includes draining the platform, removing parts and supports from the platform, removing supports, cleaning the part of excess resin, and final curing of the part. Some DLP parts are fully cured when removed from the platform and only need draining, removing from the platform, cleaning supports, and rinsing the parts. The platform must be cleaned and made ready for the next production build. It is good practice to have more than one platform, to decrease production downtime. During the cleaning process, proper PPE and environmental, health, and safety (EHS) controls should be in place to protect the user and the environment (see the section “Safety Issues with Feedstock Handling” in this article). After the build, the VP machine can automatically, or with user input, raise the part(s) out of the resin vat. After raising the platform out of the vat, there is excess resin on the part(s) and platform. The platform should be allowed to drain until the excess resin is almost completely drained from all parts. When setting up the platform for the build and

arranging the parts, it is good practice to plan for draining, along with supports and build time. The SLA machines have features on the elevator to facilitate draining, where the platform can be tilted to help drain the resin. For DLP, the platform can be raised so that all parts are out of the resin. After draining, remove the entire platform from the machine and transport it to a cleaning station. The cleaning station should consist of a work bench, rinsing apparatus, and sink. The area must be ventilated with engineering controls in accordance with an EHS plan. Platforms and parts can be rinsed manually or automatically in a bath. To rinse parts automatically, simply put the whole platform, or all of the parts, into a parts washer with agitation. The parts washer will run for a specific time related to the number and size of parts and the resin used. For manual rinsing, remove the parts from the platform and remove as many supports as reasonable before rinsing. Use brushes to remove the resin and some type of agitation, depending on the scale of the operation. Agitation can include manual swooshing, ultrasonic cleaners, or a parts washer faucet. After rinsing, let the parts dry. Pressurized air drying can atomize resin and cleaning solution and must be assessed for safety and environmental impact. When the parts are dry, some DLP parts are fully cured and can be lightly sanded, finished, and/or delivered. Most parts will need postprocessing to fully cure the parts. The machine/ resin supplier and personal experience will inform the user if postcuring is needed. A UV light box, purchased for this purpose, is used for postcuring. Postcuring usually takes between 15 min and 1 h, depending on the part design, mass, amount of residue, and the resin used. Closed volumes and hidden areas will need proper placement to direct UV light to hidden areas. Turn the part(s) during the curing cycle to facilitate curing. Having tacky parts after postcuring indicates possible problems in the production process that should be investigated. Possible causes of tacky parts are low laser power, cleaning solution that is saturated with resin, and an insufficient cleaning process. Investigate the process by checking the simplest or most obvious step fist. Cleaning liquid resin off of the part is a meticulous process and can be automated to facilitate efficiencies. Parts should be inspected prior to postcuring if they are either cleaned manually or with automation. The cleaning solution for VP parts should be assessed for saturation also. If the cleaning solution is too saturated with resin, it can cause parts to be coated with resin rather than being washed. Some tests for saturation include visually looking at the cleaner for cloudy fluid or performing a viscosity test of the cleaning solution. Keep the cleaning solution covered to help prevent evaporation. Work with the resin and machine suppliers to help solve production issues.

38 / Polymer Additive Manufacturing Processes Finishing of VP parts can be as varied as the skills, expertise, and equipment available. The VP resins have a variety of finished part properties that will affect sanding, finishing, and painting. The part properties for a prototype can be matched closely to a final material for a particular design but need only satisfy the purpose of the prototype. For finished-goods production using VP, manufacturing engineering, design engineering, and quality groups must select the proper material and process for the application. The design of the part, orientation during building, and postprocessing should be optimized to produce a quality part and quantities. For manufacturing of finished goods, the design of the part, orientation during building, and postprocessing require process controls to ensure consistent quality. Manufacturing engineering should help to design and control the process. Processes and procedures should be followed according to the International Organization for Standardization/ASTM International standards when applicable. Some possible secondary finishing techniques, while not an exhaustive list, may include: sanding, filing, gluing, sandblasting, smoothing, media tumbling, painting, plating, and dipping. The VP parts can also be used as part of a secondary process, such as room-temperature vulcanized (RTV) rubber molding. The RTV molding can be used to increase production capacity and/or to attain specific part properties or simulate a molded material. The RTV molding can eliminate the need to sand and paint each production part to specification, because the mold can be made from a single finished part, called a pattern or master. The master is used to create the RTV mold, and subsequent parts are molded from the RTV mold. The RTV molding can produce from 25 to 100 parts. Parts that are larger than the machine build volume can be glued together to create a larger part. Specific glues using UV-curable resins or cross-linking polymers (i.e., two-part epoxy) are available. QuickCast (3D Systems, Inc.) is a lost pattern process whereby the SLA part is made with a matrix rather than as a solid part. The part is sealed, watertight, sanded, and dipped in ceramic. The ceramic is fired in a kiln, and the SLA pattern is burned and cleaned out of the casting (lost), leaving a cavity for metal to be poured into. This method is used when a metal part is being developed and prototype or low quantities are required. There are also waxy VP resins available for casting that are used as lost patterns.

polypropylene or an elastomer for prototyping, and some resins have unique properties that make them a preferred material for manufacturing. Material for specific secondary processes, such as QuickCast and plating, are also available. Those who prefer to use SLA for RTV master patterns can use a resin that is easily sanded and finished. Machine manufacturers are also the source for proprietary resins that work best with their AM machine(s). Process parameters are tuned for a particular resin that exhibits desired properties. Table 2 shows utilizations rather than properties of different resins, to show the wide variety of available uses for VP resins. Because the resins are thermosetting resin, parts with thick cross sections and large flat areas can warp or swell. Thick cross sections can be eliminated by coring out the section and filling with a matrix or similar design element. The matrix can also serve as a support structure rather than a void, if properly planned. If the cross section is a closed volume, it will need to have a drain mechanism for the liquid resin, postprinting. Liquid resin cannot be left in a finished part, because it will not fully cure and will be a possible health hazard. For large, flat areas, use finite-element analysis to determine the most beneficial design. Possibly add ribs or design elements that will stiffen the flat feature. Determine which resin would be most beneficial for the design. Tune the machine to ensure parameters are correct for the resin, and check resin for contamination and expiration. Machine parameters and support structures can be designed for specific layers, as needed. Postprocessing can also have an effect on the final part properties. Green parts (prior to postcuring) must be in a controlled process during cleaning and curing. Parts that are not supported can warp during cleaning and

Table 2

Vat polymerization parts have a wide variety of part properties. Improvements in resin design have all but eliminated early issues with brittle parts. A typical resin can mimic

Special Topics Design for AM is an important aspect to consider, and AM should not be overlooked as a final manufacturing option. Using AM as a finished manufacturing option is possible with careful consideration of the options available. Manufacturing parts in one’s prototype shop may not be possible, but there are companies that manufacture AM parts as a business, and they should be interviewed and consulted. Cost analysis for AM finished goods is different than more common manufacturing

Utilization of parts made from various resins

Material

Accura25(a) Accura Xtreme White 200(a) Accura Clear View(a) Accura PEAK(a) FLGPCL04(b) FLTOTL05(c) FLDUCL02(c) Carbon Resin RPU 70(d) Carbon Resin EPX 82(e) Carbon Resin FPU 50(f)

Part Properties and Common Defects

postprocessing and be unacceptable from a usability or quality standpoint. Fragile parts may need a custom form or pallet during cleaning and curing to maintain their shape and dimensional quality. Some part-washing processes can accommodate the complete build platform, supports, and parts during postprocessing. The parts and supports can be left on the elevator platform. This is an efficient method to keep the part secure during cleaning and curing, although resin from the supports and platform will be dissolved into the cleaning solution along with resin on the part. The VP parts have a wide range of surface accuracy dependent on the layer thickness, image accuracy, and processing. Parts built with thicker layers (i.e., 0.1 mm, or 0.004 in.) will have layer lines and stair stepping on curved or sloping surfaces. These parts can be used as-is if possible, for example, for a prototype, or a secondary process of abrasive finishing can be employed. The material and manufacturing process should be considered during the design phase, and secondary finishing should be employed as needed.

ABS Flex White(g) E-Clear Series(formerly E-Shell 450)(h) E-Poxy(i)

Utilization

General-purpose models, snap-fit assemblies, and room-temperature vulcanized (RTV) master patterns. Durability and accuracy are reported as “better.” General-purpose models, RTV master patterns, and “best” for snap-fit assemblies. Accuracy is reported as “good,” with durability and moisture resistance reported as “better.” General-purpose models, snap-fit assemblies, and RTV master patterns. Moisture resistance and optical clarity are reported as “best,” with durability reported as “better.” Automotive “under the hood” and wind tunnel testing. Reported as “better” for accuracy, high temperature, and moisture resistance General-purpose, strong, precise concept models. Clear resin that can be used for optics, lighting, and fluidics applications. No postcuring is needed. Sturdy prototypes, interference- and press-fits, assemblies. Reported as high impact strength and compliant Consumer packaging, bushings/bearings, snap-fits, and living hinges. High impact strength with low modulus and high elongation make it highly resistant to deformation. Parts requiring strength, toughness, and moderate heat resistance Automotive and industrial prototypes and consumer applications. Toughness, stiffness, and temperature resistance Parts requiring repetitive stress, such as living hinges or friction-fits. Impact-, abrasion-, and fatigue-resistant semirigid material Flexible, acrylonitrile-butadiene-styrene-like three-dimensional printing material. Ideal for applications including snap-fit assemblies. Capable of high-speed builds Strong, tough, water-resistant parts. Remains clear without yellowing with age. Ideal for water-resistant, high-humidity applications Partially biosourced, tough, dual-cure material that delivers strong, thin-walled final products

Note: Utilization of parts was researched from literature from the resin supplier. (a) Source: Ref 3. (b) Source: Ref 18. (c) Source: Ref 19. (d) Source: Ref 5. (e) Source: Ref 6. (f) Source: Ref 7. (g) Source: Ref 8. (h) Source: Ref 9. (i) Source: Ref 10

Vat Polymerization / 39 techniques, such as machining and molding. Additive-manufactured parts have built-in opportunities that make comprehensive cost analysis imperative. Cost analysis developed for traditional processes, such as molding and machining, will not apply to AM and will give a skewed result. For example, AM can produce single components that are an assembly that would otherwise need to be created as separate parts. Design for AM can be used in eliminating the need for features, time, and errors associated with assembly processes. Other opportunities are lead time, customization, and design freedom. With AM, the lead time to produce finished goods can be pushed very close to the launch date, as compared to molding. The design, material, and manufacturing process can be tested and validated prior to finalizing the design. Design changes that do not affect validation can be made close to launch, if planned correctly. If applicable, design validation can be accomplished with AM parts rather than molded parts. Another benefit of finished goods manufacturing with AM is customization. Customization can become a feature of the product, for example, a scanned photograph that becomes a finished plaque. The plaque surface, base, and the technique to create the image are always the same, but the photo image can change. Another example is a manufacturing process designed for several versions of the same component. Additive-manufactured parts can be used as fixturing for different versions or as finished custom components. Additive manufacturing offers design freedoms that are not available when processing parts via injection molding or machining. The obvious design opportunity is the ability to create features that are not in the line of site of a tool or parting line. Other design opportunities include closed volumes, lattice

structures, and variable wall thicknesses. When the opportunities of AM are combined with the variety of materials and available manufacturing platforms, AM can be considered a viable manufacturing option. For prototyping, the variability of materials and processes is such that a wide variety of components can be prototyped via AM. The selection of materials and processes makes it imperative to research companies and suppliers to supplement the capabilities of a prototype shop. Finishing techniques, materials, and processes vary, and AM prototypes can be used for more than firstrun samples. Utilization includes patterns for casting, manufacturing tooling, finished models and displays, and engineering test articles. If the utilization and output of one’s machine is complex and the geometry has wide variability, then outside vendors should be consulted to supplement the operation. One should grow the operation as needed.

REFERENCES 1. “Right-Side-Up SLA,” Formlabs, https:// archive-media.formlabs.com/upload/rightside-up-sla.png 2. “Digital Light Processing (DLP),” Mechanical, Electrical, and Electronic Engineering Services, https://meee-services.com/wpcontent/uploads/2018/06/Digital-LightProcessingDLP.jpg 3. “3D Systems Material Selection Guide for Stereolithography,” 3D Systems, Inc., 2018 4. “Photopolymer Resin for Form 1+ and Form 2: Material Data Sheet,” Formlabs, prepared Sept 18, 2018; revision 06.01.22.2019 5. “Document 103213, Revision D,” Technical Data Sheet, Carbon 3D 6. “Document 107172, Revision B,” Technical Data Sheet, Carbon 3D

7. “Document 103215, Revision C,” Technical Data Sheet, Carbon 3D 8. “2018-ABS-Flex-White.pdf,” Material Data Sheet, EnvisionTEC, https://envisiontec. com/3d-printing-materials/ 9. “2018-E-Clear-.pdf,” Material Data Sheet, EnvisionTEC, https://envisiontec.com/3dprinting-materials/ 10. “2018-E-Poxy-.pdf,” Material Data Sheet, EnvisionTEC, https://envisiontec.com/3dprinting-materials/ 11. “09 Projet6000/7000PrinterOpsCourse 3D Printing Theory 20170901.pptx,” Training Manual, 3D Systems Corporation 12. “Hardware and Material Safety 20170901. pptx,” Training Manual, 3D Systems 13. “Clear Photoreactive Resin for Formlabs 3D Printers: Safety Data Sheet,” Formlabs, Oct 15, 2016, https://archive-media. formlabs.com/upload/Clear__Resin_SDS_ EU.pdf 14. “Durable: Safety Data Sheet,” Formlabs, Jan 15, 2017, https://archive-media.formlabs. com/upload/Durable_Resin_SDS_EU.pdf 15. “Tough Photoreactive Resin for Form 1+, Form 2: Safety Data Sheet,” Formlabs, Sept 26, 2016, https://archive-media.formlabs.com/upload/Tough-SDS.pdf 16. “DSM Safe Handling Guide for UV Curable Materials,” DSM Functional Materials, Inc., effective April 1, 2011 17. “08 Projet6000/7000PrinterOpsCourse mpire Supports 20170901.pptx,” Training Manual, 3D Systems Corporation 18. “Material,” Formlabs, https://formlabs. com/store/form-2/materials/ 19. “3D Printing Materials for Engineering, Manufacturing, and Product Design,” Formlabs, https://formlabs.com/materials/ engineering/

Copyright # 2020 ASM InternationalW All rights reserved www.asminternational.org

ASM Handbook, Volume 24, Additive Manufacturing Processes D. Bourell, W. Frazier, H. Kuhn, M. Seifi, editors DOI 10.31399/asm.hb.v24.a0006580

Material Extrusion Additive Manufacturing Systems David A. Prawel, Colorado State University

MATERIAL EXTRUSION SYSTEMS (Ref 1) are the most common types of additive manufacturing systems, also known as threedimensional (3D) printers. These systems are so named because they are based on a very common and well-understood fabrication process and technology called extrusion. The extrusion process is based on pushing material through a shaped hole called a die, which imparts its shape on the cross section of the material that is being extruded. The material is moved through the shaped hole by a process called ram extrusion, which pushes the material through the hole with a flat device that applies direct force to the material (Fig. 1) or uses gas (air) pressure, or by helical extrusion, where pelletized material is moved through the hole by a worm gear mechanism (Fig. 2). There are hot extrusion processes, cold extrusion processes, and chemical-based processes, each enabling a broad range of materials that can be extruded at various levels of accuracy and speed. As the names imply, hot extrusion uses thermal energy to soften the material so that it can extruded. As the material softens, its viscosity decreases, enabling it to be pushed through the die at a given level of accuracy and speed. Cold extrusion does not use heat and therefore depends more on other mechanisms to enable forcing the material through the die. Chemical extrusion uses some chemical reaction to change the viscosity of

the material, thereby enabling it to be pushed through the die. As material is extruded, the as-printed row or track that is created is referred to as a “road.” Extrusion can be a very complicated process when one considers the huge variety of material types, the complexity of the desired cross sections, and the desired quality and quantity of the resulting objects. Bulk material properties such as ductility, viscosity, and crystallinity can have a profound impact on extrusion processes, which is discussed later. Similarly, the printing process itself is comprised of a delicately tuned choreography of many printing and environmental factors, such as head speed, material flow rate, temperature, humidity, and so on. Therefore, each extrusion method has associated with it a set of advantages and disadvantages that may dictate specific material/process/quality/quantity combinations. When this article refers to quality, it means the fidelity of a printed part to the intended object, as defined by the original (digital) computer-aided design (CAD) file and other properties of the object, such as surface texture. Extrusion printers are generally categorized as industrial printers selling for more than $5000 and desktop printers selling for less than $5000 (Ref 4). This article focuses on the general 3D printing processes as can be demonstrated and manipulated in desktop printers. Industrial printers package the product so as

to produce the best parts they can produce in a given material at a given set of printing parameters. To accomplish this, many months, or even years, of testing have gone into determining the best set of material/parameters combinations to control the huge amounts of variability that make 3D printing the complex, challenging process it is. Essentially, industrial printers hide the printing process from the user in order to deliver the highest possible part quality at the lowest possible cost (time, material, etc.).

Melt Extrusion 3D Printing Melt extrusion systems are the most common types of 3D printers (Ref 4). These are the devices found in libraries, schools, and maker labs around the world. This process was first commercialized in 1991 by Stratasys, who referred to their melt extrusion system as Fused Deposition Modeling, or FDM, an acronym which has become synonymous with melt extrusion 3D printing and is a trademark of Stratasys. As the name suggests, these devices rely on a hot extrusion process to heat a material to sufficiently soften it to enable extrusion. The most common materials used in melt extrusion 3D printing are amorphous plastics, which are discussed in detail later. In melt extrusion, a plastic is softened with heat and pushed through a circular die. As with

Hopper Plastic pellets Heaters

Polymer melt

Breaker plato

Screw Barrel

Die Extrudate

v

Feed section

Fig. 1

Ram extruder. Source: Ref 2

Fig. 2

Helical extrusion. Source: Ref 3

Compression section

Meeting section

Material Extrusion Additive Manufacturing Systems / 41 extrusion in general fabrication, there are many ways to extrude material in 3D printing. The most common method of melt extrusion uses a ram extrusion mechanism to feed a plastic filament from a spool into a print head. There it is heated and softened, then pushed through a die and onto a build platform. The fundamental process is controlled by managing the phase changes and associated properties of the materials in a precise fashion. To understand the melt extrusion process, it is useful to think of a hot glue gun, a device almost everyone has some experience with. A hot glue gun heats a plastic rod of material. When the material is warm enough, a user can apply pressure, usually using a thumb (the ram), which causes softened material to exit the heated tip of the glue gun. If the material gets too hot and/or too much pressure is applied, more plastic will be extruded, perhaps too much. If the material is not hot enough and/or less pressure is applied, less material extrudes from the gun, perhaps too little. If a user is actively extruding plastic and the gun is moved too quickly, a thinner bead of material will extrude. If the gun is moved more slowly, the bead of extruded material will be wider. In this way, the best balance of heat, pressure, and motion will result in the best control of the softened plastic and the best end result. Melt extrusion 3D printing operates in much the same way, with precise control of these three main physical processes. More precise devices provide better control of these processes and produce objects that better match the digital representation, usually a CAD model, more accurately and usually cost more. In melt extrusion 3D printing, the bead of material is called a road.

Components of Melt Extrusion 3D Printers Melt Extrusion Print Heads The print head is one of the most important components of a 3D printer. It controls the deposition of material onto the build platform. The quality of the printed object depends in large part on the precision of this materialdeposition process, coordinated with the motion-control system. A critical component of a melt extrusion 3D printer is the extruder. It provides control of the feed rate and temperature of the material, which controls the state of the material, fundamentally its viscosity. Melt extrusion is actually a misnomer. The material is not actually heated to a melted state. It is heated above the glass transition temperature (Tg), at which point the material begins to soften, but kept below the melting temperature (Tm), at which point the material is liquefied. One cannot print at or above the Tm because the shape of the material cannot be controlled in a melted state, and one cannot print below the Tg because the material must be soft enough to extrude it through the nozzle.

Industrial extrusion printers deploy proprietary print heads, while desktop printers use open source or proprietary heads. In the end, the components that produce the 3D-printed part are the same. Referring to the earlier hot glue gun analogy, the harder one pushes the glue stick (i.e., higher feed rate), the more material moves through the print head, and vice versa. In a 3D printer, the feed rate is precisely balanced with material state and motion control to deliver a desired amount of material deposited in the desired state on the build platform. The extruder is capable of starting, stopping, and reversing the material feed to enable controlled deposition of material precisely where it is desired and not in places it is not desired. Different materials and associated material properties affect the rate that the printer is able to advance and retract material, thereby affecting the quality and rate at which a part is printed. This is discussed in detail later. Ram extrusion is the most common print head configuration found in melt extrusion 3D printing. A representative diagram of a ram extrusion print head is shown in Fig. 3. Other configurations, such as helical drive, are discussed subsequently. Numerous methods exist for ram extrusion, but in general, a filament-shaped material is gripped between two gears that are turned by a stepper motor to advance and retract the material by pushing and pulling it (thus “ram” extrusion). A shear force is created between the gear faces and the filament surface. The accuracy of this filament feed process, in both directions, is dependent on the pressure the gears place on the filament and the friction of the interface between the gears and the filament. Various strategies are employed to minimize the risk of slippage of this gear/filament interface. For example, the gear faces are sometimes knurled

Fig. 3

Diagram of a ram extrusion print head. Source: Ref 5

to provide a higher-friction interface. In the least expensive printers, the gear faces are simply flat, and slippage risk is higher. Material is pushed into and through a heated region of the head, to soften and compress it, and out to the build platform through a nozzle with an orifice. This portion of the head is called the hot end, for obvious reasons. There are many types of nozzles for many purposes, such as for higher-temperature materials, different-diameter materials, composite materials, and so on. The hot end is equipped with a thermocouple to control the temperature of the hot end, thereby controlling the viscosity of the material. There is a material feed system where the ram pressure, the pressure of forward pressure, relates to the filament diameter that is created. Also, there are shear forces involved. The shear force is highest outside the tube. In essence, it forms a wave front, where the shear is really high on the outside and much lower on the inside. The nozzle angle is pretty much standard on 3D printers. The primary driving forces that define the quality of the printed part, which includes both its aesthetics and its strength/durability, are convective heat transfer between the newly deposited material and the already deposited material (or the build platform), and conductive heat transfer between the material and the environment. This is discussed in detail in the section “Solidification” in this article. To achieve the desired end result, the circuitry and control systems on the printer control a delicate balance between the material feed rate and the temperature in the print head, the head speed relative to the build platform (see the section “Motion Control Systems” in this article), and the bed/chamber temperature (if either/both are heated). On-Head versus Off-Head Extruders As discussed previously, the extruder in typical ram extrusion systems contains the heating and feedback components and the ram extrusion components. This results in a significant amount of mass. As printers and build platforms become larger, higher mass becomes a design constraint, because it affects the speed and accuracy at which the print head can move and deposit material. Other print head configurations have emerged to address this challenge. For example, a Bowden design (Fig. 4) places the ram extrusion mechanism off-head, separate from the heating/feedback mechanism. The hot end alone has much lower mass and consumes less space, enabling, for example, multiple heads and larger printer build platforms. Many other configurations also exist. Cooling Fans The thermal conductivity of materials makes them cool more or less quickly than other materials. In these cases, it is common to mount a small fan on the print head to pass air directly at the material as it leaves the

42 / Polymer Additive Manufacturing Processes nozzle. This results in a faster relative cooling process, enabling the material to acquire the appropriate temperature to maintain the desired shape while achieving a good bond with prior roads.

Single/Multiple Material Extrusion It is common to 3D print multiple materials simultaneously, for example, creating objects of different colors, either as separate parts, each with specific colors, or as single objects with some predetermined color gradation within the object. It is also common to 3D print discrete support structures in a different material than the main object. For example, some materials will dissolve in water, presenting a potentially valuable method to reduce the time/cost/effort of removing support materials (discussed later). Many melt extrusion 3D printers are, or can be, configured to extrude multiple materials concurrently, for example. Some extrude multiple materials through individual heads for each material, such as the dual-material extruder shown schematically in Fig. 5, while others mix the warm material in the hot end and extrude the combined material through a single extruder nozzle. Motion-Control Systems Every 3D printer provides some mechanism for controlling the motion of the object being created (on the build platform) relative to the print head. Many configurations exist to create 3D dimensions of motion, each with advantages and disadvantages and personal preferences. The most common configuration, referred to as a Cartesian system, is to move the print head in two dimensions over a build platform that moves in one dimension. Some 3D printers provide additional degrees of freedom in the build platform, allowing it to tilt

Fig. 4

Bowden design places the ram extrusion mechanism off-head, separate from the heating/ feedback mechanism. Source: Ref 5

Fig. 5

Schematic of a dual-material extruder. Source: Ref 6

and move, which enables creation of more types of objects with less support material but adds complexity and maintenance cost. A delta system deploys a completely different configuration, with three independent-motion arms to position and tilt the print head as desired, as shown in Fig. 6. Motion-control systems include controller circuitry, data input interface, helical threaded rods and belts turned by stepper motors, build platforms, and a variety of frame configurations. Hopper Feeders Hopper feeders are sometimes deployed on melt extrusion 3D printers to enable a much larger variety of material to be printed at much lower cost. Countless plastics are available in pelletized form, which can be poured into the hopper and printed instead of being reformed as filament. The hopper grinds and heats the pellets, and the soft plastic passes through the extruder as with filament material. Pelletized plastics are available at much lower cost than filament material. Hopper feeders also enable custom mixtures of different plastics or particulate additives such as ceramics or powdered metals, to create custom formulations of materials to melt extrude. The grinding and mixing processes are quite complex, and quality 3D printing requires consistency, so these systems require significant expertise and patience. Three-dimensional

Fig. 6

SeeMeCNC Rostock MAX v3.2 three-dimensional printer

Material Extrusion Additive Manufacturing Systems / 43 printing is itself a complex process, and anything that makes it more so is to be considered carefully. Hopper feeders are also used in conjunction with helical extrusion methods, as discussed in the following section, when extrusion of high volumes of material is required. Another advantage of hopper feed systems is that many more materials are available in pelletized form than as filament. Countless combinations of plastics and composites of micro- and nanoparticles in the plastic matrix are also available, for example, metal, ceramic, and wood particles. Helical Extruders Helical extrusion systems provide an alternative feed mechanism sometimes found in melt extrusion printing. Rather than using a spool of filament, small pellets of material are fed into a hopper that funnels the pellets into the heated print head, where they are softened and pushed out the orifice, as in ram extrusion systems. The mechanism of the print head, such as the heating element and temperature-management systems, are the same as in ram extrusion mechanisms. Helical extrusion systems have been adapted for use in 3D printing, but the mechanism and process are relatively unchanged from that which has been in use for centuries. The first main component of the system is the hopper (Fig. 2), which receives the (usually) pelletized plastic material. Sophisticated mechanisms are often deployed in the hopper, such as air jets, to assure adequate mixing of the pellets as they enter the hopper and to reduce the risk of packing of pellets, which restricts the flow of pellets into the helical extrusion chamber. As the pellets fall into the extrusion chamber, they enter a space between the screw and the heated outer wall of the cylindrical extrusion chamber. Helical bands called flights are attached to the main extruder axle to form the screw, which turns to propel the material forward into and through the heated region of the chamber. The angle of attack of the flights (the pitch) and the rotation speed of the axle are calculated to optimize the speed and flow rate for the type of material being extruded. As the material is propelled forward toward the orifice, a continuous, highly controlled process of heating and compression of the material proceeds as the now-warm material is densified as it continuously moves forward, propelled by the flights. Heating elements surround the chamber to continuously and gradually heat and compress the material. As the material heats and compresses, pressure-relief mechanisms release expelled gases (outgassing) to reduce pressure and the risk of trapping bubbles in the softened plastic. The attack angle, the rotation rate of the axle, and the heating rate control the rate of progression and densification of the material, which is different for each material. When the material is sufficiently heated and densified to the desired viscosity, it approaches

the end of the extrusion chamber, which is enclosed by a die. The die defines the crosssectional shape of the extrusion. The complexity of this shape depends on the ability of the particular material to be forced through the die, each material being different. The angle of the inner face of the die as it approaches the orifice also plays a role in the additional compression, pressure requirements, and ultimate quality of the extruded parts. The advantages of helical extruders with hopper feed systems are high volume and low cost. These benefits drive the use of hopper feed and helical extruders in both melt and viscous extrusion 3D printing. These systems are common in high-volume production, for example, of plastic medical devices such as syringes and the first 3D-printed car (Fig. 7), produced by Local Motors. Hopper-fed helical extrusion is the best method to deliver sufficient material at a fast enough rate to produce the strength required of a road-worthy, motorized vehicle in a cost-effective manner. Hopper feed/helical extrusion was also used by Oak Ridge National Laboratory to produce its minibus (Fig. 8) and excavator (Fig. 9).

The Process of Melt Extrusion This section covers the melt extrusion solidification (during cooling) process, the underlying mechanism of road bonding, and the factors affecting good part quality. Process Overview In 3D printing, as in any fabrication process, reproducibility and repeatability of the outcome is a critical determinant of part quality at a given quantity and consequently whether a product can be produced and used/sold costeffectively. Part quality depends fundamentally

Fig. 7

on constant, predicable road dimensions and as-printed layer adhesion. In melt extrusion, road dimension and layer adhesion depend primarily on flow rate at a given head speed (motion of the print head relative to the build platform), material feed rate, thermal interactions (road cooling and bonding), and environmental conditions (e.g., ambient temperature and humidity). Road cooling and bonding are thermal conduction between adjacent materials and convective heat transfer between the asprinted material and the immediately surrounding environment. In viscous extrusion, road dimension and layer adhesion depend on flow rate, viscosity of material, length and radius of the extrusion orifice, and many other factors. Because the variety of solidification processes in viscous extrusion is so varied and because viscous extrusion is much less common, viscous extrusion solidification is not covered in this article. Material feed rate is determined by the extrusion process (usually ram) discussed earlier. This is an important parameter of the printing process. Its accuracy is a key determinant of part quality, because it defines how much material is extruded at a given rate and temperature. The ram pressure (at a given filament diameter) counters the shear forces generated by plastic flow along the inner faces of the heated nozzle and the back force pressure caused by the restriction of the nozzle orifice. Ultimately, these forces are overcome by enough ram pressure, and the material flows outward from the extruder toward the destination. Material flow is a viscous flow mechanism wherein mass flow depends on the pressure drop resulting from nozzle geometry and the viscosity of the material. In melt extrusion, viscosity in flow is controlled by nozzle temperature. Viscosity at the time the material is

The first three-dimensional-printed car, produced by Local Motors. Source: Ref 7

44 / Polymer Additive Manufacturing Processes deposited at the destination depends on environmental conditions. After the softened material leaves the print head, it immediately begins a complex cooling process, leading to solidification of the material. A variety of material characteristics come into play to determine how quickly the material cools, to result in a desired solid object on the build platform. These material characteristics are discussed later. In general terms, if the material solidifies too quickly (i.e., before it lands on the build platform), too little material will be deposited on the build platform, and the newly deposited material will

not bond properly to the prior material already extruded onto the build platform or to the build platform itself (if it is the first layers extruded). Conversely, if the material solidifies too slowly, too much material will be deposited on the build platform, and it will not assume the desired shape of the object being printed. Consider again the hot glue gun analogy. If the material is too hot, it cannot be controlled; too much material is extruded at even the slightest pressure, making accurate placement of the material almost impossible. One counters this by making adjustments, such as

reducing the pressure on the material, moving the gun more quickly, and so on. Conversely, if the material is too cool, too little material is extruded, and the opposite corrective actions are required. Similar adjustments are made in the melt extrusion process to produce accurate representations of the desired (CAD) object on the build platform. Accurate melt extrusion requires a delicate balance between the many factors that dictate the temperature of the material as it lands on the destination (prior material or build platform). The material must be soft enough to form a good-quality bond with the prior material (or the platform itself) when it hits the destination, and ideally, the prior material has not completely cooled at the same time. However, the material must not be so soft that it will not maintain shape upon landing at the destination. At the “best” temperature, the material will form the “best” bond with the prior material or the platform. This ideal temperature is always between its Tg, the temperature at which the material starts to soften, and its Tm, the temperature at which it liquefies. The two main factors that dictate the quality of the bond of new material to existing material are the convection of heat from the as-printed object to the environment surrounding it, and the conduction of heat within the as-printed material at the destination. These factors depend on material properties and environmental conditions. Bonding to prior material is different than bonding to the platform (discussed later), because these materials are very different. However, successful shape management in both cases depends on printing parameters, environmental conditions, and material properties. Because the best temperature to print a material lies between its Tg and Tm, amorphous or semicrystalline plastics are best for successful melt extrusion printing. A broad range

Fig. 8

Minibus from Oak Ridge National Laboratory. Courtesy of D.A. Prawel

Fig. 9

First three-dimensional-printed excavator, built by Oak Ridge National Laboratory (ORNL). ABS, acrylonitrile-butadiene-styrene. Source: Ref 8

Material Extrusion Additive Manufacturing Systems / 45 between Tg and Tm as well as distinct material state transitions at and around Tg dictate the ease with which a particular material will successfully print objects that are truest to their original digital (CAD) representation. Generally, amorphous materials are easiest to 3D print with melt extrusion. Some semicrystalline materials are also relatively easy to print, depending on the amount of crystallinity and other material properties. The higher the crystallinity, the more difficult a given material is to print. Material properties and their effect on printability and print quality are discussed in more detail later. The two most important physical processes that affect the quality of as-printed parts in melt extrusion are the solidification dynamics of the material as it cools and the bonding of printed roads on the build platform. As the material is extruded from the print head (hot end), it cools and solidifies. Convective and conductive heat transfer defines the shape of the extruded road, the strength of the bond between roads, and ultimately the quality of the as-printed object. The bonding process between roads is a sintering process wherein the roads of material coalesce without fully melting. This sintering process is a viscous flow mechanism. Surface energy/tension is minimized, driven by thermal energy, and the surfaces of each road go through a process called necking. Ultimately, when the sintering process is complete, an entropy-driven intermixing of molecular contents occurs between the two roads, creating a homogeneous (ideally) end product. These processes are discussed in more detail subsequently. Many parameters affect the way different plastics interact and ultimately the bond quality between roads and the shape of the asprinted road. Operational, machine, and material specifications as well as part geometry affect the success of the melt extrusion process. Controlling as many of these parameters as possible for repeatability will ultimately be the best strategy to produce consistently highquality parts. Consideration of each parameter would exceed the size limits of this article, so it is therefore left to the reader. Readers are referred to the “CURA User Manual” (Ref 9) for a good overview of the parameters that can be controlled in replicating rapid (RepRap) prototype devices. Commercial 3D printers usually have proprietary software that prevents manipulation of most of these parameters. This comes in exchange for more consistency in print quality from these printers, because the vendor has tested and standardized most of the parameters. For example, consider that just the nozzle geometry, the shape, the attack angle, and the profile of the orifice all directly impact critical parameters such as the extrusion pressure (amount of material extruded) and the crosssectional shape of the material as it leaves the die, which in turn affect the bonding quality and as-printed surface smoothness

(appearance). Material properties also play a large role. For example, the coefficient of friction affects the shear forces that must be overcome as the material passes through the heated region of the print head, and material behaviors such as shear thinning and viscosity directly affect the extrusion pressure. Solidification The extruded material cools and solidifies as it lands at the destination (onto the print bed or onto prior-deposited material). Continuously controlling the repeatability of this process is the key to success in melt extrusion. As mentioned earlier, convection and conduction of heat are the dominant thermodynamic factors affecting the solidification process and therefore part quality. The temperature difference between newly deposited material and its destination, both the environment and already-printed material, directly affects the accuracy and predictability of road dimensions, which, in turn, directly affect the as-printed part quality. Greater temperature difference between newly deposited material and the environment lead to faster cooling, which leads to distortion of the roads. This can be remedied by using strategies such as heating the build chamber and/or the build platform or by using an enclosure around the build area to minimize fluctuations in the ambient environmental temperature. Greater temperature difference between newly deposited material and already-printed material leads to inconsistent bonding between roads, which results in anisotropic structural properties, weak bonds and parts, and poor print quality. As a road is printed, the temperature is highest as the material leaves the print nozzle and immediately begins cooling and thereby solidifying. The rate of cooling and solidification depends both on the environmental conditions and the physical parameters of the build process. A variety of material properties come into play in determining the ideal printing and environmental characteristics. The most important are Tg, mentioned earlier, and thermal diffusivity, which defines how quickly (or slowly) a material convects (loses or gains) heat energy. A material with high thermal diffusivity will cool more quickly than one with low thermal diffusivity. In this case, one may want to use an enclosure, for example, to maintain higher ambient temperature on the process, to achieve a more consistent cooling rate. Because some portions of the as-printed object are denser than others, depending on the design being printed, this cooling rate will vary also by part density. Similarly, one would want to cool a semicrystalline material more slowly to achieve a stronger part, enabling the semicrystalline regions on the polymer to arrange ideally as the material cools. Cooling happens at a faster rate on the outer surfaces of the road, slowing nonlinearly toward the center of the road cross section.

This factor further complicates the modeling of the physical process and makes the consistent production of high part quality very challenging. To alleviate the impact of these complexities, 3D printing researchers have developed elaborate scan paths, which are designed to optimize the thermodynamics of the heating/ cooling processes within a particular part shape. Scan paths were originally developed in photopolymer printing but provide an equally valuable mechanism in melt extrusion. Scan paths are discussed in more detail subsequently. As the roads are printed and begin to cool, they solidify on the build platform or on prior-printed roads. The ultimate quality of the as-printed object depends on the quality of the bond that forms between as-printed and newly printed roads. Road Bonding In melt extrusion, bonding of roads is a sintering process. Surface energy is minimized through thermodynamic processes involving necking and coalescence, as in other sintering processes. Thermal energy overcomes the surface tension of each road, leading to formation of an opening between the roads referred to as a neck, which increases due to the natural tendency toward minimal surface energy. The neck enlarges, enabling an entropic molecular intermixing between the two roads, and the bond forms between the roads as the material cools and solidifies the material. Given an ideal balance of time (cooling) and temperature, an equilibrium state would be reached wherein the roads are fully bonded at the molecular level due to a homogeneous intermixing of the material between the roads. Typically a road is printed over the top of another road, so gravity also plays a role in bringing the roads together. The shape and quality of the bond is temperature dependent, because warmer material will sinter and self-level more than cooler material. These processes are very complex and difficult to reproduce consistently, which is the reason melt extrusion 3D printing is highly prone to variability and requires substantial trial and error to perform repeatably with high quality. Theoretical models have been developed in an attempt to model these processes and enhance repeatability, but they assume that the materials are Newtonian and isotropic, both of which are rarely, if ever, true. The physical processes are so complex that these assumptions are necessary to get the models to converge with reasonable outcomes. Unfortunately, the models are not sufficiently robust to use in actual part production. The more physical parameters that can be locked down and standardized, the more likely a reproducible family of parts can be produced. One pays more for a melt extrusion that consistently produces high-quality parts, because a substantial amount of research and development has gone

46 / Polymer Additive Manufacturing Processes into the process of standardizing these parameters and more. In the end, the goal is to repeatably produce high-quality parts. Quality includes strength, surface texture, and visual appeal. To do so, one must understand and manage conductive heat exchange between an as-printed road and a prior road, and convective heat exchange with the surrounding environment. Repeatably high-quality printed parts will result from precise calibration of the device and precise control of environmental conditions and printing parameters. Environmental Conditions Understanding the main environmental factors and their impact on as-printed parts is key to successfully producing melt-extruded objects. This section discusses only the most important: consistent build chamber temperature, oxidation, and humidity. As discussed previously, solidification and road-bonding processes are very complex and prone to high variability. Temperature plays a fundamental role in all these processes. Management of consistent temperature is probably the most important factor in successful, highquality melt extrusion. This includes precise temperature control on the build platform and the ambient temperature of the build chamber. A heated print bed and an enclosure for the build chamber are important, perhaps essential, to printing a wide variety of materials consistently, in high quality. It is ideal if printing can be done in a temperature-controlled build chamber that allows manipulation of the ambient temperature throughout the build process, not just on the build platform. Most polymer filaments will oxidize, which results in poor bonding between roads. Oxidation usually results from storage in exposure to air. Plastic filaments are porous and can also absorb water. Humidity affects printing processes through two mechanisms: by affecting the properties and behavior of the polymer filament and by affecting road bonding. Many 3D printing polymers are water sensitive. Some, such as polylactic acid, actually break down slowly in a humid environment. For these reasons, 3D printing filament is sold in vacuum-sealed packaging. If filament is left out for extended periods of time, it is advisable to vacuum dry it in low heat for at least a few hours. The highest possible vacuum pressure is preferable, but heat should not exceed approximately 60  C (140  F). The reader is advised to consult the literature for recommendations for specific materials.

Support Material and Postprocessing in Melt Extrusion After the thermoplastic is extruded, it remains soft for some period of time while it cools. This period depends primarily on the properties of the material and the ambient temperature of the surrounding environment, as

discussed earlier. During this time, if the warm material is not supported in some way, usually by some other material, it will deform (sag) due to gravity. The amount of this deformation depends on the properties of the material and the ambient temperature. To maintain the desired geometric shape during this period, support material is used. This is additional printed material that is deposited for the express purpose of supporting various portions, or all, of the object being printed. A variety of approaches are used to provide support. The most common is to use the same material as is being printed. This is referred to as primary support. The material that comprises the desired object being printed is the primary material. Most printers have only a single print head, requiring that part geometry and support structure be printed simultaneously in any given layer. The object and the support are printed at the same time in the same material. Because support material must ultimately be removed, this often becomes a major source of cost in creating a final object using melt extrusion 3D printing. Cost refers to both person time and wasted material. Moreover, some geometries require deep pockets, channels, or internal cavities that cannot be reached to enable adequate removal of primary support materials. Therefore, various strategies are employed to reduce the amount of support required or to ease support removal in regions that are difficult to access. Because support removal can be such an expensive process, it is especially important to consider it in the original design of the object being printed. This design forethought is referred to as design for additive manufacturing and is essential to the effective use of additive manufacturing. In additive, removal of support materials must be considered as part of the design process. The purpose of the object being designed, the scale, surface roughness, how it will be produced, and other important factors should be considered before an object is printed. Support structure is a necessary evil in 3D printing, and a good design of any additive-manufactured object carefully considers removal of supports. The finish and roughness of a surface may be part of the requirement specification, so these characteristics must also be considered. Removal of supports often leaves behind remnants where the supports were attached, which result in surface roughness. This becomes particularly important as scale decreases. For example, the surface roughness of the walls of small internal channels in a printed object may affect the rheometry, which plays a critical role in microfluidic and heat-exchange devices. These types of devices can be rendered unusable if the support material cannot be completely removed. Careful planning and designing expressly for additive manufacturing are key to success. The most common method of easing support removal is to print a secondary support

material that has different properties than the primary material, ideally making it easier to separate from the object being printed. This method requires the use of a second print head. The most popular secondary support material for this method is polyvinyl alcohol (PVA), a water-soluble thermoplastic that is easy to print. Thus, supports can be dissolved off the part by using a simple water bath. As an added benefit, complex parts with deep pockets and overhangs can be printed using PVA as a support material, then soaked in water to dissolve it from these difficult-to-reach areas. Secondary support materials can also be softer than the primary material or sensitive to specific chemicals that will not corrode or erode the primary material. Another method of reducing the cost of support structure removal is to change printing parameters to affect the bonding of the support material with the desired object in the very small areas where support material meets the object being printed. The most common parameter to modify is to increase the print head speed in a very small, highly localized area. This will decrease road-bonding potential, as discussed earlier, only in this small area, making it easier to remove the supports. Other parameters, such as varying the head temperature or issuing a brief pulse of cool air from a fan, are not effective due to poor response time. Thermal inertia slows the temperature change to the extent that total print time becomes prohibitively long. A third approach to reduce the cost of support removal is to use a material with different material properties that reduce the amount of sag over a given span distance between supports. A material with a higher thermal diffusivity will sag less between supports than a material with a lower thermal diffusivity, because it cools faster. Printing materials with higher thermal diffusivity will enable roads to span greater distances while maintaining shape. Less support material is required because the material can span greater distances between supports, which reduces the number of supports required in a given area, thereby reducing the cost of support removal. Another method of easing support material removal is to use a different material at the interface between the primary material of the object and the support material. This requires a second print head that prints the interface material in the space between where a support structure meets the desired part geometry. If the secondary material is more brittle than the primary material and the support material, it will fracture easily and usually be completely removable.

Additional Road-Quality Considerations In addition to heat transfer, physical printing parameters, and environmental

Material Extrusion Additive Manufacturing Systems / 47 effects, additional physical phenomena relating to the melt extrusion process and the layering of roads are often seen in the 3D printing process, both of which affect the as-printed part quality. As stated earlier, quality in this context refers to adherence to the original digital CAD geometry representation. Die Swelling and Ooze As material is extruded through the hot end and nozzle, it is compressed. When it leaves the nozzle, it expands, as shown in Fig. 10. This phenomenon, referred to as die swelling, results in variability of the as-printed road width. Die swelling can play a profound role in reducing the predictability of as-printed part size compared to the expected CAD geometry, ultimately affecting final part quality. Because it affects road diameter, die swelling can have a significant effect on feature quality in the X, Y plane, the plane of the build platform. Objects with features that have critical dimensions in the X, Y plane, such as holes, slots, and so on, will be more susceptible to die swelling. It can also affect the height of roads in the Z-dimension, creating a phenomenon referred to as stack-up error, which is discussed in more detail subsequently. The amount of die swelling that occurs, and therefore the actual precise diameter of an asprinted road, results from a balance between material properties, nozzle diameter, extrusion pressure, and the difference in thermal gradient between the as-printed road and the ambient temperature. Material properties such as thermal expansion coefficient and diffusivity affect the amount of die swelling that occurs. Environmental conditions, especially as they pertain to amplitude and variation between the extrusion temperature and the ambient temperature in the build volume, can also affect the extent of die swelling that occurs. It is always necessary to minimize both

in order to enhance as-printed part quality and consistency between parts. Nozzle diameter also plays an important role, balanced against the other factors mentioned. More material compressed in the hot end results in more material expanding as it leaves the nozzle. The pressure required to extrude a particular material also depends on its material properties, including those mentioned previously as well as the frictional coefficient. Higher extrusion pressure creates a pressure head that results in higher relative compression and expansion of the material as it enters and exits the nozzle. Therefore, one material will swell more or less than another, creating variability and unpredictability in as-printed part quality. All of these factors are affected by each particular environmental temperature gradient, further confounding predictability. As seen in Fig. 11, the distance the nozzle is physically positioned above the build platform affects the dimensional accuracy of an asprinted road. As hot material is exiting the nozzle, the actual as-printed width of the road is dictated by where the new material intersects an existing road. Very small variations in Z-height of the nozzle will affect the accuracy and consistency in road width. In addition to head/nozzle height, variations in height of the build platform from corner to corner will also directly affect the precise height of the print head/nozzle relative to the build platform. The greater the distance above the build platform, the narrower the as-printed road diameter, by very small amounts. This effect is mitigated by careful leveling of the build platform with each printing project. Some of the more expensive RepRap printers and all commercial printers do this automatically.

Fig. 11 Fig. 10

As material leaves the nozzle, it expands, which is referred to as die swelling. Source: Ref 10

The distance the nozzle is physically positioned above the build platform affects the dimensional accuracy of an as-printed road. (a) The road shape is distorted when the print head is above the deposited road. (b) The print head in contact with the road helps to ensure a flat top road surface. Source: Ref 10

Ooze is another important concern in asprinted part quality. When the print head moves along a path in a print and then changes course to, for example, turn a corner or reverse course to print parallel roads, material gathers as the print head slows or stops to change direction. When the head slows or stops, the material is still extruding, probably at the rate it was moving in midroad. Extra material gathers when the print head slows. Still more gathers when the head stops to reverse course. This excess material builds up, either on the inside and outside faces of the object in that location or as an increase in road height in the general area of head speed variation. This additional material creates a variation in layer width and height, significantly affecting the as-printed quality of the object being printed. Settings exist in most good slicing tools to slow the print head before and after changes in direction, but these are inaccurate and imperfect corrections. Variation and error are created, and the as-printed quality is diminished. Gravity also pays a variable role in part quality. As a road is deposited, it is hot and soft. The force of gravity causes a settling of the material that flattens the original circular extrusion profile. As the road flattens, yet still contains the same amount of material, the roads also spread. Depending on how quickly cooling occurs, the road flattens more or less. The amount and rate of flattening depends in part on the properties of the material, particularly thermal diffusivity. The faster a material loses its latent heat, the less likely it is to flatten. If material is printed more densely in one portion of an object than another, there is less convective heat loss to the surrounding environment. Conversely, if one part of an object is printed with less density than another, it loses heat more quickly. These variations in density result in poor quality and inconsistent dimensional conformation to the desired CAD geometry. For similar reasons, the path of the print head as it prints a given layer (i.e., the scan path) can also affect part quality. Scan paths are discussed in more detail subsequently. If a new road is immediately printed on top of an existing road, the existing layer is warmer than if time is allowed to pass before the new layer is printed. Printing smaller, denser stacks of roads creates shapes that retain heat longer than printing more open, porous shapes, due to the density of material that has been deposited. The smaller shapes would remain softer longer than the more open shapes, because they retain heat and would therefore be more prone to settling due to gravity and to compression due to the weight of subsequent layers. Many other factors control print quality, and each has an associated parameter setting in the software that controls the print setup process. A comprehensive listing of the various parameters that can be controlled in a tool called Slic3r provides a good reference (Ref 11).

48 / Polymer Additive Manufacturing Processes Slic3r is a popular software tool for slicing CAD models to create layers that ultimately are printed. Considering these complex factors, it can easily be seen why as-printed part quality in extrusion 3D printing is very difficult to predict at a high level of precision, making it difficult to create high-quality parts consistently. These confounders are countered by carefully controlling the printing parameters and leveling the bed as precisely as possible. The reader is again reminded of the economic aspects of these various adjustments. Low-cost 3D printers leave much of this fiddling to the user to figure out. More-expensive printers are packaged with parameters for each material that are very carefully tuned to the material and used through specially designed proprietary software, so that users obtain parts printed with maximum success and minimal hassle. Stack-Up Error and Nozzle Interactions As discussed earlier, there is a small amount of natural variation in the actual printed height of each layer. With each successive layer, this small error stacks up to become a larger error. Stack-up error refers to the very small variability in road height that occurs with each printed layer. Material properties such as thermal expansion coefficients and diffusivity play an important role in the amount of stack-up error that occurs. Die swelling, mentioned earlier, can cause small variations in the Z-dimension, which also affect the as-printed road height. Again referring to Fig. 11, the distance the head/nozzle is physically positioned above the build platform affects the as-printed road height, because in lower positions, more new material is deposited on the existing road. This results in a slightly higher Z-height of the newly deposited road, which happens over a variable distance. This variability amplifies as more layers are printed, resulting in greater amounts of error in taller parts. Some extruders can tend to pulse the material, ever so slightly, as it leaves the nozzle, further exacerbating this effect. Mechanical systems have inherent variability in performance, which is usually less impactful in more-expensive RepRap printers and commercial printers. Variations in the Z-height of a road can occur as the material expands and shrinks after it leaves the nozzle, as shown in Fig. 11. This variance depends on the extrusion pressure and the material properties. As the nozzle becomes closer or farther from the existing road, the road height varies, which is multiplied with each successive layer. This can result in the cumulative height of the asprinted object being taller than the nozzle Zheight, which can result in the nozzle plowing into a newly extruded road. If the nozzle interacts with the warm material comprising the asprinted object, the object is damaged and the nozzle quickly becomes coated with arm

plastic, which adheres to the nozzle and is dragged along as the nozzle moves. This usually results in a clogged nozzle and quite a mess of stringy plastic (worse if the user is not paying attention) all over the build platform as the head moves. More-expensive commercial melt extrusion printers provide technology to level the effective Z-height of the as-printed part with some preset number of layers. These devices traverse the build platform and “shave” the asprinted object to adjust the actual height to the intended dimension, as defined in the CAD representation. A class of 3D printers, called hybrid printers, has emerged that provides a subtractive milling tool in addition to the additive printing apparatus. Hybrid devices enable additive fabrication concurrent with subtractive dimensional adjustments for accuracy. After some preset number of layers, the mill machines the as-printed object according to the actual height dimension specified in the CAD file.

Infill in Melt Extrusion One of the key advantages of additive manufacturing is the light-weighting of objects while maintaining the intended balance of form, fit, function, features, and finance (cost). In subtractive manufacturing methods, objects are made from solid material. As material is removed subtractively, the remaining object is still solid material. Three-dimensional printing, being an additive process, enables a hollow, completely closed object to be created. Whether or not this makes sense would be dictated by the product requirements, but usually some amount of structural strength is required. Infill can be added to the hollow spaces inside an object to give it structural strength. As would be expected, there are many different infill patterns to suit many geometric and part strength requirements. For example, different patterns would be better than others to design the amount of material in the corners of an object, which may be different again if there was a hole in a corner that needed extra strength. There are many adjustments one can make to specific settings in any given infill pattern. For example, the spacing between infill roads and so on can be modified to increase the density of a given pattern, to increase part stiffness. A good reference for information about infill patterns and settings can be found in Ref 11. The best choices of patterns and settings for each part design are driven by the ideal structural and accuracy requirements for the object being printed. For example, there is likely an ideal weight-per-strength ratio for a particular design. A choice of infill pattern may also affect the quality and accuracy of an object. This is discussed in the section “Path Planning in Melt Extrusion” in this article. Cost is also a factor in determining the best choices of patterns and settings.

Just as there are many choices for infill patterns and settings for each pattern, there are also many trade-offs to be considered when defining infill. Adding infill density increases the weight of the as-printed part. It also adds to print time. The biggest potential risk associated with using too much infill is the increased potential for part warpage that is associated with too much part density. A denser volume of hot plastic is more likely to distort than something less dense as it cools. Low, flat shapes are especially risky if they are too dense. Material type and the associated properties can affect infill decisions. Because the goal is usually to maximize strength with minimal material usage, dual-headed printers can also be used to print less of an especially stiff material as infill and a very precise polymer as the primary material. This would result in an object that may be especially appealing visually, with higher structural strength than would otherwise be possible with the precise polymer. This would be dependent on the particular geometry and topology.

Path Planning in Melt Extrusion Infill patterns can have a dramatic effect on the quality and accuracy of a printed object. Printing some roads closer to or farther from others results in a variable amount of material extruded in specific regions, as discussed earlier. This results in a higher density of material in regions with more material, which, in turn, results in variations in the heat-transfer dynamics in local regions of the as-printed object. This is another source of print variability that causes poor accuracy and lower consistency over multiple objects. Printing some regions of a part before others is a method of optimally distributing material, thereby reducing the variability caused by localized changes in part density. Printing certain parts of an object before others enables longer or shorter relative cooling periods in local regions, before more roads are printed over the same area. Defining the route that the print head takes in a given print layer is a process called path planning, and the route is referred to as the scan path. Scan paths are often planned in advance to take these factors into account. Sophisticated planning algorithms were originally developed for photopolymer printing processes, but the concept maps easily to extrusion printing. As an example of how scan path planning can have a profound effect on part quality, consider the diagrams in Fig. 12. Recall that part quality refers to both aesthetics and strength. The outer wall and infill are indicated. Three different possible scan paths are shown, each of which will produce profoundly different effects on as-printed object quality. In Fig. 12(a), the scan path results in the infill intersecting with the outer wall and then

Material Extrusion Additive Manufacturing Systems / 49

Fig. 12

Three different possible scan paths, each of which will produce profoundly different effects on as-printed object quality. Source: Ref 12

moving in the other direction. In Fig. 12(b), the scan path runs parallel to the outer wall, while in Fig. 12(c), the scan path meets tangentially with the wall. Figure 12(b) is the strongest. It has the largest interface boundary. It is no different than putting a road on top of another road. There is a larger interfacial boundary, thereby providing greater bonding between the infill and the wall. A scan path can be defined to achieve this situation. Note, however, that because there is overlap between the infill and the wall, additional material will build up, creating the ooze problem discussed previously. This will create a bumpy surface on the outer wall of this object, because of the excess material at the intersection locations. Conversely, Fig. 12(a) would be the weakest, because of a smaller overlap region between the infill and the wall, but it would produce the smoothest outer surface of the object. The ideal scan path would be that shown in Fig. 12(c), because it maximizes the interfacial boundary while minimizing excess material. Not all software tools enable this complex type of scan path, sometimes requiring custom editing of the G-code. Scan paths can be developed to optimize many situations, minimizing predicable error while maximizing part quality. The available infill patterns are limited to the most common situations. In situations where features such as holes are close to the outer surfaces of an object or in other areas of high detail, it is often necessary to edit the scan paths to optimize infill placement, thus minimizing part distortion. Each case is unique.

Material Properties Numerous characteristics help predict how the materials will behave when extruded with heat. Some of these material properties have a greater impact on melt extrusion than others. Amorphous materials have a Tg at which they begin to soften and a Tm at which they liquefy. This enables fine control of the shape of the heated plastic as it forms on the build platform. As examples, the Tg for polylactic acid (PLA) is approximately 60 to 65  C

(140 to 150  F); for acrylonitrile-butadienestyrene (ABS), it is approximately 94 to 105  C (200 to 220  F). Polymers list a range of Tg values because of variations in crystallinity and proprietary mixtures with fillers. High-quality polymers are expensive. A high-quality polymer is more expensive because it has very low contamination, fillers, and so on and therefore demonstrates a more predictable behavior. Another very important material property is thermal conductivity. Polycaprolactone cools slowly, has low thermal conductivity, and holds more heat, so it is difficult to print because it does not take shape (cool) quickly. A fan on the nozzle at output is usually required to cool it faster. On the other hand, glycol-modified polyethylene terephthalate cools quickly, which enables it to span relatively larger gaps between supports. The coefficient of friction also plays an important role in the printability of a material. For example, ABS has a lower coefficient of friction than PLA, so it is easier to print. This is why many people start with ABS when they are not concerned about other parameters, such as the fact that it warps more. So, it is all a balance. Among other important properties is the coefficient of thermal expansion, which relates to how much a material will expand when heated. This has direct implications on the stack-up error discussed previously. The heatdeflection temperature (HDT, or Vicat softening point) characterizes the lowest temperature where a particular polymer begins to deform when heated. The HDT also helps characterize the maximum working temperature that an object can withstand in operation. Thus, it is sometimes referred to as the maximum working temperature. Many other properties provide standard material strength information, such as tensile and flexural modulus and strength, which help predict how an as-printed object will perform. Commercial vendors characterize their materials and provide these specifications. It is best to start with a manufacturer’s recommendations for printer settings and then vary slightly, considering each specific situation. An excellent table from Simplify3D (Ref 13) provides a good overview of many of the most

important properties for the most popular melt-extruded materials. Stratasys also provides a useful overview of their materials, along with a good general reference to the important material properties, in comparison to nonproprietary suppliers (Ref 14). It is very important to note that these material properties are measured using standard methods, such as the American National Standards Institute, International Organization for Standardization, or ASTM International. These standards work from standard specimen dimensions. These specimens are usually molded into the required shapes, not 3D printed. Therefore, these specifications provide bulk material properties, not asprinted material properties. Despite the most optimum balance of all heat-transfer factors discussed earlier, with very few exceptions, 3D-printed parts in general and definitely melt-extruded parts are anisotropic by nature. As the layers of roads are printed atop each other, as-printed parts assume different structural properties in different directions. For example, parts are always much weaker when bent in alignment with the roads than when bent against the alignment of the roads. These differences are geometry and feature-size dependent and are always a factor to be considered. No matter how complete the bonding of roads, as discussed previously, as-printed objects usually fail in this plane before they fail in other dimensions. It is important to remember that the properties provided with a material are only a rough estimate of how the as-printed part will perform in production. Thorough testing must be completed to evaluate the properties of each object and to predict actual performance. These testing methods and protocols should be statistical in nature, with sample sizes and geometric cases commensurate with the importance, duty cycle, and safety requirements of each part.

Materials for Melt Extrusion Hundreds of materials are available for melt extrusion. Coverage of this topic in any detail would be too big to cover in this article, so

50 / Polymer Additive Manufacturing Processes further investigation is left to the reader. The table referred to in the prior section (Ref 13) provides properties for the most popular meltextruded materials. Web searches will uncover many more, along with their material properties. A useful table from Simply3D (Ref 15) provides an excellent look at the most popular materials for melt extrusion.

Viscous Extrusion 3D Printing Viscous extrusion is the extrusion of a viscous, often honeylike material through a nozzle. It does not rely on heat to soften the material. Pushing caulk with a caulking gun is a form of viscous extrusion. A trigger generates pressure on a plunger that pushes a pastelike material through a hole. Too much pressure, too much material, or too fast movement results in a gap in a road. Viscous extrusion systems rely on the viscosity of the material to enable its extrusion. These devices extrude some form of paste, gel, or slurry by using a bulk material and a solvent or carrier to thicken it to a controllable/desired viscosity. When extruded, this solidifies into a shape as the solvent or carrier evaporates. Viscous extrusion systems may also include some use of heat for secondary processing, but it is usually low heat and is not the primary means of creating an object. Viscous extrusion systems rely on similar forms of pressure generation as in melt extrusion, for example, ram, helical, or pneumatic. The most common viscous extrusion method also uses a ram extrusion method, in this case, a syringe pump that pushes on the plunger of a syringe to extrude viscous material, such as a paste or gel, through the syringe orifice or a dull (pointless) needle, as shown in Fig. 13. Figure 14 shows a helical (auger)-type system for paste extrusion. Helical extrusion systems provide an alternative feed mechanism for viscous extrusion. A helical worm gear turns inside a syringe tube, propelling the material forward through the orifice (needle). Gas pressure (pneumatic) is also used in viscous extrusion. Helical (auger) extruders are best for rapid printing and retraction, as in highly detailed printing. Extrusion pressure and part quality are best controlled with an auger extruder. Because of the control the screw provides, it can go forward and back more quickly under higher pressures. The other methods that depend on vacuum are not as quick to respond because of the back pressure in the nozzle. When printing viscous materials, the materials are often designed to be shear thinning. Because shear is created by increasing pressure, it thins, so it cannot come through the needle faster and easier. When the shear decreases, it thickens again. Bioprinting is done mostly by viscous extrusion. Exceptions are emerging as the field quickly expands. The best way to 3D print with live cells is to use a bioprinter that can print

Fig. 13

A syringe pump pushes on the plunger of a syringe to extrude viscous material, such as a paste or gel, through the syringe orifice or a dull (pointless) needle. Source: Ref 16

Fig. 15 Fig. 14

Helical (auger)-type system for paste extrusion. PTFE, polytetrafluoroethylene. Source: Ref 17

and create shapes without heat, which kills cells and denatures proteins and bioactive reagents. The material will behave very differently depending on where it is printed. This is called extrusion on demand. There are many methods to optimize extrusion on demand, for example, controlling the attack-retraction mechanism. Auger (helical extruder) printers are generally more precise. Syringe pumps introduce problems because they do not move material fast enough. Figure 15 shows a helical extruder from Autodesk. Many processes in viscous extrusion combine gels and pastes with other materials to

Helical extruder from Autodesk

make composites. This is one of the most powerful reasons to use this printing process. As more material is mixed into the base material (the matrix), the viscosity increases. If the viscosity becomes too high, it cannot be extruded. The pressure required depends on a balance between the viscosity, matrix content, and material content, as represented in Fig. 16. If the material content is too low, the printed object will have poor mechanical properties. If the matrix content is too low, the object will be too brittle to remove from the build platform. If the viscosity is too low, the material will not create a final shape. If the viscosity is too high, the material will not extrude unless the pressure is sufficiently high, which sometimes is not viable. This relationship between

Material Extrusion Additive Manufacturing Systems / 51

Fig. 16

The pressure required depends on a balance between the viscosity, matrix content, and material content.

pressure and viscosity also sets limits on the accuracy of printed objects. Much more pressure is required to extrude a viscous material through a smaller-radius needle than a larger radius. These relationships are defined by the HagenPoiseuille equation. At a given flow rate and needle length, the pressure required to extrude a material is directly proportional to the viscosity and inversely proportional to the radius to the fourth power. Thus, it is it very difficult to obtain good accuracy and resolution with a fairly viscous material. High viscosity is often required to obtain satisfactory mechanical properties in the as-printed object. REFERENCES 1. “Standard Terminology for Additive Manufacturing—General Principles— Terminology,” 52900:2015(E), ISO/ ASTM International, 2015

2. J.R. Wagner Jr., E.M. Mount III, and H.F. Giles Jr., Extrusion: The Definitive Processing Guide and Handbook, 2nd ed., Elsevier, 2014, https://doi.org/10.1016/ B978-1-4377-3481-2.00003-X 3. “Polymer Screw Extrusion Introduction,” PTFE Machinery, Feb 22, 2017, https://ptfe machineryblog.wordpress.com/2017/02/ 22/polymer-screw-extrusion-introduction/ 4. “Wohlers Report 2019: 3D Printing and Additive Manufacturing State of the Industry,” Wohlers Associates, 2019 5. B. Shaqour, “Developing an Additive Manufacturing Machine,” 2016, https:// www.researchgate.net/figure/The-Bowdenconfiguration-on-the-left-vs-the-Direct-con figuration-on-the-right_fig4_317974747 6. A. Sidambe, “Biocompatibility of Advanced Manufactured Titanium Implants—A Review,” 2014, https://www. researchgate.net/figure/Fused-depositionmodelling-schematic_fig10_269836411

7. “Local Motors Strati 3D Printed Car,” Local Motors, 2014, https://www.thingiverse.com/thing:476016 8. M. Mueller, “Digging In: Oak Ridge National Laboratory Creates World’s First 3D Printed Excavator,” Office of Energy Efficiency and Renewable Energy, 2017, https://www.energy.gov/eere/articles/digging-oak-ridge-national-laboratory-createsworlds-first-3d-printed-excavator 9. “CURA 13.11.2 User Manual: Ultimaker’s Software for Making 3D Parts,” Ultimaker, https://ultimaker.com/download/170/ Cura_User-Manual_v1.0.pdf 10. B.N. Turner, R. Strong, and S.A. Gold, Review of Melt Extrusion Additive Manufacturing Processes, Part I: Process Design and Modeling, Rapid Prototyp. J., Vol 20 (No. 3), 2014, p 192–204, DOI 10.1108/RPJ-01-2013-0012 11. “Slic3r Manual,” https://manual.slic3r.org/ expert-mode/print-settings 12. M.K. Agarwala et al., Structural Quality of Parts Processed by Fused Deposition, Rapid Prototyp. J., Vol 2 (No. 4), 1996, p 4–19 13. “Filament Properties Table,” Simplify3D, https://www.simplify3d.com/support/mate rials-guide/properties-table 14. “Our Materials,” Stratasys, https://www. stratasys.com/materials/search?technologies= ff37d7b8297c4e43977c155d765f3305sort Index=0 15. “Ultimate 3D Printing Materials Guide,” Simplify3D, https://www.simplify3d.com/ support/materials-guide/ 16. W. Li et al., “Methods of Extrusion-onDemand for High Solids Loading Ceramic Paste in Freeform Extrusion Fabrication,” Solid Free-Form Fabrication Symposium (Austin, TX), 2015, https://www.researchgate.net/figure/Schematic-of-paste-extrusion-for-the-freeform-extrusion-fabrication-process-and_fig1_304252577 17. D. Drotman et al., “Control-Oriented Energy-Based Modeling of a Screw Extruder Used for 3D Printing,” ASME 2016 Dynamic Systems and Control Conference, https://www.researchgate.net/figure/ Experimental-3D-printer-extruder-used-forprinting-from-raw-pellets_fig3_313786675

Copyright # 2020 ASM InternationalW All rights reserved www.asminternational.org

ASM Handbook, Volume 24, Additive Manufacturing Processes D. Bourell, H. Kuhn, W. Frazier, M. Seifi, editors DOI 10.31399/asm.hb.v24.a0006543

Powder Bed Fusion of Polymers David K. Leigh, EOS North America David Bourell, University of Texas at Austin

POWDER BED FUSION (PBF) of polymers is a collection of additive manufacturing (AM) processes that melt and fuse polymer in a powder bed. The general process for laser sintering using a scanning laser beam on the surface of the powder bed is illustrated in Fig. 1. A thin layer of powder, ~100 mm, is spread over the surface, typically using a recoating blade or counter-rotating cylinder. The laser beam spot is moved over the surface with the assistance of scanning galvanometer mirrors, melting the powder and causing adherence of the melt pool to the previously deposited layer. The process is repeated to build up the part. Another method for PBF, shown in Fig. 2, involves selectively depositing a fusing agent and/or detailing agent on the surface of the powder bed after powder is deposited. Here, an infrared light source bathes the surface of the powder bed with light, but fusion occurs only where the fusing agent is present and where there is no detailing agent. Because jetting is significantly faster than laser scanning, areal PBF is considerably faster than laser-based PBF.

Fig. 1

Schematic of laser-based powder bed fusion (Ref 1)

In contrast to most other AM processes, the powder bed serves as a support structure for the part, enabling the creation of support-free parts. Both the material cost of the support structure and the postprocessing cost of removing supports are eliminated. Beyond economics, the ability to design parts without consideration of the support structures enables geometric freedom in design. Here, the only substantive design rule is that parts with internal features must have an exit point for powder to be removed. Another feature of supportless manufacturing is the ability to nest parts in a single build easily, which increases part-packing density; parts also build faster without supports. Polymer PBF feedstock is generally certain types of semicrystalline polymers, with the most popular being polyamide (nylon), with or without composite additions such as flame retardants and glass. Other feedstocks for polymer PBF include polypropylene, polyetheretherketone (PEEK), and polyetherketoneketone (PEKK). The attractive mechanical properties of these AM thermoplastics explain why PBF is widely used for polymeric functional prototyping and service parts. Polymer PBF is used in aerospace

Fig. 2

Illustration of areal powder bed fusion (Ref 2)

applications (e.g., ductwork using polyamide), biomedical applications (implants using PEEK), consumer parts, automotive applications, and legacy parts (parts that are no longer manufactured and for which drawings may or may not be available, such as replacement parts for antique automobiles) (Ref 3). Polymer PBF competes on a cost basis with traditional “lowcost” AM processes such as materials extrusion, particularly for production runs because the build rate is higher for PBF due to both the elimination of supports and high build speed. Very few powder feedstocks are considered “neat” materials (raw with no additives). Additives to powder feedstocks either facilitate superior processing in the PBF process or complement an end-use application. Additives that improve PBF processing include whitening agents, antioxidants, and flow agents. Additives used to enhance the end-use part include ultraviolet stabilizers; custom pigmentation; glass, mineral, and carbon fibers to improve mechanical properties along a single axis; flame retardants; and glass, mineral, and ceramic powder to improve heat deflection and stiffness.

Powder Bed Fusion of Polymers / 53 In 2018, polymer powders for AM represented 26.9% of all feedstock sales for AM (402.1 million USD for polymer powder of 1.495 billion USD total sales) (Ref 4). Fig. 3 shows worldwide sales of polymer powders for PBF from 2008 to 2018 (Ref 4). The increases in powder sales for 2016, 2017, and 2018 were 24.7, 29.1, and 37.9%, respectively. ASTM F3091 (Ref 5) provides useful information for classifying polymer PBF parts, part ordering, materials specification, test coupons, feedstock issues, production builds, inspection, part retesting and rejection, certification, identification, and packaging.

Depending on the thermal stability of the polymer, unused powder can be recycled from previous builds. Laser-based PBF is accomplished at elevated temperature, where usually the powder bed temperature is a few degrees below the melting point. Preheating powder prior to depositing into the powder bed build area is necessary to ensure that the previously melted layer is not crystallized when the next layer of powder is supplied to the powder bed. However, due to thermal degradation considerations and their impact on recycling, the feedstock is typically preheated to a lower temperature than the powder bed. Thermal degradation of polymer occurs through both oxidation and cross linking. For the latter, increases in the molecular chain length lower the flow viscosity of the melt. Eventually, the polymer feedstock flow characteristics degrade to the point where surface irregularities (e.g., “orange peel”) appear, or the part quality is significantly compromised. For polyamides, feedstock for production is often a blend of used and fresh powder. If the polymer thermal degradation is kinetically rapid, as is the case for PEEK powder, then the powder cannot be recycled. Areal PBF is conducted with localized heating, but the effects of the fusing and detailing agents must be considered if powder is recycled. Polymers for laser sintering are typically semicrystalline thermoplastics. In PBF, it is desirable for the melting point of solid material on heating to be higher than the crystallization temperature of the melt on cooling. This is illustrated in Fig. 4 for polyamides (Ref 6). As for metals, when polymers are run in laser sintering at room temperature, significant curling or distortion occurs if no support structure is used. Polymers, unlike metals, stress relax at relatively low temperature, so the distortion can be eliminated by heating the powder bed. Typically, the powder bed is heated to just below the melting point. For polyamide 11 (Fig. 4), the typical powder bed temperature is ~180
C. Table 1 shows typical powder bed temperatures and thermal processing windows (melting temperature minus crystallization temperature) for several commercial polymer powders.

Fig. 3

Sales of polymer powder feedstock for powder bed fusion (Ref 3). USD, United States dollar

175 2

210

250

285

Temperature, °F 320 355

390

430

465

220

240

161.4 °C (322.5 °F) 1 Heat flow, W/g

Thermal Issues

0

–1

191.7 °C (377.1 °F) –2

Fig. 4

Table 1

80

100

120

140

160 180 Temperature, °C

200

Calorimetry plot of heat flow versus temperature for polyamide 11. The melting temperature is ~191
C (~377
F), and the crystallization temperature is ~161
C (~322
F) (Ref 6).

Powder bed fusion polymer thermal and mechanical properties Elastic modulus

Material/formula

PA12, Neat PA12, glass beads PA12, carbon fiber PA12, mineral fiber PA11, neat PA11, carbon fiber TPE TPU PEKK PEKK, carbon fiber PPS PA6

Product description

Tensile strength

Elongation at break, %




C




F

Deflection temperature at 0.46 MPa (0.07 ksi)


C




F

Deflection temperature at 1.8 MPa (0.26 ksi)


C




F

MPa

ksi

MPa

ALMPA 650 ALMPA 614-GS

1720 3650

249 529

47 41

7 6

20 5

181 358 186 367

178 177

353 351

86 122

187 252

ALMPA 603-CF

5403

929

75.5 11

5

185 365

179

354

173

344

ALMPA 620 MF

5725

830

51

7

5

184 363

184

363

179

354

ALMPA 850/860 PA 802-CF

1496 5205

217 755

48 70

7 10

49 17

201 394 185 365

184 186

363 367

44 179

111 354

75

11

8

1

200

150 302

…

…

86 3000

12 435

18 88

3 13

276 3

… … 302 576

… …

… …

… 139

… 282

6894 1000 110

16

2

307 585

…

…

…

…

7 11

4 13

295 563 220 428

220 199

428 390

116 100

241 212

EOS PrimePart ST 2301 Farsoon 1092A Arkema Kepstan 6002 OPM Oxfab ESD Farsoon 8100 PPS BASF Ultrasint X028

… …

… …

47 78

ksi

Melting point

…

PA, polyamide, TPE, thermoplastic elastomer; TPU, thermoplastic polyurethane; PEKK, polyether ketoneketone; PPS, polyphenylene sulfide

54 / Polymer Additive Manufacturing Processes greater than the energy required for melting, and consequently, the material continues to stay above the critical temperature even after the laser has transitioned to another area. Pointsource PBF also relies on a second heat source to heat the powder bed region not being scanned by the laser. A main issue in areal PBF is that due to the lower processing temperature, large flat areas tend to curl after melting.

Safety Considerations While polymeric materials generally do not pose particular safety concerns, when in powder form, certain issues can arise. For example, powder can become airborne, particularly when feedstock is loaded at the start of a build and parts are broken out at the end of a build. Operators should wear appropriate inhalation masks to avoid powder inhalation or take other precautions including forced air ventilation to remove particulate safely. Organic aerated powder may pose an explosion hazard given certain circumstances, so the airborne content should be controlled within established explosion limits. Finally, fine particulate may settle on walkways, posing a slip hazard. In industrial settings, daily mopping of the workspace helps to avoid creation of slip hazards.

Manufacturing Issues During planning of a part build, determining the orientation of parts in the build chamber,

particularly minimizing the z height of the build, is important. This results in the fastest build time and uses the minimum amount of powder. For PBF of polyamide using optimized commercial processing parameters, part strength is rather isotropic, but the ductility is on average ~50% lower in the z direction than in the plane of the build. As mentioned in the section “Thermal Issues,” parts built for structural applications are typically placed in the center of the build volume, the “sweet spot” where mechanical properties are optimized (Fig. 5). Regarding orientation effects on surface finish, true vertical and horizontal walls produce the best finish. Otherwise, the potential for stair stepping is present; this is illustrated schematically in Fig. 6. Surface finish may be limited to approximately the powder particle size, typically ~50 mm. Setting manufacturing parameters is critical to optimize part properties. Optimizing surface finish entails small layer thickness, fine particulate feedstock, and usually high scan speed and nominal hatch spacing. For laser sintering, laser power is controlled to minimize over-melting of surfaces (“part growth”), which negatively impacts part dimensions. For optimized part strength, a minimum threshold macroscopic volume energy density is required (Ref 9), where the energy density ED is given by: ED ¼

4000 28 (4000) 3500

21 (3000)

3000

14 (2000) 7 (1000)

2500

0 –1

2000 1

3

1500 5 7 9

X location, in

11 13

–1 1

3

5

7

9

11

1000

W_ ðHSÞðtÞðvÞ

25

20 20

15 10

15

5 0 –1

1

10 3

5 7

500 Y location, in

X location, in 0

(Eq 1)

25 Elongation break, %

Yield strength, MPa (psi)

On cooling immediately after laser scanning, the molten laser-sintered material cools back to the powder bed temperature in a matter of seconds, but, because the powder bed temperature is higher than the crystallization temperature, the layer remains molten. This persists for multiple layers before the molten material finally cools to the crystallization temperature due to heat conduction through the build chamber side walls, which typically takes several minutes. This is beneficial, because time is allowed for thermal equilibrium to take place in the part, which in turn minimizes residual stress and distortion during cooling. In the laser sintering powder bed, cooling occurs primarily by conduction through the unheated metallic build chamber side walls. Due to the insulating characteristics of powder beds, cooling to room temperature after a large build may require 20 to 30 h. For example, the measured thermal conductivity of polyamide 12 powder has been measured to be ~0.1 W/mK, ~200 times less than the thermal conductivity of steel (Ref 7). Parts built near the edge of the powder bed therefore cool much quicker than parts built in the middle of the powder bed. The increased time at temperature in the middle of the build chamber may not affect part strength, but it is beneficial to certain mechanical properties, particularly ductility. This is shown in Fig. 5, a plot of the (a) strength and (b) ductility of z-oriented ASTM D638 (Ref 8) tension specimens as a function of their location in the build cylinder for a commercial polyamide 11 build (Ref 6). To obtain good ductility, parts must be built no closer than about 50 mm (1.97 in.) from the edge of the powder bed. This central area where optimal mechanical properties are obtained is referred to as the laser sintering chamber “sweet spot.” Due to convective cooling of the top surface of a build, this area ends approximately 10 to 30 mm (0.39 to 1.18 in.) from the top surface of the build. Beneficial effects of time at temperature on part ductility have several origins. First, polymer chain diffusion occurs, destroying weak interfaces between prior particle boundaries. Residual porosity is reduced by extended exposure to elevated temperature. Additionally, viscous sintering allows flow and bonding of the polymer melt. Thermal issues are less extensive for areal PBF of polymers compared with point-source PBF. Areal PBF exposes a region to a highintensity infrared lamp that heats the broader area while the infrared absorber that defines the part geometry absorbs sufficient energy to allow it to melt. This process results in less thermal exposure to the powder bed than with pointsource PBF, because powder bed preheating occurs while the part is being melted. In addition, it takes about 1 to 2 s for the material to respond to the applied heat and start to flow. This time corresponds with the energy curve of areal PBF, while point-source PBF energy is applied for only a fraction of a second. Therefore, the energy applied in point-source PBF is much

9

11

13 –1

1

3

5

7

9

11

5

Y location, in 0

Fig. 5

Plot of the (a) yield strength and (b) elongation at break of polyamide 11 (ALM FR-106) z-oriented test specimens as a function of the x-y location in the build chamber. Plots are averages of approximately 68,000 ASTM D638 tension specimens in a commercial build environment (Ref 6).

Fig. 6

Schematic showing the effect of build orientation on stair stepping from layering

Powder Bed Fusion of Polymers / 55 where W_ is the laser power, HS is the hatch spacing, t is the layer thickness, and v is the scan speed. Fig. 7 shows a typical plot of the effect of energy density on strength and ductility of polyamide 12. For polyamide, the threshold energy density is ~0.1 to 0.15 J/mm3. The correlation of energy density to strength is much stronger than the effect on ductility. This is due to the additional consideration of accumulated thermal exposure throughout the entire PBF process chain if part ductility is to be optimized. The energy density effect is operational within a fixed power-velocity range, assuming constant hatch spacing and layer thickness. At low energies (high scan speed and low laser power), there is insufficient energy to adequately bond powder particles to each other and to the preceding layer. This results in delamination for extremely low values and in fusion errors otherwise. At high power and low scan speed, the energy density may become so high that material degradation occurs. This may take the form of chain scission or porosity formation due to polymer decomposition. An additional deleterious factor at high energy density is part growth, which negatively impacts part dimensions. Polymers for PBF oxidize at temperatures below their melting temperature. Therefore, oxygen exposure in the powder bed must be limited during the build. Typically, due to economic considerations, nitrogen is used as a carrier gas. Gas pressure in the chamber is typically held at slightly above 1 atm, so that any leaks result in nitrogen flow out of the build chamber rather than air flow into the build chamber. Shielding gas flow must be monitored, because changes in the flow due to appearance of leaks may negatively impact the flow over the powder bed. This may result in localized cooling, which may lead to diminished accumulated thermal exposure and reduced mechanical properties at that location.

Postprocessing and Finishing

Common Defects and Part Properties

Polymer parts created using PBF typically are postprocessed after the build is completed. At the least, the loose powder surrounding parts (i.e., part cake) must be removed and refurbished if it is to be recycled. Residual loose powder may be removed by mechanical means with scraping tools and brushes. Use of a pressurized air supply with or without abrasive media can facilitate removal of powder from hard-to-reach locations, but care must be taken to control exposure to airborne particulate. Part surface finish is improved by traditional surface finishing techniques for polymers. A common method is to tumble the part in hard media. This approach is effective regardless of part geometry, but care must be taken respecting fine features that may be lost. Polymers for PBF typically have rather low machinability due to smearing of machined surfaces. This is ameliorated by first coating the polymer with a thermoset polymer, typically a room-temperature curing epoxy or cyanoacrylate. After the surfaces are suitably coated and finished, PBF parts may be primed, filled, and painted or plated, typically using methods similar to finishing of polymeric parts in the automotive or aerospace industries. PBF parts are typically first sealed using a low-viscosity polymer such as cyanoacrylate. A sandable auto primer is then applied. As necessary, surfaces and defects may be removed using a suitable filler such as Bondo (3M, St. Paul, MN, USA). For painted parts, any high-quality automotive or aerospace body paint may then be applied. In some cases, a finishing clear coating is applied to provide ultraviolet stabilization, improve scratch resistance, and limit atmospheric corrosion. As an alternative to painting, standard coating techniques such as electroplating and physical vapor deposition are possible.

Principal defects in PBF polymer parts are interlayer interfaces, prior particle boundaries, and porosity. Polymeric PBF parts are increasingly used in structural applications where mechanical properties are important. (Table 1 lists typical mechanical properties for a variety of PBF polymers.) Relative to mechanical properties of injection molded resins, generally the strength of PBF parts is comparable and isotropic, but ductility is decreased and anisotropic. This is attributed to the fact that prior particle and layering interfaces exist in PBF polymeric parts. These interfaces are not intersected by long-chain molecules and serve as points for fracture. Ductility in PBF parts is comparable with values for compression-molded resins, as the latter processing results in similar compromised interfaces as well (Ref 11). The considerable mixing and shear deformation associated with injection molding destroys the prior particle interfaces, resulting in parts with randomly oriented long-chain molecules and comparatively high ductility. Under optimized processing conditions, PBF part strength is relatively isotropic. This is illustrated in Fig. 8 for laser-sintered polyamide 11. Elongation to fracture does show significant anisotropy. For the data shown in Fig. 8, the elongation to failure in the build (z) direction is about half the value in the build plane (x and y). The loss in ductility is generally attributed to layer interface weakness. As discussed in the section “Thermal Issues,” parts built in the center of the build volume, away from the build volume side walls and top surface, have optimal mechanical properties (Fig. 5). It is therefore

60

40

Enough energy to fully melt powder

6

35

5

30

4

25

4

20

3

15

2

10

Blend

2

5

Virgin

1

0.25

0.30

0.5

1.0

1.5

2.0

Elongation, %

2.5

20

3.0

20 blend 0

virgin 15

55 (8000) Strength, MPa (psi)

40

0.0

Yield stress,ksi

7

Elongation at break, %

Yield stress, MPa

Energy-melt ratio 45

10

5

Tensile strength

41 (6000)

28 (4000) Yield strength 14 (2000)

0

0 0.00

(a)

Fig. 7

0.05

0.10

0.15

0.20

Energy density, J/mm

3

0

0

X 0.00

(b)

0.05 0.10 0.15 0.20 Energy density, J/mm3

0.25

Y

Z

0.30

Effect of volume energy density (Eq 1) on (a) strength and (b) elongation at break for polyamide 12. While the effect of energy density on strength is monotonic, there is considerable scatter in ductility (Ref 10).

Fig. 8

ASTM D638 tension test results for lasersintered polyamide 11 as a function of tensile coupon orientation. Squares represent the average values for the 3851 samples tested (Ref 12).

56 / Polymer Additive Manufacturing Processes common, per ASTM F3091, for certification purposes to “cage” parts by surrounding them with tension coupons to ensure that parts have minimum mechanical properties, as illustrated in Fig. 9. While polymer PBF parts processed under optimal conditions have reasonable strength and ductility (Table 1), a transition in mechanical behavior may occur when parts are processed under less-than-optimal conditions. In practice, deviation from optimal processing conditions may occur due to processing issues such as laser window fogging, expansion of the laser spot size, and degradation of the laser power with age. As the defect density increases, as manifested primarily as porosity and weakened layer interfaces, the part stress intensity increases. This may result in extreme conditions in a shift from a full plasticity mode of failure to one driven by fracture mechanics (illustrated in Fig. 10 for a laser-sintered polyamide 11 foot prosthetic that shattered when flexed). Fig. 11 shows a series of stress-strain curves for laser-sintered polyamide 12 specimens with varying levels of engineered porosity. According to a fracture mechanics argument, the fracture stress varies linearly with the intrinsic material toughness KIc and inversely with a function of the extent

pffiffiffiffiffiffi of the defect structure Y pa, where Y is a stress intensity factor and a is a nominal crack size. Increasing the defect structure lowers the fracture stress, as shown in the Fig. 11 plots, where the primary difference between the various stress-strain curves is not the general loaddeflection path but rather the point at which fracture occurs. When the fracture stress drops below the full plasticity yield stress, the elongation to fracture can become quite low (0–3%).

Case Studies in Polymer Powder Bed Fusion Part Count Reduction and Lightweighting Bell Helicopter developed a flight certification for the use of laser-sintered nylon in their rotary wing aircraft. The primary issue that drove the adoption of a new manufacturing process was the ability to design for function such that very complex geometries were made in a lightweight fashion. The laser-sintered parts, pictured in Fig. 12, were much thinner than typical roto-molded geometries, and the design-for-function approach allowed many of the assemblies to be combined into a single geometry, eliminating the weight of the adhesives and fasteners as well as reducing labor cost for assembly. This lightweighting allowed for the aircraft’s functional payload to be increased for applications that required additional equipment and personnel in situations such as life-flighting critical care patients.

nylon 11 material produced on 3D Systems equipment. The V-22 Osprey had been grounded from operations in Afghanistan and Iraq. Increased deployment of the Osprey in a challenging environment eventually caused a component in the nacelle to fail. The sand in the region was highly abrasive and damaged the rotors of the turboprop engine during take-off and landing, and when the craft was hovering close to the ground. To mitigate the effect of the sand, a high-speed centrifugal fan was used to displace the sand before the air was pulled into the main engine. When the centrifugal fan bearings failed, the hydraulic system over-pressurized and caused a rupture in the line at the air intake. Since the centrifugal fan was in the air intake duct, the leaking hydraulic fluid ignited and was pulled into the engine. This failure resulted in the destruction of several airship nacelles before an investigation and failure analysis pinpointed the root cause. One of the solutions involved shunting the hydraulic fluid away from the engine intake and toward the tail end of the nacelle. Two sets of exterior-mounted tubing on each nacelle were designed, built through laser sintering, and tested. These parts are the light-colored runners indicated by the arrow in Fig. 13. Production-qualified parts were in the field within six weeks from the time of the initial order.

Part Count Reduction The Boeing Company used its experience in deploying 3D-printed solutions for prototypes and military applications to drive the adoption

Rapid Deployment In the summer of 2007, Bell Helicopter was required to use the laser sintering process to rapidly deploy a solution in the field. Traditional manufacturing methods could not meet the compressed time frame, and Bell Helicopter had just created and adopted new production specifications for the use of a

Fig. 9

50

Caging of ASTM D638 tension coupons around parts in the powder bed fusion build chamber

Stress, MPa

40

30

20

10

0 0

Fig. 10

Laser-sintered polyamide 11 prosthetic foot part that shattered when loaded due to lessthan-optimal processing conditions

Fig. 11

10

20

30 Strain, %

40

50

60

Laser-sintered polyamide 12 stress-strain curves of specimens with varying amounts of engineered porosity (Ref 11)

Fig. 12

Laser-sintered Bell Helicopter parts used to reduce part count, assembly cost, and weight. The line indicates the black powder bed fusion parts. Courtesy of EOS

Powder Bed Fusion of Polymers / 57

Fig. 13

Laser-sintered parts on the nacelle of a Bell/Boeing V-22 Osprey (Ref 13)

Fig. 14

Demonstrator air duct showing a reduction of ten components/fasteners to a single powder bed fusion polyamide part. Courtesy of Stratasys

of machines, materials, and processes during the development of the 787 Dreamliner. The Dreamliner features almost 40 air ducts manufactured using PBF of fire-retardant polyamide 11 (Ref 14). The PBF air ducts replaced hundreds of parts composing a complex assembly. Fig. 14 illustrates the concept for an aerospace air duct in which a ten-part assembly is replaced by a single PBF polyamide part.

Time-Sensitive Biomedical Customization The precision required to successfully ablate brain tumors using medical lasers led to the development of an application using lasersintered polyamide. A jig was created using a surface scan of the patient’s skull and a computed tomography scan locating the region to be ablated (Fig. 15). This jig was then used to align the medical laser head to a precise orientation and distance from the region to be ablated. This tool was fabricated in a matter of hours and delivered directly to the operating room to assist the surgical team during the procedure. REFERENCES 1. J.J. Beaman, J.W. Barlow, D.L. Bourell, R.H. Crawford, H.L. Marcus, and K.P. McAlea, Solid Freeform Fabrication: A New Direction in Manufacturing, Kluwer Academic Press, 1997 2. S. Dent, HP Wants to Be a 3D Printing Giant with New ‘Multi Jet Fusion’ Tech, Engadget, 29 Oct 2014, https://www. engadget.com/2014/10/29/hp-multi-jetfusion-3d-printer/ 3. M.A. Waller and S.E. Fawcett, Click Here to Print a Maker Movement Supply Chain: How Invention and Entrepreneurship Will Disrupt Supply Chain Design, J. Bus.

Fig. 15

4. 5. 6.

7.

8. 9.

Powder bed fusion polyamide customized surgical template for laser-ablating brain tumors. The black precision aligner is inserted into the white powder bed fusion part. Courtesy of EOS

Logist., Vol 35, 2014, p 99–102, doi: 10.1111/jbl.12045 “Wohlers Report,” Wohlers Associates, Ft. Collins, CO, 2019 “Standard Specification for Powder Bed Fusion of Plastic Materials,” F3091/ F3091M-14, ASTM International, 2014 D.K. Leigh, “Improving Process Stability and Ductility in Laser Sintered Polyamide,” doctoral dissertation, The University of Texas at Austin, 2019 M. Yuan and D. Bourell, Thermal Conductivity of Polyamide 12 Powder for Use in Laser Sintering, Rapid Prototyping Journal, Vol 19 (No. 6), 2013, p 437–445 “Standard Test Method for Tensile Properties of Plastics,” D638, ASTM International, 2014 D. Bourell, J. Coholich, A. Chalancon, and A. Bhat, Evaluation of energy density measures and validation for powder bed fusion of polyamide, CIRP Annals Manufacturing Technology, Vol 1 (No. 66), Aug 2017, p 217–220

10. T. Starr, T. Gornet, and J. Usher, The Effect of Process Conditions on Mechanical Properties of Laser Sintered Nylon, Rapid Prototyping Journal, Vol 17 (No. 6), 2011, p 418–423 11. D.K. Leigh, A Comparison of Polyamide 11 Mechanical Properties between Laser Sintering and Traditional Molding, 23rd Annual International Solid Freeform Fabrication Symposium Proceedings, The University of Texas at Austin, 2012, p 574– 605, https://sffsymposium.engr.utexas.edu/ 2012TOC 12. D.L. Bourell and D.K. Leigh, The Development of Mechanical Properties in Laser Sintered Polyamide, Polymers Moulds Innovations: Proceedings of the 5th International PMI Conference, Center for Polymer and Material Technologies, Ghent University, 2012, p 25–30 13. R. Kruse, “Photos of Military Helicopters and Tiltrotors,” 2010, https://richardkruse. com/Photos-Aviation-Helicopter.html 14. Boeing Engineer Honored for Dreamliner 787 Components, Design News, 18 May 2010

Copyright # 2020 ASM InternationalW All rights reserved www.asminternational.org

ASM Handbook, Volume 24, Additive Manufacturing Processes D. Bourell, W. Frazier, H. Kuhn, M. Seifi, editors DOI 10.31399/asm.hb.v24.a0006551

Material Jetting of Polymers Christopher B. Williams and Lindsey B. Bezek, Virginia Polytechnic Institute and State University

Process Overview Material jetting (MJ) is a classification of additive manufacturing (AM) processes that involves the selective jetting and subsequent solidification of liquid droplets onto a substrate in a layerwise manner. While the broad MJ classification encompasses selective jetting of any material, including ceramics (Ref 1), metals (Ref 2), and conductive and dielectric composites (Ref 3), this article focuses solely on MJ of polymers. The first commercialized MJ systems jetted molten waxy thermoplastics, which solidified due to cooling following deposition, to fabricate patterns suitable for lost wax casting. Later commercial MJ platforms were designed to fabricate objects from thermoset photopolymer resins (Ref 4). For these systems, an ultraviolet (UV) irradiation source (e.g., xenon arc lamp) facilitates cross linking of the jetted photopolymer resin droplets. Continued irradiation during subsequent printing passes provides sufficient UV dosage to promote interlayer adhesion and fully cure the printed parts; thus, the printed parts do not require additional postprocess UV curing. An MJ system features a print block with inkjet print heads, each containing hundreds of jetting nozzles. By accommodating multiple print heads, multiple different and/or colored materials can be simultaneously deposited within a single build layer. As the print block traverses the x-axis of the substrate, liquid droplets of build and sacrificial support materials are jetted onto the substrate. Using drop-ondemand (DoD) deposition mechanics, either thermal or piezoelectric pulses trigger the droplet ejection. In thermal jetting, a heater vaporizes the ink to form a bubble and drive a droplet out of the nozzle. In piezo DoD, a piezoelectric actuator physically deforms to displace the ink from the nozzle. Commercial photopolymer-based MJ systems generally use piezoelectric DoD print heads. For photopolymer MJ, shown in Fig. 1, jetted droplets are immediately flattened by a rotating roller as the print block moves along the x-axis. Wax-based MJ systems, such as those from Solidscape, have an incorporated machining tool that planes each layer. Both approaches

Fig. 1

Schematic of photopolymer material jetting process. UV, ultraviolet

provide a smooth surface for the deposition of the subsequent layer. To manufacture parts larger than the width of the array of print heads, the print block shifts incrementally along the y-axis. Once the layer is completed, the build plate lowers (z-axis) by the distance of one layer thickness to accommodate the jetting of the subsequent layer. Like other AM processes that do not use a powder bed, MJ requires the use of a sacrificial support material to structurally support overhanging features and downward-facing surfaces as they are constructed layer by layer. However, MJ is unique in that this sacrificial support material is applied to all part surfaces regardless of their overhang angle. As shown in Fig. 2, the support material in MJ provides a boundary on the contour of each layer that prevents the jetted liquid droplets from spreading and reducing part quality. Once printing is complete, this support material must be removed. Support-removal techniques vary depending on process, material composition, and part geometry. 3D Systems’ photopolymer MJ systems require freezing the build plate to enable the

Fig. 2

Sample parts fabricated on a Stratasys system. Once finished printing, parts are fully encased with support material in addition to having support within internal cavities.

separation of parts from the build plate, followed by a heat treatment to melt away the waxy thermoplastic support material (VisiJet S). Stratasys offers three types of photocurable gellike support materials. SUP705 is a hydrophilic material that must be removed manually or with a pressurized waterjet. SUP 706 and SUP 707 are soluble in

Material Jetting of Polymers / 59 sodium hydroxide and water, respectively. Parts from wax-based MJ systems (e.g., Solidscape) may be gently removed from the build plate before being placed in a stirred and heated solution bath to dissolve the supporting wax. Once the support material is removed, additional finishing steps are sometimes administered. To make translucent models (e.g., Stratasys’ VeroClear resin), photobleaching parts with fluorescent lamps is recommended to amplify clarity prior to polishing and lacquering the parts (Ref 5). For model strengthening, a quick dip in glycerol solution is optional. Models may also be polished or coated. Painting models may also be desired, and it is recommended to first prime the surface, followed by sanding if needed. Paint, dye, and lacquer can then be applied. Additional postprocess thermal treatments can increase the heat-deflection temperature for certain materials (e.g., Stratasys’ High Temperature resin). A summary of the specifications of select commercial photopolymer MJ systems is provided in Table 1. Both Stratasys and 3D Systems, who commercialize their technologies as PolyJet and MultiJet Printing, respectively, offer several machine options at different price points and capabilities, and representative systems are presented for comparison. 3D Systems currently offers a variety of high-definition modes (e.g., “Ultra High” or “Xtreme High”), and these modes define the build volume, build plane resolution, and layer thickness. For example, for the ProJet MJP 5600, “Ultra High Definition” achieves 600
600
1600 dots per inch and 16 mm layers, while “Xtreme High Definition” achieves 750
750
2000 dots per inch and 13 mm layers. Stratasys distinguishes High Speed, High Quality, and High Mix modes that vary in print time and layer thickness. For the J750, High Speed and High Mix modes print 27 mm layers, and High Quality mode prints 14 mm layers. For High Speed mode, multiple print heads are designated to run the same material, thus reducing the number of print block passes required to complete a single layer.

Process Characteristics An MJ system has specific, functional characteristics that distinguishes it from other AM technologies. Specifically, each jetting head has individually addressable nozzles capable of dispensing picoliter-scale droplets, typically 40 to 60 mm in diameter, of different materials. Using an array of inkjet heads enables the production of large parts with high-resolution features and quality surface finish from multiple materials and multiple colors.

Multiple Materials Multiple, individually addressable inkjet heads enable MJ systems to deposit multiple build materials (and a sacrificial support material) within a single layer, which affords the opportunity to fabricate multimaterial parts. While other AM processes are more recently providing multimaterial capability, most instances are AM system modifications that amplify complexity. Extrusion systems, for example, offer multimaterial printing by increasing the number of extrusion heads but must account for additional controls to start and stop material flow and to ensure no collisions occur between nozzles and deposited materials. Powder-bed fusion and vat photopolymerization processes require much more intricate materials handling to keep powders or resins separate until needed for layer formation. MJ systems, on the other hand, feature multiple inkjet heads within the printing block that, due to separate cartridges and fluid lines, easily keep unique materials separate until jetted. When these materials are deposited simultaneously within the same print pass, multimaterial layers are achieved in a single build. In addition to the commercial offerings, research laboratories have also developed multimaterial jetting systems, such as the MultiFab system, which can deposit up to ten different materials simultaneously (Ref 12). Case 1 of Fig. 3 presents an example of a multimaterial MJ part that can be directly

printed: a kitchen spatula that features a rigid blade (material A) and a soft, conforming handle (material B). Due to the limitations of the stereolithography (STL) file format, a multimaterial component is fabricated via assembly of multiple STL files, wherein each file is dedicated to a different material assignment. The faces of the geometries represented in the STL files must be coincident for the different material regions to cure together during printing. Some MJ systems also allow the user to select from a set of predefined material compositions that vary the relative concentrations of two or more build materials. Similar to the manner in which grayscale images are converted to black and white in image processing for two-dimensional inkjet printing, MJ uses dithering algorithms to create intermediate properties from discrete droplets of different build materials. To disperse the two materials for this dithering effect, an algorithm assigns discrete droplets to predetermined locations to match the requested concentrations. There is no physical mixing of the materials involved; instead, base material droplets are jetted side by side, which causes the final part to behave similar to a composite material (Ref 13). Because of the high resolution of the MJ systems, the dithered part acts as a homogeneous material at the macro scale and is treated as a single material assignment. By changing the relative concentrations, one could make the aforementioned spatula blade slightly more flexible (e.g., 70/30 material A/B ratio, shown in case 2 of Fig. 3) or very flexible (e.g., 30/70 material A/B ratio, shown in case 3 of Fig. 3). While most MJ systems only allow the user to choose from predetermined material ratios, recent advancements allow users to precisely define the location and material of every jetted voxel (i.e., three-dimensional pixel) of a part. The advantage of this principle is the ability to design any multimaterial ratio to extend part performance capabilities. In this manner, materials can be assigned to form a slow gradient shift from one material to the other (e.g., case 4 of Fig. 3), instead of relying on the predefined

Table 1 Comparison of commercial polymer material jetting systems

Build size x
y
z, in.

Layer thickness, min in.

mm

Accuracy(a)

3D Systems ProJet MJP 5600 (Ref 6) 3D Systems ProJet MJP 3600 Max (Ref 7) Mimaki 3DUJ-553 Solidscape S500

20.4
15
11.8

0.0005

13

11.2
7.3
8

0.0006

16

20
20
12 6
6
4

0.0008 0.002

20 51

Stratasys Objet1000 Plus

39.3
31.4
19.6

0.0006

16

Stratasys J750

19.3
15.35
7.9 0.00055

14

±0.001–0.002 in. (25–50 mm) per inch of part dimension ±0.001–0.002 in. (25–50 mm) per inch of part dimension Within ±3% of part dimension ±0.005 in. (127 mm) for first inch, ±0.001 (25 mm) for each additional inch ±0.024 in. (610 mm) for large, rigid material builds ±0.008 in. (200 mm) or 0.06% of part length for large, rigid material builds

Commercial system

(a) The measurement of accuracy varies based on part geometry, part size, part orientation, and build parameters.

Build materials

Support material

Rigid opaque/transparent, rubberlike

Melt-away wax

Rigid opaque/transparent/blue, flexible, castable Colored rigid Wax, castable

Melt-away wax Water soluble Melt-away wax

Rigid opaque/transparent, rubberlike

Remove with high-pressure water

Rigid opaque/transparent, rubberlike, colored rigid and flexible

Remove with high-pressure water or soluble in water or sodium hydroxide

60 / Polymer Additive Manufacturing Processes

Fig. 4

Multicolor material jetting systems enable unsurpassed color matching compared to other additive manufacturing systems. Used with permission from Mimaki

Fig. 3

Schematic of multimaterial assignment applied to a spatula. Case 1 shows the two-part spatula, with each part designated as a unique base material A or material B. Cases 2 and 3 depict different concentrations of both base materials selectively distributed within the same region to create new homogeneous and compositelike materials. Case 4 illustrates a material gradient from the handle to the tip of the spatula.

composite material definitions. Thus, when the deposited materials have distinct mechanical properties, functionally graded mechanical properties are feasible that can be tailored to certain performance needs. This multifunctionality has been used in many areas, including biomedical and four-dimensional printing applications, and is driving future developments for MJ (see the section “The Future of Polymer MJ” in this article).

Multiple Colors The ability of MJ to selectively jet droplets of multiple materials also enables the fabrication of multicolored parts. Multiple colored variants of the same resin can be dithered together in all three dimensions to print millions of colors (Ref 14–16). For example, Mimaki’s commercial MJ system offers more than 10 million color combinations (Ref 8); an example of their highresolution color matching is provided in Fig. 4. The Stratasys J750 MJ platform offers its users certified Pantone color matching (Ref 17). Color remains a desired feature, especially for medical models, clothing and shoes, and finite-element stress or temperature fields.

higher because the jetted layer can be compressed by a rolling mechanism (e.g., as small as 13 mm in 3D Systems’ machines) or a planing mechanism (e.g., 50 mm in Solidscape systems). This small layer thickness provides printed parts with a much smoother surface finish compared to other AM systems (e.g., root mean square of 1.6 mm in Solidscape systems, Ref 9). The small droplet size also enhances the accuracy of final parts, making MJ a good choice for detailed prototypes and models.

Speed and Scalability Throughput of AM processes is often bottlenecked due to the use of a point-source scanning/deposition tool (e.g., laser, extrusion head), the need to heat (and then cool) large print chambers and/or raw materials, recoating/ spreading of the build material (e.g., powder, resin), and/or postprocess thermal or UV-cure treatments. Material jetting systems do not suffer from these process bottlenecks because:  Banks of print heads can print ~76 mm

(3 in.) wide bands of a layer per print pass.

 Thermal energy is not required.  Both material deposition and solidification

occur during the jetting pass.

High Resolution

 Parts are fully cured during printing.

Due to the small size of its printed primitives, MJ systems offer high resolution in all three orthogonal directions. Resolution in the xy-plane (i.e., the build plane) is primarily constrained by the diameter of the jetted droplet, which is typically in the range of 40 to 60 mm. The resolution in the z-plane (i.e., the layer thickness) is even

However, the speed gains of MJ systems from rapid layer deposition are somewhat muted due to its small layer thickness, which is roughly one-third the size of the layer thicknesses of other AM processes. Furthermore, although wide bands of material are deposited in each x-axis print pass, each printed band

requires multiple passes along the x-axis to complete the deposition and deliver sufficient UV energy to cure. Each pass is slightly offset along the y-axis to compensate for gaps between jetting nozzles. As a result, the throughput of MJ systems is similar to that of other polymer AM processes. Because the print block must travel along the x-axis in printing a layer of a part, it does not require substantially more time to continue the motion and print a layer of an adjacent part within the same band. In this way, total print time does not scale linearly with the number of objects on the build plate, as seen in other AM processes. Instead, print time scales with the width of components; adding additional parts with cross sections wider than the band (i.e., the width of the print head nozzles) increases the number of y-axis increments and consequently the total number of print passes. Ultimately, this increases total build time. Material jetting is a scalable process because additional print heads can be efficiently added to the systems without diminishing resolution. This results in facile scale up part size, material throughput, and printing speed. Furthermore, without the presence of heating or fusion functionality, large parts can be formed without concern of challenges caused by managing heat transfer and solidification distortion, as in other AM processes (e.g., extrusion, powder-bed fusion). Largescale MJ systems are commercially available, as seen in Stratasys’ Objet1000 Plus (39.3 by 31.4 by 19.6 in.) (Ref 10) and the Massivit 1800 (57 by 44 by 70 in.) systems (Ref 18). Research regarding impact of increases in material ejection and print head speed is limited because these parameters are not directly adjustable in commercially available MJ systems. Optimizing speed remains one of the complexities of jetting because a trade-off exists between print block speed, droplet velocity, and accurate placement and formation of deposited droplets. Speed also affects the flattening of droplets and the rate of curing.

Material Jetting of Polymers / 61

Materials Polymer MJ systems incorporate both build material and support material. Wax resins are designed to be fully castable, and wax support material is meant to be melted away or dissolved. As thermoplastics, waxy materials are meant to have a linear or branched molecular structure, which allows repeated melt and solidification cycles. Alternatively, thermoset MJ photopolymers rely on the cross linking of small molecules (monomers) into larger ones (polymer chains) via UV irradiation. Material jetting photopolymers consist of a photoinitiator and liquid monomers, which are either acrylate- or epoxy-based and are driven by free-radical or cationic photopolymerization, respectively. UV irradiation initiates the reaction that leads to the cross linking of polymer chains to form a strong covalently bonded network (Ref 4). While both MJ and vat photopolymerization processes involve photocuring polymers, MJ systems offer a more “office-friendly” process for operators. Compared to vat photopolymerization, in which materials must be carefully poured into a vat, MJ materials are stored in cartridges that are inserted into and extracted from the machine, thus minimizing contact with hazardous uncured liquids. Material jetting parts are also fully cured following printing. Vat photopolymerization parts, on the other hand, require additional postprocess UV curing and thus more care in handling, since operators should not be directly exposed to uncured resins. Because MJ parts are fully cured, no posttreatments are required outside of support cleaning. Regardless, protective gloves should be worn when handling MJ parts before they are cleaned.

Material Properties Commercially Available Material Properties In general, MJ photopolymers exhibit rigid (polypropylene-like) or flexible (elastomerlike) characteristics. ACEO, for example, uses

silicone in its custom DoD MJ system (Ref 19). Current commercial MJ materials offer a multitude of different material properties; a sampling of material offerings from Stratasys and 3D Systems is provided in Table 2. Multimaterial Properties Parts made from composite depositions of multiple materials yield mechanical properties falling in between those of the base build materials (Ref 22), thus expanding the spectrum of materials appropriate for mimicking composites or biocompatible materials. Table 3 provides a sampling of properties for multimaterial parts created with Stratasys’ VeroWhitePlus and TangoBlackPlus and 3D Systems’ Visijet CR-WT and Visijet CE-BK. As can be seen from Tables 2 and 3, a variety of properties are available; however, these ranges impose limits on the ability for MJ to produce end-use parts. Flexible MJ photopolymers mimic rubber with regard to shore hardness but do not replicate its strain or hysteresis behaviors. These materials still cannot compete with hydrogels for use as biomaterials, nor can they attain the structural durability of performance plastics. One of the main limiting factors is the low heat-deflection temperatures of photopolymers, as well as their sensitivity to light, heat, and humidity, which often makes the rigid polymers brittle. In addition, it is important to note that the shelf life of the resin (noted as an expiration date on the material cartridge) can impact material properties, as well as batch differences, material formulation changes, or age of the printed part (Ref 24). Fatigue Properties Similar to the rigid materials, the currently available soft materials for MJ only mimic the behavior of their traditional counterparts. Specifically, the soft material offerings are characterized in terms of hardness (Shore A values) and do not exhibit true elastomeric behavior. While not as well studied as the rigid materials, the printed soft materials have been shown to have limited strain and fatigue life. A relationship between strain and fatigue

life for a flexible material (Stratasys’ TangoBlackPlus) has been constructed in Fig. 5. It indicates that longer fatigue life may be achieved when specimens are maintained at elongations under 20%. In general, improved fatigue life relies on these low strains, minimized shear loading at multimaterial interfaces, and avoidance of necklike regions that end up inducing stress concentrations (Ref 25). Glossy surface finish, discussed more in the section “Part Design Considerations” in this article, has also been shown to improve fatigue life by 37% on average (Ref 26).

Material Design Considerations The performance of MJ materials is lacking when compared to other polymer AM material properties. This is predominately due to the physics of the jetting process, which impose a strict constraint on the viscosity, and thus the molecular weight (and commensurate mechanical properties), of the resin. The jettability of novel materials can be predicted by the inverse of the Ohnesorge (Oh) number of the fluid, a parameter termed Z (Ref 27): Z¼

1 Re ¼ pffiffiffiffiffiffiffi ¼ Oh We

pffiffiffiffiffiffiffiffi gra Z

(Eq 1)

where g is the surface tension, r is the density, a is a characteristic length, and Z is the dynamic viscosity. The Ohnesorge number describes resistance to spreading, and it can also be represented as a function of the Reynolds number (Re, defining laminar or turbulent flow) and the Weber number (We, defining impact- or capillary-driven spreading). Successful jetting is generally defined as when 1 < Z < 10. Generally, at higher Reynolds numbers, satellite drops could form, while at lower Reynolds numbers, the fluid is too viscous to jet. At high Weber numbers, droplets are prone to splashing, while at lower Weber numbers, there is insufficient energy for droplet formation (Ref 27). Therefore, a balance

Table 2 Comparison of commercial polymer material jetting material specifications

Tensile strength Company

Stratasys

Material

VeroWhitePlus, other Vero materials High Temperature TangoBlackPlus, TangoPlus Agilus30 3D Systems Visijet CR-BK Visijet CR-WT 200 Visijet CR-CL 200 Visijet CE-NT, Visijet CE-BK Source: Ref 20, 21

Color

White, blue, gray, cyan, magenta, yellow White Black, clear Black, clear Black White Clear Natural, black

Modulus of elasticity

Heat-deflection temperature (at 1.82 MPa, or 0.264 ksi)

MPa

ksi

MPa

ksi

Elongation at break, %

50–65

7–9

2000–3000

290–435

10–25

45–50 113–122

D: 83–86

3200–3500 ... ...

464–508 ... ...

10–15 170–220 220–270 5–11 12–22 14–22 160–230

55–57 131–135 ... ... ... ...

D: A: A: D: D: D: A:

70–80 10–12 0.8–1.5 0.1–0.2 2.4–3.1 0.4–0. 5 37–48 5–7 33–40 5–6 30–43 4–6 0.2–0.4 0.03–0.06

1800–2500 261–363 1500–2000 218–290 1400–2100 203–305 0.27–0.43 0.04–0.06



C



F

48–49 118–120 40–44 104–111 40–44 104–111 ... ...

Shore hardness, scale: value

87–88 26–28 30–35 78–83 77–80 77–80 27–33

62 / Polymer Additive Manufacturing Processes Table 3

Comparison of commercial polymer material jetting multimaterial properties from rigid to flexible

Stratasys

VeroWhitePlus RGD8505

Material Tensile strength, MPa (ksi) Flexural modulus, MPa (ksi) Tear resistance, kN/m (lbf/in.) Shore hardness, scale: value

50–65 (7–9)

40–60 (6–9)

2200–3200 (319–464)

RGD8510

RGD8515

RGD8520

RGD8525

RGD8530

FLX9895

FLX9885

40–60 (6–9)

40–60 (6–9)

40–60 (6–9)

35–45 (5–7)

29–38 (4–6)

8.5–10.0 (1.2–1.5)

5.0–7.0 (0.9–1.0)

...

...

1500–2500 1500–2500 1500–2500 1500–2500 1400–1800 1200–1500 (218–363) (218–363) (218–363) (218–363) (203–261) (174–218)

...

...

D: 83–86

...

D: 81.1–85.5

...

D: 81.1–85.5

...

D: 81.1–85.5

...

D: 81.1–85.5

...

D: 79.5–83.5

FLX9870 3.5–5.0 (0.5–0.7) ...

FLX9860 2.5–4.0 (0.4–0.6) ...

FLX9850 1.9–3.0 (0.3–0.4) ...

FLX9840 TangoBlackPlus 1.3–1.8 (0.2–0.3) ...

41–44 23–25 15.5–17.5 11–13 7.5–9.5 5.5–7.5 (0.23–0.25) (0.13–0.14) (0.09–0.10) (0.06–0.07) (0.04–0.05) (0.03–0.04)

D: A: 92–95 76.1–81.7

A: 80–85

A: 68–72

A: 57–63

A: 45–50

A: 35–40

0.8–1.5 (0.18–0.22) ...

2.0–4.0 (0.01–0.02)

A: 26–28

3D Systems Material Tensile strength, MPa (ksi) Flexural modulus, MPa (ksi) Tear resistance, kN/m (lbf/in.) Shore hardness, scale: value

CR-WT 200 RWT-EBK- RWT-EBK- RWT-EBK- RWT-EBK- RWT-EBK- RWT-EBK- RWT-EBK- RWT-EBK- RWT-EBK- RWT-EBK- RWT-EBK- CE-BK D75 D70 D65 D60 D55 A90 A80 A70 A60 A50 A40 33–40 19–27 12–16 8–10 4–5 2–3 1.4–1.9 1.3–1.7 0.75–1.1 0.48–0.77 0.35–0.48 0.23–0.32 0.2–0.4 (5–6) (3–4) (1.7–2.3) (1.2–1.5) (0.6–0.7) (0.3–0.4) (0.2–0.3) (0.19–0.25) (0.11–0.16) (0.07–0.11) (0.05–0.07) (0.03–0.05) (0.03–0.06) ...

1200–1700 (174–247)

450–750 (65–109)

350–550 (51–80)

150–250 (22–36)

70–180 (10–26)

30–80 (4–12)

...

...

...

...

...

...

D: 77–80

D: 70–80

D: 65–75

D: 60–70

D: 55–65

D: 50–60

...

...

...

...

...

...

44–62 25–32 18–23 11–17 6.6–9.3 6.5–8.5 3.1–3.7 (0.25–0.35) (0.14–0.18) (0.10–0.13) (0.06–0.10) (0.04–0.05) (0.04–0.05) (0.018–0.021)

A: 86–92

A: 75–85

A: 65–75

A: 55–65

A: 45–55

A: 35–45

A: 27–33

Source: Ref 20, 23

zx xz

z

cing

Pa

x y

yz

rt ppo t: su d Par move re

in x

a r t sp

xy yx

h

Pa r

nis e fi



tt Ma

ts pa

cin g

h

in

ssy

y

Glo

finis

ne rface pla e xz - rial int ate m i t l mu

Fig. 5



zy

e ne pla erfac xy - rial int ate ltim mu

Fatigue profile for Stratasys’ TangoBlackPlus, with the dashed lines denoting the 95% confidence interval. Source: Ref 25

must be struck between these parameters in order to achieve a jettable material. Additional considerations for designing printable MJ inks include:



 Process

parameters: Droplet velocity, droplet frequency, and the speed of the print head are coupled to deliver droplets without splashing and unnecessary spreading.



A trade-off exists between speed of printing and accuracy. Rheology: The rheological properties of the fluid, especially viscosity, determine the characteristics of droplet formation, including what process parameters are necessary to develop repeatable, spherical droplets without the formation of satellite droplets or long trails behind the main droplet that could impede deposition accuracy. Nozzles: The nozzle diameter dictates resolution and jettability, and the distance between the nozzles and substrate, usually a few millimeters, plays a role in droplet flight. Additionally, for print heads with hundreds of nozzles, even a few clogged nozzles can affect dimensional accuracy and mechanical properties of parts. Automated purging and nozzle-cleaning routines are used to mitigate errors from clogged nozzles. Mimaki offers in situ monitoring of the nozzles as well as compensation for clogs, but this technique is not implemented with other systems. Material compatibility: The ink must suitably wet the printing substrate as well as previously deposited layers. Good surface tension is a significant factor to ensure droplets remain stable in their deposited location. Solidification/curing rate: The ink droplets must coalesce and then phase change from

Material Jetting of Polymers / 63 liquid to solid through curing or evaporation. This must not initiate during droplet flight but must occur after impact at a critical time during the droplet spreading (Ref 28). Solidification must be uniform and quick to ensure that material is precisely placed. Along with expanding material offerings for advanced applications, manufacturing systems need extended flexibility to tune process parameters. Commercial systems significantly limit a user’s ability to address droplet velocity, print speed, and spacing between the nozzles and build plate. Some commercial systems require the sole use of their proprietary resins, and further complexity arises when their firmware or material compositions are altered without notification to the consumer. Most research in novel jetting materials involves a single print head or nozzle on a gantry that allows adaptability of the droplet frequency and other parameters.

Part Design and Processing Considerations Clear understanding of process-structureproperty relationships is critical for AM processes because they influence adoption of the technology, qualification and certification, and thus design considerations and final application space. The distinct capabilities (e.g., discrete jetting of multimaterial droplets) of MJ are accompanied by unique process-imposed effects on final part properties and overall quality. Detailed in this section are these process-related effects as well as corresponding design considerations.

Process Impact on Material Properties Orientation As already presented, MJ materials and multimaterial counterparts offer a spectrum of mechanical properties (Tables 2 and 3). Studies of MJ photopolymer tensile properties are well covered in published literature, and discrepancies in published mechanical properties have arisen due to differences in material, sample geometry, processing, and testing variation. Here, general trends for photopolymer MJ are discerned. Twelve process-related factors (ranging from positioning to timing) have been explored through a design of experiments, and interlayer and intralayer effects have been shown to most significantly influence mechanical properties (Ref 24); these effects are directly correlated to print orientation. Figure 6 provides a sampling of build orientations for reference. Rigid and flexible tensile specimens are weakest in the zx-orientation (such that tensile loads are placed across printed layers), due to poorer interlayer adhesion. Tensile specimens exhibit

slight weakness in the yx-orientation because of the many nozzle paths causing intralayer delamination. Specimens in the xy-orientation are generally strongest because the tensile load aligns with the jetting direction (Ref 22). Furthermore, under some conditions, specimens oriented in xz are strongest (Ref 29), although this orientation choice increases print time. Rigid xy specimens typically have the lowest modulus of elasticity and the highest elongation at break (Ref 22), which intuitively makes sense when considering how loading is perpendicular to the printed layers, enabling additional strain before failure. Part Spacing Spacing of parts in the build plane, illustrated in Fig. 6, likewise has a statistically significant effect on the mechanical properties of printed parts. Rigid parts printed closer together have an increased tensile strength and modulus of elasticity and reduced elongation (Ref 29), which indicates material strengthening coupled with becoming more brittle. This effect has been attributed to the additional UV exposure that parts receive when printed closer together (due to the overlapping UV irradiation from the print block UV lamps between printing passes), which can lead to overcuring and embrittlement of the acrylate material (Ref 24, 29). Aging Materials have been shown to exhibit timedependent mechanical properties regardless of storage condition (Ref 22, 24). Specifically, parts appear to cure further over time, which is typical of printed photopolymers. For rigid tensile specimens, the ultimate tensile strength of printed parts increases as parts age; statistically significant increases are observed after five weeks for specimens both exposed to laboratory lighting conditions (with a small UV dosage) and confined to a dark desiccant box.

Fig. 6

Similarly, elongation at break decreases over time, statistically significant after five weeks (Ref 22). This behavior stresses the importance of mechanical testing parts at a consistent time after manufacturing. Multimaterial Interfaces Similar in the field of composites, adhesion at multimaterial interfaces is a critical aspect that determines the performance capabilities of the structure. For multimaterial parts with discrete regions of flexible and rigid materials (e.g., case 1 in Fig. 3), the interface between the materials proves as strong as the compliant material during tensile testing (Ref 30). However, this interface occasionally fails before the softer elastomer-like material in fatigue load cases (Ref 25). The strength of interfaces between rigid and composite material sections (e.g., cases 2 and 3 in Fig. 3) varies depending on print orientation; however, in general, interfacial failures have reduced strength by up to 50% (Ref 30). Examining crack behavior at the multimaterial interface has provided insight into fracture toughness for different interfacial designs, although results are highly dependent on specimen shape, strain rate, and orientation. Fracture toughness is highest when the multimaterial interface is designed in the xy-plane, as referenced in Fig. 6. The interfacial fracture pattern resembles mud cracking. Specimens printed with the multimaterial interfacial plane in the xz-orientation see lower fracture toughness, and the fracture pattern remains solely within the flexible material, indicating a weaker material relative to the interface (Ref 31). Visual inspection and nanoindentation also confirm higher interfacial bond strength relative to the weakest base material. The study also finds that a rigid-flexible interface spans approximately 40 mm to get from pure rigid material to pure flexible material; furthermore, a smooth transition in hardness

Representation of various material jetting layout and processing decisions, including orientation, spacing, glossy or matte surface finish, and orientation of a multimaterial interface

64 / Polymer Additive Manufacturing Processes and modulus of elasticity is observed over that interface (Ref 13). This knowledge of interfacial interaction will help to guide future multiphysics models for predicting multimaterial part performance; one such preliminary study involves using finite-element analysis to model strains of functionally graded MJ parts (Ref 32). Some models have been developed to enable the design of multimaterial patterns with tailored fracture properties (Ref 33). However, the nature of multimaterial interfaces relies on the full understanding of all significant factors. While orientation has been investigated more thoroughly, the viscoelasticity of the materials, among other factors, also contributes to the complexity of evaluating interfacial behavior (Ref 31).

Part Design Considerations This section provides design-for-AM considerations specifically for MJ. These guidelines serve as design, build plate configuration, and handling tips to reduce print time, maintain geometric accuracy, and ensure the highest chance of achieving optimal properties. Orientation When a part is loaded onto an MJ system, the user can translate or rotate the part on the virtual build plate of the software. Usually, parts are oriented to save time and/or material. By default, time-savings is prioritized by placing the smallest dimension along the z-axis and aligning multiple parts along the x-axis to minimize the total number of printing passes. Material consumption can be minimized by orienting the part to reduce the volume under overhanging features. From the discussed orientation effects on mechanical properties (see the section “Process Impact on Material Properties” in this article), it is generally recommended to orient longer dimensions and highest load-bearing dimensions along the jetting direction (x-axis). Because systems have different resolutions in each dimension, orientation decisions could impact minimum feature size, and surfaces with fine features are often oriented as the topmost surface. Matte versus Glossy Finish Stratasys offers the option to select matte or glossy finish, which modifies the support material distribution. As previously stated, a typical MJ part has every face coated with support material, which gives the surfaces a uniform surface roughness and aesthetic quality (Ref 26). Stratasys terms this matte finish, which is illustrated in Fig. 6, where the part in the middle of the build tray prior to cleaning is fully encased in translucent support. However, because support material is not needed structurally on the topmost surface, the user can instead choose to leave the top surface free of support material (glossy finish), also shown in Fig. 6. This saves print time and material

because it reduces the overall build height. Glossy mode is best if a smooth, shiny top surface is desirable, although matte finish is typically selected to provide a uniform surface finish for the entire part. Build Layout To increase geometric accuracy (Ref 24) and to reduce print time (by reducing total print block travel distance), parts are typically placed in the corner of the MJ build plate closest to the print head homing location. Printing multiple parts as a batch is achieved by locating the parts across the MJ build tray. Unlike powder-bed AM systems, it is not advised to stack parts in the z-direction because the stacked parts suffer from lower resolution due to error propagation from being built on a soft support material matrix. As discussed in the sections “Process Characteristics” and “Process Impact on Material Properties” in this article, part spacing on the MJ build tray has a significant, direct impact on build time and mechanical properties, respectively. Therefore, designers should place parts closely spaced (default automatic placement) to maximize print time and accuracy but should consider spaced-out layouts for uniform UV exposure if material properties are a prominent concern. Photopolymer Support Removal Although using high-pressure water is most efficient for quick support removal, consideration must be taken for feature survivability and internal cavities (Ref 34). Even though fine geometric features may successfully print, they may not survive the pressurized water cleaning. The minimum survivable feature depends primarily on the material used; for nonsoluble support, VeroWhitePlus features greater than 1 mm (0.04 in.) survive pressurized water cleaning, while TangoBlackPlus features greater than 3 mm (0.12 in.) are needed to survive cleaning (Ref 34). To design for MJ, certain geometric features should be avoided. For example, an internal cavity lacking access to the external surface will fill with support material that will not be removable. Another feature of concern is a long tube or channel. Depending on the internal diameter, wall thickness, and length of tube, the pressure from the waterjet could damage the part, and tools may not be adequate for reaching all remaining support material. Soluble supports can also take a long time to dissolve in long, small-diameter channels. Finally, complex geometries must be designed with support removal in mind. A part with intricate internal cavities and thin walls poses the risk for pieces of support to be stuck in hard-to-reach places and would likely need at least some manual cleaning. Multimaterial When designing parts composed of multiple materials, the results in the section “Process Impact on Material Properties” in this article

indicate that factors including orientation and materials selection drive interfacial performance. It is recommended to orient a part such that loading is placed perpendicular to the material interface if high fracture toughness is desired (Ref 30), although this sometimes leads to interfacial failure (Ref 31). Alternatively, orienting a part with loading not across the material interface will have lower fracture toughness but higher interfacial strength (i.e., failure will occur at the weaker material) (Ref 31). When printing with composite materials (i.e., a dithered combination of two or more materials; Table 3), it is important to consider that there is an added resolution constraint on designed features. Due to the use of dithering algorithms, there exists a spatial limit at which the dithering of discrete droplets cannot effectively represent the desired global composite material properties. Specifically, these microlevel resolution constraints, rising from the interaction of material droplets with different properties, result in the ineffectiveness of dithered multimaterial regions under ~5 mm (0.2 in.) (Ref 13). The use of voxel-level control to functionally grade the multimaterial interface can improve the interfacial strength. By using voxel patterns to generate stepwise or continuously graded interfaces, better fatigue performance is found with shorter gradient lengths overall, although longer stepwise gradients also increase fatigue life (Ref 35). Continued research is needed to simulate, predict, and design effective interfacial voxel patterns with regard to fracture toughness. Voxel-controlled deposition remains a limited feature for photopolymer jetting, but the design technique has potential for expanding future multifunctional MJ capability. Changing the dithering ratio of two base materials in a part alters the distribution pattern of the materials. Sometimes, this distribution pattern causes the multimaterial region to behave like the dominant base material (a local material property) instead of the desired multimaterial characteristics (a global material property) (Ref 13, 36). Additionally, adjusting voxel sizes and patterns directly impacts the mechanical properties of the resulting composite structure (Ref 37).

Applications Due to the extensive capabilities offered by MJ, a broad spectrum of applications exists. However, the material limitations imposed by the process physics (discussed in the section “Materials” in this article) have thus far limited the use of MJ to mainly modeling, prototyping, and tooling applications. Wax MJ parts have primarily been used in jewelry and dental industries, and much of the more recent work in photopolymer MJ has gravitated toward biomedical applications. Adoption into

Material Jetting of Polymers / 65 fashion and animation extends the influence of MJ to the arts. Electronics and smart materials applications take advantage of the multimaterial, multifunctional abilities afforded by the technology and are motivating future application spaces (discussed in the section “The Future of Polymer MJ” in this article).

Concept Modeling and Prototyping Due to its ability to print detailed, large parts with smooth surface finish, MJ has been frequently used to quickly produce prototypes to aid designers in the iterative process of product realization. The use of an MJ system allowed Trek’s designers to rapidly prototype and test a large variety of aerodynamic bicycle frame cross sections via wind tunnel testing. Lead times for these prototypes were less than one day, compared to a week or more for machined parts and several days for vat photopolymerization parts (Ref 38). The added functionality of printing multiple materials enables designers to explore product designs featuring both soft and hard materials of various Shore durometers before committing to the expensive tooling found in injection molding and overmolding processes. Similarly, multicolor printing is useful in visualizing the final product design aesthetic. For example, the ability to quickly generate accurate multimaterial taillight prototypes has allowed Audi to reduce design time by up to 50%, since traditionally, taillights would be milled or injection molded. As shown in Fig. 7(a), their taillights can be made to tolerance with exact, transparent colors (Ref 39).

whereas the same wax pattern can be printed in a matter of hours or days (Ref 9).

Anatomical Modeling High-precision wax printers have also been used in the dental market for printing models of crowns and bridges. Models of complex dental cases assist visualization and planning. Photopolymer MJ model accuracy has surpassed that of all other AM processes in dentistry (Ref 42), and Stratasys and 3D Systems have dedicated materials particularly for dental and orthodontics modeling and end-use parts such as custom aligners (Ref 43). Multiple aligners can be arranged on a single build plate for efficient print times. Physical medical models provide a handson method to understand complex anatomy, practice or plan interventional procedures, and communicate with other medical professionals, patients, and families (Fig. 7c). High-resolution MJ models from magnetic resonance imaging data have been shown to reflect accuracy of anatomy and are used as preoperative training tools for medical education (Ref 44). Multicolor MJ improves communication in models such as an upper limb, where colors effectively distinguish muscles, bones, arteries, and nerves (Ref 45). In addition, the multimaterial capability of MJ enables printing of more realistic medical models. Starting from computed tomography images, Maragiannis et al. represented calcium distributions as rigid material within a

clear and flexible arterial section (Ref 46). By dithering materials of different hardnesses at the voxel level, Vukicevic et al. more accurately mimicked the mechanical properties of the mitral valve within a cardiovascular model. The model was used as a tool for patient-specific surgical planning and communication (Ref 47). Although multimaterial parts exhibit more realistic characteristics, it has been shown that the limited material property space in MJ (Ref 22) fails to match the flexural properties for certain human tissues (Ref 48).

Art and Design Drop-on-demand jetting and multimaterial/ color capability facilitates control of color and material composition at the voxel level, which provides the extraordinary flexibility for designing complex structures and material combinations, since both can be assigned in threedimensional space with high resolution and precision. Designers have used these unique capabilities in fashion, such as Ganit Goldstein’s vibrantly colored printed shoes with voxel-level patterns (Ref 49). LAIKA, a stop-motion animation studio, has used Stratasys’ multimaterial printing capability to rapidly print the faces of their stop-motion animation puppets (Fig. 7d). To animate the characters, thousands of unique facial expressions are designed and printed for each character and are (re)placed onto the puppet’s head for each still-image frame capture (Ref 50). With an eye toward multifunctionality,

Lost Wax Casting—Jewelry and Dental Applications Wax-based MJ systems have predominately been used to fabricate detailed sacrificial patterns for lost wax (i.e., investment casting) applications. In the lost-wax casting process, a printed wax pattern is first covered in a ceramic slurry. Once dried, the resulting part is placed into a furnace to sinter the ceramic shell and melt the wax pattern. The resulting hollow ceramic shell is then filled with molten metal. Once solidified, the ceramic shell is broken to reveal the metal casting. Solidscape markets a line of wax MJ systems specifically designed for fabricating jewelry patterns because the demand of the market for fine features and high-quality surface finish can be met via the high resolution and small layer thickness of MJ; an example is presented in Fig. 7(b). The jetted wax materials are advantageous because they offer good surface finish and a clean burnout, which translates to accurate and consistent casted parts. Additionally, the printing of wax molds significantly speeds the investment casting process, since the creation of patterns and wax copies via traditional tooling processes can take more than six weeks,

Fig. 7

Sample applications of material jetting. (a) Audi uses material jetting to rapidly prototype taillights (Ref 39). Courtesy of Stratasys. (b) Material jetting complex geometries with wax leads to high-resolution jewelry (Ref 40). Courtesy of Solidscape. (c) Cardiovascular model from a patient’s computer tomography scans used for presurgical planning. Courtesy of J. Foerst, Carilion Clinic, Roanoke, VA. (d) LAIKA designs thousands of faces per character in stop-motion animation (Ref 41). Courtesy of LAIKA

66 / Polymer Additive Manufacturing Processes Dr. Neri Oxman and Dr. W. Craig Carter designed a unique chaise lounge with attention to aesthetics and an acoustically quiet environment. The chaise lounge incorporates both voxel-controlled photopolymer materials and milled wood (Ref 51).

Tooling and End-Use Products High resolution, good surface finish, and the ability to mimic polypropylene are some of the attractions for using MJ for tooling and enduse products. Recognizing the opportunity for lighter materials, assembly reduction, and end-use parts, Nissan Motorsports has manufactured functional, complex geometries of a Gurney flap and air pump housing to improve aerodynamics and cooling in its V8 Supercars (Ref 52). Daimler Trucks uses MJ to fabricate go/no-go gages for quality assurance in preproduction (Ref 53). Overall, MJ for end-use products exists in limited capacity compared to other AM processes. Materials selection, machine reliability, part accuracy, and part consistency remain critical challenges that the MJ community must continue to address for the technology to be considered for widespread production of multifunctional parts in such applications as automotive or aerospace. It is expected that the continued development of multifunctional materials and advanced systems will improve MJ such that new materials and techniques will more frequently become desired alternatives to traditional manufacturing practices.

The Future of Polymer Material Jetting Polymer MJ systems offer easy-to-use, officefriendly machines that print accurate, consistent, fully cured parts with high feature resolution, good surface finish, and many color options. MJ systems are inherently scalable; without the need to manage thermal energy, the systems can have large build volumes and increased throughput by simply increasing the number of inkjet modules. Multimaterial modes provide the opportunity to create multifunctional parts with discrete or functionally graded mechanical properties throughout the structure. Despite these strengths, the lack of jettable engineering-grade materials and consequently the small spectrum of available material properties have thus far prevented the use of MJ to fabricate many functional end-use parts. Material jetting currently finds its value with quick, geometrically accurate prototypes under little loading; tooling/injection molding; and multimaterial/functionally graded parts. For polymer MJ to be implemented as a manufacturing technology, focus must be placed especially on expanding materials selection and performance. Recent research efforts in expanding MJ material capability

Fig. 8

Origami airplanes produced by material jetting. The black material is Stratasys’ TangoBlackPlus, and the white material is Stratasys’ VeroWhitePlus. (a) A flat, three-hinged structure, when cooled, folds into (b), an airplane with a 0 angle in the middle hinge and 90 angles for the side hinges. (c) A flat, five-hinged structure, when cooled, folds into (d), an airplane similar to (b) but with two additional winglets. Both structures return to flat orientation when heated. Source: Ref 55. Used by permission from the authors

and leveraging the voxel-level multimaterial control have resulted in unique functionality that points toward a bright future as a versatile fabrication platform. Efforts such as:  Tunable structures: Shape memory poly-

mers, often considered smart materials, change shape in response to a stimulus such as temperature or light. Stratasys’ MJ materials have been found to exhibit shape memory characteristics. The materials exhibit a large transition region centered around the glass transition temperature (Ref 54), permitting deformation through a wide range of temperatures. Figure 8 displays this concept applied to four-dimensional printing (when a material responds to an external field and changes shape) (Ref 55). In this case, the temperature is such that the flexible material changes shape while the rigid material remains unaffected, thus enabling controlled shape change. With multimaterial patterning, MJ can generate tailored metamaterials used for robotics (Ref 56) and auxetic structures (Ref 57).  Printed hybrid electronics: Although most direct jetting of conductive traces for printed electronics applications involves jetting metal or carbon nanosuspensions, polymer field-effect transistors have been fabricated (Ref 58). Off-the-shelf electronic components (e.g., sensors, batteries, antennae) have been embedded in multimaterial MJ parts to directly fabricate parts with embedded sensing and actuation (Ref 59). Hydraulically actuated robots have also been directly printed via Stratasys’ MJ systems by leveraging the multimaterial capability and jetting the cleaning fluid of the system into enclosed

volumes to serve as the fluid power (Ref 60). In addition, 3D Systems’ materials have been used in the fabrication of fluidic capacitors, diodes, and transistors (Ref 61).  New jettable inks: Most novel material development has occurred on custom machines that provide flexibility to fine-tune process parameters. Research in new materials has included jettable support (Ref 62), polyimide dielectric ink for direct fabrication of printed circuit boards (Ref 63), electromagneticresponsive ink (Ref 64), tablets with tailored drug release (Ref 65), and liquid latex with micronized rubber powder (Ref 66). ACKNOWLEDGMENTS L. Bezek is supported by the National Science Foundation Graduate Research Fellowship Program under Grant No. 1324586. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation. REFERENCES 1. XJet, https://xjet3d.com 2. “Disrupting Manufacturing with 3D Printing,” Xerox Corporation, https://www. xerox.com/en-us/innovation/insights/additive-manufacturing-3d-printing 3. Nano Dimension, https://www.nano-di.com 4. I. Gibson, D. Rosen, and B. Stucker, Additive Manufacturing Technologies—Rapid Prototyping to Direct Digital Manufacturing, Springer, New York, 2010 5. “Guide to Basic Post-Printing Processes for PolyJet 3D Models,” Stratasys Ltd.,

Material Jetting of Polymers / 67

6. 7. 8. 9. 10. 11. 12.

13.

14.

15.

16. 17.

18. 19. 20.

21.

22.

23.

http://intelligy.com/archivos/guia-postporceso-impresion-modelos-polyjet.pdf “ProJet MJP 5600,” 3D Systems, https:// www.3dsystems.com/3d-printers/projet-mjp5600 “ProJet MJP 3600 Series,” 3D Systems, https://www.3dsystems.com/3d-printers/ projet-mjp-3600-series “3DUJ-553,” Mimaki Global, https:// mimaki.com/product/3d/3d-inkjet/3duj-553/ specification.html Solidscape, Inc., https://www.solidscape. com/high-precision-investment-castingwax-3d-printer/ “Objet1000 Plus,” Stratasys Ltd., https:// www.stratasys.com/3d-printers/objet1000plus J735 and J750, Stratasys Ltd., https:// www.stratasys.com/3d-printers/j735-j750 P. Sitthi-Amorn et al., MultiFab: A Machine Vision Assisted Platform for Multi-Material 3D Printing, ACM Trans. Graph., Vol 34 (No. 4), July 2015, Article 129; pages 1–11 J. Mueller, D. Courty, M. Spielhofer, R. Spolenak, and K. Shea, Mechanical Properties of Interfaces in Inkjet 3D Printed Single- and Multi-Material Parts, 3D Print. Addit. Manuf., Vol 4 (No. 4), Dec 2017, p 193–199 A. Brunton, C.A. Arikan, T.M. Tanksale, and P. Urban, 3D Printing Spatially Varying Color and Translucency, ACM Trans. Graph., Vol 37 (No. 4), July 2018, p 1–13 A. Brunton, C.A. Arikan, and P. Urban, Pushing the Limits of 3D Color Printing: Error Diffusion with Translucent Materials, ACM Trans. Graph., Vol 35 (No. 1), 2015, p 1–13 “Cuttlefish: The Universal 3D Printing Driver,” Fraunhofer IGD, https://www. cuttlefish.de/ “Investor Relations,” Stratasys Ltd., http:// investors.stratasys.com/news-releases/newsrelease-details/stratasys-and-pantone-accelerate-design-realism-rapid “3D Printers,” Massivit 3D, https://massivit3d.com/solutions/3d-printers/ “ACEO—3D Printing with Silicones,” Wacker Chemie AG, https://www.aceo3d. com/3d-printing/ “ProJet MJP 5600,” 3D Systems, https:// www.3dsystems.com/sites/default/files/ 2018-11/3d-systems-mjp-5600-tech-specsa4-us-2018-11-13-a-web.pdf “Our Materials,” Stratasys Ltd., https:// www.stratasys.com/materials/search?technologies=731e07a1a51b42419acd1cb7514 2dfe6&sortIndex=0 L.B. Bass, N.A. Meisel, and C.B. Williams, Exploring Variability of Orientation and Aging Effects in Material Properties of Multi-Material Jetting Parts, Rapid Prototyp. J., Vol 22 (No. 5), Aug 2016, p 826–834 “Digital Materials Data Sheet,” Stratasys Ltd., https://www.stratasys.com/-/media/ files/material-spec-sheets/mss_pj_digitalmaterialsdatasheet_0617a.pdf

24. J. Mueller, K. Shea, and C. Daraio, Mechanical Properties of Parts Fabricated with Inkjet 3D Printing through Efficient Experimental Design, Mater. Des., Vol 86, Dec 2015, p 902–912 25. J.P. Moore and C.B. Williams, Fatigue Properties of Parts Printed by PolyJet Material Jetting, Rapid Prototyp. J., Vol 21 (No. 6), 2015, p 675–685 26. L.K. Gillespie, Design for Advanced Manufacturing: Technologies and Processes, McGraw-Hill Education, New York, NY, 2017 27. B. Derby, Additive Manufacture of Ceramics Components by Inkjet Printing, Engineering, Vol 1 (No. 1), March 2015, p 113–123 28. K. Picha, C. Spackman, and J. Samuel, Droplet Spreading Characteristics Observed during 3D Printing of Aligned Fiber-Reinforced Soft Composites, Addit. Manuf., Vol 12, Oct 2016, p 121–131 29. M.W. Barclift and C.B. Williams, “Examining Variability in the Mechanical Properties of Parts Manufactured via PolyJet Direct 3D Printing,” International Solid Freeform Fabrication Symposium (Austin, TX), Aug 6–8, 2012 30. T.S. Lumpe, J. Mueller, and K. Shea, Tensile Properties of Multi-Material Interfaces in 3D Printed Parts, Mater. Des., Vol 162, Jan 2019, p 1–9 31. I.Q. Vu, L.B. Bass, C.B. Williams, and D. A. Dillard, Characterizing the Effect of Print Orientation on Interface Integrity of Multi-Material Jetting Additive Manufacturing, Addit. Manuf., Vol 22, Aug 2018, p 447–461 32. E. Salcedo, D. Baek, A. Berndt, and J.E. Ryu, Simulation and Validation of Three Dimension Functionally Graded Materials by Material Jetting, Addit. Manuf., Vol 22, Aug 2018, p 351–359 33. L.S. Dimas, G.H. Bratzel, I. Eylon, and M.J. Buehler, Tough Composites Inspired by Mineralized Natural Materials: Computation, 3D Printing, and Testing, Adv. Funct. Mater., Vol 23 (No. 36), Sept 2013, p 4629–4638 34. N. Meisel and C. Williams, An Investigation of Key Design for Additive Manufacturing Constraints in Multimaterial Three-Dimensional Printing, J. Mech. Des., Vol 137 (No. 11), Oct 2015, 111406 35. D.V. Kaweesa and N.A. Meisel, Quantifying Fatigue Property Changes in Material Jetted Parts due to Functionally Graded Material Interface Design, Addit. Manuf., Vol 21, May 2018, p 141–149 36. N.A. Meisel, D.A. Dillard, and C.B. Williams, Impact of Material Concentration and Distribution on Composite Parts Manufactured via Multi-Material Jetting, Rapid Prototyp. J., Vol 24 (No. 5), July 2018, p 872–879 37. D.V. Kaweesa and N.A. Meisel, Material Property Changes in Custom-Designed

38.

39.

40.

41.

42.

43. 44.

45.

46.

47.

48.

49.

Digital Composite Structures due to Voxel Size, Int. Solid Free. Fabr. Symp., 2018, p 1499–1510 “Trek Wind Tunnel 3D Printing Case Study,” Stratasys Ltd., https://www.stratasys.com/resources/search/case-studies/ trek-wind-tunnel?resources=cd10362db310-4078-a6ea-82a25813ef06 “Audi to Save 50% on Lead Times by Switching to 3D Printing for Prototyping,” 3D Printing Industry, https://3dprintingindustry.com/news/audi-save-50-lead-timesswitching-3d-printing-prototyping-134355/ “Solidscape Announces Winners of 2017 Baselworld 3D Printed Jewelry Competition,” 3D Printing Media Network, https:// www.3dprintingmedia.network/solidscapetannounces-winners-2017-baselworld-3dprinted-jewelry-design-competition/ “How LAIKA Used 3D Printing and Cuttlefish Software to Bring ‘Missing Link’ to Life,” 3D Printing Media Network, https://www.3dprintingmedia.network/ laika-3d-printing-cuttlefish-missing-link/ S.-Y. Kim, Y.-S. Shin, H.-D. Jung, C.-J. Hwang, H.-S. Baik, and J.-Y. Cha, Precision and Trueness of Dental Models Manufactured with Different 3-Dimensional Printing Techniques, Am. J. Orthod. Dentofacial Orthop., Vol 153 (No. 1), Jan 2018, p 144–153 “Stratasys J700 Dental,” Stratasys Ltd., https://www.stratasys.com/3d-printers/j700dental J.P. Costello et al., Utilizing ThreeDimensional Printing Technology to Assess the Feasibility of High-Fidelity Synthetic Ventricular Septal Defect Models for Simulation in Medical Education, World J. Pediatr. Congenit. Heart Surg., Vol 5 (No. 3), July 2014, p 421–426 S.R. Mogali et al., Evaluation by Medical Students of the Educational Value of Multi-Material and Multi-Colored ThreeDimensional Printed Models of the Upper Limb for Anatomical Education: 3D Printed Upper Limb in Anatomical Education, Anat. Sci. Educ., Vol 11 (No. 1), Jan 2018, p 54–64 D. Maragiannis et al., Replicating PatientSpecific Severe Aortic Valve Stenosis with Functional 3D Modeling, Circ. Cardiovasc. Imaging, Vol 8 (No. 10), 2015, e003626 M. Vukicevic, D.S. Puperi, K.J. GrandeAllen, and S.H. Little, 3D Printed Modeling of the Mitral Valve for Catheter-Based Structural Interventions, Ann. Biomed. Eng., Vol 45 (No. 2), Feb 2017, p 508–519 G. Biglino, P. Verschueren, R. Zegels, A. Taylor, and S. Schievano, Rapid Prototyping Compliant Arterial Phantoms for InVitro Studies and Device Testing, J. Cardiovasc. Magn. Reson., Vol 15 (No. 2), 2013 “Stratasys Unveils New 3D Printed Woven Shoes Designed by Fast-Rising

68 / Polymer Additive Manufacturing Processes

50. 51.

52.

53.

54.

Fashion Design Graduate Ganit Goldstein,” WhichPLM, https://www.whichplm. com/stratasys-unveils-new-3d-printedwoven-shoes-designed-by-fast-rising-fashion-design-graduate-ganit-goldstein/ LAIKA, https://www.laika.com/videos/ behind-the-scenes S. Keating, Beyond 3D Printing: The New Dimensions of Additive Fabrication, Designing for Emerging Technologies: UX for Genomics, Robotics, and the Internet of Things, O’Reilly Media, 2014, p 379–405 “Australia’s Nissan Motorsports Uses Additive Manufacturing to Rev Up Performance,” 3D Systems, https://www.3dsystems.com/learning-center/case-studies/ australias-evok3d-uses-direct-digital-manufacturing-rev-performance “Daimler Trucks Revs Up Design Verification with MultiJet Printing,” 3D Systems, https://www.3dsystems.com/learning-center/case-studies/daimler-trucks-drives-projet-5000-design-verification M. Wagner, T. Chen, and K. Shea, Large Shape Transforming 4D Auxetic Structures, 3D Print. Addit. Manuf., Vol 4 (No. 3), Sept 2017, p 133–142

55. Q. Ge, C.K. Dunn, H.J. Qi, and M.L. Dunn, Active Origami by 4D Printing, Smart Mater. Struct., Vol 23 (No. 9), Sept 2014, 094007 56. T. Chen, O.R. Bilal, K. Shea, and C. Daraio, Harnessing Bistability for Directional Propulsion of Soft, Untethered Robots, Proc. Natl. Acad. Sci., Vol 115 (No. 22), May 2018, p 5698–5702 57. K. Wang, Y.-H. Chang, Y. Chen, C. Zhang, and B. Wang, Designable DualMaterial Auxetic Metamaterials Using Three-Dimensional Printing, Mater. Des., Vol 67, Feb 2015, p 159–164 58. Y. Liu, K. Varahramyan, and T. Cui, LowVoltage All-Polymer Field-Effect Transistor Fabricated Using an Inkjet Printing Technique, Macromol. Rapid Commun., Vol 26 (No. 24), 2005, p 1955–1959 59. R.L. Dumene, P. Kennedy, C.B. Williams, D. Sweeney, and G. Earle, Creating Embedded Radiofrequency Structures Using Polyjet Material Jetting, Int. Solid Free. Fabr. Symp., 2015, p 1769–1787 60. R. MacCurdy, R. Katzschmann, Y. Kim, and D. Rus, Printable Hydraulics: A Method for Fabricating Robots by 3D Co-Printing Solids and Liquids, 2016 IEEE

61.

62.

63.

64.

65.

66.

International Conference on Robotics and Automation (ICRA), 2016, p 3878–3885 R.D. Sochol et al., 3D Printed Microfluidic Circuitry via Multijet-Based Additive Manufacturing, Lab. Chip, Vol 16 (No. 4), Feb 2016, p 668–678 Y. He et al., A Tripropylene Glycol Diacrylate-Based Polymeric Support Ink for Material Jetting, Addit. Manuf., Vol 16, Aug 2017, p 153–161 F. Zhang et al., Reactive Material Jetting of Polyimide Insulators for Complex Circuit Board Design, Addit. Manuf., Vol 25, Jan 2019, p 477–484 E. Saleh et al., 3D Inkjet-Printed UVCurable Inks for Multi-Functional Electromagnetic Applications, Addit. Manuf., Vol 13, Jan 2017, p 143–148 M. Kyobula et al., 3D Inkjet Printing of Tablets Exploiting Bespoke Complex Geometries for Controlled and Tuneable Drug Release, J. Control. Release, Vol 261, Sept 2017, p 207–215 M.A. Quetzeri-Santiago, C.L. Hedegaard, and J.R. Castrejo´n-Pita, Additive Manufacturing with Liquid Latex and Recycled End-of-Life Rubber, 3D Print. Addit. Manuf., Vol 6 (No. 3), June 2019, p 149–157

Copyright # 2020 ASM InternationalW All rights reserved www.asminternational.org

ASM Handbook, Volume 24, Additive Manufacturing Processes D. Bourell, H. Kuhn, W. Frazier, M. Seifi, editors DOI 10.31399/asm.hb.v24.a0006546

Modeling for Polymer Additive Manufacturing Processes Neil Hopkinson, XAAR3D David Rosen, Georgia Institute of Technology

Introduction The four main industrial additive manufacturing approaches used to create polymer parts include:





Material extrusion Powder-bed fusion Vat polymerization Material jetting

Material extrusion and powder-bed fusion are used to produce thermoplastic polymer parts, and vat photopolymerization and material jetting are used to produce thermosetting polymer parts. Each approach has different subapproaches, some of which are discussed in detail in this article.

Material Extrusion Material extrusion (MEX) modeling takes the form of lumped-parameter models to describe material flow through the extruder and high-fidelity continuum models of temperature distributions during filament distribution, or mechanical responses as deposited filaments cool.

Lumped-Parameter Material Flow In most MEX machines, the extrusion device consists of a filament drive system, a liquefier that heats and melts the filament, and a nozzle through which the fine filament is extruded to deposit polymer. Material flow through the nozzle is similar to capillary flow and is controlled by the pressure drop between the chamber and the surrounding atmosphere. Reference 1 provides a good review of lumped-parameter modeling. A simple volumetric flow-rate model provides a good starting point to describe material flow. The volumetric flow rate (Q) of solid polymer into the liquefier is given by:

Q ¼ vf pr2f

(Eq 1)

where vf is the filament feed velocity, and rf is the filament radius. (Subscript “f ” denotes variables related to the filament fed into the extruder, and “r” denotes variables associated with roads being deposited.) At the nozzle, the volumetric flow rate is: Q ¼ vr WH

(Eq 2)

where vr is the deposition velocity, W is the width of the deposited road, and H is the road height, assuming that the deposited filament is approximately rectangular in cross-sectional shape. By equating the flow rates, one of the velocity terms can be expressed as a function of the other quantities. For example, the feed velocity is given as: vf ¼

vr WH prf2

(Eq 3)

Feed rate can be related to motor drive speed through vf = opRp, where op is the angular velocity of the pinch rollers, and Rp is the pinch-roller radius. If the pressure drop, DP, through the liquefier can be estimated, the force required to push the filament through the extrusion head is given by: F ¼ DPA

(Eq 4)

where A is the cross-sectional area of the filament. Pressure drop is determined through a momentum flux balance model that uses a model of viscosity as a function of temperature and shear rate (Ref 2), which is beyond the scope of this article. Given the force, the torque and power required to drive the filament can be computed as: G ¼ FRp , Pmot ¼ op G

(Eq 5)

This model assumes that one motor drives the pinch rollers through a gear train. If two

motors are used (one for each roller), then the torque term should be divided by 2.

Temperature-Distribution Model Higher-fidelity models were developed to estimate deposited filament shape, temperature distribution, residual stress, and deformation. Time-dependent temperature distributions provide insight into filament behavior, deposited shapes, cooling rates, and filamentfilament bonding. Bonding is particularly important, because it directly corresponds to bond strength and overall part mechanical properties. However, to date (2019), no MEX models have been able to directly predict bond strength, despite many years of effort (Ref 3). Finite-element analysis software capable of simulating fluid flow is required for MEX deposition modeling. Watanabe et al. (Ref 4) captured thermal processes experienced during material deposition by using ANSYS Polyflow and ANSYS Mechanical simulation software, wherein they developed several simulations, and the sequential simulations were linked to one another through the temperature profiles developed in previous steps. The first simulation model was the deposition and cooling of the first layer of filament deposited onto the build platform, which was at a constant temperature of 80  C (175  F). The geometry and mesh before deposition is shown in Fig. 1. Volumetric flow rate at the nozzle was given by Eq 5 in the vertical direction. The build platform was translated horizontally to achieve relative motion between the nozzle and platform. The remeshing technique was used in Polyflow to achieve good deposition simulation results. For this simulation, deposition temperature was 220  C (430  F), deposition speed was 20 mm/s (0.75 in./s), and layer height was 0.2 mm (0.008 in.), which represent typical values of process variable settings representative of the MEX process for polypropylene. Figure 2 shows the evolution of temperature

70 / Polymer Additive Manufacturing Processes distribution and deposited filament shape during the first layer deposition. The 5 mm (0.2 in.) deposit took 0.25 s, and the first part of the deposited filament was already cooling down to approximately 200  C (390  F). In the simulation, the deposited filament was allowed to cool for 1.25 s after deposition was completed, which was long enough for the entire filament to cool to the substrate temperature. Temperature distribution and deposited filament shape after depositing the first layer were exported from the previous stage of the simulation model. The build platform temperature was applied to the bottom surface of the first layer, and the other surfaces cooled due to convection with air. The second simulation model was deposition of the second filament layer on top of the first layer and cooling of both layers. The temperature distribution after first layer cooling was exported from the previous simulation. The procedure for this simulation model was similar to that for the first layer deposition and cooling. However, it was crucial to simulate conduction heat transfer between the two layers, which was accomplished by using the fluid-to-fluid contact capability in Polyflow. Figure 3 shows a sample of the second layer deposition simulation, with deposition at 0.2 s. Note that much of the first filament was heated significantly above the build-platform temperature. Temperature at the interface between filaments was above the melt temperature of 151  C (304  F), which helps ensure a good bond between filaments. The last simulation model determined the residual stress and warpage/deformation of the two deposited filament layers. The exported temperature distributions and deposited filament shapes were imported into ANSYS Mechanical to conduct structural analyses. By fixing the midpoint on the bottom surface of the first layer to the build platform and applying zero force everywhere else, residual stress and warpage during the cooling process were computed. The computational domain and boundary conditions are shown in Fig. 4. Residual stress and warpage/deformation of the two deposited filament layers at steady state are shown in Fig. 5. Because the simulation model was used to predict thermally induced residual stress and part warpage caused by material crystallization during cooling, the two-layer cooling simulation model results were linked to conduct the structural analyses in ANSYS Mechanical. These simulation model results were validated experimentally (Ref 4, 5).

Polymer Powder-Bed Sintering/ Fusion Polymer powder-bed sintering and/or fusion processes involve creating parts layer by layer from powder feedstock using selectively

Fig. 1

Geometry and mesh before first layer deposition in material extrusion deposition modeling. Source: Ref 4

Fig. 2

Temperature distribution and deposited filament shape during first layer deposition in material extrusion deposition modeling. Source: Ref 4

Fig. 3

Intermediate stage of second layer deposition showing temperature distribution in material extrusion modeling. Source: Ref 4

applied heat to induce bonding between particles in selected areas, while areas not selected remain in powder form. After all layers are sintered or fused, a powder bed is created in which parts are encased in powder; parts are revealed by removing the powder. Powder

not sintered or fused at the end of a build can, with some restrictions, be reused in subsequent builds. There is much debate about the use of the term sintering in powder-bed additive manufacturing. Conventional sintering is defined as the

Modeling for Polymer Additive Manufacturing Processes / 71 formation of solid parts from powder feedstock, wherein particles coalesce under the influence of elevated temperature and/or pressure. Sintering processes are generally considered to fall into two categories:
Solid-state sintering: Where powder mate-

rial is subjected to a temperature below its melting point, Tm (typically ~Tm/2 on the Kelvin scale), for an extended time to enable formation of bonds across particle boundaries. Pressure can also be applied to increase the rate and extent of mass transfer and interparticle bonding.
Liquid-phase sintering: Where the feedstock consists of at least two different powder materials, with one material having a lower melting temperature than the other. Sintering occurs by heating the feedstock to a temperature where the lower-meltingtemperature material melts, and the molten material through-capillary action encases the solid material to form a solid structure after cooling. In conventional sintering, solid-state sintering is the dominant process in industrial application. Two different mechanisms by which particles coalesce in polymer powder-bed sintering or fusion are:

where Pl is the laser power (W), SS is the scan spacing (mm), and Sb is the beam speed (mm/s). Caulfield et al. (Ref 7) used Eq 6 to determine the influence of energy density on mechanical properties of parts produced (Fig. 6). Figure 6 shows that for polyamide 12 (PA12), part ductility, measured by percent elongation at break, increases as energy density increases up to ~0.023 J/mm2, decreasing thereafter. It is concluded that increased energy helps to achieve more complete melting and consolidation of particles up to a point, after which increasing energy leads to thermally induced degradation of the polymer and subsequent reduction in mechanical properties.

A drawback of Eq 6 is that it is an areal approach and does not consider the volume of material processed by the laser; specifically, it does not include layer thickness or the temperature of the bed prior to application of the laser. Gibson and Shi (Ref 8) addressed these drawbacks by developing an equation that takes into account layer thickness and bed temperature and also considers material characteristics, including powder packing density, reflectivity, specific heat, and latent heat of fusion of the powder polymer material, given as: Pl ¼ ðSb ÞðpÞðDb Þðhl Þ½CðTm  Tb Þ þ Lf =ð1  RÞ (Eq 7)

Fig. 4

Geometry and mesh before the residual-stress and warpage simulation of two deposited filament layers in material extrusion deposition modeling. Source: Ref 4

Fig. 5

Residual stress at steady-state (left) and warpage/deformation (right) results from simulation in material extrusion deposition modeling. Source: Ref 4

Fig. 6

Effect of energy density on ductility of polyamide 12 parts in selective laser sintering. Source: Ref 7


Melting and fusion: Part of each particle, or

the entire particle, is heated to above its melting temperature and coalesces according to the Frenkel equation.
Second-order thermal coalescence: Polymer particles are heated to a temperature higher than the glass transition temperature, and bonds form across particle boundaries by resetting secondary bonds (e.g., hydrogen or van der Waals bonds) across particle boundaries. A majority of applications of powder-bed sintering or fusion use melting or fusion for particle coalescence; thus, this mechanism has been researched and is best understood and is the focus of the following discussion. Different levels of scale used to address modeling and the impact of process settings include:
Thermodynamics at the powder-bed surface
Consolidation of adjacent particles in the

fusion process


Fusion and molecular-level behavior within

particles

Thermodynamics at the Powder-Bed Surface In the first published work about modeling laser energy applied to a powder-bed surface in laser sintering, Nelson et al. (Ref 6) devised an expression to determine energy density as: Energy density ðJ=mm2 Þ ¼ Pl =SSðSb Þ

(Eq 6)

72 / Polymer Additive Manufacturing Processes where Pl is the required laser power, p is the powder density, Db is the laser beam diameter, hl is the layer height, C is the specific heat of the polymer, Tm is the polymer melting temperature, Tb is the powder-bed temperature, Lf is the latent heat of fusion of the polymer powder, and R is the reflectivity of the polymer powder. Starr et al. (Ref 9) combined and expanded this work to help understand the process by developing an equation to determine energy melt ratio, which effectively is the amount of energy applied by the machine divided by the theoretical minimum energy required to fuse the powdered material. They considered energy applied according to Eq 6 but included layer thickness, given by the expression: Energy density ¼ Pl =ðSSÞðSb Þðhl Þ

(Eq 8)

They determined the energy-to-melt ratio by dividing the applied energy density from Eq 8 by the minimum energy required to melt the powder, derived from Eq 7: Energy-to-melt ratio ¼ Energy density=Energy to melt the layer (Eq 9)

Fig. 7

Effect of energy-to-melt ratio on ductility of laser-sintered polyamide 12 parts. Source: Ref 10

Fig. 8

Progression of coalescence of polymer particles as they melt and become liquid, which is described by the Frenkel-Eshelby model. Source: Ref 11

Starr et al. (Ref 10) subsequently established a link between the energy-to-melt ratio and the mechanical properties of laser-sintered PA12 parts, as shown in Fig. 7.

Consolidation of Adjacent Particles in the Fusion Process As polymer particles melt and become liquid, they coalesce with a progression of coalescence described by the Frenkel-Eshelby model, as depicted in Fig. 8 and the equation: y ¼ a




 Gt 0:5 ma

(Eq 10)

where y is the half-neck radius, a is the initial particle radius (assuming a1 = a2 from Fig. 8), G is the surface energy, m is the viscosity, and t is the sintering or fusing time. The model shows how two adjacent particles coalesce in a process driven by surface free energy and are restricted by viscosity. Because polymers have low surface free energy and high melt viscosity, the process is relatively slow. However, a smaller particle radius can reduce the time required to achieve full coalescence. Haworth et al. (Ref 12) modeled the progression of coalescence for PA12 powder using various blends of virgin and used powder. Used powder from the part cake in polymer powder-bed fusion has an increased molecular chain length and thus increased viscosity over that of virgin powder. Figure 9 shows the progression of sintering modeled by using the measured viscosity of different

blends of virgin and used powder. It demonstrates the impact of viscosity on the time required to achieve full coalescence, with virgin powder expected to achieve full coalescence after ~0.5 s and used powder achieving full coalescence after ~4 s. Vasquez (Ref 13) established that scanning lasers in laser sintering using typical process parameters results in each particle being irradiated for a duration of 0.000066 to 0.0004 s, indicating that there is a clear mismatch of

1000 to 10,000 times between the time in which heat for fusing is applied when using a scanning laser and the time over which coalescence can occur.

Fusion and Molecular-Level Behavior within Particles In polymer powder-bed fusion, it is common to have incomplete melting of powders; many

Modeling for Polymer Additive Manufacturing Processes / 73

1.4

1.2 Dimensionless neck radius, x/r

particles used to form the part are only partially melted. A possible reason is because smaller particles achieve full melting more readily than larger particles, and overheating can lead to issues such as loss of part-feature resolution. Zarringhalam et al. (Ref 14) used differential scanning calorimetry (DSC) to measure the degree of particle melting in a manufactured sample. The method works with semicrystalline PA12 because the raw (unmelted) powder has a certain percentage of crystallinity (crystal in the alpha form), which melts at ~190  C (375  F). However, when the molten material recrystallizes in the powder-bed fusion process, it forms a gamma-phase crystal structure, which melts at ~180  C (355  F). The measured DSC curve shows different endothermic melt peaks for the two phases. Figure 10 shows an optical microscopy image from a microtomed slice cut from a laser-sintered PA12 part, with outlines indicating the edge of a particle (melted and (re)crystallized region) and, within it, the edge of the zone that remained unmelted during the sintering process (unmelted particle core). The graph in Fig. 10 shows the DSC endotherm for virgin powder material with a melt peak of 188  C (370  F) and a DSC trace showing two endotherms for the different crystalline phases in a manufactured part, with the unmelted (gamma) phase having a slightly higher melt peak of 189  C (372  F) and the melted (alpha) phase having a larger melt peak but at a significantly lower temperature of 181  C (358  F). The relative sizes of the melt peaks from the sintered part, measured by integrating the curve of the peaks, provide a calculation for the degree of melted material, expressed as a percentage (Ref 15):

Virgin

1

10% 20%

0.8

30% 40%

0.6

50% 60%

0.4

70% 80%

0.2

90% 100%

0 0

1

2

3

4

5

6

7

8

Sintering time, s

Fig. 9

Effect of sintering time on achieving full particle coalescence of different blends of virgin and used polyamide 12 powder. (Legend indicates percent of used powder in the blend.) Source: Ref 12

Fig. 10

Optical microscopy image of microtomed sample of polyamide 12 laser-sintered part together with associated differential scanning calorimetry melt peaks for the part and for virgin powder. Source: Ref 14

% crystallinity ¼

Heat of melting ðsampleÞ Heat of melting ð100% crystalline specimenÞ (Eq 11)

and DPMð% meltedÞ %C of sample%C of unmelted material ¼ %C of melted regions%C of unmelted regions (Eq 12)

Hopkinson et al. (Ref 15) also established a relationship between the derived degree of particle melting (DPM) by using the derivation method described previously and the measured mechanical properties of the produced part (Fig. 11).

Vat Photopolymerization In vat photopolymerization (VPP) processes, light is projected in a pattern into a vat of liquid photopolymer resin, similar in many ways to irradiation in powder-bed

fusion, where the patterned light causes a polymerization reaction to cure the photopolymer. Typically, curing occurs at the resin surface and forms a layer of the part. Two broad classes of VPP processes have been developed based on:


Laser scanning, where scan vectors of a

laser scan over the resin surface and “draw” the part cross section to form the layer
Pattern projection, where patterned light is projected onto the resin surface so the entire part cross section is illuminated

74 / Polymer Additive Manufacturing Processes capture the effects of laser angle of incidence and refraction in the resin (Ref 18). Simulations of laser-scanning VPP processes proceed by simulating all laser scans for fabricating a part and determining the accumulation of exposure in each point of interest in the vat by using Eq 16. Given a grid of points in the vat, P, and a set of scans, Si, for each layer i, a simple simulation algorithm can be given as: For each layer, i ¼ 1:N

Depth = Layer thickness  (i  1) For each scan, s in Si For each point, p ¼ ðpx , py , pz Þ in P

Fig. 11

Effect of degree of particle melting (DMP) on ductility of polyamide 12 parts. ■ = differential scanning calorimetry method; ● = optical microscopy method. Source: Ref 15

and causes the part layer to cure everywhere simultaneously Two primary approaches to modeling VPP processes include a lumped-parameter approach to estimate cured regions in the vat, known as the Jacobs model (Ref 16), and a high-fidelity, continuum approach that uses finite-element methods.

Lumped-Parameter VPP Simulation The focus in this section is on the simulation of laser-scanning VPP, but pattern-projection VPP is discussed briefly. Laser-Scanning VPP As a laser beam is scanned across the resin surface, it cures a line of resin in the shape of a parabola, the size of which depends on factors including resin characteristics, laser-energy characteristics, and scan speed. The main idea is to model the radiation distribution spatially and monitor the total radiation received at each point in the vat to determine whether that total is enough to cause curing. In other words, radiation exposure is assumed to be additive. Irradiance is the radiant power of the laser per unit area, H(x,y,z). As the laser scans a line, radiant power is distributed over a finite area, because beams are not infinitesimal. Assuming that the laser scans along the x-axis, while the z-axis is oriented perpendicular to the resin surface and into the resin, and that radiation is attenuated in the resin according to the BeerLambert law, then the general form of the irradiance equation for a Gaussian laser beam is: Hðx, y, zÞ ¼ Hðx, y, 0Þez=Dp 2 2 2 2 ¼ H0 e2x =W0 e2y =W0 ez=Dp

(Eq 13)

where Dp is a resin parameter called the penetration depth; H0 is the maximum laser irradiance at the center of the laser beam,

typically given as 2PL/(pW02); and W0 is the 1/e2 Gaussian half-width of the beam spot. For z = Dp, then Dp is the depth into the resin at which the irradiance is 37% (because e1 = 0.36788) of the irradiance at the surface. However, the exposure at a point in the resin determines the extent of cure, not simply the irradiance. Exposure is the energy received per unit area. When exposure at a point in the resin vat exceeds a critical value, called Ec, the resin is assumed to have cured. Exposure can be determined at point p by appropriately integrating Eq 13 along the scan line, from time 0 to time tb, when the laser reaches some point b. In this case, integration over distance is far more convenient than trying to integrate over time. For a constant laser scan velocity, Vs, in the x-direction, the exposure received at a distance y from the scan vector (local x-axis) is: Eðy, 0Þ ¼

2PL 2y2 =W2 0 e pVs W02

Z

x¼b

e2x

2 =W 2 0

dx

(Eq 14)

x¼0

Integration can be carried out through an appropriate change of variable, assuming a laser scan longer than approximately five laser-beam diameters, given by: Eðy, 0Þ ¼

rffiffiffi 2 PL 2y2 =W 2 0 e p W0 Vs

(Eq 15)

The fundamental general exposure equation is obtained by combining Eq 15 with Eq 13 (Ref 17): rffiffiffi 2 PL 2xdist2 =W 2 2ðyys Þ2 =W 2 z=Dp 0e 0e Eðx, y, zÞ ¼ e p W0 Vs (Eq 16)

where s = (xss ! xse, ys), denoting the scan vector from point (xss, ys) to point (xse, ys); and xdist denotes the x-distance between x and the scan vector for points x, y, and z in the vat. More sophisticated models of laser-scanning VPP

8 < xss  px , if px < xss xdist ¼ px  xse , if px > xse : 0, otherwise y ¼ py z ¼ pz  Depth Exposurep ¼ Exposurep þ Eq 16

The last line indicates that exposure is assumed to additive; that is, exposure from a single scan is added to the exposure already received from other scans. For the simple wedge-shaped part in Fig. 12(a), simulation results using reasonable values of resin, laser, and scanning parameters yield the simulated cross-sectional shape shown in Fig. 12(b). The zoomed-in region shows laser scan locations as small red circles. Pattern-Projection VPP In pattern-projection VPP using a digital micromirror device (DMD) chip, the pattern displayed on the chip is projected into the resin vat to cure a part cross section. Broadly, cured shapes are controlled by the projected pattern, radiation intensity (irradiance), and the time the pattern is projected (Ref 19). In the simplest case, the exposure received at the resin surface from a single micromirror is the irradiance multiplied by the time that the micromirror is flipped “on,” E = H(t). For a gray-scale DMD, irradiance from each micromirror can be varied over time, so a more general expression for exposure received is given as: Eðx, yÞ ¼ a Hðx, yÞt

(Eq 17)

where irradiance can have a spatial distribution, and a is the gray-scale level (0 to 1). Simulation of pattern-projection VPP proceeds in a manner similar to that of laser scanning. At a given point in time, the irradiance of the projected pattern is given by the process plan, and the irradiance pattern is integrated over a time step to determine exposure at all sample points; when total exposure received at a point is greater than Ec, that point is assumed to be cured solid.

Modeling for Polymer Additive Manufacturing Processes / 75 motion depends on the fluid pressure and vice versa. An iterative loop is used between the user subroutine and the CFD code to update fluid motion based on the coupled behavior of the diaphragm.

Droplet Impact

Fig. 12

Laser scan simulation results showing effect of surface roughness nonuniform exposure distribution typical of vat photopolymerization (VPP). (a) Simple wedge-shaped part 20  8  10 mm (0.8  0.3  0.4 in.). (b) VPP simulation results. Source: Ref 18

Continuum Model of VPP High-fidelity, continuum modeling of VPP processes takes the form of a finite-element model, where each element models the chemical reactions that occur, as well as other physical phenomena that the modeler deems relevant. The phenomena are typically heat generation and transfer, diffusion of reacting species, and possibly diffusion of oxygen (which functions as an inhibitor for freeradical polymerization), shrinkage of cured resin, and mechanical response, to include residual stresses. Just as in lumped-parameter modeling, a model of irradiance is required that captures the spatial and temporal distributions of incident radiation that occur using either laser-scanning or pattern-projection processes. The most common commercial software package for such modeling is COMSOL, which has a built-in chemical engineering module with appropriate first-order differential equations to model reactions. Concentrations of photoinitiator, monomer, polymer, and oxygen are computed, based on initiation, polymerization, and termination reaction rates. Because polymerization reactions tend to be exothermic, heat generation is also modeled; models of thermal conduction are used to compute temperature distributions over time (Ref 20). More sophisticated, or secondary, models can also model mechanical phenomena to include size changes (e.g., thermal expansion contractions and shrinkage of polymerization) (Ref 21).

Material Jetting For material jetting, simulations are used to study droplet generation at the nozzle and

droplet impact. Approximate, computationally efficient approaches are of interest for droplet impact to enable modeling of deposited shapes. The material-jetting additive manufacturing process benefits from extensive research conducted on droplet behavior in many applications, including two-dimensional printing, printed electronics, thin-film research, surface coating, and fluid behavior in heat-transfer applications, among others.

Droplet Generation Inkjet printheads for material-jetting processes typically use standard nozzle arrays. Fundamental droplet-generation technology relies on generating a pressure wave in a small fluid reservoir that serves to eject a droplet of fluid through an orifice that connects the reservoir to the ambient environment. Two basic mechanisms used to generate the pressure wave are thermal bubbles and piezoelectric transducers (Ref 22). The mechanics of the actuator play a large role in modeling pressure and velocity fields within the reservoir and the emerging droplet. The approach of Pan et al. (Ref 23) is typical of droplet-generation simulation methods, in which a three-dimensional computational fluid dynamics (CFD) code based on volume of fluid (VOF) methods was linked with a user subroutine to model the behavior of the diaphragm actuator. In addition to the requisite Navier-Stokes and mass continuity equations, the traction condition at the free surface in the aperture area is modeled by using an expression that relates the rate-of-strain tensor with pressure, surface tension, and curvature. The diaphragm equation of motion is coupled to the fluid pressure, indicating that diaphragm

Control and predictability of the dropletdeposition process is essential for part quality and accuracy. Models to explain and predict droplet impingement dynamics have been studied as early as the 19th century (Ref 24). Various phenomena, including droplet oscillations, splash, and contactangle hysteresis, have been analyzed. More recent emphasis on high-fidelity numerical models in the fluid dynamics field resulted in accurate predictions of droplet behavior under a wide range of situations. These models have become of interest in the additive manufacturing community. Macroscale continuum models of droplet impact are commonly used for high-fidelity simulations. They implement the governing Navier-Stokes and mass-continuity equations but differ depending on their treatment of the interfaces between phases in a multiphase flow problem. The challenge is to track the evolution of interfaces between two fluids (i.e., droplet and air) as the droplet interacts with the substrate. Several methods have been used to simulate multiphase flow, including the VOF method (Ref 25), the level-set method (Ref 26), the phase-field method (Ref 27), and the front-tracking method. The VOF and level-set methods treat the interface between fluid phases as sharp (i.e., infinitesimally thin). The methods differ in that the VOF method requires a careful geometric reconstruction of the interface shape so the interface curvature can be computed accurately. Both methods depend on the fluid domain being discretized into a grid of elements. The VOF method uses a function of volume fraction, VF, to represent the fluid interface (Ref 25), where the volume fraction of each phase is determined within each grid element. In contrast, the level-set method introduces a continuous, smooth level-set function that represents the signed normal distance from the interface. By updating the level-set function using fluid velocity, the method tracks the interface as the zero set of the function. The VOF method has been implemented in many commercial and open-source codes, such as ANSYS Fluent, Flow3D, and STAR-CCM, while the level-set method has been implemented in other codes, most notably, COMSOL. In the phase-field method, the fluid interface is modeled as a finite-thickness transitional region with continuous variations of an order parameter (e.g., density) from one material phase to another. The order parameter of the phase-field method generally is a physical quantity governed by physical laws, in contrast

76 / Polymer Additive Manufacturing Processes to the distance-field approach used in the levelset method. In a multiphase flow, a typical order parameter is the composition of one fluid, where the composition variable is driven by the gradient of chemical potential and governed by the Cahn-Hilliard equation (Ref 27). Phase-field methods are incorporated into COMSOL and some open-source codes. Mesoscale, particle-based methods have emerged as an alternative to the continuumbased methods summarized previously and are arguably better at capturing smaller-scale phenomena that occur in gas and fluid dynamics, particularly in multiphase flows. The lattice Boltzmann method (LBM) is typical of these methods and has been applied to model droplet impingement. An underlying concept of the LBM is the construction of a lattice grid in the fluid domain; fluid particles are constrained to move along predefined paths in the lattice. Particle distribution in the lattice space is modeled probabilistically by using a particle distribution function rather than tracking the exact locations and momenta of individual particles, as in molecular dynamics. An advantage, and major challenge, with the LBM is that various complex macroscopic phenomena are modeled by using a particleinteraction model. For many fluid dynamics phenomena, these interaction models are well developed, based on the Boltzmann transport equation. However, for some phenomena for certain complex flows and fluids, capturing accurate fluid behaviors through interaction models is challenging and remains the subject of research. Research using the LBM illustrates some droplet-impingement simulation results. The objective was to develop a computationally efficient method to simulate impingement of simultaneous droplets by using a wide range of fluid properties, including high-viscosity and high-surface-tension fluids. Zhou et al. (Ref 28) advanced LBM modeling of dropletsubstrate interface physics by performing simulations of single droplets, two droplets, a line of droplets, and an array of droplets. They used an LBM lattice of 100 by 100 by 70 units for single droplets, where the droplet radius was 25 units. Simulations were run on a standard personal computer (in 2013) using a single thread and were completed within 20 h. In contrast, it is estimated that running similar simulations on COMSOL for the same modeling resolution would require approximately one month of computing time. Results from one simulation are shown in Fig. 13 for the case of an array of droplets impacting at 10 m/s (33 ft/s), 50 mm diameter, 0.0723 N/m surface tension, 0.008 Ns/m2 viscosity, 1000 kg/m3 density, and 65 mm droplet spacing. Corresponding Weber and Ohnesorge numbers were approximately 70 and 2, respectively, indicating that fluid behavior is inertiadriven balanced by viscosity effects. Stated differently, droplets are expected to be deformed considerably on impact, but splashing should not occur. This behavior is observed in the time

sequence of images taken from the simulation shown in Fig. 13. Although the droplets interact and deform significantly, the dynamics eventually settle down and a thin film is formed. Such simulation models can provide important insight into droplet behavior and the effects of changes in fluid characteristics and impact conditions. While high-fidelity simulations provide very accurate models of droplet behavior, they are still too computationally demanding to support the simulation of meso- and macroscale features fabricated by material jetting. To simulate part and feature fabrication, approximate dropletdeposition models are required that capture droplet impact and interaction behavior only to the extent needed to determine final droplet shapes. Lu et al. (Ref 29) offered an empirical height-change model applied to a fixed printing grid—a simple approach that seems to be reasonably good for additive manufacturing materials. Each droplet contributes a predetermined amount of material (height) to the grid point at which it is printed and also contributes a lesser amount to its neighboring grid points. This model is powerful enough to yield edge effects observed in real printed parts, where edges expand due to droplet flow and are rounded or lower than bulk regions. However, because it is developed using fixed-neighbor contributions, it cannot correctly model a wide variety of small features. An approach of intermediate sophistication is based on a quasi-static analysis, which seeks to balance the pressure drop across the liquid droplet interface with the surface tension of the interface based on its curvature. Thompson et al. (Ref 30) used the Young-Laplace equation: Z V¼

ðhðx, yÞ  sðx, yÞÞdxdy

(Eq 18)

Ω

to compute the droplet surface shape by iteratively changing the pressure drop, computing

Fig. 13

curvatures, and integrating over the droplet domain subject to a constraint on the droplet volume. That is, the Young-Laplace equation and the volume constraint equation: Dp ¼ sr2 hðx, yÞ ¼ sðhxx þ hyy Þ

(Eq 19)

are solved iteratively to determine the height profile, h(x,y), of each deposited droplet, where s is the surface tension, and s(x,y) denotes the height profile of the substrate on which the droplet is deposited (Ref 31). This model has been applied to several shapes and features, demonstrating good correlation with shapes fabricated by using a Stratasys Connex 260 printer. Fabrication of a small 1 by 1 mm2 by 0.1 mm (0.0016 by 0.0016 in.2 by 0.004 in.) high part with a 0.2 mm (0.008 in.) diameter hole was simulated by using a 30 pL droplet with a 45 static contact angle. A total of 3020 droplets were “deposited” over the five layers by using a droplet and line spacing of 42.3 mm. The simulation took approximately 8 min on a standard personal computer; the resulting part is shown in Fig. 14. As was expected, the part exhibits rounding along all edges, with the diameter of the hole at the substrate smaller than desired but becoming significantly larger than designed due to edge rounding. It is expected that simulation models such as these will continue to be developed so that predictive models will become available with which jetting process plans, new materials, and feature shapes can be investigated without resorting to extensive experimentation programs.

REFERENCES 1. B. N. Turner, R. Strong, and S. A. Gold, A Review of Melt Extrusion Additive Manufacturing Processes, Part I: Process

Sequence of images from droplet array simulation. Impact velocity = 10 m/s (33 ft/s); droplet diameter = 50 mm; surface tension = 0.0723 N/m; viscosity = 0.008 Ns/m2; density = 1000 kg/m3; and droplet spacing = 65 mm. Source: Ref 28

Modeling for Polymer Additive Manufacturing Processes / 77

Fig. 14

2.

3.

4.

5.

6.

7.

8.

9.

Part shape from quasi-static material-jetting simulation. Part dimension: 1  1 mm2  0.1 mm (0.0016  0.0016 in.2  0.004 in.) high, with 0.2 mm (0.008 in.) diameter hole. Source: Ref 31

Design and Modeling, Rapid Prototyp. J., Vol 20 (No. 3), 2014, p 192–204 A. Bellini, S. Guceri, and M. Bertoldi, Liquefier Dynamics in Fused Deposition, J. Manuf. Sci. Eng., Vol 126 (No. 2), 2004, p 237–246 M.A. Yardimci and S. Guceri, Conceptual Framework for the Thermal Process Modelling of Fused Deposition, Rapid Prototyp. J., Vol 2 (No. 2), 1996, p 26–31 N. Watanabe, M. Shofner, N. Treat, and D.W. Rosen, “A Model for Residual Stress and Part Warpage Prediction in Material Extrusion with Application to Polypropylene Composite Materials,” Solid Freeform Fabrication Symposium (Austin, TX), Aug 8–10, 2016 E.R. Fitzharris, N. Watanabe, D.W. Rosen, and M.L. Shofner, Effects of Material Properties on Warpage in Fused Deposition Modeling Parts, Int. J. Adv. Manuf. Technol., Vol 95 (No. 5), 2018, p 2059–2070, DOI: doi.org/10.1007/s00170-017-1340-8 J.C. Nelson, S. Xue, J.W. Barlow, J.J. Beaman, H.L. Marcus, and D.L. Bourell, Model of the Selective Laser Sintering of Bisphenol-A Polycarbonate, Ind. Eng. Chem. Res., Vol 32 (No. 10), 1993, p 2305– 2317, https://doi.org/10.1021/ie00022a014 B. Caulfield, P.E. McHugh, and S. Lohfeld, Dependence of Mechanical Properties of Polyamide Components on Build Parameters in the SLS Process, J. Mater. Process. Technol., Vol 182, 2007, p 477–488 I. Gibson and D. Shi, Material Properties and Fabrication Parameters in Selective Laser Sintering Process, Rapid Prototyp. J., Vol 3 (No. 4),1997, p 129–136 T.L. Starr et al., “Laser Sintering of PA-11 and PA-12 for Direct Digital Manufacturing,” Solid Freeform Fabrication Symposioum (Austin, TX), 2009

10. T.L. Starr, T.J. Gornet, and J.S. Usher, The Effect of Process Conditions on Mechanical Properties of Laser-Sintered Nylon, Rapid Prototyp. J., Vol 17 (No. 6), 2011, p 418–423 11. G.M. Vasquez et al., Targeted Materials Selection Process for Laser Sintering, Add. Manuf., Vol 1–4, Oct 2014, p 127–138, https://doi.org/10.1016/j.addma.2014.09.003 12. B. Haworth, N. Hopkinson, D.J. Hitt, and X. Zhong, Shear Viscosity Measurements on Polyamide-12 Polymers for Laser Sintering, Rapid Protoyp. J., Vol 19 (No.1), Jan 2013, p 28–36, doi: 10.1108/ 13552541311292709 13. M. Vasquez, “Analysis and Development of New Materials for Polymer Laser Sintering,” Ph.D. thesis, Loughborough University, 2012 14. H. Zarringhalam, C.E. Majewski, and N. Hopkinson, Degree of Particle Melt in Nylon-12 Selective Laser Sintered Parts, Rapid Prototyp. J., Vol 15 (No.2), 2009, p 126–132, DOI: 10.1108/135525409109 43423 15. N. Hopkinson, C.E. Majewski, and H. Zarringhalam, Quantifying the Degree of Particle Melt in Selective Laser Sintering , CIRP Ann., Vol 51 (No. 1), 2009, p 197– 200, doi: 10.1016/j.cirp.2009.03.001 16. P.F. Jacobs, Rapid Prototyping and Manufacturing: Fundamentals of Stereolithography, Society of Manufacturing Engineers, New York, NY, 1992 17. I. Gibson, D.W. Rosen, and B. Stucker, Additive Manufacturing Technologies: 3D Printing, Rapid Prototyping, and Direct Digital Manufacturing, 2nd ed., Springer, 2015, ISBN: 978-1-4939-2113-3 18. B. Sager and D.W. Rosen, Use of Parameter Estimation for Stereolithography Surface Finish Improvement, Rapid Prototyp. J., Vol 14 (No. 4), 2008, p 213–220

19. A. Bertsch, S. Zissi, J. Jezequel, S. Corbel, and J. Andre, Microstereolithography Using Liquid Crystal Display as Dynamic MaskGenerator, Microsyst. Technol., Vol 3 (No. 2), 1997, p 42–47 20. Y. Tang, C. Henderson, J. Muzzy, and D.W. Rosen, Stereolithography Cure Modeling and Simulation, Int. J. Mater. Prod. Technol., Vol 21 (No. 4), 2004, p 255–272 21. M.M. Emami and D.W. Rosen, “An Improved Vat Photopolymerization Cure Model Demonstrates Photobleaching Effects,” Solid Freeform Fabrication Symposium (Austin, TX), Aug 13–15, 2018 22. H.P. Le, Progress and Trends in Ink-Jet Printing Technology, J. Imag. Sci. Technol., Vol 42, Jan 1, 1998, p 49–62 23. F.X. Pan, J. Kubby, and J.K. Chen, Numerical Simulation of Fluid-Structure Interaction in a MEMS Diaphragm Drop Ejector, J. Micromech. Microeng., Vol 12, Jan 2002, p 70–76 24. A.M. Worthington, On the Forms Assumed by Drops of Liquids Falling Vertically on a Horizontal Plate, Proc. R. Soc. (London), Vol 25, 1876, p 261–271 25. C.W. Hirt and B.D. Nichols, Volume of Fluid (VOF) Method for the Dynamics of Free Boundaries, J. Comput. Phys., Vol 39 (No. 1), 1981, p 201–225 26. S. Osher and J.A. Sethian, Fronts Propagating with Curvature-Dependent Speed: Algorithms Based on Hamilton-Jacobi Formulations, J. Comput. Phys., Vol 79 (No. 1), 1988, p 12–49 27. D. Jacqmin, Calculation of Two-Phase Navier-Stokes Flows Using Phase-Field Modeling, J. Comput. Phys., Vol 155 (No. 1), 1999, p 96–127 28. W. Zhou, D. Loney, A.G. Fedorov, F.L. Degertekin, and D.W. Rosen, Lattice Boltzmann Simulations of Multiple Droplet Interaction Dynamics, Phys. Rev. E, Vol 89, March 2014, Paper 033311, doi: 10.1103/PhysRevE.89.033311 29. L. Lu, J. Zheng, and S. Mishra, A Layerto-Layer Model and Feedback Control of Ink-Jet 3-D Printing, IEEE/ASME Trans. Mechatron., Vol 20 (No. 3), 2015, p 1056–1068 30. A.B. Thompson, C.R. Tipton, A. Juel, A. L. Hazel, and M. Dowling, Sequential Deposition of Overlapping Droplets to Form a Liquid Line, J. Fluid Mech., Vol 761, 2014, p 261–281 31. C. Hume and D.W. Rosen, “Low Cost Numerical Modeling of Material JettingBased Additive Manufacturing,” Solid Freeform Fabrication Symposium (Austin, TX), Aug 13–15, 2018

Ceramic Additive Manufacturing Processes Division Editor: Ming Leu, Missouri University of Science and Technology

Vat-Photopolymerization-Based Ceramic Manufacturing . . . . . Vat Photopolymerization . . . . . . . . . . . . . . . . . . . . . . . . . . . . Vat-Photopolymerization-Based Ceramic Fabrication . . . . . . . . Postprocessing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Property Identification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Summary and Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

81 82 83 86 88 92

Material Extrusion Based Ceramic Additive Manufacturing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98 Post-Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 Mechanical Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 Future Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 Innovation Opportunities . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 Material Jetting of Ceramics . . . . . . . . . . . . . . . . . . . . . . . . . 112 Ink Properties and Delivery . . . . . . . . . . . . . . . . . . . . . . . . . 112 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114 Binder Jetting of Ceramics . . . . . . . . . . . . . . . . . . . . . . . . . . 118 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118 Processability Considerations for Binder Jetting Additive Manufacturing of Ceramics . . . . . . . . . . . . . . . . . . . . . . . 118

Binder Jetting Additive Manufacturing Technology Potential. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128 Directed-Energy Deposition for Ceramic Additive Manufacturing . . . . . . . . . . . . . . . . . . . . . . . . . . . . Directed-Energy Deposition Equipment for Ceramic Additive Manufacturing . . . . . . . . . . . . . . . . . . . . . . . . . . Directed-Energy Deposition Materials for Ceramic Additive Manufacturing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Directed-Energy Deposition Process . . . . . . . . . . . . . . . . . . . Microstructure and Properties of Typical Directed-Energy Deposition Ceramics . . . . . . . . . . . . . . . . . . . . . . . . . . . . Powder Bed Fusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Powder Bed Fusion for Ceramics . . . . . . . . . . . . . . . . . . . . . Ceramics Processed by Powder Bed Fusion . . . . . . . . . . . . . . Challenges and Potential in Powder Bed Fusion of Ceramics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Applications of Powder Bed Fusion Ceramics . . . . . . . . . . . . Conclusion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

131 133 134 135 136 152 152 153 154 159 160

ASM Handbook, Volume 24, Additive Manufacturing Processes D. Bourell, W. Frazier, H. Kuhn, M. Seifi, editors DOI 10.31399/asm.hb.v24.a0006578

Copyright # 2020 ASM InternationalW All rights reserved www.asminternational.org

Vat-Photopolymerization-Based Ceramic Manufacturing Xiangjia Li and Yong Chen, University of Southern California

CERAMICS are one type of inorganic and nonmetallic material, and their crystalline structure and chemical composition result in excellent physical properties, including mechanical, thermal, chemical, optical, electrical, and magnetic properties (Ref 1). Due to these versatile properties, ceramics are widely used in many different applications, such as thermal-protection shields in aerospace, bioimplants in biomedical engineering, pump filters and catalysts in the chemical industry, sensors and dielectrics in electronics, and cutting tools in manufacturing (Fig. 1) (Ref 1). With the

Fig. 1

increasing use of ceramics in different fields, various forming methods were developed over the last 100 years. However, with conventional manufacturing techniques, such as injection molding, die pressing, tape casting, throwing, and so on, it is difficult to satisfy the processing demand of ceramic parts with complex geometric design due to the limited fabrication capability (Ref 2). For example, bioceramicbased scaffold is customized based on the shape of the defect, and it is filled with interconnected holes, making it impossible to be built using the aforementioned manufacturing techniques

Current additive manufacturing processes developed to fabricate ceramic components. IR, infrared; PZT, lead zirconate titanate; SLA, stereolithography; SLS, selective laser sintering; MJP, multijet printing; SLM, selective laser melting; LOM, laminated object manufacturing; TPP, two-photon polymerization; DIW, direct ink writing; FDM, fused-deposition modeling

(Ref 2). In the last 30 years, the manufacturing community has benefited from additive manufacturing (AM), also known as three-dimensional (3D) printing, which enables the fabrication of a 3D object from scratch using a wide range of materials (Ref 3). Additive manufacturing shows excellent advantages in building 3D objects with complex geometry, and this unique manufacturing method makes it possible to fabricate customized ceramic structure with high accuracy that is difficult for conventional manufacturing techniques (Ref 4). Several ceramic AM processes, including vat photopolymerization (VPP), two-photon polymerization (TPP), fused-deposition modeling (FDM), direct ink writing (DIW), inkjet printing, powder binder jetting, selective laser sintering/ melting (SLS/M), and laminated object manufacturing, have been developed (Fig. 1) (Ref 5, 6). With the development of ceramic 3D printing, different types of ceramics may be fabricated to study their chemical, electrical, optical, thermal, and mechanical functionalities, giving designers and engineers more flexibility to be innovative without being limited by traditional manufacturing methods (Ref 5, 6). According to the physical phase of the printing material, ceramic AM processes can be broadly divided into two main categories: powder-based ceramic printing and slurrybased ceramic printing (Ref 2). In powderbased ceramic printing, ceramic powder is first spread by a paving mechanism on the powder bed, and the printing process, after the ceramic powder is spread, varies in principle. Major powder-based ceramic printing processes are inkjet printing (Ref 7), binder jetting (Ref 8), and SLS/M (Ref 9). For example, binder jetting is a technology in which binder is selectively deposited over the spread ceramic powder layer, and, with the help of binder, the ceramic powder is accumulated layer by layer to form a 3D shape (Fig. 2) (Ref 8). For binder jetting, postprocesses that include postcuring, depowdering, sintering, infiltration, annealing, and finishing are necessary to obtain highquality ceramic parts (Ref 5). Because solid

82 / Ceramic Additive Manufacturing Processes powder plays the role of support, binder jetting exhibits strength for fabricating components that have complex inner structures, and the process is widely used in the fabrication of bioceramic-based scaffolds for tissue engineering (Ref 8). Without adding extra binder, SLS is widely used to fabricate powder-based material, such as metal, ceramics, and plastics (Ref 9). In SLS, a high-power laser beam is used to irradiate the selected ceramic powder, and the targeted powders will be heated up and sintered to the bulk joining point (Ref 9). Unlike binder jetting, which requires postprocessing to remove the binder, SLS can directly obtain the 3D-printed ceramic part without postprocessing. In general, powderbased ceramic 3D printing shows advantages in the fabrication of parts with complex cellular structures (Ref 2, 7–9). In slurry-based ceramic printing, highviscous slurry composed of ceramic particles and other liquids is deposited in a layer-bylayer manner to form a 3D object (Ref 2). For example, in FDM-based ceramic printing, a filament is fed and heated to a molten or semimolten state. The melted slurry extruded from the nozzle under the pressure of a piston will deposit and fuse with the adjacent part that has already been deposited (Ref 10, 11). However, it is difficult to directly shape ceramics into filaments, due to their brittle mechanical property. Therefore, to use FDM to fabricate ceramic components, the ceramic particles are mixed with thermoplastic binders to form filament feedstock (Fig. 2) (Ref 12). After finishing the printing process, debinding and sintering must be conducted to accomplish binder removal and densification (Ref 11, 12). Ceramic transducers (Ref 12) and bioceramicbased scaffolds (Ref 13–16) with lattice structures having spatial resolutions smaller than 100 mm can be fabricated by using FDM-based ceramic printing. Ceramic-based slurry can also be fabricated by using DIW, in which an ink-deposition nozzle is used to deposit the

Fig. 2

slurry to generate the 3D architecture and composition (Fig. 2) (Ref 17). Compared with other ceramic 3D printing processes, DIW is a cheap and fast manufacturing process that has the capability of fabricating numerous different structures, ranging from solid monolithic parts (Ref 18) to complex porous scaffolds (Ref 19). However, the density of a ceramic part fabricated by the DIW process is not as high as that of the powder-based approaches. This is because the viscosity of ink goes up dramatically with the increase in ceramic particle concentration, and the highviscous slurry is hard to extrude from the ink-deposition nozzle. In addition, the ceramic particle size should be much smaller than the nozzle tip; otherwise, the ceramic particles will easily block the nozzle tip (Ref 17). Most of the slurry-based ceramic fabrication using 3D printing in modern times has evolved to be a combination of AM and heat-treatmentbased consolidation (Ref 2). Such a process begins with green-part fabrication of a ceramicpolymer composite by using various types of 3D printing processes, such as FDM (Ref 13–15), DIW (Ref 17–19), TPP (Ref 2), and VPP (Ref 9). A heat treatment procedure, including debinding and sintering, is necessary to form the consolidated part (Ref 2). The ceramic-polymer composite is also constantly investigated to find the best degree of manufacturability, where the viscosity of the slurry is adjusted to provide a sufficient density of the ceramic ingredient within the mixture (Ref 2, 4).

Vat Photopolymerization The AM revolution has brought new possibilities in designing and building objects with complex freeform surfaces (Ref 20, 21). Various AM processes were developed to overcome the difficulty of traditional manufacturing in terms of 3D fabrication, and a

Current additive manufacturing processes developed to fabricate ceramic components. Source: Ref 2

large range of materials, including difficultto-process material, can now be formed into 3D shapes using AM technologies (Ref 22, 23). Due to its advantages, AM technologies have been widely used in every area of social life, such as aerospace, biomedical engineering, civil engineering, electronics, and so on (Ref 20, 21). Among all the AM processes, VPP shows unique capability and demonstrates superiority in a wide variety of applications (Ref 24, 25). Vat photopolymerization was developed based on one type of chemical reaction, called photopolymerization, in which the photocurable polymer is cross linked by the initiation of light exposure (Ref 23). The material phase transforms from liquid to solid, forming a linear or cross-linked polymer structure in the photopolymerization process (Ref 23). In VPP, a vat of liquid resin, which is composed of monomers, oligomers, and photoinitiators, undergoes the aforementioned chemical reaction and selectively accumulates into a 3D shape defined by the exposed ultraviolet (UV) light (Ref 23). Various VPP processes have been developed to improve printing capability, including laser writing stereolithography (LWSL) (Ref 26), mask-image-projection-based stereolithography (MIP-SL) (Ref 27–31), continuous light interface process (Ref 32), physical-field-assistedstereolithography (Ref 33), and TPP (Ref 34). In LWSL, a laser beam with controllable power and wavelength is reflected by highspeed-scanning galvo mirrors and is further focused on the surface of the liquid resin (Ref 26). Following the generated tool path, the laser beam solidifies the liquid polymer into a special two-dimensional (2D) pattern. After that, the platform moves down a distance of one layer thickness, and liquid resin fills back to the fabrication area to facilitate the fabrication of the next layer (Ref 23). To achieve multiscale fabrication, the laser beam size can be dynamically adjusted based on the geometric shape of the printed object

Vat-Photopolymerization-Based Ceramic Manufacturing / 83 (Ref 26). In MIP-SL, a digital micromirror device is used to generate a 2D-patterned light beam, and the whole layer of resin can be cured with only a single exposure (Ref 23, 32, 33). Based on the oxygen-inhibition layer and two-way movements, centimeters-tall objects can be printed within several minutes, which is 100 times faster than other AM approaches (Ref 30, 31). In TPP, an ultrafast laser beam is used to irradiate the photopolymer, which requires much high energy to be cross linked (Ref 34, 35). Because TPP is a volume-based 3D printing process, there is no topological constraint, and complex 3D shapes with submicrometer features can be directly printed using TPP (Ref 35). Due to its unique capability, TPP has been applied to many applications, such as micro/nanophotonics, microelectromechanical systems, microfluidics, biomedical implants, and microscale devices (Ref 34). A large range of materials, including plastics (Ref 36), ceramics (Ref 2), metals (Ref 37), and composite materials (Ref 38), are used in VPP. When using VPP to fabricate noncurable materials, a mixture composed of both noncurable material and photopolymer is formed into 3D shape during photopolymerization of the photocurable polymer (Ref 2, 32, 37). For example, a nanoscale graphene platelet based composite material was formed into nacre-inspired structures by using electrically assisted MIP-SL, and the printed material showed mechanical reinforcement and electrical self-sensing capabilities (Ref 32). To obtain pure noncurable material, the photopolymer in the fabricated object can be removed by postprocessing methods, such as chemical dissolution (Ref 39), thermal decomposition (Ref 3), or microwave sintering (Ref 40). For example, both pure metallic and ceramic microscale lattices, which were extraordinary light and stiff, were fabricated by using microscale MIP-SL. After the printing process, thermal decomposition and sintering were applied to remove the polymer and to densify the metallic and ceramic (Ref 39). Similarly, 3D-shaped fused silica glass designed with complex inner structures was obtained after debinding and sintering the polymerized composite fabricated by the VPP process (Ref 41). Due to its unique strengths, such as high detail accuracy (Ref 29), geometric complexity (Ref 39), smooth surface quality (Ref 25, 42), large materials selection (Ref 32, 33, 36, 37), and fast speed (Ref 27, 30, 31), VPP is one of the more popular 3D printing techniques (Ref 43, 44).

Vat-Photopolymerization-Based Ceramic Fabrication Similar to polymers, a ceramic part can be fabricated by the VPP-based 3D printing process with high resolution and surface quality (Ref 45). The procedure of ceramic fabrication by using the VPP process is shown in Fig. 3. First, the microscale or nanoscale ceramic

Fig. 3

Flowchart of vat photopolymerization (VPP)-based ceramic fabrication

particles are mixed with the photocurable resin; then, using the slurry mixture, the green part is prepared by VPP-based ceramic printing (Ref 3). In the VPP-based ceramic printing process, the photopolymer mixed with the ceramic particles undergoes a chemical reaction (photopolymerization) and further forms a solid part defined by the radiation light in the UV range of wavelengths (Ref 45). After that, general postprocessing, including lowtemperature debinding and high-temperature sintering, is necessary to remove the inner polymer and fuse the ceramic particles together (Ref 3). Information and methods of material preparation, green-part fabrication, postprocessing, property identification, and polymerderived ceramics are introduced in this section.

Material Preparation To print ceramic components using the VPP-based 3D printing process, a ceramic composite slurry is prepared by mixing ceramic filler particles with photocurable resin (Ref 45). During the fabrication process, the slurry is selectively cured by light exposure, and the resin serves as a binder to bond ceramic particles into the desired 3D shape (Ref 3). As the first step, the preparation of the slurry is critical to the whole fabrication process. Specifically, the ceramic powder must be nonagglomerating in an azeotropic mixture, with the dispersant added by ball milling to obtain a homogeneous distribution (Ref 3). Often, the dispersant can simply be added to the monomer solution. Figure 4(a) shows a flowchart for slurry preparation. After evaporation of the solvent, the dried ceramic particles with dispersant adsorbed onto their surface will be mixed with a photocurable resin. Then, the premixed resin and ceramic particles go through the ball-milling process again until

the mixture becomes a homogeneous slurry (Ref 3). Homogeneity of the slurry, which determines the quality of a 3D-printed part, is an important factor that must be considered during the manufacturing process. The inhomogeneity of ceramic particles in the slurry caused by sedimentation, which results from the aggregation of ceramic particles, will generate defects in the final ceramic part after removing the polymer (Fig. 4b) (Ref 3). To control the slurry homogeneity, one must understand the slurry-preparation mechanism. The slurry is a mixture of ceramic particles and photocurable resin. These micro- or nanoscale ceramic particles inside the slurry are extremely easy to aggregate due to van der Waals attractive forces (Ref 3). Due to particle aggregation, the inhomogeneity of the slurry causes nonisotropic distribution of ceramic particles inside the final green part (Ref 3). Hence, defects and failure, such as cracking and delamination, may occur during postprocessing due to the nonuniform stress inside the green part, caused by the inhomogeneous distribution of ceramic particles (Fig. 4b) (Ref 3). To eliminate this problem, the ceramic-based slurry should be prepared according to the procedures discussed previously.

Curing Characteristics After the slurry is prepared, the curing performance of the ceramic slurry should be studied to obtain the best fabrication parameter values. Compared to the pure photocurable resin, the curing characteristics of the ceramic composite slurry are quite different (Ref 46). When light travels through the slurry, the direction of the light will be changed due to the scattering effect of the ceramic particles inside the ceramic slurry (Ref 47). Therefore, the

84 / Ceramic Additive Manufacturing Processes penetration depth of the light and the curing depth for the slurry are dramatically decreased. Based on a study (Ref 3), the cure depth for ceramic composite slurry is mainly affected by the resin sensitivity and the ceramic refractive index. The penetration depth of light, Dp, can be determined by (Ref 3): 1 ¼ eP cP þ eD cD þ ðb  eP cP þ eD cD Þf Dp
 b f2  2fmax

(Eq 1)

where eP and eD are molar extinction coefficients of the photoinitiator and dye, respectively; cP and cD are the concentrations in volume unit of photoinitiator and dye in the slurry, respectively; b is the variation in scattering, determined by the refractive index

contrast; f is the volume fraction of ceramic particles; and fmax is the maximum concentration of ceramic inside the slurry. Overall, the cure depth of the slurry is determined by the refractive index difference between the filler ceramic particles and the photocurable resin, the particle size, and the solid loading of ceramic particles in the slurry (Ref 3, 46). After the cure depth is decided, the layer thickness should be smaller than the cure depth of slurry, so that the newly cured layer can be attached on the surface of the previously cured layers.

Green-Part Fabrication The viscosity of the ceramic composite slurry is dramatically increased with the increment of ceramic particles concentration. To

fabricate by using the layer-by-layer approach, the rheological characteristic of the ceramic slurry should be studied in the VPP-based ceramic printing process (Ref 48). Due to relatively low viscosity and good flowability, the pure photocurable resin is able to refill small gaps by itself in the normal VPP process (Ref 49). However, because of its poor flowability, the ceramic composite slurry with high viscosity cannot self-refill a gap if driven only by air pressure and self-gravity (Ref 48). To solve the filling problem, a doctor-blade-based material feeding system, which can recoat a thin layer of slurry with desirable thickness, was introduced in the VPP-based ceramic printing process (Ref 3, 49). The basic design of a VPP-based ceramic fabrication machine is shown in Fig. 5(a). To fabricate the green

Fig. 4

Material preparation in vat-photopolymerization-based ceramic fabrication. (a) Flowchart of slurry-preparation procedure. (b) Homogeneity problem in slurry fabrication

Fig. 5

Green-part fabrication by using the vat photopolymerization (VPP)-based ceramic printing process. (a) Diagram of VPP-based ceramic printing. DMD, digital micromirror device; LED, light-emitting diode. (b) Set of projection images generated by slicing the digital model in VPP-based ceramic printing. (c) Process planning of green-part fabrication by using the VPP-based ceramic printing process

Vat-Photopolymerization-Based Ceramic Manufacturing / 85 part, a series of 2D mask images are generated by slicing the digital model of the building part (Fig. 5b) (Ref 3). The projection light defined by each mask image is exposed on the surface of the recoated slurry in sequence. After one layer is cured solid, a small amount of slurry is fed, and a uniform layer of slurry will be recoated on the glass plate by the doctor blade. After that, the light coming from the bottom will project onto the surface of the slurry, and a new layer of ceramic will solidify and attach onto the surface of the precured layers (Fig. 5c). By repeating the process layer by layer, a green part with desired 3D shape is formed by VPP-based ceramic printing.

Process Design and Planning Because the properties of the slurry significantly influence the performance of the VPP-based ceramic printing process, the relationship between material properties and process parameters must be studied in order to set process parameters to successfully build a ceramic component (Ref 3). For each layer, a small amount of slurry is deposited on the fabrication surface behind the doctor blade; then, the blade pushes the slurry to form a thin layer of material with desirable thickness. However, to achieve the desired thickness, several process parameters, including gap distance between the blade and the synthetic fluorine-containing resin film, the moving speed of the film collector, and others, must be considered (Fig. 6b). Because the pressure in the dispensed slurry can be ignored, the blade recoating procedure can be modeled with a plane Couette flow pattern (Ref 48, 50). As shown in Fig. 6(b), the bottom plate moves at a constant speed of Vr relative to the blade. To

Fig. 6

generate the uniform thickness, d, of a slurry layer, the gap, gblade, of the blade should be:

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ½sD2 0 < Vr < 0 8rCD d L

(Eq 3)

0

gblade >

2ar d rr

(Eq 2)

where r is the slurry density, and r0 is the density of the newly formed layer, and r is the width ratio of the slurry layer before and after recoating. To ensure layer bonding, a is a safety factor to ensure that neighboring layers will be bonded (a > 1). When the ceramic slurry passes the doctor blade, it has a lower viscosity, caused by a higher shear rate due to the shear thinning behavior of the ceramic slurry (Ref 3). Thus, sedimentation of ceramic particles in the recoated layer can be avoided, and good homogeneity of the fabricated green parts can be achieved (Ref 3). With the same blade gap, g, a higher recoating speed, Vr, or a larger slurry viscosity will generate a bigger shear rate, g (Ref 48). In the printing process, the platform goes up a small distance (smaller than the cure depth of the slurry) from the fabrication surface after the fabrication of each layer. Then, the film collector moves in the x-axis with a speed, Vr, to detach the newly cured layer from the fabrication surface, and a new layer of slurry is coated at a thickness of d0 (Ref 48). As shown in Fig. 5(c), when the part slides from left to right, the portion of the cylinder immersed in the slurry layer experiences a drag force, F, in the slurry moving direction as the average force, q (q = F/d0 ) (Ref 3). Suppose a cylindrical shape is being fabricated whose diameter is D and the current building height is L. The sliding speed of Vr should be set in the range shown in Eq 3 to avoid damage to the printed features:

where s is the bending stress of the cured material; r is the mass density of the slurry; and CD is the drag coefficient, which is related to the Reynolds number, Re, of the slurry and can be identified by experiments. The conventional material recoating method is shown in Fig. 5(c); the film collector must move from left to right for layer recoating and then move from right to left for the generation of a new layer. This method spends more time on the motion of the film collector (Ref 3). A fast recoating method was developed to reduce the motion time and to speed up the material feeding process in the VPP-based printing process (Fig. 6a) (Ref 33). In this continuous feeding process, the film collector was designed to be a circular plate, which was mounted on the rotary stage, and the layer recoating can be achieved by rotation the plate (Ref 33). The slurry was continuously extruded from the gap generated by the doctor blade and spread to a thin film during the continuous rotation in one direction (Ref 33).

Support Development To fabricate ceramics with complex inner structures such as overhangs, supports must be designed and fabricated with the structure. However, a support brings many problems to the 3D printing of a single material (Ref 1). For example, it is very difficult to remove supports without causing extra damage to parts with delicate features. To overcome these challenges, many research activities have been done to optimize support generation in the VPP-based ceramic printing process. New

Material recoating process in the vat-photopolymerization-based ceramic printing process. (a) Slurry recoating process based on rotation. (b) Side view of slurry recoating. (c) Top view of slurry recoating

86 / Ceramic Additive Manufacturing Processes methods for support detection were developed for the VPP process in the past decades. For example, a surface element (surfel) convolutional neural-network-based approach was developed to reduce the number of supports and increase the accuracy of the support position during support detection (Ref 51). This method produced a local surfel image, which contained the local topology information of the sampling point defined by the layered depth-normal image, and a set of models with ground-truth support regions were used to train the deep neural network (Ref 51). In addition, researchers have studied some novel approaches to developing a highly removable support structure for the fabrication process. Researchers used solid ice as the support material, which can be easily removed after fabrication without causing any damage to the component surface (Ref 52). During the 3D printing process, a cooling device was used to freeze water, and then the photocurable resin that is spread on the ice surface is solidified by projection light (Ref 52). Both approaches can be applied to VPP-based ceramic fabrication for support generation. Furthermore, some support-free fabrication methods have been developed to improve the quality of 3D-printed green parts. A highyield-stress ceramic slurry was used as the feedstock material, which exploits the elastic force of the material to support overhanging features without the need to build additional support structures (Ref 53). Based on that, Song et al. developed a new ceramic stereolithography (CSL) method that avoids the need for additional support (Ref 54). In this CSLbased approach, an elastoviscoplastic ceramic suspension that has inherently strong

Fig. 7

interparticle force was used as a support bed, so that an additional support structure is not necessary when fabricating a part with overhang features (Fig. 7) (Ref 54).

process parameters of VPP-based ceramic printing can be further optimized to achieve better quality control.

Quality Control and Optimization

Postprocessing

To achieve high quality and reproducibility, it is important to have quality control in the VPP-based ceramic printing process. In addition, due to the complexity of the printing and postprocessing processes, the optimization of process parameters is also indispensable. Many research studies have been conducted to study quality control and optimization in VPP. For example, Xu et al. studied part-fabrication mechanisms and determined that three most significant sources, such as over- or underexposure, light blurring, and phase change, induced shrinkage or expansion of the part printed by the VPP process (Ref 55). Zhou et al. increased the printing accuracy of the VPP process by optimizing process parameters (Ref 56). By measuring the printed part quality, the manufacturing process parameters, including layer thickness, resultant overcure, hatch space, blade gap, and part location, were optimized to improve the printing quality (Ref 56). Polymerization shrinkage and thermal cooling effect are two major factors that lead to curl distortion in the VPP process. Kai et al. studied the photocuring temperature during the MIP-SL process by using a highresolution infrared camera. After the experimental study, they discovered the temperature distribution during the printing process and further developed some new exposure strategies that can effectively reduce curl distortion (Ref 57). Based on these previous studies, the

Debinding The debinding process is aimed at thoroughly removing the polymer ingredient, which is used as the binder in the material mixture in order to form the desired part geometry in the VPP-based ceramic printing process (Fig. 8a) (Ref 58). Particularly, the photocurable polymer is mixed with the desired ceramic particles and cured to the intended shape, which is called the green part, under a projected light or laser (Ref 3). Subsequently, the debinding process is conducted on the green part to pyrolyze the polymer inside through a controlled process of raising the temperature to 600  C (1110  F) in a furnace environment (Ref 59). The green part is initially heated from room temperature, followed by a process of slowly ramping up the temperature to prevent heat-formed cracking (Ref 59). At a certain threshold temperature, the value is held for a certain amount of time to heat the part thoroughly before the temperature is again raised to a higher stage. For example, barium titanate (BTO) is heated at a rate of 1  C/min (1.8  F/min), with incremental pauses for 30 min at temperature thresholds of 200, 300, 400, and 500  C (390, 570, 750, 930  F), respectively. The temperature eventually reaches and is maintained at 600  C, upon which the polymer continues to pyrolyze for 3 h toward full vaporization (Fig. 8d) (Ref 60).

No-support ceramic fabrication by the vat-photopolymerization-based ceramic printing process. (a–b) Schematic diagram of suspension-enclosing projection stereolithography. (c) Procedure for one-layer fabrication. (d) Supporting mechanism for overhang feature fabrication. (e) Three-dimensional-printed ceramic parts without added supports. Source: Ref 54

Vat-Photopolymerization-Based Ceramic Manufacturing / 87 Currently, major debinding approaches are mainly categorized into thermal- (Ref 59), solvent- (Ref 61), and microwave- (Ref 62) based methods (Ref 3). In thermal-based debinding methods, vacuum debinding eliminates damage caused by oxidization of the materials and features a reduced cycle time for the process (Ref 60). Typically, debinding is done in air. The choice of debinding atmosphere is often determined by the powder. Inert-gas-based debinding can have a similar effect as the vacuum debinding process. Gases, such as argon and nitrogen, are applied in the debinding environment to prevent oxidization and the corresponding damage to the fragile green part. In both vacuum- and inert-gas-based debinding processes, a vital process specification is evacuation of the vaporized polymer (Ref 60). For solid parts at relatively large scale, the vaporized polymer is prone to be trapped inside the object, causing defective parts with cracks, notches, and unpredictable fractures, whereas a part with a thin-wall feature, interconnected and porous structure, or simply smaller scale of dimension would have a superior performance with respect to gas evacuation of the vaporized polymer. In practice, vaporized and evacuated polymers would turn into carbon, which, in turn, attaches to the surface of a part, blackening the part in high temperature while leaving pores in the meantime (Fig. 8b).

Sintering By definition, the sintering process of ceramic is to fuse the ceramic particles

Fig. 8

together to fill the porosity left by pyrolyzed polymer (Fig. 8c). There are two major approaches currently being adopted: thermaland microwave-based sintering processes (Ref 40, 60). Thermal-based sintering is normally a process that heats the part from more than 1000  C (1830  F) to achieve grain growth. Similar to the debinding process, the procedure begins with heating the part from room temperature, followed by faster temperature increments usually at a rate that is higher than the one used in the debinding process. Similarly, the temperature is then held at certain levels to heat the part thoroughly; eventually, it is raised to and kept at a sufficient level for grain growth to take place. The highest sintering temperature and the time needed are determined by the ceramic material to be fabricated (Ref 63). As an example, hydroxyapatite/ tricalcium phosphate (HA/TCP) is heated at a rate of 3  C/min (5.4  F/min), and the temperature is held for 30 min, respectively, at 200, 300, 600, 1000, and 1100  C (390, 570, 1110, 1830, and 2010  F), with the last temperature step being kept at 1250  C (2280  F) for 3 h (Fig. 8d). In comparison, BTO exhibits a different heating curve than HA/TCP, with a higher temperature of 1330  C (2425  F) being held for 6 h for the final sintering step (Ref 60). The microwave sintering process, compared with the thermal sintering process, is of higher power at the same level of energy consumption (Ref 40). Microwave energy penetrates the green part, is absorbed by the matter, and is uniformly transformed into heat inside the

part, while thermal sintering must inefficiently transport heat from outside to inside (Ref 40). As a result, microwave sintering requires less sintering time and has controllable grain growth, which leads to less porosity and higher densification. The major drawback of microwave sintering, however, is the cost and the equipment capability in handling different scales of parts (Ref 63). The equipment of microwave sintering is relatively more expensive than that of thermal sintering and is usually restricted in size capability, which would be a significant limitation in some dimensionally focused cases.

Shrinkage and Compensation The phenomenon of shrinkage is inevitably associated with the removal of polymer in the debinding process and the densification in the sintering process (Fig. 8) (Ref 63). The vaporization of polymer creates voids between grains, and the sintering of particles diminishes those voids as much as possible. However, some voids remain that may cause shrinkage upon the removal of external energy (Ref 63). In practice, BTO has a measured shrinkage of 26.7% on the xy plane and 34.3% along the z-direction after sintering at 1330  C (2425  F) (Ref 60); similarly, alumina has a total shrinkage of 22.7% after sintering at 1650  C (3000  F) (Ref 64), and HA/TCP possesses a shrinkage of 30% after sintering at 1250  C (2280  F) (Ref 93) (Fig. 9a–c). Shrinkage not only leads to changes in size of the 3D-printed part but also produces residual stress, which incurs

Postprocessing of a green part printed by the vat-photopolymerization-based ceramic printing process. (a) Ceramic part before and after postprocessing. (b) Scanning electron microscopy (SEM) images of brown part after the debinding process. HA/TCP, hydroxyapatite/tricalcium phosphate. (c) SEM images of final part after the sintering process. (d) Temperature curves of the debinding and sintering processes for fabrication of a bioceramic part

88 / Ceramic Additive Manufacturing Processes deformation on the part. With the shrinkage problem being the predominant cause of fabrication failures, some compensation methods are considered and investigated to assist in retaining the desired shape during the sintering process. Various approaches for size compensation in the design phase have been introduced based on modeling of the shrinkage problem. To compensate for uniform shrinkage, the commonly used method is to apply a dimensional offset to the part design at a volume of the statistically computed or actually observed shrinkage scale (Fig. 9d). Therefore, the design must be expanded with the offset operation, to leave stock for a predictable or computable shrinkage that will occur during or after sintering and thus to be close to the truly desired part (Fig. 9e) (Ref 66). In a more complicated but more common case, nonuniform shrinkage is desired; hence, a modern method is mostly based on machine learning techniques, where the dimensional difference between the printed and designed parts is demystified by the mass volume of data training. Accordingly, the dimensional input dumped to the model can derive a predictable result in the shrinkage, and a nonuniform offset can be applicable to the input design (Ref 67–69). Other than the design approaches, there are also some process-related methods to mitigate the shrinkage issue and its consequences. For instance, to compensate the deformation caused by shrinkage and its consequential residual stress, the

projection mask image parameters in the MIPSL process, for example, light intensity, gray scale, and so on, can be altered and fine-tuned to purposely reinforce or weaken suspicious deformation zones (Ref 70, 71).

Property Identification Porosity Pores are mostly generated at the debinding process when polymer is pyrolyzed (Ref 58). In the sintering afterward, the highest temperature, usually up to the range from 1200 to 1500  C (2190 to 2730  F), would cause densification of the ceramic part (Ref 3, 63). Sintering at such high temperatures is applied to the grain boundaries so that most of the pores disappear during grain growth, while some pores remain and may be difficult to eliminate due to their diffusion paths that are excessively long among the large grains. In most cases, the sintering process diminishes the pores to some extent and also alleviates the negative influences of pores on material properties such as strength (Ref 63). However, there are also certain applications where porosity is desired and important in order to address functional needs, such as cell attachment to scaffolds (Ref 40). Porosity is determined primarily by sintering temperature, time, and ceramic particle size (Ref 63). Several methods, such as

mechanical test, density measurements, porosimetry, gas absorption for surface area, quantitative microscopy, and so on, can be used to gage or evaluate porosity and its variation (Ref 63). In real-world situations, the geometry and arrangement of ceramic particles, although measurable, are usually unpredictable and uncontrollable due to interaction and overlapping among localized mechanisms (Ref 3). Inconsistency and randomness are always associated with porosity, even in the situation where parts are built in the same way using the same material and equipment. Although deviation widely exists, adjusting the sintering temperature and ceramic particle size are proven to be significant factors that can be used to manipulate porosity (Ref 58–60). Temperature is negatively proportional to the porosity level, meaning that higher sintering temperature usually results in lower porosity, whereas ceramic particle dimension addresses porosity in an opposite way, for example, larger particle size results in larger porosity. Additionally, with all things being equal, the process of sintering has a significant impact on porosity as well. The microwave sintering process outperforms the traditional thermal sintering process in downgrading porosity at the same condition (Ref 40). As a general example of the aforementioned variables, porosity is measured for a part with a desired open porosity of 27% for the sintering of ceramic fabricated by a particle size of 500 mm. At a thermal sintering temperature of 1150  C (2100  F), open porosity is approximately 57.87%; at a thermal sintering temperature of 1250  C (2280  F), open porosity is reduced to 54.11% (Ref 40). Moving to the case of microwave sintering, open porosity is further reduced to 51.4% at a temperature of 1150  C (Ref 40). Another designed part has a particle size of 750 nm. In thermal sintering with a temperature of 1150  C, open porosity is approximately 63.1%; in thermal sintering with a temperature up to 1250  C, open porosity is also reduced to 58.61%. In comparison, open porosity is approximately 59.76% in microwave sintering at a temperature of 1150  C (Ref 40). Lastly, adding dissolvable particles to the material mixture is another effective way of adjusting porosity for the sintering process. An experimental case, where sugar was added to the material, significantly brought the porosity to 50% in the fabrication, based on photocurable resin (Fig. 10) (Ref 72). In addition to sugar, other dissolvable particles such as salt have a similar effect.

Bioceramics Fabrication Fig. 9 Shrinkage and compensation in vat photopolymerization (VPP)-based ceramic printing. (a) Ceramic parts before and after postprocessing. (b) Shrinkage of ceramic part fabricated by microscale VPP-based ceramic printing. (c) Final ceramic part after postprocessing. (d) Compensation by redesign of the build-part digital model based on the shrinkage ratio. (e) Green parts before and after compensation. Source: Ref 65

With the development of tissue engineering and regenerative medicine, 3D scaffolds with a porous structure have been used for hard tissue regeneration. The porous structure provides enough space for cells to attach and

Vat-Photopolymerization-Based Ceramic Manufacturing / 89 enables blood vessels to deliver nutrition through the inner connective pore network. Bioceramics have been widely used in the fabrication of 3D scaffolds and bone implants due to their good biocompatibility and high mechanical performance (Ref 73). Bioceramics such as hydroxyapatite, calcium phosphate (CAP), bioglass, and graphite show promising properties for tissue engineering applications (Ref 74). Bioceramics fabricated by traditional methods, such as freeze casting (Ref 75), foam

replica methods (Ref 76), high-pressure pressing (Ref 77), and particle leaching (Ref 78), can achieve a large range of porosity, but only simple structures can be formed. The VPPbased ceramic printing process can solve these challenges, and biomimetic hierarchical porous structures with high mechanical strength can now be fabricated by using a multiscale VPP process (Ref 79). For example, a shell-shaped scaffold was designed and fabricated by using MIP-SL

for long-bone critical-defect regeneration (Fig. 11a). A photocurable polymer was mixed with 30% (weight/weight) CAP, and the viscosity of this CAP-based slurry was 5375 MPa  s, which is difficult to fabricate using the extrusion-based AM processes. As shown in Fig. 11(a), a 3D scaffold with complex geometric structures was designed based on a digital model of the bone defects. Hierarchical features, ranging from the microscale porous structure (diameter: 20 to 1000 mm) to interconnected small pores (50 vol%) aqueous ceramic paste containing 1 to 4 vol% of organic additives is extruded in a freezing environment (99% TD 99% TD 82% TD 90% TD >95% TD ... ... ... >93% TD

2011 2012 2000 2017 2016 2018 2004 2016 2006 2007 2007

86 94 95 96 97 98 99 100 101 102 102

200 30 510

~1% porosity ... 92–98% TD

2002 103 2013 104 2017 105

(a) Finest resolution of reported samples. (b) Usually measured using Archimedes principle.; TD, theoretical density; PEI, polyethyleneimine; APA, ammonium polyacrylate; HPMC, hydroxypropyl methylcellulose; EC, ethylcellulose; PVA, poly(vinyl alcohol); PEG, poly(ethylene glycol); PVB, poly(vinyl butyral); PLGA, poly(lactic-co-glycolic) acid; DCM, dichloromethane. EGDE: ethyleneglycol butylether; DBP: dibutyl phthalate; water: deionized water. Source: Ref 82

1

2

2

3

1

5 mm

(a) a1 b1 c1 a2 d3 c2 d4 a3 d5 c3 d6 a4 d7 c4 d8 a5 d9 c5 d10

3

1

2

2 mm

(b)

Fig. 6

Robocast BaTiO3-Ni multimaterial structures: (a) optical images, (b) partial cross section. Adapted from Ref 102 with permission from Wiley

102 / Ceramic Additive Manufacturing Processes

BioGlass

200 µm

200 µm

(a)

200 µm

(b)

400 µm

(e)

Fig. 7

(d) ZnO

Y2O3/ZrO2

100 µm

500 µm

1 cm

(c)

AI2O3

Si3N4

SiC

HA

Ba TiO3

500 µm

300 µm

(f)

(g)

(h)

SEM images of sintered ceramic structures fabricated by robocasting: (a) bioglass, (b) hydroxyapatite (HA), (c) silicon carbide, (d) silicon nitride, (e) alumina, (f) yttriastabilized zirconia, (g) zinc oxide, and (h) barium titanate. Adapted from Ref 82 with permission from Wiley-vch.de

(rapid freezing) was used to maintain the shape. The frozen ceramic paste offered a higher strength compared with parts printed by robocasting. Moreover, the absence of nonuniform drying enabled the FEF process to print nonsparse structural ceramic components without the risk of warping, cracking, and delamination. The FEF process has been used to fabricate ceramic components from various materials, such as Al2O3 and ZrB2 (Fig. 9), ZrO2 (Ref 111), ZrC, and W (Ref 113). Besides fabricating ceramic parts using a single extruder, the FEF system was also used to fabricate parts by extruding multiple pastes. For example, Al2O3 parts with overhangs were fabricated by using two discrete print heads; one nozzle for depositing Al2O3 paste and the other for depositing support material (methylcellulose solution) (Ref 114). Functionally graded material (FGM) specimens have also been fabricated by extruding different ceramic pastes at a changing ratio of extrusion speeds into a mixing chamber, followed by depositing the mixed paste through a single nozzle. Figure 10 shows two FGM parts, in which the change of the color along the vertical build direction reflects the changing of material ratio in the FGM sample. Despite the advantages of good green part strength and low risk of part warpage and cracking, the FEF process suffers from issues caused by freezing. The motion stage and extruders are required to work at a freezing temperature (typically between
10 and
20  C, or 14 and
5  F), while the paste flow path must be kept at room temperature to avoid the paste being frozen before deposition. As a result, the cost of the FEF system was higher

Fig. 8

(a)

Fig. 9

Apparatus for the freeze-form extrusion fabrication (FEF) process in a thermo-insulated chamber equipped with a triple-extrusion device for extruding and blending multiple pastes. Source: Ref 111

45 mm

(b)

10 mm

(c)

10 mm

Sintered ceramic parts fabricated using the freeze-form extrusion fabrication (FEF) process: (a) Al2O3 cones, (b) ZrB2 cones. Source: Ref 111. Reprinted with permission from Elsevier. (c) Al2O3 cuboid with through holes fabricated using sacrificial support material. Source: Ref 112

Material Extrusion Based Ceramic Additive Manufacturing / 103 than normal MECAM systems. Also, more sophisticated process control was needed to rapidly freeze the material after extrusion from the nozzle, yet at the same time to prevent clogging of the nozzle in the freezing environment. Numerical analysis and experimental validation (Ref 115–117) of the FEF process have been carried out to optimize process parameters, as well as precise control of paste extrusion (Ref 118, 119). Ceramic parts with ~92% relative density after sintering were achieved (Ref 119). The formation of ice crystals in the part and the difficulty of accurate control of paste flow rate were the main factors that limited the relative density and strength of the FEF fabricated parts (Ref 119). Ceramic-On-Demand Extrusion The use of aqueous ceramic pastes eliminates the issue of binder burnout in the robocasting (RC) and freeze-form extrusion fabrication (FEF) processes. This is an advantage compared with extrusion freeform fabrication (EFF) and fused deposition of ceramics (FDC). However, the challenges of part deformation for bulk structures in RC and ice crystal formation from FEF have limited the techniques to fabricating highly dense (>99% relative density), bulky ceramic components for use in structural applications. The ceramic on-demand extrusion (CODE) process was proposed and developed (Ref 51, 53–55, 57, 58) to address the challenging issues associated with the RC and FEF processes. In the CODE process, aqueous ceramic pastes are deposited on a substrate located in a tank designed to hold a fluid medium (typically light mineral oil). After deposition of each layer, a liquid feeding subsystem pumps oil into the tank surrounding the layer to preclude undesirable water evaporation from the sides of the deposited layers (Ref 121). The liquid level is maintained at a level just below the top surface of the part being fabricated. Infrared radiation is used to uniformly dry the deposited layer for shape retention. The layered uniform radiation drying approach minimizes the water content gradient in the fabricated part, and, thus, enables producing crack-free thick ceramic parts with complex geometries using the CODE process (Ref 57–59, 62). A schematic of the CODE process and an image of the CODE process setup is shown in Fig. 11. Using the FDC, EFF, RC, and FEF processes to fabricate parts having both high density (~99% of TD) and large part thickness presented challenges leading to research to improve the precision of extrusion deposition, improve paste homogeneity and degassing, and implement more sophisticated extrusion mechanisms to optimize the CODE process. Use of a high-precision auger extruder and fine-tuning paste preparation with the CODE process enabled achieving >99% TD in bulk ceramics with mechanical properties such as the hardness, flexural strength, and fracture toughness comparable to those of parts made using conventional shaping techniques

(Ref 123). Figure 12 shows examples of CODE-fabricated nonsparse ceramic samples, each of which achieved at least 98% of relative

density. The improvement in the precision of paste extrusion in the CODE process also benefited the fabrication of ceramic parts with

Fig. 10

Parts made of functionally graded material (FGM) fabricated using the freeze-form extrusion fabrication (FEF) process: (a) Cylinder fabricated from pink and green colored CaCO3 pastes, (b) thick wall graded from 100% Al2O3 to 50% Al2O3 + 50% ZrO2 pastes in different colors. Source: Ref 111. Reprinted with permission from Elsevier

(a)

(b)

Fig. 11

(a) Schematic diagram of the ceramic on-demand extrusion (CODE) process, (b) image of the CODE apparatus equipped with dual extruders. Source: Ref 122. Reprinted with permission from Elsevier

Fig. 12

Highly dense (nonsparse) solid ceramic samples fabricated using the ceramic on-demand extrusion (CODE) process: (a) ZrO2 (3 mol% Y2O3) block, (b) ZrO2 (8 mol% Y2O3) gear, (c) ZrO2 (3 mol% Y2O3) helical gear and cross section of its CAD model. Source: Ref 59. Reprinted with permission from Elsevier.

104 / Ceramic Additive Manufacturing Processes intricate geometries and fine features, such as the gears in Fig. 12 and fine structures shown in Fig. 13. Besides enhancing fabrication precision research to develop sacrificial support materials has greatly broadened the range of part geometries that can be fabricated by MECAM. Li, et al. (Ref 90) proposed calcium carbonate (CaCO3) as a sacrificial material, which decomposes during sintering and can be removed by dissolving in water or acid after sintering. Aqueous CaCO3 paste was prepared

as the support material and was concurrently deposited with Al2O3 paste using a workhead with dual nozzles for the part and sacrificial materials. A multistep sintering technique was developed for Al2O3 parts due to the favorable phase equilibrium between Al2O3 and CaCO3 within the sintering temperature range. The support material was removed by dissolving in water before the final sintering step. Figure 14 shows an Al2O3 turbine blower-housing part fabricated by the CODE process using CaCO3 as the support material.

Besides this inorganic support material, several organic materials have been investigated for use as sacrificial support materials in the CODE process including polycaprolactone (PCL) and petrolatum. They can be removed by melting at a temperature higher than their melting temperatures of ~100  C (~210  F). Using these support materials, ZrO2 parts with overhanging features were fabricated by the CODE process, such as the gear with doublesided grooves shown in Fig. 15. Li, et al. (Ref 124) further developed the CODE process by adding a printhead with a dynamic mixer (Fig. 16), through which nonsparse FGM specimens were fabricated grading from Al2O3 and ZrO2. Figure 17 shows an Al2O3-ZrO2 FGM specimen fabricated using the CODE process, in which the volume ratio of Al2O3 in the mixed paste gradually changed from 100% to 50% by 5% increments. The material composition of each layer in the sintered FGM specimens was validated by measuring the atomic percentage of Al and Zr by energy dispersive spectroscopy (EDS) compared with the original design compositions, with an average of 1% error in the material compositions observed. Other FGM blocks with different material gradients were printed and sintered, and different amounts of deformation (warping) among these blocks were observed after sintering. A smoother (reduced) gradient of material compositions reduced the amount of deformation, and, thus, the risk of part failure.

Fig. 13

Fine structures fabricated using the ceramic on-demand extrusion CODE process showing fine surface finish: (a) ZrO2 (3 mol% Y2O3) pendant, (b) ZrO2 (3 mol% Y2O3) thin-wall structure. Source: Ref 59. Reprinted with permission from Elsevier

Fig. 14

Al2O3 turbine blower housing part fabricated by the ceramic on-demand extrusion (CODE) process using CaCO3 as the support material: (a)-(d) Printing process and printed part and (e) sintered final part. Source: Ref 122. Reprinted from with permission from Elsevier

Material Extrusion Based Ceramic Additive Manufacturing / 105

Material A inlet

Material B inlet

Extruder A

Servo motor

Extruder B

Active mixing blade

Mixture outlet

Mixing chamber Mixing chamber Nozzle

(a)

Fig. 16

(b) (a) Schematic of dynamic mixer in workhead of ceramic on-demand extrusion (CODE) system. (b) Actual workhead mounted on the system. Source: Ref 124

30 µm

Fig. 15

ZrO2 gear with double-sided grooves fabricated using polycaprolactone (PCL) support material: (a) cross section of CAD model, (b) sintered part. Source: Ref 63

30 µm

(a)

Post-Processing Post-processing involves processing the asprinted (green) ceramic parts to obtain the final dense ceramic components. All MECAM techniques require post-processing, which can vary depending on the feedstocks used and the shape retention methods. Feedstocks used in the FDC and EFF processes contain high concentrations (typically >35 vol%) of organic additives. Post processing is carried out in two stages: removing binders and sintering. To remove organic additives (binders), the liquid binder content is removed via capillary action at low temperatures ( 4. Furthermore, upon impact with the processing surface (e.g., substrate, powder-bed surface), the binder droplet should exhibit minimal splashing, which can be satisfied when We1/2Re1/4 > 50. Figure 2 summarizes the characteristic rheological ranges of the binder liquid based on previously described criteria. Binder jetting additive manufacturing is a well-established AM technology that has been

adopted in various applications. However, the physics that govern the interaction between binder droplets and the porous powder bed are complicated and still being investigated. On contact with the powder-bed surface, the binder droplet can exhibit inertia-driven behavior, including spreading, splashing, receding, and rebounding, as well as surface-energy-driven behavior, including penetration and coalescence (Ref 10, 11). Figure 3 illustrates an experimental study of real-time droplet evolution upon impact on a flat porous surface, as well as how it is influenced by the Weber number of the fluid, which further shows the challenge of accurate process control (Ref 12). The droplet/powder-bed interaction for BJAM not only influences binder-powder compatibility and binder printability but also determines various process characteristics, such as minimum resolution, raster track overlap, maximum/minimum layer thickness, and maximum printing speed (Ref 13–15). After the initial spreading on the powder-bed surface

Fig. 2

Binder liquid characteristic rheological ranges. Source: Ref 7

Fig. 3

Effect of Weber number (We) on droplet dynamics on a porous surface at 150  C (300  F). Source: Ref 12

120 / Ceramic Additive Manufacturing Processes (typically in microsceconds) (Ref 16), the binder droplets migrate into the pores of the powder bed. The migration process is primarily driven by surface tension and capillary force, and after equilibrium is achieved, the droplet permeation profile typically exhibits semicircular or semielliptical cross section, as shown in Fig. 4(a); the shape of the profile can vary significantly depending on the powder-bed characteristics. For a single droplet, the liquid/powder-bed interaction can be roughly divided into three phases: 1. Initial kinetic-energy-driven phase, in which the droplet rapidly spreads on the substrate surface and exhibits partial or complete rebound behavior, as shown in Fig. 4(b) 2. Droplet-spreading and pinning phase, driven primarily by capillary force, as shown in Fig. 4(b)

3. Droplet-absorption phase, driven by capillary force and liquid viscosity, as shown in Fig. 4(b) With most liquid inks and porous surface conditions, the individual phases occur at significantly different time scales and therefore can generally be considered to be temporally decoupled. Also, it has been suggested that splashing and complete rebound would not occur for most binder/powder-bed systems, because the We number of the binder is typically