Inkjet-based Micromanufacturing

Citation preview

Edited by Jan G. Korvink, Patrick J. Smith, and Dong-Youn Shin Inkjet-based Micromanufacturing

Related Titles Bechtold, T., Schrag, G., Feng, L. (eds.)

Hierold, C. (ed.)

System-level Modeling of MEMS

Carbon Nanotube Devices Properties, Modeling, Integration and Applications

2012 ISBN: 978-3-527-31903-9

2008 ISBN: 978-3-527-31720-2

Hutchings, I. Gerlach, G., Dotzel, W.

Inkjet Technology for Digital Fabrication 2012

Introduction to Microsystem Technology A Guide for Students

ISBN: 978-0-470-68198-5 2008

Bolic, M., Simplot-Ryl, D., Stojmenovic, I. (eds.)

RFID Systems Research Trends and Challenges

ISBN: 978-0-470-05861-9

Tabata, O., Tsuchiya, T. (eds.)

Reliability of MEMS Testing of Materials and Devices

2010 ISBN: 978-0-470-74602-8

2008 ISBN: 978-3-527-31494-2

Saile, V., Wallrabe, U., Tabata, O. (eds.)

LIGA and its Applications

Hopkinson, N., Hague, R., Dickens, P. (eds.)

2009

Rapid Manufacturing

ISBN: 978-3-527-31698-4

An Industrial Revolution for the Digital Age 2005 ISBN: 978-0-470-01613-8

Edited by Jan G. Korvink, Patrick J. Smith, and Dong-Youn Shin

Inkjet-based Micromanufacturing

The Editors Prof. Dr. Jan G. Korvink University of Freiburg Department of Microsystems Engineering IMTEK Freiburg Institute for Advanced Studies - FRIAS Georges-Koehler-Allee 103 79110 Freiburg Germany

All books published by Wiley-VCH are carefully produced. Nevertheless, authors, editors, and publisher do not warrant the information contained in these books, including this book, to be free of errors. Readers are advised to keep in mind that statements, data, illustrations, procedural details or other items may inadvertently be inaccurate.

Dr. Patrick J. Smith University of Sheffield Department Mechanical Engineering Mappin Street Sheffield S1 3JD UK

Library of Congress Card No.: applied for

Dr. Dong-Youn Shin Pukyong National University Department Printing and Information Engineering San 100, Yongdang-dong, Nam-gu Busan, 608-739 South Korea

Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available on the Internet at .

British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library.

© 2012 Wiley-VCH Verlag & Co. KGaA, Boschstr. 12, 69469 Weinheim, Germany

All rights reserved (including those of translation into other languages). No part of this book may be reproduced in any form – by photoprinting, microfilm, or any other means – nor transmitted or translated into a machine language without written permission from the publishers. Registered names, trademarks, etc. used in this book, even when not specifically marked as such, are not to be considered unprotected by law. Composition Laserwords Private Ltd., Chennai, India Printing and Binding betz-druck GmbH, Darmstadt Cover Design Schulz Grafik-Design, Fußg¨onheim Printed in the Federal Republic of Germany Printed on acid-free paper Print ISBN: 978-3-527-31904-6 ePDF ISBN: 978-3-527-64711-8 oBook ISBN: 978-3-527-64710-1

V

Contents List of Contributors 1 1.1 1.2 1.2.1 1.2.2 1.3 1.4 1.5 1.5.1 1.5.2 1.5.3 1.6 1.6.1 1.6.2 1.6.3 1.6.4 1.7

2

2.1 2.2 2.3 2.4

XIII

Overview of Inkjet-Based Micromanufacturing 1 David Wallace Introduction 1 Inkjet Technology 1 Continuous Mode Inkjet (CIJ) Technology 2 Demand Mode Inkjet Technology 2 Fluid Requirements 3 Pattern Formation: Fluid/Substrate Interaction 5 Micromanufacturing 6 Introduction 6 Limitations and Opportunities in Micromanufacturing Benefits of Inkjet in Microfabrication 8 Examples of Inkjet in Micromanufacturing 9 Chemical Sensors 9 Optical MEMS Devices 10 Bio-MEMS Devices 12 Assembly and Packaging 13 Conclusions 14 Acknowledgments 14 References 14

7

Combinatorial Screening of Materials Using Inkjet Printing as a Patterning Technique 19 Anke Teichler, Jolke Perelaer, and Ulrich S. Schubert Introduction 19 Inkjet Printing – from Well-Defined Dots to Homogeneous Films 20 Thin-Film Libraries Prepared by Inkjet Printing 25 Combinatorial Screening of Materials for Organic Solar Cells 28

VI

Contents

2.5

Conclusion and Outlook 34 References 35

3

Thermal Inkjet 41 Naoki Morita History of Thermal Inkjet Technology 41 Market Trends for Inkjet Products and Electrophotography 42 Structures of Various TIJ Heads 43 Research on Rapid Boiling and Principle of TIJ 44 Inkjetting Mechanism of TIJ 47 Basic Jetting Behavior of TIJ 48 Input Power Characteristics 48 Frequency Characteristics 49 Dependency on Temperature 49 TIJ Behavior Analysis Using Simulation 51 Cylindrical Thermal Propagating Calculation Based on the Finite Element Method (Software Name: Ansys) 51 Fluidic Free Boundary Calculation Based on the Finite Differentiation Method (Software name: Flow3D) 51 Issues with Reliability in TIJ 53 Present and Future Evolution in TIJ Technology 54 References 55

3.1 3.2 3.3 3.4 3.5 3.6 3.6.1 3.6.2 3.6.3 3.7 3.7.1 3.7.2 3.8 3.9

4 4.1 4.2 4.3 4.4 4.5 4.6 4.7

5 5.1 5.2 5.3 5.4 5.5 5.6 5.7

High-Resolution Electrohydrodynamic Inkjet 57 Park Jang-Ung and John A. Rogers Introduction 57 Printing System 57 Control of Jet Motions 59 Drop-on-Demand Mode Printing 60 Versatility of Printable Materials and Resolutions 62 Applications in Electronics and Biotechnology 64 High-Resolution Printing of Charge 69 References 70 Cross Talk in Piezo Inkjet 73 Herman Wijshoff Introduction 73 Electrical Cross Talk 73 Direct Cross Talk 74 Pressure-Induced Cross Talk 76 Acoustic Cross Talk 78 Printhead Resonance 81 Residual Vibrations 83 References 84

Contents

6 6.1 6.1.1 6.1.2 6.1.3 6.2

7 7.1 7.2 7.2.1 7.2.2 7.2.3 7.2.4 7.2.5 7.3 7.3.1 7.3.2 7.3.3

8

8.1 8.1.1 8.1.2 8.2 8.2.1 8.2.2 8.2.3 8.2.4 8.2.5 8.3

9 9.1 9.2

Patterning 87 Patrick J. Smith and Jonathan Stringer Introduction 87 Droplet Impact and Final Droplet Radius 88 Evaporation of Inkjet-Printed Droplets at Room Temperature 90 Morphological Control for Ink Droplets, Lines, and Films 91 Conclusion 94 References 95 Drying of Inkjet-Printed Droplets 97 Hans Kuerten and Daniel Siregar Introduction 97 Modeling of Drying of a Droplet 98 Fluid Model 98 Lubrication Approximation 99 Solute Concentration 101 Evaporation Velocity 102 Numerical Method 103 Results 103 Droplet Shape Evolution 104 Layer Thickness 106 Effect of Diffusion 108 Acknowledgments 109 References 109 Postprinting Processes for Inorganic Inks for Plastic Electronics Applications 111 Jolke Perelaer Introduction 111 Inkjet Printing 111 Printed Electronics 111 Inkjet Printing and Postprinting Processes of Metallic Inks 112 Choice of Metal 112 Postprinting Processes to Convert Inorganic Precursor Ink 115 Conventional Sintering Techniques 116 Alternative and Selective Sintering Methods 116 Room-Temperature Sintering 119 Conclusions and Outlook 121 Acknowledgments 122 References 122 Vision Monitoring 127 Kye-Si Kwon Introduction 127 Measurement Setup 127

VII

VIII

Contents

9.3 9.4 9.5 9.6

Image Processing 130 Jetting Speed Measurement 134 Head Normalization and Condition Monitoring 139 Meniscus Motion Measurement and Its Application 141 References 144

10

Acoustic Monitoring 145 Herman Wijshoff Introduction 145 Self Sensing 145 Measuring Principle 146 Drop Formation, Refill, and Wetting Dirt 152 Air Bubbles 153 Printhead Control 156 References 157

10.1 10.2 10.3 10.4 10.5 10.6 10.7

11 11.1 11.1.1 11.1.2 11.1.3 11.1.4 11.1.5 11.1.5.1 11.1.5.2 11.2 11.2.1 11.2.2 11.2.3 11.2.4 11.2.5

12 12.1 12.2

150

Equalization of Jetting Performance 159 Man-In Baek and Michael Hong Equalization of the Droplet Volume on the Fly 160 Components of a Drop Watcher 160 Equalization through Volume Control 160 Results of the Droplet Volume Measurement and Equalization Process 161 Speed Equalization 164 Problems with the Droplet Equalization Methods on the Fly 164 Distortion of the Captured Droplet Images 166 Relation between Droplet Volume and Speed 166 Droplet Volume Equalization with Sessile Droplets 166 Equalizing the Droplet Volume with the Measurement of Sessile Droplets 167 Results of the Sessile Droplet Measurement and Equalization Process 168 Usefulness of the Sessile Droplet Measurement and Equalization Process 169 The Droplet Volume Equalization Process Using Light Transmittance 170 Result of the Droplet Volume Equalization Process Using Light Transmittance 171 Further Reading 171 Inkjet Ink Formulations 173 Alexander Kamyshny and Shlomo Magdassi Introduction 173 Ink Formulation 174

Contents

12.2.1 12.2.2 12.2.2.1 12.2.2.2 12.2.3 12.2.4 12.3 12.3.1 12.3.2 12.3.3 12.3.4 12.3.5 12.3.6 12.3.7 12.3.8 12.4 12.4.1 12.4.2 12.4.3 12.4.4 12.5 12.6 12.7

Functional Materials 176 Solvents 177 Solvent-Based Inks 177 Water-Based Inks 178 Hot-Melt (Phase-Change) Inks 178 UV-Curable Inks 178 Ink Parameters and Additives 179 Rheology Control 179 Surface Tension Modifiers 180 Electrolytes and pH 180 Foaming and Defoamers 181 Humectants 181 Binders 181 Biocides 182 Examples of Inkjet Ink Formulations 182 Jetting Performance 182 Drop Formation 183 Ink Latency 183 Recoverability 184 Ink Supply 184 Ink Interaction with Substrates 185 Nongraphic Applications 186 Conclusions 187 References 187

13

Issues in Color Filter Fabrication with Inkjet Printing 191 Dong-Youn Shin and Kenneth A. Brakke Introduction 191 Background 191 Comparison of Printing Technologies 195 Printing Swathe due to Droplet Volume Variation 199 Subpixel Filling with a Designed Surface Energy Condition Other Technical Issues 212 Conclusion 213 References 213

13.1 13.2 13.3 13.4 13.5 13.6 13.7

14

14.1 14.2 14.3 14.3.1 14.3.2 14.3.3

204

Application of Inkjet Printing in High-Density Pixelated RGB Quantum Dot-Hybrid LEDs 217 Hanna Haverinen and Ghassan E. Jabbour Introduction 217 Background 218 Experimental Procedure and Results 220 Role of Droplet Formation 221 Atomic Force Microscopy 222 Electroluminescence 225

IX

X

Contents

14.4 14.5

Inkjet-Printed, High-Density RGB Pixel Matrix Conclusion 234 Acknowledgment 234 References 234 Further Reading 236

15

Inkjet Printing of Metal Oxide Thin-Film Transistors 237 Jooho Moon and Keunkyu Song Introduction 237 Materials for Metal Oxide Semiconductors 237 Inkjet Printing Issues 239 Ink Printability 239 Influence of Substrate Preheat Temperature 242 Solution-to-Solid Conversion by Annealing 247 All-Oxide Invisible Transistors 251 Summary 254 References 254

15.1 15.2 15.3 15.3.1 15.3.2 15.4 15.5 15.6

16 16.1 16.2 16.3 16.4 16.4.1 16.4.2 16.5 16.5.1 16.5.2 16.6 16.7 16.8 16.9 16.10 16.11 16.12

17 17.1 17.2 17.3 17.4

229

Inkjet Fabrication of Printed Circuit Boards 257 Thomas Sutter Introduction 257 Traditional Printed Circuit Board Processes 257 Challenges for Inkjet in Printed Circuit Boards 258 Legend-Marking Processes 261 Cost Comparison 262 Materials for Legend Printing 262 Innerlayer Copper Circuit Patterning 263 Materials for Copper Etch Resists 264 Substrate Modification 265 Copper Plating Resist 266 Waste Reduction Using Inkjet Printing 268 Solder Mask Printing 269 Metallic Inks 273 Theoretical Printing Example for PCB Manufacturing 275 Digital Printing Alternatives to Inkjet Fabrication 276 Future Applications for Inkjet in Printed Circuit Boards 276 References 277 Photovoltaics 279 Heather A.S. Platt and Maikel F.A.M. van Hest Introduction 279 Device Structures 280 Small- and Large-Area Printing for Photovoltaics 283 Commercial Inkjet for Photovoltaics 289

Contents

17.5

Summary and Perspective References 292

18

Inkjet Printed Electrochemical Sensors 295 Aoife Morrin Introduction 295 Printed Sensor Manufacturing 297 Inkjet Printing of Sensor Components 298 Substrates 299 Conducting Tracks 300 Transducer Materials 300 Biomolecules 305 Inkjet-Printed Sensor Applications 306 Future Commercial Projection 306 Abbreviations 309 References 309

18.1 18.2 18.3 18.3.1 18.3.2 18.3.3 18.3.4 18.4 18.5

19 19.1 19.1.1 19.1.1.1 19.1.2 19.1.2.1 19.1.2.2 19.2 19.2.1 19.2.2 19.2.3 19.2.4 19.2.4.1 19.2.4.2 19.2.4.3 19.3

20 20.1 20.2 20.2.1 20.2.2 20.3 20.4 20.5

291

Antennas for Radio Frequency Identification Tags 313 Vivek Subramanian Introduction 313 Introduction to RFID 313 RFID Tag Classification 314 Applications of Printing to RFID Antenna Production 317 An Overview of RFID–HF versus UHF 318 Silicon-Based RFID Tag Construction – from Chip to Tag 319 Printed Antennas 319 HF Tag Antenna Considerations 320 UHF Tag Antenna Considerations 321 Application of Printing to Antenna Fabrication 322 Materials for Printed Antennas 323 Metallic Pastes 324 Particle-Based Inks 325 Organometallic Precursors 326 Summary of Status and Outlook for Printed Antennas 327 References 328 Inkjet Printing for MEMS 331 K. Pataky, V. Auzelyte, and J. Brugger Introduction 331 Photolithography and Etching 331 Photolithography 332 Etching 332 Direct Materials Deposition 333 Optical MEMS 336 MEMS Packaging 339

XI

XII

Contents

20.6 20.7

Functionalization and Novel Applications Conclusion 342 References 342

21

Inkjet Printing of Interconnects and Contacts Based on Inorganic Nanoparticles for Printed Electronic Applications 347 Jolke Perelaer and Ulrich S. Schubert Introduction 347 Inkjet Printing of Metallic Inks for Contacts and Interconnects 348 Inkjet Printed Contacts and Interconnects for Microelectronic Applications 348 Inkjet Printing in High Resolution 351 Surface Wetting and Ink Modifications 351 Reduced Printed Droplet Diameter 353 Physical Surface Treatment 357 Inkjet-Printed Ionogels 359 Conclusions and Outlook 361 Acknowledgments 362 References 362

21.1 21.2 21.2.1 21.3 21.3.1 21.3.2 21.3.3 21.3.4 21.4

Index

365

340

XIII

List of Contributors V. Auzelyte Microsystems Laboratory Ecole Polytechnique Federale de Lausanne (EPFL) 1015 Lausanne Switzerland

Hanna Haverinen University of Oulu P.O. Box 4500 90014 Oulu Finland

Man-In Baek LG Electronics Inc. PRI. 19-1 Cheongho-ri Jinwi-myeon Pyeongtaek-si Gyeonggi-do 451-713 South Korea

King Abdullah University of Science and Technology (KAUST) Materials Science and Engineering, Electrical Engineering Solar and Photovoltaics Engineering Research Center Thuwal 23955-6900 Kingdom of Saudi Arabia

Kenneth A. Brakke Susquehanna University Mathematics Department 514 University Avenue Selinsgrove, PA 17870-1164 USA J. Brugger Microsystems Laboratory Ecole Polytechnique Federale de Lausanne (EPFL) 1015 Lausanne Switzerland

and

Maikel F.A.M. van Hest National Center for Photovoltaics National Renewable Energy Laboratory 1617 Cole Boulevard Golden, CO 80401-3393 USA Michael Hong LG Electronics Inc. PRI. 19-1 Cheongho-ri Jinwi-myeon Pyeongtaek-si Gyeonggi-do 451-713 South Korea

XIV

List of Contributors

Ghassan E. Jabbour University of Oulu P.O. Box 4500 90014 Oulu Finland and Arizona State University School of Mechanical, Aerospace, Chemical and Materials Engineering 7700 South River Parkway Tempe, AZ 85284 USA and Arizona State University Advanced Photovoltaics Center 7700 South River Parkway Tempe, AZ 85284 USA and King Abdullah University of Science and Technology (KAUST) Materials Science and Engineering, Electrical Engineering Solar and Photovoltaics Engineering Research Center Thuwal 23955-6900 Kingdom of Saudi Arabia Park Jang-Ung UNIST (Ulsan National Institute of Science and Technology) School of Nano-Bioscience and Chemical Engineering School of Mechanical and Advanced Materials Engineering 100 Banyeon-ri, Eonyang-eup Ulju-gun 689-798 Ulsan South Korea

Alexander Kamyshny The Hebrew University of Jerusalem Institute of Chemistry Casali Institute of Applied Chemistry Edmond Safra Campus Givat-Ram Jerusalem 91904 Israel Hans Kuerten Eindhoven University of Technology Department of Mechanical Engineering Den Dolech 2 P.O. Box 513 5600 MB Eindhoven The Netherlands Kye-Si Kwon Soonchunhyang University Department of Mechanical Engineering 646, Eupnae-ri, Shinchang-myeon Asan-si, Chungcheongnam-do 336-745 South Korea Shlomo Magdassi The Hebrew University of Jerusalem Institute of Chemistry Casali Institute of Applied Chemistry Edmond Safra Campus, Givat-Ram Jerusalem 91904 Israel

List of Contributors

Jooho Moon Department of Materials Science and Engineering Yonsei University 50 Yonsei-ro Seodaemun-gu Seoul, 120-724 Korea

Heather A.S. Platt National Center for Photovoltaics National Renewable Energy Laboratory 1617 Cole Boulevard Golden, CO 80401-3393 USA

Naoki Morita Fuji Xerox Co., Ltd. Marking Technology Laboratory 2274 Hongo Ebina, 243-0494 Japan

John A. Rogers University of Illinois at Urbana-Champaign Department of Materials Science and Engineering Department of Chemistry Department of Mechanical Science and Engineering 1304 W. Green Street 61801 Urbana USA

Aoife Morrin Dublin City University School of Chemical Sciences National Centre for Sensor Research Collins Avenue Dublin 9 Ireland K. Pataky Microsystems Laboratory Ecole Polytechnique Federale de Lausanne (EPFL) 1015 Lausanne Switzerland Jolke Perelaer Friedrich-Schiller-University Jena Laboratory of Organic and Macromolecular Chemistry (IOMC) Humboldtstraße 10 D-07743 Jena Germany

Ulrich S. Schubert Friedrich-Schiller-University Jena Laboratory of Organic and Macromolecular Chemistry (IOMC) Humboldtstraße 10 D-07743 Jena Germany Dong-Youn Shin Pukyong National University Department Printing and Information Engineering San 100, Yongdang-dong Nam-gu Busan, 608-739 South Korea

XV

XVI

List of Contributors

Daniel Siregar Eindhoven University of Technology Department of Mechanical Engineering Den Dolech 2 P.O. Box 513 5600 MB Eindhoven The Netherlands Patrick J. Smith The University of Sheffield Department of Mechanical Engineering Kroto Research Institute Broad Lane, Sheffield S3 7HQ UK Keunkyu Song Department of Materials Science and Engineering Yonsei University 50 Yonsei-ro Seodaemun-gu Seoul, 120-724 Korea Jonathan Stringer The University of Sheffield Department of Mechanical Engineering Kroto Research Institute Broad Lane, Sheffield S3 7HQ UK

Vivek Subramanian University of California Berkeley Department of Electrical Engineering and Computer Sciences Berkeley, CA 94720-1770 USA Thomas Sutter Dow Electronic Materials 455 Forest Street Marlborough, MA 01752 USA Anke Teichler Friedrich-Schiller-University Jena Laboratory of Organic and Macromolecular Chemistry (IOMC) Humboldtstraße 10 D-07743 Jena Germany David Wallace Vice President MicroFab Technologies, Inc. 1104 Summit Ave. ste 110 Plano TX 75074 USA Herman Wijshoff Oce` Technologies B.V. P.O. Box 101 5900 MA Venlo The Netherlands

1

1 Overview of Inkjet-Based Micromanufacturing David Wallace

1.1 Introduction

Inkjet technology has come to prominence in the past decade as the dominant printer technology in the combined home and small office/home office (SOHO) markets, and inkjet is now familiar to the general public. Also, in the past decade, inkjet technology has become recognized in the technical community as a highly capable tool for manufacturing, particularly micromanufacturing [1]. As an introduction to the more detailed discussions in the chapters to follow, this chapter will, briefly, give a background on inkjet technology; discuss fluid requirements and pattern formation fundamentals; discuss the characteristics of microfabrication and the potential of inkjet as a tool; discuss the breadth of applications being addressed with inkjet technology, including a few specific examples; and discuss issues, challenges, and possible future applications.

1.2 Inkjet Technology

Inkjet technology is not a single technology, but a family of very different technologies that have a similar function: the precise generation of free-flying fluid droplets. Precise refers to the volume of the droplet, the time at which it is produced, the velocity with which it travels, and the direction of travel. The range of diameter, velocity, and frequency of generation of the droplet obtained, along with the precision, will vary considerably depending on the specific technology employed. A complete discussion of the different inkjet technologies and their relationship to each other is beyond the scope of this chapter and only a brief overview is given below. Excellent overviews of the physics [2, 3] and practical applications [4] of inkjet are available for those interested in more detailed reviews. Inkjet-based Micromanufacturing, First Edition. Edited by J. G. Korvink, P. J. Smith, and D.-Y. Shin. © 2012 Wiley-VCH Verlag GmbH & Co. KGaA. Published 2012 by Wiley-VCH Verlag GmbH & Co. KGaA.

2

1 Overview of Inkjet-Based Micromanufacturing

1.2.1 Continuous Mode Inkjet (CIJ) Technology

Repeatable drop formation from a cylinder of liquid was noted as early as 1833 by Savart [5] and described mathematically by Lord Rayleigh for inviscid jets [6, 7], and by Weber [8] for viscous jets. In continuous mode inkjet (CIJ) technology, fluid under pressure is forced through an orifice, typically 50–80 μm in diameter, and breaks up into uniform drops under the action of surface tension, by the amplification of diameter perturbations [9, 10] or surface tension perturbations [11]. Piezoelectric actuators have been used as the source of diameter perturbations, but microheaters have been used as the source of surface tension perturbations [12]. A drop breaks off from the jet in the presence of a varying electric field, and thus acquires a charge. The charged drops are directed to their desired location, either the catcher or one of several locations on the substrate, by a static electric field [13]. This type of system is generally referred to as ‘‘continuous’’ because drops are continuously produced and their trajectories are varied by the amount of charge applied. CIJ printing systems produce droplets that are approximately twice the orifice diameter of the droplet generator, which is an important limitation in the use of CIJ in micromanufacturing. Droplet generation rates for CIJ systems are usually in the 80–100 kHz range, but systems with operating frequencies up to 1 MHz have been commercialized [14]. Droplet sizes can be as small as 20 μm in a continuous system, but 150 μm is typical. CIJ systems are currently in widespread use in product labeling of food and medicines. They have high throughput capabilities, and are best suited for applications where they are in continual use. Very few CIJ systems are multicolor (multifluid), but two-color systems are in use. CIJ systems require unused drops to be recirculated or wasted, another limitation in using CIJ for micromanufacturing. Applications where CIJ is used include Massachusetts Institute of Technology (MIT’s) 3D printing rapid prototyping technology [15], metal jetting technology [16], and medical diagnostic test strip production [17]. Figure 1.1 illustrates a schematic of a continuous inkjet printer and a jet of water breaking up into 120 μm droplets at 20 kHz. 1.2.2 Demand Mode Inkjet Technology

In demand mode inkjet technology, a piezoelectric transducer, or heated resistor that produces a bubble [18], creates pressure/velocity waves in a fluid, which starts at essentially atmospheric pressure, and these waves propagate to produce a drop at an orifice [19–21] or free surface [22]. Since a drop is created only when desired, these types of systems are referred to as drop-on-demand or demand mode. Demand mode systems are conceptually far less complex than continuous mode systems. On the other hand, demand mode droplet generation requires the transducer to deliver three or more orders of magnitude greater energy to produce a droplet, compared to the continuous mode. Driven by the need for multiple orifices to

1.3 Fluid Requirements

Piezoelectric transducer

Charging electrode

Deflection plates

+V

Orifice

Pump Charge driver Transducer driver

Ink reservoir

Gutter Substrate

Data

Figure 1.1 Schematic of a continuous inkjet printer and a jet of water breaking up into 120 μm droplets at 20 kHz.

meet the throughput requirement in printing applications, and constrained by the transducer energy requirements, there are many ‘‘elegant’’ (i.e., complex) array demand mode printhead designs [23–26]. Demand mode inkjet systems have been used primarily in desktop printers, and now dominate the low-end printer market (HP’s DeskJets, Cannon’s Bubble Jets, and Epson’s Stylus). Demand mode inkjet systems have no fluid recirculation requirement, and this makes their use as a general fluid microdispensing technology more straightforward than continuous mode technology. Almost all of the results discussed in this book have been obtained with demand mode systems. Figure 1.2 illustrates the schematic of a demand mode system and drops of an organic solvent being formed at 4 kHz.

1.3 Fluid Requirements

The fluid property requirements for demand mode inkjet dispensing are viscosity ∼20 dyn/cm. Low viscosities usually lead to satellite formation (formation of multiple drops when one is desired) and low damping of post-drop oscillations, limiting the upper operating frequency. If the fluid is heated or cooled, the critical properties are those at the operating temperature of the office,

3

4

1 Overview of Inkjet-Based Micromanufacturing

Piezoelectric transducer

Piezo driver

Orifice

Ink ~atmospheric pressure

Substrate

Data

Figure 1.2 Schematic of a demand mode system and drops of an organic solvent being formed at 4 kHz.

not room temperature. Higher viscosities can be tolerated in the fluid delivery system if this does not create a pressure drop that limits the desired maximum frequency due to restriction of the flow into the printhead. For high density fluids, such as molten metals [27] (e.g., solders, tin, mercury, rubidium, and lithium), the fluid properties should be converted to kinematic values to determine if the fluid properties are acceptable. Newtonian behavior is not strictly required, but non-Newtonian effects are never beneficial. Viscoelastic behavior causes significant performance problems by increasing the amount of deformation the fluid can withstand without breaking off to form a drop [28]. Particle suspensions, such as pigmented inks, are acceptable as long as the particle/agglomerate size and density do not cause the suspension to depart from the fluid properties range given above. Particles that are >5% of the orifice diameter (e.g., cells; [29]) will cause at least some instability in drop generation behavior, but still may be acceptable in low concentrations. The ‘‘window’’ of fluids and suspensions that can be dispensed using inkjet technology has been enlarged by heating, cooling, stirring, wiping, purging, preoscillating, diluting, and other methods. However, this window is unavoidably narrowed as orifice diameter decreases, frequency increases, and the number of jets in an array increases. The diversity of fluids that have been dispensed using inkjet technology is impressive, given the fluid property restrictions described above. Inks by themselves represent a broad class of materials. Dye and pigment aqueous dispersions are the most commonly used in conventional printing, volatile, and low-volatility solvent inks are all in common use, as are UV-curing inks. Other materials that have been dispensed using inkjet technology are shown in Table 1.1.

1.4 Pattern Formation: Fluid/Substrate Interaction Table 1.1

Materials that have been deposited using inkjet technology.

Electronic/optical Biological fluids materials

Organic solvents

Particle suspensions

Other

Liquid metals Fluxes Photoresists Epoxies Polyimides

DNA Nucleic acids Amino acid Cells Proteins

Alcohols Keytones Aliphatics Aromatics Dipolar solvents

Pigments Cells Latex spheres Metals Teflon

Electroactive polymers Organometallics

Lipids

Phosphors

Sol-gels Thermoplastics Thermosets Acrylics Photographic developer Fuels

Biosorbable polymers

Ferrites Zeolyts

Aqueous adhesives Odorants

1.4 Pattern Formation: Fluid/Substrate Interaction

Except for the applications where inkjet technology is used to meter fluid, as in filing a well or a region bounded by a nonwetting coating, inkjet deposition processes are used to produce a pattern of material on a substrate. The interaction between the fluid properties, jetting parameters (drop size, velocity, and frequency), substrate properties, printing grid (dots per inch), and printing sequence (interleave, overprinting, sequence of fluid, etc.) is a multivariate optimization in the development of all inkjet printing systems/applications. For the conventional inkjet case of liquid ink on paper, the porosity of the paper and the low viscosity of the ink represented a major challenge in the initial development of inkjet printers. Rapid spreading of liquid ink through the fibers can cause the spot size to become much larger than the drop size, decreasing the optical density of the spot and resulting in irregular spots that degrade the quality of characters, lines, and so on. Ink formulations that produce good print quality on a wide range of papers have been a cornerstone in the wide acceptance of inkjet printers in the marketplace. Most micromanufacturing applications of inkjet technology deposit liquid onto nonporous substrates, similar to printing an overhead transparency. Control of the spreading is essential if the desired resolution is to be obtained. Phase change inks were developed for conventional inkjet printers for precisely this reason, since they solidify quickly after impact. In micromanufacturing applications, solders for electronic manufacturing [30] and thermoplastics for free-form fabrication [31] are examples of phase change materials. The control of spreading by solidification is a beneficial aspect of phase change materials if the goal is to limit spreading and obtain the smallest spot for a given drop size. However, if the goal is deposition of a uniform layer, solidification into a bump is a problem, not a feature.

5

6

1 Overview of Inkjet-Based Micromanufacturing

Many organic liquids, such as isopropanol, acetone, and acetonitrile, are of very low viscosity, low contact angle/surface tension, and volatile. Their ability to wet most surfaces and low viscosity allows these fluids to spread rapidly. As with a phase change material’s lack of spreading, this is either a feature or a problem, depending on the application. If one is trying to write a small conductive line using an organometallic ink, or create a pixel in a light-emitting polymer display, it is definitely a problem. In many cases, surface features, such as the wells commonly used in flat panel displays, provide a barrier to spreading and help physically define the feature size. In other cases, surface treatments, such as plasma cleaning or application of a nonwetting coating, are used to control spreading [32]. Volatility of a solution with dissolved or suspended solids can cause operational issues, and ink drying in the orifice is one of the most common failure modes with office inkjet printers. In addition, volatility can cause nonuniform distribution of the solid on the substrate after drying [33]. Solutions to this problem have been many and diverse: reactive substrate, covalent binding of the solid to the substrate, cosolvents that are lower volatility, UV or thermal cross-linking, and so on. Pattern or image formation in its simplest implementation can be just the selection of pixel (picture element) size and spacing, then using inkjet dispensing to fill the desired pixels, ratering out the image as is done in a conventional inkjet printer. However, even the lowest cost inkjet printers have complicated print modes that are used to increase print quality. Rows of spots are interleaved to hide coherent errors from a single jet, colors are printed so as not to bleed together in wet state, multiple passes are made over an area to increase color saturation of the printed area, and operating frequency is decreased for high-quality printing. All of these methods, and more, have applicability to micromanufacturing applications of inkjet.

1.5 Micromanufacturing 1.5.1 Introduction

Micromanufacturing today has evolved from microelectromechanical systems (MEMS) fabrication technology, developed initially in the 1980s with the goal of integrating electromechanical sensors and actuators with their conditioning electronics. By adding silicon micromachining and deposition of metals and oxides to silicon-based analog integrated circuit (IC) fabrication technology, very small electromechanical sensors and actuators could be fabricated at very low cost and in high volume [34, 35]. In addition, these MEMS-based devices were more robust and reliable than their larger conventional counterparts. MEMS-based accelerometers are the most widely used MEMS products. They have enabled the use of airbags in automobiles and can currently be found in cell phones, computers, cameras, golf clubs, skis, and, of course, video games. MEMS-based pressure transducers

1.5 Micromanufacturing

have also been in widespread use for decades, primarily in automobiles, airplanes, process control equipment, and disposable biomedical sensors. In most applications, the benefits of a high degree of system integration are readily apparent. The success of MEMS-based accelerometers and pressure sensors, along with the general drive toward small, low-cost, high-volume products has led to an explosion of process technologies that can be used in MEMS fabrication and applications targeted by these process technologies. Inclusion of digital electronics was a fairly obvious step. Expansion to optical emitters, detectors, and switches has resulted in micro-optical electro-optical mechanical systems (MOEMS) [36]. Examples include micromirror devices that are used in televisions and projectors, and solid-state laser devices that have enabled the growth of the telecommunication and data communication industries [37]. Inclusion of biological functions has resulted in Bio-MEMS devices [38]. Lab-on-a-chip devices for point-of-care diagnostics are one of the few examples of available Bio-MEMS products [39, 40], although a large number of applications are being developed, including implantable sensors (e.g., glucose) [41] and microneedle-based transdermal drug delivery devices [42]. Finally, by including fluid flow with thermal or piezoelectric actuators, many inkjet printheads fall into the category of MEMS devices and inkjet printers have long been ‘‘adopted’’ by analysts when reporting MEMS industry revenue figures [43]. Many optical devices now included in the MOEMS category do not have a mechanical function, being more properly categorized as electro-optics (EO) devices. Other devices cited above have similar issues when it comes to categorization. This illustrates the difficulty in using the term MEMS in any strict sense at this point in time, since it has been broadened and generalized to refer to micromanufacturing in general, partly as a result of its successes and familiarity. In this chapter, the term ‘‘MEMS’’ is used in the loosest sense, referring to miniaturized devices and systems that must be micromanufactured. 1.5.2 Limitations and Opportunities in Micromanufacturing

MEMS device fabrication methods grew out of the silicon-based semiconductor industry, so most rely on photolithography. Photolithographic processes are particularly well suited for high-volume manufacturing of devices with high feature density and low diversity of fabrication processes and fabricated features [44, 45]. The prime example is a dynamic random access memory (DRAM) device with repetition of the same features millions of times using a limited number of fabrication processes. MEMS fabrication has successfully built on the huge microelectronics equipment and technology base, adding the feature diversity required to create a ‘‘system’’ through a limited number of additional compatible processes. However, photolithography and other ‘‘IC-like’’ fabrication processes are severely limited in the types of materials that can be used. In addition, there are technical and cost limitations that limit the number of ‘‘layers’’ that can be created in fabricating an MEMS device.

7

8

1 Overview of Inkjet-Based Micromanufacturing

Materials limitations in conventional MEMS fabrication fall into two categories, compatibility and cost. First, materials must be compatible with photolithography, which means that they must survive the application, patterning, and removing of photosensitive masking materials, and, in general, must be compatible with the creation of one or more additional layers. Functional materials such as ones that are biologically active, chemically active/receptive, or optically active (emitters and receivers) are typically difficult or impossible to use in photolithographic or other ‘‘IC-like’’ fabrication methods. Yet it is these types of materials that could enable a broad range of new small integrated devices for medicine, security, communications, and so on. Some of the most interesting materials are rare or expensive and thus would be cost prohibitive to use in the current subtractive processes employed in MEMS fabrication. Many biologically active materials fall into this category. Finally, there is an inherent conflict between the desire to increase the number of functions in MEMS devices and the cost of each ‘‘layer’’ in a photolithographic or other ‘‘IC-like’’ fabrication method. Even if the functional materials are compatible with multiple layers of photolithography and are not particularly expensive, each layer has a substantial total cost (equipment, labor, facility, materials, yield, testing, etc.) associated with it. Thus, additional functions always have a strong cost counterweight in MEMS device design based on current manufacturing methods. 1.5.3 Benefits of Inkjet in Microfabrication

Inkjet printing technology has a number of attributes that can overcome some of the inherent limitations of photolithographically based microfabrication methods. Since it is an additive method, material is only deposited where it is desired. Usage of rare or expensive materials can thus be conserved. The net savings can be significant for even moderately priced materials if the amount of area required to be covered is small compared to the total area of the substrate (i.e., low feature density). In addition to cost savings because of low materials usage, additive methods have little or no waste, so they are much more environmentally friendly. In contrast, subtractive methods waste large amounts of the functional material, plus the photosensitive masking material (if different) and the cleaning and etching solvents. Since inkjet is noncontact, interaction between different materials when they are deposited is eliminated or greatly reduced, as is the requirement to consider each material to be deposited as an expensive additional ‘‘layer.’’ Deposition can occur on nonplanar surfaces, eliminating planarization steps that are sometimes required in photolithography. Even very fragile or sensitive surfaces such as released layers (thin structures not supported by any solid material underneath them) can be printed onto using inkjet because of the extremely low inertial force exerted by a deposited drop. Active layers (detector, emitter, biological, reactive, etc.) can be deposited onto without degrading their function. Thick films can be created by

1.6 Examples of Inkjet in Micromanufacturing

overprinting the same location multiple times, and locally layered structures can be created without having to process the entire substrate area. Lastly, since inkjet is a fundamentally digital, data-driven process, the cost and time associated with producing the masks required by photolithographically based microfabrication methods is eliminated. This removes both the cost and turnaround time associated with fixed tooling. The ability of inkjet to deposit onto individual dye and partial wafers, and to perform parametric process development experiments under digital control, can accelerate the development time for a microfabricated device. Several examples of the use of inkjet in micromanufacturing are discussed below to illustrate the characteristics and issues of inkjet deposition discussed earlier.

1.6 Examples of Inkjet in Micromanufacturing 1.6.1 Chemical Sensors

Chemical sensors represent a fairly new and broad area of research and development for MEMS devices, driven by the need for large numbers of low-cost sensors for explosive, chemical warfare agents, drugs of abuse, industrial gases, residential gases, and many others. A majority of these sensors use materials that are electrically or photonically active, or more simply have surfaces that cause the molecules of interest to temporarily adhere to them. Not surprisingly, most of these sensing materials are ‘‘sensitive,’’ meaning delicate, and cannot be photolithographically processed. Also, because they are sensitive, they are applied in the last or nearly last fabrication process; typically, this is onto nonplanar, feature-rich surfaces that can be very fragile. All these factors make MEMS chemical sensor manufacturing an area that is broadly exploring the use of inkjet fabrication technology. Chemoresistive materials, ones that change resistance when exposed to specific molecules of interest, are one of the oldest and most broadly used sensing materials classes used in MEMS sensor devices [46]. Recent developments in nanomaterials and MEMS structures have expanded the number of materials and sensor structures being developed [47]. An example of an MEMS chemoresistive sensor is one that is being developed at Carnegie Mellon University to detect volatile organic compounds (VOCs) in respirators, indicating end of life [48]. The basic sensor structure, shown in Figure 1.3a, is a pair of spiral electrodes in a 250 μm circle that is in a 350 μm diameter SU-8 well. Multiple sensing and reference elements, which, in general, could contain multiple sensing materials, are incorporated on a 2.65 mm die that also contains all the required control electronics, as shown in Figure 1.3b. The die is assembled into a TO-5 package (Figure 1.3c), which is commonly used for optical devices. The sensing material, gold-thiolate nanoparticles, are suspended in a carrier fluid (5–10 mg ml−1 ) and deposited onto the sensing area. Although not visible

9

10

(a)

1 Overview of Inkjet-Based Micromanufacturing

(b)

(c)

Figure 1.3 Carnegie Mellon chemoresistive sensor: (a) sensing element configuration showing 250 μm diameter dual electrode spiral in a 350 μm SU-8 well; (b) multiple sensing and reference elements on a 2.65 mm

(d) die; (c) sensor device in a TO-5 package; and (d) sensor element printed with 225 nominally 30 pl drops of solution containing gold-thiolate nanoparticles. (Images courtesy Carnegie Mellon University.)

in Figure 1.3a, 15 drops of nominally 30 pl volume have been deposited onto the sensor using an inkjet device. Figure 1.3d shows the sensing area after 225 nominally 30 pl drops of solution have been deposited, producing an average film thickness of 1.5 μm. It is interesting to note the use of two wetting ‘‘stops’’ in the sensing area. The SU-8 well contains the initial fluid volume dispensed, preventing undesired wetting onto other areas of the die. In addition, the fluid dewets from the outer portion of the well during drying so that all of the particles are deposited onto the electrode region. This self-centering behavior results in an impedance variation of 99% waste generation, inkjet printing has a yield of >90% (in case micropipettes are utilized). Furthermore, thin-film libraries can be created with inkjet printing and characterized in a combinatorial approach. Less material is thereby required, while the reproducibility as well as the reliability of the characterization process increases. Inkjet printing is a noncontact processing technique that is compatible with various substrates and does not require expensive masks. Therefore, inkjet printing leads to a decreased production time and, thus, to a reduction of the production costs as well. However, suitable inks have to be developed and optimized for their drying behavior, that is, by preventing the coffee-ring effect, which will result in homogeneous dried structures [18]. The use of combinatorial printing techniques in the field of biology applications, for example, the preparation of functional patterns of glucose oxidase [19], proteins [20] as well as for drug screening [21, 22], was described as well in literature. Furthermore, inkjet printing was used in combinatorial materials research in the field of ceramics [23, 24] and sensor applications [25]. Owing to their semiconducting properties, it is highly important to optimize conjugated polymers for their use in organic conducting polymers. Therefore, this chapter mainly discusses the development of polymer thin-film libraries using high-throughput solution, film preparation, and materials screening techniques. The first section discusses the need for a combinatorial setup in the area of materials research that covers the whole experimental workflow: from solution preparation and ink deposition to the characterization of the as-printed feature properties. Furthermore, the advantages of using inkjet printing as a deposition technique are discussed. The next section introduces the requirements of the inkjet printing process in order to prepare well-defined dots, uniform lines, and homogeneous films. The third and fourth sections show investigations that have been performed on different fields of research, including the preparation of thin-film libraries that consist of poly(p-phenylene-ethynylene)-alt-poly(p-phenylene-vinylene)s (PPE-PPVs) copolymers, and CdTe nanocrystals (NCs). Furthermore, the screening of polymers for optoelectronic applications, such as light-emitting diodes and solar cells, is discussed as well. Finally, the last chapter concludes with a view of future research directions.

2.2 Inkjet Printing – from Well-Defined Dots to Homogeneous Films

Using inkjet printing, functional features such as dots, lines, and films can be prepared in a simple manner. With increased feature size, that is, when going from single dots to films, the printing properties that influence the quality of the

2.2 Inkjet Printing – from Well-Defined Dots to Homogeneous Films

(a)

(b)

Figure 2.1 (a) Dispenser printhead with attached reservoir and (b) micropipette that aspirates inks from a microtiter plate. (Reprinted from Ref. [28].)

printed structures become more complex. In order to understand the influence of individual printing parameters, fluid characteristics as well as printed structure properties need to be investigated step by step, starting from the smallest printing feature. De Gans et al. [26] reviewed the requirements for the inkjet printing equipment as well as for the used solutes in order to prepare well-defined features and arrays. One of the most important findings was that the printhead needs a positioning accuracy of the printed droplets smaller than 10 μm, in order to provide a sufficient reproducibility. Piezoelectric drop-on-demand (DoD) inkjet printers can be divided into two categories, according to their operation mode; the ink is either supplied from a reservoir that contains several milliliters of the ink (Figure 2.1a), or the printhead aspirates several microliters of ink from a separate reservoir, for example, from a microtiter plate (Figure 2.1b). The first (dispensing) mode is not suitable for the usage in a combinatorial experimentation process: the reservoir can contain only a single ink that requires manual handling when one would like to print another ink (composition). This additional process increases the complexity of the overall process by introducing manual cleaning steps. The second mode, referred to as micropipette mode, is a more suitable approach for the preparation of arrays and compound libraries because of a much more simple exchange of the ink. Before changing the ink, the nozzle still requires a cleaning step, but this is done in an automated manner by programming a sequence of purging and aspirating steps with a cleaning solvent from one of the wells in the microtiter plate [27]. On the other hand, the inks also require certain physical criteria for the application in the inkjet printing process [29]. Fromm introduced the dimensionless Z-number, which is the inverse of the Ohnesorge number (Oh) [30, 31]: Z = (dργ )1/2 /η = Oh−1

21

22

2 Combinatorial Screening of Materials Using Inkjet Printing as a Patterning Technique

6.0 10

Z (μm)

8

600.0

Y

0

X 0.0

4 2

600.0

(a)

6

(b)

0

200

400

600

X (μm) Figure 2.2 (a) Array of polymer dots printed at a mutual distance of 150 μm, formed by droplets of a 1 wt% solution of polystyrene in acetophenone and (b) cross section in the x-direction, corresponding to the line in (a). (Reprinted with permission from Ref. [33], copyright 2011, the American Chemical Society.)

where η, ρ, and γ are the viscosity, density, and surface tension of the liquid, respectively, and d is a characteristic length, which is the diameter of the nozzle aperture. Fromm predicted that when the Z-number is larger than two, the ink would be printable. Fromm’s prediction was investigated by Derby et al. [32], who studied a range of concentrated alumina wax suspensions and found that DoD printing was only possible in a range of 1 < Z < 10. The authors also observed that the droplet volume increases with Z in the range of 1–14, which is consistent with the predictions of Fromm. De Gans et al. [27] found that systems with Z numbers up to 91 were printable. The main factor that affected the printability was the vapor pressure, with unstable droplets, and no droplets being produced for solvents with vapor pressures higher than approximately 13 kPa. Starting from the smallest and most simple printed structures, de Gans et al. [33] inkjet printed a 4 × 4 array of dots (Figure 2.2). The deposition accuracy on the substrate could be improved by a reduced distance between nozzle and substrate, which subsequently reduces the disturbance to the flying droplet caused by the airflow. When printing from a single solvent, ringlike drying patterns were formed, caused by the well-known coffee-ring effect [18], whereas by using a solvent mixture of acetophenone/ethyl acetate dotlike structures were formed. The height of the printed features had a variation of 1–3.5%, which shows the reproducibility of the ejected droplet volume. Therefore, inkjet printing was found to be a suitable patterning technique for highly reproducible processes. When single droplets are printed close to each other, the droplets will merge and build structures, such as lines or films. For the formation of poly(3,4-ethylenedioxy-thiophene):poly(styrene-sulfonate) (PEDOT:PSS) lines, Soltman et al. [34] described the various morphologies that may appear, ranging from individual droplets to stacked coins, depending on the dot spacing of the deposited droplets as well as on the time delay between each droplet deposition

2.2 Inkjet Printing – from Well-Defined Dots to Homogeneous Films

100 μm

(a)

(b)

(c)

(d)

(e)

Figure 2.3 Examples of principal printed line behaviors: (a) individual drops, (b) scalloped, (c) uniform, (d) bulging, and (e) stacked coins. The dot spacing decreases from left to right. (Reprinted with permission from Ref. [34], copyright 2011, the American Chemical Society.)

(Figure 2.3). The smoothest and, therefore, most preferable topography was formed when the line was uniform and revealed a constant and defined line width as well as height (Figure 2.3c). Printing droplets with a large dot spacing led to individual dots (Figure 2.3a), while decreasing the dot spacing revealed scalloped lines, as shown in Figure 2.3b. In contrast, when too much material was deposited, that is, with small dot spacings, bulges (Figure 2.3d) or even a stacked coins configuration appeared (Figure 2.3e). The latter situation were also observed when the delay time between each droplet was increased. Stringer et al. [35] developed a model to predict a stable line formation that depends on the impact and impression of the droplet on the substrate. The formation of continuous lines is a result of inkjet printing droplets in one direction using a dot spacing that is equal to or smaller than the droplet’s diameter. Subsequently, when choosing the dot spacing both in the x- and y-directions smaller than the droplet’s diameter on the substrate, the formation of a continuous film will occur. Tekin et al. [36] investigated the influence of various parameters on the reproducibility of inkjet-printed films, including solvent mass ratio, printhead velocity, and dot spacing. The authors presented a detailed study of how defined and homogeneous films could be prepared by inkjet printing. Furthermore, the authors showed that the coffee-drop effect of the films could be reduced by using a

23

24

2 Combinatorial Screening of Materials Using Inkjet Printing as a Patterning Technique 4.00 3.45

4

2.90

1.80 646.85

μm

2.35

2

1.25 0.70 6787.50

0.15

0

−0.40

(a)

0.00

0

2000

4000

6000

μm

10.00 6.25

40

2.50 −1.25 −8.75 633.28

20

−12.50 7453.13

−16.25 −20.00

(b)

μm

−5.00

0.00

0 0

2000

4000

6000

μm

Figure 2.4 Confocal scanning microscopy images and corresponding cross sections of polystyrene films (2 wt%) inkjet printed from (a) butyl acetate and (b) from a solvent mixture consisting of 5/95 acetophenone/isopropyl acetate. (Reprinted with permission from Ref. [36], copyright 2011, the Royal Society of Chemistry.)

solvent mixture that consists of a low- and a high-boiling solvent, instead of using a single solvent (Figure 2.4). Once the right solvent mixture was found, further parameters that influence the film quality were systematically varied and optimized. In order to investigate the reproducibility of inkjet-printed thin-film libraries (Figure 2.5), the UV–vis absorption spectra were recorded for eight adjacent films that were printed with equal settings; an error of the absorption maximum of approximately 4% was observed (Figure 2.5b), similar to the results obtained from the height measurements from single droplets by de Gans et al. [33]. This relatively small error confirms the high reproducibility of the inkjet printing process. These results represent an important step toward the application of inkjet printing for the controlled deposition of functional materials, such as polymers, into thin-film libraries that can be used in a combinatorial experimental workflow. Furthermore, the inkjet-printed thin-film libraries can be used for a systematic screening of the printing parameters.

2.3 Thin-Film Libraries Prepared by Inkjet Printing

1.0

A Variation in printing parameters

B

0.8

C 0.6 A

D E

0.4

F 0.2

G H

0.0 400

(a)

25

Reproducibility

Figure 2.5 (a) Polystyrene films, inkjet printed using a solution containing 2 wt% polystyrene and 0.05 wt% Disperse Red 1, to improve the contrast of the films. The films were printed from a methyl benzoate/ethyl acetate solvent mixture. Print parameters and

(b)

500 l(nm)

solvent mixture composition vary from row to row. (b) Overlay of the absorbance spectra of eight adjacent films from row B in (a). (Reprinted with permission from Ref. [36], copyright 2011, the Royal Society of Chemistry.)

2.3 Thin-Film Libraries Prepared by Inkjet Printing

Using the approach of inkjet printing thin-film libraries, as discussed in the previous section, parameters that influence the quality of the films can be investigated, including the utilized solvents and the film thickness. Tekin et al. [37] studied the optical properties of the inkjet-printed films, which were made from various PPE-PPVs copolymers, as a function of the film thickness. PPE-PPVs with tailored alkoxy side chains are attractive materials because of their tunable band gap properties and, thus, their variable emission colors [38–40]. Moreover, these polymers are also promising electron donor materials for solar cell applications [41, 42]. Furthermore, the side chains also provide remarkable changes in the optical, electronic, and transport properties of the conjugated polymers in the solid state [39]. Figure 2.6 shows a thin-film library of six alkoxy-substituted PPE-PPVs inkjet printed with a systematically varied film thickness. The optical properties of the printed libraries were screened utilizing high-throughput methods [37]. It was found that the emission color of the investigated polymers strongly depends on the interchain interactions, which increased with increasing film thickness. Furthermore, it was shown that longer side chains revealed a redshift in the emission color of the PPE-PPVs. Wang et al. [43] also reported the influence of the film thickness on the optical properties of conjugated polymers. The authors found similar changes in the photoluminescence with increased film thickness for both inkjet-printed and spin-coated films. However, a better control over the film thickness was observed

600

26

2 Combinatorial Screening of Materials Using Inkjet Printing as a Patterning Technique

a

b

c

d

e

f

1 2 3 4 5 6

Figure 2.6 Inkjet-printed PPE-PPV thin-film library under UV illumination (wavelength: 390 nm). Each row displays images of a different derivative of PPE-PPV, printed with varying thickness from 50 nm (column a) to 150 nm (column f ). (Reprinted with permission from Ref. [37], copyright 2011, the Royal Society of Chemistry.)

when using inkjet printing and, thereby, also an enhanced examination of the optical properties of the resulting films. Besides conjugated polymers, semiconducting NCs are also of importance for the use in optoelectronic devices [44, 45]. Schubert and coworkers studied the emission properties of CdTe NCs that were embedded in a poly(vinylalcohol) (PVA) matrix. The properties were screened in a parallel manner by using inkjet printing to create thin-film libraries and by high-throughput characterization methods for a fast evaluation of structure–property relationships [46]. The inkjet-printed thin-film library, as depicted in Figure 2.7a, shows the influence of the PVA/CdTe ratio (columns) and the NC size (rows) on the emission properties. Besides a variation of the emission color with changing the NC size from 2.6 (row A) to 3.8 nm (row D), an increased emission intensity up to a maximum value was observed with increased PVA amount in the blend. The authors ascribe this effect to an increased interparticle distance, which prevents particle–particle interactions and subsequently self-quenching of the photoluminescence. Furthermore, the effects of the particle–particle interactions were determined from the emission spectra of the inkjet-printed films by mixing green- and red-emitting CdTe NCs (Figure 2.7b). A reduced emission intensity of the green-emitting particles was observed with an increased amount of red-emitting particles in the blend, indicating that an energy transfer took place from the smaller green to the larger red particles. A particular research area where combinatorial screening techniques prove their efficiency is in the fabrication and optimization of optoelectronic devices. In particular, the replacement of the expensive ITO as anode material by an

2.3 Thin-Film Libraries Prepared by Inkjet Printing

A B C D Increasing PVA /CdTe NCs ratio

(a) G (100%)

R (100%)

40

G/R (100/0) G/R (95/5) G/R (90/10) G/R (85/15) G/R (80/20) G/R (60/40) G/R (50/50) G/R (0/100)

PL intensity

30

20

10

0 500 (b)

550

600 650 l (nm)

Figure 2.7 (a) Photograph of an inkjetprinted thin-film library of different-sized CdTe NCs that emit at different wavelengths, including a systematic variation of PVA content from 0 to 1.4 wt% (in steps of 0.2 wt% and from left to right, respectively).

700

750

(b) Inkjet-printed library of green (G)- and red (R)-emitting CdTe NCs mixtures and the corresponding photoluminescence spectra. (Reprinted with permission from Ref. [46], copyright 2011, Wiley-VCH Verlag GmbH & CoKGaA.)

environmental-friendly polymeric material, such as PEDOT:PSS, in OLEDs and OPVs has gained increased interest in the past years [7, 47, 48]. Jabbour et al. [49] showed a simple modification of the sheet resistivity of conducting polymer anodes using combinatorial inkjet printing techniques. Different oxidizing agents were tested and inkjet printing was used for the patterning of a PEDOT:PSS anode as well as to obtain predefined shapes and control over the sheet resistivity [50]. Figure 2.8a shows a photograph of an OLED, including the patterned PEDOT:PSS anode. The

27

28

2 Combinatorial Screening of Materials Using Inkjet Printing as a Patterning Technique

Sheet resistivity (ohm /sq.)

1.00E + 06

1.00E + 05

(b)

16

22

45

57

Black 80 100

1.00E + 04

1.00E + 03

1.00E + 02 (a)

White 0 10

0

20

40

60

80

100

Ratio of the printed ink (%)

Figure 2.8 (a) Photograph of an OLED that was prepared by inkjet printing an oxidizing agent onto a PEDOT:PSS anode, resulting in a change in conductivity. (b) Measured sheet resistivity for different ratios of printed oxidizing ink. (Reprinted with permission from Ref. [49], copyright 2011, Wiley-VCH Verlag GmbH & CoKGaA.)

sheet resistivity of the PEDOT:PSS anode was varied by printing the oxidizer on top of the polymer layer using different concentrations, which was realized by altering the gray value of the printing pattern (Figure 2.8b). This approach showed that a library of electrodes with different sheet resistivities can be created within a short time using combinatorial inkjet printing techniques.

2.4 Combinatorial Screening of Materials for Organic Solar Cells

As a last topic in this chapter, examples are discussed where inkjet printing was used for the screening of a number of (printing) parameters and compounds for organic solar cells, in particular, bulk heterojunction solar cells, by using the thin-film library approach, as mentioned in the previous section [16, 51]. A basic device structure is displayed in Figure 2.9a. This type of solar cell is based on a layered structure, where the active layer consists of a mixture of electron–donor and electron–acceptor materials (Figure 2.9b). Critical parameters for the performance of bulk heterojunction solar cells are the donor/acceptor ratio, the film thickness, as well as the morphology of the resulting films. In particular, the morphology is highly important for the efficiency of an organic solar cell since excitons have to reach a donor–acceptor interface within a few nanometers after being created; in addition, the created charges need to reach the electrodes [52, 53]. As a consequence a good intermixing of donor and acceptor material is required for a good device functionality. Besides the donor/acceptor ratio, the most important parameters that influence the nanoscale morphology are the utilized processing solvent, the solute concentration, and the method of film preparation [54, 55].

2.4 Combinatorial Screening of Materials for Organic Solar Cells

Aluminum LiF Photoactive layer PEDOT-PSS ITO GLASS (a) m

n

O

O

O

O

2+

O

N N N Ru N N N

O 2 SbF6− Electron

(b)

Figure 2.9 (a) Design of a bulk heterojunction organic solar cell with a photoactive layer sandwiched between the electrodes ITO and aluminum. (b) The photoactive layer consists of a polymer with Ru-bipyridine

moieties, acting as the electron-donating material, and a fullerene derivative, which acts as the electron-accepting material. (Reprinted with permission from Ref. [56], copyright 2011, Wiley-VCH Verlag GmbH & CoKGaA.)

By inkjet printing thin-film libraries, the influence of ink composition, substrate properties, and different printing parameters to the film properties was studied systematically and in a fast, reproducible, and simple way with a high materials efficiency. Being a noncontact process, inkjet printing also enables large area and roll-to-roll (R2R) processing [57, 58]. Therefore, once a suitable candidate has been identified for the preparation of an organic solar cell, inkjet printing can be used as well to prepare thin and homogeneous layers of the active materials. Brabec and coworkers [59, 60] showed that inkjet printing can be used for the preparation of highly efficient solar cells by printing the active layer of poly(3-hexylthiophene) (P3HT) and (1-[3-(methoxycarbonyl)propyl]-1-phenyl)-[6,6]C61 (PCBM). The authors studied the influence of the solvent formulation on the film morphology and investigated the resulting device performances. An inhomogeneous and rough film surface was formed when using the single solvent tetralene (Figure 2.10a). In contrast, when using a solvent mixture of ortho-dichlorobenzene (o-DCB) and mesitylene (Figure 2.10b), homogeneous films were obtained. The authors found that the drying time of the inkjet-printed layer strongly affects the morphology of the photoactive layer and, hence, the power conversion efficiency (PCE) of the final device. Subsequently, the improved film properties,

29

2 Combinatorial Screening of Materials Using Inkjet Printing as a Patterning Technique

μm 0

10

20

30

nm 180 160 140 120 100 80 60 40 20 0

40

0 5 10 μm

15 20 25 210 μm

30 35

41.0 μm

40

41.3 μm

Length = 41.3 μm Pt = 129 nm Scale = 120 nm

nm

50 0 −50 0

5

10

(a)

15

20 μm

25

30

35

40

μm 0

10

20

30

40

0 5 10 μm

15 20 25 30

111 μm

35 40

41.3 μm

41.8 μm

nm 180 160 140 120 100 80 60 40 20 0

Length = 41.3 μm Pt = 14.1 nm Scale = 120 nm nm

30

60 40 20 0 −20 −40 0

5

10

15

(b) 4m 2m 0 −2m −4m

20 μm

25

30

35

40

Tetralene formulation −2 Jsc = 4.73 mA cm Voc = 0.45 V FF = 0.63 PCE = 1.29 % o-DCB/Mesitylene formulation −2 Jsc = 8.4 mA cm Voc = 0.535 V FF = 0.64 PCE = 2.9%

−6m −8m −10m −0.1 (c)

0.0

0.1

0.2

0.3

Voltage (V)

0.4

0.5

0.6

Figure 2.10 Atomic force microscopy images of P3HT/PCBM blends prepared by inkjet printing using the solvent (a) tetralene and (b) o-DCB/mesitylene; and (c) resulting solar cell performances. (Reprinted with permission from Ref. [60], copyright 2011, Wiley-VCH Verlag GmbH & CoKGaA.)

2.4 Combinatorial Screening of Materials for Organic Solar Cells

when inkjet printed from the solvent mixture, resulted in an enhanced solar cell performance with a PCE of 2.9% (Figure 2.10c). Lange et al. [61] reported a PCE of 2.4% for P3HT/PCBM when inkjet printed from a solvent system consisting of chlorobenzene and trichlorobenzene in a ratio of 55/45 by weight. Hoth et al. [62] showed the potential of inkjet printing for the production of solar cells with high efficiencies, but a comparative screening of the prepared films was not presented. Although many morphology studies of the active layer are reported in the literature, knowledge of the detailed working principle of an organic solar cell is still lacking [63, 64]. Newly discovered donor/acceptor materials require a large number of combinations to be tested for their behavior in bulk heterojunction solar cell applications [65, 66]. For this purpose, an enormous number of samples need to be screened for the evaluation of promising polymer/fullerene combinations and good processing conditions, which obviously consumes a lot of resources, both in terms of time and related costs. Instead of studying potential materials in a sequential and one-by-one manner, for example, by spin coating as recently shown by Renz et al. [67], inkjet printing of thin-film libraries allows a much faster screening of materials in parallel. Moreover, synergies that may exist between various parameters would only be discovered in a combinatorial approach. Walter et al. [68] presented a highly automated screening platform for functional materials, based on spin coating as the film preparation technique, but their method was not material efficient. In contrast, Marin et al. used a combinatorial approach, including inkjet printing, for the investigation of ruthenium(II)-containing polymers as electron-donating materials. The polymer was combined with two different electron-accepting materials, including PCBM and 1,1 -diheptyl-4,4 -bipyridine dibromide (heptyl viologen) [56]. The schematic representation of a 20-array film library prepared by inkjet printing is displayed in Figure 2.11a. A fast evaluation of the most promising donor/acceptor combinations and ratios using high-throughput screening techniques was obtained by measuring the quenching of the polymer emission by adding the acceptor material in certain amounts. An efficient charge transfer was observed from the polymer, which is the light-absorbing compound, to the acceptor material. The authors concluded that this is a hint for a good intermixing of the two components in the bulk layer, since the created excitons survived over a length of 10 nm. Furthermore, the authors showed the fluorescence quenching as a function of the component ratios in the polymer/PCBM and polymer/viologen blends and reported a poor quenching efficiency when using the viologen as acceptor material. In contrast, when using PCBM in a ratio of 1 : 2 led to an efficient charge transfer from the ruthenium(II)-containing polymer to the fullerene (Figure 2.11b). Recently, a straightforward screening method was presented by Teichler et al. [69]. The described experimental workflow, as shown in Figure 2.12, allowed a fast and simple evaluation of potential donor/acceptor combinations for usage in OPV. Using high-throughput solution preparation, solution deposition as well as property screening techniques, the loop of synthesizing and screening new materials was closed.

31

2 Combinatorial Screening of Materials Using Inkjet Printing as a Patterning Technique T1

T2

T1

T2

1 : 0.0 1 : 0.5 1 : 1.0 1 : 1.5

RuPMMA : quencher molar ratios

1:2

(a)

PC60BM

1000

750 Emission (a.u.)

32

Viologen

RuPMMA:PCBM (1 : 0.0) RuPMMA:PCBM (1 : 0.5) RuPMMA:PCBM (1 : 1.0) RuPMMA:PCBM (1 : 1.5) RuPMMA:PCBM (1 : 2.0)

500

250

0 500 (b)

600 700 Wavelength (nm)

800

Figure 2.11 (a) Layout of an inkjet-printed polymer/fullerene library. (b) Emission quenching as a function of the polymer/fullerene ratio. (Reprinted with permission from Ref. [56], copyright 2011, Wiley-VCH Verlag GmbH & CoKGaA.)

A pipetting robot was utilized to prepare donor/acceptor solutions in a standard 96-well plate, while varying systematically the compound ratio, the solvent mixture, as well as the concentration. The as-prepared solutions in the well plate were then used as small reservoirs for the inkjet printing process. The thin-film libraries were printed onto the substrate in a microtiter plate layout, that is, each film measured 5 × 5 mm2 and the distance of the rectangles corresponds to the distance between the wells of a 96-well plate. These printed libraries were then able to be characterized in a fast and efficient manner, using standard industrial analytical tools, including UV–vis and Fourier transform infrared (FTIR) spectroscopy, and Ramanplate readers. Using the described combinatorial setup, two conjugated polymers, (poly[2, 6-(4,4-bis(2-ethylhexyl)-4H-cyclopenta[2,1-b;3,4-b]dithiophene)-alt-4,7(2,1,3-benzo-

2.4 Combinatorial Screening of Materials for Organic Solar Cells

New compounds O O

S

S

n

Normalized intensity at max (counts)

N

10

10

-2

10

-3

N S

100

-1

CH3

Bladed reference (100 nm) Inkjet printed (150−200 nm)

Poly

4:1

1:1

1:2

Solution preparation with robot

1:4

PCPD TBT:mono-PCBM ratio

Screening of film properties

Inkjet printing of film libraries Figure 2.12 Experimental workflow for the screening of donor/acceptor film libraries for organic solar cell applications using high-throughput experimentation techniques. (Reprinted with permission from Ref. [69], copyright 2011, Wiley-VCH Verlag GmbH & CoKGaA.)

thiadiazole)] (PCPDTBT); Figure 2.13a) and (poly[(4,4 -bis-(2-ethylhexyl)-dithieno (3,2-b;2,3 -d)silole]-2,6-diyl-alt-(2,1,3-benzothiadiazole)-4,7-diyl]) (PSBTBT); Figure 2.13b), and two fullerene derivatives, mono-PCBM and bis-PCBM, were used to prepare different blend compositions. A combinatorial and reproducible screening of PCPDTBT/bis-PCBM, PCPDTBT/mono-PCBM, and PSBTBT/mono-PCBM blends was realized by inkjet printing thin-film libraries with a systematically varied film thickness, concentration, solvent ratio, and blend composition. The polymer PCPDTBT revealed a smooth film formation in all blends (Figure 2.13c). In contrast, the polymer PSBTBT, which is almost identical to PCPDTBT, apart from the bridging C-atom in the backbone that is replaced by a Si-atom, showed aggregate formation due to a reduced solubility of the polymer in organic solvents that is caused by a higher crystallinity (Figure 2.13d) [70–72]. Selected from the combinatorial study, two blends were tested for their solar cell activity. A maximum PCE of 0.64 and 1.48% was obtained for the PSBTBT/mono-PCBM, and PCPDTBT/mono-PCBM system, respectively.

33

34

2 Combinatorial Screening of Materials Using Inkjet Printing as a Patterning Technique

Si

S

S

n

S N

N

N S

(a) 4:1

n

S

(b) 1:1

nm

N S

1 : 3.4 nm

nm

250

200

200

150

150

100

100

50

50

50

0

0

0

−50

−50

−100

−100

−150

−150

−200

−200

−200

−250

200 150 100

−50 −100 −150

(c) 4:1

1:1 nm

1:2 nm

nm

150

150

100

100

50

50

0

0

−50

−50

−100

−100

−150

−150

200 150 100 50 0 −50 −100 −150 −200

(d)

Figure 2.13 Schematic representation of the chemical structures of (a) PCPDTBT and (b) PSBTBT; optical profiler images of inkjet-printed films from different (c) PCPDTBT/mono-PCBM, and (d) PSBTBT/mono-PCBM ratios prepared from

chlorobenzene (CB)/o-DCB 90/10 and a concentration of 0.5 wt%. The scale bar corresponds to 1 mm. (Reprinted with permission from Ref. [69], copyright 2011, Wiley-VCH Verlag GmbH & CoKGaA.)

2.5 Conclusion and Outlook

This chapter summarized the advantages of using inkjet printing as a solute deposition and patterning technique in combinatorial screening approaches. In the first and second parts, the need for a fast, simple, and material-efficient combinatorial workflow, as well as the requirements concerning instruments and inks, was discussed. The optimization for the preparation of well-defined and, most importantly, reproducible printed features, including dots, lines, and films, was obtained by using a solvent system that consist of a low- and a high-boiling solvent.

References

This combination of solvents reduced the coffee-drop effect and revealed smooth printed structures. By incorporating inkjet printing into a combinatorial workflow, a significant step was realized toward a fast screening of important materials properties for applications such as OLEDs and OPVs. The third section discussed the investigation of important structure–property relationships using inkjet-printed thin-film libraries. A more detailed understanding regarding the chemical structure of a compound and the resulting material properties by the systematic investigation of inkjet-printed thin-film libraries was demonstrated. As an example, PPE-PPV libraries were prepared with variations in the polymer side groups, which led to significant changes in the optical properties. Furthermore, it was shown that a library of polymeric anodes of the conductive polymer PEDOT:PSS with different sheet resistivities can be created within a short time using inkjet printing of an oxidative ink. In the fourth section, the experimental setup and a combinatorial study were presented to enable a fast and efficient screening of polymer/fullerene blends in the area of OPV. It was shown that using the combinatorial screening setup including inkjet printing led to optimized film properties, PCEs up to 2.9% could be reached. The effects of additives on the device characteristics can also be screened with the described combinatorial workflow, which would further optimize the device properties. Moreover, additional future directions should also include automated device-producing steps in the combinatorial experimentation workflow to evolve direct relationships between layer properties and device performances to reveal a more detailed understanding of the working principle of the devices. This chapter summarized a selection of examples where inkjet printing was used for combinatorial materials research. Inkjet printing of thin-film libraries shows significant advantages over other methods to screen materials, for example, spin coating. Moreover, it closes the loop from materials synthesis, deposition, and characterization toward optimization. Therefore, the inkjet printing technique may be used more often in the near future in order to screen solid-state compound properties in a fast and more efficient materials usage. In particular, research areas where materials optimization requires many and complex synthesis steps, for example, involving the synthesis of polymers or polymer complexes, an efficient compound handling represents a prerequisite, because then the amount of waste materials can be reduced significantly and small amounts of materials are sufficient. It is foreseen that this materials research is used not only for plastic electronics applications, as most examples were taken from this area, but also for other (new) areas, such as biology, for proteins and drug screening, as well as for ceramics and sensor applications.

References 1. Yim, Y., Park, J., and Park, B.

(2010) Solution-processed flexible ITO-free organic light-emitting

diodes using patterned polymeric anodes. J. Display Technol., 6 (7), 252–256.

35

36

2 Combinatorial Screening of Materials Using Inkjet Printing as a Patterning Technique 2. Kim, T., Suh, M., Kwon, S.J., Lee,

3.

4.

5.

6.

7.

8.

9.

10.

T.H., Kim, J.E., Lee, Y.J., Kim, J.H., Hong, M., and Suh, K.S. (2009) Poly(3,4-ethylenedioxythiophene) derived from poly(ionic liquid) for the use as hole-injecting material in organic light-emitting diodes. Macromol. Rapid Commun., 30 (17), 1477–1482. Thompson, B.C. and Frechet, J.M.J. (2008) Organic photovoltaics – polymer-fullerene composite solar cells. Angew. Chem. Int. Ed., 47 (1), 58–77. Dennler, G., Scharber, M.C., and Brabec, C.J. (2009) Polymer-fullerene bulk-heterojunction solar cells. Adv. Mater., 21 (13), 1323–1338. Madec, M.B., Smith, P.J., Malandraki, A., Wang, N., Korvink, J.G., and Yeates, S.G. (2010) Enhanced reproducibility of inkjet printed organic thin film transistors based on solution processable polymer-small molecule blends. J. Mater. Chem., 20 (41), 9155–9160. Choi, M.H., Han, S.H., Lee, S.H., Choo, D.J., Jang, J., and Kwon, S.K. (2009) Effect of active layer thickness on environmental stability of printed thin-film transistor. Org. Electron., 10 (3), 421–425. Do, H., Reinhard, M., Vogeler, H., Puetz, A., Klein, M.F.G., Schabel, W., Colsmann, A., and Lemmer, U. (2009) Polymeric anodes from poly(3,4-ethylenedioxythiophene):poly (styrenesulfonate) for 3.5% efficient organic solar cells. Thin Solid Films, 517 (20), 5900–5902. Nielsen, T.D., Cruickshank, C., Foged, S., Thorsen, J., and Krebs, F.C. (2010) Business, market and intellectual property analysis of polymer solar cells. Sol. Energy Mater. Sol. Cells, 94 (10), 1553–1571. Krebs, F.C. (2009) Fabrication and processing of polymer solar cells: a review of printing and coating techniques. Sol. Energy Mater. Sol. Cells, 93 (4), 394–412. Zoombelt, A.P., Leenen, M.A.M., Fonrodona, M., Nicolas, Y., Wienk, M.M., and Janssen, R.A.J. (2009) The influence of side chains on solubility and photovoltaic performance of dithiophene-thienopyrazine small

11.

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

band gap copolymers. Polymer, 50 (19), 4564–4570. Egbe, D.A.M., Roll, C.P., Birckner, E., Grummt, U.W., Stockmann, R., and Klemm, E. (2002) Side chain effects in hybrid PPV/PPE polymers. Macromolecules, 35 (10), 3825–3837. Zhang, H.Q., Hoogenboom, R., Meier, M.A.R., and Schubert, U.S. (2005) Combinatorial and high-throughput approaches in polymer science. Meas. Sci. Technol., 16 (1), 203–211. Schubert, U.S., de Gans, B.J., and Kazancioglu, E. (2004) Combinatorial and high-throughput polymer research: composition of complete workflows. Abstr. Pap. Am. Chem. Soc., 227, U547. Hoogenboom, R., Meier, M.A.R., and Schubert, U.S. (2003) Combinatorial methods, automated synthesis and high-throughput screening in polymer research: past and present. Macromol. Rapid Commun., 24 (1), 16–32. Webster, D.C. and Meier, M.A.R. (2010) Polymer libraries: preparation and applications. Polym. Libr., 225, 1–15. Gunes, S., Neugebauer, H., and Sariciftci, N.S. (2007) Conjugated polymer-based organic solar cells. Chem. Rev., 107 (4), 1324–1338. Chang, C.C., Pai, C.L., Chen, W.C., and Jenekhe, S.A. (2005) Spin coating of conjugated polymers for electronic and optoelectronic applications. Thin Solid Films, 479 (1–2), 254–260. Deegan, R.D. (2000) Pattern formation in drying drops. Phys. Rev. E, 61 (1), 475–485. Arrabito, G., Musumeci, C., Aiello, V., Libertino, S., Compagnini, G., and Pignataro, B. (2009) On the relationship between jetted inks and printed biopatterns: molecular-thin functional microarrays of glucose oxidase. Langmuir, 25 (11), 6312–6318. Delaney, J.T., Smith, P.J., and Schubert, U.S. (2009) Inkjet printing of proteins. Soft Matter, 5 (24), 4866–4877. Arrabito, G. and Pignataro, B. (2010) Inkjet printing methodologies for drug screening. Anal. Chem., 82 (8), 3104–3107.

References 22. Lemmo, A.V., Rose, D.J., and Tisone,

23.

24.

25.

26.

27.

28. 29.

30.

31.

32.

T.C. (1998) Inkjet dispensing technology: applications in drug discovery. Curr. Opin. Biotech., 9 (6), 615–617. Wang, J., Mohebi, M.M., and Evans, J.R.G. (2005) Two methods to generate multiple compositions in combinatorial ink-jet printing of ceramics. Macromol. Rapid Commun., 26 (4), 304–309. Wang, J. and Evans, J.R.G. (2005) Library preparation using an aspirating-dispensing ink-jet printer for combinatorial studies in ceramics. J. Mater. Res., 20 (10), 2733–2740. Li, B., Santhanam, S., Schultz, L., Jeffries-EL, M., Iovu, M.C., Sauve, G., Cooper, J., Zhang, R., Revelli, J.C., Kusne, A.G., Snyder, J.L., Kowalewski, T., Weiss, L.E., McCullough, R.D., Fedder, G.K., and Lambeth, D.N. (2007) Inkjet printed chemical sensor array based on polythiophene conductive polymers. Sens. Actuators B-Chem., 123 (2), 651–660. de Gans, B.J. and Schubert, U.S. (2003) Inkjet printing of polymer micro-arrays and libraries: Instrumentation, requirements, and perspectives. Macromol. Rapid Commun., 24 (11), 659–666. de Gans, B.J., Kazancioglu, E., Meyer, W., and Schubert, U.S. (2004) Ink-jet printing polymers and polymer libraries using micropipettes. Macromol. Rapid Commun., 25 (1), 292–296. http://www.microdrop.de (accessed December 8 2010). Jang, D., Kim, D., and Moon, J. (2009) Influence of fluid physical properties on ink-jet printability. Langmuir, 25 (5), 2629–2635. Derby, B. (2010) Inkjet printing of functional and structural materials: fluid property requirements, feature stability, and resolution. Annu. Rev. Mater. Res., 40, 395–414. Fromm, J.E. (1984) Numerical-calculation of the fluid-dynamics of drop-on-demand jets. IBM J. Res. Dev., 28 (3), 322–333. Reis, N., Ainsley, C., and Derby, B. (2005) Ink-jet delivery of particle suspensions by piezoelectric droplet ejectors. J. Appl. Phys., 97 (9), 094903.

33. de Gans, B.J. and Schubert, U.S. (2004)

34.

35.

36.

37.

38.

39.

40.

41.

42.

Inkjet printing of well-defined polymer dots and arrays. Langmuir, 20 (18), 7789–7793. Soltman, D. and Subramanian, V. (2008) Inkjet-printed line morphologies and temperature control of the coffee ring effect. Langmuir, 24 (5), 2224–2231. Stringer, J. and Derby, B. (2010) Formation and stability of lines produced by inkjet printing. Langmuir, 26 (12), 10365–10372. Tekin, E., de Gans, B.J., and Schubert, U.S. (2004) Ink-jet printing of polymers - from single dots to thin film libraries. J. Mater. Chem., 14 (17), 2627–2632. Tekin, E., Wijlaars, H., Holder, E., Egbe, D.A.M., and Schubert, U.S. (2006) Film thickness dependency of the emission colors of PPE-PPVs in inkjet printed libraries. J. Mater. Chem., 16 (44), 4294–4298. Tekin, E., Egbe, D.A.M., Kranenburg, J.M., Ulbricht, C., Rathgeber, S., Birckner, E., Rehmann, N., Meerholz, K., and Schubert, U.S. (2008) Effect of side chain length variation on the optical properties of PPE-PPV hybrid polymers. Chem. Mater., 20 (8), 2727–2735. Egbe, D.A.M., Ulbricht, C., Orgis, T., Carbonnier, B., Kietzke, T., Peip, M., Metzner, M., Gericke, M., Birckner, E., Pakula, T., Neher, D., and Grumm, U.W. (2005) Odd-even effects and the influence of length and specific positioning of alkoxy side chains on the optical properties of PPE-PPV polymers. Chem. Mater., 17 (24), 6022–6032. Egbe, D.A.M., Carbonnier, B., Birckner, E., and Grummt, U.W. (2009) Arylene-ethynylene/arylene-vinylene copolymers: synthesis and structure-property relationships. Prog. Polym. Sci., 34 (10), 1023–1067. Hoppe, H., Egbe, D.A.M., Muhlbacher, D., and Sariciftci, N.S. (2004) Photovoltaic action of conjugated polymer/fullerene bulk heterojunction solar cells using novel PPE-PPV copolymers. J. Mater. Chem., 14 (23), 3462–3467. Hoppe, H., Sariciftci, N.S., Egbe, D.A.M., Muhlbacher, D., and Koppe, M. (2005) Plastic solar cells based on

37

38

2 Combinatorial Screening of Materials Using Inkjet Printing as a Patterning Technique

43.

44.

45.

46.

47.

48.

49.

50.

51.

52.

novel PPE-PPV-copolymers. Mol. Cryst. Liquid Cryst., 426, 255–263. Wang, M., Yang, G.Z., Wang, M., and Liu, T.X. (2010) Effect of film thickness controlled by ink-jet printing method on the optical properties of an electroluminescent polymer. Polym. Adv. Technol., 21 (6), 381–385. Kim, J.Y., Ingrosso, C., Fakhfouri, V., Striccoli, M., Agostiano, A., Curri, M.L., and Brugger, J. (2009) Inkjet-printed multicolor arrays of highly luminescent nanocrystal-based nanocomposites. Small, 5 (9), 1051–1057. Kim, H.S., Kang, J.S., Park, J.S., Hahn, H.T., Jung, H.C., and Joung, J.W. (2009) Inkjet printed electronics for multifunctional composite structure. Compos. Sci. Technol., 69 (7–8), 1256–1264. Tekin, E., Smith, P.J., Hoeppener, S., van den Berg, A.M.J., Susha, A.S., Rogach, A.L., Feldmann, J., and Schubert, U.S. (2007) Inkjet printing of luminescent CdTe nanocrystal-polymer composites. Adv. Funct. Mater., 17 (1), 23–28. Carter, S.A., Angelopoulos, M., Karg, S., Brock, P.J., and Scott, J.C. (1997) Polymeric anodes for improved polymer light-emitting diode performance. Appl. Phys. Lett., 70 (16), 2067–2069. Lai, S.L., Chan, M.Y., Fung, M.K., Lee, C.S., and Lee, S.T. (2003) Concentration effect of glycerol on the conductivity of PEDOT film and the device performance. Mat. Sci. Eng. B-Solid, 104 (1–2), 26–30. Yoshioka, Y., Calvert, P.D., and Jabbour, G.E. (2005) Simple modification of sheet resistivity of conducting polymeric anodes via combinatorial ink-jet printing techniques. Macromol. Rapid Commun., 26 (4), 238–246. Yoshioka, Y. and Jabbour, G.E. (2006) Inkjet printing of oxidants for patterning of nanometer-thick conducting polymer electrodes. Adv. Mater., 18 (10), 1307–1312. Brabec, C.J., Sariciftci, N.S., and Hummelen, J.C. (2001) Plastic solar cells. Adv. Funct. Mater., 11 (1), 15–26. Yang, X. and Loos, J. (2007) Toward high-performance polymer solar cells:

53.

54.

55.

56.

57.

58.

59.

60.

61.

62.

the importance of morphology control. Macromolecules, 40 (5), 1353–1362. Hoppe, H. and Sariciftci, N.S. (2006) Morphology of polymer/fullerene bulk heterojunction solar cells. J. Mater. Chem., 16 (1), 45–61. Peet, J., Senatore, M.L., Heeger, A.J., and Bazan, G.C. (2009) The role of processing in the fabrication and optimization of plastic solar cells. Adv. Mater., 21 (14–15), 1521–1527. Barrau, S., Andersson, V., Zhang, F.L., Masich, S., Bijleveld, J., Andersson, M.R., and Inganas, O. (2009) Nanomorphology of bulk heterojunction organic solar cells in 2D and 3D correlated to photovoltaic performance. Macromolecules, 42 (13), 4646–4650. Marin, V., Holder, E., Wienk, M.M., Tekin, E., Kozodaev, D., and Schubert, U.S. (2005) Ink-jet printing of electron donor/acceptor blends: towards bulk heterojunction solar cells. Macromol. Rapid Commun., 26 (4), 319–324. Bundgaard, E., Hagemann, O., Manceau, M., Jorgensen, M., and Krebs, F.C. (2010) Low band gap polymers for roll-to-roll coated polymer solar cells. Macromolecules, 43 (19), 8115–8120. Krebs, F.C. (2009) All solution roll-to-roll processed polymer solar cells free from indium-tin-oxide and vacuum coating steps. Org. Electron., 10 (5), 761–768. Hoth, C.N., Schilinsky, P., Choulis, S.A., and Brabec, C.J. (2008) Printing highly efficient organic solar cells. Nano Lett., 8 (9), 2806–2813. Hoth, C.N., Choulis, S.A., Schilinsky, P., and Brabec, C.J. (2007) High photovoltaic performance of inkjet printed polymer: fullerene blends. Adv. Mater., 19 (22), 3973–3978. Lange, A., Wegener, M., Boeffel, C., Fischer, B., Wedel, A., and Neher, D. (2010) A new approach to the solvent system for inkjet-printed P3HT:PCBM solar cells and its use in devices with printed passive and active layers. Sol. Energy Mater. Sol. Cells, 94 (10), 1816–1821. Hoth, C.N., Choulis, S.A., Schilinsky, P., and Brabec, C.J. (2009) On the effect of poly(3-hexylthiophene) regioregularity

References

63.

64.

65.

66.

67.

68.

on inkjet printed organic solar cells. J. Mater. Chem., 19 (30), 5398–5404. van Bavel, S.S., Barenklau, M., de With, G., Hoppe, H., and Loos, J. (2010) P3HT/PCBM bulk heterojunction solar cells: impact of blend composition and 3D morphology on device performance. Adv. Funct. Mater., 20 (9), 1458–1463. van Bavel, S., Sourty, E., de With, G., Veenstra, S., and Loos, J. (2009) Three-dimensional nanoscale organization of polymer solar cells. J. Mater. Chem., 19 (30), 5388–5393. Troshin, P.A., Hoppe, H., Renz, J., Egginger, M., Mayorova, J.Y., Goryochev, A.E., Peregudov, A.S., Lyubovskaya, R.N., Gobsch, G., Sariciftci, N.S., and Razumov, V.F. (2009) Material solubility-photovoltaic performance relationship in the design of novel fullerene derivatives for bulk heterojunction solar cells. Adv. Funct. Mater., 19 (5), 779–788. Zoombelt, A.P., Mathijssen, S.G.J., Turbiez, M.G.R., Wienk, M.M., and Janssen, R.A.J. (2010) Small band gap polymers based on diketopyrrolopyrrole. J. Mater. Chem., 20 (11), 2240–2246. Renz, J.A., Keller, T., Schneider, M., Shokhovets, S., Jandt, K.D., Gobsch, G., and Hoppe, H. (2009) Multiparametric optimization of polymer solar cells: a route to reproducible high efficiency. Sol. Energy Mater. Sol. Cells, 93 (4), 508–513. Walter, H., Basset, G., Beierlein, T., Von Muhlenen, A., and Nisato, G.

69.

70.

71.

72.

(2010) Combinatorial approach for fast screening of functional materials. J. Polym. Sci. Part B: Polym. Phys., 48 (14), 1587–1593. Teichler, A., Eckardt, R., Hoeppener, S., Friebe, C., Perelaer, J., Senes, A., Morana, M., Brabec, C.J., and Schubert, U.S. (2011) Combinatorial screening of polymer:fullerene blends for organic solar cells by inkjet printing. Adv. Energy Mater., 1 (1), 105–114. Chen, H.Y., Hou, J.H., Hayden, A.E., Yang, H., Houk, K.N., and Yang, Y. (2010) Silicon atom substitution enhances interchain packing in a thiophene-based polymer system. Adv. Mater., 22 (3), 371–375. Scharber, M.C., Koppe, M., Gao, J., Cordella, F., Loi, M.A., Denk, P., Morana, M., Egelhaaf, H.J., Forberich, K., Dennler, G., Gaudiana, R., Waller, D., Zhu, Z.G., Shi, X.B., and Brabec, C.J. (2010) Influence of the bridging atom on the performance of a low-bandgap bulk heterojunction solar cell. Adv. Mater., 22 (3), 367–370. Morana, M., Azimi, H., Dennler, G., Egelhaaf, H.J., Scharber, M., Forberich, K., Hauch, J., Gaudiana, R., Waller, D., Zhu, Z.H., Hingerl, K., van Bavel, S.S., Loos, J., and Brabec, C.J. (2010) Nanomorphology and charge generation in bulk heterojunctions based on low-bandgap dithiophene polymers with different bridging atoms. Adv. Funct. Mater., 20 (7), 1180–1188.

39

41

3 Thermal Inkjet Naoki Morita

3.1 History of Thermal Inkjet Technology

It is said that the moment that thermal inkjet (TIJ) technology was invented was in the late 1970s at a Canon Inc. (Canon) research lab. During an experiment conducted by Ichiro Endo, a hot soldering iron accidentally came in contact with the needle of a syringe that was filled with ink, and the ink fired out with force. However, Ricoh Co., Ltd. (Ricoh) was quicker to submit a patent application of this technology by three days. Ricoh filed an application on 30 September 1977 (JP-S54-51837), while Canon filed an application on 3 October 1977 (JP-S54-59936). This coincidence must have occurred as many researchers in the 1970s were pursuing the principle of this new marking technology. Although it has been said that the episode with the hot soldering iron is a story that was made up by Canon, the author chooses to believe this fun episode by Canon instead. In terms of commercialization, following the invention of TIJ by J. Vaught in 1979, Hewlett-Packard Development Company L.P. (HP) introduced a TIJ printer named ‘‘ThinkJet’’ in 1984, and Canon introduced the ‘‘BJ80’’ to the market in 1985. Even with regard to colorization, HP was the first to introduce ‘‘PaintJet’’ in 1986. After this, HP, which did not possess its own electrophotographic technology, aspired to produce products for office use and released the ‘‘DJ1200’’ in 1993, a printer for the small office/home office (SOHO) market equipped with an ink-drying heater, and the CM8050/8060 in 2007, a TIJ multifunction printer (MFP) that can completely replace electrophotographic machines. In the meantime, since office electrophotography was already an existing business for Canon, it was not active in producing inkjets for office use. With regard to piezoelectric inkjets, the competing technology to TIJ, Seiko Epson Corporation (Epson) introduced the PM700 as a device for outputting photos using glossy paper and photo ink in 1996, in association with the popularization of digital cameras. Canon, for which photos are essentially their important pillar of business, has commenced intense competition with Epson in terms of photo image quality in the area of photo printers, but during this time, silver halide photography itself has declined, and the photo printing market has not grown as expected. Inkjet-based Micromanufacturing, First Edition. Edited by J. G. Korvink, P. J. Smith, and D.-Y. Shin. © 2012 Wiley-VCH Verlag GmbH & Co. KGaA. Published 2012 by Wiley-VCH Verlag GmbH & Co. KGaA.

42

3 Thermal Inkjet

The wide format market can be mentioned as a market that is adapted to inkjet technology. Following their appearance in the early 1990s, wide printing for printing of graphical posters, machine drawings using A2 or larger paper, and printing banners used for decorating the exterior of buildings, became an unrivaled territory for inkjets. In addition, attempts to apply inkjet technology to the publishing market are being carried out actively at the moment. In 2004, Riso Kogaku Corporation/Olympus Corporation gathered some success in releasing the ‘‘Orphis HC5000’’ with full-width heads that stretch across the width of the paper using Xaar-type heads (Xaar plc: piezoelectric head manufacturer and licenser) manufactured by Toshiba TEC Corporation, and carries out high-speed printing of 100 A4 sheets per minute or more using oil-based ink. As an application of inkjet technology to publishing machines that can print on A2 and larger paper, Dainippon Screen Mfg. Co., Ltd. was the first to produce the ‘‘Truepress Jet 520,’’ which comprises full-width Epson heads, in 2006. Furthermore, a series of inkjet devices, such as the ‘‘Jet Press 720’’ by FujiFilm Corporation that uses Fujifilm Dimatix Inc. (Spectra) heads, were announced at Drupa 2008. Since various companies announced their inkjet devices, the exhibition was called Inkjet Drupa. One characteristic of these devices is the use of piezoelectric technology, rather than TIJ technology. The primary characteristic of TIJ technology is that since the actuator for inkjetting is a heater, it can be manufactured by integrated circuit (IC) manufacturing processing, thus enabling the provision of compact printers at low prices. As a result, such technology gathered large success in popularization among households. On the other hand, the piezoelectric method is becoming more predominant in the research and development area, since TIJ ink is limited to only fluids that can be ‘‘well’’ boiled, and since the lifetime of the head is short in comparison.

3.2 Market Trends for Inkjet Products and Electrophotography

Figure 3.1 shows unit numbers of both inkjet and electrophotography printers sold all over the world from 2007 to 2009. The number of inkjet machines sold is 90−80 million units, which is about twice that of electrophotography machines. Both are slightly decreasing in volume, which indicates that the market is matured and starting to shrink. Figure 3.2 shows the total inkjet machines sold in 2009, where HP holds an overwhelming share of 46%, and Canon holds the second position at 23%. Epson and Brother Industries Ltd. occupy the third and fourth positions with their piezoelectric printers, holding 19 and 5% respectively. In total, TIJ occupies 76% of all inkjet printers.

3.3 Structures of Various TIJ Heads 100,000 90,000 80,000 70,000 60,000 50,000 40,000

Printer

30,000

MFP

20,000 10,000 K units Inkjet

Laser

Inkjet

2007

Laser

2008

Inkjet

Laser

2009

Figure 3.1 Worldwide printer and MFP sold for both inkjet and electrophotography. (Data source: IDL.) Lexmark 3,447

Dell 810

Kodak 1,151

Others 149

Brother 3,563 [k units] Epson 14,380 HP 35,615

Canon 17,656 Samsung 236

Figure 3.2 Companies and numbers of inkjet printers sold in 2009. (Data source: IDL.)

3.3 Structures of Various TIJ Heads

Table 3.1 shows the types of TIJ printhead structures, classified according to the configuration of the heater in the ink chamber and the direction of the jet to be fired. Major players adopt the ‘‘roof shooter’’ structure, as shown in Figure 3.3, where jets are fired in the same direction as the bubble growth, which is most effective in applying kinetic energy to inks. Sony Corporation realized the deflection of jets by utilizing two heaters that have different currents applied to each heater. This makes each heater generate bubbles at different timings, so that the bubbles deflect and result in jet deflection.

43

44 Table 3.1

3 Thermal Inkjet

Printhead structure.

Name

Roof shooter

Side shooter

Back shooter

Floating

Double surrounding

Single narrow, long

Heater(s) configuration

Single rectangular

Double rectangular

Single rectangular

Employing company

HP Canon third generation Lexmark Kodak

Sony

Canon first, BenQ second generation Xerox

Silverbrook

Head structure

Jet Nozzle Bubble Heater Substrate

Figure 3.3

Printhead configuration (roof shooter/single heater).

Xerox Corporation and Canon first and second generation printheads had in the past incorporated the ‘‘side-shooter’’ structure, which was suitable for mechanical manufacturing, such as in processing by dicing. Ben Q Corporation adopts a unique ‘‘back shooter’’ structure, which has a series of two heaters placed behind the nozzle plate to surround both sides of the nozzle; jets are fired opposite to the bubble growth [1]. The product is sold mainly in China. Silverbrook Research Pty Ltd. also announced a unique ‘‘floating’’ structure or ‘‘bridgelike’’ heater inside the ink chamber that can generate bubbles from both surfaces of the heater, which can realize twice the effectiveness compared to other methods [2]. 3.4 Research on Rapid Boiling and Principle of TIJ

Generally, in order to form vapor bubbles in liquid and boil the liquid, it is necessary for the liquid to be in a superheated state. It is known that when existing air bubbles, such as vapor trapped in pits, and so on, on a heat-transfer surface that

3.4 Research on Rapid Boiling and Principle of TIJ

maintain the thermal equilibrium with the liquid phase, serve as the nuclei, the superheating temperature is several degree Celsius or dozens of degree Celsius. In such a case, the process by which air bubbles are formed is based on activation of preexisting nuclei, and when the preexisting nuclei within the system satisfy the conditions for activation, the bubbles start to grow. On the other hand, in cases where such kinds of bubble nuclei do not already exist, and when nuclei are generated by liquid molecules through heat perturbation under high superheating temperatures, explosive boiling is known to occur. The superheating temperature in such a case is extremely high compared to when there are preexisting nuclei, and since this serves as a high-speed phenomenon associated with rapid vaporization, it corresponds industrially to the sudden boiling phenomenon, vapor explosion, and so on. In such cases, the process by which bubbles are formed is referred to as spontaneous nucleation, and bubbles are formed through a statistical process such as fluctuations in their own density. Furthermore, spontaneous nucleation is categorized into homogeneous nucleation, in which bubble nuclei are generated in the mother liquid, and heterogeneous nucleation, in which bubble nuclei are formed on heat-transfer surfaces and the surfaces of boundaries with other fluids that do not become mixed. Incidentally, ink is jetted by the volume expansion of the bubbles that are formed by the boiling of the ink due to the electrical pulse heating of the heater in TIJ as a driving source; dots are arranged on the paper surface and an image is formed. At this time, since the required size of the heater is approximately the same as the value of the ink drop volume on two-dimension conversion, the TIJ heater is an efficient actuator regardless of its extremely small size in comparison to the piezoelectric method. The bubble formation phenomenon in TIJ is thought to be an extremely explosive, instantaneous vaporization phenomenon and accordingly, it is considered that the boiling phenomenon, which serves as the operating principle of TIJ, is based on spontaneous nucleation and further thought to be categorized as formation of heterogeneous nuclei that exist in the heat-transfer surface and ink interface. An example of experimental research conducted in the past relating to boiling based on the rapid heating method is the research conducted by Skripov et al. [3–7]. Skripov et al. used a thin platinum wire with a diameter of 20 μm as a heater, carried out rapid heating at a maximum heating rate of 107 K s−1 , and measured the temperature at which boiling started for several types of organic fluids and water. In addition, Derewnicki [8] used a thin platinum wire with a diameter of 25 μm, heated water at a maximum heating rate of 107 K s−1 , made the pressure fluctuate between 0.1 and 1.0 MPa, and measured the temperature on the heat-transfer surface and temporal changes in the thermal flux. Using liquefied gas, Sinha et al. [9] reports that the temperature at which changes in the temperature curve of the heat-transfer surface is nearly the same as that for homogeneous nucleation occurs using thin platinum wires and thin copper wires. In this series of experiments using thin platinum wires, and so on, which had been conducted until now, a heating rate of up to 107 K s−1 had been used. By

45

46

3 Thermal Inkjet

lida, Okuyama et al., however, a series of studies have been carried out based on a rapid heating method that realizes a heating rate of 107 K s−1 or greater than those used in the past by approximately 1 power. A tiny heat-transfer surface of 0.1 mm × 0.25 mm was produced and observations of the boiling state were made using organic fluid and water, to provide reports on boiling behavior and explanations based on the theory of nucleation. In addition, comparisons were made with theoretical values for the boiling inception temperature that is indicated based on the theory of homogeneous nucleation, and results that demonstrate a good match for organic fluids, and a temperature approximately 19 ◦ C lower than the 312.5 ◦ C presented by Skripov for water were obtained. The differences between the theoretical and measured values are suggested as being due to wettability [10, 11]. In experimental studies using TIJ devices, the phenomenon that occurs with TIJ heads is observed [12], and has demonstrated the superheated fluid layer vaporization theory. Bubble formation and growth, which is a primordial issue in boiling, are based on energy stored in the superheated fluid layer adjacent to the heat-transfer surface [13]. Meanwhile, in using the boiling phenomenon in inkjets, Asai et al., who represent the founding group of Canon, have carried out the following study. A heater, 50 × 50 μm2 in size, was made of HfB2 . The observation of the bubble phase was carried out after pulse heating for several tens of microseconds. The boiling inception temperature was calculated with the superheating limit and flow channel geometry. The growth of bubbles and the inkjetting state were studied [14]. They deemed that there is a correlation between fluctuations in the inkjetting velocity and the state of formation of bubbles, categorized boiling into a film state and nuclear boiling, and suggested that each is equivalent to spontaneous nucleation and to foaming from a rough surface at a slow heating rate, respectively [15]. The effect of the dissolved air on the nucleation probability was investigated. At the saturated mole fraction of air in water at room temperature, it was found that the incipient boiling time and temperature are only 0.01 μs and 0.4 K smaller. In cases where the wettability of the heat-transfer surface becomes bad, the boiling temperature decreases and the pressure obtained from boiling decreases; therefore, the wettability of the heater surface should be well maintained [16]. In his research, what Asai [17] focuses on from the perspective of understanding the boiling phenomenon is to promote the analysis of bubble behavior, such as by solving the lifetime of bubbles by taking heat flux into consideration, based on Skripov’s findings, and comparing the measurement results of actual bubble behavior. In his article, Asai presents research results that are experimentally superior as well, such as by observing bubble formation from horizontal directions, rather than being limited to the theoretical considerations mentioned above. Asai carried out activities related to TIJ until 1992; after announcing studies using the Navier–Storks equations to calculate the behavior of ink fluid by making calculations and experimental comparisons, taking the viscosity and surface tension of ink into consideration, and applying these to flow channel design [18], Asai has not carried out any prominent activities.

3.5 Inkjetting Mechanism of TIJ

Although Canon is deemed as being the inventor of TIJ, HP, which is said to have started technological developments for TIJ at nearly the same time, is overwhelmingly superior in all markets throughout the world. However, there are not many publications by HP regarding their research activities. Although it can be perceived that their corporate culture is one that places emphasis on industrial activities rather than research activities, the considerations of Allen et al. [19] that were described in the HP Journal in 1985 are full of suggestions as an article that relates to boiling in the dawn of TIJ technology, as it includes observations such that the boiling phenomenon of TIJ cannot be explained with conventional boiling, but rather with the fluctuation of molecules based on superheating, that the surface boundary between the heater and ink is heterogeneous, and that selective boiling from cavities on the heater surface is likely.

3.5 Inkjetting Mechanism of TIJ

Figure 3.4 shows an outline of transitional changes in the behavior of heat and pressure generated from pulse superheating. In this figure, the heater is driven by pulse waveforms and the temperature of the heat-transfer surface rises sharply up to more than 300 ◦ C; together with the generation of large pressure, bubbles start to form and grow, and afterwards, the grown bubbles shrink. When the bubble volume becomes 0, a large pressure is generated once again, and the action of the bubbles at this time is explained as follows. Boiling incipience

5

Temparature [102 °C] Volume [105 μm3] Pressure [MPa]

Electrical power [W]

6

4 3 2 1

−5

0 0

5

10

15

Time (μs) Temperature

Bubble volume

Heating pulse

Pressure at heater

Figure 3.4 TIJ jetting mechanism in relation with pulse heating, temperature, pressure, and bubble volume.

47

48

3 Thermal Inkjet

Owing to rapid heating, the superheating limit of the fluid (water) was reached, resulting in the internal pressure of the bubbles achieving several megapascals, acting on the entire heat-transfer surface as an impulse force [17]. The bubbles start growing based on this force and continue expanding due to inertia, thus resulting in the inkjetting from the nozzle. The internal pressure of the bubbles is of a vacuum state, falling below atmospheric pressure; because of the difference with the external atmospheric pressure and fluid resistance, the bubbles stop their movement and start collapsing. When the bubbles disappear, the high-speed gas–liquid interface of the ink collides with the heat-transfer surface and impulse force starts acting once again. This is the cavitation [20] phenomenon, which occurs in places where bubbles disappear, or in other words, in the central part of the heat-transfer surface. This phenomenon repeatedly acts on the heat-transfer surface, causes mechanical deterioration and leads to heater destruction.

3.6 Basic Jetting Behavior of TIJ

Basic jetting characteristics of TIJ consist of (i) the momentum of jetted ink in relation with the energy input into the heater, (ii) refilling speed of ink after jetting, or in other words, the repeated frequency response, and (iii) environmental dependence, or temperature dependence. There are also various other evaluation items, such as the ink-drying speed around nozzles in relation to the temperature and humidity, the dependence on the number of ejected jets, the directionality of the jets, the ink-retention performance in relation to back pressure (water head height) of the ink, variability in manufacturing, lifetime, the removal of the bubbles that exist in the ink channel, and so on. This chapter describes (i), (ii), and (iii) mentioned above as representative characteristics. 3.6.1 Input Power Characteristics

Figure 3.5 shows the results of measuring the ink drop volume and ink drop velocity in relation to the power input into the heater. The general characteristic for TIJ is that after jetting starts from the low-power side, both the ink drop volume and velocity increase suddenly, and then saturation occurs. The shorter pulse width but the higher heating pulse, which enables higher heating speed, contributes to the sudden rise with respect to drop volume and velocity just after the commencement of jetting; in other words, the heating speed plays a particularly significant role in the change rates of drop volume and velocity. Normally, with piezoelectric jets and thermal printers, the printing dot diameter changes as it is dependent on the input energy, but with TIJ, the dot diameter does not change because of the above-mentioned characteristic. This saturation

3.6 Basic Jetting Behavior of TIJ Figure 3.5 Jetting characteristics on applied power (a) drop volume and (b) drop velocity.

60

Drop volume (pl)

50 40 30 20 10 (a)

0 16

Drop velocity (ms−1)

14 12 10 8 6 4 2 0 3.2 (b)

3.6 4.0 Applied power (W)

4.4

characteristic is a favorable characteristic that indicates that there are small changes in relation with noise and variation. On the other hand, in order to change the dot size and achieve gradation, naturally, adjustment of the input power for bubble generation is not suitable. 3.6.2 Frequency Characteristics

Figure 3.6 shows the changes in ink drop volume and velocity in relation to changes in frequency. In accordance with an increase in frequency, the ink drop volume increases slightly outside of the stable range, and then starts decreasing. In the same manner, the ink drop velocity increases. The slight increase in ink drop volume immediately before it starts decreasing is considered as being due to the acoustic resonance and structural vibration caused by the repeated inkjetting. In addition, the increase in ink drop velocity is caused by the elongation of the tip of the jet. 3.6.3 Dependency on Temperature

Figure 3.7 shows the results of measuring the jetted ink drop volume based on heating the base onto which the printheads are installed and increasing the ink temperature together with that of the printheads as a whole. The ink drop

49

3 Thermal Inkjet

60

Drop volume (pl)

50 40 30 20 10 (a)

0

Drop velocity (ms−1)

20 16 12 8 4 0

0

(b)

2

4

6

8

10

12

14

Driving frequency (kHz)

Figure 3.6

Frequency characteristics (a) drop volume and (b) drop velocity.

60

Drop volume (pl)

50

55

50

45

40 25

Figure 3.7

35 45 Head temperature (°C)

55

Temperature dependence of drop volume.

volume in TIJ possesses the attribute of increasing at a constant rate in relation to temperature. Accordingly, adjustment of the ink drop volume for the modulation of dot diameter as mentioned above is possible through temperature control. There are multiple pulses for doing so; in other words, by preheating the ink with a prepulse,

3.7 TIJ Behavior Analysis Using Simulation

it is possible to adjust the ink drop volume, and to maintain the ink drop volume on the high-temperature side even if the environmental temperature changes.

3.7 TIJ Behavior Analysis Using Simulation

In actuality, designing printheads is carried out based on a regressive approach where heads are fabricated and evaluated. However, changes in the flow path dimensions and reviews of common differences, or in other words, when studying the effects on jetting when design values are intentionally shifted, simulation is an extremely efficient tool in terms of reducing time and costs. In TIJ, it is natural that focus is placed on heat and fluid analysis, and the following tools are generally used in simulation as well. 3.7.1 Cylindrical Thermal Propagating Calculation Based on the Finite Element Method (Software Name: Ansys)

In order to analyze the design of heater dimensions, layer configuration, heat efficiency, rise in temperature of chips as a whole, and so on, in TIJ, it is effective to carry out heat simulation using Ansys. In this simulation, the temperature distribution T (z,t) can be obtained by solving the basic equation for heat conduction below, with the heat density that is generated in the unit of time as Q (z,t). λ 1 ∂T (z, t) − T (z, t) = Q (z, t) ∂t cρ cρ Here, T = temperature, t = time, λ = heat conductivity, c = specific heat, and p = density. The right side of the equation indicates the heater as a heat source, and when calculating the temperature of parts other than the heater, temperature is determined at each level based on using 0 for the value of the right side of the equation. Figure 3.8 shows a model for heat calculations. The layers that are subject to heat calculation consist mainly of the heater, the electrical insulation layer in the direction of the fluid side above the heater (z direction), protective layer tantalum (Ta), ink, and thermal resistance layer in the direction below the heater and silicon (Si) substrate below. As boundary conditions, a spatial expansion to an extent where the heat that is generated by the heater does not reach the boundary within the calculated time period is set, serving as an isothermal boundary. 3.7.2 Fluidic Free Boundary Calculation Based on the Finite Differentiation Method (Software name: Flow3D)

With regard to jetting of ink based on the formation and growth of bubbles, it is efficient to conduct fluid simulation using Flow3D in order to analyze the form

51

52

3 Thermal Inkjet

z

Ink Protective layer (Ta) Electrical insulator Heater

Fluid side r

Thermal resistance layer

Below heater

Si substrate

Figure 3.8

Simulation model for thermal propagation (Ansys).

design for the flow path including the nozzle diameter, the ink drop jetting volume, jetting velocity, and other ink fluids. Flow3D is a product developed at the Los Alamos National Laboratory by Flow Science, which was established by C.W. Hirt, who proposed the volume of fluid (VOF) method [21]. The VOF is a method that predicts dynamic behavior of the air–liquid interface. The basic formula in fluid analysis is a combination of the use of Euler’s continued formula based on the conservation of mass as shown below: ∂ρ + div ρv = 0 ∂t

(ρ = density, t = time, v = flow velocity vector)

and the Navier–Stokes equation as follows:       1 ∂v  + v · grad v = −grad P + ρF + μv + η + μ grad div v ρ ∂t 3 

Here, P = pressure, F = external force per unit mass, μ = shear viscosity, and η = volume viscosity; by incorporating the VOF method, it is possible to calculate the ejection of ink drops in air. The conditions for numerical analysis of ink drop behavior based on the finite difference method are indicated in Figure 3.9, where a flow path, nozzle, and atmospheric region into which ink drops are ejected and set up on a two-dimensional orthogonal mesh, with the free entrance/exit boundary on the atmospheric side, the constant pressure boundary on the ink entrance side of the flow path, and rigid wall boundaries for other sides. The size of the atmospheric region is determined in accordance with the jetting area of the ink drops to be analyzed. As initial conditions, the vapor bubbles on the heater surface are presumed as having foamed based on the Clapeyron–Clausius equation under a certain temperature, vapor pressure, and volume.

3.8 Issues with Reliability in TIJ

Boundary for constant pressure

Air region

Jet Boundary for free flow in and out Nozzle

Bubble

Figure 3.9 Simulation model for inkjetting (Flow 3D).

3.8 Issues with Reliability in TIJ

Ikeda [22] raise the following seven points as issues of reliability with TIJ printer heads. 1) 2) 3) 4) 5) 6) 7)

Heater breakdown due to cavitation Heater breakdown due to heat stress Ink scorching onto heater surface (kogation) Residual air/gas bubbles inside nozzle Electrode breakdown caused by heat storage and ink corrosion Clogging of nozzle by ink Jetting failure due to ink viscosity increase.

Ikeda et al. classified (1), (2), (3), and (5) as hard failures, or failures that cannot be recovered, and (4), (6), and (7) as soft failures that can be recovered through maintenance, and so on. From among these issues, for (1), (2), and (5), major measures consist of improvements in printhead manufacturing technology, and lifetime has progressed dramatically to this date. Maintenance techniques and technology to prevent drying of the ink have also developed, and similarly, improvements for (4), (6), and (7) are also advancing. In particular, a unique problem for TIJ is (3), or ‘‘kogation,’’ which leads to the degradation of printing quality. Kogation is based on the Japanese term ‘‘koge,’’ or ‘‘scorch,’’ and represents a term used by Canon, HP researchers, and Chang [23] that has become universalized. In other words, this is a problem where, since a rapid boiling phenomenon for ink that occurs based on conducting pulse heating of the heater is used in TIJ, residue accumulates on the heat-transfer surface due to thermal decomposition of ink components associated with repeated heating of high-frequency cycles, the residue becomes scorched, as shown in Figure 3.10, and the jetting force declines, thus inducing degradation in image quality as shown in Figure 3.11.

53

54

3 Thermal Inkjet

10.0 kV

X 20.0 K

1.50 μm

Figure 3.10

Example of ink residue scorched onto heater.

Figure 3.11

Print quality degradation due to kogation phenomenon.

In TIJ, since heat generation based on a heater is the basic operating principle, heating ink, which is a recording fluid, is a fundamental action that cannot be avoided, and kogation can be considered a primordial issue peculiar to TIJ. This issue has been dealt with through various means, such as selecting and adjusting ink components, making the head and ink together to be disposable by replacing them before the issue arises as seen with earlier TIJ products. However, the decrease in the degree of freedom when selecting materials as a recording fluid and the increase in costs resulting from purification of ink are disadvantageous in price competition, and single use based on replacements is not favorable from the perspective of consideration to the global environment. In response to this issue, Morita et al. [24] provide technology for longer lifetime based on applying a recovery pulse for removing kogation residue.

3.9 Present and Future Evolution in TIJ Technology

With respect to the competition for higher image quality based on smaller ink drops, as of 2002, Epson had achieved 1.8 pl and Canon had achieved 2 pl. In the latter half of the 1990s, when inkjets became popular, making ink drops smaller for the purpose of higher image quality progressed rapidly, but stagnated after falling down to approximately 2 pl. The leading explanation is that making drops

References

even smaller became unnecessary, as the dot diameter had reached the resolution capability of the human eye. In such a way, higher image quality using smaller drops, special paper for inkjets, diluted and RGB ink, and so on, faster speeds by full-width array of the heads, and low costs based on the evolution of IC technology were all realized by TIJ. As the next stage, studies on energy saving such as based on the elimination of a protective layer [25] had been carried out in opposition to piezoelectric technology, but TIJ technology remained limited to the area of applied research as represented by the release of TIJ-MFP by HP in 2007, and came to a technical stagnation. However, in 2007, Australia’s Silverbrook made a sensational announcement regarding TIJ [2]. This was the announcement of an ultracompact, high-speed, 1600 dpi printer using integrated heads of five colors. The world was first astounded at its structure in which the heater floats in the ink, and was further amazed with the fact that the width of the head devices that print in all colors is 0), the sum of P and 12 ε0 ·E 2 needs to exceed 4γ /dN . Therefore, the minimum voltage for initiation of jetting can be approximated from Eq. (4.1). As the applied voltage raises gradually above the minimum voltage required to initiate jetting, the frequency of jet pulsation (f ) increases, and can reach the kilohertz regime. This pulsation is related to the capillary wave caused by the imbalance between the surface tension force and electrical force on the surface of the charged meniscus apex. The relationship between f and E under constant P conditions can be approximated by the following scaling law [15, 19]:  3 1.5 4 ε0 E (4.2) f ∝ 0.75 2 ρ γ dN where ρ is the density. Also, f can be affected byP.f in the kilohertz range can increase further with P under constant E. However, the increase in P also causes widening the diameter of the jet and a corresponding increase in the droplet size [15]. The aspects described in this section illuminate some of the underlying physics and also suggest engineering routes to improve the performance. For optimized resolution and printing speed, it is ideal to use high-frequency pulsations. Low-pressure-driven flow is preferred to achieve high printing resolutions, even though such arrangements require high electric fields to maintain required pulsation frequencies.

4.4 Drop-on-Demand Mode Printing

In the previous section, it was seen that the frequency of jet pulsation (f) is controlled simply by adjusting the magnitude of a fixed, DC voltage. In this case, the distances between neighboring printed droplets are determined by the substrate moving speed and f as the nozzle scans over the substrate. Placement of each individual droplet is not actively controlled [11]. To solve this challenge, the driving voltage can be turned on and off using pulsed signals, instead of the constant DC voltages, so that ejection of each individual droplet is under direct computer control. This drop-on-demand printing mode can improve the droplet positioning. The development of this type of printing mode without retarding f is a key engineering challenge. For frequencies in the kilohertz range, a brief, high-voltage pulse superimposed on a fixed baseline voltage can be used [14], as illustrated in Figure 4.4. The high pulse-peak voltage induces droplet ejection, while the low baseline voltage holds the charge in the meniscus for accelerating the overall periods of charge accumulation, cone formation, and relaxation. Here, this baseline voltage needs to be low enough not to eject a jet without the pulse-peak voltage, and the pulse-peak voltage must be high enough to ensure that the ejection occurs within the short temporal duration of the pulse.

4.4 Drop-on-Demand Mode Printing

Voltage Pulse spacing

Pulse width

Pulse peak voltage

Baseline voltage Time Figure 4.4 Schematic time plot of the voltage profile for high-frequency drop-on-demand electrohydrodyanmic inkjet.

In this drop-on-demand mode, the distance (sd ) between the printed droplets can be adjusted by pulse spacing and the substrate movement speed; sd = (pulse spacing) × (substrate movement speed). In addition, the diameter of the printed droplet can be controlled using pulse width as well as the nozzle size and the pressure drop. By increasing the pulse width, several jets in a row can be ejected within the width of the single pulse. In this case, multiple droplets printed on a substrate combine to create a larger droplet. Hence, the final droplet diameter is controlled by the number of the droplets released per the high-voltage pulse peak and consequently, the pulse duration. As an example, Figure 4.5 shows an optical

100 μm

20 μm

Figure 4.5 Printed pattern using a 5 μm internal diameter nozzle, with on-the-fly droplet control by changing pulse width. UV-curable polyurethane prepolymer was used as an ink, and droplet diameters are ∼4 and ∼8 μm

with constant center-to-center distance of ∼16 μm. (Reprinted with permission from Ref. [14] Mishra S. et al. (2010) J. Micromech. Microeng., 20 (9), 095026, Copyright IOP Publishing Group.)

61

62

4 High-Resolution Electrohydrodynamic Inkjet

micrograph of an alphabetic pattern printed in the drop-on-demand mode with two different droplet sizes by adjusting the pulse width (500 μs for ∼4 μm diameter droplet and 2500 μs for ∼8 μm diameter droplet) under constant pulse spacing. In this way, the printing resolution in the local area of a pattern can be tuned on-the-fly without changing pressure and nozzle sizes.

4.5 Versatility of Printable Materials and Resolutions

Mobile charges must be present in inks in order for them to be printable using electrohydrodyanmic flow effects. In practice, even tiny charge concentrations are enough to induce this ejection process. For example, ejection is possible with inks that have electric conductivities that span 10 orders [20], from 10−13 to 10−3 S m−1 . This versatility enables various functional organic and inorganic inks, such as suspensions of solid nanomaterials, to be used with the electrohydrodynamic inkjet method [10–12]. Figure 4.6a,b presents dot matrix alphabetic letters printed using an aqueous suspension of poly(3,4-ethylenedioxythiophene):poly(styrene-sulfonate) (PEDOT:PSS) conducting polymer and a dielectric polyurethane prepolymer, respectively. Suspensions of a Si nanoparticles PEDOT/PSS ink

SWNT ink

200 μm

(a) Polyurethane

Si nanoparticles (c) Si rods

(b)

100 μm

(d)

50 μm

Aligned SWNTs

(e)

Alignment direction

10 μm

(f)

Figure 4.6 Optical micrographs and scanning electron microscopy (SEM) images of various images printed with different inks. (a) Letters printed with the PEDOT:PSS conducting polymer. The average dot diameter is 10 μm. (b) Letters printed with a photocurable polyurethane polymer with dot diameters of 10 μm. (c) Fluorescence micrograph of Si nanoparticles (average particle diameter of 3 nm) that emit light at 680 nm. The diameter of the printed dots is ∼4 μm.

500 μm

100 μm

(d) Optical micrograph of single-crystal Si rods printed from a suspension in 1-octanol. (e) SEM image of aligned SWNTs grown by CVD on an ST-cut quartz wafer using printed patterns of ferritin as a catalyst. (f ) Image of a flower printed with dots (∼8 μm diameters) of SWNTs from an aqueous suspension. (Reprinted with permission from Ref. [10] Park, J.-U. et al. (2007) Nature Mater., 6 (10), 782–789, Copyright Nature Publishing Group.)

4.5 Versatility of Printable Materials and Resolutions

(average diameter: 3 nm) and a single-crystal Si rod (length: 50 μm, width: 2 μm, and thickness: 3 μm) in 1-octanol can be printed, as shown in Figure 4.6c,d. In addition, single-walled carbon nanotubes (SWNTs) aligned in parallel on an annealed ST-cut quartz substrate can be synthesized using the chemical vapor deposition (CVD) after printing an aqueous suspension of ferritin catalyst using the inkjet (Figure 4.6e). Instead of printing the catalyst, SWNTs can be suspended in water using surfactants as an ink, and this SWNT ink can be inkjet printed, as presented in Figure 4.6f. For the results in Figure 4.6, 30 μm inner diameter nozzles were used and yielded dot diameters of approximately 10 μm. The printing resolution can be controlled by nozzle sizes, and it can extend into the submicron range using small inner diameter nozzles [10, 11]. For example, Figure 4.7a presents an optical micrograph of the portrait printed using dots of Polyurethane dots (diameter, 490 nm)

100 μm

(a)

2 μm

DNA dots (diameter, ~100 nm) 2 μm 30 nm 4 nm 98 nm

(b)

1μ

m

m 2μ 0

1 μm

Figure 4.7 High-resolution electrohydrodynamic inkjet printing. (a) Optical micrograph of an image printed using polyurethane ink with dots of ∼490 nm in diameter. (Reprinted with permission from Ref. [10] Park J.-U. et al. (2007) Nature Mater., 6 (10), 782–789, Copyright Nature Publishing

2 μm

Group.) (b) Atomic force microscopy (AFM) images of ∼100 nm diameter dots printed using an aqueous suspension of DNA (37 mer, single stranded). (Reprinted with permission from Ref. [11] Park J.-U. et al. (2008) Nano Lett., 8 (12), 4210–4216, Copyright American Chemistry Society.)

63

64

4 High-Resolution Electrohydrodynamic Inkjet

∼490 nm in diameter with a 500 nm inner diameter nozzle and polyurethane ink. The dot size can decrease further to ∼100 nm using a 300 nm inner diameter nozzle, as shown in Figure 4.7b. These resolutions, which approach the nanoscale regime, significantly exceed conventional thermal/piezoelectric inkjet systems.

4.6 Applications in Electronics and Biotechnology

The extremely high resolution and compatibility with diverse inks make electrohydrodynamic inkjet attractive for printed electronics. Key device elements including source, drain, and gate electrodes for transistors and interconnect lines, all of which can be inkjet printed with complex designs. As an example, Figure 4.8a presents an electrode pattern of a ring oscillator circuit [10]. In this case, a polymer etch resist served as an ink, printed on predeposited metal layers. After etching the metal areas not protected by the resist and removing the polymer, fine electrode patterns with minimum line widths of ∼1 μm were produced (Figure 4.8a–c). Short channel lengths (distances between sources and drains) are important because they enhance the switching speeds and output currents. A minimum channel length of ∼900 nm was achieved, as shown in Figure 4.8c, with direct printing, thereby avoiding the topological wetting assist features that are required with conventional inkjet techniques. This scale is comparable to the channel length (∼2 μm) of amorphous-silicon thin-film transistors employed in commercial active-matrix liquid crystal displays fabricated using large-area photolithography [13, 21]. Printing such fine electrode dimensions using conventional inkjet printers would be difficult or impossible. Unlike printing of etch resist, which requires separate steps for metal deposition, etching, and resist removal, inks that consist of suspensions of metal nanoparticles can yield patterned electrode features, directly. A simple heating step fuses the printed nanoparticles into homogeneous metallic lines. The high surface-to-volume ratio of nanoparticles significantly reduces their melting temperatures, and hence even flexible plastics can be used as substrates (e.g., the melting temperature of 3 nm diameter Ag nanoparticles: ∼130 ◦ C). Ag source/drain electrodes with minimum line width of ∼2 μm and channel length of ∼1 μm were formed on a flexible polyethylene naphthalate (PEN) substrate in this way [13]. Flexible transistor arrays can be fabricated using electrohydrodynamic inkjet printing of electrodes, as demonstrated in Figure 4.9 [10]. In this example, the aligned SWNT arrays were used for the transistor channel, and source/drains were printed with different channel lengths (L). Figure 4.9a,b shows a schematic illustration of the device structure and an SEM image of aligned SWNTs with the patterned source/drain, respectively. The density of the tube array is ∼2.5 SWNTs per 10 μm. Figure 4.9c presents typical ID – VG curves of the devices. The ID output increases with 1/L, and Ion /Ioff ratios are in the range between ∼1.5 and ∼4.5. Semiconducting tubes coexist with metallic tubes in the aligned SWNT-array channel area, and this metallic tube population results in the low Ion /Ioff values.

4.6 Applications in Electronics and Biotechnology

(a)

10 μm

100 μm

300 μm

(b)

10 μm

130 nm

(c)

5 μm

Figure 4.8 Electrode geometries for a ring oscillator and isolated transistors patterned by electrohydrodynamic inkjet printing. (a) The printed electrode pattern for a ring oscillator circuit. The insets show magnified images. (b) Array of source/drain electrode pairs for transistors. The inset shows an

910 nm electrode pair separated by ∼1 μm channel length of a transistor. (c) AFM image and depth profile of a portion of this pair. (Reprinted with permission from Ref. [10] Park J.-U. et al. (2007) Nature Mater., 6 (10), 782–789, Copyright Nature Publishing Group.)

However, the low ratios can be enhanced beyond 104 by an electrical breakdown process to remove metallic tubes in the channel selectively [10, 22]. Mobilities of the devices in the linear regime are between 20 and 141 cm2 V−1 s−1 with L in the range of 1 ∼ 42 μm (Figure 4.9d), and reduce with L due to the contact resistance between SWNTs and source/drain electrodes. (Here, an accurate capacitance model that considers the electrostatic capacitance coupling of the gate to the SWNT arrays were used for the mobility calculation [10, 22].) The transistors built on a flexible

65

−20 (c)

0

50

100

(a)

Gate

Dielectric

−10

0

100 10 1

10

20

10 20 30 40 L (μm)

Flexible plastic

Aligned SWNTs

0 VG (V)

−ID (μA)

Drain

(d)

0

40

80

120

160

10

20 30 L (μm)

(b)

40 (e)

20 μm

3 mm

Figure 4.9 Fabrication of aligned SWNT transistors on a plastic substrate. (a) Schematic illustration of the transistor layout, where the source/drain are patterned by electrohydrodynamic inkjet printing. (b) SEM image of the aligned SWNTs connected by source/drain electrodes. (c) ID –VG curves measured from transistors with channel lengths, L = 1, 6, 12, 22, and 42 μm, from top to bottom, and channel widths, W = 80 μm at a source/drain voltage, VD = −0.5 V. The inset shows on and off currents (black and gray lines, respectively) as a function of L. (d) Linear regime device mobilities (μdev ), as a function of L. (e) Photograph of an array of flexible, SWNT transistors. (Reprinted with permission from Ref. [10] Park J.-U. et al. (2007) Nature Mater., 6 (10), 782–789, Copyright Nature Publishing Group.)

−ID (μA)

Source

μ dev (cm2 Vs−1)

66

4 High-Resolution Electrohydrodynamic Inkjet

4.6 Applications in Electronics and Biotechnology

substrate of polyimide shown in Figure 4.9e can operate in bent conditions. No significant change in the mobility and Ion /Ioff ratio is detected for bending to radii of curvature as small as 2 mm (bending-induced strain: ∼0.6%). Another area of application is in biotechnology. For example, microarrays for biochips, which present surfaces that have multiple probe sites (e.g., DNAs and proteins) where each site can bear specific reagent, are of interest. Binding of a complementary molecule to the reagent generates a signal that can be detected by an imaging technique such as fluorescence. To perform massively parallel analyses of complex biological samples, increasing the number of probe sites per unit area is important. Thus, technologies that can print small dots at small spacings are of interest. In this way, the microarray can benefit from the high-resolution capability of electrohydrodynamic inkjet printing. As an example, Figure 4.10a presents 14 × 14 microarrays of DNA (single-stranded, and fluorescently labeled with Alexa546 dye (emission wavelength: 573 nm)) printed using a 2 μm inner diameter nozzle in drop-on-demand mode [11]. The dot diameter and spacing are 2 and 5 μm, relatively. This DNA dot size is approximately 10 times smaller than that produced using conventional thermal/piezoelectric inkjet system. Even at these small probe-site dimensions, the fluorescence intensity is readily detectable with standard fluorescence microscope. As demonstrated in Figure 4.7b, the diameters of the DNA dots can be substantially decreased down to ∼100 nm, by reducing the nozzle diameters. Although these nanometer-scale dots have limited utility because the associated fluorescence signal from a single dot is too small to be detected effectively using a conventional microscope, such printing resolution can be important for applications such as nanomaterials assembly and fundamental study of molecular interactions. One widely used application of DNA microarrays is to recognize base sequences of unknown strands for genomics. Recent advances in synthesizing more functional DNA strands extend their uses to detect much broader range of analytes, including organic small molecules, inorganic ions, cancer cells, or viruses, with high selectivity [23, 24]. Especially, DNA aptamers are an interesting class of functional DNAs that bind to specific molecules and induce conformation changes [24]. A growing number of reports describe synthesizes of new aptamers and their uses for application in biosensors to detect adenosine, cocaine, thrombin, and so on. Figure 4.10b,c demonstrates the use of electrohydrodynamic inkjet to form an adenosine-DNA aptamer microarray, printed into a complex geometry, as a fluorescence biosensor for adenosine [11]. The sensor fabrication process begins with inkjet printing the single-stranded DNA functionalized with the fluorescence dye and biotin. (This biotin group enables the DNA to be immobilized on a streptavidin-coated substrate.) Subsequently, the printed DNA is hybridized with two complementary strands of (i) the DNA aptamer strand that can selectively react with adenosine and (ii) the other strand that contains the quencher label. After this hybridization, the fluorescence becomes quenched and hence dark (left insets in Figure 4.10c). After exposure to an aqueous adenosine solution (5 mM), however, the fluorescence signal increases significantly again by the release of the quencher

67

4 High-Resolution Electrohydrodynamic Inkjet

700 nm

25

μm

10 μm (a)

0

68

Adenosine aptamer

Adenosine Q

Dye

Quencher

Adenosine

Surface adenosine F

Substrate

(b)

Before hybridization

After hybridization

Hybridized

(c)

50 μm

Figure 4.10 Biosensing using DNA microarrays. (a) Fluorescence micrograph (left) and AFM image (right) of DNA microarrays. (b) Schematic representation of the detection of adenosine using the printed aptamer patterns. (c) Fluorescence micrograph of

After adenosine treatment Adenosine

50 μm the alphabetic pattern before (left) and after (right) reaction with adenosine molecules. (Reprinted with permission from Ref. [11] Park J.-U. et al. (2008) Nano Letters, 8 (12), 4210–4216, Copyright American Chemical Society.)

4.7 High-Resolution Printing of Charge

strand (right image in Figure 4.10c). This chip can work as a turn-on sensor for adenosine. The advantages of high-resolution electrohydrodynamic inkjet printing for these applications over the conventional inkjet systems lie mainly in the high levels of resolution that can be achieved. The submicron-scale printed features of various functional materials are suitable for use in not only the demonstrations described here but also in other application areas. Exploring these possibilities and further developing the printing techniques by increasing the number of nozzles for high-speed manufacturing appear to be promising for future work.

4.7 High-Resolution Printing of Charge

An interesting feature of the electrohydrodyanmic inkjet process is that the printed droplets contain an overall net charge. This property can be exploited to produce a charge printer able to form complex patterns of positive and/or negative ionic charge, with high resolution, on any dielectric surface [12]. As an example, Figure 4.11 shows the Kelvin force microscopy (KFM) images, which reveal that the printed dots have positive (white color in the images) or negative potentials (black color), compared to the unprinted substrate areas that have negligible potentials. Switching the direction of the electric field used to initiate jetting reverses the charge polarity of the printed droplets, without significant change in the droplet sizes. Therefore, complex patterns with both polarities can be printed simply by controlling the bias direction, as demonstrated in Figure 4.11a–d. These results show that preexisting patterns of charge have little effect on the printing process. Also, various functional inks with a wide range of physical properties and pH values can be printed in both positive and negative potentials on a single substrate. Figure 4.11a presents patterns of dots with potentials of about +5.5 V and −5.5 V (peak values), using an aqueous sodium phosphate solution (10 mM, pH ∼ 7) as the ink. Figure 4.11b shows an array of lines patterned using the polyurethane (pH ∼ 5). Here, an array of charged lines at −1.3 V was printed first and subsequently another set of lines at +1.3 V. At the crossing points, the negative and positive charges balance one another, thereby reducing the potentials in these regions to values close to 0 V. Figure 4.11c shows a pattern of dots at ∼ ±2.4 V using an organic base, quinoline, as the ink. Charged lines printed using an ink of polyethylene glycol diacrylate (Figure 4.11d) exhibited potentials that scale with multiple printing cycles in the expected way, from ∼ −0.2 V for a single pass to ∼ −1 V for five cycles. Additional cycles can further increase the potentials. Both positive and negative patterns of charge persist for times that depend on environmental factors including humidity. In ambient conditions, the potentials decrease continuously with lateral spreading of charge, due to water molecules in the environments. (The condensed water can neutralize some of the printed charge and facilitate its diffusion on the surface.) By storing the printed charge patterns in an environment with low humidity (e.g., H2 O ∼ 0.3 ppm), nearly complete

69

70

4 High-Resolution Electrohydrodynamic Inkjet

20 μm

(a)

20 μm

(c)

(b)

Number of printing times 5

4

3

2

20 μm

Number of printing times

1

5

4

3

2

1

+1.2 V 0V −1.2 V

20 μm

(d)

0 μm

27 μm

54 μm

(e)

Figure 4.11 Printing and dissipation of positive and negative charges controlled by electric field direction. (a) Potential mode KFM images of aqueous sodium phosphate solution (pH ∼ 7). (b) Polyurethane (pH ∼ 5). (c) Quinoline (pH > 8). (d) Potential control

by printing multiple times with an ink of (polyethylene glycol diacrylate). (Reprinted with permission from Ref. [12] Park J.-U. et al. (2010) Nano Letters, 10 (2), 584–591, Copyright American Chemical Society.)

retention of potentials and sizes in patterns of printed dots is possible over one week [11]. This capability to print charge patterns with controlled polarities and magnitudes can be used for electrostatic doping to manipulate the threshold voltages and on-current levels of transistors [11]. In addition, invisible, printed security codes, guided assembly of charged particles, and modulation of activity in biological systems represent promising application areas.

References 1. Wang, Y. and Bokor, J. (2007)

Ultra-high-resolution monolithic thermal bubble inkjet print head. J. Micro/Nanolith. MEMS MOEMS, 6 (4), 043009. 2. Park, S.K., Kim, Y.-H., and Han, J.-I. (2009) High-resolution patterned nanoparticulate Ag electrodes toward all printed organic thin film transistors. Org. Electron., 10 (6), 1102–1108. 3. Stringer, J. and Derby, B. (2009) Limits to feature size and resolution in ink

jet printing. J. Eur. Ceram. Soc., 29 (5), 913–918. 4. Sanaur, S., Whalley, A., Alameddine, B., Carnes, M., and Nuckolls, C. (2006) Jet-printed electrodes and semiconducting oligomers for elaboration of organic thin-film transistors. Org. Electron., 7, 423–427. 5. Creagh, L.T. and McDonald, M. (2003) Design and performance of inkjet print heads for non-graphic-arts applications. MRS Bull., 28, 807–811.

References 6. Mills, R.N. (1999) in Recent Progress in

7.

8.

9.

10.

11.

12.

13.

14.

Ink Jet Technologies II Chapter 5, (ed. Eschbach, R.) Society for Imaging Science and Technology, Washington, pp. 286–290, ISBN / ISSN: 0-89208-220-8. Nakao, H., Murakami, T., Hirahara, S., Nagato, H., and Nomura, Y. (1999) Head design for novel ink-jet printing using electrostatic force. IS&Ts NIP15: International Conference on Digital Printing Technologies, pp. 319–322. Choi, D.H. and Lee, F.C. (1993) Continuous-tone color prints by the electrohydrodynamic ink-jet method. Proceeding of IS&T’s Ninth International Congress on Advances in Non-Impact Printing Technologies, Yokohama, October 4, pp. 33–35. Kawamoto, H., Umezu, S., and Koizumi, R. (2005) Fundamental investigation on electrostatic ink jet phenomena in pin-to-pin discharge system. J. Imaging Sci. Technol., 49 (1), 19–27. Park, J.-U. et al. (2007) High-resolution electrohydrodynamic jet printing. Nat. Mater., 6 (10), 782–789. Park, J.-U., Lee, J.H., Paik, U., Lu, Y., and Rogers, J.A. (2008) Nanoscale patterns of oligonucleotides formed by electrohydrodynamic jet printing with applications in biosensing and nanomaterials assembly. Nano Lett., 8 (12), 4210–4216. Park, J.-U. et al. (2010) Nanoscale, electrified liquid jets for high-resolution printing of charge. Nano Lett., 10 (2), 584–591. Sekitani, T., Noguchi, Y., Zschieschang, U., Klauk, H., and Someya, T. (2008) Organic transistors manufactured using inkjet technology with subfemtoliter accuracy. Proc. Natl. Acad. Sci. U.S.A., 105 (13), 4976–4980. Mishra, S., Barton, K.L., Alleyne, A.G., Ferreira, P.M., and Rogers, J.A. (2010) High-speed and drop-on-demand

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

printing with a pulsed electrohydrodynamic jet. J. Micromech. Microeng., 20 (9), 095026. Choi, H.K., Park, J.-U., Park, O.O., Ferreira, P.M., and Rogers, J.A. (2008) Scaling laws for jet pulsations associated with high resolution electrohydrodynamic printing. Appl. Phys. Lett., 92 (12), 123109. Chen, C.H., Saville, D.A., and Aksay, I.A. (2006) Scaling laws for pulsed electrohydrodynamic drop formation. Appl. Phys. Lett., 89 (12), 124103. Eyring, C.F., Mackeown, S.S., and Millikan, R.A. (1928) Field currents from points. Phys. Rev., 31 (5), 900–909. Marginean, I., Nemes, P., and Vertes, A. (2006) Order-chaos-order transitions in electrosprays: the electrified dripping faucet. Phys. Rev. Lett., 97 (6), 064502. Marginean, I., Nemes, P., Parvin, I., and Vertes, A. (2006) How much charge is there on a pulsating Taylor cone? Appl. Phys. Lett., 89 (6), 064104. Jayasinghe, S.N. and Edirisinghe, M.J. (2004) Electric-field driven jetting from dielectric liquid. Appl. Phys. Lett., 85 (18), 4243–4245. Menard, E. et al. (2007) Micro-and nanopatterning techniques for organic electronic and optoelectronic system. Chem. Rev., 107 (4), 1117–1160. Kang, S.J. et al. (2007) High-performance electronics using dense, perfectly aligned arrays of single-walled carbon nanotubes. Nat. Nanotechnol., 2 (4), 230–236. Chen, Y., Liu, H.P., Ye, T., Kim, J., and Mao, C.D. (2007) DNA-directed assembly of single-wall carbon nanotubes. J. Am. Chem. Soc., 129 (28), 8696–8697. Braun, E., Eichen, Y., Sivan, U., and Ben-Yoseph, G. (1998) DNA-templated assembly and electrode attachment of a conducting silver wire. Nature, 391, 775–558.

71

73

5 Cross Talk in Piezo Inkjet Herman Wijshoff

5.1 Introduction

The actuation in piezo inkjet is done with a piezoelectric element, which deforms the channels. Cross talk is the effect that neighboring channels can also be influenced by the actuation of a channel. Cross-talk effects via the structure of the printhead are electrical cross talk, direct cross talk, and pressure-induced cross talk. The latter cross-talk effect goes also through the ink domain and another ink-domain-related cross-talk effect is acoustic cross talk via the ink supply. Furthermore, disturbing resonances in the printhead structure can be excited. Finally, the variations in drop properties can become very large when cross-talk effects interfere with residual vibrations in the ink domain.

5.2 Electrical Cross Talk

Generally, the basis of piezo inkjet printers is the use of a piezoelectric element to convert an electrical driving voltage into a mechanical deformation of an ink chamber, which generates the pressure required for the drop formation from a nozzle. The main discriminator between the piezo inkjet printheads is the used dominating deformation mode of the piezoelectric element, together with the geometry of the ink channels. It is obvious that only the channel to which a driving voltage is supplied should be deformed by the piezo element. However, in many cases some of the neighboring channels are also deformed. When the electric field expands to other channels, an electric cross-talk effect is generated. So, the electric field has to be restricted to just one channel [1]. Other electrical cross-talk effects from, for example, parasitic capacitances in the electronic driving circuit have to be avoided too. Generally, electric cross talk can easily be avoided with a proper design of the actuator and the driving circuit.

Inkjet-based Micromanufacturing, First Edition. Edited by J. G. Korvink, P. J. Smith, and D.-Y. Shin. © 2012 Wiley-VCH Verlag GmbH & Co. KGaA. Published 2012 by Wiley-VCH Verlag GmbH & Co. KGaA.

74

5 Cross Talk in Piezo Inkjet

5.3 Direct Cross Talk

In the shear mode, the strong shear deformation component in piezoelectric materials is used to deform an ink chamber wall. The electric field, the one-direction, is perpendicular to the polarization direction, the three-direction, and the d15 mode is used to deform the channels via a shear mode deformation. This results in a shear deformation parallel to the direction of the polarization, as shown in Figure 5.1 [2]. In this figure, two versions of the shear mode inkjet are shown, the shared-wall principle, where the piezo ceramic is also the channel plate, and the shear mode with the actuator as a separate outer layer, which covers the channel plate. With the shared-wall principle, the channels are diced in the piezo material itself, and covered with a solid deck [3, 4]. The polarization is directed along the height of the channel walls and the electric field is directed across the channel walls. With the shared-wall design of the shear mode actuator, a very high direct cross talk results from the fact that now the walls between two channels are used to deform the channel cross section. The deformation of the channel cross section in both neighboring channels is about 50% of the deformation of the actuated channel and even a counter pulse on the neighboring channels is often used to keep the cross talk below a certain level. It is not possible to actuate two neighboring channels simultaneously. Only one of every three channels can be actuated at one time [5, 6]. With the actuator as outer deforming layer, this problem does not occur and the direct cross-talk effect is therefore much less. Only some electric cross talk is generated because a small part of the electric field is along the polarization direction, which results in extension of the actuator and thus in a deformation of the channel walls [7]. In the bend mode, the bending of a wall of the ink chamber is used to eject a drop [8, 9]. In the first printheads, an outer wall was made of a diaphragm with

P

P E

P E α x

Figure 5.1 The shear mode principle and the calculated deformation in the actuation direction after applying an electric field between the center and the outer electrodes. The two versions used in piezo printheads are shown.

5.3 Direct Cross Talk

a piezo ceramic bonded onto it. The electrical field is applied in the polarization direction of the piezo material. Not only the deformation in the poling direction is used but also the deformation perpendicular to the poling direction. Since the piezo ceramic is bonded onto a passive membrane, the actuator will bend. The bend mode actuator with only one active layer is also called a unimorph actuator [10]. There are several types of bend mode actuators. The bimorph actuator is made with two piezoelectric layers. This actuator can be operated either in a series connection with the poling of both layers in the same direction, or in a parallel connection with opposite poling [11]. A variant is the bimorph actuator with a metal shim to increase the reliability and the mechanical strength. Actuators with a monolithic structure are also called monomorphs [12]. In all these cases, the actuator bends because of a difference in elongation of the different layers. The shape and location of the deformation is more or less similar to the shear mode as outer layer. So, the direct cross talk is therefore also very low with bend mode actuators. Now there is only a bending moment on the channel walls. With soft channel plate materials such as graphite, this can still lead to some direct cross talk. With the push mode, also referred to as bump mode, a piezoelectric element pushes against an ink chamber wall to deform the ink chamber. The electrical field is applied in the poling direction and the deformation is in the same direction or perpendicular to the poling direction. In order to be capable of deforming an ink channel, the piezo elements have to be supported by a substrate, which supplies the reaction force. An important consequence of this is that mechanical stresses will be generated, which decrease the effective piezoelectric behavior of the piezo element. One way to reduce the loss in effective piezoelectricity is to make use of the shape factor [1]. At least one dimension in the direction perpendicular to the height of the piezo element should be much less than the height of the element. Then, the mechanical constraints from the substrate only extend over a similar small part of the piezo element height and the loss in effective piezoelectricity will be much less, as used in several commercial printheads [13, 14]. As an example, an actuator is used with 500 μm high PZT elements on a 1 mm AlOx substrate. The actuator is attached to a channel block with channels at the same resolution as the piezo elements, for example a resolution of 75 channels per inch. A 25 μm tantalum foil covers a 220 μm wide channel. This foil has to be displaced over several tens of nanometers. With the actuated piezo element supported by a substrate, the reaction force of the substrate will be guided to the nonactuated elements. The channels that are not actuated will deform too. This is shown in the first example of Figure 5.2 for a 120 μm high channel, made in brass. The resulting deformation of the neighboring channels is opposite to the deformation, which is necessary for firing a drop. This results in a direct cross-talk effect of almost 50% [1]. The drop speed will be much lower when a neighboring channel is actuated simultaneously. The reaction force of the substrate has to be guided to another part of the printhead to suppress the direct cross-talk effect. A simple way is to use the channel

75

76

5 Cross Talk in Piezo Inkjet

220 μm

338 μm 169 μm 600 μm

500 μm

(a)

(b) Figure 5.2 Front views of a bump mode channel structure and the calculated deformations of the channel structure. The color scale light → dark is 13 nm → −27 nm in the first example (a) and the deformation is magnified 2000 times in the plot of the deformed structure. The reaction force of the substrate is guided to the neighboring elements and this results in an opposite deformation of the neighboring channels. In the second example (b), the calculated deformation of 220 μm wide channels in graphite,

500 μm 169 μm

(c) covered with a 10 μm poly-imide foil, is shown. The color scale corresponds now to a displacement range of 0–60 nm. The reaction force of the substrate is guided to the channel walls and this eliminates the direct cross-talk effect. In the third example (c), the calculated deformation of 220 μm wide channels in graphite is now magnified 10 000 times in the plot. The pressure in the middle channel is 1 bar. The resulting displacement of the foil in the actuation direction is about 2 nm bar−1 .

walls. With 220 μm wide channels, there is enough space for doubling the piezo element resolution. The number of piezo elements is doubled after reducing the piezo element width and spacing from 169 to 84.5 μm. One half of the piezo elements is used as actuators and the other half as supports against the channel walls. This creates a force loop around each actuated channel. The stiffness of the force loop must be much higher than the stiffness of the channel structure, which has to be deformed. The stiffness of a 25 μm tantalum foil is in this example of the same order as the stiffness of 500 μm high piezo elements and the cross-talk effect in the first neighboring channel is still 10% [1]. So, the flexibility of the foil is important. With a 10 μm thin polyimide foil, the stiffness of the foil in the actuation direction is much less and the direct cross-talk effect can be eliminated. This is shown in the second example of Figure 5.2. The actuation of piezo element results in 40 nm displacement of the foil of the actuated channel. The displacements in the neighboring channels are less than 1 nm.

5.4 Pressure-Induced Cross Talk

The actuation of a channel results in pressure waves with an amplitude of at least 1–2 bar. These pressure waves will also deform the structure, especially with a soft channel plate material such as graphite. In the third example in Figure 5.2, the deformation is shown of 220 μm wide channels in graphite, covered with a 25-μm tantalum foil. A positive pressure inside the actuated channel results in a deformation of all channels, which is called the pressure-induced cross-talk effect.

5.4 Pressure-Induced Cross Talk

The deformation of the channels has two main components, an elongation and a bending component. The actuator is pushed away from the channels by the elongation component, which is counteracted by the piezo elements that are connected to the channel walls. The elongation component of the channel deformation results in an enlargement of all channels, thus resulting in a lowering of the channel pressure and the drop speed for all channels. The amount of elongation depends on the stiffness in the actuation direction of the foil between the piezo elements and on the stiffness of the channel walls. A 25 μm tantalum foil has a stiffness comparable to the elongation stiffness of the piezo elements and the channel walls. The channel walls and the piezo elements will be deformed a lot, as shown in Figure 5.2. With a 10 μm poly-imide foil, the stiffness of the foil is much lower. The pressure waves will deform mainly the foil of the channel itself. The deformation of the first neighboring channel is now about 10% of the deformation of the channel, which is under pressure. This still results in a cross-talk effect, which reduces the drop speed in the neighboring channels significantly. For the pressure-induced cross-talk effect, the bending stiffness of the channel walls is important too. The bending stiffness of the piezo elements is negligible. The bending component reduces the cross section of the first neighboring channel and can compensate the effect of the elongation deformation component. However, the lowering in drop speed, resulting from the remaining pressure-induced cross-talk, can still add to a total decrease in drop speeds of several meters per second, when more channels are firing simultaneously [1]. To prevent the deformation of the channel plate, a very hard channel plate material is necessary, for example, silicon [15]. Another option results from the fact that the pressure-induced cross-talk effect has the same dynamics as the pressure waves. The meniscus in the neighboring, not actuated, channel moves approximately in the opposite direction of the meniscus in the actuated channel (Figure 5.3). The meniscus movement in the opposite direction explains the lower drop velocities when more channels are firing simultaneously. The frequency of the meniscus movement is the same as the frequency of the acoustic resonance of the ink channel. The simulations are done with an acoustic model, which takes into account the acoustoelastic interactions, additional to the finite element simulation of the deformation of the printhead structure [16]. The resulting deformations of all channel cross sections Ai can be written as the sum of all contributions of the neighboring channels from the direct cross-talk effect αij and the pressure-induced cross-talk effect βij :   Ai = αij Vj + βij pj A0

(5.1)

j

With this equation as input for the acoustic part, which simulates the wave propagation in narrow channels, the meniscus movement in many channels can be calculated.

77

5 Cross Talk in Piezo Inkjet

8

Measurements

10x neighbor ch.

6 Meniscus speed [m/s]

8.5

Actuated channel

8 Drop speed [m/s]

78

4 2 0 −2

0

50

100

7.5 7 6.5

−4

6

−6 −20

−8 Delay time [μs] (a)

5.5 −10 0

10

20

30

40

50

Delay time [μs] (b)

Figure 5.3 (a) The calculated meniscus speed as function of time in the nozzle of the actuated channel and the neighboring channel. The meniscus speed in the neighboring channel is 10 times enhanced in the picture to show more clearly that this meniscus moves in the opposite direction. This results in a lower drop speed when both channels are actuated simultaneously. (b) The measured effect on drop speed of

a delay time between the reference and the first neighboring channel. The delay time has a large impact on the pressure-induced cross-talk effect. The channel cross section of the 8 mm long pressure channel is 220 × 120 μm, shaped in a graphite channel-block and covered with a 25 μm poly-imide foil. The nozzle has a diameter of 32 μm in a 50 μm electroformed nickel nozzle plate.

So, the meniscus speed in all channels resonates with the channel resonance frequency. With a time delay between the start of the actuation in neighboring channels, we get a large effect on the drop speed as shown in Figure 5.3(b). With a time delay corresponding to a half period of the channel resonance, 11 μs for this 8 mm channel with a resonance frequency of 44 kHz, the meniscus movement in the neighboring channel will be in-phase with the meniscus movement in the actuated channel. Then, even a higher drop velocity will result from the local cross-talk effect. A time delay corresponding to one-fourth of a period of the channel resonance eliminates the effect of local cross talk on drop speed. This can be used to minimize the dot positioning errors on the paper [17]. The positioning errors of the dots in a scanning printer concept are proportional to the differences in the speed of the drops.

5.5 Acoustic Cross Talk

A complete reflection of the pressure waves inside the ink channel at the ink supply or reservoir is very important. Without a complete reflection of the pressure waves, a part of the pressure waves will travel through the supply and the reservoir. The transmitted part of the pressure waves from the actuated channels will enter all

5.5 Acoustic Cross Talk

other channels as a traveling wave from the supply side of the channels, leading to an acoustic cross-talk effect throughout the whole printhead. The pressure waves propagating in the channel direction will be reflected when the characteristic acoustic impedance Z of the channel changes. The acoustic impedance of the channel depends on the size of the channel cross section A and the speed of sound c as ρc (5.2) Z= A with ρ the density of the ink. The speed of sound is influenced by the compliance β of the channel cross section [18]:  1 (5.3) c = c0 1 + ρc02 β The reflection and transmission coefficients at the interface between domain 1 and domain 2 are 2Z2 Z2 − Z1 T= (5.4) R= Z1 + Z 2 Z1 + Z2 When the compliance does not change, the following relationship holds: R=

A1 − A2 A1 + A2

T=

2A1 A1 + A2

(5.5)

One option is to make the acoustic impedance of the supply as low as possible. Since the acoustic impedance is inversely proportional to the cross section of the channel, the supply channel must have much larger dimensions than the pressure channel. A first step is not to use separate supply channels for all individual channels, but one large supply for all channels as shown with a 3D finite element acoustic model of the ink domain, Figure 5.4 at the left. The minimal height of the supply in order to reduce the acoustic cross talk can be calculated with this model. When only a single channel is actuated, a 1 mm high supply is sufficient to get an almost complete reflection. However, when all channels are actuated, a significant part of the pressure wave is transmitted to the supply. Only the difference in height between the pressure channel and the supply can now contribute to a difference in acoustic impedance. For a channel height of 100 μm, the height of the supply must be of the order of 10 mm to reduce the transmitted pressure wave to less than 1% of the incoming wave. This is not always possible, since the dimensions of the printhead must remain within certain limits. Another way to reduce the impedance of the supply is to increase the compliance of the supply. The most simple way to increase the compliance is to cover the supply with a thin foil [19, 20]. The compliant foil will deform when a pressure wave reaches the supply. With a supply of 1 mm height at an angle of 45◦ , the compliant foil covers a gap of 1.4 mm. A 25 μm poly-imide foil deflects 300 nm when a pressure wave with an amplitude of 0.8 bar reaches the supply, Figure 5.4 at the right. This movement can compensate the complete displacement of the acoustic pressure wave, eliminating the acoustic pressure (as would happen with

79

80

5 Cross Talk in Piezo Inkjet

Su p

Connection channel 8 mm

Nozzle 1 mm

ply

Supply Channel block Z

MX

Y Y

X

X MM

Pressure channel

zo

Pie (a)

Substrate

(b)

Figure 5.4 (a) A 3D Ansys model of the ink domain with multiple channels connected via an ink supply slit. At the (b) the channel structure around the supply with the deformation of the compliant foil when a pressure wave reaches the supply, calculated with a 3D Ansys model of a single channel. A 25 μm poly-imide foil covers a gap of 1.4 mm

and bends almost 300 nm when an acoustic pressure wave with an amplitude of 0.8 bar reaches the supply. This results in a complete reflection of the acoustic pressure wave. The ink domain is deleted from the plot to make the movement of the foil clearly visible.

a very large supply). For the behavior of a single channel, a compliant foil has no effect at all, the reflection was already complete. The effect of the compliance is the same as adding extra volume Vc to the volume of the supply channel Vs . With the bulk compliance ρc2 of the liquid in the supply and the compliance defined with respect to the undeformed volume of the supply Vs we get for the added volume: Vc = βρc2 Vs

(5.6)

With a 25 μm poly-imide foil over a gap of 1.5 mm, the compliance is 4 · 10−9 m2 N−1 , and the extra volume from the compliance Vc is six times larger than the volume of the supply Vs itself. Another option is to use a closed boundary. With a very small inlet at the supply, acoustic cross talk can be eliminated, too [21–25]. With an inlet located at the beginning of the pressure channel longer than 10% of the pressure channel length and smaller than 25% of the channel cross section, less than 1% of the pressure waves from the channel are transmitted [1]. The cross-sectional area of the inlet must be smaller or the length must be larger than these values to prevent acoustic cross talk. However, acoustic cross talk can still play a role via the slosh mode. When the inlet channel has a certain minimum acoustic impedance, resonance modes are possible like a Helmholtz resonator as described in [26]. For these modes, the relative large volume of the pressure channel can act as a spring, with the ink in the nozzle as vibrating mass. A parasitic mode is with the ink in the inlet as vibrating mass, which can transmit acoustic pressure wave into the ink supply or reservoir. The inlet inductance must therefore be several times larger than the nozzle inductance to suppress this mode.

5.6 Printhead Resonance

5.6 Printhead Resonance

To generate pressure waves in the ink channels, the speed of the deformation of the printhead must be on the same time scale as the channel acoustics. This requires a driving waveform at frequencies of several tens of kilohertz. However, the fast movement of the piezo elements can also excite resonances in the printhead structure, especially when actuating many piezo elements simultaneously. In Figure 5.5, the response of a meniscus surface on the actuation of all channels at the other side of a two-sided channel block is depicted. The channels at the other side are actuated with a sinusoidal driving waveform with an amplitude of 10% of the nominal driving amplitude of a bump mode printhead [1]. The speed of the meniscus movement in a nozzle in the middle of this printhead is measured with a laser-Doppler equipment. Exciting many channels at one side of the printhead influences the movement of the meniscus surface of a nozzle at the other side of the channel block, especially at frequencies around 35 and 100 kHz and at a frequency range between 130 and 170 kHz. This can only be caused by resonances in the printhead structure and the effect of the actuation will then be spread over the whole printhead. The deformations inside the printhead cannot be measured. Therefore, the dynamic response of the printhead structure has to be simulated with a 3D model of the complete printhead structure. A 3D CAD model of the printhead can be imported into a commercial code such as Ansys as the basis of a finite element model. Because of the very small and thin parts in the structure, a very fine grid would be required and the number of grid elements would be very large. So, a first simplification of the model is to strip all the small parts. It is not likely that these small parts will have any effect. Secondly, the meshing is simplified by replacing small round geometries with rectangular geometries.

Meniscus speed [m/s]

1.4 1.2 1 0.8 0.6 0.4 0.2 0 0

50

100

150 200 250 Frequency [kHz]

Figure 5.5 The meniscus speed as function of the frequency of the harmonic actuation of all channels at the other side of the channel block. The meniscus speed of a nozzle in the middle is measured with a

300

350

400

laser-Doppler. The actuation of many piezo elements results in resonances in the printhead structure, especially at frequencies up to 200 kHz.

81

82

5 Cross Talk in Piezo Inkjet

F is the load vector and m, c, and k are the structural mass, damping, and stiffness matrices. The equation of motion for a displacement vector x is F = m x¨ + c x˙ + kx

(5.7)

However, solving this equation in a full transient analysis with a model of the complete printhead would require a huge computational effort. Therefore, one can start with a modal analysis and perform a mode superposition analysis to get the complete transient behavior. From experience we know that the amount of damping in the printhead structure is much less than the damping in the ink channels. The damping will be neglected in the modal analysis of the printhead structure. For a linear system, free vibrations will be harmonic of the form x = φ i cos(ωi t), where φ i is the shape of the eigenmode at angular frequency ωi . The equation of motion without a load vector, which resolves the eigenmodes, reduces for each mode to (−ωi2 m + k)φ i = 0

(5.8)

The mesh element sizes are maximized with the frequency range as criterion. For most applications, only modes with a frequency less than 250 kHz are expected to have an impact on the channel acoustics and the drop formation. The corresponding minimum wavelengths of the relevant modes are about 1 mm. With the criterion that at least 10 elements are required to capture the shape of the modes, the maximum element size becomes 100 μm. The smallest elements are necessary to describe the smallest details, for example the foil, which has a thickness of 25 μm. The voltage on all electrodes is set to zero in the modal analysis. With the modal analysis, all resonances in the printhead structure are identified. Now we have to calculate which modes are excited the most. First, the deformation S, which results from the electrical loading (the electric field E applied on the piezo elements with piezoelectric coefficients d and compliance s) and the mechanical loading T (the constraints from the substrate and the channel block): S = sT + dE

(5.9)

is translated into a pure mechanical loading to reduce the computational effort further. The deformation as the sum of the mechanical and piezoelectric strain is translated into an elastic stress with T = s−1 S

(5.10)

Only the components of the stress acting in the actuation direction on the top and bottom of the piezo elements have a significant contribution. These normal components of the elastic stress are multiplied with the surface area to get the load vector F. The displacement x in the printhead structure can be written as  x= ξi φ i (5.11) i

with ξi the modal coordinates, or the contribution of every eigenmode to the resulting deformation of the printhead. With the orthogonal condition: φ Tj m φ i = 0

j = i

(5.12)

5.7 Residual Vibrations

φ Tj k φ i = 0

j = i

(5.13)

and the normalization: φ Ti m φ i = 1

(5.14)

The modal coordinates can be resolved in the time domain with the following equation: ξ¨i + ωi2 ξi = fi

(5.15)

where fi is defined as fi = φ Ti · F

(5.16)

After resolving the modal coordinates ξi and the eigenmodes φ i , the transient response of the printhead structure can be resolved with Eq. (5.11), the mode superposition analysis. To model the impact of the resonances in the printhead structure on the pressure waves inside the ink channels, a finite element model of the ink domain has to be used. Normally, the deformation of the channel cross sections by the actuation of a piezo element, Eq. (5.10) is the boundary condition. Now, the deformation of the channel walls from the modal analysis and the mode superposition analysis is used as boundary condition. To determine which modes are contributing to the pressure waves inside the channels, the resulting channel wall deformations per mode have to be taken as boundary condition. It turns out that the strongest excited modes do not necessarily have the most impact on the channel acoustics and the drop formation. The frequency characteristics of the channels, which are strongly affected by the length over which the pressure channel is deformed, are important for the acoustic coupling between resonances of the printhead structure and their impact on the meniscus movement and drop formation [1].

5.7 Residual Vibrations

In order to print patterns on a substrate, not only multiple channels have to be actuated but also each channel has to be actuated many times with a short time interval, that is a high drop repetition rate. When the next droplet is fired before the acoustic pressure waves have come to rest, which can take typically 100 μs [18], the next drop formation cycle will start with a nonzero velocity of the meniscus surface. Depending on whether the residual movement of the meniscus is in-phase or out-of-phase with the next drop formation cycle, the drop speed of the next droplet will be higher or lower, respectively. This results in oscillations in drop speed and drop size at a drop repetition rate higher than 10 kHz. The frequency at which the drop speed reaches the highest value corresponds to the channel resonance frequency. The other peaks are found at fractions

83

84

5 Cross Talk in Piezo Inkjet

(1/n, n = 1, 2, 3, . . .) of the resonance frequency, corresponding to a time period between the consecutive drops of n resonance periods respectively. The amplitude of the oscillations is mainly determined by the amount of damping in the channel acoustics. The most important contribution to the damping is the viscous dissipation in the nozzle, when a pressure wave reflects at the nozzle side. Traveling through the channels gives only a minor contribution because the speed of the ink in the large channel cross section is much lower than the speed in the small nozzle cross section. Therefore, the nozzle area and the nozzle length have a large impact on damping, and thus on the amplitude of the oscillations in drop speed. The impact of residual vibrations on the drop properties is also influenced by cross-talk effects. The residual vibrations and cross talk do influence each other at high drop repetition rates. This can result in very large differences in drop speed between a single actuated channel and the same channel with the neighboring channels are firing as well [1, 20].

References 1. Wijshoff, H. (2010) The dynamics of the

2.

3.

4.

5.

6.

7.

8.

piezo inkjet printhead operation. Phys. Rep., 491, 77. McDonald, M. (1996) Scaling of piezoelectric drop-on-demand jets for high resolution applications. Proceedings IS&T’s Nip12, San Antonio, p. 53. Manning, H.J. and Harvey, R.A. (1999) Xaar greyscale technology. Proceedings IS&T’s Nip15, Orlando, p. 35. Zapka, W., Kaack, R., Voit, W., Schulz, M., Zimmermann, M., Levin, L., and Ahlemeir, W. (2002) Large drop volumes from Xaar-type inkjet printheads. Proceedings IS&T’s Nip18, San Diego, p. 161. Beurer, G. and Kretschmer, J. (1997) Function and performance of a shear mode piezo printhead. Proceedings IS&T’s Nip13, Seattle, p. 621. Kretschmer, J. and Beurer, G. (1997) Design parameters of a shear mode piezo printhead for a given resolution. Proceedings IS&T’s Nip13, Seattle, p. 626. McDonald, M. (1999) Crosstalk study of a high speed shear mode piezo inkjet printhead. Proceedings IS&T’s Nip15, Orlando, p. 40. Stemme, E. and Larsson, S.G. (1973) A piezoelectric capillary injector. IEEE Trans. Electron. Dev., 20, 14.

9. Kyser, E.L., Collins, L.F., and Herbert,

10.

11.

12.

13.

14.

15.

16.

N. (1981) Design of an impulse ink jet. J. Appl. Photogr. Eng., 7, 73. Germano, C. (1971) Flexure mode piezoelectric transducers. IEEE Trans. Audio Electroacoust., 19(1), 6. Wang, Q.M. and Cross, L.E. (1998) Performance analysis of piezoelectric cantilever bending actuators. Ferroelectrics, 215, 187. Uchino, K., Yoshizaki, M., Kasai, K., Yamamura, H., Sakai, N., and Asakura, H. (1987) Monomorph actuators using semiconductive ferroelectrics. Jap. J. Appl. Phys., 26(7), 1046. Takahashi, T. (2001) Adaptability of piezoelectric inkjet head. Proceedings IS&T’s Nip17, Fort Lauderdale, p. 323. Chen, T. (2006) Piezoelectric inkjet print head technology for Precision Dispensing Application. Proceedings DF2006, Denver, p. 66. Maekawa, S., Yoshida, T., Kinpara, S., Eguchi, H., and Ohta, Y. (2007) Wide print-head with high-stiffness and control method drive. Proceedings IS&T’s Nip23, Anchorage, p. 310. Wijshoff, H. (2004) Free surface flow and acousto-elastic interaction in piezo inkjet. Proc. NSTI Nanotech2004, 2, Boston, 215.

References 17. Howkins, S.D., Willis, C.A., and

18.

19.

20.

21.

22.

Nishimura, H. (2004) Cross-talk reduction by smart delay firing. Proceedings IS&T’s Nip20, Salt Lake City, p. 845. Wijshoff, H. (2006) Manipulating drop formation in piezo acoustic inkjet. Proceedings IS&T’s Nip22, Denver, p. 79. McDonald, M. (2001) Continuous improvement: Performance and reliability in shear mode piezo ink jet printing. Proceedings IS&T’s Nip17, Fort Lauderdale, p. 287. Duby, T.G. (2001) Performance improvements for commercial piezo prinhead. Proceedings IS&T’s Nip17, Fort Lauderdale, p. 328. Kitahara, T. (1995) Ink jet head with multi-layer piezoelectric actuator. Proceedings IS&T’s Nip11, Hilton Head, p. 346. Berger, S.S., Burr, R.F., Padgett, J.D., and Tence, D.A. (1997) Ink manifold

23.

24.

25.

26.

design of phase change piezoelectric ink jets. Proceedings IS&T’s Nip13, Seattle, p. 703. Noto, N., Torii, T., Suematsu, S., and Kugai, K. (1998) A new compact high resoluation solid ink print head and its application to a plate making printer. Proceedings IS&T’s Nip14, Toronto, p. 50. Brady, A.L., McDonald, M., Theriault, S.N., and Smith, B. (2005) The impact of Silicon MEMS on the future of ink jet printhead design and performance. Proceedings IS&T’s Nip21, Baltimore, p. 264. Letendre, W., Halwawala, S., and Smith, B. (2006) Jetting and imaging performance of the M-300/10 jet module. Proceedings IS&T’s Nip22, Denver, p. 83. Dijksman, J.F. (1999) Hydro-acoustics of piezoelectric driven ink-jet print heads. Flow Turb. & Comb., 61, 211.

85

87

6 Patterning Patrick J. Smith and Jonathan Stringer

6.1 Introduction

Like other printing and manufacturing techniques, inkjet printing is judged by the quality of the final product. If the product looks good and performs well, then the manufacturing technique is judged successful. (Obviously, the time taken to produce the final product is also a factor in determining success.) Inkjet-printed features are prepared from a series of droplets being deposited in an overlapping manner on a substrate. From a graphical point of view, inkjet-printed droplets need only be positioned close enough together so that the human eye perceives a continuous print. However, if the printed material serves a function, such as the silver particles that form a printed conductor, then the constituent particles must have a sufficient degree of contact with each other in order to perform that function satisfactorily. This chapter discusses how droplets of functional material, such as conductive polymers or metal nanoparticles, can be placed, or patterned, in order to obtain the most successful print. The behavior of droplets on the substrate, from the moment of their impact through to the final feature being obtained by the evaporation of carrier solvent or the droplet’s solidification, is reviewed since this allows one to predict and control the appearance of the final print [1]. The main substrate used in the following discussion is glass, which is usually cleaned, although in some cases a coating, which either increases or decreases the contact angle, has also been applied. Two types of droplets have been considered. The first are droplets of solvent, which may also describe droplets of solution below the solution’s critical concentration. (This type of droplet can also describe droplets of solvent that solidify on coming into contact with a substrate that is set at a temperature below the solvent’s melting point.) The second are suspension droplets that can also describe droplets of solution after the critical concentration has been exceeded and are useful in explaining droplets formed from nanoparticle ink. As well as considering single droplets, large arrays of droplets, such as lines and films, are also discussed. Inkjet-based Micromanufacturing, First Edition. Edited by J. G. Korvink, P. J. Smith, and D.-Y. Shin. © 2012 Wiley-VCH Verlag GmbH & Co. KGaA. Published 2012 by Wiley-VCH Verlag GmbH & Co. KGaA.

88

6 Patterning

6.1.1 Droplet Impact and Final Droplet Radius

In their investigation into droplet impact, van Dam and Le Clerc [2] used an inkjet printer to produce water droplets with diameters in the range of 36–84 μm. The substrate was glass, which had been thoroughly cleaned and treated to produce ◦ advancing contact angles of 15◦ , 35◦ , or 70 . They divided the impact of a droplet with glass into three stages: contact, spreading, and recoil. After the droplet has made contact with the substrate, a thin, circular film forms around it. The radius of the circular film expands to an order of magnitude greater than that of the in-flight droplet radius. As the film expands, liquid travels radially outwards and mass is accumulated at the boundary, forming a liquid ring. Kinetic energy is dissipated partly because of viscous flow in the thin film. The film recoils when a maximum is reached. The fluid comes to rest after a series of inertial oscillations that are dampened by viscous dissipation. Splashing has been seen to occur in droplets with Weber (We) numbers of 100–1000, and fingers have been observed in droplets that have a Reynolds (Re) number of 15 000 and a We of 1000. Re = (ρ × u × d)/μ and We = (ρ × u2 × d)/σ , where ρ, u, d, μ, and σ are the liquid density, droplet impact velocity, droplet diameter, liquid viscosity, and liquid surface tension, respectively. As the drops used in inkjet printing typically have diameters below 100 μm, values of Re = 2.5–2000, and We = 2.7–1000 can be expected. It is important to understand how the impact and spreading of a droplet on the substrate controls the evolution of the size of the droplet on the substrate. This was looked at by van Dam and Le Clerc, who studied the impact of inkjet-printed water droplets on surface-modified glass using high-speed flash photography. On comparing the maximum radii of the impacted droplets with a number of predictions made by several preexisting models, they found that the maximum was most accurately predicted by the model of Pasandideh-Fard et al. [3]: Rf /Ro = ([We + 12]/[3(fs − cos θd ) + 4(We/Re0.5 )])0.5 , where Rf = final radius, R0 = initial radius, and fs = the ratio of fluid–vapor surface to fluid–solid surface, which can be taken as 1 for small contact angles, and θd = the contact angle formed by the droplet when it reaches its maximum radial extent. The limitation of this model was that the assumptions made were only valid at relatively high impact speeds, below which the final radius was more accurately predicted by a volume conservation model of a spherical cap. Further experiments were performed by van Dam and Le Clerc that looked at the size of deposits left by silver salt solutions, which were found to be largely invariant over for all but the highest impact velocity. This indicates the tendency for the capillary spreading phase to control the final radius of the deposited droplet. This has been shown experimentally in other work by Stringer and Derby [4] and by Hsiao et al. [5]. While the capillary spreading phase after initial impact may have a significant influence on the final deposit diameter, it is important to consider the full evolution of the droplet during deposition. This is particularly true for functional applications

6.1 Introduction

Spreading factor bmax (−)

5 Pasandideh-Fard (3) Ukiwe (4) Bennett (5) Park (6) Experimental Exponential fit

4

3

2 0

100

200

300

400

500

600

Molecular weight (kD) Figure 6.1 Droplet diameter after impact on the surface for different molecular weights of polystyrene/toluene solutions. The closed symbols represent the experimental data, while the errors bar in x-direction show the

molecular weight distribution. The dashed line is a first order exponential decay fit (R2 = 0.92). The open symbols show the calculated diameters for various models. (From Ref. [4], © 2009 Wiley.)

where any cross contamination between neighboring droplets could result in a loss of function; an example of this would be a droplet adjoining a neighboring droplet and forming a short circuit in a conductor. The final radii of inkjet-printed droplets of a polystyrene/toluene solution, in which the molar mass of polystyrene was varied from 1.5 up to 545 kDa, have also been measured [6]. In this work, a number of models were used to predict the final radius and these values were compared to the actual measured radius. No one model was found to predict the final radius for the entire range of molecular weight (Figure 6.1). The extent to which a droplet spreads was found to be a function of the polymer’s molar mass, with the higher values causing an increase in viscosity, which resulted in smaller dried droplet radii. The increase in viscosity can slow down the capillary spreading, as well as impact, and that the spreading might be ‘‘frozen’’ by the phase change. For molten droplets, their behavior is greatly dictated by the freezing of the contact line, which has a large influence on the final solid shape of the droplet [7]. For low values of We, spreading is driven by interfacial forces that take place at the contact line rather than by impact phenomena. For low We droplet deposition, the Ohnesorge number (Z = μ(ρ × σ × d)−0.5 ), which measures the ratio of the viscous and inertial resistances to spreading, is of use; its value decreases with increasing droplet size. For a droplet that is solidifying, the Stefan number (S = c(Tf − Tt )/L, where c is the specific heat, Tf is the solidification temperature, Tt is the substrates temperature, and L is the latent heat of melting) is important.

89

90

6 Patterning

6.1.2 Evaporation of Inkjet-Printed Droplets at Room Temperature

` Bourges-Monnier and Shanahan [8] identified three regimes that a solvent droplet on a smooth substrate undergoes during evaporation: constant diameter, then constant contact angle before the final chaotic regime. In the constant diameter phase, the droplet loses mass through evaporation, while the droplet’s contact angle and height decrease until the receding angle is reached. Thereupon, the contact angle remains constant, while the droplet’s diameter shrinks. Finally, both the droplet’s diameter and contact angle shrink until evaporation is complete. As a general rule, droplets with smaller initial contact angles have higher evaporation rates and mass loss is linear. Solvent droplets on rough substrates only undergo two regimes; the constant contact angle regime is not entered, which is due to the droplet’s contact line being pinned. The evaporation of a droplet is also affected by whether the substrate is a thermal insulator or a conductor [9]. If the substrate is a perfect thermal insulator, then the evaporation rate is altered by changes in the droplet–vapor interface area. However, if the substrate is a perfect thermal conductor, then the evaporation rate is also affected by a second mechanism, namely, the heat transfer between the substrate and the droplet. In this second situation, the evaporation rate is higher than that of a droplet sitting on a thermal insulator. An evaporating, sessile droplet of suspension on a smooth substrate can be thought of as behaving similarly to a solvent droplet on a rough substrate. Initially, the droplet’s diameter remains constant, while the droplet’s contact angle and height decrease. However, as the droplet is composed of suspended particles (or precipitate in the case of a solution droplet) and carrier solvent, its diameter remains constant because some of these particles are deposited close to the contact line, which pin the droplet [10]. As evaporation continues, a replenishing flow of solvent, carrying suspended material with it, travels from the droplet’s center to its pinned edge. This process continues until evaporation is complete, resulting in a feature that is commonly called a coffee stain. Deegan et al. [11] gave three conditions that need to be met by the droplet in order for coffee staining to occur: the solvent meets the substrate at a nonzero contact angle, the contact line is pinned, and that the solvent is volatile. These conditions were subsequently refined to being a pinned contact line and evaporation from the edge of the droplet [8]. Although the exact conditions that cause coffee staining continue to generate debate [1], it is generally agreed that when a ring of material forms from an evaporating droplet of suspension, the particles that form the ring are carried there by a strong replenishing flow, which originates from the droplet’s center and is driven by evaporation of carrier solvent at the stationary contact line. The strength of the replenishing flow has been demonstrated by Magdassi et al. [12], who found that the electrical conductivity was 15% of bulk silver in rings formed from the room temperature evaporation of aqueous droplets of silver nanoparticles.

6.1 Introduction

6.1.3 Morphological Control for Ink Droplets, Lines, and Films

Inkjet-printed features are composed of overlapping droplets. Just as a single droplet can dry to leave a ‘‘coffee stain,’’ so an inkjet-printed film or line can dry to leave a similar feature. For example, a square film can dry to leave a dark frame [13]. The approach to preventing coffee staining in droplets can be used with equal success in films. Studies performed with polymer solutions [14] demonstrate that the use of a binary mixture of solvents can eliminate the formation of ring stains, if one of the solvents has a much higher boiling point than the other (Figure 6.2). The explanation for why coffee staining is prevented, using a two-solvent approach, is that the percentage of the higher boiling point solvent at the contact line increases. This shift causes a decrease in the evaporation rate at the contact line and establishes a surface tension gradient; a flow travels from low surface tension regions toward high surface tension regions when M, the Marangoni number, is sufficiently large [12]. (M = −(dγ /dT)LT/(ηα), where dγ /dT is the change in surface tension with temperature, L is the length scale involved, T is the difference in temperature, η is the dynamic viscosity, and D is the thermal diffusivity.) Lines are of interest since they are printed to form contacts and interconnects; with a number of ways to reduce line width and control morphology being investigated. Line widths can be predicted if the equilibrium contact angle is known (and a good rule to adopt is to ensure that this value is known). Derby et al. [15] printed a silver solution on to five different substrates (ranging from glass, 5.9◦ to Teflon™, 58.7◦ ) and used the equation, w2 = πd3 /(6x)/(θ/[4 sin2 θ] − cos θ/[4 sin θ]) to calculate predicted track width (where w = predicted line width, x = dot spacing, and d = droplet diameter). They found that their predictions compared well to measured values. Schiaffino and Sonin [16] found that an inkjet-printed line of molten wax forms with parallel contact lines, which freeze while the majority of the bead is still largely liquid. The still-molten material is stable when the contact angle is less than 90◦ but unstable above this. An inkjet-printed line is similar to a rivulet of liquid, for which Davis [17] derived a stability criterion by dividing the nature of the contact line into three groups: (i) lines with a fixed equilibrium contact angle and mobile contact lines; (ii) lines whose contact angle depends on the contact line speed, but reduces to an equilibrium value at zero speed; and (iii) lines whose contact lines are arrested in a parallel state while the contact angle is free to change. Davis showed that cases (i) and (ii) will always be unstable at some disturbance wavelengths, but that case (iii) will be stable if θ < 90◦ , which is representative of a liquid bead with strong contact angle hysteresis (θa > θr , where θa is the advancing contact angle and θr is the receding contact angle). Case (i) was represented by water, which was printed onto polymethyl methacrylate [14] for a set frequency at various substrate speeds, all of which were low enough for the deposited droplets to overlap. Large, unconnected sessile drops were formed instead, with their size found to be dependent on substrate speed and

91

X (μm)

100

150

0

2

3

4

5

0

50

x -direction y -direction

1

0

0.0

25.0

1

2

3

4

5

x

125.0

0

300.0

y

100

0.0

X (μm)

x

200

1.25

300

x -direction y -direction

300.0

0

2

4

6

8

10

0

125.0

y

50

0.0

X (μm)

x

100

5.0

150

x -direction y -direction

125.0

Figure 6.2 Profiles (a) and cross sections (b) of droplets formed from a 1 wt% solution of polystyrene. On the left-hand side, the solvent used was acetophenone, which boils at 202 ◦ C; in the middle the solvent used was ethyl acetate, which boils at 77 ◦ C. The right-hand image is of a droplet of 80/20 wt% ethyl acetate/acetophenone solution, which illustrates the influence of solvent ratios on final droplet morphology. (From Ref. [12], © 2004 The American Chemical Society.)

(b)

Z (μm)

(a)

y

Z (μm)

125.0

3.5

Z (μm)

92

6 Patterning

6.1 Introduction

deposition frequency. Duineveld [18] observed the same phenomenon for aqueous droplets of PEDOT:PSS (poly(3,4-ethylenedioxythiophene:poly(styrene-sulfonate)) doped with polystyrene sulfonic acid) printed onto CF4 -treated glass (θa = 97◦ and θr = 32◦ ). Duineveld’s main concern was investigating lines formed from liquids with zero receding angles, namely, PEDOT:PSS on a substrate with θa = 66 or θa = 24◦ . Many of these lines exhibited a series of regularly spaced liquid bulges that were connected by a ridge of liquid. The instabilities occurred when the calculated initial angle formed by the deposited droplet is larger than the advancing contact angle, θa , of the line causing a bulge to form. Ink is pumped through the ridge due to a pressure difference that is generated by the advancing liquid front at the head of the line, which causes the contact angle in the ridge to be smaller than the advancing contact angle. A new bulge forms when θa is exceeded. The bulges were found to depend on substrate velocity and applied liquid volume. The distance between bulges decreased as substrate velocity and applied liquid volume increased. Stringer and Derby [19] adapted the model proposed by Duineveld to make the model of bulging more amenable to graphical representation by means of a stability map. This was achieved by considering whether the liquid in a newly deposited droplet was liable to spread because of capillarity on impact or be pumped into the already deposited bead directly behind the droplet, which was modeled as a stable bead using an earlier model [13]. Using this method, rather than trying to model the full evolution of the bulge, they were able to group the relevant variables into two distinct groups. This enabled them to construct a stability map that not only predicted bulging behavior but could also be used to predict whether drops would successfully merge into a bead with parallel contact lines, and placed a limit on the morphologies that could be observed with a given combination of printing platform and ink. When printing lines, one must consider the drying time experienced by each droplet [20]. Soltman and Subramanian investigated the variation in printed line morphology as a function of dot spacing and substrate temperature. They found that bulges occurred when there was a short delay between droplet deposition and a small dot spacing (Figure 6.3). Bulges also occurred for low temperatures and small dot spacings. When low temperatures are used, the ink droplets remain liquid for longer, with successive deposited droplets adding to the ink body. Similarly, a decrease in dot spacing can be thought of as an increase in the amount of ink per unit area. The extreme of going to higher temperatures and/or longer delays leads to each individual droplet drying before the next droplet is deposited, which also produces an unwanted morphology, stacked coins. Subramanian et al. [21] have also investigated the stability of inkjet-printed rectangular films, which is a common building block for printed electronic structures such as transistors, displays, and sensors. They note that using fixed dot spacing resulted in both irregular and circular features that deviated significantly from the targeted rectangle. They addressed this problem by employing a variable line spacing approach that maintained the bead’s contact angle between its advancing

93

6 Patterning

Stacked coins

Delay

94

Uniform line

Scalloped line

Isolated drops

Bulging

Dot spacing (Dx ) Figure 6.3 Typical printed line behavior of PEDOT:PSS ink at intermediate temperatures (approximately 30 ◦ C). (From Ref. [18], © 2008 The American Chemical Society.)

and receding angles as it is printed. They also demonstrated the usefulness of prepatterning a border, which helped pin the feature’s contact line. For example, for a square they printed the four corners before dispensing a further five droplets in the center. Meier et al. [22] combined a number of strategies to produce 25 μm wide lines on untreated and unstructured polyimide using a 10 pl printhead. Although the ink used formed a low equilibrium contact angle with polyimide (29◦ ), they were able to dispense droplets that were 2 pl in volume by setting the printhead temperature to 55 ◦ C, which lowered ink viscosity and allowed a lower voltage to be used that resulted in a smaller droplet. They also tailored the jetting waveform to remove the typical tail that appears in inkjet-printed droplets. By using the equation derived by Derby et al. [15], they successfully predicted the final widths of their lines. This enabled them to print arrays of split-ring resonators that can form metamaterial layers designed for gigahertz to terahertz frequencies [23]. In order to reduce bulging, they set the printer’s dot spacing value to 10 μm and used a dotted line pattern that resulted in an effective dot spacing of 20 μm.

6.2 Conclusion

The topic of patterning involves the understanding of ink droplet behavior on the substrate, and how this behavior can be beneficially influenced. Typically, inkjet-printed droplets are composed of a carrier solvent and a functional material. Such a system can lead to coffee-stained features. For the prevention of coffee staining, the use of a mixture of solvents is useful, although the exact ratio needs to

References

be optimized for the particular system under investigation. Lines can also exhibit bulging, which occur when the calculated initial angle formed by each deposited droplet is larger than the advancing contact angle of the line. This can be somewhat rectified by optimizing the print head velocity and dot spacing.

References 1. Smith, P.J. (2009) The behaviour of

2.

3.

4.

5.

6.

7.

8.

9.

an ink droplet on the substrate, in The Chemistry of Inkjet Inks (ed. S. Magdassi), World Scientific Publishing, Singapore, 55–72. van Dam, D.B. and Le Clerc, C. (2004) Experimental study of the impact of an ink-jet printed droplet on a solid substrate. Phys. Fluids, 16, 3403–3414. Pasandideh-Fard, M., Qiao, Y.M., Chandra, S., and Mostaghimi, J. (1996) Capillary effects during droplet impact on a solid surface. Phys. Fluids, 8, 650–659. Stringer, J.E. and Derby, B. (2009) Limits to feature size and resolution in ink jet printing. J. Eur. Ceram. Soc., 29, 913–918. Hsiao, W.-K., Hoath, S.D., Martin, G.D., and Hutchings, I.M. (2009) Ink jet printing for direct mask deposition in printed circuit board fabrication. J. Imaging Sci. Technol., 53, 050304. Perelaer, J., Smith, P.J., van den Bosch, E., van Grootel, S.S.C., Ketelaars, P.H.J.M., and Schubert, U.S. (2009) The spreading of inkjet-printed droplets with varying polymer molar mass on a dry solid substrate. Macromol. Chem. Phys., 210, 495–502. Schiaffino, S. and Sonin, A.A. (1997) Molten droplet deposition and solidification at low Weber number. Phys. Fluids, 9, 3172–3187. ` Bourges-Monnier, C. and Shanahan, M.E.R. (1995) Influence of evaporation on contact angle. Langmuir, 11, 2820–2829. David, S., Sefiane, K., and Tadrist, L. (2007) Experimental investigation of the effect of thermal properties of the substrate in the wetting and evaporation of sessile drops. Colloids Surf. A: Physicochem. Eng. Aspects, 298, 108–114.

10. Deegan, R.D., Bakajin, O., Dupont, T.F.,

11.

12.

13.

14.

15.

16.

17.

18.

19.

Huber, G., Nagel, S.R., and Witten, T.A. (2002) Contact line deposits in an evaporating drop. Phys. Rev. E, 62, 756–765. Deegan, R.D., Bakajin, O., Dupont, T.F., Huber, G., Nagel, S.R., and Witten, T.A. (1997) Capillary flow as the cause of ring stains from dried liquid drops. Nature, 389, 827–829. Magdassi, S., Grouchko, M., Toker, D., Kamynshny, A., Balberg, I., and Millo, O. (2005) Ring stain effect at room temperature in silver nanoparticles yields high electrical conductivity. Langmuir, 21, 10264–10267. Tekin, E., de Gans, B.-J., and Schubert, U.S. (2004) Ink-jet printing of polymers – from single dots to thin film libraries. J. Mater. Chem., 14, 2627–2632. de Gans, B.-J. and Schubert, U.S. (2004) Inkjet printing of well-defined polymer dots and arrays. Langmuir, 20, 7789–7793. Smith, P.J., Shin, D.-Y., Stringer, J.E., Derby, B., and Reis, N. (2007) Direct ink-jet printing and low temperature conversion of conductive silver patterns. J. Mater. Sci., 41, 4153–4158. Schiaffino, S. and Sonin, A.A. (1997) Formation and stability of liquid and molten beads on a solid surface. J. Fluid Mech., 343, 95–110. Davis, S.H. (1980) Moving contact lines and rivulet instabilities. Part 1. The static rivulet. J. Fluid Mech., 98, 225–242. Duineveld, P.C. (2003) The stability of ink-jet printed lines of liquid with zero receding contact angle on a homogeneous substrate. J. Fluid Mech., 477, 175–200. Stringer, J. and Derby, B. (2010) Formation and stability of lines formed

95

96

6 Patterning by inkjet printing. Langmuir, 26, 10365–10372. 20. Soltman, D. and Subramanian, V.S. (2008) Inkjet printed line morphologies and temperature control of the coffee ring effect. Langmuir, 24, 2224–2231. 21. Soltman, D., Smith, B., Kang, H., Morris, S.J.S., and Subramanian, V.S. (2010) Methodology for inkjet printing of partially wetting films. Langmuir, 26, 15686–15693.

22. Meier, H., L¨ offelmann, U., Mager, D.,

Smith, P.J., and Korvink, J.G. (2009) Inkjet printed, conductive 25 μm wide silver tracks on unstructured polyimide. Phys. Status Solidi A, 206, 1626–1630. 23. Walther, M., Ortner, A., Meier, H., L¨offelmann, U., Smith, P.J., and Korvink, J.G. (2009) Terahertz metamaterials fabricated by inkjet printing. Appl. Phys. Lett., 95, 251107.

97

7 Drying of Inkjet-Printed Droplets Hans Kuerten and Daniel Siregar

7.1 Introduction

After an inkjet-printed droplet collides with a solid, horizontal surface, two subsequent processes take place. First, its kinetic energy will cause spreading of the droplet on the surface. Depending on the wetting properties of the substrate and on the viscosity of the liquid, the droplet will spread and possibly oscillate until an equilibrium shape is reached in which the forces on the droplet balance. The characteristic timescale for this initial phase is given by the inertial timescale τ = R0 /U0 , where R0 is the initial radius of the droplet and U0 the impact velocity. Usually, this characteristic timescale is less than 1 ms for droplets with a size typical for inkjet printing applications. The initial phase is followed by the drying phase, in which the liquid contained in the droplet evaporates and after total drying a solid residual layer remains on the substrate. The typical timescale for drying depends on the ambient temperature and humidity and on the droplet size. Under normal circumstances, it is on the order of a few seconds for droplets with a size typical for inkjet printing applications. This implies that the two phases of the process, spreading and drying, hardly influence each other and can be modeled separately. Processes happening during the drying phase are of crucial importance to the properties of the deposited structure resulting after drying. In the traditional graphical applications, the visual appearance depends on the drying process, while in more recent applications for depositing proteins in biosensors [1], or electro-optical materials in the electronic industry [2] the distribution of the layer thickness determines the functionality of the deposited layer [3]. This distribution is determined by the redistribution of material during the drying process [4]. In this chapter, we focus on the drying process on a nonporous substrate. Also, a considerable amount of research has been performed on the behavior of droplets on porous substrates, both experimentally and by analytical and numerical modeling [5, 6]. In the next section, a suitable model is derived and a numerical method to solve the equations of the model is presented. Section 7.3 shows some typical results. A distinction will be made between situations in which the contact line Inkjet-based Micromanufacturing, First Edition. Edited by J. G. Korvink, P. J. Smith, and D.-Y. Shin. © 2012 Wiley-VCH Verlag GmbH & Co. KGaA. Published 2012 by Wiley-VCH Verlag GmbH & Co. KGaA.

98

7 Drying of Inkjet-Printed Droplets

of the droplet is pinned and where it is unpinned. Which of the two happens depends on the spreading and receding contact angles and on the smoothness of the substrate.

7.2 Modeling of Drying of a Droplet

An important contribution to the understanding of the phenomena that take place during drying of a droplet has been made by Deegan [7–9], who explained the typical shape of a coffee stain on a table, where all coffee has moved to the rim of the original droplet after the water has evaporated. Deegan’s model assumes that because of surface tension the droplet maintains the shape of a spherical cap during drying. Since evaporation is larger at the edge of the droplet, a convective flow from the center to the edge of the droplet appears and this transports the solute. Diffusion counteracts this process, but is often less important, especially for large solute molecules that have a very small diffusion coefficient. An important extension of this model has been derived by Fischer [10], who did not assume the spherical shape of the droplet during drying, but derived an evolution equation for the shape, based on the forces acting on it. This extension also makes it possible to incorporate, for instance, concentration-dependent viscosity, which can play a large role in cases where the initial solute concentration is already high [3]. 7.2.1 Fluid Model

In the remainder of this section, we first derive suitable governing equations for a droplet and the concentration of solute during drying, which are valid in case of micrometer-sized droplets. The starting point of the derivation is the Navier–Stokes equation for the droplet. If we assume an incompressible liquid in the droplet, the Navier–Stokes equation can be written as ρ

∂u + ρu · ∇u = −∇p + η∇ 2 u − gez ∂t

(7.1)

and the continuity equation: ∇ ·u=0

(7.2)

where u is the fluid velocity, ρ and η the mass density and dynamic viscosity of the fluid and p the pressure. The acceleration of gravity is denoted by g and it points in the negative z direction. The order of magnitude of the different terms in Eqs. (7.1) and (7.2) can be assessed by introducing characteristic values for horizontal size of the droplet, R and vertical size of the droplet, H. Suitable values are the radius of the droplet and its height after the initial spreading phase. If the typical velocity of the liquid in the vertical direction is equal to the evaporation velocity averaged over the surface of

7.2 Modeling of Drying of a Droplet

the droplet, ve,av , the typical value for the horizontal velocity equals: R H which is large compared to the typical vertical velocity, if the droplet’s radius is large compared to its height. Hence, we scale horizontal velocity components with V and vertical velocity with ve,av . The most suitable timescale is R/V. Next, we split the velocity in its horizontal component u and a vertical component w: u = u + wez . In this way, the horizontal and vertical components of the Navier–Stokes equation (7.1) can be written in the form:     ∗ ∂ 2 u∗ ∂u R ∗ ∗ H2 2 ∗ ∗ ∗ (7.3) + u · ∇u = − ∇ p + Ca + 2 ∇ u  We ∂t∗ Rc ∂z∗2 R V = ve,av

and



We

∂w∗ + u∗ · ∇w∗ ∂t∗

 =−

  2 ∗ R3 ∂p∗ ∂ w H2 2 ∗ − Bd + Ca + ∇ w H2 Rc ∂z∗ ∂z∗2 R2 

(7.4)

Here we introduced asterisks for scaled variables and the symbol ∇ denotes the horizontal component of the gradient operator. Moreover, Ca is the capillary number, defined as ηV R2 σ H2 We is the Weber number, defined as Ca =

ρV 2 R σ and Bd is the Bond number, defined as We =

(7.5)

(7.6)

ρgR2 (7.7) σ Finally, Rc is the typical value of the radius of curvature, which yields σ/Rc as the typical value for pressure, where σ is the surface tension. Bd =

7.2.2 Lubrication Approximation

In order to estimate the order of magnitude of the terms in the Navier–Stokes equation, we consider a typical case at the start of the drying with V = 1 × 10−6 m s−1 , R = 1 × 10−4 m, H = 1 × 10−5 m, Rc = 1 × 10−3 m, σ = 0.03 N/m, ρ = 1 × 103 kg/m3 , and η = 1 × 10−3 Pa s. This gives We = 3 × 10−12 , Ca = 3 × 10−6 and Bd = 3 × 10−3 . Depending on the initial concentration of the solute, the viscosity may increase several orders of magnitude during the drying process, which leads to a proportional increase in capillary number. The small Bond and Weber number justify, however, that the present scaling is valid. For the typical values of the parameters chosen here, the convective terms on the left-hand side of Eqs. (7.3) and (7.4), the gravity term, the horizontal derivatives in the viscous terms

99

100

7 Drying of Inkjet-Printed Droplets

in Eq. (7.3) and the complete viscous term in the vertical Navier–Stokes equation (7.4) can safely be neglected. Therefore, the governing equations can be simplified to ∂p =0 ∂z

(7.8)

and Rc ∂ 2 u Ca 2 (7.9) R ∂z where, for convenience, the asterisks have been dropped. These equations are similar to those derived by Fischer [10] for the axially symmetric case. The approximation used here, where horizontal length scales are large compared to vertical scales, is known as the lubrication approximation. From Eq. (7.8), it follows that the pressure is a function of the horizontal coordinates x and y only. Then, u can be found from Eq. (7.9) with a no-slip boundary condition u (z = 0) = 0 on the substrate and a no-stress boundary condition ∂u /∂z = 0 on the interface given by z = h(x, y) as   1 R 1 2 z − hz ∇ p (7.10) u = Ca Rc 2 ∇ p =

The time evolution of the interface height h(x, y) is determined by the continuity equation integrated over the vertical direction. The result is ∂h 1 R = −∇ (hu ) − ve = ∇ (h3 ∇ p) − ve ∂t 3Ca Rc

(7.11)

where the brackets denote the average over the droplet height. Moreover, in the final step, Eq. (7.10) has been substituted for the fluid velocity and ve denotes the local evaporation velocity. This equation can be solved if the pressure in the droplet can be expressed in terms of the height and if the local evaporation velocity is known. In the lubrication approximation, where gravity and hence hydrostatic pressure are negligible, two contributions to the pressure have to be taken into account: p = pL + . One is the pressure caused by surface tension, which depends on the local curvature of the interface and is in the lubrication approximation given by Rc H 2 ∇ h (7.12) R2  The other contribution to the pressure is the disjoining pressure, which accounts for the molecular interaction and is necessary to describe a moving contact line. For the case of an unpinned contact line, we will use the form proposed by Schwartz and Eley [11] given by  ∗ n  ∗ m  h h (7.13)  = −B − h h pL = −

where B is a positive constant, n and m with n > m are positive integer constants, and h∗ denotes the thickness of the precursor film, which is assumed to cover the dry parts of the substrate. It can be seen that the disjoining pressure only plays a

7.2 Modeling of Drying of a Droplet

role if the height of the droplet is of the same order of magnitude as the thickness of the precursor film and not constant. The first term in Eq. (7.13) describes the repulsive forces between liquid and solid, whereas the second term is the attractive force. Constant B is related to the equilibrium contact angle θe . A force balance in static conditions yields: B=

1 (n − 1)(m − 1) σ (1 − cos θe ) h∗ n−m

(7.14)

For small contact angles, this can be approximated by B=

1 (n − 1)(m − 1) 2 θe 2h∗ n−m

(7.15)

For n and m, various possibilities have been suggested and used in literature. Changes in n and m change the depth of the potential well and the strength of the repulsive force for small distances. The values n = 9 and m = 3 apply to the well-known Lennard-Jones potential. Teletzke et al. [12] used n = 3, n = 2 for simulations of water droplets spreading on a solid substrate. These lower values have the advantage of a better stability of the numerical algorithm. For a pinned contact line, the disjoining pressure can be disregarded, as the interaction of the air and the droplet needs no modeling. In that case, the evolution equation for the droplet height is found by substituting Eq. (7.12) in Eq. (7.11) as 1 H ∂h =− ∇ (h3 ∇ ∇2 h) − ve ∂t 3Ca R

(7.16)

7.2.3 Solute Concentration

The concentration of solute in the droplet changes by convection with the fluid velocity and by diffusion. Conservation of mass of the solute results in an evolution equation: ∂(ch) + ∇ (cu h) = ∇ (Dl h∇ c) ∂t

(7.17)

where c is the solute concentration and Dl the diffusion coefficient of the solute in the liquid. In this convection–diffusion equation, the vertical derivatives of the concentration have been neglected. This assumption can be made when diffusive transport in the vertical direction is dominant over advective transport in the horizontal directions, which is typically true if VH2 /RDl  1. For the typical parameter values mentioned above, this is satisfied if Dl  1 × 10−12 m2 /s. If the same scaling quantities for length and timescales are used as for the droplet shape, the nondimensionalized diffusivity equals: D∗ =

Dl RV

In the examples shown in the next section, we consider for simplicity an axially symmetric droplet. If r denotes the radial coordinate, the interface of the droplet is

101

102

7 Drying of Inkjet-Printed Droplets

written as z = h(r). Then the evolution equation for the droplet height is given by ∂h 1 H1 ∂ =− ∂t 3Ca R r ∂r



∂ 1 ∂ rh ∂r r ∂r 3

  ∂h r − ve ∂r

(7.18)

Note that the constant parameter in the first term on the right-hand side H/(CaR) has a value large compared to 1 for typical values of the physical parameters. The term between parentheses is equal to zero in case the curvature of the droplet is constant along the interface. Hence, a small departure from the shape of a spherical cap will result in a large restoring force. That is the reason why Deegan [8] assumed that the droplet remains a spherical cap during the drying process. In that case, the shape of the droplet and the convection velocity can be solved analytically in case of axial symmetry and only the concentration equation needs to be solved numerically. Close to the edge of the drop, however, and in case the contact line is not pinned, the deviations from a spherical shape can become large, so that the general shape of the interface considered here has to be taken into account. 7.2.4 Evaporation Velocity

In order to be able to solve Eq. (7.11) for the droplet height, an expression for the evaporation velocity ve as a function of the horizontal coordinates is required. The form of this velocity depends on the slowest step in the evaporation process, which can be either the transfer rate over the interface between droplet and air or the diffusion of the saturated vapor immediately above the interface. Even for drops of micrometer size, the typical time for diffusion of vapor is of the order of 10−6 s, whereas the typical time for the transfer rate above the interface is typically of the order to 10−10 s. We can therefore assume that the diffusion of vapor is the limiting step. In that case, the evaporation velocity follows from the solution of a diffusion equation for vapor concentration in the air surrounding the droplet. The relevant boundary conditions are that the vapor concentration equals the concentration of saturated air at the interface and equals the bulk vapor concentration far away from the interface. Even for the relatively simple geometry of a droplet with the shape of a spherical cap, this problem is quite hard to solve analytically [7, 9]. In case of a very thin drop, which is consistent with the lubrication approximation, the evaporation velocity can, in the case of an axially symmetric droplet, be written as [13, 14] ve =

2 Dv (ρs − ρ∞ ) √ π ρ R2 − r 2

(7.19)

where Dv is the diffusion coefficient for vapor in air, ρs and ρ∞ are the mass densities of saturated vapor and in the bulk and ρ denotes the liquid mass density, as before.

7.3 Results

Integration of Eq. (7.19) over the droplet interface yields the total evaporation rate dV/dt as: ρs − ρ∞ dV = −4Dv R dt ρ

(7.20)

which illustrates that the evaporation rate is proportional to the droplet radius and not to its surface area. This is a consequence of the fact that diffusion is the limiting step in the evaporation process. In case the droplet evaporates in vacuum, the transfer through the interface is the limiting step and the total evaporation rate is proportional to the surface area of the droplet. 7.2.5 Numerical Method

As can be seen, Eq. (7.18) is a nonlinear partial differential equation with a maximum fourth-order derivative with respect to the radial coordinate. Therefore, a numerical solution method has to be chosen with some care in order to guarantee stability without restricting the time step size too much. Using the method of lines, we transform the partial differential equation for the droplet height into a system of ordinary differential equations. To this end, the space domain is discretized by a finite volume method with central differences to approximate the spatial derivatives. In this way, mass conservation is also guaranteed on the discrete level. In order to deal with the large range of eigenvalues of the system of nonlinear ordinary differential equations, the time integration is solved by a fourth-order accurate Gear method [15], which is suited for stiff problems. For a problem in cylindrical coordinates, special care has to be taken near the axis of symmetry, r = 0. It turns out that all numerical problems are circumvented if the values of droplet height and pressure are taken in the centers of the cells and the values of velocity in the grid points and proper symmetry conditions are applied. Once the height of the droplet and hence the velocity are known, numerical solution of convection–diffusion equation (7.17) for the solute concentration is quite easy, since the equation is linear in the concentration and not stiff. Therefore, standard numerical methods can be used for the solution of this equation. For both convective and diffusive terms, a second-order accurate central finite volume method is applied. This ensures that no numerical diffusion is added. However, the grid size should be chosen sufficiently small to prevent numerical instabilities. Time integration is performed with the second-order accurate implicit Crank–Nicolson method. In one spatial dimension, only a tri-diagonal matrix, which results from the spatial discretization, has to be solved.

7.3 Results

In this section, we show several results of numerical simulations based on the lubrication approximation, both for pinned and unpinned contact lines. In almost

103

7 Drying of Inkjet-Printed Droplets

all examples, we retain the values of the typical parameters as mentioned above in the justification of the lubrication approximation. Results are presented in nondimensional form, where the initial droplet radius R and height H are used as length scales in the horizontal and vertical direction and time is scaled with R/V. The initial droplet shape is always chosen as the equilibrium shape in case there is no evaporation. For a pinned contact line, this is a spherical cap, that is the shape with constant curvature, but in the lubrication approximation this is approximated as an elliptic paraboloid. The initial shape of the droplet with an unpinned contact line is determined by the equilibrium of the pressure induced by surface tension and the disjoining pressure. In order to allow the movement of the contact line, a precursor film with thickness h∗ = 0.01H is added on the dry part of the substrate next to the droplet [16]. The parameters in the disjoining pressure (Eq. 7.13) are set to n = 3 and m = 2. Parameter B determines the contact angle and is given a value in agreement with the small contact angle of the initially shallow droplet. 7.3.1 Droplet Shape Evolution

Figure 7.1 shows the evolution of the shape of the droplet without solute for the case on a pinned contact line. The figure shows a gradual decrease of the droplet height during evaporation. The shape remains very close to a spherical cap. Only in the later stages of the evaporation the droplet becomes so shallow, that convection 1

h/H

0.8 0.6 0.4 0.2 0

0

0.5

1

r/R

Figure 7.1 Evolution of the shape of a droplet without solute with a pinned contact line. The lines correspond to equidistant times.

0.25 0.2 V/ 2π

104

0.15 0.1 0.05 0

0

0.005 t/t 0

0.01

Figure 7.2 Volume of a droplet without solute with a pinned contact line as a function of time.

7.3 Results

cannot keep pace with evaporation near the contact line. As a result, the contact line moves inward. This is an effect of the finite capillary number. For zero capillary number, the droplet would keep its spherical shape throughout the evaporation process and for larger capillary numbers the droplet cannot maintain the pinned contact line at an earlier time. In Figure 7.2, the volume of this droplet is plotted as a function of time. This figure clearly shows that the evaporation rate of this droplet is constant in time as long as the contact line stays in the same position, which is consistent with Eq. (7.19) for a droplet with constant radius. In the later stages of the evaporation process, the radius of the droplet decreases and the evaporation rate decreases as well. The corresponding results for the case of an unpinned contact line are shown in Figures 7.3 and 7.4. The shape of the droplet interface near the contact line clearly differs from the parabolic profile of the pinned contact line due to the presence of the disjoining pressure. The contact angle remains constant during the evaporation. Moreover, since the total evaporation rate is proportional to the radius of the droplet according to Eq. (7.20), the evaporation rate reduces in the course of time. For a perfectly spherical cap, the droplet volume would be proportional to (t0 − t)3/2 with t0 the total drying time. However, owing to the effect of disjoining pressure this is not exactly true, although initially this behavior is found.

1

h/H

0.8 0.6 0.4 0.2 0

0

0.5

1

r/R

Figure 7.3 Evolution of the shape of a droplet without solute with an unpinned contact line. The lines correspond to equidistant times.

0.25

V/ 2 π

0.2 0.15 0.1 0.05 0

0

0.005

0.01 t/t 0

0.015

0.02

Figure 7.4 Volume of a droplet without solute with an unpinned contact line as a function of time.

105

7 Drying of Inkjet-Printed Droplets

7.3.2 Layer Thickness

Next, we introduce the solute and solve equation (7.17) along with the equation for the droplet height. As an initial condition, we take a uniform solute concentration. The evaporation process depends on the value of the initial concentration, since the solute concentration cannot become larger than a certain limiting value above which the solute forms a close packing. In order to model this, convection of the solute and liquid is stopped at positions where this maximum value of the solute concentration is reached. If the initial solute concentration is large, the maximum concentration is reached earlier. We will compare results for a high initial concentration, which is 0.1 of the maximum possible concentration, with results for a low initial concentration, which is 10 times smaller. First, we consider again a droplet with a pinned contact line. Figures 7.5 and 7.6 show the layer thickness after evaporation for the high and low initial solute concentration, respectively, for a very small value of the diffusion coefficient. Effects of variations in the diffusion coefficient are discussed later. Both results show the typical coffee-stain effect with a thick layer near the edge of the droplet and a much thinner layer near the center. Since the evaporation velocity is larger near the edge of the droplet than in the center and the curvature of the interface has to remain uniform, the large evaporation has to be compensated by a convective velocity from the center to the edge of the droplet. The solute is convected with this velocity 1.5

ch /c 0H

1

0.5

0

0

0.5

1

Figure 7.5 Layer thickness after evaporation for the case of a pinned contact line and high initial solute concentration as a function of radial coordinate.

1

Figure 7.6 Layer thickness after evaporation for the case of a pinned contact line and low initial solute concentration as a function of radial coordinate.

r/R

5 4 ch/ c 0H

106

3 2 1 0

0

0.5 r/R

7.3 Results

and moves away from the center. The smaller the initial solute concentration, the larger the time before the maximum possible concentration is reached. Therefore, the coffee-stain effect is largest for the smaller initial solute concentration. In the present examples, it has been assumed that the viscosity of the fluid does not depend on the solute concentration. If a significant concentration-dependent viscosity is adopted, convection of solute is impeded in regions where the concentration is high and the coffee-stain effect is suppressed [3]. Figures 7.7 and 7.8 show the layer thickness after evaporation for droplets with an unpinned contact line for the high and low initial solute concentration, respectively. In this case, the disjoining pressure, which is important near the edge of the droplet, will try to maintain a constant contact angle. Hence, the contact line will move toward the center of the droplet, as shown in Figure 7.3. Therefore, the convection velocity is smaller than in the pinned case. This results in a smaller increase in concentration near the edge of the droplet and a more uniform layer thickness. The results shown in the figures are slightly counterintuitive. One would expect a thicker layer near the edge of the droplet for the lowest initial solute concentration, just as in the case of a pinned contact line. However, for the higher initial solution concentration, the maximum possible concentration is quickly reached near the edge of the droplet. Therefore, the shape evolution of the droplet looks more like the pinned contact line case shown in Figure 7.1 than like the unpinned case of Figure 7.3. For the lower initial solute concentration, the maximum possible concentration is not reached before the contact line starts moving toward the center

ch/c 0H

1.5

1

0.5

0

0

0.5 r/R

1

Figure 7.7 Layer thickness after evaporation for the case of an unpinned contact line and high initial solute concentration as a function of radial coordinate.

1

Figure 7.8 Layer thickness after evaporation for the case of an unpinned contact line and low initial solute concentration as a function of radial coordinate.

1

ch /c 0H

0.8 0.6 0.4 0.2 0

0

0.5 r /R

107

7 Drying of Inkjet-Printed Droplets

of the droplet and the shape evolution is not changed by the presence of the solute. Compared to the pinned contact line, the unpinned contact line leads to a flatter shape of the layer, especially if the initial solute concentration is low. 7.3.3 Effect of Diffusion

Finally, we study the effect of the diffusion coefficient. In the results shown above, the diffusion coefficient is so low that the effects of the diffusion are negligible. This happens in practical applications, for instance, for DNA molecules in water, for which the diffusion coefficient is on the order of Dl = 10−12 m2 /s. For smaller molecules in water, the diffusion coefficient is larger by several orders of magnitude. Diffusion will try to counteract the occurrence of concentration gradients caused by convection of the liquid from the center of the droplet to the edge. It can therefore be expected that for larger diffusion coefficients the coffee-stain effect, visible in Figures 7.5 and 7.6, is diminished. The diffusion will lead to lower solute concentrations near the edge of the droplet and therefore postpone the moment in which the maximum possible solute concentration is reached. Figure 7.9 shows the final layer thickness after evaporation for the case of a pinned contact line and a high initial solute concentration. Results at various values of the diffusion coefficient ranging between Dl = 1 × 10−10 m2 /s and Dl = 1 × 10−5 m2 /s are included. The profile of the layer thickness for the lowest value of Dl 2.5

1×10−5

ch/c 0H

2 1×10−6

1.5

1×10−10

1×10−7

1 0.5

−8

1×10

1×10−9

0

0

0.5

1

r/R

Figure 7.9 Layer thickness after evaporation for the case of an unpinned contact line and high initial solute concentration as a function of radial coordinate for various values of the diffusion coefficient; the values of D in m2 /s are indicated in the figure.

8 1 ×10

6 ch / c 0H

108

1 ×10

−5

−6

1 ×10

4

1 ×10 1 ×10

−7

2 0

−10

−9

1 ×10

0

0.5 r/R

−8

1

Figure 7.10 Layer thickness after evaporation for the case of an unpinned contact line and low initial solute concentration as a function of radial coordinate for various values of the diffusion coefficient; the values of D in m2 /s are indicated in the figure.

References

collapses with the result shown above and deviations with the result for a 10 times higher value of Dl are still small. A further amplification of Dl with a factor of 10 clearly shows the influence of diffusion, which reduces the coffee-stain effect. For the two highest values of Dl shown in the figure, the effect of diffusion is so strong that the solute concentration is almost uniform throughout the evaporation process. The maximum possible solute concentration is therefore reached at the same instant of time over the entire radius of the droplet. This leads to a final layer thickness, which has the shape of a spherical cap. Note, however, that such large diffusion coefficients are not realistic for water as solvent. Figure 7.10 shows the corresponding results for a low initial solute concentration. Again, the result for the lowest value of Dl is the same as the one shown in Figure 7.6 and a 10 times higher value of Dl yields approximately the same result. For higher values of Dl , the effect of diffusion is visible and leads to a more uniform solute concentration. In contrast with the previous example with a high initial solute concentration, the maximum possible solute concentration is not reached near the edge of the droplet before the contact line starts to move toward the center of the droplet because of the finite capillary number. This effect can also be seen in Figure 7.1. At a later time, the maximum concentration is reached in the region where there is still liquid present. This explains the deviation of the final layer thickness shown in Figure 7.10 from the spherical cap shape. In conclusion, it can be stated that for realistic values of the diffusion coefficient the effect of diffusion is limited. However, for very low values of Dl , which are applicable to large molecules like DNA solved in water, it can no longer be assumed that the concentration variations in vertical direction can be disregarded. The consequence is that a two-dimensional convection–diffusion equation for the solute concentration should be solved.

Acknowledgments

The research described in this chapter is supported by the Dutch Technology Foundation STW, applied-science division of NWO (Dutch Organisation for Scientific Research), and the Technology Program of the Ministry of Economic Affairs of the Netherlands. The authors wish to thank Oc´e and Philips Research for many stimulating discussions on the topic of this chapter.

References 1. Dufva, M. (2005) Fabrication of high

3. Van Dam, D.B. and Kuerten, J.G.M.

quality microarrays. Biomol. Eng., 22(5– 6), 173–184. 2. Hjelt, K.T., Van den Doel, R., Lubking, W., and Vellekoop, M.J. (2000) Measuring liquid evaporation from micromachined wells. Sens. Actuators, 85(1– 3), 384–389.

(2008) Modeling the Drying of Ink-Jet-Printed Structures and Experimental Verification. Langmuir, 24(2), 582–289. 4. Rieger, B., Van den Doel, L.R., and Van Vliet, L.J. (2003) Ring formation in nanoliter cups:

109

110

7 Drying of Inkjet-Printed Droplets

5.

6.

7.

8.

9.

Quantitative measurements of flow in micromachined wells. Phys. Rev. E, 68, 036312. Dijksman, J.F. and Pierik, A. (2008) Fluid dynamical analysis of the distribution of ink jet printed biomolecules in microarray substrates for genotyping applications. Biomicrofluidics, 2, 044101. Siregar, D.P., Kuerten, J.G.M., Wijshoff, H.M.A., and Van der Linden, L.T.M. (2010) Numerical simulation of the absorption of a droplet in a porous medium. Porous Media and Its Applications in Science, Engineering, and Industry: 3rd International Conference. AIP Conference Proceedings, Montecacini, Italy, 1254, pp. 135–140. Deegan, R.D., Bakajin, O., Dupont, T.F., Huber, G., Nagel, S.R., and Witten, T.A. (1997) Capillary flow as the cause of ring stains from dried liquid drops. Nature, 389, 827–829. Deegan, R.D. (2000) Pattern formation in drying drops. Phys. Rev. E, 61(1), 475–485. Deegan, R.D., Bakajin, O., Dupont, T.F., Huber, G., Nagel, S.R., and Witten, T.A.

10.

11.

12.

13.

14.

15.

16.

(2000) Contact line deposits in an evaporating drop. Phys. Rev. E, 62(1), 756–765. Fischer, B.J. (2002) Particle convection in an evaporating colloidal droplet. Langmuir, 18(1), 60–67. Schwartz, L.W. and Eley, R.R. (1998) Simulation of droplet motion on low-energy and heterogeneous surfaces. J. Colloid Interface Sci., 202, 173–188. Teletzke, G.F., Davis, H.T., and Scriven, L.E. (1987) How liquid spreads on solids. Chem. Eng. Commun., 55, 41–82. Guena, G., Poulard, C., and Cazabat, A.-M. (2007) The dynamics of evaporating sessile droplets. Colloid J., 69(1), 1–8. Popov, Y.O. (2005) Evaporative deposition patterns: spatial dimensions of the deposit. Phys. Rev. E, 71, 036313. Gear, C.W. (1970) The simultaneous numerical solution of differential-algebraic equations, SLAC-PUB-723. Alleborn, N. and Raszillier, H. (2004) Spreading and sorption of a droplet on a porous substrate. Chem. Eng. Sci., 59, 2071–2088.

111

8 Postprinting Processes for Inorganic Inks for Plastic Electronics Applications Jolke Perelaer

8.1 Introduction 8.1.1 Inkjet Printing

Inkjet printing of conductive precursor materials – usually metallic nanoparticles (NPs) or metal organic complexes – has been used as a relatively fast technique that might enable roll-to-roll (R2R) production [1–3]. However, the sintering step that is necessary to render the precursor compounds conductive typically requires >30 min and/or higher temperatures (>250 ◦ C). In particular, the long sintering time is not scalable to R2R production lines and the high sintering temperatures are not compatible with paper or common polymer foils that have a relatively low glass transition temperature (Tg ), such as polyethylene terephthalate (PET). This limits the choice to more expensive polymers. Both the temperature and the time required for sintering clearly need to be reduced, and this has, therefore, been the topic of research for many scientists over the past few years [1, 2, 4, 5]. 8.1.2 Printed Electronics

Inkjet printing of metal precursor inks has been used in many applications, such as interconnections for a circuitry on a printed circuit board [6], disposable displays and radio frequency identification (RFID) tags [7], organic thin-film transistors [8], and electrochromic devices [9]. Furthermore, printing large areas is also a possibility [10]. Even though there have been many successes in the inkjet printing of metal precursor inks as a rapid fabrication tool for plastic electronic applications, some challenges still remain. First, the processing temperature needs to be below the Tg of the substrate materials and the decomposition temperature of protecting materials, such as barriers or insulating materials. Second, the current resolution of inkjet printing is on the order of micrometers and, therefore, cannot be compared with the resolution Inkjet-based Micromanufacturing, First Edition. Edited by J. G. Korvink, P. J. Smith, and D.-Y. Shin. © 2012 Wiley-VCH Verlag GmbH & Co. KGaA. Published 2012 by Wiley-VCH Verlag GmbH & Co. KGaA.

112

8 Postprinting Processes for Inorganic Inks for Plastic Electronics Applications

of lithographic techniques, including soft-, photo-, and nanoimprint lithography, which have a resolution in nanometers [11]. The current industrial state-of-the-art feature size of 32 nm has been achieved as a consequence of an enormous financial and intellectual effort over the past 60 years. While this technology is needed for state-of-the-art, high-frequency circuits, the appeal of inkjet printing comes into play when the focus is on properties other than speed or feature density. The great advantage of inkjet printing is that it is a ‘‘maskless’’ process; it can quickly switch from one design to another without the need for a new set of expensive masks, which enables a more flexible processing flow [12]. It is also not limited to a few rigid substrates, such as silicon or gallium arsenide. Therefore, a trade-off exists between resolution and flexibility. Finally, conductivity values need to reach a certain application-dependant value. Typically, the obtained conductivity after the sintering step is only a fraction of the bulk metal conductivity. The conversion of the precursor ink into bulk material is affected by the processing temperatures, which are well below the melting temperature of bulk metal. This can be compensated for by printing multiple layers [13], which comes at the cost of more needed material and time. Therefore, another trade-off appears between the processing conditions and the feature’s conductivity. The first part of this chapter discusses the inkjet printing of conductive precursor ink, including inorganic NPs and metal–organic decomposition (MOD) inks and the advantage of using these materials. The second section introduces various selective postprinting processes to convert the precursor inks into conductive material in such a way that the substrate is not affected. Conventional sintering techniques, such as standard convection ovens, are compared with new, alternative, and more selective methods. Moreover, the mechanisms involved in the sintering process are discussed. Finally, the chapter concludes with a summary and some comments on the future of the field.

8.2 Inkjet Printing and Postprinting Processes of Metallic Inks 8.2.1 Choice of Metal

Metal-containing inks are preferred for inkjet printing of conductive features, such as contacts and interconnects, because of their superior conductivity values compared to conductive polymers [14]. Two main types of inks can be distinguished: the first is MOD ink [15], based on a metal salt dissolved in a suitable solvent. MOD inks have been used for inkjet printing since the late 1980s [16]. The second type of ink is a suspension of NPs, and is referred to as nanoparticle ink [17]. In contrast to MOD inks that are based on complexes, hence seen as solutions, NP inks contain metallic NPs with a diameter between 1 and 100 nm. It was found by Buffat and Borel in 1976 [18] that gold NPs with a diameter below 10 nm reveal a

8.2 Inkjet Printing and Postprinting Processes of Metallic Inks

Tm (°K)

m.p. bulk

1300

1000

500 300

0

50

(a)

100

150

200

Diameter (Å) 1.0

0.95 Tm To 0.90

Indium Lead Bismuth

0.85

Tin 0.80 (b)

0

0.05

0.10

0.15

0.20

0.25

Inverse radius (nm−1)

Figure 8.1 Influence of gold (a) and lead, bismuth, tin, and indium (b) particle diameter on their melting temperature. (Reprinted (a) and (b) with permission from Refs. [18] and [19], copyright 2011, American Physical Society and Elsevier Science, respectively.)

significant reduction in their melting temperature, as depicted in Figure 8.1 a, from their bulk melting temperature of 1064 ◦ C to well below 300 ◦ C when the diameter is below 5 nm. Allen et al. [19] showed later that this reduction in the melting temperature is also valid for other metals, including tin, lead, and bismuth. In a graph (Figure 8.1b) of the melting temperature against the reciprocal of the particle radius, the data exhibit near-linear relationships. It was also found that plates instead of spheres do not show a reduced melting temperature. This suggests that

113

114

8 Postprinting Processes for Inorganic Inks for Plastic Electronics Applications 2Ag+

Ag+

N2H4

Cu2+

Cu2+

Cu0

CuNP

2Ag+ + Cu0 → 2Ag0 + Cu2+

CucoreAgshell NP

Figure 8.2 Schematic illustration of a single Cu nanoparticle synthesis and the formation of a silver shell by a transmetalation reaction. The surface copper atoms serve as reducing agents for the silver ions. (Reprinted with permission from Ref. [21], copyright 2011, the Royal Society of Chemistry.)

the dependence of melting on size in particles is related to the internal hydrostatic pressure caused by the surface stress and by the large surface curvature of the particles, but not by the planar surfaces of platelets. Each type of ink has its advantages as well as disadvantages. For example, MOD inks are solutions that reduce nozzle clogging and do not require colloidal stabilizers. NP inks usually have a higher particle loading, are more widely available commercially, and have been reported to have lower contact resistances [8]. Ideally, a metal-containing ink would be cheap, easy to prepare, store, and inkjet print, and would give high conductivity values after deposition and postprocessing. The bulk resistivity of silver is the lowest, 1.59 × 10−8 m, followed by copper, 1.72 × 10−8 m, and then gold, 2.44 × 10−8 m, which makes silver the most preferred material to use. In terms of price, however, an ounce of gold costs about $1300, whereas an ounce of silver costs $27 and 1 ounce of copper is about 30 cents [20]. Despite its low price, copper has the disadvantage of oxidizing spontaneously in air, which makes handling of copper material rather complex. Magdassi et al. [21] encapsulated copper NP cores in silver shells, as depicted in Figure 8.2, which allowed them to use the ink in air. This approach appears to be stable as no oxides were found to have formed on the inkjet-printed copper patterns after several months. Gold, on the other hand, is almost prohibitively expensive. Currently, the use of silver has been reported the most, owing to the relatively simple synthesis of silver NPs [22]. Recently, Van Hest et al. reported an inkjet-printable formulation of a Ni-MOD ink, containing a Ni salt and a low-molar-mass complexing agent in a reducing solvent. This ink was printed into 80-μm-wide lines on a 200 ◦ C glass substrate [23]. Although this temperature is too high for common polymer foils, the choice of metal is of particular interest, since it exhibits a good conductivity, and its price is only a fraction of that of silver.

8.2 Inkjet Printing and Postprinting Processes of Metallic Inks

8.2.2 Postprinting Processes to Convert Inorganic Precursor Ink

After a metal-containing precursor ink has been deposited onto a substrate, both MOD and NP inks require an additional processing step to render the as-printed patterns conductive: a process called sintering. In the case of MOD inks, a conversion of the metal-containing complex usually takes place at a relatively low temperature below 200 ◦ C [24], although temperatures as low as 150 ◦ C have been reported as well [2]. After the conversion of the metal complex, metal NPs are created in situ, which need to sinter and form larger aggregates. When heating an NP ink, the NPs lose their organic shell and start showing conductance by direct physical contact. Conductivity only arises when metallic contact between the particles is present and a continuous percolating network is formed throughout the printed feature. An organic layer between the silver particles as thin as a few nanometers is sufficient to prevent electrons from moving from one particle to another [1]. When metal NPs are created, either by removal of the organic shell or by conversion of the metal complexes, larger particles form through Ostwald ripening; [25] a process by which the surface energy is reduced because of the particles’ large surface-to-volume ratio. The process of Ostwald ripening stalls when a particle diameter of approximately one and a half times their original size is reached, leaving behind a porous structure, and lower conductivity values than the bulk material are obtained. The transport of material between particles during Ostwald ripening is nonconvective and purely of a diffusive nature [26]. Larger particles are formed because of particle migration and coalescence. During sintering, the relative density increases with increased temperature or pressure but decreases with increased particle size [27]. Full densification, that is, closure of pores, can be obtained either with a fast diffusing gas or with a high pressure. Gamerith et al. found that, on sintering, the surface morphology for both the MOD and NP ink materials strongly depends on the sintering temperature as well as time. With increased sintering temperature, the crystallite size and homogeneity of silver-precursor-based films increased, while for the NP-based material an unfavorable growth of large isolated crystallites was observed [8]. When sintering metal-containing inks, two properties are important: first, the lowest temperature that can convert the ink into its conductive counterpart, which is mainly determined by the amount of organics present in the ink [28], and second, obtaining the lowest possible resistivity of the printed features at the lowest possible temperature. In order to achieve a low resistivity, sintering of the particles is required to transform the initially very small contact areas to thicker necks and, eventually, to a dense layer. High conductivities can then be obtained through the formation of large necks, hence densification, which decrease constriction resistance and eventually form a metallic crystal structure with a low number of grain boundaries.

115

116

8 Postprinting Processes for Inorganic Inks for Plastic Electronics Applications

8.2.3 Conventional Sintering Techniques

The conversion of nonconductive precursor inks has been reported mostly by simply applying heat. However, typical sintering temperatures are above 200 ◦ C [14], which is not suitable for common polymer substrates that have a relatively low glass transition temperature (Tg ), such as PET and polycarbonate [3]. In fact, only the expensive high-performance polymers, such as polyimide and polytetrafluoroethylene, can be used at high temperatures, which represent a drawback for implementation in large-area production and are not favorable in terms of costs. Lee et al. [29] have described an aqueous silver colloid dispersion, which is mainly prepared with environmentally friendly chemicals such as water and diethylene glycol. Although short times of only 10 min are necessary, high temperatures up to 300 ◦ C are required to obtain good conducting lines. Schubert and coworkers [1] have formulated an aqueous silver ink with a low curing temperature by using a very low amount of organic additives. The silver ink was prepared using silver powder from Mitsui and weakly adsorbing binding materials, such as poly(ethylene oxide) or poly(vinyl alcohol), instead of amines, amides, or mercapto groups, which usually are strong binding groups toward silver particles. The low content of organic additives already revealed conductivity at a temperature of 80 ◦ C. Similarly, low sintering temperatures have previously been reported by Huang et al. [4] for gold inks, again formulated using low-molar-mass organic additives. A serious drawback of using low content of binding materials is that the resulting ink is hard to inkjet print. Organic binders promote the inks’ printability and are often added to assure mechanical integrity and adhesion to the substrate [30]. Steric stabilization of the particles in solvents substantially screens van der Waals attractions and introduces steep steric repulsion between the particles at contact, which avoids agglomeration [31]. 8.2.4 Alternative and Selective Sintering Methods

As a first alternative, Allen et al. [32] have used a DC current as a selective sintering technique. This technique requires that the features are slightly conductive before sintering. The authors obtained good conductivity values up to 60% of bulk silver within a very short time of 2 μs and a power density of at least 100 nW μm−3 . The conductivity of the printed NP layer increases by more than 5 orders of magnitude during the sintering process. Furthermore, the authors demonstrated contactless sintering by using a high-voltage probe with an alternating field (100 V, 300 MHz) above a silver NP layer and a ground plate placed beneath, as shown schematically in Figure 8.3. Another selective sintering technique, which was developed by Reinhold et al. [33], is to expose the printed features to low-pressure argon plasma. This process decomposes the organic moieties around the NPs within the printed feature from

8.2 Inkjet Printing and Postprinting Processes of Metallic Inks

Nanoparticle layer

Substrate (a) 5 mm

5 mm

(b)

(c)

Figure 8.3 (a) Schematic representation of contactless AC sintering between a probe above the nanoparticle layer and a ground plate beneath the printing substrate. Infrared (λ < 10 μm) images of (b) pulse and (c) square-shaped (closed-loop) conductor

patterns with UDC = 5 V voltage applied over the conductors for visualization purposes. (Reprinted with permission from Ref. [32], copyright 2011, Institute of Physics Publishing.)

top to bottom, which can be followed by a growing skin layer in time. After a sufficient amount of sintering time, the as-printed features are converted into bulk material. The authors confirmed the formation of a skin layer by applying adhesive tape to the partially sintered printed structures, which removes the conductive top layer, while leaving the designated unsintered part of the material behind. Removing the upper layer revealed a characteristic bluish appearance of NPs, which did not show conductivity. The crust that was transferred to the adhesive tape, however, was conductive and showed similar resistance compared to the complete track before. Subsequent plasma processing of the sample again yielded conductivity. Figure 8.4 shows, on the left-hand side (I and II), scanning electron microscopy (SEM) images of the transition between two different regions. These regions are a result of a twofold application of the adhesive tape to a sample with repeated plasma sintering. The structural change in the ink is obvious in the lower left picture (II), where sintered and unsintered materials meet. The SEM images on the right-hand side show that the unsintered material looks smooth (V), while obvious grain growth has taken place during sintering (III and IV). The comparison of the microstructures of the samples plasma processed once and twice shows no noticeable difference. Using this technique, a conductivity of 30–40% of bulk silver was obtained. A third example of selective sintering is by using microwave radiation [34]. Typically, highly conductive materials, for example, metals, have a penetration depth of 1–2 μm at a microwave frequency of 2.45 GHz. It is believed that the conductive particle interaction with microwave radiation, that is, inductive

117

118

8 Postprinting Processes for Inorganic Inks for Plastic Electronics Applications

III

I 2× sintered

IV 1× sintered II Unsintered

V

Figure 8.4 Skin layer removal and microstructures after repetitive sintering. (Reprinted with permission from Ref. [33], copyright 2011, the Royal Society of Chemistry.)

coupling, is mainly based on Maxwell–Wagner polarization, which results from the accumulation of charge at the materials interfaces, electric conduction, and eddy currents [35]. In contrast to the relatively strong microwave absorption by the conductive particles, the polarization of dipoles in thermoplastic polymers below Tg is limited, which makes the polymer foil’s skin depth almost infinite, hence transparent, to microwave radiation. Exposing metallic NPs to microwaves not only reveals that a sintering process is taking place but also decreases the sintering time by a factor of 20, while conductivity values are similar when using thermal sintering. By the application of conductive antenna structures around features that exhibit a small conductance, the exposure times could be reduced to only 1 s, as described by Perelaer et al. [36] (Figure 8.5). In fact, the antenna structures do not require a physical contact with the unsintered features, which makes recycling of the antennae possible. This process can then be implemented into R2R production. After sintering, the features revealed a conductivity of up to 34% when compared to the bulk silver value. As a final example, photonic sintering is listed here, which uses a high-intensity white light beam to sinter metal precursor inks (Figure 8.6). This relatively new technique is, for example, commercialized by NovaCentrix [37]. The tool produces very short pulses of white light with a maximum of 100 kW cm−2 to deliver the energy to the target material. By careful control of the duty cycle of the lamps, the energy delivery to the inks can be stopped at the right moment, that is, just before enough energy is delivered to convert the ink into its conductive form and to prevent substrate damage. Conductivity values of 25–30% of bulk silver were obtained with translational speeds of up to 100 m min−1 . Other techniques that are reported in the open literature and have been used for sintering include exposure to UV radiation [38], pulse electric current sintering [39], high-temperature plasma sintering [40], and LASER sintering [41].

8.2 Inkjet Printing and Postprinting Processes of Metallic Inks

Inkjet printed line

5 mm

2 mm

(a)

Electrodes/antennae

105 A = 0 mm2 A = 20 mm2

Resistance (Ω)

104

A = 31 mm2 A = 44 mm2

103 102 101 100 0

10

20

(b)

30

40

50

60

Time (s)

Figure 8.5 Schematic representation of the printed template onto a polyethylene naphthalate (PEN) substrate (a), with four silver antennae in gray and a single silver line inkjet printed on top of the antennae in black. Influence of the total surface of the

four electrodes on the template for an initial line resistance of 100  (b) on microwave flash exposure for 1–60 s. (Reprinted with permission from Ref. [36], copyright 2011, Wiley-VCH Verlag GmbH & Co.KGaA.)

8.2.5 Room-Temperature Sintering

In recent years, much research has been performed to sinter metallic inks at room temperature. Magdassi et al. [42] discovered that silver NPs behave as soft particles when they come into contact with oppositely charged polyelectrolytes and undergo a spontaneous coalescence process at room temperature. Triggered by these findings, the authors have inkjet printed a solution containing the cationic

119

120

8 Postprinting Processes for Inorganic Inks for Plastic Electronics Applications

Figure 8.6 Roll-to-roll photonic sintering tool from NovaCentrix. (Reprinted with permission from PulseForgeTM brochure, http://www.novacentrix.com.)

(a)

50 nm

50 nm

(b)

(c) Figure 8.7 (a) Schematic illustration showing what happens when a droplet of poly(diallyldimethylammonium chloride) (PDAC) solution is printed on silver NP array. (b–d)Scanning electron microscopy (SEM) image of a printed drop zone (c) and

(d) the magnified images of NP arrays after contact with PDAC outside (b) and inside (d) the droplet zone. (Reprinted with permission from Ref [42], copyright 2011, the American Chemical Society.)

polymer poly(diallyldimethylammoniumchloride) onto an as-printed film of silver NPs that are stabilized by poly(acrylic acid), which lead to the sintering of silver NPs and the formation of conductive films without further heating, as depicted in Figure 8.7. The obtained conductivity was approximately 20% compared to bulk silver.

8.3 Conclusions and Outlook

Owing to the sintering of NPs at room temperature, the formation of conductive patterns on polymer substrates with a low thermal stability and even on paper is possible. The authors demonstrate the usability of this process by the fabrication of a transparent PET-based electroluminescent device. Another room-temperature sintering method uses UV light to create ‘‘latent images’’ in inkjet-printed MOD ink, which is then chemically reduced [43]. The process is rapid and results in conductivities that are 10% of bulk silver. An interesting point while using MOD inks is that they generate NPs in situ. However, they often require heat to remove the unwanted organic component of the salt. Instead of inkjet printing, a readymade NP or MOD ink is used and subsequently postprocessed after pattering; this is preferable for creating conductive materials on demand and in situ on the substrate. The above technique is called reactive inkjet printing [44, 45]. Calvert et al. [46] have reported the preparation of conductive copper lines by using reactive inkjet printing. Conductive copper lines were directly written on paper through inkjet printing of a copper salt and sodium borohydride (NaBH4 ) as a reducing agent sequentially from two separate compartments of a multicolor HP cartridge. After successive printing, the obtained conductivity was a small percentage of bulk copper. Reactive inkjet printing further reduces the number of processing steps and may become revolutionary, although it is currently applicable only at the laboratory scale.

8.3 Conclusions and Outlook

This chapter provided an overview of how inkjet printing in combination with metal-containing inks can be used for the fabrication of conductive features on polymer substrates for printed electronics applications. The main obstacle is to convert the metal from its nonconductive state in the ink to a final conductive print. Typically, heat is required for the conversion, which, however, is not compatible with common, temperature-sensitive polymer foils. The heat is used either to remove the organic moieties that are present around the NPs to stabilize the NPs against agglomeration or to burn off the organic component of the metal salt used in MOD inks. In recent years, the main focus of research has been on decreasing the sintering temperatures as well as times in order to obtain the highest conductivity compared to the bulk metal values in a short duration. For NP inks, a reduction in the sintering temperature can be achieved by lowering the amount of organic additives that are present in the ink. However, when only small amounts of stabilizing materials are present in the ink, the ink becomes more difficult to print. Therefore, a balance exists between printing and sintering a metal-containing ink that needs to be found in order to achieve stable, good printing results as well good conductivity values. Another concern when using metal-containing inks is the price of raw metals. Although gold is among the best conductors, its price is extremely high compared

121

122

8 Postprinting Processes for Inorganic Inks for Plastic Electronics Applications

to other metals. However, although copper, nickel, and aluminum may cost only a fraction of the price of gold, their sensitivity toward oxygen makes the material hard to handle. Therefore, the use of silver has been mostly reported. Currently, promising strides have been made with inkjet-printable, air-stable Cu/Ag core/shell NPs. When using polymer foils, which are favored as they reduce production costs and enable R2R productions with a low glass transition temperature, the temperature during processing cannot exceed the polymers’ Tg or the foil will be permanently deformed. A number of alternative sintering methods that do not use direct heat have been reported during the last years. Rapid electrical sintering and microwave exposure are very fast sintering techniques, where conductivity values between 34 and 60% of bulk silver can be obtained within 1 s or less. The features, however, require a certain conductivity before the final flash sintering can take place successfully. Plasma sintering, that is, exposure to low-pressure argon plasma, shows an evolution starting from a sintered top layer into bulk material and yielding a conductivity up to 40% in 15–30 min. Photonic sintering uses high-energy white light pulses to convert the precursor materials in milliseconds to conductive features and is a promising candidate that can be used in R2R production for printed electronic applications. Recent developments have enabled the conversion of metal inks at room temperature. This was done either by a physical reaction, whereby the negatively charged stabilizer is removed simply by overprinting a cationic polymer solution, or by reactive inkjet printing, which couples reducing agents with metal precursors in a single print run. The combination of low-temperature and flash sintering techniques may be explored more in the near future, which ultimately makes fast processing and a good conductivity possible for R2R production processes.

Acknowledgments

The European Community’s Seventh Framework Programme (FP7/2007–2013) under grant agreement no 248816 is greatly acknowledged for financial support.

References 1. Perelaer, J., de Laat, A.W.M., Hendriks,

C.E., and Schubert, U.S. (2008) Inkjet-printed silver tracks: low temperature curing and thermal stability investigation. J. Mater. Chem., 18 (28), 3209–2115. 2. Smith, P.J., Shin, D.-Y., Stringer, J.E., Reis, N., and Derby, B. (2006) Direct ink-jet printing and low temperature

conversion of conductive silver patterns. J. Mater. Sci., 41 (13), 4153–4158. 3. Perelaer, J., Hendriks, C.E., de Laat, A.W.M., and Schubert, U.S. (2009) One-step inkjet printing of conductive silver tracks on polymer substrates. Nanotechnology, 20 (16), 165303. 4. Molesa, S., Redinger, D.R., Huang, D.C., and Subramanian, V. (2003) High-quality inkjet-printed multilevel

References

5.

6.

7.

8.

9.

10.

11.

12.

13.

interconnects and inductive components on plastic for ultra-low-cost RFID applications. Mater. Res. Soc. Symp. Proc., 769, H8.3.1–H8.3.6. Lu, C.-A., Lin, P., Lin, H.-C., and Wang, S.F. (2007) Characterization of the low-curing-temperature silver paste with silver 2-ethylhexanoate addition. Jpn. J. Appl. Phys., 46 (1), 251–255. Szczech, J.B., Megaridis, C.M., Gamota, D.R., and Zhang, J. (2002) Fine-line conductor manufacturing using drop-on-demand PZT printing technology. IEEE Trans. Electron. Packag., 25 (1), 26–33. Subramanian, V., Fr´echet, J.M.J., Chang, P.C., Huang, D., Lee, J.B., Molesa, S.E., Murphy, A.R., Redinger, D.R., and Volkman, S.K. (2005) Progress toward development of all-printed RFID tags: materials, processes, and devices. Proc. IEEE, 93 (7), 1330–1338. Gamerith, S., Klug, A., Schreiber, H., Scherf, U., Moderegger, E., and List, E.J.W. (2007) Direct ink-jet printing of Ag-Cu nanoparticle and Ag-precursor based electrodes for OFET applications. Adv. Funct. Mater., 17 (16), 3111–3118. Argun, A.A., Aubert, P.H., Thompson, B.C., Schwendeman, I., Gaupp, C.L., Hwang, J., Pinto, N.J., Tanner, D.B., MacDiarmid, A.G., and Reynolds, J.R. (2004) Multicolored electrochromism polymers: structures and devices. Chem. Mater., 16 (23), 4401–4412. Calvert, P. (2001) Inkjet printing for materials and devices. Chem. Mater., 13 (10), 3299–3305. Menard, E., Meitl, M.A., Sun, Y.G., Park, J.U., Shir, D.J.L., Nam, Y.S., Jeon, S., and Rogers, J.A. (2007) Micro- and nanopatterning techniques for organic electronic and optoelectronic systems. Chem. Rev., 107 (4), 1117–1160. Tekin, E., Smith, P.J., and Schubert, U.S. (2008) Inkjet printing as a deposition and patterning tool for polymers and inorganic particles. Soft Matter, 4 (4), 703–713. Van Osch, T.H.J., Perelaer, J., De Laat, A.W.M., and Schubert, U.S. (2008) Inkjet printing of narrow conductive

14.

15.

16.

17.

18.

19.

20. 21.

22.

23.

24.

tracks on untreated polymeric substrates. Adv. Mater., 20 (2), 343–345. Perelaer, J., Smith, P.J., Mager, D., Soltman, D., Volkman, S.K., Subramanian, V., Korvink, J.G., and Schubert, U.S. (2010) Printed electronics: the challenges involved in printing devices, interconnects, and contacts based on inorganic materials. J. Mater. Chem., 20 (39), 8446–8453. Dearden, A.L., Smith, P.J., Shin, D.-Y., Reis, N., Derby, B., and O’Brien, P. (2005) A low curing temperature silver ink for use in ink-jet printing and subsequent production of conductive tracks. Macromol. Rapid Commun., 26 (4), 315–318. Teng, K.F. and Vest, R.W. (1987) Liquid ink jet printing with mod inks for hybrid microcircuits. IEEE Trans. Compon. Hybrid., 10 (4), 545–549. Meier, H., Loffelmann, U., Mager, D., Smith, P.J., and Korvink, J.G. (2009) Inkjet printed, conductive, 25 μm wide silver tracks on unstructured polyimide. Phys. Status Solidi A, 206 (7), 1626–1630. Buffat, P. and Borel, J.-P. (1976) Size effect on melting temperature of gold particles. Phys. Rev. A, 13 (6), 2287–2298. Allen, G.L., Bayles, R.A., Gile, W.W., and Jesser, W.A. (1986) Small Particle Melting of Pure Metals. Thin Solid Films, 144 (2), 297–308. http://www.metalprices.com (accessed 27 January 2011). Grouchko, M., Kamyshny, A., and Magdassi, S. (2009) Formation of air-stable copper-silver core-shell nanoparticles for inkjet printing. J. Mater. Chem., 19 (19), 3057–3062. Schmid, G. (2004) Nanoparticles: From Theory to Applications, Wiley-VCH Verlag GmbH, Weinheim, pp. 185–238. Van Hest, M.F.A.M., Curtis, C.J., Miedaner, A., Pasquarelli, R.M., Kaydanova, T., Hersh, P., and Ginley, D.S. (2008) Direct-write contacts: metallization and contact formation. 33rd IEEE Photovoltaic Specialists Conference, San Diego, CA, 04922798. Wu, Y., Li, Y., and Ong, B.S. (2007) A simple and efficient approach to a printable silver conductor for printed

123

124

8 Postprinting Processes for Inorganic Inks for Plastic Electronics Applications

25. 26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

electronics. J. Am. Chem. Soc., 129 (7), 1862–1863. Ostwald, W. (1896) Lehrbuch der Allgemeinen Chemie, Vol. 2, Part 1, Leipzig. Goeke, R.S. and Datye, A.K. (2007) Model oxide supports for studies of catalyst sintering at elevated temperatures. Top. Catal., 46 (1–2), 3–9. Kang, S.-J.L. (2005) Sintering: Densification, Grain Growth, and Microstructure, Elsevier Butterworth-Heinemann, Burlington, pp. 37–77. Liang, L.H., Shen, C.M., Du, S.X., Liu, W.M., Xie, X.C., and Gao, H.J. (2004) Increase in thermal stability induced by organic coatings on nanoparticles. Phys. Rev B., 70 (20), 205419. Lee, H.-H., Chou, K.-S., and Huang, K.-C. (2005) Inkjet printing of nanosized silver colloids. Nanotechnology, 16 (10), 2436–2441. Lovinger, A.J. (1979) Development of electrical conduction in silver-filled epoxy adhesives. J. Adhesion, 10 (1), 1–15. B¨onnemann, H. and Richards, R.M. (2001) Nanoscopic metal particles – synthetic methods and potential applications. Eur. J. Inorg. Chem., 10, 2455–2480. Allen, M.L., Aronniemi, M., Mattila, T., Alastalo, A., Ojanper¨a, K., Suhonen, M., and Sepp¨a, H. (2008) Electrical sintering of nanoparticle structures. Nanotechnology, 19 (17), 175201. Reinhold, I., Hendriks, C.E., Eckardt, R., Kranenburg, J.M., Perelaer, J., Baumann, R.R., and Schubert, U.S. (2009) Argon plasma sintering of inkjet printed silver tracks on polymer substrates. J. Mater. Chem., 19 (21), 3384–3388. Perelaer, J., De Gans, B.-J., and Schubert, U.S. (2006) Ink-jet printing and microwave sintering of conductive silver tracks. Adv. Mater., 18 (16), 2101–2104. Agrawal, D. (2006) Microwave sintering of ceramics, composites and metallic materials, and melting of glasses. Trans. Ind. Ceram. Soc., 65 (6), 129–144. Perelaer, J., Klokkenburg, M., Hendriks, C.E., and Schubert, U.S. (2009) Microwave flash sintering of inkjet-printed

37.

38.

39.

40.

41.

42.

43.

44.

silver tracks on polymer substrates. Adv. Mater., 21 (47), 4830–4834. Farnsworth, S., Rawson, I., Martin, K., Schroder, K.A., and Pope, D. (2009) Evaluation of tool for processing metallic conductive inks. Proceedings of the 42nd International Symposium on Microelectronics, San Jos´e, CA, pp. 1012–1015. Jahn, S.F., Blaudeck, T., Baumann, R.R., Jakob, A., Ecorchard, P., Ruffer, T., Lang, H., and Schmidt, P. (2010) Inkjet printing of conductive silver patterns by using the first aqueous particle-free mod ink without additional stabilizing ligands. Chem. Mater., 22 (10), 3067–3071. Xie, G.Q., Ohashi, O., Yamaguchi, N., and Wang, A.R. (2003) Effect of surface oxide films on the properties of pulse electric-current sintered metal powders. Metall. Mater. Trans. A, 34A (11), 2655–2661. Groza, J.R., Risbud, S.H., and Yamazaki, K. (1992) Plasma activated sintering of additive-free AlN powders to near-theoretical density in 5 minutes. J. Mater. Res., 7 (10), 2643–2645. Ko, S.H., Pan, H., Grigoropoulos, C.P., Luscombe, C.K., Fr´echet, J.M.J., and Poulikakos, D. (2007) All-inkjet-printed flexible electronics fabrication on a polymer substrate by low-temperature high-resolution selective laser sintering of metal nanoparticles. Nanotechnology, 18 (34), 345202. Magdassi, S., Grouchko, M., Berezin, O., and Kamyshny, A. (2010) Triggering the sintering of silver nanoparticles at room temperature. ACS Nano, 4 (4), 1943–1948. Valeton, J.J.P., Hermans, K., Bastiaansen, C.W.M., Broer, D.J., Perelaer, J., Schubert, U.S., Crawford, G.P., and Smith, P.J. (2010) Room temperature preparation of conductive silver features using spin-coating and inkjet printing. J. Mater. Chem., 20 (3), 543–546. Kr¨ober, P., Delaney, J.T., Perelaer, J., and Schubert, U.S. (2009) Reactive inkjet printing of polyurethanes. J. Mater. Chem., 19 (29), 5234–5238.

References 45. Delaney, J.T., Liberski, A.R., Perelaer, J.,

and Schubert, U.S. (2010) Reactive inkjet printing of calcium alginate hydrogel porogens-a new strategy to open-pore structured matrices with controlled geometry. Soft Matter, 6 (5), 866–869.

46. Li, D.P., Sutton, D., Burgess, A.,

Graham, D., and Calvert, P.D. (2009) Conductive copper and nickel lines via reactive inkjet printing. J. Mater. Chem., 19 (22), 3719–3724.

125

127

9 Vision Monitoring Kye-Si Kwon

9.1 Introduction

Vision-based inkjet measurement techniques are discussed in this chapter. Inkjet technology has been used as a manufacturing tool in printed electronics applications such as large area displays, radio frequency identification (RFID), and printed circuit boards (PCBs). As inkjet applications broaden, various types of jetting materials are required to be precisely dispensed from inkjet heads. Droplet behavior from the inkjet head must be measured properly to evaluate and control jetting behavior. Vision-based measurement techniques are widely used in inkjet-based manufacturing systems, since physical insight into jetting behavior can be obtained from visual images. The droplet jetting speed and droplet volume are the most frequently measured jetting performance parameters obtained from droplet images. A high-speed camera capable of recording up to 160 000 images per second can be used in an inkjet vision measurement system [1]. Using a high-speed camera, drop-by-drop variances in jetting behavior can be measured. However, there are two drawbacks in using high-speed cameras for inkjet measurements: they are expensive, and real-time image processing for measuring jetting behavior can be difficult. Therefore, standard charge-coupled device (CCD) cameras are widely used to measure jetting behavior [2–7]. By using light-emitting diode (LED) lights synchronized to the firing signal, droplet images appear to be frozen in the acquired CCD camera image. As a result, the image processing used to measure ink droplet location and size can be straightforward. In this chapter, vision-based measurements using standard CCD cameras for inkjet applications are discussed in detail.

9.2 Measurement Setup

Figure 9.1 shows a schematic of a vision monitoring system used to measure ink jetting performance from a piezo inkjet head. It should be noted that the vision Inkjet-based Micromanufacturing, First Edition. Edited by J. G. Korvink, P. J. Smith, and D.-Y. Shin.  2012 Wiley-VCH Verlag GmbH & Co. KGaA. Published 2012 by Wiley-VCH Verlag GmbH & Co. KGaA.

9 Vision Monitoring

Inkjet head

CCD camera

Driver

LED LED driver Digital pulses

Waveform generator

PC

Figure 9.1

Schematic of vision monitoring system.

Rising part Voltage

128

Falling part Dwell part

Time Figure 9.2

Typical driving waveform for inkjet head [8, 9].

monitoring system shown in Figure 9.1 can easily be modified to measure the droplet jetting behavior from drop-on-demand (DOD) inkjet heads other than piezo inkjet printheads. A proper waveform must be used to drive a piezo inkjet head. Figure 9.2 shows a typical waveform. The driving waveform for a piezo inkjet head can significantly alter jetting behavior. Design methods for waveform parameters have been discussed in the literature [8, 9]. The waveform data, stored in the memory of the waveform generator, waits for trigger signals to generate the desired waveform. A high-voltage power amplifier is used to amplify the waveform voltage generated from the waveform generator and drive the piezo actuator for jetting. The diameter of an ink droplet from the inkjet head can vary from 20 to 100 µm depending on the jetting conditions and nozzle diameter. Therefore, a zoom lens with an amplification factor of about 4–8 × must be used to magnify the image of the droplet. To obtain a frozen droplet image, the LED light is synchronized with respect to the jetting signal [2]. Two digital pulse trains are typically used for the synchronization, as shown in Figure 9.3 [2].

9.2 Measurement Setup

(a)

(b) Trigger delay

Figure 9.3 Two digital pulse trains for droplet image measurement. (a) Trigger signal for driving waveform and (b) digital pulse train for LED light control.

The first digital pulse train, shown in Figure 9.3a, is used as a trigger signal to generate the jetting waveform. The second pulse train, shown in Figure 9.3b, is used to control the LED light. The second pulse is triggered from the first pulse. Here, by adjusting the time delay of the second pulse with respect to the first pulse, the droplet image at the delayed time can appear to be frozen in the acquired image. Here, the duty ratio (on-time) of the second pulse, shown in Figure 9.3b, can change the image brightness by varying the light intensity of the LED. For example, Figure 9.4 illustrates the use of the digital pulse duty ratio for LED light control. The digital pulse for the LED shown in Figure 9.4a has a longer on-time than the digital pulse shown in Figure 9.4b. If the on-time for the LED light increases, then the image brightness increases. However, we note that droplet movement can result in droplet image blur if the light flash is not short enough. The image will be brighter, but there will be motion blur if the digital pulse shown in Figure 9.4a is used rather than the digital pulse shown in Figure 9.4b. Thus, there is a trade-off relationship in the LED pulse duty ratio in terms of image brightness and motion blur. On

Off (a) On

Off (b) Figure 9.4 Digital pulses for LED light control: (a) high-duty ratio and (b) low-duty ratio.

129

130

9 Vision Monitoring

9.3 Image Processing

Using the LED light delay adjustment, a frozen image at a trigger delay time can be obtained from the CCD camera, as shown in Figure 9.5a. The acquired image is then processed to extract inkjet drop information. Normally, the acquired images from the CCD camera have 8 bit image pixels that can appear as gray images with values ranging from 0 to 255 according to the brightness of the image. Most image processing techniques focus on extracting droplet location and size information from the droplet image. In general, the most common method for image processing is based on the binary droplet image. Note that a droplet image is likely to be darker than the background image. As a result, to extract droplet information based on converted binary images, it is preferred that the droplet has a value of 1 and the background has a value of 0. Therefore, we recommend that image values that are higher than the threshold value for binary conversion be mapped to 0, and image values lower than the threshold value be mapped to 1 in the converted binary image, as shown in Figure 9.5b. Note that in addition to the ink droplet image, other structures, such as parts of the printhead or fixtures for the head, can have a value of 1, as shown in Figure 9.5b. Also, the converted binary image of a droplet may have a hole in the center when the center of the droplet image is brighter than the outer part of droplet image, as shown in Figure 9.5b. An unfilled hole in the droplet binary image can result in size errors if the size of the droplet is calculated by counting the number of pixels with an image value of 1. To solve these problems, two steps can be taken to extract the appropriate information from droplet images. First, the region of interest (ROI) on the image can be used such that only the ROI area in the image is analyzed by excluding other parts of image that have nondroplet structures, as shown in Figure 9.6 [2]. Next, the binary image can be modified by filling the hole in the droplet image to measure the droplet information accurately, as shown in Figure 9.7.

Image value: 1

Droplet location Size Image value: 0

(a)

Figure 9.5 age.

(b)

Binary conversion of a gray droplet image. (a) Gray image and (b) binary im-

9.3 Image Processing

x

y

ROI

Figure 9.6 Region of interest (ROI).

Image value: 1

Image value: 0

Figure 9.7 Processing of binary image: morphology.

From an analysis of binary images, the number of droplets, including main drops and satellites, can be obtained. The location of an ink droplet is identified in terms of pixel locations in the CCD camera image. Assuming that the jetting direction is downward, the pixel location in the y-direction Py can be used for the droplet location. Note that the pixel location is not a real distance. Thus, the identified pixel distance of the droplet location needs to be multiplied by a converting factor F (in units of meter per pixel) to obtain the actual distance. For the droplet jetting speed measurement, droplet image locations must be obtained from two acquired droplet images at LED triggering delay times t1 and t2 with respect to the jetting signal. Then, jetting speed v can be calculated as v=

Py2 − Py1 ×F t2 − t 1

(9.1)

where Py2 and Py1 give the identified droplet center location in terms of pixel numbers in the y-direction at t1 and t2 , respectively, as shown in Figure 9.8. In addition to jetting speed, the droplet volume is also a frequently measured vision measurement. To measure droplet volume, the three-dimensional (3D) shape of the droplet must be assumed, whereas measurement of the jetting speed

131

132

9 Vision Monitoring

Py1 Py 2

(a) Figure 9.8

(b) Jetting speed measurement from two images: (a) t1 and (b) t2 .

requires only the location of the droplet. On the basis of an assumption of the 3D droplet shape (e.g., spherical), the droplet volume can be calculated from the two-dimensional (2D) drop image. We note that the droplet volume could be difficult to measure if there is a long tail in the droplet. Recently, Hutchings et al. [4] have developed image processing techniques to obtain the volume of ink droplets with ligaments. The sliced width of a ligament of a droplet image in the lateral direction was used for their droplet volume measurement, as shown in Figure 9.9. Note that the diameter of the sliced cylinder, which can be identified from the image as the horizontal length at a sliced location, varies along the jetting direction, as shown in Figure 9.9. The ink droplet volume can be calculated by a summation of sliced volumes of cylindrically shaped elements. Using this method, the volume of a nonspherical droplet can be calculated. The droplet volume should be measured correctly because the amount of functional material is related to device performance when inkjet technology is used in printed electronics applications. However, it is difficult to measure droplet volume accurately because the CCD camera has limited pixel resolution. As a result, the measured droplet size and droplet locations may have errors. To understand the pixel resolution effect, the droplet image shown in Figure 9.6 was magnified as shown in Figure 9.10. Figure 9.10 shows the uncertainty in determining droplet size due to optical resolution. In addition to pixel resolution, the measured droplet size can be significantly affected in the gray-to-binary image conversion process. As shown in Figure 9.10, there is smooth transition of image value near the boundary between the droplet

1 Pixel slice Slice diameter

Figure 9.9

Droplet volume measurement with a long ligament [4].

9.3 Image Processing

One pixel Figure 9.10 Magnified droplet gray image.

Image value: 1

Image value: 0 (a)

(b)

Figure 9.11 Threshold effects on droplet size. (a) Threshold value of 25 and (b) threshold value of 75.

image and background. The smooth transition of image value depends on various conditions such as lighting brightness and lens focus. As a result, the identified droplet size will differ according to the threshold value for binary conversion. Figure 9.11 shows binary images that were converted from the gray image in Figure 9.10, where two different threshold values were used for comparison. When a threshold value of 25 was used for binary conversion, the diameter of the droplet can be identified as 20 pixels. On the other hand, a threshold value of 75 can result in a diameter of 24 pixels. Therefore, it is difficult to measure an accurate inkjet droplet volume (or size) from the images. As shown in Figure 9.11, the errors related to binary conversion are much larger than the 1-pixel resolution of the CCD camera. To reduce measurement errors, a higher zoom magnification with an optical lens may be used. If the 1-pixel size is significantly small compared to the droplet, the pixel-related error can be minimized. However, high magnification might not be good for drop formation measurements, since it could narrow the image view by focusing on a very small area. A high-resolution camera can be used to reduce the pixel errors. However, there are cost issues in using a high-resolution camera.

133

9 Vision Monitoring

9.4 Jetting Speed Measurement

Droplet jetting speed and droplet volume are the parameters that are most frequently measured from vision images. As previously mentioned, the measured droplet size can vary during the process of binary conversion. These binary conversion errors are likely to be the same during the entire drop formation, assuming that measurement conditions such as lighting, lens focus, and threshold value are unchanged. Thus, the errors can be cancelled out in calculating jetting speed because of the subtraction of two droplet locations to calculate the travel distance during the two timings. As a result, the measured jetting speed of the droplet is considered to be more reliable than the measured droplet volume. Jetting speed is frequently used to determine the optimal jetting waveform by using the dwell time and jetting speed relationship [9]. Figure 9.12 shows typical experimental results for the dwell time and jetting speed relationship when the typical waveform shown in Figure 9.2 is used. To obtain the dwell time and jetting speed relationship, waveform parameters such as rising time, falling time, and driving voltage were set as fixed values for the experiment. Then, the jetting speed was measured by varying the dwell time. A waveform using a dwell time that results in maximum jetting speed is recommended for the driving inkjet head [9]. However, it should be noted that the measured jetting speed may be affected by droplet jetting behavior. As the jetting speed increases, the inkjet droplet does not remain spherical because the droplet is likely to have ligaments and satellites. During drop formation, a satellite can be merged into the main droplet, or the long ligament can change its shape into a single spherical droplet. The jetting speed of the droplet is likely to be affected by such a drop formation. As a result, it may 3 2.5 Jetting speed (m s−1)

134

2

Efficient dwell time

1.5 1 0.5 0

0

5

10

15

20

25

30

35

Dwell time (µs) Figure 9.12

Relationship between jetting speed and dwell time.

40

9.4 Jetting Speed Measurement

be difficult to design an efficient waveform from the jetting speed and dwell-time relationship when ligaments and satellites exist. Most previous measurement techniques for jetting speed used the two timings shown in Figure 9.8 [2, 3]. We note that there are two major drawbacks in the conventional speed measurement method using Eq. (9.1). First, it cannot measure the jetting speed variation properly during drop formation and the measured jetting speed can differ according to the selection of the two timings. Thus, it is very difficult to define a representative jetting speed via the conventional measurement technique. Second, it is difficult to understand the relative jetting speed of the main droplet with respect to satellites during drop formation. The conventional method is focused on the jetting speed of main droplet, neglecting the jetting speeds of satellites. However, it is important to measure the relative jetting speed of the main droplet with respect to the satellites. This is because the relative jetting speed provides useful information as to whether or not the satellite will merge with the main droplet. Satellites that do not merge with the main droplet could generate placement errors during printing. For a better understanding of jetting speed measurement issues, jetting images were acquired over time, as shown in Figure 9.13. Note that some of the droplets measured at different times were not spherical. In the case of a nonspherical inkjet droplet, the droplet location at the lowest point in the jetting direction was used instead of the center location for the measurement of jetting speed. Table 9.1 shows the measured jetting speeds, which were calculated from two images at two different times. As shown in Table 9.1, the measured jetting speeds differed over time. The jetting speed variation according to the selection of two timings is normally larger than the measurement errors due to pixel resolution. The measurement error of jetting speed is related to pixel resolution and the time interval, and is less related to the threshold value for binary image conversion. To roughly estimate the measurement errors, the droplet images shown in Figure 9.13 were used to evaluate such errors. In the figure, the pixel resolution and time interval were 1.4 µm and 40 µs, respectively. Then, the jetting speed uncertainty due to pixel resolution was calculated to be about 0.035 m s−1 . The speed variation (1.6–2.5 m s−1 ) during the drop formation given in Table 9.1 is much larger than the errors due to pixel resolution. Since the jetting speed varies during the drop formation, as shown in Table 9.1, it is difficult to define a representative jetting speed. Consequently, the conventional Table 9.1

Case 1 Case 2 Case 3 Case 4

Jetting speeds using conventional methods. Time 1 (µs)

Time 2 (µs)

Jetting speed (m s−1 )

60 140 180 220

100 180 220 260

2.5 1.6 2 2.3

135

136

9 Vision Monitoring

60 µs

100 µs

140 µs

260 µs 180 µs Figure 9.13

220 µs

Jetting images at different times.

jetting speed measurement method clearly has a limitation in defining the speed. To overcome this limitation, an instantaneous jetting speed curve was recently proposed such that jetting behavior can be fully understood during drop formation [10]. The instantaneous jetting speed measurement differs from conventional methods in that jetting speeds are measured during a short period of time. Also, many jetting speeds over time in a sequential order were measured to understand jetting speed variation during drop formation. Similarly, many sequential images have been used in attempts to explain drop formation. Dong and Carr [5] used the so-called DOD drop formation curve, which is discussed in detail later in this section, to explain inkjet drop formation such as satellite and ligament behavior. Image processing techniques used to measure DOD drop formation curves and instantaneous jetting speed curves are based on binary images [10, 11]. From a binary image converted from a gray image, the maximum and minimum locations of the kth droplet in the y-direction, denoted as Pkmax (td ) and Pkmin (td ), can be obtained, as shown in Figure 9.14. The superscripts ‘‘max’’ and ‘‘min’’ denote maximum and minimum locations in the jetting direction, respectively. The DOD drop formation curve was updated by adding the calculated maximum and minimum locations of each droplet Pkmax (td ) and Pkmin (td ), respectively, to the graph before acquiring the next sequential image, as shown in Figure 9.15a. The instantaneous jetting speeds of each droplet, Vkmax (td ) and Vkmin (td ) of the kth droplet, can be obtained using the following relations:

9.4 Jetting Speed Measurement Figure 9.14 Droplet locations in a droplet image.

min

P2

max

P2

min

P1

max

P1

Pkmax (td ) − Pkmax (td −
td ) and
td Pmin (td ) − Pkmin (td −
td ) k = 1, 2, . . . , n Vkmin (td ) = k
td

Vkmax (td ) =

(9.2)

where
td is the incremental time of trigger delay between the two consecutive images, and Pkmax (td ) − Pkmax (td −
td ) and Pkmin (td ) − Pkmin (td −
td ) are the travel distance from the nozzle of the maximum and minimum locations of the kth droplet, respectively, during the time duration
td . In most printing applications, the maximum location of droplet Pkmax (td ) is important, since it is placed on the substrate first. Therefore, the behavior of the jetting speed at the maximum location Vkmax (td ) rather than Vkmin (td ) needs to be measured. From the measured curves shown in Figure 9.15, the drop formation can be analyzed as follows: Phase 1 (from 30 to 110 µs): In phase 1, part of the droplet is extruded from the nozzle. During the jetting process of phase 1, the instantaneous jetting speed of the extruded part, Vkmax (td ) is significantly reduced from 4.0 to 1.9 m s−1 . This effect may be due to the viscoelasticity of the fluids: the still-attached fluid at the nozzle may pull back the extruded fluid from the nozzle before the fluid is pinched off from the nozzle. Here, the time required for the pinch-off, and the velocity profile before the pinch-off, can be used to evaluate the viscoelastic effect of ink on jetting. 2) Phase 2 (from 110 to 130 µs): In phase 2, after pinch-off at 110 µs, the ink droplet is in a condition of free-flying jetting but still has a ligament. The instantaneous jetting speed of the leading end, V1max (td ), was about 1.6–1.8 m s−1 , as shown in Figure 9.15b. The jetting speed variation at the free-flying condition was small compared to the jetting speed variation in phase 1. 3) Phase 3 (from 130 to 180 µs): In phase 3, the single droplet is separated into a spherical main droplet and a lengthy satellite with a ligament length of 120 µm

1)

137

9 Vision Monitoring

Phase 2 Phase 1

Phase 3

Phase 4

0 Distance from nozzle (µm)

min

(a)

P2

100

max

max

200

P2

P1

300 400 min

500 600

P1 0

50

100

150

200

250

300

350

250

300

350

4 Jetting speed (m s−1)

138

3 2 1 min

0 −1

(b)

min

V1

V2 0

50

100

150

200 Time (µs)

Figure 9.15 Drop formation measurement. (a) DOD drop formation curve and (b) instantaneous jetting speed curve.

at 130 µs. From the instantaneous jetting speed, it is seen from Figure 9.15b that the jetting speed of the satellite droplet was slightly negative (−0.8 m s−1 ) during the breakup process. Note that the negative jetting speed resulted from the speed measurement at the leading end of the droplet. If the speed is measured at the centroid of the droplet in the jetting direction, the jetting speed could be positive. Note that the jetting speed variation of satellites during the breakup is also related to jetting conditions such as jetting material properties and driving waveform. The satellite ligament length can be determined from the difference in distance between P2max (td ) and P2min (td ). Note that the length of satellite ligament decreased as the tail end of the satellite ligament was faster than the leading end. 4) Phase 4 (from 180 µs): In phase 4, the satellite droplet became spherical at 180 µs. The jetting speed of the satellite increased significantly (from −0.8 to 2 m s−1 ) when it became spherical. Then, the jetting speeds of both

9.5 Head Normalization and Condition Monitoring

spherical droplets tended to be constant until the two droplets merged into a single droplet. Therefore, when measuring the jetting speed by the conventional method using two different timings, it was required to measure the speed when the droplet became spherical. Otherwise, the measured speed may not represent the jetting performance, since the measured jetting speed can be different according to the selection of the two timings, as shown in Table 9.1. Also, the relative jetting speed of the satellite with respect to the main droplet should be measured properly. If the jetting speed of the satellite is slower than the main (or first) droplet, then the two droplets will not merge and there will be placement errors on printed substrate because of the satellite. Therefore, it is recommended to find jetting conditions where the jetting speed of the satellite is faster than that of the main droplet. The measured instantaneous jetting speed curve can be effectively used to measure the relative jetting speed of the satellite with respect to the main droplet. Figure 9.15b shows an instantaneous jetting speed curve. The curve shows that the satellite droplet was slightly faster than the main droplet, thus indicating that the two droplets will eventually merge into the main droplet, as shown in Figure 9.15b. Jetting speed can be affected by drop formation. Therefore, the instantaneous jetting speed curve has advantages because the speed variation during drop formation can be fully understood. Also, from the measured instantaneous jetting speed, the relative jetting speed of the satellite relative to the main drop can be investigated. The instantaneous jetting speed curve can be used to evaluate and control jetting performance [11]. However, the required time for the measurement can be significant compared to the conventional jetting speed measurement, even though in situ measurement techniques described in [10] can reduce the measurement time. Therefore, the method may not be practical for monitoring the jetting status of many nozzles in a multinozzle head.

9.5 Head Normalization and Condition Monitoring

For inkjet-based manufacturing, control and monitoring of the jetting condition of the printheads are important because the jetting condition affects productivity and reliability. In most applications, multinozzle heads are commonly used for high throughput. There are two different issues in this case: normalization of the multinozzle head to obtain uniform jetting performance among nozzles, and monitoring the multinozzle head to determine if there are any misfiring nozzles in the printhead. The first issue is related to the quality of inkjet patterning and the second issue is related to reliability. The jetting performance of each nozzle in a printhead may differ from the other nozzles. However, jetting uniformity among nozzles in the printhead is important because this will affect device performance unless corrected. For example, for a display application, the droplet volume and the droplet jetting speed from each nozzle should be uniform among nozzles in a printhead [12, 13]. For this purpose,

139

9 Vision Monitoring

Time

Volume (jetting speed)

(a)

Time

Volume (jetting speed)

Voltage (nozzle 1)

droplet volume and jetting speed are obtained by vision-based measurement [11, 12]. The driving voltage is then adjusted for uniform performance, as shown in Figure 9.16. In this study, for the voltage adjustment, we used a drive-per-nozzle (DPN) driver that drove each nozzle separately with a different driving voltage. For example, consider a jetting volume (or jetting speed) for nozzle number 1 that is measured to be greater (or higher) than the target value, as shown in Figure 9.16a. Then, the magnitude of the driving voltage for nozzle 1 should be reduced from a nominal voltage to decrease the droplet volume (or jetting speed). The level of voltage reduction is determined by the extent to which the measured jetting performance deviates from the target jetting performance. Once the jetting performance of a specific nozzle meets the target jetting performance, the same process is repeated for the next nozzle, as shown in Figure 9.16b. In the figure, nozzle number 2 has a lower jetting speed (or volume) compared to a target value. Thus, the magnitude of the driving voltage for nozzle 2 should be increased to increase the jetting speed (or volume) to the target value. This requires significant time because the scanning process is repeated to measure the jetting performance of all the nozzles, and adjust driving voltages for each nozzle to obtain uniform jetting performance. However, a vision-based monitoring system may not be successful in normalizing the jetting performance because of the limit of optical resolution, especially when measuring the droplet volume for normalization. The measured jetting speed can be used for the normalization, but this method might have limitations because the jetting speed might not be directly proportional to droplet volume. Alternative approaches have been tried in order to replace the vision-based normalization techniques [14].

Voltage (nozzle 2)

140

(b)

1 Target speed Target volume

Nozzle number

Target speed Target volume 2 Nozzle number

Figure 9.16 Normalization of a multinozzle head. (a) Adjustment of driving voltage for nozzle number 1 and (b) adjustment of driving voltage for nozzle number 2.

9.6 Meniscus Motion Measurement and Its Application

Inkjet print head CCD camera

Lens

LED strobe

Fixed mirror

Drain reservoir Rotatable mirror

Figure 9.17 Rotating mirror-type scanning system. (This illustration was kindly provided by Dr. Dong-Youn Shin.)

The scanning mechanism for normalizing a multinozzle head can be used for condition monitoring. If there is no jetting or misfiring, the malfunction needs to be detected and fixed immediately. To monitor multinozzle heads, a quick measurement mechanism (hardware) and an algorithm (software) for detecting faulty nozzles are required. The hardware of scanning mechanisms has been improved to reduce the time required for vision-based monitoring. For example, a fast scanning mechanism with XY stages and a high-speed frame grabber capable of acquiring 240 fps were used in [13]. However, the improvement of conventional methods might not be able to meet industry requirements; thus, new mechanisms need to be developed. Recently, a rotating mirror-type scanning system was reported in [15] in which the linear stage is not required, as shown in Figure 9.17. By using a rotating mirror system, the scanning time was reported to be significantly reduced compared to the linear XY stages. Owing to the speed limitations of vision-based scanning methods, methods for using a piezo self-sensing capability are under development to replace conventional vision-based approaches [16, 17]. Condition monitoring using piezo self-sensing can be much faster because it only uses electrical signals, and does not require mechanical alignment of the camera with respect to the monitoring nozzle. Nonetheless, the vision-based measurement approach is important in inkjet-based manufacturing systems because it is straightforward, and physical insight can be obtained from the visual images.

9.6 Meniscus Motion Measurement and Its Application

Since the experimental study by Bogy and Talke [8], jetting phenomena have been known to be related to a pressure wave of ink inside the inkjet head. In Bogy’s

141

142

9 Vision Monitoring

work, meniscus protrusion behavior was discussed in relation to pressure waves. However, an in-depth study of meniscus motion behavior might be difficult because Bogy’s meniscus measurement relied on the postprocessing of a prohibitively large number of CCD camera images. Recently, a simple vision processing algorithm was used to obtain the meniscus motion resulting from the pressure wave inside an inkjet head [9]. Sequential images must be acquired to obtain the meniscus motion. Also, low input voltage was used to measure meniscus motion such that there was no jetting from the nozzle. To identify the meniscus location from sequential images, an edge detection technique that identifies abrupt changes in pixel intensity values along the ROI line was used. As shown in Figure 9.18, ROI line AB is defined in the jetting direction starting from the center of the nozzle. Figure 9.19 shows the image values along the ROI line AB. By using the threshold value, the meniscus location can be determined, since there is an abrupt change in image values at the boundary between the meniscus and the background image. The same image processing was repeated throughout the sequence of images to obtain the meniscus behavior with respect to time, as shown in Figure 9.20. The use of edge detection differs from the binary image approach in that it does not require conversion of the original gray image to a binary image. Furthermore, the image processing technique is computationally efficient and straightforward because it deals with a one-dimensional (1D) array (a straight line) rather than a 2D (area) image. There are two different applications for meniscus motion: condition monitoring to find the faulty nozzle [17], and designing an efficient driving voltage for jetting [9]. If the jetting condition varies, then the pressure wave can be affected and the meniscus motion can change accordingly. By measuring meniscus motion, the causes of the malfunction can be understood. Possible changes from the normal condition include frequency, amplitude, and damping ratio of the meniscus motion, as shown in Figure 9.20 [17]. However, many sequential images are required to obtain the meniscus motion. Therefore, owing to time constraints, this method may not be an appropriate way to scan many nozzles to find a faulty nozzle. A 1

Detected edge B

Figure 9.18

Meniscus location by edge detection techniques.

9.6 Meniscus Motion Measurement and Its Application

Line profile 255 200

100 Threshold 0 A

B Detected edge location in pixel

Figure 9.19 Image pixel values along the ROI line.

30 Normal case Malfuction

Meniscus motion (µm)

25 20 15 10 5 0 −5 −10

0

50

100 Time (µs)

150

200

Figure 9.20 Condition monitoring using meniscus motion.

Meniscus behavior is related to the ink material as well as the waveform voltage. Thus, measured meniscus motion can be used for ink evaluation as well as waveform design. Figure 9.2 shows a typical waveform for jetting. When the rising (or falling) section of the waveform is applied to a piezo inkjet head, negative (or positive) pressure waves are generated, and then propagated inside the inkjet head [9]. However, the dwell region of the waveform is not related to the generation of pressure waves, but rather to the amplification or cancellation of pressure waves generated from the rising and falling sections of the waveform. Ink droplet jetting becomes easier if the positive pressure wave is amplified by means of adjusting the dwell time. Thus, the waveform design issue is focused on determining the optimal value for the dwell time [9]. The optimum value for a dwell time of L/C, where L is the dispenser tube length and C is the speed of sound in the ink, was recommended in [9] for the amplification of a positive pressure wave of ink inside an inkjet head. In practice, finding optimal time requires significant experimental effort. Also, if there is no jetting, a trial and error method might be the only option to find the waveform for jetting. The use of meniscus motion for waveform design has advantages, since the optimal dwell time for the waveform can be determined

143

144

9 Vision Monitoring

without actual jetting. Note that the period of meniscus motion is related to the speed of sound in the ink. After determining optimal dwell time based on meniscus motion, the magnitude of the waveform needs to be determined as a final step such that the target droplet jetting speed can be satisfied. The magnitude of meniscus motion is related to the jettability of the ink. If this magnitude of the meniscus motion is small, a higher waveform voltage may be needed for jetting.

References 1. Wijshoff, H. (2010) The dynamics of the

2.

3.

4.

5.

6.

7.

8.

9.

10.

piezo inkjet printhead operation. Phys. Rep., 491, 77–177. Kwon, K.S. (2009) Speed measurement of ink droplet by using edge detection techniques. Measurement, 42 (1), 44–50. Kipman, Y. (2009) Three methods of measuring velocity of drops in flight using jetXpert. Proceedings of NIP25 and Digital Fabrication, Kentucky, September 20–24, pp. 71–74. Hutchings, I.M., Martin, G.D., and Hoath, S.D. (2007) High speed imaging and analysis of Jet and drop formation. J. Imaging Sci. Technol., 51, 438–444. Dong, H. and Carr, W.W. (2006) An experimental study of drop-on-demand drop formation. Phys. Fluids, 187, 072102. Dong, H., Carr, W.W., and Morris, J.F. (2006) Visualization of drop-on-demand inkjet: drop formation and deposition. Rev. Sci. Instrum., 77, 085101. Jang, D., Kim, D., and Moon, J. (2009) Influence of fluid physical properties on ink-jet printability. Langmuir, 25, 2629–2635. Bogy, D.B. and Talke, F.E. (1984) Experimental and theoretical study of wave propagation phenomena in drop-on-demand ink jet devices. IBM J. Res. Dev., 28 (3), 314–321. Kwon, K.S. (2009) Waveform design methods for piezo inkjet dispensers based on measured meniscus motion. J. Microelectromech. Syst., 18, 1118–1125. Kwon, K.S., Go, J.K., Kim, J.W., and Oh, D. (2010) In-situ measurement of instantaneous jetting speed curve.

11.

12.

13.

14.

15.

16.

17.

Proceedings of NIP26 and Digital Fabrication, Texas, USA, September 20–23, 2010. Kwon, K.S. (2010) Experimental analysis of waveform effects on satellite and ligament behavior via in situ measurement of the drop-on-demand drop formation curve and the instantaneous jetting speed curve. J. Micromech. Microeng., 20, 115005. Creagh, L.T. (2003) Advances in deposition of OLEP materials via piezoelectric ink jet. Proceedings of 36th Annual International Symposium on Microelectronics (IMAPS), Boston, MA, November 16–20, 2003. Allbertalli, D. (2005) Gen 7 FPD inkjet equipment development status. Proceedings of SID (Society for Information Display), Boston, MA, May 22–27, 2005. Kim W., Kim, S.J., Kim, S.W., Kwon, K.S., and Kim, S.I. (2006) Normalization method of ink drops to ensure uniformity of amount of ink ejected from nozzles of inkjet head. US Patent, Application No. 11/585,254, filed Sept. 24, 2006. Shin, D.-Y. (2010) High speed inkjet monitoring module for jetting failure inspection. J. KSME, doi: 10.3795/KSME-AB.2010.34.0.000. Jong, J.D., Bruin, G.D., Reinten, H., Berg, M.V.D., Wishoff, H., Versluis, M., and Lohse, D. (2006) Air entrapment in piezo-driven inkjet printheads. J. Acoust. Soc. Am., 120 (3), 1257–1265. Kwon, K.S. (2009) Methods for detecting air bubble in piezo inkjet dispensers. Sens. Actuators A, 153 (1), 50–56.

145

10 Acoustic Monitoring Herman Wijshoff

10.1 Introduction

The possibilities of monitoring the printhead operation without an additional measuring setup is discussed in this chapter. With piezo inkjet, the actuator can also be used as a sensor during normal printhead operation. The acoustic signal detects variations in the drop formation process, the refill of the nozzle, wetting of the nozzle plate surface, disturbances from dirt particles, and the presence of air bubbles. Self-sensing also enables the active control of the driving waveform.

10.2 Self Sensing

The measurements, described in the previous section, give details on the ink flow outside the printhead. The phenomena inside the channels are difficult to measure. Only when using special transparent printheads, for example, channels in or covered by a glass plate, and flow tracing particles the flow inside can be measured [1]. The most suitable method for opaque piezo heads uses the actuator also as a sensor. The results described in this section are a summary of some form of already published results of the research performed at Oc´e Technologies and the group of Prof. Detlef Lohse from the University of Twente. A more complete description can be found in [2]. The driving force to fire a droplet with a piezo inkjet printhead is generated by an actuator, which deforms the structure through the inverse piezoelectric effect. The piezoelectric effect (electricity from an applied mechanical stress) was discovered by Pierre and Jacques Curie in 1880 [3]. Their experimental demonstration consisted of a conclusive measurement of surface charges appearing on specially prepared crystals, which were subjected to a mechanical stress. In 1881, Lippmann deduced mathematically the inverse piezoelectric effect (stress in response to an applied electric field). The Curie brothers immediately confirmed the existence of this Inkjet-based Micromanufacturing, First Edition. Edited by J. G. Korvink, P. J. Smith, and D.-Y. Shin. © 2012 Wiley-VCH Verlag GmbH & Co. KGaA. Published 2012 by Wiley-VCH Verlag GmbH & Co. KGaA.

146

10 Acoustic Monitoring

property. In the following years, the 20 natural crystal classes in which piezoelectric effects occur and all possible macroscopic piezoelectric coefficients were defined [4].

10.3 Measuring Principle

So, a piezo can be used as actuator and as sensor by using the inverse (actuator) and direct (sensor) piezoelectric effect. The former comprises the effect that when an electrical voltage V is applied to the piezo unit, a displacement y of the piezo unit results. The latter refers to the effect that when a force F is applied to the surface of a piezo, an electric charge Q results. Together, this behavior can be described as      y d 1/k V = (10.1) Q C d F with C the capacitance of the piezo element, d the piezoelectric charge constant, and k the stiffness of the piezo element. Switching the piezo elements from the electronic driving circuit to a measuring circuit gives an accurate recording of the total force on the piezo element, which is a measure of the average pressure inside the ink channel, which will from now on be called the acoustic signal (Figure 10.1). First the driving waveform is applied, which takes 2–20 μs. After that, the current from the piezo element can be measured until the next actuation cycle starts. The main problem for this setup is that the amplitude of the current for charging the piezo is at least two orders of magnitude larger than the acoustic signal, which is typically 100 μA. The big discharge current from the piezo can disturb the measurement, so the piezo must be completely decharged before the acoustic measurement starts. With a reference signal, however, an acoustic measurement during actuation is even possible. The measured signal Q is made up of two contributions. The first is that of the applied actuation voltage V via the piezo’s capacity C and is referred to as the direct path. The second contribution originates from the force F exerted by the ink in the channel via the piezoelectric charge constant k and is referred to as the indirect path. As only this second contribution is the required signal,

Pulse

Piezo Readout

Figure 10.1 Outline of acoustic measurement. Switching the piezo elements between an electronic driving circuit and a measuring circuit enables both the actuation of the channels and the measurement of the pressure variation inside the channels.

10.3 Measuring Principle Direct-path ‘piezo’

Indirect-path ‘ink’ + d +

v

+

q

u

1 k

C +

Zc

‘Piezo’ +

d

147

F

‘Ink’

−

‘Piezo’

=

+

Full channel

Empty channel

Figure 10.2 The measuring block with the ink channel as impedance and the basic principle to obtain the actuation and sensor signal simultaneously.

it has to be extracted from the measured signal Q. Therefore, the direct signal, which results from the voltage on the piezo element, has to be subtracted from the indirect signal, as shown in Figure 10.2. One option is to measure the difference over a bridge structure with a reference capacitance [5] or even better with another channel as reference. The measured signal of a full ink channel comprises both the direct and indirect path. The measured signal of an empty ink channel only consists of the contribution of the direct path. Again, by subtracting both measured signals, the indirect path or sensor signal can be obtained. The hardware compensation in the latter requires that both piezo elements are exactly the same. However, small differences due to drift or production tolerances are always present and the resulting signal is not as accurate as the signal measured with switching. To minimize the effects of piezo capacity differences, the following measures can be taken: • Temperature differences. Differences in piezo capacity occur due to temperature differences of both piezo-units. By isolating the piezo printhead, these differences have to be minimized. • Differences in piezo capacity. Matching the impedance of various piezo elements usually results in a satisfactory pair. • Influence of structural effects on the sensor measurement. Even though the ink channel is empty, a small contribution due to the deformation of the structure may be present in the indirect path. This effect can be neglected when the cross-talk level is low enough (Chapter 5). Another option is to calculate the contribution of the direct path to the acoustic signal and subtract this from the measurement [6, 7]. The main drawback of the computational compensation is related to the required accuracy of the piezo model. The nonlinear behavior of the piezo elements is very difficult to model accurate enough so that the error is at least significantly smaller than the sensor signal that one is trying to obtain. Barium titanate (BaTiO3 ), the first piezoelectric ceramic with a perovskite structure (a tetragonal/rhombohedral structure very close to cubic), was found around

‘Ink’

148

10 Acoustic Monitoring

1943. S. Roberts detected the piezoelectric effect in BaTiO3 in 1947. In 1954, the discovery of the piezoelectric ceramic Pb(Zrx Ti1−x )O3 , lead zirconate titanate (PZT), was reported by B. Jaffe (US Patent 2,708,244). In the following years, PZT became the main industrial product in piezoelectric ceramic materials [8]. Many actuator applications have emerged for the use of PZT [9]. Ceramic perovskites have a cubic structure that is stable at temperatures above their Curie temperature. When the temperature decreases and falls below the Curie temperature, the structure changes. The cubic structure becomes either a rhombohedral structure with a low grade of Ti or a tetragonal structure at high Ti grades. The rhombohedral structure can be viewed as a cubic structure, stretched diagonally along a body axis of the unit cell. The tetragonal structure can be viewed as a cubic structure, stretched along one of the lattice vectors. So at a temperature below the Curie temperature, the O2− and the Pb2+ ions are moved from their cubic positions and the Ti4+ and Zr4+ ions are moved from the center of the cube. The positive and negative charge sites do not coincide anymore, which results in a dipole. Because the charge sites do not coincide at a temperature below the Curie temperature, an external applied electric field will tend to deform the structure. This is the inverse piezoelectric effect. In general, a uniform alignment of the electric dipoles only occurs in certain regions of a crystal, while in other regions the polarization may be in the reverse direction. Such regions are called ferroelectric domains. In order for the material to become piezoelectric, it has to be poled. Poling is the imposition of a DC-voltage across the material. The ferroelectric domains align to the field resulting in a net piezoelectric effect. Not all the domains become exactly aligned. Some of them align only partially and some do not align at all. This results in a nonlinear behavior with hysteresis. In general, linear relationships are used for simulations such as D = dT T + ε E

(10.2)

where D is the electric displacement field (or the charge density), E the applied electrical field, and T the applied mechanical stress. The matrices with the piezoelectric coefficients d for the piezoelectric and the inverse piezoelectric effect are equal and ε is the matrix with the dielectric constants. With the acoustic measurement, the electric current from the piezo elements is measured. The Paint current Ip from a piezo element with area Ap is given by the equation: dD dQ = Ap (10.3) dt dt with D the electric displacement or charge density in the polarization and actuation direction of the piezo element. With the electric field applied in the actuation direction, E = V/hp , with hp the height of the piezo element, we get with Eq. (10.2): Ip =

dT dE + Ap (10.4) dt dt with deff the effective piezoelectric coefficient in the actuation direction, taking into account, for example, the mechanical constraints of the piezo element [2]. The Ip = deff Ap

10.3 Measuring Principle

normal stress component T is generated by the pressure P in the ink channel. With V the voltage on the electrodes, this equation becomes: Ip = deff Ap

Ap dV dP + dt hp dt

(10.5)

The capacitance of the piezo element is given by Cp = Ap /hp . After integrating the pressure, which can show local variations, over the length lp of the piezo element, we get:  lp dV dP Ip = Cp + deff bp dz (10.6) dt dt 0 with bp the width of the piezo element. The acoustic signal is now known in terms of the actuator voltage and the channel pressure. With the general homogeneous solution for the acoustic channel pressure and the particular solution, which is the pressure due to the actuator, in the frequency domain per frequency according to d’Alemberts solution: 2 V P(z, t) = Pr ei(ωt−kz) + Pl ei(ωt+kz) − αρceff

(10.7)

we can express the acoustic signal in terms of propagating waves toward the nozzle, subscript r, and coming from the nozzle, subscript l, with wave number k and frequency ω. Inserting this equation in Eq. (10.6) results in:  ω  iklp 2 Ip = iω(Cp − αdeff Ap ρceff Pl e − Pr e−iklp eiωt )V + deff bp (10.8) k With this measurement, we have only limited access to the interior of the printhead. The use of the actuator as sensor is an option that has no consequences for the manufacturing of the printhead, except for the electronics, which has now also to include a measuring circuit. With some more modifications of the printhead, another piezo-based possibility would be to use separate sensors such as a small extra piezo element with the local channel pressure as parameter. This option can give some more details on the pressure wave propagation. However, this is difficult to implement without disturbing the acoustics and the drop formation. Another alternative option is a built-in measurement method of the meniscus position, which also allows measurements in the jetting situation. This can be offered by a sensing nozzleplate with a capacitive layer inside the nozzle. The capacitance between the electrodes depends on the amount of ink in the nozzle [10, 11]. This is also difficult to implement without disturbing the wetting properties of the nozzle and the drop formation process. We need more information on the phenomena preceding the drop formation for a better understanding of the operating principles of the piezo printhead and a correct interpretation of the measured acoustic signal. Details on ink flow inside the channels and acoustic pressure waves are only available through modeling [12]. Therefore, the modeling of the physical phenomena with available commercial codes and the development of dedicated special models are an essential part in the development of a new inkjet technology. Added to the measurements, the modeling

149

150

10 Acoustic Monitoring

reveals the chain of processes, which lead to drop formation. This enables a faster and better development of new printheads [13]. When modeling inkjet printheads, we have to face many challenges, as with the modeling of all micro electromechanical systems (MEMS) [14]. The management of small-scale flows is a common denominator in MEMS [15]. The modeling comprises a multiscale simulation from nanometers to meters of multiphysics: solid mechanics, fluid dynamics, electromagnetism, materials science, electronic circuit design, mechatronics, and so on. Most of the models are derived from continuum mechanics, although in some cases the continuum mechanics are pushed to its limits. The continuum theory requires that variables such as density, pressure, and velocity are defined by some averaging process and determined by the solution to some system of equations. We must keep this in mind when going to the smallest length scales. The domains of interest are the structural dynamics, including piezoelectricity and acoustoelastic interaction, and the fluid dynamics [2].

10.4 Drop Formation, Refill, and Wetting

The phenomena that have an influence on the pressure waves inside the ink channels can be detected with acoustic measurement. The drop formation process changes the amount of ink inside the nozzle by jetting away a certain amount of fluid. The mass of ink in the nozzle has a direct impact on the acoustic properties. The nozzle acts as a partial closed boundary condition for the pressure waves inside the ink channel. Less ink in the nozzle results in a lower acoustic impedance of the nozzle, that is, the reflection of the acoustic pressure waves will change toward a reflection at an open end. The reflection of the pressure waves at a nozzle with less ink will result in a lower pressure amplitude at the nozzle and a higher amplitude inside the channels. In order to fire many drops continuously, the nozzle has to get filled again after one drop formation cycle. At a short time scale, that is, as long as the acoustic pressure waves are not damped out, the asymmetric acceleration of the ink in the nozzle is the main driving force behind the refill of the nozzle. Before a positive pressure peak reaches the nozzle, the meniscus surface is retracted, and the small mass of ink in the nozzle gets a large acceleration outward. Before a negative pressure peak reaches the nozzle, the meniscus surface is displaced outward, and the large mass of ink in the nozzle gets a small acceleration inward. This results in a net displacement outward, which fills the nozzle at a short time scale, that is, the time scale over which the pressure fades away, which is typically less than 100 μs. The surface tension of the ink γ generates a capillary pressure pc in a nozzle with radius with Rn : 2γ cos θ (10.9) pc = Rn with cos θ equal to 1 at the nozzle exit with sharp edges. The capillary pressure is another driving force behind the refill of the nozzle. This pressure will drive

10.4 Drop Formation, Refill, and Wetting

the meniscus surface to its equilibrium position, also after the pressure waves are damped out. With the pressure from a Poiseuille flow profile in a nozzle with length Ln and u = ∂zn /∂t: 8πηLn u An

pv =

(10.10)

with η the viscosity of the ink, we get the Washburn equation for the meniscus position: γ Rn cos θ ∂zn (t) = ∂t 4η(Ln + zn (t))

(10.11)

This results in a rather slow filling of the nozzle, for example, a filling time up to several hundred microseconds. Both the effects result in a variation of the filling of the nozzle during normal printhead operation. In Figure 10.3a, a top view on a jetting nozzle is shown, for a case that results in overfilling of the nozzle [12]. The filling level at the start of a drop formation cycle has a direct influence on the resulting drop properties. Jet of drops

Rim of ink 100 μm (b)

Signal (mA)

(a)

(c)

1.0 0.8 0.6 0.4 0.2 0.0 −0.2 −0.4 −0.6 −0.8

0

100

200 Time (μs)

Figure 10.3 (a) Top view of a firing 32 μm nozzle at a cycle time of 100 μs. The overfilling results in a rim of ink, accumulated around the nozzle opening. (b) The same nozzle, when the overfilling leads to wetting, that is, a layer of ink around the nozzle exit.

300

400

(c) The effect of a decreasing ink layer thickness on the acoustic signal. The ink layer thickness is 20 μm at the start of the first drop formation cycle and decreases to 16, 12, 8, 4, 3, 2, and 1 μm at the beginning of the next cycles.

151

152

10 Acoustic Monitoring

The free surface flow in the nozzle and the acoustics in the ink channels have to be designed to give a good refill of the nozzle. This makes high drop repetition rates possible. However, a strong refill process also results in overfilling of the nozzle and overfilling of the nozzle can result in wetting of the nozzle plate, especially when the area around the nozzles has mechanical and chemical defects such as scratches or depositions of ink components (Figure 10.3b). A layer of ink on the nozzle plate around a nozzle has a large effect on the drop formation process [16]. An ink layer results in a larger mass of ink at the nozzle. More mass of ink at the nozzle results in more inertia, which slows down the speed of the drop [17]. The extra mass of ink at the nozzle also changes the pressure waves inside the ink channels. As with overfilling because of the refill effect, the impedance of the nozzle increases. The reflection of the pressure waves at the wetted nozzle will result in higher pressure amplitudes at the nozzle and lower amplitude inside the channels, but the effect is much stronger as with overfilling. In Figure 10.3c, the effect of a disappearing ink layer on the amplitude of the acoustic signal is shown.

10.5 Dirt

A very important requirement for today’s productive drop-on-demand inkjet printers is the stability of the jetting process. Large dirt particles with a radius more than 15 μm can block a nozzle opening completely and there will be no drop formation at all. A nozzle can also be partially blocked by a large dirt particle. This results in severe deviations of the drop speed and size, the jetting angles and the drop shape [2]. Small dirt particles do also influence the drop formation process, at least during one drop formation cycle [18]. The distortion of the drop formation and the variation in the acoustic signal I, defined as the variance σ 2 during a certain measuring time window T:  1 T  2 (10.12) I(t) dt σ2 = T 0 are shown in Figure 10.4. The amplitude of the acoustic signal varies typically by 5–10% during a distortion, which occurs during drop formation cycles 20 to 25. The amplitude of the acoustic signal over 25 drop formation cycles before and after the distortion varies by less than 0.5%. A variation of a few micrometers in the position of the meniscus results in a comparable variation of the acoustic signal [2]. To explain that the amplitude is first several percent lower and then several percent larger, the meniscus must first be protruded out of the nozzle and second the meniscus must be retracted. Also a variation in the drop speed is observed during a distortion. The first droplet is 0.1–0.2 m s−1 faster and the second droplet up to 0.4 m s−1 slower, with often a deviating shape and jetting angle [18]. The variation in the drop speed, as a result of the different meniscus positions, that is, because of a variation in the refill

10.6 Air Bubbles

t = 50 μs

t = 100 μs Drop 21

t = 150 μs

Drop 22

153

t = 200 μs

Drop 23

Drop 24

s2

0.14

0.13

0.12 5

10

15

20

25

30

35

40

Droplet number

Figure 10.4 The drop formation recorded at 100 kfps, shows a disturbance. Until drop 20, the drop formation is regular. Drop 21 displays a slight deviation in the tail. Drop 22 shows a large disturbance being jetted out. From drop 23, the droplet formation is regular again. Below the recording, the variation in the acoustic signal is shown.

level, corresponds to the observed variations during the distortion of the acoustic signal. So the variation in meniscus position seems to be the main effect on the drop formation process and the reflection of the pressure waves. However, the dirt particles itself can also be the source of the observed variations in drop speed, since the drop speed varies with the particle at different positions under the surface [2].

10.6 Air Bubbles

The variation in the amplitude of the acoustic signal, as shown in Figure 10.4, is the most common example of a distortion. After the disturbance occurred, jetting keeps going on as usual. But sometimes after a disturbance, an air bubble is entrapped, which becomes later visible as a larger deviation of the acoustic signal. Then, there is also a disturbance of the drop formation visible [19], which finally can even lead to the complete breakdown of the drop formation process [20, 21]. Wetting itself can also result in air entrainment at the nozzle when the ink layer thickness has a critical value [18]. Suppressing the generation of air bubbles and controlling the behavior of air bubbles is therefore crucial in realizing a maximum jetting stability.

45

10 Acoustic Monitoring

0.11

0.10 s2

154

0.09

0.08 10 20 30 40 50 60 70 80 90 100 Droplet number

0.6 0.4 0.2 0 −0.2 −0.4 −0.6 −0.8 20

120 80 30

40 35

40

Time (μs

45

)

Figure 10.5 The amplitude of the acoustic signal when a distortion occurs before cycle 20 and air is entrained. The amplitude of the acoustic signal deviates more than 10% within 100 drop formation cycles at a

50

0

et

pl

o Dr

r

be

m nu

repetition rate of 20 kHz. The development of the actual acoustic signal after an air bubble is entrained is shown at the right. The z axis is the amplitude of the signal.

Not only the drop formation, the refill, the wetting of the nozzle plate, and dirt particles change the reflection of the acoustic pressure wave but also an air bubble has a large influence on the pressure waves coming from the nozzle. This enables the acoustic measurement to monitor the jetting stability also [18]. With the acoustic measurement, only the global acoustic response can be measured. When air is entrained, the amplitude of the Paint signal will show an increasing deviation from its nominal value after the distortion, as shown in Figure 10.5. An increase or a decrease in the amplitude indicates that something is happening. At a repetition rate of 20 kHz, the variance deviates more than 10% within 100 drop formation cycles of the printhead used in this experiment. First of all, a bubble will oscillate in an acoustic field as described by the Rayleigh–Plesset equation [22]. The oscillation of the bubble itself is not recorded

10.6 Air Bubbles

as disturbance of the acoustic signal, but the change in the reflection of the pressure waves. Therefore, the acoustic properties of the ink channel will act as a filter for the acoustic measurement [2]. We see the influence of a bubble as a 170-kHz distortion because this frequency is next to the basic channel resonance at 50 kHz, the next strong peak in the frequency characteristics of the printhead used in this experiment. The frequency characteristics describe the relation between the electric activity in the actuator and the meniscus movement and acts also as a filter for the acoustic measurements. The high-frequency components are more sensitive to small changes in the reflection conditions of the pressure waves and, therefore, the higher frequency components are altered more than the low-frequency components. Many forces are acting on a bubble in an acoustic pressure field. The movement of a bubble is determined by the competition between the acoustic and the hydrodynamic forces [23]. These acoustic and hydrodynamic forces will result in complicated movement patterns. Furthermore, a bubble in an acoustic pressure field will grow by rectified diffusion when the acoustic pressure variations are strong enough [22]. At pressure maxima, air is squeezed out of the bubble, but this loss is overcompensated at the pressure minima when the bubble expands. This results in a net gas diffusion into the bubble. Rectified diffusion is a result of a surface effect and a shell effect, for example, an expanded bubble can absorb more air because of its larger surface area and the higher concentration gradient of the dissolved air in the liquid around the bubble, which is compressed by the expanded bubble. The bubble size increases fast and saturates at an equilibrium size after 1000–5000 actuation cycles [2]. The growing by rectified diffusion is then balanced by the dissolution rate of the bubble, which dissolves mainly between the actuation cycles. The growing bubble will have an increasing influence on the acoustic properties of the ink channel, visible in the increasing deviation of the Paint signal in Figure 10.5. At equilibrium, the bubble volume is comparable to the total volume displacement of the acoustic pressure wave. The volume displacement of acoustic pressure field can now be counteracted by the change in bubble size. There is no driving force left for the ink movement in the nozzle and the drop formation comes to an end. This will also change the acoustic properties of the channel completely. Without an air bubble, the nozzle acts as a partial closed boundary condition for the pressure waves inside the ink channel. The channel will act almost as a 14 λ resonator in the printhead used for these experiments. A large bubble can counteract the pressure buildup. This results in a complete open reflection. The channel will act now as a 12 λ resonator. The main resonance frequency will increase and also the amplitude of the pressure wave will increase. This is again visible in the acoustic signal, Figure 10.6(a). A numerical model is developed, which describes this impact very accurately [24]. This model is the basis for a derivation of the bubble size and position from the deviation of the Paint only, Figure 10.6(b). The experimental validation is done with special transparent printheads to enable a direct optical recording of the air

155

10 Acoustic Monitoring 0

50

100

150

200

1 0.8

Actuation pulse

Time (μs)

0.6 0.4 0.2 0 −0.2

Normal signal

−0.4 −0.6

with air bubble

−0.8 (a)

−1 140 120 Bubble volume (pI)

156

100 80 60 40 20 0

0

(b)

50

100

150 Time (s)

Figure 10.6 (a) The measured effect of a large air bubble on the Paint signal. A large air bubble acts as an open boundary condition for the pressure. The acoustic properties of the channel shift to a 12 λ resonator with a higher frequency. (b) The direct optically

200

250

300

measured bubble volumes (the solid line with the white area indicating the error margins) and the bubble volumes derived with the model from the acoustic Paint signal (the dotted line with the gray area indicating the error margins).

bubbles [25]. The acoustically derived bubble volumes deviate less than 12% from the direct optically measured bubble volumes.

10.7 Printhead Control

The acoustic measurement also enables feed-forward control of the driving waveform [26]. For most designs, the input waveform is manually shaped, based on physical insight in the working of a printhead. Normally, the actuation pulse

References

is tuned to the first eigenfrequency of the ink channel. In addition, somewhat more complex waveforms are designed for purposes such as smaller droplets and damping of the residual vibrations. A control framework enables the systematic exploration of better driving waveforms to enhance the performance of the printhead, without having to perform a redesign of the printheads. Iterative learning control (ILC) as feed-forward control is very effective in improving the performance of a process that performs repetitive tasks [27, 28]. So more specifically, given the highly repetitive character of the jetting process, ILC is a logical choice as control strategy for inkjet printheads [26] to reduce cross-talk effects [29] or the impact of residual vibrations [30]. Another suitable feed-forward control strategy is optimization-based control to reduce the effect of residual vibrations [31].

References 1. Meinhart, C.D. and Zhang, H. (2000)

2.

3.

4.

5.

6.

7.

8.

9.

10.

The flow structure inside a microfabricated inkjet printhead. J. MEMS Syst., 9, 67. Wijshoff, H. (2010) The dynamics of the piezo inkjet printhead operation. Phys. Rep., 491, 77. Curie, J. and Curie, P. (1880) Development, par pression, de l’electricite polarise dans les crystaux hemiednes et fares inclines. Comp. Rend., 91, 294. Ballato, A. (1995) Piezoelectricity: old effect, new thrust. IEEE Trans. Ultrason. Ferroelectr. Freq. Control, 42, 916. Kwon, K.S. and Go, J.K. (2009) Methods for detecting jetting failures in inkjet dispensers. Proceedings IS&T’s Nip25, Louisville, p. 382. Dosch, J.J., Inman, D.J., and Garcia, E. (1992) A self-sensing piezoelectric actuator for collocated control. J. Intell. Mater. Syst. Struct., 3, 166. Anderson, E.H. and Hagood, N.W. (1994) Simultaneous piezoelectric sensing/actuation: analysis and application to controlled structures. J. Sound Vibrat., 174(5), 617. Jaffe, B., Cook, W.R., and Jaffe, H. (1971) Piezoelectric Ceramics, Acedemic Press, London. Uchino, K. (1997) Piezoelectric Actuator and Ultrasonic Motors, Kluwer Academic Publishers, Boston, MA. vd Velden, M., Spronck, J.W., Munnig Schmidt, R.H., Wei, J., and Sarro, P.M. (2007) Characterization of a nozzle-integrated capacitive sensor

11.

12.

13.

14.

15.

16.

17.

18.

for microfluidic jet systems. Proceedings IEEE Sensors 2007, Atlanta, p. 1241. Wei, J., vd Velden, M., and Sarro, P.M. (2007) Fabrication of vertical electrodes on channel sidewall for picoliter liquid measurement. Proceedings Transducers 2007, Lyon, p. 1613. Wijshoff, H. (2004) Free surface flow and acousto-elastic interaction in piezo inkjet. Proc. NSTI Nanotech2004, Boston, 2, 215. Wijshoff, H. (2007) Better printheads via simulation: Flow3D helped double the speed of a new wide-format printer without sacrificing quality. Desktop Eng., 13(2), 46. Palesko, J.A. and Bernstein, D.H. (2003) Modeling MEMS and NEMS, Chapman & Hall CRC Press. Gad-el-Hak, M. (1999) The fluid mechanics of microdevices - The Freeman scholar lecture. J. Fluids Eng., 121, 5. de Jong, J., Reinten, H., Wijshoff, H., van den Berg, M., Delescen, K., van Dongen, R., Mugele, F., Versluis, M., and Lohse, D. (2007) Marangoni flow on an inkjet nozzle plate. Appl. Phys. Lett., 91, 204102. Wijshoff, H. (2007) Drop formation mechanisms in piezo-acoustic inkjet. Proc. NSTI’s Nanotech2007, 3, 448. de Jong, J., Reinten, H., van den Berg, M., Wijshoff, H., Versluis, M., de Bruin, G., and Lohse, D. (2006) Air entrapment in piezo-driven inkjet printheads. J. Acoust. Soc. Am., 120, 1257.

157

158

10 Acoustic Monitoring 19. de Jong, J., Jeurissen, R., Borel, H.,

20.

21.

22.

23.

24.

25.

van den Berg, M., Wijshoff, H., Reinten, H., Versluis, M., Prosperetti, A., and Lohse, D. (2006) Entrapped air bubbles in piezo-driven inkjet printing: their effect on the droplet velocity. Phys. Fluids, 18, 121511. Brock, J.D., Cohen, I.M., Ivanov, I.P., Le, H.P., and Roy, J. (1984) Oscillations of an air bubble in an ink jet. J. Imaging Sci. Technol., 10, 127. Hine, N.P. (1991) Deaeration system for a high-performance drop-on-demand ink jet. J. Imaging Technol., 17, 223. Brennen, C.E. (1995) Cavitation and Bubble Dynamics, Oxford University Press. Rensen, J., Bosman, D., Magnaudet, J., Ohl, C.D., Prosperetti, A., Toegel, R., Versluis, M., and Lohse, D. (2001) Spiraling bubbles: how acoustic and hydrodynamic forces compete. Phys. Rev. Lett., 86, 4819. Jeurissen, R., de Jong, J., Reinten, H., van den Berg, M., Wijshoff, H., Versluis, M., and Lohse, D. (2008) Effect of an entrained air bubble on the acoustics of an ink channel. J. Acoust. Soc. Am., 123, 2496. Jeurissen, R., van der Bos, A., Reinten, H., van den Berg, M., Wijshoff, H., de Jong, J., Versluis, M., and Lohse, D. (2009) Acoustic measurement of bubble

26.

27.

28.

29.

30.

31.

size in an inkjet printhead. J. Acoust. Soc. Am., 126, 2184. Groot Wassink, M.B. (2007) Inkjet printhead performance enhancement by feedforward input design based on two-port modeling. PhD thesis. Delft University of Technology. Moore, K.L. (1998) Iterative learning control: an expository overview. Appl. Comp. Contr., Signal Proc. Circ., 1, 151. Longman, R.W. (2000) Iterative learning control and repetitive control for engineering practice. Int. J. Control, 73, 930. Groot Waasink, M.B., Bosgra, O., and Koekebakker, S.H. (2006) Minimization of crosstalk for an inkjet printhead using MIMO ILC. Proceedings 2006 ACC, Minneapolis, p. 964. Groot Wassink, M.B., Bosgra, O.H., Koekebakker, S.H., and Slot, M. (2006) Minimizing residual vibrations and cross-talk for inkjet printheads using ILC designed simplified actuation pulses. Proceedings IS&T’s Nip22, Denver, p. 69. Khalate, A., Bombois, X., Babuska, R., Wijshoff, H., and Waarsink, R. (2011) Performance improvement of a drop-on-demand inkjet printhead using an optimization based feedforward control method. Contr. Eng. Eng. Pract., 19, 771.

159

11 Equalization of Jetting Performance Man-In Baek and Michael Hong

The most important physical properties of an inkjet printhead are the droplet volume and speed from its nozzles. However, the deviation of jetting performance across nozzles cannot be avoided because of the imperfect manufacturing process of an inkjet printhead. This deviation causes the difference in the volume and speed of droplets out of individual nozzles even when the same operating voltage and wave form are applied to all the nozzles of an inkjet printhead. Such deviation eventually degrades the printing quality and generates a serious smear on the printed device. Since the droplet volume determines the thickness and line width of the pattern formed on a substrate, it affects resistance uniformity of the printed electrodes in electronic applications and visible scan stains, called printing swath marks, in display applications while the jetting speed deviation of droplets causes positional errors on the substrate. Therefore, there is a need for separate equipment and methods to control the droplet volume and speed out of each nozzle in an inkjet printhead. The printing swath mark, which is generally recognized as the most serious problem when applying inkjet technology to manufacture display panels, is mainly led by the droplet volume deviation than by the positional error of the droplet. To control the droplet volume out of each nozzle, therefore, it is important to determine the baseline physical quantity. Measurement methods for equalizing droplet volumes across nozzles in an inkjet printhead include (i) measuring the speed and volume of droplets out of nozzles on the fly and (ii) measuring the size and volume of sessile droplets on the substrate. Especially in case of the color filer fabrication for thin-film transistor liquid crystal displays (TFT LCDs), the measurement of light transmission deviation from colored subpixels is used to indirectly detect the droplet volume deviation out of each nozzle. In this section, the strengths and weaknesses as well as the applications of each method are described.

Inkjet-based Micromanufacturing, First Edition. Edited by J. G. Korvink, P. J. Smith, and D.-Y. Shin.  2012 Wiley-VCH Verlag GmbH & Co. KGaA. Published 2012 by Wiley-VCH Verlag GmbH & Co. KGaA.

160

11 Equalization of Jetting Performance

11.1 Equalization of the Droplet Volume on the Fly

Methods of equalizing the droplet volume out of each nozzle on the fly include measuring the droplet speed ejected from each nozzle and using the correlation between the speed and volume of a droplet to control the voltage and pulse width applied to each nozzle; and also by capturing the images of instantaneously formed droplets using a high-speed camera and processing the images to calculate the droplet volume and then using it to control the voltage and pulse width applied to each nozzle. Generally, a drop watcher is used to measure the droplet speed ejected out of each nozzle and measuring the droplet volume from the captured images. 11.1.1 Components of a Drop Watcher

A drop watcher, which is used in measuring the speed and volume of droplets on the fly, mainly consists of an operation unit that drives the inkjet printhead, a supply unit that delivers the ink to the inkjet printhead, a measurement unit that captures the images of ejected droplets out of each nozzle with a charge-coupled device (CCD) and a control unit that calculates the droplet speed and volume using the captured images of droplets and then controls the system. The operation unit consists of an inkjet printhead and controller, which applies the voltage and pulse to the inkjet printhead. The supply unit consists of the meniscus module, which applies a slightly negative pressure to hold the meniscus in position with ink provided during the droplet inspection period as well as the printing period, and a subreservoir, which continuously supplies the appropriate amount of ink to the inkjet print head. The measurement unit consists of the high-speed camera, the light source module that inspects the droplets by synchronously reacting to the jetting event, and the driving unit that can focus on a set of nozzles of each inkjet printhead. Lastly, the control unit calculates the droplet speed and volume using the captured images received from the measurement unit and then uses them to control the whole system. Figure 11.1 shows a typical drop watcher system and main screen of the control unit. 11.1.2 Equalization through Volume Control

To equalize the droplet volume ejected from each nozzle, the volume of a droplet on the fly out of each nozzle after applying the same voltage must be measured using a drop watcher. To calculate the droplet volume, the instantaneously captured image of a droplet is binary coded to extract the boundary of the droplet, as shown in Figure 11.2. The extracted two-dimensional image of the droplet is divided into a set of layers with the finite thickness in the vertical direction, and the volume of each layer is summed to calculate the entire droplet volume by rotating each layer on its axis. In accordance with the deviation of the measured droplet volume from

11.1 Equalization of the Droplet Volume on the Fly

Inkjet printhead Nozzle

Ink droplet Strobe CCD Camera

Figure 11.1 A drop watcher system and the main screen of the control unit.

Figure 11.2 Binary coding of the captured droplet images on the fly.

the baseline volume, a proportional coefficient is calculated and multiplied to the previously applied voltage for the voltage adjustment. Then, the calculated voltage or pulse width for the inspected nozzle is downloaded to the inkjet printhead controller. By repeating such processes through all the nozzles, the droplet volume ejected from each nozzle of the inkjet printhead can be equalized. For this purpose, different voltages need to be applied to individual nozzles, which is called a drive-per-nozzle (DPN) function. Figure 11.3 shows the sequence diagram to equalize the droplet volumes on the fly. 11.1.3 Results of the Droplet Volume Measurement and Equalization Process

Figure 11.4 shows the measurement results of droplet volumes ejected out of each nozzle after applying an initial voltage of 80 V to an inkjet printhead by Fuji Dimatix. The droplet volume measurement and equalization process was repeated three times on the fly. As shown in the figure, the deviation of drop volumes out of the nozzles was reduced from 24 to 12% after three repetitions of the measurement

161

162

11 Equalization of Jetting Performance

Start

Voltage setting and ink jetting

Image capturing and volume calculation

Volume adjustment

Volume deviation < Specification

No

Yes End Figure 11.3 Sequence diagram for the droplet volume measurement and equalization process of an inkjet printhead with a drive-per-nozzle (DPN) function.

60

Before DPN After DPN

Volume (pl)

55 50 45 40 35 30

1

11

21

31

41

51

61

71

81

91

101

111

121

Nozzle number Figure 11.4 Measured droplet volume before and after the droplet volume measurement and equalization process.

and equalization process. The initial droplet volume deviation of 24% is 10% larger than the characteristic deviation of an inkjet printhead by the Fuji Dimatix; and the droplet volume deviation of 12% even after the droplet volume measurement and equalization is still too large to be adapted to for the color filter fabrication process with inkjet printing. The deviation was not lower than 12% even after increasing the number of repetitions of the droplet volume measurement and equalization process. This is due to the errors caused by the inherent uncertainty in capturing

11.1 Equalization of the Droplet Volume on the Fly

Nozzle surface

Figure 11.5 Droplet formation along with the distance from the nozzle.

and processing the images, the inherent limitations of the drop watcher, and the characteristics of the inkjet itself. The image acquired by the drop watcher is not an instantaneous single image of the droplet but a superimposed one from tens to hundreds of images. Since the jetting conditions such as the voltage, pressure drop, and nozzle wetting of the inkjet minutely changes from time by time, the droplets are not generated in exactly the same position in all images, and the distortion of the superimposed image cannot be avoided. As shown in Figure 11.5, the length of the droplet tail differs according to the position of each droplet from the moment it is detached from the nozzle surface. The tail length increases as it is closer to the nozzle surface, and a spherically shaped droplet is eventually formed as it moves away from the nozzle. In Figure 11.5, the droplet with a long tail has a clear tail image, while the image becomes blurrier as the droplet tail becomes shorter. This yields the vision recognition error when extracting the droplet boundary during image processing. In other words, the droplet tail length affects the measured droplet volume. Since the droplet tail length changes when the voltage is adjusted during the droplet volume measurement and equalization process, it causes the measurement error. The measurement error can be reduced by measuring the droplet volume under conditions in which the droplet tail is not generated around the nozzle surface where the spherical droplet is formed. However, this would be meaningless, since such a condition is different from the actual jetting conditions. As the droplet tail during printing is related to the straightness of the droplet flight, jetting to the accurate position is enabled when the droplet tail has a certain length. Although the droplet speed of 4–5 m s−1 is appropriate for printing, most inks produce an elongated droplet tail at such a droplet speed. Therefore, the droplet volume equalization will be feasible only when performed with ink that does not form a droplet tail around the nozzle surface when jetted under typical inkjet printing conditions.

163

164

11 Equalization of Jetting Performance

V = ∆D

Delay time = t1

(t2 − t1)

∆D Delay time = t2 Figure 11.6

Droplet speed measurement.

11.1.4 Speed Equalization

According to the laboratory data of inkjet printhead makers and studies, the droplet volume and speed are proportional at certain jetting conditions. Therefore, drop volume can be equalized by measuring and controlling the droplet speed without measuring the droplet volume. This method is called speed equalization, since it equalizes the jetting speeds of droplets out of the nozzles. The droplet speed measurement and equalization process is the same as that of the droplet volume measurement and equalization process except that droplet speed is measured instead of droplet volume, as shown in Figure 11.3. As in the droplet volume measurement and equalization process, the captured image can be distorted during the droplet speed measurement and equalization process, too. However, since only the two bottommost droplet positions are needed for measuring the droplet speed, as shown in Figure 11.6, the error from image distortion is relatively smaller, typically around 1 pixel distortion. With the unit pixel size of 1.7 µm, the droplet speed deviation is around 0.08 m s−1 under typical inkjet jetting conditions. Figure 11.7 shows the droplet speeds out of the nozzles after applying the initial voltage of 80 V to the inkjet printhead by Fuji Dimatix. After repeating the droplet speed equalization process three times, the initial droplet speed deviation of 36% is reduced to 5.6%. As the droplet speed deviation is smaller than the droplet volume deviation, the droplet speed measurement and equalization process can be considered a more reliable equalization method than the droplet volume equalization method. As shown in Figure 11.8, droplets are not aligned before the droplet speed measurement and equalization process but they are aligned on the same line after the process. Therefore, droplet speed equalization not only helps the droplet volume equalization but also improves the droplet’s positional accuracy. 11.1.5 Problems with the Droplet Equalization Methods on the Fly

Equalizing the droplet volume by measuring the droplet volume and speed using a drop watcher reveals several problems, which are described as follows.

11.1 Equalization of the Droplet Volume on the Fly

165

6 Before DPN After DPN

Velocity (m s−1)

5.5 5 4.5 4 3.5 3

1

11

21

31

41

51

61 71 81 Nozzle number

Figure 11.7 Droplet speed deviation before and after the droplet speed measurement and equalization process.

(a)

(b)

Before droplet speed equalization

After droplet speed equalization

Figure 11.8 Comparison of droplet speed uniformity.

91

101

111

121

166

11 Equalization of Jetting Performance

11.1.5.1 Distortion of the Captured Droplet Images As explained before, the droplet images on the fly acquired by the drop watcher are not just a single image of a droplet but the superimposed one from tens to hundreds of instantaneous images frozen by the synchronization of the jetting frequency and the flashing moment of a strobe light-emitting diode (LED) while the camera iris is open. Therefore, there will be no image distortion if the droplet position is exactly the same in all images. However, there are minute differences among the jetted droplets because of the jetting condition changes (voltage, pressure drop, nozzle area wetting, acoustic pressure residual in the head, etc.) at each jetting moment, and the droplet position changes at a certain moment. As a result, the superimposed droplet image has an unwanted distortion. Since diffraction of the light depends on the brightness and position of the LED, the distortion of the captured images can occur on the edge of the droplet. It is the same phenomenon encountered as when it is difficult to see an object in front of a bright light. Therefore, unless the same lighting is applied in parallel with the camera on the drops in an image, there will be distortion on the edge of each droplet. Furthermore, an LED will remain lighted for a certain period, typically around 1 µs, when it is triggered. Since the droplet continues moving during the LED on state and the ejected droplet speed is around 5 m s−1 , the droplet will move 5 µs while the LED remains lighted so that the motion blurring affects the sharpness of the captured droplet image. 11.1.5.2 Relation between Droplet Volume and Speed As mentioned before, the most important attribute when manufacturing various display devices with inkjet printing is to minimize the droplet volume deviation across the nozzles of each head. Therefore, the linear relation between the droplet volume and speed is the essential factor when performing the droplet speed measurement and equalization process. However, previous studies indicate that the droplet volume is not linearly proportional to the droplet speed. In other words, the droplet speed deviation of 1%, for example, does not necessarily equate with the droplet volume deviation of 1%.

11.2 Droplet Volume Equalization with Sessile Droplets

Various methods have been proposed to solve the problem in equalizing the droplet volume out of the nozzles in the inkjet print head by measuring the droplet volume and speed on the fly. The biggest problem with the drop watcher is the distortion of the captured images because the droplets are not stationary. Therefore, the size measurement of stationary sessile droplets on a substrate has been proposed as an alternative. Figure 11.9 illustrates the stationary sessile droplet on the substrate, with the assumption that the sessile droplet on the substrate constitutes a spherical form,

11.2 Droplet Volume Equalization with Sessile Droplets

h

q R

r q

Figure 11.9 Illustration of a sessile droplet on a substrate.

where R is the radius of the sessile droplet and θ is the contact angle of ink against the substrate; the droplet volume can be calculated by the following equation. π Volume = h2 (3r − h) 3 π 3 (1 − cos θ)(2 + cos θ) (11.1) = R 3 sin θ(1 + cos θ) Assuming that the contact angle of ink against the substrate does not deviate above a certain limit over the entire area of the substrate; the above equation indicates that the droplet volume can be equalized by controlling the sessile droplet diameter on the substrate. 11.2.1 Equalizing the Droplet Volume with the Measurement of Sessile Droplets

To equalize the droplet volume with the measurement of sessile droplets, the surface properties of the substrate must be uniform. Generally, glass is most widely used as the substrate for the sessile droplet measurement and equalization process. Metalized/indium tin oxide (ITO) glass rather than a bare one without any surface treatment or photoresist (PR)-coated substrate provides more uniform surface conditions. A hydrophilic substrate is not recommended as the ink spreads too widely and individual circles will not be formed. There should be no foreign substances to distort the wetting state of sessile droplets on the substrate, as the surface should always maintain the uniform wetting state. Figure 11.10 shows an image of sessile droplets captured with a typical black and white camera. The sizes of all sessile droplets in the camera focus of view (FOV) of each nozzle are averaged and then the jetting voltage of each nozzle is adjusted accordingly. This sessile droplet measurement and equalization process is repeated several times until all the sessile droplets fall within the target specification.

167

11 Equalization of Jetting Performance

Nozzle #n

Nozzle #n+1

Nozzle #n+2

Figure 11.10 Camera image of sessile droplets.

36

32

Volume (pl)

168

28

24

20

R-Square: 0.99754 SD: 0.55567 N: 5 P: 0.000146

16 76

80

84 88 Sessile droplet size (µm)

92

96

Figure 11.11 Relationship between the sessile droplet size and volume (SE-DPN, Fuji Dimatix).

11.2.2 Results of the Sessile Droplet Measurement and Equalization Process

Figure 11.11 shows the measurement results of the sessile droplet size and volume with an inkjet printhead, SE-DPN, by Fuji Dimatix. It is found that the relationship between the sessile droplet size and volume can be approximated to have a linear relationship in the minute length scale range, despite Eq. (11.1). Therefore, the volume can be equalized using size equalization. Figure 11.12 shows the initial sessile droplet sizes with the initial voltage of 80 V to the inkjet printhead and the equalized sessile droplet sizes after the sessile droplet measurement and equalization process was performed three times.

11.2 Droplet Volume Equalization with Sessile Droplets

84 Before DPN After DPN

82

Diameter (µm)

80 78 76 74 72 70 0

20

40

60

80

100

120

Nozzle number Figure 11.12 Sessile droplet size change before and after the sessile droplet measurement and equalization process.

It shows the initial sessile droplet size deviation of 5.52% is reduced to 0.85% after iterative equalization. On visual inspection of the inkjet-printed patterns, one can observe unwanted printing swath marks between nozzles before but they disappear after the sessile droplet measurement and equalization process. 11.2.3 Usefulness of the Sessile Droplet Measurement and Equalization Process

The droplet size measurement and equalization processes are valid when the jetting conditions are the same as the actual jetting conditions. As explained before, the drop watcher must have the inkjet printhead aligned at 0◦ where the nozzles are arranged in the direction perpendicular to the optical axis of the CCD camera. However, an inkjet printhead is rotated at a certain degree, called ‘‘saber angle,’’ to meet the required print resolution in actual cases. As a result, the droplets from each nozzle of the inkjet printhead are jetted with a certain amount of delay time corresponding to the saber angle. This unsynchronized jetting creates the cross talk among the nozzles. Therefore, when the droplets are jetted at a certain saber angle, different from 0◦ , after the droplet volumes of the inkjet printhead are equalized at the saber angle of 0◦ , the droplet volume from the nozzle could be different from that under the equalized condition. Figure 11.13 shows the change in droplet sizes when applying the voltage condition, adjusted at the saber angle of 0◦ to have the droplet volume variation within 1%, to the inkjet printhead at the saber angle of 45◦ . The droplet size uniformity became worse and hence another sessile droplet measurement and equalization process was conducted. The sessile droplet size uniformity at the saber angle of 45◦ was 0.92% better than the equalization result at the saber angle of 0◦ .

169

11 Equalization of Jetting Performance 82 (A): DPN at head 0° (A): applied to head 45° (B): DPN at head 45°

80 Diameter (µm)

170

78

76

74 0

20

40

60

80

100

120

Nozzle number

Figure 11.13 Change in droplet size due to the inkjet printhead rotation, (a) just after the droplet volume equalization at the saber angle of 0◦ , (b) at the saber angle of 45◦ , and (c) after the reequalization at the saber angle of 45◦ .

Move Camera

Glass Move Light Figure 11.14 Mimetic diagram of equalization using transmittance.

11.2.4 The Droplet Volume Equalization Process Using Light Transmittance

When a hydrophobic pattern is formed on a transparent substrate, as shown in Figure 11.14, the jetted ink can be confined inside the space of the formed pattern. This is a typical process with inkjet printing for the fabrication of a TFT LCD color filter. The droplet volume equalization process using light transmittance equalizes the ink volume jetted from the nozzles of the inkjet printhead by first checking the deviation in light transmittance with a bottom light and top camera and then compensating for the deviation. The accuracy of this method is very high because the inspection is performed under the same conditions as actual jetting.

Further Reading

171

272.5 272 271.5 271 270.5 270 269.5 269 268.5 268

1H_2 scan 1H_3 scan 1H_4 scan 1H_5 scan 1H_6 scan

2H_7 scan 2H_8 scan 2H_9 scan 2H_10 scan 2H_11 scan

Figure 11.15

Light transmittance deviation after the droplet volume equalization process.

Furthermore, the measurement precision is improved because the inspection is performed after depositing the multidroplets inside the pattern. However, the droplet volume equalization process using light transmittance requires a prepattern with certain sections and must use a transparent substrate such as glass to transmit the light from the bottom light source. Therefore, its use is limited to such applications as TFT LCD or polymer light emitting diode (PLED). 11.2.5 Result of the Droplet Volume Equalization Process Using Light Transmittance

When applying the droplet volume equalization process using light transmittance to a TFT LCD color filter, it has the benefit of fitting the color coordinates with the target value. It will improve the deviation of the droplet volume of nozzles and retain the target total volume of droplets in the subpixel at the same time. As the adjustment range can be arbitrarily set according to the level of transmittance deviation at 1024 gray level, this method is very useful compared to other droplet volume equalization methods. Figure 11.15 shows the result of the droplet volume equalization process using light transmittance to retain the target light transmittance level through repeated inspections.

Further Reading Kalaaji, A., Lopez, B., Attane, P., and Soucemarianadin, A. (2003) Breakup length of forced liquid jets. Phys. Fluids, 15, 2469. Gonzales, H. and Garcia, F.J. (2006) Comment on ‘‘Breakup length of forced liquid

jets’’. Phys. Fluids, 18, 019101 [Phys. Fluid, 15, 2469 (2003)]. Ritter, R.C., Zinner, N.R., and Sterling, A.M. (1974) Analysis of drop intervals in jets modeling obstruction of the urinary tract. Phys. Med. Biol., 19, 161.

61

55

49

43

37

31

25

19

13

7

he

ad

#1

267.5

172

11 Equalization of Jetting Performance Quian, S.X., Snow, J.B., Tzeng, H.M., and Chang, R.K. (1986) Lasing droplets: highlighting the liquid-air interface by laser emission. Science, 231, 486. Vago, N., Spiegel, A., Couty, P., Wagner, F.R., and Richerzhagen, B. (2003) New technique for high speed microjet breakup analysis. Exp. Fluids, 35, 303. Doring, M. (1982) Inkjet printing. Philips Tech. Rev., 40, 192. Martin, G.D., Hoath, S.D., and Hutchings, I.M. (2008) Inkjet printing – the physics of manipulating liquid jets and drops. J. Phys. Conf. Ser., 105, 012001. Zhou, Y. (1998) Measurement of the displacement of a shear mode piezoelectric transducer using laser Doppler vibrometer. Proceedings of the IS&T NIP14, p. 23. Kwon, K.S. (2008) Inkjet Status monitoring using meniscus measurement Proceedings of the IS&T NIP24 and DF2008, p. 134. Kwon, K.S. and Kim, W. (2007) A waveform design method for high speed inkjet printing based on self sensing measurement. Sens. Actuators A, 140 (1), 75. Creagh, L., McDonald, M., and Letendre, W. (2004) Inkjet printhead as a precision deposition tool in manufacturing FPDs. SEMICON China, FPD Manufacturing Conference. MicroFab Technologies, Inc. (1999) Drive Waveform Effects on Inkjet Device Performance, Microfab technote, p. 99. Bogy, D.B. and Talke, F.E. (1984) Experimental and theoretical study of wave propagation phenomena in drop on demand inkjet devices. IBM J. Res. Dev., 28, 314. Chen, A.V. and Basaran, O.A. (2002) A new method of significantly reducing drop on demand drop production. Phys. Fluids, 14 (1), L1.

Kwon, K.S. (2009) Speed measurement of ink droplet by using edge detection techniques. Measurement, 42, 44. Kim, Y.D., Kim, J.P., Kwon, O.S., and Cho, I.H. (2009) The synthesis and application of thermally stable dyes for inkjet printed LCD color filters. Dyes Pigm., 81, 45. Tsuda, K. (1993) Color filter for LCDs. Displays, 14, 115. Ohmi, T. (2004) Manufacturing process of flat display. JSME Int. J. Ser. B (Jpn. Soc. Mech. Eng.), 47, 422–428. Takamatsu, T., Ogawa, S., and Ishii, M. (1991) Color Filter fabrication technology for LCDs. Sharp Tech. J., 50, 69. Son, Y. and Kim, C. (2009) Spreading of inkjet droplet of non-Newtonian fluid on solid surface with controlled contact angle at low Weber and Reynolds numbers. J. Non-Newtonian Fluid Mech., 162, 78. Koo, H.S., chen, M., Pan, P.C., Chou, L.T., Wu, F.M., Chang, S.J., and Kawai, T. (2006) Fabrication and chromatic characteristics of the greenish LCD Color Filter layer with nano particle ink using inkjet printing technique. Display, 27, 124. Worthington, A.M. (1877) On the forms assumed by drops of liquids falling vertically on a horizontal plate. Proc. R. Soc. Lond., 25, 261. Calvert, P. (2001) Inkjet printing for materials and devices. Chem. Mater., 13, 3299. Meinhart, C.D. and Zhang, H. (2000) The flow inside a microfabricated inkjet printhead. J. Microelectromech. Syst., 9 (1), 67. Dong, H. and Carr, W.W. (2006) Visualization of drop on demand inkjet: Drop formation and deposition. Rev. Sci. Instrum., 77, 085101.

173

12 Inkjet Ink Formulations Alexander Kamyshny and Shlomo Magdassi

12.1 Introduction

Inkjet printing is a nonimpact, dot-matrix technology. It is a low-cost, reliable, quick, and convenient approach for printing digital files. Inkjet has been utilized since the 1950s and has seen commercial success since the 1970s [1]. In the past two decades, inkjet printing has grown to be an important topic in scientific research and technology [1–10]. The main advantages of inkjet printing, compared to other deposition methods such as pad printing, screen printing, spraying, and photolithographic printing, are one-step processing, low-cost and compact equipment, and applicability to various substrates. In addition to conventional graphic applications, inkjet printing has been adapted for nongraphic applications, such as microfabrication of various devices, for example, transistors, integrated circuits, conducting polymer devices, structural polymers, and ceramic parts [10, 11]; biomaterials and even printed scaffolds for growth of living tissues [4, 10]; as well as for building complicated 3D objects [3] and microelectromechanical systems (MEMS) [12]. In the electronics industry, manufacturing electronic devices such as flexible displays, radio frequency identification (RFID) tags, sensors, organic light-emitting diodes (OLEDs), photovoltaic (PV) devices including solar cells (SCs), batteries, and printed circuit boards (PCBs) by inkjet printing of conductive inks can provide low-cost means of manufacturing large-area electronics on a wide range of substrates (paper, polymers, glass, metals, ceramics, etc.) and is attracting tremendous interest [2, 5, 9, 11, 13–16]. In theory, inkjet is simple: droplets of ink are jetted from a small orifice in a printhead directly to a specified position on a substrate as a result of pressure developed after an electronic signal has been sent to the printhead to create an image (most of the inkjet printers are based on the drop-on-demand (DOD) methods, mainly thermal and piezoelectric generation of droplets [1]). However, reliable operation depends on careful design, implementation, and operation of a whole system, and very often is determined by the specific application of the inkjet printing. One of the main challenges is proper ink formulation, which must have physicochemical properties specific to the various printing devices and Inkjet-based Micromanufacturing, First Edition. Edited by J. G. Korvink, P. J. Smith, and D.-Y. Shin.  2012 Wiley-VCH Verlag GmbH & Co. KGaA. Published 2012 by Wiley-VCH Verlag GmbH & Co. KGaA.

174

12 Inkjet Ink Formulations

substrates, and it should also be suitable for obtaining specific functional printing [17, 18]. Owing to the complex nature of inkjet inks, their design and preparation is often very complicated. In addition to the conventional requirements, such as long shelf life and proper color properties, the ink must have physicochemical properties that are specific to the various printing devices. For example, each printhead has a specific range of surface tension and viscosity, which enable proper jetting (e.g., piezoelectric printheads usually function at ink viscosity in the range of 8–15 cP, while thermal printheads require viscosity below 2 cP). Proper selection of the ink vehicle is also very important. Such selection can be affected tremendously not only by the requirements regarding the quality of the printed pattern on a specific substrate but also by the end use and the printing environment. For inks that bring a sophisticated functional property beyond graphic performance, additional difficulties can arise. For example, in addition to usual inkjet ink requirements, conductive inks should provide good electrical conductivity of the printed patterns. Obtaining such a functional property often presents conflicting directions for the scientists who prepare the inks: in conductive ink, which is composed of metallic nanoparticles, the best way to stabilize the dispersion against coagulation of the particles is by using a polymeric stabilizer. However, in order to obtain good electrical conductivity after printing, the metal particles should form continuous percolation paths between the particles in the printed pattern. Very often, it is difficult to achieve this percolation because of the presence of the polymeric layer, which acts as an insulator between the particles. Another example of conflicting requirements is UV-curable inks. To obtain high throughput in an industrial printing system, curing of the printed pattern should be rapid, providing good printing resolution. However, rapid curing does not favor the spreading of the ink droplets over large areas, so the ink coverage is low, which may be a problem in certain applications. Overcoming the low coverage would require placing more ink droplets on the substrate, which obviously means greater ink consumption and cost. Therefore, while formulating new inkjet inks, one has to take into account the effect of each component on the overall performance of the ink, from shelf life in the cartridge, through jetting, to its behavior on the substrate and the effect on health and environment at the stages of manufacturing and printing. In this chapter, the principles of inkjet ink formulations and the main parameters providing optimal overall performance of the inks, such as ink preparation methods and compositions, which affect jetting performance, and interaction of ink with substrates, are discussed.

12.2 Ink Formulation

Inkjet ink is composed of a functional material and a liquid vehicle, which is a carrier of this functional material (this definition also holds for hot-melt ink, in

12.2 Ink Formulation

175

Inkjet ink formulations

Solvent based

Solution of functional material

Water based

Solution of functional material

Dispersion of functional material

Dispersion of functional material

Heterophase system 100% solids

UV curable

Hot-melt

Figure 12.1 General types of inkjet ink formulations.

which the vehicle is solid before and after printing, but is liquid during printing). The vehicle is composed of liquids, additives that bring a specific function (e.g., surfactant, preservative, and photoinitiator), and usually also a polymer, which enables the binding of the functional molecules (e.g., colorants and conductive particles) to the substrate after printing. The selection of various components of the vehicle is tailored according to the printing technology and the final function of the printed pattern. It is common to divide the inks according to the nature of the vehicle. In general, there are three main types of inks (Figure 12.1): aqueous (water-based inks), nonaqueous organic (solvent-based inks), and 100% solid inks. A common feature of the first two types of inks is that once the ink droplet is placed on the substrate, the liquid vehicle should be removed, usually by evaporation. In the third type, there is no solvent evaporation; the ink undergoes a phase transformation from solid at room temperature to liquid (hot-melt ink) at jetting temperature, and again to solid on the substrate, or is initially liquid (UV ink) but becomes solid after polymerization of the liquids on the substrate (UV-curable inks). Heterophase ink may be based on a water-dissipatable polymer (e.g., polyester) carrying bound water-insoluble colorant [19], micellar system with dye solubilized in micelle interior [19–21], oil-in-water microemulsions, and miniemulsions containing water-insoluble dye dissolved in an organic phase [21–23], and liposomes [24]. Functional materials can be dissolved in the ink vehicle (dye-type inks) or dispersed in it in the form of micro- or nanoparticles (pigment-based inks) (Figure 12.1). In the last case, the addition of a stabilizing agent to prevent aggregation of particles and to provide the colloidal dispersion stability is required. When used as colorants, pigment-based inks are superior to dye-based inks for lightfastness and waterfastness. In principle, dye inks, which are solutions,

176

12 Inkjet Ink Formulations

are thermodynamically stable, while most pigment inks are only kinetically stable. In the following sections, we discuss the components of various inkjet ink compositions, mainly for DOD with regard to their functionality and effect on ink performance. 12.2.1 Functional Materials

At present, inkjet technology, in addition to graphic applications, has found numerous applications in many fields of science and technology. The number of functional materials, which give the ink its primary function, is very high: colorants (pigments or dyes) for graphic art; metal nanoparticles, organometallic compounds, and carbon nanotubes for printing conductive patterns; polymers for printing light-emitting diodes; UV-curable monomers for printing 3D structures; ceramics; biomedical materials; and even living cells for printing an organ [1–7, 11, 18, 25]. These functional materials can be soluble in the ink vehicle (e.g., colorants, polymers, organometallic compounds, and biomaterials) or dispersed in it (Figure 12.1). The content of the functional material in ink formulation is determined by its solubility (for dye-type inks) or dispersibility and stability (for pigment-type inks) in the presence of all the ink components as well as by the whole inkjet system performance, and can vary in a wide range from 0.5 to 10% by weight for colorant up to about 60% for metal nanoparticles in conductive inks [7, 19, 26]. For inks that do not contain undissolved material, the instability is due to interactions between the ink components such as polymerization of monomeric components, phase separation due to temperature variations, and adsorption of ink components onto the container walls. For inks containing a high concentration of dispersed material, the main problem is aggregation of particles followed by irreversible coagulation and sedimentation if the density of particles is high compared to the density of the liquids (e.g., metallic particles in solvent-based inks) and carbon black, which tends to form large aggregates in water-based inks. In order to prevent aggregation, a mechanism to overcome the attraction is required. Electrical repulsion, which can be obtained if the surface of the pigment particle bears electrical charges, is such a mechanism. For example, if an anionic surfactant, such as sodium dodecyl sulfate (SDS), is adsorbed on the surface of hydrophobic particles (e.g., carbon black), it will impart negative charges to the pigment surface, and electrical repulsion between the particles will prevent their aggregation as described well by the DLVO theory [27]. The electrostatic stabilization mechanism is effective in systems having a high dielectric constant, and therefore is mainly important for water-based inks. Stabilization of particles can also be achieved by a steric mechanism [27, 28], in which polymeric molecules, which are adsorbed onto the surface of the pigment, provide steric repulsion. For example, stabilization of carbon black particles can be achieved using a polymer with hydrophobic groups, which can bind to its surface, and also has hydrophilic

12.2 Ink Formulation

segments that are soluble in water. This stabilization mechanism is very effective for both aqueous and nonaqueous inks, and there is a large variety of commercially available polymeric dispersants, such as Efka, Tegsperse, Solsperse, and Disperbyk [29, 30]. In order to achieve a stable dispersion of pigment-containing ink, various dispersing agents and their optimal concentrations should be evaluated. The selection should provide a dispersant, which has anchoring groups enabling its adsorption on the pigment surface, while enabling the other groups of the dispersant to be extended into the solution. The dispersant concentration is of great importance: there is an optimal concentration of the dispersant for good ink performance. For example, a high concentration of polymeric dispersant in metal-based conductive ink will prevent close contact of nanoparticles in the printed pattern even after postprinting treatment, thus making it impossible to obtain a conductive pattern. In this case, the use of as low a concentration of polymeric stabilizer as possible is recommended. The dispersant concentration has a crucial effect on the viscosity of the dispersion, which has to be tailored according to the printhead requirements. It should be noted that to achieve stable dispersions of functional material in the ink vehicle, the particles should be small enough, typically less than 200 nm. 12.2.2 Solvents

The solvents are the primary ink vehicles that dissolve or suspend the functional material and other components of the ink. The solvent weight content in inkjet inks is usually in the range of ∼25–80 wt%. Typical solvents are organic liquids (solvent-based inks) and water (water-based inks). 12.2.2.1 Solvent-Based Inks Typical solvents are oxygenated organic compounds such as alcohols, methyl ethyl ketone, ethyl acetate, propylene carbonate, ethylene glycol butyl ether acetate, diethylene glycol monobutyl ether, and more rarely hydrocarbons such as toluene and tetradecane [2, 19, 31]. In many formulations, mixtures of various solvents are used, which enable tailoring ink properties such as viscosity, evaporation rate, and surface tension. Solvent-based inkjet inks have been widespread for many years in industrial graphic applications because of their exceptional print quality, image durability, fast drying time, and compatibility with a wide range of substrates (metal, glasses, ceramics, plastics, wood, leather, foods, etc.). Solvent inks can be formulated with either soluble functional materials (e.g., dyes) or as dispersions of submicron and nanoparticles. For example, solvent-based conductive inks containing dispersed silver and copper nanoparticles were recently demonstrated to be applicable for microfabrication of electronic devices and SCs [7]. Disadvantages of solvent-based inks include environmental and health issues (hazardous volatile organic compounds and strong odor), as well as the

177

178

12 Inkjet Ink Formulations

possibility of blocking the printhead nozzles while using fast-drying, volatile organic liquids. 12.2.2.2 Water-Based Inks Water-based or aqueous inks are prevalent in office and home printers. They are relatively inexpensive and environmentally friendly, but their application is limited for several reasons. Water-based inks require porous or specially treated substrates or even lamination to impart durability, and the ink tends not to adhere to nonporous and plastic substrates. In addition, many piezoelectric industrial printheads are incompatible with water-based ink formulations, although this is changing in some part because of market demand for printing, for example, water-based biological or food contact fluids [1]. 12.2.3 Hot-Melt (Phase-Change) Inks

Hot-melt, or phase-change, inks are usually based on natural waxes or synthetic compounds containing long hydrocarbon chains, such as stearic acid, behenic acid, didodecyl sebacate, and stearamide [19]. Hot-melt inks are melted before jetting, and typically the operating temperatures are in the range of 50–120 ◦ C. Hot-melt ink formulations, especially loaded with colorants, mainly pigments (e.g., titanium dioxide), can be easily designed to have good covering power that is especially important for printing on nonporous and dark substrates, for example, black plastics. Advantages of phase-change inks are their very fast solidifying rate and better stability than pigment inks, since the dispersing matrix is a solid phase during storage. In addition, they are environmentally friendly. It is also relatively easy to control the quality of the print because they do not tend to spread because of rapid solidification. Their primary disadvantages are the lack of durability and poor abrasion resistance. 12.2.4 UV-Curable Inks

UV-curable inks and coatings have been used in printing markets for many years, and inkjet is now an established and robust deposition tool for UV-curable fluids [1]. UV-curable inks are designed to remain as a stable liquid until irradiated with light of a particular wavelength and intensity. UV ink formulations contain monomers and oligomers (usually acrylate derivatives), functional material (e.g., pigments of dyes dispersed or dissolved in the reactive carrier), photoinitiators, and various additives, which aid the UV curing process [25]. UV curing can proceed by two mechanisms: free radical polymerization, which is the predominant route (typical photoinitiator α-hydroxyketone) or cationic polymerization (typical photoinitiator benzophenone).

12.3 Ink Parameters and Additives

UV inks are now successfully employed in a variety of inkjet applications, such as product coatings and package labeling. The main advantages of UV-curable inks are little or no volatile organic chemicals or hazardous air pollutants, and durability and abrasion resistance of printed patterns because of the cross-linked nature of the film [25]. Current limitations are in edible and food contact applications. Disadvantages include mainly cost of materials and requirements of the UV curing hardware.

12.3 Ink Parameters and Additives 12.3.1 Rheology Control

The rheology of the ink is of great importance for its performance during jetting and spreading on the substrate, and is affected by many parameters, such as solvent composition, the presence and concentration of polymeric additives, surfactants, humectants, and quality of the dispersion. Most current inkjet inks are Newtonian, that is, they have a constant viscosity over a wide range of shear rates, but non-Newtonian inks are also available [18]. The viscosity of inkjet inks is very low, usually below 20 cP, depending on the printhead (typically below 3 cP for thermal printheads). Typical viscosity can be controlled by proper selection of additives, such as long-chain glycols at concentrations of 1–3 wt%, and soluble, high-molecular-weight polymers [18, 19]. The viscosity of the ink may change during its storage. In the case of pigment-based inks (particles of functional material), the viscosity may increase because of flocculation of the particles. Since the ink has a low viscosity, sedimentation of pigment would occur if the pigment particles are large and/or have a high density (such as metals or ceramics). Therefore, great effort is made to obtain particles as small as possible. An increase in viscosity during storage of UV-curable inks is usually a result of polymerization reactions. In such cases, formation of oligomers is sufficient to cause a viscosity increase that would significantly affect the overall performance of the ink during printing. An interesting approach to prevent particle sedimentation is based on controlling the viscosity, so that the ink has a high viscosity during storage and low viscosity during jetting (at elevated temperatures). The extreme example of such an ink is the hot-melt ink, which is a solid at room temperature and a low-viscosity liquid above its melting point. By using this concept, sedimentation of pigment particles during storage is prevented. Taking advantage of the very different shear rates encountered during storage and during jetting opens the possibility of a non-Newtonian ink [18], namely, an ink with pseudoplastic behavior: during storage, the ink will have a high viscosity (low shear rate) and a low viscosity during jetting (very high shear rate). Obviously, such behavior would affect other ink properties such as ink flow through the printing system, jetting, and drop breakup.

179

180

12 Inkjet Ink Formulations

12.3.2 Surface Tension Modifiers

Surface tension of the ink is one of the primary factors that determines droplet formation in the printhead and spreading of the ejected drop on the substrate on contact [18, 32]. The surface tension is controlled by proper solvent composition and addition of surfactants. For example, the use of propanol as cosolvent in water-based inks results in a large decrease in surface tension, from 72.8 dyn cm−1 (pure water) to below 30 dyn cm−1 , depending on the propanol concentration. Significant decrease in surface tension with the addition of a cosolvent is usually obtained at relatively high cosolvent concentrations. Surfactants are usually used at very low concentrations, below 1 wt% and very often even below 0.1 wt%. If the surface tension is determined by the composition of the liquid medium, it will not change over time, and its value will be that of equilibrium conditions. However, if the surface tension is controlled by surfactants, the dynamic surface tension should also be considered. This parameter is important in cases in which a new surface is formed and is not yet covered with the surfactant molecule (such as during drop formation or spreading of a drop on a substrate); thus, initially the new surface has a high surface tension. Diffusion of the surfactant to the interface will result in a decrease in the surface tension until it reaches the equilibrium value. It should be emphasized that the resulting surface tension (static and dynamic) depends on all the components present in the ink and can be affected by interactions between the components, such as binding of surfactant to the dissolved polymer [33] and even due to migration of a plasticizer from the ink container. 12.3.3 Electrolytes and pH

The presence of electrolytes can cause stability problems during storage in the case of pigment-type functional material because of compression of the electrical double layers surrounding the particles that may result in their flocculation. Therefore, the concentration of electrolytes should be as low as possible. This is especially important for electrolytes composed of multivalent ions, such as Ca2+ . A typical chelating agent used in aqueous formulations is ethylenediaminetetraacetic acid (EDTA) at concentrations of 0.1–0.5 wt%. The pH is also important for water-based inks, since it can significantly affect the solubility of various components and the stability of the dispersed particulate functional material. For example, the solubility effect is often crucial when the ink contains a polymeric binder such as acrylic resin, which is insoluble at low pH. Since stabilization of dispersed materials is often achieved by adsorption of charged polymeric molecules, and their charge is a function of pH, this parameter is very important for ink stability. To control the required pH value, some ink formulations contain buffers. A typical buffer used in water-based inks is Trizma Base [19].

12.3 Ink Parameters and Additives

12.3.4 Foaming and Defoamers

Foaming is often observed in inks, which contain surfactants and polymers, and imposes a severe problem in inkjet performance. A solution to this problem is the addition of defoamers, which are molecules that cause the breakdown of the foam. The defoamers suppress stable foam formation by two main mechanisms: (i) reducing the surface tension in a local area to very low values, thus causing these areas to be thinned rapidly, and (ii) promoting drainage of liquid from the lamellae. Additives performing through the first mechanism should have limited solubility in the ink vehicle, and usually contain an immiscible moiety (e.g., a silicon derivative). A typical example of a defoamer performing through the second mechanism is tributylphosphate, which reduces surface viscosity [18]. In the case of an insoluble defoamer, the effect of phase separation can be significant during jetting (changing the wettability of the printhead) and after printing (forming surface defects due to low surface tension spots). Therefore, one should avoid, if possible, the use of defoamers or select defoamers that do not separate out during prolonged storage. In any case, the defoamer should be used at minimal concentrations. 12.3.5 Humectants

Humectants, which are used in water-based inks, are low-volatility liquids, which, when added to ink formulation, retard the evaporation of the ink in the printhead and prevent orifice clogging [17]. The humectants should be soluble in water. Typical humectants for aqueous inks are glycerol, diethylene glycol, polyethylene  glycols, and propylene glycol methyl ethers (Dowanols ). In addition to preventing the evaporation of ink in the nozzle of the printhead, humectants may sufficiently increase the time of drying of the printed image. Therefore, while selecting the optimal concentration of humectant, one should take into account these two contradicting effects. Typically, the humectant holds a substantial fraction of the solvent portion, in the range of 10–30 wt%. 12.3.6 Binders

Ink formulations usually include a binder, which provides good adhesion of the printed material to the substrate and may protect against abrasion. The binder is usually a polymeric resin that is soluble or dispersed in the ink vehicle and can be heat cured or UV cured. The choice of a binder is determined by good adhesion and substrate matching [2]. The molecular weight of resins, which are used as binders, is usually below 100 000, and often below 50 000, since the inks should have low viscosity. Common resins used are vinyl chloride/vinyl acetate copolymers, acrylic resins, and polyketone resins [31].

181

182

12 Inkjet Ink Formulations

12.3.7 Biocides

Since most dyes are organic molecules, they are good media for the growth of both bacteria and fungi, especially in aqueous solutions, and their colonies can cause orifice clogging. To prevent their growth, a biocide should be added to the ink. The choice of a biocide is determined by the species of the growing organisms. Ideally, the biocide should be very effective and have a broad spectrum of antimicrobial/antibacterial activity. Commonly used biocides are 1,2-benzisothiazolin-3-one and 2,6-dimethyl-m-dioxan-4-ol acetate. Biocides are usually relevant to water-based inks, and are added in small quantities, about 0.1–0.5 wt% [17]. 12.3.8 Examples of Inkjet Ink Formulations

• Solvent-based inkjet ink: Solvent mixture (glycol ether 10%, hydroxyketone 10%, alkyl lactate 40%, and acetoacetate 22%); blue pigment (Blue 44 or Blue 45, 3.5%); diluting agent (methanol, ethanol, or 2-propanol, 10%); surfactant (polyacrylate or polysiloxane, 0.05%); resin (vinyl chloride/vinyl acetate copolymer, 4%); and UV absorber (benzophenone or benzotriazole, 0.5%) [34]. • Water-based inkjet ink: Cosolvent mixture (2-pyrrolidone, 10% and tetraethylene glycol, 5%); carbon black pigment (Cabojet 300, 4%); binders (SMA 2000, copolymer of maleic anhydride, and a vinyl aromatic compound, 0.7%; and Mace 85-302-1, polyurethane, 0.7%); surfactants (nonionic Surfonol 465, 0.05%; lithium carboxylate anionic fluorosurfactant, 0.1%; Phospholan 9NP, and phosphate ester, 0.3%); and chelating agent (EDTA, 0.05%) [26]. • Hot-melt ink: Solid vehicle (stearone, 48%; methylenestearic acid amid, 30%; and amide resin, 20%) and a pigment, Neopen Yellow, 2% (the melting temperature of the composition is 110 ◦ C) [35]. • UV-curable ink: Polymerizable monomers and oligomers (dipropylene glycol diacrylate, 23.5%; neopentyl glycol propoxylate diacrylate, 18%; laurylacrylate, 17%; dipentaerythritol hexaacrylate, 3%; trimethyl propane ethoxylated triacrylate, 6%); free radical stabilizer (Genorad 16, 0.5%); photoinitiators (1-hydroxycyclohexyl phenyl ketone, 4%; Genocure LTM with absorption peaks at 253 and 368 nm, 4%; and Genocure PMP with absorption peak at 307 nm, 4%); and colorant (magenta dispersion SPF 586, 20%) [25].

12.4 Jetting Performance

While formulating the ink for inkjet printing, one should take into consideration the compatibility of each ink component with all the printer materials it meets. These materials can be very different: metallic parts (orifice plate, sensors, filters, etc.), which are sensitive to high or low pH of the ink vehicle, and plastic parts, which

12.4 Jetting Performance

may dissolve or swell on contact with organic solvent or monomers. Therefore, evaluation of the effect of individual components of ink formulation on various parts of the printing device is not sufficient, and the compatibility testing should be performed with the final ink. Usually, such evaluations are performed by dipping the various parts of the printhead and the ink supply system into the ink for a prolonged time under accelerated conditions. Once the ink meets the physicochemical requirements for a specific printhead, it should be tested for jetting performance [18, 36]. This section focuses on the general requirements that are crucial for obtaining an ink formulation with good jetting performance. 12.4.1 Drop Formation

Drop formation is probably the most important performance issue, since it affects the overall performance of the printing process. The jet stability and break-off behavior with respect to the fluid properties are stated in well-known theories (Navier–Stokes equations and Rayleigh theory [37]). During recent years, many computer simulations have been performed for predicting the jetting process in specific printheads and for establishing methodologies for selection of ink additives [38–40]. However, the optimal ink formulations are still based on empirical approaches. The liquid ejected from the 10 µm size nozzles travels with a specific momentum dictated by the kinetic energy of the droplet. The main parameters that govern the jetting process are the surface tension and the rheological properties of the ink. Depending on the printhead, typical inkjet inks should have a surface tension in the rage of 25–50 dyn cm−1 , and viscosity in the range of 1–25 cP. The relations between these parameters, together with ink density and nozzle diameter, give a ‘‘window of operation’’ in terms of Reynolds numbers [32, 41], Weber numbers [41], and the combined parameter, which is the Ohnesorge number [42]. As described above, the surface tension can be controlled by proper selection of the ink vehicle (for example, addition of iso-propanol to water causes a significant decrease in surface tension), or by adding surfactants that are effective at very low concentrations (usually silicon or fluorosurfactants). The viscosity can be controlled by the liquid composition, the presence of soluble polymeric additives (such as binders), and the quality of the dispersion (dispersant type and concentration). 12.4.2 Ink Latency

When the ink reaches the nozzle, the volatile components may evaporate, and the composition of the vehicle in the vicinity of the nozzle can differ from that of the bulk ink. This may result in changing the physicochemical parameters of the ink (e.g., an increase in viscosity, decrease in surface tension, and precipitation of some components), which cause the shifting away from the required properties, and the crucial problem of proper jetting after prolonged idle time arises

183

184

12 Inkjet Ink Formulations

(‘‘first-drop problem’’). The time that inkjet ink can successfully wait in an individual orifice, without jetting, is termed latency [19]. This period varies from a few seconds to a few minutes in commercial solvent- and water-based inks, which are used in DOD printers, and reaches many days for inks that do not contain volatile solvents, such as 100% UV inks. To obtain good latency, one should follow the following general guidelines [18]: 1) 2) 3)

4)

5)

For solvent-based inks, by using less volatile solvents and solvents with lower evaporation rates (the evaporation rate is usually given relative to butyl acetate) For water-based inks, by adding humectants, which are cosolvents capable of delaying the loss of water by binding water molecules (hydration) For water- and solvent-based inks, by maximizing the solubility of solids in the liquid by selecting solvent composition and cosolvents (e.g., N-methyl pyrrolidone for water-based inks), which enable good dissolution even after a fraction of the liquid medium has evaporated For pH-sensitive inks (required to keep high solubility or high zeta potential), by using nonvolatile pH control agents (e.g., an amine with a high boiling point) For pigment-containing inks, by selecting polymeric stabilizers that would keep the viscosity low enough even at high pigment load

12.4.3 Recoverability

Since any solvent- and water-based ink must eventually be fixed and dried on the substrate, it should have some volatile components. Therefore, from time to time the ink may not be jetted after idle periods. To restore the inkjet ink in the nozzle region to the initial jettable composition, it is usually ejected into a waste collector (‘‘spittoon’’) in the maintenance station. While formulating the inkjet ink, one should make sure that it has the capability to redissolve the ink crust, and is able to redisperse aggregated particles. This ability can be preliminarily tested by drying an ink sample, and testing how fast the dried ink returns to its original properties on contact with fresh ink. In practice, this test is also performed by measuring the quantity of ink that has to be ejected until proper jetting is enabled, and is termed recoverability. 12.4.4 Ink Supply

For inks to be properly jetted from the printhead, they pass through tubes and various filters. Therefore, for optimal performance of an inkjet printer, two parameters are especially important: rheology of the ink and the particle size in pigment-containing formulations. Most of the inks are Newtonian and have sufficiently low viscosity to enable the flow through millimeter-size diameter tubes. Filters in the ink supply system and printheads are aimed to prevent arrival of

12.5 Ink Interaction with Substrates

large particles to the nozzles. Aggregation of pigment particles usually causes an increase in viscosity, which can interfere with the ink flow through the ink supply system [29], block the filters, and thus decrease the flow rate over time. Therefore, to obtain good flow of the ink and good performance regarding the clogging issue, the size of pigment particles should be much smaller than the orifice diameter, optimally one-hundredth the size (e.g., for a typical orifice diameter of 40 µm, the particle size should be less than 400 nm). Control of the rheology of the ink can be achieved by proper selection of the components that affect it most significantly, such as polymers (binders, dispersants) and the phase fraction of the dispersed particles.

12.5 Ink Interaction with Substrates

Hieght (µm)

The behavior of ink drops on substrates depends on the physicochemical properties of both substrate and ink. For example, on the top of a nonporous substrate, the ink may cause clusters in the case of underwetting, high ‘‘dot gain’’ in the case of overwetting, and ‘‘bleeding’’ at the interface of two adjacent colors. In the case of porous substrates, the ink will penetrate into the pores very quickly, causing low optical density of the image, or even penetrate to the other side of the substrate [43]. Therefore, for obtaining good printing quality, matching the ink to the substrate is of crucial importance, and both substrate and ink should be tailored according to the specific applications. Tailoring of the substrate is achieved by specific coating, for example, by coating with a thin layer capable of effective absorption of the ink droplets after their contact with the substrate [43]. For optimal spreading and wetting (large dots), the ink should have a surface tension lower than the surface energy of the substrate [44]. Low surface tension values can be achieved by proper selection of the liquid vehicle and by adding wetting agents, which are available from many companies. Figure 12.2 shows the profile of a printed pattern on a glass slide (water-based inks, 25 wt% Ag dispersion, 17 nm average) in the presence and absence of a wetting agent. Without a wetting agent, most of the silver nanoparticles are located at the edges due to the well-known 6 5 4 3 2 1 0

A

B

0

200

400

600

800

Width (µm) Figure 12.2 Height profile of lines printed with silver dispersion, without (A) and with (B) a wetting agent.

1000

1200

1400

185

186

12 Inkjet Ink Formulations

‘‘coffee stain effect.’’ Addition of a wetting agent, Byk 348, at a concentration of 0.1 wt%, resulted in a much smoother profile with almost constant height [8]. If small dots are required (for better resolution, increased height of the printed pattern in conductive inks, etc.), the surface tension should be as high as possible in order to prevent fast wetting. The ink viscosity is also very important for spreading. A high-viscosity ink will lead to a small dot size compared to a low-viscosity ink, within a given time for the spreading process. The spreading usually stops when the viscosity becomes very high, for example, after UV curing, solvent evaporation, or solidification of hot-melt ink. Prevention of drop spreading can be achieved by including volatile solvents in the ink, which would cause a fast increase in viscosity on evaporation.

12.6 Nongraphic Applications

As mentioned, nowadays inkjet printing is performed to fabricate functional devices. An example is a photoluminescent device constructed in our laboratory by inkjet printing of the conductive pattern. The ink utilized for this application was produced by synthesis of silver nanoparticles by chemical reduction, and these nanoparticles were the ink ‘‘pigment.’’ The aqueous ink also contained additives such as humectant and a wetting agent (as described previously), and was printed by a low-cost office printer on a plastic substrate. However, fabrication of conductive wiring by inkjet printing of metal-based inks requires postprinting sintering, usually by heating to temperatures above 200 ◦ C [28]. Such a treatment is not suitable for most plastic substrates (e.g., polyethylene BaTiO3 ZnS ITO

PDAC

PET

Figure 12.3 Schematic illustration of the EL device and the printing process (top) and photograph of the EL device with printed silver electrodes (bottom). (Reproduced with permission from Ref. [9],  2010 American Chemical Society.)

References

terephthalate, PET) because of their sensitivity to high temperatures. We have recently reported on a new approach to achieve sintering of metallic nanoparticles at room temperature and obtaining high electrical conductivity of printed patterns on various substrates [9]. Such sintering is a result of spontaneous 3D coalescence of negatively charged metallic nanoparticles when they come into contact with oppositely charged polyelectrolyte, poly(diallyldimethylammonium chloride) (PDAC) deposited onto the substrate (preprinting treatment). Figure 12.3 presents a flexible, transparent electroluminescence (EL) device with silver electrodes fabricated by inkjet printing of a silver dispersion onto PDAC-precoated upper BaTiO3 layer. The voltage (100 V) applied between the indium-tin-oxide (ITO) and the Ag electrodes resulted in light emission (90 cd m−2 ) following the Ag printed pattern [9].

12.7 Conclusions

Inkjet printing is considered an innovative technique for high-quality and low-cost patterning. In addition to conventional graphic applications, inkjet printing is now adapted for microfabrication of various structures and devices.In the electronics industry, manufacturing electronic devices such as flexible displays, RFID tags, sensors, OLEDs, PV devices including SC, batteries, and PCBs by inkjet printing of conductive inks can provide low-cost means of manufacturing large-area electronics on a wide range of substrates. Obviously, the requirements from functional inkjet inks should be tailored according to the final required function, beyond the conventional industrial inks. Optimal inkjet ink formulation should meet the requirements of each step of the printing process, including manufacture and storage. Although inkjet printing is simple in concept, it is very difficult to satisfy all requirements for perfect performance. Some of the requirements of one step may contradict the requirements of another step, and therefore sometimes the optimal compositions are compromises between contradicting properties.

References 1. Hudd, A. (2010) in The Chemistry of

Inkjet Inks (ed. S. Magdassi), World Scientific, New Jersey, London, Singapore, pp. 3–18. 2. Nir, M.M., Zamir, D., Haymov, I., Ben-Asher, L., Cohen, O., Faulkner, B., and De La Vega, F. (2010) in The Chemistry of Inkjet Inks (ed. S. Magdassi), World Scientific, New Jersey, London, Singapore, pp. 225–254. 3. Napadensky, E. (2010) in The Chemistry of Inkjet Inks

(ed. S. Magdassi), World Scientific, New Jersey, London, Singapore, pp. 255–267. 4. Deravi, L.F., Wright, D.W., and Summerel, J.L. (2010) in The Chemistry of Inkjet Inks (ed. S. Magdassi), World Scientific, New Jersey, London, Singapore, pp. 269–282. 5. Subramanian, V. (2010) in The Chemistry of Inkjet Inks (ed. S. Magdassi), World Scientific, New Jersey, London, Singapore, pp. 283–317.

187

188

12 Inkjet Ink Formulations 6. Guttler, S. and Gier, A. (2010) in

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

The Chemistry of Inkjet Inks (ed. S. Magdassi), World Scientific, New Jersey, London, Singapore, pp. 319–339. Kamyshny, A. and Magdassi, S. (2010) Metal-based inkjet inks for printed electronics. Open Appl. Phys. J., 4, 19–36. Magdassi, S., Kamyshny, A., and Grouchko, M. (2006) Making connections: Aqueous dispersions of silver nanoparticles form conductive inkjet inks. Eur. Coat. J., 11, 54–59. Magdassi, S., Grouchko, M., Berezin, O., and Kamyshny, A. (2010) Triggering the sintering of silver nanoparticles at room temperature. ACS Nano, 4, 1943–1948. Calvert, P. (2001) Inkjet printing for materials and devices. Chem. Mater., 13, 3299–3305. De Gans, B.-J., Duineveld, P.C., and Schubert, U.S. (2004) Inkjet printing of polymers: state of the art and future developments. Adv. Mater., 16, 203–213. Fuller, S.B., Wilhelm, E.J., and Jacobson, J.M. (2002) Ink-jet printed nanoparticles microelectromechanical systems. J. Microelectromech. Syst., 11, 54–60. Forrest, S.R. (2004) The path to ubiquitous and low-cost organic electronic appliances on plastic. Nature, 428, 911–918. Ko, S.H., Pan, H., Grigoropoulos C.P., Luscombe, C.K., Fr´echet, J.M.J., and Poulikakos, D. (2007) All-inkget-printed flexible electronics fabrication on a polymer substrate by low-temperature high-resolution selective laser sintering of metal nanoparticles. Nanotechnology, 18, 345202-1–345202-8. Van Osch, T.H.J., Perelaer, J., De Laat, A.W.M., and Schubert, U.S. (2008) Inkjet printing of narrow conductive tracks on untreated polymeric substrates. Adv. Mater., 20, 343–345. Perelaer, J., Klokkenburg, M., Hendriks, C.E., and Schubert, U.S. (2009) Microwave flash sintering of inkjet-printed silver tracks on polymer substrates. Adv. Mater., 21, 4830–4834. Andreottola, M.S. (1991) in Handbook of Imaging Materials (ed. A.S. Diamond), Marcel Dekker, New York, pp. 525–544.

18. Magdassi, S. (2010) in The Chemistry of

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

Inkjet Inks (ed. S. Magdassi), World Scientific, New Jersey, London, Singapore, pp. 19–41. Pond, S.F. (2000) Inkjet Technology and Product Development Strategies, Torrey Pines Research, Carlsbad, CA, pp. 153–204. Moffat, J.R., Bedford, E.T., and Lauw H.P. (1999) Ink-jet inks for improved print quality. US Patent 5,935,309, filed April 20, 1998 and issued Aug. 10, 1999. Ben-Moshe, M. and Magdassi, S. (2010) in The Chemistry of Inkjet Inks (ed. S. Magdassi), World Scientific, New Jersey, London, Singapore, pp. 203–221. Magdassi, S. and Ben Moshe, M. (2003) Patterning of organic nanoparticles by ink-jet printing of microemulsions. Langmuir, 19, 939–942. Kamyshny, A., Ben-Moshe, M., Aviezer, S., and Magdassi, S. (2005) Ink-jet printing of metallic nanoparticles and microemulsions. Macromol. Rapid Commun., 26, 281–288. Breton, M.P., Noolandi, J., Isabella, M., Birkel, S., and Hamer, G.K. (1997) Ink compositions containing liposomes. US Patent 5,626,654, filed Dec. 5, 1995 and issued May 6, 1997. Edison, S.E. (2010) in The Chemistry of Inkjet Inks (ed. S. Magdassi), World Scientific, New Jersey, London, Singapore, pp. 161–176. Schmid, C. (2010) in The Chemistry of Inkjet Inks (ed. S. Magdassi), World Scientific, New Jersey, London, Singapore, pp. 123–140. Tadros, T.F. (2007) in Colloid Stability: The Role of Surface Forces, Part I (ed. T.F. Tadros), Wiley-VCH Verlag GmbH, Weinheim, pp. 1–22. Kamyshny, A. and Magdassi, S. (2010) in Nanoscience: Colloidal and Interfacial Aspects (ed. V. Starov), CRC Press, Boca Raton, London, New York, pp. 747–778. Spinelli, H.J. (1998) Polymeric dispersants in ink jet technology. Adv. Mater., 10, 1215–1218. Wong, R., Hair, M.L., and Croucher, M.D. (1988) Sterically stabilized polymer

References

31.

32.

33.

34.

35.

36.

37.

38.

colloids and their use as ink-jet inks. J. Imaging Technol., 14, 129–131. Samuel, J. and Edwards, P. (2010) in The Chemistry of Inkjet Inks (ed. S. Magdassi), World Scientific, New Jersey, London, Singapore, pp. 141–159. Kang, H.R. (1991) Water-based ink-jet ink. I. Performance studies. J. Imaging Sci., 35, 195–201. Nizri, G., Lagerge, S., Kamyshny, A., and Magdassi, S. (2008) Polymer-surfactant interactions: binding mechanism of sodium dodecyl sulfate to poly(diallyldimethylammonium chloride). J. Colloid Interface Sci., 320, 74–81. King, K.-Y., Ling, C.-K., Wang, S.-J., and Lu, Y.-C. (2008) Environmental-friendly and solvent-based inkjet ink composition. US Patent Application 2008/0006175A1, filed Nov. 9, 2006, Pub. date Jan. 10, 2008. Sawada, H. (1999) Hot-melt ink composition. US Patent 5,922,114, filed Jun. 12, 1997 and issued July 13, 1999. Lee, H.P. (1998) Progress and trends in ink-jet printing technology. J. Imaging Sci. Technol., 42, 49–62. Kang, H.R. (1991) Water-based ink-jet ink. I. Formulation. J. Imaging Sci., 35, 179–188. Chen, P.H., Chen, W.C., Ding, P.P., and Chang, S.H. (1998) Droplet formation of

39.

40.

41.

42.

43.

44.

a thermal sideshooter inkjet printhead. Int. J. Heat Fluid Flow, 19, 382–390. Wu, H.C., Lin, H.J., Kuo, Y.C., and Hwang, W.S. (2004) Simulation of droplet ejection for a piezoelectric injet printing device. Mater. Trans., 45, 893–899. Clay, K., Gardner, I., Bresler, E., Seal, M., and Speakman, S. (2004) Direct legend printing (DLP) on printed circuit boards using piezoelectric inkjet technology. Circuit World, 28, 24–31. Noguera, R., Dossou-Yovo, c., Lejeune, M., and Chartier, T. (2005) 3D fine scale PZT skeletons of 1–3 ceramic polymer composites formed by ink-jet prototyping process. J. Phys. IV: Proc., 126, 133–137. Girard, F., Attane, P., and Morin, V. (2006) Anew analytical model for impact and spreading of one drop: application to inkjet printing. Tappi J., 5, 24–32. Frenkel, M. (2010) in The Chemistry of Inkjet Inks (ed. S. Magdassi), World Scientific, New Jersey, London, Singapore, pp. 73–97. Podhajny, R.M. (1991) in Surface Phenomena and Fine Particles in Water-Based Coatings and Printing Technology (eds M.K. Sarma and F.J. Micale), Plenum Press, New York, pp. 41–58.

189

191

13 Issues in Color Filter Fabrication with Inkjet Printing Dong-Youn Shin and Kenneth A. Brakke 13.1 Introduction

The success of thin film transistor liquid crystal display (TFT LCD) industries was dependent on the capabilities to mobilize enormous and decisive capital investments and to employ highly skilled workforces. The infrastructure of semiconducting industries was also a key to success, although display industries pursued the ‘‘large area’’ rather than the ‘‘highly integrated’’ criterion. Only a few countries, namely the three Far East countries Japan, South Korea, and Taiwan, could supply such key elements at a low cost, and they prevailed in the display market. There was no more room for new players in this fully mature market. However, this market became too competitive and harsh, even for old players. Certainly, attempts were made to reduce the overall fabrication costs to survive, but all efforts based on the conventional fabrication technology, photolithography, have almost reached their limits. This is a strong motivation for seeking a new technology to replace photolithography. Simultaneously, new players, who could not but would love to join the race, have sought new display gadgets. They have found good potentials in e-Paper and organic light-emitting diode (OLED) displays, but have also found that they could not win the race with the same production tool, photolithography, with which old players were very familiar. This is why printing technologies are in the spotlight. We introduce the very basics of TFT LCD and the reasons for the demand to replace photolithography in the fabrication of TFT LCD color filters in Section 13.2. In Section 13.3, various printing technologies and the reasons why inkjet printing has been considered amongst others are described. Jetting and wetting issues for the fabrication of a TFT LCD color filter with inkjet printing is covered in Sections 13.4 and 13.5. Finally, some remaining issues related to inkjet printing are briefly described in Section 13.6. 13.2 Background

As mentioned above, printing technologies have attracted industries’ attention, especially in the field of large area displays, because of their unique ability to make Inkjet-based Micromanufacturing, First Edition. Edited by J. G. Korvink, P. J. Smith, and D.-Y. Shin.  2012 Wiley-VCH Verlag GmbH & Co. KGaA. Published 2012 by Wiley-VCH Verlag GmbH & Co. KGaA.

192

13 Issues in Color Filter Fabrication with Inkjet Printing

Photolithography 1. Substrate preparation

Printing

5. Developing

1. Substrate preparation Substrate

Substrate 2. Functional material deposition Functional material substrate

Functional material substrate 6. Etching

3. Spin coating of photoresist Substrate

Positive photoresist functional material substrate 4. UV exposure through photomask

7. Stripping

2. Printing

Photomask Substrate Functional material substrate

Figure 13.1

Substrate

Comparison of photolithography and printing technologies.

fine patterns on the order of tens to hundreds of micrometers over a substrate, the area of which reaches several square meters. Their fabrication step principle is much simpler than that of conventional photolithography, as shown in Figure 13.1. When photolithography and printing technologies are compared, the former is based on sequential subtractive processes, which means that most of the precious materials are wasted by dry and wet etching processes, and the latter is an additive process, which means that a functional material is only deposited where it is required. This pattern-wise different approach has been the driving force for printing technologies gaining a good deal of attention from large area display industries, because materials cost takes around 60% and over of the overall cost structure of a TFT LCD, the most popular large area display. Among many other components, such as glass, polarizer, liquid crystal (LC), backlight unit (BLU), and driver IC, a color filter (CF) takes up to 25% of the overall material cost, as can be seen in Figure 13.2. A color filter is a principle constituent of a TFT LCD, which converts white light from a BLU into three primary colors, red, green, and blue, as shown in Figure 13.3 [2]. The light emitted from a cold cathode fluorescent lamp (CCFL) or a light-emitting diode (LED) in the BLU passes through a diffuser sheet to homogenize its intensity, through prism sheets to collimate and enhance the light, and through the first polarizer. The role of the alignment layers is to guide the alignment orientation of the LC, and their gap is maintained by spacers. After the light passes through the color filter composed of subpixels colored red, green, and

13.2 Background

193

Target, 2% Glass, 10%

Chemicals, 4%

PCB, etc, 14%

Inverter, 5%

Color filter, 19% Backlight unit, 23% Polarizer, 10%

Liquid crystal, 6%

Other cell materials, 2%

Driver IC, 5%

Figure 13.2 Materials cost structure of 32 in. high-definition (HD) TFT LCD TV [1]; PCB, printed circuit board. Second polarizer

Black matrix

Subpixel (red, green and blue)

Glass Overcoat and common electrode B

G

R

B

G

R

B

G

R

Color filter Alignment layer Liquid crystal

Spacer White light

Alignment layer

Protector sheet Prism sheet (H) Prism sheet (V)

TFT array Glass

CCFL

First polarizer

CCFL reflector Backlight unit

Reflector sheet Diffuser sheet Light guide plate

Figure 13.3 Structure of a TFT LCD.

blue, it passes through the second polarizer, where only the light twisted by the LC passes through. Finally, the true color image emerges. Conventional color filter fabrication steps with photolithography are shown in Figure 13.4 [2], where each color is formed with seven repeated steps of cleaning, deposition, drying, prebaking, ultraviolet (UV) exposure, development, and postbaking. On the other hand, printing technologies have much simpler fabrication

194

13 Issues in Color Filter Fabrication with Inkjet Printing

UV exposure

Glass

Photomask

Cleaning Black matrix

Red Color photoresist deposition

Development

Drying and prebaking

Postbaking

Green

Blue

(a) Glass Black matrix Printing (baking)

Printing (baking)

Printing (baking)

Red

Green

Blue

(b) Figure 13.4

Fabrication steps of a color filter by (a) photolithography and (b) printing.

steps, where each color is formed with two repeated steps of printing and drying. When the total number of fabrication steps is counted, photolithography needs 21 steps but printing technologies require 6 steps. Besides the great simplification of fabrication steps, the material usage to form each color is greatly reduced in printing technologies, since they deposit the colorant ink only where required. According to rough calculations, around 70% of the colorant ink could be saved in an ideal situation, and other materials such as developer, purified water for cleaning and rinsing, photomask, and so on, are not necessary for color filter fabrication. It

13.3 Comparison of Printing Technologies

should be noted that the usage of printing technologies also takes less space for equipment, which contributes to the cost reduction of running a facility.

13.3 Comparison of Printing Technologies

The advantages of printing technologies over conventional photolithography for the fabrication of a TFT LCD color filter were described in Section 13.2. Various printing technologies, which have been recently considered by display industries, are briefly introduced and their advantages and disadvantages discussed. Screen printing is an age-old printing technology and has been used in display industries for a long time, especially for plasma display panels (PDPs). A screen-printable paste is spread on a screen, which is composed of a mesh and emulsion, and then transferred through the patterned openings by squeegeeing, as shown in Figure 13.5. Screen printing has been used for fabricating color filters for TFT LCD [3] as well as organic light-emitting devices [4]. However, screen printing is generally suitable where there is thick film formation because of the rheological characteristics of screen-printable pastes [5]. To hold the shape of a paste deposited on a substrate, the screen-printable paste normally exhibits a Bingham plastic behavior with a certain yield stress, like toothpaste. This makes screen printing hard to be employed for a thin film process. If the thickness of a colored subpixel is above a certain level, then it exceeds the planarization capability of the overcoating (OC), and subsequent processes, such as the formation of alignment layers and spacers, have less margin and would lead to a nonuniform birefringence of light through the LC. In addition, the screen deformation due to the squeegee contact and motion alters the critical dimension and positional accuracy of patterns. Therefore, this might not be suitable for patterning works that not only require high resolution but also positional accuracy with good thickness uniformity. Roll printing has been employed in graphic art and press industries. It utilizes a relief-patterned plate, which is rolled into a cylinder. If the ink on top of the protruded regions is transferred to a substrate, as shown in Figure 13.6a, then it is categorized as ‘‘flexography printing’’ and ‘‘letterpress printing.’’ Flexography printing has a relief plate made of an elastic rubber and hence the deformation of Squeegee Opening Mesh

Squeegee motion Ink

Emulsion Pattern Substrate Figure 13.5 Schematic illustration of screen printing.

195

196

13 Issues in Color Filter Fabrication with Inkjet Printing

Engraving

Protrusion

Ink

Pattern

Doctor blade Ink

Substrate (a)

(b)

Blanket roller

Ink

(c) Ink coating

Remaining ink

Blanket roller

(d)

Pattern

Cliché

Substrate

Figure 13.6 Schematic illustrations of roll printing types. (a) Flexography, (b) gravure, (c) gravure offset, and (d) reverse offset.

a soft relief plate generally limits its printing resolution, although its conformal nature might fit a nonuniform substrate at a high resolution [6]. On the other hand, it is called ‘‘gravure printing’’ if the ink in the engraved region is transferred to a substrate, as shown in Figure 13.6b [7, 8]. Generally, the relief plate is made of a hard material, so its countersubstrate must be soft. Otherwise, either the relief plate or the countersubstrate may be damaged because both must be pressurized to seamlessly remain in contact with each other for good ink transfer. To avoid unwanted damage to a hard substrate such as glass, gravure offset printing employs a blanket roller between the gravure cylinder and substrate, as can be seen in Figure 13.6c. However, the ink transfer from the engraved region of a relief plate to the blanket roller is problematic and hence reverse offset printing was devised, as seen in Figure 13.6d [9, 10]. Reverse offset printing starts with a uniform coating of the ink on a blanket roller and then transfers ink to a clich´e, where a patterned relief is formed. Because the unwanted ink on a blanket roller is transferred to protruded regions of a clich´e, the remaining ink forms a positive pattern image, which is subsequently transferred to a substrate.

13.3 Comparison of Printing Technologies

Unlike conventional gravure offset printing, the printing quality of reverse offset printing is comparable to photolithographically formed patterns, and a prototype 15 in. TFT LCD panel has been demonstrated [11]. Despite the great success of reverse offset printing, the use of a blanket roller has several drawbacks. The printed line width tends to increase as printing takes place [12]. The solvent of the ink is absorbed into the blanket roller and hence the change in the blanket roller’s physical properties such as stiffness, surface energy, and solvent absorption capability leads to a variation in the line width. To avoid this problem, a blanket roller either has to be regularly replaced with a new one after a certain number of printings or it has to be regularly dried. However, these fixes increase the running cost of reverse offset printing. The employment of a soft blanket roller also deteriorates the positional accuracy. The critical issue of reverse offset printing is to maintain the cleanness of an expensive clich´e. If ink with dispersed nanoparticles such as pigments is used, then the engraved regions of the clich´e might be contaminated with such particles. After each printing run, the used clich´e needs to be rinsed, which might need a couple of auxiliary clich´es and a special facility for the cyclic cleaning of clich´es. Despite all the above mentioned issues, the usage of ink is not advantageous, compared to that in other printing technologies. Most of the ink remains on the protruded regions of the clich´e and it must be washed away. In the case of a color filter, approximately 70% of each colorant ink will be wasted. For much finer patterning work, for example metal electrodes of the TFT array, where the required line width is as fine as 10 µm, approximately 90% of expensive silver nanoparticulate ink will be consumed for nothing. Another fine patterning method being used for years is laser transfer. As can be seen in Figure 13.7, a donor film, where ink is coated, is laminated onto a substrate and then laser imaging is performed. The laser-induced heat selectively expels the ink coated on the donor film onto the substrate and patterns are formed after the Donor film Laminating Substrate Laser Laser imaging

Peeling off

Transferred pattern Figure 13.7 Schematic illustration of laser transfer.

197

198

13 Issues in Color Filter Fabrication with Inkjet Printing

laser-imaged donor film is peeled off [13, 14]. This laser-induced transfer exploits the high resolution of lasers and has successfully demonstrated its capability to fabricate organic electronics [15], an active matrix organic light-emitting diode (AMOLED) display [16–18], and phosphor patterns for PDP [19]. The donor film has to be carefully formulated to balance cohesion and adhesion forces between the donor layer and the substrate. In addition, the coated organic materials must withstand the laser-induced heat impulse. Inherently, the laser transfer is sensitive to the fluctuation of laser power and surface energy states of the donor film and substrate. This generally does not contribute to the cost reduction of materials, although the usage of materials for the fabrication of a color filter could reach 67% or more in an ideal case where the donor film is recycled. The last printing technology to be considered is inkjet printing, which is very familiar to us. Basically, the fabrication of a color filter with inkjet printing employs a piezo drop-on-demand (DOD) inkjet print head, because of reliability issues. Compared to a thermal inkjet print head, a piezo DOD inkjet print head is much less limited in the selection of solvents, especially those with low vapor pressure and high boiling point. It is free from cogation, which is the accumulation of decomposed crud on a thermal heater due to cyclic, high-temperature bubble generation. Ideally, inkjet printing can pattern three primary colors simultaneously, although one color at a time is patterned in practice, as shown in Figure 13.8. This is due to the inherent nature of inkjet printing, which does not need contact with a substrate to form patterns. Other printing technologies based on direct contact with a substrate cannot conduct patterning works of more than two colors simultaneously. Since inkjet printing does not need contact with a substrate, there is no contamination. Compared to other printing technologies, inkjet printing does not require any indirect means such as a relief plate, screen, or donor film. Instead, it uses virtual image data to print out. As a result, it is the easiest method to scale up the printing area as well as cope with model changes from time to time in the production line of color filters. Inkjet print head

Nozzles Black matrix (hydrophobic)

Ink droplets

R

G

B

R

G

B

Substrate (hydrophilic) Figure 13.8

Schematic illustration of the fabrication of an inkjet-printed color filter.

13.4 Printing Swathe due to Droplet Volume Variation Table 13.1

Comparison of printing technologies. Viscosity [20] (Pa s)

Screen Flexography Gravure (offset) Reverse offset Laser transfer Inkjet

Resolution [21] (µm)

Thickness [21] (µm)

100 40 15 10 [22] 20 [15] 50

3–15 0.8–2.5 0.8–8 – – 0.3–20

0.5–50 0.05–0.5 0.05–0.2 – Not applicable 0.001–0.003

Scale up Bad Bad Bad Bad Good Best

So far, various printing technologies have been introduced and their pros and cons have been discussed, as summarized in Table 13.1. Amongst other printing technologies, inkjet printing has been considered to be the most promising one in display industries because of its unique ability to handle a large area substrate with less auxiliary lead time for model changes. In the next section, process simulations to extract the requirements of a piezo DOD inkjet print head and materials are introduced from a practical viewpoint in more detail.

13.4 Printing Swathe due to Droplet Volume Variation

The droplet volume variation across nozzles either in one inkjet print head or among multiple inkjet print heads affects the color uniformity [23]. If more droplet volume is deposited in a subpixel, a denser color appears in that subpixel compared to adjacent subpixels. This is truly a human color cognition issue, which determines the acceptable tolerance of color variation, that is, the acceptable tolerance of the droplet volume variation. Conventional approaches to synchronize droplet volumes across nozzles have mainly used vision-based systems, as shown in Figure 13.9. Nozzles

Inkjet print head

Nonuniform droplets in volume Synchronize droplet velocities Volume or area calculation

Surface profile measurement for volume calculation Substrate

Figure 13.9 Various droplet volume calibration methods.

199

200

13 Issues in Color Filter Fabrication with Inkjet Printing

(a)

(b)

75 µs (d)

(c)

125 µs (e)

175 µs (f)

23 µm 34.5 µm Main droplet 225 µs

Satellite 1 48.3 µm

275 µs

Satellite 2

11.5 µm

32.2 µm 20.7 µm 225 µs

Figure 13.10 Vision captured image with a CCD camera.

The straightforward method is to detect either the edges of droplets to compute their volumes or the areas of droplets to compare with each other. Another method is to synchronize droplet velocities across nozzles. A third method is to measure the surface profiles of deposited droplets on a substrate and extract their volumes with a confocal laser-scanning microscope, white light interferometer, and so on. Typically, droplet images are, captured with a charge coupled device (CCD) camera and LED strobe (Figure 13.10), the LED impulse of which is synchronized with the jetting trigger signal. The typical impulse time of a LED is a couple of microseconds, and hence insufficient light falls on CCD sensor in such a short period of time. On the other hand, the typical shutter speed of a CCD is tens of milliseconds. Therefore, one frozen image of a droplet jetted at a few to tens of kilohertz actually comes from tens to hundreds of superimposed images. This superposition complements the light deficiency but the subtle difference in each jetted droplet smears the overall image quality. As a result, the edge of a droplet is blurry because of the superposition of multiple images of the captured droplets, as can be seen in Figure 13.10f. Another source of image blur is the motion of droplets. When a droplet is ejected out of the nozzle, its velocity is a few to a dozen meters per second. Because the typical impulse time of a LED is a couple of microseconds, a droplet moves a few to dozens of micrometers during the LED flashes.

13.4 Printing Swathe due to Droplet Volume Variation

Diameter: 53.1 µm Volume: 78.7 pl

(a)

Diameter: 50.7 µm Volume: 68.2 pl

(b)

Figure 13.11 Influence of the light intensity of an LED strobe on the detected droplet volume.

The light intensity of a LED strobe also affects the size measurement of a droplet. As can be seen in Figure 13.11, the measured droplet size differs as the light intensity of a LED strobe changes, although the same jetting conditions are applied. The most critical uncertainty issue in calibrating the droplet volume is the length scale to detect an abnormal droplet volume. If it is assumed that the acceptable tolerance of color variation is around ±2% and the droplet diameter is 35 µm, then the acceptable droplet diameter ranges from 34.8 to 35.2 µm. This means that the length scale required to detect a droplet volume out by ±2% is almost comparable to the UV wavelength, 232–235 nm, which far exceeds the optical resolution limit of a vision system by a couple of micrometers at best but practically tens of micrometers. In such a case, the surface profile measurement with a confocal, laser-scanning microscope or white light interferometer might be a better option to calculate the volume of a sessile droplet on a substrate. However, it is noteworthy that their lateral resolutions are as poor as the CCD-based vision system, although the vertical resolution is as high as tens of nanometers. In addition, their surface profile measurement at the steep edge is not accurate and the measurement speed is generally sluggish. All the above mentioned sources of uncertainty make current vision-based systems inaccurate to detect and calibrate droplet volumes out of nozzles in order to match them. Therefore, it is crucial to employ an inkjet print head with an inherently better volume consistency. In addition, nozzle diameter reduction is required to place the ejected droplet in a tiny subpixel, 45 by 120 µm in width and length respectively, for a small size polymer light-emitting diode (PLED) display by Philips PolyLED [24]. In order to deposit a droplet in this tiny subpixel with a sufficient clearance, the size of a droplet should be as small as 27 µm or less. As a result, inkjet print head manufacturers have attempted to improve the quality of inkjet print heads for years, and currently silicon microelectromechanical systems (MEMSs) are being used for the fabrication of inkjet print heads [25]. With improved fabrication tools, the nozzle diameter has gotten smaller, as shown in Figure 13.12 [26].

201

202

13 Issues in Color Filter Fabrication with Inkjet Printing

Volume: 280 pl Diameter: 81.2 µm

Volume: 100 pl Diameter: 57.6 µm Year 1984

Year 1994 Volume: 220 pl Diameter: 74.9 µm

Volume: 35 pl Diameter: 40.6 µm

Volume: 5 pl Diameter: 21.2 µm Year 1999 Volume: 1 pl Diameter: 12.4 µm

Figure 13.12 Trend of the droplet reduction in size and volume.

In color filter fabrication of large area TFT LCDs, the previous requirement to reduce the nozzle diameter required a change. The subpixel size of a TFT LCD, even with the smallest 32 in. screen, is big enough to neglect the placement error of a droplet, as seen in Figure 13.13. If the droplet volume out of a nozzle is small, then more droplets are required to fill the subpixel. If the printing speed is the same, then more droplets must be ejected at the same printing length with smaller nozzles, which implies an increase in the jetting frequency. As the jetting frequency increases, complex acoustic resonance and antiresonance behaviors appear [27, 28], Droplet (28 µm in diameter)

Subpixels of large area TFT LCDs 32′′

Subpixel (Philips PolyLED)

37′′

42′′

47′′

55′′

Figure 13.13 Comparison of subpixel sizes against screen inches.

13.4 Printing Swathe due to Droplet Volume Variation

which could make inkjet printing harder. Moreover, not only the cost of an inkjet print head with smaller nozzles but also its maintenance gets harder. Therefore, whether the development of an inkjet print head with smaller nozzles is really a better choice in terms of volume consistency, productivity, cost, and maintenance for the fabrication of a TFT LCD color filter needs to be revaluated. Because there are no means to fabricate such nozzles with a given tolerance, ranging from −2 to +2 µm at an increment of 0.25 µm, numerical simulations were carried out with a MicroFab inkjet print head (MicroFab Technologies, Inc., Plano, TX, USA), where nozzle diameters were chosen as 30, 40, and 50 µm [23]. The driving conditions were set at 3 µs for rising and falling times, 15 µs for dwell time, 30 V for driving voltage, and 2 kHz for jetting frequency. The

Normalized volume error (%)

15 10 5 0 −2

−1.5

−1

−0.5

0

0.5

1

1.5

1

1.5

30 um 40 um 2 50 um

−5 −10 −15 Nozzle tolerance (µm)

(a)

Normalized velocity error (%)

9

(b)

4

−2

−1.5

−1

−0.5

−1 0

0.5

−6

−11 Nozzle tolerance (µm)

Figure 13.14 Normalized errors against nozzle tolerance. (a) Droplet volume and (b) droplet velocity.

2

203

204

13 Issues in Color Filter Fabrication with Inkjet Printing

physical properties of ethylene glycol used as ink for numerical simulations were 50 mN m−1 for surface tension, 20 mPa for dynamic viscosity, 1113 kg m−3 for density, and 1680 m s−1 for speed of sound at room temperature. As can be seen in Figure 13.14, the droplet volume variation against the nozzle tolerance tends to decrease as the nozzle diameter increases. On the other hand, the droplet velocity variation tends to increase as the nozzle diameter increases. This means that any abnormality in the droplet volume is more easily detectable with a larger nozzle diameter. These results imply that the previous approach to fabricating a large area TFT LCD color filter with inkjet print heads having small nozzles needs to be reconsidered. Rather than a reduction in the nozzle diameter, it is more desirable to enlarge the nozzle diameter for better consistency in the droplet volume and better detection of abnormal droplets by velocity measurement. Such inkjet print heads are not only cheaper but also have easier maintenance and longer life expectancy.

13.5 Subpixel Filling with a Designed Surface Energy Condition

Once droplets are ejected out of nozzles, they are required to fill subpixels, as shown in Figure 13.15a. If the contact angle of the colorant ink against the black matrix (BM), θBM , is not high enough, the colorant ink crosses over the black matrix and trespasses onto the adjacent subpixels, as shown in Figure 13.15b. If the contact angle of the colorant ink against glass, θglass , is not low enough, then the colorant ink does not fill the subpixel completely, as shown in Figure 13.15c. Therefore, the fundamental question arises on what those contact angles of the colorant ink against the black matrix and glass should be. In general, the required contact angles of the colorant ink against the black matrix and glass have been taken to be above 80◦ and below 10◦ respectively [29]. If the contact angle of the colorant ink against the black matrix is too high, however, Black matrix Colorant ink

(b) Ink overspill

(b) Ink underfilling qGlass

qBM

Substrate

(a) Ideal ink filling Figure 13.15 Subpixel filling with ink. (a) Ideal ink filling, (b) ink overspill, and (c) ink underfilling.

13.5 Subpixel Filling with a Designed Surface Energy Condition

a subsequent coating process such as an overcoat layer in Figure 13.3 would be problematic without any additional surface treatment such as plasma ashing or UV/O3 surface treatment. Another problem related to the use of a hydrophobic chemical compound is the thermal-induced surface diffusion of a hydrophobic chemical compound over the hydrophilic region, which could turn the hydrophilic region into a hydrophobic region and hence lead to incomplete filling of the colorant ink in a subpixel. As a result, the contact angle requirements of the colorant ink should be known to chemists, who formulate the colorant ink and black matrix, and to process engineers, who are in charge of the quality control of the TFT LCD color filter production line. The most popular approach for calculating the required contact angles of the colorant ink against the black matrix and glass has been computational fluid dynamics (CFD) [30, 31]. CFD computes the velocity and pressure fields of a fluid and provides the predicted motion and shape of a fluid. However, the computation of the intermediate fluidic behavior such as spreading and recoiling of a droplet, as shown in Figure 13.16a, is of little importance to determine the required contact angle of the colorant ink against the black matrix and glass. The greatest concern in determining the required contact angle of the colorant ink against the black matrix and glass is the final state of the colorant ink in a subpixel, that is, whether it fills the subpixel without any overspill or underfill under given contact angles, θBM and θglass , as shown in Figure 13.16b. Another issue is the size of a computational domain, especially the mesh cell size. The thickness of the black matrix is just a couple of micrometers but the length scale of a subpixel in length and width is on the order of hundreds of micrometers. If the details of ink spreading along the border between the black matrix and glass are taken into account, then locally refined cells occur along the border and this irregularity in the cell size might affect the convergence of numerical solutions. On the other hand, if the size of cells in the spatial domain is uniform, the required number of cells is so huge that this approach may be computationally too expensive. As can be seen in Figure 13.17, the simulation result from CFD with 200 000 cells for a subpixel, 143 µm by 466 µm, does not show a better resolution than that from the Surface Evolver, where the cell size is approximately 3.5 µm, which is 3.5 times bigger than the minimum feature size of a subpixel, that is, the thickness of the black matrix. If the cell size is adjusted to be one-third of the thickness of the black matrix, the number of uniform cells reaches around 316 million. The Surface Evolver takes less than 10% of the CPU time compared to a conventional CFD in case the number and size of cells in the computational domain for CFD are approximately 100 000 and 6 µm, which is six times more than the minimum feature size of the subpixel, that is, the thickness of the black matrix. If the cell size decreases while maintaining uniform cell size, then the required CPU time dramatically increases for conventional CFD. However, the Surface Evolver has an ability to adapt to the cell size, and the irregularity of cell size does not affect its computation result. Therefore, the CPU time of the Surface Evolver is not quite proportional to the computational domain. In this sense, the advantage of the surface evolution technique over conventional CFD in computing

205

13 Issues in Color Filter Fabrication with Inkjet Printing

0

Pressure Contours (z multiplied by 1.e+0.4) 574 1149 1724 2298

2873

3448

1.0

Drop impact (fluidic behaviour)

0.5

1 0.8 z

z

Initial shape

0.6 0.4 0.2 0 1

0.0 0.0

.00003

.00006

.00009

.00012

.00015

0.5

x

0

Pressure Contours (z multiplied by 1.e+0.4) -15324

-9960

y

-4595

769

-0.5 -1

6134

0 -0.2 -0.4 -0.6 -0.8 -1

0.8 0.6 0.4 0.2

x

z

1.0

Drop spreading (fluidic behaviour)

0.5

0.0 0.0

.00003

.00006

.00009

.00012

.00015

x Pressure Contours (z multiplied by 1.e+0.4) 962

1050

1158

1256

1354

Equilibrium shape 1452

1550

z

1.0

Equilibrium status z

206

0.5

0.8 0.6 0.4 0.2 0 -0.8 -0.6 -0.4 -0.2 x

(b) 0.0

(a)

0.0

.00003

.00006

.00009

.00012

.00015

x

Figure 13.16 Comparison of the numerical approaches. (a) CFD and (b) the Surface Evolver.

(a)

(b)

Figure 13.17 Comparison of the numerical simulation results. (a) CFD with 200 k cells and (b) Surface evolver.

0 0.2 0.4 0.6 0.8

0 -0.2 -0.4 -0.6 -0.8

0.8 0.6 0.4 0.2 y

1

13.5 Subpixel Filling with a Designed Surface Energy Condition Subpixel sizes of various TFT LCD screens.

Table 13.2

Inch (µm)

207

BM width x (µm)

32 42 52 65

14 19 23 29

Subpixel y (µm)

109 143 177 221

355 466 576 721

(i) The aperture ratio of each screen is assumed to be 85%. (ii) The thickness specification of the black matrix and the colorant film is assumed to be 1 µm.

the equilibrium state of a liquid in a subpixel is obvious for various numerical tests of the contact angles of the colorant ink against the black matrix and glass before ink formulation. Table 13.2 lists subpixel sizes of various TFT LCD screens. In the first step, the minimum contact angle to confine the colorant ink in a subpixel is simulated, as shown in Figure 13.18, where the size of a TFT LCD

200 150 100 50 0 -200 -150 -100 -50 0 x( µm 50 ) 100 150 200

Figure 13.18

z (µm)

200 150 100 50 0 -200 -150 -100 -50 0 x( 50 µm 100 ) 150 200

0 ) -50 µm -100 y( -150 -200

0

200 150 100 50

) -50 µm -100 y( -150 -200

200 150 100 50 0 -200 -150 -100 -50 x( 0 µm 50 ) 100 150 200

200 150 100 50

z (µm)

z (µm)

z (µm)

Initial volume

200 150 100 50 0 -200 -150 -100 -50 x( µm 0 50 ) 100 150 200

Surface evolution of the liquid blob located in the subpixel.

200 150 100 50 0 ) -50 µm -100 y( -150 -200

200 150 100 50 0 ) -50 µm -100 y( -150 -200

208

13 Issues in Color Filter Fabrication with Inkjet Printing

30° Maximum overflow distance 40°

z (µm)

50° y (µm) x (µm)

Figure 13.19 Ink overflow over the black matrix depending on the contact angle.

screen ranges from 42 to 65 in. and the solid content of the colorant ink is between 5 and 10% in volume. When the solid content of the colorant ink is known in volume, then the average dried colorant film thickness can be approximated. If the average dried colorant film thickness needs to be the same as the thickness of the black matrix, for example, 1 µm, so as to minimize the thickness difference between the black matrix and the subpixel area, the initial volume of the colorant ink to be deposited in the subpixel can be approximated, as shown in Figure 13.18. As the computation of surface evolution continues, the number and size of surface cells evolve and the final shape of the colorant ink is calculated. Here, the inner subpixel is completely filled with colorant ink by setting the contact angle of the colorant ink against glass equal to zero. However, the contact angle of the colorant ink against the black matrix varies from 30◦ to 70◦ , as shown in Figure 13.19. If the maximum overflow distance exceeds the half width of the black matrix, then it is assumed that the contact angle of the colorant ink against the black matrix is not high enough because the colorant ink can trespass onto the adjacent subpixel. In this case, an ink chemist needs to reformulate the colorant ink or bestow a higher hydrophobic characteristic to the black matrix by blending a proper hydrophobic chemical compound. As can be seen in Figure 13.20, the maximum overflow distance varies with the solid content and contact angle of the colorant ink. It is noteworthy that the required contact angle of the colorant ink against the black matrix becomes higher with the 5 vol% solid content of colorant ink, as the size of a TFT LCD screen decreases. If model changes of TFT LCD screens in the production line are to be considered, the contact angle of the colorant ink against the black matrix must be in common for 32–65 in. and hence it needs to be as high as 60◦ . This high contact angle of the colorant ink against the black matrix is achievable by the reformulation of the colorant ink or black matrix, or the incorporation of a surface treatment process such as CF4 plasma. However, the surface modification should be carefully assessed because it could affect the processability of postprocesses. For example,

13.5 Subpixel Filling with a Designed Surface Energy Condition 32′′ 42′′ 52′′ 65′′

Maximum overflow distance (mm)

45

(a)

35

25

15

5

−5

30

35

40

45 50 55 Contact angle (°)

60

65

32′′ 42′′ 52′′ 65′′

Maximum overflow distance (mm)

45

(b)

70

35

25

15

5

−5

30

35

40

45

50

55

60

65

70

Contact angle (°)

Figure 13.20 Simulation results of the maximum overflow distance with various contact angles of the colorant ink against the black matrix. (a) 5 vol% and (b) 10 vol%.

the postbaking process of the colorant ink could diffuse a hydrophobic chemical compound over the hydrophilic subpixel area, and hence no sufficient wetting in the adjacent subpixel might occur at the next printing step with another colorant ink. The addition of a hydrophobic characteristic to the black matrix can also lead to failure of the OC process shown in Figure 13.3. As the solid content of the colorant ink increases, on the other hand, the required contact angle of the colorant ink against the black matrix decreases. Because the

209

13 Issues in Color Filter Fabrication with Inkjet Printing

105.00

32′′ 42′′ 52′′

Surface ratio (µm)

100.00

65′′

95.00 90.00 85.00 80.00 75.00 70.00 5

10

15

(a)

20 25 30 Contact angle (°)

35

40

45

105.00 32′′ 42′′ 52′′

100.00 Surface ratio (µm)

210

65′′

95.00 90.00 85.00 80.00 75.00 70.00 5

(b)

10

15

20 25 30 Contact angle (°)

35

40

45

Figure 13.21 Simulation results of the subpixel filling with various contact angles of colorant ink against glass. (a) 5 vol% and (b) 10 vol%.

required volume of the colorant ink required to meet the specified dried colorant film thickness reduces, the confinement of the colorant ink in a subpixel has more tolerance. Therefore, the intermediate conclusion is drawn that colorant ink with a higher solid content is desirable for the issue of confinement in a subpixel. As the second step, another set of simulations to check the processability of subpixels filling with various contact angles of the colorant ink against glass was carried out and plotted in Figure 13.21. The contact angle of the colorant ink against the black matrix was set to 60◦ in accordance with the simulation results of the first step. The colorant ink should completely cover the entire area of a subpixel without

13.5 Subpixel Filling with a Designed Surface Energy Condition

Metastable state, C

Stable state, A

Stable state, B Metastable state, C

Black matrix

(b) a qBM qBM

Stable state, A

b Substrate

(a) (c)

Figure 13.22 Metastability of colorant ink in a subpixel. (a) Cross section of the deposited colorant ink, (b) metastable state, C, and (c) stable state, A.

any void, so the surface ratio of the colorant-ink-filled surface area to the subpixel area was computed with the contact angle of the colorant ink against glass from 5◦ to 45◦ . In Figure 13.21a, the surface ratio of a 32 in. TFT LCD screen below 40◦ exceeds 100%, but this comes from the slight bulge of the colorant ink, as shown in Figures 4.18 and 4.19. It is noteworthy that the surface ratio with the solid content of 5 vol% suddenly decreases, especially in case of screen sizes of 32 and 42 in. This is due to the metastability of the colorant ink in a subpixel, as shown in Figure 13.22 [32, 33]. It is assumed that the contact angle of the colorant ink against glass, θglass , is low enough to fill the entire area of a subpixel and the contact angle of the colorant ink against the black matrix, θBM , high enough to confine the colorant ink in a subpixel. Therefore, the initial surface of the colorant ink in a subpixel might be formed along C, as shown in Figure 13.22a. However, this state might not satisfy θBM and hence the surface of the colorant ink would take either of two stable states, A or B, each of which satisfies θBM . However, it is physically impossible to take state B from state C, unless the volume of the colorant ink diminishes because of drying or the colorant ink overflowing into the nearby subpixels, which violates the assumption that the contact angle of the colorant ink against the black matrix, θBM , is high enough to confine the colorant ink in the subpixel. Therefore, the fluidic motion labeled b in Figure 13.22a is unlikely to occur. As long as the contact angle of the colorant ink against the glass is low enough, the colorant ink wets the entire

211

212

13 Issues in Color Filter Fabrication with Inkjet Printing

subpixel and remains in state C. This state is regarded as the ‘‘metastable’’ state. On the other hand, it is physically possible to reach state A from state C while maintaining a constant volume of the colorant ink by contact line recession. This fluidic motion, labeled a, occurs when the contact angle of the colorant ink against glass is not low enough. If the contact angle of the colorant ink against glass is not low enough and there is any unfilled region by the colorant ink in the subpixel, the colorant ink suddenly retreats and reaches the stable state A, as shown in Figure 13.22c. The higher the solid content of the colorant ink, the lower is the contact angle required for the colorant against glass, as can be seen in Figure 13.21. If the solid content of the colorant ink is 5 vol%, then the required contact angle of the colorant ink against glass is approximately 20◦ for all TFT LCD screens. If the solid content of the colorant ink is 10 vol%, then the required contact angle becomes as low as 10◦ . Maintaining such a low contact angle of the colorant ink against glass might not be desirable because a surface having such a low contact angle implies that its surface energy is so high as to be easily contaminated. In addition, a hydrophobic chemical compound blended in the black matrix to increase the contact angle of the colorant ink against the black matrix could easily trespass over such a surface with a high surface energy, especially during the postbaking process. Therefore, another intermediate conclusion drawn from the second step is that the solid content of the colorant ink should be low, which is against the first intermediate conclusion. As a matter of fact, parameters such as the solid content and contact angles of the colorant ink against the black matrix and glass have to be carefully manipulated because they are bound to come up in a trade-off relationship.

13.6 Other Technical Issues

Many technical issues have still been left unsolved, although inkjet printing is being used for somewhat simpler processes such as the formation of alignment layers in the production lines of display industries. For example, the nonuniformity of colorant film thicknesses between neighboring printing swathes is problematic. This problem is relatively easy to solve because it is known to be caused by different degrees of solvent vapor concentration in the air [34]. The solution to this problem is to change the solvent composition in the colorant ink. However, not all the sources of the existing problems are known. A more difficult problem is a way to calibrate the droplet volumes across nozzles. Currently, droplet volumes are calibrated by comparing the measured color coordinate of each subpixel. If the measured color coordinate of a subpixel is out of specification, then the driving voltage of the nozzle is adjusted. This approach is more reliable than vision-based methods. However, the color coordinate measurement of each subpixel nibbles the production time away. Besides, the calibration could become invalid due to external disturbances such as meniscus pressure

References

fluctuation and the change in ink properties. In such a case, this time-consuming calibration has to be repeated. Therefore, a better, faster, more reliable, and concise way to calibrate droplet volumes quickly is still sought for. One of the critical issues in inkjet printing for display industries is the sporadic jetting failure of a couple of nozzles among several hundreds. The exact cause of this sporadic jetting failure has not been completely understood. One might suspect that nozzle clogging is due to unfiltered contaminants or fragments built up inside the channel of an inkjet print head. In this case, clogged nozzles could be tagged and excluded in the printing work. However, sporadic jetting failure differs from normal nozzle clogging in the sense that problematic nozzles are random. This might require a statistical approach for better production reliability, which requires the efforts of researchers in inkjet printing. 13.7 Conclusion

This paper has discussed jetting and wetting issues in the fabrication of a TFT LCD color filter with inkjet printing. In addition, some other technical issues in inkjet printing for display industries have been briefly introduced. Although inkjet printing has a long history in graphic art industries and much scientific research has been carried out, there are still many questions to be answered. Some might remain unsolved because they could be intrinsic problems of inkjet printing itself. However, other auxiliary techniques could diminish such problems and make inkjet printing even more useful. Because the fabrication of a TFT LCD with inkjet printing is not very different from the fabrication of a large area PLED screen, the understanding of these technical issues could help in the employment of inkjet printing in PLED displays and anchor inkjet printing as a de facto standard in manufacturing. References 1. DisplaySearch (2008) Quarterly TFT

5. Lin, H.-W., Chang, C.-P., Hwu, W.-H.,

LCD Cost Forecast Reports with Annual and Quarterly Projections. 2. Sabnis, R.W. (1999) Color filter technology for liquid crystal displays. Displays, 20, 119–129. 3. Lee, T.-M., Choi, Y.-J., Nam, S.-Y., You, C.-W., Na, D.-Y., Choi, H.-C., Shin, D.-Y., Kim, K.-Y., and Jung, K.-I. (2008) Color filter patterned by screen printing. Thin Solid Films, 516, 7875–7880. 4. Pardo, D.A., Jabbour, G.E., and Peyghambarian, N. (2000) Application of screen printing in the fabrication of organic light-emitting devices. Adv. Mater., 12 (17), 1249–1252.

and Ger, M.-D. (2008) The rheological behaviours of screen-printing pastes. J. Mater. Process. Technol., 197, 284–291. 6. Michel, B., Bernard, A., Bietsch, A., Delamarche, E., Geissler, M., Juncker, D., Kind, H., Renault, J.-P., Rothuizen, H., Schmid, H., Schmidt-Winkel, P., Stutz, R., and Wolf, H. (2001) Printing meets lithography: soft approaches to high-resolution patterning. IBM J. Res. Dev., 45, 697–719. 7. Lahti, M., Lepp¨avuori, S., and Lantto, V. (1999) Gravure-offset-printing technique

213

214

13 Issues in Color Filter Fabrication with Inkjet Printing

8.

9.

10.

11.

12.

13.

14.

15.

16.

for the fabrication of solid films. Appl. Surf. Sci., 142, 367–370. Jiang, W., Liu, H., Ding, Y., Tang, Y., Shi, Y., Yin, L., and Lu, B. (2009) Investigation of ink transfer in a roller-reversal imprint process. J. Micromech. Microeng., 19, 015033. Yoo, S.S., Chang, Y.G., and Cho, H.L. (2006) Printing plate, method of fabricating the same, and method of fabricating flat panel display using the same. US Patent 0,048,667, filed Apr. 28, 2006 and issued Mar. 1, 2007. Nam, S.-H. and Yoo, S.-S. (2007) Pattern transcription device and method of fabricating clich´e for the same. US Patent 0,190,307, filed Dec. 28, 2007 and issued Aug. 14, 2008. Chang, Y.-G., Nam, S.-H., Kim, N.-K., Kook, Y.-H., Kim, J., Yoo, S.-S., Kim, C.-D., Kang, In.-B., and Chung, I.-J. (2008) A study of roll-printing technology for TFT-LCD fabrication. J. Soc. Inf. Disp., 17 (4), 301–307. Lee, T.-M., Noh, J.-H., Kim, C.H., Jo, J., and Kim, D.-S. (2010) Development of a gravure offset printing system for the printing electrodes of flat panel display. Thin Solid Films, 518, 3355–3359. Blanchet-Fincher, G. (1995) Doner element and processes for laser-induced thermal transfer. US Patent 5,523,192, filed Aug. 2, 1995 and issued Jun. 4, 1996. Mizuno, K., Staral, J.S., and Pokorny, R.J. (1999) Thermal transfer with a plasticizer-containing transfer layer. US Patent 6,228,543, filed Sept. 9, 1999 and issued May 8, 2001. Blanchet, G.B., Loo, Y.-L., Rogers, J.A., Gao, F., and Fincher, C.R. (2003) Large area, high resolution, dry printing of conducting polymers for organic electronics. Appl. Phys. Lett., 82 (3), 463–465. Chin, B.D., Suh, M.C., Kim, M.H., Kang, T.M., Yang, N.C., Wong, M.W., Lee, S.T., Kwon, J.H., Chung, H.K., Wolk, M.B., Bellmann, E., and Baetzold, J.P. (2003) High efficiency AMOLED using hybrid of small molecule and polymer materials patterned by laser transfer. J. Inf. Disp., 4 (3), 1–5.

17. Suh, M.C., Chin, B.D., Kim, M.-H.,

18.

19.

20.

21.

22.

23.

24.

25.

26.

Kang, T.M., and Lee, S.T. (2003) Enhanced luminance of blue light-emitting polymers by blending with hole-transporting materials. Adv. Mater., 15 (5), 1254–1258. Lee, J.Y. and Lee, S.T. (2004) Laser-induced thermal imaging of polymer light-emitting materials on poly(3,4-ethylenedioxythiophene): silane hole-transport layer. Adv. Mater., 16 (1), 51–54. Lee, J.H., Yoo, C.D., and Kim, Y.-S. (2007) A laser-induced thermal spray printing process for phosphor layer deposition of PDP. J. Micromech. Microeng., 17, 258–264. Kipphan, H. (2001) Handbook of Print Media, Springer-Verlag and Heidelberg GmbH & Co. KG, Berlin. Huebler, A., Hahn, U., Beier, W., Lasch, N., and Fischer, T. High volume printing technologies for the production of polymer electronic structures. 2nd International IEEE Conference on Polymers and Adhesives in Microelectronics and Photonics, Zalaegerszeg, Hungary, June 23–26, 2002. Yoo, S.-S. (2010) Development of printing technology for the CF arrays of TFT-LCDs. Printed Electronics & Photovoltaics Europe, Dresden, Germany, April 13–14. Shin, D.-Y. and Smith, P.J. (2008) Theoretical investigation of the influence of nozzle diameter variation on the fabrication of thin film transistor liquid crystal display color filters. J. Appl. Phys., 103, 114905. Letendre, W. (2004) Advances in piezoelectric ink jet micropumps for precision deposition using silicon nozzles. IS&T NIP20, Salt Lake City, USA, October 31–November 5. Menzel, C., Bibl, A., and Hoisington, P. (2004) MEMS solutions for precision micro-fluidic dispensing applications. IS&T NIP20, Salt Lake City, USA, October 31–November 5. Keszler, D.A. (2006) Solution-based processing of inorganic oxide electronics. Printed Electronics USA, Phoenix, AZ, December 5–6.

References for FPD manufacturing. 2nd Advanced Technology Workshop on Printing an perimental and theoretical study of Intelligent Future, Boston, MA, March wave propagation phenomena in 16–19. drop-on-demand ink jet devices. IBM J. Res. Dev., 28 (3), 314–321. 32. Brinkmann, M. and Lipowsky, R. (2003) Wetting morphologies on substrates Dijksman, J.F. (1984) Hydrodynamics with striped surface domains. J. Appl. of small tubular pumps. J. Fluid Mech., Phys., 92 (8), 4296–4306. 139, 173–191. LINK ISD (2003) Production technology 33. Shin, D.-Y. and Brakke, K.A. (2009) on plastic substrate for flexible display Modelling and analysis of the wetting and electronics. characteristics of ink for display applications with the surface evolution Creagh, L.T., Heun, S., and Tallant, N. technique. Modell. Simul. Mater. Sci. (2003) Recent advances in deposition Eng., 17, 045002. of OLEP materials via piezoelectric ink jet. International Display Manufacturing 34. Shimoda, T., Morii, K., Seki, S., and Conference, Taipei, Taiwan, February Kiguchi, H. (2003) Inkjet printing of 18–21. light-emitting polymer displays. MRS Bull., 28 (11), 821–827. Creagh, L.T., Heun, S., and Tallant, N. (2003) Improving piezo ink jet and LEPs

27. Bogy, D.B. and Talke, F.E. (1984) Ex-

28.

29.

30.

31.

215

217

14 Application of Inkjet Printing in High-Density Pixelated RGB Quantum Dot-Hybrid LEDs Hanna Haverinen and Ghassan E. Jabbour

14.1 Introduction

The market in portable electronics is strongly heading toward thinner, faster, and affordable devices. A very critical part in these devices is the display, which needs to be bright and with as pure primary colors as possible to obtain the largest color gamut. Recent advances in display technology have shown a highly attractive display with its active light-emitting elements made of organic light-emitting diodes (OLEDs). Such devices have been the fruits of a breakthrough starting with the work of Tang and Van Slyke in 1987 [1]. Among the many positive attributes, OLEDs are bright, have excellent color contrast, and fast response time, which contribute to overall better display features than current liquid crystal displays (LCDs). Owing to such attractive characteristics, OLEDs have been used in many cell phones, and will soon be used in other portable devices with relatively larger displays. However, some shortcomings of OLEDs are still being addressed, including the lack of pure colors and longer lifetime for more demanding applications such as solid-state lighting. Of particular challenge is the blue color OLED, which is currently characterized by an emission slated toward the greener part of the spectrum. Therefore, it is of no surprise that the search for long lifetime, efficient pure blue organic emitters is of strong interest to the industry [2, 3]. OLEDs have light emission with a relatively large full-width-at-half-maximum (FWHM is typically more than 70 nm), which compromises color purity, thus complicating the design and development of high-definition televisions (HDTVs) [4]. Recently, an intriguing solution to obtain better color purity has been to introduce inorganic emissive quantum dots (QDs) into an otherwise OLED structure [5, 6]. The efficient QDs have an emissive core surrounded by a wider band gap shell used to improve quantum yield by eliminating surface traps due to unsatisfied bonds. However, such particles have very poor to no solubility in a given solvent. In order to allow better solubility, they are capped with an organic ligand, which also reduces surface traps present on the shell outer layer. However, currently these QDs remain poor charge transporters (Figure 14.1), with most of their use limited to the light-emission layer. In most demonstrated efficient quantum dot Inkjet-based Micromanufacturing, First Edition. Edited by J. G. Korvink, P. J. Smith, and D.-Y. Shin. © 2012 Wiley-VCH Verlag GmbH & Co. KGaA. Published 2012 by Wiley-VCH Verlag GmbH & Co. KGaA.

218

14 Application of Inkjet Printing in High-Density Pixelated RGB Quantum Dot-Hybrid LEDs Vacuum level (0 eV)

Vacuum level (0 eV)

−

CB Core

Core

VB

− Core

Energy (−eV)

Energy (−eV)

LUMO

−

H T L

LUMO E T L

+ Anode +

− Cathode

−

HOMO

+

Shell

HOMO

Ligands

(a)

Distance

(b)

Figure 14.1 (a) Typical energy band structure of multilayer of QDs. Even when QDs are intimately touching, dot-to-dot charge transport is relatively weakened by the presence of the shell and ligand. (b) The use

Distance

of organic layers facilitates the process of charge transport to and from the QDs. CB, conduction band; valence band; HOMO, highest occupied molecular orbital; LUMO, lowest unoccupied molecular orbital.

light-emitting diodes (QDLEDs), organic compounds are used as charge transport layers, indium tin oxide (ITO) is used as a hole-injecting electrode, and a metal (e.g., Al, Mg, or Ca) is used as an electron-injecting electrode. Having an FWHM of less than 20 nm, QDs promise to allow for better color gamut to be obtained than the best OLEDs available. The ease of processing of QDs stems from the above-mentioned organic capping ligand, which renders the QDs soluble and thus readily printable via inkjet or other higher throughput printing techniques. The success of such a manufacturing approach will revolutionize the display industry and have a positive effect on other relevant technologies.

14.2 Background

Even though some of the first studies of QD confinement in II–VI semiconductors were reported in the 1960s [7, 8], it was not until three decades later when the first reports of hybrid electroluminescent (EL) QDLEDs were published. Colvin et al. [5] dipped ITO/p-paraphenylene vinylene (PPV), which was functionalized with hexane dithiol, into dispersion of QDs and toluene. With magnesium as a cathode, the resulting devices emitted 100 cd m−2 and had external quantum efficiency (EQE) of 0.01−0.001%. Kumar et al. [9] used a similar approach but with cadmium sulfide (CdS) QDs spin coated onto p-PPV film, which was then capped with an aluminum cathode. The QD layer in these devices functioned as a charge transporter with emission dominated by PPV. Luminance reached 150 cd m−2 and EQE ∼1%.

14.2 Background

A remarkable step forward in the research on QDLEDs was in 2002 when Coe et al. reported a 25-fold increase in power efficiency. The new approach was to embed the QDs between a hole transport layer (HTL) and an electron-transport layer (ETL). To achieve this, the QDs were dispersed in a N,N  -bis(3-methylphenyl)-(1,1 -biphenyl)-4,4 -diamine (TPD) solution and spin coated. Phase separation in the spin-coated film drove the QDs to the top of the TPD layer, forming a monolayer [10]. In these devices, tris(8-hydroxyquinolinato) aluminum (Alq3) was used as an ETL. Such an approach of using organic HTL and ETL to sandwich the QD layer has been used for many years thereafter. The enhanced efficiency of 0.52% reported by Coe et al. was explained by improved exciton formation in the QDs, mainly due to a F¨orster energy transfer from HTL–ETL excitons and direct charge injection. The same research group later optimized the quality of the QD film, resulting in doubling of the EQE [11]. By 2005, an EQE of about 2% and luminance close to 7000 cd m−2 were reported [12]. Spin coating remained the most popular method to deposit the QD layer because of its convenience, simplicity, and ease of accessibility. While depositing the QD interlayer, it is critical to ensure that no dissolution of the underneath layer occurs. In this regard, Chaudhary et al. [13] have used solvents with incompatible polarity during device fabrication. In this case, the HTL and ETL were polymers dissolved in nonpolar solvents, while QDs were dispersed into a polar aqueous solution. An EQE of 0.22% and luminance of 500 cd m−2 were reported. A more robust and reliable approach was introduced by Zhao et al. who used thermally cross-linkable poly-TPD as HTL, thus rendering it insoluble on the application of the QD solution on top of it. Devices made with such materials showed nearly 1% EQE and a luminance of 1000 cd m−2 . Further thermal treatment of the device layers doubled the EQE and allowed for a luminance of nearly 6000 cd m−2 [14, 15]. In addition, enhancements in QD synthesis and processing have also led to an improved performance of QDLEDs [16, 17]. The latest reports at the time of writing this manuscript indicate the possibility of QDLEDs with maximum brightness exceeding 100 000 cd m−2 [18]. Other QD deposition approaches such as mist coating [19] and dip coating [20], have been demonstrated. However, the drawback in all of these approaches is that they do not allow for selective and controlled patterning of QDs. The inherent deficiency of spin coating in multicolor patterning renders it impractical in display manufacturing. Moreover, the relatively high cost of QDs and the loss of more than 96% of the solution during spin casting are strong reasons to search for an alternative deposition route capable of multicolor patterning and consuming minimum amounts of QD materials. Microcontact stamping can be used for high-definition patterning. In this method, typically, polydimethylsiloxane (PDMS) is deposited onto the molding mask and released after curing, resulting in a stamp resembling the original mask. The stamp can then be used to transfer a thin layer of QDs onto the substrate. Indeed, this method was used by Bulovic et al. and Gigli et al. [21–26], who reported that monochromatic RGB and white QDLEDs give good performance. However, a serious drawback that renders this approach unattractive is the use of spin coating to deposit the QD layer on the stamp before its consequent transfer to the substrate.

219

220

14 Application of Inkjet Printing in High-Density Pixelated RGB Quantum Dot-Hybrid LEDs

Perhaps the most attractive solution to accurately place various colors of QDs in predefined places on the substrate is inkjet printing. In recent years, there has been an increase in the use of inkjet printing in various areas of materials research and device fabrication [27, 28]. Some of the works that utilize the inkjet printing approach include on-demand modification of sheet resistivity of conducting polymers, monochrome and full-color polymers, light-emitting displays [29, 30], and biological materials applications [31], to mention a few. Haverinen et al. [32, 33] have reported the first demonstration of DC EL inkjet-printed QDs. AC-driven hybrid QDLEDs, in which the dots have been inkjet printed, were reported by Wood et al. [34]. The emphasis of this report is to present a simple discussion of the first attempts to fabricate high-density, pixelated (quarter video graphics array (QVGA) format), monochromatic and RGB QDLEDs, where inkjet printing is used to deposit the light-emitting layer of QDs. We show some of the factors that have to be considered in order to achieve the desired accuracy and printing quality.

14.3 Experimental Procedure and Results

We start by demonstrating a monochrome high-density pixel array. Here, an organic HTL of cross-linkable poly-TPD (poly (N,N  -bis(4-butylphenyl-N,N  -bis (phenyl)benzidine)) was used as received from American Dye Source (Montreal, Canada). Poly TPD (18 mg) was left to stir overnight in 3 ml of chloroform at room temperature. The solution was then filtered and spin coated on precleaned glass/ITO at a substrate speed of 2000 rpm for 60 s. After the polymer deposition step, the substrates were exposed to UV light (245–254 nm) for 10 min in order to allow the polymer to cross-link. Measurements of the cross-linked layer, using Dektak 3030 stylus profilometer, revealed a 20 nm thickness. Upon cross-linking, the QD solution (4 mg in 1 ml of chlorobenzene) was deposited to obtain the light-emitting layer of the device. The QDs were used as received from Ocean Nanotech, LLC. The information received from the supplier specified a core-shell type QD structure of CdSe (core) and CdS/ZnS (double shell), with a total average dot diameter of about 9 nm. This value includes a 3 nm coating of octadecylamine, which drastically enhances the solubility of the QDs as opposed to bare particles that tend to aggregate [35]. A Fuji-Dimatix DMP 2800 materials printer was used to deposit the QD solution. The QD ink was placed into a cartridge having 16 piezoelectrically actuated nozzles capable of dispensing a liquid drop volume of 10 pl. Optimum deposition conditions were obtained after carefully adjusting the bias amplitude and electrical waveform driving the actuator, the print head, and the substrate holder temperature. Formulation of the proper solution for inkjet process is critical for the success of our experiment. A simple comparison between two solvents, mainly toluene and chlorobenzene, will help justify our choice of solvent: Toluene has low viscosity of 0.590 cP and room temperature vapor pressure of 2.9 kPa. On the other hand,

14.3 Experimental Procedure and Results

chlorobenzene has relatively higher viscosity of 0.8 cP and lower vapor pressure of 1.6 kPa. A lower vapor pressure is needed in order to extend the drying time and to assist in better film formation through the reduction of coffee-ring effects [36, 37]. It is critical to eliminate coffee-ring effects especially for light-emitting layers in order to avoid any variation in the deposited layer thickness. Variation in thickness of the printed QD layer leads to varying electric field distribution over the device active area, with higher currents propagating through thinner locations of the sample, leading to uneven light emission over the device area and in turn jeopardizing overall device reliability. 14.3.1 Role of Droplet Formation

In addition, proper droplet formation is also critical in inkjet printing. Figure 14.2 reveals the camera view of droplets right after their ejection from the nozzle. In case of toluene, droplet formation was unfavorable as satellite drops were observed during the jetting process despite optimizing the driving conditions of the piezoelectric. Small droplets following the main drop lead to inaccuracy in droplet positioning on the substrate. Moreover, the amount of materials deposited becomes harder to control, resulting in unwanted film inhomogeneity. However, when chlorobenzene is used as a solvent, printing conditions can be easily optimized and controlled droplet formation and jetting without any satellites can be obtained, Figure 14.2. It is also critical to present favorable conditions at the substrate surface in order to obtain acceptable film quality. For chlorobenzene, nearly complete wetting is obtained on the poly-TPD surface with a wetting angle of about 0◦ , while the same

Mirror image Nozzle

Nozzle

(a)

Flying drop

Flying drop

(b)

Figure 14.2 (a) Toluene solvent showing poor jetting properties, indicating the formation of satellite drops (smaller drops lagging behind the main leading drop) and (b) chlorobenzene solvent showing perfect drop jetting.

221

222

14 Application of Inkjet Printing in High-Density Pixelated RGB Quantum Dot-Hybrid LEDs

Contact angle 78.4°

(a) Figure 14.3

Contact angle ~0°

(b) Contact angle of QD/chlorobenzene ink on (a) PVK:TPD and (b) poly-TPD.

solvent forms a wetting angle of 78.4◦ on the PVK:TPD surface, as measured in air and at room temperature (Figure 14.3). On printing the QD solution, the substrates (1 in. × 1 in.) were then dried and moved into a deposition chamber situated in a nitrogen glove box. Here, a layer of electron-transport and hole-blocking TPBi (1,3,5-tris(2-N-phenylbenzimidazolyl) benzene) was deposited at a thickness of 20 nm, followed by the deposition of LiF(1.5 nm)/Al (180 nm) cathode [38, 39]. 14.3.2 Atomic Force Microscopy

The morphology of printed QD layers on top of poly-TPD were studied via atomic force microscopy (AFM). As shown in Figure 14.4, a close-packed arrangement of the QDs with short-range order can be clearly seen. In order to shed more light

100 nm (a)

100 nm (b)

Figure 14.4 Atomic force microscope (AFM) height image (a), and phase image (b) showing that there is no wide-range regularity in packing. Peak-to-valley distance is 5.22 nm [32].

14.3 Experimental Procedure and Results

Figure 14.5 Mesa structure of printed QD film. Inset shows the height variation in the marked region (white line).

on the morphology aspect of the printed film, phase scan was gathered as shown in Figure 14.4b. The exceptionally sensitive approach to detect minor changes in morphology, especially when relatively smooth samples are studied, allows for a higher resolution to be obtained with phase imaging. Owing to the subnanometer difference in the nominal height, it is normally difficult to obtain a detailed picture of the QD packing by simple analysis of the scanner’s movements on the z-axis (height). A phase image provides information that points to differences in composition, as well as additional information regarding surface packing of the QDs. In our case, the darker regions seen in the height image are clearly resolved in the phase image, revealing the close-packed structure of QDs. Further analysis of the AFM height image indicates the presence of a mesa-type structure, which is a clear indication of the three-dimensional nature of the printed layer as to a single continuous monolayer (Figure 14.5). The measured average roughness is 0.58 nm with a peak-to-valley height of 5.22 nm. The AFM scans indicate that the extent of the structural order is limited to very small regions. The quality of the polymer surface and the relatively inhomogeneous QD size distribution are some of the reasons that prevent the growth of larger ordered grains of QDs in the printed film. This size variation, along with the layered nature of the printed film, makes it difficult to assess accurately the diameter of the QDs. In order to facilitate the use of AFM to accurately measure the size of the QDs, a monolayer spread over an ultrasmooth substrate surface is needed. Using polished silicon wafer as the substrate and after optimizing the deposition conditions of a solution with concentration of less than 0.5 mg ml−1 , we were able

223

14 Application of Inkjet Printing in High-Density Pixelated RGB Quantum Dot-Hybrid LEDs

224 8.0 nm 0.0

m

200

500 nm

500 nm

400

400

300

300

200

200

100

100

(a)

100

200

300 400

500 nm

10 5 0 −5 nm 100 200 300 400 500 600 nm

(b)

(c) Figure 14.6 AFM analysis of quantum dot size (a) 3D height image of spin-coated quantum dots (500 nm scan size) reveals aggregates and uneven surface of quantum dot thin film. (b) Typical size based on AFM

was ∼5–6 nm. (c) Phase image gives better perspective for the size and packing of the dots. Dashed circle points to aggregation region. Whitish large circular areas are regions of exposed silicon surface.

to obtain a monolayer of QDs with voids exposing the silicon surface. Although at first the voids might not look attractive, they are very critical in acting as a reference during height measurement. Before using the AFM, the QD film was plasma ashed to remove all organic ligands, thus exposing the outer shell surface of the QDs. Owing to the spherical apex of the tip, distortion in scan lines will be present. This does not estimate properly the size information obtained in lateral dimension of the scan, and thus must be ignored. Accurate information in this case comes from the vertical direction of the scan, which is not affected by the tip geometry. The results of the AFM measurement, as shown in Figure 14.6, indicate a QD size of about 5–6 nm. This is similar to what is observed in the transmission electron microscopy (TEM) measurements shown in Figure 14.7, which point to a diameter of 4–6 nm for bare QDs (without capping organic ligand), a value in accord with that reported by the supplier. As previously mentioned, the quality of the underlying polymer surface also affects the growth of the QD layer. In this case, we have studied the impact of the solvent in inkjet versus spin coating. Chlorobenzene was deposited on top of the TPD polymer layer using spin coating and inkjet printing. In both the cases, the polymer surface morphology was studied using AFM before and after solvent deposition. The results indicate virtually no significant effect of the solvent on the surface roughness of the poly-TPD layer, while noticeable changes are observed in the polymer film in the inkjet printing process. The average roughness of the cross-linked poly-TPD surface is 0.58 nm, which remains roughly the same

14.3 Experimental Procedure and Results

5 nm

(a)

5 nm

(b)

Figure 14.7 TEM images of the quantum dots reveal nonuniform size distribution of the dots. (a) 285 000 magnification and (b) 400 000 times magnification. Both images are from different areas.

(0.51 nm) upon spin coating of the solvent. However, the roughness seems to nearly double (0.97 nm) after inkjet printing with the appearance of scattered islands of various sizes. The dimension of such features can reach 6 nm in height and several microns in width. Such features can present a strong barrier, preventing QDs packing into larger ordered grains, and constraining the growth to a rather shorter range order. The absence of such features in the spin-coating process is due to the relatively faster dynamics of the process, which does not allow enough time for the diffusion and aggregation of the dissolved polymer material. On the contrary, the higher residual solvent volume in the inkjet process provides enough time for the non-cross-linked parts of the polymer to diffuse and aggregate, forming the observed islands upon drying, Figure 14.8. 14.3.3 Electroluminescence

Figure 14.9 shows the EL properties of the QDLEDs, where the QD layer was printed continuously on a 1 in. × 1 in. area of the substrate. A circular cathode (0.14 cm2 ), which defined the active area of the device, was used (inset of Figure 14.9). The sharp emission peak at 600 nm is due to the QDs, while the wider emission around 450–470 nm is a mixed contribution of poly-TPD (about 410 nm) and exciplex formation between the poly-TPD and TPBi. The presence of exciplex emission is an indication of the intimate contact between the two organic materials pointing to the presence of pinholes in the printed QD layer. The maximum EQE is 0.19 %, at a bias voltage of 14.2 V, as shown in Figure 14.10, with more than 87 % of the emission originating from the QDs. The maximum forward light output is 381 cd m−2 , at 15.9 V. To appreciate the inkjet printing approach, similar devices having a spin-coated layer of QDs were made and tested. In this case, the QDLEDs

225

226

14 Application of Inkjet Printing in High-Density Pixelated RGB Quantum Dot-Hybrid LEDs

10 nm

6.0 nm 0.0

30 μm

30 μm

0.0

30 μm

30 μm

20

20 20

20

10

10 10 (a)

10 10

30 μm

20

(b)

10

6.6 nm

0.0 30 μm

30 μm 20

20 10

10 (c)

10

20

Figure 14.8 Three AFM height images of poly-TPD layers after several solvent treatments. (a) Cross-linked poly-TPD layer before solvent treatment, roughness 0.577 nm. (b) Spin coating of chlorobenzene, roughness 0.514 nm. (c) Inkjet printing case results in roughness of 0.972 nm. All scans are 30 μm on one side.

1.0

Intensity (a.u.)

0.8 0.6 0.4 0.2 0.0 100 200 300 400 500 600 700 800 900 Wavelength (nm) Figure 14.9 EL of the QDLEDs showing emission from the organic layers (400–500 nm) and quantum dot emission (near 600 nm). Of the emission, 87.7% comes from the dots; inset shows a photo of QDLEDs having an active area of 0.14 cm2 [32].

30 μm

20

30 μm

14.3 Experimental Procedure and Results

400 0.20 EQE(%)

Luminance (cd m−2)

350 300 250

0.15 0.10 0.05 0.00 2

4

200

6

8 10 12 14 16 Bias (V)

150 100 50 0

0

2

4

6

8 10 Bias (V)

12

14

16

18

Figure 14.10 Luminance versus bias voltage. The maximum brightness is 381 cd m−2 , at 15.9 V. Inset shows EQE versus bias voltage [32].

showed a maximum EQE of 0.15% at an applied bias voltage of 20.7 V, and a maximum brightness of 244 cd m−2 at a bias voltage of 18.7 V. The following example demonstrates our first attempt at using inkjet to print a monochrome QD solution over a pixelated substrate surface having a QVGA (640 × 480) pixels). The individual pixels consisted of an L-shaped photoresist bank having a dimension of 160 μm × 160 μm (Figure 14.11), which was photolithographically patterned on top of an ITO/glass. Inkjet printing of the QD solution was controlled to take into account the spacing between the pixels so that the ink is dispensed only within pixel boundaries. The number of jetted drops per pixel was optimized in order to obtain best coverage of the pixel surface area. The cathode covered nearly 243 pixels, leading to their simultaneous operation on applying a bias voltage (Figure 14.12). The number of drops per pixel was varied to obtain various devices, which allowed us to optimize the characteristics of the emitted spectrum (Figure 14.13). It is clear from the figure that a count of 15 drops per pixel leads

Figure 14.11 Geometry of the pixels used in this work. Not drawn to scale (L shape is 160 μm × 160 μm).

227

14 Application of Inkjet Printing in High-Density Pixelated RGB Quantum Dot-Hybrid LEDs Figure 14.12 Photograph of DC-driven QDLED having about 243 pixels under a common cathode [32].

Normalized intensity (a.u.)

1.0

5 drops/pixel 10 drops/pixel 15 drops/pixel

0.8 0.6 0.4 0.2 0.0 0

100

200 300 400 500 Wavelength (nm)

(a)

600

700

200 Luminance (cd m−2)

228

10 drops/pixel 15 drops/pixel

150

100

50

0 −2 (b)

0

2

4

6

8 10 12 14 16 18 20 22 Bias (V)

Figure 14.13 Exciplex emission can be controlled (a) by increasing the amount of drops per pixel. (b) Changes in film thickness increases the turn on voltage.

14.4 Inkjet-Printed, High-Density RGB Pixel Matrix

to the dominance of QD emission in the EL spectrum, indicating a near complete coverage of the printed film. The calculated thickness of QDs in this case is about 9 nm, equivalent to monolayer QD coverage. Figure 14.13 also shows the luminance versus bias voltage of QDLEDs having 10 and 15 drops per pixel. In this case, a brightness of 1 cd m−2 can be obtained between 5–6 and 10.6 V, respectively. The maximum brightness (150 cd m−2 ) is seen at about 12 V (10 drops per pixel) and 18 V (15 drops per pixel). The CIE coordinates of such devices are (0.424; 0.294) and (0.587; 0.328), respectively.

14.4 Inkjet-Printed, High-Density RGB Pixel Matrix

Having demonstrated the potential of inkjet printing to allow for accurate deposition of the light-emitting layer in a high-density monochrome QDLED pixel format, we now show that it is possible to use the same approach to deposit RGB subpixels with high accuracy (using the same format and density as discussed before), thus enabling the fabrication of the first DC-driven full-color QVGA QDLED pixel format [33]. The first step was to optimize device characteristics of each color as done in the above discussion using the same pixelated substrate format. QDs of various diameters were used in order to obtain the R, G, and B colors. All have the structure of CdSe core, ZnS passivating shell, and an organic monolayer. The QD solution for each color had a concentration of 2.5 mg ml−1 . The solvent used was a 1 : 1 ratio of toluene: dichlorobenzene. A solvent other than chlorobenzene was used because of the different nature of the organic ligand on the R, G, and B dots, which was not disclosed to us by the supplier. The R QDs having a size of 5.0 nm emit light at a wavelength of 600 nm ± 10 nm, G QDs (3.4 nm) emit light at 540 nm ± 10 nm, and B QDs (3.2 nm) showed light emission at 490 nm ± 10 nm. A 10 pl cartridge was used for each solution (color). During printing, only one nozzle at a time was used. Within each pixel, similar drop spacing was maintained during the deposition of the R and G QDs. However, smaller drop spacing was needed to obtain an acceptable layer quality of the B QDs. The device fabrication proceeded by spin coating poly[3,4-ethylenedioxythiophene] doped with poly[styrenesulfonate] (PEDOT: PSS) at 5000 rpm for 60 s on top of the ITO electrode. The resulting layer was then postbaked for 30 min at 150 ◦ C in air, resulting in a final film thickness of caabout 30 nm. PEDOT:PSS functioned as a hole-injection layer. On moving the substrates into a nitrogen-filled glove box, they were reheated to 120 ◦ C for 5 min to get rid of any water absorbed during the transport step. A solution of photocross-linkable poly-TPD [poly-(N,N-bis(4-butylphenyl-N,N-bis(phenyl)benzidine)] was then spin coated on top of the PEDOT:PSS. The HTL solution had the concentration of 6 mg ml−1 , where chloroform was used as a solvent. The spin speed of the HTL deposition process was 2000 rpm for 1 min. The HTL polymer was then cross-linked by exposing it to UV radiation having a wavelength of 245–254 nm for 10 min. The

229

230

14 Application of Inkjet Printing in High-Density Pixelated RGB Quantum Dot-Hybrid LEDs

140 nm

Al

1.5 nm LiF

− 40 nm +

TPBi

40 nm P-TPD

30 nm PEDOT:P

SS

TO Glass/I

Figure 14.14 Schematic of a QDLED device structure [33].

168 μm 80 μm

50

μm

16

3μ

m

80 μm

60 μm 108 μm

Figure 14.15 Schematic showing the printing sequence of RGB subpixels along with the relevant dimensions.

process resulted in a total HTL thickness of 40 nm. This was followed by inkjet printing in air of the R, G, and B QD solutions. Figure 14.14 depicts the QDLED structure made in this study. For the R, G, and B monochrome devices, the printing was carried out on each and every pixel of the patterned substrate totaling in an area of 1 in. × 1 in. A sequence of RGBRGBRGB . . . was used to deposit the three colors on the same substrate, as shown in Figure 14.15. Here, we would like to point out that it is not possible for the cartridge of the inkjet used to hold three different solutions (R, G, B), simultaneously. Therefore, each solution must be printed in a separate step with accurate control of the placement of the printed drops in order to avoid color contamination. A heating step of 180 ◦ C for 10 min in nitrogen helped eliminate water and organic ligands from the printed QD pixels [8]. The substrates were then placed in a vacuum chamber where a 40 nm layer of TPBi 1,3,5-tris(2-N-phenylbenzimidazolyl)benzene followed by 1.5 nm LiF and 140 nm Al cathode [9, 10] were deposited.

Normalized intensity (a.u.)

14.4 Inkjet-Printed, High-Density RGB Pixel Matrix

1.0 0.8 0.6 0.4 0.2 0.0 300

400

500 600 Wavelength (nm)

700

Figure 14.16 QDLED emission showing EL originating mostly from QDs. The R and G spectra were taken at 15 V and that of B at 20 V [33].

In Figure 14.16, the EL spectra of each of the R, G, and B QDLEDs are shown. A common cathode allowing simultaneous operation of nearly 243 pixels was used, as described earlier. All devices operated at a relatively low bias between 4 and 5 V indicating a very thin QD layer (on the order of a monolayer). At a lower voltage, no indication of emission from the organic layers was observed. However, as the applied voltage increased, slight emission from organic layers and their exciplex was observed, as seen in Figure 14.16, which was taken at a relatively high operating voltage (15 V for R and G, and 20 V for B). In this case, the peak emission and (CIE) are 605 nm (0.57, 0.37), 550 nm (0.34, 0.56), and 475 nm (0.12, 0.13), for R, G, and B, respectively. Relatively high EQE can be seen at a low operating voltage (1.5% R, 0.6% G, and 0.26% B). However, the forward light output is very low (few cd m−2 ) and not useful for display application. It is worth noting that the R QDLEDs had the best performance, while the blue had the worst. Figures 14.17 and 14.18 show that 400

Brightness (cd m−2)

350 300 250 200 150 100 50 0

0

5

10

15

20

Bias (V)

Figure 14.17 Brightness versus bias voltage for R, G, and B devices having printed QD layer [33].

231

14 Application of Inkjet Printing in High-Density Pixelated RGB Quantum Dot-Hybrid LEDs

1.6 1.4 1.2 EQE (%)

232

1.0 0.8 0.6 0.4 0.2 0.0

4

6

8

10 12 14 16 18 20 22 24 26 Bias (V)

Figure 14.18 EQE versus bias voltage measured for each device having R, G, and B QD printed layers. Note that R QDLEDs are most efficient at any given voltage [33].

(a)

(b)

(c)

Figure 14.19 Photographs taken with a commercial camera of monochrome QVGA QDLEDs operating under DC voltage where the QD layer was inkjet printed. The quality of the camera did not allow us to produce the actual colors (especially for B) seen in Figure 14.16 [33].

a maximum brightness of 352 cd m−2 (EQE about 0.23%) at 17.5 V, 270 cd m−2 (EQE 0.15%) at 18 V, and 122 cd m−2 (EQE about 0.1%) at 16.5 V, can be seen for R, G, and B QDLEDs, respectively. Figure 14.19 shows a photograph of R, G, and B pixels under a DC bias. A stronger overlap of the absorption of R and G QDs and the emission of organic materials contributes to the better performance of such devices. This is not the case for the B QDs, where weaker overlap between the absorption of B QDs and the emission of p-TPD (and TPBi) is present. In this case, incomplete energy transfer from the organics to the QDs along with possible exciplex formation due to pinhole defects in the film contributes negatively to the quantum efficiency of QDLEDs. Figure 14.20 shows an inset depicting the photograph of an RGB QDLED operating at a forward DC voltage of 10 V. The apparent color imbalance seen in

14.4 Inkjet-Printed, High-Density RGB Pixel Matrix

Intensity (a.u.)

1.0 0.8 0.6 0.4 0.2 0.0 300

400

500

600

700

Wavelength (nm) Figure 14.20 RGB spectrum measured at 10 V. Inset: photograph showing strong emission from red pixels, which is stronger than that of G and B pixels. The overall device area is 0.14 cm2 , with both ITO and metal as common electrodes to all array elements [33].

400 1.0 0.8 EQE (%)

Luminance (cd m−2)

350 300 250

0.6 0.4 0.2 0.0

200

5

15

10

20

Bias (V)

150 100 50 0

0

5

10 Bias (V)

15

20

Figure 14.21 Luminance of RGB device versus bias voltage. A peak brightness of 350 cd m−2 is recorded at 17.5 V [33].

the photograph is due to the use of common electrodes and the low sensitivity of the digital camera used [33]. As shown for the monochrome devices, one anticipates that the R QDLED emission will dominate that of the other two colors, which is the case as shown in Figure 14.20. As the voltage increased, the overall shape of the EL spectrum did not change. The operational voltage, where whitish light was visible, was 5.2 V. The maximum brightness of the device was 350 cd m−2 (EQE is 0.14%), measured at 17.5 V, as depicted in Figure 14.21. In this case, a video brightness of 100 cd m−2 required only a bias of 9.3 V (EQE 0.24%).

233

234

14 Application of Inkjet Printing in High-Density Pixelated RGB Quantum Dot-Hybrid LEDs

14.5 Conclusion

In this chapter, we have demonstrated the feasibility of using the inkjet printing approach to deposit QDs in a predefined pixel pattern having a QVGA format. The successful operation of the RGB printed devices indicates the potential of the inkjet printing approach in the fabrication of full-color QDLEDs for display application. However, further optimization of print quality is still needed in order to eliminate the formation of pinholes, thus maximizing energy transfer from organic layers to the QDs and in turn increasing the performance of the devices.

Acknowledgment

Dr. H.M. Haverinen thanks TEKES (Finnish Funding Agency for Technology and Innovation), Graduate School of Modern Optics and Photonics for funding, Dr. Xiaohui Yang and Mr. Rafal Sliz for technical support. Prof. G.E. Jabbour thanks the Academy of Finland, Distinguished Professor of Finland program, and Fuji Dimatix for their support. The authors also thank Mr. Scott Ageno of the Flexible Display Center at ASU for providing the patterned substrates.

References CdSe nanocrystals. J. Appl. Phys., 83, Organic electroluminescent diodes. Appl. 7965–7974. Phys. Lett., 51, 913–915. 7. Bergstresser, T.K. and Cohen, M.L. (1967) Electronic structure and optical Lee, J.H., Ho, Y.H., Lin, T.C., and Wu, properties of hexagonal CdSe, CdS and C.F. (2007) High efficiency fluorescent ZnS. Phys. Rev., 164, 1069–1080. blue organic light emitting device with balanced carrier transport. J. Electrochem. 8. Lawson, W.D., Smith, F.A., and Young, Soc., 154, 226–228. S.A. (1960) Influence of crystal size on the spectral response limit of evaporated Li, R.Z. and Meng, H. (2007) Organic PbTe and PbSe photoconductive cells. J. Light Emitting Materials and Devices, Electrochem. Soc., 107, 206–210. Taylor & Francis Group, LLC. Xing, Y., Xu, X., Zhang, P., Tian, W., 9. Kumar, N., Joshi, M.P., Friend, Yu, G., Lu, P., Liu, Y., and Zhu, D. C.S., and Prasad, P.N. (1997) (2005) Carbazole-pyrene based organic Organic-inorganic heterojunction emitters for electroluminescent device. light emitting diodes based on Poly Chem. Phys. Lett., 408, 169–173. (p-Phenylene Vinyle/Cadmium Sulfide thin films. Appl. Phys. Lett., 71, Colvin, V.L., Schlamp, M.C., and 1388–1390. Alivisatos, A.P. (1994) Light-emitting diodes made from cadmium selenide 10. Coe, S., Woo, W.K., Bawendi, M., and nanocrystals and a semiconducting Bulovic, V. (2002) Electroluminescence polymer. Lett. Nat., 370, 354–357. from single monolayers of nanocrystals in molecular organic devices. Nature, Dabbousi, B.O., Radzilowski, L.H., 420, 800–803. Thomas, E.L., Bawendi, M.G., and Rubner, M.F. (1998) Electrolumi11. Coe-Sullivan, S., Woo, W.K., Steckel, J., Bawendi, M., and Bulovic, V. nescence from heterostructures of (2003) Tuning the performance of Poly(Phenylene Vinylene) and inorganic

1. Tang, C.W. and Van Slyke, S.A. (1987)

2.

3.

4.

5.

6.

References

12.

13.

14.

15.

16.

17.

18. 19.

20.

hybrid organic/inorganic quantum dot light-emitting devices. Org. Electron., 4, 123–130. Coe-Sullivan, S., Steckel, J.S., Woo, W.K., Bawendi, M.G., and Bulovic, V. (2005) Large-area ordered quantum-dot monolayers via phase separation during spin casting. Adv. Funct. Mater., 15, 1117–1124. Chaudhary, S., Oxkan, M., and Chan, W.C. (2004) Trilayer hybrid polymer-quantum dot light emitting diodes. Appl. Phys. Lett., 84, 2925–2928. Zhao, J., Bardecker, J.A., Munro, A.M., Liu, M.S., Niu, Y., Ding, I.K., Luo, J., Chen, B., Jen, A.K.Y., and Ginger, D.S. (2006) Efficient CdSe/CdS diodes using a thermally polymerized hole transport layer. Nanoletters, 6, 463–467. Niu, Y.H., Munro, A.M., Cheng, Y.J., Tian, T., Liu, M.S., Zhao, J., Bardecker, J.A., Plante, I.J.L., Ginger, D.S., and Jen, A.K.Y. (2007) Improved performance from multilayer quantum dot light emitting diodes via thermal annealing of the quantum dot layer. Adv. Mater., 19, 3371–3376. Bae, W.K., Kwak, J., Park, J.W., Char, K., Lee, C., and Lee, S. (2009) Highly efficient green light-emitting diodes based on CdSe/ZnS quantum dots with a chemical-composition gradient. Adv. Mater., 21, 1690–1694. Bae, W.K., Kwak, J., Lim, J., Lee, D., Nam, M.K., Char, K., Lee, C., and Lee, S. (2009) Deep blue light emitting diodes based on Cd1-xZnx/ZnS quantum dots. Nanotechnology, 20, 075202. Lee, C.H. (2010) MRS Fall 2010, Boston, MA. Zhu, T., Shanmugashundaram, K., Price, S.C., Tuzyllo, J., Zhang, F., Xu, J., Mohney, S.E., Zhang, Q., and Wang, A.Y. (2008) Mist fabrication of light emitting diodes with colloidal nanocrystals quantum dots. Appl. Phys. Lett., 92, 023111–023113. Lee, C.L., Nam, S.W., Kim, V., Kim, J.J., and Kim, K.B. (2009) Electroluminescence from monolayer of quantum dots formed by multiple dip-coating processes. Phys. Status Solidi B, 246, 803–807.

21. Anikeeva, P.O., Halpert, J.E., Bawendi,

22.

23.

24.

25.

26.

27.

28.

29.

30.

M.G., and Bulovic, V. (2007) Electroluminesence from a mixed red-green-blue colloidal quantum dot monolayer. Nanoletters, 7, 2196–2200. Anikeeva, P.O., Madigan, C.F., Halpert, J.E., Bawendi, M.G., and Bulovic, V. (2008) Electronic and excitonic processes in light-emitting devices based on organic materials and colloidal quantum dots. Phys. Rev. B, 78, 085434–085441. Kim, L., Anikeeva, P.O., Coe-Sullivan, S., Steckel, J.S., Bawendi, M.G., and Bulovic, V. (2008) Contact printing of quantum dot light-emitting devices. Nanoletters, 8, 4513–4517. Steckel, J.S., Snee, P., Coe-Sullivan, S., Zimmer, J.P., Halpert, J.E., Anikeeva, P.O., Kim, L.A., Bulovic, V., and Bawendi, M.G. (2006) Color-saturated green-emitting QD-LEDs. Angew. Chem. Int. Ed., 45, 5769–5799. Mazzeo, R.A., Biasiucci, M., Cingolani, R., and Gigli, G. (2008) White electroluminescence from a microcontact-printing-deposited CdSe/ZnS colloidal quantum-dot monolayer. Small, 4, 2143–2147. Rizzo, A., Mazzeo, M., Palumbo, M., Lerario, G., D’Amone, S., Cingolani, R., and Gigli, G. (2008) Hybrid light-emitting diodes from microcontact-printing double-transfer on colloidal semiconductor CdSe/ZnS quantum dots onto organic layers. Adv. Mater., 20, 1886–1891. Yoshioka, Y. and Jabbour, G.E. (2006) Desktop inkjet printer as a tool to print conducting polymers. Synth. Met., 156, 779–783. Singh, M., Haverinen, H.M., Dhagat, P., and Jabbour, G.E. (2010) Inkjet printing – process and its applications. Adv. Mater., 22, 673–685. Yoshioka, Y., Calvert, P.D., and Jabbour, G.E. (2005) Simple modification of sheet resistivity of conducting polymeric anodes via combinatorial ink-jet printing techniques. Macromol. Rapid Commun., 26, 238–246. Hebner, T.R., Wu, C.C., Marcy, D., Lu, M.H., and Sturm, J.C. (1998) Ink-jet printing of doped polymers for organic

235

236

14 Application of Inkjet Printing in High-Density Pixelated RGB Quantum Dot-Hybrid LEDs

31.

32.

33.

34.

35.

36.

light emitting devices. Appl. Phys. Lett., Poly[2-methoxy-5-(2’-ethylhexyloxyl)– 72, 519–521. 1,4-phenylenevinylene] (MEH– PPV) by Calvert, P., Yoshioka, Y., and Jabbour, ink-jet printing. Adv. Funct. Mater., 17, G.E. (2004) Learning from Nature How 277–284. to Design New Implantable Biomaterials: 37. Deegan, R.D., Bakajin, O., Dupont, T.F., From Biomineralization Fundamentals Huber, G., Nagel, S.R., and Witten, T.A. to Biomimetic Materials and Processing (1997) Capillary flowasthe cause of ring Routes, Kluwer Academic, p. 169. stains fromdried liquid drops. Nature, 389, 827–829. Haverinen, H.M., Myllyl¨a, R.A., and Jabbour, G.E. (2009) Inkjet printing of 38. Hung, L.S., Tang, C.W., and Mason, light emitting quantum dots. Appl. Phys. M.G. (1997) Enhanced electron injecLett., 94, 073108–073110. tion in organic electroluminescence devices using an Al/LiF electrode. 70, Haverinen, H.M., Myllyl¨a, R.A., and 118344–118346. Jabbour, G.E. (2010) Inkjet printed RGB quantum dot-hybrid LED. IEEE J. Disp. 39. Jabbour, G.E., Kawabe, Y., Shaheen, Technol., 6, 87–89. S.E., Morrell, M.M., Wang, J.F., Kippelen, B., and Peyghambarian, N. Wood, V., Panzer, M., Chen, J., Bradley, (1997) Highly efficient and bright orM.S., Halpert, J., Bawendi, M.G., ganic electroluminescent devices with an and Bulovic, V. (2009) Inkjet-printed aluminum cathode. 71 (2). quantum dot-polymer composites for full-color AC-driven displays. Adv. Mater., 21, 2151–2155. Xie, R., Kolb, U., Li, J., Basche, T., Further Reading and Mews, A. (2005) Synthesis and characterization of highly luminescent Yoshioka, Y. and Jabbour, G.E. (2006) Inkjet CdSe-core CdS/Zn0.5Cd0.5S/ZnS mulprinting of oxidants for patterning of tishell nanocrystals. J. Am. Chem. Soc., nanometer-thick conducting polymer 127, 7480–7488. electrodes. Adv. Mater., 18, 1307–1312. Tekin, E., Holder, E., Kozonaev, D., and Schubert, U.S. (2007) Controlled pattern formation of

237

15 Inkjet Printing of Metal Oxide Thin-Film Transistors Jooho Moon and Keunkyu Song

15.1 Introduction

Metal oxide semiconductors have attracted considerable attention because of their superior material properties, including their band gap, transparency, and high field-effect mobility compared to conventional a-Si:H and organic semiconductors in applications requiring transparent thin-film transistors (TFTs). These materials meet the combined requirements of high-performance semiconducting active layers and low-temperature processing capabilities for the development of flexible electronics [1–7]. Metal oxide semiconductor thin films have been deposited primarily using vacuum-based physical vapor deposition techniques. TFTs based on an oxide semiconductor exhibit electron mobilities that exceed 10 cm2 V−1 s−1 , even for a channel grown near room temperature [8, 9]. While these semiconductors are considered to be emerging materials for transparent TFTs and unconventional electronics applications, vacuum processing significantly increases the manufacturing costs and poses major obstacles that prevent the realization of modern, large-area, inexpensive electronics. In contrast, solution-processed TFTs can offer low-cost TFT array/circuits via roll-to-roll processes using a combination of coating and printing techniques [6, 7, 10–12]. In particular, inkjet printing is amenable to the selective deposition of semiconducting layers for large-area printed transistors [12]. Inkjet-printed organic semiconductor-based TFTs have been extensively discussed in the literature [13, 14]. However, inkjet-printed, metal-oxide–based TFTs have been only recently examined. In this chapter, several fundamental issues associated with the inkjet printing of metal oxide TFTs are discussed.

15.2 Materials for Metal Oxide Semiconductors

The materials science community has paid considerable attention to inorganic semiconductors that are solution processable to realize high-performance devices. Inkjet-based Micromanufacturing, First Edition. Edited by J. G. Korvink, P. J. Smith, and D.-Y. Shin. © 2012 Wiley-VCH Verlag GmbH & Co. KGaA. Published 2012 by Wiley-VCH Verlag GmbH & Co. KGaA.

238

15 Inkjet Printing of Metal Oxide Thin-Film Transistors

They include carbon nanotubes, chalcogenide semiconductors, liquid silicon, and metal oxides. Among these, ZnO-based materials have been extensively researched as solution-processed and high-performance semiconductor materials. They are promising as channel layers for TFTs that are stable in air and are employed in transparent flexible electronics [10–12]. Unlike conventional covalent semiconductor charge carrier transporters (i.e., silicon), ZnO-based materials contain posttransition metal cations that do not depend on the atomic arrangement in space (i.e., amorphous or crystalline). The neighboring large ns orbitals directly overlap in the constituent cations (here, n was the principal quantum number) [1, 15]. This confers several attractive characteristics on amorphous oxide semiconductors (AOSs) such as good transparency, a low processing temperature, and high mobility. If transparent AOSs are applied to flexible devices, low-temperature processing can be realized, allowing for the highly desirable all-solution processes. Oxide semiconductors investigated thus far include ZnO (zinc oxide), ZIO (zinc indium oxide), ZTO (zinc tin oxide), IZTO (indium zinc tin oxide), GZTO (gallium zinc tin oxide), GZO (gallium zinc oxide), AZO (aluminum zinc oxide), and IGZO (indium gallium zinc oxide). There are two types of functional solutions, that is, inks, from which ZnObased thin films are printed. The first type is a colloidal dispersion in which nanoparticles are suspended. The nanoparticulate-derived deposition is suitable for obtaining high-purity crystalline films at low temperatures, but it adversely affects the performance of a TFT because of grain boundary scattering and surface roughness as the particulate materials turn into granular films [16]. The other type is a chemical solution including a sol–gel, a metallo-organic solution, and a chelate process. In particular, sol–gel methods offer the greatest control over the nature of the solution precursor species. The sol–gel solution is appropriate for various coating and printing techniques such as spin coating and inkjet printing. Sol–gel synthesis contains chemical precursors, solvents, and stabilizing agents. Most recent works utilize metal chlorides, acetates, and acetylacetonates dissolved in an organic solvent such as acetonitrile and 2-methoxyethanol. Stabilizing agents such as formamide, acetic acid, or ethanolamine are added to improve the precursor solubility and solution stability. In 2007, Chang’s group first demonstrated solution processed metal–insulator– semiconductor field-effect transistors (MISFETs) by the inkjet printing of the ZIO thin films as channel layers. Later, they demonstrated inverted staggered IZTOTFTs with high field-effect mobility (μsat ∼ 30 cm2 V−1 s−1 ), which is 1 order of magnitude higher than the previously reported value for inkjet-printed oxidebased transistors. The precursor was prepared by metal chlorides in acetonitrile followed by annealing at 600 ◦ C for 1 h in air [17, 18]. Schneider et al. investigated a molecular zinc complex as a single-source precursor that contains an oximato ligand, which allows low-temperature conversion to nanocrystalline-dense ZnO. They demonstrated inkjet-printed ZnO lines with a height of 300 nm and a width of 500 μm on a polyethylene terephthalate, followed by thermal consolidation at 150 ◦ C. The printed lines showed good mechanical stability and excellent adherence to a polymer substrate [19]. Recently, Kim’s group reported that IGZO ink can be

15.3 Inkjet Printing Issues

prepared using metal acetate and nitrate in 2-methoxyethanol as a solvent with monoethanolamine as a stabilizer. The active channel region was inkjet printed on the substrate at a resolution of 300 dpi and a head frequency of 350 Hz, followed by annealing at 450 ◦ C for 1 h in air. The resulting inkjet-printed IGZO-TFTs with a conventional inverted staggered structure exhibited an on-to-off current ratio of ∼5 × 104 , a μsat value of 0.03 cm2 V−1 s−1 , a threshold voltage (Vth ) of 6.2 V, and a subthreshold slope value of 1.50 V dec−1 [20]. In contrast to organic solvent-based sol–gel solutions, Meyers et al. [21] employed an aqueous Zn precursor based on a metal ammine complex in an ammonium hydroxide solution. This approach relies on accelerating M–OH interactions within the aqueous precursor solution, resulting in low-temperature conversion to crystalline ZnO. The aqueous precursor approach can reduce the need for a high volume of organic ligand, thereby leading to smooth and dense films through dehydration and condensation reactions [22]. In a similar approach, Fleischhaker et al. [23] demonstrated a ZnO field-effect transistor on a polyethylene naphthalate (PEN foil using ZnO powder and aqueous ammonia. Our group [24] also synthesized an aqueous inorganic precursor by the direct dissolution of zinc hydroxide in an ammonium hydroxide solution. With the combined use of microwave annealing, solution-processed ZnO-TFTs at 140 ◦ C showed a device mobility value of ∼1.7 cm2 V−1 s−1 .

15.3 Inkjet Printing Issues 15.3.1 Ink Printability

A major challenge in applying inkjet techniques for the deposition of metal oxides is the formulation of suitable inks. The chemistry and formulation type of the ink not only determine the drop ejection characteristics and stability but also the quality of the printed patterns [7, 17, 18, 25]. The drop-on-demand (DOD) method is most commonly used for modern industrial applications. It deposits precise quantities of functional inks in the form of droplets on an arbitrary surface by applying a short pressure pulse through a nozzle, which is typically 20–50 μm in diameter [26, 27]. The jetting operation mechanism involves the generation of pressure waves in a fluid-filled pathway behind an orifice. At the end of the orifice, the fluid meniscus is maintained by surface tension. A piezoelectrically induced pressure wave can propagate against the surface tension of the fluid, forming a small droplet, which is ejected from the nozzle. Under suitable electrical conditions, the ejected fluid develops into a single droplet for quality inkjetting. However, appropriately functional ink materials are limited in availability. Inappropriate ink will lead to unstable inkjetting in which long-lived filaments form, connecting the ejected droplet to the nozzle [28]. The length and lifetime of the filament influence the positional accuracy and resolution of the printing as well as the printability of the

239

240

15 Inkjet Printing of Metal Oxide Thin-Film Transistors

inks. Fluid dynamics involved in inkjet printing have been studied [29–32], and an atomistic understanding of inkjet dynamics is emerging [33, 34]. The important physical parameters of printing fluids are viscosity, density, and surface tension. These fluid properties influence the drop formation mechanism and subsequent drop size at a given voltage. The inkjetting behavior is characterized in terms of the inverse of the Ohnesorge number (Z), which is related to the viscosity, surface tension, and density of the fluid. We have redefined the printable range as 4 ≤ Z−1 ≤ 14 by considering several characteristics of printing including the single-droplet formability, the minimum standoff distance, positional accuracy, and maximum allowable jetting frequency. Printing using a fluid with a low Z −1 value (