Organic Field-Effect Transistors

8080_C000.fm Page i Friday, April 6, 2007 10:20 AM Half Title Page Organic Field-Effect Transistors 8080_C000.fm Pa...

Author: Zhenan Bao | Jason Locklin

191 downloads 1619 Views 21MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

8080_C000.fm Page i Friday, April 6, 2007 10:20 AM

Half Title Page

Organic Field-Effect Transistors

8080_C000.fm Page ii Friday, April 6, 2007 10:20 AM

OPTICAL SCIENCE AND ENGINEERING

Founding Editor Brian J. Thompson University of Rochester Rochester, New York

1. Electron and Ion Microscopy and Microanalysis: Principles and Applications, Lawrence E. Murr 2. Acousto-Optic Signal Processing: Theory and Implementation, edited by Norman J. Berg and John N. Lee 3. Electro-Optic and Acousto-Optic Scanning and Deflection, Milton Gottlieb, Clive L. M. Ireland, and John Martin Ley 4. Single-Mode Fiber Optics: Principles and Applications, Luc B. Jeunhomme 5. Pulse Code Formats for Fiber Optical Data Communication: Basic Principles and Applications, David J. Morris 6. Optical Materials: An Introduction to Selection and Application, Solomon Musikant 7. Infrared Methods for Gaseous Measurements: Theory and Practice, edited by Joda Wormhoudt 8. Laser Beam Scanning: Opto-Mechanical Devices, Systems, and Data Storage Optics, edited by Gerald F. Marshall 9. Opto-Mechanical Systems Design, Paul R. Yoder, Jr. 10. Optical Fiber Splices and Connectors: Theory and Methods, Calvin M. Miller with Stephen C. Mettler and Ian A. White 11. Laser Spectroscopy and Its Applications, edited by Leon J. Radziemski, Richard W. Solarz, and Jeffrey A. Paisner 12. Infrared Optoelectronics: Devices and Applications, William Nunley and J. Scott Bechtel 13. Integrated Optical Circuits and Components: Design and Applications, edited by Lynn D. Hutcheson 14. Handbook of Molecular Lasers, edited by Peter K. Cheo 15. Handbook of Optical Fibers and Cables, Hiroshi Murata 16. Acousto-Optics, Adrian Korpel 17. Procedures in Applied Optics, John Strong 18. Handbook of Solid-State Lasers, edited by Peter K. Cheo 19. Optical Computing: Digital and Symbolic, edited by Raymond Arrathoon 20. Laser Applications in Physical Chemistry, edited by D. K. Evans 21. Laser-Induced Plasmas and Applications, edited by Leon J. Radziemski and David A. Cremers

8080_C000.fm Page iii Friday, April 6, 2007 10:20 AM

22. Infrared Technology Fundamentals, Irving J. Spiro and Monroe Schlessinger 23. Single-Mode Fiber Optics: Principles and Applications, Second Edition, Revised and Expanded, Luc B. Jeunhomme 24. Image Analysis Applications, edited by Rangachar Kasturi and Mohan M. Trivedi 25. Photoconductivity: Art, Science, and Technology, N. V. Joshi 26. Principles of Optical Circuit Engineering, Mark A. Mentzer 27. Lens Design, Milton Laikin 28. Optical Components, Systems, and Measurement Techniques, Rajpal S. Sirohi and M. P. Kothiyal 29. Electron and Ion Microscopy and Microanalysis: Principles and Applications, Second Edition, Revised and Expanded, Lawrence E. Murr 30. Handbook of Infrared Optical Materials, edited by Paul Klocek 31. Optical Scanning, edited by Gerald F. Marshall 32. Polymers for Lightwave and Integrated Optics: Technology and Applications, edited by Lawrence A. Hornak 33. Electro-Optical Displays, edited by Mohammad A. Karim 34. Mathematical Morphology in Image Processing, edited by Edward R. Dougherty 35. Opto-Mechanical Systems Design: Second Edition, Revised and Expanded, Paul R. Yoder, Jr. 36. Polarized Light: Fundamentals and Applications, Edward Collett 37. Rare Earth Doped Fiber Lasers and Amplifiers, edited by Michel J. F. Digonnet 38. Speckle Metrology, edited by Rajpal S. Sirohi 39. Organic Photoreceptors for Imaging Systems, Paul M. Borsenberger and David S. Weiss 40. Photonic Switching and Interconnects, edited by Abdellatif Marrakchi 41. Design and Fabrication of Acousto-Optic Devices, edited by Akis P. Goutzoulis and Dennis R. Pape 42. Digital Image Processing Methods, edited by Edward R. Dougherty 43. Visual Science and Engineering: Models and Applications, edited by D. H. Kelly 44. Handbook of Lens Design, Daniel Malacara and Zacarias Malacara 45. Photonic Devices and Systems, edited by Robert G. Hunsberger 46. Infrared Technology Fundamentals: Second Edition, Revised and Expanded, edited by Monroe Schlessinger 47. Spatial Light Modulator Technology: Materials, Devices, and Applications, edited by Uzi Efron 48. Lens Design: Second Edition, Revised and Expanded, Milton Laikin 49. Thin Films for Optical Systems, edited by Francoise R. Flory 50. Tunable Laser Applications, edited by F. J. Duarte 51. Acousto-Optic Signal Processing: Theory and Implementation, Second Edition, edited by Norman J. Berg and John M. Pellegrino

8080_C000.fm Page iv Friday, April 6, 2007 10:20 AM

52. Handbook of Nonlinear Optics, Richard L. Sutherland 53. Handbook of Optical Fibers and Cables: Second Edition, Hiroshi Murata 54. Optical Storage and Retrieval: Memory, Neural Networks, and Fractals, edited by Francis T. S. Yu and Suganda Jutamulia 55. Devices for Optoelectronics, Wallace B. Leigh 56. Practical Design and Production of Optical Thin Films, Ronald R. Willey 57. Acousto-Optics: Second Edition, Adrian Korpel 58. Diffraction Gratings and Applications, Erwin G. Loewen and Evgeny Popov 59. Organic Photoreceptors for Xerography, Paul M. Borsenberger and David S. Weiss 60. Characterization Techniques and Tabulations for Organic Nonlinear Optical Materials, edited by Mark G. Kuzyk and Carl W. Dirk 61. Interferogram Analysis for Optical Testing, Daniel Malacara, Manuel Servin, and Zacarias Malacara 62. Computational Modeling of Vision: The Role of Combination, William R. Uttal, Ramakrishna Kakarala, Spiram Dayanand, Thomas Shepherd, Jagadeesh Kalki, Charles F. Lunskis, Jr., and Ning Liu 63. Microoptics Technology: Fabrication and Applications of Lens Arrays and Devices, Nicholas Borrelli 64. Visual Information Representation, Communication, and Image Processing, edited by Chang Wen Chen and Ya-Qin Zhang 65. Optical Methods of Measurement, Rajpal S. Sirohi and F. S. Chau 66. Integrated Optical Circuits and Components: Design and Applications, edited by Edmond J. Murphy 67. Adaptive Optics Engineering Handbook, edited by Robert K. Tyson 68. Entropy and Information Optics, Francis T. S. Yu 69. Computational Methods for Electromagnetic and Optical Systems, John M. Jarem and Partha P. Banerjee 70. Laser Beam Shaping, Fred M. Dickey and Scott C. Holswade 71. Rare-Earth-Doped Fiber Lasers and Amplifiers: Second Edition, Revised and Expanded, edited by Michel J. F. Digonnet 72. Lens Design: Third Edition, Revised and Expanded, Milton Laikin 73. Handbook of Optical Engineering, edited by Daniel Malacara and Brian J. Thompson 74. Handbook of Imaging Materials: Second Edition, Revised and Expanded, edited by Arthur S. Diamond and David S. Weiss 75. Handbook of Image Quality: Characterization and Prediction, Brian W. Keelan 76. Fiber Optic Sensors, edited by Francis T. S. Yu and Shizhuo Yin 77. Optical Switching/Networking and Computing for Multimedia Systems, edited by Mohsen Guizani and Abdella Battou 78. Image Recognition and Classification: Algorithms, Systems, and Applications, edited by Bahram Javidi

8080_C000.fm Page v Friday, April 6, 2007 10:20 AM

79. Practical Design and Production of Optical Thin Films: Second Edition, Revised and Expanded, Ronald R. Willey 80. Ultrafast Lasers: Technology and Applications, edited by Martin E. Fermann, Almantas Galvanauskas, and Gregg Sucha 81. Light Propagation in Periodic Media: Differential Theory and Design, Michel Nevière and Evgeny Popov 82. Handbook of Nonlinear Optics, Second Edition, Revised and Expanded, Richard L. Sutherland 83. Polarized Light: Second Edition, Revised and Expanded, Dennis Goldstein 84. Optical Remote Sensing: Science and Technology, Walter Egan 85. Handbook of Optical Design: Second Edition, Daniel Malacara and Zacarias Malacara 86. Nonlinear Optics: Theory, Numerical Modeling, and Applications, Partha P. Banerjee 87. Semiconductor and Metal Nanocrystals: Synthesis and Electronic and Optical Properties, edited by Victor I. Klimov 88. High-Performance Backbone Network Technology, edited by Naoaki Yamanaka 89. Semiconductor Laser Fundamentals, Toshiaki Suhara 90. Handbook of Optical and Laser Scanning, edited by Gerald F. Marshall 91. Organic Light-Emitting Diodes: Principles, Characteristics, and Processes, Jan Kalinowski 92. Micro-Optomechatronics, Hiroshi Hosaka, Yoshitada Katagiri, Terunao Hirota, and Kiyoshi Itao 93. Microoptics Technology: Second Edition, Nicholas F. Borrelli 94. Organic Electroluminescence, edited by Zakya Kafafi 95. Engineering Thin Films and Nanostructures with Ion Beams, Emile Knystautas 96. Interferogram Analysis for Optical Testing, Second Edition, Daniel Malacara, Manuel Sercin, and Zacarias Malacara 97. Laser Remote Sensing, edited by Takashi Fujii and Tetsuo Fukuchi 98. Passive Micro-Optical Alignment Methods, edited by Robert A. Boudreau and Sharon M. Boudreau 99. Organic Photovoltaics: Mechanism, Materials, and Devices, edited by Sam-Shajing Sun and Niyazi Serdar Saracftci 100. Handbook of Optical Interconnects, edited by Shigeru Kawai 101. GMPLS Technologies: Broadband Backbone Networks and Systems, Naoaki Yamanaka, Kohei Shiomoto, and Eiji Oki 102. Laser Beam Shaping Applications, edited by Fred M. Dickey, Scott C. Holswade and David L. Shealy 103. Electromagnetic Theory and Applications for Photonic Crystals, Kiyotoshi Yasumoto 104. Physics of Optoelectronics, Michael A. Parker 105. Opto-Mechanical Systems Design: Third Edition, Paul R. Yoder, Jr. 106. Color Desktop Printer Technology, edited by Mitchell Rosen and Noboru Ohta 107. Laser Safety Management, Ken Barat

8080_C000.fm Page vi Friday, April 6, 2007 10:20 AM

108. Optics in Magnetic Multilayers and Nanostructures, Sˇtefan Viˇsˇnovsky’ 109. Optical Inspection of Microsystems, edited by Wolfgang Osten 110. Applied Microphotonics, edited by Wes R. Jamroz, Roman Kruzelecky, and Emile I. Haddad 111. Organic Light-Emitting Materials and Devices, edited by Zhigang Li and Hong Meng 112. Silicon Nanoelectronics, edited by Shunri Oda and David Ferry 113. Image Sensors and Signal Processor for Digital Still Cameras, Junichi Nakamura 114. Encyclopedic Handbook of Integrated Circuits, edited by Kenichi Iga and Yasuo Kokubun 115. Quantum Communications and Cryptography, edited by Alexander V. Sergienko 116. Optical Code Division Multiple Access: Fundamentals and Applications, edited by Paul R. Prucnal 117. Polymer Fiber Optics: Materials, Physics, and Applications, Mark G. Kuzyk 118. Smart Biosensor Technology, edited by George K. Knopf and Amarjeet S. Bassi 119. Solid-State Lasers and Applications, edited by Alphan Sennaroglu 120. Optical Waveguides: From Theory to Applied Technologies, edited by Maria L. Calvo and Vasudevan Lakshiminarayanan 121. Gas Lasers, edited by Masamori Endo and Robert F. Walker 122. Lens Design, Fourth Edition, Milton Laikin 123. Photonics: Principles and Practices, Abdul Al-Azzawi 124. Microwave Photonics, edited by Chi H. Lee 125. Physical Properties and Data of Optical Materials, Moriaki Wakaki, Keiei Kudo, and Takehisa Shibuya 126. Microlithography: Science and Technology, Second Edition, edited by Kazuaki Suzuki and Bruce W. Smith 127. Coarse Wavelength Division Multiplexing: Technologies and Applications, edited by Hans Joerg Thiele and Marcus Nebeling 128. Organic Field-Effect Transistors, Zhenan Bao and Jason Locklin

8080_C000.fm Page vii Friday, April 6, 2007 10:20 AM

Organic Field-Effect Transistors

Title Page

Zhenan Bao Jason Locklin

Boca Raton London New York

CRC Press is an imprint of the Taylor & Francis Group, an informa business

8080_C000.fm Page viii Friday, April 6, 2007 10:20 AM

CRC Press Taylor & Francis Group 6000 Broken Sound Parkway NW, Suite 300 Boca Raton, FL 33487-2742 © 2007 by Taylor & Francis Group, LLC CRC Press is an imprint of Taylor & Francis Group, an Informa business No claim to original U.S. Government works Printed in the United States of America on acid-free paper 10 9 8 7 6 5 4 3 2 1 International Standard Book Number-10: 0-8493-8080-4 (Hardcover) International Standard Book Number-13: 978-0-8493-8080-8 (Hardcover) This book contains information obtained from authentic and highly regarded sources. Reprinted material is quoted with permission, and sources are indicated. A wide variety of references are listed. Reasonable eﬀorts have been made to publish reliable data and information, but the author and the publisher cannot assume responsibility for the validity of all materials or for the consequences of their use. No part of this book may be reprinted, reproduced, transmitted, or utilized in any form by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying, microﬁlming, and recording, or in any information storage or retrieval system, without written permission from the publishers. For permission to photocopy or use material electronically from this work, please access www. copyright.com (http://www.copyright.com/) or contact the Copyright Clearance Center, Inc. (CCC) 222 Rosewood Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-proﬁt organization that provides licenses and registration for a variety of users. For organizations that have been granted a photocopy license by the CCC, a separate system of payment has been arranged. Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identiﬁcation and explanation without intent to infringe. Library of Congress Cataloging-in-Publication Data Organic ﬁeld-eﬀect transistors / edited by Zhenan Bao and Jason Locklin p. cm. -- (Optical science and engineering series) Includes bibliographical references and index. ISBN-13: 978-0-8493-8080-8 (alk. paper) ISBN-10: 0-8493-8080-4 (alk. paper) 1. Organic ﬁeld-eﬀect transistors. I. Bao, Zhenan. II. Locklin, Jason. III. Series. TK7871.95.O734 2007 537.6’22--dc22 Visit the Taylor & Francis Web site at http://www.taylorandfrancis.com and the CRC Press Web site at http://www.crcpress.com

2006038167

8080_C000.fm Page ix Friday, April 6, 2007 10:20 AM

Contents Section 1.1 Theoretical Aspects of Charge Transport in Organic Semiconductors: A Molecular Perspective ..........................................1 Demetrio A. da Silva Filho, Yoann Olivier, Veaceslav Coropceanu, Jean-Luc Brédas, and Jérôme Cornil Section 2.1 Charge Carrier Transport in Single-Crystal Organic Field-Effect Transistors ..........................................................................................27 Vitaly Podzorov Section 2.2 Charge Transport in Oligomers..........................................................73 Gilles Horowitz Section 2.3 Charge Transport Physics of Solution-Processed Organic Field-Effect Transistors....................................................................103 Henning Sirringhaus Section 2.4 Contact Effects in Organic Field-Effect Transistors .......................139 Matthew J. Panzer and C. Daniel Frisbie Section 3.1 Design, Synthesis, and Transistor Performance of Organic Semiconductors ................................................................................159 Abhijit Basu Mallik, Jason Locklin, Stefan C. B. Mannsfeld, Colin Reese, Mark E. Roberts, Michelle L. Senatore, Hong Zi, and Zhenan Bao Section 3.2 Dielectric Materials: Selection and Design .....................................229 Ashok Maliakal Section 4.1 Grazing Incidence X-Ray Diffraction (GIXD) ...............................253 Tae Joo Shin and Hoichang Yang

8080_C000.fm Page x Friday, April 6, 2007 10:20 AM

Section 4.2 Near-Edge X-Ray Absorption Fine Structure (NEXAFS) Spectroscopy ....................................................................................277 Dean M. DeLongchamp, Eric K. Lin, and Daniel A. Fischer Section 4.3 Scanning Probe Techniques .............................................................301 Hoichang Yang Section 5.1 Vacuum Evaporated Thin Films ......................................................341 Alex C. Mayer, Jack M. Blakely, and George G. Malliaras Section 5.2 Solution Deposition of Polymers.....................................................371 Hoichang Yang Section 5.3 Solution Deposition of Oligomers ...................................................403 Howard E. Katz and Chad Landis Section 5.4 Inkjet Printed Organic Thin Film Transistors .................................419 Ana Claudia Arias Section 5.5 Soft Lithography for Fabricating Organic Thin-Film Transistors ........................................................................................433 Kimberly C. Dickey, Kwang Seok Lee, and Yueh-Lin Loo Section 6.1 Radio Frequency Identification Tags ...............................................489 Vivek Subramanian Section 6.2 Organic Transistor Chemical Sensors..............................................507 Luisa Torsi, M. C. Tanese, Brian Crone, Liang Wang, and Ananth Dodabalapur Section 6.3 Flexible, Large-Area e-Skins ...........................................................529 Takao Someya, Takayasu Sakurai, and Tsuyoshi Sekitani Section 6.4 Organic Thin-Film Transistors for Flat-Panel Displays..................551 Michael G. Kane Index ......................................................................................................................595

8080_C000.fm Page xi Friday, April 6, 2007 10:20 AM

Preface Since the early 1990s, remarkable progress has been made in the development of organic thin film transistors (OTFTs). The performance of the best organic materials now rivals that of the amorphous silicon TFTs commonly used as the pixel-switching elements in active matrix flat-panel displays. OTFTs also are of great interest from a technological standpoint, since a major advantage of organic materials is that they can be deposited onto substrates at low temperatures, thus providing for compatibility with plastic substrates. The purpose of this book is to provide a comprehensive survey of present theory, synthetic methodology, materials characterization, and current applications of organic field-effect transistors. The book is divided into six sections dealing with different aspects of organic transistors. Section 1 provides a theoretical description of charge transport in organic semiconductors at the molecular level. Understanding the influence of molecular parameters on charge transport is extremely important in designing new organic semiconductors. The conduction mechanism for organic materials is different from that found in traditional inorganic semiconductors. The molecules are held together by weak intermolecular forces and, except for a few single-crystal devices, the carriers usually move through the material by hopping instead of band transport. Section 2 is divided into four subsections describing the current understanding of charge transport in single-crystal devices, small molecules and oligomers, conjugated polymer devices, and charge injection issues in organic transistors. Section 3 begins with a detailed description of the synthetic methodologies used for organic semiconductors. An in-depth look at molecular design rules and rationales is given for each material system to tailor device characteristics of p- and n-type organic semiconductors. A survey of all reported molecules and correlations between their structure and transistor performance are presented. The second part of the section deals with the design of materials used as dielectric gate insulators in organic thin film transistor devices. Several requirements of dielectrics in OTFTs, including processability, high capacitance, high dielectric strength, high on/off ratio, and low hysteresis, are discussed. Different processing methods for dielectrics are surveyed. The performance of the best OTFTs, such as those fabricated with pentacene, depends not only on the molecular electronic properties of the organic semiconductor but also on their microstructure. Section 4 provides an overview of various characterization techniques used to probe interfacial ordering, microstructure, molecular packing, and orientation crucial to device performance. The section focuses on characterizing the orientation of organic semiconductors at interfaces, primarily through the use of grazing incidence x-ray diffraction (GIXD) and near-edge x-ray absorption fine structure (NEXAFS). These techniques provide valuable electronic and structural information about atoms, molecules, and local chemical functional-

8080_C000.fm Page xii Friday, April 6, 2007 10:20 AM

ities. The section also includes a discussion of scanning probe techniques used to probe local morphology, microstructure, and electrical properties of thin film devices. Unless they are carefully designed, conjugated oligomers are typically insoluble and thin films can only be deposited by vacuum evaporation. Imparting solubility by the introduction of chemical functionality disrupts the molecular π-system’s natural tendency to pack, and most semiconductor molecules that rely heavily on π-orbital overlap to achieve high carrier mobility suffer as a result of functionalization. Organic polymers are typically more soluble, since their chemical structure can be modified in a regular manner with organic side chains and has better filmforming ability than that of small molecules. Section 5 is divided into five parts describing the different processing techniques and associated organic semiconductor thin film growth mechanisms for molecules deposited by vacuum and solution. The last part includes a section on different printing techniques and nonconventional patterning methods by soft lithography. Section 6 provides some current technological examples that utilize OTFTs in their operation. This section includes specific examples of radio-frequency ID tags, chemical and pressure sensors, flexible scanners, display technology, and circuit design. Zhenan Bao Jason Locklin

8080_C000.fm Page xiii Friday, April 6, 2007 10:20 AM

Acknowledgments We would like to thank Stefan C. B. Mannsfeld for his advice and critical reading of the manuscript. Also, Jordi Mata Fink, Ajay Virkar, Anna Reichardt, and Fei Qui are acknowledged for their help in formatting and editing.

8080_C000.fm Page xiv Friday, April 6, 2007 10:20 AM

8080_C000.fm Page xv Friday, April 6, 2007 10:20 AM

The Editors Zhenan Bao received her Ph.D. in chemistry from the University of Chicago in 1995. She is an associate professor in the Department of Chemical Engineering at Stanford University. Jason Locklin received his B.S. from Millsaps College in 1999 and Ph.D. from the University of Houston in 2004. He is currently an assistant professor in the Department of Chemistry and a member of the faculty of engineering at the University of Georgia.

8080_C000.fm Page xvi Friday, April 6, 2007 10:20 AM

8080_C000.fm Page xvii Friday, April 6, 2007 10:20 AM

Contributors Ana Claudia Arias Palo Alto Research Center Palo Alto, California Zhenan Bao Stanford University Stanford, California Jack M. Blakely Cornell University Ithaca, New York Jean-Luc Brédas Georgia Institute of Technology Atlanta, Georgia University of Mons-Hainaut Mons, Belgium Jérôme Cornil Georgia Institute of Technology Atlanta, Georgia University of Mons-Hainaut Mons, Belgium Veaceslav Coropceanu Georgia Institute of Technology Atlanta, Georgia

Dean M. DeLongchamp National Institute of Standards and Technology Gaithersburg, Maryland Kimberly C. Dickey University of Texas at Austin Austin, Texas Ananth Dodabalapur University of Texas at Austin Austin, Texas Daniel A. Fischer National Institute of Standards and Technology Gaithersburg, Maryland C. Daniel Frisbie University of Minnesota-Twin Cities Minneapolis, Minnesota Gilles Horowitz ITODYS University Denis-Diderot Paris, France Michael G. Kane Sarnoff Corporation Princeton, New Jersey

Brian Crone Los Alamos National Laboratory Los Alamos, New Mexico

Howard E. Katz Johns Hopkins University Baltimore, Maryland

Demetrio A. da Silva Filho Georgia Institute of Technology Atlanta, Georgia

Chad Landis Johns Hopkins University Baltimore, Maryland

8080_C000.fm Page xviii Friday, April 6, 2007 10:20 AM

Kwang Seok Lee University of Texas at Austin Austin, Texas

Vitaly Podzorov Rutgers University Piscataway, New Jersey

Eric K. Lin National Institute of Standards and Technology Gaithersburg, Maryland

Colin Reese Stanford University Stanford, California

Jason Locklin University of Georgia Athens, Georgia Yueh-Lin Loo University of Texas at Austin Austin, Texas Ashok Maliakal Bell Laboratories Murray Hill, New Jersey George G. Malliaras Cornell University Ithaca, New York Abhijit Basu Mallik Stanford University Stanford, California Stefan C. B. Mannsfeld Stanford University Stanford, California

Mark E. Roberts Stanford University Stanford, California Takayasu Sakurai University of Tokyo Tokyo, Japan Tsuyoshi Sekitani University of Tokyo Tokyo, Japan Michelle L. Senatore Stanford University Stanford, California Tae Joo Shin Pohang Accelerator Laboratory Pohang, Korea Henning Sirringhaus Cavendish Laboratory University of Cambridge Plastic Logic Ltd. Cambridge, United Kingdom

Alex C. Mayer Cornell University Ithaca, New York

Takao Someya University of Tokyo Tokyo, Japan

Yoann Olivier University of Mons-Hainaut Mons, Belgium

Vivek Subramanian University of California, Berkeley Berkeley, California

Matthew J. Panzer University of Minnesota-Twin Cities Minneapolis, Minnesota

M. C. Tanese Università degli Studi di Bari Bari, Italy

8080_C000.fm Page xix Friday, April 6, 2007 10:20 AM

Luisa Torsi Università degli Studi di Bari Bari, Italy

Hoichang Yang Rensselaer Polytechnic Institute Troy, New York

Liang Wang University of Texas at Austin Austin, Texas

Hong Zi Stanford University Stanford, California

8080_C000.fm Page xx Friday, April 6, 2007 10:20 AM

8080_book.fm Page 1 Tuesday, April 3, 2007 1:47 PM

1.1

Theoretical Aspects of Charge Transport in Organic Semiconductors: A Molecular Perspective

Demetrio A. da Silva Filho, Yoann Olivier, Veaceslav Coropceanu, Jean-Luc Brédas, and Jérôme Cornil CONTENTS 1.1.1 Introduction....................................................................................................1 1.1.2 A Primer on Electron-Transfer Theory .........................................................3 1.1.3 Electron-Vibration Coupling and Reorganization Energy ............................5 1.1.3.1 Intramolecular Reorganization Energy ...........................................5 1.1.3.2 Intramolecular Reorganization Energy of Oligoacenes..................8 1.1.4 Electronic Coupling .....................................................................................10 1.1.4.1 Inﬂuence of Intermolecular Separation ........................................12 1.1.4.2 Inﬂuence of Long- or Short-Axis Displacements ........................13 1.1.5 From Molecular Parameters to Carrier Mobilities ......................................16 1.1.5.1 Inﬂuence of the Electric Field ......................................................17 1.1.5.2 Inﬂuence of the Reorganization Energy .......................................18 1.1.5.3 Inﬂuence of Intermolecular Distance............................................18 1.1.5.4 Inﬂuence of Molecular Translations .............................................20 1.1.5.5 Introduction of a Gaussian Disorder.............................................21 1.1.6 Concluding Remarks....................................................................................22 References................................................................................................................22

1.1.1 INTRODUCTION The development of the ﬁeld of organic electronics has beneﬁted from the unique set of characteristics offered by π-conjugated oligomers and polymers. These

1

8080_book.fm Page 2 Tuesday, April 3, 2007 1:47 PM

2

Organic Field-Effect Transistors

materials combine the electrical properties of semiconductors with the properties typical of plastics: low cost, versatility of chemical synthesis, ease of processing, and ﬂexibility. In organic ﬁeld-effect transistors, the key steps of operation involve charge injection and formation of a conducting channel within the organic semiconductor due to application of a gate voltage; upon application of a drain voltage, the charges migrate across the organic layer and are collected at the drain electrode. Charge injection and collection processes and, in most instances, charge transport actually correspond to redox (electron-transfer) reactions. Much success in gaining a better understanding of charge-transport phenomena in organic materials has come recently from extending the theory of electron-transfer reactions, originally formulated by Marcus for the description of redox reactions in solution, to organic semiconductors [1–5]. The charge-transport properties in conjugated materials critically depend on the packing of the chains and degree of order in the solid state [6] as well as on the density of impurities and structural defects [7]. As a result, the measured mobility values can largely vary as a function of sample quality [8]. Overall, the transport mechanism results from a balance between the energy gained by electron delocalization in an electronic band and the energy gained by geometric relaxation and polarization around a charge on an oligomer or polymer segment to form a polaron [9]. In highly puriﬁed molecular single crystals, transport at low temperature can be described within a band picture, as shown by Karl and coworkers [10]. As a general rule of thumb, (effective) bandwidths of at least 0.1 eV are needed to stabilize a band regime [9]. In that case, the positive or negative charge carriers are fully delocalized and their mobilities are a function of the width of the valence or conduction band, respectively (i.e., of the extent of electronic coupling between oligomer or polymer chains). When temperature increases, the mobilities progressively decrease as a result of scattering processes due mainly to lattice phonons, as is the case in metallic conductors. Transport can then be described on the basis of effective bandwidths that are smaller than the bandwidths obtained for a rigid lattice. At elevated temperatures, localization steps in and transport operates via a thermally assisted polaron hopping regime where charge carriers jump between adjacent molecules or chains, as described, for instance, by Conwell and coworkers [11]. The hopping regime generally applies in the presence of signiﬁcant static disorder, dynamic ﬂuctuations [12], and/or impurities; this transport mechanism is thus expected to be operative in most organic ﬁeld-effect transistors. At the microscopic level, polaron hopping can be viewed as a self-exchange electron-transfer reaction where a charge hops from an ionized site to an adjacent neutral site. In that context, the carrier mobilities are a direct function of the self-exchange electron-transfer reaction rates. In this chapter, we focus on the hopping regime and start with a primer on electron-transfer theory in Section 1.1.2. This section will underline the three major parameters that enter the expression of the electron-transfer rate: reorganization energy, electronic coupling, and driving force. We then discuss some examples of the impact of chemical structure and packing mode on these parameters. Section

8080_book.fm Page 3 Tuesday, April 3, 2007 1:47 PM

Theoretical Aspects of Charge Transport in Organic Semiconductors

3

1.1.3 deals with reorganization energy, and Section 1.1.4 is devoted to electronic coupling. In Section 1.1.5, the role of the driving force (due to the application of an external electric ﬁeld) is incorporated. This section provides an illustration of how the information on electron-transfer rates gathered at the intramolecular and intermolecular levels can translate into charge carrier mobilities at the macroscopic level.

1.1.2 A PRIMER ON ELECTRON-TRANSFER THEORY Electron-transfer processes, as well as energy-transfer processes, can be viewed as special cases of the nonradiative decay of an electronic state. In the framework of perturbation theory [1,2], the probability for a transition from a discrete initial state ψi (corresponding to the reactants) to a discrete ﬁnal state ψf (corresponding to the products of the reaction) writes under application of a perturbation V to ﬁrst order:

Pif =

1 < ψi V ψ f > 2

2

⎡ sin(ω fi t / 2 ) ⎤ ⎥ ⎢ ⎢⎣ ω fi / 2 ⎥⎦

2

(1.1.1)

where t denotes time, ωfi the transition energy between the electronic states i and f, and <ψi|V|ψf> is the corresponding electronic coupling matrix element. To account for a continuous distribution of ﬁnal (vibrationally coupled) electronic states, Equation 1.1.1 can be recast by introducing the density of ﬁnal states ρ(Ef) and summing over all probability densities. Assuming that the function |<ψi|V|ψf>|2 ρ(Ef) varies slowly with energy, the transition probability per unit time (or transition rate) adopts, in the long-time limit, the simple and widely exploited Fermi’s golden rule form:

kif =

2 2π < ψ i V ψ f > ρ( E f )

(1.1.2)

The transition mechanism involves vibrational motions driving the reaction coordinates from reactants to products. The expression for the rate obtained within the Frank–Condon approximation factorizes into an electronic and a vibrational contribution as:

kif =

2π Vif

2

( FCWD )

(1.1.3)

Here, Vif = <ψi|V|ψf> is the electronic coupling matrix element and (FCWD) denotes the Franck–Condon weighted density of states. In the high-temperature regime — that is, when assuming that all vibrational modes are classical ( ω i << kB T ), the FCWD obeys a standard Arrhenius type of equation:

8080_book.fm Page 4 Tuesday, April 3, 2007 1:47 PM

4

Organic Field-Effect Transistors

1

( FCWD ) =

4πλkB T

exp[−( ΔG 0 + λ)2 / 4λkB T ]

(1.1.4)

and the rate takes its semiclassical Marcus theory expression [3–5]:

kif =

2π Vif

1

2

4πλkB T

exp[−( ΔG 0 + λ)2 / 4λkB T ]

(1.1.5)

where λ denotes the reorganization energy induced by the electron or energy transfer and ΔG° is the variation of the Gibbs free energy during the reaction. When the reorganization energy λ is cast into contributions of both classical modes for the surrounding medium [(λ0); ω h << kB T ] and intramolecular highfrequency vibrational modes [(λi); ω i >> kB T ], the rate kif becomes in the context of the Bixon and Jortner model (for details, see the review in Jortner et al. [2]):

kif =

2π Vif

2

1 4πλ 0 kB T

∞

∑ exp(− S ) Sn! exp[− (ΔΔG n i

i

n =0

0

+ λ 0 + n ω i )2 ] (1.1.6) 4 λ 0 kB T

Here, a single effective quantum mode, ωi, is assumed to contribute to λi. The Huang–Rhys factor, Si = λ i ω h , is a measure of the electron-vibrational coupling interaction. Equations 1.1.5 and 1.1.6 can be further generalized to account for the inﬂuence of external electric ﬁeld and disorder. Thus, in the presence of an electric ﬁeld (and when neglecting the entropy contributions), the free energy associated with the hop of an electron from site i to site j is given by: ΔGij0 = E j − Ei − e( F ⋅ dij )

(1.1.7)

Here, Ei and Ej are the site energies of molecules i and j; F is the vector associated with the electric ﬁeld; and dij is the vector connecting the i to site j. Energy disorder in the form of diagonal disorder [13] can be simply introduced by deﬁning a distribution (usually taken as Gaussian) for the site energies. Off-diagonal disorder, which arises from ﬂuctuations in the electronic coupling due to the positional and orientational disorder of the molecules or chains, can also be incorporated. For instance, in the case of through-space interaction, the electronic coupling decays exponentially (vide infra) with intermolecular separation dij: Vij = V(0 )ij ⋅ exp(− γ ij dij )

(1.1.8)

8080_book.fm Page 5 Tuesday, April 3, 2007 1:47 PM

Theoretical Aspects of Charge Transport in Organic Semiconductors

5

Positional disorder can then be modeled by assuming a distribution of dij (or, alternatively, γij) values. We note that Equations 1.1.5 and 1.1.6 were obtained in the framework of perturbation theory and are therefore strictly applicable only in the limiting case of weak electronic coupling (nonadiabatic electron-transfer regime) [14,15]. In a more general case (any Vij < 0.5λ), the thermal activation barrier also depends on electronic coupling while the prefactor is a function of the attempting nuclear frequency νn (frequency for nuclear motion along the reaction coordinate) and the electronic frequency ν e that is equal to the prefactor in Equations 1.1.5 and 1.1.6 (i.e., νe = 2 π Vij

2

4π 2 λkB T ). The limiting case when the electronic coupling is

dominant (i.e., the condition νe >> νn is satisﬁed) is referred to as the adiabatic electron-transfer regime. In this case, the hopping rate for a self-exchange reaction (ΔG° = 0) is given by [14,15]: kif = ν n exp[−(λ − 2 Vif )2 / 4λkB T ]

(1.1.9)

This equation can also be further modiﬁed to accommodate other effects, such as site energy ﬂuctuations, external ﬁelds, or disorder. In this chapter, however, we will limit ourselves to the nonadiabatic electron transfer regime. In the next two sections, we review some recent work that addresses, at the molecular level, the nature of the main parameters that govern electron-transfer processes in π-conjugated oligomers and polymers. This molecular approach contrasts with many models developed earlier for organic materials (where these processes were described on a phenomenological basis) and from a macroscopic perspective, thereby masking the actual chemical structures of the systems behind effective parameters.

1.1.3 ELECTRON-VIBRATION COUPLING AND REORGANIZATION ENERGY 1.1.3.1 INTRAMOLECULAR REORGANIZATION ENERGY As emphasized in Section 1.1.2, the reorganization energy is one of the key quantities that control the rates for electron (or energy) transfer. From the rate expression given in Equation 1.1.6, it is clear that the lower the reorganization energy, the higher the rate. The reorganization energy is usually expressed as the sum of inner and outer contributions. The inner (intramolecular) reorganization energy arises from the change in equilibrium geometry of the donor (D) and acceptor (A) sites consecutive to the gain or loss of electronic charge upon electron transfer (ET). The outer reorganization energy is due to the electronic and nuclear polarization/relaxation of the surrounding medium. It is important to bear in mind that, due to the weakness of the van der Waals interactions among organic molecules, the separation of the reorganization energy

8080_book.fm Page 6 Tuesday, April 3, 2007 1:47 PM

6

Organic Field-Effect Transistors

D

2

A

2

ii

(D2) λi

(A2) λi

Energy

ΔQA 1

i

(D1) λi

i

1

ii

QD

(A1) λi QA

FIGURE 1.1.1 Sketch of the potential energy surfaces (in the monomer coordinate representation) related to electron transfer, showing the vertical transitions, the normal mode displacement ΔQ, and the relaxation energies λi(1) and λi(2).

into inter- and intramolecular contributions remains largely valid even in the case of molecular crystals. We note that in most instances the outer contribution to the reorganization energy is expected to be of the same order of magnitude as the inner part (see Section 1.1.5); it is also expected to be less sensitive to the chemical structure of the constituents than the inner contribution is. The formalisms used to estimate the outer reorganization energy have been mainly developed to describe electron-transfer processes in solution and apply to isotropic media [16]; thus, it is desirable to extend these standard models to account for the anisotropy in the solid state. In this section, we focus on the intramolecular reorganization energy and its description in terms of vibrational modes. In order to illustrate the physical meaning of the intramolecular reorganization energy, we represent in Figure 1.1.1 the potential energy surfaces (PES) of the donor and acceptor involved in an intermolecular ET reaction of the type D + A+ → D+ + A. In the ﬁgure, the electronic states D1 or A1 and D2 or A2 correspond to the neutral and cation states of the donor or acceptor, respectively. The ET process can be formally divided into two steps: Step 1 is the simultaneous oxidation of D and reduction of A+ at frozen reactant geometries. (In Figure 1.1.1, this step corresponds to a vertical transition from the minimum of the D1 surface to D2 and a similar A2 to A1 transition.) Step 2 corresponds to the relaxation of the product nuclear geometries. As seen from Figure 1.1.1, the overall intramolecular reorganization energy upon electron transfer consists of two terms [1–5,17–21]: λ i = λ i( A1) + λ i( D 2 )

(1.1.10)

λ (i A1) = E ( A1) ( A + ) − E ( A1) ( A)

(1.1.11)

8080_book.fm Page 7 Tuesday, April 3, 2007 1:47 PM

Theoretical Aspects of Charge Transport in Organic Semiconductors

λ (i D 2 ) = E ( D 2 ) ( D ) − E ( D 2 ) ( D + )

7

(1.1.12)

Here, E(A1)(A+) and E(A1)(A) are the energies of the neutral acceptor A at the cation geometry and optimal ground-state geometry, respectively; E(D2)(D) and E(D2)(D+) are accordingly the energies of the radical cation D+ at the neutral geometry and optimal cation geometry. The vertical transitions involved in Figure 1.1.1 and Equations 1.1.10–1.1.12 are consequences of the Franck–Condon principle that requires the nuclear conﬁgurations of the system immediately before and after electron transfer to coincide. However, it is important to note that in addition to the Franck–Condon principle, the principle of energy conservation should also be satisﬁed for electron transfer to occur [1–5]. In the case of optically driven electron transfer, the mismatch between the electronic vertical transitions (see the lines labeled “i” in Figure 1.1.1) is balanced by the absorption of light. In the case of thermal (dark) electron transfer (which is our main concern here), to satisfy both principles, thermal ﬂuctuations from the equilibrium nuclear conﬁgurations of the reactants are needed prior to electron transfer [3–5]. The contribution of each vibrational mode to λi can be obtained by expanding the potential energies of the neutral and cation states in a power series of the normal coordinates (denoted here as Q1 and Q2). In the harmonic approximation, the relaxation energy λi writes [1–5,17–21]: λi =

λj =

kj 2

∑ λ = ∑ ω S j

ΔQ 2j ,

j

j

S j = λ j / ω j

(1.1.13)

(1.1.14)

where the summations run over the vibrational modes ΔQj represents the displacement along normal mode j between the equilibrium positions of the two electronic states of interest kj and ωj are the corresponding force constants and vibrational frequencies Sj denotes the Huang–Rhys factor. The numerical procedure to obtain the reorganization energy consists of the following steps. First, the normal-mode coordinates and force constants of the electron donor and acceptor are determined. The standard rectilinear normal modes Q1(2) are obtained as a linear combination of Cartesian displacements [22]: Q1(2 ) j =

∑L

1( 2 ) kj

k

(q1(2 ) k − q1((02)) k )

(1.1.15)

8080_book.fm Page 8 Tuesday, April 3, 2007 1:47 PM

8

Organic Field-Effect Transistors

The matrix L1(2) connects the 3n-6 (n is the number of atoms in the [nonlinear] molecule) normal coordinates with the set of 3n mass-weighted Cartesian coordinates q1(2); the vectors q1(0) and q2(0) correspond to the stationary points on the adiabatic potential surfaces of states 1 and 2, respectively. Then, the normal mode displacements ΔQ1(2) are obtained by projecting the displacements Δq = q1(0) – q2(0) onto the normal-mode vectors. Finally, substituting the calculated quantities into Equations 1.1.10, 1.1.13, and 1.1.14 provides the total relaxation energy.

1.1.3.2 INTRAMOLECULAR REORGANIZATION ENERGY OF OLIGOACENES We now describe how the numerical approaches discussed previously were applied to the case of oligoacenes containing from two to ﬁve rings: naphthalene, anthracene, tetracene, and pentacene [15,23–27]. These oligomers are of high current interest because of their high charge mobilities in the crystalline state, in particular tetracene, pentacene, and their derivatives such as rubrene [28,29]. We note that all calculations including geometry optimizations and normal-mode analyses were performed at the DFT (density functional theory) level with the hybrid B3LYP functional using the standard 6-31G** basis set. The bond-length modiﬁcations upon positive ionization show a consistent trend along the series. Naphthalene displays the largest geometry relaxations, with changes in C–C bond lengths on the order of 0.03 Å. This value is reduced to ca. 0.02, 0.015, and 0.01 Å in anthracene, tetracene, and pentacene, respectively. The geometry distortions, as well as the changes in atomic charge densities (Mulliken populations), are found to spread over the entire molecule. The theoretical estimates of the relaxation energies and total reorganization energies obtained from the normal mode analysis are in excellent agreement with the values computed directly from the adiabatic potential energy surfaces [30]. The calculated reorganization energies evolve from 187 meV in naphthalene to 137 meV in anthracene, 113 meV in tetracene, and 97 meV in pentacene [30]. These values are also in good agreement with the results of previous calculations by Kato and Yamabe [31–35] and Klimkans and Larsson [36]. Our results indicate that the main contribution to the relaxation energy comes from high-energy vibrations. This high-energy contribution is in fact divided over several vibrational modes with wave numbers in the range of 1200–1600 cm–1. The contribution to λi from low-energy vibrations is negligible in anthracene and tetracene and is very small in naphthalene and pentacene. We have also carried out the Frank–Condon simulation of the shape of the ﬁrst ionization peak of anthracene, tetracene, and pentacene using the DFT/B3LYP estimates of the frequencies and Huang–Rhys factors obtained from normal-mode calculations. The vibrational frequencies and Huang–Rhys factors calculated for pentacene along with the Frank–Condon simulation of the shape of the ﬁrst ionization peak are shown in Figure 1.1.2. In general, the positions and shapes of the peaks are very well reproduced. Similar results were obtained for the oligoacenes and their derivatives [24,26,27,37,38]. These results underline the importance of multimode effects to obtain a detailed understanding of the UPS band shapes in oligoacenes.

8080_book.fm Page 9 Tuesday, April 3, 2007 1:47 PM

Theoretical Aspects of Charge Transport in Organic Semiconductors

9

Intensity (arb. units)

1 0.8 0.6 0.4 0.2 0

6.4

6.6

6.8

7

Energy (eV) 0.125 0.1

S

0.075 0.05 0.025 0

0

250

500

750 Frequency

1000

1250

1500

1750

(cm−1)

FIGURE 1.1.2 (Top) DFT/B3LYP simulation (dashed lines) of the vibrational structure of the UPS ﬁrst ionization peak of pentacene (solid lines) and (bottom) computed Huang-Rhys factors as a function of the frequency. The normal modes of the cation species with the largest Huang-Rhys factors ((10 modes) have been used for the simulations. A scaling factor of 0.9613 has been applied to the computed frequencies. The transition intensities were convoluted with Lorentzian functions with full-width at half-maximum (FWHM) of 0.060 eV. The HOMO wavefunctions obtained at the DFT/B3LYP level is also illustrated.

Overall, the intramolecular reorganization energies in tetracene (0.11 eV) and pentacene (0.10 eV) rank among the smallest λi values that have been calculated or measured for molecules. A smaller value of λi (0.045 eV) has been found only in the case of phthalocyanine [39]. The tetracene and pentacene values are about three times as small as in TPD (0.29 eV), which is a hole-transport material widely used in organic molecular devices. Interestingly, side-chain derivatizations of pentacene in the form of ethynylsilyl substitutions have been reported by Anthony and coworkers [40]. We have found that such substitutions actually lead to a signiﬁcant increase (by about 50%) in the intramolecular reorganization energy, due to the involvement of the side chains in the geometry relaxation process upon ionization. In contrast, Wudl and coworkers have synthesized a tetra-methyl derivative of pentacene with the goal of improving the processability of the material [41]. These authors calculated the reorganization energy in the same way as described earlier and found that it remains exactly the same as in pentacene because, in this instance, the substituents have a saturated nature and do not couple to the geometry relaxations of the conjugated backbone [42].

8080_book.fm Page 10 Tuesday, April 3, 2007 1:47 PM

10

Organic Field-Effect Transistors

The origin of the small reorganization energy values in tetracene and pentacene can be traced to a combination of macrocyclic rigidity and full delocalization of the frontier molecular orbitals [24,26,27]. Accordingly, other molecules that have been found to present small intramolecular λi values are fullerenes, as described by Devos and Lannoo [43], phthalocyanines [39,44], or discotic macrocycles [45]. We note that the reorganization energy λi is directly related to such quantities as the polaron binding energy (Epol = λi/2) and the dimensionless electron–phonon parameter λe–ph (λe–ph = λi N(EF), where N(EF) is the density of states at the Fermi level [15,23–25,43]. The electron–phonon parameter is a key value in the conventional theory of superconductivity. Therefore, the results discussed before are especially relevant in the development of adequate polaron models to understand superconductivity and charge transport in organic molecular systems.

1.1.4 ELECTRONIC COUPLING A number of computational techniques, based on ab initio or semiempirical methodologies, have been developed to estimate the electronic coupling Vif; they have recently been reviewed in references 1, 2, and 46. A robust approach to compute Vif is to describe the diabatic states of the reactants and products by means of a Slater determinant and to compute their splitting at the transition state [47,48]. Li and coworkers have applied this approach to benzene and biphenyl dimers using concerted linear reaction coordinates to deﬁne the geometry of the transition state [47,48]. Another approach is to use Koopman’s theorem and to estimate (in the context of a one-electron picture) the transfer integrals t for holes or electrons as half the splitting of the highest occupied molecular orbital (HOMO) or lowest unoccupied molecular orbital (LUMO) levels, respectively, in a system made of two chains in the neutral state. In the case of benzene and biphenyl dimers, a good quantitative agreement is observed between the two approaches [49]. The applicability of Koopman’s theorem was conﬁrmed in a study by Pati and Karna [50]; we note that the direct coupling Vif was not evaluated at the transition-state geometry, on the basis that the coupling amplitude hardly depends on the actual nuclear conﬁguration, a feature often referred to as the Condon approximation. The transfer integrals can be estimated in a yet simpler approach from the spatial overlap between the two molecular orbitals in interaction [51,52]. All these considerations explain why many theoretical studies have made use of Koopman’s theorem to estimate electronic couplings [51–57]. We note that much care has to be taken when Koopman’s theorem is used to estimate the transfer integrals in asymmetric dimers, as has been extensively discussed elsewhere [58,59]. In such instances, part of the electronic splitting can simply arise from the different local environments experienced by the two interacting molecules, which create an offset between their HOMO and LUMO levels prior to their interaction due to polarization and/or electrostatic effects. In order to evaluate the effective couplings, this offset can be accounted for by performing calculations using molecular orbitals localized on the individual units as basis set [59] or by applying an electric ﬁeld to promote the resonance between the electronic levels, as done by Jortner and coworkers [56]. This artifact can also be prevented when

8080_book.fm Page 11 Tuesday, April 3, 2007 1:47 PM

Theoretical Aspects of Charge Transport in Organic Semiconductors

11

E LUMO

Conduction band

HOMO

Valence band

FIGURE 1.1.3 Illustration of the bonding-antibonding interactions between the HOMO/ LUMO levels of two ethylene molecules in a cofacial conﬁguration; we also illustrate the formation of the valence and conduction bands when a large number of stacked molecules interact.

estimating the electronic coupling from the overlap between the molecular orbitals of the isolated molecules. The electronic splittings reported next have been calculated within Koopman’s theorem using the semiempirical Hartree–Fock INDO (intermediate neglect of differential overlap) method; interestingly, the INDO method often provides transfer integrals of the same order of magnitude as those obtained with DFT-based approaches [45,60]. It is of interest to note that when building inﬁnite one-dimensional stacks of chains, the widths of the corresponding valence and conduction bands are usually found to be nearly equal to four times their respective t integrals; this indicates that in most instances the tight-binding approximation is relevant. In the remainder of this section, our goal is simply to lay out a basic understanding of how molecular packing and molecular size affect the electronic coupling between neighboring oligomers [61]. As a ﬁrst step, it is useful to consider the simple example of a dimer made of two perfectly superimposed ethylene molecules. In an isolated ethylene molecule, the HOMO corresponds to the bonding situation between the two π-atomic orbitals (zero-node case), while the LUMO corresponds to the (fully) antibonding situation. In a cofacial dimer, the interaction between the two molecules leads to a splitting of the HOMO level and a splitting of the LUMO level, as illustrated in Figure 1.1.3. The HOMO splitting is very large (0.54 eV for an intermolecular distance of 4 Å); the reason is that the interaction between the HOMO wave functions of the two molecules is either fully bonding (which leads to the much stabilized HOMO-1 level of the dimer) or fully antibonding (leading to the much destabilized dimer HOMO level). For the LUMO level, the splitting is found to be much smaller (0.15 eV) because direct bonding interactions are compensated by diagonal antibonding interactions in the dimer LUMO wave

8080_book.fm Page 12 Tuesday, April 3, 2007 1:47 PM

12

Organic Field-Effect Transistors

function or direct antibonding interactions are compensated by diagonal bonding interactions in the LUMO + 1 level (see Figure 1.1.3). Thus, as a qualitative rule in the case of perfectly cofacial conﬁgurations, one can state that the lower the number of nodes in the wave function of a given frontier level of an isolated chain is, the larger the splitting of that level in a dimer (or larger clusters) will be. Since in isolated π-conjugated chains the LUMO wave function has usually one more node than the HOMO wave function, the LUMO splitting is expected to be smaller than the HOMO splitting. Qualitatively, for large clusters, this will translate into larger HOMO bandwidths. This feature is what gave rise earlier to the conventional wisdom that, in crystals or crystalline ﬁlms of π-conjugated chains, the hole mobility is expected to be higher than the electron mobility. We will see later that this concept breaks down in numerous instances.

1.1.4.1 INFLUENCE

OF INTERMOLECULAR

SEPARATION

We now consider perfectly cofacial dimers made of two tetracene molecules and examine the evolution of the electronic splittings of the HOMO (Figure 1.1.4a) and LUMO (Figure 1.1.4b) levels as a function of the distance, d, between the molecular planes (see inset in Figure 1.1.5). Although fully cofacial conﬁgurations are rarely encountered in actual crystal structures, it is of interest to start by studying such geometries since they provide a highly symmetric reference point and lead to the largest electronic splittings (here, for instance, 280 and 200 meV for the HOMO and LUMO levels, respectively, for an interchain distance of 4 Å). The results are illustrated in Figure 1.1.5. Consistent with our previous discussion, the HOMO splitting is calculated to be larger than the LUMO splitting whatever the interchain separation may be. The amplitudes of the electronic splittings are observed to decay exponentially when the interchain distance is increased; this simply translates the exponential decay in intermolecular overlap of the π-atomic orbitals when the two oligomers are pulled apart. By ﬁtting the HOMO/LUMO

(a)

(b)

FIGURE 1.1.4 B3LYP/6-31G(d,p) HOMO (a) and LUMO (b) wavefunctions for tetracene.

8080_book.fm Page 13 Tuesday, April 3, 2007 1:47 PM

Theoretical Aspects of Charge Transport in Organic Semiconductors

13

1400

1200 d

Splitting (meV)

1000

800

600

400 HOMO 200

LUMO

0 3.3

3.6

3.9

4.2

4.5

4.8

5.1

5.4

5.7

d spacing (Å)

FIGURE 1.1.5 Evolution of the INDO-calculated electronic splittings of the HOMO and LUMO levels in a cofacial dimer made of two tetracene molecules as a function of the intermolecular separation.

splitting (Figure 1.1.5) to Equation 1.1.8, we obtained for holes or electrons V0 = 0.717 × 106 meV or 0.622 × 106 meV, and γ = 2.137 Å–1 or 2.187 Å–1, respectively. An important result is that the electronic splittings vary by as much as a factor of three to four between 3.4 and 4.0 Å. This range corresponds to the typical intermolecular distances found in organic conjugated crystals and thin ﬁlms.

1.1.4.2 INFLUENCE

OF

LONG-

OR

SHORT-AXIS DISPLACEMENTS

In many instances, cofacial packing involves the displacements of adjacent molecules along their long and/or short molecular axes. Figure 1.1.6 describes the evolution of the HOMO and LUMO splittings in dimers where the top tetracene molecule is translated along its long axis (while keeping the interchain distance ﬁxed at 3.4 Å). As expected, the overall effect of this displacement is to reduce the overlap and thus the electronic coupling. The most interesting result is the appearance of oscillations in the values of the splitting. The important consequence of this difference in oscillation period is that small translations can lead to situations where the electronic splitting is larger for the LUMO than for the HOMO and hence where electrons can possibly be more mobile than holes. For instance, for a shift of about 4.5 Å, the reversal in the relative amplitude of the splittings is very signiﬁcant: We calculate a LUMO splitting of 381 meV and a HOMO splitting of 17 meV.

8080_book.fm Page 14 Tuesday, April 3, 2007 1:47 PM

14

Organic Field-Effect Transistors

Short axis displacement

Long axis displacement

HOMO LUMO

1000 800

1000 800

600

600

400

Splitting (meV)

Splitting (meV)

HOMO LUMO

200 0 −200

400 200 0

−400

−200

−600

−400

−800

−600 0.0

2.5

5.0

7.5

Long axis displacement (Å)

10.0

12.5

0 1 2 3 4 5 6 Short axis displacement (Å)

FIGURE 1.1.6 Evolution of the INDO-calculated electronic splittings of the HOMO and LUMO levels in a dimer formed by two tetracene molecules separated by 3.4 Å as a function of the degree of translation of one molecule along its long axis (left) and short axis (right).

The calculated evolutions can be rationalized in terms of the phase and the nodal properties of the HOMO and LUMO orbitals of a single tetracene molecule [61,62]. In the HOMO level, the distribution of the positive and negative LCAO (linear combination of atomic orbitals) coefﬁcients shows a change in the sign of the wave function of every monomer unit (see Figure 1.1.4a). This pattern leads to extrema in the calculated electronic splittings for degrees of translation corresponding roughly to multiples of the monomer unit size. Large electronic splittings are dominated by full bonding or antibonding interactions between the π-atomic orbitals localized over the carbon–carbon bonds within the benzene rings. In contrast, small splittings are calculated for geometries where the global overlap (and hence the HOMO splitting) is considerably reduced by the compensation of bonding and antibonding interactions between the two bonds of one chain and the two adjacent bonds of the other chain. The evolution of the LUMO splitting also displays maxima and minima and can be explained once again by considering the shape of the LUMO orbital. Large or negligible small values are observed when the benzene rings of one molecule overlap the rings or half of the ring of the second molecule, respectively. The oscillation

8080_book.fm Page 15 Tuesday, April 3, 2007 1:47 PM

Theoretical Aspects of Charge Transport in Organic Semiconductors

15

period of the curve is roughly the same as that calculated for the HOMO splitting, although an increasing dephasing is observed; when the displacement is about 6.0 Å, the LUMO splitting has opposite phase with respect to the HOMO splitting. Note that an overall decrease in the HOMO and LUMO splittings occurs for increasing translational shifts. This simply results from the progressive reduction in the overall extent of spatial overlap between the two oligomers. In the case of the long-axis displacement, when the spatial overlap of the two molecules is zero (e.g., displacements bigger than the size of the molecule), the splitting is also close to zero. We now turn to the impact of translating the top oligomer along its short axis. The HOMO and LUMO splittings decay to zero when the molecule is shifted by more than 6.0 Å. This is more then twice the lateral width of the tetracene molecule (see Figure 1.1.6), illustrating clearly that the estimate of the coupling based purely on the “spatial overlap” of the two molecules can be misleading. In the case of the long-axis displacement, zero spatial overlap does mean negligible HOMO/LUMO splitting, while in the case of short-axis displacement, the orbital splitting is still reasonably large in the situation where the spatial overlap tends to zero. Again, the oscillations in splittings can be rationalized by looking at the wave functions. The antibonding character over the bonds along the short molecular axis found in the HOMO level leads to the appearance of a minimum in the course of translation. In contrast, the LUMO wave function does not change sign along the short axis; thus, the translation preserves dominant bonding or antibonding interactions in the LUMO + 1 or LUMO level of the dimer, respectively. These interactions attenuate as the translational shift increases, leading to a decrease in splitting without change in sign. The case of rubrene — a tetraphenyl derivative of tetracene lately gaining much attention due to reported ﬁeld-induced hole mobility of up to 20 cm2/Vs at room temperature [63] — illustrates how the results presented earlier can prove useful to rationalize the charge transport properties of this system. The largest electronic coupling is found for dimers along one of the crystallographic directions (see Figure 1.1.7). A combination of zero short-axis displacement and a reasonably large longaxis displacement (ca. 6.13 Å) still results in a large electronic coupling between the molecular units [37]. This result can be rationalized in terms of the oscillatory behavior of the electronic coupling as a function of the long-axis displacement (see Figure 1.1.6). Although the overall coupling decreases with the increase in displace-

a 3.74Å 6.13Å

FIGURE 1.1.7 Rubrene dimer along the a direction. The π-stacking distance and long-axis displacement are indicated.

8080_book.fm Page 16 Tuesday, April 3, 2007 1:47 PM

16

Organic Field-Effect Transistors

ment — around 6.13 Å — an extremum exists and thus the splitting observed at this conﬁguration is still a considerable fraction of the maximum splitting computed for the cofacial conﬁguration. To conclude this section, we underline that combining the results of calculations on transfer integrals (electronic couplings) and on intramolecular reorganization energies (electron- and hole-vibration constants) allows one to gain a basic understanding of the impact of molecular parameters on the intrinsic electron and hole mobilities in π-conjugated materials. In the following section, we adopt a pragmatic approach and, assuming reasonable values for the outer reorganization energy, we show how it is possible to make the connection between the molecular parameters described earlier and the carrier mobilities in model one-dimensional stacks of pentacene molecules.

1.1.5 FROM MOLECULAR PARAMETERS TO CARRIER MOBILITIES In this section, our goal is to make the connection between molecular scale and mesoscopic scale. We evaluate charge transport over large distances, in the presence of an electric ﬁeld, by inserting into Monte Carlo simulations the electron-transfer rates computed at the quantum-chemical level within the Bixon and Jortner model (Equation 1.1.6) for pairs of interacting molecules. We illustrate this approach on model systems made of one-dimensional stacks of pentacene molecules. The restriction to one-dimensional stacks allows us to rationalize easily the variations calculated in charge mobilities when: (1) the relative positions of the molecules are changed; (2) structural disorder is introduced; and (3) the external electric ﬁeld is modulated. This methodology can, however, be readily extended to two- and three-dimensional structures. A detailed description of the methodology is given elsewhere [64]. Brieﬂy, the electron- or hole-transfer rates calculated at the INDO quantum-chemical level are injected into Monte Carlo (MC) simulations to evaluate the propagation of a single charge along the pentacene one-dimensional stacks. MC methods rely on the use of random variables; in our case, the random variable is the occurrence of a hop between two adjacent molecules separated by a given distance. This distance is counted as positive if the hole propagates in the direction of the electric ﬁeld and as negative in the opposite direction; this counting scheme is reversed for electrons. In the MC algorithm, the ﬁrst step is to choose randomly the direction along which charge hopping takes place in a given iterative cycle (i.e., a jump to the left or right nearest neighbor in the one-dimensional stack). The second step is to calculate the probability of transferring the charge in the chosen direction (to the right in Equation (1.1.16)) as: p→ =

k→ k→ + k←

(1.1.16)

where k→ and k← correspond to the transfer rates in the right and left directions, respectively; these are different in the presence of the electric ﬁeld.

8080_book.fm Page 17 Tuesday, April 3, 2007 1:47 PM

Theoretical Aspects of Charge Transport in Organic Semiconductors

17

Finally, a random number is generated between 0 and 1 and is compared to p→. If the random number is lower than the transfer probability, the transfer in the right direction is accepted. Otherwise, the transfer is rejected, the charge carrier remains in its initial position, and a new iterative loop is performed. After a large number of iterative cycles (typically between 1010 and 1013), the mobility can be estimated directly as: μ = d/(τ F)

(1.1.17)

with F the magnitude of the electric ﬁeld, d the total distance traveled by the charge (summed as positive and negative contributions depending on the direction of the hop with respect to the ﬁeld), and τ the total time calculated as the sum of the inverse of the transfer rates. The electric ﬁeld is varied here from 104 to 107 V/cm; these are reasonable values for devices a few hundred nanometers thick under the application of 10–100 V between the electrodes (note that we assume a linear voltage drop across the device). Thus, this approach offers a way of connecting molecular parameters such as λ, ΔG°, and t to macroscopic quantities such as charge mobility. There is no limitation in size since the system is built progressively around the charge when moving along the stacks. We recall that the inner reorganization energy λi of pentacene has been calculated to be 0.10 eV for holes and 0.13 eV for electrons [25]. Since the electric ﬁeld is consistently applied in the direction perpendicular to the molecular plane of the pentacene oligomers, the magnitude of λi is assumed to be ﬁeld independent. This is further validated by recent calculations showing that λi is not affected in the presence of an external electric ﬁeld applied along the long molecular axis of pentacene molecules [65]. The outer reorganization energy λs has been estimated on the basis of a modiﬁed expression, taking into account the actual shape of the donor and acceptor molecules; values on the order of 0.3–0.4 eV are obtained when considering the dielectric characteristics of organic matrices [66].

1.1.5.1 INFLUENCE

OF THE

ELECTRIC FIELD

We report in Figure 1.1.8 the ﬁeld dependence of the hole and electron mobilities calculated for a one-dimensional array of pentacene molecules in a cofacial conﬁguration, with an intermolecular distance ﬁxed at 4 Å. The simulations have been performed at 300 K for an external reorganization energy of 0.4 eV. The transfer integrals associated with such an arrangement amount to 0.13 and 0.10 eV for holes and electrons, respectively. The mobility increases by about a factor of four for both holes and electrons when the electric ﬁeld is varied from 104 to 107 V/cm. Since the magnitude of the electric ﬁeld appears in the denominator of the mobility expression, the results further indicate that the charge velocity (d/τ) increases faster than the magnitude of the ﬁeld. The ratio between the hole and electron mobilities at any ﬁeld perfectly matches the ratio of the corresponding transfer rates. It is interesting to note that the calculated mobility values (in the range 10–1–1 cm2/Vs) are very reasonable; the mobilities reported for pentacene thin ﬁlms, and crystals are usually in the range of 0.1–5 cm2/Vs [67].

8080_book.fm Page 18 Tuesday, April 3, 2007 1:47 PM

18

Organic Field-Effect Transistors

0.8 0.7 4Å

Mobility (cm2/ Vs)

0.6 0.5 0.4 0.3 0.2 0.1

Hole Electron

0 0

1

2

3

4 Field

5

6

(106

V/cm)

7

8

9

10

FIGURE 1.1.8 Evolution of the hole mobility as a function of the magnitude of the electric ﬁeld in a one-dimensional array of pentacene molecules separated by 4 Å.

1.1.5.2 INFLUENCE

OF THE

REORGANIZATION ENERGY

Since the external reorganization energy may vary signiﬁcantly in speciﬁc cases (for instance, when the molecules are located close to structural defects or in the vicinity of metallic electrodes), we have investigated the way the hole mobility evolves as a function of the magnitude of λs. Monte Carlo simulations have been performed on the same one-dimensional array of pentacene molecules as discussed earlier, with an electric ﬁeld ﬁxed at 5 × 106 V/cm and at 300 K. Figure 1.1.9 illustrates the mobility drop with λs. The mobility is reduced by about a factor of ﬁve for an increase of λs from 0.2 to 0.4 eV or from 0.4 to 0.6 eV; it is strongly hampered beyond 0.7 eV. In this case, the evolution simply reﬂects that for the transfer rate (see inset of Figure 1.1.9).

1.1.5.3 INFLUENCE

OF INTERMOLECULAR

DISTANCE

Changes in the distance between molecules can be modulated by substituents attached to conjugated backbones. Accordingly, we have examined the evolution of hole mobility in the one-dimensional array of pentacene molecules when varying the intermolecular separation between 3.4 and 5 Å; see Figure 1.1.10. In the remainder, unless stated otherwise, the simulations have been achieved with λs = 0.4 eV and for two different magnitudes of the electric ﬁeld (106 and 5 × 106 V/cm). In all cases, the mobilities drop exponentially with an increase in the intermolecular distance and are shifted rigidly depending on the magnitude of the electric ﬁeld; this evolution has to be traced back to that of the transfer integrals since the overlap between the HOMOs decreases exponentially with distance.

8080_book.fm Page 19 Tuesday, April 3, 2007 1:47 PM

Theoretical Aspects of Charge Transport in Organic Semiconductors

1.8

Transfer rate (s−1)

1.6

Mobility (cm2/ Vs)

1.4 1.2 1 0.8

19

4.50E + 14 4.00E + 14 3.50E + 14 3.00E + 14 2.50E + 14 2.00E + 14 1.50E + 14 1.00E + 14 5.00E + 14 0.00E + 00 0.2

0.3

0.4 0.5 λs (eV)

0.6

0.6

0.7

0.4 0.2 0 0.2

0.25

0.3

0.35

0.4

0.45

0.5

0.55

0.6

0.65

0.7

λs (eV)

FIGURE 1.1.9 Evolution of the hole mobility as a function of the magnitude of the external reorganization energy in a one-dimensional array of pentacene molecules separated by 4 Å.

10 F = 106 V/cm F = 5 × 106 V/cm 1 Mobility (cm2/ Vs)

3.4

3.6

3.8

4

4.2

4.4

4.6

4.8

5

0.1

d 0.01

0.001 Intermolecular distance (Å)

FIGURE 1.1.10 Evolution of the hole mobility as a function of intermolecular distance in a one-dimensional array of pentacene molecules.

8080_book.fm Page 20 Tuesday, April 3, 2007 1:47 PM

20

Organic Field-Effect Transistors

Transfer integral (eV)

0.45 0.4

Mobility (cm2/ Vs)

0.35 0.3 0.25

0.14 0.12 0.1 0.08 0.06 0.04 0.02 0

d

4Å

0

0.2

0.5

1

1.5 2 Shift (Å)

2.5

3

3.5

F = 106 V/cm F = 5 × 106 V/cm

0.15 0.1 0.05 0 0

0.5

1

1.5

2

2.5

3

3.5

Shift (Å)

FIGURE 1.1.11 Evolution of the hole mobility as a function of the shift applied to every other molecule in a direction perpendicular to the stacking axis in a one-dimensional array of pentacene molecules separated by 4 Å. The inset shows the evolution of the corresponding transfer integral.

1.1.5.4 INFLUENCE

OF

MOLECULAR TRANSLATIONS

We have looked at the impact of translating molecules in a direction parallel or perpendicular to the stacking axis. Figure 1.1.11 shows the variation of the hole mobility when translating every other molecule by a distance d perpendicularly to the stacking axis (note that a single transfer integral value characterizes the full system). The conventional wisdom here would be that the mobility goes down with an increase in displacement due to the progressive reduction in spatial overlap between two adjacent molecules. However, as discussed previously, the calculated values globally decrease with distance but in an oscillating way [61]. Again, this evolution fully reﬂects that of the corresponding transfer integrals. Once more, we emphasize that the balance between the number of bonding versus antibonding interactions in the electronic overlap between the wave functions of the two molecules dictates the magnitude of the transfer integral and thus of the hole mobility. Maxima are observed when one kind of interaction dominates and minima when a compensation occurs between them. We have also shifted every other molecule along the stacking axis by a distance ranging from 0.1 to 0.5 Å with respect to the initial situation where all the molecules are separated by 4 Å. In such an arrangement, the charge has to hop alternatively over a distance larger or shorter than 4 Å. For various magnitudes of the applied electric ﬁeld, we obtain parallel variations that point to a reduction in the mobility when the displacement is ampliﬁed (Figure 1.1.12); the impact is much more pro-

8080_book.fm Page 21 Tuesday, April 3, 2007 1:47 PM

Theoretical Aspects of Charge Transport in Organic Semiconductors

21

1

Mobility (cm2/ Vs)

F = 106 V/cm F = 5 × 106 V/cm

4Å − Δx 4Å + Δx

0.1

0.01 0

0.1

0.2

0.3

0.4

0.5

Shift (Å)

FIGURE 1.1.12 Evolution of the hole mobility as a function of the shift applied to every single molecule over two in a direction parallel to the stacking axis.

nounced for large shifts. This behavior can be understood by comparing the transfer rates associated with the short and long hops. When the distance is smaller or longer than 4 Å, the transfer rate increases or decreases, respectively, as discussed in Section 1.1.5.3. However, since the evolution with respect to the initial distance is not symmetric, the time required to make two consecutive hops in the same direction (and thus to travel in all cases a distance of 8 Å) increases with the degree of translation. Interestingly, ﬂuctuations in the separation as large as 0.5 Å reduce the mobility only by a factor of ﬁve.

1.1.5.5 INTRODUCTION

OF A

GAUSSIAN DISORDER

We generalize the simulations carried out in the previous section by introducing a Gaussian distribution of the intermolecular distances d along the cofacial stack, randomly among the pairs of interacting molecules. The corresponding transfer integrals are estimated from an analytical expression of the results obtained in Section 1.1.5.3; we further assume here that λs is not affected by variations in the intermolecular distances. The Gaussian distribution g(d) is centered around an average value d0 set equal to 4 Å and is characterized by a standard deviation σ set equal to 0.05, 0.1, and 0.2 Å, respectively, in the simulations:

g(d ) =

⎡ ( d − d0 )2 ⎤ exp ⎢ − ⎥ 2σ2 ⎦ σ 2π ⎣ 1

(1.1.18)

With this expression, 99% of the generated distances lie in the interval [d0 – 3σ, d0 + 3σ]. Figure 1.1.13 shows that, as expected, the mobility is reduced when the standard deviation increases, whatever the magnitude of electric ﬁeld might be. However, this reduction is very moderate and points to the weak impact of such

8080_book.fm Page 22 Tuesday, April 3, 2007 1:47 PM

22

Organic Field-Effect Transistors

1

Mobility (cm2/ Vs)

F = 106 V/cm F = 5 × 106 V/cm

0.1

0

0.05

0.1

0.15

0.2

σ (Å)

FIGURE 1.1.13 Evolution of the hole mobility as a function of the width of the Gaussian distribution of the intermolecular distances in the one-dimensional array of pentacene molecules.

types of ﬂuctuations on the charge-transport properties. The calculated evolution is due to the fact that the transfer rate increases faster below 4 Å than it decreases above 4 Å.

1.1.6 CONCLUDING REMARKS A major message resulting from our discussions is that the amplitudes of the transfer integrals depend on both the relative positions of the interacting molecules/oligomers and the shape (bonding–antibonding pattern) of their frontier molecular orbitals. Thus, the transfer integral amplitudes can hardly be predicted from a simple examination of molecular packing. Fortunately for theoretical chemists, this underlines the useful role that quantum chemistry can play by providing a molecular-scale understanding of the charge-transport parameters in conjugated systems. However, the role of the interactions with lattice phonons, which modulate the transfer integrals and deﬁne the external reorganization energy, and of the induced electronic polarization and electrostatic effects, which introduce an offset between the electronic levels and hence another contribution to ΔG°, need to be better incorporated, as shown by a number of recent studies [59,68–70]. Work along these lines is necessary to be able to estimate carrier mobilities accurately.

REFERENCES 1. Balzani, V., Electron transfer in chemistry, Wiley–VCH: Weinheim, NY, 2001. 2. Jortner, J., Bixon, M., Prigogine, I., and Rice, S.A., Electron transfer: From isolated molecules to biomolecules, John Wiley & Sons: New York, 1999.

8080_book.fm Page 23 Tuesday, April 3, 2007 1:47 PM

Theoretical Aspects of Charge Transport in Organic Semiconductors

23

3. Marcus, R.A., Electron transfer reactions in chemistry. Theory and experiment, Rev. Mod. Phys., 65, 599, 1993. 4. Marcus, R.A., On the theory of oxidation-reduction reactions involving electron transfer. I, J. Chem. Phys., 24, 966, 1956. 5. Marcus, R.A. and Sutin, N., Electron transfers in chemistry and biology, Biochim. Biophys. Acta, 811, 265, 1985. 6. Dimitrakopoulos, C.D. and Malenfant, P.R.L., Organic thin ﬁlm transistors for large area electronics, Adv. Mater., 14, 99, 2002. 7. Jurchescu, O.D., Baas, J., and Palstra, T.T.M., Effect of impurities on the mobility of single crystal pentacene, Appl. Phys. Lett., 84, 3061, 2004. 8. Fichou, D., Structural order in conjugated oligothiophenes and its implications on opto-electronic devices, J. Mater. Chem., 10, 571, 2000. 9. Duke, C.B. and Schein, L.B., Organic solids: Is energy-band theory enough? Phys. Today, 33, 42, 1980. 10. Warta, W., Stehle, R., and Karl, N., Ultrapure, high-mobility organic photoconductors, Appl. Phys. A: Mater. Sci. Process., 36, 163, 1985. 11. Wu, M.W. and Conwell, E.M., Transport in α-sexithiophene ﬁlms, Chem. Phys. Lett., 266, 363, 1997. 12. Troisi, A., and Orlandi, G., The hole transfer in DNA: Calculation of electron coupling between close bases, Chem. Phys. Lett., 344, 509, 2001. 13. Bassler, H., Charge transport in disordered organic photoconductors A Monte Carlo simulation study, Phys. Status Solidi B, 175, 15, 1993. 14. Demadis, K.D., Hartshorn, C.M., and Meyer, T.J., The localized-to-delocalized transition in mixed-valence chemistry, Chem. Rev., 101, 2655, 2001. 15. Coropceanu, V., André, J.M., Malagoli, M., and Brédas, J.L., The role of vibronic interactions on intramolecular and intermolecular electron transfer in p-conjugated oligomers, Theor. Chem. Accounts, 110, 59, 2003. 16. Marcus, R.A., On the theory of electron-transfer reactions. VI. Uniﬁed treatment for homogeneous and electrode reactions, J. Chem. Phys., 43, 679, 1965. 17. Reimers, J. R., A practical method for the use of curvilinear coordinates in calculations of normal-mode-projected displacements and Duschinsky rotation matrices for large molecules, J. Chem. Phys., 115, 9103, 2001. 18. Silinsh, E.A., Klimkans, A., Larsson, S., and Capek, V., Molecular polaron states in polyacene crystals. Formation and transfer processes, Chem. Phys., 198, 311, 1995. 19. Pope, M., and Swenberg, C.E., Electronic processes in organic crystals and polymers, 2nd ed., Oxford University Press: New York, 1999. 20. Malagoli, M. and Brédas, J.L., Density functional theory study of the geometric structure and energetics of triphenylamine-based hole-transporting molecules, Chem. Phys. Lett., 327, 13, 2000. 21. May, V. and Kühn, O., Charge and energy transfer dynamics in molecular systems: A theoretical introduction, 1st ed., Wiley–VCH: Weinheim, NY, 2000. 22. Wilson, E.B., Decius, J.C., and Cross, P.C., Molecular vibrations: The theory of infrared and Raman vibrational spectra, Dover Publications: New York, 1980. 23. Gruhn, N.E., da Silva, D.A., Bill, T.G., Malagoli, M., Coropceanu, V., Kahn, A., and Brédas, J.L., The vibrational reorganization energy in pentacene: Molecular inﬂuences on charge transport, J. Am. Chem. Soc., 124, 7918, 2002. 24. Malagoli, M., Coropceanu, V., da Silva, D.A., and Brédas, J.L., A multimode analysis of the gas-phase photoelectron spectra in oligoacenes, J. Chem. Phys., 120, 7490, 2004.

8080_book.fm Page 24 Tuesday, April 3, 2007 1:47 PM

24

Organic Field-Effect Transistors 25. Coropceanu, V., Malagoli, M., da Silva, D.A., Gruhn, N.E., Bill, T.G., Brédas, J.L., Hole- and electron-vibrational couplings in oligoacene crystals: Intramolecular contributions, Phys. Rev. Lett., 89, 275503, 2002. 26. da Silva, D.A., Friedlein, R., Coropceanu, V., Ohrwall, G., Osikowicz, W., Suess, C., Sorensen, S.L., Svensson, S., Salaneck, W.R., and Brédas, J.L., Vibronic coupling in the ground and excited states of the naphthalene cation, Chem. Comm., 15, 1702, 2004. 27. Kwon, O., Coropceanu, V., Gruhn, N.E., Durivage, J.C., Laquindanum, J.G., Katz, H.E., Cornil, J., and Brédas, J.L., Characterization of the molecular parameters determining charge transport in anthradithiophene, J. Chem. Phys., 120, 8186, 2004. 28. Nelson, S. F., Lin, Y.Y., Gundlach, D.J., and Jackson, T.N., Temperature-independent transport in high-mobility pentacene transistors, Appl. Phys. Lett., 72, 1854, 1998. 29. Sundar, V. C., Zaumseil, J., Podzorov, V., Menard, E., Willett, R.L., Someya, T., Gershenson, M.E., and Rogers, J.A., Elastomeric transistor stamps: Reversible probing of charge transport in organic crystals, Science, 303, 1644, 2004. 30. Brédas, J.L., Beljonne, D., Coropceanu, V., and Cornil, J., Charge-transfer and energytransfer processes in conjugated oligomers and polymers: A molecular picture, Chem. Rev., 104, 4971, 2004. 31. Kato, T. and Yamabe, T., Vibronic interactions and superconductivity in acene anions and cations, J. Chem. Phys., 115, 8592, 2001. 32. Kato, T. and Yamabe, T., Electron–intramolecular phonon coupling and possible superconductivity in negatively charged coronene and corannulene, J. Chem. Phys., 117, 2324, 2002. 33. Kato, T. and Yamabe, T., Electron–intramolecular–vibration interactions in positively charged phenanthrene-edge-type hydrocarbons, J. Chem. Phys., 120, 3311, 2004. 34. Kato, T. and Yamabe, T., Electron–phonon coupling in negatively charged cubane and octasilacubane, J. Chem. Phys., 118, 3300, 2003. 35. Kato, T. and Yamabe, T., Electron–phonon interactions in charged cubic ﬂuorocarbon cluster, (CF)8, J. Chem. Phys., 120, 1006, 2004. 36. Klimkans, A. and Larsson, S., Reorganization energies in benzene, naphthalene, and anthracene, Chem. Phys., 189, 25, 1994. 37. da Silva, D.A., Kim, E.G., and Brédas, J.L., Transport properties in the rubrene crystal: Electronic coupling and vibrational reorganization energy, Adv. Mater., 17, 1072, 2005. 38. Sánchez-Carrera, R.S., Coropceanu, V., da Silva Filho, D.A., Friedlein, R., Osikowicz, W., Murdey, R., Suess, C., Salaneck, W.R., and Brédas, J.L., Vibronic coupling in the ground and excited states of oligoacene cations, J. Phys. Chem. B, 110, 18904, 2006. 39. Tant, J., Geerts, Y.H., Lehmann, M., De Cupere, V., Zucchi, G., Laursen, B.W., Bjornholm, T., Lemaur, V., Marcq, V., Burquel, A., Hennebicq, E., Gardebien, F., Viville, P., Beljonne, D., Lazzaroni, R., and Cornil, J., Liquid crystalline metal-free phthalocyanines designed for charge and exciton transport, J. Phys. Chem. B, 109, 20315, 2005. 40. Anthony, J.E., Brooks, J.S., Eaton, D.L., and Parkin, S.R., Functionalized pentacene: Improved electronic properties from control of solid-state order, J. Am. Chem. Soc., 123, 9482, 2001. 41. Meng, H., Bendikov, M., Mitchell, G., Helgeson, R., Wudl, F., Bao, Z., Siegrist, T., Kloc, C., and Chen, C.H., Tetramethylpentacene: Remarkable absence of steric effect on ﬁeld effect mobility, Adv. Mater., 15, 1090, 2003.

8080_book.fm Page 25 Tuesday, April 3, 2007 1:47 PM

Theoretical Aspects of Charge Transport in Organic Semiconductors

25

42. Chen, H.Y. and Chao, I., Effect of perﬂuorination on the charge-transport properties of organic semiconductors: Density functional theory study of perﬂuorinated pentacene and sexithiophene, Chem. Phys. Lett., 401, 539, 2005. 43. Devos, A. and Lannoo, M., Electron-phonon coupling for aromatic molecular crystals: Possible consequences for their superconductivity, Phys. Rev. B, 58, 8236, 1998. 44. Sun, S.-S. and Sariciftci, N.S., Organic photovoltaics: Mechanism, materials, and devices, Taylor & Francis: Boca Raton, FL, 2005. 45. Lemaur, V., da Silva Filho, D.A., Coropceanu, V., Lehmann, M., Geerts, Y., Piris, J., Debije, M.G., Van de Craats, A.M., Senthilkumar, K., Siebbeles, L.D.A., Warman, J.M., Brédas, J.L., and Cornil, J., Charge transport properties in discotic liquid crystals: A quantum-chemical insight into structure–property relationships, J. Am. Chem. Soc., 126, 3271, 2004. 46. Newton, M.D., Quantum chemical probes of electron-transfer kinetics: The nature of donor–acceptor interactions, Chem. Rev., 91, 767, 1991. 47. Li, X.-Y., Tang, X.-S., and He, F.-C., Electron transfer in poly(p-phenylene) oligomers: Effect of external electric ﬁeld and application of Koopmans theorem, Chem. Phys., 248, 137, 1999. 48. Li, X.Y. and He, F.C., Electron transfer between biphenyl and biphenyl anion radicals: Reorganization energies and electron transfer matrix elements, J. Comp. Chem., 20, 597, 1999. 49. Li, Z.C., Xu, L., Sun, H., Xiao, Y.M., and Zhang, J., Investigation on performances of non-loss storage for cryogenic liqueﬁed gas, Cryogenics, 44, 357, 2004. 50. Pati, R. and Karna, S.P., Ab initio Hartree–Fock study of electron transfer in organic molecules, J. Chem. Phys., 115, 1703, 2001. 51. Wolfsberg, M. and Helmholz, L., The spectra and electronic structure of the tetrahedral ions MnO–4, CrO–4, and ClO–4, J. Chem. Phys., 20, 837, 1952. 52. Pietro, W.J., Marks, T.J., and Ratner, M.A., Resistivity mechanisms in phthalocyanine-based linear-chain and polymeric conductors: Variation of bandwidth with geometry, J. Am. Chem. Soc., 107, 5387, 1985. 53. Paulson, B.P., Curtiss, L.A., Bal, B., Closs, G.L., and Miller, J.R., Investigation of through-bond coupling dependence on spacer structure, J. Am. Chem. Soc., 118, 378, 1996. 54. Liang, C.X. and Newton, M.D., Ab initio studies of electron transfer: Pathway analysis of effective transfer integrals, J. Phys. Chem., 96, 2855, 1992. 55. Palenberg, M.A., Silbey, R.J., Malagoli, M., and Brédas, J.L., Almost temperatureindependent charge carrier mobilities in liquid crystals, J. Chem. Phys., 112, 1541, 2000. 56. Voityuk, A.A., Rosch, N., Bixon, M., and Jortner, J., Electronic coupling for charge transfer and transport in DNA, J. Phys. Chem. B, 104, 9740, 2000. 57. Grozema, F.C., van Duijnen, P.T., Berlin, Y.A., Ratner, M.A., and Siebbeles, L.D.A., Intramolecular charge transport along isolated chains of conjugated polymers: Effect of torsional disorder and polymerization defects, J. Phys. Chem. B, 106, 7791, 2002. 58. Valeev, E.F., Coropceanu, V., da Silva Filho, D.A., Salman, S., and Brédas, J.L., Effect of electronic polarization on charge-transport parameters in molecular organic semiconductors, J. Am. Chem. Soc., 128, 9882, 2006. 59. Senthilkumar, K., Grozema, F.C., Bickelhaupt, F.M., and Siebbeles, L.D.A., Charge transport in columnar stacked triphenylenes: Effects of conformational ﬂuctuations on charge transfer integrals and site energies, J. Chem. Phys., 119, 9809, 2003.

8080_book.fm Page 26 Tuesday, April 3, 2007 1:47 PM

26

Organic Field-Effect Transistors 60. Mattheus, C.C., Polymorphism and electronic properties of pentacene, Ph.D. dissertation, University of Groningen, Netherlands, 2002. 61. Brédas, J.L., Calbert, J.P., da Silva, D.A., and Cornil, J., Organic semiconductors: A theoretical characterization of the basic parameters governing charge transport, Proc. Natl. Acad. Sci. USA, 99, 5804, 2002. 62. Kazmaier, P.M. and Hoffmann, R., A theoretical study of crystallochromy. Quantum interference effects in the spectra of perilene pigments, J. Am. Chem. Soc., 116, 9684, 1994. 63. Podzorov, V., Menard, E., Borissov, A., Kiryukhin, V., Rogers, J.A., and Gershenson, M.E., Intrinsic charge transport on the surface of organic semiconductors, Phys. Rev. Lett., 93, 086602, 2004. 64. Olivier, Y., Lemaur, V., Brédas, J. L., and Cornil, J., Charge hopping in organic semiconductors: Inﬂuence of molecular parameters on macroscopic mobilites in model one-dimensional stacks, J. Phys. Chem. A, 110, 6356, 2006. 65. Sancho-Garcia, J.C., Horowitz, G., Brédas, J.L., and Cornil, J., Effect of an external electric ﬁeld on the charge transport parameters in organic molecular semiconductors, J. Chem. Phys., 119, 12563, 2003. 66. Lemaur, V., Steel, M., Beljonne, D., Brédas, J.L., and Cornil, J., Photoinduced charge generation and recombination dynamics in model donor/acceptor pairs for organic solar cell applications: A full quantum-chemical treatment, J. Am. Chem. Soc., 127, 6077, 2005. 67. Facchetti, A., Yoon, M.H., and Marks, T.J., Gate dielectrics for organic ﬁeld-effect transistors: New opportunities for organic electronics, Adv. Mater., 17, 1705, 2005. 68. Johansson, A. and Stafstrom, S., Interchain charge transport in disordered π-conjugated chain systems, Phys. Rev. B, 66, 085208, 2002. 69. Troisi, A. and Orlandi, G., Charge-transport regime of crystalline organic semiconductors: Diffusion limited by thermal off-diagonal electronic disorder, Phys. Rev. Lett., 96, 086601, 2006. 70. Leontyev, I.V. and Tachiya, M., The reorganization energy of electron transfer in nonpolar solvents: Molecular level treatment of the solvent, J. Chem. Phys., 123, 224502, 2005.

8080_C002.fm Page 27 Friday, April 6, 2007 10:21 AM

2.1

Charge Carrier Transport in Single-Crystal Organic Field-Effect Transistors

Vitaly Podzorov CONTENTS 2.1.1 Introduction: The Field Effect in Small-Molecule Organic Semiconductors............................................................................................28 2.1.2 Fabrication of Single-Crystal OFETs..........................................................30 2.1.3 Charge Transport on the Surface of Organic Single Crystals ....................38 2.1.3.1 Basic FET Operation.....................................................................38 2.1.3.2 The Multiple Trap-and-Release Model.........................................46 2.1.3.3 Anisotropy of the Mobility ...........................................................48 2.1.3.4 Longitudinal and Hall Conductivity in Rubrene OFETs..............50 2.1.3.5 Comparison with the Holstein–Peierls Model and Transport Measurements in the Bulk of Organic Crystals............................54 2.1.3.6 Tuning the Intermolecular Distance..............................................55 2.1.3.7 Surface versus Bulk Transport ......................................................56 2.1.3.8 Photoinduced Processes in Single-Crystal OFETs.......................58 2.1.4 Defects at the Surface of Organic Crystals .................................................59 2.1.4.1 Bulk and Surface Electronic Defects in Organic Crystals ...........61 2.1.4.2 Density of Defects in Single-Crystal OFETs ...............................63 2.1.4.3 Single-Crystal OFETs as Tools to Study Surface Defects...........64 2.1.5 Conclusion ...................................................................................................65 Acknowledgments....................................................................................................67 References................................................................................................................67

27

8080_C002.fm Page 28 Friday, April 6, 2007 10:21 AM

28

Organic Field-Effect Transistors

2.1.1 INTRODUCTION: THE FIELD EFFECT IN SMALL-MOLECULE ORGANIC SEMICONDUCTORS Organic semiconductors represent a large class of solids consisting of organic oligomers or polymers. This chapter focuses on crystals of small organic molecules (mostly, polyacenes containing typically 2–10 benzene rings) held together in a solid by van der Waals forces. These small-molecule organic semiconductors, together with polymers, represent the material basis for the rapidly developing field of organic electronics [1–5]. Due to the weak van der Waals bonding, many electronic properties of these materials (e.g., the energy gap between the highest occupied and lowest unoccupied molecular orbitals — HOMO and LUMO, respectively) are determined by the structure of an isolated molecule [6–8]. The weak intermolecular overlap of electronic orbitals results in narrow electronic bands (a typical bandwidth, W ~ 0.1 eV, is two orders of magnitude smaller than that in silicon), a low mobility of carriers (μ ~ 1–10 cm2/Vs at room temperature) and strong electron-lattice coupling. The anisotropy of the transfer integrals between adjacent molecules reflects the low symmetry of molecular packing in organic molecular crystals (OMCs). It is believed that the most adequate description of charge transport in these semiconductors is based on the concept of polarons — the electronic states resulting from interaction of charge with lattice polarization at a length scale comparable to or greater than the lattice constant [6,7,9,10]. After several decades of intensive research, our basic understanding of charge transport in small-molecule organic semiconductors remains limited. The complexity of transport phenomena in these systems is due to the polaronic nature of charge carriers and a strong interaction of small polarons with defects [6]. An especially challenging task is to develop an adequate model of high-temperature polaronic transport. At room temperature, which is typically comparable to or even higher than the characteristic phonon energies, the lattice vibrations might become sufficiently strong to destroy the translational symmetry of the lattice. In this regime, the fluctuation amplitude of the transfer integral becomes of the same order of magnitude as its average value [11], the band description breaks down, and a crossover from the band-like transport in delocalized states to the incoherent hopping between localized states is predicted with increasing temperature. At low enough temperatures (T), when the band description is still valid, the polaronic bandwidth, W, “shrinks” as T increases, leading to a decrease of the carrier mobility μ with T [12–18]. The benchmark for the study of charge transport in organic semiconductors was established by time-of-flight (TOF) experiments with ultrapure polyacene crystals, such as naphthalene and anthracene [19]. These experiments have demonstrated that the intrinsic (not limited by static disorder) charge transport can be realized in the bulk of these crystals. This transport regime is characterized with a rapid growth of the carrier mobility with decreasing temperature and a pronounced anisotropy of the mobility, which reflects the anisotropy of the intermolecular transfer integrals [12,13]. Numerous applications, however, are dependent on the charge transport on

8080_C002.fm Page 29 Friday, April 6, 2007 10:21 AM

Charge Carrier Transport in Single-Crystal Organic Field-Effect Transistors

29

the surface of organic semiconductors. The most important example is the organic field-effect transistor, in which field-induced charges move along the interface between an organic semiconductor and a gate dielectric. In these devices, conduction truly occurs at the surface because the thickness of the conducting channel does not exceed a few molecular layers [20–22]. The transport of field-induced carriers on organic surfaces may differ from the bulk transport in many respects. For instance, the density of carriers in the fieldeffect experiments can exceed that in the bulk TOF measurements by many orders of magnitude, approaching the regime when the intercharge distance becomes comparable with the size of small polarons [23,24]. Interactions between the polaronic carriers may become important in this regime. The motion of charge carriers in the field-induced conduction channel might be affected by the polarization of the gate dielectric [25]. Finally, molecular packing at organic surfaces could be different from that in the bulk. The investigation of polaronic transport on organic surfaces is crucial for a better understanding of the fundamental processes that determine operation and ultimate performance of organic electronic devices. This is an important issue. On the one hand, the first all-organic devices (e.g., the active matrix displays based on organic light-emitted diodes and organic transistors) are expected to be commercialized within a few years. On the other hand, our knowledge of the transport properties of organic semiconductors is much more limited than it is for their inorganic counterparts. This paradoxical situation contrasts sharply to the situation in inorganic electronics in the mid-1960s, when the first Si metal-oxide semiconductor field-effect transistors (MOSFETs) were developed [26]. Difficulties in fundamental research have been caused by the lack of a proper tool for exploring the polaronic transport on surfaces of organic semiconductors. The most common organic electronic device, whose operation relies on surface transport, is the organic thin-film transistor (TFT). Over the past two decades a large effort in the development of TFTs has resulted in an impressive improvement of the characteristics of these devices [27] so that, currently, the best organic TFTs outperform the widely used amorphous silicon (α-Si:H) transistors. However, even in the best organic TFTs, charge transport is still dominated by the presence of structural defects and chemical impurities. As a result, it has been concluded that TFTs cannot be reliably used for the studies of basic transport mechanisms in organic materials [28]. The recently developed single-crystal organic transistors with significantly reduced disorder [29–37] provide unique opportunities to explore fundamental processes that determine the operation and reliability of organic electronic devices. For the first time, these single-crystal organic field-effect transistors (OFETs) have enabled the observation of intrinsic (not limited by static disorder) transport of fieldinduced charges at organic surfaces [35,38,39]. The carrier mobility in these devices is an order of magnitude greater than that in the best organic TFTs [32]. Equally important, the single-crystal OFETs are characterized by a very good reproducibility: Devices fabricated in different laboratories exhibit similar characteristics. This reproducibility, which is crucial for the investigation of electronic properties of organic semiconductors, has never been achieved with thin-film devices, whose electrical

8080_C002.fm Page 30 Friday, April 6, 2007 10:21 AM

30

Organic Field-Effect Transistors

characteristics are strongly dependent on the details of fabrication processes and handling environment. In this chapter, we present a brief overview of the experimental results obtained with single-crystal OFETs over the last four years. In Section 2.1.2, we briefly describe the crystal growth and OFET fabrication techniques that preserve the high quality of pristine surfaces of as-grown crystals. Section 2.1.3 focuses on the intrinsic transport characteristics of surfaces and interfaces of organic crystalline devices. Electronic defects at organic surfaces and mechanisms of their formation are discussed in Section 2.1.4. Section 2.1.5 outlines several basic issues that can be experimentally addressed in the near future owing to the availability of singlecrystal OFETs.

2.1.2 FABRICATION OF SINGLE-CRYSTAL OFETS The first step in the fabrication of single-crystal OFETs is the growth of ultrapure organic crystals. The best results to date have been obtained with physical vapor transport (PVT) growth in a stream of ultrahigh purity argon, helium, or hydrogen gases, similar to the method suggested by Laudise et al. [40]. A PVT furnace consists of a quartz tube with a stabilized temperature profile created along the tube by external heaters (Figure 2.1.1). The temperature gradient can be achieved by resistively heated wire unevenly wound on the quartz tube or, more conveniently, by coaxially enclosing the quartz tube in a metal tube (copper, brass, or stainless steel) with two regions of stabilized temperature: high T on the left and low T on the right (Figure 2.1.1). In such a design, the good thermal conductivity of the metal creates a linear temperature distribution between the heated and cooled regions, with the temperature (Tset) (Tgrowth) Water coil

Heater H2

Copper tube

Temperature proﬁle

T (°C)

200 150 100 50 0

10

20 30 x (cm)

40

50

FIGURE 2.1.1 A sketch of the physical vapor transport (PVT) growth furnace (top) and an example of the temperature profile along the axis of the quartz tube (bottom).

8080_C002.fm Page 31 Friday, April 6, 2007 10:21 AM

Charge Carrier Transport in Single-Crystal Organic Field-Effect Transistors

31

gradient inversely proportional to the length of the metal tube. This helps to create a smooth temperature profile along the growth reactor. In addition, the outer metal tube reduces the access of ambient light to the organic material, which might be important in the case of photosensitive organic compounds. Starting material is loaded into the high-temperature zone and maintained at temperature Tset, where sublimation takes place, and molecules are carried by the gas stream into the region of lower temperature. For a given concentration of evaporated molecules, defined by the temperature Tset, there is a point located downstream at a lower temperature, Tgrowth, where crystallization occurs. At this point, the crystallization rate (proportional to the density of molecular vapor) becomes slightly greater than the rate of sublimation from crystal facets kept at temperature Tgrowth. Although both crystallization and sublimation occur at a facet simultaneously, the growth prevails and free-standing crystals grow. In the region to the left of the growth zone (upstream), sublimation prevails and no growth occurs; in the region to the right (downstream), the density of molecular vapor decreases and crystallization also does not occur. In this region, only smaller molecular weight impurities condense. If the temperature Tset is maintained very low (near the sublimation threshold of the material), heavier molecular impurities do not sublime and are retained in the load zone. This creates a 2- to 3-cm wide crystallization region that is typically well separated from the original material and from the impurities. Therefore, PVT process results in the crystal growth and material purification at the same time. For better separation of the crystals from the impurities, the temperature gradient along the tube should be sufficiently small (e.g., ~5–10°C/cm). Several factors affect the morphology and the quality of the grown crystals. Important parameters are, for instance, the temperature of the sublimation zone, Tset, and the carrier gas. For each material and each furnace, the optimal set of parameters has to be determined empirically. At least one common tendency has been observed for the common compounds, such as 7,7′,8,8′-tetracyanoquinodimethane (TCNQ), tetracene, and rubrene: The slower the growth process is, the higher the field-effect mobilities obtained in the resultant OFETs are. For this reason, Tset should be adjusted close to the sublimation threshold of the material. Typical Tset resulting in very slow growth of bulky crystals with large and flat facets is: 200, 210, and 300°C for the growth of TCNQ, tetracene, and rubrene, respectively. At such conditions, typical growth duration is 40–70 hours for 100–300 mg loaded material and H2 gas flow rate ~ 100 cc/min. Large, high-purity organic crystals can be obtained by the PVT technique (Figure 2.1.2). Most of the organic crystals are shaped as thin platelets or needles. The crystal shape is controlled by the anisotropy of intermolecular interactions: For many materials, the largest crystal dimension corresponds to the direction of the strongest interactions and, presumably, the strongest overlap of π-orbitals of adjacent molecules. For this reason, the direction of the fastest growth of elongated rubrene crystals (b axis) coincides with the direction of the highest mobility of field-induced carriers (see Section 2.1.3). In platelet-like crystals, the largest natural facet typically corresponds to the a–b plane. In-plane dimensions range from a few square millimeters

8080_C002.fm Page 32 Friday, April 6, 2007 10:21 AM

32

Organic Field-Effect Transistors

Rubrene

Tetracene

FIGURE 2.1.2 Single crystals of rubrene and tetracene grown from the vapor phase.

to a square centimeter. The crystal thickness also varies over a wide range and, in most cases, can be controlled by stopping the growth process at an early stage. For example, the thickness of the tetracene crystals grown for 24 hours ranges between ~10 and ~200 μm [41]; however, it is possible to produce crystals of submicron thickness by interrupting the growth after ~10 min. According to the atomic force microscopy (AFM) studies (Figure 2.1.3) [42], the slow crystal growth proceeds by the flow of steps at a very low growth rate (≤10 μm/hour in the direction perpendicular to the a–b facet) and results in molecularly flat facets with a low density of molecular steps, separated by relatively wide (0.5–1 μm) terraces. Several ultrahigh-purity gases have been used as a carrier agent. In de Boer, Klapwijk, and Morpurgo [31], the highest mobility of tetracene-based devices, μ = 0.4 cm2/Vs, was realized with argon, whereas other groups reported slightly higher mobilities in tetracene grown in hydrogen (0.8–1 cm2/Vs) [36,43]. The best reported mobilities in rubrene have been measured in the crystals grown in ultrahigh-purity

500 nm

1 μm

FIGURE 2.1.3 AFM images of the surface of uncleaved vapor-grown rubrene (left) and TCNQ (right) crystals, showing the molecular growth steps. The height of the steps is consistent with the lattice parameter along the c axis in these crystals.

8080_C002.fm Page 33 Friday, April 6, 2007 10:21 AM

Charge Carrier Transport in Single-Crystal Organic Field-Effect Transistors

33

(UHP) H2. The carrier gas might also influence the size and morphology of the crystals. For instance, growth of tetracene in helium gas yields very thin and wide crystals, inappropriate for fabrication of free-standing OFETs, though useful for lamination on hard substrates. On the other hand, slow growth in H2 or Ar yields much thicker (bulkier) and robust tetracene crystals that can be used to create free-standing devices using parylene gate dielectric with mobilities up to ~1 cm2/Vs [43]. Similar tendencies in morphology with the carrier gas have been observed for rubrene. At present, it is still unclear how, exactly, the transport gas affects the crystal quality, and more systematic studies are required. Poorly controlled factors such as parts per million (ppm) levels of water, oxygen, and other impurities in UHP gases that might create charge traps in organic material could complicate such studies. Another poorly controlled parameter is the purity of the starting material. Empirically, different grades of material with the same nominal purity might result in crystals of quite different quality (in terms of the field-effect mobility). Normally, the density of impurities can be greatly reduced by performing several regrowth cycles, in which previously grown crystals are used as a load for the subsequent growths. However, after a gradual increase of μ with the number of purifications, the mobility saturates already after two or three cycles. This indicates that some of the impurities cannot be effectively removed from the material or new defects might be forming during the growth process. Clearly, the higher the purity of the starting material is, the fewer regrowth cycles are required. In Podzorov et al. [30], the rubrene OFETs with μ > 5 cm2/Vs have been fabricated from the “sublimed grade” rubrene (Sigma–Aldrich) after only one or two growth cycles. Besides the growth from a vapor phase, other techniques, such as Bridgman growth from a melt or crystallization from a solution, can be used to produce organic crystals. For instance, vapor-Bridgman growth from a saturated vapor in a sealed ampoule has been used to grow large tetracene crystals for TOF studies (Figure 2.1.4) [44]. Crystallization from a solution usually results in mobilities substantially lower than those obtained in vapor-grown crystals. A clear demonstration of this has been recently obtained with OFETs based on single crystals of halogenated tetracene derivatives that are soluble in common organic solvents and can also be sublimed without decomposition [45]. OFET mobilities in the vapor-grown crystals were as high as 1.6 cm2/Vs, while the solution growth resulted in devices with μ ~ 10–3 cm2/Vs. A rare example of a high-mobility solution-grown crystalline system is dithiophene-tetrathiafulvalene (DT-TTF), with field-effect mobilities of up to 1.4 cm2/Vs [46]. It is worth noting that the mobility in single-crystal devices might be substantially improved if a zone refining process is used for prepurification of the starting material. Indeed, in the time-of-flight studies of organic crystals, the highest mobilities have been obtained after multiple cycles of zone-refinement purification. This process enabled reduction of impurity concentration in the bulk down to the part-per-billion level. It has to be noted that zone refinement cannot be applied to all organic materials, since this technique requires the existence of a coherent liquid phase (i.e., the melting temperature of a substance has to be lower than the temperature of decomposition of its molecules).

8080_C002.fm Page 34 Friday, April 6, 2007 10:21 AM

34

Organic Field-Effect Transistors

FIGURE 2.1.4 A tetracene crystal grown by the vapor-Bridgman technique and used for the time-of-flight studies. (From Niemax, J. et al., Appl. Phys. Lett., 86, 122105, 2005.)

X-ray diffraction studies show that most of the PVT-grown crystals are of excellent structural quality; they are characterized by a very small mosaic spread, typically, less than 0.050 [47] (in rubrene this value has been found to be even smaller, ~0.0160) [48]. Rubrene crystallizes in an orthorhombic structure with four molecules per unit cell and the lattice parameters a = 14.44 Å, b = 7.18 Å, and c = 26.97 Å [49] (crystallographic data for several other polyacenes have been reported in Campbell et al. [50]). The crystals are usually elongated along the b axis; the largest flat facet of the crystal corresponds to the (a,b)-plane. The possibility of surface restructuring or existence of a “surface phase” on free facets of organic crystals has not been addressed yet and remains to be studied experimentally. Deviations of the surface structure from the bulk phase that might be important for the charge transport in OFETs might occur similarly to the thin-film phase in monolayer-thick pentacene films studied by grazing incidence x-ray diffraction [51,52]. Recently, it has been demonstrated that scanning tunneling microscopy (STM) could be used to study the molecular organization at the surfaces of bulk crystals in certain cases of high-mobility systems, such as rubrene [53]. Figure 2.1.5 shows the first molecular-resolution STM image of the surface of a bulk organic crystal (rubrene) at room temperature. The common problem of charging of an insulating surface with the tunneling electrons is avoided here because the high mobility of carriers in rubrene facilitates fast removal of the tunneled electrons through the crystal into the conducting substrate. The surface quality of these crystals is unprecedented; the density of surface defects is very low, resulting in a low-noise image and the opportunity to resolve individual molecules at the surface. The packing motif observed with an STM at the surface is consistent with the bulk packing obtained by crystallography. It is a “herringbone” type of structure with a stack forming along the b axis.

8080_C002.fm Page 35 Friday, April 6, 2007 10:21 AM

Charge Carrier Transport in Single-Crystal Organic Field-Effect Transistors

35

a b a

b 2 mm

FIGURE 2.1.5 Scanning tunneling microscope images of a–b facet of a thick, as-grown rubrene crystal. The herringbone molecular organization at the surface, consistent with the bulk structure (shown in the lower left corner), is evident. (From Menard, E. et al., Adv. Mater., 18, 1552, 2006.)

Fabrication of field-effect structures on the surface of organic crystals is a challenge because the conventional processes of thin-film technology (such as sputtering, photolithography, etc.) introduce a high density of defects on vulnerable organic surfaces. For this reason, the first single-crystal OFETs have been realized only recently, after development of the two innovative fabrication techniques briefly described next. The first technique is based on the use of an unconventional gate dielectric: thin polymeric film of parylene, which can be deposited from a vapor phase on the surface of organic crystals at room temperature, producing a defect-free semiconductor–dielectric interface [29]. Conformal parylene coating is a well developed technology used commercially in electronic packaging applications [54]. A homebuilt setup for parylene deposition is depicted in Figure 2.1.6 (commercially available parylene coaters are not recommended for this research purpose because of their large volume and high cost). The reactor consists of a 20-mm ID quartz tube blocked at one end and a twozone furnace for sublimation and pyrolysis of the commercially available parylene dimers. The quartz tube extends from the high-T (700°C) section of the furnace by about 40 cm to the right; the sample(s) with prefabricated contacts and leads are placed in this portion of the tube, which is then connected to the mechanical pump through a liquid N2 trap. After evacuating the reactor to approximately 10–2 torr, the temperatures in the sublimation and pyrolysis zones are set to 100 and 700°C, correspondingly. Parylene dimers, sublimed at 100°C, split into monomers at 700°C and polymerize as they enter the room temperature section of the tube, producing a clear pinhole-free insulating coating on the sample’s surface.

8080_C002.fm Page 36 Friday, April 6, 2007 10:21 AM

36

Organic Field-Effect Transistors

CH2

CH2

CH2

CH2

CH2

2CH2

CH2

CH2 n

Di-para-xylylene (dimer)

Sublimation at 100 °C

Para-xylylene (monomer)

Pyrolysis at 700 °C

Poly(para-xylylene) (polymer)

Polymerization To liquid–N2 at 25 °C tap & mech. pump

Dimer powder

Sample

FIGURE 2.1.6 Reactions involved in parylene deposition process (top): sublimation of dimers at ~100°C, splitting into monomers at ~700°C, and polymerization at room temperature. Inexpensive system for parylene deposition (bottom) consists of a two-zone tube furnace and a 20-mm ID quartz tube containing parylene dimer powder and connected to a one-stage mechanical pump through a liquid nitrogen trap. A sample with prefabricated contacts and attached wire leads is placed in the tube at about 30 cm from the furnace.

The advantages of this technique are the following: • • • • •

•

High-energy charged particles, inherent to plasma-based deposition techniques (e.g., sputtering) and detrimental for the organic surfaces, are avoided. The sample is maintained at room temperature throughout the entire process. A high vacuum is not required, which, in combination with cheap parylene precursors, makes this a very low cost technique. The deposition process is fast: Growth of a 1-μm thick film lasts ~20 min. The physical properties of parylene films are remarkable: Parylene is a very good insulator with an electrical breakdown strength of up to 10 MV/cm, superior chemical stability, and high optical clarity. The coating is truly conformal, which allows working with crystals that have sharp features on their surfaces (e.g., steps, edges, etc.), without having problems with shorts.

The conformal properties of parylene coating are especially important in devices with colloidal graphite contacts that have rough surfaces. The parylene coating is the only technique available to date for the fabrication of free-standing single crystal OFETs. In comparison with the laminated devices, this has several advantages, such as elimination of substrate-related strains and a possibility to perform studies of photoinduced effects in OFETs by illuminating the conduction channel through the transparent parylene dielectric and a (semi)transparent gate electrode [55]. The

8080_C002.fm Page 37 Friday, April 6, 2007 10:21 AM

Charge Carrier Transport in Single-Crystal Organic Field-Effect Transistors

37

OFETs with parylene dielectric are very stable. For example, the characteristics of rubrene/parylene transistors remained unchanged after storing the devices for more than two years in air and in the dark. For the free-standing devices, deposition of metal contacts directly on the surface of organic crystal is necessary. While painting colloidal graphite contacts (e.g., Aquadag Colloidal Graphite, Ted Pella, Inc.) on the crystal surface works well and results in low contact resistance, it is difficult to create complex contact geometries and well-defined features with this simple technique. An alternative high-vacuum thermal evaporation of metals through a shadow mask is challenging due to several factors related to generation of defects at the organic surfaces as a result of (1) infrared radiation from evaporation sources; (2) contamination of the channel area with metal atoms able to penetrate under the shadow mask; and (3) interaction of organic surfaces with free radicals produced by hot filaments and high-vacuum gauges (the gauge effect) [56]. Nevertheless, devices with evaporated contacts have been successfully fabricated in Podzorov et al. [30] by using an optimized deposition chamber — a technique that might be useful for more complex contact geometries in OFETs, such as four-probe or Hall geometry. In the second technique of single-crystal OFET fabrication, the transistor circuitry is prefabricated by conventional microfabrication (lithography) methods on a substrate (this structure can be called a “stamp”), and the organic single crystal is subsequently laminated to it. This technique eliminates the need for deposition of metal contacts and dielectrics directly onto organic crystals. Hard inorganic (e.g., Si) and flexible elastomeric (polydimethylsiloxane = PDMS) stamps have been used for this purpose. In the first case, a heavily doped Si wafer with a thermal SiO2 plays the role of an insulated gate electrode [31,33]. After the deposition of gold contacts, a thin organic crystal can be laminated to such a stamp owing to van der Waals attraction forces. Similarly, field-effect transistor (FET) structure can be fabricated using PDMS substrates and spin-coated PDMS films [35]. The elastomeric stamps compare favorably with the Si stamps in two respects. First, the slightly conformal properties of PDMS enable establishing a good contact even with crystals that are not perfectly flat. Conversely, the use of hard Si stamps is restricted to perfectly flat crystals or to very thin and “bendable” crystals that could conform to hard substrate. Second, for the robust and bulky crystals such as rubrene, the PDMS stamps provide a unique opportunity to re-establish the contact many times without breaking the crystal and without degradation of the crystal’s surface. However, the achievable density of field-induced charges is typically greater in the Si-based stamps, especially if these stamps utilize high-ε gate insulators [23,57]. This is important for the exploration of the regime of high carrier densities, in which novel electronic phases might emerge (see Section 2.1.3.7). Even though the lamination of crystals on prefabricated substrates enables a “low-impact” probing of charge transport on organic surfaces, this impact may still be too strong for chemically reactive organic materials (e.g., a strong electron acceptor TCNQ). To minimize these effects and to preserve the pristine surface of organic crystals, modification of the PDMS stamping technique has been recently introduced that avoids these complications simply by eliminating the direct contact between the crystal and the gate dielectric [39]. The idea of these so-called vacuum-

8080_C002.fm Page 38 Friday, April 6, 2007 10:21 AM

38

Organic Field-Effect Transistors

PDMS stamp

Pr

Peel back; ﬂip over; coat with Ti/Au; laminate crystal

Si

Single Crystal

Ti/Au

D

S G

FIGURE 2.1.7 Casting and curing procedures for fabrication of the “air-gap” transistor stamps. The recessed gate electrode is separated from the conductive channel by a micronsize gap. (From Menard, E. et al., Adv. Mater., 16, 2097, 2004.)

gap stamps is illustrated in Figure 2.1.7. In these devices, the conventional dielectric is replaced by a micron-size gap between the gate electrode and the surface of the organic semiconductor. A thin layer of gas (e.g., air) or vacuum between the bottom surface of the crystal and the recessed gate electrode plays the role of gate dielectric. This approach eliminates surface defects introduced in the process of lamination and enables studies of the effect of different gases and other environmental agents on the conduction channel in OFETs [56].

2.1.3 CHARGE TRANSPORT ON THE SURFACE OF ORGANIC SINGLE CRYSTALS In this section, after a brief introduction of the OFET operation principles, we outline the main signatures of the intrinsic polaronic transport observed in the experiments with single-crystal OFETs. We compare them to the results of TOF and space-charge limited current (SCLC) experiments that probe charge transport in the bulk.

2.1.3.1 BASIC FET OPERATION Contemporary OFETs are based on undoped organic semiconductors, and mobile charges in these devices must be injected from the metallic contacts. These devices can potentially operate in the electron- and the hole-accumulation modes, depending on the polarity of the gate voltage (the so-called ambipolar operation). Often, however, the injection barrier at the contact or the field-effect threshold for either n- or

8080_C002.fm Page 39 Friday, April 6, 2007 10:21 AM

Charge Carrier Transport in Single-Crystal Organic Field-Effect Transistors

39

Vg EVAC

G

S ϕM

VSD

D ISD

EHOMO(Vg < 0) EF ϕB

Metal

EHOMO(Vg = 0)

Semiconductor

FIGURE 2.1.8 Schematic energy diagram of a metal–organic interface (the contact). Evac is the vacuum energy level, EF is the Fermi energy of metal, EHOMO is the energy of the band edge of the semiconductor. The inset shows a two-probe OFET circuitry: S and D are the source and drain contacts, G is the gate electrode; Vg and VSD are the gate and source-drain voltages, respectively, and ISD is the source-drain current.

p-type conductivity is so large that an FET operates in a unipolar mode. For this reason, we will mainly discuss p-type conductivity, which is more commonly observed in OFETs. We will start the discussion with charge injection from contacts. An energy diagram of a hole injecting metal-semiconductor contact and a generic field-effect transistor circuit are schematically shown in Figure 2.1.8. The hole injection occurs through the interfacial Schottky barrier of height ϕB; the formation of the barrier is a complex process that depends on the metal work function ϕM, the ionization energy of the semiconductor, and the interfacial dipole moment formed due to a charge transfer at the interface. For a comprehensive review of the energetics of metal– organic interfaces, see, for example, the paper by Cahen et al. [58]. While the maximum height of the barrier, ϕB, remains fixed due to the pinning of energy levels at the interface, its width can be modified by an external electric field associated with either VSD or Vg. Figure 2.1.8 shows that when a negative Vg is applied, the effective width of the Schottky barrier for hole injection decreases. This results in a decrease of the contact resistance (RC), which depends on the barrier height ϕB, its effective thickness, and temperature. The “triangular” shape of the Schottky barrier allows the carrier injection via thermally activated excitation above the barrier and via tunneling under the barrier (the latter process does not require thermal excitation, but it is limited by Vgdependent barrier width). Both processes are possible, and the resultant injection mechanism, called thermionic emission, typically causes an exponentially fast increase of the contact resistance RC with lowering T and a strong dependence of RC on the gate voltage. The contact resistance enters the equations of OFET operation because the source-drain circuit is represented by two resistors connected in series — the contact

8080_C002.fm Page 40 Friday, April 6, 2007 10:21 AM

40

Organic Field-Effect Transistors

resistance, RC, and the channel resistance, RCH — so that the total source-drain resistance is RSD ≡ VSD/ISD = RC + RCH. The Schottky contact resistance in OFETs is typically high; in many cases, RC ≥ RCH, especially in short-channel TFTs, and the operation of such devices is contact limited. This is also the case for shortchannel single-crystal OFETs. However, if the channel is long enough (0.2–5 mm), rubrene and tetracene devices with either graphite or laminated gold/PDMS contacts are not contact limited at room temperature. The devices with evaporated silver or gold contacts generally have higher contact resistances. Nevertheless, RC typically decreases with VSD, and at large enough VSD and Vg < 0, even the devices with evaporated contacts are not dominated by contacts. However, it is important to be able to directly measure RC and RCH independently in each individual OFET in the entire temperature range of interest, because of a strong T-dependence of the contact resistance. In the prior studies, contact resistance was estimated by measurements of twoprobe OFETs and fitting the data with the Schottky model at different T and Vg [59] and by performing the channel length scaling analysis [60,61]. While these methods provide useful information about the contacts of a particular system, they do not allow for the direct and model-independent measurement of the contact resistance in each individual device. Recently introduced OFETs with four-probe contact geometry (source, drain, and two voltage probes in the channel) can be used to address this problem [29,30]. Before describing the four-probe measurements, let us introduce the operation of a conventional two-probe OFET, assuming that the contact resistance is negligible compared to the channel resistance. (This is practically valid for some cases of single-crystal OFETs with long channels at room temperature.) With an increase of the gate voltage |Vg| towards the threshold value |Vgth|, the carriers injected from the metallic contacts fill localized in-gap states of the organic semiconductor, associated with impurities and defects in the channel, whose energy is separated from the edge of the HOMO band by more than a few kBT (the deep traps; see Figure 2.1.9) (this simplified model assumes the existence of the HOMO band; this assumption may be violated at high temperatures) [12,62]. As the result, the Fermi level at the organic surface, E F, initially positioned within the HOMO–LUMO gap, approaches the edge of the HOMO band, EHOMO, which corresponds to the zero energy in Figure 2.1.9. As soon as EF – EHOMO becomes smaller than ~kBT, the OFET’s conductance increases by several orders of magnitude due to the thermal excitation of the carriers from the localized states into the HOMO band. As a result, a conduction channel is formed at the interface between the semiconductor and the gate dielectric. Overall device operation depends, to a large extent, on the energetics of the semiconductor bands and metal contacts, and therefore studies of the electronic structure of molecular interfaces are important [58]. Figure 2.1.10 shows the transconductance characteristics (i.e., the dependence of the source-drain current on the gate voltage, ISD(Vg), measured at a constant sourcedrain voltage, VSD) and ISD(VSD) characteristics typical for the p-type rubrene singlecrystal OFETs [29,30,39]. The channel conductance per square,

8080_C002.fm Page 41 Friday, April 6, 2007 10:21 AM

Charge Carrier Transport in Single-Crystal Organic Field-Effect Transistors

41

ε(eV) LUMO

2.2

~0.1 Deep traps Shallow traps

Few kBT

0 HOMO ν(ε)

FIGURE 2.1.9 The schematic diagram of the energy distribution of localized electronic states in the energy gap between the HOMO and LUMO bands in the rubrene single-crystal OFETs. (From Podzorov, V. et al., Phys. Rev. Lett., 93, 086602, 2004.)

μb = 12.3 cm2/Vs

ISD(μA)

3

VSD= 5 V

2

1 μa = 5.2 cm2/Vs

0

−50

−25 VG(V)

2.0

0

VG= −30 V

−ISD(μA)

1.5 1.0

−25 V

0.5

−20 V −15 V

0.0

−10 V 0

10

20 −VSD (V)

30

FIGURE 2.1.10 The transconductance ISD(Vg) (the upper panel) and ISD(VSD) (the lower panel) characteristics of rubrene single-crystal OFET (see, for example, Podzorov et al. [30] and Menard et al. [39]).

8080_C002.fm Page 42 Friday, April 6, 2007 10:21 AM

42

Organic Field-Effect Transistors

σ≡

I SD L , VSD W

increases linearly with Vg at |Vg| > |Vgth|. (Here, L and W are the length and width of the conduction channel, respectively.) This indicates that the carrier mobility [63]

μ≡

σ ⎛ 1 ⎞ ⎛ dI SD ⎞ L = en ⎜⎝ C iVSD ⎟⎠ ⎜⎝ dVg ⎟⎠ W

(2.1.1)

does not depend on the density of carriers field-induced above the threshold

(

)

n = C i Vg − Vgth / e

(2.1.2)

Here, Ci is the capacitance per unit area between the gate electrode and the conduction channel. A density-independent μ has been observed in devices based on single crystals of rubrene [30,35,36], pentacene [33,34], tetracene [31], and TCNQ [39]. This important characteristic of single-crystal OFETs contrasts sharply with a strongly Vg-dependent mobility observed in organic TFTs [64] and α-Si:H FETs [65]. In the latter case, the density of localized states within the gap is so high that the Fermi level remains in the gap even at high |Vg| values. The observation of Vg-independent mobility in single-crystal OFETs suggests that the charge transport in these structures does not require thermal activation to the mobility edge and the mobile field-induced carriers occupy energy states within the HOMO band. This is consistent with an increase of the mobility with cooling observed in high-quality single-crystal OFETs (see Section 2.1.3.4). (For comparison, μ decreases exponentially with lowering temperature in organic and α-Si:H TFTs.) The pronounced difference in the Vg- and T-dependences of the mobility in these two types of devices clearly indicates that the theoretical models developed for the charge transport in α-Si:H and organic TFTs [65] are not applicable to singlecrystal OFETs. In the cases when contact resistance is not negligible, four-probe OFETs are used to measure the channel and the contact resistances independently. In the fourprobe OFET geometry (Figure 2.1.11), in addition to ISD, voltage between a couple of extra probes located in the middle of the channel, V4w, can be measured as a function of Vg, VSD, and T. Gate voltage dependences of ISD and V4w for a typical four-probe OFET are shown in the upper panel of Figure 2.1.12. In the four-probe geometry, conductivity of the section of the channel between the voltage probes per square is: σ≡

I SD D V4w W

8080_C002.fm Page 43 Friday, April 6, 2007 10:21 AM

Charge Carrier Transport in Single-Crystal Organic Field-Effect Transistors

43

D

Vg

W L S

V4w

G

D

VSD ISD

FIGURE 2.1.11 Four-probe field-effect transistor with measurement circuitry: V4w is the voltage measured between the two voltage probes located in the middle of the channel. The inset shows the channel geometry: L and W are the channel length and width; D is the distance between the voltage probes.

Note that the voltage V4w does not necessarily remain constant when Vg or T is varied because the total VSD voltage applied to the device is distributed between the channel resistance, RCH, and the contact resistance, RC, both of which vary with Vg and T. Using the relationship σ = enμ, and n from Equation (2.1.2), we obtain for the contact-resistance-corrected channel mobility, μ4w: ⎛ 1 ⎞ ⎛ d ( I SD / V4w ) ⎞ D μ 4w = ⎜ ⎟W ⎜ dVg ⎝ C i ⎟⎠ ⎝ ⎠

(2.1.3)

and for the contact resistance RC: RC =

VSD L V4w − I SD D I SD

(2.1.4)

Typical Vg- and VSD-dependences of the contact resistance, normalized to the channel width, RCW, are shown in the lower panel of Figure 2.1.12. In agreement with the Schottky model, RC for a p-type device decreases with a positive VSD applied to the hole-injecting source contact and with a negative Vg applied to the gate. Interestingly, the relatively large magnitude of the contact resistance in Figure 2.1.12 (≥100 kΩ⋅cm) can be greatly reduced down to 1–2 kΩ⋅cm by treating the gold contacts with trifluoromethylbenzenethiol before the crystal lamination or by using nickel instead of gold, which has been recently reported to result in a remarkably low contact resistance ~ 0.1–0.4 kΩ⋅cm [61]. For several important applications in plastic optoelectronics, including the possibility of electrically pumped organic lasers, it would be very important to achieve an ambipolar operation in OFETs, with high electron and hole mobilities. Gatecontrolled electroluminescence from organic small-molecule thin-film transistors

8080_C002.fm Page 44 Friday, April 6, 2007 10:21 AM

44

Organic Field-Effect Transistors

2 10

1

5

V4w

2

ISD 0

0

400 RCW (kΩcm)

RCW (kΩcm)

107

105

V4w (V)

ISD (μA)

7

300 200 100 10

20 30 VSD (V)

103

101

−75

−50

−25 Vg (V)

40

0

25

FIGURE 2.1.12 Four-probe OFET characteristics, ISD(Vg) and V4w(Vg) (top), and the corresponding contact resistance RCW(Vg) (bottom). The inset shows the dependence of the contact resistance on VSD. (From Sundar V. C. et al., Science, 303, 1644, 2004.)

and, more recently, from single-crystal OFETs based on thiophene/phenylene cooligomers has been observed (see, for example, Nakamura at al. [66] and references therein). However, because these devices have not been optimized yet, hole and electron currents were not balanced, and only unipolar (p-type) electrical characteristics have been observed. Interestingly, Vg-controlled electroluminescence and ambipolar characteristics have been recently observed in conjugated polymer OFETs [67,68], which indicates a balanced electron and hole injection. However, low hole and electron mobilities (~10–3 cm2/Vs), typical for polymer semiconductors, limit the channel current and therefore may present a serious problem for realization of electrically pumped polymer lasers. For this reason, ordered small-molecule organic semiconductors with higher mobilities are very promising for research in this direction.

8080_C002.fm Page 45 Friday, April 6, 2007 10:21 AM

4-probe channel conductivity, σ = IDS/V4W (μS)

Charge Carrier Transport in Single-Crystal Organic Field-Effect Transistors

0.05

45

With “vac-gap” stamp μa4w ~ 1.6 cm2/Vs Ci = 0.19 nF/cm2

0.04

L

0.03

0.02

D

0.01

0.00

−20

−10

0

10 20 Vgate (V)

30

40

50

60

FIGURE 2.1.13 Channel conductivity along the a-axis of TCNQ single crystal measured in the “vacuum-gap” OFET. The mobility of n-type carriers is 1.6 cm2/Vs. (From Menard, E. et al., Adv. Mater., 16, 2097, 2004.)

Most of the small-molecule organic FETs operate in the p-type mode, and examples of n-type operation with high mobility are rare [39,69]. This “asymmetry” between n- and p-type carriers is due to several factors: the HOMO bandwidth is typically larger than the LUMO bandwidth [17], a stronger trapping of n-type polarons [19], and a larger Schottky barrier for electron injection into organic semiconductors from the most commonly used high work function metals. Figure 2.1.13 illustrates the n-type operation in a single-crystal TCNQ transistor. The surface of TCNQ, a semiconductor with a very high electron affinity, can be easily damaged (e.g., a direct contact of the crystal with PDMS dielectric in the contact stamps, such as those used in Sundar et al. [35], results in a very poor transistor performance with electron mobilities ~ (2–3)·10–3 cm2/Vs). The “air-gap” PDMS stamps [39] help to solve the problem. The observed carrier mobility ~ 1.6 cm2/Vs in the linear regime is significantly higher than in most of the n-channel organic TFTs. This value, however, is still limited by trapping (see Section. 2.1.3.2); more work is required to approach the fundamental limit of performance of n-type OFETs. In practice, realization of high-mobility ambipolar operation is a challenge because two difficult problems must be solved simultaneously: (1) the density of both n- and p-type traps should be minimized at organic/dielectric interfaces; and (2) an effective injection of both n- and p-type carriers from the contacts into the organic semiconductor must be realized. Among inorganic FETs, only devices based on carbon nanotubes [70] and single crystals of transition metal dichalcogenides (e.g., WSe2 and MoSe2) [71] demonstrated high-mobility ambipolar operation. The number of organic materials in which the ambipolar operation has been demonstrated is limited as well [67,69,72,73]. The organic single crystals, with their intrinsically

8080_C002.fm Page 46 Friday, April 6, 2007 10:21 AM

46

Organic Field-Effect Transistors

low density of traps, offer a unique opportunity to realize the ambipolar operation with a relatively high mobility of both types of carriers. Ambipolar operation has been recently observed in the single-crystal OFETs based on metal phthalocyanines (MPc), namely, FePc and CuPc [74] (Figure 2.1.14) and rubrene [75]. Because of the reduced density of electron traps at the interface and a relatively small HOMO–LUMO gap in the case of MPc (<1.5 eV), both electrons and holes can be injected in the organic crystal from the contacts. Interestingly, in these ambipolar single-crystal OFETs, both source and drain contacts were made of the same high work function metal (gold in MPc FETs and silver in rubrene FETs). Note that the CuPc-based TFTs with gold contacts demonstrate only unipolar p-type operation (presumably because of a high density of traps for electrons in thin films) [76]. Although the performance of these single-crystal OFETs in the n-type regime is still dominated by traps, the mobilities observed for holes and electrons (0.3 and 0.03 cm2/Vs in MPc FETs, respectively, and 1.8 and 0.011 cm2/Vs in rubrene FETs, respectively) compare favorably with that in the corresponding thin-film transistors employing Ca electrodes [72]. This indicates a good potential of the single-crystal OFETs for the studies of ambipolar charge injection, transport, and recombination processes. In general, the performance of FET devices is characterized by many parameters, including the mobility, the threshold voltage, the ON/OFF ratio, and the subthreshold slope [32,63]. Next, we will focus on the first two parameters — the mobility, μ, and the threshold voltage, Vgth — that are the most relevant to the physics of charge transport on the surface of organic semiconductors. Note that, with respect to other parameters (e.g., the subthreshold slope), single-crystal OFETs also compare favorably with the best organic and inorganic thin-film transistors (see, for example, Podzorov et al. [30]).

2.1.3.2 THE MULTIPLE TRAP-AND-RELEASE MODEL Although the density of defects in the conduction channel of single-crystal OFETs is significantly lower than in organic TFTs, the defects are still present. These defects create localized electronic states in the HOMO–LUMO gap schematically shown in Figure 2.1.9 (the electronic defects will be discussed in more detail in Section 2.1.4). The effect of these states on charge transport depends on their energy. If the energy of a localized state is separated from the mobility edge (or, in other words, from the edge of HOMO(LUMO) for p(n)-type carriers) by more than a few kBT, the state acts as a deep trap: once trapped in a deep trap, the charge cannot be released by thermal excitations. For the pristine surface of rubrene single crystals at room temperature, for instance, the density of deep traps can be as low as 1010 cm–2. Conversely, trap states with energies within a few kBT of the mobility edge (shallow traps) are characterized by a finite trapping time; after being trapped for a characteristic time τtr, a polaron can be thermally activated and released to the band. At a phenomenological level, the effect of shallow traps on the channel conductivity can be described by the multiple trap and release (MTR) model [27,77]. According to this model, which helps to illustrate the distinction between “intrinsic”

8080_C002.fm Page 47 Friday, April 6, 2007 10:21 AM

Charge Carrier Transport in Single-Crystal Organic Field-Effect Transistors

47

10−7 FePc

S

ISD (A)

10−8

D

FePc

b

10−9

10−10 VSD = −25V 10−11

−20

−10

0

10

20

VG (V) 10−6 CuPc N 10−7

N

ISD (A)

N

N M

N

10−8

N N

N 10−9

10−10 VSD = −25 V 10−11 −30

−20

−10

0 VG (V)

10

20

FIGURE 2.1.14 The transconductance characteristics of FePc and CuPc single-crystal FETs with gold electrodes. The hole mobilities reach ~0.3 cm2/Vs for both FePc and CuPc; the electron mobilities are 3⋅10–2 cm2/Vs and 10–3 cm2/Vs, respectively. (From de Boer, R. W. I. et al., Appl. Phys. Lett., 86, 262109, 2005.)

and “trap-dominated” transport, not all the charges field induced above the threshold (at |Vg| > |Vgth|) contribute to the current flow at any given moment of time. Some of the mobile charges can be momentarily trapped by shallow traps; the number of these charges depends on the density of shallow traps and temperature. Within the MTR model, the effect of trapping can be described using two approaches. In the first approach, one can assume that all carriers field induced above

8080_C002.fm Page 48 Friday, April 6, 2007 10:21 AM

48

Organic Field-Effect Transistors

the threshold, n (see Equation 2.1.2), contribute to the current flow at any moment of time, but their effective mobility μeff is reduced in comparison with its intrinsic, trap-free value μ0:

μ eff = μ 0 (T )

τ(T ) τ(T ) + τtr (T )

(2.1.5)

Here, τtr(T) is the average trapping time on shallow traps and τ(T) is the average time that a polaron spends diffusively traveling between the consecutive trapping events. In the alternative second approach, one can assume that only a fraction of the carriers field induced above the threshold voltage are moving at any given moment of time:

neff = n

τ(T ) τ(T ) + τtr (T )

(2.1.6)

However, these charges are moving with the intrinsic trap-free mobility μ0. These two approaches are equivalent for describing the channel conductivity σ = enμ, which depends only on the product of n and μ. The distinction between these approaches becomes clear in the Hall effect measurements, in which the density and the intrinsic mobility of truly mobile carriers can be determined independently (see Section 2.1.3.4). According to Equation 2.1.5, the intrinsic regime of conduction is realized when τ >> τtr. In this case, the dependence σ(T) reflects the temperature dependence of intrinsic mobility μ0(T). In the opposite limit τ << τtr, the transport is dominated by shallow trapping processes. It will be shown below that the intrinsic regime is characterized by a large anisotropy of the charge transport, an increase of the mobility with decreasing temperature, and a conventional “nonactivated” Hall conductivity.

2.1.3.3 ANISOTROPY

OF THE

MOBILITY

Polyacenes typically form crystals with a herringbone packing of molecules (the molecular packing in rubrene crystals is shown in Figure 2.1.15). Transfer integrals between the adjacent molecules in these crystals vary significantly depending on the crystallographic direction [78–80]. This leads to a strong anisotropy of transport properties of organic crystals, which has been well documented in the TOF experiments [19]. However, prior to the development of single-crystal OFETs, the anisotropy had never been observed in the field-induced transport on the surface of organic semiconductors. Several types of single-crystal OFETs based on rubrene demonstrate anisotropy of surface conductivity [35,38,48]. In rubrene devices based on PDMS stamps, the mobility along the crystallographic b axis exceeds the mobility along the a axis by a factor of ~ three (Figs. 2.1.10 and 2.1.16) [35]. Similar μ0b/μ0a ratio has been observed for rubrene transistors with parylene gate dielectric [48]. A clear correlation

8080_C002.fm Page 49 Friday, April 6, 2007 10:21 AM

Charge Carrier Transport in Single-Crystal Organic Field-Effect Transistors

49

b

c a

0

b

a c a

FIGURE 2.1.15 Packing of molecules in rubrene crystals. The crystal facets used for the FET fabrication correspond to the a–b crystallographic plane; the largest μ is observed for the charge transport along the b axis. (From Sundar V. C. et al., Science, 303, 1644, 2004.)

between the mobility and molecular packing has also been found recently in a family of tetrathiafulvalene derivatives [81]. A small density of shallow traps in single-crystal rubrene OFETs facilitated the observation of mobility anisotropy. However, even in these devices the anisotropy of μeff vanishes at lower temperatures (Figure 2.1.17), where charge transport becomes trap dominated. To explain vanishing of the mobility anisotropy within the MTR model, one should take into account that τx, the time of travel between shallow traps along a certain crystallographic direction x, is inversely proportional to the intrinsic mobility along this direction, μ0x. In the trap-dominated regime (τ << τtr), Equation 2.1.5 is reduced to μ eff = μ 0x

τx τtr

and μeff becomes independent on the crystallographic direction. The higher the shallow trap density is, the narrower is the temperature range where the mobility anisotropy can be observed. The observed anisotropy of μ in rubrene can be explained qualitatively on the basis of the molecular packing in these crystals (Figure 2.1.15). Due to the cofacial orientation of molecules in the stacks along the b axis, the charge motion along the stacks is facilitated in comparison with that in the perpendicular direction. Recent calculations of the band structure of rubrene based on the methods of quantum chemistry confirmed that the value of transfer integrals reaches a maximum for the b axis [17]. For the quantitative description of the mobility anisotropy and its

8080_C002.fm Page 50 Friday, April 6, 2007 10:21 AM

50

Organic Field-Effect Transistors

Eﬀective mobility (cm2/Vs)

90 4 2 0

360

180

2 4 270

Channel conductivity (μS)

5 4 3 2

W⋅Rcontact(Ω cm)

(a) 106

D W

L 105

−80 −60 −40 −20

0

1 0

b direction a direction −80 −60 −40 −20 0 Gate Voltage (V)

20

(b)

FIGURE 2.1.16 (a) Polar plot of the mobility at the rubrene a–b surface (the angle is measured between the b axis and the direction of current flow). (b) The four-probe measurements of the channel conductivity and the contact resistance (inset) as a function of Vg along the b and a axes; the μ values measured along the b and a axes are 15.4 cm2/Vs and 4.4 cm2/Vs, respectively. (From Sundar V. C. et al., Science, 303, 1644, 2004.)

temperature dependence, ab initio calculations that take into account intra- and intermolecular vibrations are needed; currently, such calculations are available only for bulk conduction in crystals of linear polyacenes (naphthalene, anthracene, and tetracene) [12,13].

2.1.3.4 LONGITUDINAL AND HALL CONDUCTIVITY RUBRENE OFETS

IN

The intrinsic and trap-dominated transport regimes in single-crystal OFETs can be identified by measuring the conductivity over a wide temperature range. The

8080_C002.fm Page 51 Friday, April 6, 2007 10:21 AM

Charge Carrier Transport in Single-Crystal Organic Field-Effect Transistors

51

30

μ (cm2/Vs)

20

10

b-axis 5

100

a-axis

150

200 T (K)

250

300

FIGURE 2.1.17 Temperature dependence of the mobility in rubrene OFET extracted from the four-probe measurements of the conductivity along the a and b axes. (From Podzorov, V. et al., Phys. Rev. Lett., 93, 086602, 2004.)

longitudinal conductivity of rubrene single-crystal OFETs, measured at different T as a function of Vg, is shown in Figure 2.1.18. Assuming that all charges, n, field induced above the threshold are moving with a nonzero effective velocity at any moment of time, one can attribute the observed T-variations of the slope dσ /dVg to the nonmonotonic temperature dependence of the effective mobility μeff = σ/en (Figure 2.1.17). 0.16

Source Gate

0.12

σ (μS)

Source

Drain

Rubrene

t

PDMS D L

Drain

300 K

W

280 K

0.08

240 K 180 K 160 K

0.04

146 K 120 K 0.00 −40

−30

−20 −10 Vg (V)

0

FIGURE 2.1.18 Sheet conductivity σ of the “vacuum-gap” rubrene single-crystal OFET, measured as a function of Vg at different T using the four-probe technique. (From Podzorov, V. et al., Phys. Rev. Lett., 93, 086602, 2004.)

8080_C002.fm Page 52 Friday, April 6, 2007 10:21 AM

52

Organic Field-Effect Transistors

In the past, two signatures of intrinsic band-like polaronic conduction have been observed in the bulk of ultrahigh-purity organic crystals in TOF measurements [19]. These signatures are the mobility increase with cooling below 300 K and a pronounced anisotropy of μ. The increase of μeff with cooling and the mobility anisotropy observed in OFETs in the temperature range 200–300 K (Figure 2.1.17) indicate that this regime corresponds to the intrinsic (i.e., not trap dominated) polaronic transport. (The effective mobility μeff(T) in this temperature range coincides with the intrinsic mobility of polarons μ0(T).) The rapid drop of μeff at T ≤ 160 K and the vanishing of the anisotropy indicate a crossover to the trap-dominated regime, where μeff < μ0. Note that, for the device whose conductivity is shown in Figure 2.1.18, the density of shallow traps, Ntrsh, which can be estimated from the linear temperature dependence of the threshold voltage, Vgth(T), is relatively low (Ntrsh ~ 1010 cm–2). For devices with higher Ntrsh, the crossover between the intrinsic and trap-dominated regimes occurs at higher temperatures. As a result, devices with Ntrsh ≥ 1011 cm–2 demonstrate an activated temperature dependence of μeff even at room temperature, and the effective mobility is smaller than μ0 [29]. A better understanding of the intrinsic transport can be gained from measurements of the transverse (Hall) conductivity. In the Hall effect, a voltage across the conducting channel is induced by an application of a magnetic field (B) perpendicular to the channel (Figure 2.1.19). The origin of the Hall effect in a conventional band semiconductor is the Lorentz force, acting on a charge carrier propagating along the channel in a transverse magnetic field. This force is proportional to the microscopic velocity of the carrier (when it is moving between shallow traps). For trapped carriers, the Lorentz force is zero, and therefore trapped carriers do not contribute 6 0.66

4

0.65

0

B (T)

VH (V)

2

–2

0.64

1

2

S

–4

D –6

3 0.63

0

1

2

3 t (h)

4

5

6

FIGURE 2.1.19 Hall effect in rubrene OFETs (the inset shows the contact geometry). Symbols: the Hall voltage, VH, measured between the contacts 1 and 3; line: magnetic field, B, applied perpendicular to the channel. (From Podzorov, V. et al., Phys. Rev. Lett., 95, 226601, 2005.)

8080_C002.fm Page 53 Friday, April 6, 2007 10:21 AM

Charge Carrier Transport in Single-Crystal Organic Field-Effect Transistors

53

to the Hall signal. Hall measurements have the following important advantages compared to the longitudinal conductivity measurements: •

• •

The mobility determined from the Hall effect is the intrinsic (i.e., trapindependent) mobility μ0(T), even if a considerable trapping occurs in the channel. Hall measurements allow for an independent determination of the density of instantaneously mobile charges and their intrinsic mobility μ0. Observation of the classical (i.e., band) Hall effect strongly suggests a “bandlike” motion of carriers as a fundamental conduction mechanism in smallmolecule organic semiconductors, as opposed to the incoherent hopping. Indeed, in the hopping regime, conduction occurs via tunneling of charge carriers from site to site, and microscopic velocity cannot be introduced.

From the Hall data, the density of mobile carriers in the OFET’s channel can be directly determined for the first time, without the assumptions regarding the gatechannel capacitance Ci. Hall effect studies in other organic semiconductors (e.g., pentacene and tetracene) are highly desirable because most of the OFETs based on those materials still operate in the trap-dominated regime and the intrinsic mobilities at the surface of these semiconductors, μ0(T), are unknown. The Hall measurements, however, are complicated by a very high sheet resistance of the conduction channel typical for OFETs, which often exceeds 10 MΩ/square. For this reason, the first demonstration of the Hall effect appeared only recently, due to the high carrier mobility in rubrene [82,83]. The quantity nH determined in the Hall measurements is the density of charges that are moving at any given moment of time (i.e., nH coincides with neff given by Equation 2.1.6). The charges temporarily trapped in shallow traps do not contribute to the Hall voltage because the Lorentz force, proportional to the carrier velocity, is zero for these charges. Figure 2.1.20 shows the temperature dependence of nH normalized to the density of charge carriers, n, field-induced in the channel above the threshold (determined using the FET capacitance Ci according to Equation 2.1.2). The ratio nH/n is close to unity in the high-temperature (intrinsic) regime (see also Takeya et al. [83]) and decreases with cooling in the trap-dominated regime. The mobility μH determined from the Hall effect measurements is the intrinsic, trap-free mobility μ0, even if the charge transport is significantly affected by trapping. Again, this reflects the fact that the Hall voltage is proportional to the velocity of charge carriers moving between the trapping events. Therefore, in contrast to μeff, the Hall mobility μH continues to increase with decreasing T even at low temperatures, at which the longitudinal conduction is in the trap-dominated regime. In the experiment of Podzorov et al. [82], the increase of μH with cooling could be traced down to ~150 K; at lower temperatures, the Hall measurements were hindered by a rapid enhancement of 1/f noise of the channel conductivity with T decreasing below the crossover to the trap-dominated regime. The observation of a “band” Hall effect suggests that charge transport on the surface of rubrene single crystals occurs via delocalized states over the whole studied temperature range (for a discussion of the Hall effect, see, for example, Pope and Swenberg [7]).

8080_C002.fm Page 54 Friday, April 6, 2007 10:21 AM

54

Organic Field-Effect Transistors

25

μ (cm2/Vs)

20 15 10 5 0

nHIn

1.0 0.8 0.6 0.4

175

200

225 250 T (K)

275

300

FIGURE 2.1.20 Upper panel: the temperature dependences of the Hall mobility, μH, (solid circles) and the effective mobility, μeff, extracted from the conventional FET equations — that is, from the longitudinal FET conductivity and the density of charges, n, field induced above the threshold (Equation 2.1.2) (open circles). Lower panel: the temperature dependence of the ratio of the Hall carrier density, nH, to the density n. (From Podzorov, V. et al., Phys. Rev. Lett., 95, 226601, 2005.)

2.1.3.5 COMPARISON WITH THE HOLSTEIN–PEIERLS MODEL AND TRANSPORT MEASUREMENTS IN THE BULK OF ORGANIC CRYSTALS A microscopic theory of finite-density charge transport on the surface of organic crystals has yet to be developed. However, several models have been proposed for the analysis of low-density (i.e., single particle) intrinsic transport in the bulk of organic crystals observed in TOF experiments by Karl et al. [19,84] and, more recently, in the space charge limited current measurements by Jurchescu et al. [85] and de Boer et al. [41] Signatures of the intrinsic band-like transport have also been observed in the experiments on subpicosecond transient photoconductivity [86,87]. The results of these experiments can be interpreted as an increase of the mobility with decreasing temperature (μ ∝ T–γ, with γ ~ 0.3), assuming that the efficiency of carrier photogeneration is temperature independent. Note, however, that in these experiments, unlike the transport measurements, hot (i.e., nonthermalized) optically excited carriers are probed. The ab initio calculations of polaron mobility on the basis of the Holstein–Peierls model including a nonlocal electron-lattice coupling (Hannewald and Bobbert [12,13]; see references to earlier work therein) reproduced the temperature-

8080_C002.fm Page 55 Friday, April 6, 2007 10:21 AM

Charge Carrier Transport in Single-Crystal Organic Field-Effect Transistors

55

γ=2

T−γ, γ = 2

100

μ (cm2/Vs)

μ (cm2/Vs)

20 10

1 0.1

μa μb μc′

10 200

300

0.01 20

T (K)

100 T (K)

400

FIGURE 2.1.21 Left: the Hall mobility versus T for a rubrene OFET in the double-log scale (compare with Figure 2.1.20). (From Podzorov, V. et al., Phys. Rev. Lett., 95, 226601, 2005.) Right: the calculated T-dependences of the hole mobility for different crystallographic directions in tetracene. (From Hannewald, K. and Bobbert, P. A., AIP Conf. Proc. 772, 1101, 2005.)

dependent mobility measured in single crystals of naphthalene along different crystalline directions [19,84]. According to this theory, the mobility for p- and n-type carriers in polyacenes should exhibit a “metallic” behavior for all T, up to room temperature (with the exception of n-type carriers in naphthalene). Similar behavior was predicted by a semiclassical model developed for the high-temperature regime by Troisi and Orlandi [88]. Interestingly, the theory of Hannewald and Bobbert also agrees semiquantitatively with the temperature dependence of μ observed in the “intrinsic” regime for rubrene OFETs. Fitting the Hall mobility data for rubrene OFETs (Figure 2.1.21) by power-law dependence μ(T) ~ T–γ yields the value γ ~ 2 at high temperatures, which agrees with the calculations for anthracene and tetracene. Similar temperature dependences of the hole mobility (μ ~ T–γ with γ ≈ 2–2.9) were obtained in the TOF experiments with bulk ultrapure crystals of naphthalene and perylene by Karl et al. [84] and in the SCLC measurements with ultrapure pentacene single crystals by Jurchescu et al. (Figure 2.1.22) [85].

2.1.3.6 TUNING

THE INTERMOLECULAR

DISTANCE

The availability of OFETs operating in the intrinsic conduction regime provides the opportunity to experimentally measure the dependence of polaronic mobility on the intermolecular distance of the crystal lattice, d. Variations of the polaron mobility with d can only be observed in the intrinsic regime, where trapping is not dominating (i.e., only when τtr << τ). Recently, the effect of continuous “tuning” of the intermolecular distance on mobility was observed in rubrene single-crystal OFETs with application of high pressure, P, of up to 0.5 GPa (Figure 2.1.23) [89]. The estimates show that, at such pressure, the intermolecular distance is decreased by Δd ~ 1.5%. It was observed that the mobility increased linearly with pressure over the range

8080_C002.fm Page 56 Friday, April 6, 2007 10:21 AM

56

Organic Field-Effect Transistors

60 Pentacene single crystal

μ (cm2/Vs)

50 40 30 20 10 0

220

240

260

280 300 T (K)

320

340

Mobility (cm2/Vs)

FIGURE 2.1.22 Temperature dependence of the hole mobility in ultrapure pentacene crystals extracted from the SCLC measurements. The open symbols correspond to the values of μ calculated under an assumption of uniform current flow across the crystal thickness; solid symbols take into account the anisotropy of conductivity in pentacene. Below room temperature, the mobility increases with decreasing T as μ ~ T–γ, where γ ≈ 2.4. (From Jurchescu, O. D. et al., Appl. Phys. Lett., 84, 3061, 2004.) VDS = −1V

10 8

u = 8.00 × P + 6.07

6

(a)

VT (V)

0.44 0.40 0.36

(b)

0.32 0.00

0.15 0.30 0.45 Pressure (GPa)

FIGURE 2.1.23 Pressure dependence of the field-effect mobility (a) and the threshold voltage (b) in single-crystal rubrene OFETs (solid and open symbols correspond to the increasing and decreasing pressure). (From Rang, Z. et al., Appl. Phys. Lett., 86, 123501, 2005.)

0–0.5 Gpa. This observation is in line with the expectations based on polaronic models: The mobility, which is proportional to the square of the transfer integral for the Holstein type of small polaron, should depend linearly on pressure for small variations of the intermolecular distance [89].

2.1.3.7 SURFACE

VERSUS

BULK TRANSPORT

A semiquantitative agreement between the carrier mobilities obtained from OFET measurements and the models developed for the bulk transport may suggest that

8080_C002.fm Page 57 Friday, April 6, 2007 10:21 AM

Charge Carrier Transport in Single-Crystal Organic Field-Effect Transistors

57

μ (cm2/Vs)

25

μ (cm2/Vs)

20 15

10

1

10 1

10

ε

5 0 0

5

10

15

20

25

ε

FIGURE 2.1.24 The dependence of the mobility of p-type carriers in rubrene single-crystal OFETs on the dielectric constant of the gate insulator, ε. The bars represent the spread in the mobility values. (From Stassen, A. F. et al., Appl. Phys. Lett., 85, 3899, 2004.)

there are no differences between polaronic conduction in the bulk and at the surface of organic crystals. However, recent experiments with single-crystal OFETs [57] revealed strong dependence of carrier mobility on the dielectric constant ε of the gate insulator. Figure 2.1.24 shows that the room-temperature mobility in these devices varies approximately as ε–1 over a wide range ε = 1 – 25. Earlier, a similar trend was observed for the organic TFTs based on soluble polymers [90]. (For a recent review of gate dielectrics, including inorganic, polymeric, and ultrathin self-assembled molecular layers, see Facchetti et al. [91]). In single-crystal OFETs these observations cannot be attributed to a difference in the morphology of the organic semiconductor, since the crystals used to obtain the data in Figure 2.1.24 were grown under the same conditions, irrespective of the gate insulator. Thus, the observed μ(ε) trend indicates that the mobility of charges at the interface between an organic semiconductor and an insulator is the property of the interface (a combination of the semiconductor and insulator), rather than the organic material alone. It has been predicted that the intrinsic mobility of polarons at an interface with a highly polarizable dielectric may decrease due to an increase of the effective polaronic mass [25]. Though this prediction is in line with the experimental observations, it is worth noting that the mobilities measured in the high-ε OFETs by Stassen et al. [23] and de Boer et al. [57] might be trap dominated and therefore could significantly differ from the corresponding intrinsic μ0(ε) (e.g., because of a larger densities of shallow traps at interfaces with greater ε). More experiments are needed for a better understanding of μ(ε) dependence and differences between bulk and interfacial transport in organic semiconductors. For example, measurements of the Hall mobility, μH, in highε OFETs could shed more light on the intrinsic μ0(ε) dependence. Development of single-crystal OFETs allows studies of charge transport in organic semiconductors in a much broader range of charge carrier densities, inaccessible with the bulk time of flight of SCLC measurements. Approximately one

8080_C002.fm Page 58 Friday, April 6, 2007 10:21 AM

58

Organic Field-Effect Transistors

carrier per molecule at the interface can be induced using high-ε gate dielectrics. At such density, polaron–polaron interactions may play a significant role, which can lead to the formation of novel electronic phases. Indeed, it is known that at a sufficiently high density of chemically doped carriers, the potassium-doped fullerene KxC60 exhibits superconductivity (x = 3) [92] and a Mott–Hubbard insulating state (x = 4) [93]. In rubrene single-crystal OFETs, stable operation at a carrier density n = 5 ⋅ 1013 cm–2 (corresponding to 0.1 carriers per molecule) has been demonstrated by using high-quality Ta2O5 gate dielectric [23]. With a breakdown field of EB = 6.6 MV/cm and a dielectric constant of ε = 25, this insulator should allow reaching an even higher density of charge carriers, ~1014 cm–2. However, the presence of moderate leakage currents which induce irreversible degradation in the devices currently prevents the operation of Ta2O5 gate insulators too close to the breakdown. In another interesting approach for achieving high carrier densities in OFETs (>1015 cm–2), a conventional gate dielectric is replaced with a polymer electrolyte [24,94]. In OFETs with high-ε dielectrics and with polyelectrolytes, a nonmonotonic dependence of channel conductivity on charge carrier density has been observed, with a maximum conductivity at ~1015 charges per square centimeter. This observation might indicate filling of the conduction band in these narrow-band semiconductors when the density of field-induced carriers approaches ~ one carrier per molecule. These field-effect experiments are in line with the experiments on chemical doping of organic thin films, where the insulator–metal–insulator transition has been observed with increasing dopant concentration. At the current stage of the exploration of high carrier density regime in OFETs, reproducible fabrication of high-capacitance devices has been achieved, and notable deviations from the standard device characteristics have been observed at high carrier densities [23,94]. Despite the first successful demonstrations, much more work is needed in this direction in order to distinguish intrinsic effects from the effects that might be associated with defects and interface morphology. For example, at these large densities, molecular steps at the crystal surface could significantly affect the longitudinal transport [95].

2.1.3.8 PHOTOINDUCED PROCESSES

IN

SINGLE-CRYSTAL OFETS

Despite the fact that organic transistors are intended to operate back to back with organic light-emitting diodes (OLEDs) as switches for display pixels (see, for example, Forrest [3]), surprisingly little is known about photoinduced effects in OFETs. In thin-film organic transistors, light illumination produces a weak photoconductivity response [96] and reduces the threshold voltage [97]. One of the explanations for this paradox is that thin-film organic transistors are ill suited for fundamental research in this direction due to a large density of defects that could trap light-generated carriers and/or act as recombination centers. By contrast, singlecrystal organic transistors with a low density of defects provide a unique opportunity to investigate intrinsic photoinduced processes at organic surfaces and interfaces. A better understanding of these phenomena is crucial for optimization of organic electronic devices.

8080_C002.fm Page 59 Friday, April 6, 2007 10:21 AM

Charge Carrier Transport in Single-Crystal Organic Field-Effect Transistors

59

Several novel light-induced effects have been recently observed using the singlecrystal OFETs as an experimental platform. In experiments with back-gated rubrene OFETs [98], switching of devices from an OFF state into a persistent ON state by a short pulse of light was observed. The effect is caused by interplay between the surface built-in conducting channel and nonequilibrium carriers photogenerated in the bulk. The self-latching character of this effect provides an opportunity for using these devices as optically addressed memory elements. The standard, front-gate transistor geometry with a transparent parylene insulator and a semitransparent gate electrode makes it possible to illuminate the critical semiconductor–insulator interface with photons and to study the OFET’s response (Figure 2.1.25). Light- and gate-controlled shift of the field-effect threshold voltage has been observed in these experiments. The effect is caused by the photoinduced charge transfer across the interface between an ordered semiconductor and a disordered polymeric gate insulator [55]. This class of interfaces is especially important for a wide range of all-organic flexible devices. When the interface is illuminated with photons of energy greater than the absorption edge of the organic semiconductor, nonequilibrium charges are generated near the interface. Driven by the transverse gate electric field at the interface, these carriers are transferred into the polymer and become immobilized by its deep traps. Either holes or electrons can be transferred, depending on the sign of the gate voltage applied during the illumination, Vgillum. This causes a nonvolatile shift of the fieldeffect threshold in OFETs that can be used for a direct measurement of the charge transfer rate. Potential applications of this effect range from organic photosensors and optical memory devices to lithography-free patterning of the conduction channel in OFETs with light exposure. More studies using optical and nonlinear optical tools are required to investigate and characterize photoexcitations and charge transport in organic single crystals and related devices. Time-resolved photoluminescence and photocurrent studies, for example, can provide information on charge-carrier excitation mechanisms, their mobility, and lifetimes. In addition, such optical studies are complementary to the efforts on design, fabrication, and studies of organic electronic devices. For instance, the absorption edge in the spectral characteristics of rubrene single crystals, shown in Figure 2.1.26 [99], coincides with the “red boundary” of the photoinduced threshold shift in rubrene OFETs demonstrated in Figure 2.1.25. This indicates that the previously mentioned effect is indeed associated with the photocarriers generated in the organic semiconductor.

2.1.4 DEFECTS AT THE SURFACE OF ORGANIC CRYSTALS Because of the small size of polarons in molecular crystals, the conduction channel in organic transistors extends in the transverse direction for only a few molecular layers [20–22]. For the same reason, polarons interact strongly with chemical impurities and structural defects. As a result, polaronic transport in organic OFETs is very sensitive to the morphology of the semiconductor surface and to the presence

8080_C002.fm Page 60 Friday, April 6, 2007 10:21 AM

60

Organic Field-Effect Transistors

6 Rate (×109 cm−2s−1)

3 × 10−6

ISD (A)

2 × 10−6

3

Electrons

0 0.2 Holes

0.0 400

1 × 10−6

500 600 λ (nm)

700

0 As prepared 1 × 10−6 Vgillum = 50 V

ISD (A)

1 × 10−7

Vgillum = −50 V

1 × 10−8 Vonset

hν 1 × 10−9

S

G

D

1 × 10−10 −60

−40

−20

0 Vg (V)

20

40

60

FIGURE 2.1.25 Gate controlled photoinduced shift of the onset voltage in OFET caused by the photoinduced charge transfer at the interface. Linear (top) and semilog (bottom) plots of ISD(Vg) of a rubrene OFET, measured in the dark after illumination of the device at fixed applied gate voltages, Vgillum. The top inset shows that the effect exhibits a “red boundary.” The bottom inset schematically shows the device geometry. (From Podzorov, V. et al., Phys. Rev. Lett., 95, Z26601, 2005.)

of electronic defects at the semiconductor–insulator interface. Carrier trapping, charge doping, molecular reorientation, dipole formation, and a range of possible chemical interactions are among the many phenomena that can occur at the semiconductor–insulator interface and they can affect the electrical characteristics of single-crystal OFETs. For example, the localized electronic states within the HOMO–LUMO gap impair the performance of the field-effect transistors by increasing the field-effect threshold voltage and reducing the effective mobility of charge carriers [100]. The surface density of electronic defects in high-quality single-crystal OFETs can be less than 1010 cm–2 [38], which corresponds to interdefect distances of ~ 0.1 μm.

8080_C002.fm Page 61 Friday, April 6, 2007 10:21 AM

Charge Carrier Transport in Single-Crystal Organic Field-Effect Transistors

Wavelength (nm) 500 550 600

450

650

700

1.0

α (104 cm−1)

μs-PLE

α

0.8

800 1.0

PL

0.6 0.5 0.4

PL (arb. units)

400

61

PLE, 620 nm 0.2

PLE, 570 nm 0.0

3.0

2.8

2.6

2.4 2.2 Photon energy (eV)

2.0

1.8

1.6

0.0

FIGURE 2.1.26 b-Polarized absorption spectrum (α), photoluminescence spectrum (PL), cw luminescence excitation spectrum (PLE) — monitored at 570 nm (dashed line) and 620 nm (dotted line) — and excitation spectrum of the microsecond luminescence transient induced by b-polarized light (open circles) in rubrene single crystals. (From Najafov, H. et al., Phys. Rev. Lett., 96, 056604, 2006.)

This low density of surface defects is the major factor that determines the record performance of single-crystal OFETs and enables exploration of the fundamental limits of charge carrier transport in organic materials. In addition, these devices provide an efficient tool for studying the polaron-defect interactions. This section focuses on defects that can be formed in the process of crystal growth, OFET fabrication, and as a result of the interaction with ambient environment.

2.1.4.1 BULK AND SURFACE ELECTRONIC DEFECTS ORGANIC CRYSTALS

IN

The density of electronic defects in organic crystals might significantly vary depending on the crystal growth methods. Niemax et al. performed TOF mobility measurements and concluded that vapor-grown tetracene single crystals are of a higher quality than vapor-Bridgman-grown crystals [44]. The distribution of localized states in the bulk of organic crystals can be probed by photoconductivity [6,7,101] and space charge limited current (SCLC) measurements [8,41,85]. Figure 2.1.27 shows the photocurrent spectra obtained at room temperature for pentacene crystals in recent experiments by Lang et al. [101]. These data indicate a broad (~1 eV wide) amorphous-like distribution of localized states within the HOMO–LUMO gap near the HOMO edge, which resembles the exponential “tails” of valence and conduction bands in strongly disordered inorganic semiconductors. This observation raises the question of the origin of such an amorphous-like distribution of localized states because it is not in line with the typically high crystallinity of vapor-grown organic crystals.

8080_C002.fm Page 62 Friday, April 6, 2007 10:21 AM

62

Organic Field-Effect Transistors

100

Photocurrent yield (e/pH)

10−1 10−2 10−3

Photocurrent Scaled photoquenching Silinsh PC (E-2.25)2.5/E2 Scaled exciton Exp (−β(2.25 − E)); β = 5.7 eV-1

10−4 10−5 10−6 10−7 1.0

1.5

2.0

2.5

3.0

3.5

Photon energy (eV)

FIGURE 2.1.27 Energy distribution of localized states in HOMO–LUMO gap of singlecrystal pentacene. The HOMO band corresponds to energies > 2.25 eV. (From Lang, D. V. et al., Phys. Rev. Lett., 93, 086802, 2004.)

The large density of in-gap trap states in these highly purified pentacene crystals might originate from the tendency of pentacene to react with oxygen and water from the environment. Such sensitivity to gases and vapors has been used in TFT-based gas-sensing applications; it is argued that a diffusion of molecules to the conduction channel along the grain boundaries in thin films is crucial for the sensing effect [102,103]. Recently, however, based on infrared absorption, mass spectrometry, and SCLC measurements, it was suggested that gas molecules from the ambient can diffuse even into the single crystals of pentacene as deeply as tens of microns, creating doping centers associated with O2 and traps associated with H2O molecules [104]. In particular, it was shown that 6,13-pentacenequinone (the product of pentacene oxidation) forms electronic defects in pentacene, and careful purification of the commercially available material by vacuum sublimation is required to achieve a significant reduction of this impurity [85]. Following an extensive purification procedure, it has been estimated from SCLC measurements that the bulk density of traps in ultrapure pentacene crystals can be as low as ~2 ⋅ 1011 cm–3. However, the density of traps can be considerably larger close to the crystal surface, which is normally exposed to the environment during the OFET assembly. Rubrene also reacts with oxygen and forms rubrene endoperoxide in the process of self-sensitized photo-oxidation [98,105]. Oxidation of rubrene is restricted to a thin surface layer [106]. The role of the surface endoperoxide in the charge transport properties of rubrene OFETs is still unclear and more work is needed in this direction. For example, surface endoperoxide might help to protect the conduction channel

8080_C002.fm Page 63 Friday, April 6, 2007 10:21 AM

Charge Carrier Transport in Single-Crystal Organic Field-Effect Transistors

63

from interaction with ambient gases. Recent experiments with the “air-gap” rubrene single-crystal OFETs, in which the conduction channel is fully exposed to environmental species, have shown that the channel conductivity is not sensitive to the presence of such gases as O2, H2O, N2, and H2, and even to the saturated vapors of acetone and propanol [56]. This remarkable stability of rubrene OFETs is in a sharp contrast to typically fast degradation of pentacene devices.

2.1.4.2 DENSITY

OF

DEFECTS

IN

SINGLE-CRYSTAL OFETS

Photoconductivity and SCLC experiments have been used to obtain information on localized states in the bulk of organic crystals. However, information about the surface states that are relevant to the OFET operation is still very limited. Fortunately, the calculation of several basic OFET parameters does not require a detailed knowledge of the trap spectrum. For example, the field-effect threshold voltage is determined by the total density of deep traps Ntrdeep: Vgth = eNtrdeep/Ci

(2.1.7)

The density of deep traps is a temperature-dependent quantity: The borderline between deep and shallow traps shifts closer to the band edge with decreasing T, and any temperature-driven changes of the bandwidth of organic semiconductor, W, would affect the density and spectrum of traps. The temperature dependence of the threshold voltage of p-type “air-gap” rubrene single-crystal OFETs is shown in Figure 2.1.28 [38]. At room temperature, the threshold voltage is small; it corresponds to the deep-trap density on the pristine crystal surface ~ 7 × 109 cm–2. The concentration of deep traps increases quasilinearly with cooling up to 2 × 1010 cm–2 at 150 K. Assuming that the thickness of the conduction channel does not exceed one or two molecular layers, the three-dimensional density of traps near the surface at 300 K can be estimated as ~2 × 1016 cm–3. 0

Vth (V)

−5

b-axis a-axis

−10 −15 −20 100

150

200 T (K)

250

300

FIGURE 2.1.28 Temperature dependence of the threshold voltage of a rubrene OFET measured along a and b axes in the basal plane of an orthorhombic single crystal. (From Podzorov, V. et al., Phys. Rev. Lett., 93, 086602, 2004.)

8080_C002.fm Page 64 Friday, April 6, 2007 10:21 AM

64

Organic Field-Effect Transistors

Defects in the molecular materials that act as traps can also be generated during device operation. The creation of such defects has been observed in tetracene, perylene, and rubrene single-crystal transistors with Ta2O5 gate dielectric at large gate electric fields; this results in irreversible decrease of ISD of an operating OFET [23]. Systematic experiments with differently prepared gate dielectrics have shown that the defects originate from a leakage current through the gate insulator. The leakage at high gate fields induces a degradation of the molecular material by creating a large density of deep taps, which causes a substantial increase of the threshold voltage and a decrease of the source-drain current. The precise nature of the leakageinduced traps has not been established yet, and more work is necessary to address this issue.

2.1.4.3 SINGLE-CRYSTAL OFETS SURFACE DEFECTS

AS

TOOLS

TO

STUDY

It is well known that x-rays create defects in organic semiconductors by breaking the molecules and creating new chemical species. Using this effect, Podzorov et al. [38] controllably increased the defect density in rubrene single crystals and studied how this process affected charge transport in the corresponding air-gap OFETs (Figure 2.1.29). The x-ray treatment increases the field-effect threshold, and thus, the density of deep traps. Interestingly, the mobility, which is proportional to the slope of ISD(Vg), and its temperature dependence μ(T) were not affected by x-ray irradiation. This suggests that the deep traps, being filled above the threshold, do not scatter mobile polarons. The defects in OMCs can be induced not only during the crystal growth and OFET fabrication, but also as a result of the handling of crystals in a high-vacuum 3.0 ΔtX-ray

2.5

b-axis a-axis Δt = 0 90 s 210 s 330 s

ISD (μA)

2.0 1.5 1.0 0.5 0.0 −30

−25

−20

−15

−10 Vg (V)

−5

0

5

FIGURE 2.1.29 The effect of a gradual increase of x-ray exposure on the transconductance characteristics of rubrene OFET. Upon increasing the x-ray dose, the threshold voltage increases, while the mobility remains unchanged. (From Podzorov, V. et al., Phys. Rev. Lett., 93, 086602, 2004.)

8080_C002.fm Page 65 Friday, April 6, 2007 10:21 AM

Charge Carrier Transport in Single-Crystal Organic Field-Effect Transistors

65

1.0 2.5 (×10−6 torr)

ISD/ISD0

0.8

Gauge ON

0.6

5.0 Crystal

0.4

S

0.0

15.0

G PDMS stamp

0.2

0

8.0

D

200

Vacuum gap 400 t (s)

600

800

FIGURE 2.1.30 Demonstration of the “gauge effect”: time evolution of ISD in a vacuum-gap OFET (the channel is exposed to the environment), operating in a vacuum chamber at different pressures: 2.5, 5, 8, and 15 × 10–6 torr. The hot-cathode high-vacuum gauge was turned on at t = 250 s. (From Podzorov, V. et al., Appl. Phys. Lett., 87, 093505, 2005.)

environment. In the experiments with “vacuum-gap” single-crystal OFETs, where the conduction channel is exposed to environmental agents, it was observed that deep and shallow traps are generated at the organic surface in vacuum due to interactions with chemically active species produced by high-vacuum gauges or hot surfaces such as resistively heated filaments and evaporation sources — the “gauge effect” [56]. Figure 2.1.30 shows that a rapid decrease of the source-drain current of an operating device occurs when a high-vacuum gauge is turned on. The effect has been attributed to the interaction of the organic surface with electrically neutral free radicals. These species can be produced at the surface of hot filaments in highvacuum gauges in a process of hydrocarbon cracking that has relatively low activation energy, Ea ~ 2.5 eV (240 kJ/mol). Clearly, minimization of the damage induced through this mechanism is important for optimizing the performance of a wide range of thin-film organic devices that are fabricated or characterized in high vacuum. The experiments described earlier demonstrate the great potential of singlecrystal OFETs as a diagnostic tool for investigations of phenomena affecting the performance of organic transistors, as well as for studying surface-limited reactions in organic semiconductors.

2.1.5 CONCLUSION Within a short four-year span, the development of single-crystal organic field-effect transistors significantly advanced our understanding of transport processes on organic surfaces and shed light on the fundamental limits of organic devices (Figure 2.1.31). Single-crystal OFETs enabled the systematic study of polaronic effects in a broad range of parameters due to their reproducibility, bridging the gap between

8080_C002.fm Page 66 Friday, April 6, 2007 10:21 AM

66

Organic Field-Effect Transistors

103

Mobility (cm2V−1s−1)

102 101 100

Today’s Processors

Si wafers Rubrene OFETs Poly-Si

Hybrids

Low-cost ICs

Organics

Smart cards, displays

a-Si:H

10−1

E-paper

10−2 10−3

Polythiophenes Thiophene oligomers Pentacene Organic/inorganic hybrid

10−4 10−5 10−6

1988

1992

1996 2000 Time (years)

2004

2008

FIGURE 2.1.31 Realization of charge transport not limited by static disorder is crucial for better understanding of the fundamental limits of organic electronics. The red dot on the figure borrowed from the IBM Journal for Research and Development, January 2001, represents the room-temperature mobility in rubrene single-crystal OFETs, the only “intrinsic” mobility found in the field-effect experiments to date. (From Shaw, J. M. and Seidler, P. F., IBM J. Res. Dev. 45, 3, 2001.)

the fundamental research on ultrapure bulk crystals and applied research on thinfilm transistors. The potential of single-crystal OFETs goes far beyond the basic research on polaronic transport. These devices are very useful for studies of surface reactions and defect formation mechanisms on organic surfaces [56], light-induced effects in organic semiconductor devices [55,98], and work-function engineering with self-assembly monolayers [107]. Although the single polaron problem was one of the early subjects of interest for condensed matter physics, polaronic conduction at a finite density remains a complex and poorly understood topic even these days, with many open questions. The organic single-crystal OFETs represent a unique tool to address these questions in a class of materials relevant for the emerging field of plastic electronics. For example, the band structure calculations that can be found in the literature for different molecular crystals do not include the interaction between the electronic and vibrational degrees of freedom, so it remains to be understood how polaronic effects affect the density of states and the width of HOMO and LUMO bands. These quantities are expected to exhibit nontrivial temperature dependences due to the strongly T-dependent parameters of polarons in Holstein-like models. Another important issue is transport across metal/organic interfaces [58]. For inorganic semiconductors, our understanding of contacts with metals is based on the concept of a Schottky barrier. For organic semiconductors based on small molecules, the experimental data remain, to a large extent, unclear and irreproducible because of the poor interface control. Narrow bandwidth of organic semiconductors and weak van der Waals molecule–molecule and molecule–metal coupling make the behavior of metal/organic interfaces qualitatively different from that of their inorganic counterparts. Thanks to their great reproducibility, the single-crystal OFETs have a

8080_C002.fm Page 67 Friday, April 6, 2007 10:21 AM

Charge Carrier Transport in Single-Crystal Organic Field-Effect Transistors

67

potential to clarify the situation in this field. Note that these questions are directly related to the issue of contact resistance in organic transistors that impose serious limitations on the downscaling of organic thin-film devices [108]. Increasing the purity of organic crystals and reducing the contact resistance in OFETs is another challenging direction of future experimental work, which will help to extend the temperature range where the intrinsic polaronic transport can be studied. The development of more advanced techniques for purification of molecular materials will enable the expansion of the intrinsic transport regime to much lower temperatures, where the effects of quantum statistics and polaron–polaron interactions should become experimentally accessible. The development of high-quality single-crystal OFETs will be crucial for the comparative studies of different molecular materials and, in particular, for testing the transport properties of newly synthesized molecular compounds in their crystalline form. Such tests are essential for supporting the effort to synthesize novel organic molecules that could form high-mobility ordered aggregates from solutions — one of the most attractive goals of research in organic electronics. Although some soluble derivatives of linear acenes and the corresponding OFETs have been demonstrated, much more experimental work needs to be done to optimize these systems for solubility and charge carrier mobility [45,109,110]. Systematic effort in this direction will require the investigation of a much broader variety of new molecular compounds. Single-crystal OFETs provide a unique tool for the express analysis of transport characteristics of new molecular materials with defect densities much smaller than those in TFTs. Therefore, even though large-scale applications will ultimately rely on thin films, research on single-crystal OFETs can play an important role in the material selection for applied devices. The case of rubrene perfectly illustrates this point. The unprecedented quality of OFETs based on vapor-grown rubrene crystals has stimulated work on the deposition of rubrene thin-films from solution: StingelinStutzmann et al. have recently demonstrated solution-processed rubrene TFTs with high mobility (up to 0.7 cm2/Vs at room temperature) [111].

ACKNOWLEDGMENTS I am indebted to all my coworkers and collaborators for their significant contributions to the work presented here, especially to M. E. Gershenson, A. F. Morpurgo, J. A. Rogers, D. Fichou, I. Biaggio, H. Najafov, M. F. Calhoun, V. M. Pudalov, E. Loginova, R. W. I. de Boer, E. Menard, V. C. Sundar, J. Zaumseil, Z. Rang, P. P. Ruden, M. I. Nathan, C. D. Frisbie, O. Ostroverkhova, B. Yakshinskiy, T. Madey, Y. Chabal, and E. Garfunkel. VP at Rutgers University is supported in part by the NSF grants DMR-0405208 and ECS-0437932.

REFERENCES 1. Shaw, J. M., and Seidler, P. F., Organic electronics: Introduction, IBM J. Res. Dev., 45, 3, 2001.

8080_C002.fm Page 68 Friday, April 6, 2007 10:21 AM

68

Organic Field-Effect Transistors 2. Hoppe, H., and Sariciftci, N. S., Organic solar cells: An overview, J. Mater. Res., 19, 1924, 2004. 3. Forrest, S. R., The path to ubiquitous and low cost organic electronic appliances on plastic, Nature, 428, 911, 2004. 4. Katz, H. E., Recent advances in semiconductor performance and printing processes for organic transistor-based electronics, Chem. Mater., 16, 4748, 2004. 5. Agranovich, V. M., and Bassani, G. F., Electronic excitations in organic based nanostructures, Elsevier Academic Press, New York, 2003. 6. Silinsh, E. A., and Cápek, V., Organic molecular crystals: Interaction, localization, and transport phenomena, AIP Press, New York, 1994. 7. Pope, M., and Swenberg, C. E., Electronic processes in organic crystals and polymers, 2nd ed., Oxford University Press, New York, 1999. 8. Kao, K. C., and Hwang, W., Electrical transport in solids, vol. 14, Pergamon Press, New York, 1981. 9. Holstein, T., Polaron motion. II. “Small” polaron, Ann. Phys., 8, 343, 1959. 10. Emin, D., Small polarons, Phys. Today, 35, 34, 1982. 11. Troisi, A., Orlandi, G., and Anthony, J. E., Electronic interactions and thermal disorder in molecular crystals containing cofacial pentacene units, Chem. Mater., 17, 5024, 2005. 12. Hannewald, K., and Bobbert, P. A., Ab-initio theory of charge-carrier conduction in ultrapure organic crystals, Appl. Phys. Lett., 85, 1535, 2004. 13. Hannewald, K., and Bobbert, P. A., Ab-initio theory of charge transport in organic crystals, AIP Conf. Proc., 772, 1101, 2005. 14. Wu, M. W., and Conwell, E. M., Transport in α-sexithiophene films, Chem. Phys. Lett., 266, 363, 1997. 15. Deng, W.-Q., and Goddard, W. A., Predictions of hole mobilities in oligoacene organic semiconductors from quantum mechanical calculations, J. Phys. Chem. B, 108, 8614, 2004. 16. Kenkre, V. M. et al., Unified theory of the mobilities of photoinjected electrons in naphthalene, Phys. Rev. Lett., 62, 1165, 1989. 17. da Silva Filho, D. A., Kim, E.-G., and Brédas, J.-L., Transport properties in the rubrene crystal: Electronic coupling and vibrational reorganization energy, Adv. Mater. 17, 1072, 2005. 18. Fratini, S., and Ciuchi, S., Dynamical mean-field theory of transport of small polarons, Phys. Rev. Lett., 91, 256403, 2003. 19. Karl, N., Charge-carrier mobility in organic crystals, in Organic electronic materials, eds. Farchioni, R., and Grosso, G., pp. 283–326, Springer–Verlag, Berlin, 2001. 20. Dodabalapur, A., Torsi, L., and Katz, H. E., Organic transistors: Two-dimensional transport and improved electrical characteristics, Science, 268, 270, 1995. 21. Dinelli, F. et al., Spatially correlated charge transport in organic thin film transistors, Phys. Rev. Lett., 92, 116802, 2004. 22. Kiguchi, M. et al., Electric-field-induced charge injection or exhaustion in organic thin film transistor, Phys. Rev. B, 71, 035332, 2005. 23. de Boer, R. W. et al., Influence of the gate leakage current on the stability of organic single-crystal field-effect transistors, Appl. Phys. Lett., 86, 032103, 2005. 24. Panzer, M. J., and Frisbie, C. D., Polymer electrolyte gate dielectric reveals finite windows of high conductivity in organic thin film transistors at high charge carrier densities, J. Am. Chem. Soc., 127, 6960, 2006. 25. Houili, H., Picon, J. D., Bussac, M. N., and Zuppiroli, L., Polarization effects in the channel of an organic field-effect transistor, J. Appl. Phys., 100, 023702, 2006.

8080_C002.fm Page 69 Friday, April 6, 2007 10:21 AM

Charge Carrier Transport in Single-Crystal Organic Field-Effect Transistors

69

26. Riordan, M., and Hoddeson, L., Crystal fire: The birth of information age, W. W. Norton & Co., New York, 1997. 27. Horowitz, G., Organic thin film transistors: From theory to real devices, J. Mater. Res., 19, 1946, 2004. 28. Nelson, S. F., Lin, Y.-Y., Gundlach, D. J., and Jackson, T. N., Temperature-independent transport in high-mobility pentacene transistors, Appl. Phys. Lett., 72, 1854, 1998. 29. Podzorov, V., Pudalov, V. M., and Gershenson, M. E., Field-effect transistors on rubrene single crystals with parylene gate insulator, Appl. Phys. Lett., 82, 1739, 2003. 30. Podzorov, V. et al., Single-crystal organic field effect transistors with the hole mobility ~8 cm2/V s, Appl. Phys. Lett., 83, 3504, 2003. 31. de Boer, R. W. I., Klapwijk, T. M., and Morpurgo, A. F., Field-effect transistors on tetracene single crystals, Appl. Phys. Lett., 83, 4345, 2003. 32. de Boer, R. W. I. et al., Organic single-crystal field-effect transistors, Phys. Stat. Sol., 201, 1302, 2004. 33. Takeya, J. et al., Field-induced charge transport at the surface of pentacene single crystals: A method to study charge dynamics of two-dimensional electron systems in organic crystals, J. Appl. Phys., 94, 5800, 2003. 34. Butko, V. V. et al., Field-effect transistor on pentacene single crystal, Appl. Phys. Lett., 83, 4773, 2003. 35. Sundar V. C. et al., Elastomeric transistor stamps: Reversible probing of charge transport in organic crystals, Science, 303, 1644, 2004. 36. Goldmann, C. et al., Hole mobility in organic single crystals measured by a “flipcrystal” field-effect technique, J. Appl. Phys., 96, 2080, 2004. 37. Aleshin, A. N. et al., Mobility studies of field-effect transistor structures based on anthracene single crystals, Appl. Phys. Lett., 84, 5383, 2004. 38. Podzorov, V. et al., Intrinsic charge transport on the surface of organic semiconductors, Phys. Rev. Lett., 93, 086602, 2004. 39. Menard, E. et al., High-performance n- and p-type single-crystal organic transistors with free-space gate dielectrics, Adv. Mater., 16, 2097, 2004. 40. Laudise, R. A. et al., Physical vapor growth of organic semiconductors, J. Cryst. Growth, 187, 449, 1998. 41. de Boer, R. W. I. et al., Space charge limited transport and time of flight measurements in tetracene single crystals: A comparative study, J. Appl. Phys., 95, 1196, 2004. 42. Menard, E., Jiang, W., Podzorov, V., Garfunkel, E., Rogers, J. A., and Gershenson, M. E., unpublished. 43. Calhoun, M.F., Hsieh, C., and Podzorov, V., Effect of shallow traps on polaron transport at the surface of organic semiconductors, cond_mat/0610851. 44. Niemax, J., Tripathi, A. K., and Pflaum, J., Comparison of the electronic properties of sublimation- and vapor-Bridgman-grown crystals of tetracene, Appl. Phys. Lett., 86, 122105, 2005. 45. Moon, H. et al., Synthesis, crystal structure, and transistor performance of tetracene derivatives, J. Am. Chem. Soc., 126, 15322, 2004. 46. Mas-Torrent, M. et al., High mobility of dithiophene-tetrathiafulvalene single-crystal organic field effect transistors, J. Am. Chem. Soc., 126, 984, 2004. 47. Zeis, R., Siegrist, T., and Kloc, C., Single-crystal field-effect transistors based on copper phthalocyanine, Appl. Phys. Lett., 86, 022103, 2005. 48. Zeis, R. et al., Field effect studies on rubrene and impurities of rubrene, Chem. Mater., 18, 244, 2006. 49. Henn, D. E., Williams, W. G., and Gibbons, D. J., Crystallographic data for an orthorhombic form of rubrene, J. Appl. Cryst., 4, 256, 1971.

8080_C002.fm Page 70 Friday, April 6, 2007 10:21 AM

70

Organic Field-Effect Transistors 50. Campbell, R. B., Monteath Robertson, J., and Trotter, J., The crystal structure of hexacene, and a revision of the crystallographic data for tetracene, Acta Cryst., 15, 289, 1962. 51. Fritz, S. E. et al., Structural characterization of a pentacene monolayer on an amorphous SiO2 substrate with grazing incidence x-ray diffraction, J. Am. Chem. Soc., 126, 4084, 2004. 52. Ruiz, R. et al., Structure of pentacene thin films, Appl. Phys. Lett., 85, 4926, 2004. 53. Menard, E., Marchenko, A., Podzorov, V., Gershenson, M. E., Fichou, D., and Rogers, J. A., Adv. Mater., 18, 1552, 2006. 54. For specifications, see the web site of Specialty Coating Systems: http://www.scscoat ings.com/parylene_knowledge/specifications.cfm. 55. Podzorov, V., and Gershenson, M. E., Photoinduced charge transfer across the interface between organic molecular crystals and polymers, Phys. Rev. Lett., 95, 016602, 2005. 56. Podzorov, V. et al., Interaction of organic surfaces with active species in the highvacuum environment, Appl. Phys. Lett., 87, 093505, 2005. 57. Stassen, A. F. et al., Influence of the gate dielectric on the mobility of rubrene singlecrystal field-effect transistors, Appl. Phys. Lett., 85, 3899, 2004. 58. Cahen, D., Kahn, A., and Umbach, E., Energetics of interfaces between molecules and conductors, Mater. Today, Jul./Aug., 32, 2005. 59. Chwang, A. B., and Frisbie, C. D., Field effect transport measurements on single grains of sexithiophene: Role of the contacts, J. Phys. Chem. B, 104, 12202, 2000. 60. Zaumseil, J., Baldwin, K. W., and Rogers, J. A., Contact resistance in organic transistors that use source and drain electrodes formed by soft contact lamination, J. Appl. Phys., 93, 6117, 2003. 61. Hulea, I. N. et al., Reproducible low contact resistance in rubrene single-crystal fieldeffect transistors with nickel electrodes, Appl. Phys. Lett., 88, 113512, 2006. 62. Troisi, A., and Orlandi, G., Charge-transport regime of crystalline organic semiconductors: Diffusion limited by thermal off-diagonal electronic disorder, Phys. Rev. Lett., 96, 086601, 2006. 63. Sze, S. M., Physics of semiconductor devices, Wiley, New York, 1981. 64. Horowitz, G., Hajlaoui, M. E., and Hajlaoui, R., Temperature and gate voltage dependence of hole mobility in polycrystalline oligothiophene thin film transistors, J. Appl. Phys., 87, 4456, 2000. 65. Shur, M., Hack, M., and Shaw, J. G., A new analytic model for amorphous silicon thin-film transistors, J. Appl. Phys., 66, 3371, 1989. 66. Nakamura, K. et al., Light emission from organic single-crystal field-effect transistors, J. Appl. Phys., 44, L1367, 2005. 67. Chua, L. L. et al., General observation of n-type field-effect behaviour in organic semiconductors, Nature, 434, 194, 2005. 68. Zaumseil, J., Friend, R., and Sirringhaus, H., Spatial control of the recombination zone in an ambipolar light-emitting organic transistor, Nat. Mater., 5, 69, 2006. 69. Chesterfield, R. J. et al., High electron mobility and ambipolar transport in organic thin-film transistors based on a π-stacking quinoidal terthiophene, Adv. Mater., 15, 1278, 2003. 70. Misewich, J. A. et al., Electrically induced optical emission from a carbon nanotube FET, Science, 300, 783, 2003. 71. Podzorov, V. et al., High-mobility field-effect transistors based on transition metal dichalcogenides, Appl. Phys. Lett., 84, 3301, 2004.

8080_C002.fm Page 71 Friday, April 6, 2007 10:21 AM

Charge Carrier Transport in Single-Crystal Organic Field-Effect Transistors

71

72. Yasuda, T., and Tsutsui, T., Organic field-effect transistors based on high electron and ambipolar carrier transport properties of copper–phthalocyanine, Chem. Phys. Lett., 402, 395, 2005. 73. Meijer, E. J. et al., Solution-processed ambipolar organic field-effect transistors and inverters, Nat. Mater., 2, 678, 2003. 74. de Boer, R. W. I. et al., Ambipolar Cu- and Fe-phthalocyanine single-crystal fieldeffect transistors, Appl. Phys. Lett., 86, 262109, 2005. 75. Takahashi, T. et al., Ambipolar organic field-effect transistors based on rubrene single crystals, Appl. Phys. Lett., 88, 033505, 2006. 76. Bao, Z., Lovinger, A. J., and Dodabalapur, A., Organic field-effect transistors with high mobility based on copper phthalocyanine, Appl. Phys. Lett., 69, 3066, 1996. 77. Bube, R. H., Photoconductivity in solids, Wiley, New York, 1960. 78. Brédas, J. L. et al., Organic semiconductors: A theoretical characterization of the basic parameters governing charge transport, Proc. Nat. Acad. Sci., 99, 5804, 2002. 79. Cheng, Y. C. et al., Three-dimensional band structure and bandlike mobility in oligoacene single crystals: A theoretical investigation, J. Chem. Phys., 118, 3764, 2003. 80. de Wijs, G. A. et al., Anisotropy of the mobility of pentacene from frustration, Synth. Met., 139, 109, 2003. 81. Mas-Torrent, M. et al., Correlation between crystal structure and mobility in organic field-effect transistors based on single crystals of tetrathiafulvalene derivatives, J. Am. Chem. Soc., 126, 8546, 2004. 82. Podzorov, V. et al., Hall effect in the accumulation layers on the surface of organic semiconductors, Phys. Rev. Lett., 95, 226601, 2005. 83. Takeya, J. et al., Hall effect of quasi-hole gas in organic single-crystal transistors, J. Appl. Phys., 44, L1393, 2005. 84. Karl, N. et al., Fast electronic transport in organic molecular solids? J. Vac. Sci. Technol. A, 17, 2318, 1999. 85. Jurchescu, O. D., Baas, J., and Palstra, T. T. M., Effect of impurities on the mobility of single crystal pentacene, Appl. Phys. Lett., 84, 3061, 2004. 86. Thorsmølle, V. K. et al., Ultrafast conductivity dynamics in pentacene probed using terahertz spectroscopy, Appl. Phys. Lett., 84, 891, 2004. 87. Ostroverkhova, O. et al., Bandlike transport in pentacene and functionalized pentacene thin films revealed by subpicosecond transient photoconductivity measurements, Phys. Rev. B, 71, 035204, 2005. 88. Troisi, A., and Orlandi, G., Charge-transport regime of crystalline organic semiconductors: Diffusion limited by thermal off-diagonal electronic disorder, Phys. Rev. Lett., 96, 086601, 2006. 89. Rang, Z. et al., Hydrostatic pressure dependence of charge carrier transport in singlecrystal rubrene devices, Appl. Phys. Lett., 86, 123501, 2005. 90. Veres, J. et al., Low-k insulators as the choice of dielectrics in organic field-effect transistors, Adv. Funct. Mater., 13, 199, 2003. 91. Facchetti, A., Yoon, M.-H., and Marks, T. J., Gate dielectrics for organic field-effect transistors: New opportunities for organic electronics, Adv. Mater., 17, 1705, 2005. 92. Hebard, A. F. et al., Conducting films of C60 and C70 by alkali-metal doping, Nature, 350, 600, 1991. 93. Kochanski, G. P. et al., Electrical resistivity and stoichiometry of KxC60 films, Science, 255, 184, 1992.

8080_C002.fm Page 72 Friday, April 6, 2007 10:21 AM

72

Organic Field-Effect Transistors

94. Panzer, M. J., and Frisbie, C. D., High charge carrier densities and conductance maxima in single-crystal organic field-effect transistors with a polymer electrolyte gate dielectric, Appl. Phys. Lett., 88, 203504, 2006. 95. Wehrli, S., and Helm, C., Interface steps in field effect devices, J. Appl. Phys., 95, 5621, 2004. 96. Narayan, K. S., and Kumar, N., Light responsive polymer field-effect transistor, Appl. Phys. Lett., 79, 1891, 2001. 97. Hamilton, M. C., and Kanicki, J., Organic polymer thin-film transistor photosensors, IEEE J. Selected Topics Quantum Electronics, 10, 840, 2004; Noh, Y.-Y. et al., Highphotosensitivity p-channel organic phototransistors based on a biphenyl end-capped fused bithiophene oligomer, Appl. Phys. Lett., 86, 043501, 2005; Saragi, T. P. I. et al., Organic phototransistor based on intramolecular charge transfer in a bifunctional spiro compound, Appl. Phys. Lett., 84, 2334, 2004. 98. Podzorov, V., Pudalov, V. M., and Gershenson, M. E., Light-induced switching in back-gated organic transistors with built-in conduction channel, Appl. Phys. Lett., 85, 6039, 2004. 99. Najafov, H. et al., Primary photoexcitations and the origin of the photocurrent in rubrene single crystals, Phys. Rev. Lett., 96, 056604, 2006. 100. Schmechel, R., and von Seggern, H., Electronic traps in organic transport layers, Phys. Stat. Sol., 201, 1215, 2004. 101. Lang, D. V. et al., Amorphouslike density of gap states in single-crystal pentacene, Phys. Rev. Lett., 93, 086802, 2004. 102. Crone, B. et al., Electronic sensing of vapors with organic transistors, Appl. Phys. Lett., 78, 2229, 2001. 103. Zhu, Z.-T., Mason, J. T., Dieckmann, R., and Malliaras, G. G., Humidity sensors based on pentacene thin-film transistors, Appl. Phys. Lett., 81, 4643, 2002. 104. Jurchescu, O. D., Baas, J., and Palstra, T. T. M., Electronic transport properties of pentacene single crystals upon exposure to air, Appl. Phys. Lett., 87, 052102, 2005. 105. Nardello, V. et al., Photochemistry without light: Oxidation of rubrene in a microemulsion with a chemical source of singlet molecular oxygen (1O2, 1Dg), J. Chemical Education, 76, 1285, 1999. 106. Käffer, D., and Witte, G., Growth of crystalline rubrene films with enhanced stability, Phys. Chem. Chem. Phys., 7, 2850, 2005. 107. Takeya, J. et al., Effects of polarized organosilane self-assembled monolayers on organic single-crystal field-effect transistors, Appl. Phys. Lett., 85, 5078, 2004. 108. Bürgi, L. et al., Close look at charge carrier injection in polymer field-effect transistors, J. Appl. Phys., 94, 6129, 2003. 109. Payne, M. M. et al., Organic field-effect transistors from solution-deposited functionalized acenes with mobilities as high as 1 cm2/V·s, J. Am. Chem. Soc., 127, 4986, 2005. 110. Klare, J. et al., Cruciform systems for molecular electronics applications, J. Am. Chem. Soc., 125, 6030, 2003. 111. Stingelin-Stutzmann, N. et al., Organic thin-film electronics from vitreous solutionprocessed rubrene hyperereutectics, Nat. Mater., 4, 601, 2005.

8080_C002.fm Page 73 Friday, April 6, 2007 10:21 AM

2.2

Charge Transport in Oligomers

Gilles Horowitz CONTENTS 2.2.1 Introduction..................................................................................................73 2.2.2 Operating Mode of the Organic Thin-Film Transistor ...............................75 2.2.3 Charge Transport in Conjugated Oligomers ...............................................77 2.2.3.1 Band Transport ..............................................................................78 2.2.3.2 Polaron Transport ..........................................................................80 2.2.3.2.1 Polarization in Molecular Crystals..............................80 2.2.3.2.2 Molecular Polaron .......................................................82 2.2.3.2.3 Marcus Model..............................................................83 2.2.3.3 Hopping Transport.........................................................................85 2.2.4 Trap Limited Transport in Organic Transistors...........................................86 2.2.5 Parameter Extraction....................................................................................89 2.2.5.1 Threshold Voltage..........................................................................91 2.2.5.2 Contact Resistance ........................................................................91 2.2.5.2.1 Contact Resistance Extraction.....................................91 2.2.5.2.2 Origin of Contact Resistance ......................................94 2.2.5.3 Mobility Degradation ....................................................................95 2.2.6 Concluding Remarks....................................................................................97 Acknowledgments....................................................................................................97 References................................................................................................................99

2.2.1 INTRODUCTION Solids are made of atoms, which in turn are composed of an ion core (i.e., the nucleus and those electrons so strongly bound to it that they cannot move around) and valence electrons (electrons in the outermost shell). Solid-state physics textbooks [1] teach us that the most important physical distinction in solids, as established on the basis of the configuration of valence electrons, is that between metals and insulators. The difference depends on whether there is (metals) or is not (insulators) any partially filled energy band. In perfect crystals at zero temperature, this is a rigorous criterion leading to unambiguous categories. The concept of (intrinsic) semiconductor only emerges at nonzero temperature in insulators with small energy

73

8080_C002.fm Page 74 Friday, April 6, 2007 10:21 AM

74

Organic Field-Effect Transistors

gap, where thermal excitation of electrons from the valence band to the conduction band may occur. Because their electrical properties so strongly depend on temperature, intrinsic semiconductors are of no practical use for realizing microelectronic devices, which instead are based on the concept of extrinsic semiconductors, in which charge carriers are induced by intentionally added impurities. Based on the configuration of the valence electrons, it is of common practice to make a distinction among four classes of solids [1]: 1. The outstanding example of molecular crystals is the solid noble gases. They have completely filled electronic shells, so there is little electronic density between ion cores; all the electrons remain in the vicinity of their parent ions. For this reason, molecular crystals are insulators. 2. Like in molecular crystals, the electronic charge distributions in ionic crystals are highly localized in the neighborhood of the ion cores. However, some electrons are now bound to the component of the opposite type. As such, ionic crystals can be viewed as molecular crystals in which the constituents are not atoms or molecules, but ions. Ionic crystals are also insulators. 3. In metals, unlike the two previous cases, the density of electrons remains appreciable even in between the ion cores, leading to the fact that metals are electronic conductors. 4. The situation in covalent crystals is something in between that of insulators and metals. In covalent crystals, valence electrons are not sharply localized near the ion cores. However, the density of electrons is not uniform; instead, it concentrates along certain preferred directions, leading to chemical bonds. It is interesting to note that conventional semiconductors (and especially silicon) are covalent crystals. The electronic charge distribution in the four categories of solids is schematically displayed in Figure 2.2.1. Organic molecular crystals, that is, crystals made of organic molecules [2], obviously belong to the class of molecular crystals. It may therefore sound paradoxical to call some of these crystals — namely, those made of conjugated molecules — organic semiconductors [3,4]. The expression actually emerged when it was found that the electrical resistance of crystals of various phthalocyanines, measured between two metal electrodes, presented the thermally activated behavior characteristic of intrinsic semiconductors [5,6], an observation that was extended later to other conjugated organic solids [7–10]. Although the interpretation of the thermally activated resistivity was probably flawed because the presence of impurities and crystal defects was not taken into account, these findings opened the way to further investigations on the subject. The practical reason for such a development was the prospect of joining organic chemistry with electronics. The phenomena of photoconductivity, electroluminescence, superconductivity, and the photovoltaic effect have all been identified in organic crystals [26]. However, the launch of organic electronics had to wait until it was demonstrated that real devices, such as light-emitting diodes [12] and field-effect transistors [13], could be indeed fabricated with organic semiconductors.

8080_C002.fm Page 75 Friday, April 6, 2007 10:21 AM

Charge Transport in Oligomers

75

18+ 18−

18+ 18−

18+ 18−

18+ 18−

17+ 18−

19+ 18−

17+ 18−

19+ 18−

18+ 18−

18+ 18−

18+ 18−

18+ 18−

19+ 18−

17+ 18−

19+ 18−

17+ 18−

(a)

(b)

Molecular (Ar)

Ionic (KCl)

6+

6+

6+

6+

19+

19+

19+

19+

6+

6+

6+

6+

19+

19+

19+

19+

(c)

Covalent (C)

(d)

Metallic (C)

FIGURE 2.2.1 Schematic two-dimensional view of the electronic space distribution in the four basic solid categories. The small circles represent the nuclei and the shaded areas the regions where electron density is significant. (a) Molecular (argon); (b) ionic (potassium chloride); (c) covalent (carbon-diamond); (d) metallic (potassium).

Because conjugated organic solids are more like insulators than semiconductors, charge transport in these materials is much less efficient than in conventional semiconductors. The problem is more crucial in transistors, where charges have to travel along much longer ways than in diodes. At the current state of the art, mobility in organic thin-film transistor (OTFT) ranges between 0.01 and 10 cm2/Vs, which is still much lower than what found in inorganic semiconductors (mobility is around 1,000 cm2/Vs in silicon), but substantially higher than the typical numbers for organic light-emitting diodes or photovoltaic cells. “High” mobility in OTFT is the result of large research efforts at improving structural order in the organic semiconductor film. It is worth pointing out that the highest mobility is achieved with organic single crystals. However, the physical origin of high mobility in organic solids is still an unresolved theoretical question. It is the aim of this chapter to account for the theoretical efforts that have been made to understand charge carrier transport in organic oligomers and small molecules.

2.2.2 OPERATING MODE OF THE ORGANIC THIN-FILM TRANSISTOR The basic idea that guides the insulated-gate field-effect transistor (FET) traces back to the mid-1920s [14], but it was not until 1960 that this early concept could be successfully demonstrated, with the invention of the metal-oxide-semiconductor FET (MOSFET) [15]. Field-effect measurements on copper phthalocyanine films were

8080_C002.fm Page 76 Friday, April 6, 2007 10:21 AM

76

Organic Field-Effect Transistors

L Semiconductor

Z

Drain Insulator Source Gate Substrate

FIGURE 2.2.2 Three-dimensional view of an organic thin-film transistor.

reported as early as 1964 [16]. However, organic transistors that could indeed be used in practical electronic circuits only appeared in the late 1980s [13,17]. Figure 2.2.2 shows a three-dimensional view of an organic thin-film transistor. The active part of the device is constituted of an organic semiconductor thin film equipped with two electrodes: the source and the drain. The distance between the source and the drain is called the channel length L ; the transverse dimension of the structure is the channel width Z . A third electrode, the gate, is laid out along the channel between source and drain; this electrode is electrically isolated from the semiconductor film by a thin insulating film, hence forming a metal-insulatorsemiconductor (MIS) structure. By applying a voltage VG between source and gate, one induces the formation of an accumulation layer at the semiconductor–insulator interface, thus forming a conducting channel between these two electrodes. A unique feature of FETs is that, unlike diodes, these devices are two dimensional. That is, they are governed by two independent voltages perpendicular to each other. The role of the gate voltage VG is to induce charges in the conducting channel, while the drain voltage VD drives these charges from source to drain. The most popular equations to describe the current-voltage curves of an FET are very simple, but this simplicity hides several assumptions that are not usually fulfilled in organic devices. In short, I–V characteristics can be drawn by either varying the drain voltage at a constant gate voltage (output characteristics) or changing the gate voltage at a fixed drain voltage (transfer characteristics). In the former case, the curves are divided into a linear regime at low VD that convert into the saturation regime when VD > VG . The current in both regimes is given by Equations (2.2.1) and (2.2.2) [18]: I Dlin =

⎡ Z V2 ⎤ Ciμ ⎢(VG − VT ) VD − D ⎥ 2 ⎦ L ⎣

(2.2.1)

Z 2 Ciμ (VG − VT ) 2L

(2.2.2)

I Dsat =

Here, C i is the capacitance of the insulator, μ the mobility in the semiconductor, and VT the threshold voltage. The meaning of the latter parameter will be detailed later. In short, VT is the gate voltage beyond which the conducting channel forms.

8080_C002.fm Page 77 Friday, April 6, 2007 10:21 AM

Charge Transport in Oligomers

77

Apart from showing the performance of the device, I–V curves are used to extract its basic parameter, primarily the mobility and threshold voltage. A widely used method for parameter extraction consists of plotting the square root of the saturation current as a function of the gate voltage. As seen in Equation (2.2.2), this is supposed to give a straight line, the slope of which gives μ while its extrapolation to the VG axis corresponds to the threshold voltage. Equations (2.2.1) and (2.2.2) rests on the following assumptions: (1) The transverse electric field induced by the gate voltage is largely higher than the longitudinal field induced by the gate bias (so-called gradual channel approximation); (2) the mobility is constant all over the channel. Assumption (1) is justified by the geometry of the device; that is, the distance from source to drain is most often much larger than the thickness of the insulator. Assumption (2) is more or less fulfilled in a conventional inorganic semiconductor. However, this is far from true in organic solids, as will be shown in this chapter. For this reason, the use of Equation (2.2.2) to extract the mobility may lead to an incomplete, if not erroneous, description of charge transport in organic semiconductors. Alternative approaches to circumvent this difficulty will be presented in the following sections.

2.2.3 CHARGE TRANSPORT IN CONJUGATED OLIGOMERS In contrast to its parent elements of column IV of the periodic table (Si, Ge …), carbon presents the unique feature of being able to exist under three different hybridization configurations: namely, sp, sp2, and sp3. The latter one is found in the so-called saturated compounds that are the constituting element of plastics. In this configuration, each carbon atom is linked to its neighbors by four strong σ bounds that point to the four verges of a tetrahedron. Because σ bounds are so strong, the distance between the bonding and antibonding energy levels (also called highest occupied and lowest unoccupied molecular orbitals — HOMO and LUMO) is high, which has two consequences: Plastics are transparent to visible light, and they are electrically insulating. All the organic compounds designated as semiconductors are those made of sp 2 hybridized carbons, also called conjugated organic materials. Under such circumstances, each carbon is linked to its neighbors by three σ bonds resulting from the hybridization of 2s , 2 p x , and 2 py orbitals, while the remaining 2 pz orbital forms a π bond, which presents significantly less overlap than σ bonds. For this reason, the energy distance between the bonding and antibonding orbitals is somewhat reduced, which has two consequences: The materials absorb visible light (dyes are conjugated materials) and may behave as a semiconductor at nonzero temperature. This concept is illustrated in Figure 2.2.3 in the case of ethylene C2H2. In larger molecules, typically benzene, the π orbitals become delocalized and form a π system that extends all over the molecule. The HOMO–LUMO gap becomes smaller with increasing delocalization. In the case of a long chain of carbon atoms, the π bonds delocalize over the whole chain and form a one-dimensional electronic system. The resulting

8080_C002.fm Page 78 Friday, April 6, 2007 10:21 AM

78

Organic Field-Effect Transistors σ+ π+ pz

pz π−

sp2

sp2

σ−

FIGURE 2.2.3 Energy scheme of ethylene C2H2. . . n Conduction band (π∗) pz sp2

Valence band (π)

FIGURE 2.2.4 Molecular and energy schemes of poly-para-phenylene-vinylene (PPV).

one-dimensional band has substantial band width, and the chain can be viewed as a one-dimensional semiconductor with a filled valence band originating from the HOMO and an empty conduction band coming from the LUMO. Figure 2.2.4 illustrates this image with the molecule of poly-para-phenylene-vinylene (PPV). The image depicted in Figure 2.2.4 gives rationale for why charge carriers can be injected and reside in a conjugated molecule. However, the limiting step for charge transport in a solid is not within single molecules; rather, it involves electron transfer between molecules or molecular chains; because orbital overlap between molecules is low, the phenomenon of charge transport in conjugated solids requires further investigations.

2.2.3.1 BAND TRANSPORT Band transport refers to the mechanism occurring in crystalline inorganic solids like metals and semiconductors. Band theory can be found in numerous textbooks [1] and will not be detailed here. In short, energy bands in solids form because when a very large number of interacting atoms are brought together, their energy levels become so closely spaced that they become indistinct. Any solid has a large number of energy bands, but not all these bands are filled with electrons. The likelihood of any particular band to be filled is given by the Fermi–Dirac statistics, Equation

8080_C002.fm Page 79 Friday, April 6, 2007 10:21 AM

Charge Transport in Oligomers

79

(2.2.3), so that at zero temperature, bands are filled up to the so-called Fermi energy E F . f (E ) =

1 1 + exp E −kTEF

(2.2.3)

On this basis, solids can be divided into insulators, in which the highest occupied band (the valence band) is completely filled, while the lowest unoccupied band (the conduction band) is completely empty and metals present a partly empty and partly filled band (the conduction band). Semiconductors are a particular case of insulators where the energy gap between the top of the valence band and the bottom of the conduction band is small enough that, at nonzero temperature, the smoothing out of the Fermi–Dirac distribution causes an appreciable number of states at top of the valence band to be empty and an equivalent number of states at bottom of the conduction band to be filled. Note that the conductivity in semiconductors is highly temperature dependent. The simplest model of charge transport in delocalized bands is the Drude model, which assumes the carriers are free to move under the influence of an applied electric field, but subject to collisional damping forces. Note that the scattering centers are not the nuclei of the background material, but rather phonons (lattice vibrations) or impurities. A statistical equation for estimating the mean drift velocity of the carriers in the direction of the electric field Fx may be written as d q 1 vx = ∗ Fx − vx dt m τ

(2.2.4)

where q is the elemental charge and m ∗ the effective mass. τ is the mean free time between two collisions (or relaxation time). Steady state corresponds to dtd v x = 0 . Under such circumstances, the solution of Equation (2.2.4) writes vx =

qτ Fx = μFx m∗

(2.2.5)

which defines the mobility μ . It is important to note at this stage that the model is only valid when the mean free path λ — that is, the mean distance between two collisions — is much larger than a characteristic distance that can be the de Broglie length of the charge carrier, or the distance between two atoms in the crystal. The mean free path is given by λ = vth τ

(2.2.6)

where vth = 3 kT / m ∗ is the electron thermal velocity (~105 m/s at room temperature). From Equations (2.2.5) and (2.2.6) the mobility can also be defined as

8080_C002.fm Page 80 Friday, April 6, 2007 10:21 AM

80

Organic Field-Effect Transistors

μ=

qλ m∗vth

(2.2.7)

The temperature dependence of the mobility depends on the nature of the scattering centers (acoustical or optical phonons, charged impurities …) However, in all cases, it is found that the dependence follows the general law given by Equation (2.2.8) μ(T ) ∝ T − n

(2.2.8)

In most practical cases, n is positive, so the mobility increases when temperature decreases. Evidence for “band transport” is often claimed to be brought when the temperature dependence in Equation (2.2.8) is observed. The most celebrated example for such a behavior is that by Karl and coworkers on highly pure crystals of acenes [19]. However, as pointed out by Silinsh and Cápek [20], the argument does not resist further analysis because, at least for temperatures above 100 K, the mean free path calculated from Equation (2.2.5) falls below the distance between molecules in the crystal, which is not physically consistent with diffusion limited transport; so the exact nature of charge transport in these crystals is still unresolved for the time being.

2.2.3.2 POLARON TRANSPORT 2.2.3.2.1 Polarization in Molecular Crystals The main reason why the band model is unable to account for charge transport in organic semiconductors is that it fails to account for a crucial phenomenon in these materials: polarization. The occurrence of polarization in organic solids has been analyzed in detail by Silinsh and Cápek [20]. The principle is the following. A charge carrier residing on a molecular site tends to polarize its neighboring region. As the resulting formed polarization cloud moves with the charge, the traveling entity is no longer a naked charge, but a “dressed” charge, and the formed species is called a polaron. In conjugated solids, the main polarization effect is that on the clouds formed by the π-electrons. The principle is illustrated in Figure 2.2.5, where the conjugated molecules are symbolized by benzene rings; the hexagons represent the (fixed) core of the six carbon atoms, while the circles stand for the delocalized π-electrons. Under the effect of the positive charge on the central molecule, the π-electron rings tend to move towards the central molecule, thus creating an electric dipole, the magnitude of which is the greater as the molecule is closer to the center. In order to estimate the stability of the polaron, we can define two typical times: (1) the residence time τres corresponds to the average time a charge will reside on a molecule; (2) the electronic polarization time τel is the time it takes for the polarization cloud to form around the charge. An order of magnitude for both times can be estimated by using Heisenberg’s uncertainty principle

8080_C002.fm Page 81 Friday, April 6, 2007 10:21 AM

Charge Transport in Oligomers

81

+

FIGURE 2.2.5 The figure is a schematic representation of the formation of a polaron when a positive charge is placed on a molecule in a conjugated organic solid. The hexagons symbolize the core of the nuclei, while the circles represent the delocalized π-electrons.

τ< ~

ΔE

(2.2.9)

where ΔE is a characteristic energy. For the residence time, the pertinent energy is the width W of the allowed band, typically 0.1 eV in an organic semiconductor and 10 eV in an inorganic semiconductor, which gives a residence time of 10 −14 s for the former and 10 −16 s for the latter. As for the electronic polarization time, the corresponding energy is that of an electron transition — that is, the energy gap (~1 eV) — so the time of the order of 10 −15 s in both cases. Similarly to electronic polarization, other polarization mechanisms can be invoked: molecular polarization, which concerns the displacement of the nuclei of the molecule where the charge resides, and lattice polarization, which involves movements of the entire lattice. The energies and times corresponding to these processes are estimated from the intramolecular and lattice vibration frequencies. The energy and time of the various polarization processes are summarized in Table 2.2.1. Table 2.2.1 reveals a striking difference between inorganic and organic semiconductors. In the former, the localization time is shorter than all the possible polarization times. In other words, electrons and holes move so fast that the polarization cloud does not have enough time to form. This is actually included in the band theory through the so-called rigid-band approximation, which states that the band structure remains uncharged when a charge is injected in the solid. The situation is drastically different in organic materials. Here, the electronic polaron has long enough time to form, so the energy levels of a charged molecule are significantly shifted with respect to that of a neutral molecule, as shown in Figure 2.2.6. As for the molecular polarization, it forms in a time comparable to the residence time, so the situation varies from one compound to the other. Finally, the formation time of the lattice polaron is too long, so its occurrence is unlikely under all circumstances. The pertinent parameters in the energy diagram in Figure 2.2.6 are the polarization energies for positive P + and negative charge carriers P −. Both are composed

8080_C002.fm Page 82 Friday, April 6, 2007 10:21 AM

82

Organic Field-Effect Transistors

TABLE 2.2.1 Residence Time and Various Polarization Times

Residence τres Polarization

Energy (eV)

Time (s)

10 0.1 1 0.1 < 0.01

10–16 10–14 10–15 10–14 > 10–13

Inorganic SC Molecular SC Electronic τel Molecular τv Lattice τl

Note: The reference energy is bandwidth for residence time, energy gap for electronic polarization, molecular vibration (~1,000 cm–1) for molecular polarization, and lattice vibration (<100 cm–1) for lattice polarization.

VL EAG EAC

LUMO P−

IPC

IPG EG P+ HOMO Gas phase

Solid

FIGURE 2.2.6 Energy scheme of the electronic polaron in a molecular crystal. VL is the vacuum level, EA the electronic affinity, and IP the ionization potential. P+ and P– are the polarization energy for positive and negative charge, respectively, and EG the transport band gap. (From Silinish, E.A. and Cápek, V., Organic molecular crystals: Interaction, localization, and transport phenomena, AIP Press, New York, 1994.)

of various contributions that render their estimation complex. A detailed analysis of the process of electronic polarization in crystals of polyacenes can be found in Reference 21, together with theoretical and experimental determinations of the corresponding energies. Representative data are sketched in Table 2.2.2. 2.2.3.2.2 Molecular Polaron Complementary to the electronic polarization, a charge carrier created in a molecular solid also polarizes the intramolecular vibration modes of the molecule on which it is located as well as dipole active modes of the neighboring molecules, thus forming an extended ionic state. As already mentioned, the corresponding relaxation time τ v is comparable to the residence time τres . The new quasiparticle associated with this

8080_C002.fm Page 83 Friday, April 6, 2007 10:21 AM

Charge Transport in Oligomers

83

TABLE 2.2.2 Calculated and Measured Polarization Energy in Polyacenes P+ (eV)

P– (eV)

Crystal

Experimental

Calculated

Experimental

Calculated

Anthracene Tetracene Pentacene

1.65 1.72 1.63

1.52 1.52 1.48

1.09 1.10 1.17

1.16 1.07 1.02

Source: Silinish, E.A. and Cápek, V., Organic molecular crystals: Interaction, localization, and transport phenomena, AIP Press, New York, 1994.

process is termed molecular polaron. The relaxation energy Eb gained in the process corresponds to the difference between the optical (direct) and adiabatic energy gaps. An important feature in the rationalization of the temperature-dependent mobility of highly pure polyacenes is the temperature dependence of the effective mass expressed by meff (T ) = meff (0 )eT /T0

(2.2.10)

where meff (0 ) is the molecular polaron effective mass at T = 0 and T0 a characteristic temperature [20]. The great interest of the molecular polaron model is that it can account for the apparent contradiction between the band-like temperature-dependent mobility and a small mean-free path. The weak point of the approach developed by Silinsh and coworkers is that it is phenomenological. It is worth mentioning that a more consistent analytical theory has been formulated by Kenkre and coworkers [22]. The presentation of this model would go to far outside the scope of this chapter and will not be developed here. 2.2.3.2.3 Marcus Model An alternative approach to describe charge transport in polarizable media uses Marcus’s electron transfer (ET) theory. Originally, the theory aimed at describing electron transfer from an electron donor to an electron acceptor in solution, according to D + A → D+ + A −

(2.2.11)

It is of usual practice to represent the energy involved in the reaction in a diagram where the energy of the reactants VR (left-hand side in the equation) and the products VP (right-hand side of the equation) are given along a generalized coordinate q by a parabola:

8080_C002.fm Page 84 Friday, April 6, 2007 10:21 AM

84

Organic Field-Effect Transistors

V VP

VR

λ 2t DG ≠ qR

qC

qP

q

FIGURE 2.2.7 Potential energy of the charge transfer reaction in the dimer coordinate representation. λ is the reorganization energy, t the transfer integral, and ΔG≠ the energy barrier height.

VR =

p p (q − qR )2 , VP = (q − qP )2 2 2

(2.2.12)

Here, the subscripts R refer to the reactant and P to the product, and p is an oscillator strength. In the case where the electron transfer is faster than molecular reorganization (nonadiabatic transfer), the path for electron transfer in the dimer coordinate representation illustrated in Figure 2.2.7 can be decomposed in a vertical activation from the minimum of VR to the VP curve, followed by a relaxation to the equilibrium configuration of the product. Accordingly, the Marcus theory of electron transfer [23] introduces the reorganization energy λ λ=

f ( q R − q P )2 2

(2.2.13)

If we now introduce the transfer integral t (dotted line in Figure 2.2.7), the energy barrier height for reaction Equation (2.2.11) is given by ΔG ≠ =

( λ − 2 t )2 λ − t, t < λ 4λ 4

(2.2.14)

The transfer integral t critically depends on the structural organisation (i.e., crystal structure) of the organic solid. Several computational techniques have been developed to estimate the transfer integral; a recent review can be found in Newton [24]. Assuming a classical Arrhenius behavior, the electron transfer rate kET is given by ⎡ λ / 4 −t⎤ kET = A exp ⎢− ⎣ kT ⎥⎦

(2.2.15)

8080_C002.fm Page 85 Friday, April 6, 2007 10:21 AM

Charge Transport in Oligomers

85

TABLE 2.2.3 Calculated Reorganisation Energy and Transfer Integral in Polyacenes Naphthalene

Anthracene

Tetracene

Pentacene

187 37 74

137 48 96

113 69 138

97 98 196

Reorganisation energy (meV) Transfer integral (meV) (meV)

Source: Taken from Brédas, J.-L. et al., Chem. Rev., 104, 4971, 2004.

where A is a constant that depends on the frequency at which the electron gets across the energy barrier. Equation (2.2.15) poses an obvious problem; that is, it predicts that the electron transfer approaches zero as the temperature approaches zero, which is at variance with experiment. The reason for such a discrepancy is that the classical theory does not account for tunneling effects that occur at low temperature. A semiclassical treatment can overcome this problem; it leads to an equation of the form [25]: kET =

2π

⎡ λ / 4 −t⎤ t2 exp ⎢− ⎣ kT ⎥⎦ 4 πλkT

(2.2.16)

A connection can be established between the Marcus theory and the polaron model by noting that the mobility is connected to the electron transfer rate by [26] μ=

qa 2 kET kT

(2.2.17)

where a is the lattice parameter of the molecular crystal. A generalization of the Marcus theory establishes an important criterion for activationless (“band-like”) or localized transport; namely, the former occurs when 2t > λ , while the latter dominates when 2t < λ . A connection with the molecular polaron model developed before can be derived from the fact that the reorganization energy is linked to the the molecular polarization time, and the transfer integral to the residence time, so the first inequality can also write τres < τ v and the second one τres > τ v , which is precisely what was established previously. Table 2.2.3 gives values of calculated reorganization energy and transfer integral for the polyacene series [27]. It can be seen that localized transport is expected for naphthalene and anthracene and delocalized transport for pentacene; tetracene is located in between these two extreme cases.

2.2.3.3 HOPPING TRANSPORT The problem with hopping transport is that dozens of different models have been proposed, based on different physical principles and approximations. Moreover, most

8080_C002.fm Page 86 Friday, April 6, 2007 10:21 AM

86

Organic Field-Effect Transistors

of these models only give a qualitative description of charge transport, thus preventing the possibility of computational treatment. However, hopping models have proved useful in rationalizing charge transport in disordered materials, such as polymers. Time of flight (TOF) measurements have revealed that the carrier mobility in these organic materials is thermally activated. Another ubiquitous feature relates to the field dependence of μ obeying a ln μ ∝ F law, where F is the magnitude of the electric field. The current practice is to interpret this behavior in terms of Gill’s [28] or Poole–Frenkel (PF)-like equation − Δ0 −β F ) / kTeff μ = μ0e (

(2.2.18)

where 1 / Teff = 1 / T − 1 / T ∗ and β is the PF factor. The problem with Equation (2.2.18) is that it presents several physical inconsistencies, among which are the lack of physical meaning for the effective temperature and the fact that the actual values of the PF factor are far from that predicted by the conventional PF theory. The disorder model developed by Bässler [29] rests on the following assumptions: (1) Because of the randomness of the intermolecular interactions, the electronic polarization energy of a charge carrier located on a molecule is subject to fluctuations; (2) transport is described in terms of hopping among localized states; in analogy to optical absorption profiles, the DOS is described by a Gaussian distribution of variance σ ; (3) charge transport is random walk described by a generalized master equation of the Miller–Abrahams form [30]: ν = ν0 e−2 γΔRij e−Δεij / kT

(2.2.19)

where ΔRij is the intersite distance and Δεij the energy difference of the sites; and (4) in addition to the energetic disorder, there exists a position disorder with a Gaussian distribution of variance Σ (the so-called off diagonal disorder). From a Monte Carlo simulation, Bässler arrives at a universal law relating the mobility to the degree of both diagonal and off diagonal disorder: ⎫ ⎧ ⎡ ⎤ 2 ⎡ ⎛ 2 σ ⎞2⎤ ⎪⎪ ⎪⎪ ⎢⎛ σ ⎞ 2 ⎥⎥ ⎢ μ = μ 0 exp ⎢−⎜ ⎟ ⎥ exp ⎨⎪C ⎢⎜ ⎟ − Σ ⎥ F ⎬⎪ ⎝ ⎠ ⎢⎣ ⎝ 3 kT ⎠ ⎥⎦ ⎥⎦ ⎪⎭ ⎪⎩ ⎢⎣ kT

(2.2.20)

where C is an empirical constant.

2.2.4 TRAP LIMITED TRANSPORT IN ORGANIC TRANSISTORS In real organic transistors, charge transport is most of the time limited by localized states induced by defects and unwanted impurities. Clear evidence for such a process is given by the fact that the performance of the devices is strongly sample dependent.

8080_C002.fm Page 87 Friday, April 6, 2007 10:21 AM

Charge Transport in Oligomers

Transport band Localized levels

87

E DOS

FIGURE 2.2.8 Principle of charge transport limited by multiple trapping and thermal release.

Two useful models for accounting for such a trend are the multiple trapping and thermal release (MTR) and variable range hopping (VRH) models. While hopping transport is appropriate to describe charge transport in disordered materials, the MTR model [31] applies to well-ordered materials, prototypes of which are vapor deposited small molecules like pentacene or the oligothiophenes, where thermally activated mobility is often observed. The basic assumption of the model is a distribution of localized energy levels located in the vicinity of the transport band edge. During their transit in the delocalized band, the charge carriers interact with the localized levels through trapping and thermal release (Figure 2.2.8). The model rests on the following assumptions: (1) Carriers arriving at a trap are instantaneously captured with a probability close to one; and (2) the release of trapped carriers is controlled by a thermally activated process. The resulting effective mobility μ eff is related to the mobility μ 0 in the transport band by an equation of the form μ eff = μ 0αe−( Ec −Et )/ kT

(2.2.21)

where Ec is the energy of the transport band edge. In the case of a single trap level of energy Et and density of state (DOS) N t , the total charge-carrier concentration ntot splits into a concentration of free carriers n f = N c e − ( Ec − E F )/ kT , where N c is the effective density of states at transport band edge, and a concentration of trapped carriers nt = N t e − ( Et − E F )/ kT . The ratio of trapped to total densities is given by [32] Θ=

nt 1 N = t e−( Ec −Et )/ kT nt + n f 1 + NNct e( Ec −Et )/ kT N c

(2.2.22)

In that instance, the effective mobility is μ eff = Θμ 0 , so that Et in Equation (2.2.21) is the energy of the single trap level and α the ratio of the trap DOS to the effective density of states at transport band edge. If traps are energy distributed, distribution-dependent effective values of Et and α must be estimated, as will be exemplified in the following. In all circumstances, whichever the actual energy distribution of traps, the main feature predicted by the MTR model is thermally activated mobility. An important outcome of the MTR model is that in the case of an energy distributed DOS, mobility is gate voltage dependent. The mechanism at work is schematically pictured in Figure 2.2.9.

8080_C002.fm Page 88 Friday, April 6, 2007 10:21 AM

88

Organic Field-Effect Transistors

VS

EC EF

DOS VG − VT = 0

VG − VT ≠ 0

FIGURE 2.2.9 Gate voltage dependent mobility induced by an energy distributed density of traps.

When a bias is applied to the gate, a potential Vs develops at the insulator–semiconductor interface, which results in shifting by the same amount the Fermi level towards the transport band edge, thus partly filling the distribution of localized states. Accordingly, the energy distance between the filled traps and the transport band edge is reduced, so trapped-carrier release is made easier, and the effective mobility increases. Such a gate-voltage dependence of mobility has indeed been reported on several occasions [32,33]. The shape of the gate voltage dependence depends on that of the DOS. We have seen before that in hopping models, the preferred DOS distribution is a Gaussian distribution. By analogy with what is found in hydrogenated amorphous silicon [31], the trap distribution used in the MTR model is an exponential band tail. This is because the trap distribution no longer corresponds to the transport band itself (as in the case of the hopping model); instead, the DOS is a tail to a delocalized transport band. Note, however, that an exponential tail distribution can also be associated to a Gaussian transport DOS. One of the major interests of the exponential distribution is that it leads to an analytical form of the gate voltage dependence of the mobility. The general form of an exponential distribution of traps is given by

Nt (E ) =

N t 0 −( Ec −E )/ kT0 e kT0

(2.2.23)

where N t0 is the total density (per unit area) of traps and T0 a characteristic temperature that accounts for the slope of the distribution. The previously defined trapped charge is connected to the density of traps through

nt = q

∫

+∞ −∞

N t ( E ) f ( E ) dE

(2.2.24)

where f ( E ) is the Fermi distribution. If N t ( E ) is a slowly varying function, the Fermi distribution can be approximated to a step function; that is, it equals zero for E < E F and one for E > E F . The integration of Equation (2.2.24) leads to

8080_C002.fm Page 89 Friday, April 6, 2007 10:21 AM

Charge Transport in Oligomers

89

nt qN t ( EF 0 + qVs ) = nt 0 eqVs / kT0

(2.2.25)

As stated earlier, we have made use of the fact that the Fermi level E F is shifted towards the band edge Ec from the value E F 0 at zero gate bias by an amount qVs (see Figure 2.2.9). nt 0 = N t 0 e − ( Ec − E F 0 )/ kT0 is the density of trapped charge at zero gate voltage. Making use of the defined effective mobility and assuming n f << nt , we finally obtain [34] T0 −1

μ eff = μ 0

N c ⎛ Ci (VG − VT ) ⎞ T ⎜ ⎟ N t 0 ⎝ qN t 0 ⎠

(2.2.26)

Note that the gate voltage dependence has the form of a power law in VG − VT [32]. It is worth pointing out that in transistors made with single crystal, the mobility is found to be very seldom gate voltage dependent [35], which indirectly confirms that the VG dependence originates from localized levels associated with chemical and physical defects.

2.2.5 PARAMETER EXTRACTION Parameter extraction is a central issue in settling the debates between charge transport mechanisms in organic transistors. Before depicting in more details the different methods developed for parameter extraction, let us first go back to Equations (2.2.1) and (2.2.2) and their relevance to actual current-voltage curves. Because the drain current I D depends on two independent voltages — the drain voltage VD and the gate voltage VG — the current-voltage (I–V) curves can be plotted in two ways: the output characteristic, where a set of drain current versus drain current curves is drawn for various values of the gate voltage, and the transfer characteristic, in which the drain current is plotted as a function of the gate voltage for a given drain voltage. Both representations are shown in Figure 2.2.10. Parameter extraction mainly involves mobility and threshold voltage. Because of the very simple analytical form of Equation (2.2.2), the most widespread technique to extract mobility and threshold voltage consists of plotting the square-root of I Dsat as a function of gate voltage in the saturation regime (that is, measured at a drain voltage higher than the gate voltage). However, it must be remembered that the simplicity of Equation (2.2.2) implies two assumptions: (1) gradual channel approximation; and (2) constant mobility. There are now several pieces of evidence that the second criterion is not fulfilled in organic transistors. One is the previously developed trapping model, which predicts that mobility increases with gate voltage. A second one, which will be analyzed in the following, is mobility degradation, a process by which mobility tends to decrease as the gate voltage increases. Why the saturation regime poses a problem to extracting parameters under these circumstances is illustrated in Figure 2.2.11.

8080_C002.fm Page 90 Friday, April 6, 2007 10:21 AM

90

Organic Field-Effect Transistors

−2.5 10−6

VG = 0 V VG = −2 V

Drain current (A)

−2 10−6

VG = −4 V VG = −6 V

−1.5 10−6 −1 10−6 −5 10−7 0 0

−1

−2 −3 −4 Drain voltage (V)

−5

−6

10−6

Drain current (A)

10−7 10−8 10−9 10−10 10−11 10−12 10−13

5

0

−5 Gate voltage (V)

−10

FIGURE 2.2.10 Output (top) and transfer (bottom) characteristics of an organic transistor.

Conducting channel Source VG

Drain VG – VD ≈ VG Gate Linear regime

Source VG

Drain VG – VD ≈ 0 Gate Saturation regime

FIGURE 2.2.11 Charge distribution along the channel in a thin-film transistor. In the linear regime (left), the voltage at source and drain is nearly identical, so the charge density is constant along the channel. In the saturation regime (right), the voltage is VG at source and practically zero at drain, and the charge density drops to zero when passing from source to drain. Note that the thickness of the channel represents the charge density, not its spatial extension.

8080_C002.fm Page 91 Friday, April 6, 2007 10:21 AM

Charge Transport in Oligomers

91

The transistor must be thought of as a two-dimensional structure, a crucial consequence being that the voltages at source and drain are not necessarily equal; more precisely, the voltage is VG at source and VG − VD at drain, so, depending on the respective values of VG and VD , the electrical potential at the insulator–semiconductor interface may not be constant all along the channel. In the linear regime, VD << VG , the potential does not noticeably vary. This is not the case in the saturation regime, where VD ≥ VG . When reaching the so-called pinch-off voltage (VD VG − VT ), the voltage at the drain drops to zero, and the potential along the channel deceases from VG to zero when passing from the source to the drain. Beyond the pinch-off voltage, the pinch-off point gradually moves towards the source, and the potential variation along the channel retains this magnitude. If the mobility is gate-bias dependent, an exact estimation of the saturation current would require integrating the mobility all along the channel, which does not appear easily feasible because the gate-voltage dependence of the mobility for a given device is not known beforehand. For this reason, parameter extraction from the linear regime is the preferred technique in conventional MOSFETs and will be the one adopted here.

2.2.5.1 THRESHOLD VOLTAGE In conventional MOSFETs, the threshold voltage is an important technological parameter that plays a crucial role in circuit modeling. For this reason, numerous methods have been developed to extract the threshold voltage; most of them are performed in the linear regime. A recent review of these methods can be found in Ortiz-Conde et al. [36]. As already stated, the quasi-universal method used in OTFTs consists of plotting the square root of the saturation current as a function of the gate voltage. According to Equation (2.2.2), this should result in a straight line that crosses the VG axis at VG = VT . However, because of several effects, such as gate-voltage dependent mobility and contact resistance, the actual curve is not a real straight line. Instead, it presents an upward curvature at low gate voltages and a downward curvature at high voltages. A representative example is given in Figure 2.2.12. On this kind of curve, the extracted mobility and threshold voltage largely depend on the voltage range used for the linear fitting. The most widespread usage is to center the linear regression curve around the inflection point, but the basis for doing so is purely empirical. To illustrate an alternative method to extract VT in the linear regime, we have selected the transconductance change (TC) method developed by Wong and coworkers [37], which consists of plotting the second derivative of the linear drain current as a function of gate voltage (Figure 2.2.13). In this case, VT is given by the peak of the curve. Importantly, such a determination is insensitive to both the contact resistance and the gate-voltage dependence of the mobility.

2.2.5.2 CONTACT RESISTANCE 2.2.5.2.1 Contact Resistance Extraction Contact resistance extraction cannot be performed from simple current-voltage curve. At this stage, specific techniques that allow an independent access to the

8080_C002.fm Page 92 Friday, April 6, 2007 10:21 AM

92

Organic Field-Effect Transistors

0.004 0.0035

ID0.5 (A0.5)

0.003 0.0025 0.002 0.0015 0.001 0.0005 0

0

–5

–10 Gate voltage (V)

–15

–20

FIGURE 2.2.12 Representative example of an actual I Dsat versus VG curve. The dotted line is a linear fit to the data ranging between –10 and –15 V. Note the upward curvature of the curve at low gate voltage and its downward curvature at high voltage.

∂2/D /∂VG2 (AV–2)

2 10–8 SAM L = 10 μm

2 10–8

0

VT

–1 10–8 –2 10–8

5

0

–5 –10 –15 Gate voltage (V)

–20

–25

FIGURE 2.2.13 TC method for the extraction of the threshold voltage from the linear regime.

channel and contact resistances are needed. A first technique is the transfer line method (TLM) [38–41], a method adapted from a classical technique to estimate contact resistance, and first adopted in the case of the amorphous silicon thin-film transistor [42]. The method consists of measuring the channel resistance of similar devices with various channel lengths. The measured resistance is actually the sum of the channel and contact resistances. As long as the measurement is performed in the linear regime (small drain voltage), the channel resistance is proportional to L (see Equation (2.2.1)) and the width-normalized ( R × W ) total resistance is given by R ×W =

L + Rc × W C i μ(VG − VT )

(2.2.27)

8080_C002.fm Page 93 Friday, April 6, 2007 10:21 AM

Charge Transport in Oligomers

93

16 14

RonW (Ωcm) × 104

12 10 8 6 4

−l 0

2 0 –20 –10 0 10 20 30 40 50 Channel length (μm)

FIGURE 2.2.14 Illustration of the transfer line method. (From Zaumseil, J. et al., J. Appl. Phys., 93, 6117, 2003.)

The contact resistance is extracted by plotting the width-normalized resistance as a function of channel length. The extrapolation to zero length readily gives the contact resistance, while the slope of the curve can be used to extract device parameters. The method is exemplified in Figure 2.2.14. Each line corresponds to a given gate voltage. Figure 2.2.14 shows that the contact resistance is actually gate voltage dependent; that is, it decreases when gate voltage increases. The reason for such a behavior will be given in the next section. However, the TLM presents several drawbacks. First of all, it requires measurements on different devices, and it cannot be taken for granted that the channel and contact resistances are strictly similar for all of them, even if they are prepared during the same run. This is the reason why scattering appears when plotting the data, as shown in Figure 2.2.14. Next, the validity of Equation (2.2.27) requires that the contact resistance follows Ohm’s law. In other words, the method cannot be used in the case of nonlinear contact resistance. It must also be noted that, as the method requires measurements in the linear regime (that is, at low drain voltages), it is very sensitive to leaks through the insulator. Finally, the method cannot make a distinction between the contact resistance at the source and that at the drain. This last point has its importance; theoretical modeling predicts that in the case of ideal contacts, all the ohmic drop should occur at the source electrode. An alternative method to TLM is the four-point probe, which consists of introducing in the conducting channel two additional electrodes [43,44]. Being a conservative magnitude, the current remains the same all along the channel and the voltage drop between these two additional electrodes is not affected by the contact resistance, thus giving access to the true channel resistance. Moreover, as shown in Figure 2.2.15, the contact resistance at each side of the channel can now be independently estimated.

8080_C002.fm Page 94 Friday, April 6, 2007 10:21 AM

94

Organic Field-Effect Transistors

Source

V1

V2

Drain

VD ΔVD ΔVS V2

V1 VS = 0

FIGURE 2.2.15 Principle of the four-probe technique.

2.2.5.2.2 Origin of Contact Resistance The most commonly used image to describe source and drain contacts is that of a metal–semiconductor junction. According to the conventional Mott–Schottky (MS) model [18], contacts are expected to be ohmic when the work function of the metal is close to the HOMO or LUMO level of the semiconductor, depending on whether the semiconductor is p- or n-type. If the reverse situation prevails, an energy barrier forms at the metal–semiconductor interface, leading to poor charge injection. From this standpoint, the Au–pentacene interface would be a good candidate as lowresistance contact because the ionization potential IP of pentacene, which measures the energy distance between the HOMO level and the vacuum level E vac , is close to the work function Φ of gold (see Figure 2.2.16). In practice, the actual resistance is rather high. The mechanism of barrier formation at metal–organic semiconductor interfaces has been studied in great detail in the field of organic light-emitting diodes Δ = 1.05 eV

IP = 5.2 eV

Φ = 5.4 eV

0.85 eV Au

Evac

LUMO 2.2 eV EF HOMO

Pentacene

FIGURE 2.2.16 Actual energy level alignment in the Au-pentacene junction as determined by UV photoelectron spectroscopy. (Adapted from [45].)

8080_C002.fm Page 95 Friday, April 6, 2007 10:21 AM

Charge Transport in Oligomers

95

(OLEDs), where contact resistance is a crucial issue too. Ultraviolet photoelectron spectroscopy (UPS) and inverse UPS have been used to precisely determine the energy levels at both sides of the interface [45]. A typical result for the Au–pentacene interface is shown in Figure 2.2.16, which clearly shows that the actual interface strongly deviates from the MS model. Instead, the interface exhibits an additional “dipole” barrier Δ that shifts the vacuum level upward by more than 1 eV, hence increasing the barrier height by the same amount. The rather large interface dipole is explained by the fact that the electron density at a metal surface presents a tail that extends from the metal free surface into vacuum, thus forming a dipole pointing at the metal bulk. Molecules deposited on the metal tend to push back this tail, thus reducing the surface dipole and decreasing the work function of the metal. The previously described four-probe technique [43,44] allows a separate determination of the source and drain contact resistances. If contacts would behave as Schottky barriers, one would expect the voltage drop at source to be substantially higher than that at drain. This is what is indeed observed with “bad” contacts. However, “good” contacts show comparable drops at both electrodes. A possible origin of this behavior has been recently put forward [46]. The model assumes that the regions immediately adjacent to the electrodes are made of organic material of quality different from that of the rest of the conducting channel, with very low mobility. It is worth pointing out that top contacts usually offer lower contact resistance than bottom contacts. The asymmetry of the organic-metal contact, depending on whether the organic film is deposited on the metal or the metal on the organic layer, has been studied from both the theoretical [47] and experimental [48] points of view. For instance, combined UPS and XPS measurements have demonstrated that the deposition of gold atop a pentacene layer shows signs of metal penetration coupled with the formation of metal clusters, leading to a substantial reduction of the interface barrier from 1 to 0.3 eV.

2.2.5.3 MOBILITY DEGRADATION Gate-voltage dependent mobility has already been dealt with in the case of traplimited transport; in this circumstance, mobility increases with gate voltage. Another well-documented source of dependence in conventional MOSFETs is the so-called mobility degradation, which results in a reverse effect; that is, mobility decreases with gate voltage. Mobility degradation is attributed to various scattering agents in the vicinity of the insulator interface: namely, charged centers, surface phonons, and surface roughness [49]. A usual way to describe this behavior is to introduce the socalled mobility degradation factor θ , which in its simplest formulation leads to a mobility that looks like μ=

μ0 1 + θ(VG − VT )

(2.2.28)

8080_C002.fm Page 96 Friday, April 6, 2007 10:21 AM

96

Organic Field-Effect Transistors

A similar trend has been recently also observed in OTFT [43,50,51]. To account for this decrease, a model has been developed that takes advantage of the particular structure found in semiconductor films made of short molecules. Such molecules can be viewed as rigid rods, which in the solid state assemble parallel to each other, thus forming parallel layers, the thickness of which equals the length of the molecule (minus a small correction to account for a tilt angle). Taking into consideration that (1) each molecule can bear only one charge carrier and (2) the charge practically extends over the entire molecule [52], one can assimilate the film to a stack of n dielectric layers of thickness d , each layer bearing a uniform density of carriers ni . The calculation of the charge distribution has been performed by Horowitz and coworkers [53]. Layers are numbered starting from the insulator–semiconductor interface. To calculate the charge-carrier density ni (per unit area) in the ith layer, let us apply Gauss’s law to a cylinder of unit cross-section limited by the boundary between the (i – 1)th and the ith layers and that between the ith and the (i + 1)th layers. Within the gradual channel approximation, the electric field F is perpendicular to the layer so that Fi−1 − Fi = −

qni εs

(2.2.29)

where Fi is the module of the electric field at the boundary between the ith and the (i + 1)th layers. Because the charge density is constant in each layer, the electric field varies linearly between two boundaries, which in turn implies that the potential is quadratic within the same limits. Furthermore, assume that charge transfer between adjacent layers is sufficiently efficient so that the distribution of charge in the whole film is at thermal equilibrium. This yields ⎡ kT ⎤ ni+1 = exp ⎢− (Vi − Vi+1 )⎥ ni ⎣ q ⎦

(2.2.30)

After some manipulation, the following series of equations is found: ⎡ ⎢ ⎢ ⎢ ⎢ ⎢ ⎣

⎛

dq 2 ⎜⎜ ni+1 ni = ni+1 exp + ⎜ kT ε s ⎜⎜ 2 ⎝

n

∑n j =i+2

⎞⎤ ⎟⎥ ⎟⎥ j ⎟⎥ ⎟⎥ ⎟ ⎠⎥⎦

(2.2.31)

The gate-voltage dependent charge distribution across the conducting channel can be obtained by starting from an arbitrary value of the density of charge in the nth layer and cascading down to the first layer. The gate voltage is then calculated from the total density of charge carriers. n

∑ n =CV

qntot = q

i

1

i G

(2.2.32)

8080_C002.fm Page 97 Friday, April 6, 2007 10:21 AM

Charge Transport in Oligomers

97

Figure 2.2.17 shows the calculated relative distribution of charge in the case of a 30-nm thick pentacene film. Note that with a monolayer of around 1.5 nm, 30 nm of pentacene represents 20 layers. It is immediately seen that as the gate voltage increases, the charge gradually concentrates in the first layer. The decrease of the effective mobility is then attributed to the fact that the mobility in this first layer is likely to be degraded by the various effects mentioned earlier. The layers can now be regarded as parallel conductors, so the total conductance of the film equals the sum of the mobility times the density of charge in each layer. If the mobility is lower in the first layer, the effective mobility decreases with gate voltage because the proportion of charge carriers in the first layer increases. An alternative way to describe mobility degradation is to say that charge trapping is more pronounced at the insulator–semiconductor interface, thus reducing the density of free carriers.

2.2.6 CONCLUDING REMARKS Charge transport in organic semiconductors is the result of an interplay between intramolecular and intermolecular effects. The most remarkable consequence of such an interplay is that, unlike what occurs in their inorganic counterparts, charge carriers in organic semiconductors can no longer be viewed as bare particles; instead, they are dressed by a polarization cloud that follows their movement, thus leading to the formation of polarons. Polarons are the main reason why mobility is substantially lower in organic semiconductors than in their inorganic counterpart. Another important consequence of this interplay is that the electronic properties of a conjugated organic solid depend on the chemical structure of the individual molecules and are also strongly determined by the crystalline structure of the solid. Because of that, theoretical calculations on charge transport in conjugated organic materials are intricate, and a predictive theory is still lacking. Additional limitations to charge transport occur in thin-film conductors. These limitations concentrate at the interfaces of the device: namely, the metal–semiconductor and insulator–semiconductor interfaces. The former is responsible for limitations by contact resistance, while problems with the latter arise from the fact that defects tend to concentrate at this interface, thus leading to a degradation of charge transport. Under these circumstances, parameter extraction is not straightforward and often requires measurements on devices with various geometries. Recent results on organic transistors made of highly pure and carefully crystallized materials have shown that mobility can reach the 10 cm2/Vs range [54], thus giving hope for further improvements.

ACKNOWLEDGMENTS Most of the experimental work on which this chapter is based was performed while the author was at the Laboratoire des Matériaux Moléculaires, CNRS, directed by Francis Garnier. The author gratefully acknowledges his past and present coworkers Abderrahim Yassar, Zhi-Gang Xu, Xue-Zhou Peng, Denis Fichou, Bai Xu, Didier Delabouglise, Mohamed Hmyene, Françoise Deloffre, Fayçal Kouki, Philippe Lang,

8080_C002.fm Page 98 Friday, April 6, 2007 10:21 AM

98

Organic Field-Effect Transistors

1

VG = 1 V

ni/n1

0.8 0.6 0.4 0.2 0

1

1 2 3 4 5 6 7 8 9 1011121314151617181920 Layer number

VG = 5 V

ni/n1

0.8 0.6 0.4 0.2 0

1

1 2 3 4 5 6 7 8 9 1011121314151617181920 Layer number VG = 25 V

ni/n1

0.8 0.6 0.4 0.2 0

1 2 3 4 5 6 7 8 9 1011121314151617181920 Layer number

FIGURE 2.2.17 Relative distribution of charge in a 20-layer (30-nm) thick pentacene film for three different values of the gate voltage.

8080_C002.fm Page 99 Friday, April 6, 2007 10:21 AM

Charge Transport in Oligomers

99

Riadh Hajlaoui, Frédéric Demanze, Pratima Srivastava, Claude Noguès, Xavier Pan, Peter Spearman, Belkassem Nessakh, Ahmed El Kassmi, Mohammad Mottaghi, and Wolfgang Kalb. He is indebted to Philippe Delannoy, Jean-Luc Brédas, Jérôme Cornil, Fabio Biscarini, Antoine Kahn, Jean-Noël Chazalviel, and Libero Zuppiroli for stimulating discussions and fruitful collaborations.

REFERENCES 1. Ashcroft, N. W. and Mermin, N. D., Solid state physics, Holt, New York, 1976. 2. Wright, J. D., Molecular crystals, 2nd ed., Cambridge University Press, Cambridge, 1995. 3. Gutman, F. and Lyons, L. E., Organic semiconductors, John Wiley, New York, 1967. 4. Meier, H., Organic semiconductors: Dark- and photoconductivity of organic solids, Verlag Chemie, Weinheim, 1974. 5. Eley, D. D., Phthalocyanines as semiconductors, Nature, 162, 819, 1948 6. Vartanyan, A. T., Semiconductor properties of organic dyes. I. Phthalocyanines. Zh. Fiz. Khim., 22, 769, 1948. 7. Vartanyan, A. T., Semiconductivity of organic dyes. II. Trypaflavine, Zh. Fiz. Khim., 24, 1361, 1950. 8. Vartanyan, A. T., Semiconductive properties of organic dyes, Izv. Akad. Nauk SSSR, 16, 169, 1952. 9. Eley, D. D. et al., Semiconductivity of organic substances. I, Trans. Faraday Soc., 49, 79, 1953. 10. Eley, D. D. and Parfitt, G. D., Semiconductivity of organic substances. II, Trans. Faraday Soc., 51, 1529, 1955. 11. Pope, M. and Swenberg, C. E., Electronic processes in organic crystals and polymers, 2nd ed., Oxford University Press, New York, 1999. 12. Tang, C. W. and van Slyke, S. A., Organic electroluminescent diodes, Appl. Phys. Lett., 51, 913, 1987. 13. Horowitz, G. et al., A field-effect transistor based on conjugated alpha-sexithienyl, Solid State Commun., 72, 381, 1989. 14. Lilienfeld, J. E., U.S. Patent 1,745,175, 1930. 15. Kahng, D. and Atalla, M. M., Silicon–silicon dioxide field induced surface devices, in IRE Solid-State Devices Research Conference, Pittsburgh, PA, 1960. 16. Heilmeier, G. H. and Zanoni, L. A., Surface studies of alpha-copper phthalocyanine films, J. Phys. Chem. Solids, 25, 603, 1964. 17. Koezuka, H., Tsumura, A., and Ando, T., Field-effect transistor with polythiophene thin-film, Synth. Metal., 18, 699, 1987. 18. Sze, S. M., Physics of semiconductor devices, 2nd ed., John Wiley, New York, 1981. 19. Karl, N. et al., High-field saturation of charge carrier drift velocities in ultrapurified organic photoconductors, Synth. Metal., 42, 2473, 1991. 20. Silinsh, E. A. and Cápek, V., Organic molecular crystals: Interaction, localization, and transport phenomena, AIP Press, New York, 1994. 21. Sato, N., Inokuchi, H., and Silinsh, E. A., Reevaluation of electronic polarization energies in organic molecular crystals, Chem. Phys., 115, 269, 1987. 22. Kenkre, V. M. et al., Unified theory of the mobilities of photoinjected electrons in naphthalene, Phys. Rev. Lett., 62, 1165, 1989.

8080_C002.fm Page 100 Friday, April 6, 2007 10:21 AM

100

Organic Field-Effect Transistors

23. Marcus, R.A., Chemical and electrochemical electron-transfer theory, Ann. Rev. Phys. Chem., 15, 155, 1964. 24. Newton, M.D., Quantum chemical probes of electron-transfer kinetics: The nature of donor–acceptor interactions, Chem. Rev., 91, 767, 1991. 25. Marcus, R.A., Electron transfer in chemistry. Theory and experiment, Rev. Mod. Phys., 65, 599, 1993. 26. Pope, M. and Swenberg, C. E., Electronic processes in organic crystals and polymers, 2nd ed., Oxford University Press, New York, 1999. 27. Brédas, J.-L. et al., Charge-transfer and energy-transfer processes in p-conjugated oligomers and polymers: A molecular picture, Chem. Rev., 104, 4971, 2004. 28. Gill, W. D., Drift mobilities in amorphous charge-transfer complexes of trinitrofluorenone and poly-n-vinylcarbazole, J. Appl. Phys., 43, 5033, 1972. 29. Bässler, H., Charge transport in disordered organic photoconductors, Phys Stat. Sol. B, 175, 15, 1993. 30. Miller, A. and Abrahams, E., Impurity conduction at low concentrations, Phys. Rev., 120, 745, 1960. 31. Le Comber, P. G. and Spear, W. E., Electronic transport in amorphous silicon films, Phys. Rev. Lett., 25, 509, 1970. 32. Horowitz, G., Hajlaoui, R., and Delannoy, P., Temperature dependence of the fieldeffect mobility of sexithiophene. Determination of the density of traps, J. Phys. III France, 5, 355, 1995. 33. Völkel, A. R., Street, R. A., and Knipp, D., Carrier transport and density of state distributions in pentacene transistors, Phys. Rev. B, 66, 195336, 2002. 34. Horowitz, G., Hajlaoui, M. E., and Hajlaoui, R., Temperature and gate voltage dependence of hole mobility in polycrystalline oligothiophene thin film transistors, J. Appl. Phys., 87, 4456, 2000. 35. Stassen, A. F. et al., Influence of the gate dielectric on the mobility of rubrene singlecrystal field-effect transistors, Appl. Phys. Lett., 85, 3899, 2004. 36. Ortiz-Conde, A. et al., A review of recent MOSFET threshold voltage extraction methods, Microelectronics Reliability, 42, 583, 2002. 37. Wong, H.-S. et al., Modeling of transconductance degradation and extraction of threshold voltage in thin oxide MOSFETs, Solid-State Electron., 30, 953, 1987. 38. Necliudov, P. V. et al., Contact resistance extraction in pentacene thin film transistors, Solid-State Electron., 47, 259, 2003. 39. Zaumseil, J., Baldwin, K. W., and Rogers, J. A., Contact resistance in organic transistors that use source and drain electrodes formed by soft contact lamination, J. Appl. Phys., 93, 6117, 2003. 40. Klauk, H. et al., Contact resistance in organic thin film transistors, Solid-State Electron., 47, 297, 2003. 41. Meijer, E. J. et al., Scaling behavior and parasitic series resistance in disordered organic field-effect transistors, Appl. Phys. Lett., 82, 4576, 2003. 42. Luan, S. and Neudeck, G. W., An experimental study of the source/drain parasitic resistance effects in amorphous silicon thin film transistors, J. Appl. Phys., 72, 766, 1992. 43. Chesterfield, R. J. et al., Variable temperature film and contact resistance measurements on operating n-channel organic thin film transistors, J. Appl. Phys., 95, 6396, 2004. 44. Yagi, I., Tsukagoshi, K., and Aoyagi, Y., Direct observation of contact and channel resistance in pentacene four-terminal thin-film transistor patterned by laser ablation method, Appl. Phys. Lett., 84, 813, 2004.

8080_C002.fm Page 101 Friday, April 6, 2007 10:21 AM

Charge Transport in Oligomers

101

45. Koch, N. et al., Conjugated organic molecules on metal versus polymer electrodes: Demonstration of a key energy level alignment mechanism, Appl. Phys. Lett., 82, 70, 2003. 46. Li, T. et al., Investigation of bottom-contact organic field effect transistors by twodimensional device modeling, J. Appl. Phys., 93, 4017, 2003. 47. Tessler, N. and Roichman, Y., Two-dimensional simulation of polymer field-effect transistor, Appl. Phys. Lett., 79, 2987, 2001. 48. Watkins, N. J., Yan, L., and Gao, Y., Electronic structure symmetry of interfaces between pentacene and metals, Appl. Phys. Lett., 80, 4384, 2002. 49. van Langevelde, R. and Klaassen, F. M., Effect of gate-field dependent mobility degradation on distortion analysis in MOSFETs, IEEE Trans. Electron Devices, 44, 2044, 1997. 50. Kalb, W. et al., Structure–performance relationship in pentacene/Al2O3 thin-film transistors, Synth. Metal., 146, 279, 2004. 51. Merlo, J. A. et al., p-Channel organic semiconductors based on hybrid acenethiophene molecules for thin-film transistor applications, J. Am. Chem. Soc., 127, 3997, 2005. 52. Sancho-Garcia, J. C. et al., Effect of an external electric field on the charge transport parameters in organic molecular semiconductors, J. Chem. Phys., 119, 12563, 2003. 53. Horowitz, G. and Mottaghi, M., Field-induced mobility degradation in pentacene thin-film transistors, Org. Electron., 7, 528, 2006. 54. de Boer, R. W. I. et al., Organic single-crystal field-effect transistors, Phys. Stat. Sol. A, 201, 1302, 2004.

8080_C002.fm Page 102 Friday, April 6, 2007 10:21 AM

8080_C002.fm Page 103 Friday, April 6, 2007 10:21 AM

2.3

Charge Transport Physics of SolutionProcessed Organic Field-Effect Transistors

Henning Sirringhaus CONTENTS 2.3.1 Introduction................................................................................................103 2.3.2 Solution-Processable p-Type Organic Semiconductors ............................106 2.3.2.1 Conjugated Polymers ..................................................................106 2.3.2.2 Solution-Processable Small Molecules.......................................110 2.3.3 Solution-Processable n-Type Organic Semiconductors ............................111 2.3.4 Gate Dielectrics..........................................................................................112 2.3.5 Charge Transport Physics ..........................................................................115 2.3.6 Charge Injection Physics ...........................................................................124 2.3.7 Defect States and Device Degradation Mechanisms.................................127 2.3.8 Outlook.......................................................................................................130 References..............................................................................................................130

2.3.1 INTRODUCTION The semiconducting properties of conjugated polymers allow realization of a range of electronic and optoelectronic devices, such as light-emitting diodes (LEDs) [1], photovoltaic diodes [2], and field-effect transistors (FETs) [3]. Following the initial demonstration of field-effect conduction in small organic molecules [4,5] and conjugated polymers [3,6,7], for several years the field-effect transistor was merely a convenient experimental tool to probe the charge transport properties of undoped polymer semiconductors, circumventing the need for chemical doping to be able to measure electrical transport. Its potential for technological applications was limited by the then insufficient field-effect mobility of polymer semiconductors of typically less than 10–4 cm2/Vs and the relatively poor environmental stability of materials such as polyacetylene and polythiophene. However, since then field-effect transistors based on solution-processable organic semiconductors have experienced impressive

103

8080_C002.fm Page 104 Friday, April 6, 2007 10:21 AM

104

Organic Field-Effect Transistors

improvements in both performance and reliability as a result of a variety of different factors, including: • • •

•

availability of polymer materials with lower density of chain defects, better chemical purity, and higher molecular weights more controlled polymer processing guided by better understanding of the required thin-film microstructure better understanding of the requirements for the gate dielectric forming the active interface with the semiconducting layer and of the charge injecting source-drain contacts development of more controlled solution and printing-based device manufacturing processes

As a result, the performance of OFETs, which is generally benchmarked against that of amorphous silicon (a-Si) thin-film transistors (TFTs) with field-effect mobilities of 0.5–1 cm2/Vs and ON–OFF current ratio of 106–108, has improved significantly. Currently, the record mobility values for OFETs are 5 cm2/Vs in the case of vacuum-deposited small molecules [8] and 0.6 cm2/Vs for solution-processed polymers [9]. Even more importantly, very significant progress has been made on developing materials that allow combining high mobilities with good materials stability under air, moisture, and light exposure. OFETs are most commonly manufactured using standard top-gate (Figure 2.3.1a) and bottom-gate TFT architectures. Figure 2.3.1(b) shows the output characteristics of a state-of-the-art, unencapsulated polymer FET. The figure compares measurements performed in ambient air and light directly after device manufacture and several weeks later, after the device had participated in a customer trial and had crossed the Atlantic twice [10]. No evidence for device degradation is observed. As a result, there is now a serious level of industrial interest in using OFETs for applications currently incompatible with the use of a-Si or other inorganic transistor technologies. One of their main technological attractions is that all the layers of an OFET can be deposited and patterned at low room temperature by a combination of low-cost, solution-processing, and direct-write printing, which makes them ideally suited for realization of low-cost, large-area electronic functions on flexible substrates (for a review, see Sirringhaus et al. [11] and Forrest [12]). The first applications in which we can realistically expect OFETs to be used within the next three to five years are flexible, active matrix electronic paper displays [10,13] and simple, lowcost radio frequency identification (RFID) tags [14] and sensing devices. In particular, in the case of electronic paper displays there seems to be an excellent match between present technological capabilities and application requirements. To illustrate the current capabilities of OFET technology in this application area, Figure 2.3.2 shows an optical micrograph of an A5 active matrix display demonstrator on a flexible polyethyleneterephtalate (PET) substrate made by Plastic Logic. The display was fabricated by laminating the OFET backplane with an E Ink® Imaging Film. The display has a resolution of 100 pixels per inch (ppi) and displays four levels of gray. It contains 480,000 solution-processed OFETs (600 × 800 rows and columns). No substrate encapsulation is needed. The display exhibits excellent

8080_C002.fm Page 105 Friday, April 6, 2007 10:21 AM

Charge Transport Physics

105

W G Dielectric

d ++++++ S D L

SC

(a) 30 Vg = −40V

25

Is (μA)

20 −30V

15 10

−20V 5 −10.0V 0

0

−5

−10 −15 −20 −25 −30 −35 −40 Vd (V) (b)

FIGURE 2.3.1 (a) Schematic diagram of top-gate OFET using a standard TFT device architecture. (b) Output characteristics of state-of-the-art, unencapsulated OFET measured in air and light (closed symbols = device measured after manufacture; open symbols = device measured two weeks later).

contrast and can be bent to a radius of curvature of 5 mm. Other, more demanding display applications such as active matrix liquid crystal or organic light-emitting diode (OLED) displays or high-performance RFID tags compatible with existing communication standards are also envisioned, but require a transistor performance with mobilities exceeding 1 cm2/Vs, which is still difficult to achieve with solutionprocessed OFETs. Through their ability to control the charge carrier concentration electrostatically, rather than chemically, FET devices provide a useful scientific tool for the wealth of fundamental scientific questions regarding the charge transport and charge injection physics of organic semiconductors and their structure–property relationships. Significant effort has been devoted to understanding the fundamental electronic structure of the organic semiconductor — in particular, at the interface with the dielectric — and how microscopic molecular-scale transport processes determine the electrical characteristics of macroscopic devices. This is a challenging task because of the complex microstructure of solution-processed organic semiconductors, which in many cases cannot be fully characterized by conventional diffraction and microscopy techniques. An important related topic is the understanding of

8080_C002.fm Page 106 Friday, April 6, 2007 10:21 AM

106

Organic Field-Effect Transistors

FIGURE 2.3.2 Photograph of A5, 600 × 800 electronic paper display on a flexible PET substrate display comprising an active matrix of OFETs on a PET substrate laminated with an E Ink® Imaging Film. (Courtesy of Plastic Logic Ltd.)

electronic defects states and associated device degradation mechanisms, which are becoming an increasingly important topic as OFETs are nearing their introduction into first products with strict reliability and lifetime requirements. This section is focused on providing an up-to-date review of the state of knowledge of the materials, device, and charge transport physics of solution-processed OFETs. Sections 2.3.2 and 2.3.3 discuss the materials physics of solution-processable p- and n-type organic semiconductors, respectively. Section 2.3.4 is devoted to the important topic of solution-processable gate dielectrics, which determine critically the performance as well as the reliability of OFETs. Section 2.3.5 focuses at a more fundamental level on the electronic structure of solution-processed organic semiconductors, the charge transport processes at the active interface, and how these are affected by disorder and molecular relaxation effects. Section 2.3.6 discusses the charge injection physics at the source-drain contacts. Finally, in Section 2.3.7, we review the current understanding of electronic defect states and degradation mechanisms in OFETs, which lead to device instabilities and threshold voltage shifts upon bias stressing and/or environmental exposure.

2.3.2 SOLUTION-PROCESSABLE p-TYPE ORGANIC SEMICONDUCTORS 2.3.2.1 CONJUGATED POLYMERS Two different approaches to high-performance, solution-processable polymer semiconductors have emerged. The first approach is based on achieving high charge carrier mobilities by designing the material to exhibit microcrystalline [15] or liquidcrystalline [16] order through self-organization or by making use of specific interactions with a templating substrate. The second approach aims to produce a

8080_C002.fm Page 107 Friday, April 6, 2007 10:21 AM

Charge Transport Physics

107

completely amorphous microstructure to provide a uniform path for charge transport along which carriers experience a minimum degree of site energy fluctuations. Although the first approach is likely to lead to higher mobilities eventually, impressive device performance and stability have been demonstrated with the second approach recently. Amorphous polymers: Early FET studies on amorphous, disordered conjugated polymers, such as regioirregular polythiophene [7] or polyacetylene [17] suggested that field-effect mobilities in amorphous microstructures might be limited to low values < 10–3–10–4 cm2/Vs. However, several groups have recently reported that amorphous polymers based on triarylamine similar to those used in xerographic applications allow achieving high field-effect mobilities of 10–3–10–2 cm2/Vs, combined with good operating, environmental, and photo stability. Veres et al. have reported high-performance FETs with field-effect mobilities of up to 1 ⋅ 10–2 cm2/Vs, low threshold voltage, and good device stability based on a range of polytriarylamine (PTAA) derivatives [18,19]. These are used in combination with apolar, low-k polymer dielectrics (Figure 2.3.3). With structurally related (9,9-dialkylfluorene-alttriarylamine) (TFB) in contact with benzocyclobutene dielectric, very stable device operation during continuous switching at 120°C without device degradation was demonstrated [20]. ∗

E

N

10−1

n∗

Dipolar disorder at interface

Bulk

10−2

Drain

Source

Organic semiconductor

Dielectric

μ (cm2V−1s−1)

N (E) 10−3

10−4

10−5

TOF FET (low k insulator 1) FET (low k insulator 2) FET PMMA

Gate 10−6

5

10

15

20

25

(1000/T)2 (1000/K)2 (a)

(b)

FIGURE 2.3.3 (a) Schematic diagram of the effect of disordered polar groups on the energetic disorder at the active interface (b) Temperature dependence of the time-of-flight and field-effect mobility of PTAA. For the field-effect mobility, data for top-gate FETs with PMMA gate dielectric and two different lower k dielectrics are shown. (From Veres, J. et al., Adv. Func. Mater., 13, 199–204, 2003. Reprinted with permission. Copyright 2003, Wiley.)

8080_C002.fm Page 108 Friday, April 6, 2007 10:21 AM

108

Organic Field-Effect Transistors

Microcrystalline polymers: A prototype microcrystalline polymer is regioregular poly(3-hexylthiophene) (P3HT) [21, 22] with which high field-effect mobilities of 0.1–0.3 cm2/Vs have been achieved. Thin films of P3HT adopt a highly microcrystalline and anisotropic lamellar microstructure comprising two-dimensional conjugated layers with strong π−π interchain interactions separated by layers of solubilizing, insulating side chains (Figure 2.3.4a) leading to fast in-plane charge transport [15]. The microcrystals have been found to have a nanoribbon shape [23–27]. The mobility of P3HT depends very sensitively on the degree of head-to-tail regioregularity [21,22] and deposition conditions [15,21,28]. The mobility of P3HT FETs has also been reported to improve by orders of magnitude upon modification of the SiO2 gate dielectric substrate by hydrophobic self-assembled monolayers (SAMs), possibly through lowering of the surface energy of the gate dielectric and removal of residual surface water and other polar groups prior to deposition of the polymer [22] or by inducing microstructural changes through specific interactions with functional groups of the polymer [29]. From studies of high molecular weight P3HT films with varying degrees of crystallinity as induced by varying the boiling point of the solvent [26], there is clear evidence that a higher degree of crystallinity generally results in higher mobility. Recently, much effort has been devoted to understanding the dependence of mobility on molecular weight [24,30–32]. Figure 2.3.4(b) illustrates the dependence of the field-effect mobility on the combined effects of film formation kinetics and molecular weight (MW) [32]. Data for three kinds of deposition conditions are shown: rapid film formation by spin-casting from low-boiling point solvent (chloroform) as well as slow film formation by drop-casting from chloroform and spincasting from a high boiling point solvent (trichlorobenzene). In this way the microstructure, polymer chain folding, and degree of crystallinity of the polymer film can be controlled. In trichlorobenzene spin-cast devices, the mobility increases sharply with MW from 10–3 (MW = 15.4 and 20 kD) up to 10–1 cm2/Vs (52 kD) and then starts to level off for even higher molecular weight. The MW dependence of mobilities in chloroform drop-cast devices behaves in a similar manner but saturates at a value slightly lower than 10–1 cm2/Vs. In contrast, in films spin-coated from chloroform, the mobility varies only little between 3 × 10–3 to 10–2 cm2/Vs in an MW range of 15 to 270 kD, but no systematic dependence on MW is found. The behavior in the low mobility regime (μ < 10–2 cm2/Vs) has been attributed to grain boundaries limiting the transport in low molecular weight samples by Kline et al. [24] as well as a more isotropic orientation of polymer crystallites [33]. Zen et al. [30] have explained a similar observation in terms of a less planar polymer backbone in the amorphous regions of the film in the case of low molecular weight fractions. Chang et al. have established a correlation between the mobility and interchain/intrachain disorder as accessed by optical and electro-optical spectroscopy, which suggests that, in the high-mobility regime (μ ≈ 10–1 cm2/Vs), the mobility is limited by the crystalline quality of the P3HT nanocrystals as opposed to grain boundaries between the crystals [32]. Since transport in an FET reflects primarily the structure of the semiconducting polymer at the interface as opposed to the bulk, it is important to have experimental probes available that are able to distinguish between interface/surface and bulk

8080_C002.fm Page 109 Friday, April 6, 2007 10:21 AM

Charge Transport Physics

109

b s s

s a

s

(300) (200) (100) (010) (a) TCB spin-cast

Mobility (cm2/Vs)

10−1 Chloroform drop-cast 10−2

Chloroform spin-cast 10−3 10

100 Mw (kD) (b)

FIGURE 2.3.4 (a) Wide-angle x-ray scattering image of high-mobility P3HT on SiO2. The inset shows the in-plane, lamellar self-organization of P3HT. (From Sirringhaus, H. et al., Nature, 401, 685–688, 1999. Reprinted with permission. Copyright 1999, Nature.) (b) Dependence of mobility of P3HT FETs with SiO2 gate dielectric as a function of molecular weight for different film formation conditions (spin-casting from trichlorobenzene, as well as spincasting and drop-casting from chloroform).

8080_C002.fm Page 110 Friday, April 6, 2007 10:21 AM

110

Organic Field-Effect Transistors

microstructure. Kline et al. have recently used x-ray rocking curves to probe selectively the orientation of crystallites at the substrate interface. An exceptional, instrument-limited degree of orientation and a clear correlation between the mobility and the degree of orientation of the P3HT crystals at the interface was found [33]. p-Type semiconducting materials with low ionization potential (typically less than 4.9–5.0 eV), such as regioregular P3HT, tend to exhibit large positive VT shifts of threshold voltage upon exposure to air, presumably due to doping of the polymer [34]. P3HT is known to form a reversible charge transfer complex with oxygen [35]. Nevertheless, encouraging shelf life stability, albeit with low ON–OFF current ratio of <103, has been reported for P3HT FETs in a top-gate configuration, which may provide some encapsulation [36]. P3HT has poor photostability when exposed to ultraviolet sunlight in the presence of oxygen, causing formation of carbonyl defects in the polymer with associated loss of conjugation and mobility degradation [37]. The oxidative stability of P3HT can be improved by increasing the ionization potential of the polythiophene backbone by disrupting its ability to adopt a fully planar conformation through the side-chain substitution pattern [38] or by incorporating partially conjugated co-monomers into the main chain [9,39]. These materials maintain the beneficial microcrystalline, lamellar self-organization motive of the parent P3HT polymer. As a result, they exhibit similar or even somewhat higher field-effect mobilities, but have significantly improved environmental and operating stability.

2.3.2.2 SOLUTION-PROCESSABLE SMALL MOLECULES An alternative route to solution-processable organic semiconductors is to use small molecule semiconductors that have been designed to be compatible with solution deposition. Precursor routes: Polycrystalline thin films of a conjugated molecule can be obtained by forming a thin film of a soluble precursor on the substrate with subsequent thermal [40] or irradiative [41] conversion into the fully conjugated form. Pentacene precursors have been shown to yield field-effect mobilities of 0.01–0.1 cm2/Vs [42] and 0.1–0.8 cm2/Vs [43] after thermal conversion at 150–200°C. Recently, it was shown that even unsubstituted pentacene can be solution deposited from high-boiling point solvents at elevated temperatures [44]. A precursor route approach to tetrabenzoporphyrin has also been developed [45] that yields field-effect mobilities on SiO2 of 0.017 cm2/Vs when converted at a temperatures of 150–200°C. Sexithiophene substituted with ester groups that can be removed by thermolysis at 150–260°C exhibits field-effect mobilities on SiO2 up to 0.07 cm2/Vs [46]. Side-chain substitution: Small molecule organic semiconductors can also be rendered solution processable by attachment of flexible side chains. The substitution pattern needs to be designed carefully so that the side chains, which are needed to impart adequate solubility and film forming properties, do not interfere with the ability of the molecule to π-stack. Katz reported side-chain-substituted small molecule semiconductors such as dihexylanthradithiophene [47,48] that can be solutiondeposited with mobilities of 0.01–0.02 cm2/Vs. In bis(hexyl-bithiophene)benzene solution-cast onto a heated SiO2 substrate, mobilities of up to 0.03 cm2/Vs were reported [49]. Due to the relatively low solubility of these molecules, growth

8080_C002.fm Page 111 Friday, April 6, 2007 10:21 AM

Charge Transport Physics

111

conditions need to be optimized carefully to prevent aggregation and crystallization of the molecules in solution. This can lead to three-dimensional film morphology with poor connectivity and orientation of the grains in the films. An interesting new strategy to solution-processable small conjugated molecules such as rubrene has recently been reported [50]. The approach is based on incorporation of a glass-inducing diluent that enables controlled crystallization from an initial amorphous, vitreous state of the organic semiconductor. This leads to high crystalline quality, high mobility of up to 0.7 cm2/Vs, and good device stability. Liquid-crystalline molecules: Small side-chain-substituted molecules that exhibit liquid-crystalline phases at elevated temperatures provide alternative routes to forming highly crystalline thin films from solution. Discotic liquid-crystalline molecules such as hexabenzocoronenes have been uniaxially aligned in thin film form with the columnar axis oriented along the transport direction in the FET by using graphoepitaxy on highly crystalline teflon alignment layers [51] or deposition by zone crystallization [52] with field-effect mobilities up to 0.01 cm2/Vs along the discotic columns. Encouraging device performance has also been obtained for contorted discotic derivatives of hexabenzocoronene [53]. Reactive mesogens based on oligothiophenes with photopolymerisable end groups have been homeotropically aligned on a substrate prior to crosslinking to fix the orientation of the molecules and used as the active FET layer [54].

2.3.3 SOLUTION-PROCESSABLE n-TYPE ORGANIC SEMICONDUCTORS Many organic semiconductors show p-type conduction only (i.e., in contact with a SiO2 gate dielectric); for example, hole accumulation layers can be readily formed for negative gate bias, provided that a source-drain metal with a work function matching the ionization potential of the organic semiconductor is used. However, ntype organic semiconductors that exhibit electron transport in contact with a suitably low work function source-drain metal upon application of positive gate bias are comparatively rare, but are needed for realization of complementary logic circuits. Electron field-effect conduction has been reported in several, relatively high electron affinity (EA > 3.5 eV) small molecule organic semiconductors deposited from vacuum phase (for a review, see Dimitrakopoulos and Malenfant [55]) and solution-processed organic semiconductors. High electron-affinity materials are less susceptible to the presence of electron-trapping impurities, since such trapping groups are more likely to be positioned in energy above the LUMO states of the organic semiconductor. It has been shown recently [56] (see Section 2.3.4) that electron conduction is in fact a generic feature of most organic semiconductors, including those with normal electron affinities of 2.5–3.5 eV, provided that the right dielectric, which avoids trapping of electrons at the interface, is used. Fluoroalkyl-substituted naphtalenetetracarboxylic diimide can be processed into thin films from fluorinated solvents to yield mobilities of 0.01 cm2/Vs (bottom gate FET with SiO2 dielectric, and gold contacts) [57]. The ladder polymer poly(benzobisimidazobenzophenanthroline) (BBL) has an electron affinity of 4.0–4.4 eV and can be solution processed into microcrystalline thin films from Lewis and methanesulfonic

8080_C002.fm Page 112 Friday, April 6, 2007 10:21 AM

112

Organic Field-Effect Transistors

acids [58]. High electron mobilities of 0.03–0.1 cm2/Vs were achieved in a bottom gate FET configuration with SiO2 dielectric measured unencapsulated in air. Solutionprocessed diperfluorohexyl substituted quinque- and quaterthiophene with electron affinities of 2.8–2.9 eV have been reported to exhibit field-effect mobilities of 4–8 · 10–4 cm2/Vs (on HMDS treated SiO2 dielectric with gold contacts). The devices suffer from a relatively high threshold voltage > 25 V due to electron trapping, which might be related to the relatively low electron affinity of fluoroalkyl substituted thiophene molecules [59]. n-Type field-effect conduction has also been reported in methanofullerene phenyl C61-butyric acid methyl ester (PCBM) [60] and 6,6-phenyl-C71-butyric acid methyl ester [61]. In the case of the C60 derivative, fieldeffect mobilities of 3–4 · 10–3 cm2/Vs were achieved in an encapsulated, bottom gate device with an organic dielectric and calcium source-drain contacts. Much lower apparent mobilities were observed with gold or aluminum contacts. For the C70 derivative, the electron mobility is somewhat lower, but the material exhibits better environmental stability than the C60 derivative. Recently, there has been growing interest in ambipolar organic semiconductors, which, in a device with suitable choice of source-drain contacts, exhibit hole accumulation for negative gate bias and electron accumulation when the gate bias is reversed. One application of ambipolar semiconductors is in light-emitting FETs, which are operated by biasing the gate voltage in between the values of the source and the drain voltage to form a hole accumulation layer near the source contact and an electron accumulation layer near the drain contact. Ambipolar conduction was established in blends of solution-processed hole (poly(methoxy-dimethyloctyloxy)phenylene vinylene OC1C10–PPV or P3HT) and electron (PCBM) transporting organic semiconductors by Meijer et al. The electron mobility in such blends (7 · 10–4 cm2/Vs) was two orders of magnitude lower than the electron mobility of a pure film of PCBM; the hole mobility was similar to that of single-component OC1C10–PPV (3 · 10–5 cm2/Vs) [62]. Similarly, Babel reported ambipolar conduction measured in air in blends of BBL and CuPc [63]. Also in this case the electron (1.7 · 10–4 cm2/Vs) and hole (3 · 10–5 cm2/Vs) mobilities were several orders of magnitude lower than that of films of the single components. Interestingly, it was possible to improve either the electron or the hole mobility by postdeposition annealing under solvent atmosphere; however, this was associated with the loss of the ambipolar conduction. Shkunov et al. investigated ambipolar transport in blends of polythiophene and PCBM [64]. Ambipolar conduction has also been reported in single-component systems such as low band gap poly(3,9-di-tert-butylindeno[1,2-b]fluorene) (PIF) [62], PCBM [65], and soluble oligothiophene/fullerene donor–acceptor triads [66]. It has been shown recently [56] (see Section 2.3.4) that ambipolar conduction is in fact a generic feature of a number of organic semiconductors, provided that the right dielectric, which avoids trapping of electrons at the interface, is used.

2.3.4 GATE DIELECTRICS The performance of organic field-effect devices depends critically on the use of high-performance dielectrics that form active interfaces with low defect densities.

8080_C002.fm Page 113 Friday, April 6, 2007 10:21 AM

Charge Transport Physics

113

In the same way as silicon MOS technology owes much to the quality of the Si–SiO2 interface, dielectrics for organic FETs have recently received significant attention (for a comprehensive review, see Facchetti et al. [67]). In comparison to inorganic heterointerfaces, many aspects of the physics of charge transport along solutionprocessed heterointerfaces are still poorly understood. In the present section we will review recent progress in the understanding of these issues and some of the general selection criteria for gate dielectric materials. For a solution-processed active interface in which the gate dielectric material is deposited from solution onto a solution-processable semiconducting material or vice versa, it is critical to avoid dissolution or swelling effects during deposition of the upper layer, which can lead to interfacial mixing and increased interface roughness. This can be avoided by cross-linking the lower layer; however, this restricts the choice of materials and requires special care to avoid introducing unwanted impurities and trapping groups [68]. The preferred approach is to choose orthogonal solvents for the deposition of the multilayer structure [69]. It has been demonstrated that in this way solution-processed interfaces can be achieved at which the fieldeffect mobility is as high as that of the corresponding organic semiconductor–SiO2 interface, for which interfacial mixing is not an issue. This is somewhat surprising since a solution-processed polymer heterointerface is never atomically abrupt; its width is determined by a balance between entropy favoring a wider interface and the unfavorable energy of interaction between the two polymers [70]. The correlation between interface roughness and mobility in solution-processed OFETs has recently been investigated by Chua et al. [71], who developed an approach for fabricating self-assembled polymer semiconductor–olymer dielectric bilayers making use of vertical phase separation in ternary solutions of semiconducting polymer–gate dielectric–solvent. By varying the speed of solvent removal, the roughness of the phase-separated interface could be varied in a controlled way. The mobility was found to be constant for low values of the interface roughness less than a critical roughness threshold. For roughness exceeding this threshold, a very rapid drop of the mobility by orders of magnitude was observed, even for roughness features of surprisingly long wavelength > 100 nm. In principle, for a given thickness of dielectric, a high-k dielectric is preferable to a low-k dielectric for an FET application, which requires the FET to exhibit a high drive current at low drive voltage. Various solution-processable high-k dielectrics for low-voltage OFETs have been used in the literature, such as anodized Al2O3 [72] (ε = 8–10), TiO2 [73] (ε = 20–41), or polyvinylphenol loaded with TiO2 nanoparticles [74] (for a review see Veres et al. [19]). Many polar, high-k polymer dielectrics, such as polyvinylphenol (ε = 4.5) or cyanoethylpullulan (ε = 12), are hygroscopic and susceptible to drift of ionic impurities during device operation and thus cannot be used for ordinary TFT applications [75]. Veres et al. have shown that the field-effect mobilities of amorphous PTAA [18] and other polymers are higher in contact with low-k dielectrics with ε < 3 than dielectrics with higher k [19]. The latter usually contain polar functional groups randomly oriented near the active interface, which is believed to increase the energetic disorder at the interface beyond what naturally occurs due to the structural disorder in the organic semiconductor film resulting in a lowering of the field-effect

8080_C002.fm Page 114 Friday, April 6, 2007 10:21 AM

114

Organic Field-Effect Transistors

mobility (Figure 2.3.3a). Low-k dielectrics also have the advantage of being less susceptible to ionic impurities, which can drift under the influence of the gate field, causing device instabilities (see Section 2.3.4). A range of techniques have been developed that allow fabrication of OFET sourcedrain electrodes with submicrometer channel length (see, for example, Hamadani and Natelson [76] and Sele et al. [77]). To ensure correct scaling of the device characteristics in such short channel devices, the dielectric thickness needs to be significantly thinner than the channel length. Ideally, the gate dielectric thickness should be one order of magnitude smaller than the channel length. In this way, saturation of the FET current when the gate voltage (corrected for the threshold voltage) exceeds the sourcedrain voltage can be achieved even for submicrometer channel lengths. Very thin, sub-20 nm organic dielectrics have been demonstrated using several approaches, such as self-assembled monolayer dielectrics [78], self-assembled molecular multilayers [79], or ultrathin polymer dielectrics [20]. Cross-linking of polyvinylphenol and polystyrene using bis(trichlorosilyl)alkyl reagents has been shown to result in improved dielectric properties and enable very thin spin-coatable polymer dielectrics [79]. Cho et al. have used self-assembled monolayers of docosyltrichlorosilane as the gate dielectric of a bottom-gate, top-contact P3HT FET with inkjet-printed conducting polymer source-drain electrodes [80]. The chemical purity and composition of the gate dielectric can have dramatic effects on interfacial charge transport. The reason for the absence of n-type fieldeffect conduction in “normal” polymers such as PPVs or P3HT with electron affinities around 2.5–3.5 eV has puzzled the community for some time because, in LED devices, many of these polymers support electron conduction. Chua et al. [56] have demonstrated that by using appropriate gate dielectrics free of electron-trapping groups, such as hydroxyl, silanol, or carbonyl groups, n-channel FET conduction is in fact a generic property of most conjugated polymers. In contact with trappingfree dielectrics such as BCB or polyethylene, electron and hole mobilities were found to be of comparable magnitude in a broad range of polymers. Some polymers, such as P3HT and OC1C10–PPV, even exhibit ambipolar charge transport in suitable device configurations (Figure 2.3.5). This demonstrates clean inversion behavior in organic semiconductors with band gaps > 2 eV. n-Type behavior has previously been so elusive because most studies were performed on SiO2 gate dielectrics for which electrochemical trapping of electrons by silanol groups at the interface occurs [56]. Light-emitting organic field-effect transistors (LEOFETs) have recently attracted much attention because they combine the switching characteristics of transistors with the light emission of diodes. Although several groups had reported lightemission from an OFET [81–85], no report of spatially resolved light emission from within the channel of an organic light-emitting FET had been made until recently. As a corollary to the realization of clean ambipolar transport in organic semiconductors at trap-free gate dielectric interfaces, light-emitting polymer field effect transistors with a well-defined recombination zone within the channel have recently been demonstrated [86,87]. Figure 2.3.6(a) shows a schematic diagram of an ambipolar OFET with a semiconducting layer of OC1C10–PPV in contact with BCB gate dielectric and two dissimilar source and drain contacts (Au and Ca) formed by an angled evaporation technique.

8080_C002.fm Page 115 Friday, April 6, 2007 10:21 AM

Charge Transport Physics

115 Si H3C CH CH3 CH3 3 Si Si O

10−6

10−6

10−7

10−7 Current (A)

Current (A)

Si

10−8 10−9 O

10−10 10−11 −10

0

10−8 10−9 10−10

n

MeO 10

20

30

40

50

60

Gate Voltage (V) (a)

n

10−11 −10

C6H13 ∗

S

0

10

n∗ 20

30

40

50

60

Gate Voltage (V) (b)

FIGURE 2.3.5 Transfer characteristics of bottom-gate OC1C10 PPV and P3HT FET with trap-free BCB gate dielectric exhibiting clean ambipolar transport (Vsd = 60V). (Courtesy of Jana Zaumseil, University of Cambridge.)

When such an ambipolar FET is biased with the gate voltage in between the source and the drain voltage, an electron accumulation layer is formed near one electrode coexisting with a hole accumulation layer near the other electrode. Electrons and holes can be observed to recombine where the two accumulation layers meet, leading to light emission from a well-defined zone, the position of which can be moved to any position along the channel by varying the applied voltages (Figure 2.3.6b). Since the semiconducting layer is unpatterned, light-emission can also be observed from the periphery of the device at distances of more than 500 μm from the edge of the electrodes (Figure 2.3.6c). The observation of a spatially resolved recombination in the channel provides an unambiguous proof of the coexisting electron and hole channels and the truly ambipolar nature of charge transport at such trap-free dielectric–organic semiconductor interfaces [86].

2.3.5 CHARGE TRANSPORT PHYSICS The electronic structure of conjugated polymer semiconductors reflects the complex interplay between intrinsic π-electron delocalization along the polymer backbone and strong electron–phonon coupling, and the existence of energetic and positional disorder in solution-processed thin films. In a hypothetical, infinitely straight polymer chain, the highest occupied molecular orbital (HOMO) and lowest unoccupied

8080_C002.fm Page 116 Friday, April 6, 2007 10:21 AM

116

Organic Field-Effect Transistors

Light A Ca Au

SiOx Holes

Electrons

Au Ca OC1C10-PPV

BCB (50 nm) Thermal SiO2 (300 nm) Gate (doped Si)

B

C

Au

Au

Ca

Ca

FIGURE 2.3.6 (A) Schematic diagram of bottom-gate, ambipolar light-emitting FET with an active semiconducting layer of OC1C10–PPV and BCB gate dielectric. (B) Photograph of light emission from within the channel of the FET (IFET = 30 nA, Vg = –75V). (C) Photograph of light emission from periphery of the device illustrating the spreading of both electron and hole accumulation layers into the unpatterned semiconductor region around the source-drain electrodes. The channel length is 80 μm. (From Zaumseil, J. et al., Nat. Mater., 5, 69–74, 2006. Reprinted with permission. Copyright 2006, Nature Publishing Group.)

molecular orbital (LUMO) states of the neutral polymer are fully delocalized along the polymer chain and, in fact, exhibit significant dispersion with calculated bandwidths of several electron volts [88]. However, as a result of the strong electron–phonon coupling and the disorder-induced finite conjugation length, charges introduced onto the polymer interact strongly with certain molecular vibrations and are able to lower their energy with respect to the extended HOMO/LUMO states by forming localized polaron states surrounded by a region of molecular distortion [89]. There is clear, experimental evidence that the charge carriers carrying the current in a conjugated polymer FET are indeed of polaronic nature. Due to the surrounding molecular distortion and electronic relaxation, the charged molecule exhibits characteristic optical transitions below the absorption edge of the neutral molecule. These can be observed in operational FETs using charge modulation spectroscopy (CMS), which detects changes of the optical transmission of a semitransparent FET device upon gate voltage induced modulation of the carrier concentration in the accumulation layer [90]. In polymers such as poly(di-octyl-fluorene-co-bithiophene) (F8T2) in which close interchain interactions are weakened by the sp3-coordinated carbon atom on

8080_C002.fm Page 117 Friday, April 6, 2007 10:21 AM

Charge Transport Physics

117

au

(a)

au ag bg bu

LUMO

bg C2

π−π∗

C2 C3 au ag bg bu

au

C1

HOMO

bg

−1 −0.5

S

S

n C8H15 C8H15

0

π−π∗

C2 0.5

ΔT/T (104)

ΔT/T (104)

CT C1

−2 ∗

1

C3'

C6H13 ∗

−1 0

C3

C2

S

n∗

π−π∗

1 2

1

1.5

2 2.5 Energy (eV)

3

1

1.4

1.8

2.2

2.6

Energy (eV)

(c)

(b)

FIGURE 2.3.7 (a) Schematic energy diagram of neutral polymer (center), polaronic absorptions in the case of isolated chains (left) and interacting chains (right); charge modulation spectra of F8T2/PMMA (b) and P3HT/PMMA (c) top-gate FETs. (Courtesy of Shlomy Goffri, University of Cambridge.)

the fluorene unit, two characteristic sub-bandgap polaronic absorptions (Figure 2.3.7b) can be accounted for by the dipole-allowed C1 (≈0.4 eV) and visible C2 (1.6 eV) transitions of a simple isolated chain model (Figure 2.3.7a) [91]. In contrast, the charge-induced absorption spectrum of P3HT (Figure 2.3.7c) can only be explained by taking into account interchain interactions [92]. In addition to the C1 (0.3 eV) and C2 (1.3 eV) transitions, the CMS spectrum of high-mobility P3HT exhibits an additional C3 transition (1.6–1.8 eV), which is dipole forbidden in the isolated chain case, and low-energy charge transfer (CT) transitions at 60–120 meV [15,93]. Polarons in P3HT are not confined to a single chain, but are spread over several π-stacked chains. As a result of their two-dimensional nature, the polaron binding energy in P3HT is much reduced. From the position of the CT transition [89], the polaron binding energy Ep can be estimated to be on the order of Ep ≈

ECT ≈ 30–60 meV 2

At sufficiently high temperatures, charge transport of polaronic carriers in conjugated polymers should be governed by the physics of electron transfer processes, which was established by Marcus for chemical reactions and biological electron transfer processes [94]. In order for the localized polaron to hop between neighboring sites, the molecular configuration of the initial (occupied) site and the final (empty) site need to be distorted to a common configuration, where the molecular distortion

8080_C002.fm Page 118 Friday, April 6, 2007 10:21 AM

118

Organic Field-Effect Transistors

of both sites is equal (Figure 2.3.8a). This leads to thermally activated transport even in the absence of disorder. In the nonadiabatic limit, where the time scale for electron hopping is longer than that of the lattice vibrations, the mobility is given by: μ=

⎛ Ep ⎞ e ⋅ a2 ⋅ ν ⋅ exp ⎜− ⎟ ⎝ 2 ⋅ k ⋅T ⎠ k ⋅T

(2.3.1)

where ν is the attempt frequency ν=

π ⋅ J2 2 ⋅ E p ⋅ kT

and a is the typical hopping distance. However, in most experimental systems, the manifestations of the polaronic character of the charge carriers are masked by the effects of disorder. In any solutiondeposited thin film, disorder is present and causes the energy of a polaronic charge carrier on a particular site to vary across the polymer network. Variations of the local conformation of the polymer backbone, presence of chemical impurities or structural defects of the polymer backbone, or dipolar disorder due to random orientation of polar groups of the polymer semiconductor or the gate dielectric result in a significant broadening of the electronic density of states. The transport of charges injected into a molecular solid dominated by the effects of disorder is well understood from the work on molecularly doped polymers and other organic photoconductors used in xerography. Assuming a disorder-broadened Gaussian density of transport states with a characteristic width σ, Bässler [95] has shown on the basis of Monte Carlo simulations that an injected carrier hopping through such an otherwise empty density of states (DOS) relaxes to a dynamic equilibrium energy ε∞ = −

σ2 kT

below the center of the DOS, leading to a characteristic logμ ∝

1 T2

temperature dependence of the mobility (Figure 2.3.8b). The model has been improved by Novikov et al. [96], who showed that the dominant source of diagonal disorder is due to charge–dipole interactions and that spatial correlations of such interactions need to be taken into account in order to explain the commonly observed Poole–Frenkel dependence of the mobility on the electrical field. These researchers derived an expression for the electric field (E) and

8080_C002.fm Page 119 Friday, April 6, 2007 10:21 AM

Charge Transport Physics

119

3

EC

2

1

EF

Hopping at EF

Hopping in the band tail 2 EV

3 Extended state conduction (a)

5

ε/kT

ρ(ε/kT)

0

(ε∞)/kT

−5

−10

−1

0

1

3

2

4

5

lg (t/to) (b)

FIGURE 2.3.8 (a) Schematic energy diagram of DOS of a disordered semiconductor with a mobility edge. (b) Relaxation of energy distribution of an injected charge carrier hopping in a Gaussian DOS as a function of time. The DOS is shown as a dashed line on the right. (From Bässler, H., Phys. Status Solidi B, 175, 15, 1993. Reprinted with permission. Copyright 1993, Wiley.)

8080_C002.fm Page 120 Friday, April 6, 2007 10:21 AM

120

Organic Field-Effect Transistors

temperature dependence of the mobility in a correlated DOS with diagonal as well as nondiagonal positional disorder: 2 ⎡ ⎛ ⎛⎛ σ ⎞3/2 ⎞ e⋅a⋅ E ⎤ 3σ ⎞ ⎥ μ = μ 0 ⋅ exp ⎢− ⎜ ⎟ + 0.78 ⋅ ⎜⎜⎜ ⎟ − 2 ⎟⎟ ⎢⎣ ⎝ 5 ⋅ k B ⋅ T ⎠ ⎥⎦ σ k T ⋅ ⎝ ⎠ B ⎝ ⎠

(2.3.2)

The model describes the transport of individual injected carriers at zero/small carrier concentrations (i.e., in principle, it should not be directly applicable to the relatively high carrier concentrations p = 1018–1019 cm–3 present in the accumulation layer of FETs). Vissenberg [97] has developed a percolation model for variable range hopping transport in the accumulation layer of an FET assuming an exponential DOS with width T0. An expression for the field-effect mobility as a function of carrier concentration p was derived: ⎡ ⎛ T ⎞4 ⎛ T ⎞ ⎤ 0 ⎢ ⎜ 0 ⎟ sin ⎜ π ⎟ ⎥ T0 σ 0 ⎢⎝ T ⎠ −1 ⎝ T0 ⎠ ⎥ = pT 3 e ⎢ (2 ⋅ α) ⋅ Bc ⎥ ⎢ ⎥ ⎣ ⎦ T /T

μ FE

(2.3.3)

where σ0 is the prefactor for the conductivity, α is the effective overlap parameter between localized states, and Bc ≅ 2.8 is the critical number for onset of percolation. Transport in this model can be effectively described as activation from a gate voltagedependent Fermi energy to a specific transport energy in the DOS. Tanase et al. [98] have shown that in a series of isotropic, amorphous PPV polymers the large difference between the low mobility values extracted from spacecharge limited current measurements in LEDs and the comparatively higher fieldeffect mobilities can be explained by the largely different charge carrier concentrations (Figure 2.3.9). It was possible to fit the temperature dependence of the zerofield LED mobility to Equation 2.3.2 and the carrier concentration dependence of the FET mobility to Equation 2.3.3 with a consistent value of σ = 93–125 meV. Building on this work, Pasveer et al. showed that at room temperature the currentvoltage characteristics are dominated by the carrier concentration dependence of the mobility, while at low temperatures and high fields the field dependence of the mobility also needs to be considered [99]. The gate voltage dependence of the FET mobility of MEH-PPV has also been analyzed by Shaked et al. [100]. In several higher mobility amorphous hole transporting materials such as PTAA [18] and TFB [20], as well as in nematic, glassy polyfluorene-co-bithiophene [16], a somewhat different behavior was observed. The field-effect mobility was found to be independent of gate voltage within the carrier concentration range of 1018–1019 cm–3. In PTAA the low-density time-of-flight and high-density field-effect mobilities are of similar magnitude, with the bulk TOF mobility even higher by a factor of two to three at room temperature than the field-effect mobility. The Gaussian disorder

8080_C002.fm Page 121 Friday, April 6, 2007 10:21 AM

Charge Transport Physics

121

10−3 P3HT 10−4

T0 = 425 K

OC1C10-PPV

10−5 0.50 0.45

10−6

Ea (eV)

μh, μFE (cm2/Vs)

T0 = 540 K

10−7

1014

LED

0.40 FET

0.35 0.30 0.25

10−8

OC1C10-PPV

1015

1016

1017 p (cm

−20

−15

1018

−10 −5 Vg (V) 1019

0 1020

−3)

FIGURE 2.3.9 Hole mobility as a function of charge carrier concentration in diode and fieldeffect transistors for P3HT and a PPV derivative. (From Tanase, C. et al., Phys. Rev. Lett., 91, 216601, 2003. Reprinted with permission. Copyright 2003, American Physical Society.)

model was used to extract significantly smaller values of σ = 57 meV and σ = 68–90 meV from the temperature dependence of the time-of-flight and field-effect mobility of amorphous PTAA, respectively (Figure 2.3.3b). The increased σ-value in the case of the FET mobility was attributed to the contribution to energetic disorder from polar disorder in the dielectric close to the charge-transporting accumulation layer. The reason for the different behavior observed in PPVs with room-temperature field-effect mobility < 10–3–10–4 cm2/Vs and the higher mobility PTAA and polyfluorene polymers (μFET = 10–3–10–2 cm2/Vs) might be related to the lower degree of energetic disorder in the latter. With narrow DOS (σ < 60–90 meV), the expected concentration dependence of the room-temperature mobility over a concentration range of 1014–1019 cm–3 spanned by LED/FET measurements is relatively weak (i.e., less than an order of magnitude) and might be masked by other effects such as differences in bulk and interface microstructure, effects of interface roughness, or disorder effects induced by polar or charged groups in the dielectric. An alternative theoretical framework for understanding the effects of disorder is the multiple trapping model, which is well established for describing transport in amorphous silicon and has been claimed to be more appropriate for describing the charge transport in microcrystalline polymers such as P3HT [22] and poly(bis(alkylthienyl-bithiophene) [101,102]. This model assumes that disorder broadening is sufficiently weak that, in a certain energy range, the DOS becomes high enough that electronic states above the so-called mobility edge are extended, while electronic states below the mobility edge remain localized (Figure 2.3.8a). The current is

8080_C002.fm Page 122 Friday, April 6, 2007 10:21 AM

122

Organic Field-Effect Transistors

assumed to be transported by carriers thermally activated into the delocalized states above the mobility edge, while carriers in localized states are effectively trapped and do not contribute to the current. Assuming a specific DOS and a mobility for carriers above the mobility edge, the FET current can be obtained by first determining the position of the quasi-Fermi level at the interface for a particular gate voltage and then calculating the number of free carriers thermally excited above the mobility edge using Fermi–Dirac statistics. Salleo et al. [101] found that the multiple trapping model explained the temperature dependence of the FET mobility of poly(bis(alkyl-thienyl-bithiophene) more consistently than the Vissenberg hopping model; the latter yielded an unphysical dependence of σ0 and T0 on the processing conditions. In spite of detailed investigations to model the charge transport in a mobility regime between 10–2 and 1 cm2/Vs, it can be difficult to distinguish between hopping and band transport models. Many of the qualitative features are common to both hopping and multiple trapping and release models, such as the mobility decreasing with decreasing temperature and the dependence of the mobility on the carrier concentration. Therefore, characterization of the charge transport by techniques that provide complementary information is needed. One of the techniques providing such information is CMS. The spectroscopic properties of polarons in P3HT have been characterized as a function of molecular weight and film deposition conditions by CMS [32]. CMS experiments on regioregular P3HT have revealed a pronounced low-energy charge transfer (CT) transition in the midinfrared spectral region [15]. This transition can be interpreted in the framework of Marcus–Hush electron transfer theory describing the transfer of electrons between neighboring molecules in the presence of strong electron–lattice interactions [103]. The process is governed by two main parameters: The relaxation energy λ (which is twice the polaron binding energy) measures the energy lowering that charged molecules can achieve by adopting a relaxed conformation as a result of the electron–lattice coupling. The transfer integral t is a measure of the strength of the interchain coupling of the electronic wave functions on neighboring molecules. In the weak coupling case (λ > 2t), the lower adiabatic potential surface has a number of minima, and the charge is localized on an individual molecule (Figure 2.3.10a). Under such conditions, a charge transfer optical transition is observed centered at an energy ωCT = λ. In contrast, in the strong coupling case (λ < 2t), the lower adiabatic potential surface has only one minimum and the charge is delocalized over a certain number of neighboring molecules. In this case, an optical charge transfer transition can also be observed; it is not centered around λ, but rather around ωCT = 2t (Figure 2.3.10b). In intermediate MW samples with mobilities > 0.05 cm2/Vs, we observe an intense CT transition centered around 0.1 eV (Figure 2.3.10c). In the highest mobility, highest MW samples, the transition is similarly intense and appears to peak at slightly lower energies below 0.08 eV, which is the low energy cutoff of our experimental setup. In contrast, in the low MW samples with mobilities less than 10–2 cm2/Vs, a much less intense CT transition is observed, and the transition peaks at significantly higher energies on the order of 0.3 eV.

8080_C002.fm Page 123 Friday, April 6, 2007 10:21 AM

Energy

123

Energy

Charge Transport Physics

λ

Δ = 2Hab

Q

Q

(a)

(b)

12

−ΔT/T (10−4)

10 8 76 kD

6 4

29 kD

2 15.4 kD 0

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

Energy (eV) (c)

FIGURE 2.3.10 Potential energy diagram as a function of configuration coordinate illustrating electron transfer between two sites in the case of weak coupling (a) and strong coupling (b). (c) Charge modulation spectra in the midinfrared spectral range of TCB spin-cast P3HT films for different MW. The spectra were obtained by subtracting the infrared absorption spectra of the device structure taken at 10 and –30 V.

The position and intensity of the CT transition appears to be very directly correlated with the field-effect mobility. In high-mobility P3HT, the strong coupling situation applies [103]. The lower intensity of the CT transition indicates a lower degree of interchain polaron delocalization in the low MW samples. A plausible explanation for the reduced intensity and higher energy of the CT transition in the low MW samples is that, due to the enhanced disorder and shorter conjugation length in these samples, the weak coupling regime might apply. Such a crossover behavior between localized and delocalized polarons as a function of MW would provide an intriguing microscopic explanation for the observed rapid increase of mobility with MW below 15–20 kD (see Figure 2.3.4b) [32].

8080_C002.fm Page 124 Friday, April 6, 2007 10:21 AM

124

Organic Field-Effect Transistors

5.2

ΦELP/SUB (eV)

5.0 4.8

PFO P10AF TFB P3HT

4.6 4.4 4.2

{ , }

4.0 3.8 4.0

4.5

5.0

5.5

6.0

ΦSUB (eV)

FIGURE 2.3.11 Dependence of work function of polymer coated substrate, ΦELP/SUB, on the work function of bare substrate, ΦSUB, for four studied materials: P3HT, TFB, poly(9-1decylundecylidene fluorene (P10AF), and polydioctylfluorene (PFO). (From Tengstedt, C. et al., Appl. Phys. Lett., 88, 053502, 2006. Reprinted with permission. Copyright 2006, American Institute of Physics.)

2.3.6 CHARGE INJECTION PHYSICS Another important aspect of the device physics, particularly in the context of short channel OFETs with L < 5 μm, is the injection of charges from a metal source-drain contact into the organic semiconductor. In contrast to inorganic semiconductors, controlled doping of organic semiconductors is still difficult, since dopants incorporated in the form of small molecule counter ions can migrate and cause device instabilities. Since most organic semiconductors that have shown useful FET performance have band gaps > 2 eV, the formation of low-resistance ohmic contacts with common metals is often challenging. The energy barrier for hole injection at the metal–poymer interface is determined by the vacuum work function of the metal contact ΦW and the ionization potential IP of the polymer. For conjugated polymer films spin-coated onto hole injecting metal electrodes, it has been reported that as long as ΦW is smaller than a critical value characteristic of the polymer, no interface dipole is formed [104]. In this case, the barrier for hole injection can be estimated simply by aligning the vacuum levels of the metal and the polymer (Mott–Schottky limit); the measured work function of the metal with the polymer deposited on top increases linearly with ΦW with a slope of one (see Figure 2.3.11). However, when ΦW exceeds said critical value a significant interface dipole can be formed. Positive charges are transferred from the metal to the semiconductor and the position of the Fermi level at the interface becomes pinned at an energy level interpreted as the hole polaron/bipolaron energy level in the polymer semiconductor. This simple picture suggests that, at least in the case of solution-deposited polymers on common hole-injecting contacts, chemical interactions between the metal and

8080_C002.fm Page 125 Friday, April 6, 2007 10:21 AM

Charge Transport Physics

125

the polymer and other defect states in the band gap of the polymer do not influence strongly the contact formation. We emphasize that this is less likely to be the case for metals deposited on top of the polymer semiconductor as well as for reactive metals employed to achieve electron injection into common organic semiconductors. It has been shown that deposition of gold contacts on top of an organic semiconductor, such as pentacene, can result in formation of trap states in the organic semiconductor [105]. There are intriguing reports of efficient charge injection in systems for which Schottky barriers calculated using Mott–Schottky theory should exceed 1 eV, such as hole injection from Ca into P3HT [56] or electron injection from Au into fluorocarbon-substituted oligothiophenes [106]. It is likely that in such systems chemical interactions and interface states are important factors that determine contact formation. In order to understand the contact injection in the OFET not only the interface electronic structure, but also the device configuration and injection geometry need to be taken into account because they determine the potential profile in the vicinity of the contact and the transport of injected charges away from the contacts. In the bottom gate device configuration, the charge injection physics can be studied directly using scanning Kelvin probe microscopy (SKPM) [107–109]. SKPM uses an atomic force microscope tip with a conducting coating operated in noncontact mode to probe the electrostatic potential profile along the channel of the OFET with a spatial resolution on the order of 100 nm (Figure 2.3.12a). The voltage applied to the conducting tip is regulated by a feedback loop so that the electrostatic force between tip and sample is minimized. For polymer TFTs in accumulation the tip potential essentially follows the electrostatic potential in the accumulation layer. Figure 2.3.12(b) shows typical SKPM potential profiles obtained for bottomgate, bottom contact P3HT devices on SiO2 with Au contacts comprising a Cr adhesion layer (Cr–Au) (inset) and pure Cr contacts. It can be seen that the contact resistance at the source and drain contacts exhibits very different behavior in the two cases. In the case of Cr–Au contacts, generally, in the case of contacts for which the Schottky barrier Φb is less than 0.3 eV, the voltage drop across the source and drain contacts is small and the contact resistance at the source contact is very similar to that of the drain contact. This is somewhat unexpected since in normal FET operation the source contact is reverse biased while the drain contact is forward biased, implying that the source contact resistance should be significantly larger than the drain contact resistance. This implies that under conditions that might be typical for high-performance OFETs, the contact resistance is not determined by the Schottky barrier at the interface, but by bulk transport processes in the semiconductor in the vicinity of the contact. Consistent with this interpretation, the contact resistance was found to depend on temperature in the same way as the mobility [107] so that the potential profiles become independent of temperature. This result was explained by invoking the existence of a depletion layer in the vicinity of the contacts. Similar results have been reported using channel length scaling analysis [76]. In contrast in the case of Cr contacts or generally for systems with Schottky barriers exceeding 0.3 eV, the voltage drops across the contacts become very significant and the source resistance is found to be larger than the drain resistance, as

8080_C002.fm Page 126 Friday, April 6, 2007 10:21 AM

126

Organic Field-Effect Transistors

Tip-height control

PSD

+

Channel

Tip Organic layer Source Insulator: SiO2

Drain Id

Gate: n+-Si Vg

Vd

Accumulation layer (a)

Drain

Source 0 0 −2 −4 −6 −8

Local potential (V)

−1 −2 −3

Source

Drain

ΔVs

145 K

−4

T = 300 k Cr-Au

−1 −2 −3

Cr 6

4

2

0

−2

190 K

−5

−4 −5

−6

−6 300 K

−7

−7

ΔVd

−8 −2

0

−1

0

1

2

3

4

5

−8 6

7

Distance from source (μm)

(b)

FIGURE 2.3.12 (a) Schematic diagram of experimental setup for scanning Kelvin probe microscopy (SKPM). (b) Profiles of an L = 5.5 μm P3HT transistor with Cr electrodes taken at three different temperatures (Vg = –40 V, Vd = –8 V). The inset shows a profile obtained after switching the source and drain contacts on the same TFT with both Cr and Cr–Au contacts Vg = –40 V, Vd = –8 V). (From Burgi, L. et al., J. Appl. Phys., 94, 6129–6137, 2003. Reprinted with permission. Copyright 2003, American Institute of Physics.)

8080_C002.fm Page 127 Friday, April 6, 2007 10:21 AM

Charge Transport Physics

127

expected. This implies that in this regime the contact resistance is determined by the injection physics at the interface [107]. It is remarkable that, in spite of the significant expected Schottky barrier height, the contact resistance shows only a very weak increase with decreasing temperature, which is even weaker than that of the field-effect mobility (i.e., the voltage drop at the contacts decreases compared to that across the bulk of the polymer with decreasing temperature and charge transport becomes less contact limited at low temperatures). This behavior cannot be explained in the framework of the commonly used diffusion-limited thermionic emission model [110], which takes into account backscattering into the metal due to the small mean-free path in the organic semiconductor and predicts the activation energy of the contact resistance to be larger than that of the mobility and larger than Φb/kT. Explanation of the experimental data required taking into account disorderinduced broadening of the density of states of the organic semiconductor, which provides carriers with injection pathways through deep states in the DOS, leading to a reduced effective barrier at low temperatures. Similar conclusions have recently been drawn on the basis of channel length scaling experiments [111].

2.3.7 DEFECT STATES AND DEVICE DEGRADATION MECHANISMS Electronic defect states in the semiconductor at the interface between semiconductor and dielectric and inside the dielectric layer can cause instabilities of the threshold voltage of the TFT. For practical applications, the threshold voltage stability is a figure of merit as important as, if not more important than, the field-effect mobility because it is closely related to the operational and shelf lifetime of the device. Most TFT technologies, including those based on a-Si, suffer from threshold voltage shifts induced by bias-temperature stress (BTS). In a-Si TFTs, generation of defect states inside the semiconducting layer, such as dangling bond defects, as well as charge injection into the SiNx gate dielectric has been found to contribute to VT shifts upon BTS; charge injection into the dielectric is the dominant mechanism in high-quality material [112]. Several groups have recently reported systematic BTS investigations and studies of organic TFT characteristics upon exposure to atmospheric conditions and humidity. In most p-type organic semiconductors, a negative shift of the threshold voltage is observed upon prolonged operation of the device in accumulation, which is generally attributed to charge trapping in the organic semiconductor and/or at the active interface. Matters et al. reported negative VT shifts due to charge trapping for a PTV precursor polymer in contact with inorganic SiO2 dielectric; these were more pronounced in the presence of water than when the device was operated in vacuum or dry air [113]. Street et al. reported significant negative VT shifts in F8T2/SiO2 bottom-gate, bottom-contact TFTs [114], which were more pronounced than reported for top-gate F8T2 devices with a polymer dielectric [16]. Street et al. also found the VT stability of PQT/SiO2 devices to be significantly better than that of F8T2/SiO2

8080_C002.fm Page 128 Friday, April 6, 2007 10:21 AM

128

Organic Field-Effect Transistors

devices. It is clear from these experiments whether the device configuration, choice of contacts, and dielectric play a crucial role in determining the device stability. There is little known at present about the nature of the electronic states involved in defect formation and device degradation. Few experimental studies have been aimed at understanding at a microscopic level the nature of defect states in organic semiconductors [115]. Device modeling has been used to understand the subthreshold characteristics of OFETs [116]. Based on an analysis of the relationship between the current decay at early times after FET turn-on and the hole concentration in the channel, Street et al. have suggested that charge trapping occurs due to formation of low-mobility bipolarons by reaction of two polarons [114,117]. However, Deng et al. performed optical spectroscopy of field-induced charge on F8T2/PMMA TFTs exhibiting significant VT shifts, but were unable to detect the spectroscopic signature of bipolarons [91]. When discussing threshold voltage shifts in OFETs, it is important to distinguish between reversible and irreversible charge trapping effects [118]. Reversible charge trapping depends on the duty cycle during operation and can be recovered by not operating the device for several minutes or hours. Threshold voltage shifts due to irreversible charge trapping are independent of duty cycle and do not recover on timescales of hours if the device is not operated while being kept in the dark. However, the irreversible threshold shift can be erased by illuminating the sample with above-bandgap light [119,120] (Figure 2.3.13a). The spectral dependence of the light-induced recovery follows the absorption spectrum of the organic semiconductor (Figure 2.3.13b). Charge traps that can be emptied in this way must be located inside the organic semiconductor or directly at the interface, but cannot be located inside the gate dielectric. It has also been reported that a positive gate voltage stress leads to a shift of VT to more positive values [119]. This has recently been explained by injection and trapping of negative electrons at the interface [56]. Zilker et al. have reported that, in films of p-type solution-processed pentacene in contact with an organic photoresist dielectric, the threshold voltage shifts to more positive values for negative gate bias stress during operation in air [121]. The VT shift was the more pronounced the smaller the source-drain voltage was. This was interpreted as the result of mobile ions drifting in the gate dielectric in the presence of water. Negative ions drifting towards the active interface cause accumulation of positive countercharges in the semiconducting layer. Only during operation in vacuum or in dry air was a negative VT shift of –3 V after application of Vg = –20 V for 1000 s observed resulting from charge trapping at or near the interface. Rep et al. have investigated the role of ionic impurities originating from the substrate on the conductivity of P3HT films [122]. On Na2O containing glass substrates, Na+ ions were found to drift towards the negatively biased contact, leaving behind negative charge centers on the glass surface. Gomes et al. have claimed recently that the bias stress instability in organic FETs is caused by trapped water in the organic semiconductor [123]. The preceding results point to the crucial role of the gate dielectric in determining the operational and shelf stability of the device. Several groups have recently reported very encouraging BTS and shelf lifetime data for solution-processed

8080_C002.fm Page 129 Friday, April 6, 2007 10:21 AM

Charge Transport Physics

129

8 × 10−8

1st meas. stressed 150 s. 350 s. 900 s.

Drain current (A)

7 × 10−8 6 × 10−8 5 × 10−8 4 × 10−8 3 × 10−8 2 × 10−8 1 × 10−8 0 −25

−20

−15

−10

−5

0

Gate voltage (V) (a) 0.08 hv

0.6

0.06

0.4

0.04

0.2

0.02

0.0

2

3

4

5

6

7

τ1/2−1 (s−1)

Absorption (arb. units)

0.8

0.00

E (eV) (b)

FIGURE 2.3.13 (a) Pulsed transfer characteristics of bottom-gate F8T2/SiO2 FET after applying a negative gate bias stress showing subsequent recovery of the threshold voltage shift after illuminating the device for different periods of time. (From Salleo, A. and Street, R.A., J. Appl. Phys., 94, 471–479, 2003. Reprinted with permission. Copyright 2003, American Physical Society.) (b) Comparison of the wavelength dependence of the time constant for the light-induced trap release in TFB/SiO2 with the absorption spectrum of the organic semiconductor. (From Burgi, L. et al., Syn. Met., 146, 297–309, 2004. Reprinted with permission. Copyright 2004, Elsevier.)

OFETs measured and stored in air without special encapsulation. PTAA combined with low-k dielectrics exhibits excellent shelf life with no discernible VT shift upon storage in air and light for periods of several months [18]. Similarly, TFTs based on TFB with BCB dielectric exhibit very good operational stability during accelerated lifetime testing at temperatures of 120°C [20]. In both cases, the good stability is believed to be related to the use of an apolar, low-k dielectric, which is less susceptible to ionic impurities, and the amorphous microstructure of the aryl-aminebased polymer semiconductor with good thermal and photostability and low degree of energetic disorder.

8080_C002.fm Page 130 Friday, April 6, 2007 10:21 AM

130

Organic Field-Effect Transistors

The group at Plastic Logic has recently reported excellent operational stability results on unencapsulated polymer TFTs fabricated on PET substrates [10]. Although, of course, significant work to assess and improve the operational and shelf life of OFETs under realistic application conditions and to understand degradation mechanisms in much more detail remains, these early results strongly suggest that solution-processed OFETs can exhibit device stability and reliability similar to if not higher than their a-Si counterparts.

2.3.8 OUTLOOK Solution-processable organic FETs have become a promising emerging technology for low-cost, large-area electronics on flexible plastic substrates. FET performance is approaching that of a-Si TFTs, and solution/printing-based manufacturing processes have been developed. Device operational and environmental stability has improved significantly recently as a result of availability of organic semiconductors with higher inherent oxidative stability, better understanding of the requirements for gate dielectrics, and more controlled manufacturing processes. In this chapter, we have reviewed recent progress in understanding the device physics of solution-processable organic semiconductors. It should be apparent from the discussion that although much progress has been made in understanding the materials physics and requirements for high-performance FETs, understanding of the fundamental excitations and processes at a microscopic level involved in charge transport and injection as well as device degradation is still much more superficial than the corresponding level of fundamental understanding available in inorganic semiconductors. Particularly, many fundamental aspects of the correlation between the structure and physics of charge transport at solution-processed organic–organic heterointerfaces remain to be explored. However, the field of organic electronics is gaining momentum. Continued breakthroughs in materials and device performance, concrete industrial applications in active matrix flexible electronic paper displays, and simple, low-cost intelligent labels are emerging on the horizon to be commercialized within the next three to five years. It is likely that new scientific discoveries and technological advances will continue to cross-fertilize each other for the foreseeable future.

REFERENCES 1. Burroughes, J.H., Bradley, D.D.C., Brown, A.R., Marks, R.N., Mackay, K., Friend, R.H., Burn, P.L. and Holmes, A.B., Light-emitting diodes based on conjugated polymers, Nature, 347, 539–541, 1990. 2. Yu, G., Gao, J., Hummelen, J.C., Wudl, F. and Heeger, A.J., Polymer photovoltaic cells: enhanced efficiencies via a network of internal donor-acceptor heterojunctions, Science, 270, 1789–1791, 1995. 3. Burroughes, J.H., Jones, C.A. and Friend, R.H., New semiconductor device physics in polymer diodes and transistors, Nature, 335, 137–141, 1988.

8080_C002.fm Page 131 Friday, April 6, 2007 10:21 AM

Charge Transport Physics

131

4. Heilmeier, G.H. and Zanoni, L.A., Surface studies of alpha-copper phtalocyanine films, J. Phys. Chem. Solids, 25, 603, 1964. 5. Horowitz, G., Peng, X.Z., Fichou, D. and Garnier, F., A field-effect transistor based on conjugated alpha-sexithyenyl, Sol. St. Commun., 72, 381–384, 1989. 6. Ebisawa, E., Kurokawa, T. and Nara, S., Electrical properties of polyacetylene/polysiloxane interface, J. Appl. Phys., 54, 3255, 1983. 7. Koezuka, H., Tsumara, A. and Ando, T., Field-effect transistor with polythiophene thin film, Synth. Met., 18, 699–704, 1987. 8. Baude, P.F., Ender, D.A., Haase, M.A., Kelley, T.W., Muyres, D.V. and Theiss, S.D., Pentacene-based radio-frequency identification circuitry, Appl. Phys. Lett., 82, 3964–3966, 2003. 9. McCulloch, I., Heeney, M., Bailey, C., Genevicius, K., Macdonald, I., Shkunov, M., Sparrowe, D., Tierney, S., Wagner, R., Zhang, W., Chabinyc, M.L., Kline, R.J., McGehee, M.D. and Toney, M.F., Liquid-crystalline semiconducting polymers with high charge-carrier mobility, Nat. Mater., 5, 328–333, 2006. 10. Reynolds, K., Burns, S., Banach, M., Brown, T., Chalmers, K., Cousins, N., Creswell, L., Etchells, M., Hayton, C., Jacobs, K., Menon, A., Siddique, S., Ramsdale, C., Reeves, W., Watts, J., Werne, T.v., Mills, J., Curling, C., Sirringhaus, H., Amundson, K. and McCreary, M.D., Printing of polymer transistors for flexible active matrix displays, Proc. Int. Display Workshop (IDW), 2004. 11. Sirringhaus, H., Sele, C.W., Ramsdale, C.R. and Werne, T.V., Manufacturing of organic transistor circuits by solution-based printing. In Organic electronics, ed. H. Klauk, Wiley–VCH, Weinheim, Germany, 2006. 12. Forrest, S.R., The path to ubiquitous and low-cost organic electronic appliances on plastic, Nature, 428, 911–918, 2004. 13. Gelinck, G.H., Huitema, H.E.A., Van Veenendaal, E., Cantatore, E., Schrijnemakers, L., Van der Putten, J., Geuns, T.C.T., Beenhakkers, M., Giesbers, J.B., Huisman, B.H., Meijer, E.J., Benito, E.M., Touwslager, F.J., Marsman, A.W., Van Rens, B.J.E. and De Leeuw, D.M., Flexible active-matrix displays and shift registers based on solutionprocessed organic transistors, Nat. Mater., 3, 106–110, 2004. 14. Clemens, W., Fix, I., Ficker, J., Knobloch, A. and Ullmann, A., From polymer transistors toward printed electronics, J. Mater. Res., 19, 1963–1973, 2004. 15. Sirringhaus, H., Brown, P.J., Friend, R.H., Nielsen, M.M., Bechgaard, K., LangeveldVoss, B.M.W., Spiering, A.J.H., Janssen, R.A.J., Meijer, E.W., Herwig, P. and de Leeuw, D.M., Two-dimensional charge transport in self-organized, high-mobility conjugated polymers, Nature, 401, 685–688, 1999. 16. Sirringhaus, H., Wilson, R.J., Friend, R.H., Inbasekaran, M., Wu, W., Woo, E.P., Grell, M. and Bradley, D.D.C., Mobility enhancement in conjugated polymer field-effect transistors through chain alignment in a liquid-crystalline phase, Appl. Phys. Lett., 77, 406–408, 2000. 17. Burroughes, J.H. and Friend, R.H., The semiconductor device physics of polyacetylene. In: Conjugated polymers, ed. J.L. Brédas and R. Silbey, Kluwer, Dordrecht, The Netherlands, 1991, 555–622. 18. Veres, J., Ogier, S.D., Leeming, S.W., Cupertino, D.C. and Khaffaf, S.M., Low-k insulators as the choice of dielectrics in organic field-effect transistors, Adv. Func. Mater., 13, 199–204, 2003. 19. Veres, J., Ogier, S., Lloyd, G. and de Leeuw, D., Gate insulators in organic fieldeffect transistors, Chem. Mater., 16, 4543–4555, 2004.

8080_C002.fm Page 132 Friday, April 6, 2007 10:21 AM

132

Organic Field-Effect Transistors

20. Chua, L.L., Ho, P.K.H., Sirringhaus, H. and Friend, R.H., High-stability ultrathin spin-on benzocyclobutene gate dielectric for polymer field-effect transistors, Appl. Phys. Lett., 84, 3400–3402, 2004. 21. Bao, Z., Dodabalapur, A. and Lovinger, A.J., Soluble and processable regioregular poly(3-hexylthiophene) for thin film field-effect transistor applications with high mobility, Appl. Phys. Lett., 69, 4108–4110, 1996. 22. Sirringhaus, H., Tessler, N. and Friend, R.H., Integrated optoelectronic devices based on conjugated polymers, Science, 280, 1741–1744, 1998. 23. Bjornholm, T., Hassenkam, T., Greve, D.R., McCullough, R.D., Jayaraman, M., Savoy, S.M., Jones, C.E. and McDevitt, J.T., Polythiophene nanowires, Adv. Mater., 11, 1218–1221, 1999. 24. Kline, R.J., McGehee, M.D., Kadnikova, E.N., Liu, J.S. and Frechet, J.M.J., Controlling the field-effect mobility of regioregular polythiophene by changing the molecular weight, Adv. Mater, 15, 1519, 2003. 25. Merlo, J.A. and Frisbie, C.D., Field effect transport and trapping in regioregular polythiophene nanofibers, J. Phys. Chem. B, 108, 19169–19179, 2004. 26. Chang, J.F., Sun, B.Q., Breiby, D.W., Nielsen, M.M., Solling, T.I., Giles, M., McCulloch, I. and Sirringhaus, H., Enhanced mobility of poly(3-hexylthiophene) transistors by spin-coating from high-boiling-point solvents, Chem. Mater., 16, 4772–4776, 2004. 27. Zhao, N., Botton, G.A., Zhu, S.P., Duft, A., Ong, B.S., Wu, Y.L. and Liu, P., Microscopic studies on liquid crystal poly(3,3″′- dialkylquaterthiophene) semiconductor, Macromolecules, 37, 8307–8312, 2004. 28. Wang, G.M., Swensen, J., Moses, D. and Heeger, A.J., Increased mobility from regioregular poly(3-hexylthiophene) field-effect transistors, J. Appl. Phys., 93, 6137–6141, 2003. 29. Kim, D.H., Park, Y.D., Jang, Y.S., Yang, H.C., Kim, Y.H., Han, J.I., Moon, D.G., Park, S.J., Chang, T.Y., Chang, C.W., Joo, M.K., Ryu, C.Y. and Cho, K.W., Enhancement of field-effect mobility due to surface-mediated molecular ordering in regioregular polythiophene thin film transistors, Adv. Func. Mater., 15, 77–82, 2005. 30. Zen, A., Pflaum, J., Hirschmann, S., Zhuang, W., Jaiser, F., Asawapirom, U., Rabe, J.P., Scherf, U. and Neher, D., Effect of molecular weight and annealing of poly (3hexylthiophene)s on the performance of organic field-effect transistors, Adv. Func. Mater, 14, 757–764, 2004. 31. Zhang, R. et al., Nanostructure dependence of field-effect mobility in regioregular poly(3-hexylthiophene) thin film field effect transistors, J. Am. Chem. Soc., 28, 3480–3481, 2006. 32. Chang, J.F., Clark, J., Zhao, N., Sirringhaus, H., Breiby, D.W., Andreasen, J.W., Nielsen, M.M., Giles, M., Heeney, M. and McCulloch, I., Molecular-weight dependence of interchain polaron delocalization and exciton bandwidth in high-mobility conjugated polymers, Phy. Rev. B, 74, 15318, 2006 (submitted). 33. Kline, R.J., McGehee, M.D. and Toney, M.F., Highly oriented crystals at the buried interface in polythiophene thin-film transistors, Nat. Mater., 5, 222–228, 2006. 34. Sirringhaus, H., Tessler, N., Thomas, D.S., Brown, P.J. and Friend, R.H., Highmobility conjugated polymer field-effect transistors. In: Advances in Solid State Physics, Vol. 39, ed. B. Kramer Vieweg, Braunschweig, Germany, 1999, 101–110. 35. Abdou, M.S.A., Orfino, F.P., Son, Y. and Holdcroft, S., Interaction of oxygen with conjugated polymers: Charge transfer complex formation with poly(3-alkylthiophenes), J. Am. Chem. Soc., 119, 4518–4524, 1997.

8080_C002.fm Page 133 Friday, April 6, 2007 10:21 AM

Charge Transport Physics

133

36. Ficker, J., Ullmann, A., Fix, W., Rost, H. and Clemens, W., Stability of polythiophenebased transistors and circuits, J. Appl. Phys., 94, 2638–2641, 2003. 37. Ficker, J., von Seggern, H., Rost, H., Fix, W., Clemens, W. and McCulloch, I., Influence of intensive light exposure on polymer field-effect transistors, Appl. Phys. Lett., 85, 1377–1379, 2004. 38. Ong, B.S., Wu, Y.L., Liu, P. and Gardner, S., High-performance semiconducting polythiophenes for organic thin-film transistors, J. Am. Chem. Soc., 126, 3378–3379, 2004. 39. Heeney, M., Bailey, C., Genevicius, K., Shkunov, M., Sparrowe, D., Tierney, S. and McCulloch, I., Stable polythiophene semiconductors incorporating thieno 2,3-b thiophene, J. Am. Chem. Soc., 127, 1078–1079, 2005. 40. Herwig, P.T. and Mullen, K., A soluble pentacene precursor: Synthesis, solid-state conversion into pentacene, and application in a field-effect transistor, Adv. Mater., 11, 480, 1999. 41. Afzali, A., Dimitrakopoulos, C.D. and Graham, T.O., Photosensitive pentacene precursor: Synthesis, photothermal patterning, and application in thin-film transistors, Adv. Mater., 15, 2066, 2003. 42. Gelinck, G.H., Geuns, T.C.T. and de Leeuw, D.M., High-performance all-polymer integrated circuits, Appl. Phys. Lett., 77, 1487–1489, 2000. 43. Afzali, A., Dimitrakopoulos, C.D. and Breen, T.L., High-performance, solution-processed organic thin film transistors from a novel pentacene precursor, J. Am. Chem. Soc., 124, 8812–8813, 2002. 44. Minakata, T. and Natsume, Y., Direct formation of pentacene thin films by solution process, Syn. Met., 153, 1–4, 2005. 45. Aramaki, S., Sakai, Y. and Ono, N., Solution-processable organic semiconductor for transistor applications: Tetrabenzoporphyrin, Appl. Phys. Lett., 84, 2085–2087, 2004. 46. Chang, P.C., Lee, J., Huang, D., Subramanian, V., Murphy, A.R. and Frechet, J.M.J., Film morphology and thin film transistor performance of solution-processed oligothiophenes, Chem. Mater., 16, 4783–4789, 2004. 47. Laquindanum, J.G., Katz, H.E. and Lovinger, A.J., Synthesis, morphology, and fieldeffect mobility of anthradithiophenes, J. Am. Chem. Soc., 120, 664–672, 1998. 48. Katz, H.E., Recent advances in semiconductor performance and printing processes for organic transistor-based electronics, Chem. Mater., 16, 4748–4756, 2004. 49. Mushrush, M., Facchetti, A., Lefenfeld, M., Katz, H.E. and Marks, T.J., Easily processable phenylene-thiophene-based organic field-effect transistors and solutionfabricated nonvolatile transistor memory elements, J. Am. Chem. Soc., 125, 9414–9423, 2003. 50. Stingelin-Stutzmann, N., Smits, E., Wondergem, H., Tanase, C., Blom, P., Smith, P. and De Leeuw, D., Organic thin-film electronics from vitreous solution-processed rubrene hypereutectics, Nat. Mater., 4, 601–606, 2005. 51. van de Craats, A.M., Stutzmann, N., Bunk, O., Nielsen, M.M., Watson, M., Mullen, K., Chanzy, H.D., Sirringhaus, H. and Friend, R.H., Meso-epitaxial solution-growth of self-organizing discotic liquid-crystalline semiconductors, Adv. Mater., 15, 495–499, 2003. 52. Pisula, W., Menon, A., Stepputat, M., Lieberwirth, I., Kolb, U., Tracz, A., Sirringhaus, H., Pakula, T. and Müllen, K., A zone-casting technique for device fabrication of field-effect transistors based on discotic hexa-peri-hexabenzocoronene, Adv. Mater., 17, 684–689, 2005.

8080_C002.fm Page 134 Friday, April 6, 2007 10:21 AM

134

Organic Field-Effect Transistors

53. Xiao, S.X., Myers, M., Miao, Q., Sanaur, S., Pang, K.L., Steigerwald, M.L. and Nuckolls, C., Molecular wires from contorted aromatic compounds, Angew. Chem.Inter. Ed., 44, 7390–7394, 2005. 54. McCulloch, I., Zhang, W.M., Heeney, M., Bailey, C., Giles, M., Graham, D., Shkunov, M., Sparrowe, D. and Tierney, S., Polymerizable liquid crystalline organic semiconductors and their fabrication in organic field effect transistors, J. Am. Chem. Soc., 13, 2436–2444, 2003. 55. Dimitrakopoulos, C.D. and Malenfant, P.R.L., Organic thin film transistors for large area electronics, Adv. Mater., 14, 99, 2002. 56. Chua, L.L., Zaumseil, J., Chang, J.F., Ou, E.C.W., Ho, P.K.H., Sirringhaus, H. and Friend, R.H., General observation of n-type field-effect behaviour in organic semiconductors, Nature, 434, 194–199, 2005. 57. Katz, H.E., Lovinger, A.J., Johnson, J., Kloc, C., Siegrist, T., Li, W., Lin, Y.-Y. and Dodabalapur, A., A soluble and air stable organic semiconductor with high electron mobility, Nature, 404, 478–481, 2000. 58. Babel, A. and Jenekhe, S.A., High electron mobility in ladder polymer field-effect transistors, J. Am. Chem. Soc., 125, 13656–13657, 2003. 59. Facchetti, A., Mushrush, M., Yoon, M.H., Hutchison, G.R., Ratner, M.A. and Marks, T.J., Building blocks for n-type molecular and polymeric electronics. Perfluoroalkyl versus alkyl-functionalized oligothiophenes (nT; n = 2–6). Systematics of thin film microstructure, semiconductor performance, and modeling of majority charge injection in field-effect transistors, J. Am. Chem. Soc., 126, 13859–13874, 2004. 60. Waldauf, C., Schilinsky, P., Perisutti, M., Hauch, J. and Brabec, C.J., Solutionprocessed organic n-type thin-film transistors, Adv. Mater., 15, 2084, 2003. 61. Anthopoulos, T.D., de Leeuw, D.M., Cantatore, E., van’t Hof, P., Alma, J. and Hummelen, J.C., Solution processable organic transistors and circuits based on a C70 methanofullerene, J. Appl. Phys., 98, 2005. 62. Meijer, E.J., De Leeuw, D.M., Setayesh, S., Van Veenendaal, E., Huisman, B.H., Blom, P.W.M., Hummelen, J.C., Scherf, U. and Klapwijk, T.M., Solution-processed ambipolar organic field-effect transistors and inverters, Nat. Mater., 2, 678–682, 2003. 63. Babel, A., Wind, J.D. and Jenekhe, S.A., Ambipolar charge transport in air-stable polymer blend thin-film transistors, Adv. Func. Mater., 14, 891–898, 2004. 64. Shkunov, M., Simms, R., Heeney, M., Tierney, S. and McCulloch, I., Ambipolar fieldeffect transistors based on solution-processable blends of thieno 2,3-b thiophene terthiophene polymer and methanofullerenes, Adv. Mater., 17, 2608, 2005. 65. Anthopoulos, T.D., de Leeuw, D.M., Cantatore, E., Setayesh, S., Meijer, E.J., Tanase, C., Hummelen, J.C. and Blom, P.W.M., Organic complementary-like inverters employing methanofullerene-based ambipolar field-effect transistors, Appl. Phys. Lett., 85, 4205–4207, 2004. 66. Kunugi, Y., Takimiya, K., Negishi, N., Otsubo, T. and Aso, Y., An ambipolar organic field-effect transistor using oligothiophene incorporated with two 60 fullerenes, J. Mater. Chem., 14, 2840–2841, 2004. 67. Facchetti, A., Yoon, M.H. and Marks, T.J., Gate dielectrics for organic field-effect transistors: New opportunities for organic electronics, Adv. Mater., 17, 1705–1725, 2005. 68. Drury, C.J., Mutsaers, C.M.J., Hart, C.M., Matters, M. and de Leeuw, D.M., Lowcost all-polymer integrated circuits, Appl. Phys. Lett., 73, 108–110, 1998. 69. Sirringhaus, H., Kawase, T., Friend, R.H., Shimoda, T., Inbasekaran, M., Wu, W. and Woo, E.P., High-resolution inkjet printing of all-polymer transistor circuits, Science, 290, 2123–2126, 2000.

8080_C002.fm Page 135 Friday, April 6, 2007 10:21 AM

Charge Transport Physics

135

70. Jones, R.A.L. and Richards, R.W., Polymers at surfaces and interfaces, Cambridge University Press, Cambridge, 1999. 71. Chua, L.L., Ho, P.K.H., Sirringhaus, H. and Friend, R.H., Observation of field-effect transistor behavior at self- organized interfaces, Adv. Mater., 16, 1609–1613, 2004. 72. Majewski, L.A., Grell, M., Ogier, S.D. and Veres, J., A novel gate insulator for flexible electronics, Org. Elect., 4, 27–32, 2003. 73. Majewski, L.A., Schroeder, R. and Grell, M., One volt organic transistor, Adv. Mater., 17, 192–196, 2005. 74. Chen, F.C., Chu, C.W., He, J., Yang, Y. and Lin, J.L., Organic thin-film transistors with nanocomposite dielectric gate insulator, Appl. Phys. Lett., 85, 3295, 2004. 75. Sandberg, H.G.O., Backlund, T.G., Osterbacka, R. and Stubb, H., High-performance all-polymer transistor utilizing a hygroscopic insulator, Adv. Mater., 16, 1112–1116, 2004. 76. Hamadani, B.H. and Natelson, D., Temperature-dependent contact resistances in highquality polymer field-effect transistors, Appl. Phys. Lett., 84, 443–445, 2004. 77. Sele, C.W., von Werne, T., Friend, R.H. and Sirringhaus, H., Lithography-free, selfaligned inkjet printing with subhundred-nanometer resolution, Adv. Mater., 17, 997–1001, 2005. 78. Halik, M., Klauk, H., Zschieschang, U., Schmid, G., Dehm, C., Schutz, M., Maisch, S., Effenberger, F., Brunnbauer, M. and Stellacci, F., Low-voltage organic transistors with an amorphous molecular gate dielectric, Nature, 431, 963–966, 2004. 79. Yoon, M.-H., Facchetti, A. and Marks, T.J., Molecular dielectric multilayers for lowvoltage organic thin film transistors, Proc. Nat. Acad. Sci., 102, 4679, 2005. 80. Park, Y.D., Kim, D.H., Jang, Y., Hwang, M., Lim, J.A. and Cho, K., Low-voltage polymer thin-film transistors with a self-assembled monolayer as the gate dielectric, Appl. Phys. Lett., 87, 243509, 2005. 81. Sakanoue, T., Fujiwara, E., Yamada, R. and Tada, H., Visible light emission from polymer-based field-effect transistors, Appl. Phys. Lett., 84, 3037–3039, 2004. 82. Cicoira, F., Santato, C., Melucci, M., Favaretto, L., Gazzano, M., Muccini, M. and Barbarella, G., Organic light-emitting transistors based on solution-cast and vacuumsublimed films of a rigid core thiophene oligomer, Adv. Mater., 18, 169–172, 2006. 83. Santato, C., Capelli, R., Loi, M.A., Murgia, M., Cicoira, F., Roy, V.A.L., Stallinga, P., Zamboni, R., Rost, C., Karg, S.E. and Muccini, M., Tetracene-based organic lightemitting transistors: Optoelectronic properties and electron injection mechanism, Syn. Met., 146, 329–334, 2004. 84. Ahles, M., Hepp, A., Schmechel, R. and von Seggern, H., Light emission from a polymer transistor, Appl. Phys. Lett., 84, 428–430, 2004. 85. Rost, C., Karg, S., Riess, W., Loi, M.A., Murgia, M. and Muccini, M., Light-emitting ambipolar organic heterostructure field-effect transistor, Syn. Met., 146, 237–241, 2004. 86. Zaumseil, J., Friend, R.H. and Sirringhaus, H., Spatial control of the recombination zone in an ambipolar light-emitting organic transistor, Nat. Mater., 5, 69–74, 2006. 87. Swensen, J.S., Soci, C. and Heeger, A.J., Light emission from an ambipolar semiconducting polymer field-effect transistor, Appl. Phys. Lett., 87, 253511, 2005. 88. van der Horst, J.W., Bobbert, P.A., Michels, M.A.J., Brocks, G. and Kelly, P.J., Ab initio calculation of the electronic and optical excitations in polythiophene: Effects of intra- and interchain screening, Phys. Rev. Lett., 83, 4413–4416, 1999. 89. Brédas, J.L., Beljonne, D., Coropceanu, V. and Cornil, J., Charge-transfer and energytransfer processes in pi-conjugated oligomers and polymers: A molecular picture, Chem. Rev., 104, 4971–5003, 2004.

8080_C002.fm Page 136 Friday, April 6, 2007 10:21 AM

136

Organic Field-Effect Transistors

90. Brown, P.J., Sirringhaus, H., Harrison, M.G. and Friend, R.H., Optical spectroscopy of field-induced charge in self-organized high mobility poly(3-hexylthiophene), Phys. Rev. B, 63, 5204–5214, 2001. 91. Deng, Y. and Sirringhaus, H., Optical absorptions of polyfluorene transistors, Phys. Rev. B, 72, 045207, 2005. 92. Beljonne, D., Cornil, J., Sirringhaus, H., Brown, P.J., Shkunov, M., Friend, R.H. and Brédas, J.L., Optical signature of delocalized polarons in conjugated polymers, Adv. Funct. Mater., 11, 229, 2001. 93. Osterbacka, R., An, C.P., Jiang, X.M. and Vardeny, Z.V., Two-dimensional electronic excitations in conjugated polymer nanocrystals, Science, 287, 839–843, 2000. 94. Marcus, R.A., Exchange reactions and electron transfer reactions including isotopic exchange. Theory of oxidation-reduction reactions involving electron transfer. Part 4 — A statistical-mechanical basis for treating contributions from solvent, ligands, and inert salt, Discuss. Faraday Soc., 29, 21, 1960. 95. Bässler, H., Charge transport in disordered organic photoconductors — A Monte Carlo simulation study, Phys. Status Solidi B, 175, 15, 1993. 96. Novikov, S.V., Dunlap, D.H., Kenkre, V.M., Parris, P.E. and Vannikov, A.V., Essential role of correlations in governing charge transport in disordered organic materials, Phys. Rev. Lett., 81, 4472–4475, 1998. 97. Vissenberg, M.C.J.M. and Matters, M., Theory of the field-effect mobility in amorphous organic transistors, Phys. Rev. B, 57, 12964–12967, 1998. 98. Tanase, C., Meijer, E.J., Blom, P.W.M. and de Leeuw, D.M., Unification of the hole transport in polymeric field-effect transistors and light-emitting diodes, Phys. Rev. Lett., 91, 216601, 2003. 99. Pasveer, W.F., Cottaar, J., Tanase, C., Coehoorn, R., Bobbert, P.A., Blom, P.W.M., de Leeuw, D.M. and Michels, M.A.J., Unified description of charge-carrier mobilities in disordered semiconducting polymers, Phys. Rev. Lett., 94, 206601, 2005. 100. Shaked, S., Tal, S., Roichman, Y., Razin, A., Xiao, S., Eichen, Y. and Tessler, N., Charge density and film morphology dependence of charge mobility in polymer fieldeffect transistors, Adv. Mater., 15, 913–917, 2003. 101. Salleo, A., Chen, T.W., Volkel, A.R., Wu, Y., Liu, P., Ong, B.S. and Street, R.A., Intrinsic hole mobility and trapping in a regioregular poly(thiophene), Phys. Rev. B, 70, 115311, 2004. 102. Street, R.A., Northrup, J.E. and Salleo, A., Transport in polycrystalline polymer thinfilm transistors, Phys. Rev. B, 71, 165202, 2005. 103. Beljonne, D., Cornil, J., Sirringhaus, H., Brown, P.J., Shkunov, M., Friend, R.H. and Brédas, J.L., Optical signature of delocalized polarons in conjugated polymers, Adv. Func. Mater., 11, 229–234, 2001. 104. Tengstedt, C., Osikowicz, W., Salaneck, W.R., Parker, I.D., Hsu, C.H. and Fahlmann, M., Fermi-level pinning at conjugated polymer interfaces, Appl. Phys. Lett., 88, 053502, 2006. 105. Pesavento, P.V., Chesterfield, R.J., Newman, C.R. and Frisbie, C.D., Gated four-probe measurements on pentacene thin-film transistors: Contact resistance as a function of gate voltage and temperature, J. Appl. Phys., 96, 7312–7324, 2004. 106. Facchetti, A., Mushrush, M., Katz, H.E. and Marks, T.J., n-Type building blocks for organic electronics: A homologous family of fluorocarbon-substituted thiophene oligomers with high carrier mobility, Adv. Mater., 15, 33–37, 2003. 107. Burgi, L., Richards, T.J., Friend, R.H. and Sirringhaus, H., Close look at charge carrier injection in polymer field-effect transistors, J. Appl. Phys., 94, 6129–6137, 2003.

8080_C002.fm Page 137 Friday, April 6, 2007 10:21 AM

Charge Transport Physics

137

108. Puntambekar, K.P., Pesavento, P.V. and Frisbie, C.D., Surface potential profiling and contact resistance measurements on operating pentacene thin-film transistors by Kelvin probe force microscopy, Appl. Phys. Lett., 83, 5539–5541, 2003. 109. Nichols, J.A., Gundlach, D.J. and Jackson, T.N., Potential imaging of pentacene organic thin-film transistors, Appl. Phys. Lett., 83, 2366–2368, 2003. 110. Scott, J.C. and Malliaras, G.G., Charge injection and recombination at the metal–organic interface, Chem. Phys. Lett., 299, 115–119, 1999. 111. Hamadani, B.H. and Natelson, D., Nonlinear charge injection in organic field-effect transistors, J. Appl. Phys., 97, 2005. 112. Kanicki, J. In: Thin film transistors, ed. P.A.C.R. Kagan, Marcel Dekker, New York, 2003. 113. Matters, M., de Leeuw, D.M., Herwig, P.T. and Brown, A.R., Bias-stress induced instability of organic thin film transistors, Syn. Met., 102, 998–999, 1999. 114. Street, R.A., Salleo, A. and Chabinyc, M.L., Bipolaron mechanism for bias-stress effects in polymer transistors, Phys. Rev. B, 68, art. no. 085316, 2003. 115. Schmechel, R. and von Seggern, H., Electronic traps in organic transport layers, Phys. Status Solidi A — Appl. Res., 201, 1215–1235, 2004. 116. Lindner, T., Paasch, G. and Scheinert, S., Influence of distributed trap states on the characteristics of top and bottom contact organic field-effect transistors, J. Mater. Res., 19, 2014–2027, 2004. 117. Salleo, A. and Street, R.A., Kinetics of bias stress and bipolaron formation in polythiophene, Phys. Rev. B, 70, art. no. 235324, 2004. 118. Salleo, A., Endicott, F. and Street, R.A., Reversible and irreversible trapping at room temperature in poly(thiophene) thin-film transistors, Appl. Phys. Lett., 86, 2005. 119. Salleo, A. and Street, R.A., Light-induced bias stress reversal in polyfluorene thinfilm transistors, J. Appl. Phys., 94, 471–479, 2003. 120. Burgi, L., Richards, T., Chiesa, M., Friend, R.H. and Sirringhaus, H., A microscopic view of charge transport in polymer transistors, Syn. Met., 146, 297–309, 2004. 121. Zilker, S.J., Detcheverry, C., Cantatore, E. and de Leeuw, D.M., Bias stress in organic thin-film transistors and logic gates, Appl. Phys. Lett., 79, 1124–1126, 2001. 122. Rep, D.B.A., Morpurgo, A.F., Sloof, W.G. and Klapwijk, T.M., Mobile ionic impurities in organic semiconductors, J. Appl. Phys., 93, 2082–2090, 2003. 123. Gomes, H.L., Stallinga, P., Colle, M., de Leeuw, D.M. and Biscarini, F., Electrical instabilities in organic semiconductors caused by trapped supercooled water, Appl. Phys. Lett., 88, 2006.

8080_C002.fm Page 138 Friday, April 6, 2007 10:21 AM

8080_C002.fm Page 139 Friday, April 6, 2007 10:21 AM

2.4

Contact Effects in Organic Field-Effect Transistors

Matthew J. Panzer and C. Daniel Frisbie CONTENTS 2.4.1 Introduction................................................................................................139 2.4.2 Definition of an Ohmic Contact ................................................................140 2.4.3 Origins of Contact Resistance ...................................................................140 2.4.3.1 Electronic Structure and Potential Barriers at Metal–Organic Interfaces .....................................................................................140 2.4.3.2 Charge Transport across Metal–Organic Interfaces ...................142 2.4.3.3 Influence of Channel Dimensions...............................................145 2.4.3.4 Influence of Device Architecture ................................................146 2.4.4 Measuring Contact Resistance...................................................................148 2.4.4.1 Extrapolation of Device Resistance to Zero Channel Length....148 2.4.4.2 Gated Four-Probe Measurements................................................149 2.4.4.3 Kelvin Probe Force Microscopy .................................................150 2.4.4.4 Measured Contact Resistance Values..........................................151 2.4.5 Contact Engineering ..................................................................................154 2.4.5.1 Chemical Modifications ..............................................................154 2.4.5.2 Ambipolar and Light-Emitting OFETs.......................................155 2.4.5.3 Channel Dimensions: How Small? .............................................155 References..............................................................................................................155

2.4.1 INTRODUCTION In an ideal organic field-effect transistor (OFET), the source and drain contacts are ohmic, meaning the value of the contact resistance is negligibly small in comparison with the electrical resistance of the semiconductor (i.e., the channel resistance). While this situation can be achieved in real devices, there are several practical considerations for fabricating OFETs that are not contact limited [1]. In this chapter, we begin by defining an ohmic contact and continue with a discussion of the origins of contact resistance. Subsequent sections cover techniques used to quantify contact resistance in working OFETs, along with tables of contact resistance values for 139

8080_C002.fm Page 140 Friday, April 6, 2007 10:21 AM

140

Organic Field-Effect Transistors

L Source Rsource

Rchannel

Drain Rdrain

Insulator Gate

FIGURE 2.4.1 Schematic of an OFET showing equivalent resistances corresponding to the source contact resistance, the channel resistance, and the drain contact resistance. The total contact resistance is Rc = Rsource + Rdrain and the total device resistance is Rtot = Rc + Rchannel.

common OFET geometries based on typical organic semiconductors. We conclude with comments on contact engineering to improve OFET performance.

2.4.2 DEFINITION OF AN OHMIC CONTACT In traversing an OFET channel from source to drain, charge carriers are (1) injected from the source contact into the semiconductor channel; (2) transported across the length of the channel; and (3) extracted from the channel into the drain. These processes can be roughly thought of as three resistors in series (Figure 2.4.1). The resistances associated with carrier injection and collection steps can be grouped into the contact resistance (Rc), while the resistance associated with crossing the channel length in the semiconductor is termed the channel resistance (Rchannel). Keeping the contact resistance small compared to the channel resistance is crucial to the realization of “ohmic contacts” in OFETs (i.e., for an ohmic contact, Rc << Rchannel). If the contacts are ohmic, then they are not bottlenecks to current flow and they can source and sink all the current that can be transported by the channel under the given bias conditions. Importantly, nominally high-resistance contacts can still be ohmic as long as they are able to source or sink the current driven through an even more resistive channel. This definition also implies that contacts that function ohmically for a given channel length may in fact no longer be ohmic as the channel length shrinks. This is because Rchannel scales proportionally with length, so making smaller OFETs results in smaller channel resistances and puts a smaller upper bound on the acceptable value of contact resistance. An additional issue in FETs is that the channel resistance is continuously lowered with increasing gate voltage, and thus contact resistances must be low compared to the channel resistance at high gate voltage in order to be considered ohmic. A key factor in determining contact resistance is the electronic structure or band lineup at the metal–organic interface. We will see later that contact resistance is also a function of contact geometry and fabrication procedures.

2.4.3 ORIGINS OF CONTACT RESISTANCE 2.4.3.1 ELECTRONIC STRUCTURE AND POTENTIAL BARRIERS METAL–ORGANIC INTERFACES

AT

From conventional semiconductor electronics, it is known that creating a lowresistance ohmic contact requires alignment of the metal Fermi level (EF) with the

8080_C002.fm Page 141 Friday, April 6, 2007 10:21 AM

Contact Effects in Organic Field-Effect Transistors

Evac

Evac

Evac

φm

φm

IP

EC-EF

EGAP VB (HOMO)

IP CB (LUMO)

CB (LUMO) EGAP

EF +

EF-EV

(a)

Δ

Evac

CB (LUMO) EF

141

EF-EV +

EF-EV

EF

VB (HOMO)

VB (HOMO)

(b)

EGAP

(c)

FIGURE 2.4.2 (a) Simple band line-up diagram for a metal-organic semiconductor interface assuming that the Mott-Schottky rule holds and that the vacuum levels for the metal and semiconductor are aligned. (b) Application of a positive bias to the metal can result in hole injection into the semiconductor by thermionic emission over the barrier. (c) Band line-up diagram in the case where an interface dipole is present, causing a shift (Δ) in the vacuum levels across the junction.

energy levels (bands) of the semiconductor. Figure 2.4.2(a) shows a simple diagram depicting energy level alignments at a metal–organic semiconductor junction. This diagram assumes that the Mott–Schottky rule holds: namely, that the vacuum levels of the metal and organic semiconductor are in registry. The outcome of this assumption is that one can easily estimate the conduction band and valence band offsets from the Fermi level. For example, the valence band offset, EF – EV, is the difference between the work function of the metal (φm) and the ionization potential (IP) of the organic semiconductor, or EF − EV = φm − IP

(2.4.1)

This offset is in turn a good estimate of the potential barrier to hole injection from the metal to the semiconductor. As shown in Figure 2.4.2(b), applying a positive bias to the metal relative to the semiconductor can result in hole injection (so-called thermionic emission) into the valence band if the holes can surmount the barrier EF – EV. A similar discussion holds for electron injection into the conduction band, but in that case the metal is biased negative and the barrier is determined by EC – EF. Thus, the electronic structure of metal–organic semiconductor interfaces plays a crucial role in determining the charge transport characteristics of the contact. In reality, however, many metal–organic semiconductor interfaces do not follow the Mott–Schottky rule and the electronic structure is significantly more complicated than depicted in Figure 2.4.2(a) [2]. Often, an interface dipole (Δ) is present that shifts the vacuum level of the semiconductor with respect to the metal, as shown in Figure 2.4.2(c). Interface dipoles have several possible origins, including charge transfer between the semiconductor molecules and the metal, reduction of the metal work function by adsorption of the organic layer (even absorption of a noble gas on a metal modifies the metal work function by pushing back the metal surface elec-

8080_C002.fm Page 142 Friday, April 6, 2007 10:21 AM

142

Organic Field-Effect Transistors

trons), and population of metal-induced mid-gap states (new energy levels) at the interface. Sometimes, simple chemical notions (e.g., high or low electron affinities) can be used to predict the sign of the dipole (i.e., whether it points to the metal or to the semiconductor), but it is a difficult computational problem to predict its magnitude. The magnitude of the potential change due to the dipole (Δ) must be included in the calculation of the valence band offset, EF – EV = φm – IP ± Δ

(2.4.2))

where the mathematical sign in front of Δ is chosen to reflect the direction of the interfacial dipole. Currently, the only way to be sure of the magnitude of Δ and thus of the energy level offsets is to measure them. Most of our quantitative knowledge of electronic structure of metal–organic interfaces comes from ultraviolet photoemission spectroscopy (UPS) and inverse photoemission spectroscopy (IPES), which probe filled and empty electronic states, respectively. These techniques allow measurement of valence band and conduction band (HOMO and LUMO) positions with respect to the metal Fermi level and the presence of any interfacial dipoles. Several research groups worldwide have undertaken measurements of the electronic structure of a variety of metal–organic semiconductor combinations; work in this area is still ongoing and current understanding is consequently somewhat fluid, though great progress has been made [3–6]. Figure 2.4.3 shows the band line-up diagram for the gold–pentacene interface determined by UPS and IPES measurements. As has been discussed elsewhere, pentacene is a widely used p-channel (hole-conducting) organic semiconductor and gold is generally used to make ohmic contacts in pentacene OFETs [7]. The diagram shows a strong dipole at the interface pointed toward the gold (shifting up the gold vacuum level). In addition, there is a surprising 0.5 eV barrier for hole injection at this contact (EF – EV = 0.47 eV) [8]. Such a large barrier does not seem consistent with low contact resistance. However, as we will point out in the next subsection, the nature of real metal–organic interfaces is complicated and the transport properties of the source and drain contacts are only partially determined by the presence of potential barriers; other factors, such as structural disorder near the contact, also play a role. One final point concerning interfacial electronic structure is that it has been found that slightly contaminated surfaces can sometimes lead to better charge injection properties [9]. The reasoning is that the contamination layer (either adventitious adsorbates or intentionally deposited interlayers) can prevent interactions between the organic semiconductor and the metal that produce unfavorable dipoles or that tend to pin the Fermi level in the gap. For example, it has been found that gold electrodes coated with a thin layer of the organic metal PEDOT:PSS are better hole injectors into polythiophene than bare-gold electrodes, even though PEDOT:PSS has a lower work function than gold.

2.4.3.2 CHARGE TRANSPORT

ACROSS

METAL–ORGANIC INTERFACES

Despite the fact that significant potential barriers (>0.3 eV) exist at many metal–organic semiconductor interfaces, it is possible to make ohmic source and

8080_C002.fm Page 143 Friday, April 6, 2007 10:21 AM

Contact Effects in Organic Field-Effect Transistors

143

∆dip = 0.60 eV

Evac

Wf = 5.05 eV

Bulk

Interface

Evac

Φe = 1.17 eV EF

Eg = 2.68 eV

Au

Bulk

Interface

Φh = 0.47 eV

Pentacene

FIGURE 2.4.3 Band line-up diagram for the Au-pentacene interface. An interface dipole results in a vacuum level shift of 0.6 eV. The diagram indicates that the valence band (HOMO level) and conduction band (LUMO level) have finite widths. The hole injection barrier (Φh) is determined by taking the difference in energy between the Fermi level and the edge of the valence band. The slight shrinking of the band gap at the interface is due to polarization effects by the metal. (Reprinted from F. Amy, C. Chan, and A. Kahn, Polarization at the gold/pentacene interface, Organic Electronics, 2005, 6, 85–91, with permission from Elsevier.)

drain contacts in OFETs. A likely reason for this is that the charge injection mechanism is probably not simple thermionic emission in which carriers must surmount the full potential barrier, as originally indicated in Figure 2.4.2(b). Instead, at large interfacial electric field strength, field emission (tunneling) through the barrier can become possible; this is a process that effectively lowers the potential barrier. Another possible injection mechanism involves defect-assisted transport in which carriers bypass the barrier by hopping through midgap states. Figure 2.4.4 shows a simple comparison of these different charge injection mechanisms. There is mounting experimental evidence that the charge injection process at the source electrode in OFETs is not simple thermionic emission [10–13]. First, measurements of the source contact resistance as a function of temperature reveal that the injection process is indeed thermally activated (which is consistent with thermionic emission), but the activation energies are generally much smaller than the estimated potential barriers determined by photoemission spectroscopy. In some cases, the activation energy associated with the source contact resistance is very comparable to the activation energy associated with the carrier field effect mobility, suggesting that transport of charge in the semiconductor near the contact is the limiting bottleneck, not the actual metal-to-semiconductor emission process. Second, channel potential measurements by Kelvin probe force microscopy and the four-probe method, which are described later in this chapter, indicate that the contact resistances and temperature dependences associated with the individual source and drain electrodes are nearly identical. From a thermionic emission

8080_C002.fm Page 144 Friday, April 6, 2007 10:21 AM

144

Organic Field-Effect Transistors

Evac

Evac

Evac

EF +

EGAP EF-EV +

VB (HOMO)

V (a) Thermionic emission

CB (LUMO)

CB (LUMO)

CB (LUMO) EF

EGAP + +

+

V (b) Field emission (tunneling)

VB (HOMO)

EGAP

EF ++ +

VB (HOMO)

V (c) Defect-assisted injection

FIGURE 2.4.4 Comparison of different charge injection mechanisms at a biased metalsemiconductor contact: (a) thermionic emission, (b) field emission (tunneling), (c) defect assisted injection.

viewpoint, one would expect the resistance at the source electrode (the injecting contact) to be much larger than the resistance at the drain (the collecting contact). The fact that the source and drain contact resistance behaviors are very similar in most devices also indicates that the bottleneck at the contacts is related to charge transport in the (disordered) organic semiconductor near the contacts and not simply due to an injection barrier at the metal–organic interface. An additional point is that, in general, the source and drain contact resistances are gate voltage dependent; specifically, they decrease with increasing gate voltage. The variation of the contact resistance with gate voltage is essentially identical for both the source and drain and it is also similar to the variation of the channel resistance. The close tracking of the gate voltage dependence on the source, drain, and channel resistances also indicates that contact resistance depends on film transport properties near the contact. The simplest picture that can explain these collective observations is that the source contact resistance is the sum of (1) resistance arising from charge injection over or through the potential barrier; and (2) resistance due to transport in the disordered depletion region near the contact. The drain resistance would simply be due to the latter part. Perhaps because the presence of a strong gate field facilitates field emission through the barrier, resistance (1) is often not limiting for the source; instead, resistance (2) dominates, and thus the source and drain resistances are comparable. This description provides a useful physical picture, but quantitative modeling of the injection and collection processes is complicated by details of the device geometry and the fabrication process, such as the evaporation of hot metal onto the organic semiconductor, which might produce many defects. Work on detailed understanding of transport at OFET contacts is ongoing.

8080_C002.fm Page 145 Friday, April 6, 2007 10:21 AM

Contact Effects in Organic Field-Effect Transistors

145

Insulator Semiconductor

W Source

Drain

L

FIGURE 2.4.5 Top view of an OFET showing the relevant channel dimensions.

2.4.3.3 INFLUENCE

OF

CHANNEL DIMENSIONS

As mentioned earlier, channel dimensions are also critical in determining the importance of contact effects in OFETs. Figure 2.4.5 shows the top view of a typical OFET with channel length (L) and channel width (W). It is useful for comparison purposes to report the normalized or specific contact resistance (Rc′) as a “raw” resistance value multiplied by the width (W) of the channel. Thus, the units of Rc′ are Ωcm. Physically, the reciprocal of this value is the contact conductance per unit width of contact. The total device resistance can be expressed as RTOT = Rchannel + Rc

(2.4.3)

sheet In terms of the channel sheet resistance ( Rchannel , units of Ω/square, and VG dependent) and the specific contact resistance (Rc′, units of Ω cm and also VG dependent), this equation becomes:

sheet RTOT = Rchannel

L Rc′ + W W

(2.4.4)

This equation facilitates understanding of how the channel dimensions (L and W) affect the relative magnitudes of the contact resistance and the channel resistance. Note that the channel resistance scales as L/W but the contact resistance scales as 1/W; it does not depend on L. Consider two different OFET devices on the same semiconductor/insulator/gate substrate: both have the same channel width (equal W), but the length of the channel of the second device is 10 times smaller than that of the first (L2 = L1/10), as depicted in Figure 2.4.6(a). Both devices have equal contact resistances Rc because W is the same. But because the channel resistance scales with L/W (the source-drain current scales with W/L), the channel resistance of the second device is 10 times smaller than that of the first device. This means that contact resistance is potentially much more important in the shorter channel device because it contributes a larger fraction of the total resistance.

8080_C002.fm Page 146 Friday, April 6, 2007 10:21 AM

146

Organic Field-Effect Transistors

Device 1

W1 = W2 L1 = 10 × L2

Device 2

Equal contact resistances; Device 2 has a lower channel resistance (by 10X) (a)

Device 3

W3 = 0.5 × W4 L3 = 0.5 × L4 (W/L)3 = (W/L)4

Device 4

Equal channel resistances; Device 4 has a lower contact resistance (by 2X) (b)

FIGURE 2.4.6 (a) Decreasing the channel length (L) at constant channel width (W) leads to lower channel resistance and increased relative contact resistance, Rc/RTOT. (b) Increasing the channel size at constant (W/L) decreases the relative contact resistance.

A second hypothetical pair of OFET devices is shown in Figure 2.4.6(b). In this case, both contact patterns share the same W/L ratio, but both W and L are twice as large for the second device of this duo. While the channel resistances are now equivalent, the contact resistance of the larger device is half that of the smaller one. The larger device in this case is less likely to be contact limited. In general, one must pay attention to the magnitude of the contact resistance when scaling OFETs to very small lengths (L) because the contact resistance of short-channel devices can quickly become the dominating resistance. As a final note, to avoid nonidealities in transistor output curves, W/L should always be ≥10 in order to avoid fringing field effects at the edges of the channel.

2.4.3.4 INFLUENCE

OF

DEVICE ARCHITECTURE

There are two main architectures to choose from in OFET fabrication: the top contact and the bottom contact configurations. The physical difference between the two is the order of fabrication steps. That is, the source/drain contacts are either deposited before or after the semiconductor layer is deposited to create a bottom contact or top contact device, respectively. One can also build the entire transistor on top of the semiconductor layer (the so-called top gate architectures), in which the insulator

8080_C002.fm Page 147 Friday, April 6, 2007 10:21 AM

Contact Effects in Organic Field-Effect Transistors

Top contacts Source

147

Bottom contacts Source

Drain

Drain

Semiconductor Insulator Gate (a)

(b) Bottom contacts/top gate

Top contacts/top gate Gate Source

Drain

Insulator

Source

Drain

Semiconductor Substrate (c)

(d)

FIGURE 2.4.7 Four possible OFET architectures (in cross-section), including (a) top contacts, (b) bottom contacts, (c) top contacts with a top gate, and (d) bottom contacts with a top gate.

and gate contact are sequentially deposited on top of either of the two contact configurations. All four of these OFET architectures are shown schematically in Figure 2.4.7. Top contact OFETs (Figure 2.4.7a) generally exhibit the lowest contact resistances. This is likely because of the increased metal–semiconductor contact area in this configuration. A major contribution to contact resistance in the top contact configuration is access resistance (see Figure 2.4.8a). Access resistance results from the requirement that charge carriers must travel from the source contact on top of the film down to the accumulation layer (the channel) at the semiconductor–insulator interface and then back up to the drain contact to be extracted. In order to minimize access resistance, the thickness of the organic semiconductor layer should not be too large. However, some researchers have proposed that access resistance is less than might be expected for top contact OFETs because the contact metal penetrates the film down to the accumulation layer (perhaps due to large peak-to-valley roughness of the semiconductor film or the nature of the metal deposition process) [11]. This scenario is shown in Figure 2.4.8(b). With the bottom contact architecture (Figure 2.4.7b), access resistance is not an issue because the contacts are in the same plane as the OFET channel. In addition, very small channel dimensions (W, L < 10 μm) can be prepatterned on the insulator using conventional photolithography. A limitation to the bottom contact configuration, however, is that film morphology in the vicinity of the contacts is often nonideal. A number of researchers have demonstrated that the organic semiconductor grain sizes are very small near the contacts, presumably due to heterogeneous nucleation phenomena [14]. Pentacene molecules, for example, prefer to “stand up” with the long axis of the molecule perpendicular to the plane of the substrate when deposited on the commonly used insulator SiO2 [15]. When deposited on top of gold contacts,

8080_C002.fm Page 148 Friday, April 6, 2007 10:21 AM

148

Organic Field-Effect Transistors

Polycrystalline semiconductor

Access resistance Source

Drain

Source

Drain

Insulator Insulator Gate (a)

(b)

FIGURE 2.4.8 (a) Access resistance in a top contact OFET represented by arrows indicating the injection and extraction of charge through the bulk semiconductor film between the source/drain contacts and the conductive channel. The plus signs represent the hole accumulation layer of a p-channel device. (b) Detailed view of the contact/semiconductor interface near the accumulation layer of a top contact OFET with a very rough, polycrystalline semiconductor thin film. Ovals highlight the penetration of the top contacts deep into the film; in such devices, access resistance can often be negligible.

however, strong interactions between the pentacene π-clouds and the metal surface lead to tiny grains at the contact and, in some cases, voids are observed [14]. Semiconductor growth at the complex triple interface (contact-semiconductor-insulator) is not very well understood, although it is clear that the bottom contact configuration almost always creates greater contact resistance than in the case of top contacts. Of the two top-gate OFET architectures (Figure 2.4.7c and 2.4.7d), the top contact/top gate configuration (Figure 2.4.7c) is the more favorable of the two because bottom contact/top gate devices suffer from access resistance. However, it should be noted that both top gate architectures face the additional concerns of semiconductor top surface roughness (since this is where the channel will form) and forming a stable interface between the insulator and the top of the semiconductor film. Solution deposition of the top insulator material, for example, may damage the underlying semiconductor film. Finally, regarding the alignment of the top gate contact to the OFET channel in top gate devices, care must also be taken to ensure that the gate reaches completely across the entire length (L) of the device. If the length of the gate electrode is less than the channel length or if the gate is simply misaligned, additional contact resistance will be introduced as a result of ungated semiconductor regions at one or both of the contacts.

2.4.4 MEASURING CONTACT RESISTANCE 2.4.4.1 EXTRAPOLATION OF DEVICE RESISTANCE CHANNEL LENGTH

TO

ZERO

Since the total resistance a charge carrier experiences during its journey from source to drain (i.e., VD/ID) is the sum of the contact resistance and the channel resistance

8080_C002.fm Page 149 Friday, April 6, 2007 10:21 AM

Contact Effects in Organic Field-Effect Transistors

149

Device resistance (kΩ⋅cm)

200

150

100

Contact resistance

50

0

0

20

40 60 80 Channel length, L (μm)

100

FIGURE 2.4.9 Example of a transmission line (RTOT · W vs. L) plot at a given VG value. Extrapolation of the data to a channel length of zero yields the specific contact resistance Rc′ as the y-intercept.

as discussed earlier, one of the simplest ways to quantitatively measure OFET contact resistance involves isolating these two resistances by extrapolation. By fabricating several pairs of source and drain contacts with different channel lengths (but constant W) on the same semiconductor film, one measures the total OFET resistance and makes a plot of resistance versus channel length. Linear extrapolation of the plot to L = 0 effectively eliminates the channel resistance and yields the contact resistance as the y-intercept. This method of measuring contact resistance is known as the transmission line method or R vs. L technique. Figure 2.4.9 shows an example of a resistance (RTOT·W) versus channel length plot with extrapolation used to determine the specific contact resistance, Rc′. Because both the channel and contact resistances are affected by the charge density in the channel, the resistance measurements are made at a single gate voltage (usually in the linear operation regime). The drain voltage could also be scaled to L in order to obtain the same source-to-drain lateral electric field for each device. While this technique is straightforward to perform and understand, it has two disadvantages. First, it can be tedious to fabricate and test several devices in order to obtain the resistance versus channel length plot. In addition, the contact resistance obtained by the transmission line technique lumps the individual source and drain resistances together.

2.4.4.2 GATED FOUR-PROBE MEASUREMENTS In order to separate the individual contributions of the source and drain contacts to the total contact resistance, a more sophisticated measurement technique is required. The gated four-probe technique utilizes two narrow, voltage-sensing electrodes situated between the source and drain electrodes and slightly protruding into the channel, as shown in Figure 2.4.10(a). During the course of normal OFET electrical characterization, these voltage-sensing probes are connected to high-input impedance electrometers that sense the channel potential at the two probe positions (V1,

8080_C002.fm Page 150 Friday, April 6, 2007 10:21 AM

150

Organic Field-Effect Transistors

15

Drain

Potential (V)

Source

10

x

V2

V1 5

ΔVS 0 0 L 2L L 3 3 (a)

ΔVD

VD = 15 V VG = 75 V

V2

V1

0

L/3 2L/3 Channel position, x

L

(b)

FIGURE 2.4.10 (a) Top view of an OFET contact pattern with two voltage sensing probes (V1, V2) penetrating slightly into the channel in order to perform gated 4-probe measurements. (b) Hypothetical extrapolated linear channel potential profile based on the voltage-sensing probe data, which reveals the voltage drops (ΔVS, ΔVD) at the source and drain contacts that are responsible for the contact resistance. The dashed line is the ideal (no contact resistance) linear potential profile for an applied drain voltage of 15 V and a gate voltage of 75 V.

V2) without passing any current. In the linear regime of OFET operation (VG >> VD), the channel should be uniform in charge carrier density with a linear drop in electrostatic potential along L from source to drain. Therefore, any drops in electrostatic potential that occur at the contacts (due to contact resistance) will be manifested upon extrapolation of the channel potential profile based on the voltage-sensing probe measurements. As depicted in Figure 2.4.10(b), contact resistance at the source and drain electrodes results in a smaller than expected slope of the potential versus channel position profile. The profile is estimated by linear extrapolation between V1 and V2. Individual source and drain contact resistances are calculated by dividing the voltage drops ΔVS and ΔVD by the source-drain current, respectively. By isolating the source and drain contact contributions to the total contact resistance, the gated four-probe technique provides more information than the transmission line technique, and it is possible to determine Rc in one device (vs. several). An important caveat for the gated four-probe technique is that the extrapolated channel potential profile will only be valid for strict linear regime OFET operation (VG >> VD), where the channel potential profile can be expected to be linear and uniform.

2.4.4.3 KELVIN PROBE FORCE MICROSCOPY While the previously described techniques both require extrapolation of measured data in order to calculate the contact resistance, Kelvin probe force microscopy (KFM, also known as scanning surface potential microscopy or scanning potentiometry) can be used to determine the source and drain contributions to the contact resistance directly. In KFM, a conductive atomic force microscope (AFM) tip is scanned over the operational OFET channel twice. On the first pass, the topography

8080_C002.fm Page 151 Friday, April 6, 2007 10:21 AM

Contact Effects in Organic Field-Effect Transistors

Semiconductor Insulator Gate L 2L 3 3

x L

Potential (V)

Drain

Source

0

15

(1) Topography (2) Surface potential

AFM tip

151

ΔVD

VD = 15 V VG = 75 V

10

5

ΔVS 0

0

(a)

L/3

2L/3

L

Channel position, x (b)

FIGURE 2.4.11 (a) Application of KFM to characterize an OFET. (b) Hypothetical channel potential profile measured by the KFM technique; voltage drops at the source and drain contacts (ΔVS, ΔVD) are measured directly. The dashed line is the ideal (no contact resistance) linear potential profile for an applied drain voltage of 15 V and a gate voltage of 75 V.

of the device is recorded; then, the tip is lifted a small distance (~10 nm) off the device and the second pass retraces the channel topology in the air (or better yet, vacuum) above the sample while the electrostatic potential is recorded (see Figure 2.4.11a). The electrostatic potential data are converted into the OFET surface potential profile by subtracting an appropriate background trace. Thus, KFM measures the full channel potential profile. Since ΔVS and ΔVD are measured directly, calculating the source and drain contributions to the contact resistance can be done without any extrapolation. Figure 2.4.11(b) shows a hypothetical channel potential profile measured by KFM. A clear advantage of KFM over the gated four-probe technique is that the entire channel potential profile is measured experimentally instead of using only two points to extrapolate a linear profile (compare Figures 2.4.10b and 2.4.11b). Thus, other bottlenecks to charge transport (e.g., potential drops at grain boundaries in the channel) can also be visualized using this technique.

2.4.4.4 MEASURED CONTACT RESISTANCE VALUES A collection of experimental OFET contact resistance values is provided in Tables 2.4.1 and 2.4.2. Table 2.4.1 presents contact resistance values for seven different polycrystalline oligomeric semiconductors, including the current benchmark material, pentacene. One must be careful, however, in making comparisons between contact resistance values reported by different groups. It has been proposed that contact resistance is caused by a combination of thermionic emission and carrier diffusion through a depletion region, with the latter dominating in some cases. As a result, contact resistance depends on the level of accumulated charge in the OFET channel, and many experiments have shown that contact resistance is inversely proportional to VG [16,17]. Figure 2.4.12 shows the typical decrease of the source, drain, and channel resistances with increasing gate voltage (carrier density) for a pentacene OFET with gold contacts.

8080_C002.fm Page 152 Friday, April 6, 2007 10:21 AM

152

Organic Field-Effect Transistors

TABLE 2.4.1 OFET Contact Resistance Values: Evaporated Oligomer Films Semiconductor Pentacene

PTCDI-C5

PTCDI-C8 PTCDI-C12 PTCDI-C13

Ant-2T-Ant Tet-2T-Tet

Ωcm)b Contact metal TC/BCa RC (Ω Linear acenes (p-channel) Au TC 3 × 104 Au TC 1 × 103 Au TC RS = 3 × 102 RD = 1 × 103 Au BC RS = 4 × 104 RD = 2 × 104 Ag TC 1 × 103 Ag TC 2 × 103 Pt TC 4 × 103 Ca TC 4 × 104 TC <2 × 104 Hg(liq) Perylene diimides (n-channel) Au TC 4 × 104 Ag TC 4 × 104 Ag TC 1 × 105 Ca TC 6 × 104 Ag TC 2 × 104 Ag TC 3 × 104 LiF/Al TC 1 × 105 Acene-capped bithiophenes (p-channel) Au TC 3 × 104 Au TC 2 × 104

Method

Ref.

R vs. L Four probe KFM

17 11 19

KFM

19

Four probe Four probe Four probe Four probe R vs. L

11 16 11 11 20

Four Four Four Four Four Four Four

probe probe probe probe probe probe probe

21 21 22 21 22 22 23

Four probe Four probe

24 24

a

OFET architecture: top contact (TC) or bottom contact (BC). Total (source resistance, RS + drain resistance, RD) contact resistance, RC, unless individual values are shown. b

Thus, while the contact resistance values summarized in Tables 2.4.1 and 2.4.2 are generally reported at large values of VG in the linear operating regime, not all of the measurements have been made at the same induced charge density in the channel. Additionally, there has been some evidence that, even with “good” contacts, contact resistance is also inversely proportional to the semiconductor mobility [9,12,17]. Again, it is not necessarily the value of the contact resistance of an OFET that is significant, but rather its value in comparison to the channel resistance. In the linear regime, the scaled OFET channel resistance (RchannelW) is given by: RchannelW =

L μQ′

(2.4.5)

where μ is the semiconductor mobility and Q′ is the accumulated charge per unit area in the channel. Inserting typical values for L (100 μm), μ (0.1 cm2/Vs), and Q′

8080_C002.fm Page 153 Friday, April 6, 2007 10:21 AM

Contact Effects in Organic Field-Effect Transistors

153

TABLE 2.4.2 OFET Contact Resistance Values: Solution-Deposited Polymers and Oligomers Semiconductor

Contact metal

Ωcm)b RC (Ω

TC/BCa

Polymer films (p-channel) TC RS = 2 × 105 RD = 5 × 104 Au BC 1 × 104 Au BC ~5 × 105 Au BC RS < 5 × 103 Ag BC RS = 2 × 104 Cr/Au BC RS = 2 × 104 Cu BC 3 × 106 Cu BC RS = 3 × 105 Cr BC 1 × 105 Cr BC RS = 5 × 106 Au BC RS = 1 × 107 Cr/Au BC RS > 7 × 107

P3HT

Au

F8T2

Soluble oligomer (n-channel) Au BC ~1 × 106

C70-PCBM

Method

Ref.

Four probe

26

R vs. L R vs. L KFM KFM KFM R vs. L KFM R vs. L KFM KFM KFM

13 18 10 10 10 13 10 13 10 10 10

R vs. L

27

a

OFET architecture: top contact (TC) or bottom contact (BC). Total (source resistance, RS + drain resistance, RD) contact resistance, RC, unless individual values are shown. b

Nfree (×1012/cm2) 0

Resistance (Ω)

1011

1.7

4.2

6.7

9.2

VD = −4V

109 RFilm RS RD

107 105 103

V0 0

VT −10

−20

−30

−40

VG (V)

FIGURE 2.4.12 Evolution of the source and drain contact resistances (RS and RD), as well as the channel resistance (RFilm) for a pentacene OFET with increasing carrier density (Nfree), calculated from the capacitance of the Al2O3 dielectric layer and the gate voltage. L = 100 μm, W = 1000 μm.

8080_C002.fm Page 154 Friday, April 6, 2007 10:21 AM

154

Organic Field-Effect Transistors

(1 μC/cm2) yields a channel resistance of 1 × 105 Ωcm, which can be compared to the numbers in the tables. Table 2.4.2 presents a collection of contact resistance values measured for a few solution-deposited polymeric and oligomeric semiconductors. These values tend to be higher than in the case of evaporated oligomeric films due to lower mobilities. It has also been observed that, in certain instances, the source contact can constitute a majority of the total contact resistance in polymeric films [10,25].

2.4.5 CONTACT ENGINEERING 2.4.5.1 CHEMICAL MODIFICATIONS As discussed previously, the fundamental reason why contact resistance is generally greater for the bottom contact OFET architecture is the poor semiconductor morphology on top of or near the source and drain contacts that hinders charge injection/extraction. Thus, one of the simplest ways to improve the quality of the semiconductor film on top of bottom contacts is to pretreat the contacts chemically before depositing the semiconductor. In the simplest case, one can use thiol-terminated hydrocarbon molecules to form a self-assembled monolayer (SAM) on top of metallic contacts [14]. The affinity of the thiol group for metal surfaces and SAM formation with such molecules have been well-studied. By forming a CH3-terminated SAM on top of the contacts, the semiconductor layer will no longer “see” the metal, but rather a hydrophobic surface with a different surface energy. As a result, the semiconductor morphology will be modified on top of the contacts — hopefully, in such a way as to improve charge injection/extraction and reduce contact resistance. Taking the idea of SAM formation one step further, one can also use molecules with various chemical functionalities (not only hydrocarbons) for pretreating the contacts. SAMs featuring both electron-withdrawing and electron-donating end groups opposite the thiol-linking ends have been used to alter the charge injection properties of bottom contact OFETs [28,29]. This strategy has the advantage of tuning the energy band line-up at the contact–semiconductor interface in addition to improving semiconductor morphology. Although not as common, there have also been attempts to improve the properties of top contacts by depositing chemical moieties through a shadow mask on top of the semiconductor, immediately prior to contact metal deposition [30]. Two additional examples of manipulating the contact chemistry to achieve lower contact resistance are noteworthy. In the case of the nonmetal contact material poly(3,4-ethylenedioxythiophene) doped with poly(styrenesulfonate), PEDOT:PSS, one group found that low contact resistances were achieved between PEDOT:PSS and the polymer semiconductor F8T2 in a bottom contact OFET [31]. It was posited that the PEDOT:PSS contacts had doped the semiconductor in the vicinity of the contacts, leading to more efficient charge injection/extraction. A second group of researchers found a clever way to make better top contacts to the single-crystal charge-transfer semiconductor DBTTF-TCNQ than with either gold or silver. They used the related metallic charge-transfer material TTF-TCNQ to form top contacts to the semiconductor with success in realizing more efficient charge injection [32].

8080_C002.fm Page 155 Friday, April 6, 2007 10:21 AM

Contact Effects in Organic Field-Effect Transistors

155

These are only a few examples that show how judicious chemical modifications to the contact–semiconductor interface can often be used to improve contact quality in OFETs.

2.4.5.2 AMBIPOLAR

AND

LIGHT-EMITTING OFETS

While the development of ambipolar (both hole- and electron-transporting) OFETs is still in the early stages, it is certainly an exciting subject within the OFET community. These devices offer not only new possibilities for complementary logic circuit design, but also the potential to control electron–hole recombination within the semiconductor channel to afford light emission. Light-emitting organic fieldeffect transistors (LEOFETs) are particularly intriguing because they possess charge carrier densities in the channel that are orders of magnitude higher than those found in organic light-emitting diodes (OLEDs) [33]. While it is simpler from a fabrication standpoint to deposit the same contact material for both the source and drain contacts (symmetric contacts), one may also consider choosing two different materials for each contact (asymmetric contacts). Based on energy band line-up considerations with the semiconductor HOMO and LUMO levels, depositing two different contact materials at either end of the transistor channel may facilitate more efficient hole and electron injection, respectively. At this point, it is unclear whether separately engineering distinct contacts for hole/electron injection in ambipolar OFETs will prevail over opting for symmetric contacts. However, there will certainly be more reports on this exciting OFET subclass in the next few years.

2.4.5.3 CHANNEL DIMENSIONS: HOW SMALL? At the laboratory scale, OFET channel lengths are typically on the order of 10–100 μm, with W/L ratios ranging from ~10 to over 1,000. Commercialization of organic electronics will lead to a push to make OFET dimensions as small as possible, since the switching speed (cut-off frequency) of an ideal transistor is inversely proportional to L2. As discussed previously, however, one must be careful to avoid becoming contact resistance-limited when shrinking L. For example, from Table 2.4.1, the specific contact resistance for gold top contacts on pentacene can be as low as 1 kΩ·cm. The channel sheet resistance for pentacene devices at high gate voltage (~5 × 1012 carriers/cm2) is about 1 MΩ/sq, assuming a mobility of 1 cm2/Vs. Thus, for a 10-μm channel length pentacene device, the contact resistance is 50% of the total device resistance in the ON state! Clearly, driving the channel lengths smaller will only exacerbate this problem. Thus, ongoing contact engineering efforts to make lower resistance contacts to organic semiconductors will remain important to OFET development and the advancement of organic electronics.

REFERENCES 1. Shen, Y. et al., How to make ohmic contacts to organic semiconductors, Chem. Phys. Chem., 5, 16, 2004.

8080_C002.fm Page 156 Friday, April 6, 2007 10:21 AM

156

Organic Field-Effect Transistors

2. Kahn, A., Koch, N. and Gao, W., Electronic structure and electrical properties of interfaces between metals and π-conjugated molecular films, J. Polym. Sci. B: Polym. Phys., 41, 2529, 2003. 3. Hill, I.G. et al., Organic semiconductor interfaces: Electronic structure and transport properties, Appl. Surf. Sci., 166, 354, 2000. 4. Seki, K. and Ishii, H., Photoemission studies of functional organic materials and their interfaces, J. Electron Spectroscopy Related Phenomena, 88–91, 821, 1998. 5. Salaneck, W.R. and Fahlman, M., Hybrid interfaces of conjugate polymers: Band edge alignment studied by ultraviolet photoelectron spectroscopy, J. Mater. Res., 19, 1917, 2004. 6. Kera, S. et al., Impact of an interface dipole layer on molecular level alignment at an organic-conductor interface studied by ultraviolet photoemission spectroscopy, Phys. Rev. B, 70, 085304, 2004. 7. Necliudov, P.V. et al., Contact resistance extraction in pentacene thin film transistors, Solid-State Electronics, 47, 259, 2002. 8. Amy, F., Chan, C., and Kahn, A., Polarization at the gold/pentacene interface, Org. Elec., 6, 85, 2005. 9. Wan, A. et al., Impact of electrode contamination on the α-NPD/Au hole injection barrier, Org. Elec., 6, 47, 2005. 10. Bürgi, L. et al., Close look at charge carrier injection in polymer field-effect transistors, J. Appl. Phys., 94, 6129, 2003. 11. Pesavento, P.V. et al., Gated four-probe measurements on pentacene thin-film transistors: Contact resistance as a function of gate voltage and temperature, J. Appl. Phys., 96, 7312, 2004. 12. Pesavento, P.V. et al., Film and contact resistance in pentacene thin-film transistors: Dependence on film thickness, electrode geometry, and correlation with hole mobility, J. Appl. Phys., 99, 094504, 2006. 13. Hamadani, B.H., Natelson, D., Nonlinear charge injection in organic fild-effect transistors, J. Appl. Phys., 97, 064508, 2005. 14. Kymissis, I., Dimitrakopolous, C.D., and Purushothaman, S., High-performance bottom electrode organic thin-film transistors, IEEE Trans. Elec. Dev., 48, 1060, 2001. 15. Dimitrakopolous, C.D., Brown, A.R., and Pomp, A., Molecular beam deposited thin films of pentacene for organic field effect transistor applications, J. Appl. Phys., 80, 2501, 1996. 16. Newman, C. R. et al., High mobility top-gated pentacene thin-film transistors, J. Appl. Phys., 98, 084506, 2005. 17. Zaumseil, J., Baldwin, K.W., and Rogers, J.A., Contact resistance in organic transistors that use source and drain electrodes formed by soft contact lamination, J. Appl. Phys., 93, 6117, 2003. 18. Meijer, E.J. et al., Scaling behavior and parasitic series resistance in disordered organic field-effect transistors, Appl. Phys. Lett., 82, 4576, 2003. 19. Puntambekar, K.P., Pesavento, P.V., and Frisbie, C.D., Surface potential profiling and contact resistance measurements on operating pentacene thin-film transistors by Kelvin probe force microscopy, Appl. Phys. Lett., 83, 5539, 2003. 20. Maltezos, G. et al., Tunable organic transistors that use microfluidic source and drain electrodes, Appl. Phys. Lett., 83, 2067, 2003. 21. Chesterfield, R.J. et al., Variable temperature film and contact resistance measurements on operating n-channel organic thin film transistors, J. Appl. Phys., 95, 6396, 2004.

8080_C002.fm Page 157 Friday, April 6, 2007 10:21 AM

Contact Effects in Organic Field-Effect Transistors

157

22. Chesterfield, R.J. et al., Organic thin film transistors based on n-alkyl perylene diimides: Charge transport kinetics as a function of gate voltage and temperature, J. Phys. Chem. B, 108, 19281, 2004. 23. Gundlach, D.J. et al., High mobility n-channel organic thin-film transistors and complementary inverters, J. Appl. Phys., 98, 064502, 2005. 24. Merlo, J.A. et al., p-Channel organic semiconductors based on hybrid acene-thiophene molecules for thin-film transistor applications, J. Am. Chem. Soc., 127, 3997, 2005. 25. Street, R.A. and Salleo, A., Contact effects in polymer transistors, Appl. Phys. Lett., 81, 2887, 2002. 26. Panzer, M.J. and Frisbie, C.D., Unpublished results. 27. Anthopoulos, T.D. et al., Solution processible organic transistors and circuits based on a C70 methanofullerene, J. Appl. Phys., 98, 054503, 2005. 28. Campbell, I.H. et al., Controlling Schottky energy barriers in organic electronic devices using self-assembled monolayers, Phys. Rev. B, 54, 14321, 1996. 29. Gundlach, D.J., Jia, L., and Jackson, T.N., Pentacene TFT with improved linear region characteristics using chemically modified source and drain electrodes, IEEE Elec. Dev. Lett., 22, 571, 2001. 30. Schroeder, R., Majewski, L.A., and Grell, M., Improving organic transistor performance with Schottky contacts, Appl. Phys. Lett., 84, 1004, 2004. 31. Wang, J.Z., Chang, J.F., and Sirringhaus, H., Contact effects of solution-processed polymer electrodes: Limited conductivity and interfacial doping, Appl. Phys. Lett., 87, 083503, 2005. 32. Takahashi, Y. et al., Tuning of electron injections for n-type organic transistor based on charge-transfer compounds, Appl. Phys. Lett., 86, 063504, 2005. 33. Zaumseil, J., Friend, R.H., and Sirringhaus, H., Spatial control of the recombination zone in an ambipolar light-emitting organic transistor, Nat. Mater., 5, 69, 2006.

8080_C002.fm Page 158 Friday, April 6, 2007 10:21 AM

8080_C003.fm Page 159 Friday, April 6, 2007 10:24 AM

3.1

Design, Synthesis, and Transistor Performance of Organic Semiconductors

Abhijit Basu Mallik, Jason Locklin, Stefan C. B. Mannsfeld, Colin Reese, Mark E. Roberts, Michelle L. Senatore, Hong Zi, and Zhenan Bao CONTENTS 3.1.1 Introduction................................................................................................160 3.1.2 p-Channel Organic Semiconductors ..........................................................161 3.1.2.1 Acenes and Non-Thiophene-Based Semiconductor ...................162 3.1.2.1.1 Linear Fused Rings....................................................162 3.1.2.1.2 Fused and Extended Heteroarenes ............................166 3.1.2.1.3 Star-Shaped Oligomers..............................................173 3.1.2.1.4 Oligoaryls ..................................................................173 3.1.2.1.5 Macrocyclics..............................................................175 3.1.2.2 Thiophene-Based Oligomers.......................................................177 3.1.2.3 Polymers ......................................................................................183 3.1.2.4 Solution Processable Semiconductors: The “Precursor Method”.......................................................................................188 3.1.2.4.1 Precusor Polymers and Small Molecules .................189 3.1.2.4.2 Polydiacetylene..........................................................191 3.1.3 N-Channel Organic Semiconductors .........................................................191 3.1.3.1 Fullerenes and Fullerene Derivatives..........................................192 3.1.3.2 Phthalocyanines ...........................................................................192 3.1.3.3 Naphthalene Diimide Derivatives ...............................................194 3.1.3.4 Perylene Diimide Derivatives .....................................................195 3.1.3.5 Quinoid Systems .........................................................................197 3.1.3.6 Thiophene Based n-Channel Oligomers .....................................197 3.1.3.7 Trifluoromethylphenyl-Based Oligomers....................................200 3.1.3.8 Polymeric Systems ......................................................................201 159

8080_C003.fm Page 160 Friday, April 6, 2007 10:24 AM

160

Organic Field-Effect Transistors

3.1.4 Outlook and Conclusions ..........................................................................202 3.1.5 Table of Mobilities.....................................................................................203 References..............................................................................................................214

3.1.1 INTRODUCTION Organic materials have been an integral part of the semiconductor industry since 1960 when Kahng and Atalla demonstrated the first metal-oxide-silicon field-effect transistor.[1] Photosensitive polymers have been the defining factor in achieving feature sizes down to and below 45 nm, while other insulating polymers have been vital for packaging chips or improving the performance of low-k interlayer dielectric materials.[2] However, organic materials had not been considered as the active semiconductor layer until 1987, when Koezuka and coworkers demonstrated a polythiophene-based field-effect transistor.[3] Since this initial breakthrough, remarkable progress has been made in organic field-effect transistors (OFETs), due largely to the design of high-performance active-layer materials. Organic materials are attractive for many components of electronic devices, particularly the active semiconductor layer, due to many fundamental advantages over their inorganic counterparts.[4] Simple, solution-based processing allows for unconventional deposition methods,[5] such as inkjet,[6-10] screen,[11,12] and microcontact printing.[5,13,14] The low temperature required for these methods, combined with the mechanical flexibility of organic materials, offers compatibility with plastic substrates,[5] leading to the possibility of flexible integrated circuits[15-17], electronic paper,[18-21] and roll-up displays.[22,23] Although organic materials are not currently suitable for applications requiring high switching speeds, their low material and fabrication costs make them ideal for large-area applications, such as displays or solar cells.[24-26] Disposable electronics is another emerging class of technology based on these advantages, including chemical sensors[27-31] and radio frequency identification cards (RFIDs).[32] Already, the performance of organic materials, such as pentacene[33] and rubrene,[34] has surpassed that of amorphous silicon, while others have been incorporated in commercially available light-emitting diode displays.[35-37] This impressive growth is evident in Figure 3.1.1, which shows a logarithmic plot depicting the evolution of reported field-effect mobility values for p- and n-channel organic semiconductors over the past 18 years. Arguably the most important feature of organic materials, however, is the ability to impart functionality by intelligent material design[5,38-40] and advanced synthetic techniques. This aspect of materials engineering is a necessary component of the success of organic electronics and the development of molecular devices. The design of materials for organic electronics has proceeded through an enormous synthetic initiative, and has attracted theoretical treatment with the prospect of the rational design of high-performance materials. A variety of approaches have given valuable insight to both the relationship between functionality and electronic properties, as well as between electron transport efficiency and molecular superstructure. While existing studies do not currently offer the prediction of performance a priori, the results presented offer the hope for true materials engineering. However, while materials have been the overwhelming focus of the organic electronics field, the

8080_C003.fm Page 161 Friday, April 6, 2007 10:24 AM

Design, Synthesis, and Transistor Performance of Organic Semiconductors

161

101

Mobility (cm2V1S–1)

100

a–Si:H

10–1 10–2 10–3 10–4 10–5

P-channel Pentacene (V) Rubrene (sc) Other small mols. (V) Small mols. (S) Polymers (S) Air-stable n-channel

1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 Year

FIGURE 3.1.1 Evolution of OFET performance with time for various p-channel (pentacene, rubrene, other small molecules, and polymers) and n-channel organic semiconductors. (v): Vacuum deposition; (s) solution deposition; (sc): single crystal. A range of mobilities for hydrogenated amorphous silicon (a-Si:H) is shown for reference. Please refer to Table 3.1.1 for the respective molecule indices.

fabrication of high-performance devices depends not only on the intrinsic properties of the semiconducting material, but also the technique employed to form the active layer. The deposition method and conditions, in addition to any dielectric or postdeposition treatments, will impact device performance by influencing the molecular organization of the semiconductor molecules.[30,41-43] For this reason, it is crucial to understand the various factors that influence the dynamics of thin-film formation. In particular, the initial growth regime — of particular importance for organic thinfilm transistors (OTFTs) — has been well-characterized, and provides a framework within which the key parameters of ordered film growth may be examined. The rate of publication and development of new materials for organic electronics suggests that the coming years will bring the realization of these unique technologies. In this review, we attempt to summarize comprehensively the vast effort toward the development of high-performance organic semiconductors for OFETs. The body of this paper will focus on the extensive collection of semiconducting organic materials synthesized to-date, compiling structural and performance data as reported in the literature.

3.1.2 p-CHANNEL ORGANIC SEMICONDUCTORS The design and synthesis of p-channel organic semiconductors is an area of fervent chemical research. Hundreds of novel molecules with unique optical and electronic properties have been synthesized, almost exclusively in the past ten years. Since the first report of a transistor employing an organic active layer in 1987,[3,44] the performance of small molecule and polymer semiconductors has been steadily

8080_C003.fm Page 162 Friday, April 6, 2007 10:24 AM

162

Organic Field-Effect Transistors

improved. In this section we survey the OFET performance of these materials, focusing on two parameters that determine largely the potential speed and efficiency of circuits containing them: the field-effect mobility and the on-off current ratio.

3.1.2.1 ACENES

AND

NON-THIOPHENE-BASED SEMICONDUCTOR

3.1.2.1.1 Linear Fused Rings Polycyclic aromatic hydrocarbons are currently among the most widely studied organic π-functional materials. Particular attention has been paid to linear aromatic hydrocarbons composed of laterally fused benzene rings, called linear acenes or oligoacenes. The highest occupied molecular orbital (HOMO) energy level in linearly condensed [n]acenes significantly increases with n, which facilitates the injection of radical cations (holes) at the interface between the source and semiconductor layer under an applied gate voltage. Furthermore their planar shape facilitates crystal packing and enhances the intermolecular overlap of π-systems. Because of these features, pentance and tetracene, both members of the linear oligoacene series, are among the most promising molecular semiconductors for OFETs. However, it is well documented that preparation of such extended benzenoid molecules is plagued by facile photo-oxidation, insolubility and most detrimental, dimerization and polymerization.[45-47] This section summarizes published data on acene molecules emphasizing the influence of the structural properties and the molecular ordering on OFET performance. Anthracene (1a)(Figure 3.1.2) is the smallest member of the acene series with reported transistor characteristics. Single-crystal OFETs have shown a mobility of 0.02 cm2V–1s–1, but only at very low temperatures.[48] However, rather high mobilities have been observed using oligoanthracene and substituted anthracene derivatives. Substitution at the 2,6–positions on the anthracene backbone is expected to give the most extended π-conjugation and also the highest degree of planarity, due to the retention of linearity and limited steric hindrance. OTFTs fabricated from thermally evaporated thin-films of this type of oligomer gave field-effect mobilities in the order 1b< 1d< 1c< 1e (0.01 cm2V–1s–1, 0.07 cm2V–1s–1, 0.13 cm2V–1s–1, and 0.18 cm2V–1s–1 respectively).[49,50] This observation suggests that the addition of alkyl groups (from 1b to 1c and 1d to 1e) is more effective in improving mobility than the extension of π-conjugation (from 1b to 1d and 1c to 1e), which is also observed for oligothiophenes.[51] Acene and thiophene oligomers represent two of the most heavily studied series of organic semiconducting compounds. Two subsequent examples present anthracene-based materials with thiophene groups centralized on the conjugated core and also on the periphery. In the first example, thiophene units flank an anthracene moiety, as shown in molecules 1f and 1g.[53] The mobility of the dihexyl-substituted oligomer, 1g (0.5 cm2V–1s–1), was nearly an order of magnitude higher than that of 1f (0.063 cm2V–1s–1), presumably due to the self-assembly promoted by the alkyl side chains.[54] OTFTs fabricated using either material showed high stability under ambient conditions. In a second example, anthracene moieties were substituted on a bithiophene core to give molecule 1h.[55] OTFTs employing this compound had a mobility of 0.12

8080_C003.fm Page 163 Friday, April 6, 2007 10:24 AM

Design, Synthesis, and Transistor Performance of Organic Semiconductors

Anthracene based (1a–1i)

163

R

n

R (1a)

n = 0, R = H (1b) n = 0, R = C6H13 (1c) n = 1, R = H (1d) n = 1, R = C6H13 (1e)

S S R

R

S

S

(1h) R = H (1f) = C6H13 (1g)

Tetracene based (2a–2f)

(1i) R

(2a)

R = Cl: R′ = H (2b) = Br: R′ = H (2c) = Cl: R′ = Cl (2d) = Br: R′ = Br (2e)

R′ S S (2f)

FIGURE 3.1.2 Chemical structure of anthracene- and tetracene-based oligoacenes.

cm2V–1s–1 with an on/off ratio of 108 and also exhibited a high degree of thermal stability. Among the two unsubstituted oligomers, 1f and 1h, the latter has higher number of acene moieties and longer effective conjugation length, resulting in a higher mobility. However, the hexyl-substituted oligomer 1g, despite of having lower π-conjugation showed a higher mobility than 1h, further illustrating the importance of alkyl side chains in the molecular packing of thin films. Bis(2-acenyl)acetylene (1i) has also been used as the active layer in OTFTs, achieving a mobility as high as 1.1 cm2V–1s–1 with an on/off ratio of 4.4x105.[56] Tetracene (2a) (Figure 3.1.2) is the next larger homologue of anthracene and therefore, has a slightly higher degree of conjugation. OFETs fabricated from single crystals of tetracene show a mobility of up to 1.3 cm2V–1s–1 with an on/off ratio of 106, [57] whereas thin-films yield a mobility of 0.1 cm2V–1s–1.[58] Semiempirical calculations have shown that substitution of bromo or chloro groups on tetracene lowers HOMO and LUMO levels. The halogen groups also promote cofacial πstacking.[59] Halogenated tetracene derivatives, 2b-e, differ in position of substitution and hence provide a useful system to investigate the effect of molecular packing on charge transport.[60] Single crystal X-ray crystallography showed that the number of halogen substitutions is a key parameter and can be used to control molecular packing. The mono-substituted tetracene derivatives 2b and 2c are shown to pack in a herringbone fashion whereas the di-substituted analogues 2d and 2e π-stack in a slipped face-to-face manner. OFETs fabricated using single crystals of these molecules gave a broad range of mobilities, ranging from 1.4 × 10–4 cm2V–1s–1 (2b)

8080_C003.fm Page 164 Friday, April 6, 2007 10:24 AM

164

Organic Field-Effect Transistors

to 1.6 cm2V–1s–1 (2d). The π-stacking structure of 2d, which enhances π-orbital overlap, may be responsible for this high mobility. Similar to the anthracenethiophene oligomer 1h, the tetracene analogue 2f also showed a high mobility (0.5 cm2V–1s–1 with on/off ratio of 108).[55] Pentacene (3a) (Figure 3.1.3) is one of the most widely studied organic semiconductors, with characterization beginning in the 1970s. The highest thin-film fieldeffect mobilities so far have been recorded for pentacene (0.3–0.7 cm2V–1s–1 on SiO2/Si substrates, 1.5 cm2V–1s–1 on chemically modified SiO2/Si substrates, and 3–6 cm2V–1s–1 on polymer gate dielectrics).[57,61-66] Despite the substantial research Pentacene Based (3a-3j) SiiPr 3

SiR 3

X

F (3a)

F R2

R1

X

F

X X

F R2

R1 SiiPr 3

SiR 3

R 1 ,R 2= Me (3b) R= Cl (3c) R 1 = Me; R2 = H (3d) R 1 = C 6H 13 ; R 2= H (3e)

R= Me ( 3f ) R= Et (3g) R= iPr (3h)

X= H (3i) X= F (3j)

Quinones (4a-4c) O

O

CH 3

CH3 (4a)

O CH 3

CH3

(4b)

O

O

O

(4c)

Rubrene derivatives (5a-5b)

( 5b)

(5a )

Polycyclic Aromatic Hydrocarbons (6a-6f) R1

R2

C12H 25O

OC 12 H25

R1-6= H (6a) R1,2,4,5 =H; R 3,6 = C 6H 13 (6b) R3 R1,4=H; R 2,3,5,6 = C 6H 13 (6c) R1-6= C 6H 13 (6d) R1-6= Ph-p-C12 H25 (6e)

R6

R5

R4

(6f)

C12H 25O

OC 12 H25

FIGURE 3.1.3 Chemical structure of pentacene- and rubrene-based oligoacenes.

8080_C003.fm Page 165 Friday, April 6, 2007 10:24 AM

Design, Synthesis, and Transistor Performance of Organic Semiconductors

165

on pentacene devices, relatively few derivatives of pentacene have been synthesized. An obvious advantage of substituted pentacene is improved solubility, which would simplify purification and transistor processing. It has been suggested that the mobility of pentacene could be further enhanced if it is forced to pack in a face-to-face manner rather than the usually observed herringbone structure. One approach to disrupting the herringbone packing is to substitute the four terminal hydrogen atoms. In such a structure, the pentacene core would remain intact while steric interactions would prohibit the herringbone arrangement. Hence, methyl groups (3b) were introduced at the four terminal carbon atoms of pentacene.[67] Thin films of Me4-pentacene (3b) were thermally evaporated and exhibited a mobility of 0.3 cm2V–1s–1 with an on/off ratio of 6.3x103. However, the incorporated methyl groups are not sufficiently large enough to disrupt the herringbone packing and the crystal structure of 3b still resembles that of pentacene. Dimethyl (3d) and dihexyl (3e) pentacene derivatives also exhibited high mobility values of 2.5 cm2V–1s–1 and 0.251 cm2V–1s–1, respectively.[61,68] Tetrachloro pentacene (3c) was also synthesized, but due to its instability, it failed to show any OFET characteristics.[46] Bulky solubilizing trialkylsilyl groups have been substituted at the 6,13-positions of the pentacene molecule (3f-3h)[69,70] in order to impart solubility and disrupt herringbone packing of pentacene core. The substituents were separated from the pentacene core by a rigid alkyne spacer to allow the closest possible contact between the aromatic rings. The functionalized derivatives are very soluble and can be crystallized from common organic solvents. Solid-state packing analysis of triisopropylsilyl (TIPS) pentacene 3h showed that the molecule stacks in a 2D columnar array with significantly larger overlap of the aromatic rings compared to unsubstituted, herringbone-packed pentacene. OTFTs fabricated with TIPS-pentacene (3h) gave a very high mobility of 0.4 cm2V–1s–1 with an on/off ratio of 106, while those fabricated with trimethylsilyl (TMS, 3f) and triethylsilyl (TES, 3g) showed low mobilities of 10–5 cm2V–1s–1 The high field-effect mobility of 3h corroborates its improved thin-film structure compared to 3f and 3g. Fluorine containing derivatives of silylethynylated pentacenes (3i-3j) have also been synthesized[71] in attempt to tune both the electronic properties and thin-film packing of the pentacene derivatives. Fluorine-substituition has been used to lower the HOMO and LUMO energy levels[72-74] while the aryl-fluoroaryl interactions are well-known packing synthons in crystal engineering.[75-77] OTFTs showed pchannel behavior for both fluorine-containing compounds, contrary to expectations based on reports of other fluorinated derivatives found to behave as n-channel semiconductors.[72] Tetrafluoropentacene (3i) showed a mobility of 0.014 cm2V–1s–1, while octafluoropentacene (3j) showed a higher mobility of 0.045 cm2V–1s–1. While the longest commercially available oligoacene, pentacene, also exhibits the highest OTFT performance, very few longer acenes or derivatives thereof have been reported. Recently silylethynylated hexacene and heptacene derivatives were synthesized (Figure 3.1.3), but their transistor characteristics were not reported.[78] The only transistor examples of higher-acenes based semiconductors are molecules containing an electron-rich acene unit and an electron-poor quinoid unit (4a4c).[79] The packing influenced by this structure is such that acene units stack over

8080_C003.fm Page 166 Friday, April 6, 2007 10:24 AM

166

Organic Field-Effect Transistors

quinoid fragments in a self-complimentary manner.[80] The highest mobility was obtained for devices fabricated from thermally evaporated thin films of hexacene quinone (4a). A field-effect mobility of 5 × 10–0.05 cm2V–1s–1 and on/off ratio of 106 were reported. Despite the strong electron-withdrawing nature of quinone, these molecules acted as hole-transporting materials. Rubrene (5a) (Figure 3.1.3), a commercially available aryl-substituted linear acene, is soluble in common solvents such as toluene, xylene, and chloroform, unlike its unsubstituted counterpart, tetracene (2a). A tetra-t-butyl derivative of rubrene (5b) has also been reported. OTFTs fabricated from thin films of 5a and 5b show low mobilities in the range of 10–3 cm2V–1s–1.[81] The mobility of thin films of rubrene, 5a can be further improved up to 0.07 cm2V–1s–1 by insertion of a pentacene buffer layer in between rubrene thin films and sapphire substrate.[82] Recently a new method was introduced to fabricate high-performance field-effect transistors using rubrene single crystals, where mobilities as high as 15.4 cm2V–1s–1 have been observed.[34,57] Furthermore a solution-processed semiconductor composite film of rubrene containing a vitrifying additive and a high-molecular weight polymer gave mobilities of up to 0.7 cm2V–1s–1 with an on/off ratio of greater than 106.[83] Recent research reported ambipolar field-effect transitors based on rubrene single crystals. Hole and electron mobilities of 1.8 and 0.011 cm2V–1s–1 were derived from saturated currents.[84] Disc-shaped aromatic hexabenzocoronene (HBC) derivatives (6a-6d) (Figure 3.1.3) have also been tested as an active layer in OFETs.[85] HBC combines structural elements of linear acenes with a disc-shaped, potentially columnar liquidcrystalline core. The mobilities of dihexyl (6b) and tetrahexyl (6c) derivatives (0.011 cm2V–1s–1 and 0.012 cm2V–1s–1 respectively) were ten-fold higher than those of unsubstituted HBC (6a) and the hexahexyl derivative (6d). This good performance was explained by the 2D self-assembling properties of dihexyl and tetrahexyl HBC derivatives. Recently, a magnetic field has also been utilized to oriented thin films of 6e. Such oriented films gave mobilities of up to 0.001 cm2V–1s–1.[86] A new type of HBC derivative, 6f, with nonplanar aromatic core, imposed from steric congestion of its proximal carbon atoms, has been found to be semiconducting.[87] The single crystal structure of 6f confirmed that the aromatic core is severely distorted and the molecule self-assembles into long columns through stacking. Field-effect transistors fabricated from spin-cast films of 6f gave a mobility of 0.02 cm2V–1s–1 and an on/off ratio of 106. 3.1.2.1.2 Fused and Extended Heteroarenes Heteroarenes are an interesting but much less studied class of π-functional materials. Substitution of one of several carbon atoms in oligoacenes with a heteroatom such as nitrogen or sulphur in different valence state may induce unique properties in these π-electron systems. The generally decreased HOMO-LUMO gap is of particular interest for many applications. Several dihydrodiazapentacenes, (7a-7d) (Figure 3.1.4) have been synthesized with nitrogens replacing carbons in the phenyl rings of pentacene,[88] in order to avoid the facile oxidation of pentacene.[47] OTFTs fabricated by thermal evaporation

8080_C003.fm Page 167 Friday, April 6, 2007 10:24 AM

Design, Synthesis, and Transistor Performance of Organic Semiconductors

Dihydrodiazapentacene (7a-7d)

Imidazolylquinoline (8a-8b)

H N N H (7a)

R= H (7b) R= CH3 (7c)

H N

R

R

N H

R

R

N N N H R= H (8a) (8b)

R=

(7d)

R=

Thioacene (9a-9g)

SiR3 S

R

R S R= R= R= R=

S

H (9a) C6 H13 (9b) C12H 25 (9c) C18H 37 (9d)

S

Bis- and oligo-(benzodithiophene) (10a-10f) S S

SiR3

R= Me (9e) R= Et (9f) R= iPr (9g)

S

(10a)

S C 6H 13

S

S

S

C 6H 13

S (10b)

S

S

S S

C 6H 13

S (10d)

S

S

S

C 6H 13

S (10c)

S

S

S

S S

S (10e)

(10f)

Dithienothiophene Derivatices (11a-11f) S

S S

S

S S

S

(11a)

S

S

S

(11d) S

S

(11b)

S S

S S

S C6 H13

S S

S S

S

S

C6H 13

(11c)

S

(11e)

S S

S S

S

Substituted Fused-Bithiophene (12a-12b) S

S S

167

S (12a)

FIGURE 3.1.4 Chemical structure of thioacene-based oligomers.

(12b)

(11f)

8080_C003.fm Page 168 Friday, April 6, 2007 10:24 AM

168

Organic Field-Effect Transistors

of 7a and 7b gave field-effect mobilities of 5 × 10–5 cm2V–1s–1 and 3–6 × 10–3 cm2V–1s–1, respectively. Two imidazolylquinoline compounds (8a-8b) (Figure 3.1.4) have also been tested in OTFTs [89,90] with thermally evaporated thin films showing mobilities of 0.038 cm2V–1s–1 and 0.148 cm2V–1s–1 for 8a and 8b, respectively. Anthradithiophene (ADT, 9a) (Figure 3.1.4), a fused heterocyclic compound similar to pentacene is expected to have a higher oxidation barrier than pentacene, which would impart increased stability.[91] OTFTs of the soluble dihexyl-substituted dihexylanthradithiophene (DHADT, 9b) were fabricated by solution-casting at different temperatures. The field-effect mobility was found to be strongly dependent on the solvent evaporation temperature. The highest mobilities, 0.01–0.02 cm2V–1s–1 were obtained for a substrate temperature of 100°C. Films of DHADT (9b) obtained by thermal evaporation were highly polycrystalline in nature and gave mobilities as high as 0.15 cm2V–1s–1. Furthermore, compound 9d has a mobility of 0.06 cm2V–1s–1 even though 70% of its molecular volume is comprised of hydrocarbon chains. These alkylated anthradithiophenes combine a pentacene-like high mobility with greater solubility and oxidative stability. As with pentacenes, silylethynylated anthradithiophene derivatives (9e-9g) (Figure 3.1.4) with substituents on the central aromatic ring have also been synthesized.[92] OTFTs fabricated with the trimethyl (9e) and triisopropyl (9g) derivatives showed negligible transistor characteristics while those with triethyl derivative (9f) deposited by solution gave a mobility of 1.0 cm2V–1s–1 with on/off ratio of 107. The excellent performance of 9f was attributed to the enhanced π-orbital overlap observed in its crystalline thin-film structure. A linearly fused heteroarene, Bis(benzodithiophene) (10a) (Figure 3.1.4), in which benzenes are embedded in a fused-ring system, was synthesized in attempt to limit the conformational freedom and thereby reducing interrupted conjugation that is possible in oligothiophenes.[38,93] The benzodithiophene system was further extended to a fully fused-ring compound with dibenzo[b,b′]thieno[2,3-f:5,4f ′]benzothiophene (DBTBT, 10d), thereby completely constraining conformational freedom.[94] OTFTs comprised of thermally evaporated thin films of 10a and 10d gave mobilities of 0.04 cm2V–1s–1 and 0.15 cm2V–1s–1, respectively. The greater mobility of 10d compared to 10a is expected, due to its extended π-conjugation. However, further substitutions of 10a with hexyl (10b) or thiophene-hexyl (10c) units gave lower mobilities on the order of 10–3 cm2V–1s–1.[38] Another two extended fused-ring compounds, isomer-pure thieno[f,f′]bis[1]benzothiophens,syn and anti with respect to the orientation of thiophenes along the long molecular axis of 10e and 10f were reported. The field-effect mobilities of both regioisomers were as high as 0.12 cm2V–1s–1.[95] With the same goal of less conformational freedom and stronger inter-and intramolecular π-overlap, a new fused compound, bis(dithienothiophene) (BDT, 11a)[96-98] and its fused counterpart pentathienoacene (PTA, 11b)[99] have been synthesized (Figure 3.1.4). PTA is attractive because it combines the molecular shape of pentacene with a thiophene monomer. The linear condensed thiophene backbone not only imparts extended π-conjugation, but is also more planar than BDT. OTFTs of 11a yielded high mobilities of 0.05 cm2V–1s–1 with a high on/off ratio of 108,

8080_C003.fm Page 169 Friday, April 6, 2007 10:24 AM

Design, Synthesis, and Transistor Performance of Organic Semiconductors

169

whereas for a PTA (11b) based OFET, a mobility of 0.045 cm2V–1s–1 and an on/off ratio of 103 were achieved. Oligothiophene molecules with a central fused thiophene dithienothiophene (11c) demonstrate reversible oxidation at low potentials with similar HOMO levels to the equivalent dihexyl-sexithiophene (28i) (Figure 3.1.10). Thermally evaporated films on silicon oxide dielectrics showed a field-effect mobility as high as 0.02 cm2V–1s–1 with an on/off ratio greater than 106.[100] Recently, a series of new organic semiconductors (11d-f) were reported, using dithienothiophene as the core. The compounds exhibit excellent field-effect performance with a high mobility of 0.42 cm2V–1s–1 and an on/off ratio of 5 × 106.[101] A structural combination of fused bithiophene with fluorene (BFTT, 12a) or biphenyl units (BPTT, 12b) has been fabricated (Figure 3.1.4).[102] Thin films of BFTT (12a) and BPTT (12b) showed mobilities of 0.06 cm2V–1s–1 and 0.093 cm2V–1s–1, respectively. The slightly higher mobility of BPTT (12b) may be due to its more planar and straight conformation, which leads to better thin-film packing. A series of novel fused heteroarenes, naphtha[1,8-bc:5,4-b′c′]dithiophene (NDT) and its derivatives (13a-13f) (Figure 3.1.5) have been synthesized and tested in OTFTs.[103,104] These fused heteroarenes are formally isoelectronic with pyrene, but their aromatic character is greatly reduced by the loss of benzene rings in the skeleton. As a result, the HOMO energy levels of these π-systems are raised, whereas the LUMO energy levels are lowered. OFETs fabricated from thermally evaporated thin films of these materials showed mobilities from 10–4–0.1 cm2V–1s–1. Although the incorporation of alkyl chains tends to improve the mobility of semiconductor due to better thin-film formation,[105] NDT with (13a) and without (13b) alkyl chains showed a low mobility of 10–4 cm2V–1s–1. However 13c, the more conjugated version of 13a, gave a higher mobility of 6 × 10–3 cm2V–1s–1. Replacing the thiophene groups with phenyl units led to improved performance. The naphthyl derivative 13f showed the highest mobility of 0.11 cm2V–1s–1 with an on/off ratio of 105. Novel butterfly-shaped pyrene derivatives have also been synthesized and tested in OTFTs. The pyrene derivative 13g does not show any field-effect behavior, while 13h exhibited a mobility of 3.7 × 10–3 cm2V–1s–1 with an on/off ratio of 104.[106] Amorphous thin films of spiro-linked compounds (14a-d) (Figure 3.1.5) have been tested in OTFTs.[107,108] The characteristic structural feature of these materials is the linkage of two hole transporting units by a spiro junction. A mobility of 7 × 10–5 cm2V–1s–1 was obtained with an on/off ratio of 106. A new class of semiconductors based on a tertiary diamine structure, 5,11-disubstituted indolo[3,2-b]carbazole (15a-15d) (Figure 3.1.5) have been reported.[109,110] The derivatives are soluble in organic solvents such as toluene, chloroform, and chlorobenzene. OTFTs using these carbazole derivatives (15a-15d) exhibited p-channel behavior with mobilities of up to 0.12 cm2V–1s–1 and a current on/off ratio of 107. Recently dichloro derivatives (15e-15f) of these carbazoles gave improved device performance with mobility of 0.14 cm2V–1s–1.[111] 2,7-Carbazolevinylene-based conjugated oligomers, 15h-15j, have also been tested in OTFTs.[112,113] The higher mobility of 15h (0.30 cm2V–1s–1 with on/off ratio of 107) compared to 15j (10–6 cm2V–1s–1 with on/off ratio of 103) is attributed to the relative coplanar structure of the former and twisted biphenyl units in latter.

8080_C003.fm Page 170 Friday, April 6, 2007 10:24 AM

170

Organic Field-Effect Transistors Naphtha[1,8-bc:5,4-b'c']dithiophene and pyrene derivatives (13a-13h) S R

F3C

R

CF3

S

S

R=

S S

S (13a)

C6 H13

S

S

(13b)

(13c)

S F3C

R= (13d)

(13e)

S

CF3

(13g)

(13h)

(13f)

Spiro-linked compounds (14a-14d) Ar

Ar N

N

N

N Ar

Ar Ar=

Ar

Ar N

N

N

N Ar

Ar

(14a)

Ar=

(14b) (14c)

(14d)

Carbazole derivatives (15a-15j) R N

C12H 25

N

CH 3

R1 N

R2

N R= octyl (15a) R R= dodecyl (15b) R= 4-octylphenyl (15c) R= 4-methylphenyl (15d)

C8 H17

R2

N R1

C 12H 25

H3 C

R 1 = Cl, R 2= H (15e) R 1 = H, R 2= Cl (15f)

N C8H 17

(15g)

R2 N R1

R1 N R2

Ph

Ph

R1 = H, R 2= C6H 13 (15h) R1 = C 6H 13, R 2= CH3 (15i)

N C 6H 13

(15j)

Thiazolothiazole and Thiazole derivatices (16a-16q) N

S

S

N

Ar

Ar S

Ar=

O (16f)

(16a) S

S

(16b)

S S S O N

S

(16g)

N

S

O

(16c)

C6 H 13

(16h) O N

S

(16d)

(16i) S

N (16e)

(16j)

OMe

S (16k) S

C6 H13

N

S S

S

S

N

N

S S

S

S S R= H (16o) = C6 H13 (16p)

S

S

S

N

R= H (16l) = C 6H 13 (16m)

R

N

S

R

R

(16n)

C6H13 S S

R

FIGURE 3.1.5 Chemical structure of heteroarene oligomers.

N S (16q)

S S

8080_C003.fm Page 171 Friday, April 6, 2007 10:24 AM

Design, Synthesis, and Transistor Performance of Organic Semiconductors

171

Thiazole rings in thiophene oligomers are known to reduce the steric interaction due to the absence of one hydrogen atom.[114] In addition, thiazole is an electronwithdrawing heterocycle and hence, imparts stability to oxygen.[115] Thiazole oligomers also form π-stacked structures.[116] Thiazolothiazole is an analogue of thiazole and its co-oligomers with thiophene (16a-16c), furan (16d, 16f, 16h), thiazole units (16e, 16g, 16i), phenyl (16j) and benzothiophene (16k) have been tested in OTFTs (Figure 3.1.5).[117-119] These oligomers are donor-acceptor compounds (donor: thiophene; acceptor: thiazolothiazole) with lowered energy gaps. The unsubstituted derivative (16a) failed to show any field-effect, whereas thiophene(16b) and hexyl-thiophene (16c) derivatives showed mobilities of 0.02 cm2V–1s–1 and 0.003 cm2V–1s–1, respectively, with on/off ratios of 104. OTFTs of 16d gave mobilities up to 2 × 10–3 cm2V–1s–1 with an on/off ratio of 104, however 16f failed to show any field-effect mobility. The thiazole (16e, 16g, 16i), phenyl (16j) and benzothiophene (16k) derivatives did not show any field-effect mobility, either. The lower mobility of furan substituted molecule (16d) compared to its thiophene counterpart (16b) was attributed to its smaller size and decreased polarizability, which increases columbic repulsion and decreases intermolecular interactions. Unsubstituted (16o) and hexyl-substituted (16p) bisthiazole-thiophene co-oligomers were also reported, but had rather low mobilities of 1 × 10–5 cm2V–1s–1 and 2 × 10–5 cm2V–1s–1, respectively.[120] With increasing effective conjugation length by incorporation of additional thiophene rings, as in 16l and 16m, mobilities of 0.011 cm2V–1s–1 and 3.5 × 10–4 cm2V–1s–1, respectively, were obtained.[120] The lower performance of thiazole-containing oligomers compared to the corresponding oligothiophenes was attributed to a larger charge injection barrier and less favorable thin-film morphologies. Tetrathiafulvalene (TTF) and its derivatives (Figure 3.1.6) have been intensively investigated as building blocks for charge-transfer salts, producing a multitude of organic conductors and superconductors.[45,121-124] It was recently shown that TTF derivatives can also be used in OFETs. Single-crystal transistors of dithiophenetetrathiafulavalene (DT-TTF, 17c) were prepared by drop casting of a warm, saturated chlorobenzene solution. The maximum mobility observed in these crystals was 1.4 cm2V–1s–1.[125-128] Similarly, the mobility of single crystal (thiophene)(thiomethylene)-tetrathiafulvalene (TTDM-TTF, 17a) was found to be 0.4 cm2V–1s–1. TTF derivatives with fused aromatic rings (benzo-17h, naphtha-17j, pyrazino-17i and quinoxalino-17k) have also been synthesized, with mobilities of 0.06 cm2V–1s–1 and 0.2 cm2V–1s–1 obtained for single crystals of 17h and 17k.[129,130] The high performance of TTF derivatives suggests that bis(1,3-dithiol-2-ylidene) compounds (Figure 3.1.6), π-extended TTF analogues with a conjugated spacer group, may also be attractive candidates for OFETs. Thin films of Bis[1,2,5]thiadiazolo-p-quinobis(1,3-dithole) (BTQBT, 18a)[131-133] show a mobility of 0.2 cm2V–1s–1 with an on/off ratio of 108. BTQBT can be modified by introducing substituents on 1,3-dithiole ring and replacement of fused dithiadiazole with different π-spacers (18b-18j).[134] OTFTs fabricated from these molecules showed zero to negligible mobility (10–7–10–4 cm2V–1s–1). The replacement of sulphur in heterocycles such as thiophene has been attempted with chalcogens from the same group, such as selenium and tellurium.

8080_C003.fm Page 172 Friday, April 6, 2007 10:24 AM

172

Organic Field-Effect Transistors TTF derivatives (17a-17k) X2 X1

S

S

S

S S

S

S S

X1 = S, X2= H (17a) X1 = H, X2= S (17b) S

S

S

S

S

S (17g)

X2 X1

S

S S

X

S

S

S

S

S

X

S

S S X= C (17h) = N (17i)

X

S

X1 = H, X2 = S (17d) X1 = S, X2 = H (17e)

DT-TTF (17c) S

S

X

S

S

S

S

S

S (17f)

S

X

S

S

X

X

S

S

X

X= C (17j) = N (17k)

Bis(1,3-dithiol-2-ylidene) compounds with conjugated spacer (18a-18j) S

R

S

R R

N

N

S

S

S N

S

S

S

N N S

S

S S

N

S

R S

X

N Se

N S

S

S

S

R R R R= H (18e) = Benzo (18f)

R= H (18c) = Benzo (18d) BenzoBTQBT (18b)

N

X

S

S R

S

S

N S

S N

N S

BTQBT (18a)

R

S

N

N

S

R

S

S R

R= H (18i) S = Benzo (18j) R

X= C (18g) = N (18h)

Chalcogenophenes (19a-19f) Se

X

Se Se

Se

X X= S (19b) X= Se (19c) X= Te (19d)

(19a)

Se R

R Se

R=H (19e) =Ph(19f)

FIGURE 3.1.6 Tetrathiafulvalene and chalcogenophene derivatives.

Because charge transfer is dependent on intermolecular orbital overlap, the addition of atoms such as tellurium or selenium may have a positive impact on the transistor performance. Despite this possibility, there are very few examples of this approach in the literature. OTFTs with quaterselenophene (19a) as an active layer have been reported, demonstrating a mobility of 3.6 × 10–3 cm2V–1s–1.[135] A new class of chalcogen-containing OFET materials, 2,6-diphenylbenzo[1,2-b:4,5-b′]dichalgenophenes (19b-19d) have also been reported.[136] All three chalcogen compounds perform as good hole-transporting materials. In particular, the benzodiselenophene derivative (19c) demonstrates a relatively high OFET performance with a mobility of 0.17 cm2V–1s–1 and an on/off ratio greater than 105. Recently, single crystal FETs of 19c were fabricated and a mobility of 1.5 cm2V–1s–1 was observed.[137] [1]Benzoselenopheno[3,2-b][1]benzoselenophene (19e) and its 2,7diphenyl derivative (19f) were synthesized and tested in OTFTs. Derivative 19f exhibited excellent p-channel field-effect properties with hole mobilities as high as 0.3 cm2V–1s–1 and current on/off ratios of ~106.[138]

8080_C003.fm Page 173 Friday, April 6, 2007 10:24 AM

Design, Synthesis, and Transistor Performance of Organic Semiconductors

173

3.1.2.1.3 Star-Shaped Oligomers Although the major emphasis so far has been on linear-structured organic semiconductors, there are a few of examples of star-shaped tritopic molecules and their use as the active material in OFETs. Compared to their linear counterparts, these materials show somewhat lower mobilities, but they possess better solubility and improved film-forming properties. Aromatic amines are frequently used as xerographic materials and incorporated in OLEDs,[139] while reports of their use as the semiconductor material in OFETs are few. Only very recently triarylamine polymers have been used in OFETs.[140,141] Star-shaped molecules with triphenylamine core and carbazole (20a, 20c-20d) or fluorene (20b) side groups have all been characterized in OFETs (Figure 3.1.7).[142] Devices containing these materials were fabricated using a drop casting technique and showed mobilities on the order of 10–4 cm2V–1s–1 with good stability under ambient conditions. The star-shaped oligothiophene, 21a gave a mobility of 2 × 10–4 cm2V–1s–1 with an on/off ratio of 102 (Figure 3.1.7).[143] A series of star-shaped oligothiophenefunctionalized truxene derivatives (21b-21d) have also been tested in OFETs.[144,145] The large truxene core is expected to extend the π-delocalized system compared to star-shaped polythiophene (21a). Thin films were obtained by spin coating from chloroform solutions, yielding mobilities as high as 1.03 × 10–3 cm2V–1s–1 for 21b with an on/off ratio of 103. Recently, two novel ‘hybrid’ systems (20e-f) consisting of a triphenylamine core carrying π-conjugated bithienyl branches have been synthesized, with 20e displaying a hole mobility of 0.011 cm2V–1s–1.[146] 3.1.2.1.4 Oligoaryls p-Quaterphenyl (p-4P, 22a), p-quinquephenyl (p-5P, 22b), and p-sexiphenyl (p-6P, 22c) are examples of other commercially available semiconductor materials (Figure 3.1.8). OTFTs fabricated from thermally evaporated films of these oligophenyls gave field-effect mobilities ranging from 0.01 cm2V–1s–1 for p-4P (22a) to 0.07 cm2V–1s–1 for p-6P (22c) with on/off ratios from 105 to 106.[147] Rod-like oligo(arylacetylene)s (23a-e) have also been studied for OTFT applications.[148-149] The arylacetylene oligomers contain electron-donating para-substituted trimethyl amine units and differ by the alkyl substituents on the silyl group. Thin films of 23a and 23b showed mobilities of 0.3 cm2V–1s–1 and 4.3 × 10–4 cm2V–1s–1, respectively.[148] The superior device performance of 23a was attributed to its better structural ordering, and hence more effective π-orbital interaction and charge transport, due to the less bulky trimethylsilyl group. Incorporation of acetylene groups was achieved in an oligothiophene molecule using microwave assisted reactions to produce 23c.[150] The thermally evaporated thin films were highly crystalline with large grains, similar to the unsubstituted pentathiophene (27b). However, the electronic performance was shown to be far inferior, with a maximum mobility of 8 × 10–4 cm2V–1s–1 at a substrate deposition temperature of 140°C. Alkylated oligo(arylacetylene)s 23d and 23e showed ambipolar charge transport with hole mobilties reaching up to 0.02 cm2V–1s–1 and maxium electron mobilities around 2 × 10–3 cm2V–1s–1.[151,149]

8080_C003.fm Page 174 Friday, April 6, 2007 10:24 AM

174

Organic Field-Effect Transistors

Triarylamines (20a-20f) R N (20a)

R=

R=

(20c)

N

N (20b) N R

R N

S

S

R

R N

n-C6 H13

(20e) S R=

R= S

(20d)

R N

O

O

N

N

n-C6 H 13

S (20f)

S

R

S R

N R

Oligothiophenes (21a-21d) H n

C10H 21

S

S

S C6H 13

C 6H 13 S H n

C 6H13

C6 H13

C 6H 13

C6 H13

n= 1 (21b) = 2 (21c) = 3 (21d)

S

S H n

S S

S (21a)

C 10 H21

S

S S C 10 H 21

FIGURE 3.1.7 Star-shaped aryl amines and thiophenes.

A modified oligo(p-phenylenevinylene) derivative, 1,4-bis[4-(4-octylphenyl)styrl]-benzene(24a) has been reported[152], showing a field-effect mobility of 0.12 cm2V–1s–1 with an on/off ratio of 106 (Figure 3.1.8). Recently, OTFTs have also been fabricated using thermally evaporated films of two oligo-p-phenylenevinylenes, 1,4-bis(4-methylstyryl)benzene (24b) and 1,4-bis(2-methylstyryl)benzene (24c). Compared with the performance of 24c, 24b showed a higher mobility of 0.13 cm2V–1s–1, attributed to better film continuity and molecular linearity compared to 24c.[153] Vinylene groups have also been incorporated in the thiophene backbone

8080_C003.fm Page 175 Friday, April 6, 2007 10:24 AM

Design, Synthesis, and Transistor Performance of Organic Semiconductors

175

Oligo(arylacetylene) (23a-23e)

Oligo(aryl) (22a-22c)

Me3N

SiR 3

n n= 2 (4P) (22a) n= 3 (5P) (22b) n= 4 (6P) (22c)

R= CH 3 (23a) R= iPr (23b) R

S

S

S

S

S

R= H (23c) =n-C 6H 13 (23d) =n-C 10 H21 (23e)

R

(23c)

Oligo(arylvinylene) (24a-24g) C 8H 17 C 8H 17 (24a)

(24b)

S S

(24c)

R2

R2

S S (24d)

S

R1

R2 S

S R2

R2 S

S R2

R2

R1 R2

R 1, R2 = H (24e) R 1= C6 H13, R2= H (24f) R 1= H, R2= C 6H 13 (24g)

FIGURE 3.1.8 Oligoaryls used in thin–film transistor devices.

to give a series of unsubstituted (24d-24e) and substituted (24f-24g) oligothienylenevinylenes.[154,155] It has been shown that these oligomers have the largest effective conjugation and smallest band gap among all classes of π-conjugated oligomers with comparable chain extension.[156] OTFTs fabricated with thin films of unsubstituted thiophene-vinylene oligomers, 24d and 24e, showed field-effect mobilities of 0.012 cm2V–1s–1 and 0.0014 cm2V–1s–1, respectively. In order to study the effect of the substitution pattern of oligothienylenevinylene core on TFT characteristics, oligomers with alkyl groups either at the terminal (24f) or central thiophene rings (24g) were synthesized. Thin films derived from 24g exhibited very low mobilities, in the range of 1.1 × 10–6 cm2V–1s–1 while compound 24f bearing hexyl chains at the end π-positions showed mobilities 0.055 cm2V–1s–1. The low mobility for 24g compared to 24f is attributed to the steric effects of the hexyl substitutents at the 3and 4-positions of the thiophene rings, which maintain intermolecular separation of the conjugated units. 3.1.2.1.5 Macrocyclics Phthalocyanines (Pc) are another of the first reported families of organic semiconductors (Figure 3.1.9).[157,158] Their structures consists of a molecular cage, into which various metals can be introduced. Highly ordered, thermally evaporated thin films of copper phthalocyanine (CuPc, 25a) are p-channel semiconductors, with a

8080_C003.fm Page 176 Friday, April 6, 2007 10:24 AM

176

Organic Field-Effect Transistors

Phthalocyanines and Porphyrins (25a-25r)

t

Bu R

N N

M= = = = = = =

N M

N

N N

N N

N

N N

N N

Cu (25a) Sn (25b) H 2 (25c) Zn (25d) Fe (25e) Pt (25f) Ni (25g)

N N

N N

t Bu C8H 17 O O C8H 17 N N N N C8H 1 7O N N N N C8H17O

N

M

C8H17 O O N

C8 H1 7O

N N N

N N N

N O

O

OC8 H1 7 OC 8H 17

O O

C8H17 O C8H17

O O

M= Tb (25k) = Lu (25l)

O

O O O

O

O

N NN N N N N

N N N N N N

O O O

O O M= Eu (25m) = Ho (25n) = Lu (25o)

O O O O O

M

O O O O

N

OC8H17 OC 8H17

C8H 17 O C OH 8 17

M

O O O

N O O

N

R= O(CH2) 10OH; M= Cu (25h) = O(CH2)10OH; M= none (25i) = NH 2; M= none (25j)

N

t Bu

C8 H1 7 C8 H1 7O

N M

O

O O

O O

N O O O O O

N N

Pt N

N

NH N

(25p)

N

HN

NH

HN H2SO 4

HN

NH

HN NH

(25q)

N (25r)

Other Coordination Compounds (26a-26c) H N (26a)

2+

H N

RH 2N

Ni N H

Pt

RH2N

N H

Cl

NH 2R Pt

Cl (26b)

S

S Ni

S

S

RH 2N RH2 N Cl Cl

FIGURE 3.1.9 Macrocyclic phthalocyanines and porphyrins.

NH 2R

Cl

2-

Cl Pt

NH2R

2+

NH2R Pt

Cl Cl

2-

R= (26c)

8080_C003.fm Page 177 Friday, April 6, 2007 10:24 AM

Design, Synthesis, and Transistor Performance of Organic Semiconductors

177

mobility strongly dependent on the thin-film morphology. The best OTFT performance obtained was a mobility of 0.02 cm2V–1s–1, with an on/off ratio of 105. [159] A single–crystal CuPc-based OFET was recently reported, with mobility as high as 1.0 cm2V–1s–1.[160] In a subsequent study, different metallophthalocyanines (25b25g) were tested in OFETs.[11] The nature of metal ions and substrate temperature were found to be especially crucial in achieving high field-effect mobilities. Besides Cu, Zn-Pc (25d) gave the highest mobility of 2.8 × 10–3 cm2V–1s–1 at a 200°C substrate temperature. Recently it was observed that the charge mobility of phthalocyanines can be increased up to 0.11 cm2V–1s–1 for OFETs having source/drain electrodes sandwiched between two layers of CuPc and CoPc.[161] Most devices based on Pc are fabricated by vacuum deposition, i.e. thermal evaporation. However, recent solution-based Langmuir-Blodgett (LB) techniques have also been used. LB-films of 25h-25l gave poor (10–6 cm2V–1s–1 for 25j) to moderate (1.7 × 10–3 cm2V–1s–1 for 25l) mobilities (Figure 3.1.9).[162-164] More recently, high mobilities of 0.24–0.60 cm2V–1s–1 have been observed for LB films of amphiphilic tris(phthalocyaninato) rare earth triple-decker complexes, 25m25o.[165] OFET characteristics have also been reported for porphyrin derivatives, 25p-q with mobilities of 2.2 × 10–4 cm2V–1s–1 and 0.012 cm2V–1s–1, respectively.[166-168] LB-prepared thin films of Cyclo[8]pyrrole (25r), an extended porphyrin-like molecule, also exhibited a high mobility of 0.68 cm2V–1s–1 , but with a low on/off ratio.[169] The coordination compounds bis(4-methyl-1,2-phenylenediamino) nickel (26a) and bis(dithiobenzyl) nickel (26b) have also been used in OFETs. The mobilities reach up to 0.013 cm2V–1s–1.[170] A platinum based polymer with poly-Pt chain structure (26c) was synthesized and fabricated as FETs under ambient conditions[171]. Mobilities of 10–3–10–4 cm2V–1s–1 were obtained with 10–103 on-off ratios. Interestingly, immersion of FETs in water at 90°C for 12 hours did not deteriorate important device characteristics.

3.1.2.2 THIOPHENE-BASED OLIGOMERS For many years oligothiophenes and their alkyl-substituents have been among the most intensely investigated organic semiconductors because of the synthetic versatility of the thiophene heterocycle.[172] The first printed organic transistor was fabricated by Garnier et al. using sexithiophene as the active semiconductor layer.[12,173,174] Recently, they have also been used to demonstrate fast integrated circuits.[16] Because of their ease in functionalization, oligothiophenes provide the opportunity to study systematic variations in molecular structure by controlling the number of repeat units in the conjugated backbone and/or varying the length and functionality of alkyl substituents. One of the most widely studied oligothiophenes is α-6T (27c) (Figure 3.1.10); since the first report, its hole mobility has improved from 10–4 cm2V–1s–1 to greater than 0.07 cm2V–1s–1. [93,105,175] It has also been shown that the orientation and morphology of thermally evaporated thin films of α-6T depend strongly on the substrate temperature.[54] Increasing substrate temperature from –204°C to 280°C improves the thin-film morphology, with mobility increasing from 0.006 cm2V–1s–1

8080_C003.fm Page 178 Friday, April 6, 2007 10:24 AM

178

Organic Field-Effect Transistors

Unsubstituted Thiophenes (27a-27d) S

S

S

S

S

S

S

S

4T (27a) S

S

S

S

S

S

5T (27b) S

S

S

S

S

S

S

6T (27c)

S

S

8T (27d)

Alkyl-substituted Thiophenes (28a-28x) S

R

S

S

S

R

R

S

R= C 6H 13 (28a) = C 10H 21 (28b) = cyclohexyl (28c) =C 12 H 25 (28d)

S

S

S

R

S

S

C6 H13

S (28e)

S

R

S

S

S

S

S

S

S

S

S

S

S

S

R= CH3 (28f) = C 10H 21 (28g) = C 6 H13 (28h) R

C6 H 13

S

S

R

R= C 2H 5 (28i) = C 6 H13 (28j) = C 10H 21 (28k) = C 12H 25 (28l) = C 18H 37 (28m)

R

R = C6 H13 (28n) R

SC 4H 9

R

ButMe2 Si

S

S S

S

S

S

S

n

=3 =4

S C 4H 9 S

(28s)

n= 2 R=C5 H 11 (28o) C12 H25

R= C 5 H11 (28p) R= C 6 H13 (28q)

S

S

S

C 6H 13

S

S

S

S

S

S

S

S

S

S

S S

C 12H 25

C 6H 13

SiMe2Bu t

C 12H 25

S

R=C6 H13 (28r)

S

n n= 4 (28t) = 5 (28u) = 6 (28v)

S

(28w)

S S C 12 H25

C6 H13

C6 H13

(28x)

FIGURE 3.1.10 Unsubstituted and alkyl-substituted oligothiophenes.

to 0.025 cm2V–1s–1. Single crystals of α-6T give mobilities as high as 0.1 cm2V–1s–1.[176] Carrier mobilities near 0.2 cm2V–1s–1 have been reported for αoctithiophene (8T, 27d) OFETs with films deposited at a substrate temperatures of 150°C and higher.[105] With oligothiophenes, as well as all other conjugated oligomers, an increase in conjugation length brings about a decrease in solubility. This characteristic makes purification and deposition more difficult using traditional wet techniques, such as column chromatography and spin-coating. One way to increase the solubility of oligothiophenes in organic solvents is to introduce alkyl substituents at the β-position

8080_C003.fm Page 179 Friday, April 6, 2007 10:24 AM

Design, Synthesis, and Transistor Performance of Organic Semiconductors

179

of the thiophene ring. Furthermore, it has been suggested that an introduction of end-substituents at the 2-position of thiophene oligomers provide stability against oxidation and polymerization while minimizing steric interactions that can prohibit co-planarity of the thiophene rings. Properly designed end-substituents may also improve crystallinity, encourage two-dimensional growth, and promote dense intermolecular packing within each layer, thereby improving charge transport. Molecular engineering of thiophene oligomers by alkyl substitution results in a remarkable improvement of the structural organization of molecules at the mesoscopic level thereby creating highly ordered thin films, resulting in enhanced mobilities.[54] Evidence of strong correlation between orientation and electrical transport properties has been given by the comparison between films of unsubstituted linear oligothiophenes (27a-27d) and oligothiophenes substituted by linear alkyl groups at terminal α-positions (28a-28b, 28d-28m) (Figure 3.1.10).[54], [177] The self-organization induced by the alkyl groups results in a preferred orientation of the substituted oligomers, with their long molecular axes roughly perpendicular to the substrate plane, while the unsubstituted molecules 27a-27d may align either parallel or perpendicular. The change in orientation induced by the alkyl chains leads to an order of magnitude increase in the field-effect mobility to 0.13 cm2V–1s–1 for 28i,[54,178] from 0.01 cm2V–1s–1 for the unsubstituted 6T (27c)[175,179,180]. For the substituted oligomers, the positions of the alkyl chains on the conjugated backbone are also important.[54] In the case of the β-substituted 28s, very low mobilities of ~10–7 cm2V–1s–1 were observed for thin films that exhibited poor structural order. Furthermore, a systematic study was performed where both the conjugated units (from four to six thiophene units) in the backbone and length of the side chains (zero, two, six and ten alkyl units) were varied.[105] It was found that OTFT performance depends critically on the length of the side chains, but is relatively insensitive to the length of the conjugated backbone. This has previously been observed for unsubstituted oligothiophenes ranging from 4 to 8 thiophene units (27a-27d).[181] For side chain longer than hexyl, reduced mobility values were observed.[105] In addition to linear alkyl chains, cyclohexyl groups have also been attached as side substituents (Figure 3.1.10).[182] Although slightly bulkier in nature, the mobility for thermally evaporated thin films of dicyclohexyl-quarterthiophene, 28c (0.038 cm2V–1s–1)[182] was found to be higher than that of the dihexyl-substituted counterpart, 28a (0.02 cm2V–1s–1) with the same device dimensions.[183] Owing to the good solubility of 28c, solution deposition was also performed by drop casting, and mobilities as high as 0.06 cm2V–1s–1 were observed. Bis-silylated versions of oligothiophenes (28t-28v) have also been tested in OFETs, exhibiting mobilities on the order of 10–5 cm2V–1s–1.[184] Recently, a novel structure of , α′dihexylpentathiophene-based swivel cruciform (28x) was synthesized and tested in an OFET. It exhibited a field-effect mobility of 0.012 cm2V–1s–1 and on/off ratio greater than 105. [185] The incorporation of ether linkages into oligothiophene side-chains has also been investigated.[186] The solubility of these oligomers increased by a factor of two compared to the corresponding alkyl substituted oligomers of the same conjugation length. The hole mobility of quarterthiophene derivative 29a was 0.003

8080_C003.fm Page 180 Friday, April 6, 2007 10:24 AM

180

Organic Field-Effect Transistors

cm2V–1s–1 while that of the sexithiophene derivatives 29b and 29c were 0.033 cm2V–1s–1 and 0.009 cm2V–1s–1, respectively (Figure 3.1.11). Although the solubility of the new α-6Ts was much improved, the incorporation of an oxygen atom appears to diminish the mobility in solution-cast films relative to alkyl substituted thiophene oligomer films. Other polar species, such as phosphonate groups, have also been incorporated as end-substituents (DBP-α-6T, 29d) (Figure 3.1.11) and used in OTFTs with films fabricated from spin-casting demonstating a mobility of 4.9 × 10–3 cm2V–1s–1 and an on/off ratio of 104.[187] OTFTs have also been fabricated with thin films of oligoethyleneoxide functionalized sexithiophene (29e). Mobilities on the order of 10–4 cm2V–1s–1 were observed.[188] Sexithiophenes bearing amide or ester groups (29f-l) were also synthesized and tested in OTFTs. The oligomer bearing the ester functional group separated from the sexithiophene core by an ethylene spacer showed a hole field-effect mobility as high as 0.012 cm2V–1s–1. [189] A series of fluorene-thiophene oligomers with an identical core but having different end functionalities ranging from unsubstituted (30e), n-hexyl (30b) to cyclohexyl (30f) groups has been investigated (Figure 3.1.11).[182,190] Cyclic voltammetry showed that these derivatives have a lower HOMO energy than the corresponding oligothiophene of the same length (–4.95 eV for DH6T, 28i and –5.36 eV for DHFTTF, 30b). Thermally evaporated, thin films of the unsubstituted (30e), n-hexyl (30b) and cyclohexyl (30f) derivatives gave mobilities of 0.08 cm2V–1s–1, 0.14 cm2V–1s–1 and 0.17 cm2V–1s–1 respectively. The higher mobility of the fluorene-thiophene derivatives (30a-f) is a result of molecular alignment perpendicular to the substrate, in contrast to the equivalent fluorenone-substituted oligothiophene (30g), which was shown to grow parallel with the substrate, resulting in a significantly reduced mobility of less than 10–7 cm2V–1s–1.[191] The crystallinity and growth mode of the oligothiophene-fluorene substituted thin-films was also highly dependent on the substrate deposition temperature and end-group substitution. Another series of fluorene-thiophene oligomers with fluorene as the core (30i-q) have been synthesized and tested in OTFTs. A room-temperature mobility of up to 1.0 × 10–3 cm2V–1s–1 was obtained for 30l.[192] In an effort to decrease reactive degradation of the fluorene, 5,5′-bis(9,9′-dialkylfluorene-2-yl)2-2′-bithiophenes (30q-s) were synthesized. The field-effect mobilities range from 10–5 cm2V–1s–1 for an amorphous film of 30s to 3 × 10–3 cm2V–1s–1 for a polycrystalline film of 30r. The high mobility remains constant after 3 months at ambient conditions, which demonstrates the high environmental stability of this class of materials.[193] Thiophene-phenyl cooligomers with various end substitutions have been synthesized by several groups (Figure 3.1.11). Unsubstituted phenylene capped oligothiophenes showed mobilities of 1.4 × 10–3 cm2V–1s–1 (31c), 3 × 10–3 cm2V–1s–1 (31h) and 0.033 cm2V–1s–1 (31l). Biphenyl-capped oligomers (31d, 31j, 31n) were synthesized, where the thin film morphology and corresponding electrical performance was found to be dependent on the conjugation length.[194] Mobility values of 7 × 10–3 cm2V–1s–1, 0.17 cm2V–1s–1, and 0.055 cm2V–1s–1 were observed for 31d, 31j, and 31n, respectively, which correlated to the degree of crystallinity observed in the films. The same group had earlier reported a mobility of 0.66 cm2V–1s–1 for

8080_C003.fm Page 181 Friday, April 6, 2007 10:24 AM

Design, Synthesis, and Transistor Performance of Organic Semiconductors

Polar Substituents (29a-29l) S

R

S

S

R

S

S

R

R=

O

(29e)

O

n

O C4 H 9

N

O (29i)

C4 H 9

O

S

S

S

S

R

S

R= (CH 2) 3OC4 H9 (29b) = (CH2 )3 OC 8H 17 (29c) = (CH2 )4 PO(OEt) 2 (29d) O O O C10H 21 C10H 21 C4 H 9 N N N (29f) H (29g) H (29h) O O C10H 21 O (29l) (29k) O O

O

R 6

S

R

R= (CH2 )3 OC 4H 9 (29a)

(29j)

Thiophene-Fluorene Oligomers (30a-30v)

S S S

R n= 1, R= C 6H 13 (30a) n= 2, R= C 6H 13 (30b) n= 3, R= C 6H 13 (30c)

(30g)

O

OR O

O

(30j)

O

n=8 R=

(30n) O

R=

S

(30p)

S

O S

R1

S

R2

S

R= n-C2 H5 (30r) n-C4 H9 (30s) sec-C 4H 9 (30t)

R

OR

(30q)

R

R

(30o)

S

O

O

O

S

R

S

S

(30k)

O

C 8H 17 C 8H 17 RO

O

O

O

(30m)

O

(30l)

S

(30h) S

O

O

S

S

(30i) O

O

n= 4, R= C 6H 13 (30d) n= 2, R= H (30e) n= 2, R= cyclohexyl (30f) H2n+1 Cn C nH 2n+1

RO

n=3 R=

R

n

R1

S

R 1= C 6 H13, R2 = flourene (30u) R 1= C 6 H13, R2 = flourenone (30v)

Thiophene-phenyl Oligomers (31a-31z) C6 H 13

C 6H 13 n

S

n=1 (31a) n=2 (31b)

R

=

R

S

n=4

C 6H 13 (31k)

R= phenyl (31l) = tolyl (31m) = biphenyl (31n)

S

S

(31f)

= S

C 6H 13 (31o)

=

S

S

C 6H 13 (31e)

=

n

S

R

R

=

F

C 10 H21 (31f')

S

(31g)

S

S S

R

R= H (31q) C 6H 13 (31r) C 10 H21(31s)

R

S

S

S R=C 6H 13 (31w)

R=H (31v)

31t C6 H13 R

C10H 21 (31u) F F

F

R

F S

C 6H 13 S

S

F

S

S S

F =

R= tolyl (31p) R

R

S

R= phenyl (31c) = biphenyl (31d)

R= phenyl (31h) = tolyl (31i) = biphenyl (31j)

n=3

S

R

n

S

S S

R

S S

S

R=H (31x)

F

S S F

F

F

F

(31z)

R=C 6H 13 (31y)

Asymmetric End-capped Thiophenes (32a-32b) R1

S S

R 1 = H, R2 =C6 H13 (32a)

S S

R2

S S

S S (32b)

FIGURE 3.1.11 Thiophene and co-oligothiophene derivatives.

C6 H13

181

8080_C003.fm Page 182 Friday, April 6, 2007 10:24 AM

182

Organic Field-Effect Transistors

crystals of 31d that were epitaxially grown on a KCl (001) substrate and transferred to SiO2.[195] The number of thiophene units have been varied from one to four (31a, 31e, 31k, 31o) with highest mobility realized from 31e (0.09 cm2V–1s–1) and 31o (0.09 cm2V–1s–1).[196] Tolyl-substituted oligothiophenes containing three (31i), four (31m), and five (31p) thiophene rings have also been prepared [197] Although different thin-film morphologies were observed for each derivative, nearly identical mobilities were reported for the three (0.03 cm2V–1s–1 for 31i, 0.03 cm2V–1s–1 for 31m, and 0.02 cm2V–1s–1 for 31p). At elevated substrate deposition temperatures, layer-by-layer growth was observed for each molecule, but the molecular tilt angle at the substrate was substantially different. Furthermore, differences in morphology were observed based on odd or even number of thiophene units. By AFM, films with an even number of thiophene units showed faceted grains with sharply defined edges, while those with an odd number of thiophenes displayed elliptical domains with rough step edges. Oligothiophenes have also been substituted at the central position of the core with phenyl based units.[198] The co-oligomers 31r, 31y exhibited improved thermal stability over oligothiophenes of the same length. The oligomers have a broader band gap and lower HOMO energies than hexyl-substituted sexithiophene (28j). Depending on the structure of the central units, the hole mobility from the top-contact device configuration was 0.042 cm2V–1s–1 (31r), 0.049 cm2V–1s–1 (31y). A series of decyl-substituted (31s, 31u, 31f′) were synthesized yielding a hole mobility as high as 0.3 cm2V–1s–1, 0.08 cm2V–1s–1 and 0.4 cm2V–1s–1 .[199,200] Top-contact devices showed a hole mobility of 0.007 cm2V–1s–1 with an on/off ratio of 104.[115] Fluorene substituted phenyl-thiophene cooligmers (31g, 31z) are p-type semiconductors with mobilities of 0.01 and 4 × 10–5 cm2V–1s–1.[201] Unsubstituted and hexyl substituted oligomers with alternating phenylene and thiophene on the backbone have also been reported. Mobilities of 0.02 cm2V–1s–1 and 0.054 cm2V–1s–1 were obtained in the case of thermally deposited thin films of 31v and 31w.[115,196] An asymmetric hexyl-substituted quarterthiophene derivative (32a) has been reported[202] and compared to unsubstituted and di-hexyl substituted analogues (Figure 3.1.11). The authors showed an increase in mobility from 2 × 10–3 cm2V–1s–1 with no side chains (27a), to 0.02 cm2V–1s–1 with one hexyl chain (32a), and 0.06 cm2V–1s–1 for the symmetric hexyl substituted quaterthiophene (28a). Fused benzothiophene end-groups (32b) also improved the mobility (0.01 cm2V–1s–1) compared to unsubstituted 4T (27a), but to a lesser extent.[202] Oligomers composed of alternating thiophene and furan rings, and those having alkyl substituents at both ends of the molecules have been reported (Figure 3.1.12). A mobility of 0.042 cm2V–1s–1 was achieved for a thienyl-furan oligomer composed of five heterocycles and having hexyl groups at the terminal rings (33c)[203] A dithiophene-substituted indenoflurene compound (34c) has also been synthesized; a relatively high hole field-effect mobility of 0.012 cm2V–1s–1 was reported.[204]

8080_C003.fm Page 183 Friday, April 6, 2007 10:24 AM

Design, Synthesis, and Transistor Performance of Organic Semiconductors

183

Furan-Thiophene Oligomers (33a-33f) R

S

O

S

O

S

R

R

S

R

3

2

R= H (33a) = C 2 H5 (33b) = C 6 H13 (33c)

R= H (33d) = C 2H 5 (33e) = C 6H 13 (33f)

Other Thiophene Oligomers (34a-34c)

S

S S

S

(34a)

S C 6 H13

S

S

S

C6 H 13

(34b) S

S (34c)

FIGURE 3.1.12 Core-substituted thiophenes.

3.1.2.3 POLYMERS Polymers are attractive materials for solution processable organic semiconductors because of their high solubility and good film forming properties. One of the first solution-processed organic semiconductors used for field-effect transistors was poly(thiophene) (35a) (Figure 3.1.13). Since then various substituents have been incorporated on the polymer backbone, e.g. to impart functionality, increase solubility, or induce self-assembly. Among them, poly(3-hexylthiophene) (P3HT) is the most widely studied. The 3-alkyl substituents can be incorporated into a polymer chain with two types of arrangements, either head-to-tail (HT) or head-to-head (HH). A polymer with a mixture of HH and HT linkages in P3HT is referred to as regiorandom (35c), while one with only HT linkages is referred to as regioregular (35b). It has been found that highly regioregular P3HT (35b) self-orients into a wellordered lamellar structure with an edge-on orientation of the thiophene rings relative to the substrate.[205] P3HT with a high regioregularity (>95% HT linkages) adopt lamellae with an edge-on orientation giving mobilities of 0.05–0.2 cm2V–1s–1 and on/off ratios close to 106.[206-208] Polymers with a low degree of regioregularity

8080_C003.fm Page 184 Friday, April 6, 2007 10:24 AM

184

Organic Field-Effect Transistors

Thiophene-based Polymers (35a-35z'') R

R

S S head-to-tail (HT)

S

n

S R

RR

(35a)

S

S S head-to-head (HH) C12H 25 O

C 6H 13

C 6H 13

C 6H 13

S S S S poly(3-hexylthiophene) P3HT

S

C 6 H13

R= C4 H 9 (35d) = C 8H 17 (35e) = C 10 H21 (35f) n = C 12 H25 (35g)

C6 H 13

Regioregular P3HT (35b) Regiorandom P3HT (35c) O

HO O

O S

S

S

S O O

(35k)

n

S

n

(35h)

C 12 H25

O O

S (35j)

n

(35i)

C6 H13

F

S S

S

C 6 H13

F

S

S

S

F F (35m)

C6 H13

(35n)

C 8H 17

n

S

C 6 H13

S

S (35o)

S S

S

S

R R = C 8H 17 (35p) = C10H 21 (35q) = C12H 25 (35r)

S

3,3"PTT-8 (35s) S

Bu S m

S

S

n

n 3',4'PTT-8 (35t) C12H 25

S

S

C 12 H 25

PQT-12 (35u)

S 2

S S

S

Bu S

C8 H 17

S

S

n

n

C 8 H17

C 8H 17

C 8H 17 C8 H17

S

n C 12H 25

n

(35l)

S

S

n

C12H 25 C12H 25

S

Bu Si Bu

n

N

S

R

C6 H13

Bu Si Bu

m n

m= 2 (35v) = 3 (35w)

n Bu

S

m= 2 (35x) = 3 (35y)

Bu S

S 2

S m

S 2

2

n

R S S

S S

n

R = C 10 H21 (35z) = C12 H25 (35z') =C 14 H 29 (35z'')

R

FIGURE 3.1.13 Thiophene-based polymers.

adopt lamellae having face-on orientation giving mobilities of 10–4 cm2V–1s–1. A variety of solvents have been used for solution deposition of P3HT.[159] The mobility was found to vary by two orders of magnitude, with 1,2,4-trichlorobenzene giving the highest mobility.[208] LB prepared films of regioregular P3HT have shown a mobility of 0.02 cm2V–1s–1.[209] A comparison of poly(3-alkylthiophene)s (P3ATs) with side chains ranging from butyl (35d) to dodecyl (35g) showed a non-monotonic dependence of field-effect mobility on alkyl chain length (Figure 3.1.13).[11,210] The average mobility varied

8080_C003.fm Page 185 Friday, April 6, 2007 10:24 AM

Design, Synthesis, and Transistor Performance of Organic Semiconductors

185

from 1.2 × 10–3 cm2V–1s–1 in butyl (35d) to 2.4 × 10–5 cm2V–1s–1 for dodecyl (35g)substituted P3AT. In addition to the variation of alkyl side-chain length, modification of the chemical nature of the side chains in regioregular substituted polythiophenes have also been investigated. Regioregular polymers having chiral (35h), branched side chains (35i) and terminal carboxylate (35j) units were synthesized to study the nature of the side chains’ influence on morphology and OFET performance.[211] The carboxylic acid side chains were incorporated in an attempt to use hydrogen bonding to aid selforientation at polar surfaces, e.g. SiO2. The polymers are soluble in common organic solvents; however, solution-cast thin-films were non-uniform and essentially amorphous (except for 35h) giving mobilities of 1 × 10–3 cm2V–1s–1, 2.8 × 10–5 cm2V–1s–1 and 2.9 × 10–4 cm2V–1s–1 for 35h, 35i and 35j, respectively. The lower field-effect mobilities of these polymers compared to unbranched regioregular P3HT (35b) is believed to be due to the longer π–π stacking distance. The introduction of branches (methyl substitution), bulky side chains (oxazoline) or carboxylic acid units prohibit the formation of crystalline films through steric interactions. Polythiophenes containing electron withdrawing alkyl carboxylates (35k-35l) have also been synthesized.[212] OTFTs fabricated with the regioregular polythiophene (35k) gave mobilities of up to 0.07 cm2V–1s–1 compared to 0.0063 cm2V–1s–1 for the regiorandom polythiophene (35l).[212] Another electron-withdrawing moiety, tetrafluorobenzene, has been substituted on the polyalkylthiophene backbone (35m). Spin-cast films of 35m from 0.5 wt % chloroform solution showed a mobility of up to 2 × 10–3 cm2V–1s–1.[213] Recently there has been a thrust in design and synthesis of novel solution processable polymeric semiconductor other than P3HT. It has been reasoned that the electrical performance of P3HT is lower in air due to oxygen, which acts as a dopant, and its stability can be improved by increasing its ionization potential (IP).[214,215] The latter is partially dependent on effective π-conjugation length of the polymer backbone, which can be controlled either sterically — by reducing πoverlap between adjacent thiophene rings, or electronically — by introducing less conjugated unit in the backbone. Researchers at Xerox, Canada have demonstrated that proper control of the extended π- — system of regioregular polythiophenes enables solution fabrication of stable OFETs in ambient conditions.[216-218] They have synthesized regioregular alkyl-substituted polythiophenes with the alkyl sidechains strategically placed along the polythiophene backbone, to tune the π-conjugation by controlling both the torsional barriers and rotational freedom of thienylene moieties. Polymers 3,3′PTT-8 (35s)[217] and 3′,4′PTT-8 (35t)[216] differ from each other only in the regiochemistry of the alkyl side-chains. OTFTs were fabricated by spin coating 0.5–1 wt % chlorobenzene solutions of polythiophenes. The two regioisomers 35s and 35t showed mobilities of 0.03 cm2V–1s–1 and 0.01 cm2V–1s–1, respectively. The “spaced-out” arrangement of alkyl side-chains along the polymer backbone in 35s resulted in 3D-lamellar π-stacking and extensive intermolecular side-chain interactions, leading to a higher mobility compared to 35t, where only two-dimensional face-to-face π-stacks exist in thin films. PQT-12 (35u),[218] however, showed a much higher mobility of 0.14 cm2V–1s–1 and improved air stability, probably due to the presence of more unsubstituted thienylene moieties that possess

8080_C003.fm Page 186 Friday, April 6, 2007 10:24 AM

186

Organic Field-Effect Transistors

substantial rotational freedom, to us reducing the π-conjugation to some extent. This reduced π-conjugation imparts sufficient oxidative stability to the system, which is reflected in the minimal decrease in mobility and on/off ratio after being stored under ambient conditions for one month. In contrast, regioregular P3HT undergoes a drastic degradation in performance for devices stored under identical conditions. Polymers 35n and 35o have sterically inhibiting alkyl groups on neighboring thiophene rings, while polymers 35p-35r have an aromatic heterocycle thieno[2,3b]thiophene in the backbone that has a central cross-conjugated double bond designed to break the conjugation.[214,215] All the polymers exhibited useful molecular weights and low polydispersities. Thin-film transistors were fabricated either by spin–coating or drop–casting chloroform, dichlorobenzene or xylene solutions. Polymers 35o and 35n gave mobilities of 0.03 cm2V–1s–1 and 6 × 10–4 cm2V–1s–1, whereas mobilities of 0.15 cm2V–1s–1 and 0.12 cm2V–1s–1 were obtained for liquid crystalline polymers 35q and 35r, respectively with on/off ratios of 105.[214,215] The mobility of 35q has been further improved to 0.6 cm2V–1s–1 which is among the highest for solution-processed polymers.[219] The devices were found to be stable to ambient light and air. Monosilanylene-oligothienlene alternating polymers (35v-35y) contain silicon linkages which were anticipitated to provide solubility and elevate the HOMO energy level via its electron donating properties. However, due to the amorphous nature of their thin films, low mobilities in the order of 10–5 cm2V–1s–1 were observed.[220] McCulloch et al reported new semiconducting liquid-crystalline thieno[3,2b]thiophene polymers (35z-z′′′). Good transitor stability under static storage and operation in a low-humidity air environment was demonstrated, with charge-carrier field-effect mobilities of 0.2–0.6cm2V–1s–1 achieved under nitrogen.[221] Recently, fluorene-based co-polymers have attracted attention as promising materials for polymer OTFTs (Figure 3.1.14). A new class of co-polymeric semiconductor having fluorene and thiophene blocks, the AB-type co-polymer, poly(9,9′dioctyl-fluorene-co-bithiophene) (F8T2, 36a) have promising semiconducting characteristics.[222,223] OTFTs fabricated by spin-coating 0.5 wt % chloroform solution of 36a gave mobilities of 3 × 10–3–5 × 10–3 cm2V–1s–1. However, the mobility of 36a was dramatically enhanced to 0.01–0.02 cm2V–1s–1 by depositing the polymer onto a mechanically rubbed, substrate followed by high temperature annealing.[7,222] The chains of 36a were aligned parallel to the rubbing direction of the underlying substrate to facilitate charge transport resulting in a higher mobility. To increase the planarity and rigidity of the polymer backbone, the thieno[3,2-b]thiophene group was introduced into a fluorene-based alternating co-polymer, F8TT (36b).[224] Thin films of 36b gave a mobility of up to 1.1 × 10–3 cm2V–1s–1. Conjugated polymers consisting of dioctyl fluorene units and low band gap donor-acceptor-donor (D-AD) units (36c) have also been tested in OTFTs.[225] The D-A-D segment includes two electron-donating thiophene rings combined with an electron-withdrawing thiadiazolo-quinoxaline unit. Thin-film transistors of 36c gave a field-effect mobility of 3 × 10–3 cm2V–1s–1 and an on/off ratio of 104. Charge-transfer type co-polymers comprised of electron-donating and electronwithdrawing units are considered to be potentially useful as ambipolar semiconductors. With this in mind, co-polymers of thiophene (-electron donor) and thiazole or

8080_C003.fm Page 187 Friday, April 6, 2007 10:24 AM

Design, Synthesis, and Transistor Performance of Organic Semiconductors

187

Poly (9, 9'-dioctyl-ﬂourene-co-bithiophene) (36a–36c) S S

S

S

C8H17 n

SN N

n

C8H17

S

C8H17 F8T2 (36a)

C8H17

Thiophene-thiazole/thiadiazole co-polymer (37a–37b) C9H19 N N N S S S S n (37b) (37a)

n

N N

F8TT (36b)

C8H17

S

C8H17 (36c)

Cyclopentadithiophene based polymers (38a–38c) R1

R1

S

S

R1

R1 R2

n

n

R1 = H, R2 = C8H17 (38a) R1, R2 = C8H17 (38b)

Poly(alkylidene ﬂuorene) (39a–39c) R R

R1

S n S R = C8H17 (38c)

Poly(p-phenylene vinylene) (PPV) (40a–30e) OR1

OR2 O

n

R = C6H13 (39a) = C8H17 (39b) = C10H21 (39c)

R1O

n

R3O

n

R1 = CH3, R2 = C18H37, R2 = H (40a) R1 = CH3, R2,3 = 3,7-dimethyloctyl (40b) R1,2 = C11H23, R3 = C18H37 (40c)

n RO R = CH3 (40d) = 3,7-dimethyloctyl (40e)

Phenoxazine-based polymers (41a–41f) C6H13 CH 6

13

N

N O (41a)

O

n

C6H13 C H 6 13 Polytriarylamines (42a)

x

(41b) (X = 0.25) (41c) (X = 0.50) (41d) (X = 0.75)

1–x n

C6H13 N

N O (41a)

C6H13 (41e)

S

N n

O (41f)

(42a)

n

n

FIGURE 3.1.14 Thiophene copolymers and non-thiophene-based polymers.

thiadiazole (-electron acceptor) were synthesized and tested in OTFTs. The thiadiazole co-polymer (37b) (Figure 3.1.14) indeed showed ambipolar behavior[226], demonstrating an electron mobility of about 5 × 10–3 cm2V–1s–1 and a hole mobility of 3 × 10–4 cm2V–1s–1. The thiazole containing co-polymer (37a), however, showed only a hole mobility of 2.5 × 10–3 cm2V–1s–1.[227] Polycyclopentadithiophenes, a class of solution processable, thiophene based analogues of the polyfluorenes, have also been investigated as charge transport

8080_C003.fm Page 188 Friday, April 6, 2007 10:24 AM

188

Organic Field-Effect Transistors

materials in OTFTs.[228-231] Alternating copolymers, 38a and 38b, of cyclopentadithiophene and dioctylfluorene were prepared by Suzuki cross coupling. Thin films of 38c, obtained from spin-casting from a 0.8 wt% chloroform solution gave a mobility of 7.9 × 10–5 cm2V–1s–1, whereas polymers 38a and 38b show field-effect mobilities of 10–6–10–7 cm2V–1s–1. Since the carbon at the 9-position of fluorene-based molecules is sp3-hybridized and hence limits the π-conjugation, polyalkylidene fluorenes (39a-39c) have been synthesized. Polyalkylidene fluorenes (39a-39c) have a sp2-hybridized carbon at the 9-position which beside increasing the effective conjugation also permits the alkyl chains to adopt a coplanar conformation relative to the polymer backbone.[232] OTFTs fabricated by spin-casting 0.4–1 wt% polymer solutions in chloroform gave a mobility of up to 2 × 10–3 cm2V–1s–1 with an on/off ratio of 106. Different alkoxy-substituted poly(p-phenylenevinylene)s (PPVs) have been examined to relate performance to the substitution on the phenyl ring (symmetric: 40b, 40e or asymmetric: 40a, 40c, 40d) and nature of side chain, either linear or branched.[233-235] The PPV polymers were soluble in common organic solvents and OTFTs were fabricated by spin-casting a 0.3 wt % solution of PPV in chlorobenzene. Unsymmetrical alkoxy-substituted PPVs gave mobilities on the order of 10–4 cm2V–1s–1, compared to 0.01 cm2V–1s–1 for symmetrically substituted polymers. Polymers containing phenoxazine (41a-41e)[236] or phenothiazine (41f)[237] moieties (Figure 3.1.14) have recently attracted much research interest because of their unique electro-optical properties and their potential in diverse applications including light-emitting diodes and chemical luminescence. The rigid and planar structure of phenoxazine ring and its low ionization potential also make it a good building block for the design of semiconductors for OTFT applications. Phenoxazine-based conjugated polymers (41a-e) have been synthesized and tested as pchannel organic semiconductors.[236] Thin films of the polymers were made either by spin-casting or drop-casting a 0.2 wt% toluene or chloroform solution. OTFTs based on poly-phenoxazine (41a) showed a mobility of 0.9 × 10–5 cm2V–1s–1 and copolymers 41c and 41e showed mobilities of 3 × 10–4 cm2V–1s–1 and 6 × 10–4 cm2V–1s–1, respectively. The sulfur counterpart to 41a, poly-phenothiazine (41f) has also been reported, with a similar field-effect mobility (0.8 × 10–4 cm2V–1s–1).[237] Polytriarylamines (42a) are another class of highly stable polymeric semiconductors.[140,141] They can be handled in air and OTFTs are stable, with mobilities of 10–3–10–2 cm2V–1s–1 under ambient conditions. Moreover, several members of this family of compounds are amorphous, simplifying their characterization.

3.1.2.4 SOLUTION PROCESSABLE SEMICONDUCTORS: THE “PRECURSOR METHOD” Fully integrated solution-processed organic devices are of interest for large area rollto-roll manufacturing. To achieve this goal, extensive research efforts have been placed on functionalizing organic semiconductors in order to impart solubility. But most chemical modifications interrupt the natural π-stacking tendency of the molecules, which can inhibit charge transport that relies heavily on π-orbital overlap. In spite of a great deal of success in solution-processable thin-film devices with soluble

8080_C003.fm Page 189 Friday, April 6, 2007 10:24 AM

Design, Synthesis, and Transistor Performance of Organic Semiconductors

189

oligomeric and polymeric semiconductors, their low solubility still remains an inherent problem. Therefore, they usually require elevated solvent temperature to render solubility and the resultant films obtained by spin-casting are non-uniform in thickness and morphology. In general, spun-cast soluble semiconductors have lower performance than their vacuum-deposited counterparts. To mitigate these disadvantages, an alternative approach known as the “precursor route” has been developed. The idea is to synthesize a precursor molecule that is soluble in organic solvents but can be converted to the insoluble counterpart by either thermal or chemical treatment after spin-casting a thin film. 3.1.2.4.1 Precusor Polymers and Small Molecules The first semiconductor that was prepared using the precursor method was polyacetylene (43a)(Figure 3.1.15)[238] using the Durham route.[239] The scheme takes advantage of the properties of an intermediate precursor polymer that is non-conjugated and readily soluble in common solvents. Upon heating this precursor polymer to 80–100°C in vacuum for 12 hours, it undergoes an elimination reaction, losing hexafluoro-o-xylene as a volatile fraction and leaving polyacetylene in a fully condensed form.[239-241] OFETs with polyacetylene fabricated by this method have shown a mobility of 10–4 cm2V–1s–1. A thiophene based polymer poly(2,5-thienylenevinylene) (PTV, 43b) has been synthesized using the precursor method.[242] The dimethylformamide solution of the PTV precursor polymer was spun-cast and then heated to 200°C for 5 min under a nitrogen stream containing a small amount of HCl gas. During the heat treatment, HCl effectively acts as a catalyst for the conversion of the precursor polymer to the PTV.[243] Mobilities as high as 0.22 cm2V–1s–1 were reported. The same polymer has been used for fabrication of integrated circuits,[244] with reported mobilities of 10–4–10–3 cm2V–1s–1. The Durham route has also been used to prepare semiconducting tetrabenzoporphyrin 43c.[245] The precursor-porphyrin derivative consists of four ethylene bridged units which imparts solubility, and hence can be purified by column chromatography. Quantitative conversion to 43c is achieved throught heating the soluble precursor between 150–200°C, which results in the elimination of four ethylene molecules. OTFTs fabricated with this method yield a mobility of 0.017 cm2V–1s–1 and an on/off ratio of 105. The preparation of a soluble precursor of pentacene (3a) that can be spun-cast and then heated to form pentacene has been reported.[246-248] The pentacene precursor (Figure 3.1.15) is soluble in dichloromethane and forms a continuous, amorphous film when spun onto transistor substrates. A simple thermal treatment converts the precursor to pentacene via the elimination of tetrachlorobenzene. Thin films of pentacene formed by this technique demonstrated field-effect mobilities ranging from 0.1–0.2 cm2V–1s–1 for films converted at 200°C. Recently a novel approach for a high yield synthesis of another soluble pentacene precursor was demonstrated.[249,250] The synthesis involves a Lewis acid-catalyzed Diels-Alder reaction of pentacene and N-sulfinylacetamide.[251-253] OTFTs fabricated by spin-casting a chloroform solution of the precursor on substrate followed

8080_C003.fm Page 190 Friday, April 6, 2007 10:24 AM

190

Organic Field-Effect Transistors Precursor Polymers and Small Molecules (43a-e) F3 C

CF3

F3 C n

n soluble polymer precursor

CF3

HCl

+

S

O

N H

n

PTV (43b)

N H N

N

N -4 C 2 H4

H N

soluble ethylenebridged precursor

S

soluble precursor of PTV

polyacetylene (43a)

N

200o C

n

H N tetrabenzoporphyrin (43c)

Cl Cl

Cl

Cl

soluble pentacene precursor

Cl

pentacene (3a)

Cl

Cl Cl

200 oC O 120-200 oC

N S

O

soluble pentacene precursor R2 O R1 S O

S S Si

O H3 C C N S O O

S

R2

soluble precursor 150-200o C S

R1

O

S

S

S

Si S

soluble precursor

+

S

S

S S

S

pentacene (3a)

cat. CH 3ReO 3 CHCl3 , reflux

S S

(43d)

TBAF, pyridine n-butanol, 20 oC

S S

S S (43e)

FIGURE 3.1.15 “Precursor polymers.” Semiconducting polymers can be made by either thermal or chemical treatment of preformed solution-cast films.

by annealing (200°C/15 min or 130°C/25 min) under nitrogen atmosphere converted the adduct to pentacene (3a). The field-effect mobility of this film was as high as 0.89 cm2V–1s–1, ranking among the highest mobilities reported for an OTFTs fabricated by solution processing. The precursor approach has also be applied to thiophene oligomers.[254,255] The synthesis of symmetrically α, ω-ester substituted sexithiophene (precursor to 43d) containing thermally removable solubilizing groups and its incorporation into OTFTs was reported. Bulky and highly soluble ester end groups allowed the oligomers to be solution-cast into thin films at room temperature. A subsequent heating

8080_C003.fm Page 191 Friday, April 6, 2007 10:24 AM

Design, Synthesis, and Transistor Performance of Organic Semiconductors

191

Polydiacetylene (44a-c) R1 C C R2

n

R 1 = -(CH 2 )11CH3 , R 2= -(CH 2)COOH (44a) R 1 ,R 2= -(CH 2) 4OCONHPh (44b) R 1 ,R 2= -(CH 2) 4OCONH(CH2 )2 CH3 ( 44c)

FIGURE 3.1.16 Small molecule precursors.

cycle induced thermolytic removal of the ester solubilizing groups, affording 43d, while permitting the molecules to re-adopt the preferred π-stacking orientation that resulted in high charge mobility. OTFTs showed an increase in hole mobility from 10–5 cm2V–1s–1 with on/off ratios of ~100 before thermolysis to >0.1 cm2V–1s–1 with on/off ratios >105 after thermolysis. Recently, an acene fused-thiophene hybrid, p-channel semiconductor using a soluble precursor in which the solubility originates from trimethylsilyl groups was reported.[256] The elimination of two trimethylsilyl groups from the soluble precursor is accomplished through treatment with tetrabutylammonium fluoride in pyridine to give 43e. OTFTs made by thermally evaporated 43e showed mobility of 10–2 cm2V–1s–1. 3.1.2.4.2 Polydiacetylene Polydiacetylene (44a-c) (Figure 3.1.16) derivatives have been synthesized and tested in OTFTs. A hole mobility of 1.3 × 10–3cm2/vs and an on/off ratio of 104 were reported for 44a, the polydiacetylene prepared from 10, 12-pentacosadiynoic acid.[257]

3.1.3 N-CHANNEL ORGANIC SEMICONDUCTORS N-channel organic semiconductors are important class of materials, due to their necessity in the fabrication of bipolar transistors and complementary logic circuits. However, most of the literature to-date has focused on the design of p-channel semiconductors. Molecules such as unsubstituted oligothiophenes or pentacene are more conducive to the injection of holes than electrons since their ionization potential (~5 eV) matches reasonably well with the work function of typical metals used as source and drain electrodes (such as Au or Ag). Lower work function metals such as Al or Ca, would better facilitate electron injection, but are difficult to work with due to oxidative instability and their tendency to form charge transfer complexes between the organic semiconductor and metal.[40] In order to lower the electron injection barrier between the LUMO level of the organic with respect to the Fermi level of the metal electrodes, strong electron-withdrawing groups are often added to the outer rings of molecules through synthetic design. This has been done successfully with several semiconductor core systems. These groups increase the electron affinity and stabilize the anionic form of the molecule, allowing for the

8080_C003.fm Page 192 Friday, April 6, 2007 10:24 AM

192

Organic Field-Effect Transistors

possibility of efficient electron injection and transport. Synthetic design still provides a formidable challenge because of the inherent instability of organic anions, which are susceptible to atmospheric oxidants such as water and oxygen. Recent spectroscopic evidence has also shown that in organic semiconductors without sufficiently large electron affinities (greater than 3.85 eV), mobile electrons can be trapped by silanol groups at the dielectric interface of oxides used in typical transistor test configurations.[258] Nonetheless, recent progress has been made in this direction, with electron mobilities of greater than 0.1 cm2V–1s–1 demonstrated in several different families of materials. The synthetic design of n-channel organic semiconductors remains an intimate balance, which involves optimizing electron transport while limiting the degradation of electronic properties.

3.1.3.1 FULLERENES

AND

FULLERENE DERIVATIVES

Fullerenes (C60 and C60/C70) and fullerene derivatives (Figure 3.1.17) were some of the first n-channel materials studied. C60 (45a) has a solid state electron affinity of ~4.5 eV, and its nearly spherical shape provides isotropic carrier transport in thin films, which is different than the anisotropic behavior observed with linear oligomers (such as pentacene or oligothiophenes). In 1993, n-channel behavior in fullerenes, a mixture of C60 and C70 in a 9:1 ratio, was observed with mobilities as high as 5 × 10–4 cm2V–1s–1.[259] Later, polycrystalline films of C60 with a grain size on the order of 6 nm showed mobilities as high as 0.08 cm2V–1s–1.[260] When the substrate was pretreated with tetrakis(dimethylamino)ethylene prior to C60 deposition, the mobility increased to 0.3 cm2V–1s–1 at the cost of a diminished on/off ratio. The device performance degraded upon exposure to ambient conditions, but could be restored by annealing the films under high vacuum. Since then, mobilities as high as 0.56 cm2V–1s–1 have been reported for C60 films fabricated by molecular beam deposition without breaking vacuum.[261] Soluble methanofullerene derivatives have also been used in n-channel OTFTs. [6,6]-Phenyl C61-butyric acid methyl ester (PCBM, 45b) is a commercially available, solution processable derivative, whose LUMO level lies at approximately 3.7 eV with respect to vacuum. This barrier to electron injection (~1.4 eV with respect to Au) is significantly reduced due to the formation of a strong interface dipole layer at the metal/semiconductor interface.[244] A similar phenomenon has been observed for pristine C60 on Au and Ag.[262] Mobilities as high as 0.02–0.1 cm2V–1s–1 have been obtained for films of 45b measured in ambient conditions, which strongly depend on the work-function of the source and drain electrodes. Similarly high mobilities of 0.2 cm2V–1s–1 were recently demonstrated on polymer gate dielectrics for devices measured under an Argon atmosphere.[263] Soluble fullerene dendron (45b′) has been fabricated as an n-channel OTFT, with field-effect mobility reaching 1.7 × 10–3 cm2V–1s–1 at 300K.[264]

3.1.3.2 PHTHALOCYANINES Unsubstituted phthalocyanine derivatives show p-channel mobility as described in the previous section. However, when substituted with electron withdrawing groups

8080_C003.fm Page 193 Friday, April 6, 2007 10:24 AM

Design, Synthesis, and Transistor Performance of Organic Semiconductors

193

Fullerenes and Fullerene Derivatives (45a-45b') O

O

OMe

O PCBM, (45b)

C 60 , (45a)

N

O O

NH

OMe O

N

HN

NH

O

N

OMe

O

(45b')

O

OMe

Phthalocyanines (45c-45k) R

R

R

R

R

R R

N

R

N N N

R

N

R

N M

R

Cu

N

N N

N

N

N

N

R R

R R= R= R= R= R=

N

N

N

R R

R

R

R

R

R = SO3 Na (45h)

F, M = Cu (45c) F, M = Zn (45d) F, M = Co (45e) F, M = Fe (45f) Cl, Me = Fe (45g) R

R N N

N Cu

N

N N

N

N

R=

R + N

Cl-

N

M

N

N

R

N N

(45i)

N N

N

M = Lu (45j) = Tm (45k)

FIGURE 3.1.17 Fullerenes and phthalocyanines.

at their periphery, such as hexadecafluoro-subsitituted copper phthalocyanine (45c) (Figure 3.1.17), these materials showed n-channel field-effect mobilities as high as 0.03 cm2V–1s–1.[265] Not only do these devices show good electronic performance, but they operate under ambient conditions with excellent stability. Other metals centers and halogenated substitutents have been studied; the highest mobilites were obtained in each case from devices fabricated with a substrate temperature of 125°C or greater. The best mobilities for the other derivatives were 1.2 × 10–3 cm2V–1s–1

8080_C003.fm Page 194 Friday, April 6, 2007 10:24 AM

194

Organic Field-Effect Transistors

(45d), 4.5 × 10–5 cm2V–1s–1 (45e), 2.1 × 10–3 cm2V–1s–1 (45f), and 2.7 × 10–5 cm2V–1s–1 (45g). The difference in mobility values was attributed to to morphological differences in the crystallinity and grain size of the films. Ambipolar mobility has also been reported from water soluble copper phthalocyanines substituted with sulfonic acid (45h) and methyl pyridinium groups (45i). Unlike typical transistors with a linear and saturation regions, the drain-source current of these devices increases non-linearly with increasing VDS at a given gate bias. Furthermore, the entire set of I-V curves shifts up with repeated scans while the on/off ratio decreases. The authors attributed this behavior to a mechanism that involved ion-modulated electrical conduction. N-channel mobilities as high as 3 × 10–4 cm2V–1s–1 were observed.[266] Bisphthalocyanines based on rare earth metals lutenium (45j) and thulium (45k) were studied as early as 1990.[267] These compounds had a field-effect mobility of ~10–4 cm2V–1s–1 under vacuum, but the electronic properties degraded rapidly upon exposure to ambient conditions.

3.1.3.3 NAPHTHALENE DIIMIDE DERIVATIVES Naphthalene tetracarboxylic dianhydride (NTCDA, 46a) and its family of imide derivatives are easily synthesized from commercially available starting materials (Figure 3.1.18). When measured in vacuum, the unsubstituted 46a showed n-channel mobility of ~10–4 cm2V–1s–1 when deposited at a substrate temperature of 25°C and 3 × 10–3 cm2V–1s–1 when it was increased to 55°C. Upon exposure to air, the mobility decreased by two orders of magnitude. The morphology of both films showed small crystalline grains with an average size of 200 nm, but the films deposited at a substrate temperature of 55°C had grains with better connectivity, accounting for the higher mobility observed.[268] Alkyl substituted versions of naphthalene tetracarboxylic diimide (NTCDI, 46b) showed mobilities of 0.16 cm2V–1s–1 (R = octyl, 46c), 0.01 cm2/V–1s–1 (R = dodecyl, 46d), and 5 × 10–3 cm2V–1s–1 (R = octadecyl, 46e) when measured under vacuum, but no observable electron mobility when measured in ambient. The incorporation of fluoroalkyl groups on the side chains stabilizes NTCDI thin films and allows for n-channel operation under ambient conditions. Mobilities as high as 0.03 cm2/V–1s–1 were obtained for 46f and 0.06 cm2V–1s–1 for 46g with on/off ratios on the order of 105.[269] An even higher mobility (0.12 cm2V–1s-1) was obtained with a substituted benzyl derivative that contained a CF3 group in the para position (46h). The striking differences in the electron mobility measured under ambient atmosphere between NTCDI derivatives containing fluorine groups in the sidechains and those without was investigated in a later publication.[270] From electrochemistry measurements, the electron withdrawing character of the fluorinated substituents (R = CH2C7F15, 46g) shifts the reduction potential by about 0.14 V in comparison to the nonfluorinated (R= octyl, 46c) NTCDI. The difference was negligible between the octyl-substituted NTCDA and the derivative with a terminal fluoromethyl group (46h). None of the fluorinated compounds had reduction potentials lower than that of NTCDA or C60, neither of which can operate in air. From this data, it was concluded that solid state packing effects must play a major role in explaining the

8080_C003.fm Page 195 Friday, April 6, 2007 10:24 AM

Design, Synthesis, and Transistor Performance of Organic Semiconductors

195

Naphthalene Diimides (46b–46l)

O

O

O

O

N

O

O

N

O

O

O

N

O

O

N

O

O N C3F7

O

O O NT CDA, (46a)

C3F7

R

H

R R = C8H17 (46c) = C12H25 (46d) = C18H37 (46e)

H NTCDI, (46b)

N O

O N

O

O

N

O

O

O N C7F15

O

(46f)

(46g) R1

O F3C

C7F15

N

CF3

O

N

O

O

N

O

(46h)

O

R2 R1 = CH2 (CH2)7CH3 R2 = NH2 (46i) = CH2 (CF2)6CF3 = NH2 (46j) = NH2 (46k) = CH2 (CF2)2CF3 = N(CH3)2 (46l) = (CH2)7CH3

FIGURE 3.1.18 Naphthalene diimide derivatives.

stability, but the origin of this behavior is still unknown. One possible explanation involves the close-packing nature of fluorined side chains, which may provide a kinetic barrier to oxidation. Asymmetric NTCDIs with alkyl and fluoroalkyl sidechains have also been synthesized (46i-46l). The highest mobility obtained with an alkyl version (46i) was 3 × 10–4 cm2V–1s–1 under vacuum and 2 × 10–4 cm2V–1s–1 with a fluoroalkyl derivative (46j).[271]

3.1.3.4 PERYLENE DIIMIDE DERIVATIVES Perylene (47a) (Figure 3.1.19) has been employed as the active-layer in OTFTs, exhibiting low p-channel mobilities and no n-channel mobility, even though theoretical calculation predicts n-channel behavior.[272] Dianhydride and diimide versions of perylene have also been studied in detail by many groups. The earliest report of n-channel mobility of perylene tetracarboxylic dianhydride (PTCDA, 47b), ~10–4 cm2V–1s–1, was obtained under vacuum or in moisture-free air.[273] The substituted perylene tetracarboxylic diimides (PTCDI, 47c-47j) can be easily synthesized in similar fashion to the naphthalene versions of the same molecule. The energy levels of alkyl substituted diimides are similar to unsubstituted versions (3.4 eV and 5.4 eV for electrons and holes referenced to vacuum level).[274,275] The single-crystal structure of 47c reveals π-stacking in a triclinic lattice.[276] The highest reported n-channel mobility for a perylene derivative to date was found from an octylsubstituted PTCDI (47d), reaching a value of 1.3 cm2V–1s–1 under vacuum, when

8080_C003.fm Page 196 Friday, April 6, 2007 10:24 AM

196

Organic Field-Effect Transistors Perylene Diimide Derivatives (47a-47k) O

O

O

O

O

O PTCDA, (47b)

(47a)

NC O

O

R N

N R

O

O

O

O

R N

N R

O

O CN

R = C5 H11 (47c) = C 8 H17 ( 47d) = C 12H 25 (47e) = C 13H 27 (47f) = C 6 H5 (47g) R= (47h)

R=

(47i)

R = n-CH2 C 3F7 R= n-C 8H 17

(47j)

(47k)

FIGURE 3.1.19 Perylene diimides.

corrected for contact resistance.[277] In the same work, PTCDI substituted with pentyl (47c) and dodecyl (47e) sidechains showed a mobility of 0.06 and 0.5 cm2V–1s–1, respectively. Thin films of these derivatives adopt a molecular packing comparable to the single-crystal structure of 47d. Recently, n-channel mobility of 0.6 cm2V–1s–1 was reported with 47f and complementary inverters were demonstrated (using pentacene as the p-channel material) that showed record gain.[275] Derivatives with phenyl[278] (47g) and cyclohexyl substitutents[182] (47h) have also been reported with mobilities of 1.5 × 10–5 cm2V–1s–1 and 1.9 × 10–4 cm2V–1s–1 respectively. Core-substituted perylene diimides have also been functionalized at the two bay positions (position 1 and 7) with electron withdrawing cyano groups (47i and 47j).[279] Both compounds showed good solubility in organic solvents and have a lower LUMO than 47d, suggesting improved n-carrier stability. Single crystals of 47j were grown by sublimation, and the crystal structure analysis revealed a polycyclic core that is slightly twisted from steric repulsion caused by the cyano groups. Furthermore, the molecules were shown to pack with a face-to-face, slipped πstacked structure with a minimum interplanar spacing of 3.4 Å. Electronic measurements were performed on compounds that were a 1:1 mixture of isomers cyanated at the 1,7- and 1,6-positions of the perylene core. Mobilities as high as 0.1 cm2V–1s–1

8080_C003.fm Page 197 Friday, April 6, 2007 10:24 AM

Design, Synthesis, and Transistor Performance of Organic Semiconductors

197

and 0.64 cm2V–1s–1 were found for 47i and 47j, respectively, at optimized substrate deposition temperatures. Owing to the good solubility, solution processed films of 47j show mobilities of 10–3–10–4 cm2V–1s–1 in the bottom contact device configuration with alkane thiol treated gold electrodes. 47i was also cast from solution onto unmodified substrates, where mobilities of 10–3–10–5 cm2V–1s–1 were observed. These two materials have excellent air stability, with on/off ratios as high as 105 when measured under ambient conditions. N, N′-bis(n-octyl)-dicyanoperylene-3,4:9,10bis-(dicarbpximide) (47k) was also used in an n-type OTFT, with a reported mobility and on/off ratio of 0.14 cm2V–1s–1 and 1.2 × 10–3. [280]

3.1.3.5 QUINOID SYSTEMS Tetracyanoquinodomethane (48a) and 11,11′,12,12′-tetracyanonaptho-2,6,quinodimethane (TCNQ, 48b) are commercially available and n-channel behavior has been observed for both compounds (Figure 3.1.20).[281] Devices fabricated using 48b display a higher mobility (10–3 cm2V–1s–1) than 48a (10–5 cm2V–1s–1), and have better air-stability than NTCDA, although both 48a and 48b have low on/off ratios due to unintentional doping. A new class of dicyanopyrazinoquinoxaline derivatives were also synthesized and tested for their electronic performance. The mobilities obtained for all the compounds (48c-48h) were very low, ranging from 10–6–10–8 cm2V–1s–1, but some of the compounds had a very unique crystal packing arrangement that the authors referred to as a molecular tape structure arising from the C-H…N intermolecular interactions.[282] Dicyanomethylene groups have been incorporated onto the periphery of a terthiophene backbone, stabilizing the all-planar quinoid form of the molecule, with n-Butyl groups were added to the central thiophene moiety for improved solubility (48i).[283]. The bulk crystal structure indicates that the molecules form a face-toface, slipped π-stack structure containing π-dimers that pack in alternating columns. Thin films deposited at elevated temperatures showed an n-channel mobility as high as 0.2 cm2V–1s–1 in high vacuum.[284] Additionally, ambipolar behavior was observed in some cases when the substrate temperature for deposition was greater than 136°C.

3.1.3.6 THIOPHENE BASED N-CHANNEL OLIGOMERS As with perylene cores, perfluoroalkyl substitution has been performed on oligothiophenes (Figure 3.1.21).[39,285,286] In this case, the substitution switches the charge carrier from holes to electrons. With oligomers 49a-49c, the HOMO-LUMO gap is nearly identical among the perfluoroalkyl substituted and the alkyl substituted versions of the same molecules, but in each case, the LUMO is approximately 0.5 eV lower for the perfluoroalkyl versions. The shift of energy levels is quite different from the naphthalene and perylene cores substituted with fluorinated sidechains, in which the perfluoroalkyl chain was not in direct communication with the π-system because of an alkyl spacer. The influence of the halogenated substitutions is also different than the halogenated phthalocyanine system, as the perfluoroalkyl chains

8080_C003.fm Page 198 Friday, April 6, 2007 10:24 AM

198

Organic Field-Effect Transistors

Quinoid systems (48a–48i) NC

CN

NC

CN

NC CN

NC

CN

(48a)

TCNQ, (48b)

R2

R1 N

N

CN

N N CN R4 R1 = R2 = R3 = R4 = H (48c) R3

N

N

CN

NC

N

N

N

N

CN

NC

N

N

(48g)

NC NC

(48h) S S

S R

R

CN CN

R1 = R4 = H (48d) R2 = R3 = CH3 R1 = R4 = H (48e) R2 = R3 = OCH3 R1 = R4 = OCH3 (48f) R2 = R3 = H

R = C4H9 (48i)

FIGURE 3.1.20 Quinoid derivatives.

only exhibit a strong σ-inductive electron-withdrawing effect, where as halogens directly attached to the phthalocyanine system exhibit both σ-inductive and resonance donating effects. Electron mobilities of 0.059 cm2V–1s–1, 0.026 cm2V–1s–1, and 1 × 10–3 cm2V–1s–1 were observed for 49a, 49b, and 49c, respectively, with on/off ratios ranging from 104–105. The same type of substitution has been performed with mixed phenylene-thiophene oligomers .[287] All of the oligomers (49d-49g) showed clear n-channel performance, with 49e demonstrating the highest mobility for the series at 0.074 cm2V–1s–1 and an on/off ratio of 6 × 106 when deposited at a substrate temperature of 110°C. The fabrication of transistors based on carbonyl functionalized quarterthiophenes (DHCO-4T (49h) and DFCO-4T (49i)) has been reported.[288] The carbonyl functionality acts as a strong electron withdrawing group that allows the possibility of further chemical functionalization. High n-channel mobilities of 0.1 cm2V–1s–1 for 49h and 0.6 cm2V–1s–1 for 49i were obtained under high-vacuum measurement conditions. Surprisingly, 49h also exhibited p-channel mobilities as high as 0.01 cm2V–1s–1. The authors claim that 49i also exhibits ambipolar behavior only after doping with I2. A more electron deficient 49j and the dioxolane protected 4T core (49k) were also reported to show n-channel mobility an order of magnitude lower than the carbonyl derivatives 49h and 49i.[288] Perfluorophenyl endgroups were attached to a quarterthiophene core (49l) and an n-channel mobility of 0.08 cm2V–1s–1 with an on/off ratio of 105 was reported.[289] However, perfluorophenyl dithiophene (49q) and tertthiophene (49p) exhibited poor mobilities compared with 49l, particularly for 49q. The crystal structure of 49q reveals that only three rings are fully conjugated, making the electronic structure of this molecule closer to lower-mobility terthiophene than quaterthiophene derivatives.[201] Phenacyl groups have been incorporated at the periphery of oligothiophene core (49m) to fabricate OTFTs with high electron mobilities.[290] With the substi-

8080_C003.fm Page 199 Friday, April 6, 2007 10:24 AM

Design, Synthesis, and Transistor Performance of Organic Semiconductors

199

Thiophene-based N-Channel Oligomers (49a-49q) S

C6 F13

S

S

C 8F 13

C 8 F13

C 6F13

S

S

x x = 1 (49a) = 1.5 (49b) = 2 (49c)

x

x = 1 (49d) = 2 (49e) = 3 (49f) = 4 (49g) O S

R

S S

R

S O

R = C6 H13 (49h) = C6 F13 (49i)

O O

O

C6 F13

S

S S

C 6 F13

S

S

C 6F 13

S

S

C6 F13

S

O

O

(49j)

O

O

(49k) F

F

F

F

S

S F

F

(49l)

O

S

S

O

S

F

O

F

F

F

F

F

F

S

S

O (49n)

F

F

S

F

F

F

F (49p)

S F

F

F

S

F F

F

F

F

F

S

(49o)

F

F

F

F

F S

F

F O F (49m) F

F

S

F

O

S

S F

F

F

F O

F

F

F F

F

F

O

F

S

S

F

F F

(49q)

F

FIGURE 3.1.21 n-Type thiophene derivatives.

tution of perfluorophenacyl groups, electron mobilities as high as 0.45 cm2V–1s–1and on/off ratios of 108 were observed with devices tested under argon. Under the same argon conditions, these materials were also cast from solution and a mobility of 0.21 cm2V–1s–1 with on/off ratios of 105 was obtained, accounting for the highest values to date for solution processed n-channel devices. The phenacyl substituted quarterthiophene without fluoro-substitution (49n) showed only p-channel characteristics, with the highest mobility value of 0.04 cm2V–1s–1. Suprisingly, the reduction potential

8080_C003.fm Page 200 Friday, April 6, 2007 10:24 AM

200

Organic Field-Effect Transistors

of the perfluorophenacyl derivative was slightly more negative than that of the pchannel derivative, as observed from cyclic voltammetry. The authors were able to grow single crystals of both materials and attributed the different electronic properties to subtle molecular conformational differences in the crystal structure. Careful analysis reveals that the dihedral angle between the carbonyl group and the thiophene core was much greater in the case of the phenacyl derivative with H atoms. The greater conjugation sustained in the perfluorophenacyl derivative should enhance stabilization of a negatively charged core in the solid state. These two molecules provide a good example where molecular energy levels are not the determining factor for the type of charge carrier, and show that even small changes in crystal packing can play a major role. An n-channel, perfluorinated version of pentacene (49o) has recently been synthesized.[72] The crystal structure of both pentacene and perfluoropentacene adopt a herringbone geometry and belong to the same symmetry group (p2gg), but the two molecules in the unit cell exhibit different azimuthal angles. The closest proximity of two intermolecular carbon atoms in pentacene is 3.64 Å while short C-C intermolecular distances ranging between 3.22Å and 3.25Å are observed in perfluoropentacene due to the electrostatic interaction between the electropositive pentacene moieties and electronegative fluorine atoms. Devices fabricated with perfluoropentacene at a substrate temperature of 50°C revealed an n-channel mobility of 0.11 cm2V–1s–1 and an on/off ratio of 106 when measured under vacuum. Inverter circuits were also fabricated using pentacene as the p-channel semiconductor.

3.1.3.7 TRIFLUOROMETHYLPHENYL-BASED OLIGOMERS Recently, the trifluoromethylphenyl endgroup has also been used as an electron withdrawing group in linear oligomers based on phenylene-thiophene (50a), thiazolothiazole (50b, 50c), and anthracene derivatives (50d) (Figure 3.1.22).[118,291] Single crystals of 50d reveal a herringbone packing arrangement similar to usual oligothiophenes, but the oligomers containing the thiazolothiazole unit (50b and 50c) have a columnar π-stacking structure. The best transistor performance reported for this family of compounds occurred for a substrate deposition temperature of 50°C. Compound 50c demonstrated an n-channel mobility of 0.30 cm2V–1s–1 and an on/off ratio of 106. 50b did not show any transistor characteristics, while 50c had an electron mobility as high as 0.30 cm2V–1s–1 and an on/off ratio of 106. The phenylthiophene co-oligomer, 50a, showed a mobility of 0.18 cm2V–1s–1. In a later publication the authors observed an electron mobility of 3 × 10–3 cm2V–1s–1 from the phenyl-anthracene cooligomer, 50d.[292] Recently, transistors fabricated based on thiazole oligomers with trifluoromethylphenyl groups demonstrated very high electron mobility.[293] Within the series of molecules (50e-50i), the electron affinity increases with increasing number of thiazole rings. Compound 50e did not show any mobility but 50f exhibited an nchannel mobility as high as 1.83 cm2V–1s–1 when deposited on OTS treated substrates at room temperature. This is the highest reported room temperature mobility to date for n-channel materials. The longer oligomers also showed good mobilities with 2.8

8080_C003.fm Page 201 Friday, April 6, 2007 10:24 AM

Design, Synthesis, and Transistor Performance of Organic Semiconductors

201

Triﬂuormethylphenyl-based oligomers (50a–50j) F3C

S

S

N S S N

F3C

CF3

(50a)

F3C

S

CF3

(50b) S

N

N

S

CF3 S

F3C

CF3

(50d)

(50c)

F3C

F3C

N S

S N

S N (50g)

S

N S

S N (50i)

N S S N (50f )

F3C

N S S N (50e)

F3C

CF3 F3C

N S

S N

CF3

CF3

N S

S

CF3

(50h) F3C

S

S

CF3

S

S

S

S

CF3

(50j)

FIGURE 3.1.22 n-Type organic semiconductors with trifluoromethyl endgroups.

× 10–3 cm2V–1s–1 for 50g, 0.085 cm2V–1s–1 for 50h, 0.018 cm2V–1s–1 for 50i cm2V–1s–1 and 0.025 cm2V–1s–1 for 50j.

3.1.3.8 POLYMERIC SYSTEMS The electron mobility has been investigated in ladder type polymers poly(benzobisimidazobenzophenanthroline) (BBL, 51a) (Figure 3.1.23).[294,295] This polymer has high thermal stability, a glass transition temperature greater than 500°C, and is insoluble in aprotic organic solvents. Films can be solution cast or spin coated from methanesulfonic acid (MSA) or Lewis acid (AlCl3, GaCl3, FeCl3)/nitromethane mixtures. Electron mobility in the saturation regime was 0.03–0.05 cm2V–1s–1 for films cast from MSA with an on/off ratio of 103. In the linear regime, mobilities as high as 0.1 cm2V–1s–1 were observed, the same order of magnitude as the best hole mobility obtained in conjugated polymer systems. The non-ladder derivative, BBB (51b), which has an identical optical band gap and absorption spectra, has also been investigated. The highest obtainable mobility was 10–6 cm2V–1s–1 when cast under the same conditions. The large differences in mobility were attributed to morphological differences in thin films between the two polymers. The BBL has more efficient π-stacking and greater intermolecular order reasoned from the high degree of crystallinity. X-ray scattering from BBL (51a) thin films exhibit a Bragg (010) d-spacing of 3.3–3.4 Å due to the — intermolecular stacking of the planar chains.[296] Films of BBB (51b) were shown to be completely amorphous.

8080_C003.fm Page 202 Friday, April 6, 2007 10:24 AM

202

Organic Field-Effect Transistors

Polymeric Systems (51a–51e) O

O

N

N

N

N

O

O

N

N

BBL, (51a)

n

C8H17 N

S

C8H17

C8H17

n

(51c)

BBB, (51b)

O C8H17

S

n

N

N

n

COONH4

OR NN

N N RO

O

n

N

(51d)

R = 2' ethylhexyl (51e)

FIGURE 3.1.23 n-Type polymeric systems.

Ambipolar behavior has also been observed in BBL and poly(thiophene-3propionic acid, ammonium salt).[297] 51a showed n-channel mobilities of 0.04–0.06 cm2V–1s–1 and 0.02–0.03 cm2V–1s–1 for p-channel operation. 51c showed values of 0.5–0.7 cm2V–1s–1 for n channel and 1.2–1.7 cm2V–1s–1 for p-channel with on/off ratios between 2 and 50 for devices operated in air. A similar mechanism involving ion-modulated electrochemical conduction, as described above for water soluble phthalocyanines was proposed by the authors. Recently, n-channel transistors of poly(9,9-di-n-octylfluorene-alt-benzothiadiazole) (51d) have been reported using thin films fabricated by spin-coating.[298] Electron mobilities ranging from 6 × 10–4–4.8 × 10–3 cm2V–1s–1 were observed with the use of a 50 nm BCB polymer layer on top of the SiO2 dielectric and Ca electrodes. Alternating copolymers containing 1,3,4-heterodiazoles and fluorenes have also been synthesized and show very low n-channel mobilities (2.2 × 10–8 cm2V–1s–1 for compound 51e).[299]

3.1.4 OUTLOOK AND CONCLUSIONS In this chapter we have reviewed advances in organic semiconductor performance, combining newly synthesized, novel materials and modifications of existing promising materials. In addition to efforts driving new material development, many groups have found that the performance of existing materials can be vastly improved with the optimization of their morphology and structural order. In addition to developing new materials, a significant effort has been focused on improving the stability of materials towards air, moisture and light exposure in addition to the electronic performance. Significant achievements have been made in developing pchannel and, to a lesser degree, n-channel semiconductors. The performance of several organic semiconductors, both n- and p-channel, has already surpassed that of amorphous silicon.

8080_C003.fm Page 203 Friday, April 6, 2007 10:24 AM

Design, Synthesis, and Transistor Performance of Organic Semiconductors

203

The growth of OFET technology has been impressive, especially considering the efforts spent in developing and optimizing the materials for the semiconductor layer alone. Transistor performance has improved rapidly as a result of worldwide activity in research groups worldwide. Applications ranging from radio-frequency identification tags to flexible displays have been proposed and are poised for introduction into commercial products. Significant progress has been made toward the correlation of the molecular electronic properties and their arrangement in ordered superstructures to the macroscopic transport characteristics. Though comprehensive descriptions of such relationships remain elusive, the vast array of empirical data and the isolation of key transport parameters offer a promising look toward the engineering of high-performance semiconducting materials. Despite this impressive set of advances, several issues remain to be addressed before commercialization can begin. A better understanding of effects of the dielectric interface, organic semiconductor growth mechanisms and methods to make ohmic contacts will enable better device performances. Long-term research efforts and innovation are needed to provide new organic semiconductors with improved performance, processability, and environmental stability to oxygen and moisture. In the review, we have limited the scope to the semiconductor material, but there has also been great attention paid to the dielectric layer and conductive electrodes necessary to complete the all-organic electronic devices. The successful development and adaptation of these new materials will require increased multidisciplinary partnerships among physicists, chemists, and engineers. As more sophisticated and versatile methods currently developed in the laboratory make their way into the manufacturing environment, we can expect that organic electronic circuits will have a profound impact on future technological advances.

3.1.5 TABLE OF MOBILITIES Table 3.1.1 contains field-effect mobilities and on/off ratios.

TABLE 3.1.1 Field-Effect Mobility and On/Off Ratio Found in the Literature for Each Organic Semiconductor Described in the Text Molecule[ref.] 1a [48] 1b [49] 1c [49] 1d [49] 1e [49] 1f [53] 1g [53]

Deposition methodb

Mobility (cm2V–2s–1)

sc v v v v v v

0.02 0.013 0.13 0.072 0.18 0.063 0.5

ION/IOFF 104 104 104 105 104 8.7 × 105 2.8 × 107 Continued

8080_C003.fm Page 204 Friday, April 6, 2007 10:24 AM

204

Organic Field-Effect Transistors

TABLE 3.1.1 (Continued) Field-Effect Mobility and On/Off Ratio Found in the Literature for Each Organic Semiconductor Described in the Text Molecule[ref.] 1h [55] 1i [56] 2a [58] 2a [57] 2b [60] 2c [60] 2d [60] 2e [60] 2f [55] 3a [62] 3a [249,250] 3b [67] 3c [46] 3d [61] 3e [68] 3f [69] 3g [69] 3h [69] 3i [71] 3j [71] 4a [79] 4b [79] 4c [79] 5a [34] 5a [81] 5a [83] 5b [81] 6a [85] 6b [85] 6c [85] 6d [85] 6e [86] 6f [87] 7a [88] 7b [88] 7c [88] 7d [88] 8a [89,90] 8b [89,90] 9a [91] 9b [91]

Deposition methodb

Mobility (cm2V–2s–1)

ION/IOFF

v v v sc sc sc sc sc v v p v v v v s s v v v v v v sc v s v v v v v s s v v v v v v v v

0.12 1.1 0.1 1.3 1.4 × 10–4 0.3 1.6 N.A.c 0.5 6 0.89 0.3 N.O.d 2.5 0.251 1 × 10–5 1 × 10–5 0.4 0.014 0.045 0.05 N.A. N.A. 15.4 10–3 0.7 10–3 1.4 × 10–4 0.011 0.012 9.5 × 10–4 0.001 0.02 5 × 10–5 0.006 0.001 5 × 10–4 0.038 0.148 0.09 0.15

108 4.4 × 105 106 106 103 102 105 N.A. 108 106 107 6.3 × 103 N.O. 2 × 106 3.15 × 103 N.A. N.A. 106 N.A. N.A. 106 N.A. N.A. 106 N.A. 106 N.A. 3.3 × 102 8.2 × 104 5.7 × 104 8.7 × 102 104 106 103 5 × 103 7 × 102 3 × 103 N.A. N.A. N.A. N.A.

8080_C003.fm Page 205 Friday, April 6, 2007 10:24 AM

Design, Synthesis, and Transistor Performance of Organic Semiconductors

TABLE 3.1.1 (Continued) Field-Effect Mobility and On/Off Ratio Found in the Literature for Each Organic Semiconductor Described in the Text Molecule[ref.] 9b [91] 9c [91] 9d [91] 9e [92] 9f [92] 9g [92] 10a [93] 10b [38] 10c [38] 10d [94] 10e [95] 10f [95] 11a [96–98] 11b [99] 11c [100] 11d [101] 11e [101] 11f [101] 12a [102] 12b [102] 13a [103,104] 13b [103,104] 13c [103,104] 13d [103,104] 13e [103,104] 13f [103,104] 13g [106] 13h [106] 14a [107] 14b [107] 14c [108] 14d [108] 15a [109,110] 15b [109,110] 15c [109,110] 15d [110] 15e [111] 15f [111] 15f [111] 15g [110]

Deposition methodb

Mobility (cm2V–2s–1)

s v v s s s v v v v v v v v v v v v v v v v v v v v v v v v v v v v v v v s v v

0.02 0.14 0.06 N.A. 1 1 × 10–4 0.04 1.6 × 10–2 10–3 0.15 0.011 0.12 0.05 0.045 0.02 0.42 0.12 0.14 0.06 0.09 3 × 10–4 1 × 10–4 0.006 0.057 0.036 0.11 N.O. 0.0037 7 × 10–5 7 × 10–5 6 × 10–5 4 × 10–5 0.003 1.2 × 10–4 0.12 1 × 10–5 0.01 0.002 0.14 0.001

ION/IOFF N.A. N.A. N.A. N.A. 107 103 105 N.A. N.A. 106 4.0 × 104 1.6 × 105 108 103 106 5 × 106 5 × 105 104 7.3 × 104 7 × 104 30 10 6 × 102 3 × 104 3 × 104 105 N.O. 104 106 106 N.A. N.A. N.A. 103 107 N.A. 105 104 107 105 Continued

205

8080_C003.fm Page 206 Friday, April 6, 2007 10:24 AM

206

Organic Field-Effect Transistors

TABLE 3.1.1 (Continued) Field-Effect Mobility and On/Off Ratio Found in the Literature for Each Organic Semiconductor Described in the Text Molecule[ref.] 15h [112] 15i [112] 15j [113] 16a [117,118] 16b [117,118] 16c [117,118] 16d [117,118] 16e [117,118] 16f [119] 16g [119] 16h [119] 16i [119] 16j [119] 16k [119] 16l [120] 16m [120] 16n [115] 16o [120] 16p [120] 16q [115] 17a [125–128] 17b [125–128] 17c [125–128] 17d [125–128] 17e [125–128] 17f [125–128] 17g [125–128] 17h [129,130] 17i [129,130] 17j [129,130] 17k [129,130] 18a [131–133] 18b [134] 18c [134] 18d [134] 18e [134] 18f [134] 18g [134] 18h [134] 18i [134] 18j [134]

Deposition methodb

Mobility (cm2V–2s–1)

ION/IOFF

v v v v v v v v v v v v v v v v s v v s sc sc sc sc sc sc sc v v v v v v v v v v v v v v

0.30 0.054 1 × 10–6 N.O. 0.02 0.003 0.002 N.O. N.O. N.O. 0.0007 N.O. N.O. 6 × 10–9 0.011 3.5 × 10–4 1 × 10–4 1 × 10–5 2 × 10–5 0.007 0.4 0.015 1.4 0.0018 1.4 × 10–4 0.062 0.0012 0.06 3.3 × 10–5 0.42 0.2 0.2 3 × 10–4 N.O. 2 × 10–6 N.O. 8 × 10–5 N.O. 6 × 10–7 N.O. 2 × 10–4

107 106 103 N.O. 104 104 104 N.O. N.O. N.O. 102 N.O. N.O. 102 104 104 8 × 10 104 104 104 N.A. N.A. N.A. N.A. N.A. N.A. N.A. 104 105 6 × 103 106 108 104 N.O. 50 N.O. 104 N.O. 103 N.O. 104

8080_C003.fm Page 207 Friday, April 6, 2007 10:24 AM

Design, Synthesis, and Transistor Performance of Organic Semiconductors

TABLE 3.1.1 (Continued) Field-Effect Mobility and On/Off Ratio Found in the Literature for Each Organic Semiconductor Described in the Text Molecule[ref.] 19a [135] 19b [136] 19c [136] 19c [137] 19d [136] 19e [138] 19f [138] 20a [142] 20b [142] 20c [142] 20d [142] 20e [146] 20f [146] 21a [143] 21b [144,145] 21c [144,145] 21d [144,145] 22a [147] 22b [147] 22c [147] 23a [148] 23b [148] 23c [150] 23d [149] 23e [149] 24a [152] 24b [153] 24c [153] 24d [154] 24d [154] 24e [154] 24f [155] 24g [155] 25a [159] 25a [160] 25b [11] 25c [11] 25d [11] 25e [11] 25f [11]

Deposition methodb

Mobility (cm2V–2s–1)

v v v sc v v v s s s s v v s s s s v v v v v v v v v v v v s v v v v sc v v v v v

0.0036 0.081 0.17 1.5 0.0073 N.A. 0.3 1 × 10–4 1 × 10–4 1 × 10–4 3 × 10–4 1.1 × 10–2 10–5 2 × 10–4 0.00103 6.5 × 10–4 2.2 × 10–4 0.01 0.04 0.07 0.3 4.3 × 10–4 8 × 10–4 0.0019 0.0019 0.12 0.13 0.13 0.012 0.0014 0.0014 0.055 1.1 × 10–6 0.02 1.0 0.0034 0.0026 0.0028 6.9 × 10–4 1.5 × 10–4

ION/IOFF N.A. 2 × 103 105 104 2 × 103 N.A. 106 104 105 103 105 2 × 102 N.A. 102 103 102 102 105 106 106 105 N.A. 10 N.A. N.A. 106 5.8 × 105 9.4 × 102 N.A. 1.5 × 102 N.A. N.A. N.A. 4 × 105 104 N.A. N.A. N.A. N.A. N.A. Continued

207

8080_C003.fm Page 208 Friday, April 6, 2007 10:24 AM

208

Organic Field-Effect Transistors

TABLE 3.1.1 (Continued) Field-Effect Mobility and On/Off Ratio Found in the Literature for Each Organic Semiconductor Described in the Text Molecule[ref.] 25g [11] 25h [162] 25i [162] 25j [164] 25k [163] 25l [163] 25m [165] 25n [165] 25o [165] 25p [167] 25q [168] 25r [169] 26a [170] 26b [170] 26c [171] 27a [39] 27b [150] 27c [105] 27c [176] 27d [105] 28a [300] 28b [105] 28c [182] 28c [182] 28d [177] 28e [39] 28f [150] 28g [105] 28h [39] 28i [105] 28j [105] 28k [105] 28l [186] 28m [186] 28n [181] 28o [177] 28p [177] 28q [39] 28r [39] 28s [301] 28t [184]

Deposition methodb

Mobility (cm2V–2s–1)

ION/IOFF

v s s s s s s s s s s s v v s v v v sc v v v v s v s v v v v v v v v v v v v v v v

5.4 × 10–5 4.0 × 10–4 4.8 × 10–5 1 × 10–6 6.4 × 10–4 0.0017 0.60 0.40 0.24 2.2 × 10–4 0.012 0.68 0.013 2.0 × 10–5 10–4–10–3 0.014 0.078 0.07 0.1 0.2 0.23 0.2 0.038 0.06 3.2 × 10–5 4 × 10–4 0.0092 0.5 0.054 1.1 1 0.5 0.016 0.0013 0.03 N.A. 0.0048 0.012 0.064 10–4 8 × 10–6

N.A. N.A. N.A. N.A. 5.5 × 104 100 N.A. N.A. N.A. 104–105 N.A. 8 × 104 10 100 102–103 20 102 102 N.A. N.A. N.A. 105 4 × 105 105 N.A. 4 × 104 103 105 104 104 104 105 N.A. N.A. N.A. N.A. N.A. 4 × 105 7 × 103 5 × 104 103

8080_C003.fm Page 209 Friday, April 6, 2007 10:24 AM

Design, Synthesis, and Transistor Performance of Organic Semiconductors

TABLE 3.1.1 (Continued) Field-Effect Mobility and On/Off Ratio Found in the Literature for Each Organic Semiconductor Described in the Text Molecule[ref.] 28u [184] 28v [184] 28w [177] 28x [185] 29a [186] 29a [186] 29b [186] 29b [186] 29c [186] 29c [186] 29d [187] 29e [188] 29f [189] 29g [189] 29h [189] 29i [189] 29 [189] 29k [189] 29l [189] 30a [190] 30b [190] 30c [190] 30d [190] 30e [190] 30f [182] 30g [191] 30h [191] 30i [192] 30j [192] 30k [192] 30l [192] 30m [192] 30n [192] 30o [192] 30p [192] 30q [192] 30r [193] 30s [193] 30t [193] 30u [198]

Deposition methodb

Mobility (cm2V–2s–1)

v v v s v s v s v s s s v v v v v v v v v v v v v v v v v v v v v v v v v v v v

3 × 10–4 4 × 10–5 3.9 × 10–5 0.012 0.008 0.003 0.033 0.01 0.009 0.001 0.0049 10–4 N.O. N.O. 6 × 10–6 N.O. 7 × 10–4 6 × 10–4 0.014 0.012 0.14 0.025 0.023 0.08 0.17 N.O. N.O. 1.1 × 10–5 1.2 × 10–5 1.7 × 10–5 1.1 × 10–3 5 × 10–5 5.4 × 10–5 2.9 × 10–5 7.5 × 10–4 5 × 10–4 3 × 10–3 2 × 10–4 10–5 0.01

ION/IOFF 103 103 N.A. 105 N.A. N.A. N.A. N.A. N.A. N.A. 1.6 × 104 103 N.O. N.O. N.O. N.O. 102 10 103 103 2 × 104 102 102 104 8 × 105 N.O. N.O. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. 106 104 104 104 Continued

209

8080_C003.fm Page 210 Friday, April 6, 2007 10:24 AM

210

Organic Field-Effect Transistors

TABLE 3.1.1 (Continued) Field-Effect Mobility and On/Off Ratio Found in the Literature for Each Organic Semiconductor Described in the Text Molecule[ref.] 30v [198] 31a [196] 31b [196] 31c [194] 31d [194] 31d [195] 31e [196] 31f [201] 31f′ [200] 31g [201] 31h [115] 31i [197] 31j [194] 31k [196] 31l [115] 31m [197] 31n [194] 31o [196] 31p [197] 31q [198] 31r [198] 31s [199] 31t [115] 31u [200] 31v [115] 31w [196] 31x [201] 31y [198] 31z [201] 32a [202] 32b [202] 33a [203] 33b [203] 33c [203] 33d [203] 33e [203] 33f [203] 34a [202] 34b [198] 34c [204] 35b [208]

Deposition methodb

Mobility (cm2V–2s–1)

ION/IOFF

v s s s v sc s v v v v v v s s v v s v s v v v v s s v v v v v v v v v v v v v v s

0.002 0.01 0.018 0.0014 0.0077 0.66 0.09 0.04 0.4 0.005 0.003 0.03 0.17 0.054 0.033 0.03 0.055 0.09 0.02 5 × 10–4 0.042 0.3 5 × 10–4 0.08 0.02 0.054 0.03 0.049 4 × 10–5 0.02 0.011 4.6 × 10–4 0.042 0.014 0.0049 0.012 0.0097 0.011 0.067 0.012 0.12

106 3 × 103 1.5 × 104 104 N.A. 105 4 × 104 106 105 104 4 × 103 107 N.A. 7.7 × 103 1.7 × 103 107 N.A. 3.5 × 103 107 5.9 × 102 104 105 590 103 5 × 104 4 × 104 105 104 104 1.3 × 102 3 × 102 N.A. N.A. N.A. N.A. N.A. N.A. 102 104 N.A. 106

8080_C003.fm Page 211 Friday, April 6, 2007 10:24 AM

Design, Synthesis, and Transistor Performance of Organic Semiconductors

TABLE 3.1.1 (Continued) Field-Effect Mobility and On/Off Ratio Found in the Literature for Each Organic Semiconductor Described in the Text Molecule[ref.] 35d [210] 35e [210] 35f [210] 35g [210] 35h [211] 35i [211] 35j [211] 35k [212] 35l [212] 35m [213] 35n [214] 35o [214] 35p [215,219] 35q [215,219] 35r [215,219] 35s [217] 35t [216] 35u [218] 35v [220] 35w [220] 35x [220] 35y [220] 35z [221] 35z′ [221] 35z″ [221] 36a [222,223] 36b [224] 36c [225] 37a [227] 37b [226] 38a [229,230] 38b [229,230] 38c [229,230] 39a [232] 39b [232] 39c [232] 40a [233–235] 40b [233–235] 40c [233–235] 40d [234]

Deposition methodb

Mobility (cm2V–2s–1)

s s s s s s s s s s s s s s s s s s s s s s s s s s s s s s s s s s s s s s s s

0.0012 3 × 10–4 8.5 × 10–5 2.4 × 10–5 0.001 2.8 × 10–5 2.9 × 10–4 0.07 0.0063 0.002 6 × 10–4 0.03 N.A. 0.15 0.12 0.03 0.01 0.14 3.4 × 10–5 4.6 × 10–5 6.9 × 10–5 6.7 × 10–5 0.3 0.3 0.63 0.02 0.0011 0.003 0.0025 3.4 × 10–4 5 × 10–6 1 × 10–7 7.9 × 10–5 4 × 10–4 0.001 0.002 1 × 10–4 0.001 0.01 4 × 10–4

ION/IOFF 6.4 × 102 40 1 × 103 3.7 × 102 60 2 5 1 × 105 1 × 104 N.A. 4 × 104 2 × 102 N.A. 1 × 105 1 × 105 1 × 106 1 × 105 2 × 107 103 103 103 103 106 106 107 N.A. 1.4 × 104 104 2 × 102 104 N.A. N.A. N.A. 104 104 106 N.A. N.A. N.A. N.A. Continued

211

8080_C003.fm Page 212 Friday, April 6, 2007 10:24 AM

212

Organic Field-Effect Transistors

TABLE 3.1.1 (Continued) Field-Effect Mobility and On/Off Ratio Found in the Literature for Each Organic Semiconductor Described in the Text Molecule[ref.] 40e [234] 41a [236] 41b [236] 41c [236] 41d [236] 41e [236] 41f [237] 42a [140,141] 43a [238] 43b [242,243] 43c [245] 43d [254,255] 43e [256] 44a [257] 44b [257] 44c [257] 45a [261] 45b [263] 45b′ [264] 45c [265] 45d [265] 45e [265] 45f [265] 45g [265] 45h [266] 45i [266] 45j [267] 45k [267] 46a [268] 46c [269] 46d [270] 46e [270] 46f [269] 46g [269] 46h [270] 46i [271] 46j [271] 46k [271] 46l [271] 47a [272] 47b [273]

Deposition methodb

Mobility (cm2V–2s–1)

ION/IOFF

s s s s s s s s p p p p v p p p v s s v v v v v s s v v v v v v v v v v v v v v v

1 × 10–3 0.9 × 10–5 0.6 × 10–5 3 × 10–4 2 × 10–4 6 × 10–4 8 × 10–5 0.01 1 × 10–4 0.22 0.017 0.1 0.01 0.0013 7.8 × 10–7 N.A. 0.56 0.2 0.0017 0.03 1.2 × 10–3 4.5 × 10–5 2.1 × 10–3 2.7 × 10–5 3 × 10–4 3 × 10–4 10–4 10–4 0.003 0.16 0.01 0.005 0.03 0.06 0.12 3 × 10–4 2 × 10–4 N.O. 10–5 N.O. 10–4

N.A. 102 102 104 104 104 103 N.A. N.A. N.A. 105 105 N.A. 104 N.A. N.A. 108 1,000 N. A. 3 × 105 N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. 4 × 105 N.A. 105 105 N.A. N.A. N.A. N.A. N.A. N.A. N.A.

8080_C003.fm Page 213 Friday, April 6, 2007 10:24 AM

Design, Synthesis, and Transistor Performance of Organic Semiconductors

TABLE 3.1.1 (Continued) Field-Effect Mobility and On/Off Ratio Found in the Literature for Each Organic Semiconductor Described in the Text Molecule[ref.] 47c [277] 47d [277] 47e [277] 47f [275] 47g [278] 47h [182] 47i [279] 47i [279] 47j [279] 47j [279] 47k [280] 48a [281] 48b [281] 48c [282] 48d [282] 48e [282] 48f [282] 48g [282] 48h [282] 48i [284] 49a [39] 49b [39] 49c [39] 49d [287] 49e [287] 49f [287] 49g [287] 49h [288] 49i [288] 49j [288] 49l [289] 49m [290] 49m [290] 49n [290] 49o [72] 49p [201] 49q [201] 50a [291] 50b [291] 50c [291]

Deposition methodb

Mobility (cm2V–2s–1)

v v v v v v v s v s v v v v v v v v v v v v v v v v v v v v v v s v v v v v v v

0.06 1.3 0.5 0.6 1.5 × 10–5 1.9 × 10–4 0.1 10–3–10–5 0.64 10–3–10–4 0.14 10–5 0.001 3.6 × 10–6 1.0 × 10–8 2.1 × 10–7 2.5 × 10–7 2.2 × 10–6 9.6 × 10–7 0.2 0.059 0.026 0.001 0.0002 0.074 0.0039 0.0025 0.1 0.6 10–2–10–3 0.08 0.45 0.21 0.043 0.11 0.003 0.00001 0.18 N.O. 0.3

ION/IOFF 106 106 106 107 N.A. 100 105 N.A. 104 N.A. 1.2 × 103 N.A. N.A. 103 10 102 102 102 102 >106 105 105 104 1 × 103 6 × 106 5 × 105 8 × 106 107 107 N.A. 105 108 105 106 106 104 100 3 × 105 N.A. 106 Continued

213

8080_C003.fm Page 214 Friday, April 6, 2007 10:24 AM

214

Organic Field-Effect Transistors

TABLE 3.1.1 (Continued) Field-Effect Mobility and On/Off Ratio Found in the Literature for Each Organic Semiconductor Described in the Text Molecule[ref.] 50d [292] 50e [293] 50f [293] 50g [293] 50h [293] 50i [293] 50j [293] 51a [294] 51b [294] 51c [297] 51d [298] 51e [299]

Deposition methodb

Mobility (cm2V–2s–1)

ION/IOFF

v v v v v v v s s s s s

3 × 10–3 N.O. 1.83 0.0028 0.085 0.018 0.025 0.1 10–6 0.5–0.7 4.8 × 10–3 2.2 × 10–8

104 N.A. N.A. N.A. N.A. N.A. N.A. 2 × 103 N.A. 2–50 N.A. N.A.

a

Highest field-effect mobility measured for each organic semiconductor as reported in literature. b sc: single crystal; v: vaccum; s: solution; p: precursor. c N.A.: transistor effect “not available.” d N.O.: transistor effect “not observed.”

REFERENCES [1] Kahng, D. and Atalla, M.M., Silicon-Silicon Dioxide Field Induced Surface Devices, in IRE Solid-State Devices Research Conference, Carnegie Institute of Technology, Pittsburgh, PA, 1960. [2] Shaw, J.M. and Seidler, P.F., Organic electronics: Introduction, IBM J. Res. Dev. 45 (1), 3-9, 2001. [3] Koezuka, H., Tsumura, A., and Ando, T., Field-Effect Transistor with Polythiophene ThinFilm, Synth. Met. 18 (1-3), 699-704, 1987. [4] Reese, C. et al., Organic thin film transistors, Mater. Today 7 (9), 20-27, 2004. [5] Bao, Z.N., Rogers, J.A., and Katz, H.E., Printable organic and polymeric semiconducting materials and devices, J. Mater. Chem. 9 (9), 1895-1904, 1999. [6] Kawase, T. et al., Inkjet printing of polymer thin film transistors, Thin Solid Films 438, 279-287, 2003. [7] Sirringhaus, H. et al., High-resolution inkjet printing of all-polymer transistor circuits, Science 290, 2123, 2000. [8] Kawase, T. et al., Inkjet printed via-hole interconnections and resistors for all-polymer transistor circuits, Adv. Mater. 13 (21), 1601, 2001. [9] Arias, A.C. et al., All jet-printed polymer thin-film transistor active-matrix backplanes, Appl. Phys. Lett. 85 (15), 3304-3306, 2004.

8080_C003.fm Page 215 Friday, April 6, 2007 10:24 AM

Design, Synthesis, and Transistor Performance of Organic Semiconductors

215

[10] Chabinyc, M.L. and Salleo, A., Materials requirements and fabrication of active matrix arrays of organic thin-film transistors for displays, Chem. Mater. 16 (23), 4509-4521, 2004. [11] Bao, Z.N., Lovinger, A.J., and Dodabalapur, A., Highly ordered vacuum-deposited thin films of metallophthalocyanines and their applications in field-effect transistors, Adv. Mater. 9 (1), 42-&, 1997. [12] Garnier, F. et al., All-Polymer Field-Effect Transistor Realized by Printing Techniques, Science 265 (5179), 1684-1686, 1994. [13] Gates, B.D. et al., New approaches to nanofabrication: Molding, printing, and other techniques, Chem. Rev. 105 (4), 1171-1196, 2005. [14] Xia, Y.N. and Whitesides, G.M., Soft lithography, Angew. Chem. Int. Ed. 37 (5), 551575, 1998. [15] Dodabalapur, A. et al., Complementary circuits with organic transistors, Appl. Phys. Lett. 69 (27), 4227-4229, 1996. [16] Crone, B.K. et al., Design and fabrication of organic complementary circuits, J. Appl. Phys. 89 (9), 5125-5132, 2001. [17] Drury, C.J. et al., Low-cost all-polymer integrated circuits, Appl. Phys. Lett. 73 (1), 108110, 1998. [18] Rogers, J.A. et al., Paper-like electronic displays: Large-area rubber-stamped plastic sheets of electronics and microencapsulated electrophoretic inks, Proc. Natl. Acad. Sci. 98 (9), 4835-4840, 2001. [19] Wisnieff, R., Display technology — Printing screens, Nature 394 (6690), 225, 1998. [20] Comiskey, B. et al., An electrophoretic ink for all-printed reflective electronic displays, Nature 394 (6690), 253-255, 1998. [21] Gelinck, G.H. et al., Flexible active-matrix displays and shift registers based on solutionprocessed organic transistors, Nat. Mater. 3 (2), 106-110, 2004. [22] Zhu, F.R. et al., Toward novel flexible display — Top-emitting OLEDs on Al-laminated PET substrates, Proc. IEEE 93 (8), 1440-1446, 2005. [23] Carts-Powell, Y., Flat-panel displays — organic electronics support rollable display, Laser Focus World 41 (8), 40, 2005. [24] Brabec, C.J., Sariciftci, N.S., and Hummelen, J.C., Plastic solar cells, Adv. Funct. Mater. 11 (1), 15-26, 2001. [25] Coakley, K.M. and McGehee, M.D., Conjugated polymer photovoltaic cells, Chem. Mater. 16 (23), 4533-4542, 2004. [26] Kline, R.J. et al., Controlling the field-effect mobility of regioregular polythiophene by changing the molecular weight, Adv. Mater. 15 (18), 1519, 2003. [27] Crone, B. et al., Electronic sensing of vapors with organic transistors, Appl. Phys. Lett. 78 (15), 2229-2231, 2001. [28] Torsi, L. et al., Correlation between oligothiophene thin film transistor morphology and vapor responses, J. Phys. Chem. B 106 (48), 12563-12568, 2002. [29] Crone, B.K. et al., Organic oscillator and adaptive amplifier circuits for chemical vapor sensing, J. Appl. Phys. 91 (12), 10140-10146, 2002. [30] Zhu, Z.T. et al., Humidity sensors based on pentacene thin-film transistors, Appl. Phys. Lett. 81 (24), 4643-4645, 2002. [31] Li, D.W. et al., Humidity effect on electrical performance of organic thin-film transistors, Appl. Phys. Lett. 86 (4), 42105, 2005. [32] Baude, P.F. et al., Pentacene-based radio-frequency identification circuitry, Appl. Phys. Lett. 82 (22), 3964-3966, 2003. [33] Jurchescu, O.D., Baas, J., and Palstra, T.T.M., Effect of impurities on the mobility of single crystal pentacene, Appl. Phys. Lett. 84 (16), 3061-3063, 2004.

8080_C003.fm Page 216 Friday, April 6, 2007 10:24 AM

216

Organic Field-Effect Transistors

[34] Sundar, V.C. et al., Elastomeric transistor stamps: Reversible probing of charge transport in organic crystals, Science 303 (5664), 1644-1646, 2004. [35] Bernius, M.T. et al., Progress with light-emitting polymers, Adv. Mater. 12 (23), 17371750, 2000. [36] Friend, R.H. et al., Electroluminescence in conjugated polymers, Nature 397 (6715), 121-128, 1999. [37] Kodak Corporation www.kodak.com, 2005. [38] Katz, H.E., Bao, Z.N., and Gilat, S.L., Synthetic chemistry for ultrapure, processable, and high-mobility organic transistor semiconductors, Acc. Chem. Res. 34 (5), 359369, 2001. [39] Facchetti, A. et al., Building blocks for n-type molecular and polymeric electronics. Perfluoroalkyl-versus alkyl-functionalized oligothiophenes (nT; n=2-6). Systematics of thin film microstructure, semiconductor performance, and modeling of majority charge injection in field-effect transistors, J. Am. Chem. Soc. 126 (42), 13859-13874, 2004. [40] Newman, C.R. et al., Introduction to organic thin film transistors and design of n-channel organic semiconductors, Chem. Mater. 16 (23), 4436-4451, 2004. [41] Ling, M.M. and Bao, Z.N., Thin film deposition, patterning, and printing in organic thin film transistors, Chem. Mater. 16 (23), 4824-4840, 2004. [42] Fritz, S.E. et al., Structural characterization of a pentacene monolayer on an amorphous SiO2 substrate with grazing incidence X-ray diffraction, J. Am. Chem. Soc. 126 (13), 4084-4085, 2004. [43] Ruiz, R. et al., Pentacene thin film growth, Chem. Mater. 16 (23), 4497-4508, 2004. [44] Tsumura, A., Koezuka, H., and Ando, T., Polythiophene Field-Effect Transistor — Its Characteristics and Operation Mechanism, Synth. Met. 25 (1), 11-23, 1988. [45] Bendikov, M., Wudl, F., and Perepichka, D.F., Tetrathiafulvalenes, oligoacenenes, and their buckminsterfullerene derivatives: The brick and mortar of organic electronics, Chem. Rev. 104 (11), 4891-4945, 2004. [46] Perepichka, D.F. et al., A one-step synthesis of a poly(iptycene) through an unusual Diels-Alder cyclization/dechlorination of tetrachloropentacene, J. Am. Chem. Soc. 125 (34), 10190-10191, 2003. [47] Roberson, L.B. et al., Pentacene disproportionation during sublimation for field-effect transistors, J. Am. Chem. Soc. 127 (9), 3069-3075, 2005. [48] Aleshin, A.N. et al., Mobility studies of field-effect transistor structures based on anthracene single crystals, Appl. Phys. Lett. 84 (26), 5383-5385, 2004. [49] Ito, K. et al., Oligo(2,6-anthrylene)s: Acene-oligomer approach for organic field-effect transistors, Angew. Chem. Int. Ed. 42 (10), 1159-1162, 2003. [50] Inoue, Y. et al., Organic thin-film transistors based on anthracene oligomers, J. Appl. Phys. 95 (10), 5795-5799, 2004. [51] Garnier, F. et al., Dihexylquaterthiophene, a two-dimensional liquid crystal-like organic semiconductor with high transport properties, Chem. Mater. 10 (11), 3334-3339, 1998. [52] Ackermann, J. et al., Effect of end-substitution of hexyl chains on the growth and electrical properties of quaterthiophene thin films, Appl. Surf. Sci. 212, 26-32, 2003. [53] Meng, H. et al., High-performance, stable organic thin-film field-effect transistors based on bis-5 '-alkylthiophen-2 '-yl-2,6-anthracene semiconductors, J. Am. Chem. Soc. 127 (8), 2406-2407, 2005. [54] Garnier, F. et al., Molecular Engineering of Organic Semiconductors — Design of SelfAssembly Properties in Conjugated Thiophene Oligomers, J. Am. Chem. Soc. 115 (19), 8716-8721, 1993.

8080_C003.fm Page 217 Friday, April 6, 2007 10:24 AM

Design, Synthesis, and Transistor Performance of Organic Semiconductors

217

[55] Merlo, J.A. et al., p-channel organic semiconductors based on hybrid acene-thiophene molecules for thin-film transistor applications, J. Am. Chem. Soc. 127 (11), 39974009, 2005. [56] Gerlach, C.P., Kelley, T.W., and Muyres, D.V., New, High-Performance Acenyl Semiconductors, in 227th ACS National Meeting, Anaheim, CA, 2004. [57] Goldmann, C. et al., Hole mobility in organic single crystals measured by a "flip-crystal" field-effect technique, J. Appl. Phys. 96 (4), 2080-2086, 2004. [58] Gundlach, D.J. et al., Thin-film transistors based on well-ordered thermally evaporated naphthacene films, Appl. Phys. Lett. 80 (16), 2925-2927, 2002. [59] Sarma, J.A.R.P. and Desiraju, G.R., The Role of Cl=Cl and C-H=O Interactions in the Crystal Engineering of 4-a Short-Axis Structures, Acc. Chem. Res. 19 (7), 222-228, 1986. [60] Moon, H. et al., Synthesis, crystal structure, and transistor performance of tetracene derivatives, J. Am. Chem. Soc. 126 (47), 15322-15323, 2004. [61] Kelley, T.W. et al., High-performance OTFTs using surface-modified alumina dielectrics, J. Phys. Chem. B 107 (24), 5877-5881, 2003. [62] Kelley, T.W. et al., High Performance Organic Thin Film Transistors, in Mater. Res. Soc. Symp. Proc., 2003, pp. L6.5.1. [63] Lin, Y.Y. et al., Pentacene-based organic thin-film transistors, IEEE Trans. Elect. Dev. 44 (8), 1325-1331, 1997. [64] Klauk, H. et al., High-mobility polymer gate dielectric pentacene thin film transistors, J. Appl. Phys. 92 (9), 5259-5263, 2002. [65] Knipp, D. et al., Pentacene thin film transistors on inorganic dielectrics: Morphology, structural properties, and electronic transport, J. Appl. Phys. 93 (1), 347-355, 2003. [66] Yang, S.Y., Shin, K., and Park, C.E., Adv. Funct. Mater. 15, 1806-1814, 2005. [67] Meng, H. et al., Tetramethylpentacene: Remarkable absence of steric effect on field effect mobility, Adv. Mater. 15 (13), 1090-+, 2003. [68] Smith, T.P., Vogel, D.E., and Vogel, K.M., 2003, Preparation of dialkylpentacenes as semiconducor materials. [69] Sheraw, C.D. et al., Functionalized pentacene active layer organic thin-film transistors, Adv. Mater. 15 (23), 2009-2011, 2003. [70] Anthony, J.E. et al., Functionalized pentacene: Improved electronic properties from control of solid-state order, J. Am. Chem. Soc. 123 (38), 9482-9483, 2001. [71] Swartz, C.R. et al., Synthesis and characterization of electron-deficient pentacenes, Org. Lett. 7 (15), 3163-3166, 2005. [72] Sakamoto, Y. et al., Perfluoropentacene: High-performance p-n junctions and complementary circuits with pentacene, J. Am. Chem. Soc. 126 (26), 8138-8140, 2004. [73] Sakamoto, Y., Komatsu, S., and Suzuki, T., Tetradecafluorosexithiophene: The first perfluorinated oligothiophene, J. Am. Chem. Soc. 123 (19), 4643-4644, 2001. [74] Heidenhain, S.B. et al., Perfluorinated oligo(p-phenylene)s: Efficient n-type semiconductors for organic light-emitting diodes, J. Am. Chem. Soc. 122 (41), 10240-10241, 2000. [75] Ponzini, F. et al., Phenyl/pentafluorophenyl interactions and the generation of ordered mixed crystals: sym-triphenethynylbenzene and sym-tris(perfluorophenethynyl)benzene, Angew. Chem. Int. Ed. 39 (13), 2323-2325, 2000. [76] Weck, M. et al., Influence of perfluoroarene-arene interactions on the phase behavior of liquid crystalline and polymeric materials, Angew. Chem. Int. Ed. 38 (18), 2741-2745, 1999. [77] Gillard, R.E. et al., A novel fluorine-containing 2 catenane, J. Org. Chem. 61 (14), 45044505, 1996.

8080_C003.fm Page 218 Friday, April 6, 2007 10:24 AM

218

Organic Field-Effect Transistors

[78] M. M. Payne, S. R. Parkin, and Anthony, J.E., Functionalized higher acenes: hexacene and heptacene, J. Am. Chem. Soc. 127, 8028-8029, 2005. [79] Miao, Q. et al., Self-assembly and electronics of dipolar linear acenes, Adv. Mater. 17 (4), 407, 2005. [80] Hunter, C.A. and Sanders, J.K.M., The Nature of Pi-Pi Interactions, J. Am. Chem. Soc. 112 (14), 5525-5534, 1990. [81] Endres, R.G. et al., Structural and electronic properties of pentacene molecule and molecular pentacene solid, Comput. Mat. Sci. 29 (3), 362-370, 2004. [82] Haemori, M. et al., Fabrication of highly oriented rubrene thin films by the use of atomically finished substrate and pentacene buffer layer, Jpn. J. Appl. Phys. Part 1 44 (6A), 3740-3742, 2005. [83] Stingelin-Stutzmann, N. et al., Organic thin-film electronics from vitreous solutionprocessed rubrene hypereutectics, Nat. Mater. 4 (8), 601-606, 2005. [84] Takahashi, T. et al., Ambipolar organic field-effect transistors based on rubrene single crystals, Appl. Phys. Lett. 88 (3), -, 2006. [85] T. Mori, H. Takeuchi, and Fujikawa, H., Field-effect transistors based on a polycyclic aromatic hydrocarbon core as a two-dimensional conductor, J. Appl. Phys. 97, 066102-1-3, 2005. [86] Shklyarevskiy, I.O. et al., High Anisotropy of the Field-Effect Transistor Mobility in Magnetically Aligned Discotic Liquid-Crystalline Semiconductors, J. Am. Chem. Soc. 127, 16233, 2005. [87] Xiao, S. et al., Molecular wires from contorted aromatic compounds, Angew. Chem. Int. Ed. 44, 7390-7394, 2005. [88] Miao, Q. et al., Synthesis, assembly, and thin film transistors of dihydrodiazapentacene: An isostructural motif for pentacene, J. Am. Chem. Soc. 125 (34), 10284-10287, 2003. [89] Chen, T.-R., Yeh, A., and Chen, J.-D., A new imidazolylquinoline for organic thin film transistor, Tet. Lett. 46, 1569-1571, 2005. [90] Chen, T.-R., Synthesis, structure and field-effect property of 2-(benzimidazol-2-yl) quinoline, Mater. Lett. 59, 1050, 2005. [91] Laquindanum, J.G., Katz, H.E., and Lovinger, A.J., Synthesis, morphology, and fieldeffect mobility of anthradithiophenes, J. Am. Chem. Soc. 120 (4), 664-672, 1998. [92] Payne, M.M. et al., Organic field-effect transistors from solution-deposited functionalized acenes with mobilities as high as 1 cm(2)/V-s, J. Am. Chem. Soc. 127 (14), 49864987, 2005. [93] J.G. Laquindanum et al., Benzodithiophene rings as semiconductor building blocks, Adv. Mater. 9 (1), 36-39, 1997. [94] H. Sirringhaus et al., Dibenzothienobisbenzothiophene-a novel fused-ring oligomer with high field-effect mobility, J. Mater. Chem. 9, 2095-2101, 1999. [95] Wex, B. et al., New organic semiconductors and their device performance as a function of thiophene orientation, J. Mater. Chem. 16 (12), 1121-1124, 2006. [96] Li, X.C. et al., A highly pi-stacked organic semiconductor for thin film transistors based on fused thiophenes, J. Am. Chem. Soc. 120 (9), 2206-2207, 1998. [97] Sirringhaus, H. et al., Bis(dithienothiophene) organic field-effect transistors with a high ON/OFF ratio, Appl. Phys. Lett. 71 (26), 3871-3873, 1997. [98] Cornil, J. et al., Charge transport versus optical properties in semiconducting crystalline organic thin films, Adv. Mater. 12 (13), 978, 2000. [99] Xiao, K. et al., A Highly -Stacked Organic Semiconductor for Field-Effect Transistors Based on Linearly Condensed Pentathienoacene, J. Am. Chem. Soc. 127 (38), 1328113286, 2005.

8080_C003.fm Page 219 Friday, April 6, 2007 10:24 AM

Design, Synthesis, and Transistor Performance of Organic Semiconductors

219

[100] Iosip, M.D. et al., New dithieno[3,2-b : 2 ',3 '-d]thiophene oligomers as promising materials for organic field-effect transistor applications, Synth. Met. 146 (3), 251-257, 2004. [101] Sun, Y.M. et al., High-performance and stable organic thin-film transistors based on fused thiophenes, Advanced Functional Materials 16 (3), 426-432, 2006. [102] Noh, Y.Y. et al., Organic field effect transistors based on biphenyl, fluorene end-capped fused bithiophene oligomers, Chem. Mater. 17 (15), 3861-3870, 2005. [103] Kunugi, Y. et al., Organic field-effect transistors using di(2-thienyl)naphthodithiophenes as active layers, Chem. Lett. (10), 958-959, 2002. [104] Takimiya, K. et al., 2,6-diarylnaphtho 1,8-bc : 5,4-b ' c ' dithiophenes as new highperformance semiconductors for organic field-effect transistors, J. Am. Chem. Soc. 127 (10), 3605-3612, 2005. [105] Halik, M. et al., Relationship between molecular structure and electrical performance of oligothiophene organic thin film transistors, Adv. Mater. 15 (11), 917, 2003. [106] Zhang, H.J. et al., Novel butterfly pyrene-based organic semiconductors for field effect transistors, Chem. Comm. (7), 755-757, 2006. [107] Saragi, T.P.I., Fuhrmann-Lieker, T., and Salbeck, J., High ON/OFF ratio and stability of amorphous organic field-effect transistors based on spiro-linked compounds, Synth. Met. 148 (3), 267-270, 2005. [108] Saragi, T.P., Fuhrmann-Lieker, T., and Salbeck, J., Comparison of charge-carrier transport in thin films of spiro-linked compounds and their corresponding parent compounds, Adv. Funct. Mater. 16 (7), 966-974, 2006. [109] Wu, Y.L. et al., Indolo 3,2-b carbazole-based thin-film transistors with high mobility and stability, J. Am. Chem. Soc. 127 (2), 614-618, 2005. [110] Wakim, S. et al., Organic microelectronics: Design, synthesis, and characterization of 6,12-dimethylindolo 3,2-b carbazoles, Chem. Mater. 16 (23), 4386-4388, 2004. [111] Li, Y.N. et al., Novel peripherally substituted indolo 3,2-b carbazoles for high-mobility organic thin-film transistors, Adv. Mater. 17 (7), 849, 2005. [112] Drolet, N. et al., 2,7-carbazolenevinylene-based oligomer thin-film transistors: high mobility through structural ordering, Adv. Funct. Mater. 15 (10), 1671, 2005. [113] Morin, J.F. et al., Syntheses and characterization of electroactive and photoactive 2,7carbazolenevinylene-based conjugated oligomers and polymers, Chem. Mater. 16 (23), 4619-4626, 2004. [114] Mitschke, U., Debaerdemaeker, T., and Bauerle, P., Structure-property relationships in mixed oligoheterocycles based on end-capped oligothiophenes, Eur. J. Org. Chem. (3), 425-437, 2000. [115] Hong, X.M. et al., Thiophene-phenylene and thiophene-thiazole oligomeric semiconductors with high field-effect transistor on/off ratios, Chem. Mater. 13 (12), 46864691, 2001. [116] Koren, A.B. et al., Intermolecular interactions in X-stacked conjugated molecules. Synthesis, structure, and spectral characterization of alkyl bithiazole oligomers, J. Am. Chem. Soc. 125 (17), 5040-5050, 2003. [117] Ando, S. et al., Characterization and field-effect transistor performance of heterocyclic oligomers containing a thiazolothiazole unit, Chem. Lett. 33 (9), 1170-1171, 2004. [118] Ando, S. et al., Synthesis, physical properties, and field-effect transistors of novel thiophene/thiazolothiazole co-oligomers, J. Mater. Chem. 14 (12), 1787-1790, 2004. [119] Ando, S. et al., Physical properties and field-effect transistors based on novel thiazolothiazole/heterocyclic and thiazolothiazole/phenylene co-oligomers, Synth. Met. 156 (2-4), 327-331, 2006.

8080_C003.fm Page 220 Friday, April 6, 2007 10:24 AM

220

Organic Field-Effect Transistors

[120] Li, W.J. et al., Field-effect transistors based on thiophene hexamer analogues with diminished electron donor strength, Chem. Mater. 11 (2), 458-465, 1999. [121] Segura, J.L. and Martin, N., New concepts in tetrathiafulvalene chemistry, Angew. Chem. Int. Ed. 40 (8), 1372-1409, 2001. [122] Nielsen, M.B., Lomholt, C., and Becher, J., Tetrathiafulvalenes as building blocks in supramolecular chemistry II, Chem. Soc. Rev. 29 (3), 153-164, 2000. [123] Bryce, M.R., Functionalised tetrathiafulvalenes: new applications as versatile pi-electron systems in materials chemistry, J. Mater. Chem. 10 (3), 589-598, 2000. [124] Bryce, M.R., Tetrathiafulvalenes as pi-electron donors for intramolecular charge-transfer materials, Adv. Mater. 11 (1), 11-23, 1999. [125] Mas-Torrent, M. et al., Temperature dependence of the electrical properties of singlecrystals of dithiophene-tetrathiafulvalene (DT-TTF), Synth. Met. 146 (3), 265-268, 2004. [126] Mas-Torrent, M. et al., Correlation between crystal structure and mobility in organic field-effect transistors based on single crystals of tetrathiafulvalene derivatives, J. Am. Chem. Soc. 126 (27), 8546-8553, 2004. [127] Mas-Torrent, M. et al., High mobility of dithiophene-tetrathiafulvalene single-crystal organic field effect transistors, J. Am. Chem. Soc. 126 (4), 984-985, 2004. [128] Bromley, S.T. et al., Importance of intermolecular interactions in assessing hopping mobilities in organic field effect transistors: Pentacene versus dithiophene-tetrathiafulvalene, J. Am. Chem. Soc. 126 (21), 6544-6545, 2004. [129] Mas-Torrent, M. et al., Single-crystal organic field-effect transistors based on dibenzotetrathiafulvalene, Appl. Phys. Lett. 86 (1), 012110/1-012110/3, 2005. [130] Naraso et al., High-performance organic field-effect transistors based on pi-extended tetrathiafulvalene derivatives, J. Am. Chem. Soc. 127 (29), 10142-10143, 2005. [131] Xue, J.G. and Forrest, S.R., Organic thin-film transistors based on bis(1,2,5-thiadiazolo)p-quinobis (1,3-dithiole), Appl. Phys. Lett. 79 (22), 3714-3716, 2001. [132] Morioka, Y. et al., Novel field-effect transistors based on bis(1,3-dithiol-2-ylidene) compounds with a conjugated spacer group, Chem. Lett. 33 (12), 1632-1633, 2004. [133] Takada, M. et al., BTQBT (bis-(1,2,5-thiadiazolo)-p-quinobis(1, 3-dithiole)) thin films; A promising candidate for high mobility organic transistors, Jpn. J. Appl. Phys., Part 2 41 (1AB), L4-L6, 2002. [134] Akiba, K.Y., Ishikawa, K., and Inamoto, N., Synthesis of 1,4-Dithiafulvenes and 1,4Dithiafulvalenes by Carbonyl Olefination Using 2-Dimethoxyphosphinyl-1,3-Benzodithiole, Bull. Chem. Soc. Jpn. 51 (9), 2674-2683, 1978. [135] Kunugi, Y. et al., Organic field-effect transistor using oligoselenophene as an active layer, Chem. Mater. 15 (1), 6-7, 2003. [136] Takimiya, K. et al., 2,6-diphenylbenzo 1,2-b : 4,5-b ' dichalcogenophenes: a new class of high-performance semiconductors for organic field-effect transistors, J. Am. Chem. Soc. 126 (16), 5084-5085, 2004. [137] Zeis, R. et al., Single-crystal field-effect transistors based on organic selenium-containing semiconductor, Jpn. J. Appl. Phys., Part 1 44 (6A), 3712-3714, 2005. [138] Takimiya, K. et al., 2,7-diphenyl[1]benzoselenopheno[3,2-b][1]benzoselenophene as a stable organic semiconductor for a high-performance field-effect transistor, J. Am. Chem. Soc. 128 (9), 3044-3050, 2006. [139] Chen, C.T., Evolution of red organic light-emitting diodes: Materials and devices, Chem. Mater. 16 (23), 4389-4400, 2004. [140] Veres, J. et al., Gate insulators in organic field-effect transistors, Chem. Mater. 16 (23), 4543-4555, 2004.

8080_C003.fm Page 221 Friday, April 6, 2007 10:24 AM

Design, Synthesis, and Transistor Performance of Organic Semiconductors

221

[141] Veres, J. et al., Low-k insulators as the choice of dielectrics in organic field-effect transistors, Adv. Funct. Mater. 13 (3), 199-204, 2003. [142] Sonntag, M. et al., Novel star-shaped triphenylamine-based molecular glasses and their use in OFETs, Chem. Mater. 17 (11), 3031-3039, 2005. [143] Ponomarenko, S.A. et al., Star-shaped oligothiophenes for solution-processible organic field-effect transistors, Adv. Funct. Mater. 13 (8), 591-596, 2003. [144] Pei, J. et al., Star-shaped polycyclic aromatics based on oligothiophene-functionalized truxene: synthesis, properties, and facile emissive wavelength tuning, J. Am. Chem. Soc. 125, 9944, 2003. [145] Sun, Y.M. et al., Oligothiophene-functionalized truxene: Star-shaped compounds for organic field-effect transistors, Adv. Funct. Mater. 15 (5), 818-822, 2005. [146] Cravino, A. et al., Triphenylamine-oligothiophene conjugated systems as organic semiconductors for opto-electronics, Chem. Mater. 18 (10), 2584-2590, 2006. [147] Gundlach, D.J. et al., Oligophenyl-based organic thin film transistors, Appl. Phys. Lett. 71 (26), 3853-3855, 1997. [148] Roy, V.A.L. et al., Functionalized arylacetylene oligomers for organic thin-film transistors (OTFTs), Adv. Mater. 17 (10), 1258, 2005. [149] van Breemen, A.J.J.M. et al., Large area liquid crystal monodomain field-effect transistors, J. Am. Chem. Soc. 128 (7), 2336-2345, 2006. [150] Ostoja, P. et al., FET device performance, morphology and X-ray thin film structure of unsubstituted and modified quinquethiophenes, Synth. Met. 146, 243-250, 2004. [151] Lengyel, O. et al., Solution-processed thin films of thiophene mesogens with singlecrystalline alignment, Adv. Mater. 18 (7), 896, 2006. [152] Gorjanc, T.C., Levesque, I., and D'Iorio, M., Organic field effect transistors based on modified oligo-p-phenylevinylenes, Appl. Phys. Lett. 84 (6), 930-932, 2004. [153] Yasuda, T. et al., Organic field-effect transistors based on oligo-p-phenylenevinylene derivatives, Jpn. J. Appl. Phys. 2 45 (8-11), L313-L315, 2006. [154] Dimitrakopoulos, C.D. et al., trans-trans-2,5-bis- 2-{5-(2,2'-bithienyl)}ethenyl thiophene: synthesis, characterization, thin film deposition and fabrication of organic field-effect transistors, Synth. Met. 89 (3), 193-197, 1997. [155] Videlot, C. et al., Field-effect transistors based on oligothienylenevinylenes: From solution pi-dimers to high-mobility organic semiconductors, Adv. Mater. 15 (4), 306310, 2003. [156] Roncali, J., Oligothienylenevinylenes as a new class of multinanometer linear piconjugated systems for micro- and nanoelectronics, Acc. Chem. Res. 33 (3), 147-156, 2000. [157] Eley, D.D. et al., The Semiconductivity of Organic Substances .1, Trans. Farad. Soc. 49 (1), 79-86, 1953. [158] Eley, D.D., Phthalocyanines as Semiconductors, Nature 162 (4125), 819-819, 1948. [159] Bao, Z., Lovinger, A.J., and Dodabalapur, A., Organic field-effect transistors with high mobility based on copper phthalocyanine, Appl. Phys. Lett. 69 (20), 3066-3068, 1996. [160] Zeis, R., Siegrist, T., and Kloc, C., Single-crystal field-effect transistors based on copper phthalocyanine, Appl. Phys. Lett. 86 (2), 022103/1-022103/3, 2005. [161] Zhang, J. et al., Organic thin-film transistors in sandwich configuration, Appl. Phys. Lett. 84 (1), 142-144, 2004. [162] Xiao, K. et al., Field-effect transistors based on Langmuir-Blodgett films of phthalocyanine derivatives as semiconductor layers, J. Phys. Chem. B 107 (35), 9226-9230, 2003. [163] Su, W. et al., Thin-film transistors based on Langmuir-Blodgett films of heteroleptic bis(phthalocyaninato) rare earth complexes, Langmuir 21 (14), 6527-6531, 2005.

8080_C003.fm Page 222 Friday, April 6, 2007 10:24 AM

222

Organic Field-Effect Transistors

[164] Hu, W.P. et al., The application of Langmuir-Blodgett films of a new asymmetrically substituted phthalocyanine, amino-tri-tert-butyl-phthalocyanine, in diodes and in all organic field-effect-transistors, Synth. Met. 104 (1), 19-26, 1999. [165] Chen, Y. et al., High performance organic field-effect transistors based on amphiphilic tris(phthalocyaninato) rare earth triple-decker complexes, J. Am. Chem. Soc.127 (45), 15700-15701, 2005. [166] Noh, Y.Y. et al., Organic field-effect transistors by a wet-transferring method, Appl. Phys. Lett. 83 (6), 1243-1245, 2003. [167] Noh, Y.Y. et al., Effect of molecular orientation of epitaxially grown platinum(II) octaethyl porphyrin films on the performance of field-effect transistors, Adv. Mater. 15 (9), 699-702, 2003. [168] Checcoli, P. et al., Tetra-phenyl porphyrin based thin film transistors, Synth. Met. 138 (1-2), 261-266, 2003. [169] Xu, H. et al., High-performance field-effect transistors based on Langmuir-Blodgett films of cyclo 8 pyrrole, Langmuir 21 (12), 5391-5395, 2005. [170] Taguchi, T. et al., Comparison of p-type and n-type organic field-effect transistors using nickel coordination compounds, Chem. Phys. Lett. 421 (4-6), 395-398, 2006. [171] Caseri, W.R. et al., "(Hot-)Water-Proof", semiconducting, platinum-based chain structures: Processing, products, and properties, Adv. Mater. 15 (2), 125, 2003. [172] Barbarella, G., Melucci, M., and Sotgiu, G., The versatile thiophene: An overview of recent research on thiophene-based materials, Adv. Mater. 17 (13), 1581-1593, 2005. [173] Servet, B. et al., Polymorphism and Charge-Transport in Vacuum-Evaporated Sexithiophene Films, Chem. Mater. 6 (10), 1809-1815, 1994. [174] Garnier, F., Materials Research in France, Adv. Mater. 2 (6-7), 277-277, 1990. [175] Garnier, F. et al., Structural Basis for High Carrier Mobility in Conjugated Oligomers, Synth. Met. 45 (2), 163-171, 1991. [176] Horowitz, G. et al., Field-effect transistor made with a sexithiophene single crystal, Adv. Mater. 8 (1), 52-&, 1996. [177] Rittner, M. et al., Oligothiophenes of DDnT and DPnT type: Morphology, field effect transistors, performance in vacuum and in air, Synth. Met. 156 (1), 21-26, 2006. [178] Dimitrakopoulos, C.D. et al., Field-effect transistors comprising molecular beam deposited alpha,omega-di-hexyl-hexathienylene and polymeric insulator, Synth. Met. 92 (1), 47-52, 1998. [179] Dodabalapur, A., Torsi, L., and Katz, H.E., Organic Transistors — 2-Dimensional Transport and Improved Electrical Characteristics, Science 268 (5208), 270-271, 1995. [180] Katz, H.E., Torsi, L., and Dodabalapur, A., Synthesis, Material Properties, and Transistor Performance of Highly Pure Thiophene Oligomers, Chem. Mater. 7 (12), 2235-2237, 1995. [181] Hajlaoui, R. et al., Organic transistors using alpha-octithiophene and alpha,omegadihexyl-alpha-octithiophene: Influence of oligomer length versus molecular ordering on mobility, Adv. Mater. 9 (7), 557-&, 1997. [182] Locklin, J. et al., Organic thin film transistors based on cyclohexyl-substituted organic semiconductors, Chem. Mater. 17 (13), 3366-3374, 2005. [183] Someya, T. et al., Vapor sensing with alpha,omega-dihexylquarterthiophene field-effect transistors: The role of grain boundaries, Appl. Phys. Lett. 81 (16), 3079-3081, 2002. [184] Barbarella, G. et al., Structural and Electrical Characterization of Processable BisSilylated Thiophene Oligomers Chem. Mater. 10 (11), 3683-3689, 1998.

8080_C003.fm Page 223 Friday, April 6, 2007 10:24 AM

Design, Synthesis, and Transistor Performance of Organic Semiconductors

223

[185] Zen, A. et al., Solution processable organic field-effect transistors utilizing an alpha,alpha '-dihexylpentathiophene-based swivel cruciform, J. Am. Chem. Soc. 128 (12), 3914-3915, 2006. [186] Katz, H.E., Laquindanum, J.G., and Lovinger, A.J., Synthesis, solubility, and field-effect mobility of elongated and oxa-substituted alpha,omega-dialkyl thiophene oligomers. Extension of "polar intermediate" synthetic strategy and solution deposition on transistor substrates, Chem. Mater. 10 (2), 633-638, 1998. [187] Afzali, A., Breen, T.L., and Kagan, C.R., An efficient synthesis of symmetrical oligothiophenes: Synthesis and transport properties of a soluble sexithiophene derivative, Chem. Mater. 14 (4), 1742-1746, 2002. [188] Sandberg, H. et al., Oligoethyleneoxide functionalised sexithiophene organic field effect transistors, Synth. Met. 137 (1-3), 885-886, 2003. [189] Dell'Aquila, A. et al., Synthesis and field-effect properties of alpha,omega-disubstituted sexithiophenes bearing polar groups, J. Mater. Chem. 16 (12), 1183-1191, 2006. [190] Meng, H. et al., Oligofluorene-thiophene derivatives as high-performance semiconductors for organic thin film transistors, Chem. Mater. 15 (9), 1778-1787, 2003. [191] Porzio, W. et al., Organic FET devices: structure-property relationship in evaporated films of three fluorenone derivatives, Synth. Met. 146 (3), 259-263, 2004. [192] Woon, K.L. et al., Electronic charge transport in extended nematic liquid crystals, Chem. Mater. 18 (9), 2311-2317, 2006. [193] Thiem, H. et al., New fluorene-bithiophene-based trimers as stable materials for OFETs, Synth. Met. 156 (7-8), 582-589, 2006. [194] Yanagi, H. et al., Comparative carrier transport characteristics in organic field-effect transistors with vapor-deposited thin films and epitaxially grown crystals of biphenylcapped thiophene oligomers, Adv. Funct. Mater. 13 (10), 767-773, 2003. [195] Ichikawa, M. et al., Organic field-effect transistors made of epitaxially grown crystals of a thiophene/phenylene co-oligomer, Adv. Mater. 14 (18), 1272, 2002. [196] Mushrush, M. et al., Easily processable phenylene-thiophene-based organic field-effect transistors and solution-fabricated nonvolatile transistor memory elements, J. Am. Chem. Soc. 125 (31), 9414-9423, 2003. [197] Mohapatra, S. et al., Organic thin-film transistors based on tolyl-substituted oligothiophenes, Adv. Funct. Mater. 14 (6), 605-609, 2004. [198] Tian, H.K. et al., Novel thiophene-aryl co-oligomers for organic thin film transistors, J. Mater. Chem. 15 (29), 3026-3033, 2005. [199] Ponomarenko, S.A. et al., 1,4-bis(5-decyl-2,2 '-bithien-5-yl)benzene as new stable organic semiconductor for high performance thin film transistors, Synth. Met. 149 (2-3), 231-235, 2005. [200] Ponomarenko, S.A. et al., Decyl-end-capped thiophene-phenylene oligomers as organic semiconducting materials with improved oxidation stability. (vol 18, pg 579, 2006), Chem. Mater. 18 (7), 2004-2004, 2006. [201] Yoon, M.H. et al., Fluorocarbon-modified organic semiconductors: Molecular architecture, electronic, and crystal structure tuning of arene-versus fluoroarene-thiophene oligomer thin-film properties, J. Am. Chem. Soc. 128 (17), 5792-5801, 2006. [202] Deman, A.L. et al., Structural effects on the characteristics of organic field effect transistors based on new oligothiophene derivatives, Synth. Met. 146 (3), 365-371, 2004. [203] Minari, T. et al., Alkyl chain length dependent mobility of organic field-effect transistors based on thienyl-furan oligomers determined by the transfer line method, Appl. Phys. Lett. 88 (8), 083514, 2006.

8080_C003.fm Page 224 Friday, April 6, 2007 10:24 AM

224

Organic Field-Effect Transistors

[204] Py, C. et al., Hole mobility and electroluminescence properties of a dithiophene indenofluorene, Journal of Vacuum Science & Technology A 24 (3), 654-656, 2006. [205] Wang, G.M. et al., Increased mobility from regioregular poly(3-hexylthiophene) fieldeffect transistors, J. Appl. Phys. 93 (10), 6137-6141, 2003. [206] Sirringhaus, H. et al., Two-dimensional charge transport in self-organized, high-mobility conjugated polymers, Nature 401 (6754), 685-688, 1999. [207] Sirringhaus, H., Tessler, N., and Friend, R.H., Integrated optoelectronic devices based on conjugated polymers, Science 280 (5370), 1741-1744, 1998. [208] Chang, J.F. et al., Enhanced mobility of poly(3-hexylthiophene) transistors by spincoating from high-boiling-point solvents, Chem. Mater. 16 (23), 4772-4776, 2004. [209] Xu, G.F., Bao, Z.A., and Groves, J.T., Langmuir-Blodgett films of regioregular poly(3hexylthiophene) as field-effect transistors, Langmuir 16 (4), 1834-1841, 2000. [210] Babel, A. and Jenekhe, S.A., Alkyl chain length dependence of the field-effect carrier mobility in regioregular poly(3-alkylthiophene)s, Synth. Met. 148 (2), 169-173, 2005. [211] Bao, Z. and Lovinger, A.J., Soluble regioregular polythiophene derivatives as semiconducting materials for field-effect transistors, Chem. Mater. 11 (9), 2607-2612, 1999. [212] Murphy, A.R. et al., Synthesis, characterization, and field-effect transistor performance of carboxylate-functionalized polythiophenes with increased air stability, Chem. Mater. 17 (20), 4892-4899, 2005. [213] Crouch, D.J. et al., Hexyl-substituted oligothiophenes with a central tetrafluorophenylene unit: crystal engineering of planar structures for p-type organic semiconductors, Chem Commun (11), 1465-1467, 2005. [214] McCulloch, I. et al., Influence of molecular design on the field-effect transistor characteristics of terthiophene polymers, Chem. Mater. 17 (6), 1381-1385, 2005. [215] Heeney, M. et al., Stable polythiophene semiconductors incorporating thieno 2,3-b thiophene, J. Am. Chem. Soc. 127 (4), 1078-1079, 2005. [216] Ong, B. et al., Polythiophene-based field-effect transistors with enhanced air stability, Synth. Met. 142 (1-3), 49-52, 2004. [217] Wu, Y.L. et al., Poly(3,3"-dialkylterthiophene)s: Room-temperature, solution-processed, high-mobility semiconductors for organic thin-film transistors, Chem. Mater. 17 (2), 221-223, 2005. [218] Ong, B.S. et al., High-performance semiconducting polythiophenes for organic thinfilm transistors, J. Am. Chem. Soc. 126 (11), 3378-3379, 2004. [219] McCulloch, I., Designing air stable thiophene polymers based on control of conjugation length, presented at the 229th ACS NAtl. Meeting, San Diego, CA, March, 2005. [220] Ohshita, J. et al., Synthesis of organosilanylene-oligothienylene alternate polymers and their applications to EL and FET materials, Organometallics 24 (18), 4494-4496, 2005. [221] Mcculloch, I. et al., Liquid-crystalline semiconducting polymers with high chargecarrier mobility, Nat. Mater. 5 (4), 328-333, 2006. [222] Sirringhaus, H. et al., Mobility enhancement in conjugated polymer field-effect transistors through chain alignment in a liquid-crystalline phase, Appl. Phys. Lett. 77 (3), 406-408, 2000. [223] Hamilton, M.C., Martin, S., and Kanicki, J., Field-effect mobility of organic polymer thin-film transistors, Chem. Mater. 16 (23), 4699-4704, 2004. [224] Lim, E. et al., Thin-film morphologies and solution-processable field-effect transistor behavior of a fluorene-thieno 3,2-b thiophene-based conjugated copolymer, Macromolecules 38 (10), 4531-4535, 2005. [225] Chen, M.X. et al., Low band gap donor-acceptor-donor polymers for infra-red electroluminescence and transistors, Synth. Met. 146 (3), 233-236, 2004.

8080_C003.fm Page 225 Friday, April 6, 2007 10:24 AM

Design, Synthesis, and Transistor Performance of Organic Semiconductors

225

[226] Yamamoto, T. et al., Ambipolar field-effect transistor (FET) and redox characteristics of a pi-conjugated thiophene/1,3,4-thiadiazole CT-type copolymer, Macromol. Rapid Comm. 26 (15), 1214-1217, 2005. [227] Yamamoto, T. et al., Alignment and field-effect transistor behavior of an alternative piconjugated copolymer of thiophene and 4-alkylthiazole, Chem. Mater. 16 (23), 46164618, 2004. [228] Turner, M.L., Coppo, P., Cupertino, D.C., Yeates, S.G., Investigation of the electronic properties of cyclopentadithiophene polymers and copolymers, Mat. Res. Soc. Symp. Proc. 771, L4.9, 2003. [229] Coppo, P., Cupertino, D.C., Yeates, S.G., Turner, M.L., Synthetic routes to solutionprocessable polycyclopentadithiophenes, Macromolecules 36, 2705-2711, 2003. [230] Turner, M.L., New routes to poly(4,4-dialkylcyclopentadithiophene-2,6-diyls), J. Mater. Chem. 12, 2597-2599, 2002. [231] Coppo, P. and Turner, M.L., Cyclopentadithiophene based electroactive materials, J. Mater. Chem. 15 (11), 1123-1133, 2005. [232] Heeney, M. et al., Alkylidene fluorene liquid crystalline semiconducting polymers for organic field effect transistor devices, Macromolecules 37 (14), 5250-5256, 2004. [233] Tanase, C. et al., Optimization of the charge transport in poly(phenylene vinylene) derivatives by processing and chemical modification, J. Appl. Phys. 97 (12), 1237031, 2005. [234] van Breemen, A. et al., High-performance solution-processable poly(p-phenylene vinylene)s for air-stable organic field-effect transistors, Adv. Funct. Mater. 15 (5), 872-876, 2005. [235] Tanase, C. et al., Charge carrier density dependence of the hole mobility in poly(pphenylene vinylene), Phys. Status Solidi A 201 (6), 1236-1245, 2004. [236] Zhu, Y., Babel, A., and Jenekhe, S.A., Phenoxazine-based conjugated polymers: a new class of organic semiconductors for field-effect transistors, Macromolecules 38(19), 7983-7991, 2005. [237] Hwang, D.H. et al., Conjugated polymers based on phenothiazine and fluorene in lightemitting diodes and field effect transistors, Chem. Mater. 16 (7), 1298-1303, 2004. [238] Burroughes, J.H., Jones, C.A., and Friend, R.H., New Semiconductor-Device Physics in Polymer Diodes and Transistors, Nature 335 (6186), 137-141, 1988. [239] Friend, R.H. et al., The Durham Route to Polyacetylene, Synth. Met. 19 (1-3), 989-989, 1987. [240] Friend, R.H. et al., Conformational Defects in Durham Polyacetylene, Synth. Met. 17 (1-3), 267-272, 1987. [241] Edwards, J.H. and Feast, W.J., A New Synthesis of Poly(Acetylene), Polymer 21 (6), 595-596, 1980. [242] Fuchigami, H., Tsumura, A., and Koezuka, H., Polythienylenevinylene Thin-FilmTransistor with High Carrier Mobility, Appl. Phys. Lett. 63 (10), 1372-1374, 1993. [243] Murase, I. et al., Highly Conducting Poly(2,5-Thienylene Vinylene) Via a Precursor Route, Polym. Comm. 28 (8), 229-231, 1987. [244] Meijer, E.J. et al., Solution-processed ambipolar organic field-effect transistors and inverters, Nat. Mater. 2 (10), 678-682, 2003. [245] Aramaki, S., Sakai, Y., and Ono, N., Solution-processible organic semiconductor for transistor applications: Tetrabenzoporphyrin, Appl. Phys. Lett. 84 (12), 2085-2087, 2004. [246] A.R. Brown et al., Precursor route pentacene metal-insulator-semiconductor field-effect transistors, J. Appl. Phys. 79 (4), 2136-2138, 1996.

8080_C003.fm Page 226 Friday, April 6, 2007 10:24 AM

226

Organic Field-Effect Transistors

[247] P.T. Herwig and Mullen, K., A soluble pentacene precursor: synthesis, solid-state conversion into pentacene and application in a field-effect transistor, Adv. Mater. 11 (6), 480-483, 1999. [248] Brown, A.R. et al., Logic Gates Made from Polymer Transistors and Their Use in Ring Oscillators, Science 270 (5238), 972-974, 1995. [249] Afzali, A., Dimitrakopoulos, C.D., and Breen, T.L., High-performance, solution-processed organic thin film transistors from a novel pentacene precursor, J. Am. Chem. Soc. 124 (30), 8812-8813, 2002. [250] Afzali, A., Dimitrakopoulos, C.D., and Graham, T.O., Photosensitive pentacene precursor: Synthesis, photothermal patterning, and application in thin-film transistors, Adv. Mater. 15 (24), 2066, 2003. [251] Weinreb, S.M., Synthetic Methodology Based Upon N-Sulfinyl Dienophile 4 + 2 Cycloaddition Reactions, Acc. Chem. Res. 21 (8), 313-318, 1988. [252] Scherer, O.J. and Schmitt, R., New Types of Thionylimide Derivatives, Chem. Ber.Recl. 101 (9), 3302-&, 1968. [253] Zhu, Z.L. and Espenson, J.H., Aqueous catalysis: Methylrhenium trioxide (MTO) as a homogeneous catalyst for the Diels-Alder reaction, J. Am. Chem. Soc. 119 (15), 35073512, 1997. [254] Murphy, A.R. et al., Organic thin film transistors from a soluble oligothiophene derivative containing thermally removable solubilizing groups, J. Am. Chem. Soc. 126 (6), 1596-1597, 2004. [255] Chang, P.C. et al., Film morphology and thin film transistor performance of solutionprocessed oligothiophenes, Chem. Mater. 16 (23), 4783-4789, 2004. [256] Nicolas, Y. et al., Synthesis of a Thermally Stable Hybrid Acene-Thiophene Organic Semiconductor via a Soluble Precursor, Org. Lett. 7 (6), 3513-3516, 2005. [257] Koyanagi, T. et al., Organic field-effect transistor based on a thin film of polydiacetylene prepared from 10,12-pentacosadiynoic acid, Chem. Lett. 35 (1), 20-21, 2006. [258] Chua, L.L. et al., General observation of n-type field-effect behaviour in organic semiconductors, Nature 434 (7030), 194-199, 2005. [259] J. Kastner, J.P., H. Kuzmany, Fullerene field-effect transistors, in Electronic Properties of High-Tc superconductors, H. Kuzmany, M. M., J. Fink Springer-Verlag, Berlin, 1993, pp. 512-515. [260] Haddon, R.C. et al., C-60 Thin-Film Transistors, Appl. Phys. Lett. 67 (1), 121-123, 1995. [261] Kobayashi, S. et al., Fabrication and characterization of C-60 thin-film transistors with high field-effect mobility, Appl. Phys. Lett. 82 (25), 4581-4583, 2003. [262] Veenstra, S.C. et al., On interface dipole layers between C-60 and Ag or Au, Appl. Phys. A 75 (6), 661-666, 2002. [263] Singh, T.B. et al., Fabrication and characterization of solution-processed methanofullerene-based organic field-effect transistors, J. Appl. Phys. 97 (8), 083714/1083714/5, 2005. [264] Kusai, H. et al., Fabrication of field-effect transistor devices with fullerodendron by solution process, Appl. Phys. Lett. 88 (17), -, 2006. [265] Bao, Z.A., Lovinger, A.J., and Brown, J., New air-stable n-channel organic thin film transistors, J. Am. Chem. Soc. 120 (1), 207-208, 1998. [266] Locklin, J. et al., Ambipolar organic thin film transistor-like behavior of cationic and anionic phthalocyanines fabricated using layer-by-layer deposition from aqueous solution, Chem. Mater. 15 (7), 1404-1412, 2003. [267] Guillaud, G. et al., Field-Effect Transistors Based on Intrinsic Molecular Semiconductors, Chem. Phys. Lett. 167 (6), 503-506, 1990.

8080_C003.fm Page 227 Friday, April 6, 2007 10:24 AM

Design, Synthesis, and Transistor Performance of Organic Semiconductors

227

[268] Laquindanum, J.G. et al., n-channel organic transistor materials based on naphthalene frameworks, J. Am. Chem. Soc. 118 (45), 11331-11332, 1996. [269] Katz, H.E. et al., A soluble and air-stable organic semiconductor with high electron mobility, Nature 404 (6777), 478-481, 2000. [270] Katz, H.E. et al., Naphthalenetetracarboxylic diimide-based n-channel transistor semiconductors: Structural variation and thiol-enhanced gold contacts, J. Am. Chem. Soc. 122 (32), 7787-7792, 2000. [271] Katz, H.E. et al., Unsymmetrical n-channel semiconducting naphthalenetetracarboxylic diimides assembled via hydrogen bonds, Chem. Lett. 32 (6), 508-509, 2003. [272] Bredas, J.L. et al., Organic semiconductors: A theoretical characterization of the basic parameters governing charge transport, Proc. Natl. Acad. Sci. 99 (9), 5804-5809, 2002. [273] Ostrick, J.R. et al., Conductivity-type anisotropy in molecular solids, J. Appl. Phys. 81 (10), 6804-6808, 1997. [274] Hiramoto, M. et al., Conduction type control from n to p type for organic pigment films purified by reactive sublimation, J. Appl. Phys. 78 (12), 7153-7157, 1995. [275] Gundlach, D.J. et al., High mobility n-channel organic thin-film transistors and complementary inverters, J. Appl. Phys. 98 (6), 064502/1-064502/8, 2005. [276] Kelley, T.W. et al., 2003. [277] Chesterfield, R.J. et al., Organic thin film transistors based on N-alkyl perylene diimides: Charge transport kinetics as a function of gate voltage and temperature, J. Phys. Chem. B 108 (50), 19281-19292, 2004. [278] Horowitz, G. et al., Evidence for n-type conduction in a perylene tetracarboxylic diimide derivative, Adv. Mater. 8 (3), 242-245, 1996. [279] Jones, B.A. et al., High-mobility air-stable n-type semiconductors with processing versatility: Dicyanoperylene-3,4 : 9,10-bis(dicarboximides), Angew. Chem. Int. Ed. 43 (46), 6363-6366, 2004. [280] Yoo, B. et al., High-mobility bottom-contact n-channel organic transistors and their use in complementary ring oscillators, Appl. Phys. Lett. 88 (8), -, 2006. [281] Brown, A.R. et al., Organic N-Type Field-Effect Transistor, Synth. Met. 66 (3), 257261, 1994. [282] Nishida, J. et al., Preparation, characterization, and FET properties of novel dicyanopyrazinoquinoxaline derivatives, Org. Lett. 6 (12), 2007-2010, 2004. [283] Pappenfus, T.M. et al., A pi-stacking terthiophene-based quinodimethane is an n-channel conductor in a thin film transistor, J. Am. Chem. Soc. 124 (16), 4184-4185, 2002. [284] Chesterfield, R.J. et al., High electron mobility and ambipolar transport in organic thinfilm transistors based on a pi-stacking quinoidal terthiophene, Adv. Mater. 15 (15), 1278, 2003. [285] Facchetti, A. et al., Tuning the semiconducting properties of sexithiophene by alpha,omega-substitution — alpha,omega-diperfluorohexylsexithiophene: The first ntype sexithiophene for thin-film transistors, Angew. Chem. Int. Ed. 39 (24), 4547, 2000. [286] Facchetti, A. et al., n-type building blocks for organic electronics: A homologous family of fluorocarbon-substituted thiophene oligomers with high carrier mobility, Adv. Mater. 15 (1), 33, 2003. [287] Facchetti, A. et al., Synthesis and characterization of diperfluorooctyl-substituted phenylene-thiophene oligomers as n-type semiconductors. Molecular structure-film microstructure-mobility relationships, organic field-effect transistors, and transistor nonvolatile memory elements, Chem. Mater. 16 (23), 4715-4727, 2004.

8080_C003.fm Page 228 Friday, April 6, 2007 10:24 AM

[288] Yoon, M.-H. et al., Organic Thin-Film Transistors Based on Carbonyl-Functionalized Quaterthiophenes: High Mobility N-Channel Semiconductors and Ambipolar Transport, J. Am. Chem. Soc. 127 (5), 1348-1349, 2005. [289] Facchetti, A. et al., Building blocks for n-type organic electronics: Regiochemically modulated inversion of majority carrier sign in perfluoroarene-modified polythiophene semiconductors, Angew. Chem.-Int. Edit. 42 (33), 3900-3903, 2003. [290] Letizia, J.A. et al., High Electron Mobility in Solution-Cast and Vapor-Deposited Phenacyl-Quaterthiophene-Based Field-Effect Transistors: Toward N-Type Polythiophenes, J. Am. Chem. Soc. 127 (39), 13476-13477, 2005. [291] Ando, S. et al., High performance n-type organic field-effect transistors based on pielectronic systems with trifluoromethylphenyl groups, J. Am. Chem. Soc. 127 (15), 5336-5337, 2005. [292] Ando, S. et al., Novel p- and n-type organic semiconductors with an anthracene unit, Chem. Mater. 17 (6), 1261, 2005. [293] Ando, S. et al., n-Type Organic Field-Effect Transistors with Very High Electron Mobility Based on Thiazole Oligomers with Trifluoromethylphenyl Groups, J. Am. Chem. Soc. 127, 14996-14997, 2005. [294] Babel, A. and Jenekhe, S.A., High electron mobility in ladder polymer field-effect transistors, J. Am. Chem. Soc. 125 (45), 13656-13657, 2003. [295] Babel, A. and Jenekhe, S.A., Electron Transport in Thin-Film Transistors from an nType Conjugated Polymer, Adv. Mater. 14 (5), 371-374, 2002. [296] Jenekhe, S.A. and Tibbetts, S.J., Ion-implantation doping and electrical properties of high-temperature ladder polymers J. Polym. Sci., Part B: Polym. Phys. 26 (1), 201209, 1988. [297] Chen, X.L. et al., Ion-modulated ambipolar electrical conduction in thin-film transistors based on amorphous conjugated polymers, Appl. Phys. Lett. 78 (2), 228-230, 2001. [298] Donley, C.L. et al., Effects of Packing Structure on the Optoelectronic and Charge Transport Properties in Poly(9,9-di-n-octylfluorene-alt-benzothiadiazole), J. Am. Chem. Soc. 127, 12890-12899, 2005. [299] Janietz, S. et al., n-Type Copolymers with Fluorene and 1,3,4-Heterodiazole Moieties, Macromol. Chem. Phys. 205, 187-198, 2004.

8080_C003.fm Page 229 Friday, April 6, 2007 10:24 AM

3.2

Dielectric Materials: Selection and Design

Ashok Maliakal CONTENTS 3.2.1 Introduction................................................................................................229 3.2.2 Fundamentals and Figures of Merit ..........................................................231 3.2.2.1 Factors Affecting the Dielectric Constant...................................232 3.2.2.2 Thickness.....................................................................................233 3.2.2.3 Dielectric Roughness...................................................................233 3.2.2.4 Film Morphology ........................................................................234 3.2.2.5 Importance of Interface between Dielectric and Semiconductor.............................................................................234 3.2.2.6 Processing....................................................................................234 3.2.2.7 Reliablity .....................................................................................235 3.2.3 Major Classes of Dielectric Materials.......................................................235 3.2.3.1 Inorganic Dielectrics ...................................................................235 3.2.3.2 Polymer Dielectrics .....................................................................237 3.2.4 Alternative Gate Dielectric Strategies .......................................................240 3.2.4.1 Gate Dielectrics through Anodization of Thin-Metal Films ......240 3.2.4.2 Surface Treatment of Inorganic Materials ..................................241 3.2.4.3 Self-Assembled Monolayers/Multilayers....................................242 3.2.4.4 Nanocomposite and Nanostructured Dielectrics ........................245 3.2.5 Summary and Conclusions ........................................................................248 References..............................................................................................................248

3.2.1 INTRODUCTION The rapid growth in research and development efforts in organic/flexible electronics underscores the importance of this new science and technology [1–4]. The field of organic electronics focuses primarily on the issues relating to the organic thin-film transistor device. This device forms the basis of organic integrated circuits. A schematic of the prototypical organic thin-film transistor (TFT) is illustrated in Figure 3.2.1. This chapter will focus on the selection and design of the gate dielectric, which is the insulating material separating the active semiconducting material from the

229

8080_C003.fm Page 230 Friday, April 6, 2007 10:24 AM

230

Organic Field-Effect Transistors

Semiconductor Source Drain Dielectric Gate Substrate

FIGURE 3.2.1 Schematic of bottom gate organic thin-film transistor.

gate electrode. This review is intended to bring researchers interested in the area of organic electronics up to speed on the factors involved in selecting a gate dielectric material. The review will discuss the basic issues involved, provide an overview on existing materials used as gate dielectrics, and finally present the newest developments in gate dielectrics. As can be seen in Figure 3.2.1, the dielectric layer in a typical thin-film transistor device is sandwiched between the gate electrode and the organic semiconductor. This geometry creates two interfaces, which must be considered when selecting a suitable dielectric material. When considering the dielectric–gate interface, one must address the issue of preventing static charge or dynamic charge injection into the dielectric, which can have adverse effects on threshold voltage (VT) as a function of time. Furthermore, complete coverage of the gate electrode is necessary to prevent leakage currents through pinhole defects. The dielectric-semiconductor interface is equally important to control, if not more so. This interface is where the conducting channel is formed. The quality of this channel is determined by interface roughness, surface energy, and charges at this interface [5]. Furthermore, in most TFT structures, the semiconductor is grown or deposited on top of the dielectric. In these instances, the dielectric serves to organize the semiconductor, especially in its first few layers. The structure of these layers is thought to be critical to the performance of organic semiconductors [6]. The basic issues in choosing a gate dielectric for an organic TFT are control of interfaces, leakage, dielectric constant, processability, stability, and reliability. Achieving high capacitances in organic TFTs is of great importance, considering the low mobilities (typically, <1 cm2V–1s–1) of organic semiconductors [1,5,7]. Keeping leakage currents at a minimum will reduce the power requirements of these devices [8]. Ease of processing of dielectric materials is an important issue when planning the commercialization of real devices. In this regard, easy, inexpensive solution-phase methods are desirable, especially if they are amenable to printing techniques such as inkjet, gravure, or offset printing [9]. Finally, the dielectric films will need to be stable (i.e., not prone to drifts in performance or threshold voltage) and exhibit minimum hysteresis. As can be seen from the large number of requirements, the dielectric has an important role in TFTs that, up until recently, has often been overlooked in the search for higher mobility organic semiconductors. Initial studies of organic TFTs were performed on devices using SiO2/n-doped Si as the gate insulator/gate electrode pair, since this system was well established from amorphous silicon (α-Si) TFTs [10]. The first polymeric insulators (polyimides) for organic thin-film transistors were used in 1990 [10,11]. In the intervening 15+ years, a number of different dielectrics

8080_C003.fm Page 231 Friday, April 6, 2007 10:24 AM

Dielectric Materials: Selection and Design

231

have been explored involving inorganic, organic, and hybrid materials with the goal of high-capacitance, low-leakage films, which are easy to process and compatible with electrode and semiconducting materials. In recent years, some excellent reviews of dielectric materials for organic TTFs have appeared [1,5,7]. These reviews cover the basic developments in both inorganic and polymeric dielectrics quite well. In light of these excellent background materials, this section will treat with some brevity the areas of homopolymer dielectrics and inorganic dielectric films. Greater emphasis will be placed on the latest developments in new dielectrics selection and design (i.e., self-assembled monolayer [SAM] dielectrics, alignment layers to control semiconductor morphology, ultrathin anodized materials, and nanocomposites). Efforts will be made to highlight the relevant figures of merit when selecting a dielectric material.

3.2.2 FUNDAMENTALS AND FIGURES OF MERIT Organic thin-film transistors typically operate in accumulation mode [10]. In this case, for a p-type organic semiconductor such as pentacene, when negative gate voltages are applied, holes are attracted to the gate dielectric–semiconductor interface to generate an accumulation layer or channel [10,12,13]. The voltage required to create this channel is determined in large part by the capacitance of the dielectric. Figure 3.2.2 illustrates a metal–insulator–semiconductor device. The charge in the gate electrode (in the case of a metallic electrode) is all contained on the surface of the metal (i.e., localized at the electrode dielectric interface). In the case of the semiconductor, since the carrier density is lower than that of a metal, the induced charge is diffused over a thicker slab of the semiconductor. This induced charge in the semiconductor defines the channel through which current is carried in TFTs. Studies indicate that, for pentacene, the channel of accumulated charge is ~six monolayers thick [6]. In an idealized depiction of the field effect in a metal oxide semiconductor, the charge induced in the semiconductor is equal and opposite to that present at the gate electrode [14]. It is important to note that Figure 3.2.2 is an idealized situation. In real cases, issues of trapped charge in the gate dielectric or at interfaces will modify this picture [8,15]. With the simplest of gate dielectrics (a vacuum or air interface), the charge induced on the semiconductor as a function of gate voltage is Q = CV = (εo A/t)V, where εo is the permittivity of free space (8.85 x Gate electrode - - - - - - - - - Gate dielectric + + + + + + + + + + Semiconductor (p-type)

–Q +Q

Charge density

FIGURE 3.2.2 A metal–insulator–semiconductor structure in accumulation mode.

8080_C003.fm Page 232 Friday, April 6, 2007 10:24 AM

232

Organic Field-Effect Transistors

× 10–12F/m), A is the area, and t is the distance between the electrode and the semiconductor. In the case where a material such as silicon dioxide is used as a dielectric, the induced charge for a given voltage is Q = CV = (Kεo A/t)V, where K is the dielectric constant of the material [14,16]. Increasing the capacitance, C, through increasing the dielectric constant, K, or reducing the film thickness, t, would reduce the voltage requirement necessary to induce the same amount of charge in the semiconductor. In the case of SiO2 (K = 3.9), for a fixed thickness, the amount of charge induced in the semiconductor would be 3.9 greater at a fixed voltage compared to vacuum. Alternatively, a device could operate at ~3.9 lower voltage using an SiO2 dielectric layer as opposed to vacuum. Optimizing these two parameters, dielectric constant K and thickness t are essential to selecting and designing an optimal dielectric film.

3.2.2.1 FACTORS AFFECTING

THE

DIELECTRIC CONSTANT

The dielectric constant of a material is a measure of its polarizability in response to an electric field. This polarizability is the result of reorganization of charge, which can be in the form of interfacial or space charge motion, ionic motion, dipolar motion, and electronic motion (see Figure 3.2.3) [14]. The timescale of the charge redistributions determines the frequency dependence of this contribution to the dielectric constant for a given material. In the case of ionic motion, the frequency range is up to 1012 Hz. One example of a dielectric material in which ionic motion is important is barium titanate BaTiO3, which is ferroelectric (i.e., the induced polarization does not decay upon the removal of the electric field). In these structures, titanium ion displacement within its octahedral sites causes extremely large polarizations (2,000–3,000) [14]. In nonferroelectric materials such as titanium dioxide, ions will return to equilibrium position upon removal of field. Electronic polarizability is greater for materials containing more electrons (i.e., heavy atoms, greater polarizability). The frequency range for electronic motion is up to 1016 Hz. In the case of materials for organic

κ

Orientational, dipolar Interfacial and space charge

Ionic Electronic

10–2

100

102

104

106 108 1010 Frequency (Hz)

1012

1014

1016

FIGURE 3.2.3 Frequency dependence of the dielectric constant and mechanisms of polarization.

8080_C003.fm Page 233 Friday, April 6, 2007 10:24 AM

Dielectric Materials: Selection and Design

233

transistor applications, electronic and ionic polarization will primarily be responsible for dielectric polarizability.

3.2.2.2 THICKNESS Minimizing dielectric thickness is an alternative to increasing the capacitance of these films subject to the constraint of acceptable leakage currents, effective insulation of the gate, and sufficient dielectric breakdown strength. As dielectric films get thinner, the potential for pinhole defects increases, resulting in larger leakage currents. In order to achieve the thinnest dielectric films, great attention must be devoted to organization of the film. Success has been achieved in generating ultrathin self-assembled monolayer films, in which densely packed alltrans aliphatic chains effectively insulate gate dielectrics [17–19]. Typically, minimum thickness are on the order of 10–20 nm for inorganics such as SiO2 and 100 nm for polymers, although examples of ultrathin cross-linked polymers have appeared (vide infra) [20]. For inorganic materials, tunneling currents can also be problematic for ultrathin dielectrics such as SiO2. Tunneling currents become problematic when band edge matching between semiconductor and dielectric occurs — for example, in the case of silicon and titanium dioxide [8]. There is some correlation between conduction band offset and band gap, and as a general rule larger band gap materials offer better insulation. However, there is a trade-off here because band gap and K are often inversely related, so improvement in insulating ability would be offset by lower K. Thinner films have a higher probability of defects due to the interfaces being very close. Highly defective dielectric films offer conduction pathways via Frenkel–Poole emission or via hopping conduction [8]. As films become thinner, dielectric breakdown becomes an issue. Breakdown strength is dependent on electric field, which, as a first-order approximation, equals the voltage divided by the thickness of the film [16]. As thickness decreases, higher electric fields are observed for a given voltage; for this reason, breakdown can occur at lower voltages.

3.2.2.3 DIELECTRIC ROUGHNESS Dielectric roughness is believed to reduce mobility in organic semiconductors due to the disorder induced in the accumulation layer. Roughness disturbs π-π-stacking, which is critical to efficient charge transport [1]. Streudel et al. [21] attempted to quantify the impact of surface roughness on mobility for organic TFTs. In this attempt, they used sputtered silica as a dielectric over various metal gate electrodes. The roughness of the metal was varied and the overlying silica gate dielectric translated this roughness to the dielectric–semiconductor interface. In these studies, it was determined that surface roughness (rms) variation, from 1.7 nm for normal thermal SiO2 to 92 nm for SiO2 over rough TiW, resulted in a 98% decrease in mobility for pentacene TFTs. The authors propose that the cause of this reduced mobility is due to increased grain boundaries in rough films as well as the irregular channel geometry. Since VT and activation energy do not change as a function of roughness, traps are excluded as a possible mechanism for reduced mobility.

8080_C003.fm Page 234 Friday, April 6, 2007 10:24 AM

234

Organic Field-Effect Transistors

3.2.2.4 FILM MORPHOLOGY The effects of film morphology on semiconductor properties have been studied with great intensity [4,22]. The importance of dielectric film morphology is also important to device performance since grain boundaries in a dielectric film can also serve as leakage pathways. Polycrystalline films can be unreliable and have microscopic variations in K [8].

3.2.2.5 IMPORTANCE OF INTERFACE AND SEMICONDUCTOR

BETWEEN

DIELECTRIC

Controlling the chemistry and physics of the dielectric–semiconductor interface is essential to controlling and optimizing the performance of organic electronic devices. There are two main issues associated with the dielectric–semiconductor interface: (1) the electrical properties of this interface, which are essential to the reliability, stability, and threshold voltage; and (2) the chemical/mechanical properties of the interface, which relate to the ability to organize the semiconductor layer for efficient charge transport. Avoiding fixed interface charge is important to ensure a low midgap density of states [8]. Lower midgap density of states results in less leakage through hopping mechanisms. Differences in interface trap densities are considered to be responsible for changes in threshold voltage as a function of device processing, as well as device lifetime. Traps have been studied using admittance spectroscopy [23]. The concentration of traps can be reduced through annealing processes (under vacuum; oxygen is a known dopant at interfaces). These studies point to the complexities of dielectric–semiconductor interface and the difficulties in controlling it. Volkel et al. studied pentacene devices on SiO2 and on Si3N4 [13]. The dielectric–semiconductor interface was found to be critical to device performance, where VT and onset voltage were dependent on the presence of acceptor states (responsible for positive onset voltages) observed for SiO2. Charge accumulation at the interface due to bias stress resulted in a negative drift of VT over time for Si3N4. One of the most revealing studies on the importance of the dielectric semiconductor interface was presented by Chua et al. [24]. They found that removing interface traps can unveil n-type mobility in organic semiconductors previously thought to show only p-type conduction. This result suggests that the semiconducting properties of an organic material are highly dependent on the nature of the dielectric and the dielectric interface.

3.2.2.6 PROCESSING Ease of processing is essential to the development of organic electronics as an inexpensive, easy to deploy technology. In this sense, manufacturing processes that avoid the expensive and cumbersome lithographic patterning steps would be highly desirable [9]. Ideally, solution-phase deposition methods that are compatible with printing techniques will be invented. At present, several solution-phase processes involving dip-coated dielectrics [25], spin-coated dielectrics [5], and layer-by-layer (LbL) deposited dielectrics have been developed [26]. Alternatively, selective dep-

8080_C003.fm Page 235 Friday, April 6, 2007 10:24 AM

Dielectric Materials: Selection and Design

235

osition of dielectrics using surface-initiated polymerization has been demonstrated in which selective deposition of polymer dielectric film occurs where thiol-modified initiators have attached to the gold gate electrode (vide infra) [27].

3.2.2.7 RELIABLITY In order to achieve useful devices, it is important that during the shelf-life and operating life of devices, the dielectrics maintain their integrity. The mechanisms of failure include charging of the dielectric, changes in dielectric resistivity/dielectric constant over time, and shorting of the dielectric. Charging of the dielectric leads to drifts in VT and could have adverse effects on semiconductor performance through generating of interface traps or doping of the semiconductor [24]. These effects can be caused by bias and/or temperature stress (BTS). It is found that BTS can be the result of induced defects and charge injection into the dielectric or into the dielectric–semiconductor interface. In the case of polar, or hygroscopic, dielectrics, fixed charge from substrates or through water present at surfaces can develop, causing large changes in VT. It is found that low surface energy dielectrics such as BCB [28] show excellent BTS properties, since low surface energy nonpolar interface does not attract charged species. Humidity can also affect the dielectric constant and resistivity of hygroscopic dielectrics such as PVA [7]. Efforts have been made to reduce this problem through the use of cross-linking agents, which reduce hygroscopicity [20].

3.2.3 MAJOR CLASSES OF DIELECTRIC MATERIALS This section will describe the standard dielectric materials that have been used for thin-film transistor applications. The two major classes of inorganic and polymer dielectrics will be described. In light of excellent recent reviews on these topics [5,7], the description of these materials will be kept brief and focus on the organic electronic applications in which these dielectrics are employed.

3.2.3.1 INORGANIC DIELECTRICS The benchmark dielectric has been silicon dioxide, which is typically grown as an amorphous thermal oxide on top of silicon. The system of thermal silicon dioxide and silicon forms the basis of complementary metal oxide semiconductor (CMOS) technology [8,15,29,30]. As a benchmark, silicon dioxide has excellent insulating properties due to its large bandgap (8.9 eV) and thermodynamic stability. As a result of intense research and development efforts, commercially available materials have very low dielectric charge density, low interface state densities, and very high dielectric breakdown strength (15 MV/cm). The ready availability of 300-nm thermal SiO2 on highly doped silicon has made this gate dielectric/gate electrode the substrate for initial evaluation of almost all new organic semiconductor materials [2,4]. Despite the excellent dielectric properties of silicon dioxide, which have served the CMOS industry faithfully for many years, silicon dioxide is not as easy to adapt to flexible applications and is not easy to print or solution process. Efforts will be described

8080_C003.fm Page 236 Friday, April 6, 2007 10:24 AM

236

Organic Field-Effect Transistors

TABLE 3.2.1 Dielectric Properties for Common Inorganics Dielectric

K

Thickness (nm)

Capacitance μF/cm–2) (Ci (μ

Leakage (Acm–2)

SiO2 (standard) SiO2 (thin) TiO2 Al2–O3 Ta2O5

3.9 3.9 80 9 26

300 2.1 8–12 4.8 5–6

0.01 1.6 6–34 1.7 4–5

10–7 0.1 High 10–8 <1

Ref. 8 8 8 8

in subsequent sections regarding attempts to use ultrathin anodized SiO2 [31], ultrathin SiO2 on alumina [32], and silica nanoparticles [26] as gate dielectric materials compatible with flexible substrates and potentially solution processable. Alternative higher K dielectrics are being considered, since silicon dioxide is approaching materials limitations. Transistors with SiO2 dielectrics of thickness as low as 10–12 now have been fabricated, but thin SiO2 transistors have large leakage currents. Higher K materials could provide equivalent performance using thicker films, and thereby provide attractive alternatives to SiO2. Several different high-K materials have been considered [8] and some representative examples are listed in Table 3.2.1. As can be seen from the table, titanium oxide has a very high dielectric constant (~80 depending on crystal structure). This high permittivity is due to strong contribution to polarization from soft phonons involving titanium ions [8,33–35]. However, titanium dioxide films tend to have high defect densities that create high leakage paths, which limit its usefulness as a dielectric. Efforts to incorporate titanium dioxide into composite and layered dielectrics for use in organic TFTs will be described in subsequent sections [35,36]. These methods use SAMs and surface treatments to reduce leakage currents in TiO2. Tantalum oxide also exhibits a high permittivity (K = 26), which makes it an attractive candidate for high-K applications [8,29]. It has also received attention from the organic TFT community in the form of an anodized gate dielectric (vide infra) [31]. Alumina is another choice for a higher K dielectric with K = 9. This material has also been investigated by the organic TFT community as an alternative in anodized films [37] and also as a substrate that can be substantially improved through self-assembled monolayers [38]. Table 3.2.1 compiles data for various high-K inorganic dielectric materials. The capacitance per unit area column is instructive. In some of the thinnest films produced, capacitances in the microfarad per square centimeter range are observed. These are excellent values, and at present few materials that are solution processable and compatible with organic TFTs achieve these high capacitances. As this review will outline, issues of compatibility of inorganic dielectrics with organic semiconductors typically have to be addressed to generate high-quality organic TFTs with these materials. Subsequent sections will describe compatibility using organic

8080_C003.fm Page 237 Friday, April 6, 2007 10:24 AM

Dielectric Materials: Selection and Design

237

TABLE 3.2.2 Dielectric Properties for Polymers Dielectric PVA PMMA PS PEO-Li perchlorate PVP BCB Parylene-C CYEPL

K

Minimum thickness (nm)

Capacitance (Ci (nF/cm–2)

Leakage (Acm–2)

Ref.

500 160 122 400 450 50 ~1,000 1200

17.8 19.5 19 60–110 μF/cm 7.4 47 2.2 —

— — 10–4–10–7 High 10–7 ~10–5 10–11 High

7 39 20 40,41 42 43 44 45

10 3.5 2.6 6.4 2.65 3.1 12

surface treatments and anodization methods that attempt to make otherwise brittle material flexible when they are kept very thin.

3.2.3.2 POLYMER DIELECTRICS Polymeric materials present a second major class of materials that have been selected as gate dielectrics for organic TFT applications. Polymers offer the advantage of being solution processable and are typically applied using dip-coating and spin-coating methods. Polymers are flexible and present no special challenge in generating flexible devices. Several polymers show excellent insulating properties with exceedingly low leakage currents. Furthermore, processing methods can generate smooth and reproducible polymer films that can form excellent interfaces with organic material and provide a template for ordering organic materials. Some common examples of polymers used as gate dielectric materials are compiled in Table 3.2.2 and illustrated in Figure 3.2.4. Despite all the excellent properties of polymeric materials, they typically do not have very high dielectric constants and, for this reason, exhibit Ci values typically orders of magnitude below those of the inorganic materials. Common polymers such as polystyrene (PS) and polymethylmethacrylate (PMMA) have been used as gate dielectric materials [7,20,46]. Their ready availability made them some of the early polymers investigated by researchers in the field [11,39,47]. However, the low capacitances of these films made them less attractive than other polymers. Poly(4-methylstyrene) has been explored as a possible polarizable gate insulator [48]. Poly(vinyl alcohol) PVA and poly(vinyl phenol) (PVP) are two of the most widely used polymer dielectrics for organic electronics applications [45,49–52]. These two polymers have the capability of being deposited onto an organic semiconductor to generate a top gate dielectric, since the solvent system for these materials (aqueous systems for PVA and ethanol for PVP) are orthogonal to the semiconductor (i.e., do not disturb the semiconductor layer). Cross-linking of these polymer films using agents such as melamine-co-formaldehyde [42,51] or

8080_C003.fm Page 238 Friday, April 6, 2007 10:24 AM

238

Organic Field-Effect Transistors

O n

O

O

Si

n

O

O

Si

OH Poly (vinyl alcohol) PVA polystyrene PS

BCB precursor

n

X O n

OH

O

Poly (vinylphenol) PVP

X RO

n

O

O

Poly (ethylene oxide)

RO

O OR

X

O n

Parylene-C (X = Cl)

O

Poly (methyl methacrylate) PMMA

Cyanopulluane CYPEL R = CH2CH2CN; or H

FIGURE 3.2.4 Polymers and polymer precursors used as gate dielectrics in organic TFTs.

hexamethylene tetraamine further improves the robustness of these dielectrics. Patterning of PVP films has been demonstrated using photolithography and oxygen plasma etching in a process to generate fully organic TFTs [49]. PVP has been reported to smooth relatively rough gate electrodes such as PEDOT on flexible substrates [42]. PEDOT forms a relatively rough surface (rms roughness = 15 nm). Treatment with spun-on PVP and cross-linking (thickness 450 nm) generates a smoother surface (rms ~3 nm) and better devices. Pentacene films grow well on this substrate (capacitance of 7.4 nF/cm2), generating devices with mobilities as high as 0.24 cm2V–1s–1 that show low leakage current on the order of 10–7 A/cm2. A study of how a dielectric constant affects mobility of pentacene TFTs was performed in which the dielectric constant of PVP was varied through a difference in the amount of cross-linking agent (poly(melamine-co-formaldehyde) [PMF]) [53]. Ratios of 40:1–1:1 PVP:PMF generated a change in dielectric constant from 4.3 to 3.6. This change also correlated with a reduction of mobility from 0.81 to 0.26 cm2V–1s–1. The researchers showed that surface energy and surface roughness were the same in both cases (surface energy = 42.2–42.6 mJ/m2; rms ~ 0.3–0.4 nm) and therefore claimed that the reduction in mobility was most likely due to the lower dielectric constant. The opposite trend has been observed in other reports, which suggest that higher K materials lower mobility in organic TFTs [5,54]. The high threshold voltages (ranging from –14.5 to –21.5 V) suggest significant charge and traps at the interface. The high-K polymer cyanoethylpullulan (K =12) has also been used in dielectric applications [45]. This polymer contains highly polar cyano groups that cause it to have permittivity (see structure in Figure 3.2.4). Deposition of CYEPL from solutions of acetonitrile and dimethyl formamide (DMF) allows for orthogonal deposition on many organic semiconductors, permitting use of CYPEL in both top and bottom gate structures. Low on/off ratios (~10) make this material less attractive.

8080_C003.fm Page 239 Friday, April 6, 2007 10:24 AM

Dielectric Materials: Selection and Design

239

The spin-on polymer precursor divinyl-tetramethyl-disiloxane-bis(benzocyclobutene) (BCB) has been used recently as a gate dielectric after thermal cure (see Figure 3.2.4 for structure) [24,43]. Films as thin as 50 nm show excellent breakdown voltage (greater than 3 MV/cm) and low leakage currents (below 10–6 A/cm2). Reasonably high capacitances (Ci = 47 nF/cm2) are achieved in spite of the low K for this polymer (K = 2.65) due to the ability to generate such thin films. One drawback of this system is the high temperatures (>230°C) required for cure. These high temperatures preclude the use of this dielectric on several flexible substrates. Efforts are under way to reduce this cure temperature through analogous polymer precursors. Low fixed charge density and interface states have been reported for this material. A recent report using BCB as a dielectric states n-type conduction has been observed for semiconducting polymers such as polyfluorene and poly(p-phenylenevinylene), which have otherwise been considered exclusively p-type semiconductors [24]. It is proposed that n-type conduction in these polymeric semiconductors has not been observed previously due to trapping of electrons at the dielectric–semiconductor interface due to hydroxyl groups. This study highlights the critical nature of the dielectric–semiconductor interface and its role in modifying the performance of the semiconductor. Parylene C has been used as a polymer precursor to generate dielectric films and is very useful in generating conformal top-gate structures since it can be deposited using CVD on top of semiconductor active layers. Early reports from IBM showed that dihexylsexithiophene (DH6T) devices made using parylene dielectrics exhibited the highest mobilities reported at that time for DH6T [39]. The structure of parylene C is illustrated in Figure 3.2.4 along with its corresponding polymer. Parylene C is vaporized at ~150–200°C under vacuum and pyrolyzed at ~700°C to generate the ring-opened diradical, which polymerizes as it is deposited upon the semiconductor close to room temperature. The conformal insulating film formed is typically ~1 μm thick and has K ~ 3.1. The capacitance of parylene films is 2.2 nF/cm2. Low leakage and high on/off ratios are possible with parylene (on/off ratio of 106) [44]. A variety of substituted parylene derivatives have been used as gate dielectrics for TFTs in a control study to investigate the effect of surface energy on channel characteristics and mobility [55]. It was found that the field effect mobility for an amorphous polymer semiconductor (MEH-PPV) increases with increasing surface hydrophobicity (a result that has been seen with amorphous poly(triarylamine)s [PTAAs] also) [5]. The ability to deposit parylene C dielectrics using vapor phase deposition methods makes it very useful for sensitive semiconductors (in the case of top gate devices), such as single crystals. Researchers at Rutgers and Bell Labs have made extensive use of this dielectric in single-crystal field-effect transistors [56,57]. Using parylene C as a top gate dielectric, a double gate TFT device was generated in which the mobility at top and bottom interfaces of a single pentacene film was measured. Top gate mobility was 0.1 compared to 0.4 for bottom gate device. The roughness of the top interface was considered as the reason for the decreased mobility of the top gate device [44].

8080_C003.fm Page 240 Friday, April 6, 2007 10:24 AM

240

Organic Field-Effect Transistors

Top gate devices composed of poly(ethylene oxide) (PEO)-lithium perchlorate have been reported [40,41]. These polymer electrolyte gate dielectrics exhibit extremely high capacitances (60–110 μF/cm2), exceeding even the best values reported for inorganic dielectrics (see previous discussion). The large capacitances are reported to be due to redox chemistry at the semiconductor–dielectric interface. However, detailed understanding of this phenomenon will require further research efforts. Hysteresis effects and low mobilities for both n- and p-type semiconductors are observed, but the authors of the study believe they can operate devices at switching speeds as high as 1 kHz [41].

3.2.4 ALTERNATIVE GATE DIELECTRIC STRATEGIES 3.2.4.1 GATE DIELECTRICS THROUGH ANODIZATION THIN- METAL FILMS

OF

In the case of the SiO2 heavily doped silicon system, the gate and dielectric come together and offer a convenient platform to construct TFT devices. In an attempt to transfer this concept to flexible systems, researchers have investigated the use of sputtered thin metal films, which are subsequently anodized to generate a gate electrode/dielectric combination that may be transferable to flexible films [31,36,37]. To this end, tantalum metal films have been sputtered onto silicon substrates and anodized to generate ~50-nm thick amorphous tantalum oxide insulating films (Ta2O5) [31]. The relatively high capacitance is considered desirable (K ~ 23). Thin film transistors have been fabricated with this gate/dielectric system using dihexylquinquthiopene (DHQT) and copper perfluorophtalocyanaine organic semiconductors. Devices with DHQT (p-channel) showed mobilities of 0.02–0.04 cm2V–1s–1. These values are slightly lower than values of mobility for thermal oxide. n-Channel devices with F16CuPc show mobilities ~ 0.02, which are comparable to thermal oxide. Leakage currents were quite low (below 10–8 A/cm2) for tantalum oxide films with breakdown strengths ~ 4–5 MV/cm. Silica was also grown from heavily doped silicon and performs comparably to thermal oxide, although larger leakage currents of 10–6 A/cm2, compared to 10–12 A/cm2 for thermal SiO2 [31], are observed. Researchers have also explored the use of anodized alumina/aluminum gate dielectric/gate combinations as well as titanium oxide/titanium [36,37]. In the alumina/aluminum system, aluminum was sputtered onto glass and anodized to generate films of alumina between 26 and 130 nm thick, which exhibited excellent gate insulation (Igate < 10–9A/cm2). However, the mobility of PTAA was not very high (2.7 × 10–5 cm2V–1s–1) and the low mobility was suggested to be the result of the high surface energy of the alumina substrate, which was expected to disrupt charge transport in the semiconductor through interface trap states. Surface treatments with self-assembled monolayers (vide infra) are typically employed to address this problem. Finally, hysteresis effects are observed with anodized alumina and are thought to be due to residual ionic species (Al(OH)3) remaining in the oxide layer after anodization [37].

8080_C003.fm Page 241 Friday, April 6, 2007 10:24 AM

Dielectric Materials: Selection and Design

241

In the study of anodized titanium oxide/titanium, commercially available polyester (Mylar) was used as the substrate with a 200-nm thick film of titanium deposited by evaporation [36]. The titanium metal was anodized in generate 7- to 8-nm thick titanium dioxide films, which showed high leakage (due to TiO2 being a wide band gap semiconductor). Capacitors were made with this dielectric, and these showed a capacitance of 2.42 μFcm–2 and an effective dielectric constant Keff = 21. Due to the high leakage currents, unmodified anodized TiO2 did not work for TFT devices. However, treatment of anodized TiO2 with an ultrathin film of poly(αmethyl styrene) (10 nm) reduced the leakage current significantly enough to generate working pentacene TFT devices with VT ~ –0.5 V, 104 on/off ratio, low hysteresis, and excellent mobilities (0.8 cm2V–1s–1). The effective capacitance drops with the use of a polymer thin insulating layer to 228 nFcm–2, but this is still a very high capacitance and results in devices that can be turned on at roughly 1 V. Anodization is an intriguing approach to generating controlled thickness dielectric films. Sputtering the metals onto flexible substrates opens up the possibility of flexible devices. Anodization is a low-temperature process that occurs in aqueous solution, which suggests potentially easy processing methods.

3.2.4.2 SURFACE TREATMENT

OF INORGANIC

MATERIALS

It has been observed in several studies that the use of organic monolayers or thin polymer phases to modify the surfaces of inorganic dielectric materials often improves the performance of these devices. This strategy has been employed for modification of SiO2 [58] and Al2O3 surfaces [38] and more complicated dielectric composite materials. Researchers at 3M introduced the use of alkyl phosphonic acid monolayers as SAMs to improve performance of pentacene on alumina TFTs [38] in which mobilities as high as 2 cm2V–1s–1 were routinely achieved with high mobilities exceeding 3 cm2V–1s–1. The compound 1-phosphonohexadecane showed the best results, forming SAMs approximately 1.8 nm thick. The threshold voltage (–5 V) is similar to that of alumina. The SAMs appear to smooth roughness in the oxide surface and improve pentacene crystalline grain growth. OTMS (octadecyl-trimethoxysilane) has also been used to smooth and enhance the performance of zirconium oxide dielectrics [59]. Surface treatment has also been used to modify the threshold voltage as well as measured mobility in pentacene and C60 TFTs [60]. These transistors have a heavily doped silicon/silicon oxide gate dielectric structure where alkyl, alkylamine, and fluoroalkyl silanes are used to modify the SiO2. Evaporated pentacene and C60 form the active p- and n-type semiconductors. The experimental effect of these monolayer treatments is to alter VT and effective mobility dramatically (see Table 3.2.3). For pentacene, the mobility decreases from –F, –CH3, untreated, –NH2, with a similar shift in VT from 17 to –11 V. The opposite trend is observed for C60, in which mobility is largest for the untreated material and smallest for the fluorinated SAM. In the case of VT, the alkylamine SAM shows the lowest VT of 5.3 V. The underlying reasons for these trends are not completely understood. What is intriguing is how dramatic

8080_C003.fm Page 242 Friday, April 6, 2007 10:24 AM

242

Organic Field-Effect Transistors

TABLE 3.2.3 Effect of SAMs on Mobility and Threshold Voltage for Pentacene (p-Type) and C60 (n-Type) Organic Semiconductor TFTs Surface treatment

Mobility (pentacene)

VT (pentacene)

Mobility (C60)

VT (C60)

F-SAM CH3-SAM Untreated NH2-SAM

0.2 0.13 0.086 0.0024

17 5 –11 –11

0.005 0.07 0.2 0.1

39 37 18 5.3

an effect SAMs can have on threshold voltage for devices — especially in the case of n-type materials such as C60 — where high VT has been a problem. A recent study, in which octadecyltrimethyoxysilane treatment of SiO2 improves mobility for F16CuPC by ~one order of magnitude (from 10–3 to 10–2), has achieved significant improvement in mobility for an n-type semiconductor [61]. No change is observed in VT ~ 50 V ± 5 V, suggesting trap states still present between the organic monolayer and SiO2. As in p-type devices, the surface treatment is shown to increase the grain size of F16CuPC, creating more continuous charge conduction pathways. Polymers have also been used as modifiers to improve the performance of TFTs by reducing dielectric roughness [62]. In this study, smooth SiO2, rough SiO2 (roughened through reactive ion etching), and polymer (polystyrene coated) SiO2 were compared as gate dielectrics with evaporated pentacene. Saturation mobilities dropped from 0.31 to 0.02 cm2V–1s–1 after surface roughening; however, thin polystyrene (10 nm) films improved saturation mobility on rough substrates to as high as 0.94 cm2V–1s–1.

3.2.4.3 SELF-ASSEMBLED MONOLAYERS/MULTILAYERS In the previous section, SAMs were used to modify the properties of other insulators. Amazingly, under careful preparative conditions, SAMs can function as a monolayer dielectric. This work was pioneered by the Vuillaume group [17–19], who investigated the use of alkyltrichlorosilanes as monolayer insulators in metal insulator silicon structures. The silicon contained a thin native oxide layer (1–1.5 nm). The use of octadecyl trichlorosilane as a monolayer was found to reduce leakage currents five orders of magnitude, to as low as 10–8 A/cm2 [19]. Leakage currents for silicon oxide at thicknesses below 3 nm are 10–3–10–1 A/cm2. The key to forming good monolayer dielectric films was found in controlling the order in the monolayer. In the case of alkyltrichlorosilanes, it is imperative to have an all trans structure for these films to maintain low leakage currents, and the quality of these films is controlled by keeping the temperature below a critical value during solution deposition of these films. Experiments performed in which the monolayers are deposited above the critical temperature result in disordered films containing gauche defects

8080_C003.fm Page 243 Friday, April 6, 2007 10:24 AM

Dielectric Materials: Selection and Design

243

and have much greater leakage currents [17]. Chain length effects are not considered as critical to the resistance of these monolayers, based on studies of C12, C16, and C18 alkyl trichlorosilanes. This suggests that as long as dense monolayers are constructed, tunneling currents will be minimal for at least C12 monolayers and greater. Recent results have appeared using this monolayer approach in which docosyltrichlorosilane is used to form a dense monolayer dielectric on n-doped silica. This resulting 2.6-nm monolayer shows a capacitance of 4.5 × 10–7 F/cm2; using P3HT, organic thin film transistors (OTFTs) were fabricated with a measured mobility of 5 × 10–3 and on/off ratio of 200. Halik et al. have also made devices using SAMs that show excellent resistance to penetration from the organic semiconductor [50]. Their devices show pentacene mobilities of 1 cm2V–1s–1, on/off ratios of 106, and excellent insulator properties (gate leakage of 10–9A/cm2). A hybrid approach between monolayer and polymer film involves the use of surface-initiated polymerization (SIP) using ring opening metathesis polymerization (ROMP) to generate dielectric films directly attached to the gate electrode [27]. In this approach, norbornene functionalized with a terminal thiol group was used to generate SAMs on gold electrodes. These moieties function as initiators for ROMP. In the presence of the appropriate ruthenium catalysts, surface-initiated polymerization generates dense polymer films ranging from 1.2 to 2.5 μm. Annealing reduces pin-hole defects and leakage currents. Pentacene TFTs showed good mobilities (0.1–0.3 cm2V–1s–1), but low on/off ratios (10–100). No hysteresis was observed. One of the benefits of this SAM approach is the ability to grow dielectric films specifically from the gate electrode. Cross-linked ultrathin polymer dielectrics present a modification to the use of homopolymers as gate dielectrics and have succeeded in generated ultrathin dielectrics with very high capacitances (300 nF/cm2) and low leakage currents (as low as 10-8 A/cm2) [20]. Yoon et al. have modified PVP with the cross-linking agent 1,6bis(trichlorosilyl)hexane (C6). This agent significantly improves the insulating quality of these films, allowing films as thin as 10–20 nm to be spun from solution onto a variety of substrates (ITO/glass, heavily doped silicon, and aluminum foil). The comparison is made between these ultrathin cross-linked films and conventional homopolymers, which need to be >>300 nm to achieve sufficient insulation, at which point capacitances drop to values of approximately 20 nF/cm2. PVP and PS at 20 nm thickness have very high leakage currents preventing the measurement of meaningful capacitances for these films. These ultrathin cross-linked films are generated by spin coating the substrate with the mixture of PVP and C6 and curing in air or under vacuum at ~110°C for 10–15 min. Chemical cross-linking (see Figure 3.2.5) of PVP and C6, as well as covalent linkages to the gate electrode, make this dielectric robust for further solution-based deposition steps. Furthermore, the substrates are shown to be patternable. In the case of evaporated gold masks, RIE or BOE can etch exposed polymer, leaving behind a patterned dielectric on removal of gold masks. The patterned structures exhibit excellent dielectric properties identical to nonpatterned films. Cross-linking of ultrathin PVP with C6 also renders this dielectric film less hygroscopic (a problem affecting PVP reproducibility). Low losses are reported (<0.1 at 10 kHz); low hysteresis and low threshold voltages (less than 2 V) suggest

8080_C003.fm Page 244 Friday, April 6, 2007 10:24 AM

244

Organic Field-Effect Transistors

n n

+

Cl3Si

SiCl3

RO

OH

Si O OR

Poly (vinylphenol)PVP RO Si O OR

Crosslinked PVP

n

FIGURE 3.2.5 Cross-linking of PVP with bis(trichlorosilyl)hexane (C6) generates ultrathin, low-leakage dielectrics.

low fixed charge in the interface. TFT devices fabricated using this novel dielectric and pentacene (evaporated), DH-6T (solution cast), and P3HT (spun on) as well as CuFPc (n-type) demonstrate the generality of this dielectric to create well ordered thin films. For example, pentacene TFTs operate at 4 V, with mobility > 0.1 and on/off ratio ~ 104. Devices tested after 13 months show negligible degradation in performance. These cross-linked ultrathin polymer films show excellent promise, and further investigation in polymers beyond PVP will no doubt expand the utility of this dielectric design concept (see Table 3.2.4). Majewski et al. have looked at ultrathin SiO2 on metallized (aluminum metal) Mylar [32]. These systems involve 60-nm aluminum on Mylar as the gate electrode, with SiO2 3.5 nm (~ three times the limit predicted by theory and experiments on SiO2/Si). In these ultrathin SiO2 dielectrics, capacitances as high as 1 μF/cm2 are achieved. Addition of an OTS (octyltrichlorosilane) SAM to this system reduces the

TABLE 3.2.4 Dielectric Data for Ultrathin Cross-Linked Polymers, SAMs, and Anodized Films

Dielectric

K

Cross-linked PVP Anodized Ta2O5 Anodized alumina Anodized TiO2 with p(AMS)

6.4 23 9 21 for TiO2

Ultrathin SiO2 on aluminum Octadecyltrichlorosilane SIP/ROMP

3.9 — —

Minimum thickness (nm)

Capacitance (Ci (nF/cm–2)

Leakage (Acm–2)

10–20 50 130 8 nm TiO2 + 10 p(AMS) 3.5 2.8 nm 1,200

300 ~400 ~60 228; ~2400 for TiO2 alone 300 450 3

10–8 10–8 10–9

10–8 —

Ref. 20 31 37 36 32 19 27

8080_C003.fm Page 245 Friday, April 6, 2007 10:24 AM

Dielectric Materials: Selection and Design

245

capacitance to 300 nF/cm2. OFETs of pentacene and rr-P3HT on this dielectric perform with high mobility (0.12 cm2/Vs for pentacene and 0.01 for rr-P3HT). Excellent breakdown voltages are achieved (greater than 7 × 108 V/m). VT is ~–1 V. The device operates below 2 V and at this voltage produces currents ID ~ 1.8 μA. Without surface treatment, much lower (up to 10×) mobilities are observed, with larger leakage currents.

3.2.4.4 NANOCOMPOSITE

AND

NANOSTRUCTURED DIELECTRICS

The challenging materials requirements for a suitable gate dielectric for flexible organic thin-film transistors are not easily met by a single material. As has been demonstrated in previous sections, the use of multilayers comprising different materials is often required to balance the strengths and weaknesses of individual polymer and inorganic materials. For example, high-K ceramics offer excellent dielectric constants, but suffer from poor interfaces with organic semiconductors. Alternatively, polymer dielectrics typically have low capacitances, but excellent film-forming properties. In this section, we will explore recent efforts to blend ceramic and polymeric materials on the nanoscale to generate a nanostructured dielectric with an intelligent blending of the desirable properties of its constituents. The general strategy has been to load a polymeric material with high-capacitance nanoparticles. Table 3.2.5 summarizes the different composite materials described in this section. Early examples of polymer ceramic nanocomposites generated using micron and larger size particles of titanium oxide and polystyrene, in which composites were made through mechanical mixing, have been reported. Poor adhesion and air pockets presented a problem to this approach [63]. With increasing interest in nanotechnology and greater understanding of the control of nanoparticle synthesis, efforts to generate polymer dielectric films loaded with high-K dielectric materials appeared in the literature. An early example involved the sol-gel synthesis of bismuth titanate nanoparticles within a polyacrylate matrix [64]. Roughly 50-nm diameter particles of amorphous bismuth titanate were observed in the polyacrylate matrix using transmission electron microscopy (TEM). Effective dielectric constants as high as 10 were reported; however, detailed characterization of the material was not reported.

TABLE 3.2.5 Dielectric Properties of Nanoparticle Based Dielectrics Material

K

Thickness (nm)

Capacitance (Ci (nF/cm2)

Leakage (A/cm2)

Bismuth titanate-polyacrylate BaTiO3-PVA BaTiO3 Luxprint/smoothing layer TiO2-PVA TiO2-PS SiO2-PDDA LbL

10 9–12 — 5.4 8.2 6

— 170 10–15 μm 600 870 300

— 62.5 ~1 — 8 —

— 10–8 — — —

Ref. 64 66 54 65 35 26

8080_C003.fm Page 246 Friday, April 6, 2007 10:24 AM

246

Organic Field-Effect Transistors

Efforts at loading titanium oxide nanoparticles in PVA (commercially available from Nanophase) have been reported [65]. In this report, titanium nanoparticles are dispersed in an aqueous solution of PVA with poly(melamine-co-formaldehyde). The solution is spun onto substrate and heated to generate a cross-linked polymernanoparticle dielectric. A modest enhancement of dielectric constant is achieved for 600-nm thick films. Thin-film transistors using this composite show excellent pentacene mobility (> 0.2 cm2V–1s–1) and reasonable on/off ratios ~ 103. VT ~ –7V is reasonably high, suggesting static charge at the dielectric–semiconductor interface. Efforts combining high-K materials such as barium titanate with lower K polymers to achieve processable flexible OFETs has been attempted by Zielke et al. [54]. In this effort, a BaTiO3-filled polymer (Luxprint 8153E) was used as a spin on dielectric to generate a top gate structure over offset printed source drain contacts, and spun on semiconductor (PTAA). The offset printed source drain electrodes were extremely rough (rms roughness of 75 nm with peaks extending as high as 3 μm). For this reason, a thick gate dielectric was required (10–15 μm thick) to smooth the roughness in the films. Barium titanate (high K) allows such thick films still to have reasonable capacitance, but BaTiO3 is highly polar and is thought to induce traps in the semiconductor, thereby reducing mobility dramatically [5]. In order to insulate the semiconductor from the BaTiO3, a second insulating layer of butylene copolymer (P105, 100–600 nm in thickness) that has a K of 2.2 is used. This low-K insulating film improves mobilities. Devices with Luxprint alone showed mobilities ~ 4 × 10–4, while mobilities in devices with the low-K interlayer increased to 3 × 10–3. The lack of control in the Luxprint structure creates a very ill-defined dielectric material, with widely varying threshold voltages (0 ± 8V) and significant hysteresis effects. However, this study clearly demonstrates the ability of high-K films to overcome problems such as rough electrode morphology. Low-K polymeric materials are not able to operate at such high thickness since the voltages required would be too high to turn on devices. Flexibility of the films is also demonstrated by repeated flexing of aluminum foil devices where no loss of function is observed. Another report of barium titanate nanoparticle composite dielectrics describes the combination of commercially available barium titanate nanopowders with PVA and also PVA-co-poly(vinyl acetate)-co-poly(itaconic acid) (PVAIA) in aqueous solution to generate a well dispersed nanocomposite that can be spun on as gate dielectric [66]. Pentacene-based transistors using this barium titanate/PVA nanocomposite show promising effective dielectric constants (11–12), low threshold voltages (–0.8–1.2 V), and high field-effect mobilities (~0.4 cm2V–1s–1). In this report, the level of loading of barium titanate was not reported. It was observed that higher dielectric constant films (~12) showed somewhat lower mobility than films with a K ~ 9 (mobility ~ 0.4 cm2V–1s–1). Prior attempts at nanocomposite dielectrics made no attempt to control the nanostructure of the polymer/dielectric material through controlling the chemistry between the nanoparticle and the dielectric. The first example of the use of a well characterized core-shell nanostructured dielectric material appeared in 2005 [35]. Researchers at Bell Labs used narrowly dispersed anatase phase titanium oxide nanoparticles (rod-shaped; ~15 × 4 nm; K = 31) as the high-K core material (see

8080_C003.fm Page 247 Friday, April 6, 2007 10:24 AM

Dielectric Materials: Selection and Design

247

Flexible high-K dielectric film Core TiO2

TiO2

TiO2

TiO2

TiO2 Polystyrene shell

TiO2 TiO2

TiO2 TiO2

100

TiO2

TiO2

TiO2

TiO2 TiO2

TiO2

TiO2

TiO2

TiO2

TiO2

Particle size distribution of TiO2-oleic

Volume

80 60 Particle size distribution of TiO2-PS

40 20 0 0

20

40 60 80 Particle diameter (nm)

100

FIGURE 3.2.6 Titanium oxide core–polystyrene (TiO2–PS) shell nanoparticle gate dielectrics. Bottom graph shows particle size distribution of as-synthesized TiO2–oleic, and polystyrene functionalized TiO2–PS particles as determined by DLS.

Figure 3.2.6). Narrowly dispersed polystyrene (synthesized by atom transfer radical polymerization [polydispersity < 1.1]) was end functionized with a phosphonate moiety that binds strongly to titanium oxide. The combination of narrowly dispersed titanium oxide and narrowly dispersed phosponate-terminated polystyrene generates a narrowly dispersed core-shell architecture as measured by dynamic light scattering, which can be spun into dielectric films. The covalent coating of polystyrene around titanium oxide is helpful at preventing aggregation of the nanoparticles in organic dispersion and in thin films. The resulting films show effective dielectric constants of ~8 with very low frequency dependence. The loading level for these core-shell films is much higher (almost 20 vol% titanium dioxide) as compared to other nanoparticle-polymer films. The low dielectric constant polystyrene shell (K = 2.5) allows for an excellent interface formation with evaporated organic semiconductors. Pentacene TFTs generated using these dielectrics show excellent mobilities (approaching 0.2 cm2V–1s–1) in unoptimized devices. On/off ratios are ~500 for unpatterned devices. The presence of titanium dioxide in these dielectric films reduces their breakdown voltage from >100 V/μm for polystyrene to ~50 V/μm for ~10 vol% films, and ~25 V/μm for 18 vol% films. Controlling the covalent chemistry between different components in a nanostructured composite dielectric is anticipated to be a general strategy to significant improvement in the dielectric material and its interface with organic semiconductors. Continued efforts to develop this core-shell technique with other high-K ceramic nanoparticles and through better tuning of polymer shells is under way at Bell Labs.

8080_C003.fm Page 248 Friday, April 6, 2007 10:24 AM

248

Organic Field-Effect Transistors

An alternative nanostructured gate dielectric involving a self-assembly process using SiO2 nanoparticles has been reported [26]. In this procedure, a layer-by-layer approach is used to build up a silica nanoparticle top gate dielectric over a pentacene FET. The polyelectrolyte, poly (dimethyldiallylammonium chloride) (PDDA), is used with a thin layer of polystyrene to generate the first layer of a multilayer structure. SiO2 nanoparticles are then deposited by dipping of the substrate into a dispersion of these particles. The procedure is repeated several times to build up multilayers of silica nanoparticles and PDDA. After several layers, 300-nm films were produced that contained a silica volume fraction of ~70% with 10% PDDA and 20% air-filled pores. The dielectric constant for this silica-PDDA multilayer structure is found to be ~6 as compared with 3.9 for silica.

3.2.5 SUMMARY AND CONCLUSIONS This review has sought to illustrate a variety of different strategies employed to achieve high-capacitance, low-leakage, stable, and reliable dielectric films. The importance of the dielectric layer to the performance of organic TFT devices cannot be overstated. The dielectric is responsible for effective charge injection into the semiconductor and organization of the channel at the dielectric interface, and is critical to electrical parameters such as threshold voltage, leakage current, on/off ratio, and mobility. Various approaches involving inorganic, polymer, SAMs, anodization, and nanocomposites have been discussed. Many of these different approaches will find practical applications, depending on the requirements of the particular application and its prerequisite processing steps. Continued efforts will no doubt generate new approaches and improve existing strategies, generating a robust library of dielectric materials to enable the development of organic electronics, especially with regard to the development of mass printable organic and flexible thin-film transistor devices.

REFERENCES 1. Sirringhaus, H., Device physics of solution-processed organic field-effect transistors, Adv. Mater. 17 (20), 2411–2425, 2005. 2. Forrest, S.R., The path to ubiquitous and low cost organic electronic appliances on plastic, Nature 428, 911–918, 2004. 3. Katz, H., Recent advances in semiconductor performance and printing processes for organic transistor-based electronics, Chem. Mater. 16, 4748–4756, 2004. 4. Dimitrakopoulos, C. and Malenfant, P., Organic thin film transistors for large-area electronics, Adv. Mater. 14, 99–117, 2002. 5. Veres, J. et al., Gate insulators in organic field effect transistors, Chem. Mater. 16, 4543–4555, 2004. 6. Ruiz, R. et al., Thickness dependence of mobility in pentacene thin-film transistors, Adv. Mater. 17, 1795–1798, 2005. 7. Facchetti, A., Yoon, M.H., and Marks, T.J., Gate dielectrics for organic field-effect transistors: New opportunities for organic electronics, Adv. Mater. 17 (14), 1705–1725, 2005.

8080_C003.fm Page 249 Friday, April 6, 2007 10:24 AM

Dielectric Materials: Selection and Design

249

8. Wilk, G., Wallace, R., and Anthony, J., High-k gate dielectrics: Current status and materials properties considerations, J. Appl. Phys. 89, 5243–5275, 2001. 9. Gamota, D. et al., Printed organic and molecular electronics, Kluwer Academic Publishers, Boston, 2004. 10. Horowitz, G., Organic field effect transistors, Adv. Mater. 10, 365–377, 1998. 11. Peng, X. et al., All-organic thin-film transistors made of alpha-sexithienyl semiconducting and various polymeric insulting layers, Appl. Phys. Lett. 57, 2013, 1990. 12. Horowitz, G., Organic thin film transistors: From theory to real devices, J. Mater. Res. 19, 1946–1962, 2004. 13. Volkel, A., Street, R., and Knipp, D., Carrier transport and density of state distributions in pentacene transistors, Phys. Rev. B. 66, 195336, 1–7, 2002. 14. Kasap, S., Principles of electrical engineering materials and devices, Irwin–McGraw–Hill, Boston, 1997. 15. Wallace, R. and Wilk, G., High-k gate dielectric materials, MRS Bull. March 2002, 192–197, 2002. 16. Ohring, M., Engineering materials science, Academic Press, San Diego, 1995. 17. Boalas, C. et al., Suppression of charge carrier tunneling through organic self-assembled monolayers, Phys. Rev. Lett. 76, 4797–4800, 1996. 18. Collet, J. and Vuillaume, D., Nanofield effect transistor with an organic self-assembled monolayer as gate insulator, Appl. Phys. Lett. 73, 2681–2683, 1998. 19. Fontaine, P. et al., Octadecyltrichlorosilane monolayers as ultrathin gate insulating films in metal insulator semiconductor devices, Appl. Phys. Lett. 62, 2256–2258, 1993. 20. Yoon, M.H. et al., Low-voltage organic field-effect transistors and inverters enabled by ultrathin cross-linked polymers as gate dielectrics, J. Am. Chem. Soc. 127 (29), 10388–10395, 2005. 21. Steudel, S. et al., Influence of the dielectric roughness on the performance of pentacene transistors, Appl. Phys. Lett. 85, 4400–4402, 2004. 22. Laquindanum, J., Katz, H., and Lovinger, A., Synthesis, morphology, and field-effect mobility of anthradithiophenes, J. Am. Chem. Soc. 120 (4), 664–672, 1998. 23. Torres, I. and Taylor, D.M., Interface states in polymer metal–insulator–semiconductor devices, J. Appl. Phys. 98 (7), 073710, 2005. 24. Chua, L. et al., General observation of n-type field effect behavior in organic semiconductors, Nature 434, 194–199, 2005. 25. Granstrom, J. and Katz, H.E., Pentacene transistors with polymer gate dielectrics on metallized optical fibers, J. Mater. Res. 19 (12), 3540–3546, 2004. 26. Cui, T.H. and Liang, G.R., Dual-gate pentacene organic field-effect transistors based on a nanoassembled SiO2 nanoparticle thin film as the gate dielectric layer, Appl. Phys. Lett. 86 (6), 064102, 2005. 27. Rutenberg, I. et al., Synthesis of polymer dielectric layers for organic thin film transistors via surface initiated ring opening metathesis polymerization, J. Am. Chem. Soc. 126, 4062–4063, 2004. 28. Chua, L. et al., High-stability ultrathin spin-on benzocyclobutene gate dielectric for polymer field effect transistor, Appl. Phys. Lett. 84, 3400–3402, 2004. 29. Kingon, A., Maria, J., and Streiffer, S., Alternative dielectrics to silicon dioxide for memory and logic devices, Nature 406, 1032–1038, 2000. 30. Plummer, J. and Griffin, P., Material and process limits in silicon VLSI technology, Proc. IEEE 89, 240–258, 2001. 31. Tate, J. et al., Anodization and microcontact printing on electroless silver: Solutionbased fabrication procedures for low voltage electronic systems with organic active components, Langmuir 16, 6054–6060, 2000.

8080_C003.fm Page 250 Friday, April 6, 2007 10:24 AM

250

Organic Field-Effect Transistors

32. Majewski, L. and Grell, M., Organic field effect transistor with ultrathin modified gate insulator, Synth. Met. 151, 175–179, 2005. 33. Braun, J.H., Titanium dioxide: A review, J. Coatings Technol. 69 (868), 59–72, 1997. 34. Campbell, S. et al., Titanium dioxide (TiO2)-based gate insulators, IBM J. Res. Develop. 43, 383–391, 1999. 35. Maliakal, A. et al., Inorganic oxide core, polymer shell nanocomposite as a high K gate dielectric for flexible electronics applications, J. Am. Chem. Soc. 127 (42), 14655–14662, 2005. 36. Majewski, L., Schroeder, R., and Grell, M., One-volt transistor, Adv. Mater. 17, 192–196, 2005. 37. Majewski, L. et al., A novel gate insulator for flexible electronics, Org. Elect. 4, 27–32, 2003. 38. Kelley, T. et al., High-performance OTFTs using surface-modified alumina dielectrics, J. Phys. Chem. B. 107, 5877–5881, 2003. 39. Dimitrakopoulos, C. et al., Field effect transistors comprising molecular beam deposited a,w-dihexylthienylene and polymeric insulator, Synth. Met. 92, 47–52, 1998. 40. Panzer, M., Newman, C., and Frisbie, C., Low-voltage operation of a pentacene field effect transistor with a polymer electrolyte gate dielectric, Appl. Phys. Lett. 86, 1035031–103503, 2005. 41. Panzer, M.J. and Frisbie, C.D., Polymer electrolyte gate dielectric reveals finite window of high conductivity in organic thin film transistors at high charge carrier densities, J. Am. Ceram. Soc. 127, 6960–6961, 2005. 42. Lee, J. et al., Flexible semitransparent pentacene thin-film transistors with polymer dielectric layers and NiOx electrodes, Appl. Phys. Lett. 87 (2), 023504, 2005. 43. Chua, L. et al., High-stability ultrathin spi-on benzocyclobutene gate dielectric for polymer field effect transistor, Appl. Phys. Lett. 84, 3400–3402, 2004. 44. Newman, C.R. et al., High-mobility top-gated pentacene thin-film transistors, J. Appl. Phys. 98 (8), 2005. 45. Becker, E. et al., All-organic thin-film transistors patterned by means of selective electropolymerization, Appl. Phys. Lett. 83, 4044–4046, 2003. 46. Angelis, F.D. et al., High field-effect mobility pentacene thin-film transistors with polymethylmethacrylate buffer layer, Appl. Phys. Lett. 86, 203505, 2005. 47. Garnier, F. et al., Molecular engineering of organic semiconductors: Design of selfassembly properties in conjugated thiophen oligomers, J. Amer. Chem. Soc. 115, 8716–8721, 1993. 48. Katz, H. et al., Organic field effect transistors with polarizable gate insulators, J. Appl. Phys. 91, 1572–1576, 2002. 49. Halik, M. et al., Fully patterned all-organic thin film transistors, Appl. Phys. Lett. 81, 289–291, 2002. 50. Halik, M., Klauk, H., and Zschieschang, U., Low-voltage organic transistors with an amorphous molecular gate dielectric, Nature 431, 963–966, 2004. 51. Parashkov, R. et al., All-organic thin-film transistors made of poly(3-butylthiophene) semiconducting and various polymeric insulating layers, J. Appl. Phys. 95, 1594–1596, 2004. 52. Parashkov, R. et al., Flexible all-organic field effect transistors fabricated by electrode peeling transfer, Jpn. J. Appl. Phys. 43, L130–132, 2004. 53. Jang, Y. et al., Influence of the dielectric constant of a polyvinyl phenol insulator on the field-effect mobility of a pentacene-based thin-film transistor, Appl. Phys. Lett. 87 (15), 2005.

8080_C003.fm Page 251 Friday, April 6, 2007 10:24 AM

Dielectric Materials: Selection and Design

251

54. Zielke, D. et al., Polymer-based organic field-effect transistor using offset printed source/drain structures, Appl. Phys. Lett. 87 (12), 123508, 2005. 55. Yasuda, T. et al., Organic field effect transistors with gate dielectric films of poly-pxylylene derivatives prepared by chemical vapor deposition, Jpn. J. Appl. Phys. 42, 6614–6618, 2003. 56. Podzorov, V., Pudalov, V., and Gershenson, M., Field effect transistors on rubrene single crystals with parylene gate insulator, Appl. Phys. Lett. 82, 1739–1741, 2003. 57. Zeis, R., Siegrist, T., and Kloc, C., Single-crystal field-effect transistors based on copper phthalocyanine, Appl. Phys. Lett. 86, 022103, 2005. 58. Gundlach, D.J. et al., Pentacene organic thin-film transistors: Molecular ordering and mobility, IEEE Electron Device Lett. 18 (3), 87–89, 1997. 59. Kim, J. et al., An organic thin-film transistor of high mobility by dielectric surface modification with organic molecule, Appl. Phys. Lett. 85, 6368–6370, 2004. 60. Kobayashi, S. et al., Control of carrier density by self-assembled monolayers in organic field effect transistors, Nature Mat. 3, 317–322, 2004. 61. de Oteyza, D.G. et al., Controlled enhancement of the electron field-effect mobility of F16CuPc thin-film transistors by use of functionalized SiO2 substrates, Appl. Phys. Lett. 87 (18), 183504, 2005. 62. Fritz, S., Kelley, T., and Frisbie, C., Effect of dielectric roughness on performance of pentacene TFTs and restoration of performance with a polymeric smoothing layer, J. Phys. Chem. B. 109, 10574–10577, 2005. 63. Khastgir, D., Maiti, H., and Bandyopadhay, P., Polystyrene–titania composite as a dielectric material, Mater. Sci. Eng. 100, 245–253, 1988. 64. Su, W. et al., Bismuth titanate nanoparticles dispersed polyacrylates, J. Mater. Res. 19, 2343–2348, 2004. 65. Chen, F.-C. et al., Organic thin-film transistors with nanocomposite dielectric gate insulator, Appl. Phys. Lett. 85 (15), 3295–3297, 2004. 66. Schroeder, R., Majewski, L., and Grell, M., High-performance organic transistors using solution processed nanoparticle-filled high-k polymer gate insulators, Adv. Mater. 17, 1535–1539, 2005.

8080_C003.fm Page 252 Friday, April 6, 2007 10:24 AM

8080_C004.fm Page 253 Tuesday, April 3, 2007 1:54 PM

4.1

Grazing Incidence X-Ray Diffraction (GIXD)

Tae Joo Shin and Hoichang Yang CONTENTS 4.1.1 Introduction................................................................................................253 4.1.1.1 Two Possible Geometrical Setups...............................................255 4.1.1.2 Bragg Peaks.................................................................................257 4.1.1.3 Bragg Rod Profile .......................................................................257 4.1.2 Interpretation of the Diffraction Data .......................................................258 4.1.2.1 Calculation of Structure Factor...................................................261 4.1.2.2 Calculation of Angle between a- and b-Axes.............................263 4.1.3 Examples....................................................................................................264 4.1.3.1 Poly(3-hexyl thiophene) (P3HT) ................................................264 4.1.3.2 Pentacene.....................................................................................269 4.1.3.3 Oligo Acene-Thiophene ..............................................................271 4.1.4 Concluding Remarks..................................................................................272 References..............................................................................................................273

4.1.1 INTRODUCTION X-rays, interacting weakly with matter, have long been used as an essential characterization tool to study the structure of bulk crystalline materials owing to their negligible multiple scattering and significant penetration depth. Recently, with the benefit of very intense x-ray sources such as synchrotron radiations, it has become possible to obtain surface and/or interface information selectively. One experimental technique is using the grazing incidence geometry (Figure 4.1.1). X-rays are known to be totally reflected on the surface of flat substances and strongly attenuated at small incident angles relative to the surface plane less than a critical angle [1]. Murra et al. used this x-ray total reflection phenomenon to develop a new technique for the structures of crystal surfaces and overplayed interfaces [2]. This method is referred to as grazing incidence x-ray diffraction (GIXD), which is also called grazing incidence x-ray scattering (GIXS), and has been commonly used to analyze “in-plane” crystal structures in a range from a few nanometers to several hundred nanometers beneath the surface on solids in air. Its applications have been reviewed in an article by Fuoss, Liang, and Eisenberger [3].

253

8080_C004.fm Page 254 Tuesday, April 3, 2007 1:54 PM

254

Organic Field-Effect Transistors

qz

GIXD

Z qtot

kin

θ

qxy

α 2θhor kout

θ

REFL

FIGURE 4.1.1 Scattering geometry for grazing incidence X-ray diffraction (GIXD) (and Xray reflectivity). The angle of incidence, θ, of the X-ray beam is less than the angle of total external reflection from the substrate. kin and kout are the wave vectors of the incident and reflected beams. The scattering vector, qxy ≈ 4πsinθhor/λ, is parallel to the substrate plane and qz = 2πsinαf /λ is perpendicular to it.

Theoretically, Vineyard described GIXD with a distorted-wave approximation in the kinematical theory of x-ray diffraction [4]. In terms of the ordinary dynamical theory of Ewald [5] and Laue [6], Afanas’ev and Melkonyan [7] worked out a formulation for the dynamical diffraction of x-rays under specular reflection conditions and Aleksandrov, Afanas’ev, and Stepanov [8] extended this formalism to include the diffraction geometry of thin surface layers. Subsequently, the properties of wave fields constructed during specularly diffracted reflections have been discussed in more detail by Cowan [9] and Sakata and Hashizume [10]. Meanwhile, a geometrical interpretation of GIXD based on a three-dimensional dispersion surface has been proposed by Höche, Brümmer, and Nieber [11]. Synchrotron radiation is very effective for GIXD measurements because of the parallel and intense beams and is often used for characterizing thin films on solids. Because of the limited availability of synchrotron radiation, conventional x-ray sources also have been adapted for the grazing incidence geometry [12]. Saito et al. have applied this method to the structural study of anodic oxide films and the thermally oxidized films on steels and obtained information on the thickness and density of the oxide films. The aim of this chapter is to systematically describe the steps involved in the elucidation of the in-plane order of thin films of organic semiconductors as a whole, not of the film surface alone. The principles of grazing incidence x-ray diffraction are described and a few examples of results obtained using GIXD are given. Organic thin films of submonolayer or monolayer are weak x-ray scatterers because they are composed of two-dimensional crystallites — namely, crystallites randomly oriented on the surface. Therefore, GIXD yields x-ray diffraction data that do not allow for a direct determination of the atomic positions as in three-dimensional crystals. Nevertheless, by making use of fixed atomic coordinate models of organic molecules, it is possible to extract valuable information on the molecular packing and orientation in the two-dimensional crystallites.

8080_C004.fm Page 255 Tuesday, April 3, 2007 1:54 PM

Grazing Incidence X-Ray Diffraction (GIXD)

255

The general approach for the analysis of GIXD experiments is based on the kinematic theory. This theoretical framework is based on two simplified assumptions: a spherically symmetrical scattering electron density of atoms and insignificant contribution from multiple scattering events. The latter assumption is justified since, for typical surface diffraction studies, the in-plane coherence of the diffracting objects is small enough to preclude multiple scattering. In addition, due to the low atomic number of the organic materials, the incident beam suffers a very small attenuation by absorption. In the general GIXD geometry, the angle of incidence of the x-ray beam is kept below the critical angle αc. This limits the penetration depth of the beam to that of the evanescent wave, and the scattering due to the subphase is efficiently eliminated. This permits a reliable measurement of the weak diffraction signal originating from the crystalline monolayer. When the substrate is disordered and thus scatters in all directions, the angle of incidence of the beam should be below the critical angle of the substrate in GIXD geometry. Conversely, if substrates are crystalline, as long as one stays away from the substrate Bragg peaks, no particular effort is needed to have the incident beam at less than the critical angle. Furthermore, this geometry increases the number of incident photons by ensuring that the whole thin film participates in the diffracted signal. From the GIXD profiles, the in-plane lattice parameters can be derived simply by using the Bragg’s law. Figure 4.1.1 shows schematic representation of grazing incidence diffraction from a horizontal surface. The diffraction condition for two-dimensional crystals lying in the xy plane is that the component of the scattering vector in the horizontal plane, labeled qxy (qxy = q|| ≈ (4π/λ)sinθhor (where 2θhor is the in-plane angle between the incident and the diffracted beam), must coincide with a reciprocal lattice vector qhk = 2π(ha* + kb*), where a* and b* are the reciprocal in-plane lattice vectors and h and k are the linear coefficients for the corresponding lattice point. There is no similar selection rule or restriction on the component of the scattering vector along the normal to the film defined as qz, whose magnitude qz = q⊥ = (2π/λ)sinα, where α is the angle between the diffracted beam and the substrate surface, shown in Figure 4.1.1. Therefore, the GIXD patterns from two-dimensional crystals (crystallites) arise from a two-dimensional array of rods, called Bragg rods (BRs), which extend parallel to qz.

4.1.1.1 TWO POSSIBLE GEOMETRICAL SETUPS As shown in Figure 4.1.2(a), the collection of the diffracted radiation by means of a one-dimensional position-sensitive detector (1-D PSD) is made by scanning the detector over a range along the horizontal scattering vector qxy (≈4πsin θhor/λ) and integrating over the whole qz window of the 1-D PSD to yield the Bragg peaks. Simultaneously, the scattered intensity recorded in channels along the 1-D PSD, but integrated over the scattering vector qxy in the horizontal plane across a Bragg peak, produces qz-resolved scans called Bragg rods. The GIXD method has been commonly used to analyze in-plane crystal structures of thin surface layers. However, measurements of “out-of-plane” structures under a grazing-incidence angle can also be performed. If two-dimensional (area) detectors

8080_C004.fm Page 256 Tuesday, April 3, 2007 1:54 PM

256

Organic Field-Effect Transistors

Side View

α f

1-D PSD

α f α

αi

f X-rays

Incident footprint

2θho r

Top View

Soller collimator

1-D PSD

(a)

Two-dimensional (area) Detector

qz α X-rays

α f

f

αi

r

2θho

qxy

(b)

FIGURE 4.1.2 (a) Top and side view of GIXD geometry. The 1-D PSD has its axis along the vertical. Only the cross-beam are contributes to the measured scattering. The Soller collimator defines the horizontal resolution of the detector. (b) 2-D GIXD geometry under which both in-plane (qxy) and out-of-plane (qz) reflections can be measured simultaneously. (From Kaganer, V.M. et al., Rev. Mod. Phys. 71, 779, 1999. With permission.)

like CCD camera and image plates are used, simultaneous measurements of both inplane and out-of-plane crystal structures become possible (Figure 4.1.2b). Minor and diffuse diffraction arcs can also be recognized more easily by using two-dimensional (area) detectors. However, the terminology “grazing-incidence x-ray diffraction technique” requires only that the incidence x-ray beam impinge the sample under a grazing angle; there is no restriction on the angle of the diffracted beam. Therefore, more information can be obtained if more generalized diffraction angles are measured. Such measurements can be easily realized with two-dimensional detectors such as image plates and a spherical-type goniometer [13] in combination with a grazing-incidence-angle synchrotron radiation beam, as shown sche-

8080_C004.fm Page 257 Tuesday, April 3, 2007 1:54 PM

Grazing Incidence X-Ray Diffraction (GIXD)

257

matically in Figure 4.1.1. In this configuration the Seemann–Bohlin geometry (out of plane: 2θout) or, similarly, grazing-incidence-angle asymmetric Bragg geometry [14] and GIXD geometry (in plane: 2θin) can be used simultaneously without scanning detectors. The average incident angles, θin, are around 0.3–0.5°, depending on the critical angles of the surface layers of the samples. Since the incident angles that can achieve a total reflection condition are usually very low, in most cases conventional divergent slits cannot effectively restrict the vertical width of the incident beam, and the beam unavoidably spills over the samples. Therefore, the sample width towards the beam direction must be carefully chosen as a compromise between spatial resolution and diffraction intensity, depending on the purpose of the analysis.

4.1.1.2 BRAGG PEAKS The reflections of Bragg peaks can be indexed by two Miller indices, hk. Their angular positions, 2θhor, yield the lattice plane spacing dhk = 2π/qhk for the twodimensional lattice structure. It is possible to have an estimate of the dimensions of the domains (DL) by analyzing the full width at half maximum (FWHM) of a GIXD peak (Δexp), corrected from the instrumental broadening with the Scherrer equation:

DL =

0.94λ 2 1/ 2 , where Δ = ( Δ 2exp − Δ slit ) Δ cos θ

and θ corresponds to the angular peak position [15,16]. The square of the molecular structure factor |Fhk|2, integrated along the Bragg rod over the window of qz seen by the detector, determines the integrated intensity in the peak. The structure factor Fhk(qz) is given by Fhk (qz ) =

∑fe j

iqhk ⋅rj iqz z j

e

(4.1.1)

j

where the sum is over atoms labeled j in the unit cell of dimensions a, b. The jth atom has a scattering factor fj and is located at the lateral vector position rj = xja + yjb and at the vertical position zj (in angstroms). Because the sample consists of a two-dimensional powder, the observed intensity Ihk for _a_ given qhk (or 2θhk) and qz position contains contributions from both the (h,k) and (h,k) reflections, with possibly several h, k integer values.

4.1.1.3 BRAGG ROD PROFILE The variation of the intensity Ihk(qz) along the Bragg rod as a function of qz is given by 2

−2 I hk (qz ) = KLPAABCD Acell V (qz ) Fhk (qz ) DWhk (qz )

(4.1.2)

8080_C004.fm Page 258 Tuesday, April 3, 2007 1:54 PM

258

Organic Field-Effect Transistors

The observed Bragg rod intensity Ihk(qz) is actually a sum over those (h,k) reflections whose Bragg rods coincide at a particular horizontal 2θhor angle or qxy position. In the upper equation, the most important variation is due to the molecular structure factor amplitude |Fhk|2. The Debye–Waller factor DWhk(qz) = exp[–(qhkUxy+ qz2Uz), where Uxy and Uz are, respectively, the mean square displacements in each horizontal direction x, y and along the vertical direction z. They account for the thermal motion of the atoms in the molecule, as well as for possible ripples on the substrate, which lead to roughness of the interface. The grazing geometry factor V(qz) describes the interference of rays diffracted upward with rays diffracted down and subsequently reflected back up by the interface. V(qz) differs from unity value only in the vicinity of q = qc/2, where it contributes a sharp peak. Corrections for the crossed-beam area (AABCD ∝ 1/sin2(2θhk)), Lorentz (L ∝ 1/sin2(2θhk)), and polarization (P = cos2(2θhk)) have been inserted; Acell is the area of the unit cell. The factor K scales the calculated to the observed intensities. For the simple linear surfactant molecules considered here, the square of the molecular structure factor |Fhk|2 is bell shaped as a function of qz and reaches its maximum when the scattering vector q = (qhk, qz) is orthogonal to the molecular axis. Thus, when the molecules are vertical or tilted in a plane perpendicular to qhk, the maximum intensity along the Bragg rod will occur at the horizon for qz = 0 Å–1. For molecules tilted otherwise, we expect the Bragg rod maximum at a finite qz, dependent upon both the magnitude and direction of the tilt relative to the in-plane scattering vector. The width of the bell-shaped Bragg rod profile is inversely proportional to the length of the molecule.

4.1.2 INTERPRETATION OF THE DIFFRACTION DATA In this section, we summarize the general approach to analyzing the GIXD data from monolayers, following Kaganer et al. [17]. In GIXD experiments, the scattered intensity is monitored as a function of two angles: the angle between incident and scattered beams in the water plane and the angle between scattered beam and the substrate. A periodicity of the molecular arrangement in the monolayer manifests itself in a peak in the distribution of the scattered intensity. There is as yet no way of controlling the mosaicity of Langmuir monolayers; in other words, the monolayers are powders within the plane. The diffraction pattern is always averaged over all domain orientations in the monolayer plane (“powder averaging”) in the area illuminated by the incident x-ray beam. As a result, of the three components qx, qy, and qz of the momentum transfer vector q, only the vertical component, qz, can be measured independently. It is not possible to determine the in-plane components qx and qy individually, but only the combination qxy qxy = (qx2 + qy2)1/2. Because lattice fluctuations at room temperature cause the peak intensities to decay rapidly with increasing momentum transfer, the first-order peaks, which correspond to the distances between neighboring molecules, are usually the most intense and frequently the only observed ones. Two first-order peaks sharing the same qxy are an indication of hexagonal packing, with equal distances between the molecules; three distinct values of qxy point to a rectangular unit cell and four peaks are due to

8080_C004.fm Page 259 Tuesday, April 3, 2007 1:54 PM

Grazing Incidence X-Ray Diffraction (GIXD)

259

an oblique unit cell. The available diffraction data are obviously not sufficient to perform structural analysis in a classical crystallographic sense. Additional knowledge of the packing of the organic semiconductor molecules in crystals may also be necessary because the packing sometimes leads to extinctions in the diffraction intensity and the aforementioned rule of thumb for first-order peaks would not apply (see discussion in Section 4.1.3). If the incident beam is coming in at less than the critical angle for total reflection, the specular beam does not carry useful information. However, if there is any lateral order at the surface (or in a monolayer at the surface) there will be diffraction peaks. The diffracted beam need not be observed only at grazing exit angles, but may be detected at any angle, resulting in a q that has both horizontal and vertical components. As shown in Figure 4.1.3b, the reciprocal lattice of a two-dimensional lattice, such as a thin film or a crystal surface, is a two-dimensional lattice of rods (since there is no periodicity in the z-direction, all qz are equivalent as far as the Bragg condition is concerned). The reciprocal lattice (and thus the real lattice) can be determined by analyzing the qx and qy spacings between these lattice rods. Notice that in the grazing incidence geometry it is never possible to go to qz = 0 Å–1 because this would require that the incident and diffracted photons make zero angles to the surface. However, a two-dimensional system is forgiving: the reciprocal lattice is in the form of rods, so the diffraction peaks can just as easily be located at slightly nonzero qz. (The scattering intensity will vary strongly along the rods if the lattice is not composed of points but of molecules, and this intensity variation contains information regarding the orientation of the molecules.) In the simplest model, a domain of the monolayer is treated as a two-dimensional crystal consisting of uniformly oriented rigid molecules. The scattering pattern in reciprocal space is then given by the product of two factors: the structure factor reflecting translational order of the molecular centers in the plane of the monolayer and the form factor of the individual molecule (see Figure 4.1.3b). The structure factor of a two-dimensional lattice consists of a set of delta-function discontinuities along lines (Bragg rods) normal to the monolayer plane. The form factor of a long rod-like molecule (such as alkane or acene) is large only on a plane normal to its long axis, which will be called the reciprocal disk of the molecule. The intersections of the first-order Bragg rods with the reciprocal disk give rise to six diffraction maxima (Figure 4.1.3c–f). If the molecules do not tilt, the reciprocal disk and hence all the peaks lie in the plane of the monolayer (Figure 4.1.3c). In a hexagonal phase, all six first-order wave vectors q have equal length and overlap completely in the powder pattern. Because of this degeneracy, the sixfold symmetry can only be concluded from the failure to see any other first-order peaks. The degeneracy is lifted in cases where the lattice is distorted from the exact symmetry hexagonal. For example, as a result of ordering of the backbone planes of the molecules, distinct peaks at different values of q are observed. If the unit cell is stretched or is compressed in the direction of the nearest-neighbor molecule, the unit cell becomes centered rectangularly. There are then two distinct first-order wave vectors on the powder averaging: one pair with ±qn and the other two with ±qd (the subscripts n and d denote nondegenerate and degenerate peaks).

8080_C004.fm Page 260 Tuesday, April 3, 2007 1:54 PM

260

Organic Field-Effect Transistors

Incident beam

Real space

Diﬀracted beam

Reciprocal space

2D lattice Rod-like molecule

Specular beam

Reciprocal disk

(a)

(b)

Untilted

NN tilt

NNN tilt

I

Kz

I

Kxy

Kxy Kz

Kz

Kxy (c)

θ

I

Kxy

Kxy Kz

Kxy (d)

Intermediate tilt

θ

θ

I

Bragg rods

Kxy (e)

Kxy (f )

FIGURE 4.1.3 (a) Schematic diagram of a GIXD experiment; (b) ‘Rods’ formed in reciprocal space by a 2-D lattice of points, and the ‘reciprocal disk’ of an extended molecule, which combines to give the diffraction pattern from a 2-D array of extended molecules; (c) realspace and reciprocal-space views and characteristic diffraction patterns of the monolayer in untilted phase; (d) NN-titled phase; (e) NNN-titled phase; (f) intermediate titled phase. (From Kaganer, V.M. et al., Rev. Mod. Phys. 71, 779, 1999. With permission.)

If the unit cell stretches in the direction of the nearest-neighbor (NN) molecule, then |qn| > |qd|; the opposite inequality indicates that the unit cell shrinks in that direction. The degeneracy may also be lifted by molecular tilt. In this case, the peaks move out of the monolayer plane by a distance qz, which depends on both the tilt magnitude and its azimuth, or direction. Since the only points of the reciprocal disk to remain in the monolayer plane are those on the line normal to the tilt direction, the only diffraction peaks that originate from Bragg rods on that line remain in the monolayer plane (see Figure 4.1.3d). This occurs for one pair of peaks when the molecules tilt towards the NN. The other four peaks move out of the plane: two upwards and two downwards (naturally, diffraction beams going into the substrate cannot be observed); see Figure 4.1.3(d). The wave vectors of the two visible out-

8080_C004.fm Page 261 Tuesday, April 3, 2007 1:54 PM

Grazing Incidence X-Ray Diffraction (GIXD)

261

of-plane peaks have equal qz components and are thus degenerate in the powder pattern. The tilt angle θ is given by tan θ = qdz/[qdxy2–(qnxy/2)2]1/2. When the molecules tilt towards a next-nearest neighbor (NNN) molecule, all the wave vectors move out of the plane. The two distinct values of qz are in the ratio qnz:qdz = 2:1, and the tilt angle is given by tan θ = qnz/qnxy (Figure 4.1.3e). In these symmetrically tilted phases, the distinction between degenerate and nondegenerate peaks is unequivocal. The ratio qnz:qdz can only be 0:1 or 2:1. In an untitled phase (Figure 4.1.3c), the distinction is not as easy. In the idealized model, where rod-like molecules are represented as cylinders of uniform electron density, the integrated intensity of the degenerate peak should be twice as large as that of the nondegenerate one, but experimentally there are often significant departures from the “ideal” 2:1 intensity ratio. Leveiller et al. calculated the molecular structure factors and intensity ratios from atomic scattering factors, assuming an all transconformation of the molecules and ideal (zero-temperature) packing; they found that the intensity ratio depends on the orientations of the backbone planes of the molecules and symmetry constraints [18]. If the tilt azimuth is intermediate between NN and NNN (Figure 4.1.3f) or if the distortion of the unit cell is asymmetrical, there are three distinct first-order peaks and no indexing problem. Each peak is described by two components of the momentum transfer, giving six measured values in all. Since the monolayer model is completely described by five parameters — three for the in-plane lattice and two for the tilt (magnitude and direction) — the measured values cannot be completely independent. The relationship between them is readily shown to be q1z + q2z = q3z, where peak 3 is the one with largest qz. The two qz ratios 0:1 and 2:1 for the symmetric tilts follow from this more general relationship as particular cases in which one of the qz values is repeated and correspond to 0 + 1 = 1 and 1 + 1 = 2, respectively. Once the three first-order peaks have been assigned, the geometry of the unit cell in reciprocal space is completely determined, and the corresponding real-space lattice is easily determined. In the literature, the peaks are sometimes labeled in terms of either a hexagonal or a centered rectangular unit cell; the latter notation is more common [19]. Denoting the basic translations of the centered rectangular unit cell containing two molecules by [10] and [01] (when the lattice is hexagonal, the length of the vector [01] is √3 times larger than [10]), one finds that, if the two molecules in the rectangular unit cell are actually equivalent (i.e., the rectangular unit cell is not the smallest unit cell describing the lattice), the reflections (01) and (10) are forbidden, as are all (hk) reflections where h + k is an odd number. The lowest order reflections are (02), which is nondegenerate, and two reflections (11) _ + (11), which have equal length and so degenerate in the powder average. In case of the hexagonal unit cell, all three reflections possess equal wave-vector magnitudes.

4.1.2.1 CALCULATION

OF

STRUCTURE FACTOR

Generally, the intensity at a particular value of qz in a Bragg rod is proportional to the square of the molecular structure factor, |Fhk(qz)|2. Thus, Bragg rod intensities can be calculated using the atomic coordinates in the unit cell. The form factor is given by

8080_C004.fm Page 262 Tuesday, April 3, 2007 1:54 PM

262

Organic Field-Effect Transistors

F (q) =

∑f e j

− iqr j

where fj is the scattering factor (or form factor) of the jth molecule in the unit cell at the position rj. If two molecules per a centered rectangular unit cell were related by translation symmetry (f1 = f2 = f) — that is, the same molecules are located at (0,0) and (a/2,b/2), then F(qx, qy) is given by F (q x , qy ) = f + f ⋅ e

i / 2 ( q x a + qy b )

At the peak positions, qx =

2π h a

qy =

2π k b

and

F(q) = F(h, k) = f + f ⋅ eiπ(h+k) vanishes whenever h + k is odd. When two molecules in the unit cell are not translationally equivalent, F(h,k) is given by F(h,k) = f1 + f2 ⋅ eiπ(h+k), where it can be observed, depending on the f1 and f2, even if h + k is odd. As discussed previously, Durbin et al. observed the (21) reflection in herringbone packing due to f1 ≠ f2 in the (21) reflection, but (10) or (01) reflections are not observed because the two form factors are identical [10]. For a unit cell with different dimension along the a, b, and c-axes, when a molecule stands with height (d) along the c-axis and width (w) along the a-axis, the scattering factor is given by

f (q x' , qz' ) =

sin(q x' w / 2 ) sin(qz' d / 2 ) − iqz' d / 2 ⋅e q x' w / 2 qz' d / 2

(4.1.3)

If the molecule is tilted, the sample coordinate and laboratory coordinate are related by q ' = R -1 ⋅ q

(4.1.4)

where R is Euler rotation matrix expressed by the tilt angle (θ), tilt azimuth (ϕ), and the rotation angle to the molecular axis (χ). q′ and q are scattering vectors in the sample coordinate and laboratory coordinate, respectively. The molecule is

8080_C004.fm Page 263 Tuesday, April 3, 2007 1:54 PM

Grazing Incidence X-Ray Diffraction (GIXD)

263

located at the sample coordinate and the unit cell with the molecules is located at the laboratory coordinate. Finally, from the equations 4.1.3 and 4.1.4, a form factor can be calculated. A structure factor of the herringbone motif can be calculated by applying –χ and +χ to each molecule.

4.1.2.2 CALCULATION

OF

ANGLE

BETWEEN A- AND B-AXES

If one assumes the real space lattice vector a = (ax, 0) and b = (bx, by), then the angle γ between the two vectors is defined as the angle from a to b and is related as follows: γ = tan −1 (by / bx )

(4.1.5)

A two-dimensional lattice can be thought of as the x–y plane of a monoclinic lattice, a* = 1 / (| a | sin( γ )) ⋅ b ⊥ / | b ⊥ |

(4.1.6)

b* = 1 / (| b | sin( γ )) ⋅ a ⊥ / | a ⊥ |

(4.1.7)

Since |a| = ax > 0, sin(γ) = by/|b|, and by ≥ 0, then a* = 1 / ( a x by ) ⋅ (by , −bx )

(4.1.8)

b* = 1 / ( a x by ) ⋅ (0, a x )

(4.1.9)

since the peak position of the (h,k) plane is defined as following q hk = 2π( h ⋅ a* + k ⋅ b* )

(4.1.10)

When the molecules in the lattice are tilted, a* and b* will have z component and qhk also should have qz component. However, the in-plane components of qhk are not dependent on the molecular orientation in the lattice at all. Thus, we can write | q hk |xy ≡ q xyhk =

2π a x by

h 2 (bx2 + by2 ) + k 2 a x2 − 2 hka x bx

(4.1.11)

Thus, if three peaks are resolved, the lattice vector could be calculated by the 11 12 linear algebra. For example, when q11 xy , q xy and q xy are resolved, they could be calculated as follows:

8080_C004.fm Page 264 Tuesday, April 3, 2007 1:54 PM

264

Organic Field-Effect Transistors

2 by = 2 π 3 / (q12 xy ) +

1 11 2 3 11 2 (q xy ) − (q xy ) 2 2

(

11 2 2 (q11 xy ) − (q xy ) 2 (2 π)2 11 2 11 2 a x = 2 π 2 / (q xy ) + (q xy ) − − 8(2π)2 by2

bx =

2 11 2 (q11 xy ) − (q xy )

4(2π)2

a x by2

(4.1.12)

)b 2

2 y

(4.1.13)

(4.1.14)

4.1.3 EXAMPLES 4.1.3.1 POLY(3-HEXYL

THIOPHENE)

(P3HT)

Self-organization in many solution-processed, semiconducting conjugated polymers results in complex microstructures in which ordered microcrystalline domains are embedded in an amorphous matrix [20,21]. This has important consequences for electrical properties of these materials: Charge transport is usually limited by the most difficult hopping processes and is therefore dominated by the disordered matrix, resulting in low charge-carrier mobilities (≤10–5cm2V–1s–1) [22]. Sirringhaus et al. have studied the microstructure of 70- to 100-nm, spin-coated, regioregular poly(3-hexylthiophene) (P3HT) films by GIXD on parts of the SiO2/Si substrates on which field effect transistor (FET) devices were fabricated [20]. Grazing-incidence x-ray diffraction measurements were performed under inert He atmosphere to minimize air scattering and beam damage. A grazing incidence angle below the critical angle of total reflection from the substrate, but above the critical angle for the film, was chosen to enhance the sensitivity to the thin polymer film. The intensities plotted versus the total scattering vector are corrected for polarization and geometric factors [23]. Depending on processing conditions, the π-conjugated crystalline planes can adopt two different orientations — parallel and perpendicular to the substrate (the mobilities of which differ by more than a factor of 100) — and can reach values as high as 0.1 cm2V–1s–1 [24]. Two different orientations of the microcrystalline P3HT domains with respect to the FET substrate are shown in Figure 4.1.4. The coexistence of two different phases is evident from the different intensity distributions of the (100) reflections due to interdigitation of hexyl side chains and the (010) reflections due to π–π-interchain stacking [25]. In samples with high regioregularity (>91%) and low molecular weight, the preferential orientation of ordered domains is with the <100>-axis normal to the film and the <010>-axis in the plane of the film (Figure 4.1.4a). In contrast, in samples with low regioregularity (81%) and high molecular weight, the crystallites are preferentially oriented with the <100>-axis in the plane and the <010>-axis normal to the film (Figure 4.1.4b) [26]. The cause of this surprising orientational

8080_C004.fm Page 265 Tuesday, April 3, 2007 1:54 PM

Grazing Incidence X-Ray Diffraction (GIXD)

265

change could be understood as a dynamic phenomenon occurring during the rapid growth of spin-coated films, affected by regioregularity or molecular weight. In films prepared by drop casting from a dilute solution, the <100>-axis is normal to the film for all polymers (gray trace in Figure 4.1.4c and d). The ability to induce different orientations allows one to establish a direct correlation between the direction of π–π-stacking and the in-plane FET mobility. At room temperature, the highest mobilities of 0.05–0.1 cm2V–1s–1 are observed for the sample with the highest regioregularity (96%) and the largest size of crystallites with in-plane orientation of the <010>-axis (~95Å, as calculated from GIXD analysis for (010) peak shape) [23]. For spin-coated samples with 81% regioregularity (<010>-axis normal to the film), the mobility is only 2 × 10–4 cm2V–1s–1 in spite of pronounced in-plane crystallinity along the <100>-axis with a grain size of 130 Å. This is consistent with the FET mobility being limited by π–π-interchain, rather than intrachain, transport. In the case of the low-regioregularity polymers, it is possible to compare mobilities directly for the in-plane (cast films) and out-of-plane orientation (spin-coated films) of the π–π-stacking direction (Figure 4.1.4d). In cast films of the 81% regioregular polymer, the mobility is higher by more than an order of magnitude than in spin-coated samples and only slightly lower than that of the highly regioregular polymers (Figure 4.1.4e). The large mobility anisotropy caused by different preferential orientations of the ordered, microcrystalline domains is clear evidence that the transport is no longer dominated by the remaining amorphous regions of the polymer film. Rather, it is starting to reflect the transport properties of charge carriers in ordered polymer domains. McGehee and coworkers observed that the charge mobility values of spin-cast RR P3HT films changed up to four orders of magnitude depending on the MW [27,28]. They observed that the crystalline morphology of low MW films (Mn < 4 kDa) showing low mobility had highly ordered short nanoribbons, while highmobility, high MW films (Mn > 30 kDa) had featureless crystalline structure. Because the charge transport in organic thin film transistors (OTFTs) occurs in the plane of the substrate, GIXD was used to characterize the in-plane π-stacking. However, because in-plane GIXD measures crystals with lattice planes perpendicular to the substrate (within about 0.7°), only a small portion of reciprocal spaces samples and it is likely that crystals not measured by in-plane GIXD can be important for charge transport. In contrast, rocking-curve x-ray experiments, where the sample is rotated in the x-ray beam while keeping the scattering angle (2θ) constant, provide a means for measuring the distribution of crystal orientations in the films throughout reciprocal space. The rocking curve x-ray experiments for different MW P3HT thin films was used to explain change in charge mobilities through dielectric surface treatment with hexamethyldisilizane (HMDS) and octadecyltrichlorosilane (OTS) and postannealing [29]. Figure 4.1.5 shows the specular diffraction obtained for the (100) peak of spin-cast films with various MWs on OTS-treated substrates. The peak intensity, width, and position vary with MW. The trend of reduced d(100) spacing in the films with decreasing MW is related to either a change in molecular tilt angle or the amount of side-chain interdigitation [28,30]. Also, the reduced peak width indicates

8080_C004.fm Page 266 Tuesday, April 3, 2007 1:54 PM

266

Organic Field-Effect Transistors

s

a

s s

s s

s

a

s s

b

Intensity (a.u.)

5 b

4.5 4 3.5 3 2.5

(010) (300)

1.5

(200)

1

(100) (010)

(100) (b)

γ

(300)

Side view (200)

0.4 0.2 0 0.0

(010)

Detector

α (100)

Intensity (a.u.)

(a)

0.6

0.5

1.5

1.0

2.0

q (Å−1)

(010)

(c)

2θ

81% cast

Top view (200)

(100)

Intensity (a.u.)

Detector 0.4

0.2

81%

96%

0 0.0

0.5

1.0

1.5

2.0

q (Å−1) (d)

μsat (cm2 V−1 s−1)

10−1 10−2 10−3 10−4 70

80 90 % head-to-tail (e)

FIGURE 4.1.4

2

2.5

0.5

8080_C004.fm Page 267 Tuesday, April 3, 2007 1:54 PM

Grazing Incidence X-Ray Diffraction (GIXD)

267

0.06

Counts (arb. units)

2θ Low-MW OTS 0.04

Medium-MW OTS 0.02

High-MW OTS

0.00 0.25

0.30

0.35

0.40

0.45

q (Å−1)

FIGURE 4.1.5 Specular diffraction of the <100> peak. Out-of-plane diffraction peaks analyzed with rocking curves shown in Figure 4.1.6 for the P3HT films with low, medium, and high molecular weights on OTS-treated substrates. The inset shows the specular diffraction geometry. (From Kline, R.J. et al., Nat. Mater. 5, 222, 2006. With permission.)

that the domain size perpendicular to the films is larger for low MW films than it is for medium and high MW films. In addition, Figure 4.1.6 shows rocking curves for the orientation of the (100) crystalline planes relative to the substrate normal. It was suggested that the sharp peaks in the rocking curves, Δω ~ 0.03°, corresponding to the instrumental resolution are related to P3HT crystals located near the substrate (with r.m.s. roughness of about 1 Å) rather than the film surface (with r.m.s roughness of about 6 Å). The contribution of the crystals at the air–film interface (with high surface roughness) should have increased at least an order of magnitude greater than the width of the measured rocking-curves peak. Specifically, the different crystalline orientation in the films cast on OTS- and HMDS-treated substrates strongly supports their interpretation for the location of the highly oriented crystals. As a result, the fact that the low MW film cast on HMDS has few highly oriented crystals probably explains why the mobility (~10–6 cm2V–1s–1) is so much lower than for the other films. The anisotropy of PHT crystals limits charge transport to two crystal directions (<010>

FIGURE 4.1.4 (See figure, facing page.) Two different orientations of ordered P3HT domains with respect to the FET substrate: (a), (b), 2-D GIXD distribution from spin-coated, 70-100 nm thick P3HT films with regioregularity of 96% (a) and 81% (b) on SiO2/Si substrates. The vertical (horizontal) axes correspond to scattering normal (parallel) to the plane of the film. The insets show schematic orientations of the microcrystalline grains with respect to the substrate. (c), (d), The change of orientation is confirmed by high-resolution synchrotron X-ray diffraction measurements for constant, grazing-incidence angle with outof-plane (c) and in-plane (d) scattering geometry. (e), Charge carrier mobility of P3HT-based field-effect transistors with different microstructures. (From Sirringhaus, H. et al., Nature 401, 685, 1999. With permission.)

8080_C004.fm Page 268 Tuesday, April 3, 2007 1:54 PM

268

Organic Field-Effect Transistors

10−4

0.0002

2θ

Counts (arb)

Counts (arb)

0.0003

10

LowMW OTS

−5

MediumMW OTS

ω

0.0001

10

−6

HighMW OTS −0.04 −0.02 0.00 0.02 0.04 ω (°)

0.0000 −1.5 −1.0 −0.5

s s s s s s s s s s s s s s s s

0.0

0.5 1.0 ω (°)

1.5 (d)

(a)

Counts (arb)

0.1

0.01

MediumMW OTS

0.001 Medium-MW HMDS

−0.2

−0.1

(e)

0.0 ω (°)

0.1

0.2

(b)

Counts (arb)

1 Annealed lowMW OTS

(f )

0.1

0.01

Low-MW OTS

Annealed lowMW HMDS

Low-MW HMDS

−0.2

−0.1

0.0 ω (°)

0.1

0.2 (g)

(c)

FIGURE 4.1.6 Rocking-curve measurements on the (100) peaks and resulting crystal orientations: (a) Rocking curves for the films (with low, medium, and high molecular weights) on OTS-treated substrates. (Left inset: the rocking curve geometry showing the angle relative to the sample normal ω. Right inset: a magnification of the ω = 0 region plotted on a logarithmic scale.) Logarithmic–scale rocking curves on films with (b) medium molecular weight on both HDMS and OTS, and (c) low molecular weight before and after annealing for both HMDS and OTS. Schematics showing the highly oriented and misoriented crystals in films with (d) low molecular weight on OTS, (e) medium and high molecular weights on OTS, (g) low molecular weight on HMDS before annealing, and (f) after annealing (drawn on a smaller scale to accommodate the larger crystals). Lines correspond to the (100) plane. The bottom circles in (e) and the first and third circles in (f) denote crystals measured by specular diffraction and the other circles denote those that are not. (From Kline, R. J. et al., Nat. Mater. 5, 222, 2006. With permission.)

8080_C004.fm Page 269 Tuesday, April 3, 2007 1:54 PM

Grazing Incidence X-Ray Diffraction (GIXD)

269

and <001>), with the insulating hexyl chains preventing charge transport in the <100> direction. A large distribution of crystal orientations increases the likelihood of having poor electronic overlap along these high-mobility directions between neighboring grains.

4.1.3.2 PENTACENE Pentacene, consisting of five fused benzene rings, has emerged as a viable candidate for use in OTFTs because of its relatively high hole mobility and high on/off ratio [31,32]. However, little is known about the crystal structure in the first few layers of a vacuum sublimed pentacene film, the most important for charge transport. It is also widely recognized that the transport properties of crystalline organic films depend strongly on the intermolecular overlap of electronic wave functions within the semiconductor layer, which is very sensitive to the molecular packing in the crystal [33]. Fritz et al. reported preliminary GIXD data for a monolayer-thick pentacene film grown on amorphous silicon dioxide (a-SiO2) [34]. The data confirmed that the monolayer is crystalline and has a structure that differs from that of bulk pentacene, which has important implications for carrier transport in pentacene-based OTFTs. The crystal structure of bulk pentacene consists of layers of pentacene molecules arranged in a herringbone packing motif with an interlayer spacing of d001 = 14.1 Å [35]. Three thin film multilayer phases with different d001 values of 14.4, 15.0, and 15.4 Å have been identified by wide-angle x-ray diffraction [36,37]. The selectivity toward these phases appears to be governed by a variety of factors, including substrate material, substrate temperature during deposition, rate of deposition, and film thickness [36,38]. The different d001 values imply dissimilar packing of the pentacene molecules in the ab plane, which is the basal plane in crystalline vacuum deposited pentacene films on inert substrates and is therefore the crystal plane crucial for charge transport. Characterization of this pentacene monolayer by GIXD at room temperature yielded a diffraction pattern (Figure 4.1.7) that could be indexed as a near rectangular in-plane unit cell with dimensions a = 5.92 Å, b = 7.59 Å, and γ = 89.95°. These values differ from the corresponding lattice parameters reported for bulk pentacene (a = 6.27 Å, b = 7.78 Å, and γ = 84.68˚), but are consistent with a packing motif resembling the (001) layers in the bulk (i.e., the pentacene molecules adopt a nearvertical orientation in the monolayer on the a-SiO2 substrate). A model of the monolayer structure was constructed using the room-temperature single-crystal structure of bulk pentacene as a starting point. The a and b lattice parameters of the bulk structure were adjusted to the monolayer values, and the interlayer d001 spacing was expanded to an arbitrarily large distance to mimic an isolated monolayer. The simulated diffraction pattern produced by this energy-minimized monolayer structure is in reasonable agreement with the GIXD data. Some of the discrepancy can be attributed to contributions from the a-SiO2 background, particularly in the lower qxy region. We anticipate that a more complete data set, particularly data collected at qz > 0, will permit precise determination of the molecular tilt and faster refinement of the thin-film structure [39]. Nevertheless, the GIXD data demonstrate unequivocally that the pentacene monolayer on the a-SiO2 dielectric layer is highly crystalline and exhibits a structure distinguishable from the bulk.

8080_C004.fm Page 270 Tuesday, April 3, 2007 1:54 PM

270

Organic Field-Effect Transistors

Bulk pentacene a b

a

Intensity (a.u.)

Expanded 3x

(1 1)

1

2

(2 3), (3 1) (1 4)

(1 3), (2 2)

(2 1)

(1 2)

(2 0)

Expanded 3x (0 2)

0

3

z b Pentacene monolayer d e

Calculated

Experimental

c

a

4

z

f

z b

a

b

z b

a

qxy (Å−1) (a)

(b)

FIGURE 4.1.7 (a) GIXD pattern (bottom of (a)) for a pentacene monolayer and a diffraction pattern (top of (a)) calculated for an energy-minimized crystal structure model based on the GIXD lattice parameters and the (001) layer motif of bulk pentacene as the starting point; (b) Normal views of the ab planes of bulk pentacene and the model monolayer structures (a and d) and the respective side views (b and e, e and f). The z-axis is the normal to the ab plane. (From Fritz, S.E. et al., J. Am. Chem. Soc. 126, 4084, 2004. With permission.)

The charge carrier mobility of organic semiconductors is sensitive to the dielectric layer surface on which it is deposited. Kelley et al. obtained very high field effect mobility (~3 cm2V–1s–1) with pentacene multilayers consisting of large twodimensional grains on a 1-phosphoneo-hexadecane-treated alumina dielectric layer [40]. The molecular orientation and grain morphology of the first layer of pentacene molecules depend on balance between pentacene–substrate interactions, which can be shifted by modification of the dielectric surface in an OTFT with a self-assembled monolayer (SAM) [41]. They have studied the correlation between pentacene ultrathin film morphology and the overall OTFT device performance and observed a direct correlation between the crystalline structure of the initial submonolayer of a pentacene film and the mobility of the corresponding thick film. The terrace-like multilayered pentacene films, grown on single crystal-like faceted islands in the first layer, have shown much higher field-effect mobility than those grown on polycrystalline dendritic islands. Figure 4.1.8 shows the two-dimensional GIXD patterns and tapping mode atomic force microscopy (TM-AFM) topographies for 60-nm thick pentacene films deposited on HMDS- and OTS-treated surfaces [42]. The occurrence of many reflection spots in the direction of qz (out of plane) at a given qxy (in plane) strongly suggests growth of three-dimensional islands. Specifically, diffraction patterns of 2-ML pentacene films yielded a pseudocentered rectangular unit cell (a = 5.90 ± 0.01 Å, b = 7.51 ± 0.01 Å, and γ = 89.92 ± 0.01°) with a herringbone packing and molecules tilted along the b-axis by 4° with respect to the surface normal.

8080_C004.fm Page 271 Tuesday, April 3, 2007 1:54 PM

Grazing Incidence X-Ray Diffraction (GIXD)

(a)

μ = 3.4 ± 0.5 cm2/Vs

271

(b)

μ = 0.5 ± 0.15 cm2/Vs

1 μm

1 μm

(0, 2) (1, 2) (2, 1)

qz

qxy (Å−1)

(1, ±1)

(2, 0)

qz

qxy (Å−1)

FIGURE 4.1.8 Tapping mode AFM topographs and 2-D GIXD patterns of 60-nm-thick pentacene films on (a) HMDS- and (b) OTS-treated SiO2/Si substrates. (From Yang, H. et al., J. Am. Chem. Soc. 127, 11542, 2005. With permission.)

Despite minute differences such as a small portion of differently oriented crystals (indicated by black arrows in Figure 4.1.8b) and smaller grain size in the OTS sample, two-dimensional GIXD supports the fact that the HMDS and OTS samples have a similar vertical conducting path for top contact devices. Mobility measurements, however, show drastically different mobilities: μ = 3.4 ± 0.5 cm2V–1s–1 on HMDS-treated surface and 0.5 ± 0.15 cm2V–1s–1 on OTS-treated surface, using 60nm thick pentacene films in the top contact OTFT. This fact suggests that the minute differences in grain sizes may be important.

4.1.3.3 OLIGO ACENE-THIOPHENE Frisbie et al. have reported the structural and electrical characterization of two new p-channel organic semiconductors: 5,5′-bis(2-tetracenyl)-2,2′-bithiophene (1) and 5,5′-bis(2-anthracenyl)-2,2′-bithiophene (2) [43]. GIXD was performed to determine the thin-film unit cell parameters of 1 and 2. Figure 4.1.9 shows the GIXD patterns (intensity vs. qxy) for 1 and 2 from which the in-plane lattice parameters (a,b) and the angle between them (γ) were determined. Films of 1 exhibited two slightly different phases as evidenced by the occurrence of several peaks in the diffraction patterns. Using the first peak of each pair of doublets, the peaks are indexed to a rectangular in-plane unit cell with dimensions a = 5.99 Å, b = 7.66 Å, and γ = 90.0°. For 2, a similar in-plane unit cell was found with a = 5.98 Å, b = 7.64 Å, and γ = 90.0°. There are two molecules per unit cell similar to the pentacene bulk crystal structure [35]. The calculated densities of 1 and 2 are 1.41 g/cm3 and 1.39 g/cm3, respectively. These density values are comparable to other common organic semiconductors such as pentacene (1.36 g/cm3) and sexithiophene (1.55 g/cm3) [35,44]. Several of the peaks seen in the spectrum of 2 (denoted by arrows in Figure 4.1.9b) are not compatible with this lattice parameter assignment and appear to be a second phase. The peaks at qxy = 0.698, 0.932, 1.396, and 1.864 Å–1 (denoted by arrows) are weaker

8080_C004.fm Page 272 Tuesday, April 3, 2007 1:54 PM

272

Organic Field-Effect Transistors

Log intensity (arb. units)

(11) (02) (12) (20)

(23)

(21)

0.5

1.0

1.5

2.0

(14)

(22)

2.5

3.0

3.5

qxy (Å−1) (a)

Log intensity (arb. units)

(11)

(02) (12) (20) (22)

(23) (14)

(21)

0.5

1.0

1.5

2.0

2.5

3.0

3.5

qxy (Å−1) (b)

FIGURE 4.1.9 GIXD patterns for (a) a 350 Å film of 5,5′-bis(2-tetracenyl)-2,2′-bithiphene on PS-SiO2 and (b) a 350 Å film of 5,5′-bis(2-anthracenyl)-2,2′-bithiphene on bare SiO2. (From Merlo, J.A. et al., J. Am. Chem. Soc. 127, 3997, 2005. With permission.)

in intensity and correspond to d-spacings of 9.00, 6.74, 4.50, and 3.37 Å, which correspond to the values of (003), (004), (006), and (008) reflections as obtained from thin-film XRD. Therefore, these peaks suggest that some fraction of the molecules is lying down (i.e., long axis of the molecule parallel to the surface). For thicker films (~750 Å) of 2, the (110) peak was observed in thin-film XRD spectra, which supports this hypothesis. The fact that a fraction of the molecules are lying down is likely to have a detrimental effect on the electronic transport properties of films of 2, since the preferred transport direction occurs within the ab plane. With a full set of qxy and qz data, the complete thin-film unit cell can be determined. Both films have triclinic cells. The d001-spacing calculated from these unit cells was 31.83 Å for 1 and 27.03 Å for 2, which is in good agreement with the values from wideangle XRD (31.83 and 26.97 Å, respectively).

4.1.4 CONCLUDING REMARKS Over the past decade, various aspects of organic semiconductor thin films have been investigated, such as film processing parameters and dielectric surface chemistry.

8080_C004.fm Page 273 Tuesday, April 3, 2007 1:54 PM

Grazing Incidence X-Ray Diffraction (GIXD)

273

Recent work has been focused on understanding the assembly of semiconducting materials in thin films. Understanding the bulk crystalline structure of these materials has gradually expanded toward ultrathin film architectures — the so-called thin film phase — near the dielectric. For these semiconducting thin films, GIXD profiles displayed by both the horizontal and vertical scattering summation of the crystals distributed in the films can provide a deep insight into the state of the crystal packing and orientation in the films with the thickness ranging from the monolayer in direct contact with the substrate through the subsequent multilayer. Monitoring the in situ crystallization of organic semiconductors through combining vacuum evaporator with GIXD setup can provide more detailed information for the crystal nucleation and vertical growth mechanism of the molecules on the substrates under various processing parameters. In semiconducting thin films, determination of crystalline structures with nearly atomic resolution can be achieved in favorable cases by a combination of GIXD experimental and computational techniques. While the ultimate analysis will usually involve the more complicated methods, the simpler considerations described here are useful in offering a general understanding and in providing suitable initial conditions for least-squares fittings and other parameters.

REFERENCES 1. James, R.W. The optical principles of the diffraction of x-rays, Ox Bow Press, Woodbrige, CT, chap. 4, 1982. 2. Marra, W.C., Eisenberger, P., and Cho, A.Y., X-ray total-external-reflection–Bragg diffraction: A structural study of the GaAs-Al interface, J. Appl. Phys. 50, 6927, 1979. 3. Fuoss, P.H., Liang, K.S., and Eisenberger, P., in Synchrotron radiation research: Advances in surface science, ed. R. Z. Bachrach, Plenum, New York, 1989. 4. Vineyard, G.H., Grazing-incidence diffraction and the distorted-wave approximation for the study of surfaces, Phys. Rev. B 26, 4146, 1982 5. Ewald, P., Zur Begründung der Kristalloptik, Ann. Phys. (Leipzig) 54, 519, 1917. 6. von Laue, M., Materiewellen und ihre Interferenzen, Akademische Verlagsgesellschaft, Leipzig, 1948. 7. Afanas’ev, A.M. and Melkonyan, M.K., X-ray diffraction under specular reflection conditions. Ideal crystals, Acta Cryst. A39, 207, 1983. 8. Aleksandrov, P.A. Afanas’ev, A.M., and Stepanov, S.A., Bragg–Laue diffraction in inclined geometry, Phys. Status Solidi A 86, 143, 1984. 9. Cowan, P.L., Diffraction of evanescent x-rays: Results from a dynamical theory, Phys. Rev. B 32, 5437, 1985. 10. Sakata, O. and Hashizume, H., Report RLEMTIT, 12, 45–57, 1987. 11. Höche, H.R., Brümmer, O., and Nieber, J., Extremely skew X-ray diffraction, Acta Cryst. A42, 585, 1986. 12. Croce, P. and Nevot, L., Etude de couches minces et des surfaces par réflexion rasante spéculaire ou diffuse de rayons X, Rev. Phys. Appl. 11, 113, 1976; Cowley, R.A. and Ryan, T.W., X-ray scattering studies of thin films and surfaces: Thermal oxides on silicon, J. Phys. D: Appl. Phys. 20, 61, 1987; Gay, J.M., Stocker, P., and Rhemore, F., X-ray scattering studies of FeSi2 films epitaxially grown on Si(111), J. Appl. Phys. 73, 8169, 1993; Saito, M., Kosaka, T., Matsubala, G., and Waseda, Y., Role of thromboxane A2 and platelet-activating factor in allergic bronchoconstriction in guinea pig airway in vivo, Mater. Trans. JIM 36, 1, 1995.

8080_C004.fm Page 274 Tuesday, April 3, 2007 1:54 PM

274

Organic Field-Effect Transistors

13. Takagi, Y., Kikuchi, T., Mizutani, T., Imafuku, M., Sasaki, S., and Mori, T., Upgrade of triple-axis/four-circle diffractometer at PF-BL3A, Rev. Sci. Instrum. 66, 1802, 1995. 14. Huang, T.C., Toney, M.F., Brennan, S., and Rek, Z., Analysis of cobalt-doped iron oxide thin films by synchrotron radiation, Thin Solid Films 154, 439, 1987. 15. Scherrer, P., Estimation of Size and Internal Structure of Colloidal Particles by Means of Röntgen Rays, Göttinger Nachrichten 2, 98, 1918. Klug, H.P. and Alexander, L.E. X-ray diffraction procedures, Wiley, New York, 1954. 16. Pietsch, U., Investigation of a semiconductor superlattice by use of grazing incidence x-ray diffraction, Appl. Surf. Sci. 54, 502, 1992. 17. Kaganer, V.M. and Loginov, E.B., Symmetry and phase transitions in Langmuir monolayers: The Landau theory, Phys. Rev. E. 51, 2237, 1995; Kaganer, V.M., Möhwald, H., and Dutta, P., Structure and phase transitions in Langmuir monolayers, Rev. Mod. Phys. 71, 779, 1999. 18. Leveiller, F., Jacquemain, D., Leiserowitz, L., Kjaer, K., and Als-Nielsen, J., Toward a determination at near atomic resolution of two-dimensional crystal structures of amphiphilic molecules on the water surface: A study based on grazing incidence synchrotron x-ray diffraction and lattice energy calculations, J. Phys. Chem. 96, 10380, 1992. 19. Als-Nielsen, J., Jacquemain, D., Kjaer, K. et al., Principles and applications of grazing incidence x-ray and neutron scattering from ordered monolayers at the air-water interface, Phys. Rep. 246, 251, 1994. 20. Sirringhaus, H., Brown, P.J., Friend, R.H. et al., Two-dimensional charge transport in self-organized, high-mobility conjugated polymers, Nature 401, 685, 1999. 21. Samuelsen, E.J. and Mårdalen, J., The Structure of Polythiophenes in Handbook of Organic Conductive Molecules and Polymers, In Handbook of organic conductive molecules and polymers, vol. 3, ed. H.S. Nalwa, Wiley, Chichester, UK, 1997, 87–120. 22. Kobashi, M. and Takeuchi, H., Inhomogeneity of spin-coated and cast nonregioregular poly(3-hexylthiophene) films. Structures and electrical and photophysical properties, Macromolecules 31, 7273, 1998. 23. Warren, B.E. X-ray diffraction, Addison–Wesley, Reading, MA, 1969, 41–50. 24. Bao, Z., Dodabalapur, A., and Lovinger, A.J., Soluble and processable regioregular poly(3-hexylthiophene) for thin film field-effect transistor applications with high mobility, Appl. Phys. Lett. 69, 4108, 1996. Sirringhaus, H., Tessler, N., and Friend, R. H., Integrated optoelectronic devices based on conjugated polymers, Science 280, 1741, 1998. 25. Prosa, T.J., Winokur, M.J., Moulton J., Smith P., and Heeger, A.J., X-ray structural studies of poly(3-alkylthiophenes): An example of an inverse comb, Macromolecules 25, 4364, 1992. 26. Fell, H.J., Samuelsen E.J., Als-Nielsen J., Grubel G., and Mardalen J., Unexpected orientational effects in spin-cast, submicron layers of poly(alkylthiophene)s: A diffraction study with synchrotron radiation, Solid State Commun. 94, 843, 1995. 27. Kline, R.J., McGehee, M.D., Kadnikova, E.N., Liu, J., and Fréchet, J.M.J., Controlling the field-effect mobility of regioregular polythiophene by changing the molecular weight, Adv. Mater. 15, 1519, 2003. 28. Kline, R.J., McGehee, M.D., Kadnikova, E.N. et al., The dependence of regioregular poly(3-hexylthiophene) film morphology and field-effect mobility on molecular weight, Macromolecules 38, 3312, 2005. 29. Kline, R.J., McGehee, M.D., and Toney, M.F., Highly oriented crystals at the buried interface in polythiophene thin film transistors, Nat. Mater. 5, 222, 2006. 30. Zen, A., Pflaum, J., Hirschmann, S. et al., Effect of molecular weight and annealing of poly(3-hexylthiophene)s on the performance of organic field-effect transistors, Adv. Func. Mater. 14, 757, 2004.

8080_C004.fm Page 275 Tuesday, April 3, 2007 1:54 PM

Grazing Incidence X-Ray Diffraction (GIXD)

275

31. Lin, Y.-Y., Gundlach, D.J., Nelson, S.F., and Jackson, T.N., Stacked pentacene layer organic thin film transistors, IEEE Electron Device Lett. 18, 606, 1997; Dimitrakopoulos, C.D. and Malenfant, P.R.L., Organic thin film transistors for large area electronics, Adv. Mater. 14, 99, 2002; Karl, N., Charge carrier transport in organic semiconductors, Synth. Met. 133, 649, 2003; Horowitz, G., Organic field-effect transistors, Adv. Mater. 10, 365, 1998. 32. Baude, P.F., Ender, D.A., Haase M.A. et al., Pentacene-based radio-frequency identification circuitry, Appl. Phys. Lett. 82, 3964, 2003. 33. Cornil, J., Calbert, J.P., and Brédas, J.L., Electronic structure of the pentacene single crystal: Relation to transport properties, J. Am. Chem. Soc, 123, 1250, 2001; Cheng, Y.C., Silbey, R.J., da Silva, D.A. et al., Three-dimensional band structure and bandlike mobility in oligoacene single crystals: A theoretical investigation J. Chem. Phys. 118, 3764, 2003. 34. Fritz, S.E., Martin, S.M., Frisbie, C.D. et al., Structural characterization of a pentacene monolayer on an amorphous SIO2 substrate with grazing incidence x-ray diffraction, J. Am. Chem. Soc. 126, 4084, 2004. 35. Holmes, D., Kumaraswamy, S., Matzger, A.J., and Vollhardt, K.P.C., On the nature of nonplanarity in the [N]phenylenes, Chem. Eur. J. 5, 3399, 1999; Mattheus, C.C., Dros, A.B., Baas, J., Meetsma, A., de Boer, J.L., and Palstra, T.T.M., Polymorphism in pentacene, Acta Crystallogr. C57, 939, 2001. 36. Mattheus, C.C., Dros, A.B., Jakob, B. et al., Identification of polymorphs of pentacene, Synth. Met. 138, 475, 2003. 37. Minakata, T., Imai, Ozaki, M., and Saco, K., Structural studies on highly ordered and highly conductive thin films of pentacene, J. Appl. Phys. 72, 5220, 1992. 38. Bouchoms, I.P.M., Schoonveld, W.A., Vrijmoeth J., and Klapwijk, T.M., Morphology identification of the thin film phases of vacuum evaporated pentacene on SiO2, Synth. Met. 104, 175, 1999; Knipp, D., Street, R.A., Volkel, A., and Ho, J., Pentacene thin film transistors on inorganic dielectrics: Morphology, structural properties, and electronic transport, J. Appl. Phys. 93, 347, 2003. 39. Kuzmenko, I., Rapaport, H., Kjaer, K. et al., Design and characterization of crystalline thin film architectures at the air-liquid interface: Simplicity to complexity, Chem. Rev. 101, 1659, 2001. 40. Kelley, T., Boardman, L.D., Dunbar, T.D. et al., High-performance OTFTs using surface-modified alumina dielectrics, J. Phys. Chem. B 107, 5877, 2003. 41. Ruiz, R., Choudhary, D., Nickel, B. et al., Pentacene thin film growth, Chem. Mater. 16, 4497, 2004; Laquindanum, J.G., Katz, H. E., Dodabalapur, A., and Lovinger, A. J., n-Channel organic transistor materials based on naphthalene frameworks, J. Am. Chem. Soc. 118, 11331, 1996; Fritz, S.E., Kelley, T.W., and Frisbie, C.D., Effect of dielectric roughness on performance of pentacene TFTs and restoration of performance with a polymeric smoothing layer, J. Phys. Chem. B 109, 10574, 2005; Nickel, B., Barabash, R., Ruiz, R. et al., Dislocation arrangements in pentacene thin films, Phys. Rev. B 70, 125401, 2004; Verlaak, S., Steudel, S., Heremans, P., Janssen, D., and Deleuze, M., Nucleation of organic semiconductors on inert substrates, Phys. Rev. B 68, 195409, 2003; Heringdorf, F.J.M.Z., Reuter, M.C., and Tromp, R.M., Growth dynamics of pentacene thin films, Nature 412, 517, 2001; Kasaya, M., Tabata, H., and Kawai, T., Scanning tunneling microscopy and molecular orbital calculation of pentacene molecules adsorbed on the Si(100)2 × 1 surface, Surf. Sci. 400, 367, 1998; Pratontep, S., Brinkmann, M., Nüesch, F., and Zuppiroli, L., Nucleation and growth of ultrathin pentacene films on silicon dioxide: Effect of deposition rate and substrate temperature, Synth. Met. 146, 387, 2004; Lim, S.C. and Kim, S.H., Surfacetreatment effects on organic thin-film transistors, Synth. Met. 148, 75, 2005.

8080_C004.fm Page 276 Tuesday, April 3, 2007 1:54 PM

276

Organic Field-Effect Transistors

42. Yang, H., Shin, T.J., Ling, M.-M. et al., Conducting AFM and 2D GIXD studies on pentacene thin films, J. Am. Chem. Soc. 127, 11542, 2005. 43. Merlo, J.A., Newman, C. R., Gerlach, C.P. et al., p-Channel organic semiconductors based on hybrid acene-thiophene molecules for thin-film transistor applications, J. Am. Chem. Soc. 127, 3997, 2005. 44. Horowitz, G., Bachet, B., Yassar, A. et al., Growth and characterization of sexithiophene single crystals, Chem. Mater. 7, 1337, 1995.

8080_C004.fm Page 277 Tuesday, April 3, 2007 1:54 PM

4.2

Near-Edge X-Ray Absorption Fine Structure (NEXAFS) Spectroscopy

Dean M. DeLongchamp, Eric K. Lin, and Daniel A. Fischer CONTENTS 4.2.1 Introduction................................................................................................277 4.2.1.1 The Importance of Structure .......................................................277 4.2.1.2 NEXAFS Background.................................................................280 4.2.1.3 NEXAFS for Organic Electronics ..............................................282 4.2.2 Experimental Considerations.....................................................................284 4.2.3 Data Analysis for Orientation....................................................................287 4.2.4 Examples of Applied NEXAFS Spectroscopy..........................................289 4.2.4.1 Pentacene.....................................................................................289 4.2.4.2 Poly(3-Hexyl Thiophene)............................................................292 4.2.4.3 NEXAFS of Oriented Liquid Crystalline Polymers...................294 4.2.4.4 NEXAFS for Molecular Electronics...........................................295 4.2.5 Future Horizons for NEXAFS Spectroscopy............................................295 References..............................................................................................................296

4.2.1 INTRODUCTION 4.2.1.1 THE IMPORTANCE

OF

STRUCTURE

Organic electronics components may soon enable the low-cost, high-volume manufacture of electronic circuitry to support emerging applications such as flexible displays, radio frequency identification (RFID) tags, and plastic photovoltaics. Highperformance organic semiconductors are required to realize these possibilities, and they have proven to be the most difficult component of organic circuitry to develop and optimize. A prime challenge of organic semiconductor development has been to understand the development of their crystalline microstructure during film

277

8080_C004.fm Page 278 Tuesday, April 3, 2007 1:54 PM

278

Organic Field-Effect Transistors

formation and the impact of that microstructure on the electronic properties of the film. This understanding would enable the intelligent design of molecules and processing strategies to produce organic semiconducting layers with robust and predictable performance for targeted applications. Organic semiconductor molecules typically feature a primary chemical structure with extended conjugation. These molecules often crystallize well but lack covalent bonds between molecular repeats, and their crystals are held together by weak van der Waals forces. The strength and geometry of these intermolecular interactions determine the degree of electronic overlap and therefore the efficiency of carrier transport between molecules within a crystal. The anisotropy of molecular packing also renders the carrier mobility within a single crystal strongly anisotropic. Furthermore, practical organic thin-film transistors (OTFTs) have channels with micronscale dimensions that often appear to be filled with many hundreds or thousands of semiconductor crystals. The orientation of crystals within the transistor channel with respect to each other and to the substrate plane will influence the field effect mobility that can be achieved in a particular OTFT. Grain boundaries, gaps, and amorphous regions will also influence carrier mobility. The size, orientation, and distribution of organic semiconductor crystals within a thin film is typically strongly dependent on the purity and processing of the semiconductor, and differences in these parameters are largely responsible for the widespread, general increase in reported organic semiconductor performance, even for materials with the same nominal primary chemical structure. The conjugated network of an organic semiconductor molecule is its most important structural motif because it is within this network that carriers reside and travel. Within this conjugated network, the extent of inter- and intramolecular overlap of the π-orbital electron clouds correlated to a carrier band is the primary concern of most structural characterization methods. At the molecular level, there are essentially three types of structure considerations: (1) the orientation of molecular conjugated planes with respect to each other; (2) the orientation of molecular conjugated planes with respect to the device plane; and (3) the orientation of the conjugated long axis for molecules, such as polymers, with significantly long axes. The first structure consideration concerns crystal packing style. The packing of small molecule organic semiconductors typically falls between a “herringbone” edge-to-face interaction [1–3] and cofacial π-stacking [4–8]. Cofacial π-stacking is considered to be the superior motif because the increased physical overlap of the π-orbitals may lead to increased electronic overlap and more efficient carrier hopping. Even so, some of the highest mobility small molecule semiconductors have been of the herringbone variety, most notably pentacene [9–11] and rubrene [12–14]. Synthetic control over this packing style and the purposeful enforcement of cofacial π-stacking has become a focus of primary chemical structure studies in organic semiconductors, with some recent successes based on the addition of bulky side groups that block the edge-to-face interaction [4–8]. Notably, semiconducting polymers with reasonable performance such as the regioregular poly(3alkyl thiophenes) (P3ATs) are believed to pack naturally in a cofacial π-stacked manner that is enforced by the segregation of their side chains into aliphatic lamellae [15,16].

8080_C004.fm Page 279 Tuesday, April 3, 2007 1:54 PM

Near-Edge X-Ray Absorption Fine Structure

279

The second structural consideration concerns the alignment of packed crystals with respect to the device plane that can be induced by chemical interactions, anisotropic crystal geometry, or specific processing steps. Single crystals typically exhibit anisotropic mobility [12–14,17–19], and it is believed that the highest mobility within a single crystal is in the direction of the greatest electronic overlap of the conjugated planes of individual molecules. Thus, for device geometries of OTFTs, where carriers move horizontally across the device plane, the preferential orientation of conjugated planes is edge-on with respect to the substrate. This orientation allows π-overlap from side to side so that carriers may travel more efficiently in the device plane. For the device geometries of organic photovoltaic (OPV) cells, where large exciton diffusion lengths through the thickness of the device plane may be advantageous, the preferential orientation of conjugated planes may be in a flat or planeon orientation (most often, the semiconductors in organic light-emitting diodes [OLEDs] are made amorphous to avoid excimer formation). The semiconductors most commonly used in specific applications appear to naturally adopt the preferred device-relative conjugated plane orientation atop common substrates. For example, pentacene and P3ATs naturally adopt an edge-on orientation atop typical dielectrics in OTFTs. The third structural consideration of long-axis orientation applies primarily to polymer semiconductors, where the orientation of the conjugated polymer long axis may result in anisotropic carrier mobility because transport along the polymer chain may be preferred to interchain hopping [20–23]. Long-axis orientation is typically imposed by processing; the methods used are borrowed from liquid crystal alignment methods and are often applied to polymer semiconductors in a liquid crystalline state. These methods may include mechanically rubbing the semiconductor film directly or casting it upon a mechanically rubbed substrate. Typically, measurable but modest increases in mobility anisotropy are observed. Long-axis orientation of electroluminescent polymers is an emerging theme in the OLED community, where it has been shown to lead to polarized light emission [24–28]. Other structural considerations exist at length scales larger than molecules. Defects such as grain boundaries, dewetting or film gaps due to materials’ incompatibilities, and chemical contamination or degradation must also be considered. These types of defects become more common when organic semiconductors are processed from a fluid, which allows bottom-up device fabrication as an alternative to photolithography. Upon application to a substrate from a fluid, a solid semiconductor layer is formed in a dynamic assembly process at the substrate interface that may include the nucleation, growth, and orientation of crystalline material. Chemistry, chemical interactions, or heat may be employed to optimize these processes. In an OTFT, the microstructure and defects that develop near the dielectric and electrode interfaces will determine the semiconducting layer’s field-effect mobility, trap depth and distribution, and injection properties. Characterizing these many aspects of microstructure is necessary to establish relationships between primary chemical structure, processing, and performance. Currently, the most commonly used methods are scanning probe microscopy techniques such as atomic force microscopy (AFM) or kelvin probe force microscopy

8080_C004.fm Page 280 Tuesday, April 3, 2007 1:54 PM

280

Organic Field-Effect Transistors

(KPFM) and crystallographic techniques such as grazing-incidence x-ray diffraction (GIXD). The microstructure information collected by these techniques can sometimes be correlated to fundamental electronic properties or device performance. AFM and KPFM provide a local probe of nanometer-scale morphology or electrical heterogeneity below the light diffraction limit, while GIXD provides a whole-film measurement of unit cell dimensions and preferential orientation for crystalline regions. We describe here a complementary technique — near-edge x-ray absorption fine structure (NEXAFS) spectroscopy, which can be used to quantify chemical composition, molecular orientation, and defects for broad classes of organic semiconductor films.

4.2.1.2 NEXAFS BACKGROUND NEXAFS was developed in the mid-1980s with a focus on the structure and chemistry of molecules physisorbed or chemisorbed at interfaces [29]. Early applications of NEXAFS were primarily for catalysis [30]. NEXAFS measures the absorption of linearly polarized soft x-rays within 30 eV of the K-shell threshold into resonant electron excitations. This x-ray absorption results in the “fine structure,” which is a collection of absorption peaks near the core shell ionization edge. The technique is element sensitive by energetic selection of the K-shell that is accessed. Most often, NEXAFS is performed on the K-edges of low-Z elements with binding energies less than 1 keV, such as carbon (285 eV), nitrogen (400 eV), oxygen (535 eV), and fluorine (685 eV) [31]. Because the NEXAFS of low-Z elements requires a tunable monochromatic incident beam in the soft x-ray range with significant flux, it must be practiced at synchrotron facilities within an ultrahigh vacuum sample environment. The excitation of K-shell (1s) electrons by soft x-rays can result in either a bound state or a continuum state. Continuum state excitations create the photoelectrons that are analyzed in the well-known x-ray photoemission spectroscopy (XPS) technique. In contrast, the bound state excitations of NEXAFS occur as either Rydberg transitions or resonant excitations of 1s electrons to unfilled (typically antibonding) molecular orbitals, which may have either π- or σ-symmetry. A potential energy diagram is shown in Figure 4.2.1(a) to illustrate these transitions. A 1s → π* transition typically occurs at energies less than the vacuum level, and a 1s → σ* transition typically occurs at energies greater than the vacuum level. Among 1s → σ* transitions, the resonance position varies systematically with σ-bond length; lower energy resonance positions are expected for longer σ-bonds. The quantized energy separation between transitions creates the fine structure of near-edge x-ray absorbance and allows precise determination of the intramolecular bonding present within complex molecules. Spectrum interpretation is best illustrated with an example. In Figure 4.2.2, the carbon K-edge NEXAFS spectrum for poly(3-hexylthiophene) (P3HT) is shown. There are four distinct peaks. The lowest energy peak, near 286 eV, is the 1s → π* peak associated with the carbon–carbon double bonds of thiophene [30] (for P3HT, they are delocalized over the backbone). The second resonance at 288 eV is a combination of carbon–hydrogen, carbon–sulfur, and Rydberg excitations [30]. The

8080_C004.fm Page 281 Tuesday, April 3, 2007 1:54 PM

Near-Edge X-Ray Absorption Fine Structure

281

Continuum states σ∗ Rydberg states

π

Vacuum

π∗

σ hν

1s

(a)

Single bond

σ∗

Double bond

Conjugated system

σ∗

σ∗

π∗

σ∗

σ∗

π∗

π∗ π∗

σ∗

(b)

FIGURE 4.2.1 a) Band energy diagram depicting a NEXAFS resonant excitation. Incident soft x-ray photons excite 1s electrons to unfilled molecular orbitals such as the π* or σ*. A 1s → π* transition, as shown here, can occur at energies below the ionization edge. b) Directional resonances are dependent on the spatial location of the final state orbital, and can be expressed as vectors or planes. (Adapted from Stöhr, J., NEXAFS spectroscopy, Springer–Verlag, Berlin, 1992.)

edge jump, where continuum states are accessed, appears at ≈289 eV. This edge jump causes a step change in the absorbance, so all energies above the edge jump exhibit a constant absorbance that has been normalized to 1 in Figure 4.2.2. The peak near 293 eV is associated with the 1s → π* of carbon–carbon single bonds, while the peak near 303 eV is associated with the 1s → σ* of carbon–carbon double bonds. Both peaks are present in conjugated systems [32]. This peak hierarchy follows Figure 4.2.1(a). The intensities of these peak shapes are equivalent to the stoichiometric bond densities within the sampled volume.

8080_C004.fm Page 282 Tuesday, April 3, 2007 1:54 PM

282

Organic Field-Effect Transistors

Partial electron yield

3.0

σ∗C-S/C-H

2.5 2.0 1.5

σ∗C-C σ∗C=C

π∗C=C

S

n

1.0 0.5 0.0 280

290

300

310

320

330

Photon energy (eV)

FIGURE 4.2.2 NEXAFS spectrum of poly(3-hexylthiophene) (P3HT) with labeled resonances. PEY standard experimental uncertainty is <2%.

Because NEXAFS peaks are typically well separated in energy, as in Figure 4.2.2, quantitative chemical analyses can be performed by fitting a sum of peak shapes to the spectrum, then integrating the peak shapes to determine the resonance intensities. These intensities are proportional to the stoichiometric number density of bonds within the sampled volume. A key advantage of NEXAFS over other common surface-sensitive chemical analysis techniques such as XPS or secondary ion mass spectrometry (SIMS) is that NEXAFS can determine bond orientation [29]. The initial state of K-edge excitations is by definition the 1s orbital, which is spherically symmetric. However, the final state is typically an antibonding orbital that is correlated to a bond and highly directional. The transition dipole matrix element will be dominated by this directional final state orbital. The maximum final-state orbital amplitude relative to the atomic center of excitation will determine the directionality of the resonance, as shown in Figure 4.2.1(b). The spatial orientation of a π*-orbital is along a π-bond. The spatial orientation of a σ*-orbital is orthogonal to a σ-bond. When more than one resonance exists, these vectors can add to form planes, as do the 1s → σ* resonances in benzene. A conjugated plane has a single 1s → π* vector that is perpendicular to the plane [31]. For simple molecules, the sum of σ*- and π*-resonance directions can often be decomposed into a series of vectors or planes. The intensity of a resonance with vector direction will be proportional to cos2δ, where δ is the angle between the electric field vector of polarized incident soft x-rays and the vector of the resonance orbital. From this simple relationship, a surface-relative bond orientation can often be determined. Often one or two bond orientations are sufficient to uniquely specify a molecular orientation.

4.2.1.3 NEXAFS

FOR

ORGANIC ELECTRONICS

NEXAFS spectroscopy holds several key advantages for investigating organic semiconductor thin film microstructure. Most importantly, it is highly sensitive to π-

8080_C004.fm Page 283 Tuesday, April 3, 2007 1:54 PM

Near-Edge X-Ray Absorption Fine Structure

283

bonds, which are detected in spectra as 1s → π* resonances. Carbon K-edge 1s → π* resonances typically appear at incident energies between 284 and 287 eV, as in Figure 4.2.2. This excitation appears at energies less than the absorption edge and is more easily deconvoluted than most peaks in a NEXAFS spectrum. Because of this sensitivity, NEXAFS is particularly well suited to characterizing the extended conjugated networks of organic semiconductors. Of the three fundamental aspects of molecular level structure discussed previously, NEXAFS is particularly well suited to characterizing substrate-relative conjugated plane orientation. Most organic semiconductors feature a single conjugated plane, and the directionality of the 1s → π* transition can typically be expressed as a single vector normal to any conjugated plane, as shown in Figure 4.2.1(b) for a benzene molecule. From a single set of NEXAFS spectra collected at different incident angles, it is straightforward to establish whether conjugated molecules within the sampled volume are oriented preferentially edge-on or preferentially plane-on with respect to the substrate. The extent of orientation can also be extracted within limits that will be discussed later. Very oriented conjugated planes imply high levels of order and crystallinity within a thin film, and in some cases NEXAFS orientation measurements can be interpreted as a secondary measurement of the extent of crystallinity. A second fundamental aspect of structure that NEXAFS can measure is the long axis orientation of semiconducting polymers. The resonance most typically used to make this measurement is again the 1s → π* transition. A long axis orientation assessment requires spectra collected at multiple sample rotations with the beam near normal incidence (as well as multiple incident angles, to specify the backbone orientation exactly). The orientation of the π*-orbitals will generally be distributed in some manner about the plane orthogonal to the long axis of a polymer. Orientational order parameters that are typically used to evaluate liquid crystal mesogen ordering along a director can be extracted from the NEXAFS measurement to quantify the degree of polymer chain alignment. Although these substrate-relative orientation measurements permit useful microstructure analysis of an organic semiconductor thin film, NEXAFS cannot evaluate crystal packing style or the orientation of molecules with respect to each other. NEXAFS, like all spectroscopic techniques, provides orientation information that is the azimuthally averaged mean of individual orientations within the sampled volume. Thus, if a bimodal molecular orientation distribution exists, as would be expected in a herringbone structure, NEXAFS provides only the first moment of that distribution (or any distribution). Because NEXAFS provides an azimuthal average, if a resonance is oriented at the “magic angle” of 54.7° with respect to surface normal, it is impossible to determine whether the resonance is oriented. Extra care must be exercised when interpreting resonance orientations near 54.7°. The controlled NEXAFS sampling volume is an additional attractive feature for organic electronics measurements. NEXAFS spectra collected in partial electron yield (PEY) mode are measured by counting Auger electrons ejected from the film. Energetic selection of electrons with a high pass grid covering the detector permits a simple depth profiling capability. A low grid bias (e.g., –50 V for the carbon edge) will admit most electrons, even those inelastically scattered after originating from

8080_C004.fm Page 284 Tuesday, April 3, 2007 1:54 PM

284

Organic Field-Effect Transistors

deep within the film. A high grid bias (e.g., –250 V) will reject most electrons except for those originating near the free surface. Bias adjustment in PEY mode allows the sampled depth to be systematically controlled from ≈2–10 nm [33,34]. This depth range compares favorably to ≈5- to 6-nm dimensions of the mobile carrier region near the dielectric in OTFTs. If the bottom interface of the semiconductor can be exposed, NEXAFS will provide a measurement of molecular orientation that is localized to the mobile carrier region. Further, the localized sampling volume and elemental sensitivity of NEXAFS allow defect detection by chemical recognition of underlying substrates such as an oxide dielectric within the sampling volume. Greater amounts of substrate detection within the sampled volume can be interpreted as thinning or dewetting of the semiconductor layer, and this defect formation can be quantified. Chemical degradation or contamination present at levels above the standard error of PEY detection — about 2% — can also be detected. NEXAFS spectroscopy brings many unique capabilities to organic electronics and, in many cases, it can succeed as a solo technique in investigations of organic semiconductor microstructure development. NEXAFS is most powerful, however, when it is combined with crystallographic techniques such as GIXD or scanning probe techniques such as AFM. When combined with GIXD, NEXAFS can identify or set limits on the orientation of the disparate chemical moieties that form a molecule (such as conjugated plane tilt vs. side chain tilt). This information can be correlated to the spacing of the periodic crystal lattice, which can then be used to assign a more exact structure to the lower order peaks that result from GIXD of thin films. When combined with scanning probe techniques, NEXAFS can assign orientations of molecules with respect to the three-dimensional shape and orientation of nano- or microscale crystalline objects such as rods, needles, or plates, which can explain how these objects assemble. A combination of complementary microstructural analysis techniques is the optimal toolset with which to develop a comprehensive and self-consistent understanding of the semiconductor film formation process.

4.2.2 EXPERIMENTAL CONSIDERATIONS Many photon absorption spectroscopies use a transmission geometry in which a beam is measured before and after passing through a sample. NEXAFS spectra of solids are rarely collected in transmission because the sample would have to be a freestanding film less than 100 nm thick, which is difficult to fabricate and maintain. The required photon flux would be large, and degradation of the film would be a likely result. An exception to this generalization is NEXAFS microscopy, which does use a tightly focused beam in transmission [35–37]. Most NEXAFS beamlines measure soft x-ray absorbance by monitoring the energetic decay of the excited state, which occurs nonradiatively as Auger electron emission and radiatively as photon fluorescence. Auger electron emission provides the most common quantification of soft x-ray absorption into bound states. Auger emission occurs when an outer shell electron from the same atom decays into a core hole formed by soft x-ray excitation. The decay energy, which is the difference between outer shell and core orbital energies, is transferred to a second outer shell

8080_C004.fm Page 285 Tuesday, April 3, 2007 1:54 PM

Near-Edge X-Ray Absorption Fine Structure

285

Surface chemistry Partial electron yield detector Tunable grid

Automated

Electron Tunable incident soft x-rays

Sample Electric field vector Photon

Fluorescent photon yield detector Bulk chemistry

Bond orientation

FIGURE 4.2.3 NEXAFS experimental apparatus at the NIST/Dow soft x-ray materials characterization facility, beamline U7A at Brookhaven National Laboratory.

electron, which is then ejected from the atom. Auger emission is preferred over photon fluorescence by over two orders of magnitude. Because the Auger electron kinetic energy is determined by a set orbital energy difference that is constant for a specific element, it will remain constant over a NEXAFS K-edge, and spectra can be collected in Auger electron yield (AEY) mode by selecting a band pass energy window to admit exclusively Auger electrons. Because the mean free path of electrons with Auger kinetic energy is small relative to typical sample thicknesses, inelastic collisions will cause many electrons to lose energy before leaving the film. This circumstance allows PEY collection mode, where the detector energy window is broadened to include electrons with lower kinetic energy. By controlling the PEY electron rejection bias on a tunable grid (see Figure 4.2.3), depth selectivity can be achieved [33]. Higher biases admit only electrons ejected near the free surface, whereas lower biases admit additional electrons ejected from deep within the film and inelastically scattered [38]. The total electron emission of a sample can also be measured from the current flowing into the grounded sample. Spectra collected in this mode, called total electron yield (TEY) [39,40], have a smaller signal-to-noise ratio because the NEXAFS is superimposed atop a large background of photoelectron emission. Fluorescent photon counting is an alternative to electron yield (EY) [41]. The strongest advantage of fluorescence yield (FY) is that the mean free path of photons through solids is much longer than that of electrons. Therefore, FY mode reveals xray absorption from deeper into the film bulk (≈100 nm) than EY modes can measure (≈10 nm). A comparison of EY and FY NEXAFS spectra can reveal surface

8080_C004.fm Page 286 Tuesday, April 3, 2007 1:54 PM

286

Organic Field-Effect Transistors

segregation and heterostructure formation [42–44]. This comparison might be used in evaluating organic semiconductor films to determine whether interface and bulk molecular orientations are different. Orbital orientation with respect to the substrate plane can be determined experimentally by varying the tilt of the sample and, by extension, the incident beam angle, as shown in Figure 4.2.3. The sample may be tilted from 20 to 90° with respect to the incident beam (by extension, with respect to the electric field vector). An orientation tendency is sometimes reported by comparing spectra collected only at two extremes. Robust discrimination between different orientations, however, is only possible if more angles are used. We find that the standard deviation of fits based on the data analysis outlined in Section 4.2.3 is minimized by including spectra collected at five to seven different incident angles spaced between 20 and 90°. In most cases, using fewer than five angles adds uncertainty, and using more than seven angles does not reduce uncertainty. The uncertainty of a two-angle orientation analysis is infinitely large. These five to seven scans may be collected at different points on the sample surface to avoid sample degradation and to confirm that orientation is homogeneous across the surface. Using many angles also ensures that no experimental artifacts influence the orientation conclusions. Sample charging is important to consider in NEXAFS measurements because it can introduce experimental artifacts [45]. Electron emission is a necessary product of soft x-ray excitation, but it does cause the sample to become positively charged. Electron emission is then reduced because of increased electrostatic binding, attenuating EY signals. The extent of charge buildup depends on the beam intensity and energy, the spot size, the exposure duration, and the quality of sample grounding. The spectral signature of charging is a decaying yield signal, often artificially lowering the postedge intensity (because it is often collected last), which leads to normalization errors. A NEXAFS spectrum is typically normalized to account for changing spot size by dividing it by the intensity in the postedge at energies higher than typical boundstate resonances (≈330 eV for carbon) [46]. If the postedge is artificially depressed due to charging, then the rest of the spectrum will be artificially increased after normalization. This increase can cause misinterpretation of orientation because charging varies with incident angle. Charging is typically strongest at 20°, where the spot is largest, and it decreases as the angle comes closer to 90°, where the spot is smallest. The normalized peak intensities may appear to vary in a systematic manner with angle, leading to a false assessment of their orientations. Sample charging that causes an artificially low postedge will result in all of the bound state intensities appearing most intense at 20° and least intense at 90°, a situation that rarely occurs in properly measured and normalized spectra. Most angular collections of NEXAFS spectra exhibit isosbestic points where the intensity variation with angle “crosses over” or multiple regions of common intensity because most molecules have some orthogonal resonances. If sample charging is suspected, a different grounding scheme using an all-metal substrate and thin organic semiconductor film can be employed to check, although it is likely that the organic will selforganize differently on a metal surface. Avoiding charging with substrates common

8080_C004.fm Page 287 Tuesday, April 3, 2007 1:54 PM

Near-Edge X-Ray Absorption Fine Structure

287

to OTFTs can be difficult because they are typically metal oxides or other insulators engineered specifically for low current leakage.

4.2.3 DATA ANALYSIS FOR ORIENTATION NEXAFS data analysis to extract orbital orientation follows simple equations that are trigonometric expansions of the cos2δ relationship described in Section 4.1.3. The typical orientation of interest for organic semiconductors is a surface-relative orientation uniformly distributed about the surface normal. In this case, the azimuthally averaged tilt of the orbital can be determined. For a vector orbital, the intensity can be expressed as [31] ⎪⎧ P ⎡ 1 ⎤ (1 − P ) 2 ⎪⎫ I = A ⋅ ⎨ ⎢1 + 3 cos 2 Θ − 1 3 cos 2 α − 1 ⎥ + sin α ⎬ ⎦ 2 ⎩⎪ 3 ⎣ 2 ⎭⎪

(

)(

)

(4.2.1)

where I is the bound state resonance intensity, the integrated area of a normalized and deconvoluted peak A is a constant, roughly analogous to the extinction coefficient of conventional optical spectroscopy P is the polarization factor of the beamline Θ is the angle of the beam relative to the substrate plane α is the angle of the vector orbital relative to substrate normal The variables in Equation 4.2.1 can be related to the physical layout of sample and experimental geometry as shown in Figure 4.2.4. The polarization factor, P, is usually known; it depends on the radiation source and is closer to 1 for insertion devices such as wigglers or undulators as compared to bending magnets. The intensity, I, is measured as a function of Θ for many incident angles. Equation 4.2.1 can therefore be reduced to a linear fit of I vs. cos2Θ, where the slope and intercept can be transformed into the two remaining unknowns, A and α. A linear fit of I versus cos2Θ or I versus sin2Θ will reveal whether orientation is consistent for the collected spectra. If no systematic deviation from linearity is observed, there are probably no gross experimental artifacts. The sample charging artifact described in Section 4.2.2, however, can sometimes result in a linear fit because the illuminated spot size increases trigonometrically with incident angle. The standard deviations of the fit parameters provide some statistical uncertainty of the orientation measurement. With a sufficient number of points, a confidence interval can be determined. Small differences in orientation can then be judged on a statistical confidence basis. The parameters A and α can also be fit directly using nonlinear least-squares methods. The angle α describes the orientation of a vector orbital, but it is important to understand the meaning of this orientation. It is the azimuthal mean orientation of an ensemble of orbitals within the sampled volume. The underlying distribution of orientations within the ensemble usually cannot be determined. This consideration

8080_C004.fm Page 288 Tuesday, April 3, 2007 1:54 PM

288

Organic Field-Effect Transistors

Surface normal

Electric field vector Θ

Resonance α orientation

Incident X-rays

δ

Θ

FIGURE 4.2.4 Geometric layout of sample and experimental geometry. Definitions are as given for Equation 4.2.1.

is important when analyzing organic semiconductors, because some of their crystallization motifs have a bimodal surface-relative orientation (e.g., some herringbone structures). It is tempting to assign α to every molecule, but this is only appropriate when reasonable, such as for a filled self-assembled monolayer (SAM) surface. The azimuthal mean gives rise to an ambiguous quantity — the “magic angle” of 54.7°. If the vector orbital is tilted at 54.7°, there will be no variation in I with Θ. Conversely, if there is no variation in I with Θ, Equation 4.2.1 will reveal that α = 54.7°. An infinite number of orientation distributions can result in an azimuthal mean orientation of 54.7°, but the most important of these is complete disorder. If there is no variation in I with Θ, one may speculate but not prove that a sample is disordered. Because of these considerations, it is impossible to separate the orientation itself from the “extent of orientation.” This information is hidden in the orientation distribution. The exception to this generalization occurs when the orbital is strongly oriented far from the magic angle. If the orbital exhibits a strong horizontal or vertical tendency, fewer orientation distributions will fit. For the extremes of perfect horizontal or vertical orientation, unique distributions of comprehensive, single-angle orbital orientation at α = 0° or α = 90° could be rigorously proven. In general, the farther α is from 54.7°, the greater is the allowable confidence that the orientation distribution is tight. The magic angle ambiguity no longer occurs if molecular orbitals are highly oriented in the substrate plane [40,43,44,47]. The tilt angle still represents a mean, but the magic angle is now 45° and only occurs if the in-plane projection of the electric field vector is aligned with that of the orbital. The three-dimensional orbital orientation is determined by collecting spectra at multiple rotations of the sample

8080_C004.fm Page 289 Tuesday, April 3, 2007 1:54 PM

Near-Edge X-Ray Absorption Fine Structure

289

about surface normal, in addition to multiple beam incident angles. Equation 4.2.1 is no longer valid, but a slightly more complex form can be employed. This measurement is potentially useful for organic semiconductor characterization because in-plane orientation between an OTFT source and drain is desirable and is often created by mechanical alignment or casting upon oriented substrates. To avoid confusion, it is often desirable to use a quantity other than α to express the orientation tendency of a distribution. One possibility is the orientational order parameter S [48,49], which traditionally describes the mean orientation of liquid crystals. Assuming a surface-normal director, S=

(

)

1 3 cos 2 α − 1 2

(4.2.2)

The parameter S will vary from +1, for a vertical orbital, to 0, for magic angle orientation, to –0.5, for a horizontal orbital. S is not symmetric about the disorder condition, and it is useful for expressing “how good” the ordering is in a purposefully aligned material (e.g., how closely orientation follows the director). For general comparisons of orientation tendency, we prefer a dichroic ratio defined as R=

I ( 90°) − I ( 0°) I ( 90°) + I ( 0°)

(4.2.3)

Intensities at 0 and 90° are extrapolated from Equation 4.2.1, fitted to spectra from at least five angles. The parameter R varies from +0.7, for a horizontal orbital (not 1, because P ≈ 0.85), to 0, for magic angle orientation, to –1, for a vertical orbital. R is roughly symmetric about the disorder condition and it is useful for general comparisons. We emphasize that α, S, and R all express the same orientation information. One parameter does not reveal more than the others.

4.2.4 EXAMPLES OF APPLIED NEXAFS SPECTROSCOPY In this section, we provide worked examples of NEXAFS analysis of commonly used OTFT semiconductors pentacene and P3HT. In addition, we discuss some of the applications of NEXAFS to other materials of interest to the organic semiconductor community: liquid crystalline conjugated polymers and molecular electronics. We aim to provide a basis for understanding the practical implementation of the technique and a foundation for interpreting results.

4.2.4.1 PENTACENE In this example, a pentacene film was thermally evaporated to 50 nm thick on the native oxide of a room temperature silicon wafer. NEXAFS spectroscopy was performed on this film at the NIST/Dow soft x-ray materials characterization facility at the national synchrotron light source (NSLS) of Brookhaven National Laboratory. The carbon K-edge was collected at several incident angles, shown in Figure 4.2.5.

8080_C004.fm Page 290 Tuesday, April 3, 2007 1:54 PM

290

Organic Field-Effect Transistors

π∗C=C 285.8

Partial electron yield

4

3

σ∗C-C 294.8

π∗C=C 284.2

σ∗C=C 301

20° 31° 40° 47° 55° 62° 70°

2

1

0 280

295 290 Photon energy (eV)

285

300

305

(a)

7

R = 0.59 ± 0.01 S = −0.35 ± 0 α = 73.4 ± 0.4

π∗ intensity

6 5 4 3 2 0.0

0.2

0.4

0.6

0.8

1.0

sin2Θ (b)

FIGURE 4.2.5 a) NEXAFS spectra of a 50-nm thick evaporated pentacene film collected at a PEY bias of –50 V. These spectra are postedge normalized to the intensity at 330 eV to account for changing spot size. b) The intensity of the π* resonances calculated by integrating the spectra from 282 to 286.5 eV, displayed versus the squared sine of incident angle. The standard experimental uncertainty of PEY and its integrated intensity are <2%.

Collection bias was –50 V, so the sampled volume penetrates as much as ≈10 nm into the top surface. The most prominent absorbance is the carbon–carbon 1s → π* transition, which is split here into two peaks at 284.2 and 285.8 eV. The resonances at 294.8 and 301.0 eV are carbon–carbon 1s → σ* transitions of the single and double bonds, respectively. All transitions exhibit angular variation, and there is a clear isosbestic point around 294 eV. The π*-orbital intensity is greatest near normal incidence, and the σ*-orbital intensity is greatest near glancing incidence.

8080_C004.fm Page 291 Tuesday, April 3, 2007 1:54 PM

Near-Edge X-Ray Absorption Fine Structure

291

σ∗ π∗

FIGURE 4.2.6 Antibonding orbital directions of pentacene. Only the carbon–carbon orbital directions are shown.

The orientation of pentacene can be extracted from Figure 4.2.5(a) by considering the pentacene carbon–carbon molecular orbital orientations. A rough sketch of these is shown in Figure 4.2.6. The σ*-direction is represented as a single vector normal to the molecule, as for benzene in Figure 4.2.1(b). The σ*-direction of pentacene is more complex. It is a plane, but there are more σ*-bonds along the long axis of the molecule than there are along the short axis. An elliptical plane is therefore the best representation of the σ*-direction. We determine the molecular orientation of pentacene by comparing the angular variation of peaks in Figure 4.2.5(a) with their respective orbital orientations in Figure 4.2.6. The π*-orbital resonance is most intense near normal incidence, where the electric field vector is in the substrate plane. Thus, the π*-orbital is preferentially oriented in the substrate plane, and pentacene is oriented edge-on in the thin film. The π*-variation indicates only the conjugated plane tilt; we cannot determine the long-axis tilt from this resonance. The long-axis tilt can be determined from the σ*-intensity variation. The σ*intensity is greatest near glancing incidence, where the electric field vector is normal to the substrate plane. Thus, the long axis of the σ*-ellipse is preferentially normal to the substrate, and the long axes of pentacene are normal to the substrate. The edge-on orientation from the π* is self-consistent with the standing-up orientation from the σ*. Determination of the average tilt angle of the long axis would require a molecular model to determine the molecule-relative orientation of all σ-bonds in the pentacene molecular skeleton, and the assumption would have to be made that all π-bonds contribute equally to the 1s → σ* resonance. The orientation of the pentacene conjugated plane can be quantified using the analysis outlined in Section 4.2.3 because it is a single vector. The π*-intensity from Figure 4.2.5(a) is integrated from 283 to 286.5 eV. The intensity I varies with Θ according to Equation 4.2.1, as shown in Figure 4.2.5(b). From Equation 4.2.1, R, S, and α, which quantify the extent of edge-on ordering in pentacene molecules, can be determined. With R = 0.59 ± 0.01*, pentacene exhibits a high degree of edge-on character, and its orientation distribution is narrow. The conjugated plane normal has an average orientation angle of about 75° away from surface normal. The strongly edge-on character of pentacene orientation is correlated to strong

8080_C004.fm Page 292 Tuesday, April 3, 2007 1:54 PM

292

Organic Field-Effect Transistors

π-interactions in the source-drain plane, consistent with pentacene’s high hole mobility in OTFT architectures. Differences in pentacene orientation with film thickness, substrate chemistry, and substrate temperature can be quantified and compared by this technique. For example, NEXAFS spectroscopy has recently been applied to investigate pentacene growth modes on a variety of substrate chemistries [50,51]. It was found that if pentacene is deposited on organic self-assembled monolayer surfaces terminated with methyl groups, terphenyl groups, or carboxyl groups, it exhibits the typical edge on alignment described previously. If pentacene is deposited on a clean bare gold surface, however, it exhibits a plane-on or flat alignment. In a similar experiment, it was found by another team that the common p-type organic semiconductor alpha-sexithienyl (α-6T) also orients flat when deposited atop a clean bare silver surface [52]. The conjugated plane of α-6T can adopt a flat or edge-on orientation, depending on temperature when deposited on a clean bare copper surface [53]. This substrate dependence of small molecule growth modes may explain why pentacene crystallinity is sometimes strongly disrupted at noble metal source and drain electrodes when it is deposited onto bottom-contact OTFT testbeds. Using analyses similar to those for pentacene, we recently completed a NEXAFS analysis of microstructure development in high-performance oligothiophene semiconductors derived from soluble precursors [54,55]. The chemical reaction coordinate of thermal conversion was measured simultaneously with the development of structural order. Defect formation from the autophobic dewetting of the semiconductor was also quantified by measuring exposure of the underlying oxide substrate. These soluble precursors undergo a complex microstructural development that is coupled to the thermally induced chemical changes. At optimal conversion temperatures, a strongly oriented product is attained that exhibits an edgeon conjugated plane and a vertical long axis. Full chemical conversion, the highest levels of molecular orientation, and the absence of critical defects clearly correlate to the optimal field effect mobility in OTFTs made from this promising new class of materials.

4.2.4.2 POLY(3-HEXYL THIOPHENE) NEXAFS orientation analysis can also be applied to polymer semiconductors. In this example, films of regioregular and regiorandom P3HT were spin-coated from chloroform to approximately 20 nm atop the native oxide of silicon, then melted at 250°C and slowly cooled under vacuum. NEXAFS carbon K-edge spectra were collected at the same conditions as for pentacene. The resultant spectra are shown in Figure 4.2.7(a–b). The regiorandom P3HT spectra exhibit no angular variation, and we may interpret this result to mean that the film is disordered. The regioregular P3HT spectra exhibit the same trend of π*-variation as pentacene, indicating again that the conjugated plane is edge-on. We quantify the π*-orientation using the fits in Figure 4.2.7(c–d). The regioregular P3HT exhibits R = 0.26 ± 0.01 for the conjugated plane. This dichroic ratio is far less than that of pentacene, and the distribution of tilts

8080_C004.fm Page 293 Tuesday, April 3, 2007 1:54 PM

Near-Edge X-Ray Absorption Fine Structure

Regiorandom

2.0 1.5 Θ = 20° 31° 40° 47° 55° 62° 70°

1.0 0.5 0.0 285

300 290 295 Photon energy (eV)

Regioregular

2.5 Partial electron yield

Partial electron yield

2.5

293

2.0 1.5

Θ = 20° 31° 40° 47° 55° 62° 70°

1.0 0.5 0.0

305

285

290 295 300 Photon energy (eV) (b)

(a)

π∗ intensity

π∗ intensity

1.8 1.6 1.4 1.2

R = 0.02 ± 0.01 S = −0.02 ± 0.01 α = 55.4 ± 0.3 0.0

0.2

Regiorandom 0.6

0.4

2Θ

sin (c)

305

0.8

1.0

1.8 1.6

R = 0.26 ± 0.01 S = −0.18 ± 0.01 α = 62.6 ± 0.4

1.4 1.2

Regioregular 0.0

0.2

0.6

0.4

0.8

1.0

2Θ

sin (d)

FIGURE 4.2.7 a): NEXAFS spectra of a spin-cast regiorandom P3HT film. b) NEXAFS spectra of a spin-cast regioregular P3HT film. These spectra are postedge normalized to the intensity at 330 eV to account for changing spot size. c) and d): The intensity of π* resonances calculated by integrating the spectra from 282 to 286 eV, displayed versus the squared sine of incident angle. The standard experimental uncertainty of PEY and its integrated intensity is <2%. The uncertainties of order parameters are standard deviations from direct fits of the parameters to intensity trends.

cannot be known. These samples are not likely represented by a single, uniform orientation; R reflects an average of the edge-on, plane-on, and amorphous orientations present. This result is consistent with reports comparing microstructures of P3HTs of varying regioregularity by other methods [15,16]. We could quantify the carbon–carbon σ*-orientation, but in this case (just as for pentacene) it is not possible to define a single orbital direction relative to the molecule, as illustrated in Figure 4.2.8. From Figure 4.2.7, the trend of the σ*-peaks with incident angle indicates a net vertical orientation of carbon–carbon σ*-bonds within the molecule, which is consistent with the expectation that these peaks arise primarily from the hexyl side chains. These side chains arrange into vertically segregated lamellae, as has been determined by diffraction [15,16]. A secondary contribution to the σ*-is from the polymer long axis, which is most likely oriented in the plane of the film. Precisely deconvoluting these two carbon–carbon σ*orientations may not be possible because their resonance energies are similar. How-

8080_C004.fm Page 294 Tuesday, April 3, 2007 1:54 PM

294

Organic Field-Effect Transistors

σ∗ (Side chain) σ∗ (Long axis) π∗ S

FIGURE 4.2.8 Antibonding orbital directions of P3HT. Only the carbon-carbon orbital directions are shown.

ever, qualitative judgments about the extent of side chain orientation may be made, and comparisons of side chain vertical order may be possible between members of a poly(3-alkylthiophene) synthetic series. We find that polymer semiconductor microstructure development is often well described by comparing R for π*. Here, the diagnostic is simply tested by comparing regioregular and regiorandom P3HT. More positive R values for the regioregular P3HT indicate a greater prevalence of edge-on orientation, which is presumably correlated to better π-stacking in the source-drain plane. In more extensive studies of regioregular P3HT, we find that R varies with the film crystallization rate as controlled by the speed of spin casting. Slower spin speed and slower crystallization resulted in a more positive R, confirming that the thermodynamically favored orientation of regioregular P3HT within polycrystalline thin films is with an edge-on conjugated plane. The presumably greater π-overlap achieved at slower spin conditions resulted in higher field-effect mobility [56].

4.2.4.3 NEXAFS

OF

ORIENTED LIQUID CRYSTALLINE POLYMERS

NEXAFS has proven to be a powerful technique to investigate the orientation of purposefully ordered liquid crystalline materials, which are typically aligned by direct mechanical shear or by application to a sheared substrate [40,57,58]. Recently, this technique has been applied to liquid crystalline systems with sufficient conjugation to behave as semiconductors. The most common liquid crystalline backbone that has been investigated so far is fluorene based. Jung et al. recently applied the NEXAFS technique to examine the preferential orientation of polymer segments at the surface of thin polyfluorene films [27]. By collecting total electron yield spectra and performing an analysis similar to that described above for P3HT, it was found that the conjugated plane of the fluorene backbone was oriented plane-on or flat with respect to the substrate. Direct mechanical shear (rubbing) of the polyfluorene resulted in a strong preferential orientation of polymer long axis along the shear vector. Heating the oriented polyfluorene into its nematic liquid crystalline state resulted in complete relaxation of the long axis orientation. The authors interpreted these results to conclude that the shear-induced

8080_C004.fm Page 295 Tuesday, April 3, 2007 1:54 PM

Near-Edge X-Ray Absorption Fine Structure

295

long axis alignment of the polyfluorene extended only a small distance from the top surface into the thickness of the film. More recently, Pattison et al. applied NEXAFS spectroscopy to investigate the orientation of liquid crystalline fluorene-thiophene copolymer thin film surfaces [28]. Partial electron yield spectra were recorded with the sample heated in situ within the measurement chamber to directly measure the orientation over the liquid crystalline phase diagram. Like Jung et al., these investigators also found that in thin polyfluorene films, the conjugated plane of the backbone lies planeon or flat with respect to the substrate. Casting and annealing the liquid crystalline polymer atop a rubbed polyimide alignment layer was found to template the polymer long-axis orientation. The quality of long axis orientation was strongly dependent on annealing temperature, and optimal temperatures led to strong longaxis dichroism.

4.2.4.4 NEXAFS

FOR

MOLECULAR ELECTRONICS

From the inception of the technique, NEXAFS spectroscopy has proven to be a valuable tool for investigating molecularly thin chemically or physically absorbed layers, with a natural extension to oriented self-assembled monolayers (SAMs) [33,59–63]. Typically, NEXAFS spectroscopy is employed to measure the tilt angle of SAMs with respect to the substrate; this measurement is directly related to the packing density and graft yield. In the molecular electronics community, small ensembles of molecules that are ideally highly oriented functional SAMs are considered to be potential units of electronic circuitry. The grafting density, packing style, and substrate-relative orientation of a functional SAM are key parameters that govern its expected performance, with many analogies to the fundamental structure considerations of larger scale organic electronics. The use of NEXAFS to provide structure information in molecular electronics has appeared only recently. Krapchetov et al. performed a NEXAFS analysis of the assembly of terphenyldithiol and quaterphenyldithiol on gold and gallium arsenide [63]. It was found that while the packing density and orientation of these conjugated SAMs were independent of solvent when applied to gold, they strongly depend on solvent when applied to gallium arsenide. Another NEXAFS study has examined the orientation of terthiophene thiols on gold [64]. Complex functional SAMs such as a C60-based thiol have also been investigated [65]. As electrical metrology capabilities in molecular electronics begin to mature, the correlation of structure to electronic properties will become an increasing priority, and the attractive orientation measurement capabilities of NEXAFS spectroscopy should prove useful to the development of this emerging technology.

4.2.5 FUTURE HORIZONS FOR NEXAFS SPECTROSCOPY Advances in the capabilities of NEXAFS spectroscopy and its impact on organic semiconductor analysis will most likely result from innovations in detector design

8080_C004.fm Page 296 Tuesday, April 3, 2007 1:54 PM

296

Organic Field-Effect Transistors

or the in situ measurement of samples under stimulus. An existing detection scheme that could provide new information is the FY detection of molecular orientation within thin organic semiconductor films. For films less than 100 nm thick, FY would provide an orientation uniformly averaged over the entire film thickness. PEY detection, on the other hand, provides a top surface weighted measure that includes contributions of up to 10 nm below the top surface. Because many functional organic films are cast at approximately 20–30 nm thick, the simple subtraction of PEY orientation from FY orientation would provide an indirect measure of the molecular orientation in the bottom of the film in the mobile channel region immediately adjacent to the dielectric substrate in OTFTs. Advances in detector technology may soon also provide band pass energy discrimination and more finely tuned depth profiling within the PEY detection range. Large-area detectors may allow simultaneous parallel angle-resolved NEXAFS measurements or the fast analysis of combinatorial gradient samples. The in situ measurement of samples under stimulus is another likely source of advancements. For example, the NEXAFS analysis of microstructure development during sample annealing may elucidate intermediate orientation states that cannot be measured in quenched samples. The application of bias during measurement may reveal changes to microstructure in driven devices, especially given their potential for ohmic heating. Although manipulation of sample environment is limited by the required high vacuum sample chamber, controlled dosing with water vapor or oxygen immediately before pump down may permit investigation of chemically degraded surface states. NEXAFS spectroscopy is only beginning to see use in the field of organic electronics, and its many unique capabilities will ensure that it has a clear place in the microstructure analysis toolbox. The combination of NEXAFS with the wealth of characterization techniques and knowledge already gained in this field will drive its further implementation. The development of clear correlations from primary chemical structure to microstructure to performance will allow the targeted synthetic design of materials for any application and may eventually facilitate the widespread commercial production of new products based on organic electronic components.

REFERENCES 1. Hotta, S. and Waragai, K., Crystal-structures of oligothiophenes and their relevance to charge-transport, Adv. Mater. 5, 896–908, 1993. 2. Baker, K.N. et al., Crystal-structures, phase-transitions and energy calculations of poly(p-phenylene) oligomers, Polymer 34, 1571–1587, 1993. 3. Horowitz, G. et al., Growth and characterization of sexithiophene single-crystals, Chem. Mat. 7, 1337–1341, 1995. 4. Anthony, J.E. et al., Functionalized pentacene: Improved electronic properties from control of solid-state order, J. Am. Chem. Soc. 123, 9482–9483, 2001. 5. Anthony, J.E., Eaton, D.L., and Parkin, S.R., A road map to stable, soluble, easily crystallized pentacene derivatives, Org. Lett. 4, 15–18, 2002. 6. Sheraw, C.D. et al., Functionalized pentacene active layer organic thin-film transistors, Adv. Mater. 15, 2009–2011, 2003.

8080_C004.fm Page 297 Tuesday, April 3, 2007 1:54 PM

Near-Edge X-Ray Absorption Fine Structure

297

7. Moon, H. et al., Synthesis, crystal structure, and transistor performance of tetracene derivatives, J. Am. Chem. Soc. 126, 15322–15323, 2004. 8. Janzen, D.E. et al., Preparation and characterization of π-stacking quinodimethane oligothiophenes. Predicting semiconductor behavior and bandwidths from crystal structures and molecular orbital calculations, J. Am. Chem. Soc. 126, 15295–15308, 2004. 9. Dimitrakopoulos, C.D., Brown, A.R., and Pomp, A., Molecular beam deposited thin films of pentacene for organic field effect transistor applications, J. Appl. Phys. 80, 2501–2508, 1996. 10. Mattheus, C.C. et al., Identification of polymorphs of pentacene, Synth. Met. 138, 475–481, 2003. 11. Knipp, D. et al., Pentacene thin film transistors on inorganic dielectrics: Morphology, structural properties, and electronic transport, J. Appl. Phys. 93, 347–355, 2003. 12. Sundar, V.C. et al., Elastomeric transistor stamps: Reversible probing of charge transport in organic crystals, Science 303, 1644–1646, 2004. 13. de Boer, R.W.I. et al., Organic single-crystal field-effect transistors, Phys. Status Solidi A — Appl. Res. 201 (6), 1302–1331, 2004. 14. Goldmann, C. et al., Hole mobility in organic single crystals measured by a “flipcrystal” field-effect technique, J. Appl. Phys. 96, 2080–2086, 2004. 15. Bao, Z., Dodabalapur, A., and Lovinger, A.J., Soluble and processable regioregular poly(3-hexylthiophene) for thin film field-effect transistor applications with high mobility, Appl. Phys. Lett. 69, 4108–4110, 1996. 16. Sirringhaus, H. et al., Two-dimensional charge transport in self-organized, highmobility conjugated polymers, Nature 401, 685–688, 1999. 17. Butko, V.Y. et al., Field-effect transistor on pentacene single crystal, Appl. Phys. Lett. 83, 4773–4775, 2003. 18. Podzorov, V. et al., Single-crystal organic field effect transistors with the hole mobility ~ 8 cm2/V s, Appl. Phys. Lett. 83, 3504–3506, 2003. 19. Jurchescu, O.D., Baas, J., and Palstra, T.T.M., Effect of impurities on the mobility of single crystal pentacene, Appl. Phys. Lett. 84, 3061–3063, 2004. 20. Dyreklev, P. et al., Aligned polymer-chain field-effect transistors, Solid State Commun. 82, 317–320, 1992. 21. Dyreklev, P. et al., Polymeric field-effect transistors using oriented polymers, Synth. Met. 57, 4093–4098, 1993. 22. Amundson, K.R. et al., An in-plane anisotropic organic semiconductor based upon poly(3-hexyl thiophene), Thin Solid Films 414, 143–149, 2002. 23. Heil, H. et al., The influence of mechanical rubbing on the field-effect mobility in polyhexylthiophene, J. Appl. Phys. 93, 1636–1641, 2003. 24. Hamaguchi, M. and Yoshino, K., Rubbing-induced molecular-orientation and polarized electroluminescence in conjugated polymer, Jpn. J. Appl. Phys. Part 2 — Lett. 34, L712–L715, 1995. 25. Grell, M. et al., Blue polarized electroluminescence from a liquid crystalline polyfluorene, Adv. Mater. 11, 671–674, 1999. 26. Whitehead, K.S. et al., Highly polarized blue electroluminescence from homogeneously aligned films of poly(9,9-dioctylfluorene), Appl. Phys. Lett. 76, 2946–2948, 2000. 27. Jung, Y. et al., Surface characteristics of polyfluorene films studied by polarizationdependent NEXAFS spectroscopy, Macromolecules 38, 867–872, 2005. 28. Pattison, L.R. et al., Temperature dependence of the ordering of semiconducting fluorene-thiophene copolymers on rubbed polyimide alignment layers, in Organic photovoltaics VI, ed. Z.H. Kafafi and P.A. Lane, SPIE, Bellingham, WA, 2005, 143–150.

8080_C004.fm Page 298 Tuesday, April 3, 2007 1:54 PM

298

Organic Field-Effect Transistors

29. Stöhr, J. and Jaeger, R., Absorption-edge resonances, core-hole screening, and orientation of chemisorbed molecules: Co, No, and N-2 on Ni(100), Physical Rev. B 26, 4111–4131, 1982. 30. Stöhr, J. et al., Desulfurization and structural transformation of thiophene on the Pt(111) surface, Physical Rev. Lett. 53, 2161–2164, 1984. 31. Stöhr, J., NEXAFS spectroscopy, Springer–Verlag, Berlin, 1992. 32. Horsley, J.A. et al., Resonances in the K-shell excitation-spectra of benzene and pyridine — Gas-phase, solid, and chemisorbed states, J. Chem. Phys. 83, 6099–6107, 1985. 33. Genzer, J., Kramer, E.J., and Fischer, D.A., Accounting for Auger yield energy loss for improved determination of molecular orientation using soft x-ray absorption spectroscopy, J. Appl. Phys. 92, 7070–7079, 2002. 34. Lenhart, J.L. et al., Probing surface and bulk chemistry in resist films using near edge x-ray absorption fine structure, J. Vacuum Sci. Technol. B 20, 2920–2926, 2002. 35. Ade, H. et al., X-ray spectromicroscopy of polymers and tribological surfaces at beamline X1A at the NSLS, J. Electron Spectrosc. Relat. Phenom. 84, 53–71, 1997. 36. Urquhart, S.G. et al., NEXAFS spectromicroscopy of polymers: Overview and quantitative analysis of polyurethane polymers, J. Electron Spectrosc. Relat. Phenom. 100, 119–135, 1999. 37. Jacobsen, C. et al., Soft x-ray spectroscopy from image sequences with sub-100 nm spatial resolution, J. Microsc.-Oxf. 197, 173–184, 2000. 38. Zharnikov, M. et al., An extension of the mean free path approach to x-ray absorption spectroscopy, J. Electron Spectrosc. Relat. Phenom. 124, 15–24, 2002. 39. Harris, M., Appel, G., and Ade, H., Surface morphology of annealed polystyrene and poly(methyl methacrylate) thin film blends and bilayers, Macromolecules 36, 3307–3314, 2003. 40. Samant, M.G. et al., NEXAFS studies on the surface orientation of buffed polyimides, Macromolecules 29, 8334–8342, 1996. 41. Fischer, D.A., Colbert, J., and Gland, J.L., Ultrasoft (C,N,O) x-ray-fluorescence detection — Proportional counters, focusing multilayer mirrors, and scattered-light systematics, Rev. Sci. Instruments 60, 1596–1602, 1989. 42. Hastie, G.P. et al., Examination of the structure and melting behavior of thin film nalkanes using ultrasoft polarized near-edge x-ray absorption spectroscopy, J. Chem. Soc.-Faraday Trans. 92, 783–789, 1996. 43. Wu, W.L. et al., A direct comparison of surface and bulk chain-relaxation in polystyrene, Euro. Phys. J. E 12, 127–132, 2003. 44. Wallace, W.E. et al., Polymer chain relaxation: Surface outpaces bulk, Macromolecules 34, 5081–5082, 2001. 45. Zimmermann, U. et al., NEXAFS and ARUP spectroscopy of an organic single crystal: Alpha-perylene, Mol. Cryst. Liq. Cryst. 339, 231–259, 2000. 46. Scholl, A. et al., Energy calibration and intensity normalization in high-resolution NEXAFS spectroscopy, J. Electron Spectrosc. Relat. Phenom. 129, 1–8, 2003. 47. Stöhr, J. et al., Liquid crystal alignment on carbonaceous surfaces with orientational order, Science 292, 2299–2302, 2001. 48. Galli, G. et al., Structural organizations in polystyrenebased semifluorinated block copolymers for low surface energy coatings, Surface Coatings Int. Part B-Coatings Trans. 87, 77–82, 2004. 49. Xiang, M.L. et al., Surface stability in liquid-crystalline block copolymers with semifluorinated monodendron side groups, Macromolecules 33, 6106–6119, 2000.

8080_C004.fm Page 299 Tuesday, April 3, 2007 1:54 PM

Near-Edge X-Ray Absorption Fine Structure

299

50. Hu, W.S. et al., Molecular orientation of evaporated pentacene films on gold: Alignment effect of self-assembled monolayer, Langmuir 21, 2260–2266, 2005. 51. Hsu, Y.J. et al., Mapping molecular orientation of pentacene on patterned Au surface, J. Electron Spectroscopy Related Phenomena 144, 401–404, 2005. 52. Yoshikawa, G. et al., Molecular orientations and adsorption structures of alphasexithienyl thin films grown on Ag(110) and Ag(111) surfaces, Surf. Sci. 559, 77–84, 2004. 53. Kiguchi, M., Yoshikawa, G., and Saiki, K., Temperature and thickness dependence of molecular orientation of alpha-sexithienyl on Cu(111), J. Appl. Phys. 94, 4866–4870, 2003. 54. Murphy, A.R. et al., Self-assembly, molecular ordering, and charge mobility in solution-processed ultrathin oligothiophene films, Chem. Mater. 17, 6033–6041, 2005. 55. DeLongchamp, D.M. et al., Direct correlation of organic semiconductor film structure to field-effect mobility, Adv. Mater. 17, 2340–2344, 2005. 56. DeLongchamp, D.M. et al., Variations in semiconducting polymer microstructure and hole mobility with spin-coating speed, Chem. Mater. 17, 5610–5612, 2005. 57. Stohr, J. et al., Microscopic origin of liquid crystal alignment on rubbed polymer surfaces, Macromolecules 31, 1942–1946, 1998. 58. Stohr, J. et al., Liquid crystal alignment on carbonaceous surfaces with orientational order, Science 292, 2299–2302, 2001. 59. Hahner, G. et al., Investigation of intermediate steps in the self-assembly of nalkanethiols on gold surfaces by soft-x-ray spectroscopy, Langmuir 9, 1955–1958, 1993. 60. Himmelhaus, M. et al., Adsorption of docosanethiol from solution on polycrystalline silver surfaces: An XPS and NEXAFS study, J. Electron Spectrosc. Relat. Phenom. 92, 139–149, 1998. 61. Hild, R. et al., Formation and characterization of self-assembled monolayers of octadecyltrimethoxysilane on chromium: Application in low-energy electron lithography, Langmuir 14, 342–346, 1998. 62. Fischer, D.A. et al., Mapping surface chemistry and molecular orientation with combinatorial near-edge x-ray absorption fine structure spectroscopy, Macromol. Rapid Commun. 25, 141–149, 2004. 63. Krapchetov, D.A. et al., Solvent-dependent assembly of terphenyl- and quaterphenyldithiol on gold and gallium arsenide, Langmuir 21 (13), 5887–5893, 2005. 64. Fan, L.J., Yang, Y.W., and Tao, Y.T., Molecular orientation and bonding of terthiophene-thiol self-assembled on Au(111): A combined NEXAFS and XPS study, J. Electron Spectrosc. Relat. Phenom. 144, 433–436, 2005. 65. Patnaik, A. et al., Polarized near-edge x-ray-absorption fine structure spectroscopy of C-60-functionalized 11-amino-1-undecane thiol self-assembled monolayer: Molecular orientation and evidence for C-60 aggregation, J. Chem. Phys. 122, 2005.

8080_C004.fm Page 300 Tuesday, April 3, 2007 1:54 PM

8080_C004.fm Page 301 Tuesday, April 3, 2007 1:54 PM

4.3

Scanning Probe Techniques

Hoichang Yang CONTENTS 4.3.1 Introduction................................................................................................301 4.3.2 Atomic Force Microscopy (AFM) ............................................................303 4.3.2.1 AFM Tip Shape...........................................................................304 4.3.2.2 Basic Principle of Intermediate Contact Mode (Tapping Mode)...........................................................................305 4.3.2.3 Applications.................................................................................306 4.3.3 Electric Force Microscopy (EFM) ............................................................310 4.3.3.1 Basic Principle.............................................................................311 4.3.3.2 Applications.................................................................................313 4.3.4 Kelvin Probe Force Microscopy (KFM) ...................................................316 4.3.5 Conducting Probe AFM (CP-AFM or C-AFM) .......................................319 4.3.5.1 Applications of CP-AFM ............................................................321 References..............................................................................................................331

4.3.1 INTRODUCTION Scanning tunneling microscopy (STM) [1], with atomic-scale imaging, and atomic force microscopy (AFM) [2], which was originally introduced for high-resolution surface topography of conducting or nonconducting materials, have been pivotal in opening up many new research directions in past years. Scanning probe microscopy (SPM) includes STM, AFM, and other application modes based on AFM, a few of which are magnetic force microscopy (MFM) [3], electric force microscopy (EFM), Kelvin probe force microscopy (KFM), conducting probe-AFM (CP-AFM), nearfield scanning optical microscopy (NSOM) [4,5], and scanning capacitance microscopy (SCM) [6,7]. Although AFM was initially introduced as a small part of STM, AFM has become an advanced and developed scanning probe technique with broad applications in academic and industrial research. It is also applied for quality control in a number of industries. Because AFM can be described as high-resolution profiling of surfaces with a sharp probe, this technique can be applied for measurements of surface textures of various materials and also for quantitative examination of shapes/profiles of tech-

301

8080_C004.fm Page 302 Tuesday, April 3, 2007 1:54 PM

302

Organic Field-Effect Transistors

(a)

(b)

(c)

2nd layer μm

nm 600

1.5

400

1.0 0.2

0.5

200 (d)

(e)

(f )

0.4

0.6

0.8

μm

2nd layer

Recorded sector

500 nm

5 μm

500 nm

FIGURE 4.3.1 AFM imaging for various materials: (a) Topograph of aggregated poly(3hexyl thiophene) (P3HT, 0.01 g/L) adsorbed onto hydrophilic Si surface inside chloroform/hexane mixture. Topographic images of pentacene thin films vacuum-deposited onto hexamethylene disilazane (HMDS)-treated SiO2/Si substrates: (b) ~1.3 ML; (c) 10 nm. (d) AFM phase images of polystyrene-polybutadiene-polystyrene (SBS) triblock copolymer obtained in light (top) and hard (bottom) tapping. (Note that most polymer films have a wetting layer of several nanometers’ thickness at the polymer/air interface. An optimum scanning force is required to observe micro- or nanoscale structures in the films. (e) Topographic (top) and magnetic force gradient (bottom) images of storage magnetic surface in a commercial zip-drive; (f) AFM topographic (top) and conducting probe current (bottom) images simultaneously obtained from ~1.3 ML pentacene thin film onto HMDS-treated doped Si substrate (sample bias = +1.5 V). (Photograph courtesy of H. Yang.)

nologically important surface structures. In the past decade, many researchers have recognized the importance of AFM as a characterization technique for various materials. However, most SPM techniques are based on surface scanning because knowledge of surface topography is essential for mapping of other (mechanical, electromagnetic, thermal, etc.) surface properties. Figure 4.3.1 demonstrates SPM techniques, based on AFM, for various materials. This chapter intends to explain, beyond the conventional technique of AFM, various SPM techniques, highlighting those used for characterizing organic semiconductors in field-effect transistors. Various organic thin film transistors (OTFTs) containing π-conjugated materials are becoming commercially available in the form of light-emitting diodes and conjugated polymer circuits [8–10]. In OTFT devices, thin films can be fabricated on dielectric substrates from mainly two processes: solution processing and vapor deposition.

8080_C004.fm Page 303 Tuesday, April 3, 2007 1:54 PM

Scanning Probe Techniques

303

Specifically, during vapor deposition, crystalline nucleation and growth of semiconducting organic molecules are significantly affected by physical interactions between molecules and dielectric substrates, resulting in different molecular orientation, crystalline phases, and crystalline grains within the first few monolayers directly in contact with the substrates. Since the performance of OTFT devices is inherently related to local ordered crystalline structure near dielectric layers [11], it is important to achieve greater control of the self-assembly process that occurs when molecules condense from vapor into an ordered crystal. Therefore, SPM techniques based on AFM are one of the most powerful tools to directly characterize crystalline morphologies and grain boundaries of organic semiconductor films fabricated onto dielectric substrates without any limitation.

4.3.2 ATOMIC FORCE MICROSCOPY (AFM) In an atomic force microscope, a sharp tip positioned at the free end of a cantilever interacts with a sample surface. Either the sample or the tip is precisely moved by a piezoelectric actuator, which can precisely adjust the probe-sample vertical separation to keep the tip-sample force at a given set-point level within the scanned area. Then, the sample topography can be evaluated with high precision and atomic-scale resolution in vertical and lateral dimensions, depending on the tip apex [12]. The tip-sample interaction causes quasistatic deflection (contact mode) or a change in the dynamic parameters of the cantilever (oscillatory modes) when it oscillates at (or near) its resonant frequency [13–16]. In this case, the cantilever motion is magnified with the use of a laser beam, which is deflected from the backside of the cantilever to a photodiode detector [17]. In the contact mode, the tip stays in permanent contact with a sample during scanning, which is a major drawback in soft material applications. However, the lateral tip-sample forces can be applied to studies of sample friction. In the oscillatory mode, known as tapping [18] or magnetic-AC (MAC) [19] mode, this problem is practically eliminated, so this mode is most common for AFM studies of organic materials. In contrast to STM [1,20–53], which provides atomic-scale resolution for various materials onto conducting substrates, AFM imaging resolution is a complicated issue. Contact mode topographic images of crystal surfaces can show atomic-scale patterns identical to the crystallographic lattices [54,55]. However, the absence of atom-scale defects in these images implies that the tip contact area is larger than the atomic size [56] because the contact probe scanning does not exclude periodical force variations consistent with the surface lattice when the tip slides across it [57]. In tapping mode (TM) [18], the intermittent tip-sample contact leads to gentler imaging without shearing inherent to the contact mode. Accordingly, tapping mode has significantly broadened AFM application even though its image resolution (~1 nm) is not extremely high. Recently, Klinov and Magonov showed TM-AFM images with molecular resolution using a probe with carbon spikes [58]. At present, tapping mode applications in air are most common and it is desirable to improve resolution by using as sharp a probe as possible.

8080_C004.fm Page 304 Tuesday, April 3, 2007 1:54 PM

304

Organic Field-Effect Transistors

Si3N4

100 nm

10 μm

L

W

∼ 20 nm

45° (a)

100 nm

(b)

10 nm

Diamond

200 nm (c)

(d)

<10 nm (e)

FIGURE 4.3.2 SEM micrographs of AFM tips: (a) Si3N4 tip; (b) etched Si tip; (c) Si tip with carbon spikes (Hi’RES probe, MikroMasch); (d) high aspect ratio tip (Sting tip, MikroMasch) (additional narrow and long extra tip grown by an electron beam deposition technique at the apex of the Si tip); (e) single diamond probe. (Reproduced from Mikromasch. With kind permission.)

4.3.2.1 AFM TIP SHAPE Success of AFM was not possible without the development of microfabricated probes [59–64]. Stiffness of the cantilevers, their resonant frequency, and the geometry of the tip are the most important factors in defining the probe applicability. Si3N4 probes with triangular cantilevers are usually applied for contact mode experiments because their stiffness can be very low (0.01 N/m). However, the tips of Si3N4 probes are not sharp (opening angle ~ 80°) and their apex is typically around 20 nm (Figure 4.3.2a) [65]. The etched Si probes for tapping mode are much sharper (opening angle ~ 30°) and their apex is in the 5 ~ 10 nm range (Figure 4.3.2b). So far, attempts to make probes with the apex close to 1 nm have only been partly successful. Sharp probes with multiple carbon spikes (apex 1 ~ 3 nm; see Figure 4.3.2c), which are grown at the apex of an etched Si tip, allow for imaging with much higher resolution than the regular Si tips (Figure 4.3.2c) [58]. However, the carbon tips are not suitable for routine high-resolution imaging and are only desirable for smooth samples with surface roughness < 20 nm. If sample surfaces have steep architectures with steps of 100 nm to several microns in height, AFM images may include a profile artifact induced by an inherent tip shape. In this case, a sharp tip with high aspect ratio (e.g., sting tip [MikroMasch]) (Figure 4.3.2d) can make visible many objects previously hidden under tip artifacts. There is a definite promise in the use of a single diamond probe with an extremely sharp geometry (the opening angle below 20°) and small apex (5 ~ 10 nm) (Figure 4.3.2e). Such probes can be useful not only for high-resolution imaging of sample topography but also for sensitive local mechanical, thermal, and electric measurement

8080_C004.fm Page 305 Tuesday, April 3, 2007 1:54 PM

Scanning Probe Techniques

305

8nm

4nm

250nm

0

0

0

500 nm (a)

500 nm (b)

500 nm (c)

FIGURE 4.3.3 Tip geometry effects in tapping mode AFM topographic imaging: (a) Pentacene submonolayer deposited on SAM-treated SiO2 substrate held at room temperature. Top and bottom images were obtained using a bad tip with double apex and a normal tip, respectively. The top image shows double shape feature (see white broken circles). (b) Pentacene submonolayer on HMDS-treated SiO2 substrate. Top and bottom images were obtained using a normal tip and a carbon spike tip (Hi’RES tip, MikroMasch), respectively. Usage of the sharp AFM tip can improve the imaging contrast. (c) Apparently 30-nm-thick oligofluorene thiophene film on SAM-treated SiO2 substrate (SAM: octadecyl silane). Top and bottom images were obtained using a normal tip and a high aspect ratio tip (Sting tip DP14/STING, MikroMasch), respectively. Top images illustrates double shape feature due to lying nanoribbons of oligofluorene thiophene derivative with steep height (~200 nm). (All samples were fabricated at deposition rate of 0.3 Å/s, 10–7 torr.) (Photographs courtesy of H. Yang.)

in AFM-related modes. Figure 4.3.3 also shows that general shapes of tips should be understood in order to exclude the profile artifact in AFM images.

4.3.2.2 BASIC PRINCIPLE OF INTERMEDIATE CONTACT MODE (TAPPING MODE) In tapping mode, a tip mounted beneath a cantilever oscillating with a constant resonant frequency scans the sample surface. Figure 4.3.4a illustrates a cantilever oscillating in free air at its resonant frequency. An electric piezo stack vertically excites the cantilever, allowing the tip to bounce up and down. As the cantilever bounces vertically, a laser beam reflected on the top side of the cantilever is deflected in a regular pattern over a photodiode detector, generating a sinusoidal electronic signal. The signal is converted to a root mean square (RMS) amplitude value, which is displayed as an AC voltage. Figure 4.3.3b shows the cantilever oscillating at the sample surface. Although the piezo stack continues to excite the cantilever with the same energy, the tip is deflected due to its encounter with the surface. The reflected laser beam (“return signal”) reveals information about the vertical height of the sample surface and some characteristics of the sample material. These material characteristics can include elasticity (“hardness”) and magnetic and/or electric forces. Many materials have heterogeneous surfaces, which are composed of regions with different chemical structures or regions of the same chemical structures but different molecular packing. Specifically, most semicrystalline polymer thin films

8080_C004.fm Page 306 Tuesday, April 3, 2007 1:54 PM

306

Organic Field-Effect Transistors

Laser beam Return signal Cantilever

(a)

Return signal (dissected)

Laser beam

Sample surface

(b)

FIGURE 4.3.4 Schematics of cantilever oscillation in tapping mode AFM: (a) in free air and (b) on sample surface.

have an amorphous wetting layer with several nanometers’ thickness when they are deposited through solution processing. During an AFM experiment, the height image obtained at low tip-sample force does not reveal any specific contrast and the phase image is featureless (see Figure 4.3.1d). The phase contrast is extremely sensitive to variations of tip-sample forces and the contrast, which reflects mesoscale morphologies in thin films. TM-AFM is a powerful tool to observe surface topography of soft materials such as self-assembled organic molecules, polymers, or biomolecules because these materials require the minimization of force interaction between the AFM tip and the sample.

4.3.2.3 APPLICATIONS In the case of organic semiconductor thin films in OTFT devices, electrical properties are strongly correlated with orientation, packing, and connection of a π-conjugated crystal plane with respect to the dielectric substrates. Accordingly, AFM combined with grazing-incidence x-ray diffraction (GIXD) providing the information of molecular orientation and crystal structure has been performed for semiconductor thin films. When vapor-deposited semiconducting molecules form crystals on organic and/or inorganic dielectric substrates, their nucleation and growth are significantly affected by surface properties of the substrate. Film formation is controlled by weak van der Waals interaction. Through scanning semiconductor layers grown at an initial deposition stage, we can determine growth mechanisms of organic semiconductors on various substrates. Here, we introduce several examples of AFM imaging techniques for organic semiconductor thin films. To determine the effect of dielectric surface energy on crystal growth of pentacene, Ruiz and coworkers observed crystal growth in pentacene films with various thicknesses ranging from the sub-ML regime to 2 ML on different energy surfaces: H-terminated low-energy and oxidized high-energy Si substrates [66]. As seen in Figure 4.3.5, similar full first layer coverage and layer-by-layer formation of dendritic pentacene crystals are observed for oxidized and H-terminated Si surfaces. However, at film surface coverage (θ) = 0.14 ML (calculated from AFM

8080_C004.fm Page 307 Tuesday, April 3, 2007 1:54 PM

Scanning Probe Techniques

307

Substrate

Substrate

1st layer

Substrate

θ = 0.14ML 0 5

(a)

θ = 0.37ML

10 0 μm

5

(b)

1st

θ = 0.85ML

10 0 μm

lay

5

er (c)

θ = 1.09ML

10 0 μm

(d)

5

θ = 1.89ML

10 0 μm

2nd layer (e)

5

10 μm

(a)

1st layer

e rat b st Su

1st layer θ = 0.18ML 0

5

(f ) θ = 0.42ML 5 10 0 μm

(g) θ = 1.02ML 10 0 5 μm

(h) θ = 1.30ML 5 10 0 μm

nd

(i) θ = 1.80ML 2 5 10 0 μm

lay

er (j) 10 μm

(b)

FIGURE 4.3.5 AFM topographic images for pentacene film formation on (a) H-terminated and (b) oxidized substrates with the total film thickness θ ranging from 0.14 to 1.89 ML, as indicated. (From Ruiz, R. et al., Phys. Rev. B 67 (12), 125406, 2003. With kind permission.)

topography), the pentacene island density per unit surface (N) of the H-terminated substrate is 0.007 μm–2, while for the oxidized substrate (θ = 0.18 ML) it is 0.7 μm–2. Thus, a striking difference in nucleation density of pentacene depending on dielectric surface energy is easily revealed through scanning of the film surfaces. In addition, when conducting molecules are vapor deposited on dielectric substrates, the molecule-substrate interaction can be controlled by controlling substrate deposition temperature (TD). As seen in Figure 4.3.6, different levels of diffusion of oligofluorene thiophene derivates [67,68] on the substrates can induce different nucleation and growth behavior in the first conducting layer, which is in direct contact with the dielectric substrate, resulting in different grains and crystal packing. It has been known that during vapor-deposition diffusion-mediated growth of molecules onto the substrates involves mainly four different steps [69,70]. At an initial stage, molecules diffuse on a substrate and, when a critical number of them meet, a stable nucleus is formed. In a second step, adsorbates still nucleate new islands but also start aggregating into existing ones. Then, in the aggregation regime, the incoming molecules aggregate into the existing islands only. Finally, islands coalesce laterally. AFM imaging for sub-ML conducting films provides the information for scaling behavior of the island size and morphology. Ruiz and coworkers demonstrated twodimensional diffusion-mediated growth of pentacene on SiO2 substrate using TMAFM measurements [71]. Also, Pratontep and Brinkmann correlated growth of pentacene on SiO2 substrates with deposition rates [72]. From scaling analysis of the island size distribution in Figure 4.3.7a, the surface coverage (θ) and island density per unit surface (N) for pentacene films with the same nominal thickness of 0.5 nm can be plotted as a function of increasing

8080_C004.fm Page 308 Tuesday, April 3, 2007 1:54 PM

308

Organic Field-Effect Transistors

(a) S S

500 nm

1 μm

1 μm

500 nm

1 μm

1 μm

500 nm

500 nm

1 μm

R (b)

S

S R = hexyl

R

R

(c) S

S R = dodecyl

R

FIGURE 4.3.6 AFM topographic images of oligofluorene thiophene derivatives sub-ML layers deposited onto OTS-treated SiO2 substrates held at 25°C (left), 90°C (middle), and 140°C (right) with a deposition rate of ~0.3 Å/s: (a) FTTF; (b) DHFTTF; and (c) DDFTTF. (Photographs courtesy of H. Yang.)

deposition rate (κ) (Figure 4.3.7b–c). As seen in Figure 4.3.7a, the fractal morphology of domains at low κ suggests that the growth proceeds by diffusion of admolecules in a diffusion limited aggregation regime. In Figure 4.3.7b, the increase in coverage with increasing κ supports re-evaporation of the molecules during deposition, particularly at low κ. As demonstrated by Venables and coworkers, the relationship between N and κ can be determined by the following equation [70]: N ∞κδ exp (βE N )

(4.3.1)

where EN is the activation energy for homogeneous nucleation and β = (kBTs)–1. EN (= Ei + (I + 1)EA – ED) is a function of the activation energy for desorption EA and the free energy difference Ei between i molecules in the island and in the adsorbed state, and ED is the activation energy for surface diffusion of pentacene molecules. Note that i denotes the critical nucleus size where an island of i molecules is the most unstable and will dissociate. Stable islands start to grow from size (I + 1).

8080_C004.fm Page 309 Tuesday, April 3, 2007 1:54 PM

Scanning Probe Techniques

2 μm 15 nm/min

5 μm−1 (a)

0.4

0.1

100 10 1 0.1

0.2 (c)

Form factor

0.6

0.2

1000

1.0 0.8

0.3

0.0 Island density (islands per 10 × 10 μm2)

0.45 nm/min

(b)

0.0 10 1 0.1

Island size (μm2)

0.4

4.5 nm/min Coverage (monolayer)

0.15 nm/min

309

0.01 100 10 1 Deposition rate (nm/min)

FIGURE 4.3.7 (a) AFM surface topographs of 0.5-nm-thick pentacene films grown at various deposition rates on a 200-nm-thick SiO2 substrate at substrate deposition temperature (TD) of 65°C, different κ. AFM topographs reveal three interesting trends as a function of increasing κ for constant nominal thickness and TD: (1) the apparent coverage of θ of the films onto the SiO2 substrates; (2) the morphology of the islands becomes more compact; (3) the number density of islands is found to increase and accordingly the domain size to decrease. (b) The κ dependence on the surface coverage measured by AFM and the form factor (ƒ). (c) The κ dependence of the N and the mean area of the islands (A). (From Pratontep, S. et al., Phys. Rev. B 69, 165201, 2004. With kind permission.)

Through this AFM analysis, the critical nucleus size was estimated to be two molecules (i.e., clusters with size i ≥ 3 will be stable whereas clusters of two molecules will have a larger probability of dissociation) [71]. On SiO2 substrates, N ∝ κδ with δ = 1.16 ± 0.10 for κ varying over two orders of magnitude (from 0.15 nm/min to 45 nm/min) was observed. The “thin film phase” of crystals in vapor-deposited semiconductor thin films shows a substrate-induced structure different from the “bulk phase” [73]. For example, it has been reported that nucleation of the pentacene bulk phase starts beyond a certain critical film thickness, which significantly decreases with an increase in TD: At room temperature it is around 100 ~ 150 nm and can go down to ~30 nm for films grown at ~90°C [74–78]. Also, this critical value depends on substrate properties (like κ). As the film thickness increases, the typical thin film phase of pentacene clearly displays terrace-like morphology in TM-AFM topographic images (see Figure 4.3.8). This terrace-like structure of the thin film phase has been reported abundantly in literature [66,74,79–87]. Specifically, at the edge of the terraces, step height in AFM imaging provides reliable information for molecular tilting with respect to the substrates, due to high resolution of the AFM height profiles (~0.2 Å noise level). AFM has been further developed for uses in various environments (vacuum, gas, liquid) and under controlled temperatures. The practical experience accumulated in its application to different materials indicates that this technique substantially com-

8080_C004.fm Page 310 Tuesday, April 3, 2007 1:54 PM

310

Organic Field-Effect Transistors

(a)

(c)

(b)

1st layer

2nd layer 3rd layer

1 μm

1 μm (e)

(d)

1 μm

1 μm (f )

1 μm

1 μm

FIGURE 4.3.8 TM-AFM tracking topographs for pentacene film formation deposited on HMDS-treated SiO2 substrates at TD = 25°C, κ = 0.3 Å/s: (a) sub-ML (θ = 0.7); (b) 1.5 ML; (c) 2.4 ML; (d) 10 nm; (e) 20 nm; (f) 100 nm. (From Yang, H.C. et al., J. Am. Chem. Soc. 127, 11542–11543, 2005. With kind permission.)

plements other microscopic methods (optical and electronic microscopy) in characterization of surface morphology and provides unique information concerning structures of soft samples (polymers, organic molecules, and biological materials) in ambient and liquid environments. Using TM-AFM with a closed liquid cell, Kiriy et al. observed solvent-induced aggregates of regioregular poly(3-alkyl thiophene)s (P3ATs) in organic solvent mixtures: hexane (for poor solvent) and chloroform (CHCl3, for good solvent) (Figure 4.3.9) [88]. From UV-vis spectroscopy analysis combined with AFM, they suggested that loading of hexane into CHCl3 solutions can induce a helical conformation of the main chain of P3ATs with 12 thiophene rings per each helical turn.

4.3.3 ELECTRIC FORCE MICROSCOPY (EFM) EFM imaging is a technique that measures variation in electric-field gradient above a sample. The sample may be conducting, nonconducting, or mixed. Since the electric-field gradient is also shaped by a sharp surface edge, which can concentrate the field gradient, large differences in surface topology can make it difficult to distinguish electric-field variations. This section describes how to perform EFM imaging and demonstrate EFM applications for various materials.

8080_C004.fm Page 311 Tuesday, April 3, 2007 1:54 PM

Scanning Probe Techniques

311

100 nm

R

S

(a)

R

R

R

S

(c)

R S R

S S

−S

R

S S

S

S

(b)

S

−S

S R

R

R

R −S

R

(e)

S S R

R R

All anti-conformation

R

(d)

50 nm

1 nm

1 nm

FIGURE 4.3.9 TM-AFM images of (a) poly(3-octyl thiophene) (P3OT, 0.001 g/L) adsorbed onto the mica surface and (b) poly(3-hexyl thiophene) (P3HT, 0.01 g/L) adsorbed onto hydrophobic Si surface from CHCl3/hexane mixture (1:5 v/v). (c) The helical conformation of P3ATs molecules. A space-filling model of the P3ATs (R = alkyl side chain) in the helical conformation: side view (d); top view (e). (From Kiriy, N. et al., Nano. Lett. 3, 707–712, 2003. With kind permission.)

4.3.3.1 BASIC PRINCIPLE Two types of EFM are available: electric field gradient imaging and surface potential imaging, known as Kelvin probe force microscopy (KFM). Figure 4.3.10 represents the principle of EFM measurements using an atomic force microscope. The acquisition of EFM data is interleaved line by line into the surface topography imaging in a two-pass measurement. For each scan line, the AFM setup is first used to record the sample topography in intermittent-contact mode (tapping mode) (Figure 4.3.10a). Then, EFM data are acquired in a second scan (interleave mode), where a metalcoated EFM tip is lifted from the sample surface and oscillated at a constant distance z over the surface (Figure 4.3.10b). In this case, the EFM tip is biased at a VEFM voltage with respect to the substrate, and the shift of the cantilever resonant frequency (Δƒ), from its nominal value (ƒo), is recorded [89–91]. This shift reflects the longrange Coulomb force gradients felt by the EFM tip. Assuming a substrate potential VS, ƒo is the cantilever resonant frequency above the substrate plane for VEFM = VS (Figure 4.3.11a) [91]. EFM signals are then of three types. The first type is the tip-substrate capacitive signal Δƒt–s (Figure 4.3.11b); proportional to (VEFM – VS)2, it corresponds to the gradient of the attractive capacitive force between the tip and the substrate, and hence Δƒt–s ≤ 0. The second component is the increase Δƒε (Figure 4.3.11c) of the capacitive force gradient when the tip is moved over the conducting domain in the linear mode: Δƒε also scales as (VEFM – VS)2. Finally, a shift ΔƒQ associated with the charge in the conducting domain (Q) (Figure 4.3.11d) is the third signal. In terms of a point-charge representation, this

8080_C004.fm Page 312 Tuesday, April 3, 2007 1:54 PM

312

Organic Field-Effect Transistors

f0 z VEFM

Electric ﬁeld

Topographic proﬁle (main scan)

Force gradient proﬁle (interleave scan)

(a)

(b)

FIGURE 4.3.10 (a) AFM topography imaging in intermittent-contact mode. (b) EFM measurement in the interleave mode. The tip is oscillated at a constant distance z on the surface. (From Melin, T. et al., Phys. Rev. B 69, 35321, 2004. With kind permission.) f0 + Δft−s

f0 VEFM = VS

Z

VEFM

Z

(a)

(b)

f0 + Δft−s + Δfε

f0 + Δft−s + Δfε + ΔfQ

VEFM

Z

(c)

Z

Q

VEFM

(d)

FIGURE 4.3.11 (a) EFM measurement with the tip biased at the surface potential (VEFM = VS); the cantilever resonant frequency is ƒo. (b) EFM measurement above the sample surface with VEFM ≠ VS; the cantilever frequency shift Δƒt–s is due to the tip-substrate capacitive force gradients. (c) Additional capacitive frequency shift Δƒε when the EFM tip passes over a conducting domain of dielectric constant ε (Δƒε = 0 if VEFM = VS). (d) Additional frequency shift ΔƒQ when an amount of charge Q is located inside the domain. (From Melin, T. et al., Phys. Rev. B 69, 35321, 2004. With kind permission.)

shift would correspond to the interaction between capacitive charges on the tip apex and the stored charges in the conducting domain. This quantity is thus proportional to both (VEFM – VS) and Q and can be made positive (repulsive force gradient) for (VEFM – VS)Q ≥ 0, or negative (attractive force gradient), provided (VEFM – VS)Q ≤ 0. In EFM imaging, the condition Q(VEFM – VS) > 0 leads to bright phase shift, while Q(VEFM – VS) < 0 leads to dark features in the phase images.

8080_C004.fm Page 313 Tuesday, April 3, 2007 1:54 PM

Scanning Probe Techniques

313

4.3.3.2 APPLICATIONS The EFM technique has gained much interest in the last decade to explore the charge or polarizability properties of conducting materials, especially semiconductors, or composite materials containing conducting fillers of potential use in electro-optical applications or floating-gate memory devices [92–98]. This versatile scanning-probe technique can indeed be used to map electric force gradients of nanoparticles embedded in or deposited on thin insulating layers, with a lateral resolution of a few tens of nanometers. Also, under ambient conditions, the EFM charge sensitivity is better than the elementary charge, as observed from the discrete uctuations in EFM charge signals of CdSe nanocrystals [99–101]. Electrostatic behavior of semiconducting materials — specifically, carbon nanotubes — is of fundamental interest for potential applications such as field-emission electron sources [102–105]. Zdrojek and coworkers have reported the electrostatic properties of individually dispersed single- or multiwalled carbon nanotubes through charge injection using an AFM probe (PtIr-coated probe) [102,103]. To locally inject charges into isolated carbon nanotubes, the EFM tip is biased at Vinj with respect to the silicon wafer, its resonant frequency (ƒo) is set to zero, and the tip is gently contacted to the carbon nanotubes with several nanonewtons of contact force for a few seconds to a few minutes. Two-dimensional distribution of charges injected into the isolated materials is then characterized by EFM, in which electric force gradients acting on the tip biased at VEFM shift ƒo of the EFM cantilever. Figure 4.3.12 represents AFM topographic and EFM phase shift images of multiwalled carbon nanotubes (MWCNT) mounted on a thermally grown 200-nm SiO2 substrate [102]. In Figure 4.3.12b, dark contrast (negative frequency shift) of an MWCNT before charge injection accounts for the local increase of the tipsubstrate capacitance when the EFM tip is moved over the MWCNT. After charge injection, the EFM image (Figure 4.3.12c) showed bright contrast (positive frequency shift) of the MWCNT, resulting from repulsive interactions between the charged MWCNT and the capacitive charge at the tip apex. Also, it was found that charges in the locally charged MWCNT were delocalized during EFM imaging, resulting from the conductive nature of the MWCNT and the charging of the MWCNTsubstrate capacitor. The transport properties in vapor-deposited organic semiconductor films strongly depend on orientation and packing of the π-conjugated planes, where molecules form the π–π-orbital overlap between the nearest neighboring molecules. It is now well known that the charge carrier transport in OTFTs strongly depends on the morphology and molecular structure of the first few molecular layers in direct contact with the dielectric substrate [106–109]. Recently, Dinelli and coworkers [107] reported that two organic monolayers are requested to obtain a charge mobility comparable to that of thicker films. Electrostatic principles clearly predict that the majority of charge carriers are confined close to the gate dielectric interface, within the first few organic monolayers [110]. Heim and coworkers [111] reported how locally injected electrons and holes in a single pentacene ML island stay localized or are able to delocalize over the island as a function of the molecular conformation (order or disorder) of these

8080_C004.fm Page 314 Tuesday, April 3, 2007 1:54 PM

314

Organic Field-Effect Transistors

AFM

EFM

100 nm

20 Hz

Charge detection Charge injection

50 nm

10 Hz

0 nm

0 Hz

Vinj

AFM topography

1 μm

VEFM

EFM - before injection

(a)

EFM - after injection

(b)

(c)

FIGURE 4.3.12 Inset (left) schematics of the charge injection with the tip biased at Vinj with respect to the substrate; (right) EFM data acquisition, consisting in recording the values of Δƒ, when biased at VEFM. (a) AFM topography image of an MWCNT with ~30 nm diameter. (b) EFM image (VEFM = –2 V) of the uncharged nanotube. (c) EFM image (VEFM = –2 V) after charge injection (Vinj = –3 V for 2 min). (From Zdrojek, M. et al., Appl. Phys. Lett. 86, 213114, 2005. With kind permission.)

islands. Through vacuum deposition, islands of pentacene were isolated on two different SiO2/Si substrates that varied in surface roughness (0.15 and 0.4 nm). They found that a rougher SiO2 surface (0.4 nm) and higher deposition rate (3.3 × 10–2 Ås–1) yielded more disordered pentacene islands than those obtained on a smooth SiO2 surface (0.15 nm) at 2.5 × 10–3 Ås–1. Local charge injection and EFM experiments were performed with an EFM probe (ƒo ~ 60 kHz and spring constant 1–3 N/m) under a nitrogen atmosphere. Charge injection in isolated pentacene islands on dielectric substrates can also be performed with an EFM probe biased at Vinj. The resulting transfer of charges into the semiconducting material is then characterized by EFM. Figure 4.3.13 shows AFM topographic and EFM images of ordered pentacene monolayer islands. These dark features in the EFM images are due to the capacitance coupling effect and/or to a small residual charge in the islands. As seen in Figure 4.3.13c and d, the charges in the pentacene islands locally injected by a point contact method are delocalized. The charge and capacitive coupling effects can be separated by measuring the relationship between EFM phase shift (ΔΦ) and VEFM [91]. Figure 4.3.14 illustrates ΔΦ in EFM images for pentacene monolayer islands before any charge injection as a function of VEFM. The ΔΦ–VEFM curve varies as VEFM2 and the curve is slightly shifted by a linear component. From a second-order polynomical fit of the ΔΦ–VEFM curve, an effective surface charge density of the ordered pentacene island is estimated as ~200 charges/μm2. Conversely, EFM analysis for disordered pentacene islands shows:

8080_C004.fm Page 315 Tuesday, April 3, 2007 1:54 PM

Scanning Probe Techniques

1.0 μm

1.0 μm

2 4 Distance (μm)

(a)

??

0

2 4 Distance (μm)

(b)

4 2 0

?? ΔΦ (°)

0.0

−0.5

0

1.0 μm

??

0.5

ΔΦ (°)

??

4 2 0

ΔΦ (°)

Height (nm)

1.0 μm

315

0

2 4 Distance (μm)

(c)

0 −4 0

2 4 Distance (μm)

(d)

FIGURE 4.3.13 From top to bottom: two-dimensional images, section profiles along the black line, and three-dimensional images of pentacene islands grown on SiO2 substrate (surface roughness ~ 0.15 nm) at a deposition rate of 2.5 × 10–3 Ås–1. (a) TM-AFM of the islands. (b) EFM images (ΔΦ, VEFM = 4 V, z = 80 nm) of the same islands before charge injection. (c) EFM images (ΔΦ, VEFM = 4 V, z = 80 nm) of the same islands after a local injection (Vinj = +2 V for 4 min) on the central island (injection point marked by the arrow). (d) EFM images (ΔΦ, VEFM = 4 V, z = 80 nm) after a subsequent local injection at Vinj = –2 V for 4 min. (All images are 5 × 5 μm). (From Heim, T. et al., Nano Lett. 4, 2145–2150, 2004. With kind permission.)

Before charge injection, ΔΦ is not always negative as expected when the capacitive force gradient dominates, meaning that a larger residual negative charge stored in the pentacene islands dominates the force gradient (Figure 4.3.15b). To inject a comparable amount of charges in the ordered islands, a higher voltage is required during the injection process (Vinj = ±5 V). The injected holes are not delocalized after the injection (Figure 4.3.15c). To summarize, the EFM technique is a useful tool to study the electronic properties of isolated conducting materials and various semiconductor films on dielectric substrates at an early stage of growth. Using charge injection from the AFM tip, it is also useful to characterize molecules of interest for molecular-scale electronics [112].

8080_C004.fm Page 316 Tuesday, April 3, 2007 1:54 PM

316

Organic Field-Effect Transistors

Phase shift (°)

0.0 −0.2 −0.4 −0.6 −0.8

−4

−2

0

2

4

6

VEFM (V)

FIGURE 4.3.14 Phase shift versus VEFM (ΔΦ – VEFM) measured before injection on the central island shown in Figure 4.3.2b. Tip-substrate distance is z = 80 nm. The line is the best fit of a polynominal law ΔΦ = a(VEFM – Vs) + b(VEFM – Vs)2, with a = 0.043°/V, b = –0.04°/V2, and a surface potential Vs = 0.02. The ratio a/b is used to calculate the residual charge in this pentacene island. (From Heim, T. et al., Nano Lett. 4, 2145–2150, 2004. With kind permission.)

4.3.4 KELVIN PROBE FORCE MICROSCOPY (KFM) Kelvin probe force microscopy (KFM) [113–116] is a well-established technique for measuring the contact potential differences (CPDs) between a reference electrode and a sample; its basic operation is similar to EFM. In KFM, two conductors (sample and tip) are arranged as a parallel plate capacitor with a small spacing. In a simple model, the contact potential between the two materials is VCPD = –(Φ1 – Φ2)/e, where Φ1 and Φ2 are the work functions of the conductors, including changes due to the adsorbed layers. A periodic vibration between the two plates at a frequency ω gives an alternating current (AC) with the same frequency ω when the two plates have different work functions. i(t) = (Vbias + VCPD) ωΔCcosωt

(4.3.2)

The technique relies on detecting the zero point of the AC while the additional bias voltage is applied between the two plates until the electric field between them disappears. Thus, CPD can be measured by VCPD = –Vbias. The KFM method has high sensitivity for the CPD averaged over the whole plate area and does not provide a high-resolution lateral image (~50 nm) of the variation in CPD on the surface at the submicron scale. Ongoing efforts to improve performance in OTFTs have focused on identifying the relationships among the device structure, film morphology, and charge transport properties. Specifically, depending on the device geometry (top contact [TC] or bottom contact [BC]), contact resistances at the source and drain electrodes can significantly affect current flow in OTFT devices [117–119]. Therefore, a powerful approach to characterizing these contact resistances is to image the potential distribution along the conducting channel during device operation. For example, at given

8080_C004.fm Page 317 Tuesday, April 3, 2007 1:54 PM

Scanning Probe Techniques

317

200 nm 210 nm

??

0 0.0

0.5 Distance (μm)

(a)

1.0

0

−1 0.0

??

VEFM = −5V

1

ΔΦ (°)

2

?? ΔΦ (°)

Height (nm)

210 nm

VEFM = +5V

0.5 Distance (μm)

(b)

1.0

0 −3 −6 0.0

0.5 Distance (μm)

1.0

(c)

FIGURE 4.3.15 From top to bottom: two-dimensional images, section profiles along the black line, and three-dimensional images of pentacene islands on rough SiO2 substrate. (a) TM-AFM images (1 × 1 μm). The upper half of the image was taken at VEFM = 5 V (z = 50 nm), the lower part at VEFM = –5 V (z = 50 nm). (c) EFM images (phase shift, VEFM = 5 V, z = 20 nm) of the same island after a local injection (Vinj = 5 V for 30 s) at the point marked by the arrow. (From Heim, T. et al., Nano Lett. 4, 2145–2150, 2004. With kind permission.)

values of VD and VG, a sharp drop in potential along the channel indicates a high resistance of charge carrier transport. Several groups have demonstrated KFM for imaging potentials in bottom-contact polythiophene- [120–122] and pentacene-based devices [123]. Figure 4.3.16 shows a schematic diagram of KFM measurement for a BC device containing a vapor-deposited pentacene film [124]. Puntambekar and coworkers [124] have also studied the potential drop between source and drain in TC and BC pentacene devices using KFM. In KFM, the surface potential was measured using a nulling technique with the help of a feedback circuit. Since the surface potential contrast strongly depends on the atmospheric conditions — particularly, the presence of an adsorbed water layer on the topmost film — KFM was performed under nitrogen. Figure 4.3.17 shows AFM topographic and surface

8080_C004.fm Page 318 Tuesday, April 3, 2007 1:54 PM

318

Organic Field-Effect Transistors

Signal access module + Signal ampliﬁer (+/− 50V)

Lock in + Feedback

VDC Tip

(2) Surface potential (interleave mode)

F(x)

VAC

(1) Topography Source

Pentacene thin ﬁlm

Drain

Gate

VD

Insulator

VG

FIGURE 4.3.16 Schematic of the KFM setup for imaging potentials in a “bottom contact” (BC) pentacene OTFT. Topographic and potential images are acquired simultaneously. After each topographic line scan (1), the tip is lifted off the surface (~10 nm) and the surface topography is retracted while recording the surface potential (2). (From Puntambekar, K.P. et al., Appl. Phys. Lett. 83, 5539–5541, 2003. With kind permission.) Bottom contact device

S

D

0V

S

2 μm

D

−10V (b)

(a) Top contact device

S

D

3 μm

S

0V

D

−5V

Topograph

Surface potential

(c)

(d)

FIGURE 4.3.17 Topographic (a), (c) and corresponding surface potential images (b), (d) for BC- and TC-OTFTs. VD = 10 V for (b), VD = –5 V for (d), and VG = 0 for both cases. Labels S and D indicate the edge between the contacts and the conducting channel. For the BCOTFT, length (l)/width (w) = 10 μm/100 μm. For the TC-OTFT, the l/w = 16 μm/300μm. (From Puntambekar, K.P. et al., Appl. Phys. Lett. 83, 5539–5541, 2003. With kind permission.)

8080_C004.fm Page 319 Tuesday, April 3, 2007 1:54 PM

Scanning Probe Techniques

319

potential images of BC- and TC-OTFTs with ~30 nm thick pentacene films grown onto SiO2 substrates. Figure 4.3.17a shows a remarkable change in the morphology in the BC device between the pentacene thin film grown in the channel and the pentacene thin film growing on the electrodes, which has been observed by others [119,125]. This discontinuity in the morphology led to substantial potential drops near both the source and drain contacts. In the potential image, almost half of the drain potential (VD = 10 V) was dropped near the contact, suggesting that the presence of the discontinuous pentacene morphology at the source and drain plays a crucial role in charge transport. Conversely, for the TC device, the AFM topograph showed uniform pentacene morphology with no discontinuous growth near the electrodes, which were fabricated onto vapor-deposited pentacene films. As a result, the surface potential image in Figure 4.3.17d showed a uniform potential distribution along the channel with a small drop at the contacts and no sharp drop within the channel.

4.3.5 CONDUCTING PROBE AFM (CP-AFM OR C-AFM) This section describes a variant of SPM where conducing probes are used to measure the current-voltage (I–V) relationship and resistance (conductance) of conducting materials. “Conducting probe AFM” (CP-AFM) is also called conducting AFM, current-sensing AFM [11,126], or scanning resistance microscopy [127,128]. The aspects of CP-AFM that are most attractive for nanoscale electrical transport measurements are as follows [129]: the ability to image samples with high resolution before, during, or after I–V measurement the ability to collect I–V relationships on samples that are highly resistant or surrounded by insulating regions straightforward interpretation of the tip position relative to the sample during a point-contact measurement (i.e., a measured repulsive force indicates intimate tip-sample contact) These characteristics make CP-AFM ideal for studying electrical transport of nanotubes, nanoparticle assemblies, micro- or nanofabricated semiconductor devices, and individual molecules. Detailed appraisal of these characterizations can be obtained by comparing CP-AFM and STM. Although CP-AFM and STM share high spatial resolution imaging capability (STM ~ 0.1 nm; CP-AFM ~ 10 nm, due to larger tip apex) that is critical in linking nanoscale structure to transport properties, an important distinction is the position of the tip with respect to the sample. In the case of CP-AFM, a metal-coated tip is directly contacted to the sample under a controlled load. This means that the measured I–V relationship is mainly affected by the electrical properties of the tip-sample contact. In contrast, mechanically formed or electrochemically etched Pt/Ir-, or W-tip, as STM tips do not physically contact the sample. Accordingly, the I–V characteristic across a tunneling gap is determined by electronic structure of the sample, instead of tip-sample contact properties. Although the reliance of STM on tunneling allows

8080_C004.fm Page 320 Tuesday, April 3, 2007 1:54 PM

320

Organic Field-Effect Transistors

Au coated tip

A Au electrode

Conducting material

SiO2

FIGURE 4.3.18 Scheme of typical point contact CP-AFM experiment in which an Au-coated AFM tip is used to probe the resistance of conducting materials contacted to an Au electrode.

for powerful conductance spectroscopy that maps the electronic density in a material, the length scale over which one can probe transport in resistive materials is limited to typical tunneling distances (1 ~ 10 nm). In addition, since STM requires detection of a tunneling current in order to position the tip in the sample, STM is not practical for characterizing small structures surrounded by insulating regions, such as OTFT devices containing semiconducting layers onto gate dielectric substrates. CP-AFM conductivity measurements can be carried out independently of force feedback, which can be varied in investigation of material conductance under a controlled stress [130–137]. As a result, electrical transport measurement can be facilitated over longer distances on samples with widely varying resistances. In addition, a repulsive force in CP-AFM during the tip approach makes it easier to detect the precise location of the tip relative to the sample. For electrical characterization, the relative merit of CP-AFM and STM certainly depends on specific goals and experimental constraints. Initially, CP-AFM studies showed the ability to make two basic kinds of electrical measurements: stationary point contact and two-dimensional resistance maps. Figure 4.3.18 shows a point-contact CP-AFM measurement in which resistance (or I as a function of V) is measured between the stationary probe and a fixed contact. The current sensitivity of CP-AFM is ~1 pA with a noise level of <0.5 pA. An important application of point-contact CP-AFM is “spreading resistance profiling” (SRP) of dopant concentration in a material [138,139]. Recent CP-AFM allows determination of dopant concentration profiles with much higher resolution than is possible with conventional point probes commonly used in the semiconductor industry. CP-AFM has also been used to measure conductance of individual semiconductor nanoparticles [140,141], Langmuir–Blodgett films [142], adsorbed molecules on graphite, and thin molecular crystals on Au. The second type of measurement, “resistance mapping,” involves simultaneous topographic imaging and resistance measurements [143]. The key factor to all CP-AFM measurements is contact resistance of the conducting probes. Reported conducting probes include heavily doped Si tips and conventional Si or Si3N4 tips coated with metal (e.g., Ag, Au, Pt, NbN) or B-doped diamond films. Thomson and Moreland provided a detailed investigation of Ag-, Au-, and Pt-coated tips and doped Si tips and achieved five orders of magnitude lower contact resistance to Au surfaces with metal-coated probes than with doped

8080_C004.fm Page 321 Tuesday, April 3, 2007 1:54 PM

Scanning Probe Techniques

321

Si probes [144]. It is also known that scanning the tip across a surface in contact mode quickly wears away the metal coating on the apex of the tip. Beale and Pease [145] also reported that metal-coated AFM tips used to make low-resistance electrical contacts on a gold surface suffered from the metal separating from the tip after surface contact. Of course, the wearing away of the metal-coated tips could be overcome by doing the initial imaging of the surface in tapping mode, which nearly eliminates the lateral forces on the tip [18].

4.3.5.1 APPLICATIONS

OF

CP-AFM

Theoretical work function on electrical transport (tunneling) in metal–molecule–metal junctions indicates that junction impedance is significantly affected by the properties of the metal–molecule contacts. Specifically, the presence of barriers to electron (or hole) injection leads to drops in electrostatic potential at the metal–molecule interface, resulting in contact impedances. Various methods are currently available for probing the electrical conductance of discrete molecules or clusters of molecules bridged between metal or semiconductor electrodes to understand how the structure and electronic properties of molecules and their associated contacts affect the current-voltage (I–V) characteristics observed for the junction [133,136,146–161]. The importance of metal–molecule or semiconductor–molecule interfaces in determining junction I–V characteristics is well recognized [162–165]. Frisbie and coworkers have characterized contact resistances in molecular junctions based on n-alkanethiol, n-alkanedithiol, and alkyl isonitrile molecules (with various alkyl chain lengths) self-assembled on Au, Ag, and Pt electrodes using CP-AFM [135,136,154,166–168]. Figure 4.3.19a shows a representation of a CP-AFM approach to junction formation in which metal-coated AFM tips (Figure 4.3.19b) were used to contact a self-assembled monolayer (SAM) on a metal substrate. In CP-AFM, contact to the monolayer is controlled by feedback electronics capable of maintaining a set-point load with several nanonewtons’ precision, and I–V characteristics were acquired by sweeping the voltage applied to the tip. For low voltage (±0.3 V), voltages were applied to the tip with a Keithley 236 electrometer operated in DC mode; for high voltage (±1.5 V), the electrometer was operated in sweep mode. Figure 4.3.20 shows that I–V traces of representative Au-S-(CH2)n-CH3/Au and Au-S-(CH2)n-S-Au junctions behave sigmoidally according to the following Simmons Equation for tunneling through a square barrier:

⎧⎪⎛ ⎛ 2s 2m qV ⎞ ⎨⎜ φ − ⎟ ⋅ exp ⎜⎜− 2 ⎠ ⎪⎩⎝ ⎝

where

φ−

I=

qA 4 π 2 s 2

qV 2

⎞ ⎛ ⎛ 2s 2m qV ⎞ ⎟⎟ − ⎜⎝ φ + 2 ⎟⎠ ⋅ exp ⎜⎜− ⎠ ⎝

φ+

qV 2

⎞⎪⎫ ⎟⎟⎬ ⎠⎪⎭ (4.3.3)

8080_C004.fm Page 322 Tuesday, April 3, 2007 1:54 PM

322

Organic Field-Effect Transistors

Tip Y

Y

Y

Y 200nm VTip

(CH2)n (CH2)n (CH2)n (CH2)n X

X

X

X

A 200nm (b)

Substrate Tip, substrate = Au, Ag, Pd, Pt X = S, N C Y = S, CH3 (a)

FIGURE 4.3.19 (a) Schematic representation of a molecular tunneling junction formed using CP-AFM. (b) SEM images of the Ag and Au tips. A metal-coated (Au, Ag, or Pt) AFM tip is brought into contact with a SAM of n-alkanethiols, n-alkanedithiols, or alkyl isonitriles of various lengths on an Au-, Ag-, or Pt-coated Si substrate. (From Engelkes, V.B. et al., J. Am. Chem. Soc. 126, 14287–14296, 2004. With kind permission.)

A is the junction area s is the width of the barrier is the reduced Planck constant m is the electron mass V is the bias applied across the molecule(s) φ is the barrier height for tunneling through the LUMO level (φ = [EF – ELUMO]) or through the HOMO level (φ = [EHOMO – EF]) (EF, Fermi level of the electrode) At low bias, Equation 4.3.3 can be used to determine the resistance (R) of the linear regime as follows:

R=

⎛ 2 2 mφs ⎞ 4 π22 s ⎟⎟ exp ⎜⎜ q A 2 mφ ⎠ ⎝ 2

(4.3.4)

Since the general exponential behavior of Equation 4.3.4 is dominant in the measured I–V traces, Equation 4.3.4 can be simplified as follows: R = R0 exp (βn )

(4.3.5)

where R0 is the effective contact resistance, n is the number of repeat units (in this case carbon atoms), and β is the structure-dependent attenuation factor (or the decay factor for alkane system) [135,137,148,152,153,169–172]:

8080_C004.fm Page 323 Tuesday, April 3, 2007 1:54 PM

Scanning Probe Techniques

323

C6

10−7

C8

Current (A)

10−8

C10

10−9 10−10 10−11

20

10−12

0

C6 C8

10−13

C6 C8 C10 A (nm2) 114 56 64 s (Å) 8.6 8.8 10.1 φ (eV) 1.6 2.1 2.2

C10 −4

−20 × 10

10−14 −1.5

−1.5 −1.0 −0.5 0.0 0.5 1.0 1.5

−1.0

−0.5

0.0

0.5

1.0

1.5

Tip bias (V) (a) C6

10−6

C8

Current (A)

10−7

C9

10−8

C10

10−9 300 C6 200 C8 C9 100 0 C10 10−11 −100 −200 10−12 −300 × 10−4 −1.5 −1.0 −0.5 0.0 0.5 1.0 1.5

10−10

10−13 −1.5

−1.0

−0.5

C6 C8 C9 C10 A (nm2) 1500 730 610 230 s (Å) 8.2 7.8 8.5 8.2 φ (eV) 1.9 2.5 2.3 2.7 0.0

0.5

1.0

1.5

Tip bias (V) (b)

FIGURE 4.3.20 Representative semilog plot of I–V traces of C6, C8, and C10 Au-alkanethiolAu (a) and Au-alkanedithiol-Au (b) junctions. (Absolute values of current are displayed.) (From Engelkes, V.B. et al., J. Am. Chem. Soc. 126, 14287–14296, 2004. With kind permission.)

β=2

2 mφ 2

(4.3.6)

According to Equation 4.3.5, measured low-bias resistances are plotted versus the number of carbon atoms in the molecular chain on a semilog axis (see Figure 4.3.21). Also, the fit parameters R0 (in the zero length intercepts in the plot) and β can be extracted experimentally from R (dV/dI⏐V=0) as a function of the number of CH2 groups in the SAM. As seen in Figure 4.3.21, the junction resistance increases exponentially with molecular chain length. Figure 4.3.22 of R0 versus the electrode

8080_C004.fm Page 324 Tuesday, April 3, 2007 1:54 PM

324

Organic Field-Effect Transistors

Ag/S-(CH2)n-CH3/Ag Au/S-(CH2)n-CH3/Au Pt/S-(CH2)n-CH3/Pt

1012

Resistance (Ω)

1010 108

R0

106 β

104 102

0

2

4

6

8

10

12

Number of carbons (a) Ag/S-(CH2)n-S/Ag Au/S-(CH2)n-S/Au Pt/S-(CH2)n-S/Pt

109

Resistance (Ω)

10

8

107 106 105 104 103 102

0

2

4 6 Number of carbons

8

10

(b)

FIGURE 4.3.21 Resistance versus chain length plots for n-alkanethiol (a) and n-alkanedithiol (b) junctions. Data points are average resistance values of about 10 experiments performed with different tips. Lines are not least-squares fits to the data but are defined by the average R0 and β values for the same set of tips. (From Engelkes, V.B. et al., J. Am. Chem. Soc. 126, 14287–14296, 2004. With kind permission.)

metal work function generated from the intercepts of Figure 4.3.21 confirms that R0 decreases with increasing electrode work function (W) and that the junction with two chemisorbed contacts (n-alkanedithiols) has a lower contact resistance than those where one contact is physisorbed (n-alkanethiols). Further, spatially resolved methods of CP-AFM can allow direct measurement of the resistance or conductivity of individual grains and grain boundaries in organic thin film transistors. Dai et al. measured resistivity of individual carbon nanotubes contacted by Au electrodes fabricated on gate dielectric SiO2 by a lithography procedure [143]. With CP-AFM, it is possible to contact electrically and measure the axial conduction through a single carbon nanotube while simultaneously recording surface topology. Conductive tips were fabricated by depositing NbN onto

8080_C004.fm Page 325 Tuesday, April 3, 2007 1:54 PM

Scanning Probe Techniques

325

107

Alkanethiols Alkanedithiols

105 104

100 4.0

4.5

5.0 Work function (eV)

Pt

Au/Pt

Au

101

Au/Ag

102

Ag/Pt

103

Ag

Contact resistance (Ω)

106

5.5

6.0

FIGURE 4.3.22 Contact resistance (R0) as a function of electrode metal work function (W) for junctions composed of n-alkanethiols (ò) and n-alkanedithiols (ô). The solid lines are guides for the eye. (From Engelkes, V.B. et al., J. Am. Chem. Soc. 126, 14287–14296, 2004. With kind permission.)

commercial Si3N4 cantilevers (superconducting NbN provides good conductivity and hardness). Resistance (RT) obtained from CP-AFM measurements can be expressed at RT = RC + RSx, where RC corresponds to the contact resistance (the sum of Aunanotube and nanotube-tip contact resistance), RS is the linear resistance of the nanotube, and x is the position along the nanotube. From CP-AFM measurements for individual nanotubes, it was found that the resistance along the nanotube axis increased linearly with distance from the Au contact. Specifically, a linear resistance (RS) of a 13.9-nm diameter nanotube is 0.06 MΩ/μm (7.8 ± 1.0 Ω⋅m calculated from the annular cross-sectional area of the sample as measured by TEM), while around the bend at the end of this nanotube the value of RS increased by more than an order of magnitude. This result supports the conclusion that defects can play a key role in the electrical properties of semiconducting materials. Since commercially available conducting probe tips have relatively thick conducting layers (>30 nm), the large apex of these tips yields poor resolution or double shapes in AFM topographic images. Through CP-AFM measurement, Frisbie and coworkers [129,173–176] have reported resistance in organic semiconductor sexithiophene crystals (6T, Egap ~ 2.3 eV) as an extensively characterized thiophene oligomer [175–180]. Thin crystallites of 6T were grown onto the SiO2/Si or the substrates with premounted source/drain gold electrodes (Cr/Au, ~15/150 Å) by vacuum sublimation at 10–4 torr. As seen in Figure 4.3.23, a variable channel length transistor could be constructed using a microfabricated Au electrode (drain) contacting one 6T grain, an Au-coated AFM tip as a positionable electrode (source), and a highly doped Si substrate as the gate. Figure 4.3.24 shows typical characteristics of Au-coated AFM probe used for CPAFM measurement. Figure 4.3.25(a) shows an AFM topograph of a single-crystal grain of 6T (6.9 nm or 3 ML thick) in contact with an Au electrode (25 nm tall) on SiO2 before

8080_C004.fm Page 326 Tuesday, April 3, 2007 1:54 PM

326

Organic Field-Effect Transistors

A. Sexithiophene (6T) S

S S

C. CP-AFM of 6T Au-coated tip

S S

S

2.3 nm

6T crystal

B. 6T unit cell Au’S or SiO2

monoclinic a = 44.708 Å b = 7.851 Å c = 6.029 Å β = 90.76 deg.

Au coated probe A

6T crystal Au wire

2.3 nm

Vprobe

Insulator (SiO2) Si

FIGURE 4.3.23 Scheme of the CP-AFM experiment in which an Au-coated AFM tip (A) is used to make electrical contact to crystal layers of sexithiophene (6T) grown by vapor deposition on SiO2/Si. (B) A microfabricated Au wire serves as a drain electrode. (C) Highly doped Si substrate is a gate. (From Kelley, T.W. and Frisbie, C.D., J. Phys. Chem. B 105, 4538–4540, 2001. With kind permission.)

recording a point contact I–V measurement. I–V traces were taken at different positions on the grain (approximately along the [011] direction), using ~100-nN applied load on the probe-6T contact. As seen in Figure 4.3.25b, the point contact I–V traces at each position are rectifying. Namely, larger currents are obtained for positive probe bias, suggesting that 6T is a hole transporter and the probe is a better hole injector than the Au wire, possibly due to the high field (large voltage drop) associated with the probe point contact. At fixed probe voltage, I decreases monotonically with increasing probe-electrode distance, reflecting increasing resistance due to the grain. As shown in Figure 4.3.26, plots of the differential resistance (dV/dI) versus probe-electrode distance under different probe voltages are linear (i.e., showing a constant grain resistance per unit length). Since the contact resistance (resistance at zero distance obtained by extrapolation of the linear fits in Figure 4.3.26 is comparable to the grain resistance), the shapes of the I–V curves in Figure 4.3.25(b) result from convolution of the I–V characteristics of the contacts and the ohmic response of the single grain crystal. As a result, the CP-AFM measurement allows us to investigate transport properties of the single grain, even though the I–V characteristics partially reflect contact properties between the probe and the sample. The recent improvement in charge mobility in vapor-deposited films of organic semiconductors gives rise to the importance of film morphology, specifically the first few molecular layers directly in contact with the dielectric surface. Accordingly, grain boundaries (GBs) are believed to be bottlenecks to transport in OTFT devices. To better understand intra- versus intergrain conductivity and electrical transport in organic thin films, Frisbie and coworkers demonstrated carrier transport though intraand inter-GBs of 6T single crystals using CP-AFM measurements. Grains of 6T were grown onto the SiO2/Si substrates with premounted source/drain electrodes (Cr/Au ~ 15/150 Å) by vacuum sublimation at 10–4 torr. In this case, two electrodes

8080_C004.fm Page 327 Tuesday, April 3, 2007 1:54 PM

Scanning Probe Techniques

327

A

B

20 μm 100

Current (nA)

50

0

Tip on graphite 50 nN 100 nN 150 nN 200 nN 250 nN Conductance

C

500 nm

−50

80

μΩ−1 vs. nN

40 0 0

−100 −10

−5

0 Voltage (mV)

100 200 Load

5

300

10

FIGURE 4.3.24 (A and B) SEM images of an Au-coated CP-AFM tip. (C) I–V characteristics of Au tips in contact with graphite as a function of compressive load. The inset shows conductance (I/V) versus load. The conductance of the tips is 2 × 10–6 Ω–1 at 50 nm applied load and increases linearly until 250 nN. The increase in conductance with increasing load is expected since the tip-graphite contact area will also increase with load. The large increase in conductance upon application of a 250-nN load is likely the result of plastic deformation of the Au coating, which significantly increases the tip-graphite contact area. (From Loiacono, M.J. et al., J. Phys. Chem. B. 102, 1679–1688, 1998. With kind permission.)

with a gap size of 1.5 ~ 2.0 μm acted as nucleation sites for 6T. Devices were images in air using a Digital Instruments Multimode AFM. If GB formation on a chip was not complete, the chip was placed back into the sublimator for further 6T deposition. The 6T deposition/AFM imaging process was continued until gain boundary formation was complete (see Figure 4.3.27). Finally, using CP-AFM, I–V measurements are performed for the samples containing a pair of grain boundaries (GBs) in air as a function of the probe position and gate field by applying negative voltages to the microfabricated electrode. Figure 4.3.28a is an AFM topograph showing two 6T grains on SiO2/Si sharing a common boundary approximately 1 μm in length. Figure 4.3.28b shows ID–IV traces as a function of VG obtained when the conducting probe as a source electrode was positioned at the “x” shown in Figure 4.3.28a: ID increases as VG becomes more negative, consistent with hole conduction.

8080_C004.fm Page 328 Tuesday, April 3, 2007 1:54 PM

328

Organic Field-Effect Transistors

3

6T

2 (011)

Current (nA)

4

1

105° Au (011) wire SiO2

8 6

1

0.1

4

0.5 1.0 1.5 2.0 2.5

2

V

0

500 nm

10

10 log l

(b)

(a)

−2

1 −1 0 Probe voltage (V)

2

FIGURE 4.3.25 (a) AFM topograph of a 6T grain contacted by a nanofabricated Au wire on SiO2. The grain is ~2 μm length and width and 6.9 nm tall (3 ML). The [011] and crystallographic directions labeled in the figure correspond to the directions of the surface lattice vectors determined in high-resolution AFM images. Point-contact I–V measurements were recorded at each of the numbered positions. (b) I–V traces obtained at each of the numbered positions in (a). Current (at fixed probe voltage) decreases with increasing probewire separation. The inset shows a semilog plot of the data for point 1, where current appears to have an exponential dependence on voltage (at low voltage) revealing a barrier to charge injection of ~0.5 eV. (From Kelley, T.W. and Frisbie, C.D., J. Vac. Sci. Technol., B 18, 632–635, 2000. With kind permission.)

600

dv/dl (MΩ)

500 400 300 200 Probe voltages: 1.0 V 1.8 V 1.4 V 2.2 V

100 0

0

400 800 Distance (nm)

1200

FIGURE 4.3.26 Differential resistance (dV/dI) versus probe-electrode separation distance for different applied probe voltages, based on the data in Figure 4.3.26(b). (From Kelley, T.W. and Frisbie, C.D., J. Vac. Sci. Technol., B 18, 632–635, 2000. With kind permission.)

8080_C004.fm Page 329 Tuesday, April 3, 2007 1:54 PM

Scanning Probe Techniques

329

a

b

c

d

e

f

500 nm

FIGURE 4.3.27 Sequence of AFM topographs tracking the formation of a single GB between 6T grains. The grain boundary in (f) is approximately 2 μm long. (From Chwang, A.B. and Frisbie, C.D., J. Appl. Phys. 90, 1342–1349, 2001. With kind permission.)

0 −5

(011)

(011)

(011)

ID (nA)

g.b.

−10 −15

(011) −20

0 Vg −2 Vg −4 Vg −6 Vg

−8 Vg

G (nS)

Electrode (drain)

2 3.0 μ = 0.01 cm /Vs VT = +0.9V 2.0

1.0 0.0

0

−10 Vg

500 nm (a)

−10

−8

−6 −4 VD (V)

2

4 6 8 10 −VG (V)

−2

0

(b)

FIGURE 4.3.28 (a) AFM topographic image of a pair of 6T grains sharing a 1-μm grain boundary. The GB and crystallographic orientations of the grains are labeled: there is 10° misalignment between the directions shown in the figure. The grain in the upper half of the figure is contacted by a 250 nm wide Au wire, which serves as the drain electrode. Terraces corresponding to single monolayers of 6T are visible. (b) The ID–VD characteristics as a function of gate voltage (VG) obtained when the AFM prove was positioned at the “x” in (a). The inset shows the conductance at VD = –3 V as a function of VG. (From Kelley, T.W. and Frisbie, C.D., J. Phys. Chem. B 105, 4538–4540, 2001. With kind permission.)

8080_C004.fm Page 330 Tuesday, April 3, 2007 1:54 PM

330

Organic Field-Effect Transistors

5 6

8 2 4

Resistance (GΩ)

g.b.

1

3

5 6 7

(a)

−4 VG −6 VG

6

−8 VG 4 −10 VG 2

600 nm

7

1 0

2 1000

3

4

2000 3000 Distance (nm)

4000

(b)

FIGURE 4.3.29 (a) TM-AFM topographic image of the GB regime. Au-coated probe was positioned at point 1–7. (b) Resistance as a function of gate voltage (VG) and probe position. The resistance is actually the differential resistance (dVD/dID) obtained as a function of VG in the linear regime of the ID–VD characteristic at each probe position. (From Kelley, T.W. and Frisbie, C.D., J. Phys. Chem. B 105, 4538–4540, 2001. With kind permission.)

Figure 4.3.28 demonstrates that the gate-modulated conductance of the crystals can be recorded with CP-AFM methodology. Another key result is shown in Figure 4.3.29. After the conducting probe was placed at points 1–7 labeled in Figure 4.3.29a, the ID–VD characteristics were recorded at each point. From the linear portion of the ID–VD traces, the differential resistances (dVD/dID) were determined as a function of VG. The GB resistance was taken to be the difference in resistance between points 4 and 5, which straddle the GB, and the resistance between points 1–4 might be an intrinsic resistance across the 6T crystal grain. The GB resistance was found to be gate voltage (VG) dependent and large, on the order of 109–1010 Ω for a 1-μm boundary length (Figure 4.3.29b). Resistances across single 6T grains were an order of magnitude lower. The results indicate that GBs can be the potential energy barrier to charge transport in polycrystalline organic semiconductor films, particularly at low gate fields. The GB barrier height was estimated to be on the order of 100 meV. Recently, Yang and coworkers have also demonstrated the effects of GBs in the first pentacene layer on charge carrier transport in OTFT devices using conducting AFM current imaging, showing localized potential barriers near GBs or less-ordered crystals (Figure 4.3.30) [11]. These few examples of ability to probe transport on local length scales by CPAFM show a broadly applicable approach to quantifying the effects of microstructure and contacts on the performance of organic thin film transistors.

8080_C004.fm Page 331 Tuesday, April 3, 2007 1:54 PM

Scanning Probe Techniques

331 −2000 pA

(a) 2nd layer 1st layer

0

−2000 pA

(b)

2nd layer 0

FIGURE 4.3.30 Simultaneously recorded contact AFM topographic (left) and C-AFM current (right) images for ~1.5-ML pentacene thin films on (a) HMDS and (b) OTS-treated Si substrates. (From Yang, H.C. et al., J. Am. Chem. Soc. 127, 11542–11543, 2005. With kind permission.)

REFERENCES 1. Binnig, G. et al., Tunneling through a controllable vacuum gap, Appl. Phys. Lett. 40, 178–180, 1982. 2. Binnig, G., Quate, C.F., and Gerber, C., Atomic force microscope, Phys. Rev. Lett. 56, 930–933, 1986. 3. Koblischka, M.R. and Hartmann, U., Recent advances in magnetic force microscopy, Ultramicroscopy 97, 103–112, 2003. 4. Guest, J.R. et al., Near-field coherent spectroscopy and microscopy of a quantum dot system, Science 293, 2224–2227, 2001. 5. Muramatsu, H., Chiba, N., and Fujihira, M., Frictional imaging in a scanning nearfield optical/atomic-force microscope by a thin step etched optical fiber probe, Appl. Phys. Lett. 71, 2061–2063, 1997. 6. Huang, Y. and Williams, C.C., Capacitance-voltage measurement and modeling on a nanometer-scale by scanning C-V microscopy, J. Vac. Sci. Technol., B 12, 369–372, 1994. 7. Zimmermann, G. et al., Application of SCM for the microcharacterization of semiconductor devices, Appl. Phys. A: Mater. Sci. Process. 76, 885–888, 2003. 8. Wang, J.Z. et al., Polymer field effect transistors fabricated by dewetting, Synth. Met. 146, 287–290, 2004. 9. Berggren, M. et al., Light-emitting diodes with variable colors from polymer blends, Nature 372, 444–446, 1994. 10. Lee, T.W. et al., Hole-injecting conducting-polymer compositions for highly efficient and stable organic light-emitting diodes, Appl. Phys. Lett. 87, 231106, 2005. 11. Yang, H.C. et al., Conducting AFM and 2D GIXD studies on pentacene thin films, J. Am. Chem. Soc. 127, 11542–11543, 2005. 12. Samori, P. et al., Nanoribbons from conjugated macromolecules on amorphous substrates observed by SFM and TEM, Nanotechnology 10, 77–80, 1999. 13. www.veeco.com.

8080_C004.fm Page 332 Tuesday, April 3, 2007 1:54 PM

332 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25.

26.

27. 28.

29. 30.

31.

32.

33. 34. 35. 36.

Organic Field-Effect Transistors www.asylumresearch.com. www.nanonics.co.il. www.molec.com. Alexander, S. et al., An atomic-resolution atomic-force microscope implemented in an optical-lever, J. Appl. Phys. 65, 164–167, 1989. Zhong, Q. et al., Fractured polymer silica fiber surface studied by tapping mode atomic-force microscopy, Surf. Sci. 290, L688–L692, 1993. www.molec.com/products_options.html#macmode. Binnig, G. et al., 7 × 7 Reconstruction on Si(111) resolved in real space, Phys. Rev. Lett. 50, 120–123, 1983. Binning, G. et al., Surface studies by scanning tunneling microscopy, Phys. Rev. Lett. 49, 57–61, 1982. Rohrer, H., Scanning tunneling microscopy, Proc. Nat. Acad. Sci. USA 84, 4666–4666, 1987. Tersoff, J. and Hamann, D.R., Theory and application for the scanning tunneling microscope, Phys. Rev. Lett. 50, 1998–2001, 1983. Tersoff, J. and Hamann, D.R., Theory of the scanning tunneling microscope, Phys. Rev. B 31, 805–813, 1985. Giancarlo, L.C. and Flynn, G.W., Scanning tunneling and atomic force microscopy probes of self-assembled, physisorbed monolayers: Peeking at the peaks, Ann. Rev. Phys. Chem. 49, 297, 1998. Giancarlo, L.C. and Flynn, G.W., Raising flags: Applications of chemical marker groups to study self-assembly, chirality, and orientation of interfacial films by scanning tunneling microscopy, Acc. Chem. Res. 33, 491–501, 2000. Poirier, G.E., Characterization of organosulfur molecular monolayers on Au(111) using scanning tunneling microscopy, Chem. Rev. 97, 1117–1127, 1997. Claypool, C.L. et al., Source of image contrast in STM images of functionalized alkanes on graphite: A systematic functional group approach, J. Phys. Chem. B. 101, 5978–5995, 1997. Eigler, D.M. et al., Imaging Xe with a low-temperature scanning tunneling microscope, Phys. Rev. Lett. 66, 1189–1192, 1991. Liang, W. et al., Electronic origin of scanning-tunneling-microscopy images and carbon skeleton orientations of normal-alkanes adsorbed on graphite, Adv. Mater. 5, 817–821, 1993. Fang, H.B., Giancarlo, L.C., and Flynn, G.W., Identification of the conformation of individual molecules by scanning tunneling microscopy, J. Phys. Chem. B. 103, 5712–5715, 1999. Li, L.Y., Chen, S.F., and Jiang, S.Y., Molecular-scale mixed alkanethiol monolayers of different terminal groups on Au(111) by low-current scanning tunneling microscopy, Langmuir 19, 3266–3271, 2003. Rabe, J.P. and Buchholz, S., Commensurability and mobility in 2-dimensional molecular-patterns on graphite, Science 253, 424–427, 1991. Herwig, K.W., Matthies, B., and Taub, H., Solvent effects on the monolayer structure of long n-alkane molecules adsorbed on graphite, Phys. Rev. Lett. 75, 3154–3157, 1995. Askadskaya, L. and Rabe, J.P., Anisotropic molecular-dynamics in the vicinity of order-disorder transitions in organic monolayers, Phys. Rev. Lett. 69, 1395–1398, 1992. Fang, H.B., Giancarlo, L.C., and Flynn, G.W., Packing of Br(CH2)10COOH and Br(CH2)(11)COOH on graphite: An odd-even length effect observed by scanning tunneling microscopy, J. Phys. Chem. B 102, 7421–7424, 1998.

8080_C004.fm Page 333 Tuesday, April 3, 2007 1:54 PM

Scanning Probe Techniques

333

37. Hibino, M. et al., Microscopic origin of the odd-even effect in monolayer of fatty acids formed on a graphite surface by scanning tunneling microscopy, J. Phys. Chem. B. 102, 4544–4547, 1998. 38. De Feyter, S. et al., Expression of chirality by achiral coadsorbed molecules in chiral monolayers observed by STM, Angew. Chem. Int. Ed. 37, 1223–1226, 1998. 39. Yablon, D.G., Giancarlo, L.C., and Flynn, G.W., Manipulating self-assembly with achiral molecules: An STM study of chiral segregation by achiral adsorbates, J. Phys. Chem. B. 104, 7627–7635, 2000. 40. Fang, H.B., Giancarlo, L.C., and Flynn, G.W., Direct determination of the chirality of organic molecules by scanning tunneling microscopy, J. Phys. Chem. B 102, 7311–7315, 1998. 41. Okawa, Y. and Aono, M., Linear chain polymerization initiated by a scanning tunneling microscope tip at designated positions, J. Chem. Phys. 115, 2317–2322, 2001. 42. Hipps, K.W., Barlow, D.E., and Mazur, U., Orbital mediated tunneling spectroscopy of VOPc, Abstr. Pap. Am. Chem. Soc. 219, U840–U840, 2000. 43. Scudiero, L. et al., Scanning tunneling microscopy, orbital-mediated tunneling spectroscopy, and ultraviolet photoelectron spectroscopy of metal(II) tetraphenylporphyrins deposited from vapor, J. Am. Chem. Soc. 123, 4073–4080, 2001. 44. Hipps, K.W. et al., A self-organized 2-dimensional bifunctional structure formed by supramolecular design, J. Am. Chem. Soc. 124, 2126–2127, 2002. 45. Lackinger, M. et al., Tunneling voltage polarity dependent submolecular contrast of naphthalocyanine on graphite. A STM study of close-packed monolayers under ultrahigh-vacuum conditions, J. Phys. Chem. B. 108, 2279–2284, 2004. 46. Qiu, X.H. et al., Alkane-assisted adsorption and assembly of phthalocyanines and porphyrins, J. Am. Chem. Soc. 122, 5550–5556, 2000. 47. Wang, H.N. et al., Chain-length-adjusted assembly of substituted porphyrins on graphite, Surf. Interface Anal. 32, 266–270, 2001. 48. Lu, X. and Hipps, K.W., Scanning tunneling microscopy of metal phthalocyanines: d(6) and d(8) cases, J. Phys. Chem. B 101, 5391–5396, 1997. 49. Lu, X. et al., Scanning tunneling microscopy of metal phthalocyanines: d(7) and d(9) cases, J. Am. Chem. Soc. 118, 7197–7202, 1996. 50. Stawasz, M.E., Sampson, D.L., and Parkinson, B.A., Scanning tunneling microscopy investigation of the ordered structures of dialkylamino hydroxylated squaraines adsorbed on highly oriented pyrolytic graphite, Langmuir 16, 2326–2342, 2000. 51. Miura, A. et al., Bias-dependent visualization of electron donor (D) and electron acceptor (A) moieties in a chiral DAD triad molecule, J. Am. Chem. Soc. 125, 14968–14969, 2003. 52. Jackel, F. et al., Prototypical single-molecule chemical-field-effect transistor with nanometer-sized gates, Phys. Rev. Lett. 92, 188303, 2004. 53. Hoeppener, S., Chi, L.F., and Fuchs, H., Formation of Au-55 strands on a molecular template at the solid–liquid interface, Nano. Lett. 2, 459–463, 2002. 54. Drake, B. et al., Imaging crystals, polymers, and processes in water with the atomic force microscope, Science 243, 1586–1589, 1989. 55. Gould, S.A.C. et al., The atomic force microscope — A tool for science and industry, Ultramicroscopy 33, 93–98, 1990. 56. Weihs, T.P. et al., Limits of imaging resolution for atomic force microscopy of molecules, Appl. Phys. Lett. 59, 3536–3538, 1991. 57. Landman, U., Luedtke, W.D., and Nitzan, A., Dynamics of tip substrate interactions in atomic force microscopy, Surf. Sci. 210, L177–L184, 1989.

8080_C004.fm Page 334 Tuesday, April 3, 2007 1:54 PM

334

Organic Field-Effect Transistors

58. Klinov, D. and Magonov, S., True molecular resolution in tapping-mode atomic force microscopy with high-resolution probes, Appl. Phys. Lett. 84, 2697–2699, 2004. 59. www.spmtips.com. 60. www.nanoworld.com. 61. www.olympus.co.jp/en/insg/probe. 62. www.nanosensors.com. 63. www.veecoprobes.com. 64. www.novascan.com. 65. Magonov, S., Atomic force microscopy at the 100-nm scale: practical, theoretical, and metrological outlook, Proc. SPIE 6002, 60 020S, 2005. 66. Ruiz, R. et al., Pentacene ultrathin film formation on reduced and oxidized Si surfaces, Phys. Rev. B 67 (12), 125406, 2003. 67. Meng, H. et al., High field-effect mobility oligofluorene derivatives with high environmental stability, J. Am. Chem. Soc. 123, 9214–9215, 2001. 68. Meng, H. et al., Oligofluorene-thiophene derivatives as high-performance semiconductors for organic thin film transistors, Chem. Mater. 15, 1778–1787, 2003. 69. Amar, J.G., Family, F., and Lam, P.M., Dynamic scaling of the island-size distribution and percolation in a model of submonolayer molecular-beam epitaxy, Phys. Rev. B 50, 8781–8797, 1994. 70. Venables, J.A., Spiller, G.D.T., and Hanbucken, M., Nucleation and growth of thinfilms, Rep. Prog. Phys. 47, 399–459, 1984. 71. Ruiz, R. et al., Dynamic scaling, island size distribution, and morphology in the aggregation regime of submonolayer pentacene films, Phys. Rev. Lett. 91, 136102, 2003. 72. Pratontep, S. et al., Correlated growth in ultrathin pentacene films on silicon oxide: Effect of deposition rate, Phys. Rev. B 69, 165201, 2004. 73. Dimitrakopoulos, C.D., Brown, A.R., and Pomp, A., Molecular beam deposited thin films of pentacene for organic field effect transistor applications, J. Appl. Phys. 80, 2501–2508, 1996. 74. Bouchoms, I.P.M. et al., Morphology identification of the thin film phases of vacuum evaporated pentacene on SIO2 substrates, Synth. Met. 104, 175–178, 1999. 75. Knipp, D. et al., Pentacene thin film transistors on inorganic dielectrics: Morphology, structural properties, and electronic transport, J. Appl. Phys. 93, 347–355, 2003. 76. Jentzsch, T. et al., Efficiency of optical second harmonic generation from pentacene films of different morphology and structure, Thin Solid Films 315, 273–280, 1998. 77. Knipp, D., Street, R.A., and Volkel, A.R., Morphology and electronic transport of polycrystalline pentacene thin-film transistors, Appl. Phys. Lett. 82, 3907–3909, 2003. 78. Mattheus, C.C. et al., Identification of polymorphs of pentacene, Synth. Met. 138, 475–481, 2003. 79. Mohapatra, S. et al., Organic thin-film transistors based on tolyl-substituted oligothiophenes, Adv. Funct. Mater. 14, 605–609, 2004. 80. Jung, S.Y. and Yao, Z., Organic field-effect transistors with single and double pentacene layers, Appl. Phys. Lett. 86, 83505, 2005. 81. Dinelli, F. et al., Thermal annealing effects on morphology and electrical response in ultrathin film organic transistors, Synth. Met. 146, 373–376, 2004. 82. Kalb, W. et al., Structure–performance relationship in pentacene/Al2O3 thin-film transistors, Synth. Met. 146, 279–282, 2004. 83. Nickel, B. et al., Dislocation arrangements in pentacene thin films, Phys. Rev. B 70, 125401, 2004. 84. Yanagisawa, H. et al., Structural and electrical characterization of pentacene films on SiO2 grown by molecular beam deposition, Thin Solid Films 464–465, 398–402, 2004.

8080_C004.fm Page 335 Tuesday, April 3, 2007 1:54 PM

Scanning Probe Techniques

335

85. Beernink, G. et al., Importance of dewetting in organic molecular-beam deposition: Pentacene on gold, Appl. Phys. Lett. 85, 398–400, 2004. 86. Inoue, Y. et al., Organic thin-film transistors based on anthracene oligomers, J. Appl. Phys. 95, 5795–5799, 2004. 87. Watkins, N.J. and Gao, Y., Vacuum level alignment of pentacene on LiF/Au, J. Appl. Phys. 94, 1289–1291, 2003. 88. Kiriy, N. et al., One-dimensional aggregation of regioregular polyalkylthiophenes, Nano. Lett. 3, 707–712, 2003. 89. Boer, E.A. et al., Models for quantitative charge imaging by atomic force microscopy, J. Appl. Phys. 90, 2764–2772, 2001. 90. Schaadt, D.M. et al., Charge storage in Co nanoclusters embedded in SiO2 by scanning force microscopy, Appl. Phys. Lett. 74, 472–474, 1999. 91. Melin, T. et al., Electric force microscopy of individually charged nanoparticles on conductors: An analytical model for quantitative charge imaging, Phys. Rev. B 69, 35321, 2004. 92. Raghavan, D. et al., Mapping polymer heterogeneity using atomic force microscopy phase imaging and nanoscale indentation, Macromolecules 33, 2573–2583, 2000. 93. Cabral, J.T. et al., Topography of phase-separated critical and off-critical polymer mixtures, Macromolecules 35, 1941–1950, 2002. 94. Lei, C.H. et al., Conductivity of macromolecular networks measured by electrostatic force microscopy, Appl. Phys. Lett. 83, 482–484, 2003. 95. Bachtold, A. et al., Scanned probe microscopy of electronic transport in carbon nanotubes, Phys. Rev. Lett. 84, 6082–6085, 2000. 96. Viswanathan, R. and Heaney, M.B., Direct imaging of the percolation network in a three-dimensional disordered conductor-insulator composite, Phys. Rev. Lett. 75, 4433–4436, 1995. 97. Boer, E.A. et al., Localized charge injection in SiO2 films containing silicon nanocrystals, Appl Phys Lett 79 (6), 791–793, 2001. 98. Boer, E.A. et al., Charging of single Si nanocrystals by atomic force microscopy, Appl. Phys. Lett. 78, 3133–3135, 2001. 99. Krauss, T.D. and Brus, L.E., Charge, polarizability, and photoionization of single semiconductor nanocrystals, Phys. Rev. Lett. 83, 4840–4843, 1999. 100. Cherniavskaya, O. et al., Photoionization of individual CdSe/CdS core/shell nanocrystals on silicon with 2-nm oxide depends on surface band bending, Nano Lett. 3, 497–501, 2003. 101. Cherniavskaya, O. et al., Quantitative noncontact electrostatic force imaging of nanocrystal polarizability, J. Phys. Chem. B 107, 1525–1531, 2003. 102. Zdrojek, M. et al., Charging and emission effects of multiwalled carbon nanotubes probed by electric force microscopy, Appl. Phys. Lett. 86, 213114, 2005. 103. Zdrojek, M. et al., Comment on “Electrostatics of individual single-walled carbon nanotubes investigated by electrostatic force microscopy,” Phys. Rev. Lett. 96, 39703, 2006. 104. Zheng, X. et al., Quantum-mechanical investigation of field-emission mechanism of a micrometer-long single-walled carbon nanotube, Phys. Rev. Lett. 92 (10), 106802, 2004. 105. Buldum, A. and Lu, J.P., Electron field emission properties of closed carbon nanotubes, Phys. Rev. Lett. 91 (23), 236801, 2003. 106. Dodabalapur, A., Torsi, L., and Katz, H.E., Organic transistors: 2-dimensional transport and improved electrical characteristics, Science 268, 270–271, 1995.

8080_C004.fm Page 336 Tuesday, April 3, 2007 1:54 PM

336

Organic Field-Effect Transistors

107. Dinelli, F. et al., Spatially correlated charge transport in organic thin film transistors, Phys. Rev. Lett. 92, 116802, 2004. 108. Horowitz, G. and Hajlaoui, M.E., Charge transport in polycrystalline oligothiophene thin film transistors, Synt. Met. 121, 1349–1350, 2001. 109. Horowitz, G. and Hajlaoui, M.E., Grain size dependent mobility in polycrystalline organic field-effect transistors, Synt. Met. 122, 185–189, 2001. 110. Horowitz, G. et al., The concept of “threshold voltage” in organic field-effect transistors, Adv. Mater. 10, 923, 1998. 111. Heim, T., Lmimouni, K., and Vuillaume, D., Ambipolar charge injection and transport in a single pentacene monolayer island, Nano Lett. 4, 2145–2150, 2004. 112. Heim, T. et al., Localization and delocalization of charges injected in DNA, Appl. Phys. Lett. 85, 2637–2639, 2004. 113. Nonnenmacher, M., Oboyle, M.P., and Wickramasinghe, H.K., Kelvin probe force microscopy, Appl. Phys. Lett. 58, 2921–2923, 1991. 114. Jacobs, H.O. et al., Surface potential mapping: A qualitative material contrast in SPM, Ultramicroscopy 69, 39–49, 1997. 115. Jacobs, H.O., Knapp, H.F., and Stemmer, A., Practical aspects of Kelvin probe force microscopy, Rev. Sci. Inst. 70, 1756–1760, 1999. 116. Nonnenmacher, M., Oboyle, M., and Wickramasinghe, H.K., Surface investigations with a Kelvin probe force microscope, Ultramicroscopy 42, 268–273, 1992. 117. Necliudov, P.V. et al., Modeling of organic thin film transistors of different designs, J. Appl. Phys. 88, 6594–6597, 2000. 118. Necliudov, P.V. et al., Contact resistance extraction in pentacene thin film transistors, Solid-State Electron. 47, 259–262, 2003. 119. Dimitrakopoulos, C.D. and Malenfant, P.R.L., Organic thin film transistors for large area electronics, Adv. Mater. 14, 99, 2002. 120. Burgi, L., Sirringhaus, H., and Friend, R.H., Noncontact potentiometry of polymer field-effect transistors, Appl. Phys. Lett. 80, 2913–2915, 2002. 121. Burgi, L., Friend, R.H., and Sirringhaus, H., Formation of the accumulation layer in polymer field-effect transistors, Appl. Phys. Lett. 82, 1482–1484, 2003. 122. Hassenkam, T., Greve, D.R., and Bjornholm, T., Direct visualization of the nanoscale morphology of conducting polythiophene monolayers studied by electrostatic force microscopy, Adv. Mater. 13, 631–634, 2001. 123. Nichols, J.A., Gundlach, D.J., and Jackson, T.N., Potential imaging of pentacene organic thin-film transistors, Appl. Phys. Lett. 83, 2366–2368, 2003. 124. Puntambekar, K.P., Pesavento, P.V., and Frisbie, C.D., Surface potential profiling and contact resistance measurements on operating pentacene thin-film transistors by Kelvin probe force microscopy, Appl. Phys. Lett. 83, 5539–5541, 2003. 125. Kymissis, I., Dimitrakopoulos, C.D., and Purushothaman, S., High-performance bottom electrode organic thin-film transistors, IEEE Trans. Electron Devices 48, 1060–1064, 2001. 126. Loiacono, M.J., Granstrom, E.L., and Frisbie, C.D., Investigation of charge transport in thin, doped sexithiophene crystals by conducting probe atomic force microscopy, J. Phys. Chem. B. 102, 1679–1688, 1998. 127. Oshea, S.J. et al., Conducting atomic-force microscopy study of silicon dioxide Breakdown, J. Vac. Sci. Technol., B 13, 1945–1952, 1995. 128. Klein, D.L. and McEuen, P.L., Conducting atomic-force microscopy of alkane layers on graphite, Appl. Phys. Lett. 66, 2478–2480, 1995.

8080_C004.fm Page 337 Tuesday, April 3, 2007 1:54 PM

Scanning Probe Techniques

337

129. Kelley, T.W., Granstrom, E.L., and Frisbie, C.D., Conducting probe atomic force microscopy: A characterization tool for molecular electronics, Adv. Mater. 11, 261, 1999. 130. Zhao, J.W. and Uosaki, K., A novel nanolithography technique for self-assembled monolayers using a current sensing atomic force microscope, Langmuir 17, 7784–7788, 2001. 131. Zhao, J.W. and Uosaki, K., Formation of nanopatterns of a self-assembled monolayer (SAM) within a SAM of different molecules using a current sensing atomic force microscope, Nano. Lett. 2, 137–140, 2002. 132. Cui, X.D. et al., Reproducible measurement of single-molecule conductivity, Science 294, 571–574, 2001. 133. Cui, X.D. et al., Making electrical contacts to molecular monolayers, Nanotechnology 13, 5–14, 2002. 134. Leatherman, G. et al., Carotene as a molecular wire: Conducting atomic force microscopy, J. Phys. Chem. B. 103, 4006–4010, 1999. 135. Beebe, J.M. et al., Contact resistance in metal–molecule–metal junctions based on aliphatic SAMs: Effects of surface linker and metal work function, J. Am. Chem. Soc. 124, 11268–11269, 2002. 136. Wold, D.J. and Frisbie, C.D., Fabrication and characterization of metal–molecule–metal junctions by conducting probe atomic force microscopy, J. Am. Chem. Soc. 123, 5549–5556, 2001. 137. Wold, D.J. and Frisbie, C.D., Formation of metal–molecule–metal tunnel junctions: Microcontacts to alkanethiol monolayers with a conducting AFM tip, J. Am. Chem. Soc. 122, 2970–2971, 2000. 138. Heddleson, J.M. et al., Profiling of silicide silicon structures using a combination of the spreading resistance and point-contact current-voltage methods, J. Vac. Sci. Technol., B 12, 317–321, 1994. 139. Dewolf, P. et al., Lateral and vertical dopant profiling in semiconductors by atomicforce microscopy using conducting tips, J. Vac. Sci. Tech., A 13, 1699–1704, 1995. 140. Chen, K.W. et al., Characterization of nano-sized Si islands in buried oxide layer of SIMOX by conducting AFM, Chem Phys Lett 376, 748–752, 2003. 141. Macpherson, J.V., de Mussy, J.P.G., and Delplancke, J.L., Conducting-atomic force microscopy investigation of the local electrical characteristics of a Ti/TiO2/Pt anode, Electrochem. Solid St. 4, E33–E36, 2001. 142. Yano, K. et al., Nanometer scale conductance change in a Langmuir–Blodgett film with the atomic force microscope, Appl. Phys. Lett. 68, 188–190, 1996. 143. Dai, H.J., Wong, E.W., and Lieber, C.M., Probing electrical transport in nanomaterials: Conductivity of individual carbon nanotubes, Science 272, 523–526, 1996. 144. Thomson, R.E. and Moreland, J., Development of highly conductive cantilevers for atomic-force microscopy point-contact measurements, J. Vac. Sci. Technol., B 13 (3), 1123–1125, 1995. 145. Beale, J. and Pease, R.F., Limits of high-density, low-force pressure contacts, IEEE Trans. Components Packaging Manuf. Technol. Part A 17, 257–262, 1994. 146. Chen, J. et al., Large on–off ratios and negative differential resistance in a molecular electronic device, Science 286, 1550–1552, 1999. 147. Chen, J. et al., Electronic transport through metal-1,4-phenylene diisocyanide-metal junctions, Chem. Phys. Lett. 313, 741–748, 1999. 148. Wang, W.Y., Lee, T., and Reed, M.A., Mechanism of electron conduction in selfassembled alkanethiol monolayer devices, Phys. Rev. B 68, 35416, 2003.

8080_C004.fm Page 338 Tuesday, April 3, 2007 1:54 PM

338

Organic Field-Effect Transistors

149. Reed, M.A. et al., Conductance of a molecular junction, Science 278, 252–254, 1997. 150. Park, H. et al., Nanomechanical oscillations in a single-C-60 transistor, Nature 407, 57–60, 2000. 151. Slowinski, K. et al., Evidence for inefficient chain-to-chain coupling in electron tunneling through liquid alkanethiol monolayer films on mercury, J. Am. Chem. Soc. 118 (19), 4709–4710, 1996. 152. Holmlin, R.E. et al., Electron transport through thin organic films in metal–insulator–metal junctions based on self-assembled monolayers, J. Am. Chem. Soc. 123, 5075–5085, 2001. 153. Selzer, Y., Salomon, A., and Cahen, D., The importance of chemical bonding to the contact for tunneling through alkyl chains, J. Phys. Chem. B 106, 10432–10439, 2002. 154. Wold, D.J. et al., Distance dependence of electron tunneling through self-assembled monolayers measured by conducting probe atomic force microscopy: Unsaturated versus saturated molecular junctions, J. Phys. Chem. B. 106 (11), 2813–2816, 2002. 155. Labonte, A.P. et al., Scanning tunneling spectroscopy of insulating self-assembled monolayers on Au(111), J. Phys. Chem. B. 106, 8721–8725, 2002. 156. Fan, F.R.F. et al., Structure-dependent charge transport and storage in self-assembled monolayers of compounds of interest in molecular electronics: Effects of tip material, headgroup, and surface concentration, J. Am. Chem. Soc. 126, 4035–4042, 2004. 157. Liu, B. et al., Electron transfer at self-assembled monolayers measured by scanning electrochemical microscopy, J. Am. Chem. Soc. 126, 1485–1492, 2004. 158. Tivanski, A.V. et al., Conjugated thiol linker for enhanced electrical conduction of gold-molecule contacts, J. Phys. Chem. B. 109, 5398–5402, 2005. 159. Ramachandran, G.K. et al., A bond-fluctuation mechanism for stochastic switching in wired molecules, Science 300, 1413–1416, 2003. 160. Rawlett, A.M. et al., A molecular electronics toolbox, Nanotechnology 14, 377–384, 2003. 161. Rawlett, A.M. et al., Electrical measurements of a dithiolated electronic molecule via conducting atomic force microscopy, Appl. Phys. Lett. 81, 3043–3045, 2002. 162. Yaliraki, S.N. and Ratner, M.A., Molecule-interface coupling effects on electronic transport in molecular wires, J. Chem. Phys. 109, 5036–5043, 1998. 163. Hong, S. et al., Molecular conductance spectroscopy of conjugated, phenyl-based molecules on Au(111): The effect of end groups on molecular conduction, Superlattice Microst. 28, 289–303, 2000. 164. Seminario, J.M., De la Cruz, C.E., and Derosa, P.A., A theoretical analysis of metalmolecule contacts, J. Am. Chem. Soc. 123, 5616–5617, 2001. 165. Skulason, H. and Frisbie, C.D., Detection of discrete interactions upon rupture of Au microcontacts to self-assembled monolayers terminated with –S(CO)CH3 or –SH, J. Am. Chem. Soc. 122, 9750–9760, 2000. 166. Salomon, A. et al., Comparison of electronic transport measurements on organic molecules, Adv. Mater. 15, 1881–1890, 2003. 167. Engelkes, V.B., Beebe, J.M., and Frisbie, C.D., Length-dependent transport in molecular junctions based on SAMs of alkanethiols and alkanedithiols: Effect of metal work function and applied bias on tunneling efficiency and contact resistance, J. Am. Chem. Soc. 126, 14287–14296, 2004. 168. Engelkes, V.B., Beebe, J.M., and Frisbie, C.D., Analysis of the causes of variance in resistance measurements on metal–molecule–metal junctions formed by conductingprobe atomic force microscopy, J. Phys. Chem. B. 109, 16801–16810, 2005.

8080_C004.fm Page 339 Tuesday, April 3, 2007 1:54 PM

Scanning Probe Techniques

339

169. Slowinski, K., Fong, H.K.Y., and Majda, M., Mercury–mercury tunneling junctions. 1. Electron tunneling across symmetric and asymmetric alkanethiolate bilayers, J. Am. Chem. Soc. 121, 7257–7261, 1999. 170. Selzer, Y., Salomon, A., and Cahen, D., Effect of molecule–metal electronic coupling on through-bond hole tunneling across metal–organic monolayer–semiconductor junctions, J. Am. Chem. Soc. 124, 2886–2887, 2002. 171. Cui, X.D. et al., Changes in the electronic properties of a molecule when it is wired into a circuit, J. Phys. Chem. B 106, 8609–8614, 2002. 172. Gu, Y. et al., Structural characterization and electron tunneling at n-Si/SiO2/SAM/liquid interface, J. Phys. Chem. B 103, 5220–5226, 1999. 173. Kelley, T.W. and Frisbie, C.D., Gate voltage dependent resistance of a single organic semiconductor grain boundary, J. Phys. Chem. B 105, 4538–4540, 2001. 174. Olbrich, A., Ebersberger, B., and Boit, C., Conducting atomic force microscopy for nanoscale electrical characterization of thin SiO2, Appl. Phys. Lett. 73, 3114–3116, 1998. 175. Kelley, T.W. and Frisbie, C.D., Point contact current-voltage measurements on individual organic semiconductor grains by conducting probe atomic force microscopy, J. Vac. Sci. Technol., B 18, 632–635, 2000. 176. Chwang, A.B. and Frisbie, C.D., Temperature and gate voltage dependent transport across a single organic semiconductor grain boundary, J. Appl. Phys. 90, 1342–1349, 2001. 177. Pellegrino, O. et al., High-resolution electron energy loss spectroscopy studies on the electronic structure of quinquethiophene (5T) and sexithiophene (6T) films, Thin Solid Films 329, 291–294, 1998. 178. Torsi, L. et al., Charge transport in oligothiophene field-effect transistors, Phys. Rev. B 57, 2271–2275, 1998. 179. Lovinger, A.J. and Rothberg, L.J., Electrically active organic and polymeric materials for thin-film-transistor technologies, J. Mater. Res. 11, 1581–1592, 1996. 180. Torsi, L. et al., Intrinsic transport properties and performance limits of organic fieldeffect transistors, Science 272, 1462–1464, 1996.

8080_C004.fm Page 340 Tuesday, April 3, 2007 1:54 PM

8080_C005.fm Page 341 Friday, April 6, 2007 10:25 AM

5.1

Vacuum Evaporated Thin Films

Alex C. Mayer, Jack M. Blakely, and George G. Malliaras CONTENTS 5.1.1 Introduction................................................................................................341 5.1.1.1 Film Growth Technology Considerations ...................................342 5.1.1.2 Describing Film Growth: Thermodynamics and Kinetics..........343 5.1.1.3 Inorganics versus Organics .........................................................345 5.1.2 Thin Film Characterization Techniques ....................................................346 5.1.2.1 Scanning Probe Techniques ........................................................347 5.1.2.2 X-Ray Scattering .........................................................................348 5.1.2.3 Electron-Based Techniques .........................................................348 5.1.3 Organic Film Growth Kinetics ..................................................................349 5.1.3.1 Thermodynamic Driving Force...................................................351 5.1.3.2 Rate Equations: Microscopic ......................................................352 5.1.3.2.1 Rate Equation Elements ............................................353 5.1.3.2.2 Dynamic Island Size Distribution .............................355 5.1.3.2.3 Beyond the First Layer: Birth–Death Models ..........358 5.1.3.3 Rate Equations: Macroscopic......................................................363 5.1.4 Effects of the Substrate..............................................................................364 5.1.4.1 Effect of Surface Energy.............................................................364 5.1.4.2 Effect of Surface Roughness.......................................................364 5.1.5 Outlook.......................................................................................................365 References..............................................................................................................366

5.1.1 INTRODUCTION Small molecules such as pentacene exhibit the highest charge carrier mobilities among organic semiconductors and are therefore of interest for organic thin film transistors (OTFTs). These materials are usually utilized in OTFTs in the form of thin films deposited on insulating substrates such as SiO2. The strong correlation between film morphology and device performance has motivated several recent studies on organic film growth. In this chapter, the fundamental mechanisms that

341

8080_C005.fm Page 342 Friday, April 6, 2007 10:25 AM

342

Organic Field-Effect Transistors

determine the morphology of organic thin films grown using physical vapor deposition will be discussed. The mode of film growth and the resulting film morphology are usually described using thermodynamic concepts. However, film growth is actually a surface phenomenon mediated by kinetic processes; therefore, the vast majority of this chapter will deal with a kinetic description of film growth processes. These concepts have appeared in many books and review articles on inorganic film growth [1–5] and more recently on organic film growth [6–10]. Even though there are several important results obtained for organic films grown on conducting substrates, this chapter will primarily discuss film growth on insulators, because these substrates are relevant for the thin film transistor community. This chapter is not meant as a survey of the recent literature, but will rather explore film growth from a didactic viewpoint.

5.1.1.1 FILM GROWTH TECHNOLOGY CONSIDERATIONS Thin films are grown by a number of different processes. The choice of the process is determined by the chemical nature of the source material and the desired properties of the film. Examples of film growth techniques include physical vapor deposition (PVD), chemical vapor deposition (CVD), pulsed laser deposition (PLD), and ion sputtering [2]. The most popular deposition technique and the subject of this chapter is PVD, which involves the deposition of molecules from the vapor phase to the solid phase onto a desired substrate. This is also the technique for which a large number of film growth theories were created. Vacuum deposition occurs as molecules are removed from a solid or liquid and then travel over some distance in a vacuum chamber and impinge on the substrate at a rate, F, usually measured in monolayers per second. A typical vacuum chamber utilized for PVD is shown in Figure 5.1.1a. The chamber is equipped with a bell

Quartz Crystal Microbalance

Shutter

Source Heater

(a)

(b)

FIGURE 5.1.1 PVD vacuum systems utilized for thin film deposition. The chamber in (b) is mounted at the A2 station at the Cornell high-energy synchrotron source and allows realtime monitoring of the film morphology and structure by in situ x-ray diffraction. (Headrick, R.L. et al., AIP Conf. Proc. (AIP, Melville, NY, 2004), 705, 1150, 2004.)

8080_C005.fm Page 343 Friday, April 6, 2007 10:25 AM

Vacuum Evaporated Thin Films

343

jar, vacuum pump, boat (where the material can be heated to the sublimation or evaporation temperature), shutter, substrate mount, and quartz crystal microbalance, which is used to measure the thickness of the film. Custom-made chambers that include advanced diagnostics play an important role in understanding film growth. A photograph of such a custom system is shown in Figure 5.1.1b. This chamber is shown mounted in a synchrotron beam line at the Cornell high-energy synchrotron source and is equipped with a beryllium window to allow the transmission of x-rays onto the substrate [11]. It should be noted that, in any vacuum deposition chamber, zero pressure is never achieved. This fact has significant implications because, at any pressure, there exists a mixture of residue molecules such as N2, O2, H2O, and CO2 that impinges on the substrate and possibly interacts with the growing film. According to the kinetic theory of gases [2,5], molecules in the ambient impinge on a substrate at a rate given by [2,5]: Famb =

p

(5.1.1)

2πkB Tm

where p is the partial pressure of the gas with molecular mass m and kB is Boltzmann’s constant. A monolayer is composed of ≈1015 molecules per square centimeter, which implies that at 300 K, the time for one monolayer of N2 to adsorb on the surface is given by [2,5]: τ ≈ 10 −6 p −1s −1

(5.1.2)

where p is the pressure in millibars and s is the “sticking coefficient.” The latter is the probability for a molecule impinging on the surface to adsorb. This translates into a minimum monolayer formation time of approximately 1 sec at 10–6 mbar, a typical pressure in vacuum chambers used in OTFT fabrication, and to around 3 h at 10–10 mbar, a typical pressure in UHV chambers used for surface science studies. The commercial success of OTFTs relies on efficient, low-cost fabrication technologies, so UHV is unlikely to be utilized in mass production. Therefore, the incorporation of impurities from the ambient remains an important consideration in device fabrication. For the case of nitrogen, the effects will be small because the sticking coefficient for inert gases is usually small. On the other hand, water and oxygen can have detrimental effects because both molecules tend to affect charge transport in organic semiconductors. The effects of water can be reduced through the use of a low-cost cold trap, by heating the substrate, or by using a highly hydrophobic substrate that reduces the sticking coefficient.

5.1.1.2 DESCRIBING FILM GROWTH: THERMODYNAMICS AND KINETICS Films evolve in one of three possible modes. The mode of growth is dependent on the growth conditions used, the strengths of the interactions between the molecules

8080_C005.fm Page 344 Friday, April 6, 2007 10:25 AM

344

Organic Field-Effect Transistors

e

er rM

d an -v k n

we

StranskiKrastenov

a Fr

(a)

(b)

Vo lm

erW eb e

r

(c)

FIGURE 5.1.2 Three growth modes for an evolving thin film: (a) layer-by-layer, (b) Stranski–Krastanov, or (c) island growth.

diffusing on the substrate (admolecules) and the substrate, and the interaction strength between the admolecules and the solid islands. These growth modes are displayed in Figure 5.1.2 [1–5], where the building blocks of the films are shown as cubes arranged in a simple cubic packing on a substrate. This idealized model of a growing crystal is known as a Kossel crystal [12] and is useful for simple calculations. Frank–van der Merwe, two-dimensional, or layer-by-layer growth is shown in Figure 5.1.2a. This mode of growth is desired for transistor applications because it leads to large, well-connected domains that facilitate charge transport parallel to the substrate. Layer-by-layer growth is thermodynamically favorable for systems where the sum of the surface energy (tension) of the adsorbate, γfilm, and the interfacial energy between the substrate and the film, γint, is less than or equal to the surface energy of the substrate, γsub [5]. As the film grows thicker, the effective value of γint may increase enough (possibly due to the buildup of strain) to make layer-by-layer growth unfavorable and eventually lead to what is known as Stranski–Krastanov growth whereby islands form on top of one or several complete monolayers as shown in Figure 5.1.2b. Finally, the growth scenario known as Volmer–Weber, three-dimensional, or island growth is shown in Figure 5.1.2c. This growth mode is reminiscent of water beading up on a lightly oiled pan and occurs when γfilm + γint is greater than γsub [5]. For most systems this growth scenario occurs when the adsorbates are more tightly bound to each other than to the substrate [5]. The ideas described previously for understanding film morphology in terms of the local equilibrium and in terms of the surface tension are useful, but film growth occurs far from equilibrium (ex vi termini). Thus, kinetic processes control the details of film growth and the final film morphology. According to the paper by Burton, Cabrera, and Frank [13], the kinetic rates and processes are controlled by the thermodynamic driving force Δμ, defined as the positive difference between the chemical potential of a molecule in the vapor phase and that in the crystal phase,

8080_C005.fm Page 345 Friday, April 6, 2007 10:25 AM

Vacuum Evaporated Thin Films

345

μv–μc. The value of Δμ is a function of the supersaturation, p/pe, where pe is the equilibrium vapor pressure and p is the ambient pressure, given by [13]: ⎛ p⎞ Δμ = kB T ln ⎜ ⎟ . ⎝ pe ⎠

(5.1.3)

According to this formalism, the details of film growth (see Section 5.1.3) are described by the relative rates of processes such as the impingement and sticking rate of the adsorbate onto the substrate, desorption, surface diffusion, nucleation, attachment to an island, and coalescence. These individual rates are dependent on the deposition conditions such as the substrate temperature and the pressure of the adsorbate in the vapor phase relative to its equilibrium value, pe. The dependence of the morphology on these parameters will be discussed in Section 5.1.3.2. The crystalline nature of a material can prevent the film from developing as planar sheets as shown in the preceding idealized figures. Instead, the film may be composed of an array of individual crystallites. Both the equilibrium crystal shape as well as the anisotropy of the growth velocity of the various crystal surfaces can determine the final shape of these crystallites. Wulff [14] proposed that in equilibrium the crystalline shape can be determined from a polar plot of surface free energy in which the length of the radius vector represents the magnitude of the surface energy, γ(rˆ ) along the surface normal. If a set of (Wulff) planes is constructed normal to the radius at a distance γ(rˆ ) from the origin, then the equilibrium crystal shape is given by the inner envelope of these planes. This treatment shows that the surfaces with the lowest surface tension will preferentially be exposed. Examples are shown in Figure 5.1.3 for a Kossel crystal (Figure 5.1.3a) and for a pentacene crystal (calculated: Figure 5.1.3b; experimental: Figure 5.1.3c) [15]. As noted earlier, films are not grown under equilibrium conditions and are confined by a substrate, which implies that these predicted shapes would not be exact. However, the directions of growth are along the lowest surface energy directions. If, however, the crystallite shape is totally determined by the anisotropy of the crystal facet growth velocity v (r ) , then it can be shown that the resulting crystal shape can be constructed from a similar inner envelope of planes drawn at a dis tance v (r ) . A crystal formed this way will expose the faces with the slowest growth velocities [16]. Real film growth is controlled by both mechanisms, and it is sometimes difficult to determine which is dominant.

5.1.1.3 INORGANICS

VERSUS

ORGANICS

In molecular crystals, the binding energies are highly anisotropic due to the complex shape of the molecules in addition to the anisotropy inherent in the crystal packing. Most organic molecules have complex shapes (bucky balls being an exception), so the representation of a molecule as a cube, as in the Kossel crystal, is not a good approximation. A van der Waals box that reflects the shape of the molecule is a better approximation.

8080_C005.fm Page 346 Friday, April 6, 2007 10:25 AM

346

Organic Field-Effect Transistors

(a)

(c)

(b) )

0 (11

0)

(01 b

d1

nd1

0)

(10 a

)

10

(1–

d2

nb

nd2

na

FIGURE 5.1.3 Predicted equilibrium crystal shapes for (a) a Kossel crystal and (b and c) a pentacene crystal. (b) Results of a calculation based on a triclinic pentacene crystal. (c) The observed crytallite shape using AFM (5 by 5 μm2). (Verlaak, S. et al., Phys. Rev. B., 68, 195409, 2003.)

An important feature of OTFT film growth versus traditional inorganic film growth involves the lack of true epitaxy with the substrate, since OTFT substrates are typically inert and amorphous materials (e.g., thermal SiO2 or polymers). In some instances, the molecules will align with a preferential orientation along one crystallographic direction (e.g., pentacene’s [001] parallel to the normal of the surface). This is not true epitaxy and is only found in some systems where the molecular anisotropy is strong enough. The case of van der Waals or quasi-epitaxy has been explored in the past [17–19] for organics deposited on metals, highly ordered pyrolytic graphite (HOPG), or exotic surfaces and will not be discussed here any further. One more important parameter that is different for organic film growth as compared to inorganic film growth is the much weaker intermolecular binding energies in the case of organic molecules. The lower binding energies, which are a consequence of the van der Waals forces, reduce the temperature range utilized in film growth by a significant amount. A limitation on the substrate temperature will have significant device implications because the sticking coefficient for water will be nonzero for most substrates used in OTFT fabrication. Despite these differences between organics and inorganics, the models generated to explain inorganic film growth offer a good starting point for understanding growth physics of thin films relevant for OTFTs.

5.1.2 THIN FILM CHARACTERIZATION TECHNIQUES In order to explore the growth dynamics and film morphology, several experimental techniques have been utilized. Experimental techniques utilized to date make use of

8080_C005.fm Page 347 Friday, April 6, 2007 10:25 AM

Vacuum Evaporated Thin Films

347

x-ray scattering, electron diffraction and emission, scanning probes, and optical methods. All of these techniques can be used to decipher parts of the puzzle, and all have inherent limitations. Each of these techniques will be briefly described next, with special attention paid to basic operating principles and limitations.

5.1.2.1 SCANNING PROBE TECHNIQUES Studies of surface morphology of organic thin films with the atomic force microscope (AFM) [20] have been the most prolific to date in terms of extracting growth laws and trends. Scanning tunneling microscopy (STM) has also been used, but the atomic resolution comes at the cost of having to use a conductive substrate and therefore is not directly useful for OTFTs, where the substrate is an insulator. By inducing a transient increase in conductivity of the organic layer [21], it could be possible to extend the STM studies to fairly thick films on insulators. The working principle of an AFM is illustrated in Figure 5.1.4. The AFM relies on a sharp tip (typically a pyramid with a tip width of ~10 nm) at the end of an oscillating cantilever brought within a distance z of the surface so that the van der Waals attractive interaction and the short-range repulsion form a potential well (solid line in Figure 5.1.4a). The AFM can be operated in a contact mode that involves dragging the tip over the sample or in a tapping mode in which the tip vibrates around z just off the resonant frequency associated with the potential well (see Figure 5.1.4a). 4 3

Normalized frequency shift Force Force gradient

Fts [nN] kts [10 N/m] γ [10 fNm0.5]

2 1 0 –1 –2 –3 –4 0.2

0.3

0.4

0.5 z [nm]

0.6

0.7

0.8

(a)

(b)

FIGURE 5.1.4 (a) Dependence of the force gradient, Fts, force gradient, kts, and amplitude normalized frequency shift, γ, on the tip-sample distance. (From Giessible, F.J., Advances in atomic force microscopy, Rev. Mod. Phys., 75, 949, 2003.) (b) Representation of an AFM tip on a sample.

8080_C005.fm Page 348 Friday, April 6, 2007 10:25 AM

348

Organic Field-Effect Transistors

The tapping mode is generally preferred for organic film morphology studies because the organic materials are soft and can be easily damaged. In tapping mode, the surface morphology is mapped by following the modulation of the frequency as the tip moves close to and away from the surface. The last decade has seen many developments in improving the sensitivity of AFM measurements that have yielded an excellent vertical sensitivity of better than 0.1 nm, but due to the shape of the tip, a lateral resolution no better than 10 nm is usually achieved. The limit in the lateral resolution can lead to artifacts when trying to estimate the exact size of islands at low coverages, but this error will be minimal when analyzing island size distributions [22].

5.1.2.2 X-RAY SCATTERING After AFM, x-ray related techniques are by far the most popular because they are compatible with insulating substrates, are nondestructive, and do not require vacuum or special sample preparation. X-ray related methods are extremely powerful in determining the degree of crystallinity, average domain size, and film orientation. Information on dislocation density and surface roughness can also be obtained from the shapes of the diffracted intensity curves. The geometries utilized in x-ray diffraction are displayed in Figure 5.1.5. This figure shows the geometry known as specular (angle of incidence equals angle of reflection in the plane of incidence) where information related to the crystalline planes parallel to the surface is extracted. The spacing between these planes along the sample normal, d, in the films can be determined through Bragg’s law [23]: λ = 2d sin θ

(5.1.4)

where λ is the wavelength of the x-ray beam. An example is shown for pentacene arranged in the “thin-film” phase [24] on SiO2 in Figure 5.1.5c. The inset shows the arrangement of the molecules and the spacing between the layers measured in the specular geometry. In-plane length scales and information about the ordering of the crystallites can be determined in this geometry by holding one angle fixed and rocking the other. Figure 5.1.5 also shows the experimental arrangement (Figure 5.1.5b) and results from thin pentacene layers (Figure 5.1.5d) for the geometry known as grazing incidence x-ray diffraction (GIXD) [24,25]. In this arrangement, information about the crystal packing parallel to the substrate is gained with a high signal-to-noise ratio evidenced by data obtained from submonolayer films shown in Figure 5.1.5d. The specular geometry would provide no structural information from films this thin.

5.1.2.3 ELECTRON-BASED TECHNIQUES A low secondary electron coefficient for carbon as well as charging of the gate insulator in OTFTs lowers the usefulness of scanning electron microscopy (SEM) to study the morphology of growing organic films. Studies of organic film nucleation and growth have been carried out very successfully, however, using the low-energy

8080_C005.fm Page 349 Friday, April 6, 2007 10:25 AM

Vacuum Evaporated Thin Films

349

z (a)

(b) k0

Ghk0

kf

θ

k0

2θ

(d)

Intensity (a.u.)

105

d001 = 15.43Å

104 c* 103

a

Intensity (a.u.)

(c)

kf αf

ψ 2θ

αi

3 × 103 (110) 2 × 103

φ (210) φ b γ (020) (200) a (120)

45°C

1 × 103

22°C

102 0.0 0.4 0.8 1.2 1.6 2.0 2.4

ks αi

0 1.0

qz (Å–1)

Sub-monolayer

0°C

1.5

2.0

2.5

3.0

3.5

qII(Å–1)

FIGURE 5.1.5 X-ray scattering experimental setups (a and b) and some results (c and d) for pentacene thin films on SiO2. (Ruiz, R. et al., Appl. Phys. Lett., 85, 4926, 2004.) Specular orientation reveals information about the ordering parallel to the substrate where pentacene films have high crystalline ordering. In (a) and (c), the momentum transfer vector is normal to the surface; in (b) and (d), the momentum transfer is mostly in the plane of the surface.

electron microscope (LEEM) [26]. In LEEM, an image is formed from a backscattered diffracted electron beam when an extremely low energy beam is incident on the sample [26]. Contrast arises from variations in the surface structural order and also from variations of surface charge and potential. Photoelectron emission microscopy (PEEM) can also be used to image and monitor the growth of organic layers. In PEEM, the emitted electrons are in response to excitation of the sample by ultraviolet light. The contrast in the image arises from differences in electronic structure from different surface regions [27]. Figure 5.1.6 shows examples of PEEM images at different initial stages of the growth of pentacene on silicon [27].

5.1.3 ORGANIC FILM GROWTH KINETICS In this section, the main factors controlling organic film growth and morphology will be discussed in detail. The discussion will follow the models and explanations developed by the inorganic film growth community with several examples where the theories have been extended to organics. The fundamentals of film growth have been addressed in several textbooks and review papers on inorganic film growth [1–6] mentioned in the introduction. This section will be arranged to follow a

8080_C005.fm Page 350 Friday, April 6, 2007 10:25 AM

350

Organic Field-Effect Transistors

15 μm

(a)

(b)

(c)

FIGURE 5.1.6 Evolution of a pentacene film deposited on Si(100) as monitored by PEEM: (a) submonolayer; (b) two layers; and (c) three layers. (Meyer Zu Heringdorf, F.-J. et al., Nature, 412, 517, 2001.)

standard kinetics treatment working from the underlying physics up to experimentally observable parameters. The two standard approaches in any treatment of kinetics [28] are to explain the system in terms of the thermodynamic driving forces (namely, ∇μ) or in terms of the fundamental rate equations. The rate equations can be further subdivided into an atomistic, or microscopic, approach that accounts for individual molecules as they go through the various processes (adsorption, desorption, diffusion, capture, and release) or a phenomenological, or macroscopic, explanation that looks for correlations and the so-called scaling laws over large distances (much larger than the lattice spacing). Our discussion of the free energy driving forces will mostly follow the discussions outlined in Venables et al. [5] and in Markov [3]. The extension of the atomistic approach to the organics will come from Verlaak et al. [15]. The elements of the rate equation will follow Venables et al. [5]. Dynamic scaling will then follow as an extension to the rate equations coupled with some assumptions about the island size distributions [23]. For deposition beyond the first layer the comparison of model predictions with experiment will follow Cohen et al. [29]. The height–height correlation description will follow Krim and Palasantzas [30]. In all treatments, it is shown that there is a strong dependence of the film growth mode and morphology on the deposition rate, F, and on the substrate temperature. Examples for OTFT-relevant materials grown at different temperatures are shown in Figure 5.1.7. The top row in Figure 5.1.7 displays AFM micrographs of pentacene grown at elevated temperatures on SiO2 with a mask covering half the substrate (a) [31] and on poly-methyl-methacrylate (PMMA) without a mask (b) [32]. As can be seen, the choice of substrate drastically affects the mode of growth. Figure 5.1.7c and d show AFM micrographs for the deposition of C60 on hydrogen-terminated Si(100) [33]. An increased substrate temperature causes the transition from two dimensions at room temperature (Figure 5.1.7c) to three dimensions at 473 K (Figure 5.1.7d). Finally, for perylene deposited on SiO2 at room temperature (Figure 5.1.7e), the transition from two to three dimensions is caused by the addition of a selfassembled monolayer (SAM) of octadecyltrichlorosilane (OTS) in Figure 5.1.7f [34].

8080_C005.fm Page 351 Friday, April 6, 2007 10:25 AM

Vacuum Evaporated Thin Films

351

(a)

(b)

5 μm

2 μm

(c)

(d)

1 μm

1 μm

(e)

(f )

2 μm

2 μm

FIGURE 5.1.7 Examples of two-dimensional (connected) and three-dimensional growth for various organic molecular systems. The transition in pentacene film growth occurs at elevated temperatures by changing the substrate from SiO2 (a) (Ruiz, R. et al., Adv. Mater., 17, 1795, 2005) to PMMA (b). (Wang, G.Z. et al., Appl. Phys. Lett., 83, 3108, 2003.) The transition from two to three dimensions for the deposition of C60 on H-Si(100) occurs by changing the substrate from room temperature (c) to 473 K (d). (Sanvitto, D. et al., Surf. Sci., 452, 191, 2000.) A perylene film deposited on SiO2 at room temperature (e) grows as two dimensions while the addition of a SAM of OTS to the substrate (f) causes the transition to three dimensions. (Park, D.S. et al., J. Vac. Sci. Technol. B, 23, 926, 2005.)

OTS is a highly hydrophobic layer commonly utilized in OTFT fabrication. The transition from two to three dimensions for PTCDA on Si also occurs by increasing substrate temperature [35].

5.1.3.1 THERMODYNAMIC DRIVING FORCE The admolecules in the first layer displayed in Figure 5.1.8 will diffuse around on the surface of the substrate until they desorb, join an existing cluster, or create a new nucleus. The creation of a new nucleus occurs when several molecules attach while diffusing on the substrate. Of course there is a complication in the fact that the free energy of a cluster with j molecules will be lowered for each particle taken from the vapor phase if it is thermodynamically favorable, but the free energy will

8080_C005.fm Page 352 Friday, April 6, 2007 10:25 AM

352

Organic Field-Effect Transistors

2

4

1

5 3

FIGURE 5.1.8 The processes involved in first layer nucleation and evolution shown for the case of pentacene: (1) adsorption; (2) desorption; (3) surface diffusion; (4) nucleation; and (5) attachment.

increase due to a term derived from surface tension considerations. The gain in free energy of a j-cluster over individual molecules is given by [1,4,6]: ΔG = − j Δμ + j 2 / 3 X

(5.1.5)

where Δμ is defined in Equation 5.1.3 and X is a geometrical term that takes into account the various surface and interaction energies. This makes it possible for clusters below some critical size to be energetically unfavorable and hence unstable with respect to disassociation. Depending on the shape of the crystal, the surface energy terms, and the supersaturation, there exists a value of j for which the free energy is at maximum and therefore the attachment of one more molecule will make the crystal stable. The size of the island corresponding to this maximum is known as the critical island size, denoted by i*. The value of j2/3X is determined by the relevant interaction potentials between the molecules and the substrate. The details are worked out explicitly in Markov [3] and Taylor et al. [16]. This formalism has been continued by Verlaak et al. [15] for the case of several important organic semiconductors (pentacene, tetracene, and perylene) grown on inert substrates. They show that the transition from three- to two-dimensional nucleation is possible when Δμ (Equation 5.1.3) is greater than a critical value [15]: Δμ cr = ψ c − ψ mol −sub

(5.1.6)

where ψc is the strength of the interatomic interaction potential in the direction normal to the surface and ψmol–sub is the strength of the interaction potential between a molecule and the subsrate. If ψc > ψmol-sub, then the thermodynamic driving force, Δμ, must be greater than twice Δμcr because, at this value, the three-dimensional nucleus that is most favorable is one monolayer high (i.e., two dimensions) [15].

5.1.3.2 RATE EQUATIONS: MICROSCOPIC In this section the dynamics of film growth will be discussed in terms of the steps mentioned earlier: adsorption, desorption, diffusion, attachment, and nucleation. The dynamics of the first monolayer can be explained by calculating the density of islands

8080_C005.fm Page 353 Friday, April 6, 2007 10:25 AM

Vacuum Evaporated Thin Films

353

that consist of j molecules, nj (referred to as j-clusters), as they nucleate and evolve in time. The details were originally worked out by Zinsmeister [36] in 1966 and have been continually developed over the years; they are summarized in various review articles and textbooks for inorganics [1,5]. Because this method is concerned with lateral distributions, the results from scanning probe microscopy, LEEM, or TEM can be used to understand the dynamics and compare with theory. Through the use of the dynamic scaling assumption [22], one can extract valuable information about the nucleation and diffusion rates by simply measuring the number of islands of a given size as a function of coverage and scaling this result with an exponent related to the critical island size and the diffusion constant. To understand the film evolution over several monolayers (sometimes even greater than 100 monolayers), it helps to consider the fractional coverage of each layer, rather than the individual islands in each layer, because the details and experimental methods become daunting otherwise. Comparing these theories with surfacesensitive scattering data such as reflection high-energy electron diffraction (RHEED) or x-ray scattering in the “anti-Bragg” configuration can yield information on the coverage of each layer with time, the critical coverage for coalescence, the island shape, and the presence and magnitude of an Ehrlich–Schwoebel barrier [37,38]. 5.1.3.2.1 Rate Equation Elements For the development of the first monolayer, we will start by considering the areal density of stable islands and single molecules diffusing on the substrate. Islands with j > i* will be considered immobile on the substrate, but can join with other islands through a process called coalescence whereby the islands grow and merge with each other. Figure 5.1.8 shows the development of the first monolayer using representations of pentacene, but it is also viable for other organic semiconductors with some consideration for anisotropic binding taken into account. Here we see molecules impinging on the substrate at the deposition rate, F, diffusing around on the substrate, and either desorbing at a fixed rate, Fdes, nucleating a new stable nucleus, or attaching to an existing j-cluster. From the preceding considerations, the simplest set of rate equations can be written as [1,5]: n1 = F −

n1 − ns τa

n1< j ≤i = 0 ns = 2U1 −

∑U

j

(5.1.7)

(5.1.8) =U i − U c

(5.1.9)

j =i

The first term in Equation 5.1.7 represents the rate of impinging admolecules arriving from the vapor phase. From this number of admolecules, some will be lost

8080_C005.fm Page 354 Friday, April 6, 2007 10:25 AM

354

Organic Field-Effect Transistors

to re-evaporation with a residence time on the substrate, τa. The admolecules that do not re-evaporate will form or join a stable island at a rate defined by the last term in Equation 5.1.7 as described in Equation 5.1.9; here Uj is the rate of capture of monomers by j-clusters. Equation 5.1.8 is a consequence of the positive free energy of an island smaller than the stable nucleus. The Us on the far right side of Equation 5.1.9 represent the rates at which admolecules join stable islands and islands coalesce by growing larger, which lowers the island density. The ns is equal to the density of stable islands of any size, j ≥ i, ns =

∑ jn

j

j ≥i

The first term in Equation 5.1.9 has a special significance because it represents the nucleation rate. This is commonly represented by J and is equal to [1,5]: J = σ i Dn1ni

(5.1.10)

where σi is known as the “capture number” and is equal to the rate at which diffusing admolecules attach to a critical island and D is the diffusion constant for admolecules on the surface. The diffusion constant is usually taken to be temperature dependent through the Arrhenius relation, D = D0exp(–ED/kBT), where ED is the energy barrier for surface diffusion. The value of D0 is also dependent on the interaction between the substrate and the admolecules. If a film is deposited at a temperature such that F ≈ Fdes, then the number of admolecules remains constant because a film in local equilibrium with its vapor will have a constant areal density of subcritical islands of j molecules, nj, relative to admolecules diffusing on the substrate n1 through the Walton Equation [1,5]: n j = ( n1 )

j

⎛

⎞

∑C (m)exp ⎜⎝ Ek (Tm) ⎟⎠ j

j

m

(5.1.11)

B

where the Cj is a constant determined by the entropy of j-clusters in some configuration, m, and Ej is the binding energy of the cluster. Equation 5.1.11 is a consequence of detailed balance. The number of stable nuclei, ns, can be solved from Equations 5.1.7 through 5.1.11 under certain conditions and with some assumptions. For instance, at high temperatures the residence time on the substrate will be very short compared with the capture rate. This is known as “incomplete condensation” and is often valid at extremely low coverages (known as “initially incomplete”). The other extreme is seen at low temperatures (high rates), where the probability for desorption is negligible and the second term in Equation 5.1.7 goes to zero. This regime is known as “complete” and is often true for the growth conditions used in OTFT fabrication.

8080_C005.fm Page 355 Friday, April 6, 2007 10:25 AM

Vacuum Evaporated Thin Films

355

TABLE 5.1.1 Functional Form of the Maximum Island Density Dependences on Deposition Rate and Temperature for Various Growth Regimes and Modes Regime Incomplete: p E Complete: p E

Three dimensional

Two dimensional

2i*/5 (2/5)(Ei + i*Ea)

i*/2 (Ei + i*Ea)

i*/(i* + 2.5) (Ei + i*Ed)/(i* + 2.5)

i*/(i* + 2) (Ei + i*Ed)/(i* + 2)

In either limit, the number of stable nuclei is found to depend on the incident rate and the substrate temperature through [1,5]: ns ≈ F p eE / kBT

(5.1.12)

where the values of p and E are dependent on the particular growth regime. These values are shown in Table 5.1.1 and are related to the critical island size and the mode of growth (two or three dimensions). Pratontep et al. [39,40] grew submonolayers (0.2–0.4 ML) of pentacene on various insulating substrates valid for OTFTs. Their data show a near-linear dependence for the nucleation density on the deposition rate (p ~ 1.16 for both SiO2 and PMMA) and an Arrhenius behavior is observed for nucleation density versus inverse substrate temperature. From this observed behavior, they concluded that under their deposition conditions, the film was evolving under the incomplete condensation regime [39,40]. Thus, a critical island size of i* = 2 corresponding to a nucleus composed of three or more molecules was stable. The unit cell is composed of two molecules, so the stable nucleus is anything larger than one unit cell. The analysis also afforded an estimate of the nucleation activation energy, E, from Equation 5.1.12 of 0.78 and 0.34 eV for pentacene deposited on SiO2 and PMMA, respectively. 5.1.3.2.2 Dynamic Island Size Distribution Another route to extract the critical island size, i*, and the diffusion constant, D, is to use images generated by various scanning probe measurements or TEM measurements on submonolayer films together with the dynamic scaling assumption [22] rather than by a solution to the rate equations with some assumptions about the relative magnitudes of diffusion, capture, and re-evaporation. In Figure 5.1.9a, a log–log plot of the densities of admolecules and stable nuclei is plotted versus total coverage for a specific ratio of the diffusion constant and the rate R = D/F = 108 (from Amar et al. [41]).

8080_C005.fm Page 356 Friday, April 6, 2007 10:25 AM

356

Organic Field-Effect Transistors

(a)

(b)

10–2

101 R = 108

Scaled densities

N1

N, N1

10–4

N

10–6 L 10–8

10–4

I 10–3

10–2 θ

A

10–1

108

100

105

10–1

109

10–2

N1

105

N

10–3

C

R = 105 – 109 10–4

10–2

100

10–1 θR

100

101

102

1/2/[In(2Rθ)]1/2

FIGURE 5.1.9 (a) Comparison of the densities of admolecules and stable nuclei as a function of coverage showing the different growth regimes. (b) Scaled densities and coverage from (a). (Amar, J.G. et al., Phys. Rev. B, 50, 8781, 1994.)

As can be seen, there are four distinct regions that correspond to a low coverage (L), an intermediate region (I), an aggregation region (A), and a coalescence region (C). In the low-coverage regime, there are few stable nuclei, but many admolecules diffusing on the surface. The intermediate regime, island density, and island size are increasing, which leads to a decreasing population of admolecules. The aggregation regime represents the stage where the number of stable nuclei remains constant. In the coalescence regime, the number of islands begins to decrease as they merge, thus allowing admolecules to diffuse around on top of the first monolayer and increasing their density. In Figure 5.1.9b, a plot of the scaled island densities N 1 = n1 R

ln(2 Rθ)

and N = ns R for the system studied in Amar et al. [41]) versus a scaled coverage for several values of R. This shows that the behavior under various growth conditions can be scaled across the various growth regimes. The dynamic scaling assumption relates the island density ns =

∑n

j

j ≥i*

to the diffusion constant and the deposition rate via [22,41,42]:

8080_C005.fm Page 357 Friday, April 6, 2007 10:25 AM

Vacuum Evaporated Thin Films

357

ns = C ( R )

−χ

(5.1.13)

where C is a proportionality constant and χ is the scaling exponent equal to i*/(i* + 2). Scaling theory goes on to show that the average island size, S, is a function of layer coverage, θ = νt by [22,41,42]:

S=

∑ jn j ≥i

ns

j

=

χ θ − n1 ≈ ( R) θ ns

(5.1.14)

Now assume further that S is the only characteristic scaling variable. That is, let the number density of j-clusters scale with S by [22,41,42]: ⎛ j⎞ n j = θS −2 f ⎜ ⎟ ⎝S⎠

(5.1.15)

where the function, f, is a scaling function that satisfies

∫

∞

f ( x ) xdx = 1

0

This dynamic scaling theory has recently been applied to submonolayer films of pentacene deposited on room temperature SiO2 by Ruiz et al. [43] and their results are displayed in Figure 5.1.10. Figure 5.1.9a displays the number of nuclei of a certain size versus the island size in micrometers for several coverages between 0.18 and 0.42 ML. The analysis of these data was carried out under the assumption of negligible desorption and with a functional form for the scaling function of [43]: f ( x ) = Ciu i* exp(−bii * u1/bi )

(5.1.16)

where the Cs and bs are geometrical normalization equations and i* is the critical island size. Figure 5.1.10b shows the results of scaling the island size distribution. Fits to the universal curves for critical cluster sizes of one and three are shown, indicating that three is a better choice. The temperature dependence of the critical island size for pentacene on SiO2 was explored by Tejima et al. [44] and the results are displayed in Figure 5.1.11. The same scaled island size distributions are shown in Figure 5.1.11a for four different substrate temperatures at a constant thickness of ~0.2 ML. The curves corresponding to critical island sizes of one, two, and three are shown and it is clear that the different substrate temperatures used result in different critical island sizes. The scaled island densities for thick pentacene layers are plotted as a function of

8080_C005.fm Page 358 Friday, April 6, 2007 10:25 AM

358

Organic Field-Effect Transistors

5 0 = 18% 0 = 27% θ = 41% 0 = 42%

N8(θ) (μm–4)

4

3

θ = 0.18 20 μm

2

1 θ = 0.42 0 0.0

0.5

1.0

1.5

2.0

s (μm2) (a) 1.6

θ = 18% 0 = 27% θ = 41% 0 = 42%

(b)

1.4 1.2 N8(θ)S2/θ

fi i = 3 1.0

fi i = 1

0.8 0.6 i=1

0.4

i=3

0.2 0.0 0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

s/S (b)

FIGURE 5.1.10 (a) Island size distribution at different submonolayer coverages; (b) the scaled island-size distributions implying a critical island size of three molecules. (Ruiz, R. et al., Phys. Rev. Lett., 91, 136102, 2003.)

inverse temperature in Figure 5.1.11b. It is clear that there is a transition temperature of ~250 K from a critical island size of two to three. 5.1.3.2.3 Beyond the First Layer: Birth–Death Models Now that the first monolayer has been described in the kinetic atomistic framework, it is important to ask which mechanisms are involved when upper layers are considered. There have been several attempts to extend the rate-equation approach to

8080_C005.fm Page 359 Friday, April 6, 2007 10:25 AM

Vacuum Evaporated Thin Films

1010

342K

1.2 i* = 2

N(s) S2/θ

1.0

i* = 1

0.8 0.6

5 μm

0.4

223K 249K 304K 342K

0.2 0.0 0.0

2.0 s/S

3.0

300

200 109 100

i* = 3 108

1.0

i* = 2

χ2

i* = 3

Saturated Island Density (/cm2)

1.4

359

0 0.003

0.004 1/T (K–1)

FIGURE 5.1.11 Effects of temperature on the scaling and the critical island size for pentacene films: (a) submonolayer; (b) multilayer. (Tejima, M. et al., Appl. Phys. Lett., 85, 3746, 2004.)

incorporate the subsequent layers, but with little success due to the increased number of parameters needed to quantify the system. However, here the attention will focus on mechanisms and models that can be tested with surface-sensitive scattering techniques: anti-Bragg x-ray scattering and reflection high-energy electron diffraction. These techniques are able to explore the interlayer transport, but yield only limited in-plane information. The activation potential profile for a diffusing molecule in the vicinity of molecular steps considered is shown in Figure 5.1.12. The shaded atoms (read molecules) are representative of diffusing molecules approaching a step edge from the lower layer or the layer above. The rates at which the molecules join the island could be unequal, as represented by the κs. The two potential energy profiles represent the existence and absence of an extra barrier generated by an inequality in attachment rates. This is the famous Ehrlich–Schwoebel barrier, EES [37,38], and may be defined as the difference between the step-down barrier and the diffusion activation energy. This barrier controls the ease at which layer-by-layer growth occurs. That is, if the additional barrier is exactly equal to 0, then the probability that a molecule joins the layer below is equal to 0.5 so that the edge will be ineffective in retaining molecules on top of the island. This is the condition that is favorable for layer-bylayer growth, but it is not sufficient. Also displayed in Figure 5.1.12 is the activation energy for diffusion along the terraces. The value of the Ehrlich–Schwoebel barrier is expected to be highly dependent on the diffusion constant (substrate dependent) and the nature of the interaction potentials. In order to model film growth accurately beyond the first monolayer, one can use the so-called birth–death models that track the relative coverages of each layer as well as interlayer transport [29]. Figure 5.1.13 highlights the major components of these models. In these models, the molecules that land on top of the nth layer (where the 0th layer is the substrate) may diffuse and incorporate at the step edge of the n + 1 layer or transfer down to the top of the n – 1 layer to be incorporated into the nth layer (see Figure 5.1.13). Thermal desorption of molecules that have

8080_C005.fm Page 360 Friday, April 6, 2007 10:25 AM

360

Organic Field-Effect Transistors

K– K+

ED

EES

FIGURE 5.1.12 A monomolecular step as seen by an approaching admolecule from above or below. Potential profiles both without (middle) and with (bottom) and additional Ehrlich– Schwoebel barrier (as described in references 37 and 38).

attached to a stable nucleus as well as transfer up to the n + 1 layer are neglected. The former is usually a good assumption for substrate temperatures used in OTFT fabrication, while the latter is usually the case due to the asymmetry in the Ehrlich–Schwoebel barrier. Cohen proposed several equation forms to deal with the rate of change of the coverage of the layers θn. In this chapter, only the distributed model will be described because it has been used to study the dynamics of pentacene growth [45,46]. The equations for the coverages can be given by [29]: dθn = ( F − Fdes ) (θn−1 − θn ) + Fα n (θn − θn+1 ) − Fα n−1 (θn−1 − θn ) dt

(5.1.17)

where the αns in Equation 5.1.17 track the probability for a molecule diffusing on top of layer n to hop down to join layer n. The admolecules can only step down when they sit at a step edge, so αn can be expressed in terms of the island perimeter, dn, of the nth layer as shown in Figure 5.1.13. Cohen et al. suggest a simple form for αn given by [29]:

α n = An

dn d n + d n +1

(5.1.18)

8080_C005.fm Page 361 Friday, April 6, 2007 10:25 AM

Vacuum Evaporated Thin Films

361

Fdes

1

F 2 3

(a)

(b)

1st ML

dn+1

dn

Substrate

1st ML

(c) dn+1 dn

Substrate

FIGURE 5.1.13 Relevant growth parameters for growth beyond the first monolayer. (a) Side view of an evolving pentacene film showing the relevant rates: the deposition rate, F, the desorption rate, Fdes, downward transfer and diffusion, 1–3. Three-dimensional views of the growing islands highlighting the island perimeters, dn, of the first layer before percolation (b) and after percolation, where the gaps become important (c).

where αn = 0 if dn = 0, even if dn+1 = 0. The parameter An accounts for the probability for a molecule sitting at the step edge of layer n + 1 to step over the Ehrlich– Schwoebel barrier [37,38] and join the nth layer. The limits of the model are perfect layer-by-layer growth for An = 1 and nondiffusive growth for An = 0. To get an expression for dn we need to explore the relationship between the cross-section of the islands in a given layer with film coverage. If s represents the area of a two-dimensional cluster and D its fractal (Hausdorff) dimension, then the island cross-section is proportional to sp1, where p1 = 1/D. D = 2 corresponds to a compact island, but for pentacene that grows by diffusion limited aggregation (DLA), D ~ 5/3. Noting from before that, prior to coalescence the island density, ns, remains constant and composed of identical islands, then the total cross-section of a layer becomes dn = nsSp1. Since θ = nsS, then dn ∝ θp1. Similarly, if after coalescence the number of holes in the film also remains constant, then the perimeter of each layer can be related to its coverage as shown in Figure 5.1.12b and c by [29]: ⎧⎪ θn p 1 dn ∝ ⎨ ⎪⎩ 1 − θn

(

)

p2

, θn ≤ θc , θn ≥ θc

(5.1.19)

The two exponents are interrelated by θcp1 = (1 – θc)p2, where θc indicates the coverage at which coalescence occurs.

8080_C005.fm Page 362 Friday, April 6, 2007 10:25 AM

362

Organic Field-Effect Transistors

(a)

4 2

(c)

5 4 3 2 0

(b)

1 2 3 4 Thickness (Monolayers)

12

Intensity (Counts/105)

Intensity (Counts/105) Intensity (Counts/105)

HTS Intensity (Counts/105)

SiO2

(a)

5

(b)

8

25°C

4

(d) 8

60°C

4 0

1 2 3 4 Thickness (Monolayers)

5

(c)

FIGURE 5.1.14 (a) X-ray anti-Bragg scattering and the results of the fit using Equations 5.1.17 and 5.1.18 to pentacene films grown on SiO2 and HTS held at 25 and 60°C. Representation of the pentacene molecular dynamics extracted from the fits for (b) SiO2 and (c) HTS. (Mayer, A.C. et al., Phys. Rev. B, 73, 205307, 2006.)

Comparing the distributed growth model with anti-Bragg scattering experiments (top panel in Figure 5.1.14), Mayer et al. [45,46] found that, for pentacene molecules adsorbed on the first and second layer, the probability of downward transfer was dependent on the substrate and independent of temperature within the range from 25 to 60°C. The analyses of the extracted layer coverages are displayed in the bottom of Figure 5.1.14. On the left is the case for growth on SiO2 where the first and second layers are grown due to the step-down probability being exactly unity. However, as the film thickens, the probability for stepping down is reduced, leading to an increase in the Ehrlich–Schwoebel barrier of the order of 70 meV. For films grown on an alkylated self-assembled monolayer (right side of Figure 5.1.14), significant desorption of pentacene molecules from the substrate forced the growth mode towards the three-dimensional limit at elevated temperatures as is shown. Microscopically, this leads to a growing step front with a height greater than a single molecule, which will have a reduced growth velocity.

8080_C005.fm Page 363 Friday, April 6, 2007 10:25 AM

Vacuum Evaporated Thin Films

363

5.1.3.3 RATE EQUATIONS: MACROSCOPIC Evolving thin films exhibiting thickness fluctuations can be characterized by the root-mean-square (rms) roughness σ = 〈(h(x,y) – 〈h〉)2〉1/2, where 〈…〉 is a spatial average and h(x,y) is the local film thickness. If the term h(x,y) – h(x′,y′) is given by a normal distribution, then the surface is termed self-affine. Information about the lateral roughness fluctuations is needed because disparate films can exhibit the same rms roughness. If the lateral dimension sampled for the roughness calculation is given by L, then the rms roughness should scale as σ(L) ∝ LH, where H is referred to as the “static roughness” exponent [47]. Self-affine surfaces are also characterized by a height difference correlation function (HCDF) as given by:

( ) (

)

g ( R ) = ⎡⎣ h x, y − h x ', y ' ⎤⎦

2

∝ R2H

(5.1.20)

with R=

( x − x ') + ( y − y ') 2

2

Equation 5.1.20 is valid for self-affine surfaces up to a correlation length (ξ), such that R >> ξ ⇒ g(R) → 2σ2. The treatment, so far, has considered static spatial variations only, but film growth is dynamic and temporal fluctuations must also be considered [30]. Family and Vicsek [48] considered the evolution of the lateral and perpendicular fluctuations in their “dynamic scaling” theory. According to dynamic scaling, the rms roughness evolves with time as σ(t) ∝ tβ if the average film height is proportional to time (〈h〉 ~ t). Combining this with the lateral fluctuations yields: ⎛ t ⎞ σ ( L, t ) = LH F ⎜ H /β ⎟ ⎝L ⎠

(5.1.21)

where F(x) → 1 as x → ∞ and σ(t) ∝ tβ as x → 0. Central to Equation 5.1.21 is a scaling of the correlation length with time ξ ∝ t1/z, where z = H/β is known as the dynamic scaling exponent. The scaling exponents have been calculated theoretically using a variety of assumptions about the ratelimiting factors [49]. The numerical estimate by Kardar–Parisi–Zhang (KPZ) ignores the effects of surface diffusion and estimates H, β, and z as 0.385, 0.24, and 1.61, respectively [49]. Other estimates, more applicable to PVD growth, incorporate diffusion and yield values of 2/3, 1/5, and 10/3, respectively [49]. Durr et al. demonstrated the first application of scaling laws to organic systems for the deposition of diindenoperylene (DIP) on SiO2 [50]. They utilized x-ray reflectivity, AFM, and diffuse x-ray scattering to analyze the temporal and spatial fluctuations of the surface roughness. To do this, they used an extended version of

8080_C005.fm Page 364 Friday, April 6, 2007 10:25 AM

364

Organic Field-Effect Transistors

Equation 5.1.20 that includes a steepening exponent λ: g(R) ∝ t2λR2H [50]. Fitting the scaling equations to their samples yields 1/z = 0.92 ± 0.20, H = 0.684 ± 0.05, β = 0.748 ± 0.07, and λ = 0.17 or 0.07 depending on how H was calculated. The values presented do not match the existing theories mentioned previously; instead, the behavior of the film growth is referred to as rapid roughening. Models are being developed to explain this phenomenon.

5.1.4 EFFECTS OF THE SUBSTRATE The morphology and OTFT performance of the active layer are dependent on several external parameters in addition to vacuum pressure, deposition rate, and substrate temperature. Specifically, the nature and strength of the interaction potential between the adsorbate and the substrate will greatly influence the arrangement of anisotropic molecules when deposited and even control, to some extent, whether the islands will nucleate and grow in as two or three dimensions. The surface roughness of the substrate as well as its surface tension, γ, will greatly influence morphology (as mentioned in Section 5.1.1.2) and device performance.

5.1.4.1 EFFECT

OF

SURFACE ENERGY

The effects of surface tension and the interaction potential between the substrate and pentacene are demonstrated in Figure 5.1.15. The left column shows AFM micrographs for pentacene deposited on poly(imide-siloxane) with varying amounts of siloxane in the polymer backbone as reported by Yang et al. [51]. The variation in siloxane shifts the surface energy from 48 mJm–2 (a) to 30 mJm–2 (c) without varying the surface roughness appreciably. As can be seen, the grain size decreases with surface energy as expected from the thermodynamic explanation of film growth (see Section 5.1.1.2). The authors also show that the mobility follows the opposite trend (not shown). The right column of Figure 5.1.15 shows the dependence of the molecule–substrate interaction as discussed originally in Verlaak et al. [15] and continued in Steudel et al. [52]. The value of the molecule–substrate interaction is plotted for a variety of common OTFT substrates and pentacene in Figure 5.1.15d [52]. In Figure 5.1.15e, the effects on the growth mode of pentacene films is shown for pentacene growth on OTS and on (heptadecafluoro-1,1,2,2-tetrahydrodecyl)trichlorosilane (FDTS). As shown, the lower interaction potential between pentacene and FDTS forces the growth mode to three dimensions.

5.1.4.2 EFFECT

OF

SURFACE ROUGHNESS

Several groups have explored the effects of surface roughness on OTFTs. Figure 5.1.16 shows the effects of increasing surface roughness on pentacene film morphology. The pentacene grain sizes for growth on SiO2 with an rms roughness of 0.2 nm (a) and rough SiO2 with an rms roughness of 1.5 nm (b) are dramatically different [53]. Figure 5.1.16c and d show the morphology of pentacene films grown on OTScoated SiO2 [54]. The last two panes in Figure 5.1.16 show the effects on pentacene film morphology of the surface roughness of SiN substrates (prepared by plasma

8080_C005.fm Page 365 Friday, April 6, 2007 10:25 AM

Vacuum Evaporated Thin Films

365

(a) (d)

Φ = 0.25Å/s

Pentacene

ψmol-sub [eV]

0.12 OTS, PVP

BUTS, UETS

0.08

CUTS, PTS, PETS 0.04

FDTS

0.00 (b)

300

320

340

360

Tsub [K] 10 nm

(e)

(b)

(c) 10 nm

100 μm

(a)

1 μm

(c)

FIGURE 5.1.15 Effects of the substrate’s surface energy as it decreases from (a) to (c) for pentacene deposited on poly(imide-siloxane). (Yang, S.Y. et al., Adv. Func. Mat., 15, 1806, 2005.) The interaction potential between the molecule and the substrate for various substrates (d) and the dependence on the growth mode on the magnitude of the interaction (e). The twodimensional connected regime is for pentacene grown on OTS and the three-dimensional portion is grown on FDTS. (Steudel, S. et al., Appl. Phys. Lett., 85, 5550, 2004.)

enhanced chemical vapor deposition) [55]. The standard SiN surface had an rms roughness of 0.5 nm (Figure 5.1.16f), while the SiN in Figure 5.1.16e was atomically smooth. The second group in Figure 5.1.16 shows the effects of surface roughness for pentacene deposited on OTS-coated SiO2.

5.1.5 OUTLOOK In the preceding sections we outlined some models from the inorganic film growth community, and early attempts to adapt and apply them to organics. This process is still at its infancy because the primary identifying characteristic of organics — namely, their anisotropic shape — has not been taken into account. This will require

8080_C005.fm Page 366 Friday, April 6, 2007 10:25 AM

366

Organic Field-Effect Transistors

Increased roughness (a)

(b)

(c)

(d)

SiO2

OTS-coated SiO2

1 μm (e)

(a) 2.5 μm

(b) (f )

2.5 μm

SiN

Sample B 70°C, 0.5 Å/s, smooth

Sample A 70°C, 0.5 Å/s, standard

FIGURE 5.1.16 Effects of substrate surface roughness on pentacene film morphology. Growth on smooth (a) and rough (b) SiO2 (Fritz, S.E. et al., J. Phys. Chem. B, 109, 10574, 2205); smooth (c) and rough (d) OTS-coated SiO2 (Steudel, S. et al., Appl. Phys. Lett., 85, 4400, 2004); and smooth (e) and rough (f) SiN. (Knipp, D. et al., Appl. Phys. Lett., 82, 3907, 2003.)

a generalization of growth models and will lead not only to better control of the morphology of organic films, but also to new and exciting growth in physics.

REFERENCES 1. Venables, J.A., Spiller, G.D.T., and Hanbucken, M., Nucleation and growth of thin films, Rep. Prog. Phys., 47, 399, 1984.

8080_C005.fm Page 367 Friday, April 6, 2007 10:25 AM

Vacuum Evaporated Thin Films

367

2. Smith, D.L., Thin-film deposition: Principles and practice, McGraw–Hill, Inc., New York, 1995. 3. Markov, I.V., Crystal growth for beginners: Fundamentals of nucleation, crystal growth, and epitaxy, World Scientific, Singapore, 1995. 4. Rosenfeld, G., Poelsema, B., and Comsa, G., Epitaxial growth modes far from equilibrium, in The Chemical Physics of Solid Surfaces, Vol. 8, ed. D.A. King and D.P. Woodruff, 66–99, Elsevier Science, New York, 1997. 5. Venables, J., Introduction to surface and thin film processes, Cambridge University Press, Cambrige, 2000. 6. Silinsh, A. and Cápek, V., Organic molecular crystals: interaction, localization and transport phenomena, AIP Press, New York, 1994. 7. Schreiber, F., Organic molecular beam deposition: Growth studies beyond the first monolayer, Phys. Sta. Sol. A, 201, 1037, 2004. 8. Ruiz, R. et al., Pentacene thin film growth, Chem. Mater., 16, 4497, 2004. 9. Locklin, J. and Bao, Z., Effect of morphology on organic thin film transistor sensors, Anal. Bioanal. Chem., 384, 336, 2006. 10. Witte, G. and Woll, C., Growth of aromatic molecules on solid substrates for applications in organic electronics, J. Mater. Res., 19, 1889, 2004. 11. Headrick, R.L., Malliaras, G.G., Mayer, A.C., Deyhim, A.K., and Hunt, A.C., Eighth international conference on synchrotron radiation instrumentation, ed. T. Warwick, H. Stohr, H.A. Padamore, and J. Arthur, AIP Conf. Proc. (AIP, Melville, NY, 2004), 705, 1150, 2004. 12. Kossel, W., Nachricten der Gesellschaft der Wissenschaften Göttingen, Mathematisch-Physikalische Klasse, Band 135, 1927. 13. Burton, W.K., Cabrera, N., and Frank, F.C., The growth of crystals and the equilibrium structure of the surfaces, Phil. Trans. R. Soc. Lond. A., 243, 299, 1951. 14. Wulff, G. and Kristallogr, Z., On the question of the rate of growth and dissolution of crystal surfaces, Mineral, 34, 449, 1901. 15. Verlaak, S., Steudel, S., Janssen, D., Deleuze, M.S., and Heremans, P., Nucleation of organic semiconductors on inert substrates, Phys. Rev. B., 68, 195409, 2003. 16. Taylor, J.E., Cahn, J., and Handwerker, C.A., Geometrical models of crystal growth, Acta Metal Mater., 40, 1443, 1992. 17. Sadowski, J.T., Nagao, T., Yaginuma, S., Fujikawa, Y., Al-Mahboob, A., Nakajima, K., Sakurai, T., Thayer, G.E., and Tromp, R.M., Thin bismuth film as a template for pentacene growth, Appl. Phys. Lett., 86, 073109, 2005. 18. Koma, A., New epitaxial growth method for modulated structures using Van der Waals interactions, Surf. Sci., 267, 29, 1992. 19. Forrest, S.R., Burrows, P.E., Haskal, E.I., and So, F.F., Ultrahigh-vacuum quasiepitaxial growth of model van der Waals thin films. II. Experiment, Phys. Rev. B, 49, 11309, 1994. 20. Maganov, S. N. and Whangbo, M.-H., Surface analysis with STM and AFM, VCH, New York, 1996; Giessible, F.J., Advances in atomic force microscopy, Rev. Mod. Phys., 75, 949, 2003. 21. Umbach, C.C. and Blakely, J.M., Development of a sub-picoampere scanning tunneling microscope for oxide surfaces, Appl. Surf. Sci., 175–176, 746, 2001. 22. Vicsek, T. and Family, F., Dynamic scaling for aggregation of clusters, Phys. Rev. Lett., 52, 1669, 1984. 23. Warren, B. E., X-ray diffraction, Addison–Wesley, Reading, MA, 1969.

8080_C005.fm Page 368 Friday, April 6, 2007 10:25 AM

368

Organic Field-Effect Transistors

24. Ruiz, R., Mayer, A.C., Malliaras, G.G., Nickel, B., Scoles, G., Kazimirov, A., Kim, H., and Headrick, R.L., Structure of pentacene thin films, Appl. Phys. Lett., 85, 4926, 2004. 25. Marra, W.C., Eisenberger, P., and Cho, A.Y., X-ray total-external-reflection–Bragg diffraction: A structural study of the GaAs-Al interface, J. Appl. Phys., 50, 6927, 1979. 26. Bauer, E., Low energy electron microscopy, Rep. Prog. Phys., 57, 895, 1994. 27. Meyer Zu Heringdorf, F.-J., Reuter, M.C., and Tromp, R.M., Growth dynamics of pentacene thin films, Nature, 412, 517, 2001. 28. Balluffi, R.W., Allen, S.M., and Carter, W.C., Kinetics of materials, Wiley Interscience, New York, 2005. 29. Cohen, P.I., Petrich, G.S., Pukite, P.R., Whaley, G.J., and Arrott, A.S., Birth–death models of epitaxy: I. Diffraction oscillations from low index surfaces, Surf. Sci., 216, 222, 1989. 30. Krim, J. and Palasantzas, G., Experimental observations of self-affine scaling and kinetic roughening at submicron length scales, Int. J. Mod. Phys. B, 9, 599, 1995. 31. Ruiz, R., Papadimitratos, A., Mayer, A.C., and Malliaras, G.G., Thickness dependence of mobility in pentacene thin-film transistors, Adv. Mater., 17, 1795, 2005. 32. Wang, G.Z., Luo, Y., and Beton, P.H., High-mobility organic transistors fabricated from single pentacene microcrystals grown on a polymer film, Appl. Phys. Lett., 83, 3108, 2003. 33. Sanvitto, D., De Seta, M., and Evangelisti, F., Growth of thin C60 films on hydrogenated Si(100) surfaces, Surf. Sci., 452, 191, 2000. 34. Park, D.S., Kang, S.J., Kim, H.J., Jang, M.H., Noh, M., Yoo, K.-H., Whang, C.N., Lee, Y.S., and Lee, M.H., Characteristics of perylene-based organic thin-film transistor with octadecyltrichlorosilane monolayer, J. Vac. Sci. Technol. B, 23, 926, 2005. 35. Möbus, M. and Karl, N., The growth of organic thin films on silicon substrates studied by x-ray reflectometry, Thin Solid Films, 215, 213, 1992. 36. Zinsmeister, G., A contribution to Frenkel’s theory of condensation, Vacuum, 16, 529, 1966; Zinsmeister, G., Theory of thin film condensation. Part B: Solution of the simplified condensation equation, Thin Solid Films, 2, 497, 1968. 37. Ehrlich, G. and Hudda, F.G., Atomic view of surface self-diffusion: Tungsten on tungsten, J. Chem. Phys., 44, 1039, 1966. 38. Schwoebel, R.L. and Chipsey, E.J., Step motion on crystal surfaces, J. Appl. Phys., 37, 3682, 1969. 39. Pratontep, S., Brinkmann, M., Nüesch, F., and Zuppiroli, L., Correlated growth in ultrathin pentacene films on silicon oxide: Effect of deposition rate, Phys. Rev. B, 69, 165201, 2004. 40. Pratontep, S., Nuesch, F., Zuppiroli, L., and Brinkmann, M., Comparison between nucleation of pentacene monolayer islands on polymeric and inorganic substrates, Phys. Rev. B, 72, 085211, 2005. 41. Amar, J.G., Family, F., and Lam, P.-M., Dynamic scaling of the island-size distribution and percolation in a model of submonolayer molecular-beam epitaxy, Phys. Rev. B, 50, 8781, 1994. 42. Stroscio, J.A. and Pierce, D.T., Scaling of diffusion-mediated island growth in ironon-iron homoepitaxy, Phys. Rev. B, 49, 8522, 1994. 43. Ruiz, R., Nickel, B., Koch, N., Feldman, L.C., Haglund, R.F., Jr., Kahn, A., Family, F., and Scoles, G., Dynamic scaling, island size distribution, and morphology in the aggregation regime of submonolayer pentacene films, Phys. Rev. Lett., 91, 136102, 2003.

8080_C005.fm Page 369 Friday, April 6, 2007 10:25 AM

Vacuum Evaporated Thin Films

369

44. Tejima, M., Kita, K., Kuno, K., and Toriumi, A., Study on the growth mechanism of pentacene thin films by the analysis of island density and island size distribution, Appl. Phys. Lett., 85, 3746, 2004. 45. Mayer, A.C., Ruiz, R., Headrick, R.L., Kazimirov, A., and Malliaras, G.G., Early stages of pentacene film growth on silicon oxide, Org. Elec., 5, 257, 2004. 46. Mayer, A.C., Ruiz, R., Zhou, H., Headrick, R.L., Kazimirov, A., and Malliaras, G.G., Growth dynamics of pentacene thin films: Real-time synchrotron x-ray scattering study, Phys. Rev. B, 73, 205307, 2006. 47. Krim, J. and Indekeu, J.O., Roughness exponents: A paradox resolved, Phys. Rev. E, 48, 1576, 1993. 48. Family, F. and Vicsek, T., Scaling of the active zone in the Eden process on percolation networks and the ballistic deposition model, J. Phys. A, 18, L75, 1985. 49. Kardar, M., Parisi, G., and Zhang, Y.-C., Dynamic scaling of growing interfaces, Phys. Rev. Lett., 56, 889, 1986; Villain, J., Continuum models of crystal growth from atomic beams with and without desorption, J. Phys. I France, 1, 19, 1991; Lai, Z.W. and Das Sarma, S., Kinetic growth with surface relaxation: Continuum versus atomistic models, Phys. Rev. Lett., 66, 2348, 1991. 50. Durr, A.C., Schreiber, F., Ritley, K.A., Kruppa, V., Krug, J., Dosch, H., and Struth, B., Rapid roughening in thin film growth of an organic semiconductor (diindenoperylene), Phys. Rev. Lett., 90, 016104, 2003. 51. Yang, S.Y., Shin, K., and Park, C.E., The effect of gate-dielectric surface energy on pentacene morphology and organic field-effect transistor characteristics, Adv. Func. Mat., 15, 1806, 2005. 52. Steudel, S., Janssen, D., Verlaak, S., Genoe, J., and Heremans, P., Patterned growth of pentacene, Appl. Phys. Lett., 85, 5550, 2004. 53. Fritz, S.E., Kelley, T.W. and Frisbie, C.D., Effect of dielectric roughness on performance of pentacene TFTs and restoration of performance with a polymeric smoothing layer, J. Phys. Chem. B, 109, 10574, 2205. 54. Steudel, S., De Vusser, S., De Jonge, S., Janssen, D., Verlaak, S., Genoe, J., and Heremans, P., Influence of the dielectric roughness on the performance of pentacene transistors, Appl. Phys. Lett., 85, 4400, 2004. 55. Knipp, D., Street, R.A., and Völkel, A.R., Morphology and electronic transport of polycrystalline pentacene thin-film transistors, Appl. Phys. Lett., 82, 3907, 2003.

8080_C005.fm Page 370 Friday, April 6, 2007 10:25 AM

8080_C005.fm Page 371 Friday, April 6, 2007 10:25 AM

5.2

Solution Deposition of Polymers

Hoichang Yang CONTENTS 5.2.1 Introduction................................................................................................371 5.2.2 Solution Casting and Spin-Coating to Form Self-Assembled Polymers.....................................................................................................374 5.2.2.1 Molecular Structure.....................................................................374 5.2.2.2 Effect of Solvent .........................................................................382 5.2.2.3 Processing Condition...................................................................387 5.2.3 Concluding Remarks..................................................................................392 References..............................................................................................................395

5.2.1 INTRODUCTION Various organic thin film transistors (OTFTs) containing semiconducting polymers with π-conjugated backbones for driving light-emitting diodes and prototype electronic polymer circuits are presently being demonstrated [1–3]. Unlike vacuum deposition, which is more costly and requires long pump-down time, solutiondeposition methods for semiconductor thin films can be done under ambient conditions on short timescales. As a result, they are compatible with large-area thin film fabrication and result in lower production cost per device. Two approaches have been used to obtain organic semiconductor thin films from solution: deposition of a soluble precursor from solution and subsequent conversion reaction direct deposition of a soluble semiconductor An example of a conversion reaction from a precursor to a polymer is shown in Figure 5.2.1 [4]. The conversion reaction for the precursor materials required for achieving high mobility is performed at a high temperature above 125°C, making it incompatible with low-cost plastic substrates, which typically have glass transition temperatures below 125°C. The additional conversion step is also time consuming. Accordingly, direct deposition of soluble semiconductors is more desirable.

371

8080_C005.fm Page 372 Friday, April 6, 2007 10:25 AM

372

Organic Field-Effect Transistors

HCl

H2 CH C S OCH3 Precursor polymer

n

Δ

S

C H

C H n

Poly(2, 5-thienylenevinylene)

FIGURE 5.2.1 Conversion reaction from a precursor polymer to poly(2,5-thienylenevinyle) (PTV).

Electrochemical polymerization has been used to synthesize and deposit polythiophenes directly into device channels [5,6]. In this case, deposited films tend to show low on/off ratios and field-effect mobility. Another interesting example involves solution deposition of an n-type ladder polymer using a proper Lewis acid acting as a chelating and a solubilizing group. After film fabrication, the Lewis acid is removed by extensive washing [7,8], resulting in highly ordered polymer thin films [8]. However, the remaining trace amount of Lewis acid yields unusual electrical characteristics and has to be completely removed in order to obtain high mobility. Accordingly, direct deposition of soluble semiconductors is more desirable. Since most semiconducting polymers with π–π-conjugated backbones are insoluble in common organic solvents, side chains have been incorporated into the molecular structure through various chemical substitutions [9–14]. For example, remarkably good solubility has been achieved for polythiophenes by substituting alkyl side chains at the 3-position of each thiophene ring because these side groups mitigate the strong π–π-interchain interaction between the polythiophene chains. Until now, various poly(3-alkylthiophene)s (PATs) [15–18], polythiophenes with end-functionalized side chains [14,19], block copolymers [20–22], combined with these polymers have been used for solution processable OTFT applications. Among these polymers, regioregular head-to-tail poly(3-hexylthiophene) (HT PHT) films have shown higher charge mobility (0.01 ~ 0.1 cm2V–1S–1) and reasonable on/off ratios (>100 in air and 106 in an inert atmosphere) in OTFTs [17,18,23,24]. The Langmuir–Blodgett (LB) technique [25,26] is another possible method to fabricate organic semiconductor thin films [27–33]. The molecular density at the air–water interface is manipulated with movable barriers in an LB film trough. By moving the barriers, the film density increases (compression) or decreases (expansion), which changes the surface tension between liquid and vapor phases, γlv, at the air–water interface. The film can then be transferred onto a substrate either as a monolayer or as multilayers. The molecules in the LB film are usually well aligned, but PHT is not amphiphilic, and it was difficult to obtain highly aligned mono- or multilayers on an air–water interface. Surfactant molecules, as well as nonamphiphilic polymers decorated with alkyl chains, can be incorporated to improve the film ordering, but at the same time these insulating molecules disrupted charge transport [29]. Conversely, amphiphilic polythiophenes [32–34] can self-assemble into π-stacked conjugated chains that form a very stable monolayer with a local structure optimized for high electrical conductivity. Generally, polymeric semiconductor thin films for OTFT applications can be fabricated through various direct solution-processing methods: spin-casting [15,17,18,35,36], drop-casting [15,23], screen printing [37,38], and ink jet printing

8080_C005.fm Page 373 Friday, April 6, 2007 10:25 AM

Solution Deposition of Polymers

373

[39–42], which allow for a single-step deposition and patterning of the active materials [43]. Both spin-casting and drop-casting are commonly used for the film deposition. The spin-casting method yields fast solvent evaporation, allowing less time for molecular ordering but more uniform films compared to drop-casting. These solution-deposited films can be annealed [15,44–48] or melt crystallized [24] afterwards to improve molecular ordering. Electrical performance of the OTFT devices is affected by the microstrucuture in these semiconductor thin films [17,23,24,35,36,44]. It has been recognized that intrachain and interchain (π-conjugated plane) transport takes place in polymer films, whereas transport in amorphous portion and between grain boundaries has a high energy barrier [49,50]. Therefore, it is important to achieve greater control of the solution deposition process that takes these polymers from a soluble state into a polycrystalline thin film. Specifically, the desire to make nanoscale devices for future electronic applications further emphasizes a preferential orientation of π-conjugated crystal planes as the main hole-transporting path of p-type semiconductors. Understanding the crystal growth mechanisms of these solution-processable polymers is required to control the self-assembled structure during film formation. Most studies on solution-deposited semiconducting polymer thin films for OTFTs pertain to PATs [6,10,15,23,24,27,30,35,36,44,46–48,51–81]. The key factors affecting self-assembly of these polymers during solution processing are as follows: molecular properties: π-conjugated polymer backbone, side chain type, regioregularity, number average molecular weight ( M n , MW), molecular weight distribution ( M w / M n , PDI) solvent nature: volatility (boiling point), solubility of the polymer processing condition: substrate treatment, solvent evaporation rate, thermal annealing (Figures 5.2.2 and 5.2.3), and nature of substrate surface Mesoscale crystalline morphology, crystallinity, and molecular orientation in these deposited thin films strongly depend on molecular properties [17,18], chemical nature of the solvent, and processing condition, resulting in very different field-effect mobilities [15,23,36]. Specifically, due to heterogeneous surface-induced (epitaxy) crystal growth as a nature of semicrystalline polymers, fine control of substrate properties and solvent evaporation rate tends to yield favorable molecular orientation of these polymers (i.e., edge-on structure with respect to dielectric substrates) in solution-deposited films [24,66]. In this chapter, we will discuss the formation of various mesoscale crystalline morphologies and molecular orientation of HT-PATs, specifically HT-PHTs in solution-deposited films. Also, we will introduce poly(3,3″′-dialkylquaterthiophene) [82,83] and poly(2,5-bis(3-alkylthiophene-2-yl)thiono[3,2-b]thiophene [84], which can grow into highly oriented crystallites with respect to the substrates and show high charge mobilities of 0.2 ~ 0.6 cm2V–1s–1 achieved under nitrogen. Finally, we will discuss nucleation and growth of these crystalline polymers during solution deposition and present various approaches to control mesoscale and nanoscale formation of these self-assemblies.

8080_C005.fm Page 374 Friday, April 6, 2007 10:25 AM

Organic Field-Effect Transistors

Meltcrystallization Annealing (at 210) (180°C, 12h)

Spin-cast

No treatment

Drop-cast

374

(a) CHCl3

(b) Toluene

(c) THF

(d) CH2Cl2

FIGURE 5.2.2 Tapping mode atomic force microscopy topographs of poly(3-octyl thiophene) thin films cast on SiO2/Si substrates with various solution processing techniques from different solvents: (a) chloroform (CHCl3); (b) toluene; (c) tetrahydrofuran (THF); (d) methylene chloride (CH2Cl2). (From Yang, H., unpublished data, 2006. With permission.)

5.2.2 SOLUTION CASTING AND SPIN-COATING TO FORM SELF-ASSEMBLED POLYMERS 5.2.2.1 MOLECULAR STRUCTURE In order to improve solubility of unsubstituted polythiophenes, which are insoluble in organic solvents, various side chains have been incorporated into these backbones. In this case, these substituted side chains may interfere with molecular ordering or increase cell dimensions. Specifically, side-alkyl stacking (d(h00)) and π–π-stacking (d(010)) distance in crystals tends to increase with an increase in side-chain length, resulting in lower crystallinity and charge-carrier mobilities [51–55,58,74,77,78,85]. Figure 5.2.4 illustrates typical two-dimensional grazing incidence x-ray diffraction (GIXD; a detailed description is given in chapter 4.1) patterns of HT-PAT films dropcast from methylene chloride in closed jars. In all drop-cast HT-PAT films, these molecules tend to maintain preferentially an edge-on structure with respect to the SiO2/Si substrates, regardless of the alkyl side chains: hexyl (–C6H13), octyl (–C8H17), and dodecyl (–C12H25). The values of d(100) and d(010) increase from 16.1 to 25.5 Å and from 3.74 to 3.90 Å, respectively, with an increase in alkyl chain length from hexyl to dodecyl [51,77]. Semiconducting polythiophene derivatives have been designed to assemble into large crystalline domains on crystallization and to possess an extended, planar

8080_C005.fm Page 375 Friday, April 6, 2007 10:25 AM

375

Meltcrystallization Annealing (at 210) (180°C, 12h)

Spin-cast

No treatment

Drop-cast

Solution Deposition of Polymers

(a) CHCl3

(b) Toluene

(c) THF

(d) CH2Cl2

(a)

1.0

1.5

(b)

(200) (100) 0.0 0.5 1.0 1.5 qxy(Å−1)

(c) (400)

(300)

0.0

0.5

qz (Å−1)

2.0

FIGURE 5.2.3 Two-dimensional grazing incidence x-ray diffraction patterns of poly(3-octyl thiophene) thin films cast on SiO2/Si substrates with various solution processing techniques from different solvents: (a) chloroform (CHCl3); (b) toluene; (c) tetrahydrofuran (THF); (d) methylene chloride (CH2Cl2). (From Yang, H., unpublished data, 2006. With permission.)

(010)

(300)

(010)

(200) (100) 0.0 0.5 1.0 1.5 qxy(Å−1)

(500) (400) (300) (200) (100)

(010)

0.0 0.5 1.0 1.5 qxy(Å−1)

FIGURE 5.2.4 Two-dimensional GIXD patterns of drop-cast HT-PAT films with different alkyl side-chain lengths: (a) PHT; (b) poly(3-octylthiophene) (POT); (c) poly(3-dodecyl thiophene) (PDDT). (From Yang, H. unpublished data, 2006. With permission.)

8080_C005.fm Page 376 Friday, April 6, 2007 10:25 AM

376

Organic Field-Effect Transistors

C12H25 S

S

* S

* n

S C12H25 PQT-12

(a)

(b)

(c)

(d)

(e)

(f )

10−15 nm m 0n

−8

30

(100)

30−80 nm 10−15 nm (010)

(001)

FIGURE 5.2.5 Top shows the chemical structure of poly(3,3′″-didodecylquaterthiophene)s (PQT-12). (a) Topography and (b) phase images (1 × 1 μm) of annealed PQT-12 thin film on unmodified SiO2 surface; (c) and (d) are, respectively, topography and phase images (1 × 1 μm) of annealed PQT-12 thin film on OTS-8-modified SiO2 surface; (e) blown-up image from (d) showing rodlike structures with domains; (f) schematic depiction of a PQT-12 lamellar π–π stack. (Adapted from Wu, Y.O. et al., Appl. Phys. Lett. 86 (14), 2005.)

π-electron system that allows close intermolecular π–π-stacking and facilitates high charge mobility. Ong and coworkers have reported poly(3,3′″-didodecylquaterthiophene) (PQT-12) (Figure 5.2.5) thin films with highly edge-on orientation of polythiophene molecules, which facilitates in-plane charge transport and leads to high charge carrier mobility [16,83]. PQT-12 shows an inverse comb-like liquidcrystalline phase with two melting transitions at ~120 and 140°C, upon heating, and two crystallization transitions at ~118 and 70°C, upon cooling. Specifically, thermal annealing for spin-cast PQT-12 films near the liquid-crystalline phase transition temperatures induces highly oriented crystalline nanofibrils with respect to

8080_C005.fm Page 377 Friday, April 6, 2007 10:25 AM

Solution Deposition of Polymers

377

the substrate, resulting in high charge mobility (~0.2 cm2V–1s–1) in OTFT devices (Figure 5.2.5). Recently, poly(2,5-bis(3-alkylthiophene-2-yl)thioeno[3,2-b]thiophenes (PBTTT) (Figure 5.2.6) incorporating a linear fused conjugated unit, thieno[3,2b]thiophene with alkylthiophenes, were designed to have reduced electron delocalization along the polymer backbone. This results in a lowering of the polymer highest occupied molecular orbital level. Furthermore, the rotational invariance of the symmetrical thieno[3,2-b]thiophene in the backbone facilitates the adoption of the lowenergy chain conformation, promoting formation of highly ordered crystalline domains. PBTTT can grow into two-dimensional crystallites through postannealing spin-cast films achieving high charge mobilities of 0.2 ~ 0.6 cm2V–1s–1 [84]. Solution-deposited films of HT-PATs show various mesoscale crystalline morphologies from an amorphous texture to nanofibrils, mainly depending on their MW, solvents used, and processing conditions [15,17,23,35,36,44,59]. HT regioregularity of substituted side chains in polythiophene backbones strongly affects the solubility and molecular ordering of the resulting polymers. As the HT regioregularity increases, the solubility of PHT decreases and its crystallinity increases. However, high HT-PHT shows low crystallinity of ~15% [23,86] compared to polythiophenes (~35%) without side chains, as calculated from differential scanning calorimetry and x-ray analyses [86,87]. Sirringhaus and coworkers have found that spin-cast PHT films with low (81%) and high (96%) HT structure adapt face-on and edge-on structure of molecules on SiO2 substrates, respectively, as seen in Figure 5.2.7 [17,18]. As a result, field-effect mobility (μ) measured for the high HT-PHT film was 0.05–0.1 cm2V–1s–1, two orders of magnitude higher than the 2 × 10–4 cm2V–1s–1 of the low HT-PHT film (Figure 5.2.7c). Interestingly, drop-casting of low HT-PHT results in an order of magnitude improvement of the mobility compared to spin-casting. This is mainly related to crystallization behaviors of HT-PHT dependent on solvent evaporation rate, which will be discussed in more detail later. It is well known that solution-deposited HT-PAT films form various mesoscale crystalline morphologies according to the nature of the solvent [23,36,44,88] and solvent evaporation rate [89]. Unlike other π-conjugated polymers with rigid-rod conformation by strong π–π-interaction even in organic solvents [90], for a “flexible coil-like” HT-PAT molecule in a solution [63], self-organization of the polymer into an ordered crystal should be related to MW. This relationship is expected because PAT is a semicrystalline polymer. MW dependence is especially evident in the viscosity of semicrystalline polymer solutions or melts where the polymer MW strongly affects homogeneous nucleation and growth rate of the polymers, resulting in different mesoscale crystalline morphologies [63,91–95]. Accordingly, the influence of MW on the crystallization behaviors of semicrystalline polymers has been studied in various articles. For example, linear crystal growth rates of poly(ethylene oxide) and poly(ethylene succinate) (PES) reach a minimum value at a critical MW. This value is related to the crystallization transition from an extended chain to a folded chain conformation [96,97], suggesting that high MW polymers require sufficient reconformation time to achieve an ordered structure. As evidence of this MW dependence of the semicrystalline polymer on

8080_C005.fm Page 378 Friday, April 6, 2007 10:25 AM

378

Organic Field-Effect Transistors

R S *

S

S S

* n

R PBTTT (a)

400 nm (b)

400 nm

FIGURE 5.2.6 Top shows chemical structure of poly(2,5-bis(3-alkylthiophen-2yl)thieno[3,2b]thiophene) (PBTTT), where R = C10H21, C12H25, or C14H29. AFM images of PBTTT with side chain of C12H25 (annealed at 180°C): (a) before and (b) after annealing above the liquidcrystal isotherm. The left images show the topography and the right images show the phase images. The dark spots in the topography of the annealed film are voids where the film partially dewetted during annealing. (Adapted from McCulloch, I. et al., Nat. Mater. 5 (4), 328–333, 2006.)

8080_C005.fm Page 379 Friday, April 6, 2007 10:25 AM

Solution Deposition of Polymers

379

(b) a a

b

Intensity (a.u.)

5 b

(a)

4.5 4 3.5 3

(010) (300)

1

(100) (100)

(010)

μsat (cm2 V−1 s−1)

2 1.5

(200)

10−1

2.5

0.5

(c)

10−2 10−3 10−4

70

80 90 % heat-to-tall

FIGURE 5.2.7 Two-dimensional GIXD patterns of spin-cast PHT films with head-to-tail regioregularity of 96% (a) and 81% (b), showing different molecular orientations with respect to the SiO2 substrate. The film thickness was 70–100 nm. (c) Dependence of the roomtemperature mobility on the regioregularity for spin-cast (downward triangles) and drop-cast (upward triangles) top-contact PHT FETs (channel length L = 75 μm, channel width W = 1.5 mm). (From Sirringhaus, H., Nature, 401(6754), 685, 1999. With permission.)

crystallization transition, Magonov and coworkers reported AFM images of the solution-grown single crystals using C60H122 (M n ~ 850 Da) and C390H782 (M n ~ 5.5 kDa): the low MW C60H122 was extended in the crystals, while the high MW C390H782 showed a folded structure [92,93]. For HT PHT crystal on a highly oriented pyrolytic graphite substrate, scanning tunneling microscopy (STM) images [61,62,98–100] have showed that hexyl side chains tended to be organized in a planar, zigzag, interdigitated fashion, and the πconjugated backbones were oriented perpendicular to these side chains [98,99]. As seen in Figure 5.2.8, the STM images indicate that the conjugated chains tend to adopt both trans-linear conformation of the thiophene units and cis-conformation, which is a prerequisite for a “hair-pin” chain folding. MW and PDI of semiconducting polymers have become a popular topic for many research groups [35,44,59,60,88,101]. Commercially available HT-PHTs synthesized by metal-catalyzed cross-coupling polymerization have a broad PDI (>2.0),

8080_C005.fm Page 380 Friday, April 6, 2007 10:25 AM

380

Organic Field-Effect Transistors

(a)

(c)

(b)

A A

a B

a

B

100 A

20 A

FIGURE 5.2.8 STM images of long-range (a) and short-range (b) ordering of HT poly(3dodecyl thiophene) (PDT) on highly oriented pyrolytic graphites (HOPGs). A closer look at the images reveals that the individual strands are linear and chain folding, in which cisconformations of the thiophene units are prerequisite for the fold is also clearly observed. Regular hair-pin folds angle can be seen (A) but also folds of 120° (B). (c) Calculated model of PDDT corresponding to the area enclosed in the white square on the STM-image in (b). (Mena-Osteritz, E. et al., Angew. Chem. Int. Ed., 39(15), 2680, 2000. With permission.)

a small portion of regiorandom PHT, and other impurities. Of course, several HTPHT fractions showing narrow MWs and high regioregularity can be physically separated from the HT-PHTs using simple methods, such as a Soxhlet extraction with various solvents [35,70]. McGehee and coworkers have reported that the mobility of spin-cast HT-PHT films differs by several orders of magnitude depending on MW [35,44,102]. Low MW-PHTs (M n < 4 kDa) were found to form a highly crystalline “rod-like” morphology, while high MW polymers (M n > 30 kDa) formed isotropic nodules with low crystallinity. Recently, they also found that, for low MW PHTs, the crystal orientation at a critical buried interface between the film and the dielectric plays an important role on field-effect mobility, using x-ray diffraction rocking analysis for the low MW-PHT films on different self-assembled monolayer (SAM)-treated substrates. They observed a 1,000-fold increase in mobility of the film with highly oriented crystals [102]. More detailed discussions about their analysis with GIXD are given in chapter 4.1. Generally, the charge transport properties are dependent on the charge carrier density [103]. Charge-carrier transport of p-type semiconducting polymers in solid states (bulk or film) is classified into three different categories: intrachain, interchain, and interfibrillar hopping [49]. In crystals with ordered backbone chains, charge carriers can in theory transport along the polymer up to its contour length [50]. The carriers also transport between π-conjugated planes in the crystals. Finally, the charge carrier can transfer across crystalline grain boundaries (GBs) through interfibrillar hopping. For spin-cast HT-P3HT films, McGehee and coworkers proposed that the high MW molecules form small ordered areas separated by disordered regions [44]. In this case, the long chains can interconnect ordered areas and prevent charge carriers from being trapped by the disordered GBs. As a result, spin-cast high MWHT-P3HT films with fewer ordered crystals and GBs show higher charge mobility than low MW films. This results in highly developed crystalline structures, which require high potential energy to transfer the charge carriers across the GBs. However, it is unclear whether the difference in mobility is due to the sharper grain interface

8080_C005.fm Page 381 Friday, April 6, 2007 10:25 AM

Solution Deposition of Polymers

(a)

381

(b)

100 nm

(c)

100 nm

(d)

500 nm

(e)

500 nm (g)

(f)

500 nm (h)

500 nm

500 nm (i)

500 nm

500 nm

FIGURE 5.2.9 TM-AFM phase images of thin films of HT-PHTs of various weight-average molecular weight ( M w ) in FET devices prepared by drop-casting from toluene. M w in (a) through (i), respectively: 2.4, 4.8, 5.1, 7.0, 11.8, 15.7, 17.3, and 18.4 kDa. (Zhang, R. et al., J. Am. Chem. Soc., 128(11), 3480, 2006. With permission.)

between rods in the low MW films or to the fact that longer chains require less hopping between molecules. Recently, for HT-PHT films drop-cast from toluene, McCullough and coworkers demonstrated a change in nanofibrillar dimensions, specifically, the lateral width of the HT-PHT nanofibrils, dependent on weight-average molecular weight (M n) [104]. Figure 5.2.9 illustrates the TM-AFM phase images of drop-cast HT-PHT films on SiO2 substrates. In all samples, HT-PHTs tend to yield nanofibrils, regardless of MW, but the lateral width of the nanofibrils (as determined by Fourier transform of AFM images and grazing-incidence small-angle x-ray scattering) initially increases linearly with M w and then levels off, as shown in Figure 5.2.10. Also, field-effect mobility of HT-PHT films increases with an increase in the lateral width of the nanofibrils with respect to M w . Specifically, below M w < 10 kDa, a direct relationship between the nanofibrillar width and the contour length (Lw) [105,106] suggests that the polymer backbone can be extended up to the Lws and aligned perpendicular

8080_C005.fm Page 382 Friday, April 6, 2007 10:25 AM

382

Organic Field-Effect Transistors

Lw-weight average contour length (nm)

30

10

20

30

40 10−2

Width Mobility

20

10−3 10−4

10

Mobility (cm2/Vs)

WAFM-Nanoﬁbrill width (nm)

0

10−5 10−6 5 0 10 15 20 Mw-weight average molecular weight (kDa)

0

FIGURE 5.2.10 Dependence of nanofibril width (WAFM) and field-effect mobility (μ) on weight-average molecular weight (Mw, bottom axis) and weight-average contour length (Lw, top axis) of HT-PHT. (Zhang, R. et al., J. Am. Chem. Soc., 128(11), 3480, 2006. With permission.)

to the nanofibril axis. The increase in lateral width of nanofibrils results in improvement of field-effect mobility of the PHT-based OTFT devices (Figure 5.2.10). Ordered π–π-conjugated structure of HT-PAT polymers correlates strongly with their excellent electrical OTFT performance. However, most semiconducting polymers still have poor mechanical and processing properties compared to other nonconjugated polymers. One approach to solve this problem is to combine these polymers with other polymer units (i.e., to synthesize a block copolymer or to prepare blends) [53]. Block copolymers [107] can self-assemble into a number of nanoscale morphologies, such as lamellar, spherical, cylindrical, and so on. McCullough and coworkers have tried to control the self-assembled structure of thiophene-based polymers chemically linked with other polymers through changing the weight fractions of the HT-PHT component [9,22]. However, these polymer films show much lower conductivity (<10%) between PHT domains than homo-PHT films, though the films are composed of a continuous fibrillar network, as shown in Figure 5.2.11.

5.2.2.2 EFFECT

OF

SOLVENT

HT PATs show reversible color change response to temperature [108–114] or to an alteration of solvent polarity [79]. Table 5.2.1 illustrates solubility of HT-PHT in common organic solvents [79]. Thiophene and 1,2,4-trichlorobenzene (TC) can dissolve HT-PHT as easily as chloroform (CHCl3); its solubility in xylene, cyclohexylbenzene (CHB), and methylene chloride (CH2Cl2) is worse, and continuous heating is required after the solutions are filtered to avoid agglomeration and gelation. In solvents such as CHCl3 and tetrahydrofuran (THF), the rigidity of HT-PHT is

8080_C005.fm Page 383 Friday, April 6, 2007 10:25 AM

Solution Deposition of Polymers

383

(a)

(b) PHT

PS O

H

S

S n

Br n

FIGURE 5.2.11 AFM topographic (a) and phase (b) images of poly(3-hexylthiophene)-bpolystyrene diblock copolymer ( M w = 30,200 g/mol, M w / M n = 1.31, weight fraction of PHT: 0.37) solvent-cast from toluene. The inset in (b) represents chemical structure of the block copolymer. The scanned region is 1.5 × 1.5 μm. (Liu, J.S. et al., Angew. Chem., Int. Ed., 41(2), 329, 2001. With permission.)

TABLE 5.2.1 Solubility of HT-PHT in Common Organic Solvents Solvent

Color

λmax (nm)

Solubility at 25°°C (gL–1)

CHCl3 THF Toluene CH2Cl2 Hexane DMF, methanol, ethanol

Deep brown Reddish brown Dark brown Deep purple Black Transparent

451 445 451 448 429 —

>30 2.3 2.1 0.4 0.015 Insoluble

Source: Yamamoto, T. et al., J. Am. Chem. Soc. 120, 2047–2058, 1998.

only two to three times higher than that of common flexible polymers such as polystyrene or poly(methyl methacrylate), despite its conjugated backbone [63,64]. The magnitude of the Huggins coefficient (κH) is reasonable for a flexible polymer in a “good” solvent. However, in dilute CHCl3 and THF solutions of HTPHT, the κH of HT PHT increases as a function of MW [63]. Consequently, CHCl3 and THF are not “good” but “moderately good” solvents for HT-PHT, and there exists a molecular limit of HT-PHT at which the molecule is no longer soluble at room temperature. Specifically, light-scattering analysis for soluble HT-PHT in CHCl3 shows that photoinduced polarizability of regioregular and regiorandom PHTs is 0.26 and 0.01, respectively, indicating that HT-PHT even has a stiff structure at dilute concentrations [79]. As a solvent is evaporated in a solution containing soluble HT-PHT, the solution concentration becomes higher, causing a solventinduced aggregation or crystallization of the polymer due to solubility differences

8080_C005.fm Page 384 Friday, April 6, 2007 10:25 AM

384

Organic Field-Effect Transistors

520 a

Absorbance

0.8

560 ih

b c

0.6 0.4

Methanol content (vol %) a 0 g b 10 fe c 20 d d 30 e 40 610 f 50 g 60 h 70 i 80

0.2 0 200

1.0 0.8 Absorbance

1.0

e f g h i

0.6

ab c d

0.4

Methanol content (vol %) a 0 b 10 c 20 d 30 e 40 f 50 g 60 h 70 i 80

0.2 300

400 500 600 Wavelength/nm (a)

700

800

0 200

300

400 500 600 Wavelength/nm

700

800

(b)

FIGURE 5.2.12 Changes in UV-vis spectra of CHCl3 solutions of (a) HT-PHT and (b) mixed PHT (HT/HH = 1/2) on addition of CH3OH at 25°C. [PHT] = 15 mgL–1 or 9 × 10–5 M (monomer unit). Volume percent of CH3OH: 0–80% for a–i as shown. (Yamamoto, T. et al., J. Am. Chem. Soc., 120(9), 2047, 1998. With permission.)

between polythiophene backbones and side chains in organic solvents. This phenomenon is similar to addition of a poor solvent into a stable polymer solution, leading to π-stacking of the polymer molecules and formation of a stable colloidal aggregate [65,68,79,110,115]. For example, addition of methanol to the CHCl3 solution of HT-PHT gives a stable colloidal solution over a wide range of CHCl3/CH3OH ratios. The colloidal solution can yield information about the stacking of the polymer in solution. Figure 5.2.12 illustrates UV-vis spectra of CHCl3 solutions of HT-PHT with various loading of CH3OH. For HT-PATs bearing chiral side groups, reconformation of main chain from nonplanar to planar is accompanied by the appearance of an intense induced circular dichroism in the UV-vis region derived from the main chain or supramolecular chirality [22,114,116–119]. HT-PHT gives the π–π*-absorption band at 450 nm in CHCl3, and addition of CH3OH to the CHCl3 solution leads to a bathochromic shift of the π–π*-absorption band. As shown in Figure 5.2.12(a), the new absorption band shows splitting, and the absorption peaks at 520, 560, and 610 nm agree with those of a film of HT-PHT [56,57,71–73,75,81,120–122], indicating that HT-PHT takes a well-stacked structure in the CHCl3/CH3OH mixture, similar to the film [76,80,111,123–125]. However, PHT containing a majority of head-to-head units (HH) does not seem to form the stacked structure in the CHCl3/CH3OH mixtures at high CH3OH concentrations (Figure 5.2.12b). From these results, Yamamoto and coworkers proposed an aggregate structure of HT-PHT in a colloidal solution (Figure 5.2.13). In this model, HT-PHT molecules aggregate along the main polythiophene chain. Specifically, light scattering analysis of the aggregates supports as the most accepted model for HT-PAT agglomeration in organic solvents the theory that the aggregate tends to grow in the direction of the y and z axes in Figure 5.2.13. The other models predict that, at higher polymer concentration or at poorer solvent quality, HT-PATs aggregate in a face-to-face fashion [126–129] into rod-like micelles [130] in order to decrease the unfavorable interaction between the solvent and the aromatic main chain.

8080_C005.fm Page 385 Friday, April 6, 2007 10:25 AM

Solution Deposition of Polymers

385

x 3.8 A

y

z

16.4 A

FIGURE 5.2.13 A model for the aggregation of HT-PHT in the colloidal solution; solvent = mixture of CHCl3 and CH3OH. The aggregate may consist of several subunits with the sideto-side stacked molecules. (Yamamoto, T. et al., J. Am. Chem. Soc., 120(9), 2047, 1998. With permission.)

On the other hand, Kiriy and coworkers observed a different type of aggregate in HT PAT colloidal solutions, as shown in Figures 5.2.14 and 5.2.15. They demonstrated kinetically driven dynamic growth of one-dimensional aggregates of P3AT adsorbed onto a hydrophobic Si substrate, from a CHCl3/hexane solution using AFM [68]. From AFM and UV-vis analyses for aggregates of HT-PATs in CHCl3/hexane, they suggest that PAT molecules adopt a helical conformation with anticonfiguration of thiophene units in which all sulfur atoms are directed inside the cavity and hydrocarbon groups are oriented outside the helix (Figure 5.2.15). However, on the substrates, dried aggregates inherently have effects of solvent evaporation and substrate-induced crystallization. CHCl3 is a commonly used solvent for film deposition of HT-PHT in field-effect transistors. However, its low boiling point (60.5–61.5°C) and rapid evaporation limit the time for crystallization during a spin-casting process. As a result, charge mobilities achieved for the HT-PHT films are typically on the order of 0.01 cm2/Vs. Recently, it was reported that trichlorobenzene with good solubility and a high boiling point (218–219°C) significantly improved the charge mobility of spin-cast HT-PHT films up to 0.12 cm2/Vs with on/off ratios of 106 [36]. AFM and GIXD analyses for HT-PHT films spin-cast from CHCl3 and TCB support that high crystalline morphology and efficient π–π-stacking parallel to the in-plane current flow in the spin-cast HT-PHT films can be developed with high-boiling-point and highsolubility solvents, resulting in high electric performance of HT-PHT n based OTFTs. As a result, the use of appropriate solvents with high boiling points for spin-casting combines the advantages of increased drying time similar to that of drop-casting while retaining high thin film uniformity. This solvent-dependent structure of

8080_C005.fm Page 386 Friday, April 6, 2007 10:25 AM

386

Organic Field-Effect Transistors

(a)

(b)

150 nm

50 nm

300 nm

Height, nm

(c)

(d)

6 4 2 0

–20

(e)

200 HM = 2.2 nm HN = 1.5 nm

–15

PDI = 1.5

–10 5 0 0

2

4

Height, nm

6

Frequency (a.u.)

0

400 30 (f )

600

nm

LM = 428 nm LN = 347 nm

20

PDI = 1.23

10 0 0

300

600

900

Length, nm

FIGURE 5.2.14 Solution AFM topographs of PHT (0.01 g/L) adsorbed onto hydrophobic Si substrates from chloroform/hexane mixture (1:5 v/v): (a, b) immediately after addition of hexane into chloroform solution of PHT; (c) 15 min after addition of hexane; (d) cross-section taken along the particle marked by aHTow. (e) Height distribution of spherical particles adsorbed from PHT solution (0.001 g/L); (f) length distribution of one-dimensional aggregates formed at 0.01 g/L of PHT. (Kiriy, N. et al., Nano. Lett., 3(6), 707, 2003. With permission.)

8080_C005.fm Page 387 Friday, April 6, 2007 10:25 AM

Solution Deposition of Polymers

387

R R R

R

S S

R S

S

S

S

R

R

S R

All anti-conformation

S

RS

(a)

S S

S S

S R

R R

(c) R

S S

R

R

R

(b)

1 nm

1 nm

FIGURE 5.2.15 (a) The helical conformation of PHT molecules. Their one-dimensional aggregation into helical nanotubes: side view (b); top view (c). (Kiriy, N. et al., Nano. Lett., 3(6), 707, 2003. With permission.)

HT-PHT has also been reported in drop-cast films from various solvents with different boiling points and solubilities (Figure 5.2.16) [23]. Figure 5.2.17 illustrates AFM phase images for HT-PHT nanocrystals grown on SiO2 substrates from different solvents: CH2Cl2 and CHCl3. With a low-boiling point (40°C) and poor solubility, CH2Cl2 did not sufficiently develop nanofibrillar structure of HT-PHT during solution deposition, specifically due to fast solvent evaporation rate, showing “nanorod” structure. HT-PHT nanofibrils in CHCl3 (boiling point: 61°C), on the other hand, grew up to several microns in length and were interconnected with each other. The charge mobilities (μ) of the corresponding 30-nm-thick HT-PHT films from CH2Cl2 and CHCl3 were ~1 × 10–4 cm2/Vs and ~0.01 cm2/Vs in bottom-contact OTFTs, respectively.

5.2.2.3 PROCESSING CONDITION As mentioned previously, mesoscale crystalline morphologies and molecular orientation of HT-PATs in solution-deposited films can be controlled through different solution-processing methods and subsequent thermal treatments [15,17,23,35,36,44,59,89]. Figures 5.2.18 and 5.2.19 represent typical examples for tunable crystalline morphologies and molecular orientations of HT-PHT in solution-deposited films. First, in terms of mesoscale morphologies, HT-PHT (M n = 18.8 kDa) films drop-cast from CHCl3 show well developed nanofibrils interconnected with each other, while the spin-cast

8080_C005.fm Page 388 Friday, April 6, 2007 10:25 AM

388

Organic Field-Effect Transistors

(a)

(b)

(c)

FIGURE 5.2.16 AFM topographs (left) and phase images (right) of HT-P3HT films dropcast from various solvents in closed petri dishes: (a) toluene; (b) THF; (c) CH2Cl2. (All scale bars represent 1 μm.) (Yang, H. et al., Adv. Func. Mater., 15(4), 671, 2005. With permission.)

8080_C005.fm Page 389 Friday, April 6, 2007 10:25 AM

Solution Deposition of Polymers

389

1 μm (a)

1 μm (b)

1 μm (c)

1 μm (d)

FIGURE 5.2.17 TM-AFM phase images of HT-PHT ( M w = 11.4 kDa, PDI = 2.2) nanofibrils grown on SiO2 substrates from 0.05 and 0.1 wt% solution with different solvents: (a) 0.05 and (b) 0.1 wt% in CH2Cl2; (c) 0.05 and (d) 0.1 wt% in CHCl3. (Yang, H. et al., Adv. Func. Mater., 15(4), 671, 2005. With permission.)

films had smooth featureless surfaces, showing similar results to literature. Visible crystalline development of HT-PHT in the spin-cast films is just observed through thermal treatment above the melting temperature (216°C) because morphological development from less ordered to ordered structure requires reconformation of polythiophene backbones (Figure 5.2.18d). On the other hand, spin-cast HT-PHT films with low MW show crystalline enhancement through thermal annealing below melting temperature [44]. As seen in Figure 5.2.19, two-dimensional GIXD can provide detailed information for crystalline structure and molecular orientation of HT-PHT in these films. First, twodimensional GIXD patterns of the drop-cast film showed intense (100) peaks with higher order peaks and (010) peak along the qz- and qxy-axes, respectively (white arrows in Figure 5.2.19a). The result indicates that the chains have an edge-on structure on the substrate, where π-conjugated polythiophene backbones are preferentially oriented parallel to the substrate (the inset in Figure 5.2.19a). Conversely, the spin-cast film had both edge-on and face-on chain orientations, as indicated from the diffraction peaks of (100) crystal planes in both the meridian (qz) and equatorial (qxy) directions. In addition, the film had much lower crystallinity than the drop-cast film, as calculated by overall integration of the diffracted crystal peaks. Specifically, face-on structure of HT-PHT in spin-cast films tends to increase with increasing substrate temperature. Most interestingly, unlike significant

8080_C005.fm Page 390 Friday, April 6, 2007 10:25 AM

390

Organic Field-Effect Transistors

(a)

(b)

(c)

(d)

FIGURE 5.2.18 AFM topographs of HT-P3HT films (Mn = 18.8 kDa) fabricated from CHCl3 under different casting conditions: (a) drop-cast; (b) as prepared; (c) high-temperature annealing (at 200°C, 12 h); (d) meltcrystallized (with a cooling rate of 2°C/min). (All scale bars represent 500 nm.) (From Yang, H., unpublished data, 2006. With permission.)

enhancement in annealed film morphologies, the molecular orientation of HT-PHT in the melt-crystallized films is not equivalent to that in the drop-cast film. It is related to the fact that semicrystalline polymers have residual local order even in the melt (termed the “melting memory effect”), suggesting that initial self-assembled structures of HT-P3HT on dielectric substrates play an important role in the electrical properties of the corresponding films. The AFM and two-dimensional GIXD results for high MW spin-cast HT PHT films strongly support the conclusion that low charge mobility (≤10–3 cm2V–1s–1) in most spin-cast films from HT-PHT soluble in CHCl3 is mainly related to the poor crystallinity and unfavorable orientation of π-conjugated backbone planes normal to the substrate (i.e., the in-plane current flow) [89]. During direct solution deposition of HT-PHTs on dielectric substrates, the solvent evaporation rate can be changed in a wide range from a few seconds to a few hours by controlling a combination of the following parameters: solvent boiling point processing mode: spin-casting, drop-casting, and controlling the substrate temperature solvent vapor pressure (νp) in film deposition facility

8080_C005.fm Page 391 Friday, April 6, 2007 10:25 AM

Solution Deposition of Polymers

391

(a)

(b)

(010) Substrate

(200)

(100) 1.0

1.5

2.0

qxy (Å−1) (c)

(010)

0.5

(100)

0

(100) (300)

(200)

(300)

(200)

qΖ

(300)

(010)

Substrate

(d)

0

0.5

1.0

1.5

2.0

qxy (Å−1)

FIGURE 5.2.19 Two-dimensional GIXD patterns of drop-cast (a) and spin-cast (b–d) HTP3HT thin films obtained from CHCl3 solution: (b) without any thermal treatment; (c) annealed at 200°C for 12 h; (d) melt-crystallized from 240°C with a cooling rate of 2°C/min. (The insets in the upper right corner of (a) and (b) represent orientation of π–π-conjugated planes in HT-P3HT chains with respect to the SiO2 substrates.) (From Yang, H., unpublished data, 2006. With permission.)

The first two parameters were discussed earlier in this chapter. Generally, the spin-casting method cannot produce nanofibrillar morphologies of HT-PHTs compared to the drop-casting method because of fast evaporation rates. However, during spin-casting, the use of appropriate solvents with high boiling points or the control of νp can provide increased drying time similar to that of drop-casting while retaining high thin film uniformity. Recently, Cho and coworkers demonstrated tunable film structures of HT-PHTs through controlling νp under a spin-casting condition (1,000 rpm) [67]. Figure 5.2.20 represents AFM images for 98% HT-PHT ( M w = 54 kDa, PDI = 1.8) films spin-cast from CHCl3 under various solvent vapor pressures. In the spin-cast films, the HT-PHT nanocrystals grew from short nanorod to long nanofibrils by increasing the νp. Specifically, nucleation and growth behavior of HTPHT crystals are clearly displayed as shown in Figure 5.2.21. On SiO2 substrates covered with the HT-PHT solution during drop-casting, low solvent vapor pressure induced by an open petri dish setup cannot provide sufficient time for HT-PHT crystal growth to occur. Instead, aggregates of short nanorods were observed. On the other hand, an increase in the solvent pressure by a closed environment yields several micron-length nanofibrils. They fabricated micro-size single-crystal wires (height: 700 nm–1.3 μm; width: 1–3 μm; length: 30–500 μm) of PHT (weight-average

8080_C005.fm Page 392 Friday, April 6, 2007 10:25 AM

392

Organic Field-Effect Transistors

(a)

(b)

(c)

(d)

(e)

(f )

FIGURE 5.2.20 TM-AFM images (2 × 2 μm2) of PHT molecular nanowires deposited on a substrates (SiOx) at the following solvent vapor pressures: atmospheric condition (a); 6.2 kPa (b); 36.5 kPa (c); 48.9 kPa (d); 53.8 kPa (e); 56.6 kPa (f). (Kim, D.H. et al., Macromol. Rapid. Commun., 36(3), 691, 2005. With permission.)

molecular weight = 54 kg mol–1; regioregularity = 98.5%) through recrystallization of a dilute PHT (0.1 mg/L) solution in CHCl3 (Figure 5.2.22) [131]. In solution-deposited films on dielectrics, the crystalline orientation and morphology depend considerably on polymer–substrate interaction, which can be controlled by modification of the dielectric surface with a self-assembled monolayer (SAM). Cho et al. have controlled the intermolecular interaction at the interface between HTPHT and the dielectric substrate by using SAMs functionalized with various groups (–NH2, –OH, and –CH3) [24]. They have found that, depending on the properties of substrate surface, HT-PHT nanocrystals in the spin-cast films adopt two different orientations of π-conjugated planes — parallel and perpendicular to the substrates. These have field-effect mobilities that differ by more than a factor of four and are as high as 0.28 cm2V–1s–1 for π-conjugated planes parallel to the substrates.

5.2.3 CONCLUDING REMARKS This chapter has described different approaches for the deposition of conjugated polymer films from solution. Many of these films have shown excellent device performance, with mobilities as high as 0.6 cm2V–1s–1. Recently, advances in device performance and device stability have been achieved through both molecular design and the optimization of processing conditions. The investigation of polymer ordering at the semiconductor–dielectric interface has led to improved understanding of the charge transport mechanisms in polymer thin film devices. With continuing improvements in device parameters and stability, conjugated polymer thin film transistors remain the leading candidate for all printed organic transistors.

8080_C005.fm Page 393 Friday, April 6, 2007 10:25 AM

Solution Deposition of Polymers

393

(a)

(b)

(c)

FIGURE 5.2.21 AFM topographs (left) and phase images (right) of HT-P3HT films dropcast from CH2Cl2 under different solvent evaporation conditions: (a) open petri dish; (b) closed petri dish; (c) closed jar. (All scale bars represent 1 μm.) (From Yang, H., unpublished data, 2006. With permission.)

8080_C005.fm Page 394 Friday, April 6, 2007 10:25 AM

394

Organic Field-Effect Transistors

(a)

(e)

c-axis

a-axis S S

S S S

S S

30 nm

S

S

30 μm

500 nm (c)

(d)

S

S S

S 8.36Å

b-axis

is

S

[010]

S

(020)

S

S

S

S

S

S

S

S

S

S

S

S

S

S

S

S

S

S S

S

S

S S S

S

S

S

S

S

S

S

S

S

W

S

S

S

S

S

S

ax

S

S

S

S S

S

S S

S

S

S

S S

S

S

S S

ire

S S

S

S

S

S S

S

S S

S

c

S S

(f) S

3 μm

S S S

S

S S

S

7.80Å

S

b

S

S

S

16.60Å

(b) S

S S

S S

3.9Å 3.9Å

S

(002)

FIGURE 5.2.22 Morphological features and structure characterization of single-crystal PHT microwires. (a) POM image of one-dimensional single-crystal PHT microwire bundles. (b) Field-emission scanning electron microscopy images of PHT wires formed on a silicon substrate modified with an ODTS dielectric layer. The inset shows a side-view image showing the rectangular cross-section of PHT microwires with well-defined facets. (c) Transmission electron microscopy image of PHT wires on a silicon nitride window modified with ODTS, showing preferential growth along the [010] direction. (d) Selected-area electron diffraction pattern of PHT microwire showing diffraction equivalent to a repeating period of 7.80 Å along the π–π-stacking direction, and a repeating period of 8.36 Å along the chain direction. (e) Molecular and crystallographic structures (orthorhombic unit cell) of PHT chains show the enhanced π–π-overlap along the b direction. (f) Schematic representation of the hierarchical self-assembly of PHT chains into single-crystal microwires along π–π-stacking direction. The height, width, and length of the wires are approximately 700 nm–1.3 μm, 1–3 μm, and 30–500 μm, respectively. (Kim, D.H. et al., Adv. Mater., 18(6), 719, 2006. With permission.)

8080_C005.fm Page 395 Friday, April 6, 2007 10:25 AM

Solution Deposition of Polymers

395

REFERENCES 1. Wang, J.Z. et al., Polymer field effect transistors fabricated by dewetting, Synth. Met. 146, 287–290, 2004. 2. Berggren, M. et al., Light-emitting-diodes with variable colors from polymer blends, Nature 372, 444–446, 1994. 3. Granstrom, M. and Inganas, O., Flexible arrays of submicrometer-sized polymeric light emitting diodes, Adv. Mater. 7, 1012, 1995. 4. Fuchigami, H., Tsumura, A., and Koezuka, H., Polythienylenevinylene thin-filmtransistor with high carrier mobility, Appl. Phys. Lett. 63, 1372–1374, 1993. 5. Tsumura, A., Koezuka, H., and Ando, T., Macromolecular electronic device — fieldeffect transistor with a polythiophene thin-film, Appl. Phys. Lett. 49, 1210–1212, 1986. 6. Mardalen, J. et al., Structure and electronic-properties of polythiophene and poly(3alkylthiophene) films electropolymerized on indium tin oxide (Ito) glass at elevated potentials, Macromol. Chem. Phys. 194, 2483–2495, 1993. 7. Chen, X.L. et al., Ion-modulated ambipolar electrical conduction in thin-film transistors based on amorphous conjugated polymers, Appl. Phys. Lett. 78, 228–230, 2001. 8. Babel, A. and Jenekhe, S.A., High electron mobility in ladder polymer field-effect transistors, J. Am. Chem. Soc. 125, 13656–13657, 2003. 9. McCullough, R.D., The chemistry of conducting polythiophenes, Adv. Mater. 10, 93, 1998. 10. Jeffries-El, M., Sauve, G., and McCullough, R.D., Facile synthesis of end-functionalized regioregular poly(3-alkylthiophene)s via modified Grignard metathesis reaction, Macromolecules 38, 10346–10352, 2005. 11. Liu, J.S. and McCullough, R.D., End group modification of regioregular polythiophene through postpolymerization functionalization, Macromolecules 35, 9882–9889, 2002. 12. Sheina, E.E. et al., Highly conductive, regioregular alkoxy-functionalized polythiophenes: A new class of stable, low band gap materials, Chem. Mater. 17, 3317–3319, 2005. 13. Zhai, L. and McCullough, R.D., Layer-by-layer assembly of polythiophene, Adv. Mater. 14, 901–905, 2002. 14. Bjornholm, T. et al., Polythiophene nanowires, Adv. Mater. 11, 1218–1221, 1999. 15. Bao, Z., Dodabalapur, A., and Lovinger, A.J., Soluble and processable regioregular poly(3-hexylthiophene) for thin film field-effect transistor applications with high mobility, Appl. Phys. Lett. 69, 4108–4110, 1996. 16. Ong, B.S. et al., High-performance semiconducting polythiophenes for organic thinfilm transistors, J. Am. Chem. Soc. 126, 3378–3379, 2004. 17. Sirringhaus, H. et al., Two-dimensional charge transport in self-organized, highmobility conjugated polymers, Nature 401, 685–688, 1999. 18. Sirringhaus, H. et al., Microstructure-mobility correlation in self-organized, conjugated polymer field-effect transistors, Synth. Met. 111, 129–132, 2000. 19. Stokes, K.K., Heuze, K., and McCullough, R.D., New phosphonic acid functionalized, regioregular polythiophenes, Macromolecules 36, 7114–7118, 2003. 20. Iovu, M.C. et al., Regioregular poly(3-alkylthiophene) conducting block copolymers, Polymer 46, 8582–8586, 2005. 21. Zhang, Z.J. et al., Synthesis and characterization of a novel water-soluble block copolymer with a rod-coil structure, Mater. Lett. 60, 679–684, 2006.

8080_C005.fm Page 396 Friday, April 6, 2007 10:25 AM

396

Organic Field-Effect Transistors

22. Liu, J.S. et al., Tuning the electrical conductivity and self-assembly of regioregular polythiophene by block copolymerization: Nanowire morphologies in new di- and triblock copolymers, Angew. Chem. Int. Ed. 41, 329, 2001. 23. Yang, H.C. et al., Effect of mesoscale crystalline structure on the field-effect mobility of regioregular poly(3-hexyl thiophene) in thin-film transistors, Adv. Func. Mater. 15, 671–676, 2005. 24. Kim, D.H. et al., Enhancement of field-effect mobility due to surface-mediated molecular ordering in regioregular polythiophene thin film transistors, Adv. Func. Mater. 15, 77–82, 2005. 25. Langmuir, I., The constitution and fundamental properties of solids and liquids. II. Liquids, J. Am. Chem. Soc. 39, 1848, 1917. 26. Blodgett, K.B., Films built by depositing successive monomolecular layers on a solid surface, J. Am. Chem. Soc. 57, 1007, 1935. 27. Pal, A.J. et al., Polymeric light-emitting diodes from molecularly thin poly(3-hexylthiophene) Langmuir–Blodgett films, Appl. Phys. Lett. 69, 1137–1139, 1996. 28. Pal, A.J., Paloheimo, J., and Stubb, H., Molecular light-emitting diodes using quinquethiophene Langmuir–Blodgett films, Appl. Phys. Lett. 67, 3909–3911, 1995. 29. Paloheimo, J. et al., Molecular field-effect transistors using conducting polymer Langmuir–Blodgett films, Appl. Phys. Lett. 56, 1157–1159, 1990. 30. Xu, G.F., Bao, Z.A., and Groves, J.T., Langmuir–Blodgett films of regioregular poly(3-hexylthiophene) as field-effect transistors, Langmuir 16, 1834–1841, 2000. 31. Cheung, J.H. and Rubner, M.F., Fabrication of electrically conductive Langmuir–Blodgett multilayer films of polyaniline, Thin Solid Films 244, 990–994, 1994. 32. Bjornholm, T. et al., Self-assembly of regioregular, amphiphilic polythiophenes into highly ordered pi-stacked conjugated polymer thin films and nanocircuits, J. Am. Chem. Soc. 120, 7643–7644, 1998. 33. Reitzel, N. et al., Self-assembly of conjugated polymers at the air/water interface. Structure and properties of Langmuir and Langmuir–Blodgett films of amphiphilic regioregular polythiophenes, J. Am. Chem. Soc. 122, 5788–5800, 2000. 34. de Boer, B. et al., Amphiphilic, regioregular polythiophenes, Macromolecules 35, 6883–6892, 2002. 35. Kline, R.J. et al., Controlling the field-effect mobility of regioregular polythiophene by changing the molecular weight, Adv. Mater. 15, 1519, 2003. 36. Chang, J.F. et al., Enhanced mobility of poly(3-hexylthiophene) transistors by spincoating from high-boiling-point solvents, Chem. Mater. 16, 4772–4776, 2004. 37. Jabbour, G.E., Radspinner, R., and Peyghambarian, N., Screen printing for the fabrication of organic light-emitting devices, IEEE J. Sel. Top. Quantum Electron. 7, 769–773, 2001. 38. Bao, Z.N. et al., High-performance plastic transistors fabricated by printing techniques, Chem. Mater. 9, 1299, 1997. 39. Sirringhaus, H. et al., High-resolution inkjet printing of all-polymer transistor circuits, Science 290, 2123–2126, 2000. 40. Kawase, T. et al., Inkjet printed via hole interconnections and resistors for all-polymer transistor circuits, Adv. Mater. 13, 1601, 2001. 41. Aernouts, T. et al., Printable anodes for flexible organic solar cell modules, Thin Solid Films 451–452, 22–25, 2004. 42. Min, G., Plastic electronics and their packaging technologies, Synth. Met. 135, 141–143, 2003. 43. Bao, Z.N., Rogers, J.A., and Katz, H.E., Printable organic and polymeric semiconducting materials and devices, J. Mater. Chem. 9, 1895–1904, 1999.

8080_C005.fm Page 397 Friday, April 6, 2007 10:25 AM

Solution Deposition of Polymers

397

44. Kline, R.J. et al., Dependence of regioregular poly(3-hexylthiophene) film morphology and field-effect mobility on molecular weight, Macromolecules 38, 3312–3319, 2005. 45. Savenije, T.J. et al., The effect of thermal treatment on the morphology and charge carrier dynamics in a polythiophene-fullerene bulk heterojunction, Adv. Funct. Mater. 15, 1260–1266, 2005. 46. Cik, G., Sersen, F., and Dlhan, L., Thermally induced transitions of polarons to bipolarons in poly(3-dodecylthiophene), Synth. Met. 151, 124–130, 2005. 47. Marchant, S. and Foot, P.J.S., Annealing behavior of conductive poly(3-hexylthiophene) films, Polymer 38, 1749–1751, 1997. 48. Aasmundtveit, K.E. et al., Orientation effect in thin layers of poly(octylthiophene) on glass studied by synchrotron x-ray diffraction, Synth. Met. 89, 203–208, 1997. 49. Shimomura, T. et al., Intrachain conduction and main-chain conformation of conducting polymers as studied by frequency-domain electric birefringence spectroscopy, Phys. Rev. Lett. 72, 2073–2076, 1994. 50. Shimomura, T. et al., Frequency-domain electric birefringence spectra of conducting polymers, Synth. Met. 69, 689–690, 1995. 51. Aasmundtveit, K.E. et al., Structural anisotropy of poly(alkylthiophene) films, Macromolecules 33, 3120–3127, 2000. 52. Aasmundtveit, K.E. et al., Structural studies of polyalkylthiophenes with alternating side chain positioning, Synth. Met. 113, 7–18, 2000. 53. Babel, A. and Jenekhe, S.A., Field-effect mobility of charge carriers in blends of regioregular poly(3-alkylthiophene)s, J. Phys. Chem. B 107, 1749–1754, 2003. 54. Babel, A. and Jenekhe, S.A., Alkyl chain length dependence of the field-effect carrier mobility in regioregular poly(3-alkylthiophene)s, Synth. Met. 148, 169–173, 2005. 55. Camaioni, N. et al., Polyalkylthiophene-based photodetector devices: Effect of sidechain length on the device performance, Mater. Sci. Eng. C-Biomimetic Supramolec. Syst. 15, 261–263, 2001. 56. Chen, T.A. and Rieke, R.D., Polyalkylthiophenes with the smallest bandgap and the highest intrinsic conductivity, Synth. Met. 60, 175–177, 1993. 57. Chen, T.A., Wu, X.M., and Rieke, R.D., Regiocontrolled synthesis of poly(3-alkylthiophenes) mediated by rieke zinc: Their characterization and solid-state properties, J. Am. Chem. Soc. 117, 233–244, 1995. 58. Fell, H.J. et al., The effect of stereoregularity on the structure of poly(octylthiophene): An x-ray-diffraction study, Synth. Met. 69, 301–302, 1995. 59. Goh, C. et al., Molecular-weight-dependent mobilities in regioregular poly(3-hexylthiophene) diodes, Appl. Phys. Lett. 86, 122110, 2005. 60. Goh, C.T. et al., Molecular weight dependent mobilities in regioregular poly(3hexyl-thiophene) diodes and transistors, Abst.Pap. Am. Chem. Soc. 227, U425–U425, 2004. 61. Grevin, B. et al., Multiscale scanning tunneling microscopy imaging of self-organized regioregular poly(3-hexylthiophene) films, J. Chem. Phys. 118, 7097–7102, 2003. 62. Grevin, B. et al., Scanning tunneling microscopy investigations of self-organized poly(3-hexylthiophene) two-dimensional polycrystals, Adv. Mater. 15, 881, 2003. 63. Heffner, G.W. and Pearson, D.S., Molecular characterization of poly(3-hexylthiophene), Macromolecules 24, 6295–6299, 1991. 64. Heffner, G.W., Pearson, D.S., and Gettinger, C.L., Characterization of poly(3-octylthiophene). 1. Molecular characterization in dilute-solution, Polymer Eng. Sci. 35, 860–867, 1995.

8080_C005.fm Page 398 Friday, April 6, 2007 10:25 AM

398

Organic Field-Effect Transistors

65. Inganas, O. et al., Thermochromic and solvatochromic effects in poly(3-hexylthiophene), Synth. Met. 22, 395–406, 1988. 66. Kim, D.H. et al., Surface-induced conformational changes in poly(3-hexylthiophene) monolayer films, Langmuir 21, 3203–3206, 2005. 67. Kim, D.H. et al., Solvent vapor-induced nanowire formation in poly(3-hexylthiophene) thin films, Macromol. Rapid Commun. 26, 834–839, 2005. 68. Kiriy, N. et al., One-dimensional aggregation of regioregular polyalkylthiophenes, Nano Lett. 3, 707–712, 2003. 69. Konestabo, O.R. et al., Role of sidechain sequence in stereoregular polyalkylterthiophene, Synth. Met. 84, 589–590, 1997. 70. Liu, J.S., Loewe, R.S., and McCullough, R.D., Employing MALDI-MS on poly(alkylthiophenes): Analysis of molecular weights, molecular weight distributions, endgroup structures, and end-group modifications, Macromolecules 32, 5777–5785, 1999. 71. McCullough, R.D. and Lowe, R.D., Enhanced electrical conductivity in regioselectively synthesized poly(3-alkylthiophenes), J. Chem. Soc. Chem. Commun. 1, 70–72, 1992. 72. McCullough, R.D. et al., Design, synthesis, and control of conducting polymer architectures: Structurally homogeneous poly(3-alkylthiophenes), J. Org. Chem. 58, 904–912, 1993. 73. McCullough, R.D. et al., Self-orienting head-to-tail poly(3-alkylthiophenes): New insights on structure–property relationships in conducting polymers, J. Am. Chem. Soc. 115, 4910–4911, 1993. 74. Qiao, X.Y., Wang, X.H., and Mo, Z.S., The effects of different alkyl substitution on the structures and properties of poly(3-alkylthiophenes), Synth. Met. 118, 89–95, 2001. 75. Wu, X.M., Chen, T.A., and Rieke, R.D., Synthesis of regioregular head-to-tail poly[3(alkylthio)thiophenes]: A highly electroconductive polymer, Macromolecules 28, 2101–2102, 1995. 76. Yamamoto, T., Unique temperature dependence of NMR and UV-visible spectra of poly(3-hexylthiophene-2,5-diyl) and its related compounds, Chem. Lett. 8, 703–704, 1996. 77. Yamamoto, T. and Kokubo, H., Selective stacking of HT-poly(3-n-alkylthiophene-2,5 diyl)s in mixed systems, J. Polym. Sci., Part B: Polym. Phys. 38, 84–87, 2000. 78. Yamamoto, T., Kokubo, H., and Morikita, T., Molecular alignment of head-to-tailtype poly(3-hexylthiophene-2,5-diyl) and related polymers and compounds on substrates, J. Polym. Sci., Part B: Polym. Phys. 39, 1713–1718, 2001. 79. Yamamoto, T. et al., Extensive studies on π-stacking of poly(3-alkylthiophene-2,5diyl)s and poly(4-alkylthiazole-2,5-diyl)s by optical spectroscopy, NMR analysis, light scattering analysis, and x-ray crystallography, J. Am. Chem. Soc. 120, 2047–2058, 1998. 80. Yamamoto, T. et al., Intermolecular pi-pi interaction of poly(4,4′-dialkyl-2,2′-bithiazole-5,5′-diyl)s and photoluminescence of the polymers, Chem. Lett. 2, 139–140, 1997. 81. Yang, C., Orfino, F.P., and Holdcroft, S., A phenomenological model for predicting thermochromism of regioregular and nonregioregular poly(3-alkylthiophenes), Macromolecules 29, 6510–6517, 1996. 82. Ong, B.S. et al., Structurally ordered polythiophene nanoparticles for high-performance organic thin-film transistors, Adv. Mater. 17, 1141, 2005.

8080_C005.fm Page 399 Friday, April 6, 2007 10:25 AM

Solution Deposition of Polymers

399

83. Wu, Y.O. et al., Controlled orientation of liquid-crystalline polythiophene semiconductors for high-performance organic thin-film transistors, Appl. Phys. Lett. 86 (14), 2005. 84. McCulloch, I. et al., Liquid-crystalline semiconducting polymers with high chargecarrier mobility, Nat. Mater. 5 (4), 328–333, 2006. 85. Wu, C.G. et al., Steric effect and mobility of the alkyl chain in regio-irregular poly3-alkylthiophenes, J. Polym. Sci., Part B: Polym. Phys. 37, 1763–1772, 1999. 86. Malik, S. and Nandi, A.K., Crystallization mechanism of regioregular poly(3-alkyl thiophene)s, J. Polym. Sci., Part B: Polym. Phys. 40, 2073–2085, 2002. 87. Mo, Z. et al., X-ray-scattering from polythiophene: Crystallinity and crystallographic structure, Macromolecules 18, 1972–1977, 1985. 88. Zhang, R. et al., Nanostructure dependence of field-effect mobility in regioregular poly(3-hexylthiophene) thin film field effect transistors, J. Am. Chem. Soc. 128, 3408, 2006. 89. DeLongchamp, D.M. et al., Variations in semiconducting polymer microstructure and hole mobility with spin-coating speed, Chem. Mater. 17, 5610–5612, 2005. 90. Yang, C.D. et al., Poly(2,7-phenanthrylene)s and poly(3,6-phenanthrylene)s as polyphenylene and poly(phenylenevinylene) analogues, Macromolecules 39, 5213–5221, 2006. 91. Degennes, P.G., Reputation of a polymer chain in presence of fixed obstacles, J. Chem. Phys. 55, 572, 1971. 92. Magonov, S.N. and Yerina, N.A., High-temperature atomic force microscopy of normal alkane C60H122 films on graphite, Langmuir 19 (3), 500–504, 2003. 93. Magonov, S.N. et al., Chain unfolding in single crystals of ultralong alkane C390H782 and polyethylene: An atomic force microscopy study, Macromolecules 36 (15), 5637–5649, 2003. 94. Ghosh, S.K. et al., Power law of molecular weight of the nucleation rate of folded chain crystals of polyethylene, Macromolecules 35 (18), 6985–6991, 2002. 95. Hocquet, S. et al., Lamellar and crystalline layer thickness of single crystals of narrow molecular weight fractions of linear polyethylene, Macromolecules 35 (13), 5025–5033, 2002. 96. Umemoto, S., Kobayashi, N., and Okui, N., Molecular weight dependence of crystal growth rate and its degree of supercooling effect, J. Macromol. Sci., Part B: Phys. 41, 923–938, 2002. 97. Umemoto, S. et al., Molecular weight dependence of primary nucleation rate of poly(ethylene succinate), J. Macromol. Sci., Part B: Phys. 42, 421–430, 2003. 98. Mena-Osteritz, E. et al., Two-dimensional crystals of poly(3-alkylthiophene)s: Direct visualization of polymer folds in submolecular resolution, Angew. Chem. Int. Ed. 39, 2680–2684, 2000. 99. Mena-Osteritz, E., Superstructures of self-organizing thiophenes, Adv. Mater. 14, 609–616, 2002. 100. Kasai, H. et al., STM observation of single molecular chains of π-conjugated polymers, Chem. Lett. 7, 696–697, 2002. 101. Meille, S.V. et al., Influence of molecular weight and regioregularity on the polymorphic behavior of poly(3-decylthiophenes), Macromolecules 30, 7898–7905, 1997. 102. Kline, R.J., Mcgehee, M.D., and Toney, M.F., Highly oriented crystals at the buried interface in polythiophene thin-film transistors, Nat. Mater. 5, 222–228, 2006. 103. Roichman, Y. and Tessler, N., Generalized Einstein relation for disordered semiconductors: Implications for device performance, Appl. Phys. Lett. 80, 1948–1950, 2002.

8080_C005.fm Page 400 Friday, April 6, 2007 10:25 AM

400

Organic Field-Effect Transistors

104. Zhang, R. et al., Nanostructure dependence of field-effect mobility in regioregular poly(3-hexylthiophene) thin film field effect transistors, J. Am. Chem. Soc. 128, 3480–3481, 2006. 105. Kratky, O. and Porod, G., Diffuse small-angle scattering of x-rays in colloid systems, J. Colloid Sci. 4, 35–70, 1949. 106. Aime, J.P. et al., Structural study of conducting polymers in solution, Synth. Met. 28, C407–C417, 1989. 107. Bates, F.S., Polymer–polymer phase behavior, Science 251, 898–905, 1991. 108. Rughooputh, S.D.D.V. et al., Chromism of soluble polythienylenes, J. Polym. Sci., Part B: Polym. Phys. 25, 1071–1078, 1987. 109. Stafstrom, S., Nilsson, J.O., and Salaneck, W.R., An experimental and theoretical study of the valence electron properties of crystal violet, Physica Scripta 37, 831–835, 1988. 110. Faid, K. et al., Chromic phenomena in regioregular and nonregioregular polythiophene derivatives, Chem. Mater. 7, 1390–1396, 1995. 111. Yue, S., Berry, G.C., and McCullough, R.D., Intermolecular association and supramolecular organization in dilute solution. 1. Regioregular poly(3-dodecylthiophene), Macromolecules 29, 933–939, 1996. 112. Langeveld-Voss, B.M.W. et al., Circular dichroism and circular polarization of photoluminescence of highly ordered poly{3,4-di[(S)-2-methylbutoxy]thiophene}, J. Am. Chem. Soc. 118, 4908–4909, 1996. 113. Leclerc, M. et al., Chromic phenomena in neutral polythiophene derivatives, Macromolec. Chem. Phys. 197, 2077–2087, 1996. 114. Apperloo, J.J. et al., Concentration-dependent thermochromism and supramolecular aggregation in solution of triblock copolymers based on lengthy oligothiophene cores and poly(benzyl ether) dendrons, Macromolecules 33, 7038–7043, 2000. 115. Sandstedt, C.A., Rieke, R.D., and Eckhardt, C.J., Solid-state and solvatochromic spectra of a highly regular polythiophene, Chem. Mater. 7, 1057–1059, 1995. 116. Schenning, A.P.H.J. et al., Supramolecular organization of alpha,alpha′-disubstituted sexithiophenes, J. Am. Chem. Soc. 124, 1269–1275, 2002. 117. Brustolin, F. et al., Highly ordered structures of amphiphilic polythiophenes in aqueous media, Macromolecules 35, 1054–1059, 2002. 118. Bidan, G., Guillerez, S., and Sorokin, V., Chirality in regioregular and soluble polythiophene: An internal probe of conformational changes induced by minute solvation variation, Adv. Mater. 8, 157, 1996. 119. Langeveld-Voss, B.M.W. et al., Inversion of optical activity of chiral polythiophene aggregates by a change of solvent, Macromolecules 31, 6702–6704, 1998. 120. McCullough, R.D., Ewbank, P.C., and Loewe, R.S., Self-assembly and disassembly of regioregular, water soluble polythiophenes: Chemoselective ionchromatic sensing in water, J. Am. Chem. Soc. 119, 633–634, 1997. 121. Chen, T.A. and Rieke, R.D., The 1st regioregular head-to-tail poly(3-hexylthiophene2,5-diyl) and a regiorandom isopolymer — Ni vs Pd catalysis of 2(5)-bromo-5(2)(bromozincio)-3-hexylthiophene polymerization, J. Am. Chem. Soc. 114, 10087–10088, 1992. 122. Arroyovillan, M.I. et al., Poly(N-(3-thienyl)alkanesulfonates): Synthesis, regioregularity, morphology, and photochemistry, Macromolecules 28, 975–984, 1995. 123. Langeveld-Voss, B.M.W. et al., Chiroptical properties of highly ordered di[(S)-2methylbutoxy] substituted polythiophene and poly(p-phenylene vinylene), Synth. Met. 84, 611–614, 1997.

8080_C005.fm Page 401 Friday, April 6, 2007 10:25 AM

Solution Deposition of Polymers

401

124. Delnoye, D.A.P. et al., Pi-conjugated oligomers and polymers with a self-assembled ladder-like structure, J. Am. Chem. Soc. 118, 8717–8718, 1996. 125. Miller, L.L. and Mann, K.R., Pi-dimers and pi-stacks in solution and in conducting polymers, Acc. Chem. Res. 29, 417–423, 1996. 126. Goto, H., Okamoto, Y., and Yashima, E., Solvent-induced chiroptical changes in supramolecular assemblies of an optically active, regioregular polythiophene, Macromolecules 35, 4590–4601, 2002. 127. Goto, H. and Yashima, E., Electron-induced switching of the supramolecular chirality of optically active polythiophene aggregates, J. Am. Chem. Soc. 124, 7943–7949, 2002. 128. Zhang, Z.B. et al., Chiroptical properties of poly{3,4-bis[(S)-2-methyloctyl]thiophene}, Macromolecules 35, 941–944, 2002. 129. Garreau, S. et al., Planar-to-nonplanar conformational transition in thermochromic polythiophenes: A spectroscopic study, Macromolecules 36, 692–697, 2003. 130. Liu, G.J., Qiao, L.J., and Guo, A., Diblock copolymer nanofibers, Macromolecules 29, 5508–5510, 1996. 131. Kim, D.H. et al., Single-crystal polythiophene microwires grown by self-assembly, Adv. Mater. 18, 719, 2006.

8080_C005.fm Page 402 Friday, April 6, 2007 10:25 AM

8080_C005.fm Page 403 Friday, April 6, 2007 10:25 AM

5.3

Solution Deposition of Oligomers

Howard E. Katz and Chad Landis CONTENTS 5.3.1 Introduction................................................................................................403 5.3.2 Conjugated Oligomers ...............................................................................404 5.3.3 Fused Ring Compounds ............................................................................409 5.3.4 Conclusion and Future Prospects ..............................................................414 References..............................................................................................................414

5.3.1 INTRODUCTION Electronic technology based on printable and largely organic conductors, dielectrics, and semiconductors is a major academic research theme and the basis of significant investment by established and emerging companies [1–4]. While the usual set of passive components such as resistors and capacitors would be needed to produce functional circuits, the key device in these technologies is the organic field-effect transistor (OFET). Platforms have been identified where organic semiconductors offer lower cost fabrication compared to silicon or provide functionality that is distinct from what can be obtained from noncovalent and rigid materials. Prototypes of OFET-based display drivers [5–10] have been demonstrated. The most compatible display media for OFET driver circuits are “electronic inks” — materials in which contrasting pigment particles or domains can be electrophoretically shifted into and out of a field of view. While requiring relatively high switching voltages, they draw little current during switching and virtually no current between switching cycles, matching the typical OFET output. These media provide convincing visual evidence of the effectiveness of the organic-based switches. However, liquid crystal and organic emitter-based materials have also been employed as the visual media. The high-throughput processes and mechanical flexibility associated with these prototypes may open the possibility for electronic newspapers and books onto which information can be downloaded, read, and reset with further information as the reader requests. A second proposed use of organic semiconductors is in radio frequency ID (RFID) tags [11,12]. Several hurdles must be overcome in order to make organic RFID practical. The first is the transistor speed, which will only be adequate for

403

8080_C005.fm Page 404 Friday, April 6, 2007 10:25 AM

404

Organic Field-Effect Transistors

frequencies > 1-kHz and 1-V operation with mobilities > 0.1 cm2/Vs and channel lengths < 100 μm. At present, these are the limits for easily printed, all-organic semiconductors. Higher frequencies, which are necessary for all but the most proximate detection schemes, will only arise from much shorter and higher mobility devices. In addition, rectification of the AC input to provide a DC logic signal is challenging in all-organic systems because the fastest diodes tend to be inorganic and not particularly amenable to printing. Schemes to accomplish rectification without diodes have been proposed. In the future, it might be possible to utilize the same kinds of organic semiconductors that have been developed for OFETs to make printable diodes. Related work aimed at printed solar cells is already in progress. Nonprinted organic solar cells with breakthrough efficiencies have recently been reported [13–20]. In addition, organic light-emitting diodes are already in commercial products [21], though the role of particularly high-mobility, printed organic semiconductors has not been extensively considered. While numerous manuscripts have appeared describing new or modified conjugated cores for hole-carrying organic semiconductor molecules and several recent publications have focused on new approaches to electron-carrying molecules in films, very few consider solution deposition of molecular solids in detail. Some recent reviews have discussed the subject relatively briefly [22–24]. Molecular solids offer potential advantages over polymers in that they are generally more ordered, leading to higher mobility, and are more readily purified, leading to better stability and reproducibility. However, the printing of a molecular solid from solution is far more challenging. Molecular solutions maintain low viscosity even as much of the solvent is removed, so film localization is problematic. As crystallites form within the deposited solution, they typically remain separated from each other rather than forming a continuous film. Conversely, at the point that a film forms on a substrate, the molecules are held so tightly in a crystal lattice that they cannot anneal into their best ordered morphology, so macroscale mobility suffers. Thus, the solution deposition of molecular solids requires attention to the molecular design, for the most favorable solvent and substrate interactions, and also the deposition conditions, so that the most favorable phases and highest degrees of continuity can be achieved. This review summarizes techniques that have evolved for the deposition of highmobility organic semiconductor films from solutions of properly designed conjugated molecules. Such molecules are distinguished from polymers by having molecular weights below 1,000. In addition, they tend to have high symmetries and a balance between the volumes of the conjugated cores and solubilizing/self-organizing side chains. The review will be presented in two main sections dealing with heterocyclic oligomers and fused ring compounds, respectively.

5.3.2 CONJUGATED OLIGOMERS The first soluble oligomer with significant field-effect mobility was 5,5′′′-dihexylquaterthiophene, described by the Garnier group in 1998 [25]. The chemical structure is shown in Figure 5.3.1. The compound exhibits a mesophase transistion below 100°C. The hexyl chains improve the ordering of the compound relative to unsubstituted quaterthiophene, which in turn is believed to lead to improved mobility as

8080_C005.fm Page 405 Friday, April 6, 2007 10:25 AM

Solution Deposition of Oligomers

C6H13

405

S

S S

S

C6H13

n Dihexylquartethiophene (n = 1) Dihexylquinquethiophene (n = 2)

R

S

S S

S S

S

R

Sexithiophenes: R = hexyl, butoxypropyl, diethylphosphonylbutyl

FIGURE 5.3.1 Chemical structure of alkyl-substituted oligothiophenes.

well. For sublimed films, the improvement is at least an order of magnitude. Importantly, a dihexylquarterthiophene film spin-coated from chloroform onto a hot, primed substrate had mobility only 60% lower. Transistors with vertical architecture were made from this solution [26]. It was shown at about the same time that the special properties of this oligomer could also be related to the large-area, single-crystal grains (Figure 5.3.2.) that could be grown at elevated temperature [27]. The ability to form films with such morphology is shared by few organic semiconductors; pentacene is a notably important example. The ability to tune the size of the grains by varying the substrate temperature during deposition was useful in demonstrating that grain boundaries are a site of chemical sensitivity in OFETs [28]. Attempts to employ solution processing on longer thiophene oligomers were limited by the drastically lower solubility of these compounds. While mobilities for compounds such as dihexyl and bis(butoxypropyl) sexithiophene were on the order of 0.01 cm2/Vs from films cast from solution, the solutions were <0.1% in concen-

FIGURE 5.3.2 Grain structure of dihexylquaterthiophene deposited at 80°C. The gap between the two horizontal, overlying electrodes is 40 μm.

8080_C005.fm Page 406 Friday, April 6, 2007 10:25 AM

406

Organic Field-Effect Transistors

S

S

S

S

S Bis (bithienylthiophene)

R

S

S S

R′

n

Alkynylthiophene oligomers: R = hexyl, R′ = hex-1-ynyl, n = 1; R = propyl, R′ = hex-1-ynyl, n = 2

FIGURE 5.3.3 Chemical structures of an oligothiophene with internal double bonds and asymmetric oligothiophene derivatives.

tration and had to be maintained at elevated temperatures [29]. Mobilities were not consistent across large areas, and the process could not be easily applied to printing. Greater solubility was obtained from sexithiophenes with side chains terminated in dibutylphosphonate [30]. Solutions could be spin-coated, and mobilities similar to those of sexithiophenes with linear side chains were obtained. However, extended annealing at elevated temperature was required in order to observe transistor action. Higher mobilities and better consistency over large areas resulted from the use of dihexylquinquethiophene on substrates made less polar through fluoroalkylation. Only moderately elevated temperatures were needed to produce highly active films of this compound [31,32]. Besides using side chains, internal double bonds were added among thiophene rings to produce a more soluble derivative. Bis(bithienylethenyl)thiophene was spincoated from N-methylpyrrolidone solution [33]. The mobility was reported as 0.0014 cm2/Vs. While most molecular solid semiconductors are end-to-end symmetric, a recent publication focuses on thiophene oligomers with two different hydrocarbon end groups (Figure 5.3.3). This work also emphasizes the desirability of mesophase transitions in the processing to make films. Terthiophene with a hexynyl and a propyl end group and quaterthiophene with a propyl and hexynyl end group both exist in smectic phases at room temperature. Time-of-flight mobilities for these phases approach 0.1 cm2/Vs [34]. The co-oligomerization of thiophenes with phenylenes was investigated in order to tune the highest occupied molecular orbital energies and make the films less susceptible to doping (Figure 5.3.4). These structures are not necessarily more soluble than all-thiophene oligomers, but the on/off ratios tend to be higher and the mobilities more reproducible. The first such compound to be deposited from solution was 1,4-bis(hexylbithienyl)benzene [35]. Mobilities ranged from 0.001 to 0.02 cm2/Vs with on/off ratios > 1,000. Cross-linked methacrylate and siloxane dielectrics were used, and active areas up to 2 cm2 were obtained. The nonchlorinated solvents xylene and toluene were usable. Extensive data were obtained from the related compound 5,5′-bis(4-hexylphenyl)-2,2′-bithiophene, abbreviated 6PTTP6 [36]. Xylene was an ideal solvent for casting films of this compound, keeping the substrate 5–30° below the boiling point.

8080_C005.fm Page 407 Friday, April 6, 2007 10:25 AM

Solution Deposition of Oligomers

407

S

C6H13

S

S

C6H13

S

C6H13

S

S

C6H13

Phenylene-thiophene oligomers: bis(hexylbithienyl) benzene and 6PTTP6

FIGURE 5.3.4 Chemical structures of phenylene-thiophene co-oligomers.

Highly ordered films were obtained on primed oxide and glass resin surfaces. The on/off ratios exceeded 10,000, with the off currents reproducibly low even without having purified the semiconductor by vacuum sublimation. Optimization of multiple processing parameters, including surface functionalization, substrate temperature, and overlying atmosphere (saturation with xylene vapor), led to a highly ordered film by virtue of the traversal of lyotropic mesophases (Figure 5.3.5) [37]. The ordering was monitored by polarized optical microscopy and x-ray scattering. The highest solution-deposited oligomer mobility yet obtained, >0.1 cm2/Vs, was derived from an optimal sample (Figure 5.3.6). This semiconductor was used to introduce the concept of solution-deposited nonvolatile organic transistors [36]. (a)

(b)

FIGURE 5.3.5 Images of lyotropic mesophases from a film of solution-deposited 6PTTP6.

8080_C005.fm Page 408 Friday, April 6, 2007 10:25 AM

408

Organic Field-Effect Transistors

0.012

–140 –120 –100 –80 –60 –40 –20 0 20

–100 V

0.010

–80 V

ISD1/2 (A1/2)

ISD (μA)

Y=A+B∗X Parameter Value Error ---------------------------------------------A –3.34643E-4 3.30694E-4 B –1.19112E-4 4.98541E-6 ---------------------------------------------R SD N P ---------------------------------------------–0.99738 3.15305E-4 5 1.60685E-4 ----------------------------------------------

–60 V –40 V –20 V 0

0.008 0.006 0.004 0.002 0

–20 –40 –60 –80 –100 Source-Drain voltage (V)

–20 –40

–60 –80 –100

Gate voltage (V)

FIGURE 5.3.6 Current-voltage characteristics of a solution-deposited 6PTTP6.

C8F5OC

S

S

S

S

COC8F5

Bis (pentaﬂuorophenacyl) quarterthiophene

FIGURE 5.3.7 Chemical structure of the first n-channel, solution-deposited, thiophene oligomer.

While all of these soluble oligomers have displayed p-channel character (hole transporting ability), there is no reason why oligomers of high enough purity and low enough energies of the lowest unoccupied molecular orbitals would not act as n-channel semiconductors. The persistent addition of fluoro and oxo substituents to end-substituted thiophene oligomers has recently led to the first n-channel solution-deposited thiophene oligomers. A mobility of 0.2 cm2/Vs was obtained by drop-casting bis(pentafluorophenacyl)quaterthiophene at 120°C (Figure 5.3.7) [38]. The compound was also successfully blended with a polymeric analog for improved printability. Evidence from dihexylquaterthiophene blended with regioregular poly(3-hexylthiophene) suggests that the blending approach may be more generally applicable [39]. The oligomer considered as an additive improved the mobility of the polymer by a factor of 10. Because of the limited solubility conferred by linear side chains and the desirability of more concentrated solutions for various printing methods, branched side chains are attractive. However, branched chains tend to detract from the ordering of molecular solid films and inhibit high mobility. A collaborative group at UC Berkeley is exploring oligothiophenes with highly branched but thermally cleavable ester side chains [40–43]. The carbon attached to the terminal alpha positions of the oligomer carries an aliphatic and a carbonyloxy substituent. The carbonyloxy is further branched. Thus, highly soluble derivatives are obtained, and the stability of the cation centered on the alpha carbon leads to thermal lability. On heating, the

8080_C005.fm Page 409 Friday, April 6, 2007 10:25 AM

Solution Deposition of Oligomers

O

S

S S

S

n

O

O

C6H13

C6H13

n=1–4

O

C4H9

409

C4H9

S

S S

S n

n=1–4 Oligothiophene ester thermolysis

FIGURE 5.3.8 Schematic showing the thermolysis of highly soluble oligothiophene derivatives.

carbonyloxy group is volatilized and a short hydrocarbon chain is left at each alpha position (Figure 5.3.8). Films with a wide variety of morphologies have been obtained, depending on the nature of the substituents, substrate surface treatments, and temperature excursions. Mobilities approaching 0.1 cm2/Vs and very high on/off ratios — into the millions — have been obtained. Materials were deposited by spincoating, dip-casting, and inkjet printing. Active films just a few monolayers thick were demonstrated.

5.3.3 FUSED RING COMPOUNDS The most studied semiconducting fused ring molecule is pentacene, due to its high mobility (1 cm2/Vs) in OFETs and good film-forming capabilities [44,45]. Unfortunately, pentacene is essentially insoluble and is therefore not suitable for solution processing, which is necessary in order to realize cheap, large-area electronics. One example of direct solution processing of pure pentacene requires deposition of a heated solution of pentacene in 1,2-dichlorobenzene and trichlorobenzene under a nitrogen atmosphere [46]. The devices prepared by this method showed a mobility of 0.45 cm2/Vs and on/off ratios of 105. However, the most popular response to the insolubility of pentacene has been the preparation of soluble precursors that can be deposited on a substrate and then converted into pentacene. The first example of a pentacene precursor was a tetrachlorocyclohexadiene adduct prepared by the Müllen group (Figure 5.3.9a) [47]. This derivative is soluble in dichloromethane and forms good films by spin-coating. After deposition, the pentacene film is formed by a thermally activated retro Diels–Alder reaction expelling tetrachlorobenzene as the by-product. The hole mobilities of OFETs prepared from these pentacene precursors depended greatly on the annealing temperature

8080_C005.fm Page 410 Friday, April 6, 2007 10:25 AM

410

Organic Field-Effect Transistors

Cl Cl

O

Cl O

Cl

(a)

N S

O

R

(b)

O O

O

N N

(c)

FIGURE 5.3.9 Structures of pentacene precursors.

used, ranging from 5 × 10–5 to 3 × 10–3 cm2/Vs [48]. More recent optimized conditions require annealing at 200°C for 5 sec; thin film transistors prepared in this manner have highest hole mobilities of 0.2 cm2/Vs and on/off current ratios of 106 [49]. This derivative has also recently been used to prepare drivers for a 3.5-cm2 flexible display [5]. More numerous examples of pentacene precursors employ sulfinylamide functionality (Figure 5.3.9b). The first of these derivatives was prepared by the Afzali group and utilized a chloroform-soluble N-sulfinylacetamide derivative (Figure 5.3.9b, R = CH3) where pentacene films were formed thermally by a retro Diels–Alder reaction [50]. There was still a strong temperature dependence between the annealing temperature of the film and the resulting mobility of the devices, with OFETs prepared by annealing at 200°C resulting in hole mobilities of 0.42 cm2/Vs, while devices prepared at 130°C demonstrated mobilties of 0.13 cm2/Vs. Other sulfinylamides have been prepared that allow for differing solubilities [51] or even derivatives that undergo photoinduced retro Diels–Alder reactions that allow for high-resolution features to be printed by curing only the desired regions [52,53]. In one example, a photosensitive precursor was used to pattern an OFET with features of 40 μm and with mobilities of 0.25 cm2/Vs [53]. However, a major problem with the pentacene precursor method is that a residual amount of the Diels–Alder adduct remains in the film and limits the carrier mobility of the final pentacene film [54]. To limit this problem, pentacene precursors with smaller or more easily eliminated Diels–Alder adducts have recently appeared, but the field-effect transistor characteristics of these new derivatives are yet to be published (Figure 5.3.9c) [55,56]. The concept of soluble precursors has also recently been extended to other fused aromatic systems. For example, tetrabenzoporphorins have been prepared from thermal annealing of a soluble precursor [57]. The precursor in this case is a tetrabicyclo derivative (Figure 5.3.10a). Upon thermal annealing at 210°C for 5 min, this derivative eliminates four ethylene molecules, resulting in the electroactive tetrabenzoporphyrin (Figure 5.3.10b). Bottom-contact OFETs prepared from this material had hole mobilities of 0.017 cm2/Vs and on/off current ratios of 105, which is similar to OFETs prepared from vacuum evaporated tetrabenzoporphyrin [58]. Another series of devices from this porphyrin derivative utilized a coating of an n-channel material to limit the effects of oxygen and moisture on the performance of the device [59]. These devices were top-contact configuration, with the n-channel material coated after the electrodes are deposited. The OFETs prepared in this manner displayed

8080_C005.fm Page 411 Friday, April 6, 2007 10:25 AM

Solution Deposition of Oligomers

411

N NH

N NH

NH

NH

N

N

(a)

(b)

FIGURE 5.3.10 Structure of tetrabenzoporphyrin (a) and its precursor (b).

hole mobilities of 5 × 10–3 cm2/Vs, similar to top-contact OFETs prepared without the n-channel overlayer [60]. There have been few examples of direct functionalization of fused aromatic systems. Work done by the Katz group used anthradithiophenes instead of pentacene because of the greater oxidative stability resulting from the two fused thiophene rings. In particular, a solution-processed OFET was prepared from the dihexylanthradithiophene derivative (Figure 5.3.11a) [61]. This derivative is soluble in hot chlorobenzene and was formed into films on substrates under vacuum held at 100°C. The OFETs prepared from this derivative had hole mobilities ranging from 0.01 to 0.02 cm2/Vs, but were difficult to reproduce. R

S C6H13

C6H13

S

R (a)

R = (μPr)3Si, Me3Si, Et3Si

(b)

SiEt3 O

O

C7F15CH2 N

N

O

O

S S

SiEt3 (c)

FIGURE 5.3.11 Structures of soluble fused ring derivatives.

(d)

CH2C7F15

8080_C005.fm Page 412 Friday, April 6, 2007 10:25 AM

412

Organic Field-Effect Transistors

The Anthony group has prepared an extensive library of soluble acene derivatives. The derivatives employ trialkylsilyl solublizing groups held from a pentacene core by ethynyl spacers at the 6 and 13 positions (Figure 5.3.11b) [62,63]. The result of this functionality is a change in the molecular order of these derivatives from the herringbone orientation of unsubstituted pentacene to a one- or two-dimensional πstacking arrangement. These derivatives also demonstrate greatly increased solubility and oxidative stability compared to unsubstituted pentacene. They are soluble in a wide variety of organic solvents and are air stable as a solution for days and as crystals for weeks. One of the best of these derivatives is 6,13-(bistriethylsilylethynyl)anthradithiophene (Figure 5.3.11c) [64,65]. OFETs prepared by toluene solution casting of a mixture of the isomers of this derivative resulted in hole mobilities of up to 1 cm2/Vs and on/off current ratios of 107. Soluble derivatives of fused aromatic systems are not limited to p-channel materials; however, they are far fewer in number. A soluble derivative of NTCDA has recently been developed that can be cast from α,α,α-trifluorotoluene (Figure 5.3.11d) [66]. Devices prepared from this derivative showed electron mobilities of ~0.01 cm2/Vs [67]. This is currently the only nonfullerene solution-cast nchannel material. Directly functionalized discotic molecules have also recently been prepared to serve as liquid crystalline materials, which could be used for electronic devices. One of the impressive points of these materials is their ability to self-assemble into columnar arrays. One particular example involves a derivative of hexabenzocoronene, which is induced into a high mesoscopic order by the introduction of a magnetic field (Figure 5.2.12a) [68]. OFETs prepared from this liquid-crystalline material showed mobility of 10–4 cm2/Vs when the magnetic field is held perpendicular to the charge transport direction and 3 × 10–6 cm2/Vs when it is held parallel to the charge transport direction. A n-channel example of these materials is a tris(pentafluorobenzylester) derivative of hexaazatrinaphthylene (HAT-NA, Figure 5.3.12b) [69]. Films of the isomer shown in the picture formed textured films by rapid cooling from a melt. While OFETs were not prepared from this derivative, the measured effective electron mobility of the film was found to be 0.07 cm2/Vs. A soluble fullerene derivative is a desirable semiconducting material due to the effective electron transporting capability of C60. The most used soluble fullerene derivative is [6,6]-phenyl-C61-butyric acid methyl ester (PCBM, Figure 5.3.13b); the majority of research into this derivative is in its use as the n-channel material in bulk heterojunction solar cells [70,71]. This derivative is soluble in many organic solvents and tends to form amorphous films that result in OFETs with electron mobilities up to 0.1 cm2/Vs [72,73]. Mobilities were improved to 0.2 cm2/Vs by using a PVA dielectric [74,75]. This derivative has also been used to prepare ambipolar OFETs [76]. More recently, another derivative has been prepared in an attempt to induce greater molecular order in spin-cast films of soluble C60 and thereby improve the electron mobility of the films. This derivative, C60MC12 (Figure 5.3.13a), employs a dodecyl alkyl chain for solubility and to induce long-range order by interacalation of the alkyl chains [77]. Spin-cast films of this derivative show greater order and OFETs prepared from it display electron mobilities of 0.067 cm2/Vs and on/off current ratios of 1.6 × 105.

8080_C005.fm Page 413 Friday, April 6, 2007 10:25 AM

Solution Deposition of Oligomers

413

C12H25

R C12H25

C12H25 N

N

N R

N N

N R

C12H25

C12H25

R = CO2CH2C6F5

C12H25 (a)

(b)

FIGURE 5.3.12 Chemical structure of two soluble discotic organic semiconductors. C12H22 O

Me

O

N

C60MC12

[6, 6]- PCBM

FIGURE 5.3.13 Structures of two soluble fullerene derivatives.

During the past decade, the limits of carrier mobility have continuously been sought. Currently, the upper limit for organic single crystals is rubrene with an intrinsic mobility of ~15 cm2/Vs (Figure 5.3.14, left) [78]. Evaporated films of rubrene have not been effective and methods of preparing thin film devices of rubrene have not been fruitful. One unique method has been developed to prepare rubrenebased OFETs. This relies on an analog of rubrene (diphenylanthracene; Figure 5.3.14, right) to act as a glass-inducing impurity to inhibit crystal growth of the rubrene until after casting of the film [79]. A 55 wt% solution of rubrene with diphenylanthracene and an additional 5 wt% of ultrahigh-molecular-weight polystyrene (UHMW-PS) was cast from toluene (the UHMW-PS acting as a mechanical support for the film). The films were held at the crystallization temperature for rubrene in this particular mixture (235°C) for 2–5 min and then cooled. Devices prepared from this method showed hole mobilities up to 0.7 cm2/Vs and on/off current rations of 107.

8080_C005.fm Page 414 Friday, April 6, 2007 10:25 AM

414

Organic Field-Effect Transistors

FIGURE 5.3.14 Chemical structure of rubrene and diphenylanthracene.

5.3.4 CONCLUSION AND FUTURE PROSPECTS This chapter has described numerous approaches to the deposition of OFET semiconductor films from solutions of organic compounds. Many of these films were demonstrated in devices with mobilities from 0.01–1 cm2/Vs and could in principle be patterned through printing processes. Factors limiting the usability of these compounds in printed logic circuits are similar to the limitations on organic semiconductor use in general: semiconductor stability, uniformity of coverage and electrical properties, and interfacial effects. In addition, physical properties of the solutions, such as surface energy and viscosity, need to be optimized for particular printing approaches. Little systematic work has been done along these lines, leaving a significant opportunity to develop these solutiondeposition methods taking the liquid-phase properties into account. Judging from the promising electronic properties reported for individual devices and the appearance of initial circuit prototypes, such development could prove particularly fruitful.

REFERENCES 1. Faupel, F. et al., Organic electronics, introduction to special section, J. Mater. Res. 19, 1887–1888, 2004. 2. Gamota, D., Brazis, P., and Kalyanasundaram, K., Printed organic and molecular electronics, Kluwer Academic Publishers, Norwell, MA, 2004. 3. Jenekhe, S., The special issue on organic electronics, Chem. Mater. 16, 4381–4382, 2004. 4. Kagan, C. and Andry, P., Thin film transistors, Marcel Dekker, New York, 2004. 5. Gelinck, G.H. et al., Flexible active-matrix displays and shift registers based on solution-processed organic transistors, Nat. Mater. 3, 106–110, 2004. 6. Blanchet, G. and Rogers, J., Printing techniques for plastic electronics, J. Imag. Sci. Tech. 47, 296–303, 2003. 7. Rogers, J.A. et al., Paper-like electronic displays: Large-area rubber-stamped plastic sheets of electronics and microencapsulated electrophoretic inks, Proc. Natl. Acad. Sci. U.S.A 98, 4835–4840, 2001. 8. Kim, Y.H. et al., Active-matrix liquid crystal display using solution-based organic thin film transistors on plastic substrates, Displays 25, 167–170, 2004.

8080_C005.fm Page 415 Friday, April 6, 2007 10:25 AM

Solution Deposition of Oligomers

415

9. Fujisaki, Y. et al., Liquid crystal display cells fabricated on plastic substrate driven by low-voltage organic thin-film transistor with improved gate insulator and passivation layer, Jpn. J. Appl. Phys. Part 1 44, 3728–3732, 2005. 10. Ohta, S. et al., Active matrix driving organic light-emitting diode panel using organic thin-film transistors, Jpn. J. Appl. Phys. Part 1 44, 3678–3681, 2005. 11. Printed plastic chips promise to cut the cost of RFID, IEE Rev. 51, 18, 2005. 12. Baude, P.F. et al., Pentacene-based radio-frequency identification circuitry, App. Phys. Lett. 82, 3964–3966, 2003. 13. Velusamy, M. et al., Organic dyes incorporating low-band-gap chromophores for dyesensitized solar cells, Org. Lett. 7, 1899–1902, 2005. 14. Ackermann, J. et al., Highly efficient hybrid solar cells based on an octithiopheneGaAs heterojunction, Adv. Funct. Mater. 15, 810–817, 2005. 15. Yang, X.N. et al., Nanoscale morphology of high-performance polymer solar cells, Nano Lett. 5, 579-583, 2005. 16. Schmidt-Mende, L. et al., Organic dye for highly efficient solid-state dye-sensitized solar cells, Adv. Mater. 17, 813, 2005. 17. Hara, K. et al., Novel conjugated organic dyes for efficient dye-sensitized solar cells, Adv. Funct. Mater. 15, 246–252, 2005. 18. Yang, F., Shtein, M., and Forrest, S.R., Controlled growth of a molecular bulk heterojunction photovoltaic cell, Nat. Mater. 4, 37–41, 2005. 19. Forrest, S.R., The limits to organic photovoltaic cell efficiency, MRS Bull. 30, 28–32, 2005. 20. Chu, C.-W. et al., Efficient photovoltaic energy conversion in tetracene-C60-based heterojunctions, App. Phys. Lett. 86, 243506, 2005. 21. Holder, E., Langeveld, B.M.W., and Schubert, U.S., New trends in the use of transition metal–ligand complexes for applications in electroluminescent devices, Adv. Mater. 17, 1109–1121, 2005. 22. Parashkov, R. et al., Large area electronics using printing methods, Proc. IEEE 93, 1321–1329, 2005. 23. Subramanian, V. et al., Progress toward development of all-printed RFID tags: Materials, processes, and devices, Proc. IEEE 93, 1330–1338, 2005. 24. Katz, H.E., Recent advances in semiconductor performance and printing processes for organic transistor-based electronics, Chem. Mater. 16, 4748–4756, 2004. 25. Garnier, F. et al., Dihexylquaterthiophene, a two-dimensional liquid crystal-like organic semiconductor with high transport properties, Chem. Mater. 10, 3334–3339, 1998. 26. Garnier, F., Hajlaoui, R., and El Kassmi, M., Vertical device architecture by molding of organic-based thin film transistor, App. Phys. Lett. 73, 1721–1723, 1998. 27. Katz, H.E., Lovinger, A.J., and Laquindanum, J.G., Alpha,omega-dihexylquaterthiophene: A second thin film single-crystal organic semiconductor, Chem. Mater. 10, 457, 1998. 28. Someya, T. et al., Vapor sensing with alpha,omega-dihexylquarterthiophene fieldeffect transistors: The role of grain boundaries, App. Phys. Lett. 81, 3079–3081, 2002. 29. Katz, H.E., Laquindanum, J.G., and Lovinger, A.J., Synthesis, solubility, and fieldeffect mobility of elongated and oxa-substituted alpha,omega-dialkyl thiophene oligomers. Extension of “polar intermediate” synthetic strategy and solution deposition on transistor substrates, Chem. Mater. 10, 633–638, 1998. 30. Afzali, A., Breen, T.L., and Kagan, C.R., An efficient synthesis of symmetrical oligothiophenes: Synthesis and transport properties of a soluble sexithiophene derivative, Chem. Mater. 14, 1742–1746, 2002.

8080_C005.fm Page 416 Friday, April 6, 2007 10:25 AM

416

Organic Field-Effect Transistors

31. Katz, H.E. et al., Solution-phase deposition of oligomeric TFT semiconductors, Synth. Met. 102, 897–899, 1999. 32. Li, W. et al., Field-effect transistors based on thiophene hexamer analogues with diminished electron donor strength, Chem. Mater. 11, 458–465, 1999. 33. Dimitrakopoulos, C.D. et al., Trans-trans-2,5-bis-[2-{5-(2,2′-bithienyl)}ethenyl] thiophene: Synthesis, characterization, thin film deposition and fabrication of organic field-effect transistors, Syn. Met. 89, 193–197, 1997. 34. Funahashi, M. and Hanna, J., High carrier mobility up to 0.1 cm(2)/Vs and a wide mesomorphic temperature range of alkynyl-substituted terthiophene and quaterthiophene derivatives, Mol. Crystl. Liq. Crystl. 436, 1179–1189, 2005. 35. Hong, X.M. et al., Thiophene-phenylene and thiophene-thiazole oligomeric semiconductors with high field-effect transistor on/off ratios, Chem. Mater. 13, 4686–4691, 2001. 36. Mushrush, M. et al., Easily processable phenylene-thiophene-based organic fieldeffect transistors and solution-fabricated nonvolatile transistor memory elements, J. Am. Chem. Soc. 125 (31), 9414–9423, 2003. 37. Katz, H.E. et al., Mesophase transitions, surface functionalization, and growth mechanism of semiconducting 6PTTP6 films from solution, J. Phys. Chem. B 108, 8567–8571, 2004. 38. Letizia, J.A. et al., High electron mobility in solution-cast and vapor-deposited phenacyl-quaterthiophene-based field-effect transistors: Toward n-type polythiophenes, J. Am. Chem. Soc. 127, 13476–13477, 2005. 39. Russell, D.M. et al., Blends of semiconductor polymer and small molecule molecular crystals for improved-performance thin-film transistors, App. Phys. Lett. 87, 222109, 2005. 40. Chang, P.C. et al., Film morphology and thin film transistor performance of solutionprocessed oligothiophenes, Chem. Mater. 16, 4783–4789, 2004. 41. Murphy, A.R. et al., Self-assembly, molecular ordering, and charge mobility in solution-processed ultrathin oligothiophene films, Chem. Mater. 17, 6033–6041, 2005. 42. Murphy, A.R. et al., Organic thin film transistors from a soluble oligothiophene derivative containing thermally removable solubilizing groups, J. Am. Chem. Soc. 126, 11750–11750, 2004. 43. DeLongchamp, D.M. et al., Direct correlation of organic semiconductor film structure to field-effect mobility, Adv. Mater. 17, 2340–2344, 2005. 44. Lin, Y.Y. et al., Stacked pentacene layer organic thin-film transistors with improved characteristics, IEEE Electron Device Lett. 18, 606–608, 1997. 45. Sheraw, C.D. et al., Organic thin-film transistor-driven polymer-dispersed liquid crystal displays on flexible polymeric substrates, App. Phys. Lett. 80, 1088–1090, 2002. 46. Minakata, T. and Natsume, Y., Direct formation of pentacene thin films by solution process, Syn. Met. 153, 1–4, 2005. 47. Brown, A.R. et al., Precursor route pentacene metal-insulator-semiconductor fieldeffect transistors, J. Appl. Phys. 79, 2136–2138, 1996. 48. Jarrett, C.P. et al., Field-effect transistor studies of precursor-pentacene thin films, Syn. Met. 85, 1403–1404, 1997. 49. Herwig, P.T. and Mullen, K., A soluble pentacene precursor: Synthesis, solid-state conversion into pentacene and application in a field-effect transistor, Adv. Mater. 11, 480–484, 1999. 50. Afzali, A., Dimitrakopoulos, C.D., and Breen, T.L., High-performance, solutionprocessed organic thin film transistors from a novel pentacene precursor, J. Am. Chem. Soc. 124, 8812–8813, 2002.

8080_C005.fm Page 417 Friday, April 6, 2007 10:25 AM

Solution Deposition of Oligomers

417

51. Afzali, A., Kagan, C.R., and Traub, G.P., N-sulfinylcarbamate-pentacene adduct: A novel pentacene precursor soluble in alcohols, Syn. Met. 155, 490–494, 2005. 52. Afzali, A., Dimitrakopoulos, C.D., and Graham, T.O., Photosensitive pentacene precursor: Synthesis, photothermal patterning, and application in thin-film transistors, Adv. Mater. 15, 2066–2070, 2003. 53. Weidkamp, K.P. et al., A photopatternable pentacene precursor for use in organic thin-film transistors, J. Am. Chem. Soc. 126, 12740–12741, 2004. 54. Jurchescu, O.D., Baas, J., and Palstra, T.T.M., Effect of impurities on the mobility of single crystal pentacene, App. Phys. Lett. 84, 3061–3063, 2004. 55. Joung, M.J. et al., New soluble pentacene precursors for the application of organic thin-film transistors, Bull. Korean Chem. Soc. 24, 1862–1864, 2003. 56. Yamada, H. et al., Photochemical synthesis of pentacene and its derivatives, Chemistry-A Eur. J. 11, 6212–6220, 2005. 57. Aramaki, S., Sakai, Y., and Ono, N., Solution-processable organic semiconductor for transistor applications: Tetrabenzoporphyrin, App. Phys. Lett. 84, 2085–2087, 2004. 58. Checcoli, P. et al., Tetra-phenyl porphyrin based thin film transistors, Syn. Met. 138, 261–266, 2003. 59. Shea, P.B. et al., Methanofullerene-coated tetrabenzoporphyrin organic field-effect transistors, App. Phys. Lett. 87, 2005. 60. Shea, P.B. et al., Electrical properties of staggered electrode, solution-processed, polycrystalline tetrabenzoporphyrin field-effect transistors, IEEE Trans. Elec. Dev. 52, 1497–1503, 2005. 61. Laquindanum, J.G., Katz, H.E., and Lovinger, A.J., Synthesis, morphology, and fieldeffect mobility of anthradithiophenes, J. Am. Chem. Soc. 120, 664–672, 1998. 62. Anthony, J.E. et al., Functionalized pentacene: Improved electronic properties from control of solid-state order, J. Am. Chem. Soc. 123, 9482–9483, 2001. 63. Anthony, J.E., Eaton, D.L., and Parkin, S.R., A road map to stable, soluble, easily crystallized pentacene derivatives, Org. Lett. 4, 15–18, 2002. 64. Payne, M.M. et al., Stable, crystalline acenedithiophenes with up to seven linearly fused rings, Org. Lett. 6, 3325–3328, 2004. 65. Payne, M.M. et al., Organic field-effect transistors from solution-deposited functionalized acenes with mobilities as high as 1 cm(2)/V-s, J. Am. Chem. Soc. 127, 4986–4987, 2005. 66. Katz, H.E. et al., A soluble and air-stable organic semiconductor with high electron mobility, Nature 404, 478–481, 2000. 67. Katz, H.E. et al., Naphthalenetetracarboxylic diimide-based n-channel transistor semiconductors: Structural variation and thiol-enhanced gold contacts, J. Am. Chem. Soc. 122, 7787–7792, 2000. 68. Shklyarevskiy, I.O. et al., High anisotropy of the field-effect transistor mobility in magnetically aligned discotic liquid-crystalline semiconductors, J. Am. Chem. Soc. 127, 16233–16237, 2005. 69. Kaafarani, B.R. et al., High charge-carrier mobility in an amorphous hexaazatrinaphthylene derivative, J. Am. Chem. Soc. 127, 16358–16359, 2005. 70. Shaheen, S.E. et al., 2.5% Efficient organic plastic solar cells, App. Phys. Lett. 78, 841–843, 2001. 71. Yu, G. et al., Polymer photovoltaic cells — Enhanced efficiencies via a network of internal donor-acceptor heterojunctions, Science 270, 1789–1791, 1995. 72. Waldauf, C. et al., Solution-processed organic n-type thin-film transistors, Adv. Mater. 15, 2084–2088, 2003.

8080_C005.fm Page 418 Friday, April 6, 2007 10:25 AM

418

Organic Field-Effect Transistors

73. Lee, T.W. et al., All-solution-processed n-type organic transistors using a spinning metal process, Adv. Mater. 17, 2180–2184, 2005. 74. Anthopoulos, T.D. et al., Ambipolar organic field-effect transistors based on a solution-processed methanofullerene, Adv. Mater. 16, 2174–2179, 2004. 75. Singh, T.B. et al., Fabrication and characterization of solution-processed methanofullerene-based organic field-effect transistors, J. Appl. Phys., 97, 083714, 2005. 76. Shkunov, M. et al., Ambipolar field-effect transistors based on solution-processable blends of thieno[2,3-b]thiophene terthiophene polymer and methanofullerenes, Adv. Mater. 17, 2608–2611, 2005. 77. Chikamatsu, M. et al., Solution-processed n-type organic thin-film transistors with high field-effect mobility, App. Phys. Lett. 87, 203504, 2005. 78. Sundar, V.C. et al., Elastomeric transitor stamps: Reversible probing of charge transport in organic crystals, Science 303, 1644–1646, 2004. 79. Stingelin-Stutzmann, N. et al., Organic thin-film electronics from vitreous solutionprocessed rubrene hypereutectics, Nat. Mater. 4, 601–606, 2005.

8080_C005.fm Page 419 Friday, April 6, 2007 10:25 AM

5.4

Inkjet Printed Organic Thin Film Transistors

Ana Claudia Arias CONTENTS 5.4.1 Introduction................................................................................................419 5.4.2 Subtractive Methods — Printing Electrodes .............................................420 5.4.3 Additive Methods — Printing Semiconductors and Encapsulation Layers.........................................................................................................422 5.4.4 Display Backplane Fabrication..................................................................427 References..............................................................................................................431

5.4.1 INTRODUCTION Materials development together with the better understanding of organic-based transistors has enabled solution-processed materials that are stable in air, show good performance, are compatible with flexible substrates, and can be used in applications where thin film transistors (TFTs) are the building blocks of electronic circuits. Suitable applications for organic TFTs are large-area devices, such as active matrix displays, where high switching speeds may not be essential. The significance of polymers in particular is that they can be deposited from solution, therefore allowing device patterning by direct marking techniques. Inkjet printing is an attractive process because it is low cost, applicable to large-area processing, and compatible with flexible substrates, and can be adapted to high-throughput manufacturing processes, such as roll-to-roll printing methods. The technique provides a drop-ondemand digital deposition process without the need for physical masks, and materials are only applied where they are needed, decreasing materials cost and environmental impact. A variety of printing approaches has been used to fabricate TFTs. Each has different attributes with advantages and disadvantages regarding registration, process temperature, and device performance. For example, contact printing has been demonstrated for patterning electronics, but tends to have poor registration accuracy for multilayer structures. Patterning techniques using a fixed printing master are ill suited to flexible substrates unless their dimensional stability can be precisely controlled. In this section, different approaches used to fabricate organic-based TFTs using inkjet printing as a patterning method and as a deposition method will be reviewed.

419

8080_C005.fm Page 420 Friday, April 6, 2007 10:25 AM

420

Organic Field-Effect Transistors

The first use of inkjet printing on the fabrication of organic-based TFTs was to pattern the electrodes [1,2]. Printed source-drain electrodes were achieved by combining inkjet printing with photolithography, achieving small features in one layer of the device while the gate electrodes were directly printed on top of the gate dielectric layer [1]. A different approach used to pattern electrodes by inkjet printing is the combination of inkjet printing with conventional wet etching processes [3,4]. In order to achieve a patterned device feature, subtractive steps are used in both methods. These techniques are reviewed in Section 5.4.2. Direct writing was used later to deposit polymeric semiconductors. In direct writing, there are no subtractive steps and materials are not wasted. However, device dimensions are limited by ejector sizes and interactions between solution and substrate. The fundamental parameters controlling inkjet printed liquids are the viscosity and surface energy [5]. The pattern formed when an ejected drop of liquid hits the surface depends in large part on the ink–surface interaction. The wetting contact angle determines the spread of a liquid drop on the surface and depends on the relative surface energy of the solid–liquid, solid–vapor, and liquid–vapor interfaces. High-energy surfaces result in a small wetting angle and an extended drop, while a low surface energy results in a smaller footprint. The surface energy and contact angle also relate to the adhesion of the liquid to the surface. Strong adhesion is associated with good wetting and low adhesion with high contact angles (poor adhesion). Unfortunately, most situations need a high contact angle to limit the spread of the printed liquid and good adhesion to the surface to allow further processing [6]. To inkjet print fine features on a flat surface is not an easy task because of the difficulty in controlling the spread of a liquid on a free surface. The relation between printed line-width and contact angle is displayed in Figure 5.4.1. The solid lines are obtained from a simple model of a small volume of liquid with a cylindrical surface. The measurements of printed nanoparticle metals on different surfaces, shown in Figure 5.4.1, follow the expected trend; the printed line width decreases as the contact angle increases [6]. The printed feature size will not only determine adhesion of material to the substrate, but will also play an important role on the design of devices. Important parameters related to design of printed TFT backplanes are discussed in Section 5.4.4.

5.4.2 SUBTRACTIVE METHODS — PRINTING ELECTRODES Inkjet printing can be used to deposit resist materials that can then replace photoresist in the fabrication of conventional TFT amorphous silicon (a-Si) processes [3,7]. In this technique, conventionally deposited materials (metal, semiconductors, and dielectrics) can be patterned by printing following the steps shown in Figure 5.4.2. In this process, a thin metal film — for example, gold — is deposited everywhere. But instead of coating with photoresist and exposing, the mask is inkjet printed directly onto the layer, as shown in Figure 5.4.2(b). The amount of wax reflow that occurs before solidification can be adjusted by controlling the substrate temperature. Drop sizes from 20–40 μm are possible with industrially produced printheads.

8080_C005.fm Page 421 Friday, April 6, 2007 10:25 AM

Inkjet Printed Organic Thin Film Transistors

(a)

421 (b) Φ = 30°

3

Line width (relative)

2.5

350 μm

2

Φ = 60°

1.5

180 μm

1

Φ = 80°

0.5 0

100 μm 0

20

40

60

80

100

Contact angle (deg)

FIGURE 5.4.1 Calculated printed line width as a function of wetting contact angle, assuming a small cylindrical liquid pattern. Square data points and the optical micrographs show measurements of printed Ag nanoparticles ink on surfaces that show different water contact angle. (From Street, R.A. et al., Mater. Today 9, 32–37, 2006. With permission.)

Digital lithography

Printed wax features

300 μm (b) Etched metal features

300 μm (a)

(c)

FIGURE 5.4.2 (a) Steps of patterning a metal feature using inkjet printing and digital lithography. (b) Optical micrographs of printed wax features on a metal film. Printed wax features are used as an etch mask for conventionally deposited materials. (c) Optical micrographs of metal features patterned by inkjet printing.

8080_C005.fm Page 422 Friday, April 6, 2007 10:25 AM

422

Organic Field-Effect Transistors

Because these heads exhibit superior directionality, features produced by gaps may be fabricated down to 5 μm. In Figure 5.4.2(b), an example of a printed wax etch mask is displayed. The line width is 40 μm and the smallest spacing between two printed lines is 5 μm. The printed wax features work as an etch mask protecting the material that needs to remain on the substrate. The exposed material is etched, and the mask is stripped. The etched features are displayed in Figure 5.4.2(c). This process is called digital lithography because it uses subtractive methods for patterning, as in conventional semiconductor processing. But the mask used here is a digital mask (a file that is printed) instead of hard masks used for photolithography. Paul et al. combined wax printing with direct writing of semiconductors to fabricate polymer-based TFTs [4]. Device performance and reproducibility are discussed in Section 5.4.3 [4,8]. Another hybrid technique used to print high-resolution features is to inkjet print the active material into a predefined well. This well is usually obtained by photolithography to create a prepatterned feature that exhibits higher resolution than what is obtained by direct writing. In this technique, the liquid spread and the drying pattern are controlled by the well. The liquid flows over the surface until it reaches the well wall, which prevents further spread. Since the resulting pattern does not depend on exactly where the liquid is inkjetted, the precision requirement for the printing system is reduced. Both polymer light-emitting diodes (PLEDs) and color filters are made using this technique and are expected to be the first applications of inkjet printing in display applications [9,10]. Sirringhaus and Kawase used this approach to confine the spreading of waterbased conducting polymer ink, PEDOT, to define the critical dimensions of a TFT (Figure 5.4.3a) [1,2]. A line of PEDOT droplets is deposited into hydrophilic wells at a distance d from a hydrophobic polyimide line that is sufficiently small for the spreading droplets to reach the repelling line. After deposition of the source-drain material, TFT devices are fabricated in a top-gate configuration (Figure 5.4.3c) by spin-coating a continuous film (150–300 Å) of the active semiconducting polymer, poly(9,9-dioctylfluoreneco-bithiophene) (F8T2), from a xylene solution. A 400- to 500-nm thick film of the gate dielectric polymer, polyvinylphenol (PVP), is spincoated from isopropanol solution. Finally, a PEDOT/PSS gate electrode line overlapping the channel is inkjet printed in air. In these devices, the semiconductor, F8T2, is not patterned. After deposition by spin-coating, the device was annealed to 265°C to bring F8T2 to its liquid crystal phase and achieve higher TFT mobility [11]. The TFT mobility extracted from the transfer characteristics was 0.01 cm2/Vs. The threshold voltage was equal to –10 V and the Ion/Ioff ratio was equal to 105 for gates voltages from –40 to 10 V and W/L of 600.

5.4.3 ADDITIVE METHODS — PRINTING SEMICONDUCTORS AND ENCAPSULATION LAYERS One of the great advantages of polymeric semiconductors is that they can be deposited from solution, therefore allowing device patterning by direct marking

8080_C005.fm Page 423 Friday, April 6, 2007 10:25 AM

Inkjet Printed Organic Thin Film Transistors

n

(a)

m

423

(c)

–

SO3 HSO3 O O O O S S nS S S O O O O H

S

S

n

O O

PVP G

PI

d

S

D

(d)

(b)

S 200 nm

50 μm

G D

μm PEDOT 10

PEDOT

PI 20

30

40

20 10 μm

Channel

FIGURE 5.4.3 (a) Schematic diagram of high-resolution inkjet printing onto a prepatterned substrate. (b) AFM showing accurate alignment of inkjet-printed PEDOT/PSS source and drain electrodes separated by a repelling polyimide (PI) line with L = 5 μm. (c) Schematic diagram of the top-gate inkjet printed TFT configuration with an F8T2 semiconducting layer (S, source; D, drain; and G, gate). (d) Optical micrograph of an inkjet printed TFT (L = 5 μm). The arrow indicates pronounced roughness of the unconfined PEDOT boundary. (From Sirringhaus, H. et al., Science 290, 2123–2126, 2000. With permission.)

techniques. The ability to pattern discrete areas of semiconductor improves performance of devices by reducing the off current and in an array by reducing deviceto-device leakage [8]. Paul et al. demonstrated printed organic TFTs that were fabricated on a layer of thermal oxide with a common silicon gate electrode [4]. The metal source and drain contacts were patterned by digital lithography using printed wax as an etch resist, as described in Section 5.4.2. The inkjet printing technology used for the polymeric organic semiconductors, acoustic inkjet printing, is described elsewhere [12]. The OTFTs demonstrated by Paul and coworkers used coplanar device geometry. First, gold contact metal (100 nm on a 2-nm adhesion layer of chromium) was deposited by evaporation onto a silicon wafer having a 300-nm layer of thermal oxide. The gold film was then patterned by digital lithography to define the source and drain contacts. Channel lengths varied between 40 and 400 μm, resulting in a range of W/L between 0.3 and 27. The thermal oxide gate dielectric was chemically modified to be hydrophobic using either octadecyltrichlorosilane (OTS) or octyltrichlrosilane (OTS-8) by solution-based deposition [4]. The semiconductor was then deposited by acoustic inkjet printing onto the patterned and treated substrate. Two polymeric organic semiconductors were studied: (1) 0.45% solution of

8080_C005.fm Page 424 Friday, April 6, 2007 10:25 AM

424

Organic Field-Effect Transistors

poly(9,9′-dioctyl fluorene-co-bithiophene), F8T2 [11,13] in xylene; and (2) a solution of a regioregular poly(thiophene) (PQT-12) synthesized at Xerox Research Centre of Canada [14]. The morphology of the printed semiconductors is shown in Figure 5.4.4. The ejected drops coalesce within a printed line; however, if the solvent evaporates before the next line is printed, the dried lines result in the striated topography seen in Figure 5.4.4. Figure 5.4.4(a) is an optical micrograph of printed PQT-12 and Figure 5.4.4(b) shows an image of printed F8T2 acquired by tapping mode atomic force microscopy. The topography of the surface shows the increased thickness of the polymer as each line is applied with a drop size of 35 μm and an overlap ratio of 50%. (a)

Print direction

400 μm

(b)

Pri

nt d i

Height (μm)

2 Au contact

rec tion

1

Printed polymer

0 20 40 Substrate

60 μm

Au contact

FIGURE 5.4.4 (a) Optical micrograph of an array of OTFTs with printed polymer PQT-12 on gold source/drain contacts defined using wax printing and etching. (b) Topography of a 0.45% solution of F8T2 printed on gold contacts acquired by atomic force microscopy in tapping mode. A drop size of 35 μm and overlap ratio of 50% were defined in the printing program. The line-to-line overlap of the printed polymer is clearly seen in the corrugation of the surface. (From Paul, K.E. et al., Appl. Phys. Lett. 83, 2070–2072, 2003. With permission.)

8080_C005.fm Page 425 Friday, April 6, 2007 10:25 AM

Inkjet Printed Organic Thin Film Transistors

425

The TFT data shown by Paul and colleagues for printed semiconductors conform well to conventional transistor models in the linear and saturation regimes, with TFT mobility equal to 0.1 cm2/Vs, Ion/Ioff ratio > 106, threshold voltage of –3 V, and subthreshold slope of 1.7 V/decade. Extracted linear mobility for the printed F8T2 on OTS treated gate dielectrics had an average value of 4 × 10–3 cm2/Vs and a subthreshold slope of ~1 V/decade for coplanar device geometry. All the data are within the range of values that were observed in the equivalent spin-cast devices, indicating that the use of additive inkjet printing for depositing polymeric organic semiconductors does not compromise device performance [1,13,15]. It has been shown that in order to achieve high device performance, the surface of the gate dielectric layer needs to be treated with monolayers that create a hydrophobic coating [15–17]. However, it is difficult to coat polymer semiconductors uniformly on hydrophobic gate dielectrics by spin-coating. Therefore, direct marking also simplifies processing on hydrophobic substrates, allowing the deposition and patterning of polymeric semiconductors onto optimized gate dielectric surfaces and achieving high-performance OTFTs in one processing step. The process used by Paul et al. can also be used to fabricate TFTs with patterned gate electrodes instead of using a doped silicon wafer as a common gate [8]. The processing steps to fabricate a bottom-gate TFT structure using inkjet printing as the only patterning technique are described in Figure 5.4.5. First, a 100-nm chromium film was deposited onto glass by thermal evaporation and patterned by digital lithography using wax ejected from a multi-ejector piezoelectric printhead [14]. The dielectric material was 30 nm of silicon oxide on a 200-nm silicon nitride film, Wax

(1)

Deﬁne S/D

(4) Cr

Substrate Deﬁne Cr Gate Deposit Semiconductor H25C12

(2) S

S S

(3) Wax

S

n

C12H25

Cr/Au (5)

Si3N4/SiO2

FIGURE 5.4.5 Process flow for the fabrication of polymer TFTs by inkjet printing. Steps 1 to 4 use a subtractive process for patterning of materials. The subtractive process prints an etch mask on a previously deposited film. Step 5 illustrates the additive inkjet printing process. The additive process simultaneously deposits and patterns the semiconductor material PQT12. (From Arias, A.C. et al., Appl. Phys. Lett. 85, 3304–3306, 2004. With permission.)

8080_C005.fm Page 426 Friday, April 6, 2007 10:25 AM

426

Organic Field-Effect Transistors

deposited using plasma-enhanced chemical vapor deposition (PECVD) at 350°C. The source-drain metal layer, 100-nm gold film with a chromium adhesion layer, was deposited by thermal evaporation and subsequently defined by digital lithography. The surface of the dielectric layer was modified by deposition of a selfassembled monolayer of octyltrichlorosilane (OTS-8) before the semiconductor material was printed on the channel region of the TFTs. An additive inkjet printing process, using a separate print head, was used to simultaneously deposit and pattern the solution-based polymeric semiconductor, completing the bottom-gate TFT device fabrication. Two different print head technologies were used for depositing the semiconductor layer. The first, used for printing small areas, is a nozzleless acoustic inkjet print head that produces features of 30–40 μm of PQT-12 on an OTS-treated surface [12]. The second is a commercially available MicroFab Technologies piezoelectric print head. Feature sizes of 80–100 μm were obtained by using a 60-μm nozzle. The size of printed features obtained with this head was comparable to the channel area of the TFTs and one drop of PQT-12 per transistor was sufficient to give good TFT performance. Figure 5.4.6(a) shows typical transfer and output current-voltage characteristics for a TFT taken from an array of TFTs. The devices exhibit a field-effect mobility between 0.05 and 0.10 cm2/V·s and an off current of ~10–12 A, giving an on/off ratio of ~106 at VSD = –40 V. The onset voltage is close to 0 V and the subthreshold slope is 75 nF.V/decade.cm2. The output characteristics show good saturation and no sign of significant contact resistance. In addition, TFTs made with PQT-12 have shown minimal gate bias stress effect [18]. All the parameters obtained for the inkjet printed TFTs are similar to those obtained for spin-coated materials on thermal oxide-coated silicon substrates [14,15]. The probe test across the diagonal of the arrays shows a high yield of working TFTs and uniformity in both the on- and off-currents. A typical series of measurements, shown in Figure 5.4.6(b), gave an average mobility of 0.06 cm2/V·s, with a standard deviation of 0.02 cm2/V·s. The deviation showed here, using inkjet printing as the only patterning technique, is very promising and comparable to the deviation obtained by the photolithographic process used by Gelinck et al. [19]. As with any solution-based technique, materials compatibility is a challenge requiring device design to prevent the solvent used in successive layers from dissolving previous layers. Using digital lithography to pattern wax resist for etching metal contacts circumvents the issue of dissolution of previous layers, but further materials engineering or device design may be required when inkjet printing organic semiconductor inks on polymeric substrates or dielectric layers. For example, TFTs must be passivated, and this is also a process that can be done by inkjet printing. It has been shown that deposition of the semiconductor and the passivation can be performed in a single-step process by printing a polymer blend [20,21]. Instead of conventionally depositing the semiconducting polymer solution directly over the channel region of the TFT, the semiconductor is blended with an insulator polymer and the blend is deposited from solution over the TFT structure. Viscosity, solubility, and surface energy are manipulated to ensure phase segregation in a direction vertical to the substrate. The final thin film consists of a self-organized double layer, formed in one step deposition, where the semiconductor segregates to the substrate forming the TFT channel and is encapsulated by the insulating polymer.

8080_C005.fm Page 427 Friday, April 6, 2007 10:25 AM

Inkjet Printed Organic Thin Film Transistors

427

(a)

(b)

VDS = –40 V

μ~0.063cm2/Vs sd 0.021 cm2/Vs

0.12

1E-6 0.10

ID(A)

1E-8 1E-9

ID(A) –6 1.0 × 10

μlinear(cm2/Vs)

1E-7

VG = –40 V

–7 8.0 × 10

1E-10 1E-11 1E-12

–7 6.0 × 10 –7 4.0 × 10 VG = –20 V

–7 2.0 × 10

1E-13 –40

0.0 0

20

30

40

0.06 0.04 0.02

VG = –10 V 10

0.08

Vd(V)

0.00 –30

–20

–10 VG(V)

0

10

0

5

10

15

20

25

30

35

Transistors

FIGURE 5.4.6 (a) Transfer and output (inset) characteristics of a typical printed TFT array with W/L ~ 2 and C = 25 nF/cm2. (b) TFT mobility obtained for different transistors of a printed array, showing an average value of 0.06 cm2/Vs. (From Arias, A.C. et al., Appl. Phys. Lett. 85, 3304–3306, 2004. With permission.)

Figure 5.4.7 demonstrates successful encapsulation of the TFTs via surfaceinduced phase separation. Devices were left in air and consecutive measurements were taken every eight hours. The device characteristics are very stable for a period of two days operating in air. The transfer characteristics, shown in Figure 5.4.7(a), do not show any change during that period of time, illustrating an improved environmental stability of polymer-based devices. Devices made with the other blend compositions and with P3HT instead of PQT show similar trends [20,21]. A small shift in the onset voltage and subthreshold slope is observed when devices are left in air for longer periods of time. The transfer characteristics of a device left in air for 20 days are shown in Figure 5.4.7(b). There was a 7-V shift of the onset voltage and the subthreshold slope increased by 0.7 V/decade. The shift in the onset voltage of devices fabricated from a blend solution in 20 days is 50% less than the shift observed in one day for nonencapsulated devices, and is similar to films separately encapsulated with PMMA [20,21]. This is one example of possible simplification of the processes that can be expected from the inkjet printing technology.

5.4.4 DISPLAY BACKPLANE FABRICATION A typical active matrix thin film transistor (AM-TFT) backplane contains one transistor per pixel in a structure having four to six separate layers composed of conductors, dielectrics, and semiconductors. The TFT works as a switch with the on and off states controlling when each pixel can be addressed. Each pixel is connected to the other through the addressing lines, gate line, and data line. The gate lines turn the TFT on or off while the data (image to be displayed) are transmitted by the data

8080_C005.fm Page 428 Friday, April 6, 2007 10:25 AM

428

Organic Field-Effect Transistors

(b) (a) Time exposed to air

1E-6

8h 24h 48h

1E-6

1E-8

ID(A)

ID(A)

1E-8 1E-10

1E-10 1E-12

1E-12

1E-14 –40

1E-14

–20

0 VG(V)

20

–40

as prepared 1.16 V/dec Von~ 0V 20 days in air 1.72V/dec Von~ 7V μ = 0.06 ± 0.02 cm2/V.s –20

0

20

VG(V)

FIGURE 5.4.7 Improved environmental stability of TFTs fabricated with self-encapsulated polymer films. (a) Transfer characteristics of a TFT exposed to air for two days. (b) Transfer characteristics of TFT as prepared and after been exposed to air for 20 days. (From Arias, A.C. et al., Adv. Mater., 18, 2900–2904, 2006. With permission.)

lines. Given the previously demonstrated ability to print polymer TFT arrays, it is important to verify whether inkjet printed features can lead to pixel designs that meet the application needs. Figure 5.4.8 shows a printed 128 × 128 pixel array, along with optical micrographs of the pixels and the equivalent circuit for the pixel. The metal lines were defined by digital lithography and the semiconductor was directly deposited by inkjet printing, as described in Section 5.4.3. The pixel is 340 μm, corresponding to display backplanes with resolution of 75 dots per inch (dpi) [8]. The TFT contacts are formed by the data line and the top plate of the storage capacitor. The metal line width of about 50 μm was determined by the size of the wax drop from the inkjet printer. However, it is shown in Figure 5.4.8(c) that the space between the data line and the storage capacitor plate is smaller than 10 μm. The ability to control high-resolution gaps is important in order to obtain short TFT channel lengths and low parasitic capacitances. The channel length of 30–50 μm was chosen such that the contact resistance that may be observed for polymer-based TFT with short channel lengths is negligible [17,22]. The printed polymer is precisely confined to the region of the gate electrode between the TFT contacts, as seen in Figure 5.4.8 [8]. The extension of the polymer beyond the gate electrode is detrimental to TFT performance because it causes higher leakage current from the ungated semiconductor. The well defined area of inkjet printed PQT-12 also prevents the formation of a continuous layer of polymer between TFTs that would otherwise form a conductive path and result in significant crosstalk between pixels.

8080_C005.fm Page 429 Friday, April 6, 2007 10:25 AM

Inkjet Printed Organic Thin Film Transistors

(a)

429

(b)

300 μm (c)

(d)

Gate line Data line

CPL Gate line

CP

CST

CST

40 μm

Data line

FIGURE 5.4.8 Optical images of a printed polymer TFT array at increasing magnification, showing the whole 128 × 128 array (a), small regions of the array (b), and a single device (c). Note the printed semiconductor confined to the channel region and figures (b) and (c). (d) Equivalent circuit for the pixel, showing the gate and data address lines, the TFT, and various capacitances. In the display pixel, the storage capacitor, CST, is connected to the next gate line. CP is the parasitic capacitance between the gate line and the pixel and CPL contains all the sources of gate line capacitance, of which the main additional capacitance is to the data line and the TFT channel. (From Arias, A.C. et al., Appl. Phys. Lett. 85, 3304–3306, 2004. With permission.)

The electrical response of a backplane is governed by its RC time constants since the electrical signals sent to the pixels are time dependent [23]. A storage capacitor (CST) is added to the pixel, shown in Figure 5.4.8(c), in order to retain the charge while data are written to the other pixels. The RC time constant for charging a pixel is given by [8]: tRC = RTFT CST = const. CST/[μ (W/L) VGS]

(5.4.1)

The overlap between gate and data levels of a backplane creates a capacitive coupling between the gate line and the pixel. Consequently, the data voltage at the pixel is reduced by an amount equal to the feedthough voltage VFT (Equation 5.4.2), where CP is the parasitic capacitance coupling the gate line to the pixel and VG is the gate voltage relative to the pixel voltage (see Figure 5.4.8c and d): VFT = VGS CP/CST

(5.4.2)

Fast addressing is obtained when tRC and VFT are minimized. These are conflicting requirements since a large value of CST reduces VFT but increases tRC, while the opposite is true for large VGS (see Equations 5.4.1 and 5.4.2). The inverse product of these parameters is a convenient combined figure of merit, F, given by

8080_C005.fm Page 430 Friday, April 6, 2007 10:25 AM

430

Organic Field-Effect Transistors

F = 1/[tRC VFT] = const. μ [W/L]/CP

(5.4.3)

where larger F corresponds to more optimal performance [8]. An optimal pixel design minimizes the parasitic capacitance CP while maximizing the mobility μ and W/L. The parasitic capacitance is largely due to the overlap between the gate and source-drain metal of the TFT and can be minimized by printing small features or by improving deposition accuracy to ensure careful positioning of the two metals. Low parasitic capacitance and large W/L are easily achieved by conventional photolithography allowing the use of semiconducting polymers with TFT mobilities of 0.02 cm2/Vs to drive electrophoretic displays [19]. Arias et al. combined positioning precision of inkjet printing techniques with the high intrinsic mobility of PQT-12 compared to other polymers to optimize the pixel design of an all-printed patterned backplane. With this printing technology, it was possible to control the placement of metal lines minimizing the parasitic capacitance and obtaining VFT as low as 1.2 V, for operating gate voltage swing of 20 V. Figure 5.4.9 shows a typical pixel response for the printed array. The pixel is charged to the data-line voltage (10 V) during the gate-ON pulse (Vgate-ON = –5 V). After the gate is switched off (Vgate-OFF = +15 V), the charge level drops slowly over several seconds, which indicates good transistor and pixel performance. The gate feedthrough voltage for the chosen voltage levels is around 1–2 V in this pixel design [24]. 16 14

Gate signal Gate OFF

12 10

Pixel signal Gate feedthrough

Voltage (V)

8 6 4 1 Sec 2 Data signal

0 –2 –4 –6

Gate ON Time (500 msec increments)

FIGURE 5.4.9 The pixel signal from a printed backplane shows good switching time, pixel charge storage, and low gate feedthrough. (From Daniel, J.H. et al., SID Symp. Dig. Tech. Papers 36, 1630–1633, 2005. With permission.)

8080_C005.fm Page 431 Friday, April 6, 2007 10:25 AM

Inkjet Printed Organic Thin Film Transistors

431

Another important parameter to consider when choosing materials to fabricate TFT backplanes is the resistance of the address lines. The time constant to address a line is tADR = 0.4 N2 RPL CPL

(5.4.4)

where RPL and CPL are the total resistance and capacitance, respectively, of the metal line for one pixel and N is the number of address lines [23]. A small time constant is easy to achieve in a small format array, but the N2 dependence on the number of lines quickly increases the time constant. Typical values are CPL ~ 1–5 pF and RPL ~ 1–10 Ω for a conventional metal or RPL ~ 10–100 kΩ for conducting polymers. With conventional metals, the time constant remains small enough, around 20 μs, even for arrays where N = 1,000, while conducting polymers could only address arrays where N ~ 30 [8]. Therefore, the development of solution-processed low-resistance conductors is fundamental for an all-additive printing process.

REFERENCES 1. Sirringhaus, H. et al., High-resolution inkjet printing of all-polymer transistor circuits, Science 290, 2123–2126, 2000. 2. Kawase, T. et al., Inkjet printing of polymer thin film transistors, Thin Solid Films 438–439, 279–287, 2003. 3. Wong, W.S. et al., Amorphous silicon thin-film transistors and arrays fabricated by jet printing, App. Phys. Lett. 80, 610–612, 2002. 4. Paul, K.E. et al., Additive jet printing of polymer thin-film transistors, Appl. Phys. Lett. 83, 2070–2072, 2003. 5. De Gennes, P.G., Wetting: Statics and dynamics, Rev. Mod. Phys. 57, 827–863, 1985. 6. Street, R.A. et al., Jet printing flexible displays, Mater. Today 9, 32–37, 2006. 7. Wong, W.S. et al., Hydrogenated amorphous silicon thin-film transistor arrays fabricated by digital lithography, IEEE Electron Device Lett. 24, 577–579, 2003. 8. Arias, A.C. et al., All jet-printed polymer thin-film transistor active-matrix backplanes, Appl. Phys. Lett. 85, 3304–3306, 2004. 9. Shimoda, T. et al., Inkjet printing of light-emitting polymer displays, MRS Bull. 28, 821827, 2003. 10. Gans, B.-J.D., Duineveld, P.C., and Schubert, U.S., Inkjet printing of polymers: State of the art and future developments, Adv. Mater. 16, 203–213, 2004. 11. Sirringhaus, H. et al., Mobility enhancement in conjugated polymer field-effect transistors through chain alignment in a liquid-crystalline phase, App. Phys. Lett. 77, 406–408, 2000. 12. Elrod, S.A. et al., Nozzleless droplet formation with focused acoustic beams, J. Appl. Phys. 65 (9), 3441–3447, 1989. 13. Brennan, D.J. et al., Polyfluorenes as organic semiconductors for polymeric field effect transistors, Proc. SPIE Int. Soc. Optical Eng. 5217 (Organic Field Effect Transistors II), 1–6, 2003. 14. Ong, B.S. et al., High-performance semiconducting polythiophenes for organic thinfilm transistors, J. Am. Chem. Soc. 126, 3378–3379, 2004.

8080_C005.fm Page 432 Friday, April 6, 2007 10:25 AM

432

Organic Field-Effect Transistors

15. Salleo, A. et al., Thin-film transistors with chemically modified dielectric interfaces, Appl. Phys. Lett. 81, 4383–4385, 2002. 16. Sirringhaus, H. et al., Two-dimensional charge transport in self-organized, highmobility conjugated polymers, Nature 401, 685–688, 1999. 17. Bürgi, L. et al., Close look at charge carrier injection in polymer field-effect transistors, J. Appl. Phys. 94, 6129–6137, 2003. 18. Street, R.A., Salleo, A., and Chabinyc, M.L., Bipolaron mechanism for bias-stress effects in polymer transistors, Physical Rev. B 68, 085316-1–085316-17, 2003. 19. Gelinck, G.H. et al., Flexible active-matrix displays and shift registers based on solution-processed organic transistors, Nat. Mater. 3, 106–110, 2004. 20. Arias, A.C., Vertically segregated polymer blends: Their use in organic electronics, J. Macromolec. Sci. Part C: Polymer Rev. 46, 103–125, 2006. 21. Arias, A.C., Endicott, F., and Street, R.A., Surface induced self-encapsulation of polymer thin film transistors, Adv. Mater., 18, 2900–2904, 2006. 22. Street, R.A. and Salleo, A., Contact effects in polymer transistors, Appl. Phys. Lett. 81, 2887–2889, 2002. 23. Tsukada, T., Active-matrix liquid-crystal displays, in Technology and applications of amorphous silicon, ed. R.A. Street, Springer Verlag, Heidelberg, 2000. 24. Daniel, J.H. et al., Flexible electrophoretic displays with jet-printed active-matrix backplanes, SID Symp. Dig. Tech. Papers 36, 1630–1633, 2005.

8080_C005.fm Page 433 Friday, April 6, 2007 10:25 AM

5.5

Soft Lithography for Fabricating Organic Thin-Film Transistors

Kimberly C. Dickey, Kwang Seok Lee, and Yueh-Lin Loo CONTENTS 5.5.1 Typical Device Structures and Conventional Fabrication Techniques .....433 5.5.2 Stamps for Soft Lithography .....................................................................435 5.5.3 Microcontact Printing (μCP) .....................................................................438 5.5.3.1 Selective Etching .........................................................................439 5.5.3.2 Selective Electroless Plating .......................................................441 5.5.3.3 Selective Chemical or Electrochemical Polymerization.............443 5.5.3.4 Stamp-and-Spin-Cast...................................................................444 5.5.3.5 Other Microcontact Printing Derivatives ....................................445 5.5.4 Nanotransfer Printing (nTP) ......................................................................447 5.5.5 Soft-Contact Lamination (ScL) .................................................................458 5.5.6 Cold Welding .............................................................................................466 5.5.7 Metal Transfer Printing..............................................................................468 5.5.8 Hot Lift-Off................................................................................................469 5.5.9 Micromolding in Capillaries (MIMIC) .....................................................469 5.5.10 Soft-Contact Optical Lithography .............................................................473 5.5.11 Laser Thermal Transfer Printing ...............................................................475 5.5.12 Imprint Lithography...................................................................................475 References..............................................................................................................483

5.5.1 TYPICAL DEVICE STRUCTURES AND CONVENTIONAL FABRICATION TECHNIQUES Organic thin-film transistors (OTFTs) can be classified as bottom-contact or topcontact devices, depending on where the electrodes make contact to the organic semiconductor layer, as shown in Figure 5.5.1 [1,2]. Bottom-contact devices (Figure 5.5.1a) — where the electrodes are prepatterned prior to the deposition of the organic semiconductor layer — are generally known to have poorer device characteristics compared to those of top-contact devices (Figure 5.5.1b), where the electrodes are 433

8080_C005.fm Page 434 Friday, April 6, 2007 10:25 AM

434

Organic Field-Effect Transistors

Organic semiconductor Source Source

Drain Dielectric

Drain Organic semiconductor Dielectric

Gate

Gate

Substrate

Substrate

Bottom-contact device

Top-contact device

(a)

(b)

FIGURE 5.5.1 Side view of (a) bottom-contact and (b) top-contact OTFTs.

directly deposited on top of the organic semiconductor [1,2]. Such disparity in device performance is speculated to result from uniformity differences in the organic semiconductor thin film between the two types of device geometries [1,3]. In the case of bottom-contact devices, the organic semiconductor is deposited on two different surfaces: on the dielectric surface and on top of the metal electrodes. Due to surface energy differences, the growth behavior of the organic semiconductor can change dramatically across the dielectric–electrode interface. The discontinuity in grains across the interface can lead to structural disorder, which can hamper charge transfer and in turn limit device performance [1,3]. In contrast, a uniform organic semiconductor film is first deposited on the gate dielectric in top-contact devices. Electrodes are then defined directly on top of the uniform organic semiconductor layer. As a result, the grains are continuous across the charge transport region. The top-contact device structure is therefore desirable to maximize device performance [4–6]. Top-contact devices are frequently fabricated by sequential deposition of functional materials. Specifically, the organic semiconductor and metal electrodes are independently and sequentially deposited through shadow masks to define the active channel regions. While this technique allows highperformance OTFTs to be fabricated, these devices are typically large, with feature sizes limited by the resolution of the shadow masks (25–30 μm) [7]. Smaller features are easily patterned using photolithography, a well-developed technique for fabricating micro- or submicron inorganic semiconductor devices. According to the 2005 International Technology Roadmap for Semiconductors (ITRS 2005), features as small as 70 nm can be patterned using photolithography [8]. Combining photolithography with lift-off provides an effective method for fabricating the electrodes for OTFTs. Photolithography, however, is rarely used to define the organic semiconductor layer because the chemicals used during the subsequent processing steps (etchants, developers, etc.) can cause degradation [9] or delamination [10] of the organic semiconductor. As a consequence, photolithography is typically limited to defining electrodes in bottom-contact devices in which the organic semiconductor is deposited postpatterning. Halik and coworkers subsequently extended this process to pattern conducting polymer electrodes, such as poly(3,4-ethylenedioxythiophene) doped with polystyrene sulfonic acid (PEDOT-PSS) [11]. In this process, PEDOT-PSS is uniformly coated on a substrate that has prepatterned photoresist. The PEDOT-PSS-coated substrate is then immersed in acetone to swell the photoresist underneath the

8080_C005.fm Page 435 Friday, April 6, 2007 10:25 AM

Soft Lithography for Fabricating Organic Thin-Film Transistors

435

PEDOT-PSS overlayer. The swollen photoresist lifts off along with the PEDOT-PSS overlayer, leaving behind patterns of PEDOT-PSS in the regions that did not originally contain any photoresist. PEDOT-PSS source, drain, and gate electrodes as small as 2 μm were fabricated in this fashion. Typical of polymer conductors, these PEDOT-PSS electrodes exhibit an electrical conductivity of around 0.1 S/cm [11]. A uniform, continuous pentacene (p-type organic semiconductor) layer is then deposited on the PEDOT-PSS source and drain electrodes to complete the bottom-contact OTFTs. Since pentacene is deposited in the final step, it is not exposed to any of the solvents (e.g., acetone) required to prepattern the PEDOT-PSS electrodes. These bottom-contact pentacene TFTs with PEDOT-PSS electrodes exhibit charge-carrier mobilities comparable to charge-carrier mobilities of typical bottom-contact pentacene TFTs with gold electrodes (0.2 cm2/Vs) [1] and on/off current ratios as high as 106. Further, these device characteristics’ are on par with requirements for driving most display applications [12,13]. As mentioned previously, patterning the active semiconductor layer photolithographically is complex since many organic semiconductors tend to degrade during this process. Despite such technical difficulties, Sheraw and coworkers [14] recently demonstrated a derivative photolithographic patterning proven to define the active pentacene layer. In this process, water-soluble, photopatternable polyvinyl alcohol (PVA) is used as the etch resist for patterning pentacene. Device fabrication begins with depositing and patterning nickel gate electrodes, SiO2 gate dielectric, and palladium source and drain electrodes on a polyethylene naphthalate (PEN) substrate using ion sputtering, and photolithography followed by lift-off. Pentacene is subsequently deposited by thermal evaporation. PVA is then spin-coated directly on the pentacene layer and is photolithographically patterned to create features that define the channel regions. Subsequent oxygen plasma etching removes pentacene outside the channel (regions that are no longer protected by PVA). Bottom-contact pentacene TFTs with a channel length of 10 μm and a channel width of 200 μm exhibit charge-carrier mobilities as high as 1.2 cm2/Vs, and on/off current ratios of 108 [14]. While this process successfully patterns pentacene active regions, it has not been successfully extended for patterning other organic semiconductors, such as poly-3-hexylthiophene (P3HT), due to mechanical cracking of the PVA etch barrier during development. While photolithography has successfully been demonstrated for OTFT fabrication, it is potentially an expensive process [15]. This process is thus not well-suited for fabricating low-cost organic electronic devices. Inexpensive, noninvasive processing technologies that are physically and chemically compatible with electricallyactive organic materials need to be developed to lower fabrication costs and maximize device performance. Among a number of promising techniques that have been demonstrated to date, soft lithography and its derivative techniques will be reviewed in this chapter.

5.5.2 STAMPS FOR SOFT LITHOGRAPHY The key component of soft lithography is the stamp that is used to transfer patterns. To achieve high-quality pattern transfer, conformal contact must be established

8080_C005.fm Page 436 Friday, April 6, 2007 10:25 AM

436

Organic Field-Effect Transistors

between the stamp and the substrate, preferably without external pressure. Such conformal contact is often achieved with the use of an elastomeric stamp, such as one made of crosslinked poly(dimethylsiloxane) (PDMS). Dow Corning offers a commercially available PDMS formulation known as Sylgard 184 [16]. Sylgard 184 is a two-component system that consists of a PDMS prepolymer and a cross-linker. To create PDMS stamps, the prepolymer and cross-linker are mixed in a 10:1 (w:w) ratio and degassed. The degassed mixture is cast and cured against a prepatterned master as shown in Figure 5.5.2. The raised and recessed features of the master are typically defined by photolithography. After curing the prepolymer mixture overnight at room temperature, or at 60°C for 2 h, the PDMS stamp is peeled from the master. To ensure that the PDMS stamp releases effectively, the master is typically pretreated with tridecafluoro(1,1,2,2,tetra-hydrooctyl)-1-trichlorosilane (FSAM) [17]. The final stamp has the negative image of the master.

Fabricate and silanize master

SiO2, Si3, N4, metals, photoresists, or wax Si

Pour PDMS prepolymer over master

PDMS

Si

Cure, peel oﬀ PDMS

PDMS h l

d

FIGURE 5.5.2 Schematic of the PDMS stamp fabrication procedure. A master is fabricated by conventional photolithography and lift-off. The master is then treated with FSAM for ~30 min to ensure that the cured PDMS releases easily from the master. The mixture of PDMS prepolymer and cross-linker (Sylgard 184, Dow Corning) is poured over the master and cured. After peeling from the master, the PDMS stamp has the negative image of the master. (From Y. Xia and G. M. Whitesides, Angew. Chem., Int. Ed., 37, 550, 1998.)

8080_C005.fm Page 437 Friday, April 6, 2007 10:25 AM

Soft Lithography for Fabricating Organic Thin-Film Transistors

437

Although photolithography is necessary initially to create the master from which the stamp is cast, once the master is prepared, many stamps can be fabricated from a single master. A stamp with the positive image can also be fabricated by using the initially generated PDMS stamp (with the negative image) as the master. To ensure that the second-generation PDMS stamp releases effectively, the PDMS master can also be treated with FSAM. While the mechanical properties of PDMS are attractive for establishing conformal contact between the stamp and the substrate, the low elastic modulus and high compressibility of the Sylgard 184 PDMS formulation limit the minimum feature size achievable to several microns [18,19]. To realize nanoscale features, a high-resolution hard stamp is required. This hard stamp can be fabricated from rigid materials, such as fused silica, silicon, or GaAs [20,21] or from composite PDMS (h-PDMS) materials [22,23]. When a rigid stamp is used, the nanometer-sized features on that stamp are defined lithographically with techniques, such as electronbeam lithography or holographic lithography [24]. The surface of the rigid stamp is made nonstick with FSAM treatment [21]. To transfer patterns at ambient pressures with a rigid stamp, the substrate being patterned must be a conformal material, such as PDMS, or a plastic substrate coated with PDMS. To eliminate pressure concerns, elastomeric, composite PDMS stamps can be fabricated. These composite stamps were initially developed by Schmid and coworkers at IBM-Zurich [23]. They formulated a siloxane polymer composite, referred to as h-PDMS (“hard” PDMS), that employs a trilayer stack (shown in Figure 5.5.3) to create conformal yet high-resolution elastomeric stamps. The composite stamp consists of a flexible glass or foil backplane to prevent long-range pattern distortion, a soft cushion of Sylgard 184 PDMS to establish conformal contact with the substrate, and a hard silane polymer optimized by Schmid and coworkers to enable nanometer feature replication (as low as 80 nm) at aspect ratios (Figure 5.5.2) ranging from 5 to 0.02. The flexible backplane, as compared to a rigid backplane, allows easier release of the stamp from the master during preparation and the stamp from the substrate after patterning. Further, as shown in the bottom half of Figure 5.5.3, the flexible backplane permits conformal contact with substrates with moderate surface topography. Odom and coworkers [22] subsequently modified the design of the h-PDMS stamp to enable 50-nm feature replication in a two-layer stamp. Specifically, the backplane is completely eliminated in favor of a thicker (~3 mm) Sylgard 184 PDMS layer that supports a thin layer of h-PDMS (30–40 μm). Eliminating the glass backplane makes the composite stamp easier to handle and to release from the master. In addition to the mechanical limitations discussed, PDMS is highly susceptible to solvent swelling, which can affect feature resolution and fidelity during patterning. Recently, DeSimone and coworkers [25–27] developed photocurable perfluoropolyethers (PFPEs) that can be used as new stamp materials for soft lithography. The mechanical properties of PFPEs are similar to PDMS (i.e., conformal contact can be established at ambient conditions). PFPEs, however, are highly resistant to solvent swelling. Further, the low surface energy of PFPEs allows replication of nanoscale features without resorting to composite or hard stamps [28]. PFPE stamps are

8080_C005.fm Page 438 Friday, April 6, 2007 10:25 AM

438

Organic Field-Effect Transistors

(a)

Hard backplane Soft cushion

Stiﬀ pattern

(b)

FIGURE 5.5.3 (a) Schematic of a trilayer h-PDMS stamp, which consists of a glass backplane, a PDMS cushion layer, and a patterned h-PDMS layer. (b) Trilayer h-PDMS stamp in contact with an uneven substrate. (From T. W. Odom et al., Langmuir, 18, 5314, 2002.)

fabricated by the same procedure as PDMS stamps, with the exception that the material cures by UV irradiation (365 nm), rather than by heat treatment.

5.5.3 MICROCONTACT PRINTING (μCP) Microcontact printing (μCP) [17,29,30] is part of a set of nonphotolithographic fabrication techniques known as soft lithography. In μCP, features are patterned with inexpensive, elastomeric PDMS stamps (see Section 5.5.2 for details on stamp fabrication). Consequently, μCP can be less expensive in terms of capital and operation costs compared to photolithography for patterning large-area micron- or submicron-sized, features [31,32]. The PDMS stamp is used to transfer inks onto rigid inorganic substrates or flexible polymer substrates. The transferred ink can serve as a molecular resist for subsequent selective etching, or as a hydrophobic/hydrophilic molecular template that allows the selective deposition of functional materials in subsequent steps. To transfer patterns, the PDMS stamp is inked with a solution typically containing molecules or colloids. The inked stamp is then brought into contact with the substrate onto which the ink is transferred. Through van der Waals forces, the raised regions of the stamp and the substrate form conformal contact. In the regions of contact, the ink binds chemically or physically to the substrate. Similar to photolithography, subsequent processing steps are required to define features. For this reason, μCP is also typically limited to the fabrication of electrodes in bottom-contact OTFTs. The subsequent processing steps can be classified into three categories: selective etching [29,33–35], selective electroless plating [36–39], and selective chemical and electrochemical polymerization [40–43]. Depending on

8080_C005.fm Page 439 Friday, April 6, 2007 10:25 AM

Soft Lithography for Fabricating Organic Thin-Film Transistors

439

PDMS (a)

1−2 μm

Photoresist pattern

Si Exposure of PDMS to thiol solution (b)

PDMS

(c)

PDMS

Gold (50–200 nm) supported on 10 nm Ti adhesion layer (d)

Alkanethiol Stamping onto The thickness of gold substrate the alkanethiolate layer is 12Å Si Etching in CN−/O2

(e)

Si

FIGURE 5.5.4 Defining gold patterns by μCP followed by selective etching. (a) A PDMS stamp is cast from a master. (b) The PDMS stamp is removed from the master and inked with an alkylthiol solution. (c) The inked-PDMS stamp is brought into contact with a goldcoated Si substrate. (d) In the regions of contact, the ink transfers to the gold. (e) The ink on the gold effectively acts as an etch resist. Subsequent etching in an aqueous basic solution removes gold only in the regions that are not protected by the ink. (From A. Kumar and G. M. Whitesides, Appl. Phys. Lett., 63, 2002, 1993.)

the processing details, the final features are created subtractively or additively. Each of these techniques is reviewed in following sections.

5.5.3.1 SELECTIVE ETCHING Kumar and Whitesides [29] first patterned gold features with μCP followed by selective etching. In this example, illustrated in Figure 5.5.4, microcontact-printed alkylthiols act as an etch resist for the subtractive patterning of gold. The procedure begins with inking the PDMS stamp with an alkylthiol solution, and contacting the inked stamp against a gold-coated substrate. In the regions of contact, the thiol end groups form covalent gold–sulfur bonds with the substrate. The covalently bound alkylthiol self-assembled monolayer (SAM) acts as an etch resist, effectively protecting the underlying gold during subsequent etching. Etching the exposed gold regions with an aqueous basic solution results in gold features only in the SAMcovered regions. Gold features, as small as 1 μm, can be produced by μCP followed by selective etching. In addition to patterning gold, this technique has been extended

8080_C005.fm Page 440 Friday, April 6, 2007 10:25 AM

440

Organic Field-Effect Transistors

to patterning silver [33], copper [34], palladium [35], and indium tin oxide (ITO) [44] features. Rogers and coworkers [44] used μCP and selective etching to fabricate gold source and drain electrodes, as well as interconnects, for active matrix backplane circuits to drive flexible electronic paper displays. Since μCP requires subsequent etching to define features, source and drain electrodes were first defined prior to the deposition of chemically and mechanically fragile organic semiconductor. The resulting OTFTs were therefore built in the bottom-contact geometry. The circuits were fabricated on a flexible, prepatterned indium tin oxide (ITO)-coated poly(ethylene terephthalate) (PET) substrate. Specifically, ITO served as the gate electrode and organosilsesquioxane spin-on glass (SOG) served as the gate dielectric. Thin layers of Ti (1.5 nm) and Au (20 nm) were deposited on the dielectric layer and patterned by μCP and selective etching to define the source and drain electrodes and interconnects. The typical channel width and channel length in these μCP-patterned circuits were ~200 and ~20 μm, respectively. By minimizing the mechanical distortion of the PDMS stamp during printing, Rogers and coworkers were able to limit the registration error between the source and drain level and the gate level to as little as 50 μm over a 16 cm × 16 cm printed footprint [44]. Subsequent deposition of the organic semiconductor layer through a shadow mask completed the bottom-contact OTFTs, which are shown in Figure 5.5.5. A variety of organic semiconductors were used as active materials, and their device performance was comparable to those of typical OTFTs fabricated with bottom-contact metal electrodes defined by conventional photolithography. Another attractive feature of μCP is its ability to pattern features on curved surfaces [45,46]. Jackman and coworkers demonstrated the versatility of μCP by patterning gold features on a cylindrical surface by μCP and selective etching [45].

100 μm

FIGURE 5.5.5 Photograph of a plastic active matrix backplane circuit. Gold source and drain electrodes and interconnects were fabricated by μCP and selective etching. An optical micrograph of a typical transistor is shown in the inset. (From J. A. Rogers et al., Proc. Natl. Acad. Sci. U.S.A., 98, 4835, 2001.)

8080_C005.fm Page 441 Friday, April 6, 2007 10:25 AM

Soft Lithography for Fabricating Organic Thin-Film Transistors

441

In their process, a glass or silicon dioxide cylinder is coated with titanium and gold. The cylinder is then rolled across an alkylthiol-inked PDMS stamp. Since the PDMS stamp makes conformal contact with the curved surface of the cylinder, the alkylthiols are transferred to the metal-coated cylinder with minimal pattern distortion. The final features are generated after an etching step. Alternatively, a cylindrical PDMS stamp can be used to transfer patterns onto a flat gold/titanium-coated substrate [47]. By mounting the PDMS stamp on a glass cylinder, high-speed, reel-to-reel fabrication methods are feasible. Rogers and coworkers demonstrated such a reel-to-reel process for fabricating gold electrodes for bottom-contact regioregular poly(3-hexylthiophene) (P3HT) TFTs using a cylindrical PDMS stamp [48]. These bottom-contact P3HT TFTs exhibited an average charge-carrier mobility of 0.02 cm2/Vs, which is comparable to those of typical bottom-contact P3HT TFTs (from 0.01 to 0.1 cm2/Vs) [1]. An on/off current ratio of 10, however, was extracted from these OTFTs. This low on/off current ratio likely stems from a high off current due to leakage since the organic semiconductor layer was not patterned. Since μCP relies on conformal contact between the stamp and the substrate and the transfer of molecular inks from the stamp to the substrate, the mechanical distortion of the PDMS stamps [49] and diffusion of molecular inks [50,51] can be problematic when printing fine features with high resolution. Controlling these factors is therefore crucial for defining submicron channels. Leufgen and coworkers [52] fabricated bottom-contact OTFTs with channel lengths as small as 100 nm using μCP followed by selective etching. To achieve submicron features, they carefully controlled the contact pressure, the contact time, and the concentration of the molecular inks during μCP. These parameters directly affect how much ink (etch resist) is transferred to the substrate during printing. In Leufgen’s process, the PDMS stamp was supported on a flat, rigid, silicon backplane, so pressure had to be applied uniformly across the stamp. Increasing contact pressures and/or times broadened the stamped regions. This ultimately resulted in narrower channels after etching. Changes in the ink concentration, however, did not have a significant effect on pattern size for a given contact time (2 sec). Figure 5.5.6 shows how the channel length (etched region) can be reduced from 700 to 200 nm simply by increasing the printing pressure from 2 to 4 bar while holding all other conditions constant. While this method provides a useful tool to study short-channel OTFTs, it is doubtful that it can be extended to reproducibly fabricate more complex features.

5.5.3.2 SELECTIVE ELECTROLESS PLATING While μCP and selective etching can potentially replace costly photolithographic tools in OTFT fabrication, expensive deposition tools are still required to deposit the metal layers. Recently, a lower cost metal electroless plating deposition technique has been explored as a potential replacement for the more expensive sputtering and evaporation processes. During electroless plating, metal is deposited onto the desired surface from solution through an autocatalytic redox process [36]. Specifically, a reductant in the

8080_C005.fm Page 442 Friday, April 6, 2007 10:25 AM

442

Organic Field-Effect Transistors

Au

SiO2/Si Low p Stamp

2 bar 10s Substrate

1.3 μm 1.8 μm

0.7 μm

0.2 μm High p Stamp

4 bar 10s

Substrate

FIGURE 5.5.6 Submicron channel devices can be fabricated by μCP and selective etching by controlling the pressure during contact. After etching, the channel length is reduced from 700 to 200 nm by increasing the stamping pressure from 2 to 4 bar. (From M. Leufgen et al., Appl. Phys. Lett., 84, 1582, 2004.)

electroless plating solution reduces the cation of the metal, thereby depositing elemental metal on the surface. The surface, however, needs to be activated by a catalyst, such as palladium, to initiate metal deposition. The activated surface can be created by microcontact printing the catalyst on the substrate surface. The combination of μCP and electroless plating has successfully generated copper [36,37] and nickel [38,39] features. For example, Hidber and coworkers [36] fabricated copper electrodes by μCP and electroless plating. Glass, silicon/silicon dioxide, and polymer substrates were patterned with palladium colloids by μCP and submerged in an electroless plating solution. During electroless plating, copper growth only occurs in the regions containing palladium colloids. Using a derivative technique, Zschieschang and coworkers [39] microcontact printed FSAMs [30] on a hydrophilic flexible polyethylene naphthalate (PEN) substrate. The patterned substrate was then activated in a palladium bath; palladium catalyst only adsorbed in the hydrophilic, unstamped regions. Subsequent electroless plating resulted in the selective deposition of nickel only in the hydrophilic regions and not in the regions, stamped with FSAM. The electroless-plated nickel electrodes served as the gate electrode in flexible OTFTs. Solution-processable polyvinylphenol (PVP) was then deposited on top of the nickel electrodes as the gate dielectric. Gold source-drain electrodes in these devices were subsequently defined by photolithography. Finally, pentacene was thermally evaporated to complete the bottom-contact OTFTs. These OTFTs exhibited a charge-carrier mobility

8080_C005.fm Page 443 Friday, April 6, 2007 10:25 AM

Soft Lithography for Fabricating Organic Thin-Film Transistors

443

of 0.06 cm2/Vs, a subthreshold swing of 1.1 V/decade, and an on/off current ratio of 106. These characteristics are typical of bottom-contact pentacene TFTs with source and drain electrodes defined by ink-jet printing or other soft lithography techniques [39]. Electroless plating, however, is limited to bottom-contact device applications since the electroless plating solutions are generally not compatible with organic semiconductors.

5.5.3.3 SELECTIVE CHEMICAL POLYMERIZATION

OR

ELECTROCHEMICAL

An alternative candidate for source and drain electrodes is conducting polymers. To define conductive polymer electrodes, μCP is used to create hydrophilic and hydrophobic patterns (e.g., a molecular template), which can facilitate the selective deposition of conductive polymer [40–43]. Huang and coworkers [40] demonstrated the selective deposition of polypyrrole and polyaniline (PANI) conducting polymers on silicon/silicon dioxide and glass substrates by chemical polymerization. In this technique, regions of hydrophilic glass and silicon substrates are rendered hydrophobic by microcontact-printing alkylsilanes on the substrates [30,51]. Immersing the patterned substrates in a monomer/oxidant solution (pyrrole or aniline) results in the polymerization of the conducting polymer. Polymerization and deposition rates of the conducting polymers were found to be much higher in the hydrophobic regions than in the hydrophilic regions. As a consequence, a thicker polymer film is deposited in the hydrophobic regions of the substrate. But the conducting polymer adheres more strongly in the hydrophilic regions. Huang and coworkers were thus able to exploit the difference in adhesion to preferentially remove the polymer from the hydrophobic regions of the substrate. Since the polymer adheres more strongly to the hydrophilic surface, the thicker polymer layer can be easily peeled from the hydrophobic regions of the substrate with Scotch™ tape. Using this technique, conducting polymer features with lateral dimensions as small as 2 μm can be produced. Similar results can be obtained by electropolymerization [41–43], in which long alkylthiols (>10 carbons in length) are microcontact-printed onto a gold-coated substrate. These long alkylthiols are thought to inhibit interfacial charge transfer, thereby significantly reducing electropolymerization rates [41]. As such, polymer growth only takes place significantly in the gold regions that had not been stamped with alkylthiols. Using this technique, Gorman and coworkers [41] grew polypyrrole features as small as 2 μm with conductivities ranging from 1 to 5 S/cm. With a similar electropolymerization procedure, Parashkov and coworkers [53] selectively grew PEDOT-PSS source and drain electrodes on a gold substrate. Again, an alkylthiol template was microcontact-printed onto a gold-coated glass substrate for the selective growth of PEDOT-PSS. The PEDOT-PSS patterns exhibited conductivities of 1–5 S/cm. Since the underlying substrate (gold) is laterally conductive, the individual PEDOT-PSS patterns must be transferred to an insulating substrate for application (otherwise, all the conducting polymer patterns are electrically shorted by the underlying gold substrate). To transfer PEDOT-PSS patterns from the

8080_C005.fm Page 444 Friday, April 6, 2007 10:25 AM

444

Organic Field-Effect Transistors

gold surface, polyimide (PI) was cast and cured directly on the PEDOT-PSS patterned substrate. Peeling the PI substrate removed the PEDOT-PSS features from the gold surface. Bottom-contact OTFTs were fabricated on the PI platform in an “upside-down” fashion. A thin layer of pentacene was evaporated onto the PEDOTPSS source and drain electrodes. Poly(vinyl alcohol) (PVA) was subsequently spincoated on top of the pentacene layer to serve as the gate dielectric. PEDOT-PSS gate electrodes were then screen-printed on the PVA layer to complete the OTFTs. These devices thus adopt a bottom-contact, top-gate device geometry. The OTFTs fabricated by this technique had channel widths from 200 μm to 3.6 mm and channel lengths ranging from 5 to 50 μm. OTFTs with a channel width of 1 mm and a channel length of 10 μm typically exhibited a charge-carrier mobility of 0.02 cm2/Vs, a threshold voltage of –2.9 V, and an on/off current ratio of 13. These values are lower than those expected for pentacene bottom-contact transistors, but may result from leakage through the PVA polymer gate dielectric. Similar to the various μCP techniques described previously, the solutions required for chemical or electrochemical polymerization are generally not compatible with the organic semiconductors of OTFTs. As a consequence, the application of these procedures is also limited to the fabrication of bottom-contact devices.

5.5.3.4 STAMP-AND-SPIN-CAST Expanding the application of conducting polymers as source and drain electrodes in OTFTs, Lee and coworkers recently demonstrated a direct patterning technique for generating conducting polymer electrodes directly on insulating substrates [54]. Consequently, conductive polymer features can be directly patterned on the gate dielectric of an OTFT. Unlike the previously mentioned techniques for patterning conducting polymers, this technique, coined “stamp-and-spin-cast,” does not require any postpolymerization steps. Such direct patterning of a conducting polymer is enabled by the use of a water-dispersible PANI [54]. Stamp-and-spin-cast begins with microcontact printing hydrophobic alkylsilanes on a hydrophilic substrate (–OH terminated), as shown in Figure 5.5.7(a). To fabricate PANI source and drain electrodes, Lee and coworkers microcontact-printed octadecyltrichlorosilane (OTS) on highly-doped silicon with a thermally-grown silicon dioxide overlayer. Subsequent spin-coating of an aqueous PANI dispersion on the patterned substrate creates conductive PANI patterns in the hydrophilic regions. PANI features as small as 5 μm were achieved by Lee and coworkers using this technique [54]. Pentacene was thermally evaporated though a shadow mask on top on the PANI source and drain electrodes to complete bottom-contact devices on a silicon/silicon dioxide platform. Bottom-contact pentacene TFTs with PANI electrodes patterned by stamp-and-spin-cast with channel widths of 800 or 1,000 μm and channel lengths ranging from 30 to 300 μm exhibited an average charge-carrier mobility of 0.016 ± 0.008 cm2/Vs and on/off current ratio as high as 104. The reduced charge-carrier mobility (compared to typical pentacene bottomcontact TFTs) stemmed from the dielectric surface roughness (OTS was stamped on the dielectric surface). Using solution deposition methods to create the molecular templates yielded much smoother dielectric surfaces, and the performance of these

8080_C005.fm Page 445 Friday, April 6, 2007 10:25 AM

Soft Lithography for Fabricating Organic Thin-Film Transistors

445

UV/Ozone treatment to produce a hydrophilic surface O H O H O H O H O H O HO H

(a)

(b)

Deposit molecules R

Transfer OTS

R

R

R

R

R

R

OTS OTS OTS OH OH OH OH OH OH OH UV/Ozone exposure

Shadow mask

R

R

OH

R OH

R OH

Spin-cast PANI-PAAMPSA aqueous solution PANI-PAAMPSA

R OH

R

R OH

R OH

FIGURE 5.5.7 Schematic of stamp-and-spin-cast. Hydrophobic patterns are created by (a) μCP or (b) subtractively removing hydrophobic molecules by UV-ozone irradiation through a shadow mask. An aqueous PANI dispersion is spin-coated on the patterned substrate. PANI features selectively deposit in the hydrophilic regions immediately after the spin-coating. (From K. S. Lee et al., Appl. Phys. Lett., 86, 074102, 2005.)

devices improved accordingly [54]. These OTFTs with PANI electrodes exhibited linear output characteristics in the low source-drain voltage regime [54]. This observation implies lower contact resistance at the PANI–pentacene interface compared to that at the gold–pentacene interface, and has recently been verified by scanning surface potential measurement experiments [55].

5.5.3.5 OTHER MICROCONTACT PRINTING DERIVATIVES In addition to using μCP to fabricate electrodes for OTFTs as discussed previously, μCP can also be used to pattern the growth of organic–inorganic hybrid semicon-

8080_C005.fm Page 446 Friday, April 6, 2007 10:25 AM

446

Organic Field-Effect Transistors

(a)

(b)

FIGURE 5.5.8 Examples of the selective growth of oligoacene crystals on SAM-templated substrates. The dark regions and the bright regions indicate oligoacene and bare gold, respectively. (a) Oligoacene crystals grow on terphenylthiol–SAM-covered squares in the dodecylthiol–SAM-covered background. (b) Anthracene crystals are deposited on dodecylthiol–SAMcovered squares terphenylthiol SAM in the background. Scale bar = 300 μm. (From A. L. Briseno et al., J. Am. Chem. Soc., 127, 12164, 2005.)

ductors [56], organic semiconductor single crystals [57], and inorganic semiconductor single crystals [58]. Kagan and coworkers [56] demonstrated the selective deposition of a soluble organic–inorganic hybrid semiconductor, (C6H5C2H4NH3)2SnI4, on microcontact-printed molecular templates. In their method, hydrophobic inks, such as alkylsilanes, fluorinated alkylsilanes, alkylphosphonic acids, or alkylhydroxamic acids, were microcontact-printed on SiO 2 or ZrO 2 . A solution of (C6H5C2H4NH3)2SnI4 was then spin-coated on the patterned substrate. The semiconductor only adsorbed in the hydrophilic bare oxide regions due to the favorable wetting properties. The ability to pattern the semiconductor layer selectively is useful for building addressable arrays of transistors for display applications. Recently, Briseno and coworkers [57] demonstrated the selective growth of organic semiconductor single crystals on microcontact-printed molecular templates. Specifically, they investigated the growth of oligoacene crystals (anthracene and 5chlorotetracene) from a THF solution on substrates prepatterned with various SAMs. Their results showed that the crystal growth of oligoacenes varied dramatically depending on the chemical functionality of the stamped SAM. In particular, terphenylthiol SAMs induced the growth of large oligoacene single crystals from THF solution, while dodecylthiol-treated surfaces suppressed crystal growth. The use of a template containing terphenylthiol and dodecylthiol patterned regions therefore yielded patterned growth of oligoacene single crystals on the terphenylthiol-printed regions as shown in Figure 5.5.8. Functional 5-chlorotetracene TFTs were demonstrated by patterning dodecylthiol and terphenylthiol on gold source and drain electrodes. These patterned OTFTs exhibited an average charge-carrier mobility of 5.7 × 10–4cm2/Vs, which is comparable to that of single-crystal 5-chlorotetracene devices (1.4 × 10–4cm2/V) [59]. Finally, the selective growth of inorganic crystals was demonstrated on a SAMtemplated silver surface. Hsu and coworkers [58] preferentially grew zinc oxide (ZnO) from an aqueous solution on silver surfaces. The silver surface was patterned by μCP with a carboxyl acid (–COOH)-terminated SAM to inhibit ZnO growth. Figure 5.5.9(b) and (c) contain micrographs that demonstrate ZnO growth only

8080_C005.fm Page 447 Friday, April 6, 2007 10:25 AM

Soft Lithography for Fabricating Organic Thin-Film Transistors

(a)

447

(b) COOH terminated SAM micropatterns

Ag Si ZnO growth from aqueous solution

2 μm (c) 200 μm

ZnO

Ag Si

FIGURE 5.5.9 (a) COOH-terminated SAM is microcontact-printed on silver. ZnO preferentially grows on the bare silver surface rather than on COOH-terminated surface. (b, c) SEM images reveal that ZnO nanorods (white) only grow in the bare silver regions. The surrounding regions are covered with HSC10H20COOH. (From J. W. P. Hsu et al., Nano. Lett., 5, 83, 2005.)

occurs in the bare silver regions of the substrate. Since ZnO has a wide band gap, potential applications of patterned ZnO microstructures include photovoltaics, active sensor platforms, and microlasers [58].

5.5.4 NANOTRANSFER PRINTING (nTP) While μCP is a highly effective patterning technique, it is still subtractive in nature, requiring etchback to define functional features. Often, the chemicals used for etchback are incompatible with organic semiconductors, thereby limiting the application of μCP to bottom-contact devices. To eliminate the materials incompatibility issues and enable top-contact device fabrication, Loo and coworkers developed an additive patterning technique called nanotransfer printing (nTP) [20,21,60–62]. Nanotransfer printing is a solventless patterning technique for directly printing functional solid materials — these can be metal features or stacks of functional materials — from the raised regions of a stamp onto a substrate. Moreover, nTP occurs at ambient conditions and allows features as small as 100 nm [20] to be patterned over large areas on both rigid and flexible substrates. Using nTP, patterns have been transferred from rigid stamps [21,63] (e.g., GaAs or fused silica) and elastomeric PDMS stamps [20,21,60–62,64,65]. Pattern transfer results from an interfacial chemistry between the substrate and the material being transferred. As a consequence, the contact between the stamp and the substrate is

8080_C005.fm Page 448 Friday, April 6, 2007 10:25 AM

448

Organic Field-Effect Transistors

critical for achieving high-quality printed patterns. Using a PDMS stamp or substrate during nTP ensures conformal contact between the stamp and substrate via van der Waals interactions. The intimate contact results in high-quality printing at ambient pressures. If a rigid stamp is used to transfer patterns to a rigid substrate, external pressure (>150 MPa [66]) is required to obtain intimate contact between the stamp and the substrate. This technique has been demonstrated with cold welding of metal features [66–68], and is discussed in Section 5.5.6 of this chapter. This section will focus on systems with at least one conformal surface so that printing can be carried out at ambient pressures. The success of nTP relies on interfacial chemistries between the solid material being transferred and the substrate. For the nTP procedure shown in Figure 5.5.10, the reactivity of titania is exploited for pattern transfer [20]. Thin films of gold (~40 nm) and titanium (~5 nm) are deposited on the raised and recessed regions of a patterned stamp (either PDMS or GaAs) via sputtering or electron beam evaporation. Collimation of the metal fluxes during deposition prevents the sidewalls of the stamp from becoming coated with metal. Exposing the titanium to ambient conditions oxidizes its surface to form titania (TiOx). The gold and titania stack can then be transferred from the raised regions of the stamp onto a plastic substrate on contact. To prepare the plastic substrate for pattern transfer, a thin film of PDMS is spin coated onto a PET substrate to create a conformal, flexible substrate. Before contacting the stamp with the substrate, titania and PDMS are exposed to oxygen plasma to create reactive –OH groups on their respective surfaces [69–71]. Contacting the stamp with the substrate results in conformal contact between the raised regions of the stamp and the substrate. In the regions of physical contact, condensation reactions between the hydroxyl groups of the titania and those on the PDMS surfaces result in permanent covalent bonds. Removing the PDMS stamp (to which gold has poor adhesion) from the plastic substrate transfers the gold/titania metal from the raised regions of the stamp to the substrate. The entire procedure occurs at ambient conditions with contact times less than 15 sec. Scotch tape adhesion tests further confirm that the transferred patterns are strongly bonded to the substrate. Since aluminum also readily forms surface oxides, the same interfacial chemistry can be used to generate aluminum patterns [20]. Figure 5.5.11 illustrates the printing quality of nTP. Scanning electron micrographs of a PDMS stamp with gold deposited on the raised and recessed regions before and after nTP are shown in Figures 5.5.11(a) and (b), respectively. These micrographs clearly illustrate that the metal stack transfers completely from the raised regions of the stamp. Further, Figure 5.5.11(c) demonstrates that the fidelity of the PDMS pattern is replicated on the PDMS/PET substrate. Combining nTP with soft contact lamination (ScL), which will be discussed in Section 5.5.5, allows functional top-contact devices to be built on plastic substrates [20,63]. Loo and coworkers fabricated organic transistors and complementary inverter circuits on PDMS/PET substrates in which the gold source/drain electrodes and interconnects were patterned by nTP [20]. Figure 5.5.12(a) shows the currentvoltage characteristics of a representative pentacene [72] (p-type) TFT fabricated by a combination of nTP and soft contact lamination. The electrical properties of these top-contact devices (charge-carrier mobility ~ 0.1 cm2/V-s, on/off current ratio ~ 104)

8080_C005.fm Page 449 Friday, April 6, 2007 10:25 AM

Soft Lithography for Fabricating Organic Thin-Film Transistors

449

Stamp

(a)

Deposit Au/Ti on stamp (b) 0.2 − 10 μm 0.1 − 10 cm 0.05 − 100 μm

(c)

(d)

Plasma oxidizes surfaces of stamp, substrate; print

Substrate

Si Ti

O

TiOx chemically bonds to substrate

Remove substrate; Au/Ti printing complete (e)

FIGURE 5.5.10 Schematic of nanotransfer printing (nTP): (a) Stamp with relief features ranging from 0.2 to 10 μm and the lateral dimensions ranging from 0.05 to 100 μm. (b) 20nm Au and 5-nm Ti are sequentially evaporated on the raised and recessed regions of the stamp. (c) The stamp and substrate are exposed to oxygen plasma to create surface (–OH) groups and brought into contact with one another. (d) An interfacial condensation reaction occurs in the regions of contact binding the metal to the substrate. (e) Peeling away the stamp transfers the pattern to the substrate via permanent covalent bonds. (From Y.-L. Loo et al., Appl. Phys. Lett., 81, 562, 2002.)

8080_C005.fm Page 450 Friday, April 6, 2007 10:25 AM

450

Organic Field-Effect Transistors

(a)

100 μm (b)

100 μm (c)

100 μm

FIGURE 5.5.11 Scanning electron micrographs of a PDMS stamp with 20-nm Au and 5nm Ti on the raised and recessed regions before (a) and after (b) nTP. (c) The resulting printed metal features on a plastic substrate. (From Y.-L. Loo et al., Appl. Phys. Lett., 81, 562, 2002.)

8080_C005.fm Page 451 Friday, April 6, 2007 10:25 AM

Soft Lithography for Fabricating Organic Thin-Film Transistors

Drain-source current (μA)

Drain-source current (μA)

2 10 102

0 10 100

-30

Vg = −100 V

−2\-2 1010 0 −50 -50 −100 -100 -20 0 Gate Voltage Gate Voltage −10

0 0

451

40

100 μm m

30

Vg = 100 V

20 10

−20 −40 −60 −80 −100

0 0

20

40

60

80

Drain-source voltage (V)

Drain-source voltage (V)

(a)

(b)

100

Vout (volts)

40 30 20 10 0 0

10

20

30

40

Vin (volts) (c)

FIGURE 5.5.12 (a) Current-voltage characteristics of pentacene thin-film transistors whose gold electrodes were printed by nTP. Inset: saturated source-drain current as a function of gate voltage (W/L = 8). (b) Switching characteristics of a complementary inverter circuit whose gold electrodes were printed by nTP. (From Y.-L. Loo et al., Appl. Phys. Lett., 81, 562, 2002.)

are comparable to top-contact transistors in which the gold electrodes are evaporated through a shadow mask directly on top of pentacene [20]. Figure 5.5.12(b) shows the transfer characteristics of a complementary organic inverter circuit with pentacene (p-type) and hexadecafluoro copper phthalacyanine [73] (n-type) organic semiconductors. Again, the performance of this circuit compares well with top-contact devices in which the electrodes and wiring are fabricated by evaporation through a shadow mask [20]. The nTP process described in the preceding paragraphs relies on interfacial condensation chemistry to transfer metal patterns from a PDMS stamp to a plastic substrate. Realizing that the interfacial chemistry can be modified according to the material to be transferred and the desired substrate, nTP can easily be extended for patterning a variety of single- and multilayer conductors, semiconductors, and dielectrics. In one such variation, SAMs serve as a covalent “glue” for transferring gold from PDMS stamps onto GaAs substrates. Figure 5.5.13 illustrates this concept. Difunctional 1,8-octanedithol molecules are deposited on freshly etched GaAs. The adsorption of the molecules on the substrate occurs such that only one of the thiol endgroups bonds to the GaAs substrate, leaving the second thiol functionality free for subsequent reaction [60].

8080_C005.fm Page 452 Friday, April 6, 2007 10:25 AM

452

Organic Field-Effect Transistors

(a) Etch oxide; deposit molecules 20 nm Au PDMS stamp

SH

SH

SH

SH

(CH2)X (CH2)X (CH2)X (CH2)X S

S

S

S

Au

Au

S

S

GaAs

(b) Bring stamp into contact with substrate

(c) Remove stamp; complete nTP

SH

SH

(CH2)X (CH2)X (CH2)X (CH2)X S

S

S

S

GaAs

FIGURE 5.5.13 Schematic of nTP on GaAs. (a) 1,8-octanedithiol molecules are deposited on freshly etched GaAs. Inset: idealized orientation of 1,8-octanedithiol molecules on GaAs. (b) A PDMS stamp with 20 nm of gold evaporated on the raised and recessed regions of the stamp is contacted against the treated GaAs surface. (c) Peeling away the PDMS stamp effectively transfers gold onto the GaAs substrate in the regions of contact. Inset: idealized orientation of 1,8-octanedithiol molecules as “glue” between GaAs and gold. (From Y.-L. Loo, et al., J. Vac. Sci. Technol., B, 20, 2853, 2002.)

An idealized schematic of the orientation of the deposited 1,8-octanedithiol layer is shown in the inset of Figure 5.5.13. Separately, a thin layer (~20 nm) of gold is deposited onto the raised and recessed regions of a PDMS stamp. The stamp is then brought into contact with the thiol-functionalized GaAs substrate. In the regions of conformal physical contact between the stamp and substrate, an instantaneous chemical reaction occurs between the unreacted thiol end groups and gold on the raised regions of the stamp, resulting in permanent gold–sulfur covalent bonds. Removing the stamp leaves behind a gold pattern, as shown in Figure 5.5.14. Since gold is permanently adhered to the substrate through covalent bonds, the printed patterns always pass Scotch tape adhesion tests. Again, this chemistry works at room temperature with contact times less than 30 sec. Further, the interfacial thiol chemistry is expected to work with all coinage metals. In fact, a similar nTP process has been demonstrated for patterning copper [62].

8080_C005.fm Page 453 Friday, April 6, 2007 10:25 AM

Soft Lithography for Fabricating Organic Thin-Film Transistors

1 μm (a)

100 μm (b)

50 μm (c)

453

100 μm (d)

FIGURE 5.5.14 Optical micrographs of gold patterns printed by nTP. (a) 100-nm holes printed with a GaAs hard stamp on a plastic substrate coated with PDMS; gold patterns printed with a PDMS stamp on (b) a PDMS substrate; (c) a silicon/silicon dioxide substrate; and (d) an ITO-coated plastic substrate. (From Y.-L. Loo, et al., J. Vac. Sci. Technol., B, 20, 2853, 2002.)

There is, however, a significant difference between printed gold and copper patterns. Unlike printed gold patterns, printing with as-cast PDMS stamps results in nonconductive copper patterns. An additional step of leaching the PDMS stamps prior to metal deposition must therefore be instituted to print conductive copper patterns. X-ray photoelectron spectroscopy carried out during depth profiling experiments revealed that uncross-linked PDMS oligomers can penetrate between copper grains, thereby disrupting the lateral conductive pathway. Leaching removes uncrosslinked PDMS oligomers from the stamps. This phenomenon — that PDMS oligomers penetrate between copper grains — appears to be related to whether the metal oxidizes at ambient conditions. Specifically, gold, which does not oxidize at ambient conditions, is always conductive whether it is printed from as-cast or leached PDMS stamps. Copper [62] and silver [74], both of which oxidize at ambient conditions, are only electrically conductive when printed from leached PDMS stamps. In addition to patterning on GaAs, SAMs can also be used to bind gold patterns to silicon substrates. Similar to the procedure for patterning on GaAs, a molecule with thiol functionality, 3-mercaptopropyltrimethoxysilane (MPTMS), is employed. The silicon wafer is first treated (with a 6:1:1 mix of water:H2O2:HCl, for 10 min at 75°C) [21] to create reactive hydroxyl surface functionality (–OH). The methoxy groups of MPTMS react with the –OH functionality to generate a

8080_C005.fm Page 454 Friday, April 6, 2007 10:25 AM

454

(a)

200 μm

Organic Field-Effect Transistors

(b)

200 μm

FIGURE 5.5.15 Optical micrographs of (a) silver dots and (b) copper lines printed on silicon by nTP with as-cast PDMS stamps.

thiol-terminated (–SH) silicon surface [21,75]. Contacting a PDMS stamp (with gold deposited on the raised and recessed regions of the stamp) with the thiolfunctionalized silicon surface transfers gold patterns in the regions of contact. Again, the gold patterns are covalently bound to the silicon substrate through permanent gold–sulfur linkages. Similar results, shown in Figure 5.5.15, can be achieved for patterning silver and copper on silicon [62,74]. As noted previously, the PDMS stamps must be leached prior to printing conductive silver and copper patterns. When leached stamps are used, the conductivities of the printed copper and silver features are comparable to evaporated thin films of copper and silver of similar thicknesses. Additionally, the same interfacial silane chemistry can be employed with any substrates that have surface hydroxyl groups or can be functionalized with surface hydroxyl groups. For example, gold patterns can be printed on PET substrates coated with organosilsesquioxane (glass resin) [21]. As discussed in Section 5.5.2, the mechanical properties of commercially-available Sylgard 184 PDMS limit the minimum feature size achievable to several microns [18,19]. Specifically, shallow relief features tend to deform, buckle, or collapse [18,19] when brought into contact with the substrate to be patterned. To generate nanoscale features with nTP, hard stamps must be used. Loo and coworkers generated submicron patterns with GaAs hard stamps, as illustrated in Figure 5.5.16 [20]. Here, gold patterns are transferred at ambient pressures onto a conformal plastic substrate coated with PDMS. The printed patterns are uniform over large areas. Figures 5.5.16(b) and (c) show the edge roughness of the printed features to be less than 15 nm, suggesting that when a stamp with smooth edges is used, the edge resolution of printed features is limited by the grain size of the evaporated metal. While rigid hard stamps are successfully used to print submicron features, elastomeric stamps remain attractive because they can be easily fabricated and they readily conform to rigid substrates at ambient pressures. The low elastic modulus and high compressibility of elastomeric PDMS stamps can be overcome with composite PDMS stamps (h-PDMS) [22,23]. As discussed in Section 5.5.2, h-PDMS stamps consist of a Sylgard 184 PDMS layer that provides conformal contact with

8080_C005.fm Page 455 Friday, April 6, 2007 10:25 AM

Soft Lithography for Fabricating Organic Thin-Film Transistors

(a)

455

(b)

5 μm

1 μm

(c)

100 nm

FIGURE 5.5.16 Scanning electron micrographs of Au/Ti patterns printed onto PDMS/PET substrates. (a) 500-nm lines printed with a fused silica stamp; (b) intersecting trenches printed with a GaAs stamp; and (c) 130-nm hole array printed with a GaAs stamp. The bright regions in the micrographs represent transferred Au/Ti. (From Y.-L. Loo et al., Appl. Phys. Lett., 81, 562, 2002.)

the substrate and a rigid h-PDMS layer that provides nanoscale feature resolution. These h-PDMS stamps have successfully been used by Zaumseil and coworkers to print three-dimensional nanostructures with features as small as 50–100 nm [64]. The ability to print complex multilayer stacks and the ability to separate the deposition and patterning steps effectively make nTP a powerful tool for building devices on plastic. For example, Loo and coworkers demonstrated the versatility of nTP by fabricating metal–insulator–metal capacitors on plastic at ambient conditions [21]. In these devices, SiNx functions as the dielectric. Direct plasma deposition of SiNx on plastic is generally limited by the low working temperature of the substrate. The nTP process, however, can be exploited to transfer SiNx onto plastic at ambient conditions. To fabricate the capacitor structure, layers of Au (50 nm), SiNx (100 nm), Ti (5 nm), and Au (50 nm) are sequentially deposited onto a lithographically patterned silicon stamp treated with FSAM. The multilayer stack is then transferred to a gold-coated plastic substrate via cold welding [66–68] (see Section 5.5.6) when the stamp is brought into contact with the substrate. The performance of the printed capacitors was comparable to that of capacitors of similar dimensions fabricated on silicon wafers by photolithography, lift-off, and etchback [21]. To this point, SAMs have been discussed in the context of glues to anchor metal patterns or metal stacks to a substrate. If we extend our thinking to consider SAMs as the electrically-active components [65,76,77], we see how nTP can be a useful

8080_C005.fm Page 456 Friday, April 6, 2007 10:25 AM

456

Organic Field-Effect Transistors

tool for building nanoscale electronic devices. Specifically, nTP can provide a nondestructive method for making electrical contact to a molecular layer. In contrast, direct evaporation of metal contacts on top of a molecular layer generally leads to electrical shorts in these devices since hot metal atoms can penetrate and/or damage the molecules during evaporation [78]. Using nTP, Loo and coworkers fabricated GaAs/1,8-octanedithiol/gold molecular junctions by nTP. In these devices, the 1,8octanedithiol is not only the glue that anchors gold on GaAs, but also the electricallyactive layer between two electrodes (gold and GaAs) [65]. The electrical properties of 1,8-octanedithiol, however, are not very interesting. Recent work in this area focuses on the assembly of conjugated glues [76,79–83]. These conjugated, electrically-active molecules are attractive because they can potentially carry charges along their molecular backbone. Combining the patternability of nTP with electrically-interesting molecules moves us one step closer to understanding the fundamentals of nanoscale electronics and perhaps even the realization of molecular electronics. In addition to the nTP procedures described in the preceding paragraphs, a noncovalent transfer process was recently demonstrated by Hur and coworkers for patterning gold and gold/titanium multilayers [61]. As the name implies, this procedure does not rely on the formation of specific covalent bonds for pattern transfer. Instead, noncovalent surface forces are exploited. The significant advantage of this procedure, however, is the ability to print electrodes directly on top of the organic semiconductor thin films without prior modification or surface treatment. Figure 5.5.17 illustrates this procedure. Similar to previously described nTP procedures, the metal to be transferred is deposited on the raised and recessed regions of a PDMS stamp. The stamp is contacted against the substrate and heated mildly (50–80°C). The heating time required for pattern transfer is a function of the surface energies, and varies depending on the metal being transferred. A high surface energy substrate, like PET or polythiophene, tends to be “stickier” and therefore requires less heating time for pattern transfer (surface energy of PDMS < surface energy of PET) than a lowenergy surface, like pentacene or polypropylene (surface energy of PDMS ≈ surface energy of pentacene). For successful patterning, the nonspecific adhesion at the metal–stamp interface must be less than the nonspecific adhesion between the metal–substrate interface (i.e., the metal must prefer to adhere to the substrate over adhering to the stamp). Since PDMS has an extremely low surface energy, this is typically not a problem with most materials. Removing the stamp from the substrate transfers the metal pattern in the regions of contact. Figures 5.5.17(b) through (e) demonstrate some of the patterns generated with this technique [61]. Deposition of metals onto PDMS stamps by electron-beam evaporation, however, can modify the PDMS surface, thereby increasing the surface energy of the PDMS above its intrinsic value. If the surface energy of the PDMS becomes greater than the surface energy of the substrate, the metal will not transfer from the PDMS stamp. This is typically not a concern with conventional nTP since it relies on specific interfacial chemistries for pattern transfer. Hur and coworkers found that the surface energy of the PDMS stamp can be restored to close to its intrinsic value by instituting a heating step. They hypothesized that this heating step facilitates reorientation and

8080_C005.fm Page 457 Friday, April 6, 2007 10:25 AM

Soft Lithography for Fabricating Organic Thin-Film Transistors

457

(a) Evaporated metal ﬁlms PDMS Stamp Contact surface

Apply heat Substrate Remove stamp

(b)

(c)

500 nm 1 cm (d)

1 μm (e)

1 cm

100 μm

10 μm

FIGURE 5.5.17 (a) Schematic of noncovalent nTP: A PDMS stamp, with metal deposited on the raised and recessed regions, is contacted against a substrate. After moderate heating, the stamp is removed, leaving behind metal in the regions of contact. (b) Optical micrograph of 30-nm thick gold dots printed on a plastic substrate. (c) Scanning electron micrograph of a small region of the same dot array. (d) Optical micrograph of Ti (2 nm)/Au (30 nm) patterns printed on a silicon/silicon dioxide substrate. (e) Optical and scanning electron micrographs of a smaller region of the same printed pattern. (From S.-H. Hur et al., Appl. Phys. Lett., 85, 5730, 2004.)

segmental motion of the polymer chains and/or diffusion of the low molecular weight components of the PDMS to the stamp surface [61]. In addition to restoring surface energy, the heating step may also improve contact between the stamp and the substrate. Heating times required for pattern transfer can also be affected by the surface roughness of the substrate. Overall, smoother surfaces lead to better contact between the stamp and the substrate and therefore higher quality pattern transfer. Hur and coworkers successfully fabricated top-contact P3HT TFTs on plastic substrates with printed multilayer gold/titanium and gold/titanium/gold electrodes

8080_C005.fm Page 458 Friday, April 6, 2007 10:25 AM

458

Organic Field-Effect Transistors

by noncovalent nTP [61]. Fabrication starts on a polyimide substrate. Gold, gold/titanium, and gold/titanium/gold electrodes are printed directly on top of P3HT to complete the circuit. The authors observed that the multilayer electrodes have fewer cracks and are therefore more electrically conductive. They attribute this observation to the improved structural integrity of the multilayer stack. Further, the performance of all of the printed devices (charge-carrier mobility ~ 3.5 × 10–3 cm2/Vs, for a device with a channel width of 200 μm, and a channel length of 50 μm) is comparable to the performance of devices in which the electrodes were directly evaporated through a shadow mask (charge-carrier mobility ~ 3.0 × 10–3 cm2/Vs, for devices with a channel width of 200 μm, and a channel length of 50 μm) [61]. The most significant advantage of nTP over traditional patterning techniques is that nTP is purely additive. This allows the functional material to be transferred directly onto the desired substrate with exact placement, eliminating the use of photoresists, etchants, and developers that are required for photolithography and microcontact printing (μCP) techniques. Also, since solid functional materials can be directly transferred, ink diffusion concerns associated with μCP are eliminated. Finally, the stamp is reusable (but does require cleaning over time), and the process allows for direct patterning of three-dimensional and multilayer structures without etching or additional processing steps [21,84].

5.5.5 SOFT-CONTACT LAMINATION (ScL) Combining μCP with lamination techniques [85,86], Loo and coworkers fabricated top-contact organic transistor arrays on plastic substrates. This procedure was subsequently called soft-contact lamination (ScL) [87,88]. ScL exploits the conformal properties of PDMS, so individual components of the device can be made separately and then laminated together in the final step. This feature of ScL is particularly useful for assembling OTFTs considering the chemical and mechanical fragility of the organic semiconductor. In SeL, the organic semiconductor can be deposited independently on one conformal substrate (which also contains the gate and gate dielectric), while the source and drain electrodes are patterned on a separate conformal substrate. The two individual substrates are then laminated together to complete the circuit. Consequently, devices fabricated by ScL adopt the top-contact geometry. Since many organic materials are incompatible with the traditional lithographic and μCP processes used to fabricate and pattern the electrical contacts, separating the deposition and patterning steps greatly simplifies the fabrication of plastic electronics and avoids exposing the organic semiconductor to harsh chemical environments. Further, ScL provides a facile method for fabricating embedded transistors that are mechanically flexible. Figure 5.5.18 illustrates the ScL procedure used by Loo and coworkers to fabricate laminated pentacene TFTs. The top substrate, containing the μCP-printed source and drain electrodes, was constructed on a PDMS-coated plastic substrate [89]. The bottom substrate was constructed on an ITO-coated plastic substrate. The ITO, patterned photo lithographically to define the gate level, was coated with an organosilesquioxane spin-on glass, which comprises the dielectric. Pentacene [44]

8080_C005.fm Page 459 Friday, April 6, 2007 10:25 AM

Soft Lithography for Fabricating Organic Thin-Film Transistors

(a)

(b)

459

Transfer cast PDMS onto PET; oxidize; deposit Ti/Au

PDMS PET

Au OH PDMS

Si

Large-area μCP; oxidize PDMS

Flip sheet over; laminate

(c)

Semiconductor, dielectric gate Contacts PET

(d)

PET

PDMS Dielectric

Interfacial bonding; complete circuit

Si

Si

FIGURE 5.5.18 Schematic of soft-contact lamination. (a) Uniform layers of Ti (~1 nm) and Au (15–20 nm) are deposited onto a PDMS-coated plastic substrate. (b) Gold source and drain electrodes and interconnects are defined by microcontact printing. (c) This sheet is laminated against a bottom plastic substrate that contains the gate, dielectric and, semiconductor levels. (d) The completed laminated circuit. Insets: magnified view of the laminated circuit and the side profile of a laminated electrode. (From Y.-L. Loo et al., Proc. Natl. Acad. Sci. U.S.A., 99, 10252, 2002.)

8080_C005.fm Page 460 Friday, April 6, 2007 10:25 AM

460

Organic Field-Effect Transistors

was evaporated on the dielectric layer through a metal shadow mask and annealed at 100°C in a nitrogen environment for 6 h to create the organic semiconductor layer. Aligning and contacting the top substrate with the bottom substrate completed the circuit [89]. The PDMS layer on the top substrate allowed conformal contact with the bottom substrate at ambient pressures through van der Waals interactions, yielding a complete circuit embedded between two sheets of plastic without the use of adhesives [71]. The conformal contact provided by the PDMS layer also provided efficient electrical contact between the source/drain electrodes and the organic semiconductor layer. Zaumseil and coworkers demonstrated that contacts formed by lamination actually exhibit lower contact resistance between the electrodes and the organic semiconductor than contacts formed by direct evaporation of gold on top of the organic semiconductor [87]. Lower contact resistance translates to lower power consumption per unit current output during device operation. Additionally, the encapsulated, laminated transistors are mechanically robust. By placing the active elements of the circuits at the zero-stress plane [90] through lamination, the circuits do not fail due to mechanical fracture of the ITO layer (the most brittle material in this system), nor do they fail due to plastic deformation of the PET substrate [89]. The encapsulation that results from lamination also has practical importance since many organic semiconductors are highly sensitive to their environment. While PDMS is not designed to form a hermetic seal, it did provide several hours of protection when the laminated circuit was immersed in soapy water [89]. ScL was also used to study charge transport in organic semiconductor single crystals [91]. Prior to ScL, organic semiconductor single-crystal transistors were fabricated by placing a thin organic semiconductor single crystal (~1 μm) against a silicon wafer with predefined electrodes [92–94]. While this technique is successful, it requires extremely thin, bendable (i.e., fragile) organic semiconductor single crystals to achieve successful lamination between the crystal and the electrodes, since this process relies purely on electrostatics between the crystal and the electrodes. ScL provides two advantages over this technique. Since the metal electrodes supported on PDMS substrates are conformal to moderate surface topography, thicker (up to a few millimeters) and more rigid organic semiconductor single crystals can be studied. Further, the ScL technique is nondestructive and reversible, so the laminated contacts can be removed from the organic semiconductor single-crystal surface and re-established many times without affecting the transistor characteristics [91]. To contact the organic semiconductor single crystal, the source, drain, and gate electrode levels are all fabricated independently on a PDMS substrate [91,95]. The PDMS substrate containing the patterned gate, source, and drain electrodes is then laminated against the organic semiconductor single crystal to complete the circuit (Figure 5.5.19). This capability allows many transistor structures — each with different electrode configurations and dimensions — to be assembled and characterized on the same region of the crystal. Figure 5.5.20(b) shows data collected from a set of transistors with different channel lengths assembled sequentially on a single region of a rubrene single crystal. As expected, the saturation current scales linearly with the width to length ratio (W/L) of the transistor channel [91]. Additionally, the ScL technique has allowed Sundar and coworkers to study how molecular anisotropy

8080_C005.fm Page 461 Friday, April 6, 2007 10:25 AM

Soft Lithography for Fabricating Organic Thin-Film Transistors

Drain

Source

461

Top view

Dielectric Gate 500 μm Substrate

(a) Organic crystal

100 μm

Initiate contact

Complete lamination

(b)

FIGURE 5.5.19 (a) Schematic of sequential deposition of gate, dielectric, and source and drain electrodes onto a PDMS substrate. Inset: top view of the PDMS transistor platform. (b) Schematic of procedure for laminating an organic semiconductor single crystal against the PDMS transistor platform. Initial contact (top frame) between the crystal and the PDMS results in a “wetting” front (middle frame) that proceeds across the entire crystal/PDMS interface (bottom frame). Insets: optical micrographs of the progress of the wetting front. (From V. C. Sundar et al., Science, 303, 1644, 2004.)

8080_C005.fm Page 462 Friday, April 6, 2007 10:25 AM

Organic Field-Effect Transistors

Source-drain current (μA)

462

−5 −4 −3 −2 −1 0

−1

0

−2

−100 −80 −60 −40 −20 0 0

−20

−60

−40

−80

−100

Source-drain current (μA)

Source-drain voltage (V) (a)

−140

−150

−120

−100

−100

−50

−80

0

−60

75 μm 100 μm 120 μm 0

3

6 W/L

9

12

170 μm

−40

220 μm

−20 0 40

20

0

−20 −40 −60 Gate voltage (V)

−80

−100

(b)

FIGURE 5.5.20 (a) Current-voltage characteristics of a laminated rubrene single crystal transistor (L = 75 μm, W = 980 μm; Ci = 0.67 nF/cm2). Inset: linear regime current-voltage characteristics. (b) Transfer characteristics measured at a source-drain voltage of –100 V in the same region of a rubrene single crystal by laminating transistor stamp structures of various channel lengths (L = 220, 170, 120, 100, and 75 μm; W = 980 μm). Inset: linear scaling of the saturation currents with the W/L ratio. (From V. C. Sundar et al., Science, 303, 1644, 2004.)

8080_C005.fm Page 463 Friday, April 6, 2007 10:25 AM

Soft Lithography for Fabricating Organic Thin-Film Transistors

W x Rcontact (Ω cm)

106

Channel conductivity (μS)

5 4 3

463

D W L

105 −80 −60

2

−40

−20

0

1 b direction a direction

0 −80

−60

−40 −20 0 Gate voltage (V)

20

FIGURE 5.5.21 Four-probe conductivity measurements of a rubrene single crystal as a function of gate voltage along the b and a axes. Intrinsic mobilities of 15.4 and 4.4 cm2/V-s were measured along the b and a axes, respectively. Inset: contact resistance along the a and b directions of the rubrene single crystal. (From V. C. Sundar et al., Science, 303, 1644, 2004.)

in the organic semiconductor single crystal affects charge-carrier mobility across the transistor channel [91]. The rubrene single crystal (p-type), for example, shows maximum hole mobility along the b axis of its unit cell (Figure 5.5.21), consistent with the expectation of stronger π-orbital overlap along the b axis due to molecular packing [91]. In addition to the transistors described earlier, Zaumseil and coworkers [88] fabricated nanoscale transistors by soft-contact lamination, as depicted in Figure 5.5.22. In this procedure, the stamp is incorporated as an element of the device. Specifically, gold evaporated on the raised regions of the PDMS stamp defines the source and drain electrodes while the recessed regions of the stamp define the transistor channel. To access a nanoscale channel, only the recessed regions of the PDMS stamp need to be on the nanometer length scale. An elastomeric nanoscale transistor platform can thus be generated by depositing titanium and gold on the raised regions of an h-PDMS stamp. The PDMS platform is laminated against a silicon substrate (gate) consisting of SiNx (dielectric), pentacene (organic semiconductor), and metal contact pads to connect to the electrodes on the PDMS substrate. Figure 5.5.23 illustrates the electrical properties of a representative nanoscale laminated transistor (channel width of 20 μm and channel length of 150 nm). The transistor exhibits a lower charge-carrier mobility and on/off current ratio compared to laminated micron-size transistors (channel width of 20 μm and channel lengths of 2.5 and 100 μm). Zaumseil and coworkers attributed the differences to contact resistances and short-channel effects [88]. Methods to lower contact resistance, which may include the clever use of monolayer chemistry and conductive polymers as electrodes, are being explored [87,88].

8080_C005.fm Page 464 Friday, April 6, 2007 10:25 AM

464

Organic Field-Effect Transistors

PDMS stamp with source and drain electrode relief; Ti/Au deposited on plasma oxidized surface

Au Au Bring stamp in close contact with semiconductor and probe pads

Semiconductor Dielectric Gate electrode Probe pads Ti/Au/Ti

FIGURE 5.5.22 Schematic of soft-contact lamination for building thin-film transistors. The electrodes are defined by the raised regions of the PDMS stamp. (From J. Zaumseil et al., Appl. Phys. Lett., 82, 793, 2003.)

Source-drain current (μA)

–0.25 150 nm

–0.20 –0.15

1 μm

–0.10 –0.05 0.00 0.05 0.0

–0.5 –1.0 –1.5 Source-drain voltage (V)

–2.0

FIGURE 5.5.23 Current-voltage characteristics of a typical laminated nanoscale pentacene transistor (W ~ 20 μm and L ~ 150 nm). Inset: scanning electron micrograph of the channel region defined by the recessed region of the PDMS stamp. (From J. Zaumseil et al., Appl. Phys. Lett., 82, 793, 2003.)

Lee and coworkers [96] also used ScL to construct organic light-emitting diodes (OLEDs). Similar to the transistors described in preceding paragraphs, the laminated OLEDs consist of two parts: conformal gold electrodes supported on PDMS and a transparent, ITO-coated substrate on which the electroluminescent organic material is deposited. Bringing the two substrates together completes the device, as shown in Figure 5.5.24. For comparison, reference OLEDs were fabricated by evaporating gold

8080_C005.fm Page 465 Friday, April 6, 2007 10:25 AM

Soft Lithography for Fabricating Organic Thin-Film Transistors

465

PDMS Ti/Au

EL layer ITO Substrate (a) PDMS EL layer ITO

(b)

(c)

FIGURE 5.5.24 Schematic of ScL for fabricating laminated OLEDs. (a) Ti (1 nm)/Au (20–60 nm) electrodes are deposited on a PDMS stamp while the electroluminescent layer (EL) is deposited on an ITO-coated transparent substrate. (b) The PDMS substrate containing the electrodes is laminated against the EL layer at ambient conditions to complete the circuit. (c) Optical micrograph of the laminated OLED. (Scale bar, 50 μm). (From T.-W. Lee et al., Proc. Natl. Acad. Sci. U.S.A., 101, 429, 2004.)

8080_C005.fm Page 466 Friday, April 6, 2007 10:25 AM

466

Organic Field-Effect Transistors

electrodes directly on top of the electroluminescent organic material. Both sets of devices exhibited uniform spatial emission. The laminated devices, however, exhibited higher quantum efficiencies. Lee and coworkers attributed the difference in quantum efficiencies to the disruption of the π–π-conjugation when hot metal atoms are directly evaporated on top of the electroluminescent material [96]. These results further demonstrate the advantages of separating the organic semiconductor deposition process from the source and drain patterning steps during the fabrication of organic devices.

5.5.6 COLD WELDING Cold welding is another contact printing technique that can be used to generate electrodes for top-contact devices [66,97,98]. Cold welding differs from nTP in that it relies on the formation of metallic bonds between two metal surfaces of similar composition for pattern transfer. Figure 5.5.25 illustrates the cold welding process. A silicon hard stamp is coated with an adhesion reduction layer (typically an organic material, such as pentacene) followed by gold. A thin sacrificial layer (strike layer) of gold is deposited on the substrate. Since a rigid stamp is used in this process, external pressures must be applied to achieve conformal contact between the two gold layers. When the stamp and the substrate are brought into contact at high pressures (~150 MPa), cold welding occurs between gold on the raised regions of the stamp and the gold strike layer on the substrate at room temperature. After transferring the gold patterns from the stamp to the substrate, the exposed gold strike layer is removed from the substrate by sputter etching in an argon environment. Kim and coworkers [66] fabricated top-contact pentacene TFTs with gold electrodes patterned by cold welding. To achieve top-contact devices, a gold strike layer was deposited directly on top of pentacene. Cold welding defines the source and drain electrodes; channels with widths of 97 μm and lengths as small as 1 μm were fabricated [66]. Figure 5.5.26 shows the output characteristics of a top-contact pentacene TFT with the gold electrodes formed by cold welding. Although the OTFT is a top-contact device, the output characteristics exhibited significant nonlinear behavior in the linear regime. Kim and coworkers attributed such nonlinear output characteristics to large contact resistance at the gold-pentacene interface [66]. Further, the electrical characteristics of this top-contact pentacene TFT were lower than expected. There are two possible explanations for this observation. First, the etching process to remove the gold strike layer from the channel may damage pentacene, thereby degrading its performance. Second, the high contact pressures required for cold welding (~150 MPa) may also damage pentacene. The bottom inset of Figure 5.5.26 reveals evidence of wrinkles in pentacene after cold welding. The same authors recently demonstrated cold welding at reduced pressures (~180 kPa) using flexible PDMS stamps supported on a glass backplane, rather than rigid stamps [98]. In this example, Kim and coworkers patterned the cathodes for OLEDs by cold welding. This process begins with the deposition of an organic light-emitting material followed by an electron-transporting layer on an ITO-coated glass substrate (anode). The cathode layers and a gold strike layer (i.e., LiF/Al/Au) are then deposited on the electron-transporting layer.

8080_C005.fm Page 467 Friday, April 6, 2007 10:25 AM

Soft Lithography for Fabricating Organic Thin-Film Transistors

Step 1

467

Stamp

Achesion-reduction layer Strike layer

Metal Substrate

Step 2 Cold welding Metal transfer

Step 3 Strike layer removal

Step 4 Pattern replication in a substrate

FIGURE 5.5.25 Schematic of cold welding. A rigid silicon stamp is coated with an adhesion reduction layer, such as pentacene, followed by a metal layer. A thin metal strike layer is also deposited onto the substrate. The stamp is then brought into contact with the substrate. In the regions of contact, the metal is transferred to the strike layer via cold welding. After separating the stamp from the substrate, the exposed strike layer is removed by subsequent etching. (From C. Kim et al., Appl. Phys. Lett., 80, 4051, 2002.)

Separately, a PDMS stamp is prepared on a glass support and is coated with an adhesion reduction layer and a gold layer. The gold layer on the PDMS stamp is then cold welded to the cathodes with the organic light-emitting underlayer. Here, the conformal contact of the PDMS stamp allows the gold layer to be transferred to the strike layer at reduced pressures, thereby preventing damage to the organic light-emitting layer. The quantum efficiency of the OLEDs fabricated by cold welding is comparable with that of OLEDs fabricated by conventional pattern definition by evaporation through shadow masks, thus implying that stamping does not degrade device performance [98]. Subsequently, Loo and workers demonstrated cold welding at ambient pressures [21]. Specifically, they fabricated a metal–insulator–metal capacitor (Au–SiNx–Au) by cold welding. An Au/SiNx/Au multilayer stack was deposited on a silicon stamp and directly printed on an Au-coated PDMS/PET substrate. The details of this device were discussed previously in Section 5.5.4.

8080_C005.fm Page 468 Friday, April 6, 2007 10:25 AM

468

Organic Field-Effect Transistors

20.0 VGS = –30 V S

ID(μA)

15.0

D

10.0

–25 V

5.0

–20 V –15 V

0.0

3

0

–3

–6 VDS(V)

–9

–12

–15

FIGURE 5.5.26 Output characteristics of a top-contact pentacene TFT with gold electrodes formed by cold welding. This TFT has a channel length of 1 μm as shown in the top inset. Although the TFT adopts the top-contact device geometry, the output characteristics exhibit significant nonlinear behavior at small source-drain voltages. The bottom inset reveals wrinkles in the pentacene layer after etching to remove the underlying Au strike layer. (From C. Kim et al., Appl. Phys. Lett., 80, 4051, 2002.)

5.5.7 METAL TRANSFER PRINTING Similar to nTP and cold welding, metal transfer printing [99] allows top-contact OTFTs to be fabricated. In metal transfer printing, the metal features are directly printed on semiconducting and insulating polymers. Wang and coworkers [99] demonstrated the patterning of gold and aluminum on a variety of polymers, including poly(methyl methacrylate) (PMMA), polystyrene (PS), and P3HT. The polymer, heated above its glass transition temperature, acts as an adhesion promoter, eliminating the need for the metal strike layer in cold welding. Specifically, a thin layer of metal is deposited on the raised and recessed regions of a PDMS stamp. The PDMS stamp is contacted against a heated polymer semiconductor (>120°C), and slight pressure (~2.9 kPa) is applied. After cooling the polymer semiconductor below its glass transition temperature, removing the PDMS stamp transfers the metal patterns onto the polymer. Using metal transfer printing, Wang and coworkers [99] fabricated top-contact P3HT TFTs. P3HT was spin-coated on a substrate containing the gate electrodes coated with a Ta2O5 film, which served as the gate dielectric. Gold electrodes were directly patterned on top of P3HT by metal transfer printing. These OTFTs had a channel width of ~400 μm and a channel length of ~20 μm and exhibited a chargecarrier mobility of 0.017 cm2/Vs, which was comparable to that of P3HT TFTs with photolithographically defined electrodes [99]. While this technique is useful for fabricating top-contact OTFTs with polymer semiconductors, it will not work for small-molecule organic semiconductors since they do not exhibit a softening temperature prior to degradation.

8080_C005.fm Page 469 Friday, April 6, 2007 10:25 AM

Soft Lithography for Fabricating Organic Thin-Film Transistors

469

5.5.8 HOT LIFT-OFF Current techniques for patterning organic semiconductors mainly rely on thermal evaporation through a shadow mask. Consequently, the feature size of organic semiconductors is limited by the resolution of the shadow mask (25–30 μm). Further reduction in the feature size can be achieved by “hot lift-off” (illustrated in Figure 5.5.27a), a technique developed by Wang and coworkers [100]. Using hot lift-off, they fabricated an array of 7- by 7-μm organic semiconductor squares. In this technique, the organic semiconductor is deposited on a substrate by conventional evaporation methods. A partially cured epoxy stamp is then contacted against the organic semiconductor with an applied pressure of 981 kPa to induce local fracture of the organic semiconductor along the patterned edges of the stamp. The applied pressure is then reduced to 196 kPa while heating (80–120°C, 20 min) to maintain conformal contact between the stamp and organic semiconductor while the stamp is further cured. After cooling to room temperature, the epoxy stamp is peeled away from the substrate. In the regions of contact, the organic semiconductor peels with the epoxy stamp, leaving behind patterned organic semiconductor in the noncontact regions on the substrate (Figure 5.5.27). Such lift-off occurs because the work of adhesion between the epoxy and the organic semiconductor is greater than the work of adhesion between the substrate and the organic semiconductor. Although this technique is subtractive in nature, it avoids exposing the remaining organic semiconductor to harsh chemical environments, such as the etching solutions used in photolithography and μCP. A variety of organic semiconductors, including copper phthalocyanine (CuPc), metal-free phthalocyanine (H2-Pc), N,N′-di(naphthalene-1-yl)-N,N′-diphenylbenzidien (NPB), and tri(8-quinolinolato)aluminum (AlQ3) were patterned by hot lift-off. Additionally, Wang and coworkers [100] fabricated functional top-contact CuPc TFTs by hot lift-off and evaporation of the gold electrodes through a shadow mask. These OTFTs exhibited a charge-carrier mobility of 0.02 cm2/Vs, a threshold voltage of –5 V, and an on/off current ratio of 104–105. Although this OTFT adopts the topcontact geometry, the device performance is not better than that of bottom-contact CuPC TFTs in which the CuPC is unpatterned [101].

5.5.9 MICROMOLDING IN CAPILLARIES (MIMIC) Micromolding in capillaries (MIMIC) [102,103] differs from the techniques discussed previously because it is not a contact printing technique. Rather, MIMIC is a molding technique that relies on the spontaneous filling of the microchannels of the mold with a fluid, which can be a solution, a suspension, or precursors of the material to be patterned. This technique is therefore useful for patterning thick polymer, ceramic, and metal features. In MIMIC, a PDMS mold is brought into contact with a rigid support, forming microchannels between the substrate and the recessed regions of the PDMS mold as shown in Figure 5.5.28. The fluid, placed at the end of mold, fills the channels of the mold by capillary action. The material is then cross-linked, crystallized, or

8080_C005.fm Page 470 Friday, April 6, 2007 10:25 AM

470

Organic Field-Effect Transistors

(a)

Epoxy stamp

Small molecule

Substrate

Microcrystalline ﬁlm

Press and heat (b)

Epoxy stamp Substrate Selective lift oﬀ Epoxy stamp

Substrate (c)

(d)

20 μm

100 μm (e) 150 nm

120.00

(f )ΔZ [nm] Distance [nm] f[°] 74.05106 –432.9674 9.70548

[nm] 0

0 5

5 10

10 15 15 [μm]

–5.00 0

[nm]

1505.077

FIGURE 5.5.27 (a) Schematic of hot lift-off. A partially cured epoxy stamp is placed on an organic semiconductor film. Heating the stamp while applying pressure creates conformal contact between the stamp and the semiconductor. After cooling, peeling the stamp lifts the organic semiconductor underlayer. A variety of organic semiconductors are patterned using hot lift-off. (b) CuPC on Si. (c) Alq3/NPB on ITO. (d) H2PC on Si. (e, f) The cross-sectional profiles of the patterns obtained by AFM reveal sharp edges of the patterns. (From Z. Wang et al., J. Am. Chem. Soc., 125, 15278, 2003.)

8080_C005.fm Page 471 Friday, April 6, 2007 10:25 AM

Soft Lithography for Fabricating Organic Thin-Film Transistors

471

PDMS master

Cut oﬀ ends

Place on a support

Place a drop of prepolymer at one end

Channels ﬁll by capillary action

Cure; remove PDMS Polymer

FIGURE 5.5.28 Schematic of MIMIC. At least one end of the PDMS stamp is cut off to create entrances to the microchannels. The PDMS mold is then placed on a support to form the microchannels. When a precursor or a prepolymer solution is placed at one end, capillary action spontaneously fills the channels with the solution. After curing, the PDMS stamp is removed, leaving behind solid polymer features on the substrate. (From E. Kim et al., J. Am. Chem. Soc., 118, 5722, 1996.)

8080_C005.fm Page 472 Friday, April 6, 2007 10:25 AM

472

Organic Field-Effect Transistors

cured before the PDMS mold is removed. The patterned material is therefore a negative image of the PDMS mold. Rogers and coworkers [104] demonstrated the utility of MIMIC for fabricating electrodes for top-contact OTFTs. Specifically, conductive carbon or conducting polymer (PANI doped with m-cresol) electrodes, as wide as 100 μm and separated by 25 μm, were patterned on top of P3HT organic semiconductor by MIMIC. In these devices, the P3HT and gate dielectric (polyimide) layers were prepatterned using screen printing [103]. The source and drain electrodes were defined by the recessed regions (microchannels) of a PDMS mold. To access the microchannels, vertical holes were created at both ends of the PDMS mold. After placing the mold on top of the P3HT layer, a conductive ink (conductive carbon or a conductive polymer, such as PANI) was injected into the access holes of the mold. The ink filled the microchannels of the PDMS mold by capillary forces. After solvent evaporation, the PDMS mold was removed, leaving behind solid conductive carbon or polymer source and drain electrodes. Top-contact P3HT TFTs with conductive carbon source and drain electrodes exhibited a charge-carrier mobility of 0.01–0.05 cm2/Vs, which is comparable with the charge-carrier mobility of bottom-contact P3HT OTFTs with photolithographically-defined gold electrodes [104]. A variation of MIMIC for patterning polymer semiconductor TFT arrays was recently reported by Salleo and coworkers [105]. This technique exploits capillary forces to pattern a solution-processable polymer semiconductor. In this method, the polymer semiconductor solution is spin-coated onto a substrate. A chemically treated PDMS stamp is then placed directly on top of the substrate coated with the polymer semiconductor solution. In the regions of contact, the PDMS stamp absorbs the solvent, leaving behind a solid polymer semiconductor film between the stamp and the substrate. In the recessed regions of the PDMS stamp, the polymer semiconductor solution wicks into the stamp due to capillary forces, effectively leaving behind a clean surface in the noncontact regions. Using this stamping process, patterned arrays of poly[(9,9′-dioctylfluorine)-cobithiophene] (F8T2) and poly[5,5′-bis(3-alkyl-2-thienyl)-2,2′-bithiophene] (PQT12) with features as small as 2 μm were fabricated. Figure 5.5.29(a) illustrates the stamp fabrication and patterning process. Optical micrographs of patterned polymer semiconducting films and a used PDMS stamp after patterning are shown in Figures 5.5.29(b) and (c), respectively. Both top- and bottom-contact TFTs were fabricated with organic semiconductor arrays patterned by this technique. To fabricate topcontact TFTs, gold electrodes are evaporated through a shadow mask onto an F8T2 organic semiconductor array patterned on thermally-grown silicon dioxide/silicon substrate. These top-contact F8T2 TFTs exhibited charge-carrier mobilities of 0.5–1 × 10–3cm2/Vs, and a threshold voltage of ~1 V, similar to those obtained from analogous top-contact TFTs fabricated by spin-coating the F8T2 layer [106]. For bottom-contact TFTs, gate electrodes (Cr/Au), gate dielectric (SiNx/SiO2 bilayer), and source and drain electrodes (Cr/Au) were defined by digital lithography [107]. F8T2 or PQT-12 organic semiconductor arrays were then patterned. While the bottom-contact F8T2 TFTs showed a charge-carrier mobility of 3 × 10–4cm2/Vs and a threshold voltage of –10 V, the PQT-12 TFTs had better device characteristics with a charge-carrier mobility of 3 × 10–3cm2/Vs and a threshold voltage of 10 V.

8080_C005.fm Page 473 Friday, April 6, 2007 10:25 AM

Soft Lithography for Fabricating Organic Thin-Film Transistors

473

F8T2 Substrate Jet-print wax pattern Mold PDMS stamp

300 μm 120 μm

Release PDMS stamp O2 plasma

600 μm

Activate stamp surface

1 mm

Benzyl-TS Treat stamp surface Recess

(b)

Raised feature Coat substrate with solution place stamp

Solvent absorption

Raised portion

Capillary action

Stamp recess

Allow solution to wick in recesses and solvent to permeate stamp

Patterned polymer

Remove stamp

(c)

(a)

FIGURE 5.5.29 (a) Schematic of the stamp fabrication and patterning process of F8T2 and PQT-12. (b) Optical micrographs of patterned F8T2 arrays on thermally grown SiO2 on a Si substrate. (c) Optical micrograph of a used stamp reveals the polymer residue in the recess regions of the stamp. (From A. Salleo et al., Adv. Funct. Mater., 15, 1105, 2005.)

These charge-carrier mobilities, however, are two orders of magnitude lower than those achieved by spin-coating or jet-printing the organic semiconductor onto a gate dielectric treated with a hydrophobic molecular layer (0.02 cm2/Vs for F8T2 and 0.1 cm2/Vs for PQT-12). Salleo and coworkers attributed the reduction in chargecarrier mobilities to the fact that these TFTs were fabricated on untreated dielectric surfaces. The observed charge-carrier mobilities are comparable to those obtained by other processing methods on untreated dielectric surfaces.

5.5.10 SOFT-CONTACT OPTICAL LITHOGRAPHY As demonstrated throughout this chapter, the unique mechanical properties of PDMS have enabled many soft lithographic techniques. Soft-contact optical lithography developed by Lee and coworkers [108] is a variation of photolithography for patterning both rigid and flexible substrates at a low cost by replacing expensive photomasks with PDMS “masks.” Since the PDMS mask is flexible, it can make conformal contact against flexible or curved surfaces. Soft-contact lithography therefore allows metals to be patterned on flexible or nonflat substrates via traditional photolithography. A schematic of soft-contact lithography is shown in Figure 5.5.30.

8080_C005.fm Page 474 Friday, April 6, 2007 10:25 AM

474

Organic Field-Effect Transistors

PDMS Oxygen plasma Ti/Au Deposition Ti/Au

PDMS

Spin cast photoresist Photoresist Ti/Au

PDMS

No prebaking hv PDMS

PDMS I: Lower dose II: Higher dose Developing PDMS

PDMS

Au etching (Kl sulution)

PDMS

PDMS

FIGURE 5.5.30 Schematic of soft-contact optical lithography. Ti (1 nm)/Au (20 nm) is uniformly evaporated on a flat PDMS substrate and coated with photoresist. A patterned PDMS mask, which is transparent to UV light, is laminated against the photoresist layer. The backside of the PDMS mask is illuminated with UV light. The exposed PR film is developed. The resulting pattern is transferred to the Ti/Au layer by etching. Controlling the exposure dose results in two distinct line-width patterns. Lower exposure doses replicate the PDMS relief pattern; higher exposure doses generate line widths from 50 to 150 nm by phase-shifts in the optical near field. (From T.-W. Lee et al., Adv. Funct. Mater., 15, 1435, 2005.)

Specifically, photoresist is spin-coated on a gold-coated PDMS substrate. After establishing conformal contact between the PDMS mask and the photoresist/gold/PDMS substrate, the assembly is exposed to 330 nm UV light. By controlling the exposure dose and developing conditions, a single PDMS mask is capable of generating two different patterns in the resist. At low-exposure doses, the regions of photoresist in contact with the PDMS mask become soluble and can be removed during developing, leaving behind a pattern that replicates the recessed regions of the PDMS mask. Higher exposure doses or longer developing times generate narrower lines (as small as 150 nm) positioned at the contact edges of the features on the mask. At these higher exposure doses, the PDMS mask acts like a near-field phase-shift mask,

8080_C005.fm Page 475 Friday, April 6, 2007 10:25 AM

Soft Lithography for Fabricating Organic Thin-Film Transistors

475

allowing nanoscale features such as periodic and curved lines, open rings, and posts on planar and curved surfaces to be patterned [22,108–110]. Combining soft-contact optical lithography with ScL enables the fabrication of laminated OLEDs with nanometer-scale electrodes [96,108].

5.5.11 LASER THERMAL TRANSFER PRINTING By definition, laser thermal transfer printing is not a soft lithography technique. It is, however, a useful technique for patterning organic components of OTFTs. Laser thermal transfer printing is a solventless thermal imaging technique [111]. One of the most attractive features of this technique is that it circumvents solvent incompatibility issues frequently found in inkjet printing or screen printing. This technique also provides rapid printing of micron-sized organic material features over large areas. Blanchet and coworkers [111] used laser thermal transfer printing for patterning conducting polymer electrodes on flexible substrates for pentacene TFTs. Figure 5.5.31 illustrates this laser thermal transfer printing process. This method uses two flexible substrates. The multilayer donor substrate consists of uniformly coated thin layers of conducting polymer and metal. The other substrate is the receiver substrate on which patterned features of organic material will be transferred. These two flexible substrates are loaded into the near infrared laser printer and held together by vacuum. Focusing the laser beam on the thin absorbing metal layer in the donor substrate converts light into heat in this layer. The generated heat decomposes the conducting polymer at the metal-conducting polymer interface into gaseous by-products. The expansion of gaseous products transfers the top layer of the conducting polymer onto the receiver layer. The transfer of organic material occurs pixelwise. Gate, as well as source and drain, electrodes are patterned from a PANI/single wall carbon nanotubes (SWNT)-coated donor substrate using this technique. Laser thermal transfer printing allows both bottom- and top-contact devices to be fabricated. Bottom-contact pentacene TFTs with printed PANI/SWNT electrodes on glass resin/ITO/Mylar substrates had channel widths of 750 μm and channel lengths ranging from 10 to 250 μm. The charge-carrier mobility of these OTFTs was 0.3 cm2/Vs, which is twice as high as that of reference OTFTs with gold electrodes (0.15 cm2/Vs), at similar channel dimensions. Interestingly, pentacene TFTs with top-contact printed PANI/SWNT electrodes exhibited nonlinear output characteristics. Analogous to the top-contact devices fabricated by cold welding (where the pentacene layer is wrinkled in Figure 5.5.26), it is speculated that pentacene may have been thermally damaged or degraded during the PANI/SWNT transfer process.

5.5.12 IMPRINT LITHOGRAPHY While imprint lithography does not involve the use of an elastomeric stamp and hence is not a soft lithography technique, it is a next-generation patterning technique that has gained tremendous attention for generating nanoscale features, which is

8080_C005.fm Page 476 Friday, April 6, 2007 10:25 AM

476

Organic Field-Effect Transistors

FIGURE 5.5.31 Illustration of laser thermal transfer printing. Two flexible substrates are held together by vacuum. One film is the donor substrate, which consists of thin layers of conducting polymer and metal uniformly coated on a substrate. The other film is the receiver onto which patterns of conducting polymer are transferred. Focusing the laser beam onto the thin absorbing metal layer generates heat at the polymer conductor interface. The generated heat decomposes the conducting polymer into gaseous by-products, which then push the conducting polymer onto the receiver layer. (From G. B. Blanchet et al., Appl. Phys. Lett., 82, 463, 2003.)

why we feel that it is important that this patterning technique be discussed. Imprint lithography provides a higher resolution and lower cost alternative for fabricating nanoscale electrical circuitry components as compared to photolithography [112]. The focus of this section is thus to provide an introduction to the technology addressing the advancements and concerns associated with this next-generation patterning technology. For a more detailed discussion on the state of imprint lithography and its prognosis for future application in integrated circuit fabrication, please refer to a recent topical review written by Guo [113]. The two leading imprint lithography techniques are nanoimprint lithography (NIL) [112,114] and step-and-flash imprint lithography (S-FIL) [115,116]. Both techniques rely on a two-step process of imprinting and pattern transfer. A schematic of the NIL process is shown in Figure 5.5.32 [114]. In the NIL process, the resist is a thermoplastic [112,117] or thermal curable polymer [118]. During NIL, the thermoplastic is heated above its glass transition temperature, allowing it to flow, before imprinting with the mold. This mold is typically fabricated on a silicon or

8080_C005.fm Page 477 Friday, April 6, 2007 10:25 AM

Soft Lithography for Fabricating Organic Thin-Film Transistors

477

Mold

1. Press mold

Resist Substrate

2. Remove mold

3. RIE

FIGURE 5.5.32 Schematic of nanoimprint lithography (NIL). (1) A rigid mold is pressed into a thermoplastic resist, heated above its glass transition temperature, to create an imprint. (2) The mold is removed, leaving behind the patterned resist. (3) An anisotropic etch removes the residual resist and transfers the pattern into the substrate. (From S. Y. Chou et al., J. Vac. Sci. Technol., B, 14, 4129, 1996.)

fused silica template using e-beam lithography. Upon mold contact, the resist is cooled below its glass transition temperature, effectively “locking in” the shape of the mold, which is subsequently removed. It is important to note that NIL is a physical deformation process, not an ink-stamping process. Consequently, the diffusion concerns associated with soft lithography techniques, such as μCP, do not exist. Further, the imprint mold is rigid, so feature collapse and deformation during imprinting are eliminated. In combination, these attributes of NIL have enabled the patterning of extremely small (below 10 nm) yet high-resolution features, as illustrated in Figure 5.5.33 [119–121]. 6 nm Half-pitch

8.5 nm Half-pitch

17 nm Half-pitch

FIGURE 5.5.33 Scanning electron micrograph of polymer resist with 6-, 8.5-, and 17-nm halfpitch gratings fabricated by NIL. (From M. D. Austin et al., Nanotechnology, 16, 1058, 2005.)

8080_C005.fm Page 478 Friday, April 6, 2007 10:25 AM

478

Organic Field-Effect Transistors

While NIL is a promising next-generation lithography technique with excellent feature resolution, several concerns still need to be addressed before the technique will have industrial implications. In particular, commercially-available molds, resists, and processes still need to be developed [113]. Since the mold plays a crucial role in imprint lithography, procedures for preventing and removing mold contamination must be addressed. Also, resist materials with suitable physical properties [113], such as low viscosity and low glass transition temperatures, must be exploited to minimize the pressure [122] and the time for heating and cooling cycles required for polymer molding — potential bottlenecks in the manufacturing process. Additionally, thermal expansion effects of the resist and the mold must be considered when heating and cooling the thermopolymer to minimize feature distortion and alignment difficulties. This being said, NIL remains a powerful patterning tool. Austin and coworkers demonstrated the usefulness of NIL for fabricating bottom-contact OTFTs with channel lengths below 100 nm [123]. Figure 5.5.34 illustrates their procedure. A negative imprint of the transistor channel region is defined on a silicon platform by NIL. An oxygen plasma etch removes residual polymer from source and drain regions. Subsequently, gold is uniformly deposited across the substrate, creating source and drain electrodes, and the polymer imprint is removed by lift-off. Deposition of P3HT polymer semiconductor completes the transistor. A protective silicon dioxide cap is evaporated onto the P3HT to protect it from moisture and oxygen. Typical current-voltage characteristics are shown in Figure 5.5.35 for transistors with channel widths of 2.5 μm and channel lengths ranging from 70 to 1,000 nm. Short-channel effects are clearly visible as the channel length is decreased. The second imprint lithography technique discussed here is S-FIL. Similar to NIL, S-FIL can be a potentially high-throughput, low-cost technique for fabricating nanometer-scale features. A schematic of the S-FIL process is shown in Figure 5.5.36 [124]. A photocurable liquid (i.e., resist) is dispensed into the gap between the transparent (to UV) template (analogous to the mold used in NIL) and the substrate. Typically, the template is fabricated on a fused silica plate using conventional phaseshift reticle processing [125,126]. Willson and coworkers, however, have developed other mask fabrication techniques that incorporate a transparent indium tin oxide layer to improve imaging of the mask during defect analysis [127–130]. When the liquid resist is dispensed between the template and the substrate, capillary forces cause the liquid to disperse into the gaps and pull the template tight against the substrate. Illumination of the mold and resist with UV radiation initiates polymerization, creating a replica of the mold. Removal of the template completes the imprint process. The pattern is then transferred into the substrate through a series of etching steps. The key difference between NIL and S-FIL is the type of the resist used for pattern transfer. In S-FIL, a low-viscosity, photocurable organosilane liquid is used with a transparent template. Due to the low viscosity of the resist, external pressure is not required for imprinting. Further, the resist photocures at room temperature, thereby eliminating the heating and cooling cycles required in NIL. The rigid, transparent S-FIL template permits flood exposure of the photocurable resist and layer-to-layer alignment through classical optical techniques [131]. These attributes

8080_C005.fm Page 479 Friday, April 6, 2007 10:25 AM

Soft Lithography for Fabricating Organic Thin-Film Transistors

NIL Mold

479

(a) Nanolmprint: Imprint polymer 5 nm Gate oxide n+ Silicon Substrate (back-side gate)

NIL Mold (b) Mold Separation

(c) Oxygen Plasma Etch Residue Imprint Polymer 70 nm Channel Length (d) Gold Source / Drain Evaporation and Lift-oﬀ

(e) Add: Gold Pads Protective SiOx (f ) Add Semiconducting Polymer P3HT: R = C6H13 S Source

m

Drain (g) Protect and Isolate the Device with SiOx cap

Gate

FIGURE 5.5.34 NIL for fabricating organic thin-film transistors. (a) A negative imprint of the transistor channel is created by pressing and (b) removing the mold. (c) Residual polymer is removed with an oxygen plasma etch. (d) Gold source and drain electrodes are evaporated. Lift-off is performed to remove the polymer imprint. (e) Additional gold pads are added for making output characteristic measurements. A layer of SiOx is also added to protect the gate oxide from probe damage during output characteristic measurements. (f) The polymer semiconducting layer, P3HT, is deposited to complete the transistor. (g) A SiOx layer is deposited over the P3HT layer to protect it from moisture and oxygen. (From M. D. Austin and S. Y. Chou, Appl. Phys. Lett., 81, 4431, 2002.)

8080_C005.fm Page 480 Friday, April 6, 2007 10:25 AM

480

Organic Field-Effect Transistors

Drain current density [nA/um]

–2

Drain current density [nA/um]

(a) 1000 nm Device –3

101

Vd = –1.0V

Vg = –3V

Vd = –1.5V

10–5 –4 –2

0

2

Gate voltage Vg = –2V

–1 Vg = –1V 0

0

–1

–2

–3

Drain voltage

Drain current density [nA/um]

–4

Drain current density [nA/um]

(b) 200 nm Device –6

10

1

Vd = –1.0V Vd = –0.5V

Vg = –3V 10–5 –4

–2

0

2

Gate voltage

Vg = –2V

–2 Vg = –1V

0

0

–1

–2

–3

Drain voltage (b) 70 nm Device Drain current density [nA/um]

Drain current density [nA/um]

–30

–20

Vd = –1.0V

0

10–5 –4

–2V –2

0

2

–1

–1V

–2 Drain voltage

FIGURE 5.5.35

Vg = –3V

Vd = –0.5V

Gate voltage

–10

0

101

–3

8080_C005.fm Page 481 Friday, April 6, 2007 10:25 AM

Soft Lithography for Fabricating Organic Thin-Film Transistors

481

Release layer Etch barrier Base layer (a)

(b)

UV

(c) Base layer

(d)

(e)

FIGURE 5.5.36 Schematic of step-and-flash imprint lithography (S-FIL). (a) A photocurable liquid (etch barrier) is dispersed between a transparent template and substrate. (b) The gap between the template and the substrate is closed to create a thin resist layer. (c) The backside of the template is illuminated with UV light to cure the resist. (d) The template is removed, leaving behind the imprinted features. (e) An anisotropic etch removes the base layer and transfers the pattern into the substrate. (From T. Bailey et al., J. Vac. Sci. Technol., B, 18, 3572, 2000.)

FIGURE 5.5.35 (See figure, facing page.) Current-voltage characteristics of P3HT TFTs with a channel width of 2.5 μm and channel lengths of (a) 1,000 nm; (b) 200 nm; and (c) 70 nm. Insets: transfer characteristics of the respective devices. (From M. D. Austin and S. Y. Chou, Appl. Phys. Lett., 81, 4431, 2002.)

8080_C005.fm Page 482 Friday, April 6, 2007 10:25 AM

482

Organic Field-Effect Transistors

(a)

(b)

40 μm

FIGURE 5.5.37 Optical micrographs of S-FIL template (a) before and (b) after two imprints. Template contamination is removed during imprinting. (From T. Bailey et al., J. Vac. Sci. Technol., B, 19, 2806, 2001.)

allow an entire wafer to be patterned in a “step-and-flash” method similar to conventional photolithography. The resolution of S-FIL is limited by the resolution of the features on the template. To date, features as small as 30 nm [132] and with aspect ratios of 14:1 [133,134] have been demonstrated with S-FIL. Additionally, S-FIL is capable of pattering surfaces with topography [133]. Similar to NIL, in order for S-FIL to become commercially successful, commercially available templates and processes still need to be developed. Again, the quality of the template is critical for successful imprinting. To ensure that the photocured polymer releases cleanly from the template, the surface of the template is treated with an FSAM to create a nonstick surface [124]. Defect susceptibility studies to quantify the effectiveness of the FSAM treatment indicate that S-FIL is actually self-cleaning [124,135], as depicted in Figure 5.5.37. The contamination from the template is entrained in the polymer resist during photopolymerization, resulting in a template that is visually clean after imprinting. More recently, a new strategy was developed by Kim and coworkers [136] to promote polymer release from the template. Specifically, a fluorinated acetate is added to the resist formulation. During polymerization, the additive migrates to the template/resist interface, reducing the separation force by ~0.9 lbf as compared to the same formulation without the fluorinated acetate. While the photocurable polymers used for S-FIL avoid the viscosity and thermal concerns associated with the thermoplastics used in NIL, they are not without their problems. During polymerization, the photocurable polymers undergo a volumetric shrinkage or densification due to chemical bond formation. Consequently, the feature size, shape, and position could be affected. Studies have shown, however, that by controlling the resist composition, the shrinkage is limited to the z-direction, resulting in a reduced aspect ratio [131,137]. While NIL and S-FIL have been shown to be effective tools for creating nanoscale patterns, it is important to keep in mind that the patterned polymer layers utilized in both techniques are sacrificial structures. Additional etchback steps are required to transfer the patterns into the substrate. Further, metal contacts and other functional materials have to be deposited separately to create functional devices.

8080_C005.fm Page 483 Friday, April 6, 2007 10:25 AM

Soft Lithography for Fabricating Organic Thin-Film Transistors

483

REFERENCES 1. C. D. Dimitrakopoulos and P. R. L. Malenfant, Organic thin film transistors for large area electronics, Adv. Mater., 14, 99, 2002. 2. C. Reese et al., Organic thin film transistors, Mater. Today, 7, 20, 2004. 3. I. Kymissis, C. D. Dimitrakopoulos, and S. Purushothaman, High-performance bottom electrode organic thin-film transistors, IEEE Trans. Electron Devices, 48, 1060, 2001. 4. P. V. Necliudov et al., Contact resistance extraction in pentacene thin film transistors, Solid-State Electron., 47, 259, 2002. 5. Y. Roichman and N. Tessler, Structures of polymer field-effect transistor: Experimental and numerical analyses, Appl. Phys. Lett., 80, 151, 2002. 6. K. P. Puntambekar, P. V. Pesavento, and C. D. Frisbie, Surface potential profiling and contact resistance measurements on operating pentacene thin-film transistors by Kelvin probe force microscopy, Appl. Phys. Lett., 83, 5539, 2003. 7. M. M. Ling and Z. Bao, Thin film deposition, patterning, and printing in organic thin film transistors, Chem. Mater., 16, 4824, 2004. 8. ITRS, International Technology Roadmap for Semiconductors, 2006 update at http://public.itrs.net. 9. D. J. Gundlach et al., Solvent-induced phase transition in thermally evaporated pentacene films, Appl. Phys. Lett., 74, 3302, 1999. 10. M. Halik et al., Fully patterned all-organic thin film transistors, Appl. Phys. Lett., 81, 289, 2002. 11. M. Halik et al., Polymer gate dielectric and conducting-polymer contacts for highperformance organic thin-film transistors, Adv. Mater., 14, 1717, 2002. 12. H. Sirringhaus, N. Tessler, and R. H. Friend, Integrated, high-mobility polymer fieldeffect transistors driving polymer light-emitting diodes, Synth. Met., 102, 857, 1999. 13. C. R. Kagan and P. Andry, Thin-film transistors, Marcel Dekker, Inc., New York, Basel, 2003. 14. C. D. Sheraw et al., Organic thin-film transistor-driven polymer-dispersed liquid crystal displays on flexible polymeric substrates, Appl. Phys. Lett., 80, 1088, 2002. 15. Y. Xia and G. M. Whitesides, Soft lithography, Annu. Rev. Mater. Sci., 28, 153, 1998. 16. www.dowcorning.com. 17. Y. Xia and G. M. Whitesides, Soft lithography, Angew. Chem., Int. Ed., 37, 550, 1998. 18. A. Bietsch and B. Michel, Conformal contact and pattern stability of stamps used for soft lithography, J. Appl. Phys., 88, 4310, 2000. 19. E. Delamarche et al., Stability of molded polydimethylsiloxane microstructures, Adv. Mater., 9, 741, 1997. 20. Y.-L. Loo et al., Additive, nanoscale patterning of metal films with a stamp and a surface chemistry mediated transfer process: Applications in plastic electronics, Appl. Phys. Lett., 81, 562, 2002. 21. Y.-L. Loo et al., Interfacial chemistries for nanoscale transfer printing, J. Am. Chem. Soc., 124, 7654, 2002. 22. T. W. Odom et al., Improved pattern transfer in soft lithography using composite stamps, Langmuir, 18, 5314, 2002. 23. H. Schmid and B. Michel, Siloxane polymers for high-resolution, high-accuracy soft lithography, Macromolecules, 33, 3042, 2000. 24. M. Campbell et al., Fabrication of photonic crystals for the visible spectrum by holographic lithography, Nature, 404, 53, 2000.

8080_C005.fm Page 484 Friday, April 6, 2007 10:25 AM

484

Organic Field-Effect Transistors

25. C. D. Wood et al., New fluoropolymer materials, J. Fluorine Chem., 125, 1671, 2004. 26. J. P. Rolland, et al., Solvent-resistant photocurable liquid fluoropolymers for microfluidic device fabrication, J. Am. Chem. Soc., 126, 2322, 2004. 27. W. C. Bunyard, T. J. Romack, and J. M. DeSimone, Perfluoropolyether synthesis in liquid carbon dioxide by hexafluoropropylene photooxidation, Macromolecules, 32, 8224, 1999. 28. J. P. Rolland et al., Perfluoropolyethers as novel materials for soft lithography, Proc. SPIE, 5751, 410–414, 2005. 29. A. Kumar and G. M. Whitesides, Features of gold having micrometer to centimeter dimensions can be formed through a combination of stamping with an elastomeric stamp and an alkanethiol “ink” followed by chemical etching, Appl. Phys. Lett., 63, 2002, 1993. 30. Y. Xia et al., Microcontact printing of octadecylsiloxane on the surface of silicon dioxide and its application in microfabrication, J. Am. Chem. Soc., 117, 9576, 1995. 31. T. W. Odom et al., Generation of 30–50 nm structures using easily fabricated, composite PDMS masks, J. Am. Chem. Soc., 124, 12112, 2002. 32. T. W. Odom et al., Improved pattern transfer in soft lithography using composite stamps, Langmuir, 18, 5314, 2002. 33. Y. Xia, E. Kim, and G. M. Whitesides, Microcontact printing of alkanethiols on silver and its application in microfabrication, J. Electrochem. Soc., 143, 1070, 1996. 34. Y. Xia et al., Microcontact printing of alkanethiols on copper and its application in microfabrication, Chem. Mater., 8, 601, 1996. 35. J. C. Love et al., Self-assembled monolayers of alkanethilates on palladium are good etch resists, J. Am. Chem. Soc., 124, 1576, 2002. 36. P. C. Hidber et al., Microcontact printing of palladium colloids: Micron-scale patterning by electroless deposition of copper, Langmuir, 12, 1375, 1996. 37. H. Kind et al., Patterned electroless deposition of copper by microcontact printing palladium(II) complexes on titanium-covered surfaces, Langmuir, 16, 6367, 2000. 38. Z. Huang et al., Selective deposition of films of polypyrrole, polyaniline, and nickel on hydrophobic/hydrophilic patterned surfaces and application, Synth. Met., 85, 1375, 1997. 39. U. Zschieschang et al., Flexible organic circuits with printed gate electrodes, Adv. Mater., 15, 1147, 2003. 40. Z. Huang et al., Selective deposition of conducting polymers on hydroxyl-terminated surfaces with printed monolayers of alkylsiloxanes as templates, Langmuir, 13, 6480, 1997. 41. C. B. Goreman, H. A. Biebuyck, and G. M. Whitesides, Fabrication of patterned, electrically conducting polypryrrole using a self-assembled monolayer: A route to all-organic circuits, Chem. Mater., 7, 526, 1995. 42. D. M. Collard and C. N. Sayre, Micron-scale patterning of conjugated polymers on microcontact printed patterns of self-assembled monolayers, Synth. Met., 84, 329, 1997. 43. C. N. Sayre and D. M. Collard, Depostion of polyaniline on micro-contact printed self-assembled monolayers of w-functionalized alkanethiols, J. Mater. Chem., 7, 909, 1997. 44. J. A. Rogers et al., Paper-like electronic displays: Large-area rubber-stamped plastic sheets of electronics and microencapsulated electrophoretic inks, Proc. Natl. Acad. Sci. U.S.A., 98, 4835, 2001. 45. R. J. Jackman, J. L. Wilbur, and G. M. Whitesides, Fabrication of submicrometer features on curved substrates by microcontact printing, Science, 269, 664, 1995.

8080_C005.fm Page 485 Friday, April 6, 2007 10:25 AM

Soft Lithography for Fabricating Organic Thin-Film Transistors

485

46. J. A. Rogers, R. J. Jackman, and G. M. Whitesides, Microcontact printing and electroplating on curved substrates. Production of free-standing three-dimensional metallic microstructures, Adv. Mater., 9, 475, 1997. 47. Y. Xia, D. Qin, and G. M. Whitesides, Microcontact printing with a cylindrical rolling stamp: A practical step toward automatic manufacturing of patterns with submicrometer-sized features, Adv. Mater., 8, 1015, 1996. 48. J. A. Rogers et al., Printing process suitable for reel-to-reel production of highperformance organic transistors and circuits, Adv. Mater., 11, 741, 1999. 49. J. A. Rogers, K. E. Paul, and G. M. Whitesides, Quantifying distortions in soft lithography, J. Vac. Sci. Technol., B, 16, 88, 1997. 50. E. Delamarche et al., Transport mechanisms of alkanethiols during microcontact printing on gold, J. Phys. Chem. B, 102, 3324, 1998. 51. N. L. Jeon et al., Structure and stability of patterned self-assembled films of octadecyltrichlorosilane formed by contact printing, Langmuir, 13, 3382, 1997. 52. M. Leufgen et al., Organic thin-film transistors fabricated by microcontact printing, Appl. Phys. Lett., 84, 1582, 2004. 53. R. Parashkov et al., Microcontact printing as a versatile tool for patterning organic field-effect transistors, Adv. Mater., 17, 1523, 2005. 54. K. S. Lee et al., Direct patterning of conductive water-soluble polyaniline for thinfilm organic electronics, Appl. Phys. Lett., 86, 074102, 2005. 55. K. S. Lee et al., High-resolution characterization of pentacene-polyaniline interfaces in thin-film transistors, submitted to Adv. Funct. Mater., 16, 2409, 2006. 56. C. R. Kagan, T. L. Breen, and L. L. Kosbar, Patterning organic-inorganic thin-film transistors using microcontact printed templates, Appl. Phys. Lett., 79, 3536, 2001. 57. A. L. Briseno et al., Patterned growth of large oriented organic semiconductor single crystals on self-assembled monolayer templates, J. Am. Chem. Soc., 127, 12164, 2005. 58. J. W. P. Hsu et al., Directed spatial organization of zinc oxide nanorods, Nano. Lett., 5, 83, 2005. 59. H. Moon et al., Synthesis, crystal structure, and transistor performance of tetracene derivatives, J. Am. Chem. Soc., 126, 15322, 2004. 60. Y.-L. Loo, et al., High-resolution transfer printing on GaAs surfaces using alkane dithiol monolayers, J. Vac. Sci. Technol., B, 20, 2853, 2002. 61. S.-H. Hur et al., Nanotransfer printing by use of noncovalent surface forces: Applications to thin-film transistors that use single-walled carbon nanotube networks and semiconducting polymers, Appl. Phys. Lett., 85, 5730, 2004. 62. K. Felmet, Y.-L. Loo, and Y. Sun, Patterning conductive copper by nanotransfer printing, Appl. Phys. Lett., 85, 3316, 2004. 63. D. R. Hines et al., Nanotransfer printing of organic and carbon nanotube thin-film transistors on plastic substrates, Appl. Phys. Lett., 86, 163101, 2005. 64. J. Zaumseil et al., Three-dimensional and multilayer nanostructures formed by nanotransfer printing, Nano Lett., 3, 1223, 2003. 65. Y.-L. Loo et al., Electrical contacts to molecular layers by nanotransfer printing, Nano Lett., 3, 913, 2003. 66. C. Kim, M. Shtein, and S. R. Forrest, Nanolithography based on patterned metal transfer and its application to organic electronic devices, Appl. Phys. Lett., 80, 4051, 2002. 67. C. Kim, P. E. Burrows, and S. R. Forrest, Micropatterning of organic electronic devices by cold-welding, Science, 288, 831, 2000. 68. G. S. Ferguson et al., Contact adhesion of thin gold films on elastomeric supports: Cold welding under ambient conditions, Science, 253, 776, 1991.

8080_C005.fm Page 486 Friday, April 6, 2007 10:25 AM

486

Organic Field-Effect Transistors

69. M. Nakamura et al., Role of terminal OH groups on the electrical and hydrophilic properties of hydro-oxygenated amorphous TiOx:OH thin films, J. Appl. Phys., 90, 3391, 2001. 70. M. Nakamura et al., In situ analysis of perfluoro compounds in semiconductor process exhaust: Use of Li+ ion-attachment mass spectrometry, J. Vac. Sci. Technol., A, 19, 1105, 2001. 71. O. J. A. Schueller et al., Reconfigurable diffraction gratings based on elastomeric microfluidic devices, Sens. Actuators, A, A78, 149, 1999. 72. D. J. Gundlach et al., Pentacene organic thin-film transistors — Molecular ordering and mobility, IEEE Electron Device Lett., 18, 87, 1997. 73. Z. Bao, A. J. Lovinger, and J. Brown, New air-stable n-channel organic thin film transistors, J. Am. Chem. Soc., 120, 207, 1998. 74. R. Patel, A. Acuna, K. Dickey, and Y.-L. Loo, unpublished data. 75. M. Hu et al., Structure and morphology of self-assembled 3-mercaptopropyltrimethoxysilane layers on silicon oxide, Appl. Surf. Sci., 181, 307, 2001. 76. D. A. Krapchetov et al., Solvent-dependent assembly of terphenyl- and quaterphenyldithiol on gold and gallium arsenide, Langmuir, 21, 5887, 2005. 77. K. Ojima et al., Printing electrode for top-contact molecular junction, Appl. Phys. Lett., 87, 234110, 2005. 78. B. de Boer et al., Metallic contact formation for molecular electronics: Interactions between vapor-deposited metals and self-assembled monolayers of conjugated monoand dithiols, Langmuir, 20, 1539, 2004. 79. J. F. Kang et al., Self-assembled rigid monolayers of 4′-substituted-4-mercaptobiphenyls on gold and silver surfaces, Langmuir, 17, 95, 2001. 80. A. Ulman, Self-assembled monolayers of 4-mercaptobiphenyls, Acc. Chem. Res., 34, 855, 2001. 81. W. Wang, T. Lee, and M. A. Reed, Electronic transport in molecular self-assembled monolayer devices, Proc. IEEE, 93, 1815, 2005. 82. F. Maya et al., Formation and analysis of self-assembled monolayers from U-shaped oligo(phenylene ethynylene)s as candidates for molecular electronics, Chem. Mater., 16, 2987, 2004. 83. D. K. James and J. M. Tour, Electrical measurements in molecular electronics, Chem. Mater., 16, 4423, 2004. 84. S. Jeon et al., Three-dimensional nanofabrication with rubber stamps and conformable photomasks, Adv. Mater., 16, 1369, 2004. 85. M. Granstrom et al., Laminated fabrication of polymeric photovoltaic diodes, Nature, 395, 257, 1998. 86. T.-F. Guo et al., High performance polymer light-emitting diodes fabricated by a low temperature lamination process, Adv. Funct. Mater., 11, 339, 2001. 87. J. Zaumseil, K. W. Baldwin, and J. A. Rogers, Contact resistance in organic transistors that use source and drain electrodes formed by soft contact lamination, J. Appl. Phys., 93, 6117, 2003. 88. J. Zaumseil et al., Nanoscale organic transistors that use source/drain electrodes supported by high resolution rubber stamps, Appl. Phys. Lett., 82, 793, 2003. 89. Y.-L. Loo et al., Soft, conformable electrical contacts for organic semiconductors: High-resolution plastic circuits by lamination, Proc. Natl. Acad. Sci. U.S.A., 99, 10252, 2002. 90. Z. Suo et al., Mechanics of rollable and foldable film-on-foil electronics, Appl. Phys. Lett., 74, 1177, 1999.

8080_C005.fm Page 487 Friday, April 6, 2007 10:25 AM

Soft Lithography for Fabricating Organic Thin-Film Transistors

487

91. V. C. Sundar et al., Elastomeric transistor stamps: Reversible probing of charge transport in organic crystals, Science, 303, 1644, 2004. 92. C. Goldmann et al., Hole mobility in organic single crystals measured by a “flipcrystal” field-effect technique, J. Appl. Phys., 96, 2080, 2004. 93. R. W. I. de Boer, T. M. Klapwijk, and A. F. Morpurgo, Field-effect transistors on tetracene single crystals, Appl. Phys. Lett., 83, 4345, 2003. 94. J. Takeya et al., Field-induced charge transport at the surface of pentacene single crystals: A method to study charge dynamics of two-dimensional electron systems in organic crystals, J. Appl. Phys., 94, 5800, 2003. 95. W. R. Childs and R. G. Nuzzo, Decal transfer microlithography: A new soft-lithographic patterning method, J. Am. Chem. Soc., 124, 13583, 2002. 96. T.-W. Lee et al., Organic light-emitting diodes formed by soft contact lamination, Proc. Natl. Acad. Sci. U.S.A., 101, 429, 2004. 97. C. Kim, P. E. Burrows, and S. R. Forrest, Micropatterning of organic electronic devices by cold-welding, Science, 288, 831, 2000. 98. C. Kim and S. R. Forrest, Fabrication of organic light-emitting devices by lowpressure cold welding, Adv. Mater., 15, 541, 2003. 99. Z. Wang et al., Metal transfer printing and its application in organic field-effect transistor fabrication, Adv. Mater., 15, 1009, 2003. 100. Z. Wang et al., Micropatterning of organic semiconductor microcrystalline materials and OFET fabrication by “hot lift off,” J. Am. Chem. Soc., 125, 15278, 2003. 101. Z. Bao, A. J. Lovinger, and A. Dodabalapur, Organic field-effect transistors with high mobility based on copper phthalocyanine, Appl. Phys. Lett., 69, 3066, 1996. 102. E. Kim, Y. Xia, and G. M. Whitesides, Micromolding in capillaries: Applications in materials science, J. Am. Chem. Soc., 118, 5722, 1996. 103. Y. Xia, E. Kim, and G. M. Whitesides, Micromolding of polymers in capillaries: Applications in microfabrication, Chem. Mater., 8, 1558, 1996. 104. J. A. Rogers, Z. Bao, and V. R. Raju, Nonphotolithographic fabrication of organic transistors with micron feature sizes, Appl. Phys. Lett., 72, 2716, 1998. 105. A. Salleo et al., Polymer thin-film transistor arrays patterned by stamping, Adv. Funct. Mater., 15, 1105, 2005. 106. A. Salleo et al., Polymer thin-film transistors with chemically modified dielectric interfaces, Appl. Phys. Lett., 81, 4383, 2002. 107. W. S. Wong et al., Hydrogenated amorphous silicon thin-film transistor arrays fabricated by digital lithography, IEEE Electron Device Lett., 24, 577, 2003. 108. T.-W. Lee et al., Soft-contact optical lithography using transparent elastomeric stamps: Application to nanopatterned organic light-emitting devices, Adv. Funct. Mater., 15, 1435, 2005. 109. J. A. Rogers et al., Using an elastomeric phase mask for sub-100 nm photolithography in the optical near field, Appl. Phys. Lett., 70, 2658, 1997. 110. J. A. Rogers et al., Generating approx. 90 nm features using near-field contact-mode photolithography with an elastomeric phase mask, J. Vac. Sci. Technol., B, 16, 59, 1998. 111. G. B. Blanchet et al., Large area, high resolution, dry printing of conducting polymers for organic electronics, Appl. Phys. Lett., 82, 463, 2003. 112. S. Y. Chou, P. R. Krauss, and P. J. Renstrom, Imprint lithography with 25-nm resolution, Science, 272, 85, 1996. 113. L. J. Guo, Recent progress in nanoimprint technology and its applications, J. Phys. D, 37, R123, 2004.

8080_C005.fm Page 488 Friday, April 6, 2007 10:25 AM

488

Organic Field-Effect Transistors

114. S. Y. Chou, P. R. Krauss, and P. J. Renstrom, Nanoimprint lithography, J. Vac. Sci. Technol., B, 14, 4129, 1996. 115. M. Colburn et al., Step and flash imprint lithography: A new approach to highresolution patterning, Proc. SPIE, 3676, 379, 1999. 116. T. C. Bailey et al., Step and flash imprint lithography: An efficient nanoscale printing technology, J. Photopolym. Sci. Technol., 15, 481, 2002. 117. S. Y. Chou, P. R. Krauss, and P. J. Renstrom, Imprint of sub-25 nm vias and trenches in polymers, Appl. Phys. Lett., 67, 3114, 1995. 118. J. Haisma et al., Mold-assisted nanolithography: A process for reliable pattern replication, J. Vac. Sci. Technol., B, 14, 4124, 1996. 119. M. D. Austin et al., 6 nm half-pitch lines and 0.04 mm2 static random access memory patterns by nanoimprint lithography, Nanotechnology, 16, 1058, 2005. 120. M. D. Austin et al., Fabrication of 5-nm linewidth and 14-nm pitch features by nanoimprint lithography, Appl. Phys. Lett., 84, 5299, 2004. 121. S. Y. Chou and P. R. Krauss, Imprint lithography with sub-10 nm feature size and high throughput, Microelectron. Eng., 35, 237, 1997. 122. M. Austin and S. Y. Chou, Fabrication of nanocontacts for molecular devices using nanoimprint lithography, J. Vac. Sci. Technol., B, 20, 665, 2002. 123. M. D. Austin and S. Y. Chou, Fabrication of 70-nm channel length polymer organic thin-film transistors using nanoimprint lithography, Appl. Phys. Lett., 81, 4431, 2002. 124. T. Bailey et al., Step and flash imprint lithography: Template surface treatment and defect analysis, J. Vac. Sci. Technol., B, 18, 3572, 2000. 125. J. B. Rolfson, US Patent 5225035, 1993. 126. N. Murata, JP Patent 07056319, 1995. 127. T. C. Bailey et al., US Patent 20050064344, 2005. 128. D. J. Resnick et al., New methods for fabricating step and flash imprint lithography templates, Proc. SPIE, 4608, 176, 2002. 129. D. J. Resnick et al., High resolution templates for step and flash imprint lithography, J. Microlithogr., Microfabrication, Microsyst., 1, 284, 2002. 130. D. J. Resnick et al., Improved step and flash imprint lithography templates for nanofabrication, Microelectron. Eng., 69, 412, 2003. 131. T. Bailey et al., Step and flash imprint lithography, in Alternative lithography: Unleashing the potentials of nanotechnology, ed. by C.M. Sotomayor Torres, Kluwer Academic/Plenum Publishers, New York, 2003, pp. 117–138. 132. D. J. Resnick et al., High-resolution templates for step and flash imprint lithography, Proc. SPIE, 4688, 205, 2002. 133. M. Colburn et al., Step and flash imprint lithography for sub-100-nm patterning, Proc. SPIE, 3997, 453, 2000. 134. M. Colburn et al., Patterning nonflat substrates with a low pressure, room temperature, imprint lithography process, J. Vac. Sci. Technol., B, 19, 2162, 2001. 135. T. Bailey et al., Step and flash imprint lithography: Defect analysis, J. Vac. Sci. Technol., B, 19, 2806, 2001. 136. E. K. Kim et al., Vinyl ether formulations for step and flash imprint lithography, J. Vac. Sci. Technol., B, 23, 2967, 2005. 137. M. Colburn et al., Characterization and modeling of volumetric and mechanical properties for step and flash imprint lithography photopolymers, J. Vac. Sci. Technol., B, 19, 2685, 2001.

8080_C006.fm Page 489 Tuesday, April 3, 2007 2:15 PM

6.1

Radio Frequency Identification Tags

Vivek Subramanian CONTENTS 6.1.1 Introduction................................................................................................489 6.1.2 An Overview of RFID Standards and Classifications...............................490 6.1.2.1 135 kHz RFID.............................................................................490 6.1.2.2 13.56 MHz RFID ........................................................................491 6.1.2.3 900 MHz and 2.4 GHz RFID .....................................................491 6.1.3 Radio Frequency Identification Using Silicon: A Review ........................491 6.1.4 All-Printed RFID Tags: Topology and Architecture Framework..............492 6.1.4.1 Antenna Stage .............................................................................493 6.1.4.2 Rectifier/Power Supply and Clamp.............................................495 6.1.4.3 Digital Section and Modulation Stage........................................497 6.1.5 An Archetypal First Organic RFID Tag ....................................................500 6.1.6 Implications of Tag Architecture on Device Considerations ....................501 6.1.6.1 Transistor Performance and Structural Implications ..................501 6.1.6.2 Circuit Issues ...............................................................................503 6.1.7 Conclusions................................................................................................504 References..............................................................................................................504

6.1.1 INTRODUCTION Radio frequency identification (RFID) tags have received substantial attention in recent years as a potential application for printed organic transistors. The primary driver for consideration of this application is cost; it is expected that the cost of an RFID tag fabricated using printed transistors will be substantially lower than that achievable using conventional technologies. In this section, the status and requirements of RFID are reviewed, with particular emphasis on the process and performance requirements of organic transistors. The section begins with a review of the state of the art of RFID standards and technologies using conventional technologies. The shortcomings in these technologies are identified and used to motivate the interest in printed RFID tags. Next, the possible topologies of RFID tags are reviewed, with emphasis on the performance constraints imposed by printing. Finally, the state of the art of organic transistors for RFID applications is reviewed;

489

8080_C006.fm Page 490 Tuesday, April 3, 2007 2:15 PM

490

Organic Field-Effect Transistors

this analysis is used to drive the identification of future needs or requirements for organic transistors, thus establishing a roadmap of activities and innovation that will drive the realization of printed RFID.

6.1.2 AN OVERVIEW OF RFID STANDARDS AND CLASSIFICATIONS In general, RFID tags may be classified based on two partitions: • •

the method by which they obtain power to operate the specific frequencies at which they communicate with the reader

In the first level of classification, tags are categorized based on whether they have an on-tag battery, or depend on the reader to provide them with power. The former are called “active tags,” while the latter are called “passive tags.” Active tags, due to their higher costs and extended ranges (several meters or more) are currently using in inventory management and high-value asset tracking applications. Passive tags do not contain a battery. Instead, power is supplied to them by the reader through electromagnetic coupling. The reader broadcasts large amounts of power, a small percentage of which is captured or “harvested” by the antenna on the tag. Tags that are close to the reader are thus able to collect enough power to become energized. Since this power transfer is extremely inefficient, the range of passive tags is usually limited; common ranges are a few centimeters to a few meters. Organic transistorbased tags are generally only considered for use in passive tags; since the cost points of active tags are higher, the costs provided by silicon-based circuitry are adequate for most active tag applications. Therefore, in this section, only passive tag architectures will be considered. In the second level of classification, (passive) RFID tags are classified based on the manner in which they communicate with the reader. This is performed based on the frequency at which the reader broadcasts information and power to the tag. In general, the frequency bands already used for RFID around the world are: (1) below 135 kHz; (2) 13.56 MHz; (3) 900 MHz; and (4) 2.4 GHz.

6.1.2.1 135 KHZ RFID Tags operating at 135 kHz operate in the near-field of the reader (i.e., they interact primarily with the magnetic component of the electromagnetic signal broadcast by the reader). In this regime, both the tag and reader antennae are inductors. When the tag is within a usable operating range of the reader, the two inductors are coupled, resulting in a mutual inductance and magnetic coupling between them. Since the inductors required to produce resonant circuits at such low frequencies are large, these inductors are typically wound coil inductors rather than planar spiral inductors [1]. Hence, the fabrication costs associated with low-frequency RFID are generally high, and their use is primarily limited to applications where low frequency is specifically advantageous.

8080_C006.fm Page 491 Tuesday, April 3, 2007 2:15 PM

Radio Frequency Identification Tags

491

Low frequencies generally do not significantly interact with water and other fluids and also work fairly well in the presence of metals. Therefore, these tags work well in environments containing liquids and/or metals. Indeed, one of the most important uses of these tags is for livestock inventory control; tags are inserted into livestock and are readable from the innards of the animal with an external reader. However, given the high costs associated with wound coils, the cost of silicon is only a small fraction of the cost of the tag and, as a result, the use of conventional silicon-based circuitry (with its associated performance and reliability) is acceptable for most low-frequency RFID applications.

6.1.2.2 13.56 MHZ RFID Arguably, the most important RFID operating frequency used today is 13.56 MHz RFID. At this frequency, power is still inductively coupled. However, given the higher frequencies used, it is possible to use planar spiral inductors. These may be fabricated at substantially lower costs than the wound coil inductors used in LF RFID; as a result, 13.56 MHz RFID has received substantial attention as a candidate frequency for low-cost RFID applications [2]. In the presence of liquids, 13.56 MHz RFID works well, but is moderately susceptible to interference from metals nearby. However, compared to other candidate frequencies, 13.56 MHz RFID is generally considered to be the most promising frequency for use in RF barcodes.

6.1.2.3 900 MHZ

AND

2.4 GHZ RFID

RFID tags operating at 900 MHz and 2.4 GHz are called ultrahigh frequency (UHF) and microwave tags. Since these tags typically operate in the far-field region of the reader electromagnetic field, they primarily interact with the electric component of the same. Backscatter systems are used for communication, and the antenna is typically a dipole configuration [3]. This is conveniently formed in a printed planar configuration; as a result, printed antennae have already been adopted in these frequency ranges. These systems are able to achieve relatively long-range, high datarate operation and are therefore considered to be extremely attractive for pallet-level and case-level tracking applications, for example. Unfortunately, such tags are extremely sensitive to liquids and metals and also show various interference phenomena due to reflections, which has resulted in reduced interest in their use for item-level tracking.

6.1.3 RADIO FREQUENCY IDENTIFICATION USING SILICON: A REVIEW RFID technology to this point has had mixed success. While RFID has been tremendously successful in high-value asset tracking applications, it has only had moderate success in the low-cost applications that offer the largest potential markets. This has primarily been limited by the cost of current RFID, which is still a little too high for the target markets [4]. This, in fact, is the primary motivation for the use of organic transistors in low-cost RFID. Therefore, to facilitate a better

8080_C006.fm Page 492 Tuesday, April 3, 2007 2:15 PM

492

Individual chips are fabricated on water, but not separated

Organic Field-Effect Transistors

Chips are separated for handling

Chips are attached to a strap Si

Straps are attached to antennae

FIGURE 6.1.1 Conceptual process for silicon-based RFID tag fabrication.

understanding of the economics therewith, an overview of RFID based on conventional silicon is provided. The cost of silicon chips rises dramatically with increase in chip size, due to a reduction in process yield with increasing chip size and also due to a high cost per unit area associated with silicon. As a result, size reduction is a dominant goal in silicon-based RFID. This has important consequences for circuit partitioning. The size of the antenna is independent of the process technology used and, in fact, can be very large, depending on the operating frequency and desired range. Therefore, in silicon-based RFID tags, the antenna is moved off-chip. In other words, everything but the antenna is fabricated on the silicon chip, which is then attached to a separate antenna, typically fabricated on a plastic inlay or strap, as shown in Figure 6.1.1. While this partitioning strategy reduces the cost of the silicon chip, it adds additional components into the cost equations associated with the overall tag. The overall cost equation for an RFID tag therefore includes three main components: (1) the cost of silicon; (2) the cost of attachment; and (3) the cost of the antenna. Coupling all of these factors together, costs for silicon-based RFID will likely fall into the range of US$0.05 per tag in the foreseeable future, but will struggle to get into the range of US$0.01–0.02 required for many RF barcode applications. This has driven the interest in printed electronics, since, in an ideal world, the cost of a fully printed RFID tag would not be significantly larger than the cost of today’s printed antennae, resulting in a dramatic reduction in overall tag cost to less than US$0.02 per tag.

6.1.4 ALL-PRINTED RFID TAGS: TOPOLOGY AND ARCHITECTURE FRAMEWORK As has been discussed in previous sections, while the economic imperatives for considering printed electronics in general and organic electronics in particular are attractive, there are clear performance trade-offs. First, the performance of printed semiconductors is much worse than the performance of silicon. Second, the line widths and layer-to-layer registration achievable by printing are substantially worse than achievable in silicon technology as well. Based on these constraints, it is

8080_C006.fm Page 493 Tuesday, April 3, 2007 2:15 PM

Radio Frequency Identification Tags

493

Modulation circuit

Clock S SET Q R CLR Q

Power supply

Digital (implements ﬁnite state machine, memory, etc.)

Antenna stage

Protection clamp

FIGURE 6.1.2 Block diagram of archetypal HF passive RFID tag.

possible to examine possible circuit architectures for printed RFID tags and then discuss the implications of printed device performance on the viability of the same. As a first constraint, recall that all passive tags require that power be derived from the incident carrier frequency broadcast by the reader. For low-cost applications, this will likely be in the high frequency (HF) or UHF bands, as discussed previously. Given the performance limitations of organic materials, it is highly unlikely that harvesting of UHF frequencies will be possible. As a result, the circuit analysis here will focus exclusively on HF tags. As discussed previously, HF tags operate at 13.56 MHz. Power is inductively coupled from the reader to the tag. A block diagram architecture for a typical HF RFID tag is shown in Figure 6.1.2. For the purposes of simplifying analysis, the tag has been broken down into several blocks, which will be individually discussed.

6.1.4.1 ANTENNA STAGE As discussed earlier, the antenna in an HF tag operates through inductive power coupling. Conceptually, the reader antenna coil acts as a primary coil in a transformer, with the coil in the tag acting as a secondary. Coupling occurs through air; if the tag is within the near-field region of the reader antenna, magnetic coupling between the two coils occurs, and some voltage is “harvested” by the tag coil. To increase the voltage generated by the harvesting process, the antenna stage on the tag is typically a resonant circuit with the antenna inductor connected in parallel

8080_C006.fm Page 494 Tuesday, April 3, 2007 2:15 PM

494

Organic Field-Effect Transistors

L1 (Inductor on reader)

V1

Rs (Resistance of L2)

C2 (Tuning capacitor)

L2 (Inductor on tag)

RL (Load resistance of RFID circuit/ regulator circuit)

FIGURE 6.1.3 Conceptual equivalent circuit for an inductively coupled RFID tag.

with a tuning capacitor. The specific values of the inductance and capacitance are chosen so as to cause resonance at 13.56 MHz. The consequence of this is that the voltage seen at the terminals of the tuned circuit is Q-boosted, where Q is the loaded quality factor of the antenna stage [1]. Typical unloaded Q-values in conventional RFID circuits range from one to two, depending on the desired range. To achieve high Q, low series resistance of the inductor metallization is desired. As a result, substantial effort in recent years has been devoted to the development of low-resistance printed metals [5,6]. This antenna series resistance is extremely important, since the series resistance of the antenna acts as a loss mechanism, falling in series in the LC circuit (Figure 6.1.3) and thus reducing the Q of the circuit. This in turn reduces the power harvested and made available to the RFID circuit. In Figure 6.1.3, L1 is the inductor on the reader, L2 is the inductor in the tag, C is the tuning capacitor on the tag, Rs is the series resistance of the inductor, and Rl is the equivalent load resistance of the RFID circuitry. As is apparent from Figure 6.1.3, the two inductors in fact form a transformer with: v1 = L 1

di1 di2 +M dt dt

(6.1.1)

di1 di2 v2 = M + L2 dt dt where M is the mutual inductance and is related to the coupling factor by: k=

M

(6.1.2)

L 1L 2

Note that the coupling factor between the two inductors L1 and L2 is typically extremely low (<<10%, depending on separation between the reader and tag), resulting in very poor overall coupling efficiency. This is obvious, of course, since the

8080_C006.fm Page 495 Tuesday, April 3, 2007 2:15 PM

Radio Frequency Identification Tags

495

RS (resistance of L2)

C2 (tuning capacitor)

V1'

V2

L2 (inductor on tag)

RL (load resistance of RFID circuit/ regulator circuit)

FIGURE 6.1.4 Simplified equivalent circuit for an inductively coupled RFID tag.

coil L1 on the reader is broadcasting power in all directions, while the tag only harvests power from the region of the electromagnetic field that intersects L2. The power available at the tag may be calculated by redrawing Figure 6.1.3 as shown in Figure 6.1.4, which allows calculation of the delivered voltage as:

v2 =

(

RL v1′ RS + RL − ω L2CRL + jω ( RL RS C + L2 ) 2

)

(6.1.3)

where v1′ = jωMi p

(6.1.4)

and ip is the current through the primary coil on the reader.

6.1.4.2 RECTIFIER/POWER SUPPLY

AND

CLAMP

The voltage from the tank is applied to a rectification circuit. At HFs using high-Q tanks, coupling high voltages (several tens of volts) is relatively common, particularly at short ranges. To see why this is true, consider that the field strength B drops as 1/d3, where d is the distance between the reader and the tag. The maximum electromagnetic field (emf) generated by an unloaded coil of area Atot within this field is: emf = ( jωB) ⋅ Atot where B is the magnetic field strength seen by the coil.

(6.1.5)

8080_C006.fm Page 496 Tuesday, April 3, 2007 2:15 PM

496

Organic Field-Effect Transistors Diode-connected TFT Thin ﬁlm diode (probably Schottky)

RL RL

FIGURE 6.1.5 Conceptual rectification strategies used in HF RFID tags.

Since the field strength varies as 1/d3, the variation in maximum unloaded emf generated by this coil varies strongly with distance. Clearly, extremely large voltages can be generated. Note that, in practice, voltages will be significantly smaller, since the coils are in fact loaded. Given the preceding analysis, however, it is clear that some voltage clamping may potentially be required to protect the tag circuitry from high voltages. Rectification is performed using diodes (or their functional equivalent such as diodeconnected transistors). The output of the rectifier is connected into a filter capacitor, which smoothes out any ripple. This power is then used to drive the digital sections of the circuit. Schematic representations of the two common configurations of rectifiers (i.e., using diodes vs. using diode-connected transistors) are shown in Figure 6.1.5. For printed RFID applications, both types of rectifiers have been studied. Diodes typically have the advantage of being faster than transistors; however, very few examples of high-performance, fully printed diodes exist, and this is an area of intensive research today. Using nonprinting processes, however, several demonstrations of rectifiers running at 13.56 MHz have been made [7]. Diode-connected transistors have also been demonstrated for use in rectifiers. Since the process steps used to fabricate these are similar to those used to fabricate the transistors used in the digital section (described later), the overall process flow is likely advantageous over diodes. However, as will be discussed later, realizing printed transistors that switch at 13.56 MHz is extremely difficult, particularly using printing. However, using evaporated materials coupled with lithography and sharing the gate overlap capacitance (discussed later) as part of the filter capacitance, there have been some initial demonstrations of transistor-based rectification at 13.56 MHz [8]. In either case, there is a significant concern with operating voltage. For longrange operation (e.g., 10 cm or more), the resonant circuits will likely be constructed with Q > 10. This will enable the generation of reasonable operating voltages for driving the RFID circuitry at the extremities of the operating range. However, as the tag moves into closer proximity with the reader, generated voltages

8080_C006.fm Page 497 Tuesday, April 3, 2007 2:15 PM

Radio Frequency Identification Tags

497

will increase dramatically, and some sort of clamping architecture to protect the RFID circuitry and the rectifying diodes/transistors will be required. This is the function of the clamping circuitry. There are no known printable approaches to address this problem currently; however, this is not an immediate concern since the initial target application of most printed RFID tags will likely be for shortrange/proximity tags. These will be run at low Q and thus not see the extremely high voltages that would exist in a tag designed for operation at long range but operating in proximity.

6.1.4.3 DIGITAL SECTION

AND

MODULATION STAGE

The digital sections of the circuit include hundreds to thousands of transistors. The main function of this circuit is to generate a unique ID as a bit string signal. In its simplest form, therefore, the digital circuit will include a memory, decoding circuit, and a counter [9]. Further complexity may be required if the circuit implements an anticollision scheme. From an architectural perspective, the digital section is essentially a finite state machine coupled to a memory. When the tag enters the reader field, the rectifier charges up the smoothing capacitor. When the voltage on this capacitor reaches a sufficient value, a power-on-reset circuit turns off, allowing the RFID circuitry to operate. In the simplest RF barcode applications, the tag circuitry will then simply broadcast a unique ID repetitively. This may be achieved using a sequentially read memory connected to the output stage (shown schematically in Figure 6.1.6). Counter

Row Decoder D1

B1

S1 S2

Reset

B8

D8

Memory

S3

t S1

S8

C1 C2

Multiplexer C3 D

B

To Modulator

FIGURE 6.1.6 Digital subcircuit architecture for a simple RFID tag.

8080_C006.fm Page 498 Tuesday, April 3, 2007 2:15 PM

498

Organic Field-Effect Transistors

Antenna stage 13.56M Divide-by-two frequency divider

6.7M

Divide-by-two frequency divider

3.3M

Divide-by-two frequency divider 1.7M

Digital circuitry

200K

Divide-by-two frequency divider

400K

Divide-by-two frequency divider

800K

Divide-by-two frequency divider

FIGURE 6.1.7 Divider-based clock generation in HF RFID tags.

More sophisticated RFID tags may require anticollision (i.e., may require a means by which numerous tags may exist and be energized within the reader field at the same time and yet enable the reader to uniquely identify each of the individual tags). There are numerous anticollision protocols currently in existence, ranging from simple tags-talk-first protocols (TTF) to more sophisticated protocols as embodied in the electronic product code specification [10]. Anticollision protocol implementations will almost certainly require more complex circuits, which will make them difficult to implement in a fully printed process. As a result, at least in the near term, it is unlikely that printed RFID tags will make use of anything more complex than TTF if they implement anticollision at all. The output of the digital stage is fed to the modulation stage. This stage typically loads the tank based on the output of the digital stage, causing the current in the tank to change. This change in tank current can be sensed by the reader and converted into the corresponding data stream. The precise circuitry of the modulation stage depends on the specific encoding methodology; however, it typically consists of at most a few transistors. The digital stage has at lease one input (i.e., a clock signal), though it may also have other inputs if more complex anticollision or communication architectures are implemented as discussed earlier. While the carrier signal is at 13.56 MHz, the clock for the digital circuit is substantially slower than this; typical HF tags have clock rates of a few hundred kilohertz. Based on current standards, HF tags generate the clock signal by dividing the carrier frequency, as shown in Figure 6.1.7. This has an interesting consequence: The only parts of the entire RFID tag that run at the carrier frequency are the rectification stage (which has to rectify at 13.56 MHz) and the first stage of the divider. This is problematic if the performance of the circuitry is such that it is not possible to implement a 13.56 MHz divider. For printed transistors, this is almost certainly going to be the case, as will be discussed later. Given the difficulties in dividing 13.56 MHz signals using low-performance printed circuitries, alternative strategies have been proposed and, indeed, have been used in some silicon-based RFID tags as well. For example, the clock signal could be generated on the tag using an oscillator running at the desired clock frequency, as shown in Figure 6.1.8. This is typically called asynchronous communication, since the tag clock is not synchronized to the carrier signal. This is somewhat

8080_C006.fm Page 499 Tuesday, April 3, 2007 2:15 PM

Radio Frequency Identification Tags

499

Local Oscillator 200 kHz Clock

Modulation Circuit

Power Supply

Digital (implements ﬁnite state machine, memory, etc.)

Antenna Stage

Protection Clamp

FIGURE 6.1.8 Asynchronous clock generation in HF RFID tags.

problematic, since simple oscillators such as ring oscillators have an oscillation frequency that depends on the applied voltage. Since the applied voltage available on a tag depends on the coupling to the tag, this results in a clock frequency that will vary depending on variations in tag manufacturing and power coupling. This will cause reading difficulties for the reader. This problem may be solved by using a voltage regulator on the tag to ensure that the tag circuitry only receives a narrow range of voltages a more sophisticated oscillator forgiving communication protocols and a more sophisticated reader to ensure that some variation in data rates is allowed The most likely implementation would be a combination of all of these methods. Another method of providing a stable clock signal would be to send the clock signal amplitude modulated onto the carrier frequency. A demodulator stage on the tag could be used to extract the clock signal from the carrier. Unfortunately, the bandwidth available at 13.56 MHz based on existing licensing regulations is extremely small and will likely limit any clock signal sent using this method to <20 kHz, as shown in Figure 6.1.9. This will limit the usability of such tag architectures to applications in which low data rates are acceptable. The final portion of the digital circuitry to be considered is the memory. This remains a complete unknown for organic/printed electronics at this time. The memory must necessarily be nonvolatile. For some RFID architectures, it will be necessary to provide EEPROMs or at least PROMs, since some programmability in the

8080_C006.fm Page 500 Tuesday, April 3, 2007 2:15 PM

500

Organic Field-Effect Transistors +60dBuA/m

+9dBuA/m −3.5dBuA/m

13.56 MHz

−10dBuA/m −150 kHz −450 kHz

−150 kHz −16dBuA/m

−450 kHz

−900 kHz

−900 kHz

FIGURE 6.1.9 European spectral mask for 13.56-MHz RFID tags.

field will be required. For others, however, a ROM architecture will suffice. For example, in the simplest RF bar code implementations, the unique ID could be factory programmed (using a printed metal mask, for example). This is advantageous since it could likely be realized using existing technology and would eliminate the need for on-tag programming circuitry. PROM architectures may be necessary for some simple tag applications since field programmability is required for many applications (for example, a storefront may want the ability to custom encode a barcode, rather than depending on factoryencoded values). Simple PROM architectures could potentially be realized using antifuses, though this technology has only been recently demonstrated in an organicsbased process [11]. While simple PROM-style memories are thus potentially realizable in a printed process, the requirement for the associated programming circuitry and control circuitry will make the overall digital architecture substantially more complicated. Overall, therefore, the digital section represents the major portion of a typical RFID tag. Including the clock generation section, this represents the vast majority of the overall transistor count, has some of the most difficult performance requirements, and also represents the portion of the tag in which substantial opportunities for innovation exist.

6.1.5 AN ARCHETYPAL FIRST ORGANIC RFID TAG Based on the preceding points, it is possible to predict a likely initial deployment point for organic RFID:

8080_C006.fm Page 501 Tuesday, April 3, 2007 2:15 PM

Radio Frequency Identification Tags

•

•

•

• •

501

It is reasonable to assume that the tags will likely use substantial amounts of printing for cost reasons; given the cost competitiveness of siliconbased RFID, it is not clear that lithographically patterned organic electronics offer any compelling reason for adoption as an RFID platform; the key advantage for organics will likely come from printing. The first printed RFID tags will almost certainly be designed for shortrange operation. This will allow the generation of relatively high operating voltages, necessary for current generations of printed devices, and will also eliminate the need for yet-to-be-developed clamping devices. Anticollision will likely not be used in first-generation printed RFID tags. At best, there may be some deployment of rudimentary TTF protocols. This will reduce transistor count requirements and will also eliminate the need for sophisticated finite state machines and memory architectures. At least initially, unique IDs will likely be factory programmed using printed ROMs. Initial deployments will likely use asynchronous communication protocols, eliminating the need for a divider with its associated high-speed circuitry.

Given this general outline, it is now possible to examine the state of the art of organic transistors and identify the critical issues to be considered with regard to the realization of printed RFID tags.

6.1.6 IMPLICATIONS OF TAG ARCHITECTURE ON DEVICE CONSIDERATIONS To proceed with an analysis of tag architecture on device implications, it is worthwhile to summarize the characteristics of organic transistors as discussed in previous sections and directly relate these to their relevant architectural implications. In general, most printed device families reported to date have the following characteristics.

6.1.6.1 TRANSISTOR PERFORMANCE AND STRUCTURAL IMPLICATIONS As discussed in previous sections, there has been substantial improvement in organic transistor technology over the last few years. However, performance is still substantially worse than silicon. Coupled with the lower carrier mobility are the implications of printing. It is highly unlikely that any manufacturable printing technology will realize the submicron patterning capability achievable using optical lithography. This has important consequences on overall device switching performance: •

•

Low carrier mobility. Typical field-effect mobilities reported in the literature are in the range of 10–2–1 cm2/V-s, which is approximately three orders of magnitude lower than field-effect mobilities achievable in silicon. Long gate length. Commercially viable printing techniques are currently incapable of producing line widths of <20 μm. They are also incapable of realizing line spaces much smaller than that, based on the bleed-out,

8080_C006.fm Page 502 Tuesday, April 3, 2007 2:15 PM

502

Organic Field-Effect Transistors

•

spreading, and placement inaccuracy associated with high-speed printing. Consequently, printed transistors with gate lengths much less than 20 μm will be difficult to realize in a manufacturing-worthy process. It should be noted, however, that several novel printing techniques are in development that may ease this constraint to a degree. Large overlap capacitance. In high-speed printing processes, achieving a layer-to-layer registration of better than 10 μm is extremely difficult and generally not considered manufacturable using commercially viable technologies today. Typical printed transistors have printed source/drain patterns and a separately printed gate pattern. The alignment of these two layers to each other is achieved using the layer-to-layer registration capabilities of the printers in question. Since there is relatively poor control of the layer-to-layer alignment between these layers, it is critical that there be enough overlap between the layers to ensure that ALL devices have at least some overlap even in the worst case misalignment, since device performance is dramatically degraded in the underlap condition, particularly using low-mobility printed materials. This results in a large overlap in typical printed devices, causing an increase in gate-to-source and gateto-drain overlap capacitance.

Given these facts, it is possible to evaluate transistor switching performance and the implications of the same. As a consequence of the aforementioned low mobility, channel length > 10 μm and large overlap capacitance printed organic transistors typically suffer from poor switching characteristics. Current generations of printed devices have transition frequencies (ft) in the range of tens of kilohertz at the most: ft =

gm 2π(C in + C F )

(6.1.6)

where gm is the transconductance at the operating point (depends on mobility and channel length) and Cin and CF are the input and feedthrough capacitances, respectively. CF depends strongly on the overlap capacitance in typical circuit topologies. Given that most digital logic architectures run at clock frequencies that are at most 20–40% of the transition frequency, this severely limits the maximum clock frequencies for the digital section of the RFID tag and also makes the realization of a divider running at 13.56 MHz highly unrealistic without fundamental improvements in both device and process technology. This suggests that nonstandard architectures will likely be required, at least in the near term, for realization of printed RFIDs. •

High operating voltage. Most printed materials are amorphous or polycrystalline at best. As a result, they typically show I–V characteristics that fit best to multiple trap-and-release or variable range-hopping models. In either case, they show a field-activated mobility behavior; organic FETs typically only deliver the stated mobilities of 0.1–1cm2/V-s at relatively high fields.

8080_C006.fm Page 503 Tuesday, April 3, 2007 2:15 PM

Radio Frequency Identification Tags

•

503

Thick gate dielectric requirements. Typical gate dielectric thicknesses in most printed processes will be several tens to hundreds of nanometers. This is required to avoid concerns with pinholes, etc., in typical highspeed printing processes.

The net consequence of these facts is that typical printed transistors will have operating voltages in the range of 10 V; indeed, most current printed transistor demonstrations have operating voltages > 20 V. Relating this back to RFID, either substantial Q-boosting will be required or the tag range will be limited to a few centimeters.

6.1.6.2 CIRCUIT ISSUES In addition to the direct device implications, there are also implications of printed device technology on circuit architecture in general. These include the general immaturity of printable NMOS relative to PMOS, as well as issues related to device stability. •

•

•

•

Logic family. Most printed material demonstrated to date is PMOS. Recently, as discussed in previous sections, there have been some demonstrations of printable NMOS materials as well, though performance and stability of these materials still lag behind that of PMOS. To achieve lowpower operation, availability of CMOS will be highly desirable; otherwise, operating range of tags will be degraded by the high power consumption requirements of the digital sections of the tag. Environmental stability. Perhaps one of the biggest concerns with organic materials to date relates to their poor environmental stability. Most organic semiconductor materials are prone to degradation on exposure to oxygen and/or moisture. This will necessitate the development of robust encapsulation processes. This should be possible, however, since, unlike in OLEDs, optical transparency is not required for RFID tags, which dramatically opens up the range of candidate encapsulation materials. Bias stability. Additionally and perhaps of a greater concern for RFID applications, most organic devices to date show substantial bias stress effect, where their threshold voltage shifts during use. The mechanisms for this are currently being debated; however, the consequence is that organic devices show a history-dependent performance, which is problematic from a circuit design perspective, for obvious reasons. Poor diode performance. As discussed previously, typical printed diodes reported to date operate in the space-charge limited conduction regime, resulting in very high diode series resistance. Additionally, due to large numbers of defects in the junction region, most printed diodes have very poor ideality factor. These two phenomena together result in very poor rectification efficiency. To overcome this problem, it is necessary to improve diode mobility or reduce diode layer thickness. In recent years, there have been several demonstrations of organic diodes operating at

8080_C006.fm Page 504 Tuesday, April 3, 2007 2:15 PM

504

Organic Field-Effect Transistors

13.56 MHz; however, they typically deliver very poor efficiency at these frequencies, limiting their use to extremely short ranges. The range may be extended by using thinner layers; however, this typically results in low breakdown voltages, which is a problem given the variation in voltage seen by the diodes as a function of reader-tag separation, as discussed earlier. Based on the preceding points, it is possible to generally summarize the circuit implications of printed organic electronics. Since rectification efficiency will be low and CMOS may not be realizable, it is likely that initial deployments of organic TFTs in RFID will likely be in short-range/near-proximity applications in which adequate power will be available to operate highly inefficient tags. Additionally, communication protocols and circuit architectures will likely have to be conservative to account for the instability of organic transistors.

6.1.7 CONCLUSIONS Radio frequency identification represents one of the most intriguing applications for organic TFTs. The key driver is clearly cost; as a result, the real interest in RFID is driven by printing. Assuming that a fully printed process is realized, it is likely that a cost advantage for printed RFIDs will exist over conventional RFID technologies. Unfortunately, from a performance perspective, RFID is also one of the most challenging applications for organic TFTs. The current performance of printing technology and organic TFTs is generally marginal for RFID. However, as materials improve and printing technology develops, it is likely that niche RFID opportunities may become available, particularly in applications requiring short-range and lowcomplexity tags such as product authentication and anticounterfeiting. Should organic transistor technology continue to evolve, the potential for printed RFID will remain strong, ushering in an area of ubiquitous printed tags in a wide range of authentication, inventory control, and general barcoding applications.

REFERENCES 1. K. Finkenzeller, RFID handbook: Fundamentals and applications in contactless smart cards and identification, Wiley, New York, 2003. 2. T. Scharfeld, An analysis of the fundamental constraints on low-cost passive radiofrequency identification system design, MS thesis, Massachusetts Institute of Technology, 2001. 3. R. Glidden et al., Design of ultralow-cost UHF RFID tags for supply chain applications, IEEE Commun. Mag., 42, 140–151, 2004. 4. C. Homes, D. Metcalfe, and S. Takahashi, Exposing the myth of the 5-cent RFID tag: Why RFID tags will remain costly this decade, Forrester Res., 2004. 5. D. Redinger, R. Farshchi, and V. Subramanian, Inkjetted passive components on plastic substrate for RFID, IEEE Trans. Electron Devices, 51, 1978, 2004.

8080_C006.fm Page 505 Tuesday, April 3, 2007 2:15 PM

Radio Frequency Identification Tags

505

6. V. Subramanian, J.M.J. Fréchet, P.C. Chang, D. Huang, J.B. Lee, S.E. Molesa, A.R. Murphy, D.R. Redinger, and S.K. Volkman, Progress towards development of allprinted RFID tags: Materials, processes, and devices, Proc. IEEE, 93, 1330–1338, 2005. 7. S. Steudel, K. Myny, V. Arkhipov, C. Deibel, S. De Vusser, J. Genoe, and P. Heremans, 50-MHz rectifier based on an organic diode, Nat. Mater., 4, 597–600, 2005. 8. R. Rotzoll, S. Mohapatra, V. Olariu, R. Wenz, M. Grigas, K. Dimmler, O. Shchekin, and A. Dodabalapur, Radio frequency rectifiers based on organic thin-film transistors, Appl. Phys. Lett., 88, 123502, 2006. 9. E. Cantatore, T. Geuns, A. Gruijthuijsen, G. Gelinck, S. Drews, and D. de Leeuw, A 13.56-MHz RFID system based on organic transponders, 2006 Int. Solid State Circuits Conf., 15.2, 2006. 10. S. Sarma, D. Brock, and D. Engels, Radio frequency identification and the electronic product code, IEEE Micro, 21, 50–54, 2001. 11. B. Mattis and V. Subramanian, Stacked low-power field-programmable antifuse memories for RFID on plastic, International Electronic Device Meeting, 2006, 11.

8080_C006.fm Page 506 Tuesday, April 3, 2007 2:15 PM

8080_C006.fm Page 507 Tuesday, April 3, 2007 2:15 PM

6.2

Organic Transistor Chemical Sensors

Luisa Torsi, M. C. Tanese, Brian Crone, Liang Wang, and Ananth Dodabalapur CONTENTS 6.2.1 Introduction................................................................................................507 6.2.2 Chemical Sensors: An Overview...............................................................508 6.2.3 Organic Thin-Film Transistor Sensors ......................................................511 6.2.3.1 Device Structure ..........................................................................511 6.2.3.2 Multiparametric OTFT Sensors ..................................................512 6.2.3.3 Interface-Dependent OTFT Responses .......................................515 6.2.3.4 Gate-Induced Response Repeatability and Enhancement ..........516 6.2.3.5 Selectivity of an OTFT ...............................................................521 6.2.3.6 Scaling Behavior of Sensing Responses.....................................522 6.2.4 Applications ...............................................................................................524 References..............................................................................................................526

6.2.1 INTRODUCTION Organic thin film transistor (OTFT) sensors were proposed for the first time in the late 1980s [1], just a few years after the first organic thin-film transistor was proposed [2]. These preliminary studies reported on OTFTs’ source-drain current changes upon exposure to different volatile molecules [1,3,4], and, although the idea of using a bottom gate organic thin-film transistor as a sensing device was put forward by this early work, it was not clarified why a three-terminal device structure should have been beneficial to the sensor performance. The field of OTFT sensors has grown in the last years thanks to the advancements achieved by different groups [5–25]. Despite the wide range of chemical sensor technology available, implementing a sensitive, selective, reliable, and inexpensive portable system for detecting volatile analytes is still an issue. OTFTs appear promising in this sense because they have been demonstrated to operate as multiparametric sensors [21] and have shown remarkable response repeatability [10]. Moreover, conductive polymer (CP)-based sensing circuits have been also proposed.

507

8080_C006.fm Page 508 Tuesday, April 3, 2007 2:15 PM

508

Organic Field-Effect Transistors

6.2.2 CHEMICAL SENSORS: AN OVERVIEW Many different types of gas sensors have been proposed so far but only a few of these systems seem to satisfy the minimum requirements to be implemented in portable sensing instruments capable of performing fast and in situ detection of volatile analytes. In addition to sensitivity, selectivity, and robustness, low power consumption and compact size are also stringent requirements that sensor devices have yet to fulfill. Current commercial instruments are of two main types: metaloxide and CP resistive sensors, and inorganic field-effect sensors. Array-based systems called electronic- or e-noses are also currently being investigated and fabricated implementing CP-based sensors in sensing active matrix. For the sake of comparison, we here briefly recall the basics of such devices, whose properties have been deeply examined in many review articles [26–29]. The sensors for combustible gas alarms mounted in most houses are metal-oxidebased chemiresistors. Such sensors have the advantage of high stability and good sensitivity but they are not very selective towards different odor molecules. These are the simplest portable sensors and are constituted by two-terminal devices employing gas-sensitive metal-oxide active layers such as SnO2, ZnO, TiO2, and WO3, as well as perovskites. A schematic overview of a chemiresistor is given in Figure 6.2.1. The conductivity of the metal-oxide active layer under an applied bias between two contact areas is here investigated in presence of a reducing agent to be detected. When the chemiresistor is in normal air, the oxygen present in the air oxidizes the active layer, removing electrons from the bulk of the metal-oxide semiconductor. The free carrier concentration is subsequently reduced, creating a depletion region at the surface exposed to the air with a consequent decrease in conductivity. When the metal-oxide film is exposed to a reducing analyte, it removes oxygen at the interface, injecting electrons into the semiconductor. The depletion layer is thus reduced, resulting in a conductivity enhancement. The gas analyte is detected by means of its effect on the electrical resistance of a metal-oxide semiconductor active layer. The covalent interactions between the analyte molecules and the metal-oxide active layer require metal-oxide-based chemiresistors to be operated at high temperatures to allow the device to be reset by release of gas molecules bonded to the metal oxide. This, in addition to the poor selectivity of metal-oxide active layer, constitutes a further limitation for this class of sensors. For this reason, the use of organic active layers such as CPs, instead of metal-oxide sensors, is being widely investigated in chemiresistors [27,30]. The wide range of organic material that can be synthesized enables the fabrication of CP chemiresistors with sensitivities over a broad range of organic compounds. CPs are also cost effective and easily synthesized, and they present fast responses to a large number of volatile analytes with low power consumption. Nevertheless, reliability in CP-based chemiresistors is still an issue. They are sensitive to humidity and have long-term drift. Another important class of sensors is constituted by inorganic field-effect gas sensors, known as chemically sensitive field-effect transistors (CHEMFETs) [28,31–35]. Their most successful application is in highly sensitive hydrogen sensors commercialized by Sensistor AB (Sweden) more than 15 years ago as leak detectors. They have proven over the years to be very reliable. A CHEMFET is essentially a

8080_C006.fm Page 509 Tuesday, April 3, 2007 2:15 PM

Organic Transistor Chemical Sensors

509

Analyte

Metal-oxide active layer

(a) In air O–2 O– O– O–2 O– O–2 O– O– Depletion region

Metal-oxide active layer (b) Reducing Analyte (R) RO–2 RO– O– O–2 O– RO–2 Depletion region

Metal-oxide active layer (c)

FIGURE 6.2.1 (a) Schematic overview of a metal-oxide chemiresistor; (b) in the ambient environment oxygen extracts electrons from the metal-oxide film; (b) in the presence of a reducing agent it reduces with the oxygen at the surface of the sensor and electrons are injected into the sensing layer.

silicon-based top gate Metal-Insulator-Semiconductor Field Effect Transistor (MISFET) top gate transistor with its gate contact exposed as a sensing element (see Figure 6.2.2) [28]. The FET channel material is usually crystalline silicon, buried underneath the gate contact and the gate dielectric layer; therefore, the transport in this region is not directly affected by the exposure to the analyte. The transduction principle is based on the linear relationship existing between the transistor threshold voltage, VtCHEMFET, and the difference between the gate material and the semiconductor work functions (φM and φS, respectively) [35]: VtCHEMFET ∝ (Φ M − Φ S )

(6.2.1)

A change of the work function, in each of these materials or in both, can take place upon exposure of the CHEMFET gate contact to an analyte and the response

8080_C006.fm Page 510 Tuesday, April 3, 2007 2:15 PM

510

Organic Field-Effect Transistors

Analyte

Source

Sensing gate

Drain

SiO2 n+

n+ p–Si

FIGURE 6.2.2 Typical structure of a sensing CHEMFET.

to the odor molecule is given as the overall variation of the VtCHEMFET. The use of metals such as palladium as gate electrode permits hydrogen detection down to the parts per billion range by exploiting catalytic-type reactions. The VtCHEMFET shift has been modeled considering the catalytic dissociation of the hydrogen molecules adsorbed at the Pd exposed surface. H atoms are thought to diffuse into the metal, reaching the metal/insulator interface and acting as dipoles that change the system work function, which in turn shifts the CHEMFET linear threshold voltage according to the linear relation given previously. The sensitivity and selectivity of such devices depends on the gate materials and on the operating temperature, which is usually between 50 and 200°C. As in the case of chemiresistors, the selectivity properties of CHEMFETs were broadened by the use of conducting polymers as gate materials. Different conducting polymers were employed as CHEMFET gate layers [33]. Conducting polymers are available with a variety of work functions and with very porous morphologies that allow easy permeation of molecules, particularly solvents. When exposed to an analyte, the work function of the polymer/gate-dielectric system changes, and the CHEMFET Vt shifts. One of the major advantages of these sensors is the ability to detect changes in Vt by passing a negligible current through the conducting polymer. This makes CHEMFETs generally more stable than chemiresistors, especially in terms of signal-to-noise ratio. Detected concentrations are generally in the high parts per million range, and responses are fairly reversible for both CP-based chemiresistors and CHEMFETs. The sensing mechanism in CP-based CHEMFETs is attributed to the changes in the gate work function φM, which has been ascribed to the formation of complexes between the polymer and organic analyte molecules such as methanol, isopropanol, or hexane. In this case a partial charge transfer on the polymer or on the analyte molecules has been postulated, depending on the relative difference between the polymer electrochemical potential and the organic molecule Mulliken electronegativity. The formation of such charge-transfer complexes that vary a bulk property such as the polymer chemical potential is considered the phenomenon responsible for the φM change [28]. The interaction of CPs with different organic vapors is still a matter of discussion. Indeed, similar experiments performed with resistive CP-based sensors have reported that the change in electrical resistance occurring at alkyl-substituted polypyrroles and polythiophenes exposed to alcohols, esters, alkenes, and some aromatic

8080_C006.fm Page 511 Tuesday, April 3, 2007 2:15 PM

Organic Transistor Chemical Sensors

511

compounds can be modeled by considering nonspecific molecular interactions of the volatile organic molecules with polymer films. In the case of alcohols, the data suggested a little shape or size selectivity for nonhomologous members of this analyte class [27,30]. Recognition of complex, distinct, and diverse odors, such as cheeses, beers, olive oil, explosives, and pathogenic bacteria, is being addressed by e-noses, which are array-based systems attempting to mimic the mammalian olfactory system [29,36]. Chemiresistors employing different CP sensing layers have been successfully implemented in e-noses, which have an enormous potential to fulfill the real needs of a multifaceted market, from food and beverage quality control to medical diagnostics [27]. However, they do not yet perform at the level required by consumers. In addition, effective miniaturization is still an issue. The fundamental requirement for the e-noses is to produce a pattern of discernibly different responses for different samples. In other words, each sensor composing the array is not required to be highly specific, but it should respond to a broad range of analytes. Moreover, the correlation between responses of different sensors into the array system should be avoided because it decreases the information content of the pattern produced by the array [37]. The use of arrays of gas sensors to quantify the concentration of gasses in mixtures was first proposed in the 1980s, and since then different types of sensors, such as amperometric, piezoelectric, metal-oxide resistors, and MOSFET sensors, have been employed in this technology [38–41].

6.2.3 ORGANIC THIN-FILM TRANSISTOR SENSORS 6.2.3.1 DEVICE STRUCTURE The typical OTFT sensor structure is shown in Figure 6.2.3. It consists of a conductive substrate covered by a thin dielectric film that is interfaced to the organic active layer. The organic active layer is generally a film of a few tens of nanometers of conducting polymers or oligomers, such as regioregular polythiophenes or pentacene molecules. It is deposited by solution processing (such as solution casting, spin coating, and Langmuir–Shäfer or Langmuir–Blodgett techniques) or thermal evaporation, depending on the molecule’s degree of solubility in organic Analyte Source

Drain

Gate Gate dielectric Silicon substrate

FIGURE 6.2.3 Structure of a sensing OTFT. It is a typical bottom gate device where the polycrystalline organic thin film acts at the same time both as active layer and sensing membrane directly exposed to the analyte to be revealed.

8080_C006.fm Page 512 Tuesday, April 3, 2007 2:15 PM

512

Organic Field-Effect Transistors

2.00 Grain surface EB 1.00

100.0 nm 50.0 nm 0.0 nm 0

1.00 μm

0 2.00

FIGURE 6.2.4 AFM micrograph of a pentacene thin film. It shows a granular morphology whose grains have linear dimensions of hundreds of nanometers. A schematic diagram of the potential barriers at grain boundaries of the sensing layer is also shown.

solvents such as chloroform or xylenes. Organic active layers are generally polycrystalline, with a granular morphology characterized by grains having linear dimensions of, at most, hundreds of nanometers and a crystalline-type degree of structural order. Polycrystalline organic active layers are generally described as formed by contiguous grains having a crystalline core and amorphous grain boundaries. An example atomic force micrograph of a polycrystalline pentacene thin film is shown in Figure 6.2.4. Gold source and drain contacts are defined by thermal evaporation through a screen mask directly over the organic active layer, while a contact on the conducting substrate is used to apply the gate bias. OTFT sensors are operated in the common source configuration (i.e., connecting the source contact to ground and biasing the gate and the drain contacts against it). The operation of such a three-terminal device is generally described in analogy to that of inorganic TFTs [42]. In fact, OTFTs show the source-drain current/voltage family of curves, Ids–Vds, characteristics, very similar to those of inorganic TFTs for each different gate bias, Vg, applied. The linear (saturation) region takes place at a Vds bias much lower (higher) than (Vgs–Vt), where Vt is the transistor threshold voltage. OTFT sensors have device structures and detection mechanisms different from those of CHEMFETs. In the case of OTFT sensors, the analyte detection is performed employing a bottom gate device structure where the active layer is directly exposed to the analyte and acts as both transistor channel material and sensing membrane.

6.2.3.2 MULTIPARAMETRIC OTFT SENSORS Bottom gate organic thin-film transistors were proposed as sensing devices soon after the first organic thin-film transistor was reported [2,3,46]. Gate bias variation can be used to improve the sensor performance [21]. This property derives directly from the fact that two distinct conductivity regimes can be established in an OTFT: a bulk or three-dimensional transport regime, at no gate bias (“off” regime), and a

8080_C006.fm Page 513 Tuesday, April 3, 2007 2:15 PM

Organic Transistor Chemical Sensors

513

LUMO Fermi level

Gate dielectric HOMO Gate contact

Trap states

p-type organic semiconductor Vg = 0 (a)

Channel region 2D transport

+ ++ +++ Vg < 0 (b)

FIGURE 6.2.5 Schematic energy band diagram for a p-type TFT: (a) no gate bias; (b) gate voltage negatively biased with respect to the source.

two-dimensional transport occurring at Vg beyond the threshold Vt (“on” regime) (Figure 6.2.5). At no gate bias and at a fixed Vds bias, the OTFT measures Ids variation caused by the interaction and/or permeation of the analyte in the active layer. In this regime, a permeation of the analyte down to the gate dielectric will result in a threedimensional conductivity variation. The response acquired in this case is exactly that of an equivalent chemiresistor. Upon application of a gate bias, on the other hand, a much higher two-dimensional Ids, confined near the gate dielectric/organic interface, is induced. In this regime the “on” Ids change can be recorded and different device parameters, such as Vt and μFET, can also be simultaneously influenced during the exposure to the analyte. Since the two conduction regimes are completely independent, two orthogonal OTFT Ids responses can be recorded and the simultaneous variation of different field-effect parameters can be monitored in the “on” regime. To obtain deeper insight into the sensing mechanism, we need to consider the transport mechanism of the charged species in polycrystalline organic TFT films.

8080_C006.fm Page 514 Tuesday, April 3, 2007 2:15 PM

514

Organic Field-Effect Transistors

Although this is still a matter of studies and discussion, thermally activated transport mechanisms such as the multiple trapping and thermal release (MTR) model, widely used to model transport in a-Si:H TFTs, seem to apply to most polycrystalline OTFTs at room temperature. The organic films are modeled as systems with a narrow delocalized conduction band and a high concentration of localized lower energy electronic states situated in the gap, which act as low-mobility trap states. Traps can be due to impurities, structural defects located in the crystalline grain, or, most importantly, to grain boundaries, as demonstrated by the increase in field-effect mobility with grain size of the polycrystalline organic active layer [47,48]. Charge transport in organic materials is modeled as the result of the contribution of two phenomena: intergrain charge transport and conduction across grain boundaries. Intergrain charge transport depends on the trapping of charges upon interaction with localized energy levels as they move from source to drain. The charges are thermally released by surmounting an activation energy, EA, of several tens of millielectron volts. The mobility generally increases with increasing the gate bias because at low gate bias the induced charges are mostly trapped in low-mobility states. As the bias is increased, the Fermi level at the insulator/organic interface eventually reaches the closest band edge. At this point, the lower energy trap states are all filled and the induced charges are now freer to move. In addition to intergrain transport, charge transport through grain boundaries must be considered. This is limited by the strength of the potential barrier between contiguous grains. The effective mobility across two grains separated by a grain boundary has been described as [48]: 1 1 1 = + μ μ C μ GB

⎛ E ⎞ with: μ GB ∝ exp ⎜− B ⎟ ⎝ kT ⎠

(6.2.2)

where μC is the single crystal mobility and μGB is the mobility across the grain boundary. The potential barrier between the grains, EB, is given by [48]: EB =

e2 nt2 8εrε0 p

(6.2.3)

where εr is the dielectric constant of the semiconductor ε0 is the permittivity of vacuum nt is the surface density of charged traps at the grain boundaries p is the carried density k and T are the Boltzmann constant and temperature, respectively Grain boundary EB values have been found to be of the order of 0.1 V for different p- and n-type organic conjugated systems [49]. Depending on the material properties, either the transport within the grain or across the boundary dominates. For example, thermally evaporated thiophene

8080_C006.fm Page 515 Tuesday, April 3, 2007 2:15 PM

Organic Transistor Chemical Sensors

515

oligomers have been reported to show a transport limited by thermoionic emission over the potential barrier at the grain boundary [50], while in thermally evaporated tetracene, the mobility is dominated by intergrain transport at room temperature, by the transport across grains in the 20- to 150-K range, and by tunneling at very low temperatures [48]. The conduction mechanisms play a crucial role in determining the sensing mechanism of an OTFT sensor. OTFT Vt and μFET depend on the volume density of trapped charges and on the potential barrier between contiguous grains, respectively [48]. Several reactive species cause charge trapping–detrapping processes to occur, enhancing or lowering barriers between grains. Therefore, Vt and μFET can be greatly influenced by the interaction of the transistor active layer with a chemical species, which results in a change of the device drain source on-current and the two-dimensional conductivity. Many active layers such as substituted thiophene-based polymers and oligomers, naphthalenes, copper phthalocyanines, and pentacene have been investigated in OTFT sensors, and different analytes such as alcohols, ketones, thiols, nitriles, esters, and ring compounds have been sensed with these systems [10]. In all the observed cases, the active layer–analyte interaction has been modeled as the analyte molecules being adsorbed, or even trapped, at the surface of the grains. This is supported by the fact that no film swelling has been detected upon exposure, suggesting that the analyte is not able to permeate the crystalline grains [18]. This is reasonable because the grains have a very compact morphology and analyte molecules can easily reach the interface with the gate dielectric through the voids between grains [25]. In addition, grain boundaries play a critical role because the OTFT sensor response increases when the grain size is reduced or channel length L is raised [19,51]. This means that the strength of the interaction with the analyte increases with increasing the grain boundaries exposed to the odor molecule per volume unit. Moreover, upon exposure to analytes, a mass uptake is recorded along with an increase or decrease in Ids [18]. The OTFT’s sensing mechanism can be plausibly depicted as the analyte molecules being trapped at the grains’ boundaries and changing the height of EB and, eventually, the film mobility. Associated charge trapping can have an effect on Vt. A minor effect of doping has been observed in specifically designed systems, and this has resulted in chemical modulation of the off and on source-drain current [25,51].

6.2.3.3 INTERFACE-DEPENDENT OTFT RESPONSES Recent observations have highlighted the great potential of OTFT sensors and are leading to a deeper understanding of the analyte recognition process. It has been found that the responses of OTFTs employing different active layers exhibit an increase or a decrease of the transient Ids current in the presence of alcohol vapors. The occurrence of these two effects strongly depends on the gate-dielectric/organic semiconductor interface [25]. Sensing experiments performed on pentacene, copper phthalocyanine (CuPC), and polythiophene OTFTs, where thermally grown SiO2 or SiO2 coated by an organic glass resin was used as gate dielectric, have shown that exposure to pentanol vapors can lead to an increase or decrease of the transient current, depending on the dielectric material and the organic layer [25].

8080_C006.fm Page 516 Tuesday, April 3, 2007 2:15 PM

516

Organic Field-Effect Transistors

The responses to pentanol vapors of pentacene and CuPC OTFTs on the twogate dielectric are reported in Figure 6.2.6 as an example [25]. The analyte injection and removal are evidenced by the gray shading and each curve in the same graph is measured at a different gate bias. The pentacene on SiO2 and the CuPC on GR/SiO2 OTFT responses (Figure 6.2.6a and b) show an irreversible current decrease at high gate voltages and a moderate current increase at lower Vg, as well as at zero gate bias. Conversely, the pentacene on GR/SiO2 and CuPC on SiO2 OTFT responses, in Figure 6.2.6(c) and (d), show a reversible current increase regardless of the gate bias applied. A current decrease is likely caused by hole trapping at the grain boundaries or at the interface of the grains with the dielectric surface. Trapping reduces the number of charges effective in transport and lowers the mobility of charges that are not trapped. A source-drain current increase, conversely, is ascribable to some kind of doping of the channel material. The charge trapping induced upon exposure to the alcohol analyte is grain-size dependent [19] and is modeled as a surface process [49]. The doping and trapping effects can also coexist as shown by the current increase observed in sensors operated at low (subthreshold) gate biases and the current decrease at higher gate biases (see Figure 6.2.6a and b). This result gives evidence that the trapping process is much more effective when an OTFT device is operated in enhanced mode, supporting the model used to describe the charge conduction in polycrystalline CP-based OTFTs. These experiments all refer to long-channel OTFTs. Short-channel pentacene OTFTs on SiO2 with channel lengths comparable to the active material nanograin size show only a doping process, which is completely reversible [51]. The trapping effect greatly affects the OTFT conduction mechanism and renders OTFT sensors potentially very sensitive devices. The occurrence of trapping and doping effects in OTFT sensors also seems to be related to the nature of the interface. While pentacene, alkyl-thiophenes, and naphthalenes are molecules with low dipole moments, CuPC consists of highly polar sites such as those associated to the copper core, which can act as coordinating sites. OTFTs with organic and gate-dielectric materials with markedly different dipole moments — namely, the highly hydrophobic pentacene deposited on the highly hydrophilic SiO2 interface and the polar CuPC deposited on SiO2 coated with glass resin — show an irreversible current decrease at high gate voltages likely due to trapping process (see Figure 6.2.6). Conversely, OTFTs formed by homologous interfaces (namely, polar/polar or nonpolar/nonpolar) lead to a reversible current increase due to doping effects (see Figure 6.2.6). These recent results show, indeed, that trapping is favored at high gate voltages when materials with different dipole moments are interfaced, probably because the low level of matching favors the generation of trap states at the interface. The nature of the interface seems, therefore, to drive the physical process of the sensing transduction mechanism.

6.2.3.4 GATE-INDUCED RESPONSE REPEATABILITY AND ENHANCEMENT The electrical behavior of most CP films is not very stable. In this respect, the OTFT sensor configuration can be very convenient, especially when operated in pulse

8080_C006.fm Page 517 Tuesday, April 3, 2007 2:15 PM

517

Pentacene/SiO2

20 18 16 14 12 10 8 6 4 2 0

CuPC/GR Vg = –70V

Vg = –40V 0.3 Ids(μA)

Ids (μA)

Organic Transistor Chemical Sensors

Vg = –30V

0.2 Vg = –60V 0.1 Vg = –30V

Vg = 0V 0

20

40

60

0.0

80 100 120

0

20

t (sec)

40 60 t (sec)

Pentacene/GR

CuPC/SiO2

0.40

1.6

0.35

1.4 Ids(μA)

1.0 0.8 0.6 0.4

Vg = –60V

0.2

Vg = –20V 0

20

40 60 t (sec) (c)

80

Vg = –40V

0.30

Vg= –70V

1.2 Ids(μA)

100

(b)

(a)

0.0

80

0.25 0.20 0.15

Vg = –30V

0.10 Vg = 0V

0.05 100

0.00

0

20

40

60 80 100 120 t (sec) (d)

FIGURE 6.2.6 Signal (solid lines) and base lines (dotted lines) transient currents showing the responses (highlighted area) of (a) pentacene on SiO2; (b) CuPC on GR/SiO2; (c) pentacene on GR/SiO2; and (d) CuPC on SiO2 OTFTs exposed to a saturated atmosphere of 1-pentanol vapors. Vds is kept at –40 V for the SiO2 dielectric-based OTFTs and at –70 V for those comprising the GR/SiO2 dielectric film. Different gate biases are applied.

mode. It has been demonstrated that with an OTFT sensor it is possible to obtain an almost complete recovery after analyte exposure by a strategic use of the gate potential applied to the gate electrode, allowing the device to operate at room temperature [45]. The electrical responses of a sensing OTFT are evaluated on the basis of a fixed Vds and Vg bias point in the saturated region of the Ids/Vds transfer curve. The transient source-drain current, Ids, in the absence (baseline) and presence (signal) of the analyte is measured with the device operating either in DC or pulse mode [5,45]. Under DC, at the Vds and Vg bias of the device, the Ids current measured during analyte delivery is the sensor signal. This response is usually fairly reversible due to the low strength of CP/organic analyte interactions. Nevertheless, when the

8080_C006.fm Page 518 Tuesday, April 3, 2007 2:15 PM

518

Organic Field-Effect Transistors

concentration of the analyte is close to the saturated vapor pressure, irreversibility is much more probable. Figure 6.2.7(a) shows an example of OTFT Ids transient responses obtained for a CuPC active layer (on GR) at different gate bias upon exposure to pentanol [25]. The shaded region indicates analyte delivery, here at its maximum concentration at room temperature. For each Vg value a different transient response is recorded and, as shown in Figure 6.2.7, none of these responses is reversible. Indeed, no recovery of the baseline signal can be observed after removing the analyte. However, applying a reverse gate bias completely resets the device, restoring the original baseline. Such a reverse gate voltage pulse apparently acts as an “untrapping bootstrap.” Just as it is possible to fill the trap states by applying a gate potential up to the threshold voltage, Vt, it is also possible to release charges trapped in the CP due to interaction with the analyte by applying an opposite gate potential. Because of this, OTFT sensors exhibit extremely good response repeatability, as shown in Figure 6.2.8 (See color insert following page 468). It has been reported that many oligothiophene-based OTFTs show responses reproducible within 2% for 70 subsequent cycles of sensing measurement when exposed to several analytes [45]. It is important to stress that the best results are obtained in the pulse mode because continuous biasing is not generally well tolerated by organic devices and small analyte response can be masked by current drifts. The level of performance currently achieved with OTFT sensors exhibits responses of 5–15% for analyte concentrations of 10–100 ppm and response times of 3–5 s. All this makes OTFT sensors a promising technology able to compete with chemiresistors in sensor arrays. With a FET sensor configuration it is also possible to enhance the response intensity simply by acting on the OTFT gate electrode. Ids responses, measured as ΔI between Ianalyte and the baseline, increase when the OTFT operates in the accumulation mode (Figure 6.2.7a and b). This property of OTFT sensors has been systematically observed on very different systems, such as pentacene, copper phthalocyanine, thiophene oligomers, and OTFT biosensors [19,23,25]. A dramatic example is the response enhancement of over two orders of magnitude obtained for a copper phthalocyanine-based OTFT sensor exposed to 1-pentanol (Figure 6.2.7b) [25]. This demonstrates that the response of an OTFT sensor can be modulated with the gate bias, corroborating the multiparametric nature of the OTFT sensor responses. The response enhancement has to be attributed to the possibility of operating OTFT sensors in enhanced mode as well as to the fact that polycrystalline CP thin films act as active layer and sensing membrane in these sensors. In the “on” regime the potential applied to the gate electrode controls the density of charges accumulated in the organic layer at the interface with the dielectric. This in turn determines the current flowing through the device, with higher gate biases giving higher source-drain currents. When the OTFT is exposed to the analyte, the analyte permeates the polycrystalline active layer through the voids between the grains and reaches the interface with the dielectric, where it interacts with the charge carriers and influences Ids. For higher applied gate biases (and hence increased charges in the channel and Ids flowing through the device), one can expect an

8080_C006.fm Page 519 Tuesday, April 3, 2007 2:15 PM

Organic Transistor Chemical Sensors

519

Ids (μA)

0.3

0.2 Vg = –70V

0.1 Vg = –60V Vg = –50V

0.0 0

20

40 60 t (sec)

80

(a)

Enhancement factor ~ 150

∆Ids (nA)

160 120 80 40 0 –20

–30

–40 –50 Vg (V)

–60

–70

(b)

FIGURE 6.2.7 (a) Source-drain transient currents at different gate biases for a CuPC-based OTFT exposed to a saturated concentration of pentanol; (b) ΔIds responses for the same device at fixed Vds values (–70 V) and different gate biases.

increase in the effect of the interaction between the analyte and the charge carriers accumulated at the interface. OTFT response repeatability and enhancement, achieved with the gate bias, are apparent advantages over traditional chemiresistor sensors, whose response is a single parameter without any possibility to be modified. A resistor is a two-terminal device giving a resistive response, which is completely determined by the bulk properties of the material used. It is of particular interest to investigate whether, besides a response enhancement, OTFTs are able to deliver a sensitivity enhancement; no clear evidence of this phenomenon has been given and it is not yet well understood which parameters control this important analytical property in OTFT sensors.

8080_C006.fm Page 520 Tuesday, April 3, 2007 2:15 PM

520

Organic Field-Effect Transistors

5 VG(V) 0 Analyte delivers trigger pulse

–5 (V) 5 0 Ids (μA) –1 0 300 400 500 600 Time (sec)

700

(a) dDDα6T/1-hexanol, Vds = –5V Ianalyte (μA)

–1.0

–0.5

0.0

0

10 Time (sec) (b)

20

Vg = –5V Vg = –4V Vg = –3V Vg = –2V Vg = –1V Vg = 0V

Cycle number

dDDα6T / 1-hexanol, on current 0.05 0.00 –0.05 –0.10 –0.15

60 40 20 0

0

10 Time (sec)

20

(c)

FIGURE 6.2.8 (See color insert following page 468.) (a) One-cycle OTFT sensors testing pulse program; (b) source-drain transient currents at different gate biases; (c) color-coded response plot at Vg = –5 V for 70 cycles. (From Torsi, L. and Dodabalapur, A., Anal. Chem., 77, 380A, 2005. With ACS permission.)

8080_C006.fm Page 521 Tuesday, April 3, 2007 2:15 PM

Organic Transistor Chemical Sensors

6.2.3.5 SELECTIVITY

OF AN

521

OTFT

OTFT sensors use organic CP sensing layers that, due to their adjustable chemical properties, are able to chemically and biologically specifically respond to target analytes. CP films exhibit very weak selectivity toward organic molecules such as alcohols or alkenes. Nevertheless, their specificity can be modulated by substituting the CP backbones with properly chosen functional groups [30]. This induces a partitioning of target analytes very much like a stationary phase in a chromatographic column [52]. The sensing mechanism is thought to involve surface-mediated weak interactions between the functionalized polymers and the analyte. The adsorption of the analyte on the organic thin-film grains appears to be modulated by the degree of chemical affinity between the organic molecule and the polymer functional groups. The grain boundaries play a crucial role in determining the responses of OTFT sensors and, together with functionalization of the active layer, represent a key driving force in tailoring the sensor’s selectivity. The strategy of using properly functionalized organic semiconductors can be exploited to obtain specific recognition involving, in this case, covalent bonding to biospecific side groups. Many studies have been conducted recently to investigate the selectivity potential of OTFTs. The first experiments involved glucose and lactic acid detection by an α-6T TFT mediated by no specific receptors [53]. Subsequently, a glucose sensor was proposed [6] involving an organic/inorganic TFT comprising a proton-sensitive outermost layer where the glucose oxidase (GOx) enzymes, which catalyze the reaction of glucose in the presence of oxygen to produce hydrogen peroxide, were used as specific recognition elements. Poly(3,4-ethylene dioxythiophene)poly(styrene sulfonic acid) (PEDOTPSS) thin-film transistors were also proposed for glucose sensing in a neutral pH buffer solution [23]. Carbon nanotube nanoscale transistors have been proposed to detect protein binding via a biotin-streptavidin model system [54]. Recently, alkoxy and alkyl substitution of polythiophenes has been demonstrated to confer chemical recognition properties to an OTFT device [18]. All this suggests that a proper chemical or biological functionalization of the organic material forming the active layer can result in a film that, through the surface of its grains, is capable of controlling the partition coefficient of a gaseous or even liquid analyte into the active layer, resulting in moderately or even highly selective sensors. A recent collection of review papers by different authors on this topic can be found in references 55 through 61. The crucial role played by the gate dielectric/organic semiconductor interface in determining the transduction mechanism of an OTFT sensor suggests that it is in principle possible to tailor the recognition properties of OTFTs also by means of a proper choice of the dielectric/organic semiconductor interface. The ability to design selectivity properties of OTFT sensors, in addition to the repeatability and gate bias enhancement of their response, will make OTFT sensors a much more flexible option than existing portable sensor systems for a large number of sensing applications.

8080_C006.fm Page 522 Tuesday, April 3, 2007 2:15 PM

522

6.2.3.6 SCALING BEHAVIOR

Organic Field-Effect Transistors OF

SENSING RESPONSES

In a polycrystalline sensor, it is expected that grain boundaries will play an important role in the sensing process. Indeed, the first studies of this aspect in OTFT sensors confirmed that as the grain size gets smaller, the relative response of the sensing FET is enhanced [19]. This was attributed to enhanced grain boundary trapping mediated by analyte molecules and the consequent reduction in drain current on account of the twin effects of reduced mobility and shifted threshold. More recent work has shown that the scaling behavior is more complex and that, as the channel length is reduced to dimensions comparable to the grain size, delivery of the analyte can result in an increase in drain current in material combinations that ordinarily result in a drain current decrease in large channel-length devices (where the channel length is significantly greater than the grain size) [51,55]. This behavior is linked to the transport mechanisms in an organic FET. In most short channel-length organic FETs, injection limitations of the contact electrode greatly influence the transport properties. In injection-limited organic devices (such as FETs and light-emitting diodes), the provision of appropriate interface dipoles between the metal and the semiconductor often results in a dramatic increase in injection current for a given voltage bias condition. In very short channellength OTFT sensors, polar analytes can play the role of interface dipoles and improve injection, resulting in current increases. This is described later for the case of pentacene transistors with 1-pentanol as the analyte. The drain current response of pentacene devices with a series of channel length and varied grain sizes (operated in saturation region) upon exposure to the saturated vapor of 1-pentanol is shown in Figure 6.2.9(a). While the long channel-length devices all exhibited a decrease in current upon delivery of the analyte, the small channel-length devices showed an increase, sometimes by a factor of more than five. There are two mechanisms influencing sensor behavior; one causes a decrease in current (dominant in large L devices) and the other causes an increase (dominant in small L devices). The crossover of response behavior depends on grain size, occurring in the interval of channel length between 150 and 450 nm for ~80-nm grain size. Under the same conditions, when the average grain size of pentacene is increased to 250 nm, the sensors exhibit the crossover behaviors at larger channel lengths (from 450 nm to 1 μm), as shown in Figure 6.2.9(b). Figure 6.2.10 shows the scanning electron microscopy (SEM) image (after measurements) of the same channel length (150 nm) with different grain sizes of pentacene layer. Figure 6.2.11(a) and 6.2.11(b) are the sensing responses of long-channel devices with pentacene grain size of 140 nm and 1 μm, respectively. For all devices with channel lengths of 2 μm or greater, Ids manifested decreasing responses upon analyte delivery. The amplitudes of decreasing signal for 2-μm channels were smaller than those of longer channels. This effect is stronger with larger pentacene grains (Figure 6.2.11b). These results are consistent with the reported work for sensing effects dependent on organic grain sizes and channel lengths in large scale [15,19]. The sensing responses shown in Figures 6.2.9 and 6.2.11 are reproducible for different devices with the same channel lengths and grain sizes, indicating that the response of pentacene transistors to the 1-pentanol vapor changes from decreas-

8080_C006.fm Page 523 Tuesday, April 3, 2007 2:15 PM

Organic Transistor Chemical Sensors

Normalized drain current

3

Grain ~ 80 nm

60 nm 150 nm 450 nm

2

1 um

1

0 0

Normalized drain current

523

10

20

30 40 Time (sec) (a)

50

60

60 nm 150 nm 450 nm 1 um

3

2

1 Grain ~ 250 nm 0 0

10

20

30

40

50

60

Time (sec) (b)

FIGURE 6.2.9 The sensing effects of pentacene transistors upon exposure to 1-pentanol. (a) Sensing data of Ids (normalized to that measured just before the analyte took effect) for 80nm pentacene grain size and different nanoscale channel lengths (same W/L = 10), measured at Vg = Vds = Vside = –2.5 V (two side guards were kept at the same potential as the drain); v (analyte flux) = 45 ml/min; d (distance from syringe nozzle to device) = 2 mm. (b) Sensing data of normalized Ids for 250-nm pentacene grain size, measured at the same conditions as (a). (From Wang, L. et al., Appl. Phys. Lett., 85, 6386, 2004. With permission. Copyright 2004, American Institute of Physics.)

ing Ids to increasing Ids when the channel length shrinks from micron to 100 nm. A crossover happening in a transition interval of channel length is related to the grain sizes of pentacene. The data in Figures 6.2.9, 6.2.10, and 6.2.11 illustrate the interplay between dipole mediated injection effects and grain boundary trapping. The first effect is dominant in short channel-length devices and the second effect in longer channellength devices. There is an intermediate range of channel lengths where the two effects cancel each other out. This regime must be avoided in the design of practical OTFT sensors. The incorporation of receptors to enhance selectivity/sensitivity will be easier in large channel-length devices because receptors are generally responsible

8080_C006.fm Page 524 Tuesday, April 3, 2007 2:15 PM

524

Organic Field-Effect Transistors

(a)

(b)

FIGURE 6.2.10 SEM image taken after sensing measurements of a 150-nm channel with different average pentacene grain sizes of 80 nm (a) and 250 nm (b); scale bar = 400 nm. The grains appearing in the figure are pentacene. (From Wang, L. et al., Appl. Phys. Lett., 85, 6386, 2004. With permission. Copyright 2004, American Institute of Physics.)

for trapping events. The reproducibility and reliability were also found to be greater for larger geometry sensor devices.

6.2.4 APPLICATIONS Even though OTFT sensors are not yet commercialized, their implementation in sensing systems has already been explored. Sensing organic circuits have been demonstrated recently that exploit the CMOS (complementary metal oxide semiconductor) implementation of OTFTs [5]. These are circuits using both n-MOS (negative polarity) and p-MOS (positive polarity) devices. Since only one of the device types in a CMOS circuit element is “on” at any given time, CMOS chips require less power than chips using just one type of transistor. There are many advantages related to using sensor circuits, such as organic CMOS-based circuits, instead of discrete transistor based sensors. One of the main OTFT sensor drawbacks is the temporal change in the DC current caused by field-induced carriers falling into energetically deeper localized states with a corresponding decrease in mobility. For this reason, it is difficult to extract the change in current caused by an odorant, especially when the change is small. One solution to this is to cycle the gate potential (with respect to the source), which induces a recovery of the drain current. This happens naturally in a ring oscillator in which an odd number of inverters are connected so that the output of one stage is connected to the input of the next stage and the output of the final stage is fed back to the input of the first. Each transistor gate is cycled between supply voltage and ground at a high frequency. This results in a flat frequency response as a function of time since each transistor has the opportunity to recover during one half of an electrical cycle. In a report on organic CMOS-sensing ring oscillator circuits, didodecyl α-sexithiophene (DDα6T) was used as the p-channel material and hexadecafluorocopperphthalocyanine (F16CuPc) as the n-channel material [5]. The analytes used in the evaluation of ring oscillators include octanol and allyl propionate. F16CuPc has a negligible response to both these odorants, whereas the DDα6T is fairly responsive, with the drain current falling in response to both odorants. In this CMOS sensor, upon a 5-sec exposure to octanol, the oscillation

8080_C006.fm Page 525 Tuesday, April 3, 2007 2:15 PM

Organic Transistor Chemical Sensors

525

1.2

Grain ~ 140 nm

Normalized drain current

1.0 0.8 0.6

2 um 4 um 6 um 12 um 36 um

0.4 0.2 0.0

0

10

20

30 40 Time (sec) (a)

50

60

1.2

Normalized drain current

Grain ~ 1 um 1.0 0.8 0.6 2 um 4 um 6 um 12 um 36 um

0.4 0.2 0.0 0

10

20

30 40 Time (sec)

50

60

(b)

FIGURE 6.2.11 Sensing data of pentacene transistors upon exposure to 1-pentanol: normalized Ids under the condition of Vg = Vds = –25 V, v = 45 ml/min; and d = 2 mm for different microscale channel lengths, with average pentacene grain size of 140 nm (a) and 1 μm (b). (From Wang, L. et al., Appl. Phys. Lett., 85, 6386, 2004. With permission. Copyright 2004, American Institute of Physics.)

frequency is lowered by about 40% and the voltage level of the oscillation is reduced by about 25%. There are several other noteworthy features regarding the data. The magnitude of change in the ring oscillator frequency (>40%) exceeds the change in current of the discrete device (~15%), supporting the view that circuit sensors that possess multiple sensor components can have superior performance in comparison with discrete device sensors. Another feature is the speedier recovery of the ring oscillator sensor compared to the discrete device sensor. In fact, in order to fully restore the original characteristics of the discrete device sensor, it is necessary to

8080_C006.fm Page 526 Tuesday, April 3, 2007 2:15 PM

526

Organic Field-Effect Transistors

subject it to a reverse gate bias for about a minute. Recently, Someya’s group reported the use of OTFT arrays as flexible pressure sensors in artificial-skin applications [13]. Pentacene TFTs were integrated with a composite rubber layer, whose conductivity changes as pressure is applied into a 16 × 16 sensor matrix that can operate even when the substrate is wrapped around a cylindrical bar.

REFERENCES 1. Laurs, H. and Heiland, G., Electrical and optical properties of phthalocyanine films, Thin Solid Films, 149, 129, 1987. 2. Tsumura, A., Koezuka, H., and Ando, T., Macromolecular electronic device: Fieldeffect transistor with a polythiophene thin film, Appl. Phys. Lett., 49, 1210, 1986. 3. Assadi, A. et al., Determination of field-effect mobility of poly(3-hexylthiophene) upon exposure to ammonia gas, Synth. Met., 37, 123, 1990. 4. Ohmori, Y. et al., Gas-sensitive Schottky gated field effect transistors utilizing poly(3alkylthiophene) films, Jpn. J. Appl. Phys., Part 2, 30, L1247, 1991. 5. Crone, B.K. et al., Organic oscillator and adaptive amplifier circuits for chemical vapor sensing, J. Appl. Phys., 91, 10140, 2002. 6. Bartic, C., Campitelli, A., and Borghs, G., Field-effect detection of chemical species with hybrid organic/inorganic transistors, Appl. Phys. Lett., 82, 475, 2003. 7. Bartic, C. et al., Monitoring pH with organic-based field-effect transistors, Sens. Actuators, B, 83, 115, 2002. 8. Bouvet, M. et al., Phthalocyanine-based field-effect transistor as ozone sensor, Sens. Actuators, B, 73, 63, 2001. 9. Bouvet, M. et al., Detection and titration of ozone using metallophthalocyanine based field effect transistors, Sens. Actuators B., 72, 86, 2001. 10. Crone, B. et al., Electronic sensing of vapors with organic transistors, Appl. Phys. Lett., 78, 2229, 2001. 11. Hu, W. et al., The gas sensitivity of a metal-insulator-semiconductor field-effecttransistor based on Langmuir–Blodgett films of a new asymmetrically substituted phthalocyanine, Thin Solid Films, 360, 256, 2000. 12. Nilsson, D. et al., An all-organic sensor–transistor based on a novel electrochemical transducer concept printed electrochemical sensors on paper, Sens. Actuators B, 86, 193, 2002. 13. Someya, T. et al., A large-area, flexible pressure sensor matrix with organic fieldeffect transistors for artificial skin applications, Proc. Natl. Acad. Sci. U.S.A., 101, 9966, 2004. 14. Someya, T. et al., Alcohol vapor sensors based on single-walled carbon nanotube field effect transistors, Nano Lett., 3, 877, 2003. 15. Someya, T. et al, Vapor sensing with α,ω-dihexylquarterthiophene field-effect transistors: The role of grain boundaries, Appl. Phys. Lett., 81, 3079, 2002. 16. Tanese, M.C. et al., Poly(phenyleneethynylene) polymers bearing glucose substituents as promising active layers in enantioselective chemiresistors, Sens. Actuators B, 100, 17, 2004. 17. Torsi, L. et al., Alkoxy-substituted polyterthiophene thin-film-transistors as alcohol sensors, Sens. Actuators B, 98, 204, 2004. 18. Torsi, L. et al., Side-chain role in chemically sensing conducting polymer field-effect transistors, J. Phys. Chem B, 107, 7589, 2003.

8080_C006.fm Page 527 Tuesday, April 3, 2007 2:15 PM

Organic Transistor Chemical Sensors

527

19. Torsi, L. et al., Correlation between oligothiophene thin film transistor morphology and vapor responses, J. Phys. Chem. B, 106, 12563, 2002. 20. Torsi, L. et al., NTCDA organic thin-film transistor as humidity sensor: Weaknesses and strengths, Sens. Actuators B, 77, 7, 2001. 21. Torsi, L. et al., Multiparameter gas sensors based on organic thin-film-transistors, Sens. Actuators B, 67, 312, 2000. 22. Nilsson, D., Kugler, T., Svensson, P.O., and Berggren, M., An all-organic sensor–transistor based on a novel electrochemical transducer concept printed electrochemical sensors on paper, Sens. Acuators B, 86, 193, 2002. 23. Zhu, Z.-T. et al., A simple poly(3,4-ethylene dioxythiophene)/poly(styrene sulfonic acid) transistor for glucose sensing at neutral pH, Chem. Commun., 13, 1556, 2004. 24. Zhu, Z-T. et al., Humidity sensors based on pentacene thin-film transistors, Appl. Phys. Lett., 81, 4643, 2002. 25. Tanese, M.C. et al., Interface and gate bias dependent response of sensing organic thin-film transistors, Biosens. Bioelectron., 21, 782, 2005. 26. Wilson, D.M. et al., Chemical sensors for portable, handheld field instruments, IEEE Sens. J., 1, 256, 2001. 27. Persaud, K.C., Polymers for chemical sensing, Mater. Today, 8, 38, 2005. 28. Janata, J. and Josowicz, M., Conducting polymers in electronic chemical sensors, Nat. Mater., 2, 19, 2003. 29. Gardner, J.W. and Bartlett, P.N., Electronic noses: Principles and application, Oxford Science Publication, 1999. 30. Bissell, R.A. et al., The influence of nonspecific molecular partitioning of analytes on the electrical responses of conducting organic polymer gas sensors, Phys. Chem. Chem. Phys., 2002, 4, 3482. 31. Bergveld, P., Development, operation and application of the ion-sensitive field effect transistor as a tool for electrophysiology, IEEE Trans. Biomed. Eng., BME-19, 342, 1972. 32. Lloyd Spetz, A. and Savage, S., Advances in FET chemical gas sensors, in Recent major advances in SiC, ed. W.J. Choyke, H. Matsunami, and G. Pensl, Springer, Berlin, 2003, 879–906. 33. Janata, J. and Josowicz, M., Chemical modulation of work function as a transduction mechanism for chemical sensors, Acc. Chem. Res., 31, 241, 1998. 34. Bergveld, P., Hendrikse, J., and Olthuis, W., Theory and application of the material work function for chemical sensors based on the field effect principle, Meas. Sci. Technol., 9, 1801, 1998. 35. Bergveld, P., Thirty years of ISFETOLOGY: What happened in the past 30 years and what may happen in the next 30 years, Sens. Actuators B, 88, 1, 2003. 36. Persaud, K.C. and Dodd, G.H., Analysis of discrimination mechanisms in the mammalian olfactory system using a model nose, Nature, 299, 352, 1982. 37. Gardner, J.W. and Bartlett, P.N., Performance definition and standardization of electronic noses, Sens. Actuators B, 33, 60, 1996. 38. Carey, W.P. et al., Selection of adsorbates for chemical sensor arrays by pattern recognition, Anal. Chem., 58, 149, 1986. 39. Ballantine, D.S. et al., Correlation of surface acoustic wave device coating responses with solubility properties and chemical structure using pattern recognition, Anal. Chem., 58, 3058, 1986. 40. Weimar, U. et al., Pattern recognition methods for gas mixture analysis: Application to sensor arrays based upon SnO2, Sens. Actuators B, 1, 93, 1990.

8080_C006.fm Page 528 Tuesday, April 3, 2007 2:15 PM

528

Organic Field-Effect Transistors

41. Sundgren, H. et al., Evaluation of a multiple gas mixture with a simple MOSFET gas sensor array and pattern recognition, Sens. Actuators B, 2, 115, 1990. 42. Sze, S.M., Physics of semiconductor devices, 2nd ed., John Wiley & Sons, New York, 1981. 43. Horowitz, G., Organic field effect transistors, Adv. Mater., 10, 365, 1998. 44. Dimitrakopoulos, C.D. and Malenfant, P.R.L., Organic thin film transistor for large area electronics, Adv. Mater., 14, 99, 2002. 45. Torsi, L. and Dodabalapur, A., Organic thin-film transistors as plastic sensors, Anal. Chem., 77, 380A, 2005. 46. Guillaud, G. et al., Field-effect transistors based on intrinsic molecular semiconductors, Chem. Phys. Lett., 167, 503, 1990. 47. Horowitz, G. and Hajlaoui, M.E., Grain size dependent mobility in polycrystalline organic field-effect transistors, Synth. Met., 122, 185, 2001. 48. Powell, M.J., Analysis of field-effect-conductance measurements on amorphous semiconductors, Philos. Mag. A: Phys. Condens. Matter: Struct., Defects, Mech. Prop., 43, 93, 1981. 49. Chesterfield, R.J. et al., Variable temperature film and contact resistance measurements on operating n-channel organic thin film transistors, J. Appl. Phys., 95, 6396, 2004. 50. Kelley, T.W. and Frisbie, C.D.J., Gate voltage dependent resistance of a single organic semiconductor grain boundary, J. Phys. Chem. B, 105, 4538, 2001. 51. Wang, L., Fine, D., and Dodabalapur, A., Nanoscale chemical sensor based on organic thin-film transistor, Appl. Phys. Lett., 85, 6386, 2004 52. Hierlemann, A. et al., Conferring selectivity to chemical sensors via polymer sidechain selection: Thermodynamics of vapor sorption by a set of polysiloxanes on thickness-shear mode resonators, Anal. Chem., 72, 3696, 2000. 53. Someya, T. et al., Integration and response of organic electronics with aqueous microfluidics, Langmuir, 18, 5299, 2002. 54. Star, A. et al., Electronic detection of specific protein binding using nanotube FET devices, Nano Lett., 3, 459, 2003. 55. Wang, L., Fine, D., Torsi, L., and Dodabalapur, A., Nanoscale organic and polymeric field-effect transistors as chemical sensors, Anal. Bioanal. Chem., 384, 310, 2006. 56. Gruner, G., Carbon nanotube transistors for biosensing applications, Anal. Bioanal. Chem., 384, 322, 2006. 57. Locklin, J. and Bao, Z., Effect of morphology on organic thin film transistor sensors, Anal. Bioanal. Chem., 384, 336, 2006. 58. Mabeck, J.T. and Malliaras, G., Chemical and biological sensors based on organic thin-film transistors, Anal. Bioanal. Chem., 384, 343, 2006. 59. Bartic, C. and Borghs, G., Organic thin-film transistors as transducers in (bio)analytical applications, Anal. Bioanal. Chem., 384, 354, 2006. 60. Bouvet, M., Phthalocyanine-based filed-effect transistors as gas sensors, Anal. Bioanal. Chem., 384, 366, 2006. 61. Iba, S. et al., Use of laser drilling in the manufacture of organic inverter circuits, Anal. Bioanal. Chem., 384, 374, 2006.

8080_C006.fm Page 529 Tuesday, April 3, 2007 2:15 PM

6.3

Flexible, Large-Area e-Skins

Takao Someya, Takayasu Sakurai, and Tsuyoshi Sekitani CONTENTS 6.3.1 Introduction................................................................................................529 6.3.2 Flexible Pressure Sensors ..........................................................................530 6.3.2.1 Device Manufacturing.................................................................531 6.3.2.2 Device Performance ....................................................................533 6.3.3 Integrated Circuits......................................................................................533 6.3.4 Stretchable e-Skins ....................................................................................535 6.3.4.1 Device Manufacturing.................................................................536 6.3.4.2 Device Performance ....................................................................537 6.3.5 Flexible Thermal Sensors ..........................................................................540 6.3.5.1 Device Manufacturing.................................................................540 6.3.5.2 Device Performance ....................................................................540 6.3.5.3 Discussion....................................................................................544 6.3.6 Bending Tests.............................................................................................544 6.3.7 Issues and Future Prospects.......................................................................547 6.3.8 Summary ....................................................................................................548 References..............................................................................................................549

6.3.1 INTRODUCTION Recent studies on organic field-effect transistors (FETs) have been motivated by two major applications. The first one is flexible displays that include paper-like displays or e-papers [1,2]. Since organic transistors can be manufactured directly onto plastic films at low (ambient) temperature, these are mechanically flexible, lightweight, and thin. Organic transistors are expected to play a key role in realizing flexible, lightweight displays. Another promising application includes radio frequency identification (RFID) tags [3,4]. The manufacturing costs of organic transistor circuits would be low if they could be fabricated using printing technologies and/or the roll-to-roll process. This feature is suitable for implementation of RFID tags on packaging. Furthermore, organic FET-based chemical sensors have recently emerged and these

529

8080_C006.fm Page 530 Tuesday, April 3, 2007 2:15 PM

530

Organic Field-Effect Transistors

FIGURE 6.3.1 A flexible, large-area pressure sensor. Organic transistor active matrix is formed on a plastic film and integrated with pressure-sensitive rubber.

are proposed to realize artificial noses and tongues [5–9]. These three are reviewed in great detail in Sections 6.1, 6.2, and 6.4 of this book. As one of the new applications, we have recently demonstrated a large-area, flexible sensor (Figure 6.3.1), where organic transistor active matrices are used to read out data from the sensor arrays [10]. The example of organic transistor-based large-area sensors is a pressure sensor matrix. The new pressure sensor could be ideal for electronic artificial skin (e-skin) applications for future generations of robots. In this section, we describe recent progress, issues, and future prospects of e-skins based on organic transistors.

6.3.2 FLEXIBLE PRESSURE SENSORS Skin sensitivity is very important for next-generation robots to work in daily life for home-care and entertainment purposes [11]. Relatively little progress has been made, however, in the field of touch recognition compared to the areas of sight and voice recognition. This is mainly because good e-skins with a large area and mechanical flexibility are not yet available. The fabrication of e-skins consisting of thousands of pressure sensors requires a flexible switching matrix that cannot be realized with present silicon-based electronics. Integration of organic transistors and rubber pressure sensors, both of which can be produced by low-cost processing technology (like large-area printing technology), will provide an ideal solution to realize a practical e-skin whose feasibility has been demonstrated recently [11]. In particular, pressure images have been taken by flexible active matrices with organic transistors whose mobility is 1.4 cm2/Vs. This value is comparable to amorphous silicon on glass (~1 cm2/Vs) and beyond that obtained on plastic films (0.4–0.6 cm2/Vs). The maximum size of a sensing area is 8 × 8 cm2 and it contains a 32 × 32 array of pressure sensors. The periodicity is 2.54 mm, which corresponds to 10 dots per inch (dpi). The manufactured large-area pressure

8080_C006.fm Page 531 Tuesday, April 3, 2007 2:15 PM

Flexible, Large-Area e-Skins

531

FIGURE 6.3.2 An image of electronic artificial skin attached on the robot surface. A plastic film with organic transistors, a pressure-sensitive rubber sheet, and a plastic film with top electrodes are laminated together to form a large-area pressure sensor. Some parts of films are removed intentionally to show the inner structures.

sensor sheets are mechanically flexible and therefore can be wrapped around robot fingers, as shown in Figure 6.3.2.

6.3.2.1 DEVICE MANUFACTURING The device is manufactured by laminating three different functional films: (1) a base plastic film with organic transistors; (2) a pressure-sensitive rubber sheet; and (3) a film suspending copper electrodes for power supply. One sensor cell (sensel) consists of one organic transistor and one pressure sensor. The device structure is schematically shown in Figure 6.3.3(a). The base film (substrate) is composed of an ultrahigh heat-resistant polyethylenenaphthalate (PEN) polymer. The first step is to drill holes though the base film with a CO2 laser. On both sides, electrode patterns consisting of 150-nm thick gold layer (with 5-nm thick chromium adhesion layer) are deposited by vacuum deposition with shadow masks. One side works as the gate electrodes of transistors, while the other works as electrodes of pressure sensors. Next, polyimide precursors are spin-coated and cured at 180°C to form 500-nm thick gate dielectric layers [12]. Spots of polyimide gate

8080_C006.fm Page 532 Tuesday, April 3, 2007 2:15 PM

532

Organic Field-Effect Transistors

Copper electrode Pressure-sensitive rubber Pentacene Source

Gate insulator

Drain

Via hole

(a) Gate

Through hole

(b)

FIGURE 6.3.3 (a) A cross-sectional view of the device structure and (b) the magnified image of one sensor cell. The device is formed by laminating four different functional films. The pressure sensors are made with sandwiching a pressure-sensitive rubber sheet between two electrodes. The one of those two electrodes is connected to the transistors via holes.

dielectric layers are removed by the laser to make contact via holes. A 50-nm thick pentacene is deposited to form a channel layer, and a 50-nm thick gold layer is evaporated using shadow masks to form the source and drain electrodes of the transistors. The channel length L and width W are 120 μm and 2.8 mm, respectively. In the last step, a pressure-sensitive rubber sheet and a copper electrode suspended by a polyimide film are laminated on the bottom side of the base film to integrate pressure sensors with transistors. The pressure-sensitive flexible layer is a 0.5-mm thick polydimethylsiloxane (PDMS) containing electrically conductive graphite particles. The resistance of the rubber sheet changes when pressure is applied on the bottom side of the device. Since no patterning is needed on the rubber sheet and copper electrodes, no alignment is required for this last lamination process; it would be very inexpensive to implement on a large scale. Figure 6.3.3(b) is a picture of a single sensor cell (sensel).

8080_C006.fm Page 533 Tuesday, April 3, 2007 2:15 PM

Flexible, Large-Area e-Skins

533

–10

Drain current IDS (μA)

–8

Transistor

VBL

30kPa

VWL –6

Sensor

20

VDD

–4

10 –2 0 0 5

0

–5

–10

–15

–20

Gate voltage VGS (V)

FIGURE 6.3.4 Transfer curves (IDS vs. VGS) are measured for a sensor cell (sensel) under application of various magnitudes of pressure from 0 to 30 kPa. VDS = –20 V is applied. The resistance of the pressure-sensitive conductive rubber changes between 10 MW and 1 kW when it is off and on. The circuit diagram of each sensel is shown in the inset.

6.3.2.2 DEVICE PERFORMANCE The DC characteristics of the organic transistors were measured in air prior to the application of the rubber pressure sensor film. The field-effect mobility is 1.4 cm2/Vs in the saturation regime. The on/off ratio is 106 if the off current is defined as the minimum drain current at VGS = +40 V, while it is 104 if the off current is measured at VGS = 0 V. The mobility of the present device is comparable to or slightly larger than that of amorphous silicon (~1.0 cm2/Vs). The complete devices including the rubber pressure sensors are characterized by applying a uniform pressure using a square block made of rubber to avoid scratching. The circuit diagram of an individual sensel is shown in the inset of Figure 6.3.4. The overall configuration is similar to a memory cell and pixel of a chargecoupled device (CCD): Gate electrodes of each line are connected to a word line (VWL), while drain electrodes of each line are connected to a bit line (VBL). Typical results of the measured change in the drain current (IDS) dependence of the gate voltage (VGS) are shown in Figure 6.3.4. Pressure is applied to the bottom side of the device and varies from 0 to 30 kPa (~300 gf/cm2), so the resistance of the rubbery sheet varies from 10 MW to 1 kW and the transconductance as well as the measured current increases.

6.3.3 INTEGRATED CIRCUITS Integrated circuits (ICs) based on organic transistors have been reported in a few publications [12–15]. In this section, a scalable integrated circuit [16,17] for e-skins is described. In order to read out pressure data from the sensor arrays, decoders and selectors connected to word lines and bit lines, respectively, are needed in addition

8080_C006.fm Page 534 Tuesday, April 3, 2007 2:15 PM

534

Organic Field-Effect Transistors

to active matrices. We have manufactured all of those integrated circuits using organic transistors and characterized the electrical performance of the system. The circuits are designed and manufactured by a manual “cut-and-paste” procedure. One of the technological challenges is to implement thousands of sensing components onto three-dimensional surfaces. In case of e-skins for humanoid robots, the fingers need small-area e-skins, while the body requires large-area e-skins. It is not cost effective to design and manufacture various shapes of e-skins. Our approach is to first manufacture large-size skins by a roll-to-roll process and then “cut” large skins in adequate size and “paste” afterwards together with circuits. That is the reason the access transistor matrix, row decoders, and column selectors are manufactured separately. In order to realize the cut-and-paste feature, all the circuits of the system must be scalable in size. The access FET matrix is scalable because it is a simple repetition of sensels. The FET matrix can be cut to any required size. We have also developed row decoders and column selectors that are equally scalable. Integrated circuits for e-skins are formed by organic transistors with a pentacene channel layer. Figure 6.3.5 shows a circuit diagram of the electronic artificial-skin system. The organic circuits have been manufactured by the similar flow described previously. The sensor matrix, column selector, and row decoder are first manufactured on separate films and then assembled together using connecting tapes, which have gold stripes with 0.1-in. spacing on PEN films. The sensor matrix consisting of 16 × 16 sensor cells has assembling electrodes with 0.1-in. spacing to glue using connecting tapes with silver paste. The smaller sensor matrix (e.g., a 4 × 4 matrix) is made by physically cutting the large matrix. Furthermore, if the nonrectangular shape of sensors is needed, it can be made by cutting as long as the shape is convex. Such a maskless process scheme should reduce the manufacturing costs drastically. The waveforms of e-skins are shown in Figure 6.3.6. When pressure is applied to some areas of the sensor matrix, the pressure-sensitive rubber of those parts becomes conductive and sensor cells pull bit lines up to the supply voltage. In case of supply voltage of 40 V, the delay from activation to bit-out is 23 ms, of which the total time needed to scan the whole 16 × 16 sensor cell takes about 1 s. The delay for readout depends on supply voltage: If the voltage is 100 V, the delay is estimated to be halved. Based on the simulation using SPICE MOS model, the scan speed is enhanced by one order of magnitude when the parasitic capacitance is suppressed by reducing the channel length and/or width of wiring of decoders and selectors. The response time of the pressure-sensitive rubber is typically of the order of hundreds of milliseconds, and the individual sensor does not respond to the higher frequency. It should be noted, however, that the readout time of entire 16 × 16 sensor cell array is not the response time of individual transistors multiplied by 16 × 16, since an active matrix scheme is used. The scan speed of the entire sensor cell array is limited by the performance of individual transistors and is independent of the frequency response of the pressure sensors. The total time to access 16 × 16 transistors is about 1 s, comparable to the response time of the pressure sensor. Thus,

8080_C006.fm Page 535 Tuesday, April 3, 2007 2:15 PM

Flexible, Large-Area e-Skins

535

Row decoder

R3R3 R2R2 R1R1 R0R0

= Pressure-sensitive rubbery

φR 0 1 2 3

NAND4 C

16 × 16 Matrix

D E F D0 D1 D2 D3

V 4 × 4 DD GND Matrix GND VDD Column selector

NAND2

φC C0C0 C1C1

FIGURE 6.3.5 The circuit diagram of the electronic artificial skins consisting of 16 × 16 access transistor matrix, column selector, and row decoder. The manufactured transistor with pentacene channel layer shows p-type conduction. R0–R3 are row addresses; C0–C1 are column addresses; the bar indicates the reverse signal; fR-bar and fC-bar are activation signals of row decoder and column selector, respectively. D0–D3 are bit out. 0–4 and C–F are column addresses. GND is the ground, while VDD is the power supply.

for a larger device with a large number of sensor cells, the organic transistors will be the limiting factor of the frequency response.

6.3.4 STRETCHABLE e-SKINS The development of flexible pressure sensor films was an important step in endowing robots with skin sensitivity, but e-skins are expected to add at least two more functionalities: thermal sensing and conformability. Without conformability, e-skins cannot be applied to three-dimensional surfaces. Stretchable e-skins for humans are commercially available, but they do not possess electric functionality. Indeed, various stretchable materials, like rubber, are used in daily activities, but they have poor

8080_C006.fm Page 536 Tuesday, April 3, 2007 2:15 PM

536

Organic Field-Effect Transistors

Decoder output (Word line)

Bit output (Dx)(Bit line)

23 ms 40 V

20 V

Pressure

Without pressure X: 10.0 ms/div Y: 10.0 V/div I Rx, Cx

II 40 V

φR, φC

50 V –20 V

III Decoder output and Bit output(Dx)

40 V

FIGURE 6.3.6 A measured waveform of electronic artificial skin. When pressure is applied to the sensor matrix, the pressure-sensitive rubber becomes conductive and bit line is pulled up to the supply voltage. (I) The input signals of column and row addresses. (II) The activation signal of the row decoder and column selector. (III) The measured waveform. The decoder output (word line) and bit output (bit line).

electrical conductivity. One of the challenges in the field of electronics has been the manufacture of stretchable active electronic elements and interconnects. In this section, we describe a solution that employs a net-shaped structure to make flexible electronic film devices conformable to three-dimensional surfaces [18]. Although the base films we presently use are of polyimide and poly(ethylenenaphthalate) (PEN) — materials that are stiff and not inherently stretchable in a rubberlike sense — our solution includes struts of network structures that twist with the application of tension, as can be seen in Figure 6.3.7. Due to this three-dimensional strut deformation, the whole network structure functions electrically with a unidirectional extension of 25%. We have implemented the pressure sensor network on the surface of an egg and have obtained pressure images in this configuration.

6.3.4.1 DEVICE MANUFACTURING In this section, we describe a manufacturing process for the pressure sensor network. As shown in Figure 6.3.8(a), the sensor cells are positioned at the center of “intersection areas” and are connected to each other by “struts” with electrical wirings. The magnified image of a transistor is shown in Figure 6.3.8(b). First, arrays (typically 12 × 12) of FETs with a pentacene channel layer are manufactured on a polyimide or a PEN film by the similar flow described in the previous section. The channel length (L) and width (W) are 50 and 1800 μm,

8080_C006.fm Page 537 Tuesday, April 3, 2007 2:15 PM

Flexible, Large-Area e-Skins

537

FIGURE 6.3.7 A conformable network of pressure sensors. A plastic film with organic transistors and pressure-sensitive rubber is processed mechanically to form a unique netshaped structure, which makes a film device extendable by 25%. A magnified view of extended net structures is also shown.

respectively. The periodicity is 4 mm, and the total area is 44 × 44 mm2. The mobility of the transistors is 1 cm2/Vs in the saturation regime. The on/off ratio is 105–106 when the off current is defined as the minimum IDS at a positive gate bias. After the manufacturing process, the films containing the organic transistors are transferred to a vacuum chamber without exposure to air, and surfaces are uniformly coated with a 2-μm-thick poly-chloro-para-xylylene (parylene) passivation layer to form a flexible gas barrier, which helps to increase the device lifetime. Indeed, when the organic transistors with parylene were kept in air for two months, the change current was less than 10% in the saturation regime. It can also prevent mechanical damage to the devices during mechanical testing. In addition, the parylene layer can suppress strains induced in the devices. To form the electronic interconnections, spots of parylene on the electrodes are removed by a CO2 laser. The transistor film is then mechanically processed by a numerically controlled (NC) cutting plotter or drilling machine to form the net-shaped structures, as shown in Figure 6.3.8. In a similar way, a pressure-sensitive rubber sheet and a copper electrode suspended by a polyimide film are cut separately to form net-shaped structures. These two films are laminated on the top of the transistor film to complete the pressure sensor network. The periodicity is 4 mm, and the width of the struts is 0.3–0.5 mm.

6.3.4.2 DEVICE PERFORMANCE In this section, the electrical performance of the pressure sensor network is described. All electric measurements were performed in an ambient environment with a semiconductor parameter analyzer, unless otherwise specified. With the application of

8080_C006.fm Page 538 Tuesday, April 3, 2007 2:15 PM

538

Organic Field-Effect Transistors

Word line (WL)

Bit line (BL) (a)

Intersection area (Sensor cell) Strut (Electrical wiring) Punched area

(b) Drain

Gate

Source

FIGURE 6.3.8 (a) A picture of the 3 × 3 sensor cells. A word line, denoted as WL, is connected to the gate electrodes, while a bit line, denoted as BL, is connected to the drain electrodes. The circuit diagram of the thermal sensor network can be obtained by replacing the resistances with diodes. Scale is 4 mm. (b) An optical microscopic image of an organic transistor before shaping the net or integrating it with sensors. The dot line indicates the semiconductor channel layer. Scale is 1 mm.

tension to the pressure sensor network, which consists of many small square holes, deformation with twisted struts appears, as shown in Figure 6.3.7. However, the magnitude of the drain current IDS with and without the application of pressure remains practically unchanged when the extension coefficient increases less than 25%. At extensions beyond this value, which corresponds to the application of a force of 1 N, the device does not function due to electric disconnects at the strut wirings. Such a robust performance of sensors against tension is realized in the present device because the tension induced in stiff intersection areas is considerably less than that in the twistable struts. We should note here that the conductive rubber, which is inherently stretchable, is net shaped: Under tension, the conductive rubber struts lengthen, so the sensor performance in intersection areas is practically unaffected. The pressure sensor network was extended and attached to the surface of an egg, as shown in Figure 6.3.9. In order to read out the data, one of the word lines is activated from a low state (0 V) to a high state (–20 V), while all the other word lines remain in a low state (0 V). Then, current levels are read out by applying VDS = –20 V to each bit line. To obtain a pressure map at room temperature, a similar process was repeated, activating each word line one by one. The lower

8080_C006.fm Page 539 Tuesday, April 3, 2007 2:15 PM

Flexible, Large-Area e-Skins

539

P (kPa) Pressed

IDS (μA)

30

6

20

4

10

2

0

0

FIGURE 6.3.9 An image (upper panel) of pressure sensor matrix put on an egg is shown together with a spatial distribution of pressure (lower panel). The current of each sensor cell is measured by applying a voltage bias of VDS = –20 V and VGS = –20 V under the application of local pressure.

panel in Figure 6.3.9 shows the current level obtained for each sensor cell when the rubber block was positioned on two sensor cells. Only the two cells where the block is positioned were activated, although a small amount of cross-talk was detected due to small leakage current through the gate dielectric layer of the organic transistors. Compared with rubber-based stretchable conductors, the mechanical weakness of the extended networks could pose a problem. We have observed that mechanical failure always occurs around the bases of the struts, which then causes electric disconnections when too much tension is applied to the device. This observation indicates that the major part of the strain is induced at the base of the struts and is rather small in the stiff intersection areas. In the present design, each transistor is positioned at the center of the intersection areas; furthermore, the area of the transistor is much smaller than that of the intersection areas to minimize strain induced in the transistors. To examine mechanical robustness of such a design, the regions of concern are the struts used for electrical wiring rather than the intersection areas where the active device components are located. Therefore, we have prepared a simplified test structure consisting of a linear array that is part of a net structure, processed from a plastic film whose surface is uniformly coated with a 100-nm thick gold layer. We measured the mechanical resistance of this structure by measuring stress cycles, where each cycle comprised an extension of the film by 20% followed by release of the stress. The device showed no significant electronic degradation even after 1,000 cycles (Figure 6.3.10). Therefore, the mechanical resistance is sufficient for many applications requiring a single expansion, such as a permanent application on a curved surface. A higher mechanical resistance, which might be required for applications such as the joints of arms, could be obtained if the strain near the bases of the struts is suppressed by optimizing structural parameters such as the thickness of the base films and the width of the struts.

8080_C006.fm Page 540 Tuesday, April 3, 2007 2:15 PM

540

Organic Field-Effect Transistors

Resistance (Ω)

150

100

50

0

0

1000 2000 The number of cycles

FIGURE 6.3.10 The results of the endurance test performed by measuring stress cycles, where each cycle comprised an extension of the film by 20% followed by the release of the stress. The resistance between both edges of the test structures is monitored as a function of the number of tension cycles. The up arrow represents the breaking of either one side or the other of the gold wire. The test structure is shown in the inset.

6.3.5 FLEXIBLE THERMAL SENSORS A flexible thermal sensor network has been developed employing organic semiconductor diodes in a manner suitable for integration with the pressure sensor network. The possible implementation of both pressure and thermal sensors on the surfaces is presented. By means of laminated sensor networks, the distributions of pressure and temperature are simultaneously obtained.

6.3.5.1 DEVICE MANUFACTURING In the present design, the thermal sensor network contains its own organic transistor active matrix for data readout. This arrangement provides for a self-contained thermal sensor when combined with a pressure sensor, yielding two electrically independent networks. The design of the active matrices is exactly the same for both networks. The process for the thermal sensor network is as follows. Organic diodes, to be used as sheet-type thermal sensors, are manufactured on an ITO-coated PEN film. A 30-nm thick p-type semiconductor of copper phthalocyanine (CuPc) and a 50-nm thick n-type semiconductor of 3,4,9,10-perylene-tetracarboxylic-diimide (PTCDI) are deposited by vacuum sublimation. A 150-nm thick gold film is then deposited to form cathode electrodes having an area of 0.19 mm2. The film with the organic diodes is coated with a 2-μm thick parylene layer and the electronic interconnections are made by the method similar to that mentioned before. The diode film is also mechanically processed to form net-shaped structures. Finally, to complete the thermal sensor network, we laminated the transistor and diode net films together with silver paste patterned by a microdispenser. This is shown in Figure 6.3.11.

6.3.5.2 DEVICE PERFORMANCE Stand-alone thermal sensors are characterized in a nitrogen environment before integration with the transistors. The temperature dependence of the current is plotted

8080_C006.fm Page 541 Tuesday, April 3, 2007 2:15 PM

Flexible, Large-Area e-Skins

541

Thermal sensor

PEN Anode CuPc (P) PTCDI (N) Cathode Parylene Au Paste Au

Transistor

Parylene S Pentacene Polyimide

D

G Base ﬁlm

FIGURE 6.3.11 Cross-sectional illustration of the thermal sensor network, which contains both the diode and organic transistor.

102 CuPc/PTCDI Normalized current

CuPc PTCDI 101

100 2.2

2.4

2.8 2.6 3 1000/T (1/K)

3.2

3.4

FIGURE 6.3.12 Thermal sensor network. Temperature dependence of current is measured under voltage bias of 2 V and data normalized by current at room temperature is plotted as a function of 1000/T for three samples: stand-alone thermal sensors, denoted by solid circles, consisting of double organic semiconductors (30-nm thick CuPc and 50-nm thick PTCDI), and single organic semiconductors (80-nm thick CuPc or 80-nm thick PTCDI, denoted by solid squares and open circles, respectively) sandwiching between ITO and Au electrodes.

8080_C006.fm Page 542 Tuesday, April 3, 2007 2:15 PM

542

Organic Field-Effect Transistors

–1.5

80°C 70°C 60°C

VBL –1.0 IDS (μA)

50°C VWL

40°C

–0.5

0 40

30°C

0

–40

–80

VGS (V)

FIGURE 6.3.13 One cell of the thermal sensor network devices consisting of the diodebased thermal sensors and transistors is characterized at various temperatures from 30 to 80°C.

as a function of 1000/T (Figure 6.3.12). For the purpose of comparison, we also prepared two structures with single organic semiconductor layers, a CuPc and a PTCDI layer. For a double-layer structure, the current is enhanced by a factor of 20 when the temperature is varied from 30 to 160°C. The use of a combination of the two layers is necessary, since the temperature dependence of the devices made of only one type of layer is much smaller. One cell of the thermal sensor network, consisting of the double-layer thermal sensor array and active matrix, was characterized at temperatures between 30 and 80°C (Figure 6.3.13). Since the measurement was performed in ambient conditions, the performance was slightly degraded due to oxygen and/or moisture exposure during measurement. The temperature dependence obtained was smaller than that for devices measured in a nitrogen environment. Further optimization of the parylene passivation layer, particularly with regard to the thickness of the parylene and the deposition conditions, is required to minimize the degradation associated with exposure to oxygen and/or moisture. Ten heat cycles between 30 and 100°C were performed with a current distribution less than ±10%. To map temperatures with the active matrix, it is very important that the thermal sensors respond strongly to temperature, while the change of performance of switching transistors remains small. Indeed, as mentioned earlier, the organic diode-based sensors exhibit a current enhancement by a factor of 20 when the temperature is varied from 30 to 160°C; under the same conditions, the current of the organic transistors changed only by a factor of three. The final step is the implementation of the thermal and pressure sensor films on the target surface. In a practical production scheme, this process can be carried out by successively laminating the sensor films one by one on the surface, so that the thermal sensors and pressure sensors form checkerboard patterns, as shown in Figure 6.3.14(a). Such a manufacturing process is possible because each sensor film has

8080_C006.fm Page 543 Tuesday, April 3, 2007 2:15 PM

Flexible, Large-Area e-Skins

(a)

543

P T

(b)

IDS T (°C) (μA)

50

0.7

40

0.5

30 (c)

0.3

IDS P (kPa) (μA) 30

6

20

4

10

2

0

0

FIGURE 6.3.14 Integration of pressure and thermal sensor networks. (a) A possible implementation of thermal and pressure sensor films. The pressure and thermal sensors are represented by P and T, respectively. Scale is 4 mm. (b) The spatial distribution of temperature that is converted from the temperature-dependent current in the thermal sensor network. A copper block (15 × 37 mm2), whose temperature is maintained at 50°C, is positioned diagonally, as is indicated by the dotted line. The sensing area is 44 × 44 mm2. (c) Simultaneously, the spatial distribution of pressure is measured with the pressure sensor network.

its own active matrix and is electrically independent. Although the present network device can be applied to a curved surface, in the following mapping experiment we mounted the thermal and pressure sensor networks on a flat board in order to achieve a good thermal contact over a wide area. In the described configuration with two sensor networks, measurements of temperature and pressure mapping were performed simultaneously. The spatial distribution of the temperature-dependent current was measured by applying individually to each sensor cell a voltage VDS = –2 V for each bit line and VGS = –40 V for each word line; simultaneously, the spatial distribution of pressure was measured by applying individually to each sensor cell a voltage VDS = –20 V for each bit line and

8080_C006.fm Page 544 Tuesday, April 3, 2007 2:15 PM

544

Organic Field-Effect Transistors

VGS = –20 V for each word line. A copper block whose temperature was maintained at 50°C was positioned diagonally. The measured temperature and pressure distributions are shown in Figure 6.3.14(b) and (c), respectively, with the dotted-line rectangle indicating the area of the copper block. In Figure 6.3.14(b) and (c), sensor cells, on which the block was positioned, were activated. Both look similar since the same positions are pressed simultaneously; however, small discrepancies between the two should be ascribed to nonuniformity of pressure and temperature applied over a wide area.

6.3.5.3 DISCUSSION There are many benefits to a net-shaped film device apart from conformability. First, good isolation of each sensor leads to more accurate sensing. In particular, for an accurate measurement of temperature, it is very important to suppress thermal perturbation due to heat flow from neighboring regions. In a net-shaped structure, such a thermal flow can be minimized by reducing the width of the struts. Second, the net-shaped structures make possible the inclusion of various types of sensors on the film surfaces, a factor which is very important to perform accurate measurements of temperature and pressure. The present scheme simplifies the fabrication process, since various types of sensors can be prepared separately and laminated together in the last fabrication step. The thickness of one of the sensor films causes small steps, which may degrade the accuracy of the sensing measurements. The step height of the present thermal sensor film is about 200 μm, but it can be reduced by one order of magnitude. Indeed, our group and another research group have manufactured organic transistor active matrices on 10- to 25-μm thick plastic films [20]. Thus, it will be possible in the near future to make an electronic skin that has functions that human skin lacks by integrating various sensors not only for pressure and temperature, but also for light, humidity, strain, or ultrasound.

6.3.6 BENDING TESTS Bending tests of organic transistors have also been carried out [19,20]. Pentacene FETs are manufactured on 13-μm thick polyimide films with polyimide gate dielectric layers encapsulated by passivation layers of poly-chloro-para-xylylene with same thickness as base films. When the bending radius decreases to 2 mm, the change in mobility is less than 3%. Further decrease of the bending radius (R) causes systematic change in mobility. However, organic FETs are functional even at the bending radius down to 0.5 mm. The change in transistor characteristics is reversible and reproducible even when the bending radius is 0.5 mm. The present device is schematically shown in Figure 6.3.15(a). A gate electrode consisting of a Cr adhesion layer and an Au layer is deposited through a shadow mask in a vacuum evaporator on a 13-μm thick polyimide film. Then, polyimide precursors are spin-coated to form 500-nm thick gate dielectric layers. A 50-nm thick pentacene is deposited, and then 50-nm thick Au drain and source electrodes are evaporated. The channel length L and width W of the FETs are 100 μm and

8080_C006.fm Page 545 Tuesday, April 3, 2007 2:15 PM

Flexible, Large-Area e-Skins

(a)

545

13-μm-thick parylene

Neutral strain position

Pentacene Au(S) Au(D)

) Polyimide Au(G 13-μm-thick polyimide FET

–30 0.5 mm

IDS (μA)

(b) –20

SS DD –10 R = 20, 10, 5, 2, 1 mm

G Inward bending

0

IDS (μA)

(c)

R = 20, 10, 5, 2 mm 1 mm 0.5 mm

–20

D

S G

–10 R 0 20

Outward bending 10

0

–10 –20 –30 VGS (V)

–40

–50

FIGURE 6.3.15 (a) The cross-sectional illustration of organic transistors on plastic films with polyimide gate dielectric layers and parylene passivation layers. Transfer characteristics with bending radii of R = 20, 10, 5, 2, 1, and 0.5 mm; these correspond to inward (b) and outward (c) bending strains. A VGS is swept from 20 to –40 V with application of VDS = –40 V.

1 mm, respectively. Finally, the base film with transistors is uniformly coated by a 13-μm thick poly-chloro-paraxylene passivation layer, hereafter referred to as a parylene layer. The FETs are placed at the neutral position by sandwiching between a base film and an encapsulation layer. The direction of source drain current paths is precisely arranged parallel to the direction of strain. A capacitor is also manufactured simultaneously on the same base film and its capacitance is measured while changing the bending radius of the base films in order to obtain precise capacitance of polyimide gate dielectric layers with bending stress. Furthermore, its capacitor characteristics can function as a strain gauge. We measured electrical properties of the FETs under various inward and outward bending strains, whose magnitudes were systematically controlled while changing the bending radius R of the base plastic films. Schematic illustrations of inward and outward bending are shown in Figure 6.3.15. The DC characteristics of transistors are measured at ambient environment in dark before the start of bending experiments. The mobility is 0.5 cm2/Vs in the

8080_C006.fm Page 546 Tuesday, April 3, 2007 2:15 PM

546

Organic Field-Effect Transistors

saturation regime, and the on/off ratio exceeds 105. We measured the transfer characteristics of the FETs under inward and outward bending strains, while changing the bending radius R of the base film, as shown in Figure 6.3.15 (b) and (c). Although inward and outward bending stresses down to 2 mm are applied to the devices, saturation currents show slight changes (less than 2% on both bending experiments). However, the saturation current increases by 10% and decreases by 25% under the inward and outward bending radius down to 0.5 mm. When R is decreased down to 2 mm, the change in mobility is less than 3% on both bending cases, which indicates that strain-induced changes in transistor performance are negligibly small down to R = 2 mm. This demonstrates that the present FETs have an excellent stability under strain. However, a further decrease in R from 2 to 0.5 mm causes systematic changes in mobility; namely, it increases by 20% on inward bending or decreases by 30% on outward bending stress. Furthermore, the changes in transistor characteristics are reversible and reproducible even when R is 0.5 mm. In order to investigate the recovery performance after stressing the FETs, a lot of inward and outward bending stresses (bending cycles) of R = 2 mm are applied to the FETs. First, the FET is bent from flat (R = ∞) to R = 2 mm and immediately released up to flat state. Again, the same FET is bent down to R = 2 mm and released. This procedure is repeated on bending cycles of 110 times for a minute. Measurements are performed at each bending cycle with applications of voltage bias of VDS = VGS = 40 V, which should be distinguished from continuous DC bias stress. Figure 6.3.16 shows the normalized IDS as a function of the number of outward bending cycles. There is no significant change even after bending cycles up to 60,000 times, while IDS decreases after 60,000 times and finally decreases by 10% at 160,000

Normalized IDS

1.1

1

0.9 Outward bending cycles 0.8

101

102 103 104 The number of bending cycles

105

FIGURE 6.3.16 Normalized source-drain current as a function of the number of outward bending cycles (R = 2 mm). The data are normalized by the start of the measurement. This procedure is repeated on bending cycles of 110 times for a minute, and measurements are performed with applications of VDS = VGS = –40 V.

8080_C006.fm Page 547 Tuesday, April 3, 2007 2:15 PM

Flexible, Large-Area e-Skins

547

times. The dependence of transistor performance under inward bending cycles is qualitatively the same. Such negligible changes in transistor characteristics under a lot of bending cycles have not yet been achieved in any other transistors, including organic, single-crystalline silicon, and amorphous-silicon FETs, so far.

6.3.7 ISSUES AND FUTURE PROSPECTS One of the issues for large-area sensors is stability and reliability of organic transistors. There are two different modes of degradation of organic transistors: chemical and electrical degradation. It is pointed out that chemical degradation is induced mainly by oxygen and/or moisture-like electroluminescent (EL) devices and therefore can be suppressed by appropriate encapsulation. Applications requiring mechanical flexibility such as e-skins need flexible substrates with low gas permeability, although plastic films usually have high gas penetration. Thus, it is one of the urgent problems to develop flexible base films with low gas permeability. On the other hand, the electrical degradation is drift of electric performance observed when DC voltage bias is applied. This is usually referred to as DC bias stress effects, which are induced mainly by shift of threshold voltage. Such drifts have been observed in amorphous silicon transistors. It is understood that such shifts originate from the deep trapping of conduction carriers. The suppression of DC bias stress effects is crucial to the reliable driving of sophisticated organic integrated circuits. We have found recently that an annealing process helps to suppress DC bias stress effects. When pentacene FETs are annealed at 140°C for 12 h in a nitrogen environment, the change in their source-drain current is 3 ± 1% even after the application of continuous DC voltage biases of VDS = VGS = 40 V for 11 h. The details of this technique are reported elsewhere [21]. It is well known that the mobility of organic semiconductors is about three orders of magnitude lower than that of silicon and a major disadvantage for applications such as video-rate display and radio frequency identification (RFID) tags, which have been recognized so far as the main motivations for the development of organic transistors. In the case of area sensors, however, the slower speed is tolerable for most applications. For e-skin in particular, the integration of pressure/thermal sensors and organic peripheral electronics allows one to take advantage of the many benefits of organic transistors, such as mechanical flexibility, large area, low cost, and relative ease of fabrication, without suffering from their drawbacks. Similar functions may be achieved by amorphous silicon on plastic films, and the present study demonstrates the general feasibility of large-area electronics from the context of sensor applications. We believe that developing e-skins using the active matrix method is practical, since the manufacturing costs of organic transistors are expected to be low, even for large-area devices. As mentioned in the introduction of this chapter, integrated circuits based on organic transistors play an important role in large-area electronics where the manufacturing cost per area must be very low. It is undoubted that one of the most important directions for future electronics is ambient intelligence or wireless sensor network. In order to realize such a network, one of the key technologies is a sensor to detect physical or chemical information distributed for large

8080_C006.fm Page 548 Tuesday, April 3, 2007 2:15 PM

548

Organic Field-Effect Transistors

areas. Large-area features of organic transistors would be suitable to realize largearea sensors for this purpose. We will see many applications of large-area pressure and thermal sensors beyond robotics, such as security, home care, entertainment, sports, and more. One example is when pressure-sensitive carpet spread on a floor in a house can be used as a security system that distinguishes family members from a stranger just from the analysis of footprints. Furthermore, a tactile bed would be able to diagnose physical conditions instantly by monitoring heart rate and breathing rate. Organic transistors are inferior to silicon from the point of speed and power consumption. However, choosing organic transistors will be reasonable when the applications require lowcost features for large areas. It is very interesting to integrate organic transistor active matrices with other sensors besides pressure or thermal sensors. In another new development of largearea electronics, we have demonstrated the first large-area, flexible, and lightweight sheet image scanner, fabricated on a plastic film and integrating organic field-effect transistors with organic photodiodes [22,23]. The new sheet scanner does not require any mechanical or optical component such as focusing lenses. As a result, the device is thin, lightweight, and mechanically flexible. Furthermore, organic transistors are suitable for applications to large-area plastic actuators. We have fabricated a novel, flexible, lightweight Braille sheet display on a plastic film by integrating high-quality organic FETs with plastic actuators [24,25]. Employing the large-area feature of organic transistors, many unique applications will be proposed and realized in the near future.

6.3.8 SUMMARY In this section, we have described large-area sensors as one of the promising applications of organic thin film transistors. In particular, we have demonstrated largearea pressure and thermal sensors by integrating an organic transistor matrix with pressure-sensitive rubber sheets and temperature-sensitive organic PN diodes. These new sensors will be useful for a variety of applications such as novel security systems, regenerative medicine, and entertainment purposes beyond robot skins. Since the manufacturing costs of organic transistors would be low even for largearea devices, organic ICs will play an important role in large-area electronics. This attribute of organic ICs is complementary to silicon and therefore the combination of silicon and organic ICs will produce new market opportunities. We hope that remaining issues related to stability, operation voltage, speed, and power consumption will be solved in the near future, pushing organic transistors into practical use.

8080_C006.fm Page 549 Tuesday, April 3, 2007 2:15 PM

Flexible, Large-Area e-Skins

549

REFERENCES 1. Gelinck, G.H., Huitema, H.E.A., van Veenendaal, E., Cantatore, E., Schrijnemakers, L., van der Putten, J.B.P.H., Geuns, T.C.T., Beenhakkers, M., Giesbers, J.B., Huisman, B.–H., Meijer, E.J., Benito, E.M., Touwslager, F.J., Marsman, A.W., van Rens, B.J.E., and de Leeuw, D.M., Flexible active-matrix displays and shift registers based on solution-processed organic transistors, Nat. Mat., 3, 106, 2004. 2. Rogers, J.A., Bao, Z., Baldwin, K., Dodabalapur, A., Crone, B., Raju, V.R., Kuck, V., Katz, H., Amundson, K., Ewing, J., and Drzaic, P., Paper-like electronic displays: Large-area rubber-stamped plastic sheets of electronics and microencapsulated electrophoretic inks, Proc. Natl. Acad. Sci. U.S.A., 98, 4835, 2001. 3. Baude, P.F., Ender, D.A., Haase, M.A., Kelley, T.W., Muyres, D.V., and Theiss, S.D., Pentacene-based radio-frequency identification circuitry, Appl. Phys. Lett., 82, 3964, 2003. 4. Baude, P.F., Ender, D.A., Kelley, T.W., Haase, M.A., Muyres, D.V., and Theiss, S.D., Organic semiconductor RFID transponders, IEDM Tech. Dig., 191, 2003. 5. Crone, B., Dodabalapur, A., Gelperin, A., Torsi, L., Katz, H.E., Lovinger, A.J., and Bao, Z., Electronic sensing of vapors with organic transistors, Appl. Phys. Lett., 78, 2229, 2001. 6. Assadi, A., Gustafsson, G., Willander, M., Svensson, C., and Inganas, O., Determination of field-effect mobility of poly(3-hexylthiophene) upon exposure to NH3 gas, Synth. Met., 37, 123, 1990. 7. Ohmori, Y., Takahashi, K., Muro, K., Uchida, M., Kawai, T., and Yoshino, K., Gassensitive Schottky gated field effect transistors utilizing poly(3-alkylthiophene) films, Jpn. J. Appl. Phys., Part 1, 30, 1247, 1991. 8. Torsi, L., Dodabalapur, A., Sabbatini, L., and Zambonin, P.G., Multiparameter gas sensors based on organic thin-film-transistors, Sens. Actuators B, 67, 312, 2000. 9. Subramanian, V., Lee, J.B., Liu, V.H., and Molesa, S., Printed electronic nose vapor sensors for consumer product monitoring, ISSCC Tech. Dig., 274, 2006. 10. Someya, T., Sekitani, T., Iba, S., Kato, Y., Kawaguchi, H., and Sakurai, T., A largearea, flexible pressure sensor matrix with organic field-effect transistors for artificial skin applications, Proc. Natl. Acad. Sci. U.S.A., 101, 9966, 2004. 11. Nicholls, H.R. and Lee, M.H., A survey of robot tactile sensing technology, Int. J. Robotics Res., 8, 3, 1989. 12. Kato, Y., Iba, S., Teramoto, R., Sekitani, T., Someya, T., Kawaguchi, H., and Sakurai, T., High mobility of pentacene field-effect transistors with polyimide gate dielectric layers, Appl. Phys. Lett., 84, 3789, 2004. 13. Crone, B., Dodabalapur, A., Lin, Y.-Y., Filas, R.W., Bao, Z., LaDuca, A., Sarpeshkar, R., Katz, H.E., and Li, W., Large-scale complementary integrated circuits based on organic transistors, Nature, 403, 521, 2000. 14. Drury, C.J., Mutsaers, C.M.J., Hart, C.M., Matters, M., and Leeuw, D.M., Low-cost all-polymer integrated circuits, Appl. Phys. Lett., 73, 108, 1998. 15. Gelinck, G.H., Geuns, T.C.T., and Leeuw, D.M., High-performance all-polymer integrated circuits, Appl. Phys. Lett., 77, 1487, 2000. 16. Someya, T., Kawaguchi, H., and Sakurai, T., Cut-and-paste organic FET customized ICs for application to artificial skin, Tech. Digest 2004 IEEE ISSCC, 288, 16(2), 2004. 17. Kawaguchi, H., Someya, T., Sekitani, T., and Sakurai, T., Cut-and-paste customization of organic FET integrated circuit and its application to electronic artificial skin, IEEE J. Solid-State Circuits, 40, 177, 2005.

8080_C006.fm Page 550 Tuesday, April 3, 2007 2:15 PM

550

Organic Field-Effect Transistors

18. Someya, T., Kato, Y., Sekitani, T., Iba, S., Noguchi, Y., Murase, Y., Kawaguchi, H., and Sakurai, T., Conformable, flexible, large-area networks of pressure and thermal sensors with organic transistor active matrixes, Proc. Natl. Acad. Sci. U.S.A., 102, 12321, 2005. 19. Sekitani, T., Kato, Y., Iba, S., Shinaoka, H., Someya, T., Sakurai, T., and Takagi, S., Bending experiment on pentacene field-effect transistors on plastic films, Appl. Phys. Lett., 86, 073511, 2005. 20. Sekitani, T., Iba, S., Kato, Y., Noguchi, Y., Someya T., and Sakurai, T., Ultraflexible organic field-effect transistors embedded at a neutral strain position, Appl. Phys. Lett., 87, 173502, 2005. 21. Sekitani, T., Iba, S., Kato, Y., Someya, T., and Sakurai, T., Suppression of DC bias stress-induced degradation of organic field-effect transistors using postannealing effects, Appl. Phys. Lett., 87, 073505, 2005. 22. Someya, T., Kato, Y., Iba, S., Kawaguchi, H., and Sakurai, T., Integration of organic FETs with organic photodiodes for a large area, flexible, and lightweight sheet image scanners, IEEE Trans. Electron Devices, 52, 2502, 2005. 23. Kawaguchi, K., Iba, S., Kato, Y., Sekitani, T., Someya, T., and Sakurai, T., IEEE Sensor J., 6, 1209–1217, 2006. 24. Braille IEDM. 25. Takamiya, M., Sekitani, T., Kato, Y., Kawaguchi, H., Someya, T., and Sakurai, T., An organic FET SRAM with back gate to increase static noise margin and its application to braille sheet display, IEEE J. Solid-State Circuits, 42, 93–100, 2007.

8080_C006.fm Page 551 Tuesday, April 3, 2007 2:15 PM

6.4

Organic Thin-Film Transistors for Flat-Panel Displays

Michael G. Kane CONTENTS 6.4.1 Introduction................................................................................................552 6.4.1.1 Active Matrix Displays ...............................................................552 6.4.1.2 Organic Electronics for Displays ................................................552 6.4.2 Important Organic TFT Parameters for Display Applications..................553 6.4.2.1 Field-Effect Mobility...................................................................553 6.4.2.2 Threshold Voltage........................................................................555 6.4.2.3 Subthreshold Swing.....................................................................556 6.4.2.4 Leakage Currents.........................................................................558 6.4.2.5 Capacitances ................................................................................559 6.4.2.6 Contact Resistance ......................................................................560 6.4.2.7 Bias-Stress Instability and Hysteresis.........................................561 6.4.2.8 Light Sensitivity ..........................................................................563 6.4.2.9 TFT Nonuniformity.....................................................................564 6.4.3 Display Technologies.................................................................................564 6.4.3.1 Liquid-Crystal and Electrophoretic Displays .............................564 6.4.3.1.1 Introduction................................................................564 6.4.3.1.2 Electro-Optic Behavior of Twisted-Nematic Liquid Crystals...........................................................567 6.4.3.1.3 Electro-Optic Behavior of Electrophoretic Materials ....................................................................569 6.4.3.1.4 Liquid-Crystal and Electrophoretic Display Architecture ...............................................................570 6.4.3.2 Active Matrix OLED Displays ...................................................577 6.4.3.2.1 Introduction................................................................577 6.4.3.2.2 Electro-Optic Behavior of Organic Light-Emitting Diodes ........................................................................578 6.4.3.2.3 OLED Display Architectures ....................................580

551

8080_C006.fm Page 552 Tuesday, April 3, 2007 2:15 PM

552

Organic Field-Effect Transistors

6.4.4 Using Organic TFTs for Integrated Drivers ..............................................589 6.4.5 Conclusions................................................................................................591 References..............................................................................................................591

6.4.1 INTRODUCTION 6.4.1.1 ACTIVE MATRIX DISPLAYS While low-resolution displays such as those in digital watches use directly driven segmented elements, higher resolution flat-panel displays require a matrix architecture in which the display is an array of identical pixels arranged in rows and columns. By definition, an active matrix display has one or more electronic switching elements in each pixel, while a passive matrix has no switching elements [1]. The substrate containing the switching elements of an active matrix display is often called the backplane. It is more costly to include switching elements than to leave them out, so active matrix displays cost more to produce than passive matrix displays. However, it is more difficult to scale passive matrix displays to large sizes than active matrix displays. As a result, today’s very competitive liquid-crystal display (LCD) market is dominated by active matrix LCDs (AMLCDs), sometimes called TFT LCDs. Passive matrix LCDs are found only in small, low-cost applications such as mobile phones and handheld games. The plasma display panels (PDPs) used for large flat-screen televisions are passive matrix because conventional switching elements cannot tolerate the required high voltages. Indeed, scaling to higher resolutions has been difficult for PDP technology. In this chapter, we consider the use of organic thin-film transistors (OTFTs) in active matrix flat-panel displays. Most AMLCDs today use amorphous silicon thinfilm transistors (a-Si TFTs) or polysilicon thin-film transistors (poly-Si TFTs) as the switching elements. a-Si TFTs are the dominant technology, but manufacturers of small- to medium-sized AMLCDs, such as digital cameras, are making increasing use of poly-Si TFTs because their higher performance permits row and column driver circuits to be integrated directly on the display glass, reducing display module cost and shortening product development times.

6.4.1.2 ORGANIC ELECTRONICS

FOR

DISPLAYS

OTFTs and organic light-emitting diodes (OLEDs) are attracting considerable attention for potential use in flat-panel displays. However, it is more difficult for OTFT technology to make its case for display applications than it is for OLEDs. OLED displays look better than LCD displays, and it is just a question of the willingness of consumers to pay (or manufacturers to absorb) additional cost for a more appealing display, at least in the low-volume stages of OLED technology when manufacturing costs are high. In contrast, OTFTs go into display backplanes, which are a hidden technology not directly visible to the consumer. As a result, OTFTs do not have the same direct consumer appeal as OLEDs. The features that will push OTFT technology into display products will be (1) mechanical flexibility and (2) low cost.

8080_C006.fm Page 553 Tuesday, April 3, 2007 2:15 PM

Organic Thin-Film Transistors for Flat-Panel Displays

553

Mechanical flexibility. OTFTs will enable flexible displays that can be rolled up or wrapped around curved objects. This is because OTFTs can be fabricated at or not much above room temperature, allowing backplanes to be made on plastic substrates. Low cost. Unlike the cost of integrated circuits, the cost of direct-view displays is not reduced by reductions in feature size. Instead, cost reductions must be achieved by reducing the cost of materials and equipment and by increasing throughput. The cost of display fabrication has a number of components that can be reduced when OTFTs are used for the active matrix. The manufacturing cost depends on process temperatures because, in general, higher temperature processing entails higher capital costs, more expensive substrate materials, and lower throughput because of the time required for temperature ramping. As a result, the fact that OTFT backplanes can be fabricated at low temperatures leads to low manufacturing costs, even if they are made on rigid glass substrates. If low temperatures permit fabrication on a plastic web, manufacturing costs may be further reduced. Furthermore, if additive, printinglike processes are used for some or all of the layers, the costs associated with materials and photolithography, two of the most expensive components of backplane manufacturing, are reduced or eliminated. In this chapter, we first discuss important OTFT characteristics and parameters for display applications. Next, we describe the use of OTFTs in the three types of active matrix displays for which they have the most relevance and hold the most promise: AMLCDs, active-matrix electrophoretic displays, and active matrix organic light-emitting diode (AMOLED) displays. Finally, we briefly discuss the use of OTFTs for integrated display driver circuits.

6.4.2 IMPORTANT ORGANIC TFT PARAMETERS FOR DISPLAY APPLICATIONS 6.4.2.1 FIELD-EFFECT MOBILITY When considering the usefulness of OTFTs, the transistor parameter that has received by far the greatest attention has been the field-effect mobility μFE. It is commonly cited as a performance metric for comparing different OTFT materials and fabrication methods. But the field-effect mobility is only one of several important OTFT parameters. In this section, we discuss the field-effect mobility as a design parameter, and in following sections we discuss other important parameters. Other treatments of the application of OTFTs to active matrix displays have been reviewed in the literature [2,3]. In the standard metal-oxide semiconductor field-effect transistor (MOSFET) drain-current equations, μFE is a proportionality factor that relates the drain current ID to the gate and drain voltages VGS and VDS, respectively; the threshold voltage Vt; the channel width W and length L; and the gate dielectric capacitance per unit area Cox. The standard drain-current equation for an n-channel device in the linear region of operation (VDS < VGS – Vt) is [4]

8080_C006.fm Page 554 Tuesday, April 3, 2007 2:15 PM

554

Organic Field-Effect Transistors

ID =

W 2 μ FE Cox ⎡⎣(VGS − Vt ) VDS − VDS / 2⎤⎦ L

(6.4.1)

In the saturation region of operation (VDS ≥ VGS – Vt) it is ID =

2 W μ FE Cox (VGS − Vt ) L 2

(6.4.2)

Often the values of μFE extracted from measurements in the two regions of operation are not equal, so one sometimes refers to μFE,lin or μFE,sat, depending on the operating region. In this chapter, we do not distinguish between the linear and saturated field-effect mobility, although if the two have significantly different values, a careful designer will do so. The field-effect mobility is distinct from the more physically fundamental carrier mobility, which is only one factor in the field-effect mobility. In this chapter, when we refer to the mobility, we are referring to the field-effect mobility, since we will not have occasion to refer to the carrier mobility. We often assume for simplicity that mobility can be treated as constant, but experimentally the mobility is found to depend on VGS and VDS. Therefore, for extracting mobility from measurements and for accurate simulations, mobility must be treated as a voltage-dependent quantity that is derived from small-signal measurements: in the linear region, for very small drain voltages (VDS << VGS – Vt), ∂I D L WCoxVDS ∂VGS

(6.4.3)

2L ⎡∂ ID ⎤ = ⎥ ⎢ WCox ⎣ ∂VGS ⎦

(6.4.4)

μ FE =

in the saturation region, 2

μ FE

Figure 6.4.1 shows the voltage-dependent mobility in the saturation region for a pentacene OTFT, derived from measurements using Equation 6.4.4. Often a single value for the mobility is cited rather than a set of curves. In this case the convention is to report the maximum mobility over a range of gate voltages for a specified drain voltage, either in the linear region or the saturation region. A similar method of specifying mobility is used for the voltage-dependent mobility of silicon MOSFETs. For creating an initial design prior to computer-aided simulations that use more accurate models, it is best to estimate an average mobility over the expected range of voltages from curves like those in Figure 6.4.1, rather than using the maximum mobility value, which represents best-case device behavior at a single voltage point.

8080_C006.fm Page 555 Tuesday, April 3, 2007 2:15 PM

Organic Thin-Film Transistors for Flat-Panel Displays

usat(Vds = –2.0V) usat(Vds = –4.0V) usat(Vds = –6.0V) usat(Vds = –8.0V) usat(Vds = –10.0V) usat(Vds = –12.0V) usat(Vds = –14.0V) usat(Vds = –16.0V) usat(Vds = –18.0V) usat(Vds = –20.0V)

0.7 0.6 0.5 μFE(cm2/V-s)

555

0.4 0.3 0.2 0.1 0.0 –15

–10

–5

0

5

10

VGS(V)

FIGURE 6.4.1 Measured voltage-dependent mobility in the saturation region for a pentacene OTFT with SiO2 gate dielectric. The mobility is plotted only for data points in the saturation region.

6.4.2.2 THRESHOLD VOLTAGE Like the mobility, the threshold voltage Vt can be defined using the standard MOSFET equations, Equations 6.4.1 and 6.4.2. Experimentally, it is found that different values of threshold voltage are extracted for the two regions of operation so that there are separate parameters Vt,lin and Vt,sat. In addition, there may be a drain-voltage dependence. Usually these details about the behavior of the threshold voltage are not important for displays. We neglect them and refer simply to a single threshold voltage Vt. The threshold voltage can be extracted from measurements in the linear region by plotting ID versus VGS and extrapolating the curve to ID = 0. Similarly, the threshold voltage can be extracted from measurements in the saturation region by plotting I D versus VGS and extrapolating to ID = 0. In either case, it is found experimentally that the drain current turns off more gradually than Equations 6.4.1 and 6.4.2 predict because the mobility is bias dependent and decreases as VGS approaches Vt. In addition, there is a subthreshold region for VGS < Vt in which a small drain current still flows. As a result, typically the extrapolation to ID = 0 is performed using a tangent to the curve at the point of maximum slope or using two points on the curve that bracket the operating region. Figure 6.4.2 shows measured data for the same pentacene OTFT as in Figure 6.4.1, but in the linear region (VDS = –0.1 V). The straight line in Figure 6.4.2 is tangent to the drain-current curve at the point of maximum slope, allowing the linearregion threshold voltage to be calculated as +6.5 V.

8080_C006.fm Page 556 Tuesday, April 3, 2007 2:15 PM

556

Organic Field-Effect Transistors

0.0E+00

ID (A)

–5.0E-09 –1.0E-08 –1.5E-08 –2.0E-08 –2.5E-08 –20

–10

0 VGS(V)

10

20

FIGURE 6.4.2 Measured linear-region drain-current data (curved line, VDS = –0.1 V) and the tangent at the point of maximum slope for the same pentacene OTFT as in Figure 6.4.1. The tangent allows the threshold voltage to be calculated as +6.5 V.

It is not uncommon for threshold voltages in OTFTs to be large and uncontrolled. Threshold voltages as high as tens of volts are not uncommon. For active matrix displays as well as other applications, low threshold voltages are required for keeping drive voltages low. It is possible that too little attention is given to the importance of low threshold voltages. The application of self-assembled monolayers such as octadecyltrichlorosilane to the gate dielectric before semiconductor deposition has been found to help reduce the magnitude of Vt and render it more controllable. Higher gate dielectric capacitances also reduce threshold voltages, although often at the expense of higher gate leakage and lower yield.

6.4.2.3 SUBTHRESHOLD SWING Equations 6.4.1 and 6.4.2 predict that for VGS = Vt, the drain current drops to zero, and the simplified physical models underlying these equations predict that the drain current is zero for all VGS < Vt. In reality, as Figure 6.4.2 indicates, these equations do not correctly represent the drain current in the subthreshold region, where the linear or quadratic dependence of drain current on gate voltage in Equations 6.4.1 and 6.4.2 gradually transitions to an exponential dependence. The standard draincurrent equation in the subthreshold region has the form ID =

(

)

W Kμ FE Cox 1 − e− qVDS / kT eqVGS /nkT L

where K is a constant that depends on materials and device structure n is the ideality factor k is Boltzmann’s constant T is the absolute temperature

(6.4.5)

8080_C006.fm Page 557 Tuesday, April 3, 2007 2:15 PM

Organic Thin-Film Transistors for Flat-Panel Displays

log ID

ID =

557

W K μCoxexp(VGS /nkT) L

VDS large and ﬁxed ID =

W μCox(VGS – Vt)2 2L

Ioﬀ Vt

VGS

FIGURE 6.4.3 The transfer characteristic for an MOSFET shows an exponential dependence of drain current on gate voltage below the threshold voltage. As the gate voltage is further reduced, the exponential characteristic merges with a leakage current characteristic. Often the value that is cited for the off-current Ioff is the minimum current.

Thermodynamics demands n ≥ 1. The subthreshold region is shown more easily using a logarithmic scale for the drain current, since plotting Equation 6.4.5 on a log scale yields a straight line, as illustrated in Figure 6.4.3. The subthreshold behavior can be specified by means of n or the subthreshold slope ∂logID/∂VGS, but it is more common to cite the value of the subthreshold swing S = [∂logID/∂VGS], the inverse of the subthreshold slope. Typically, the units used for subthreshold swing are volts/decade so that it represents the change in gate voltage needed to change the drain current by a factor of 10. Experimentally, it is found that the value of S varies through the subthreshold region, as Figure 6.4.3 illustrates. At the low end, the subthreshold characteristic smoothly merges into a leakage current characteristic, and at the upper end, it turns into the saturation characteristic. Therefore, the value typically cited is the maximum slope of the log ID – VGS characteristic, analogous to what is done with mobility in the region above threshold. The relationship between S and the ideality factor n is S = n kT ln (10 ) V /decade = 60 n mV/decade at 300 K

(6.4.6)

A subthreshold region is found in all MOSFETs, but long-channel single-crystal silicon MOSFETs have a nearly ideal subthreshold region in which the ideality factor n is very close to unity, so the change in gate voltage needed to change the drain current by a factor of 10 is only 60 mV at 300 K. Nonideal subthreshold characteristics (n > 1) occur when a change in gate voltage does not produce a corresponding equal change in the surface potential of the semiconductor. Increasing the gate dielectric capacitance reduces the subthreshold swing by improving the coupling between the gate and the semiconductor surface. Typical values of subthreshold swing at 300 K for TFT technologies are:

8080_C006.fm Page 558 Tuesday, April 3, 2007 2:15 PM

558

Organic Field-Effect Transistors

S = 200 – 500 mV/decade (poly-Si TFTs) = 500 mV – 1 V/decade (a-Si TFTs) = 500 mV – 5 V/decade (OTFTs) From a design point of view, small subthreshold swing is desirable because less gate voltage excursion ΔVGS is needed to turn the transistor from fully off to fully on. Indeed, one may view the total ΔVGS required to take the transistor from fully “off” to fully “on” as having two parts. The subthreshold slope determines the voltage excursion that must take place below the threshold voltage, and the mobility determines the required excursion above threshold. OTFTs often have subthreshold swings that are large compared to silicon devices, due to localized states in the energy gap near the highest occupied molecular orbital (HOMO) or lowest unoccupied molecular orbital (LUMO) levels. For OTFTs, the voltage excursion required below threshold often has about the same magnitude as that required above threshold, and the dielectric breakdown requirements, power supply requirements, and power dissipation are adversely affected.

6.4.2.4 LEAKAGE CURRENTS Two types of leakage current have to be considered: •

•

Drain leakage current ID when the transistor is off. That is, the gate voltage VGS biases the device below the subthreshold region. This current is often called the off-current. Gate leakage current IG under all transistor conditions, both on and off. The gate leakage current can be ignored in conventional silicon MOSFETs because thermal SiO2 is an excellent insulator. However, this is not always the case with OTFTs. Thermal SiO2 is not an option for the gate dielectric unless a conducting silicon substrate is used as the gate electrode. Other gate dielectrics may not insulate as well as thermal SiO2, and gate leakage currents can be significant.

The off-current is seen in the log ID – VGS curve of Figure 6.4.3 at gate voltages below those producing the exponential subthreshold characteristic. In general, the off-current is not independent of the gate and drain voltages. In some cases, it increases as the gate voltage is reduced further, as Figure 6.4.3 illustrates. A similar effect is seen in polysilicon TFTs, where it is attributed to field-assisted tunneling through midgap states present at grain boundaries; reducing the gate voltage increases the high fields at the drain edge of the gate, increasing the tunneling rate. The value typically cited for the off-current in TFTs is the minimum drain current at a specified drain voltage. For initial design work prior to more accurate computeraided simulations that take into account the entire drain current characteristic, a better figure to use for off-current is the worst-case leakage over the expected range of voltages, rather than the minimum leakage found at a single voltage point. Gate currents are often ignored, but for the polymer dielectrics and low-temperature inorganic dielectrics often used in OTFTs, the gate current can be significant.

8080_C006.fm Page 559 Tuesday, April 3, 2007 2:15 PM

Organic Thin-Film Transistors for Flat-Panel Displays

559

Indeed, sometimes the off-current is not dominated by drain-to-source leakage, but rather by drain-to-gate leakage. Determining whether the off-current arises from leakage to source or gate is the first step in identifying the physical source of the off-current. It is sometimes observed that gate leakage is much smaller when an OTFT is off than when it is on. This occurs because a conducting channel presents a larger conducting area for gate leakage to occur.

6.4.2.5 CAPACITANCES The significant capacitances between the terminals of an OTFT are the gate-source and gate-drain capacitances. The drain-source capacitance is much smaller and can usually be ignored. The gate-source and gate-drain capacitances are each made of two capacitances in parallel (Figure 6.4.4), one due to the parasitic overlap capacitance between gate and source/drain and the other due to the capacitance between gate and channel. The channel is not a terminal of the device, so the capacitance between gate and channel must be partitioned between source and drain to obtain device terminal capacitances. Thus, CGS = CGS,overlap + CGS,channel and CGD = CGD,overlap + CGD,channel. The overlap capacitances are voltage independent and can be approximated as fixed parallel plate capacitances, with CGS,overlap = CGD,overlap = W Loverlap Cox, where Loverlap is the overlap length (Figure 6.4.4). In a MOSFET, the channel capacitance is a complicated bias-dependent quantity. The standard model is as follows [5]. For VGS < Vt the channel capacitance is zero, but for VGS ≥ Vt it is greater than zero and depends on VDS. When VDS = 0, the channel capacitance is simply the parallel plate expression WLCox, which by symmetry is partitioned equally between source and drain terminals so that CGD,channel = CGS,channel = WLCox/2. The situation is more complicated for VDS ≠ 0 because symmetry no longer holds. In saturation, CGS,channel = 2/3 WLCox and CGD,channel = 0. For values of VDS between zero and the voltage required for saturation, there is a smooth transition between these regions. In display design, one is often concerned with the charge injected into a node of a circuit when a TFT used as a switch is shut off. Thus, one is led to consider not the small-signal capacitances, but rather the charges QD, QG, and QS associated Loverlap

Lchannel

Loverlap

Source

Drain

Organic Dielectric semiconductor

Gate CGS,overlap CGS,channel

CGD,overlap CGD,channel

FIGURE 6.4.4 The gate-source capacitance CGS and the gate-drain capacitance CGD are each composed of two components, one due to the overlap between gate and source/drain and the other due to the capacitance between gate and channel, which is partitioned between source and drain.

8080_C006.fm Page 560 Tuesday, April 3, 2007 2:15 PM

560

Organic Field-Effect Transistors

with source, gate, and drain terminals and how these charges change during voltage transitions. Prior to a switch being shut off, VDS ≈ 0, so the channel charge is partitioned equally between source and drain. When a TFT switch is shut off by switching the gate voltage from VGS,on to VGS,off, the charge injected into the source and drain terminals ΔQS and ΔQD is composed of two parts, one due to the overlap capacitance and one due to the channel charge [6]: ΔQD = ΔQS = WLoverlapCox (VGS ,on − VGS ,off ) +

WL Cox (VGS ,on − Vt ) 2

(6.4.6)

In an OTFT, some of the channel charge does not exit the transistor immediately but is trapped in deep states and leaves more gradually. Thus, charge injection from the channel may not be immediate; charge may “dribble out” over an extended period. The overlap length Loverlap can be large in OTFTs, sometimes 10 μm or more. It may be larger than the channel length L itself. Thus, the overlap contributes a large amount of injected charge in OTFTs. The reason for the overlap is that the fabrication processes used to make OTFTs, like those for a-Si TFTs, do not produce source/drain electrodes that are self-aligned to the gate. In order to ensure that the channel induced by the gate reaches the source and drain, a misalignment tolerance must be designed into the layout of these regions. This misalignment tolerance is the overlap length. For OTFTs on plastic substrates, the dimensional instability of the plastic film requires additional allowance for misalignment, so Loverlap may have to be even larger. The injected charge could be reduced dramatically if a self-aligned OTFT process could be developed like the self-aligned processes used for bulk silicon MOSFETs, eliminating the overlap capacitance. The elimination of Loverlap would also permit OTFTs to follow the integration law that has been followed by silicon microelectronics, whereby the maximum frequency of operation scales with 1/L2 so that large performance gains come from reduced gate length.

6.4.2.6 CONTACT RESISTANCE Making good electrical contact to organic semiconductors is sometimes difficult. As a result there are often large parasitic resistances in series with the source and drain [7]. The value of the contact resistance typically depends on the gate voltage, decreasing with increasing gate voltage, just like the channel resistance. Nevertheless, the contact resistances are distinct from the channel resistance because they do not scale with channel length, but rather are channel-length independent. Indeed, sometimes they can be modeled simply as a fixed added channel length. For example, the contact resistances may behave very similarly to an extra 2 μm of channel length, so an OTFT with channel length L behaves as if it had a channel length L + 2 μm. In other cases, the contact resistance is nonlinear and, for small VDS, a large fraction of VDS may be dropped across the contact resistances rather than the intrinsic FET, leading to “hooked” output characteristics (Figure 6.4.5). Since small values of VDS are used to extract linear mobility, the measured linear mobility may be significantly smaller than the measured saturation mobility.

8080_C006.fm Page 561 Tuesday, April 3, 2007 2:15 PM

Organic Thin-Film Transistors for Flat-Panel Displays

561

0.00E + 00

ID (A)

–2.00E – 08 –4.00E – 08 Id (Vgs = 0V) Id (Vgs = –5V) Id (Vgs – 10V) Id (Vgs = –15V) Id (Vgs= –20V)

–6.00E – 08 –8.00E – 08 –1.00E – 07 –20

–15

–10

–5

0

VDS (V)

FIGURE 6.4.5 Contact resistance in OTFTs can lead to “hooked” output characteristics, as in this pentacene OTFT.

The total resistance of the FET is the sum of the channel resistance and the contact resistances. Because the channel resistance scales with channel length but the contact resistances are independent of channel length, the performance of short channel devices can be strongly degraded by contact resistance. It is common to see measured OTFT mobility decrease as channel length is reduced because contact resistance comes to dominate the total resistance. The term “ohmic contact” is often used phenomenologically to describe a contact that has resistance low enough to be neglected compared to the channel resistance. Thus, “ohmic contacts” might be obtained at longer channel lengths but not at shorter lengths.

6.4.2.7 BIAS-STRESS INSTABILITY

AND

HYSTERESIS

Bias-stress instability and hysteresis are memory effects in which the DC characteristics of a transistor at a given time depend on voltages applied to the device in the past. From the standpoint of design, these memory effects are undesirable. As we shall see later, it is difficult but not impossible to design in a way that is tolerant of these effects. Generally, bias-stress instability refers to long-term changes in the transistor characteristics that do not saturate but continue without limit until the device is rendered useless. Hysteresis refers to short-term reversible shifts in the characteristics that lead to looping in the measured characteristics, depending on the direction in which the bias voltages are swept. There is no sharp distinction between biasstress instability and hysteresis, and the two may arise from the same or similar physical causes. There have been few studies of either phenomenon in OTFTs [8–10]. In general, the bias-stress instability of OTFTs fabricated on inorganic gate dielectrics behaves as follows: The primary effect of positive gate bias is to shift the threshold voltage to more positive voltages, and the primary effect of negative bias is to shift the

8080_C006.fm Page 562 Tuesday, April 3, 2007 2:15 PM

562

Organic Field-Effect Transistors

5.0 × 10–5

Oﬀ to On On to Oﬀ

1 × 10–6 1 × 10–7

3.0 × 10–5

1 × 10–8

2.0 × 10–5

–Id (A)

–Id (A)

4.0 × 10–5

1 × 10–5

1 × 10–9 1.0 × 10–5

1 × 10–10

(a)

10–11

0.0 –40

–20

0 Vgs (V)

20

40

FIGURE 6.4.6 Hysteresis leads to looping transistor characteristics, as seen in the linearregion transfer characteristics of this OTFT made using pentacene on thermal SiO2. The drain current is plotted on a linear scale (left-hand vertical axis) and a logarithmic scale (right-hand vertical axis).

threshold voltage to more negative voltages. A similar effect is observed in a-Si TFTs [11] and to a lesser extent in polysilicon TFTs [12]. In our laboratory, we have found that the on-state bias-stress instability in pentacene OTFTs is somewhat worse than in a-Si TFTs, and the off-state instability is much worse. As expected, the details depend strongly on the type of organic semiconductor used and on the presence of oxygen and water vapor, just as with OLEDs, which may lead to similar encapsulation requirements. Hysteresis can be seen in the looping pentacene OTFT characteristics in Figure 6.4.6. The same phenomenon also manifests itself as an overshoot or undershoot in the drain current when a gate voltage step is applied. Similar effects have been observed in a-Si TFTs [13]. The details of this drain-current transient have been used to determine whether the responsible trapping states in pentacene OTFTs are electron traps or hole traps [14]. When a polymer dielectric is used, there is an additional complicating factor that slow polarization of the dielectric causes an instability in a direction opposite to the bias-stress instability and the hysteresis in organic semiconductors; thus, there are two competing mechanisms, with a possible crossover between them after a certain stress period [15,16]. Slow polarization in a polymer dielectric is often due to residual polar solvent in the dielectric or water absorption from the air. It is natural to characterize this type of dielectric behavior by analyzing the frequency-dependent capacitance C(f) at sufficiently low frequencies. Figure 6.4.7 compares C(f) of two polymer dielectrics that manifest slow polarizability with a dielectric that does not. At very low frequencies (below 100 mHz), the capacitances of the dielectrics with absorbed water increase significantly. In the time domain, this leads to memory effects in which voltages applied in the past affect how the OTFT responds to applied signals in the present.

8080_C006.fm Page 563 Tuesday, April 3, 2007 2:15 PM

Organic Thin-Film Transistors for Flat-Panel Displays

563

3.00E–08

Capacitance (F)

2.50E–08 Polymer dielectrics with absorbed water

2.00E–08 1.50E–08 1.00E–08 5.00E–09 0.00E+00 1.00E–03

1.00E–02

1.00E–01 1.00E+00

1.00E+01

1.00E+02 1.00E+03

Frequency (Hz)

FIGURE 6.4.7 A comparison of the frequency-dependent capacitance C(f) of two polymer dielectrics with slow polarizability due to water absorption with that of a dielectric that does not show this problem.

As a result of hysteresis phenomena, measurements of μFE and Vt can depend strongly on the gate sweep direction. This can be seen in the way the transfer characteristics in Figure 6.4.6 depend on sweep direction. Extracted OTFT parameters have sometimes been contaminated by hysteresis effects. An IEEE standard has been written in an effort to prevent these and other errors in OTFT characterization [17]. Attempts to derive μFE and Vt in the presence of hysteresis, perhaps by averaging the on-to-off and the off-to-on characteristics, are probably doomed to failure. It is preferable to extract parameters only when the hysteresis is small.

6.4.2.8 LIGHT SENSITIVITY Thin-film transistors are sensitive to light. The off-current increases under illumination because of carriers generated in the channel region and swept to the source and drain electrodes by the applied drain voltage. As a result, TFT switches become leaky under bright illumination. Transmissive displays, such as those used in notebook computers and digital cameras, use a backlight behind the display to provide illumination, so photosensitivity is a problem. Even reflective displays must cope with ambient illumination. A metal gate electrode positioned between the active region and the source of brightest light (whether a backlight or the ambient) provides some light shielding, but “light piping” still occurs laterally in the layers of the display, so additional shielding and absorbing layers are needed. It is unclear at this time how the light sensitivity of OTFTs compares to siliconbased TFTs, but it is certain that similar light-shielding measures must be employed. Studies in our laboratory have shown that, in addition to the photocurrents generated by incident illumination, long-lived light-induced threshold shifts can occur in pen-

8080_C006.fm Page 564 Tuesday, April 3, 2007 2:15 PM

564

Organic Field-Effect Transistors

tacene OTFTs in the presence of a noninert atmosphere, suggesting that photooxidation of the pentacene molecule is taking place [18]. Light shielding may have to be combined with barrier layers in order to avoid the effects of incident illumination.

6.4.2.9 TFT NONUNIFORMITY Within limits, less than optimal TFT performance can be tolerated for all of the parameters discussed so far, as long as the parameters are uniform across the display. However, nonuniform parameters can lead to visible variations in color or gray level, and this is aesthetically displeasing and commercially nonviable. Random brightness variations across the display, known as mura, are more of a problem than smooth variations because the low spatial frequencies in smooth variations are less visible to the human visual system. This has been an issue for polysilicon TFTs, which manifest random variations in threshold voltage and off-currents due to the random grain structure of the polysilicon. In our laboratory, we have developed routines for the analysis of OTFT nonuniformity using automated probing and characterization of transistor arrays, followed by statistical analysis. Figure 6.4.8 (See color insert following page 468) shows uniformity maps of threshold voltage and mobility for a 1- × 1-cm 200-OTFT array using pentacene devices with a SiO2 gate dielectric. Each color difference represents one standard deviation. Black cells represent nonfunctional TFTs. For this array, the average Vt was +0.34 V with a standard deviation of 0.54 V, and the average mobility was 0.40 cm2/V-s with a standard deviation of 0.05 cm2/V-s. We analyzed spatial parameter variations, including parameter correlations between pairs of closely spaced OTFTs, in order to predict the performance of matched pairs such as those that may be used in current mirrors in AMOLED display pixels. The distributions show smooth variations as well as random variations that cause nonuniformity at short distance scales.

6.4.3 DISPLAY TECHNOLOGIES The flat-panel display technologies that can benefit from an OTFT backplane are AMLCDs, electrophoretic displays, and AMOLED displays. Other flat-panel technologies are not compatible with OTFT backplanes because the voltages are too high (e.g., plasma displays and inorganic electroluminescent displays) or because the area is too large for an active matrix (e.g., LED displays such as those used for video signage). In this section, we discuss the use of OTFTs for AMLCDs and electrophoretic displays together, since these display technologies are similar, followed by a discussion of OTFTs for AMOLED displays.

6.4.3.1 LIQUID-CRYSTAL

AND

ELECTROPHORETIC DISPLAYS

6.4.3.1.1 Introduction The first reported OTFT display was a 16 × 16 pixel electrophoretic display demonstrated by researchers at Bell Laboratories and E Ink Corporation in 2001 [19]. Figure 6.4.9 shows operation of the 5 in.2 display, which used pentacene TFTs on a flexible film of polyethylene terephthalate (PET). Later in 2001 AMLCDs using

8080_C006.fm Page 565 Tuesday, April 3, 2007 2:15 PM

Organic Thin-Film Transistors for Flat-Panel Displays

565

OTFT Array map of measured values R0 1T R0 1B R0 2T R0 2B R0 3T R0 3B R0 4T R0 4B R0 5T R0 5B R0 6T R0 6B R0 7T R0 7B R0 8T R0 8B R0 9T R0 9B R1 0T R1 0B

LOT ID : C0 1 3. 115E +00 2. 945E +00 2. 726E +00 2. 675E +00 1. 754E +00 1. 803E +00 1. 420E +00 1. 813E +00 1. 614E +00 8. 876E -01 1. 242E +00 8. 747E -01 8. 785E -01 7. 418E -01 1. 030E +00 6. 337E -01 7. 000E -01 5. 079E -01 8. 298E -01

042000 C 02 2 . 690E +00 2 . 326E +00 1 . 729E +00 1 . 446E +00 9 . 689E -01 1 . 107E +00 1 . 201E +00 1 . 207E +00 1. 159E +00 9 . 204E -01 1 . 324E +00 5 . 328E -01 6 . 265E -01 3 . 324E -01 4 . 862E -01 4 . 062E -01 5 . 638E -01 4 . 768E -01 6 . 500E -01

Parameter : Vtx C 04 C0 5 C 06 C0 7 C 08 C0 9 C 10 1 . 171 E+ 00 1. 532E +00 9 .5 47E -01 2 .6 21 E- 01 7. 5 34E -01 7 .8 70E -01 1 . 229 E+ 00 9. 728 E- 01 1. 216E +00 9 .4 53E -01 5 .9 95 E- 01 5. 4 61E -01 7 .3 76E -01 1 . 809E +00 1 . 422 E+ 00 1. 310 E+ 00 1. 1 43E +00 8 .1 11 E- 01 1. 0 29E +00 7 .4 70E -01 1 . 467E +00 1 . 158 E+ 00 1. 137 E+ 00 1. 774E +00 1 .1 01E +00 9 .6 33 E- 01 1. 0 00E +00 1 .0 46E +00 1 . 515E +00 1 . 070 E+ 00 9. 643 E- 01 9. 992E -01 1 .1 20E +00 4 .0 74E -01 5 .7 89E -01 9 . 070E -01 8 . 383 E- 01 7. 398E -01 9 .3 65E -01 3 .8 06 E- 01 7. 1 12E -01 4 .1 84E -01 9 . 629E -01 8 . 551 E- 01 1. 016 E+ 00 6. 955E -01 1 .0 22E +00 -1. 0 27E +00 1 .3 07E -01 8 . 066E -01 6 . 897 E- 01 9. 164 E- 01 6. 254E -01 -1. 8 33E -01 2 .5 41 E- 01 6. 3 03E -01 1 . 089E +00 9 . 137 E- 01 1. 292 E+ 00 1. 439E +00 8 .8 92E -01 -3. 332E -01 5 .6 73E -01 1 .4 24E +00 1 . 157E +00 1 . 366 E+ 00 1. 050 E+ 00 1. 638E +00 5 .6 51E -01 1 .3 86 E+ 00 5. 1 32E -01 1 .0 59E +00 7 . 971E -01 1 . 011 E+ 00 1. 594 E+ 00 1. 107E +00 1 .5 57E +00 9 .6 57 E- 01 6. 7 30E -01 9 .9 52E -01 1 . 279E +00 1 . 109 E+ 00 1. 591 E+ 00 1. 021E +00 1 .5 97E +00 1 .0 82 E+ 00 6. 3 87E -01 9 .4 50E -01 5 . 148E -01 8 . 579 E- 01 8. 943 E- 01 1. 157E +00 1 .1 82E +00 9 .4 74 E- 01 8. 0 80E -01 7 .3 07E -01 8 . 678E -02 1 . 304 E+ 00 2. 154 E- 01 1. 626E +00 1 .4 29E +00 1 .0 74 E+ 00 1. 1 04E +00 5 .7 22E -01 6 . 123E -01 5 . 899 E- 01 5. 748 E- 01 9. 824E -01 1 .0 49E +00 1 .0 59 E+ 00 9. 1 90E -01 8 .1 40E -01 6 . 476E -01 9 . 465 E- 01 7. 051 E- 01 1. 359E +00 1 .1 63E +00 1 .0 64 E+ 00 8. 7 64E -01 3 .7 88E -01 5 . 548E -01 4 . 201 E- 02 1. 092 E+ 00 1. 388E +00 6 .3 84E -01 7 .2 48 E- 01 7. 3 46E -01 6 .2 69E -01 7 . 050E -01 4 . 025 E- 01 7. 662 E- 01 7. 6 77E -01 8 .5 66E -01 2 .7 60E -01 -6. 597E -02 1 . 035 E+ 00 5. 719 E- 01 3. 294E -01 5 .0 39E -01 1 .2 19 E+ 00 6. 6 84E -01 -2. 9 04E -01 5 . 002E -01 7 . 616 E- 01 7. 272 E- 01 4. 204E -01 5 .7 79E -01 3 .4 18 E- 01 4. 9 64E -01 C0 3

Min.

Max.

1

–1.00E + 15 –6.766E – 01

3

–6.766E – 01 –1.364E – 01

14

–1.364E – 01

78

4.038E – 01

4.038E – 01 9.441E – 01

70

9.441E – 01

1.484E + 00

15

1.484E + 002

2.025E + 00

1

2.025E + 00

2.565E + 00

5

2.565E + 00

1.00E + 15

Min.

Max.

2

–1.00E + 10

2 . 543E – 01

OTFT Array map of measured values R 01T R 01B R 02T R 02B R 03T R 03B R 04T R 04B R 05T R 05B R 06T R 06B R 07T R 07B R 08T R 08B R 09T R 09B R 10T R 10B

LOT ID : C0 1 4. 949E -01 4. 880E -01 5. 025E -01 5. 358E -01 4. 973E -01 4. 483E -01 4. 271E -01 4. 503E -01 4. 705E -01 4. 168E -01 4. 322E -01 4. 116E -01 4. 312E -01 4. 509E -01 4. 395E -01 4. 601E -01 4. 787E -01 4. 487E -01 4. 647E -01

042000 C 02 5. 240E -01 5. 040E -01 4. 560E -01 4. 390E -01 4. 333E -01 4. 291E -01 4. 234E -01 4. 162E -01 4. 417E -01 4. 199E -01 4. 291E -01 4. 251E -01 4. 256E -01 4. 211E -01 4. 384E -01 4. 227E -01 4. 377E -01 4. 114E -01 3. 927E -01

Parameter : usat_max C 04 C0 5 5 . 186E -01 4 . 936E -01 4 . 032E -01 4. 888E -01 4 . 344E -01 4 . 232E -01 4 . 700E -01 4 . 543E -01 4 . 179E -01 4 . 300E -01 4 . 086E -01 3 . 874E -01 4 . 335E -01 4 . 289E -01 4 . 095E -01 3 . 779E -01 3 . 572E -01 4 . 289E -01 3 . 826E -01 3 . 798E -01 3 . 803E -01 3 . 751E -01 3 . 734E -01 4 . 172E -01 3 . 864E -01 3 . 722E -01 3 . 554E -01 3 . 889E -01 3 . 467E -01 3 . 658E -01 3 . 798E -01 3 . 619E -01 3 . 895E -01 3 . 655E -01 3 . 208E -01 3 . 937E -01 3 . 753E -01 2 . 787E -01 4 . 026E -01 3 . 907E -01 3 . 449E -01 4 . 241E -01 3 . 915E -01 3 . 726E -01 4 . 012E -01 3 . 286E -01 4 . 045E -01 4 . 116E -01 3 . 656E -01 3 . 989E -01 3 . 231E -01 4 . 071E -01 1 . 864E -01 3 . 817E -01 4 . 224E -01 4 . 212E -01 C0 3

C 06 4 . 694E -01 4 . 273E -01 3 . 812E -01 3 . 596E -01 3 . 743E -01 3 . 504E -01 3 . 679E -01 3 . 588E -01 3 . 578E -01 3 . 421E -01 3 . 337E -01 3 . 268E -01 3 . 565E -01 3 . 848E -01 3 . 779E -01 3 . 639E -01 4 . 067E -01 4 . 196E -01

C0 7 4. 1 47E -01 4. 3 32E -01 3 .8 50E -01 3. 8 27E -01 3. 5 90E -01 3. 5 59E -01 3. 4 49E -01 3. 5 70E -01 3. 2 76E -01 3. 1 58E -01 3. 4 25E -01 3. 5 83E -01 3. 7 03E -01 3. 5 45E -01 3. 8 57E -01 3. 8 13E -01 3. 8 98E -01 3 .7 30E -01 3. 9 76E -01 4. 0 88E -01

C 08 3. 999E -01 4. 014E -01 3. 823E -01 3. 948E -01 3. 600E -01 3. 589E -01 3. 402E -01 3. 552E -01 3. 612E -01 3. 506E -01 3. 608E -01 3. 709E -01 3. 794E -01 3. 970E -01 3. 876E -01 3. 901E -01 3. 981E -01 4. 200E -01

C0 9 4. 264E -01 4. 149E -01 3. 882E -01 3. 693E -01 3. 487E -01 3. 580E -01 3. 541E -01 3. 414E -01 3. 606E -01 3. 642E -01 3. 651E -01 3. 578E -01 3. 623E -01 3. 702E -01 3. 607E -01 3. 897E -01 3. 963E -01 4. 063E -01 4. 002E -01

C 10 4. 422E -01 4. 273E -01 3. 940E -01 2. 505E -01 3. 577E -01 3. 552E -01

3. 431E -01 3. 738E -01 3. 960E -01 3. 887E -01 3. 833E -01 3. 793E -01 3. 839E -01 4. 009E -01 3. 900E -01 4. 007E -01 3. 538E -01 3. 900E -01

1

2. 543E – 01

3 . 013E – 01

15

3. 013E – 01

3 . 483E – 01

84

3. 483E – 01

3 . 953E – 01

63

3. 953E – 01

4 . 423E – 01

14

4. 423E – 01

4 . 893E – 01

8

4. 893E – 01

5 . 364E – 01

0

5. 364E – 01

1.00E +15

FIGURE 6.4.8 (See color insert following page 468.) Threshold voltage and mobility uniformity maps for a 1- × 1-cm 200-OTFT array using pentacene devices with a SiO2 gate dielectric.

pentacene OTFTs with polymer-dispersed liquid-crystal (PDLC) material were reported by two independent groups at about the same time. A team from Sarnoff Corporation, Penn State University, and Kent State University reported 16 × 16 PDLC AMLCDs using pentacene OTFTs on polyethylene naphthalate (PEN) substrates [20]. The high mobility of the pentacene OTFTs permitted a refresh rate of 60 Hz, with line times deliberately shortened to the 69 μsec appropriate for a quarter VGA (QVGA, 320 × 240) display so that the measured display performance would be scaleable to a larger display. This was the first reported video-capable AMLCD on a plastic substrate. The second group, at Bell Laboratories, reported 2 × 3 PDLC displays on PET substrates, but used only static rather than dynamic switching of the driving signals [21]. Still later in 2001, Philips demonstrated a video-rate 64 × 64 PDLC display on a glass substrate using a solution-deposited polythienylenevinylene precursor film that was converted into a semiconductor (Figure 6.4.10) [22]. Recent significant results include a 4.7-in. diagonal QVGA (320 × 240) electrophoretic display using a solution-deposited pentacene precursor on PEN from Polymer Vision [23] — the

8080_C006.fm Page 566 Tuesday, April 3, 2007 2:15 PM

566

Organic Field-Effect Transistors

FIGURE 6.4.9 The first reported OTFT display, a 5-in. square 16 × 16 pixel electrophoretic display using pentacene TFTs on a flexible film of polyethylene terephthalate (PET). (From Rogers, J.A. et al., Proc. Nat. Acad. Sci., 98, 4835, 2001. With permission.)

8080_C006.fm Page 567 Tuesday, April 3, 2007 2:15 PM

Organic Thin-Film Transistors for Flat-Panel Displays

567

FIGURE 6.4.10 A video-rate 64 × 64 PDLC OTFT display on a glass substrate using a solution-deposited polythienylenevinylene precursor film that was converted into a semiconductor. (From Huitema, H.E.A. et al., Nature, 414, 599, 2001. With permission.)

highest resolution OTFT display reported to date — and a 3 in. diagonal 64 × 128 color AMLCD using pentacene OTFTs on glass from ERSO/ITRI (Figure 6.4.11) (See color insert following page 468) [24]. Table 6.4.1 lists these and other reported AMLCDs and electrophoretic displays using OTFT backplanes. 6.4.3.1.2 Electro-Optic Behavior of Twisted-Nematic Liquid Crystals Conventional AMLCDs use nematic liquid crystals in a twisted configuration between two sheets of glass or plastic, one of which is the backplane containing the TFTs. The space between the two sheets is called the cell gap. Display operation relies on some subtle features of the electro-optic behavior of nematic liquid crystals [25]. We will not discuss the physics of nematic liquid crystals here. The interested reader can consult the standard reference on the physics of liquid crystals by de Gennes and Prost [26]. From the perspective of OTFT display design, the significant fact is that the electro-optic behavior is an RMS-responding effect. Nematic liquid crystals respond to the applied electric field E through induced polarization, so the interaction is proportional to E2. The liquid crystal does not respond instantaneously to E2, but rather to the average of E2 over a time interval corresponding to the response time of the liquid crystal, which is determined by its viscoelastic properties. Thus, a liquid crystal responds to the magnitude but not the sign of the applied voltage. Figure 6.4.12 shows a typical transmission versus voltage curve. As the

8080_C006.fm Page 568 Tuesday, April 3, 2007 2:15 PM

568

Organic Field-Effect Transistors

FIGURE 6.4.11 (See color insert following page 468.) A portion of a 3-in. diagonal 64 × 128 color AMLCD using pentacene OTFTs on glass from ERSO/ITRI. (From Ho, J.-C. et al., SID Int. Symp. Dig. Tech. Papers, 1298, 2004. Copyright Society for Information Display. With permission.)

TABLE 6.4.1 Active Matrix LCDs and Electrophoretic Displays Using Organic TFTs Organization

Display type

Bell Labs Sarnoff, Penn State, Kent State Bell Labs Philips

Electrophoretic PDLC PDLC PDLC

Plastic Logic Philips ERSO/ITRI Plastic Logic, E Ink Hitachi

PDLC Electrophoretic Color LCD Electrophoretic Color LCD

Substrate

Pixel Count

Ref.

Pentacene Pentacene

PET PEN

16 × 16 16 × 16

19 20

Pentacene Polythienylenevinylene precursor Polyfluorene polymer Pentacene precursor Pentacene Polyfluorene polymer Pentacene

PET Glass

2×3 64 × 64

21 22

Glass PEN Glass PET Glass

80 × 60 320 × 240 64 × 128 80 × 60 80 × 80

61 23 24 62 63

Semiconductor

x-axis indicates, the pixel responds to the RMS average of the applied voltage. Although the transmission is a complicated nonlinear function of the applied voltage, in an AMLCD this is well characterized and the data can be “warped” to take the nonlinearity into account. It is important that the average of the applied voltage over a long time-period not have a significant DC component. Small DC components of a few hundred millivolts or less lead to the transport of ionic impurities across the liquid-crystal cell gap, allowing an ion-induced compensating DC potential to be built up on the inside walls of the display. In itself this is not harmful, but the effect of a built-in DC potential is that the liquid crystal no longer responds to the applied signal alone, but rather to the sum of the applied voltage and the built-in potential. Larger DC components can lead to irreversible electrolysis of the liquid-crystal material if the liquid crystal is in direct contact with conducting electrodes.

8080_C006.fm Page 569 Tuesday, April 3, 2007 2:15 PM

Organic Thin-Film Transistors for Flat-Panel Displays

569

% Transmission

35

0 0

1

2

3

4

5

Applied RMS voltage (V)

FIGURE 6.4.12 Electro-optic curve for a typical twisted-nematic liquid-crystal cell.

The RMS-responding property of the liquid crystal allows the requirement that DC voltage components be avoided to be easily satisfied. The display driver circuits that provide data voltages to the display alternate the polarity from one frame to the next. Various spatial inversion patterns are possible. In frame inversion, all pixels in the display are written with one polarity in one frame, then the opposite polarity in the next frame. But this can lead to visible flicker because a small fixed DC component is nearly always present on the data voltages relative to the common electrode voltage, causing alternating darker and lighter frames. The flicker frequency is half the refresh rate, or 30 Hz at a typical 60-Hz refresh rate, which is quite visible. It is more common to use row inversion, column inversion, or pixel (or dot) inversion, in which the spatial inversion pattern is horizontal stripes, vertical stripes, or a checkerboard. These inversion methods make flicker hard to see by hiding it at high spatial frequencies. Producing the appropriate alternating-polarity data signals is the responsibility of the display driver circuits. However, the pixel designer must reckon with the fact that data voltages will have alternating polarities and must consider any DC component that may be contributed by the pixel design. 6.4.3.1.3 Electro-Optic Behavior of Electrophoretic Materials Electrophoretic materials contain a charged pigment dispersed in an optically contrasting material or a mixture of oppositely charged contrasting pigments dispersed in a neutral material. By applying an electric field, the pigment particles are moved to the front or back surface of the display, producing an electro-optic effect. A display made with electrophoretic material operates in a reflective mode using ambient light. Motion of the pigment materials is needed to change the optical state, so electrophoretic displays typically do not operate at video rates, but have response times of 100–300 msec. Because the only force on the pigment particles is the applied field, the movement of the pigment, and therefore its optical state, is only a function of the time-integrated electric field

8080_C006.fm Page 570 Tuesday, April 3, 2007 2:15 PM

570

Organic Field-Effect Transistors

∫

t 0

E ( τ) d τ

(This analysis is only approximate because it ignores forces between the particles and the effect of initial pigment conditions.) When no electric field is applied, there is no applied force, so the optical state does not change, at least not over short periods. Thus, an electrophoretic material is a bistable material whose reflectivity depends on the preceding integral. Bistability can provide significant display power savings in applications that leave an image on the display for a period, such as electronic books and maps, because the display module can be powered down at these times. Typical data voltages are in the range of –15 to +15 V, where negative voltage drives the display toward one optical state and positive voltage drives it toward the opposite state. Therefore, an electrophoretic display is not an RMS-responding display like an LCD, but rather responds to opposite polarities by giving opposite optical responses. 6.4.3.1.4 Liquid-Crystal and Electrophoretic Display Architecture The architecture of OTFT liquid-crystal and electrophoretic displays is not unlike that of their a-Si TFT and polysilicon TFT counterparts. The basic design is shown in Figure 6.4.13(a), and a pixel cross-section is shown in Figure 6.4.13(b). The display consists of an array of TFT switches arranged in rows and columns, one TFT per pixel. For color displays, each pixel is divided into three subpixels, one each for red, green, and blue, and there is one TFT per subpixel. All TFTs in the same row have their gates connected to a common row line (or select line), and all TFTs in the same column have a source/drain terminal connected to a common column line (or data line). The other source/drain terminal of each TFT connects to a storage capacitor internal to the pixel and to the pixel electrode that faces the electro-optic material. The other electrode of the storage capacitor is connected to a separate ground line or, more commonly, to the previous select line as a convenient nearly DC potential. On the other side of the cell gap is an unpatterned common electrode shared by all pixels and connected to a DC voltage Vcom. For transmissive displays, the electrodes on both sides of the cell gap are made from a transparent conductor such as indium tin oxide (ITO); for reflective displays only the common electrode is transparent, while the pixel electrode can be reflecting. Figure 6.4.14 shows a portion of an OTFT backplane for a PDLC display using pentacene devices on a PEN substrate [20]. The large electrode in the middle of each pixel is the ITO pixel electrode. The storage capacitor is formed between this electrode and a grounded capacitor return line that runs over it along the row direction, with the gate dielectric lying in between. In the fabrication of OTFT backplanes there are complications related to the order in which the organic semiconductor and source/drain contacts are deposited and how they are patterned. The semiconductor must be patterned in order to avoid large leakage currents through the ungated areas. Complications arise because the

8080_C006.fm Page 571 Tuesday, April 3, 2007 2:15 PM

Organic Thin-Film Transistors for Flat-Panel Displays

571

(a) Select line

Data line

Liquid crystal

Cstorage Vcom

Cover sheet

(b) ITO common electrode Liquid crystal

Source

Passivation Organic SC Gate dielectric Gate electrode

Drain ITO pixel electrode

Backplane substrate

FIGURE 6.4.13 (a) Schematic and (b) cross-section of an OTFT AMLCD pixel (not to scale).

FIGURE 6.4.14 A portion of an OTFT backplane for a PDLC display using pentacene devices on a PEN substrate.

8080_C006.fm Page 572 Tuesday, April 3, 2007 2:15 PM

572

Organic Field-Effect Transistors

organic semiconductor is very sensitive to processing after it is deposited. This leads to two problems: •

•

It is difficult to pattern the organic semiconductor using standard thinfilm photolithographic methods because these methods require the semiconductor to withstand organic solvents. Attempts to use these methods have failed. Patterning using shadow masking is not a realistic option because the pattern is too fine and the alignments too tight. Penn State University developed a technique for use with pentacene in which a photosensitized polyvinyl alcohol (PVA) film is used as a photoresist [27]. This method was used to pattern the pentacene in the display shown in Figure 6.4.14. Another method developed by IBM uses parylene as a hard mask to protect pentacene from photolithographic chemicals [28,29]. Inkjetting a soluble semiconductor solves the patterning problem because deposition and patterning are performed at the same time. The source/drain electrode can contact the organic semiconductor from above or below. The best OTFT performance is usually obtained with top contacts (i.e., when the source/drain electrodes lie on top of the semiconductor), probably because with bottom contacts the semiconductor does not form a good film at the abrupt electrode edges. However, top contacts require source/drain deposition and patterning after the semiconductor is deposited, which is problematic. Again, shadow masking is possible in principle, but difficult in practice. In our laboratory, a process has been developed for photolithographically patterning top contacts on pentacene, but it has not been widely adopted [30]. Most OTFT backplanes have used bottom contacts and suffered the performance reduction.

The active matrix operates as follows. By applying a voltage pulse to one of the select lines, the switches in that row are turned on, and analog voltage levels applied to the data lines by the display drivers pass through the switches, charging each pixel’s internal capacitance. The select lines are pulsed sequentially, row by row, and thus all the pixels in the display are written with analog voltages. Then the process starts again for the next display frame. The duration of each pulse is the line time Tline, and the time required to write all the pixels in the display is the frame time Tframe = Nrow Tline, where Nrow is the number of rows in the display. The display is rewritten at the refresh rate Rrefresh = 1/Tframe. As Figure 6.4.13(a) shows, each display pixel has two capacitors. One capacitor represents the liquid crystal or electrophoretic material, which electrically behaves very much like a capacitance. The other capacitor is a storage capacitor that is designed into the pixel to overcome some problems that would occur without it. There are three reasons for the storage capacitor: •

As a liquid crystal responds to an applied voltage, its capacitance changes. The change can be as large as a factor of three. In the absence of the storage capacitor, the stored voltage on the pixel can change by this same

8080_C006.fm Page 573 Tuesday, April 3, 2007 2:15 PM

Organic Thin-Film Transistors for Flat-Panel Displays

•

•

573

factor during the time the switch is off. A similar voltage reduction occurs in an electrophoretic display due to the motion of the charged ink particles. The storage capacitor “ballasts” the pixel against this voltage change. When the switch is shut off, charge is injected into the pixel, shifting the pixel voltage by 0.1–1 V from the level of the applied data. Earlier we discussed the origins of charge injection. The resulting voltage shift is given various names: push down (for n-channel TFTs) or push up (for pchannel TFTs), feedthrough or kickback voltage, and pedestal voltage error. Display manufacturers compensate for the shift by adjusting the DC voltage on the common electrode Vcom. The optimum value of Vcom can be found by putting the display in frame inversion mode and adjusting Vcom to minimize visible flicker. However, charge injection is not perfectly uniform, but rather varies across the display due to capacitance nonuniformities and is also data dependent (see Equation 6.4.6). Thus, charge injection can create display nonuniformities and residual DC levels. By using a large enough storage capacitance, the voltage shift due to charge injection can be significantly reduced. The storage capacitor reduces the effect of leakage through the switch. This is usually not a serious problem in a-Si TFTs, which have very low off-current. But leakage is a problem with poly-Si TFTs and OTFTs. In these types of backplanes, the storage capacitor may have to be significantly larger than the liquid crystal or electrophoretic capacitance to reduce the effects of leakage.

The basic considerations in display design are (1) the pixel capacitance must be charged through the switch to a voltage accuracy consistent with the required grayscale resolution (or color depth) of the display; and (2) the leakage of the switch must not permit the pixel voltage to decay excessively during the time when the switch is off. Pixel Charging For most of the pixel charging time, the charging characteristic is in its tail and VDS is small. In this part of the linear operating region, the TFT behaves like a resistor R whose value depends on VGS. The charging of the pixel voltage follows an RC charging characteristic, where C is the total pixel capacitance. In a 6-bit system, about five RC time-constants are needed for voltage convergence to half a gray level. The electro-optic curve of a typical twisted nematic liquid-crystal cell has a transition region that spans a data voltage range of about 1 V (see Figure 6.4.12), so this corresponds to better than 8 mV convergence. For example, in a typical SXGA (1280 × 1024) AMLCD, the refresh rate is 60 Hz, the frame time is 16.7 msec, and the line time is 16.3 μsec. For 6-bit resolution, the RC charging time constant must be less than 3.3 μsec. The total pixel capacitance is about 0.5 pF, so the TFT on-resistance must be less than 6.6 MΩ. From Equation 6.4.1, which assumes constant mobility and no contact resistance, when VDS is small, we can derive the conductance G of the TFT:

8080_C006.fm Page 574 Tuesday, April 3, 2007 2:15 PM

574

Organic Field-Effect Transistors

G =1/ R =

W μ FE Cox (VGS − Vt ) L

(6.4.7)

For a typical pentacene TFT with a 100-nm SiO2 gate dielectric, μFE = 0.5 cm2/Vs and Cox = 3.45 × 10–8 F/cm2. Ideally, the width W and the length L have minimum dimensions, so W = L. In order not to place severe demands on the row drivers, we will limit the select voltage over threshold (VGS – Vt) to 10 V. Making these substitutions, we find that the switch on-resistance R = 5.8 MΩ, satisfying the time-constant requirement. Thus, a pentacene TFT switch meets the requirements for a typical AMLCD. This is expected because the mobility of pentacene OTFTs is similar to that of a-Si TFTs [31]. We have based this design on the assumption that the select voltage over threshold (VGS – Vt) is 10 V. It must be noted that the gate-to-source voltage VGS is the difference between the select voltage and the applied data voltage. Thus, the charging time is data dependent, with the worst case occurring when VGS has its smallest value. For a p-channel OTFT, such as a pentacene OTFT, we reverse the sign of the voltages in the expressions given previously [4]. The worst case occurs with the most negative data voltage. For example, if the most negative data voltage that must be applied to the liquid crystal is –5 V and the threshold voltage is –5 V, then the select voltage must be at least –20 V to achieve a switch resistance low enough to charge the pixel capacitance adequately under worst-case data voltage conditions. We have assumed that contact resistance effects like those shown in Figure 6.4.5 are not present. A hooked output characteristic like the one shown in the figure prevents good pixel convergence in a line time. The pixel voltage will converge to a level that differs from the data voltage by approximately the “turn-on voltage” of the hooked characteristic. If this offset error were uniform across the display, it could be compensated for all pixels together by adjusting the common electrode voltage Vcom. However, the error is very likely to be nonuniform, and OTFTs with poor contacts are unlikely to be suitable for displays. Pixel Leakage The counterpart to requiring better than 8-mV convergence during pixel charging is to permit no more than 8-mV voltage decay when the switch is off during the rest of the 16.7-msec frame time. For the example given before, this corresponds to an off-current Ioff of less than 0.25 pA. As noted earlier, off-currents of TFTs are typically cited as the minimum current over a VGS range. However, the requirement Ioff < 0.25 pA must be met over the full range of possible pixel voltages and data line voltages, even in the presence of data voltage inversion, in which the data line polarity may be the opposite of the pixel voltage polarity. Furthermore, this requirement must be met not just on average, but by all (or nearly all) of the switches in the array, since leaky switches lead to visible pixel defects. These stringent off-current requirements can be satisfied by a-Si TFTs. However, the off-currents of OTFTs are typically much higher than those of a-Si TFTs. Allowance can be made for large statistical scatter in TFT leakage by using double or triple series-gated devices, provided that the leakage through each gated region is independent of the others. But, in general, displays with high gray-scale resolution

8080_C006.fm Page 575 Tuesday, April 3, 2007 2:15 PM

Organic Thin-Film Transistors for Flat-Panel Displays

575

and color depth will be difficult to make using OTFTs until off-currents are reduced significantly while maintaining high mobilities for pixel charging. These two considerations — charging and leakage — can be viewed as a single requirement on the TFT on–off current ratio. To analyze the switch requirements in this way is less accurate than the preceding analysis because it treats the charging in terms of a capacitance charged from a current source rather than as RC charging. Nevertheless, it provides insight. The analysis is as follows: The voltage across the pixel capacitance C must be changed by a voltage ΔV using the TFT’s on-current Ion in a line time Tline. This requires that Tline ≥ CΔV/Ion. For voltage accuracy better than αΔV (e.g., α = 0.78% for half a gray level in a 6-bit system), the permissible decay when the switch is off is αΔV. The voltage will decay by this amount in a time CαΔV/Ioff. Thus, it is required that Tframe = Nrow Tline ≤ CαΔV/Ioff. If the charging and leakage requirements are combined, the requirement for the on–off current ratio is Ion/Ioff ≥ Nrow/α. For example, an SXGA display (Nrow = 1,024) with 6-bit gray scale or color depth requires Ion/Ioff ≥ 1.3 × 105. Again, this analysis is illuminating but one must not regard it as exact. The pixel’s charging characteristic is actually slower than assumed in this analysis because of the exponential charging characteristic. In addition, reported on–off current ratios typically use the minimum leakage current for Ioff, which as noted earlier may not be realistic. Charge Injection When the switch in an LCD or electrophoretic pixel is shut off, it injects charge into the pixel, leading to a potentially nonuniform voltage shift [32]. We noted earlier that, when a TFT switch is shut off by switching the gate voltage from VGS,on to VGS,off, the charge injected into each of the source/drain terminals is Q = WLoverlapCox (VGS ,on − VGS ,off ) +

WL Cox (VGS ,on − Vt ) 2

Including a storage capacitor in the pixel reduces the voltage shift but does not eliminate it. Treating the total pixel capacitance as having a fixed value C, the voltage shift is Vshift = Q/C. For a p-channel OTFT, the pixel voltage is pushed up by this amount as the switch is shut off. The shift can be partly compensated by adjusting the common electrode voltage Vcom. But adjusting Vcom does not eliminate all effects of charge injection because the charge is data dependent and nonuniform. It is therefore beneficial to use as small a TFT as possible, minimizing W and L, and to minimize the overlap length Loverlap. OTFT layouts in which the electrode connected to the data line wraps around the pixel’s internal electrode can be used to reduce the charge injection. Figure 6.4.15 shows a layout that our laboratory found to work well. Nevertheless, for very low mobility OTFTs that require large widths W to satisfy charging requirements, the TFT capacitances can become excessive. In some cases it may be difficult to shut the switch off. As the switch is deselected, charge injection pushes up the pixel voltage, and the TFT’s gate voltage may have to “chase” the pixel voltage through a large voltage range before the TFT is shut off. Therefore, when using large, low-mobility OTFTs as pixel switches, it is important to determine

8080_C006.fm Page 576 Tuesday, April 3, 2007 2:15 PM

576

Organic Field-Effect Transistors

FIGURE 6.4.15 A wraparound geometry can be used to reduce charge injection in the pixel switch.

the select line voltages in a self-consistent manner that takes into account the voltage shift induced by the shutting off of the TFT. Bias-Stress Instability In considering the effect of bias-stress instability on OTFT pixel switches it is instructive to consider analyses of a-Si AMLCDs [33] because the effect is similar. In a-Si TFTs the primary effect of positive gate bias is to shift the threshold voltage to more positive voltages, and the primary effect of negative bias is to shift the threshold voltage to more negative voltages. Another typical feature of a-Si TFTs is that the threshold shift under on-bias is more rapid than the shift under off-bias. However, TFTs used as pixel switches are in the on-condition for much less time than they are in the off-condition. For example, in an SXGA display, the TFTs are on for about 0.1% of the time and off for about 99.9% of the time. As a result, the direction of the net threshold shift under actual usage conditions can depend on the fabrication details of the TFT and must be determined experimentally. In any case, because pixel TFTs are left in the more unstable on-condition for a low duty-cycle and because they are subject to the compensating effect of the off-condition, the shift of an a-Si pixel TFT over a typical 10,000-hour lifetime is small, at most a few volts. This is taken into account by allowing some voltage margin in the select and deselect voltages applied to the select lines. The situation with OTFTs is unclear. Our studies have shown that on-state bias stress instability in pentacene OTFTs is somewhat worse than in a-Si TFTs and the off-state instability is much worse. The off-state instability is particularly troubling, since a pixel switch is usually in the off state. However, other organic semiconductors show much less instability. Polymer gate dielectrics are seeing increasing use in

8080_C006.fm Page 577 Tuesday, April 3, 2007 2:15 PM

Organic Thin-Film Transistors for Flat-Panel Displays

577

OTFTs, and slow polarization effects that can occur in these dielectrics lead in the opposite direction: The effect of positive gate bias is to shift the threshold voltage to more negative voltages and vice versa. For stable OTFTs, a stable semiconductor must be combined with a stable dielectric. Modeling must be performed using the actual voltage conditions produced by the select pulses and the applied data voltages. Once the worst-case threshold voltage shifts are known, the select and deselect voltages must have enough margin to allow for the predicted shifts. Since the select line voltages influence the threshold voltage shifts, design must be performed in a self-consistent manner. Breakdown Voltages If the data voltage were always 0 V, the select-to-deselect voltage excursion would simply be the voltage needed to take the TFT switch from its on-state with the required low resistance, down to Vt, through the subthreshold region, to the off-state. However, the data voltages cover a range of positive and negative voltages, so the select-to-deselect excursion must be augmented by the total required data voltage swing. Charge injection, bias-stress instability, and hysteresis may add additional components to the required voltage excursion. In the end, the influence of low mobility, large subthreshold swing, charge injection, bias-stress instability, and hysteresis may require total select-to-deselect voltage excursions for OTFT displays of 50 V or more. This can introduce problems with availability of driver electronics because high-voltage driver chips, if available at all, are expensive. It can also introduce breakdown voltage problems for the TFTs. Using thicker gate dielectric for higher breakdown voltage is not a good solution, since it also increases the required voltage swing. Especially for electrophoretic displays, which typically require larger data voltages than LCDs, breakdown voltage requirements may limit what is possible with OTFT technology.

6.4.3.2 ACTIVE MATRIX OLED DISPLAYS 6.4.3.2.1 Introduction Most OLED displays currently used in consumer products are passive matrix OLED (PMOLED) displays. Only a few products, such as the Kodak EasyShare LS633 digital still camera and the Sony Clie VZ-90 PDA, use AMOLED displays. But the advantages of AMOLED displays over PMOLED displays are compelling. They arise from the fact that light emission from the passive matrix occurs one line at a time, whereas in the active matrix, emission is continuous. This leads to higher power efficiency, longer operational lifetime, and fewer demands on the currenthandling capacity of the driver circuits. The first AMOLED display was a 16 × 16 array using a poly-Si TFT backplane, demonstrated in mid-1998 by a team from Sarnoff Corporation, Planar America, Kodak, and Princeton University [34]. Later that year Seiko-Epson demonstrated an 800 × 236 AMOLED display, also using poly-Si TFTs [35]. Most AMOLED backplanes since then have used poly-Si technology. The prospects were considered to be poor for making AMOLED backplanes using lower mobility transistors such as a-Si TFTs [36].

8080_C006.fm Page 578 Tuesday, April 3, 2007 2:15 PM

578

Organic Field-Effect Transistors

TABLE 6.4.2 Active Matrix OLED Displays Using Organic TFTs Organization Cambridge University Bell Laboratories University of Tokyo Pioneer Dong-A University Penn State University Victor and NHK

Semiconductor Polythiophene Polythiophene Pentacene Pentacene Pentacene Pentacene Pentacene

Substrate

Specifications

Ref.

Glass Silicon wafer Glass Glass PET PET PEN

1 × 1, single TFT 1 × 1, single TFT 1 × 1, single TFT 8 × 8, two TFT 64 × 64, single TFT 48 × 48, two TFT 16 × 16, two TFT

41 42 64 43 65 44 66

However, in 1995 our laboratory began investigating the use of a-Si TFTs for AMOLED displays as a low-cost alternative to poly-Si TFTs. We concluded that aSi TFT AMOLED backplanes can be competitive, and developed self-compensating a-Si pixels that are tolerant of bias-stress instability. In 1997 Princeton University demonstrated the integration of a single a-Si TFT with an OLED [37]. In 2003 Chi Mei Optoelectronics and IBM demonstrated a 20-in. AMOLED display based on aSi technology, proving that a-Si backplanes can be used for large-area AMOLED displays [38]. There have been other, more recent demonstrations, most notably a 40-in. a-Si AMOLED display from Samsung [39]. A review of a-Si technology for AMOLED displays has been published recently [40]. The performance of OTFTs is similar to that of a-Si TFTs, so these demonstrations imply the feasibility of largearea OTFT AMOLED displays. The first demonstrations of using OTFTs to drive OLEDs came from independent groups at Bell Laboratories and Cambridge University nearly simultaneously in 1998 [41,42]. These were single “smart pixels,” each incorporating one polymer-based OTFT and one OLED. Since AMOLED displays require at least two TFTs per pixel, these demonstrations were not examples of complete AMOLED pixels. In 2004 Pioneer reported the first OTFT AMOLED display, containing an 8 × 8 array of two TFT pixels using pentacene OTFTs on a glass substrate [43]. In 2005 Penn State University demonstrated a 48 × 48 OTFT AMOLED display on a PET substrate, also using two TFT pixels [44]. Table 6.4.2 lists these and other significant OTFT AMOLED display results. 6.4.3.2.2 Electro-Optic Behavior of Organic Light-Emitting Diodes The electro-optic behavior of OLEDs is similar to that of inorganic LEDs. The current-voltage (I–V) characteristic is diode-like. The current that flows under forward bias causes light emission. Under reverse bias, little current flows and there is no light emission. As with inorganic LEDs, the light output of OLEDs varies linearly with forward current over a wide current range. That is, the quantum efficiency is nearly constant. It is only at very low forward currents that leakage currents associated with nonradiative processes come to dominate, and at very high currents, the

8080_C006.fm Page 579 Tuesday, April 3, 2007 2:15 PM

Organic Thin-Film Transistors for Flat-Panel Displays

579

Fixed current

Voltage

Light output

5000

0 Time (hrs)

FIGURE 6.4.16 An increase in voltage and a decrease in light output at a fixed current level are the two effects limiting the operational lifetime of OLEDs.

quantum efficiency drops because of high exciton densities in the organic materials. Among the reasons for PMOLED displays having lower efficiencies than AMOLED displays are the high voltages required to bias the diodes at high currents and the low quantum efficiencies at high current levels. Depending on the materials used and the degree of encapsulation, OLEDs can have short operational lifetimes. In severe cases, OLEDs may last for less than 100 hours of operation. OLEDs using more advanced materials that are well encapsulated last more than 10,000 hours. Lifetime issues have slowed the commercialization of the technology, and a great deal of effort has been devoted to improving OLED materials and developing encapsulation methods to serve as a barrier to oxygen and water vapor. From the perspective of display design, the primary lifetime effects are: • •

Voltage increase. The voltage required to cause a given forward current to flow through an OLED increases over its operational lifetime Luminance decay. The quantum efficiency of an OLED decreases over its operational lifetime, so at a given current, its light output decreases

Figure 6.4.16 illustrates these effects. Both are also seen in inorganic LEDs, but the timescales involved are so long that they are ignored in all but the most demanding LED display applications. Because the light output is proportional to current, it is natural to drive an OLED from a current source rather than a voltage source. If an OLED were driven from a voltage source, the diode I–V characteristic would render the light output very sensitive to small errors in the data voltage — for example, due to voltage drops in the conductors of the display. Furthermore, small nonuniformities in the I–V characteristics would lead to large current differences and therefore large nonuniformities in the light output. This situation would be exacerbated by voltage increases over the operational lifetime, causing static images left on the display for

8080_C006.fm Page 580 Tuesday, April 3, 2007 2:15 PM

580

Organic Field-Effect Transistors

Select line

VDD

C

Select lines P1

P2

Data line

OLED

Data lines

(a)

(b)

FIGURE 6.4.17 (a) Passive matrix and (b) active matrix OLED display pixels.

a long time to be burned in. Therefore, like inorganic LEDs, OLEDs are always driven from current sources in the driver electronics, with each current source’s output set by the image data. 6.4.3.2.3 OLED Display Architectures Figure 6.4.17 illustrates PMOLED and AMOLED display architectures, with only one pixel of the AMOLED display shown for simplicity. The pixel in Figure 6.4.17(b) is the simplest AMOLED pixel, a two TFT pixel. It was proposed in 1975 for use with any electro-optic material that requires the flow of current [45]. Later, we describe AMOLED pixels that are more complicated than the two TFT pixel. The passive matrix consists of a set of row electrodes and column electrodes, with an OLED formed at each intersection. A row is selected for light emission by applying a select voltage pulse, forward biasing the diodes in that row. Currents proportional to the image data for the selected row are applied to the column lines by the driver circuits. All the unselected rows are reverse biased. In the active matrix, data are written into a selected row by applying a select pulse to the TFT switches in the row. Each pixel’s capacitance is charged to the data voltage, just as in an AMLCD. Thus, TFT P1 is used in the same way as the switch in an AMLCD, and our earlier discussion of the switch in an AMLCD applies here as well, with similar considerations for charging, leakage, charge injection, etc. Care must be taken to prevent light emitted by the OLED from being laterally “piped” into the active region of the TFT switch, inducing photocurrents. Unlike an AMLCD pixel, the AMOLED pixel contains a drive transistor P2 for converting the stored voltage to a current that drives the OLED continuously. The reason the pixel needs more than one TFT is that, unlike a liquid crystal, an OLED is not a capacitor that can hold the stored data voltage, but rather is a current-drawing element that would quickly dissipate stored charge if the OLED were connected directly to the switch. Therefore, one or more additional TFTs are needed in an AMOLED pixel to allow the data voltage to be held as it controls the OLED current.

8080_C006.fm Page 581 Tuesday, April 3, 2007 2:15 PM

Organic Thin-Film Transistors for Flat-Panel Displays

581

Power consumption (mW)

1000 800 AMOLED

600

PMOLED

400 200 0 80 × 101

128 × 128

144 × 176 Display Resolution

176 × 240

FIGURE 6.4.18 Power consumption of AMOLED and PMOLED displays of different resolutions. (Adapted from Haskal, E., SID Int. Symp. Seminar Lecture Notes, M-3/3, 2004. Copyright Society for Information Display. With permission.)

The advantage of AMOLED over PMOLED displays arises because emission in a passive matrix occurs one line at a time, so each OLED element operates at high peak currents and low duty-cycle. The duty-cycle in a PMOLED display is approximately equal to the inverse of the number of rows. For example, in an SXGA (1280 × 1024) display, the duty cycle is approximately 0.1%. The peak current of an OLED pixel may be 1 mA or more. High OLED currents lead to reduced power efficiency and operational lifetime and also place greater demands on the current capacity of the row driver circuits, which may have to handle currents of hundreds of milliamperes on each output (although not simultaneously). In contrast, in AMOLED displays, each OLED element operates at nearly 100% duty cycle, independent of the number of rows in the display. Because of this, it is generally agreed that high-resolution OLED displays will require an active matrix for efficiency and long lifetime. Figure 6.4.18 shows data from Philips comparing the calculated power consumption of AMOLED and PMOLED display modules [46]. The power advantage of the active matrix increases for higher resolution displays. The demands that an AMOLED display places on TFT performance are more stringent than those placed by liquid-crystal or electrophoretic displays. Table 6.4.3 compares the TFT requirements of high-resolution displays of these types. The different TFT requirements arise from the fact that LCD and electrophoretic display pixels have a single TFT that is used as a switch, while an AMOLED pixel has one or more TFTs used as a switch, plus one or more used to drive current. A transistor used as a switch typically has stringent requirements for off-current because it must not leak significant charge when it is off, but the requirements for mobility, uniformity, and stability are modest. A transistor that is used to drive current has stringent requirements for mobility, uniformity, and stability, but offcurrent requirements are lenient. Because the AMOLED pixel has both types of

8080_C006.fm Page 582 Tuesday, April 3, 2007 2:15 PM

582

Organic Field-Effect Transistors

TABLE 6.4.3 TFT Requirements of AMLCD, Active Matrix Electrophoretic, and AMOLED Displays TFT parameter

LC and electrophoretic display

Mobility Off-current Uniformity Stability

≥0.1 cm2/V-s ≤1 pA Moderately important Moderately important

AMOLED display ≥1 cm2/V-s ≤1 pA Very important Very important

Common cathode OLED organic stack

Source

Passivation Organic SC Gate dielectric Gate electrode

Drain Anode

Backplane substrate

OLED light emission

FIGURE 6.4.19 Cross-section of an OTFT AMOLED display pixel showing the drive transistor and OLED (not to scale).

transistors, the TFT technology must meet more stringent requirements than those required by liquid-crystal and electrophoretic pixels. A cross-section of the drive transistor and OLED of a two TFT pixel is shown in Figure 6.4.19. As with AMLCDs, for color displays each pixel is divided into three subpixels, one each for red, green, and blue. The different colors can be produced by using three sets of OLED materials with different emission spectra. (Alternatively, one white-emitting OLED material can be used with three color filters or one blue-emitting material with down-converting phosphors.) In the most common and simplest process, the TFT is connected to the anode of the OLED, and there is a large-area common cathode shared by all of the OLEDs. Indium-tin oxide works well as an OLED anode and is commonly used. The cathode is usually an opaque low-work-function metal. Even the simple two TFT AMOLED pixel requires that contact be made between the gate and the source/drain conductors (see Figure 6.4.17b), so, unlike an AMLCD, contact openings must be made in the gate dielectric layer between these layers in every pixel. Patterning the gate dielectric adds process complexity to the AMOLED process. Using polymer dielectrics creates special problems because it requires that

8080_C006.fm Page 583 Tuesday, April 3, 2007 2:15 PM

Organic Thin-Film Transistors for Flat-Panel Displays

583

photolithographic processing be performed on the polymer, typically followed by stripping of the photoresist mask. This can degrade a polymer dielectric. Since the common OLED structure has a transparent electrode on the backplane side and an opaque common electrode, light emission in an AMOLED display is usually through the backplane. This is called bottom emission. As a result, the TFTs get in the way of the light emission, and the use of large TFTs or many TFTs reduces the emission fill-factor of the display. The OLEDs can always be driven at high current densities to make up for poor fill-factor, but at the expense of efficiency and lifetime, for the same reasons as in PMOLED displays. This is a particular problem with a-Si TFT and OTFT AMOLED displays, since the low mobilities of these technologies require the use of wide devices. One way out of this problem is to use top-emitting OLEDs, which have a transparent cathode, so that the TFTs in the backplane do not obstruct light emission. In a-Si TFT technology only n-channel devices are available, while in OTFT technology p-channel devices have had the best performance and the greatest ease of processing. For the switch used in AMLCD and AMOLED pixels, it makes no difference whether an n-channel or a p-channel TFT is used, since they are equivalent if one inverts the polarity of the select pulse. However, for the drive transistor, a pchannel TFT is preferred over an n-channel TFT. Typically, the OLED anode is connected to the TFT. Because the anode is the terminal through which current enters a diode, connecting this terminal to the TFT ties it to the drain of a p-channel TFT but the source of an n-channel TFT. As a result, for a p-channel drive transistor, the VGS of the drive transistor equals the data voltage Vdata stored in the pixel. If the supply voltage VDD is high enough to keep the TFT in the saturation region, Equation 6.4.2 states that the TFT acts as a current source with the current set by the data voltage, which works well. However, with an n-channel drive transistor, the OLED is in the TFT’s source. The VGS of the drive transistor equals (Vdata – Vdiode), where Vdiode is the voltage across the OLED. But Vdiode is nonuniform across the display and also varies over the operational life of the display, so VGS will also be nonuniform. Furthermore, this is a configuration that uses the TFT as a source-follower voltage source, not as a current source, so applying a voltage Vdata is equivalent to setting a voltage across the OLED. We have noted earlier that driving an OLED from a voltage source is problematic. Thus, unlike poly-Si TFT and OTFT technologies that provide pchannel TFTs, a-Si technology cannot implement the two TFT pixel well. More complex pixels can overcome the difficulties from having the OLED in the source of the a-Si TFT, but the issue always remains as a complication. In this respect, the availability of p-channel devices gives OTFTs an advantage over a-Si TFTs for AMOLED displays. 6.4.3.2.3.1 Nonideal Behavior in AMOLED Pixels Ideally, the drive TFT would be a perfect voltage-controlled current source in which the current delivered to the OLED is a function of the data voltage alone, and this function is uniform across the display. Even if the current is a complicated nonlinear function of the data voltage, this function can be determined in advance and the data can be warped to take the nonlinearity into account. Unfortunately, the current is

8080_C006.fm Page 584 Tuesday, April 3, 2007 2:15 PM

584

Organic Field-Effect Transistors

not a function of the data voltage alone. For one thing, the ideal current source represented by Equation 6.4.1 is only an approximation. In reality, the current ID through the drive transistor depends on the voltage VDS across the transistor, even in the saturation region. Thus, the current varies with the OLED voltage Vdiode. The use of a second TFT in series with the drive TFT in a cascode configuration, with its gate connected to a fixed DC potential, provides a much better current source, although at the cost of higher power supply requirements. There are other important ways in which the drive TFT fails to be a perfect, uniform voltage-controlled current source. The voltage-to-current conversion performed by the drive transistor depends on the TFT’s threshold voltage and mobility, and these parameters are nonuniform across the display. This has been a particular problem for poly-Si AMOLED backplanes because poly-Si TFTs exhibit large nonuniformities due to the random nature of the polysilicon grain structure. But even a-Si TFTs must cope with this problem. While a-Si TFTs are uniform at the beginning of operational life, they quickly become nonuniform as they undergo image-dependent threshold voltage shifts from bias-stress instability. Hysteresis is a source of image-dependent nonuniformity over shorter time scales. In the present state of the technology, OTFTs exhibit initial nonuniformity as well as imageinduced nonuniformity. It is instructive to examine the sensitivity of the drain current to TFT variations. From Equation 6.4.1 one can derive that the fractional variation in the drain current equals the fractional variation in mobility ΔID/ID = ΔμFE/μFE; for example, a modest ±10% variation in mobility leads to an unacceptably large ±10% variation in current and thus in pixel brightness. One can also derive that the fractional drain current variation per volt of threshold voltage variation is ΔID/(IDΔVt) = –2/(VGS – Vt). Since the data voltage is typically a few volts over threshold, this relationship indicates that a relatively small ±0.2-V variation in threshold voltage leads to ±13% variation in current and pixel brightness. For a 6-bit system, one gray level corresponds to a 1.6% difference in brightness, so these brightness nonuniformities correspond to many gray levels. As a result, like two TFT AMOLED displays made with poly-Si TFT backplanes, those made with OTFTs can be expected to exhibit large brightness nonuniformity due to parameter variations. One way of dealing with TFT nonuniformities is to demand better initial uniformity and less bias-stress instability and hysteresis from the TFT technology. There has been progress in this direction using new OTFT materials and processes. However, an alternative is to develop pixel designs that are more tolerant of TFT nonuniformity. AMOLED pixels that compensate for TFT nonuniformities fall into two categories. Some of them use voltage to represent the image data (as the two TFT pixel does), while others use current. These two types are called voltage-programmed pixels and current-programmed pixels, respectively. Several reviews have been published on the subject [47–52]. 6.4.3.2.3.2 Voltage-Programmed AMOLED Pixels Like the standard AMLCD pixel, the two TFT AMOLED pixel is voltage programmed. That is, voltages are used to represent the image data. In 1998 Sarnoff Corporation, Planar America, Kodak, and Princeton University reported the earliest

8080_C006.fm Page 585 Tuesday, April 3, 2007 2:15 PM

Organic Thin-Film Transistors for Flat-Panel Displays

585

Apply data here Select line

VDD

C1 N1

P1 C2

Autozero line Data line

P2

N2 OLED

FIGURE 6.4.20 Sarnoff four TFT self-compensating voltage-programmed AMOLED pixel.

work on AMOLED pixels that compensate for TFT nonunformity. This team developed a self-compensating voltage-programmed pixel that uses four TFTs [34]. The four TFT voltage-programmed pixel, shown in Figure 6.4.20, has much less sensitivity to TFT nonuniformities than a two TFT pixel does. Before writing the data voltage, each pixel undergoes an autozeroing cycle in which N2 is shut off, disconnecting the OLED, and N1 and P2 are turned on, connecting the gate and drain of drive transistor P1 and applying the grounded data line to the left side of capacitor C2. During the autozero interval, the drive transistor relaxes into a condition in which its gate and drain are at a threshold voltage relative to the source. (This assumes that P1 is an enhancement-mode transistor, i.e., it is a device that is off when VGS = 0 so that it is in the saturation region when gate and drain are connected together.) Then P2 is shut off and N2 is turned back on, and the threshold voltage is stored on storage capacitor C1. Finally, the data voltage is applied to the column line and written into the pixel through N1. Capacitor C2 is used to sum the data voltage and the threshold voltage. Thus, VGS for the drive transistor is equal to (Vdata + Vt). Referring to Equation 6.4.2, the quantity that sets the drive current (VGS – Vt) is therefore equal to Vdata, and variations in threshold voltage are automatically compensated. Figure 6.4.21 shows 16 × 16 AMOLED displays using the two TFT pixel and the four TFT pixel, both fabricated with poly-Si TFT technology on the same substrate. The uniformity of the four TFT pixel is much better than that of the two TFT pixel. At typical full brightness level, the standard deviation in the brightness level of the two TFT pixel was measured to be 10.1% of the average, whereas for the four TFT autozeroing pixel, it was only 2.6% of the average. Issues with this pixel are as follows: •

It has additional TFTs and an extra control line compared to the two TFT pixel. This reduces fill factor and adds to the complexity of the row drivers

8080_C006.fm Page 586 Tuesday, April 3, 2007 2:15 PM

586

Organic Field-Effect Transistors

FIGURE 6.4.21 16 × 16 poly-Si TFT AMOLED pixel arrays using (a) two TFT pixel and (b) four TFT self-compensating voltage-programmed pixel. The additional area occupied by the circuitry of the four TFT pixel reduces the pixel fill-factor.

•

•

because of the need for additional signals. The reduction in fill-factor is apparent from a comparison of Figure 6.4.21(a) and (b). It requires a large voltage-summing capacitor C2 in addition to the storage capacitor C1. C2 must be large in order not to significantly dilute the data voltage due to the capacitive voltage divider formed by C2 and the total pixel capacitance. Thus, C2 occupies a large area, reducing fill-factor. It compensates for threshold voltage nonuniformity but not for mobility nonuniformity. However, in our laboratory we found that if one deliberately allows insufficient time for the drive transistor to converge to its threshold voltage during the autozeroing operation, then those drive transistors with lower mobility will converge to a compensation voltage slightly above the threshold voltage, providing an extra “boost” to these devices. This was found to provide some compensation for mobility nonuniformity.

6.4.3.2.3.3 Current-Programmed AMOLED Pixels Instead of relying on a TFT to convert the data voltage to the OLED current, an alternative is to represent the data by a current, to store the value of the current in the pixel, and to deliver it to the OLED. Sarnoff Corporation, Planar America, Kodak, and Princeton University proposed the current-programmed approach in 1998 using the four TFT pixel circuit shown in Figure 6.4.22 [47]. During the time when a data current Idata is written into the pixel, P2, which ties the source of drive transistor P1 to VDD, is shut off and N1 is turned on, so Idata is driven into the source of P1 through N1. At the same time, N2 is turned on, connecting gate and drain of P1. This establishes a value of VGS for P1 that will produce the programming current, and this value of VGS is stored on storage capacitor C. (This assumes that P1 is an enhancement-mode transistor so that when gate and drain are connected together it is in the saturation region.) Then, N1 and N2 are shut off and P2 is turned back on. The current through the OLED has a stored value equal to the sampled data current. In 2001 Sony demonstrated a different four TFT current-programmed approach using a current mirror, shown in Figure 6.4.23 [52]. This pixel operates similarly to

8080_C006.fm Page 587 Tuesday, April 3, 2007 2:15 PM

Organic Thin-Film Transistors for Flat-Panel Displays

587

VDD Select line P2 N1 C P1 N2

Idata

Data line OLED

FIGURE 6.4.22 Sarnoff four TFT self-compensating current-programmed AMOLED pixel.

the one in Figure 6.4.22, but the data current is sampled by a different transistor from the one that drives the OLED. During the time when the data current is sampled, N1 and N2 are on, connecting the gate and drain of P1 and tying them to the data line. The data current is drawn from P1, establishing a value of VGS for P1 that will produce the programmed current. This value of VGS is stored on capacitor C. The same VGS is applied to P2, causing it to copy the current and deliver it to the OLED. In analog circuit design this configuration of two transistors with a shared VGS is called a current mirror. The current delivered to the OLED can also be a scaled version of the sampled current by scaling the width-to-length ratios W/L of the two mirror transistors. VDD C P2 P1

N1

OLED Select line

Idata

N2

Data line

FIGURE 6.4.23 Sony four TFT self-compensating current-programmed AMOLED pixel using a current mirror.

8080_C006.fm Page 588 Tuesday, April 3, 2007 2:15 PM

588

Organic Field-Effect Transistors

Current-programmed pixels have the following advantages over voltage-programmed pixels: •

•

The threshold voltage and mobility of the drive transistor are irrelevant, since these quantities relate only to how a transistor translates voltages into currents. Indeed, all transistor characteristics are irrelevant, as long as the TFT is a good current source (i.e., ID is independent of VDS) with sufficient on-current to drive the OLED to full brightness, and the voltages that it requires are not unmanageably high. It does not require the large summing capacitor needed by the autozeroing voltage-programmed pixel.

The current-programmed pixel also has disadvantages: •

•

In a current mirror pixel, mismatch between the two mirror transistors is a source of nonuniformity. There are well-known layout methods used in bulk silicon design for obtaining good matching between neighboring transistors, but completely random variations such as those seen in polySi TFTs are not reduced by these methods. Referring to Figure 6.4.23, it is especially undesirable for P2 to have a lower threshold voltage or higher mobility than P1, since this causes the OLED current to be higher than intended and a dark pixel may appear bright. This is more visible than the opposite error. Fortunately, bias-stress instability and hysteresis depend primarily on VGS rather than VDS and the two mirror transistors share the same VGS, so they are expected to track each other even with these effects present. However, in our laboratory, we have observed that the initial parameters of closely spaced pentacene OTFTs do not match well, and at the present time this limits their use in current mirrors. A long time is needed to charge the large data line capacitance of an AMOLED display from the driver current source because it is always slower to charge a capacitance from a current source than from a voltage source. Only a line-time is available and, in a typical high-resolution display, this is too little time for charging the data line, especially at low current levels. To help with this problem, sometimes the driver circuits are designed to apply a precharge voltage to the data lines before applying the current. During the precharge interval, the data line is rapidly charged from a voltage source to a voltage estimated to be the value subsequently obtained with the current source. However, voltage precharging assumes that a good voltage estimate can be made, and this may not be a good assumption over the life of the display. Our laboratory developed a driver architecture in which the current source is in parallel with a circuit that simulates a negative capacitance, canceling most of the capacitance presented by the display [53]. This approach significantly reduces charging times without using a voltage precharge. A similar method was later developed independently by the University of Waterloo [54].

8080_C006.fm Page 589 Tuesday, April 3, 2007 2:15 PM

Organic Thin-Film Transistors for Flat-Panel Displays

589

More complicated methods have been developed for nonuniformity compensation. For example, AMOLED displays have also been demonstrated with a photodetector in each pixel to sense OLED light output and apply feedback to correct for nonuniformity [55]. It should be noted that this is the only approach that compensates for OLED luminance decay over operational life, since none of the others detects light output. However, the method introduces new sources of nonuniformity arising from the photodetector and its associated circuitry, and it is unclear whether the net effect will always improve display uniformity. Additional discussion and analysis of the use of OTFTs for AMOLED displays can be found in the literature [56,57]. To date, only the two TFT pixel has been implemented with OTFTs, although a paper design for a current-programmed OTFT AMOLED pixel has been presented [57]. It is only a matter of time before the more complex pixels that have been investigated using poly-Si and a-Si TFTs are tested with OTFTs.

6.4.4 USING ORGANIC TFTS FOR INTEGRATED DRIVERS Manufacturers of small- and medium-sized AMLCDs are making increasing use of poly-Si TFTs for the integration of row and column display drivers onto the display glass. The integration of driver circuits on the display substrate can reduce display module cost and shorten product development times. The performance of poly-Si TFTs is very good, with the possible exception of the off-currents, but low off-currents are generally not important for driver circuits. Furthermore, complementary n-channel and p-channel devices are available with poly-Si TFT technology, so complementary metal-oxide semiconductor (CMOS) design methods can be used. There have also been efforts to integrate display driver circuits using a-Si TFTs, since the cost of a-Si technology is lower than that of poly-Si technology. As long ago as 1987, the David Sarnoff Research Center and Thomson LCD developed a technology known as SASID that permitted the integration of a-Si TFT row and column drivers onto an AMLCD substrate. However, implementation of display drivers using a-Si TFTs is considerably more difficult than using poly-Si TFTs because of the lower mobility and bias-stress instability of a-Si TFTs and because a-Si technology provides only n-channel TFTs, preventing use of CMOS techniques. The major advantages of CMOS over single-channel circuits and systems include: • • • • •

Power dissipation is lower, especially static dissipation in digital systems. Signal voltages can easily swing all the way from one power supply voltage to the other. The performance of the system is more tolerant of variations in transistor parameters, so fabrication yields are higher. Circuit gain is higher, leading to larger noise margins in digital systems. Design methods are simpler.

8080_C006.fm Page 590 Tuesday, April 3, 2007 2:15 PM

590

Organic Field-Effect Transistors

The application of OTFTs to integrated driver circuits is similar to the situation with a-Si TFTs: Their usefulness is hindered by low mobility, bias-stress instability, and the lack of CMOS. However, at the present time the methods employed by the SASID technology have not been applied to OTFTs. There are two obstacles to organic CMOS: the lack of good n-channel devices and the process difficulty of integrating two types of organic TFTs on one substrate. The advantages of CMOS exist even if one transistor type is inferior to the other. But to derive the full benefit of CMOS, high performance is required from both device types; otherwise, system performance will be dominated by the lower performing device. For example, a low-mobility n-channel OTFT in a CMOS inverter must be scaled to large widths to provide pull-down capability that matches the pullup capability of the p-channel OTFT, but this means that gate input capacitances are dominated by the large n-channel devices. Rise and fall times will be balanced, but slow. A good case can be made that the benefits of a complementary technology outweigh the gains from achieving modest mobility improvements in a singlechannel process and that more effort to develop organic CMOS is warranted. A row driver consists of a shift register that shifts a select pulse down the select lines. The shift register typically operates at clock frequencies below 100 kHz, within the capability of a-Si TFTs and OTFTs. For those AMOLED architectures requiring multiple control lines for each row, the shift register may require more than one stage per row, with some combinational logic to generate the row signals. A column driver for AMLCDs and voltage-programmed AMOLED displays generates a voltage that depends on the image data for each column. External driver chips place a digital-to-analog converter (DAC) at the output for each column. Integrated column drivers typically do not put all this functionality onto the display. With poly-Si integrated drivers, a few high-speed DACs are placed on an external silicon chip, while the integrated column driver circuitry implements an analog demultiplexer containing a shift register and switches to sample the high-speed analog data sequentially onto the columns at a rate of a few megasamples per second to tens of megasamples per second. Because of the lower mobility of amorphous silicon, a-Si integrated column drivers leave both the DACs and the shift register on an external silicon chip and implement only the demultiplexing switches on the display. It is expected that OTFT column drivers will take a similar approach. A column driver for current-programmed AMOLED displays is difficult to implement using TFTs because currents do not lend themselves to simple sampleand-hold demultiplexing using switches. One approach is to use the same type of current sample-and-hold circuits described earlier for current-programmed pixels, but now placed on each column. There have been a number of reports of analog and digital OTFT circuits, but few have had a level of integration commensurate with integrated display drivers. Two of the most complex have been a 4-bit parallel-to-serial converter containing 171 p-channel OTFTs [58] and a 15-bit code generator containing 326 p-channel OTFTs [59]. The highest level of OTFT integration reported to date was a 120-stage shift register from Philips [60]. The circuit was fabricated on a plastic substrate using a soluble pentacene precursor. It contained 2130 OTFTs and operated at a

8080_C006.fm Page 591 Tuesday, April 3, 2007 2:15 PM

Organic Thin-Film Transistors for Flat-Panel Displays

591

maximum clock frequency of 2 kHz. The circuit is suitable for use as an integrated display row driver and could drive a QVGA display at a frame rate of 8 Hz.

6.4.5 CONCLUSIONS Demonstrations of AMLCDs, electrophoretic displays, and AMOLED displays have shown that OTFT technology can be used for active matrix display backplanes. However, it remains to be seen whether the performance and robustness of OTFTs are adequate and cost reductions sufficient to warrant the introduction of this new semiconductor paradigm. Because of the difficulty of disrupting entrenched TFT technologies, OTFT technology may have to be incrementally introduced into standard glassbased TFT display manufacturing lines, with PECVD chambers replaced by organic film deposition equipment. If additive printing-like processes are used for some or all of the layers, the costs associated with materials and photolithography, two of the most expensive components of TFT backplane manufacturing, will be reduced or eliminated. Introduced gradually, OTFT displays may initially appear as an alternative TFT technology for glass-based displays whose differentiating feature is low cost. Later, as OTFT backplane technology becomes mainstream, display manufacturers will invest in facilities for processing flexible substrates. Then, flexible displays and the often predicted roll-to-roll OTFT manufacturing line may become a reality.

REFERENCES 1. The origin and evolution of the active matrix as a method of display addressing are reviewed in Brody, T.P., The birth and early childhood of active matrix — A personal memoir, J. Soc. Inf. Display, 4, 113, 1996. 2. Huitema, E. et al., Polymer-based transistors used as pixel switches in active-matrix displays, J. Soc. Inf. Display, 10, 195, 2002. 3. Martin, S., Hamilton, M., and Kanicki, J., Organic-polymer thin-film transistors for active-matrix flat panel displays? J. Soc. Inf. Display, 11, 543, 2003. 4. For p-channel MOSFETs, all voltage polarities and current directions are reversed. Other MOSFET drain-current expressions have been derived, some of them specifically for OTFTs. These may permit circuit performance to be predicted more accurately. But the standard MOSFET expressions will suffice for our purposes. 5. Tsividis, Y.P. Operation and modeling of the MOS transistor, 2nd ed., Oxford University Press, New York, 1999. The classical model also includes capacitances to a fourth MOSFET terminal, the body, which does not exist for a TFT, which has only three terminals. An accurate TFT capacitance model is therefore expected to deviate from an accurate four-terminal MOSFET model. 6. A more accurate treatment will consider the effect on the partitioning of the channel charge produced by the change in VDS as the device is shut off, which can lead to current flow between source and drain, and an asymmetric partitioning of channel charge during turn-off. Strictly speaking, symmetric partitioning of channel charge during turn-off occurs only if the circuits attached to source and drain are equivalent and therefore respond identically to injected charge or if the device is shut off so slowly that source and drain equilibrate by the flow of current during the turn-off period.

8080_C006.fm Page 592 Tuesday, April 3, 2007 2:15 PM

592

Organic Field-Effect Transistors

7. It has been found that conducting polymers often make better contact to organic semiconductors than do metals. An explanation is given in Koch, N. et al., Conjugated organic molecules on metal versus polymer electrodes: Demonstration of a key energy level alignment mechanism, Appl. Phys. Lett. 82, 70, 2003. However, conducting polymers are not as conductive as metals. To obtain low OTFT contact resistance while maintaining low-resistance interconnects may require the use of conducting polymer source/drain contact pads with metal interconnects. 8. Katz, H.E. et al., Organic field-effect transistors with polarizable gate insulators, J. Appl. Phys., 91, 1572, 2002. 9. Knipp, D. et al., Pentacene thin film transistors on inorganic dielectrics: Morphology, structural properties, and electronic transport, J. Appl. Phys., 93, 347, 2003. 10. Salleo, A., Endicott, F., and Street, R.A., Reversible and irreversible trapping at room temperature in poly(thiophene) thin-film transistors, Appl. Phys. Lett., 86, 263505, 2005. 11. Powell, et al., Defect pool in amorphous-silicon thin-film transistors, Phys. Rev. B, 45, 4160, 1992. 12. Hack, M., Lewis, A.G., and Wu, I.-W. Physical models for degradation effects in polysilicon thin-film transistors, IEEE Trans. Elec. Dev., 40, 890, 1993. 13. Dresner, J., Dynamic changes in characteristics of a-Si transistors during fast pulsed operation, IEEE Trans. Elec. Dev., 38, 2673, 1991. 14. Gu, G. et al., Electron traps and hysteresis in pentacene-based organic thin-film transistors, Appl. Phys. Lett., 87, 243512, 2005. 15. Zilker, S.J. et al., Bias stress in organic thin-film transistors and logic gates, Appl. Phys. Lett., 79, 1124, 2001. 16. Jung, T. et al., Moisture induced surface polarization in a poly(4-vinyl phenol) dielectric in an organic thin-film transistor, Appl. Phys. Lett., 87, 182109, 2005. 17. IEEE standard for test methods for the characterization of organic transistors and materials, IEEE Standard P1620, IEEE, New York, 2004. 18. G. Gu, personal communication. On the other hand, it has been reported that illumination can assist in recovery from bias-stress effects in polymer TFTs. See Salleo, A. and Street, R.A., Light-induced bias stress reversal in polyfluorene thin-film transistors, J. Appl. Phys., 94, 471 2003. 19. Rogers, J.A. et al., Paper-like electronic displays: Large-area rubber stamped plastic sheets of electronics and microencapsulated electrophoretic inks, Proc. Nat. Acad. Sci., 98, 4835, 2001. 20. Kane, M.G. et al., AMLCDs using organic thin-film transistors on polyester substrates, SID Int. Symp. Dig. Tech. Papers, 32, 57, 2001. 21. Mach, P. et al., Monolithically integrated flexible display of polymer-dispersed liquid crystal driven by rubber-stamped organic thin-film transistors, Appl. Phys. Lett., 78, 3592, 2001. 22. Huitema, H.E.A. et al., Plastic transistors in active-matrix displays, Nature, 414, 599, 2001. 23. Huitema, H.E.A. et al., A flexible QVGA display with organic transistors, Proc. 2003 Int. Display Workshop, Fukuoka, Japan, 1664, 2003. 24. Ho, J.-C. et al., Pentacene organic thin-film transistor integrated with color twisted nematic liquid crystals display (CTNLCD), SID Int. Symp. Dig. Tech. Papers, 35, 1298, 2004. 25. We consider only conventional LCDs that use nematic liquid crystals as the electrooptic material. There are less common types of LCDs that use other types of liquid crystals, such as cholesteric and ferroelectric liquid crystals.

8080_C006.fm Page 593 Tuesday, April 3, 2007 2:15 PM

Organic Thin-Film Transistors for Flat-Panel Displays

593

26. de Gennes, P.G. and Prost, J., The physics of liquid crystals, 2nd ed., Oxford University Press, New York, 1993. 27. Sheraw, C.D. et al., Organic thin-film transistor-driven polymer-dispersed liquid crystal displays on flexible polymeric substrates, Appl. Phys. Lett., 80, 1088, 2002. 28. Kymissis, I., Dimitrakopoulos, C.D., and Purushothaman, S., Patterning pentacene organic thin-film transistors, J. Vac. Sci. Tech. B 20, 956, 2002. 29. DeFranco, J.A., Schmidt, B.S., Lipson, M., and Malliaras, G.G., Photolithographic patterning of organic electronic materials, Org. Elec., 7, 22, 2006. 30. Gu, G. et al., Organic thin-film transistor with photolithographically patterned top contacts and active layer, 2004 Device Research Conference. 31. A more careful analysis for LCDs must take into account the fact that during charging the pixel voltage does not simply change from one point on the electro-optic curve to another. Rather, the pixel voltage must make a larger transition from one polarity to the other. 32. Measurements of feedthrough using high-impedance probing of a-Si TFT and OTFT electrophoretic pixels are shown in Daniel, J.H. et al., Flexible electrophoretic displays with jet-printed active matrix backplanes, SID Int. Symp. Dig. Tech. Papers, 36, 1630, 2005. 33. Early work by Toshiba on the effect of a-Si TFT instability on AMLCD lifetime is reported in Ibaraki, N. et al., Threshold voltage instability of a-Si:H TFTs in liquid crystal displays, J. Non-Crystalline Solids, 115, 138, 1989. See also Chiang, C. et al., AMLCD lifetime evaluation based on AC electrical instability of amorphous silicon thin-film transistors, Euro Display ’96, 13. 34. Dawson, R.M.A. et al., Design of an improved pixel for a polysilicon active matrix organic light emitting diode display, SID Int. Symp. Dig. Tech. Papers, 29, 11, 1998. 35. Shimoda, T. et al., High resolution light emitting polymer display driven by low temperature polysilicon thin film transistor with integrated driver, Proc. Int. Display Research Conf., 217, 1998. 36. Kanzaki, K. and Sakamoto, M., Direction of low-temperature p-Si technology, SID Int. Symp. Dig. Tech. Papers, 32, 242, 2001. 37. Wu, C.C. et al., Integration of organic LEDs and amorphous Si. TFTs onto flexible and lightweight metal foil substrates, IEEE Electron Dev. Lett., 18, 609, 1997. 38. Tsujimura, T. et al., A 20-inch OLED display driven by super-amorphous-silicon technology, SID Int. Symp. Dig. Tech. Papers, 34, 6, 2003. 39. Chung, K. et al., Large-sized full color AMOLED TV: Advancements and issues, SID Int. Symp. Dig. Tech. Papers, 37, 1958, 2006. 40. Nathan, A. et al., Amorphous silicon backplane electronics for OLED displays, IEEE J. Selected Topics in Quantum Electronics, 10, 58, 2004. 41. Sirringhaus, H.,Tessler, N., and Friend, R.H., Integrated optoelectronic devices based on conjugated polymers, Science, 280, 1741, 1998. 42. Dodabalapur, A. et al., Organic smart pixels, Appl. Phys. Lett., 73, 142, 1998. 43. Chuman, T. et al., Active matrix organic light emitting diode panel using organic thin-film transistors, SID Int. Symp. Dig. Tech. Papers, 35, 45, 2004. 44. Zhou, L. et al., Pentacene TFT driven AMOLED displays, IEEE Electron Dev. Lett., 26, 640, 2005. 45. Brody, T.P. et al., A 6- × 6-in. 20-lpi electroluminescent display panel, IEEE Trans. Electron Devices, 22, 739, 1975. 46. Haskal, E., Polymer light emitting devices overview, SID Int. Symp. Lecture Notes, M-3/3, 2004.

8080_C006.fm Page 594 Tuesday, April 3, 2007 2:15 PM

594

Organic Field-Effect Transistors

47. Dawson, R.M.A. et al., The impact of the transient response of organic light emitting diodes on the design of active matrix OLED displays, Int. Electron Devices Meeting, 875, 1998. 48. Dawson, R.M.A. and Kane, M.G., Pursuit of active matrix organic light emitting diode displays, SID Int. Symp. Dig. Tech. Papers, 32, 372, 2001. 49. Fish, D. et al., A comparison of pixel circuits for active matrix polymer/organic LED displays, SID Int. Symp. Dig. Tech. Papers, 33, 968, 2002. 50. Sanford, J.L. and Libsch, F.R., TFT AMOLED pixel circuits and driving methods, SID Int. Symp. Dig. Tech. Papers, 34, 10, 2003. 51. Nathan, A., Chaji, G.R., and Ashtiani, S.J., Driving schemes for a-Si and LTPS AMOLED displays, J. Disp. Tech., 1, 267, 2005. 52. Sasaoka, T. et al., A 13.0-inch AMOLED display with top emitting structure and adaptive current mode programmed pixel circuit, SID Int. Symp. Dig. Tech. Papers, 32, 384, 2001. 53. Kane, M.G., U.S. Patent Appl. 20050068275, 2005. 54. Chaji, G.R. and Nathan, A., A fast settling current driver based on the CCII for AMOLED displays, J. Disp. Tech., 1, 283, 2005. 55. Fish, D. et al., Optical feedback for AMOLED display compensation using LTPS and a-Si:H technologies, SID Int. Symp. Dig. Tech. Papers, 36, 1340, 2005. 56. Jackson, T.N., Lin, Y.-Y., Gundlach, D.J., and Klauk, H., Organic thin-film transistors for organic light-emitting flat-panel display backplanes, IEEE J. Selected Topics in Quantum Electronics, 4, 100, 1998. 57. Aerts, W.E., Verlaak, S., and Heremans, P., Design of an organic pixel addressing circuit for an active-matrix OLED display, IEEE Trans. Electron Devices, 49, 2124, 2002. 58. Krumm, J. et al., A polymer transistor circuit using PDHHT, IEEE Electron Dev. Lett. 25, 399, 2004. 59. Drury, C.J. et al., Lost-cost all-polymer integrated circuits, Appl. Phys. Lett., 73, 108, 1998. 60. van Lieshout, P.J.G. et al., System-on-plastic with organic electronics: A flexible QVGA display and integrated drivers, SID Int. Symp. Dig. Tech. Papers, 35, 1290, 2004. 61. Sirringhaus, H. et al., Active matrix displays made with printed polymer thin film transistors, SID Int. Symp. Dig. Tech. Papers, 34, 1084, 2003. 62. Burns, S.E. et al., Flexible active-matrix displays, SID Int. Symp. Dig. Tech. Papers, 36, 19, 2005. 63. Kawasaki, M. et al., High-resolution full-color LCD driven by OTFTs using novel passivation film, IEEE Trans. Electron Dev., 53, 435, 2006. 64. Kitamura, M., Imada, T., and Arakawa, Y., Organic light-emitting diodes driven by pentacene-based thin-film transistors, Appl. Phys. Lett., 83, 3410, 2003. 65. Seong, R.-G. et al., Flexible AMOLED backplane technology using pentacene TFTs, 2005 Int. Symp. Super-Functionality Organic Devices, 146. 66. Mizukami, M. et al., Flexible AMOLED panel driven by bottom-contact OTFTs, IEEE Electron Dev. Lett., 27, 249, 2006.

8080_IDX.fm Page 595 Tuesday, April 3, 2007 2:27 PM

Index A AC, see Alternating current Access resistance, OFET, 147 Acene(s), 162–177 derivatives, soluble, 412 fused and extended heteroarenes, 166–172 linear fused rings, 162–166 macrocyclics, 175–177 oligoaryls, 173–175 star-shaped oligomers, 173 Activation energy, dielectric roughness and, 233 Active matrix display backplanes, use of OTFT technology for, 591 Active matrix LCDs (AMLCDs), 552, 564, 568 column driver for, 590 pixel, 580, 582 Active matrix organic light-emitting diode (AMOLED) display(s), 553, 591 advantage over PMOLED displays, 581 architecture, 580 cross-section of, 582 demands on TFT performance, 581 first, 578 pixel, 587 current-programmed, 586 nonideal behavior of, 583 voltage-programmed, 584, 585 Active matrix thin film transistor (AM-TFT), 427 Active tags, 490 Adiabatic electron-transfer regime, 5 Admittance spectroscopy, dielectric–semiconductor interface and, 234 AEY mode, see Auger electron yield mode AFM, see Atomic force microscopy Air-gap stamps, 38, 45 Aligothiophenes, alkyl-substituted, chemical structure of, 405 Alkane system, decay factor for, 322 Alkyl phosphonic acid monolayers, 241 Alkyl-substituted oligothiophenes, 178, 405 Alkyltrichlorosilanes, SAM and, 242 Alternating current (AC), 316 Alumina/aluminum gate dielectric/gate combinations, 240

Ambipolar operation, 38, 43 AMLCDs, see Active matrix LCDs AMOLED, see Active matrix organic lightemitting diode Amorphous polymers, 107, 239 AM-TFT, see Active matrix thin film transistor Anthracene based oligoacenes, 163 ionization peak of, 8 transistor characteristics, 162 Anthradithiophene, 168, 411 Aromatic amines, use of, 173 Arrhenius behavior, 84, 355 Artificial-skin applications, OTFT in, 526, 530 Aryl amines, star-shaped, 174 Asymmetric contacts, 155 Asynchronous communication, 498 Atomic force microscopy (AFM), 32, 150, 279–280, 301, 303 applications, 306 cantilever oscillation, 306 Digital Instruments Multimode, 327 force gradient, 347 GIXD and, 306 growth law extraction using, 347 images, vapor-grown rubrene crystals, 32 importance of, 302 intermittent-contact mode, 312 micrograph, pentacene thin film, 512 oligofluorene thiophene derivatives, 308 SPM techniques based on, 303 STM vs., 303 tapping mode, 303, 305 tip shape, 304 topographs, poly(3-octyl thiophene) thin films, 374 UV-vis spectroscopy analysis and, 310 Atomic scattering factors, intensity ratios from, 261 Auger electron yield (AEY) mode, 285 Azimuthal mean orientation, 287

B Backplanes, OTFT, 570 Band transport

595

8080_IDX.fm Page 596 Tuesday, April 3, 2007 2:27 PM

596 evidence for, 80 solids, 78 Barium titanate ionic motion in, 232 nanoparticle composite dielectrics, 246 BBL, see Poly(benzobisimidazobenzophenanthroline) BCB, see Divinyl-tetramethyl-disiloxanebis(benzocyclobutene) Bendable crystals, 37 Bending tests, OFET, 544 Benzodiselenophene derivative, OFET performance, 172 Bias stability, RFID tag, 503 -stress instability, 561, 576, 578 -temperature stress (BTS), 127, 235 Birth–death models, 358–362 Bixon and Jortner model, 4 Bottom-contact devices, characteristics, 433 Bottom emission, 583 Bottom gate organic thin-film transistor, 230, 512 BR, see Bragg rod Bragg geometry, asymmetric, 257 Bragg peaks, reflections of, 257 Bragg rod (BR), 255 diffraction peaks originating from, 260 intensity, 258, 261 profile, 257, 258 scattering factor, 262 Bragg’s law, 255, 348 Braille sheet display, 548 Breakdown voltages, electrophoretic display architecture and, 577 BTS, see Bias-temperature stress

C Cantilever resonant frequency, 311 Capture number, 354 Carbon nanotube nanoscale transistors, 521 Carrier(s) injection, contact resistance and, 140 mean drift velocity of, 79 mobility, 42 field-effect mobility and, 554 TOF measurements and, 86 CCD, see Charge-coupled device CCD camera, 255–256 Cell(s) gap, 567 sensor, e-skin, 538 thermal sensor network, 542 Chalcogenophene derivatives, 172

Organic Field-Effect Transistors Channel length, 76 decrease of, 146 zero, 148 Channel resistance(s), 140 carrier injection and, 140 equal, 146 linear regime and, 92 Charge carrier de Broglie length of, 79 electronic polarization and, 82 excitation mechanisms, time-resolved photoluminescence and, 59 mobility, nanoscale laminated transistor, 463 total resistance of, 148 Charge-coupled device (CCD), 533 Charge injection electrophoretic display architecture and, 575 mechanisms, comparison of, 144 Charge modulation spectroscopy (CMS), 116 Charge transfer (CT) salts, 171 transition(s), 117 intensity of, 123 P3HT, 122 type co-polymers, 186 Charge transport, see also Oligomers, charge transport in; Organic semiconductors, charge transport in; Single-crystal organic field-effect transistors, charge carrier transport in; Solution-processed organic fieldeffect transistors, charge transport physics intergrain, 514 metal–organic interfaces, 142 model, 79 OTFTs, 265 parameter extraction and, 89 polarons and, 28 properties, in conjugated materials, 2 static disorder and, 66 Charge traps, emptying of, 128 CHB, see Cyclohexylbenzene CHEMFETs, see Chemically sensitive field-effect transistors Chemically sensitive field-effect transistors (CHEMFETs), 508 Chemical sensors, see Organic transistor chemical sensors Chemical vapor deposition (CVD), 342 film growth and, 342 plasma-enhanced, 425–426 Chemiresistor(s) CP sensing layers of, 511 schematic overview of, 508, 509

8080_IDX.fm Page 597 Tuesday, April 3, 2007 2:27 PM

Index Chloroform, mobility in films spin-coated with, 108 CMOS, see Complementary metal-oxide semiconductor CMS, see Charge modulation spectroscopy Coalescence, 353, 356 Cold welding, 466 metal–insulator–metal capacitor fabricated by, 467 PDMS stamps and, 466 schematic of, 467 Column chromatography, oligothiophenes and, 178 Column inversion, 569 Complementary metal-oxide semiconductor (CMOS), 235, 503, 524, 590 Condon approximation, 10 Conducting probe-AFM (CP-AFM), 301 applications, 321 comparison of STM with, 319 experiment, 320, 326 gate-modulated conductance of crystals recorded with, 329, 330 I–V measurements using, 327 molecular tunneling junction formed using, 322 organic semiconductor sexithiophene crystals and, 325 resistance mapping, 320 spatially resolved methods of, 324 spreading resistance profiling, 320 tip, SEM images of, 327 Conductive polymer (CP), 507 films, electrical behavior of, 516–517 interaction with organic vapors, 510 sensing circuits, 507, 508 work functions of, 510 Conformal parylene coating, 35 Conjugated molecules, 74 Conjugated organic materials, 77 Conjugated polymers, see Polymers, conjugated Contact effects in organic field-effect transistors, 139–157 contact engineering, 154–155 ambipolar and light-emitting OFETs, 155 channel dimensions, 155 chemical modifications, 154–155 definition of ohmic contact, 140 measuring contact resistance, 148–154 extrapolation of device resistance to zero channel length, 148–149 gated four-probe measurements, 149–150 Kelvin probe force microscopy, 150–151 measured contact resistance values, 151–154

597 origins of contact resistance, 140–148 charge transport across metal–organic interfaces, 142–144 electronic structure and potential barriers at metal–organic interfaces, 140–142 influence of channel dimensions, 145–146 influence of device architecture, 146–148 Contact potential differences (CPDs), 316 Contact printing technique, cold welding, 466 Contact resistance, 140 carrier injection and, 140 -corrected channel mobility, 43 display applications and, 560 electrode metal work function and, 325 equal, 146 extraction, 91–93 flat-panel displays and, 560–561 method of measuring, 149 Ohm’s law and, 93 oligomer, 91–95 contact resistance extraction, 91–93 origin of contact resistance, 94–95 origins of, 94–95, 140–148 charge transport across metal–organic interfaces, 142–144 electronic structure and potential barriers at metal–organic interfaces, 140–142 influence of channel dimensions, 145–146 influence of device architecture, 146–148 painting of colloidal graphite contacts and, 37 Schottky, 40 thermionic emission and, 39 values, OFET, 151, 152, 153 Cooligomers phenyl-thiophene, 182 thiophene-phenyl, 180 Co-oligothiophene derivatives, 181 Co-polymers charge-transfer type, 186 fluorene-based, 186 thiophene, 187 Copper phthalocyanine (CuPC), 515 Covalent crystals, 74 CP, see Conductive polymer CP-AFM, see Conducting probe-AFM CPDs, see Contact potential differences Critical island size, 352, 355 Crystal(s) bendable, 37 bulk phase, 309 covalent, 74 gate-modulated conductance of, 329, 330 growth mechanisms, polymer, 373 rubrene, 413

8080_IDX.fm Page 598 Tuesday, April 3, 2007 2:27 PM

598 HT-PHT, 379, 380 ionic, 74 Kossel, 344, 345 lamination, treatment of gold contacts before, 43 liquid response to applied voltage, 572 twisted-nematic, 567 molecular, example of, 74 sexithiophene, 325 structures, in-plane, 253 surfaces, growth velocity of, 345 thin film phase, 309 van der Waals box, 345 vapor-Bridgman-grown crystals, 61 Crystalline planes, rocking-curve measurements, 267, 268 Crystallization rate, OFET, 31 CT, see Charge transfer CuPC, see Copper phthalocyanine Current mirror, 587 Current-programmed pixels, 586, 588 Current-voltage curves, 89 CVD, see Chemical vapor deposition Cyclohexylbenzene (CHB), 382

D DAC, see Digital-to-analog converter Data line, 570 warped, 568 DC bias stress effects, 547 DD6T, see Didodecyl -sexithiophene Debye–Waller factor, 258 Defect states, 127 Density of state (DOS), 87, 118 energy diagram, 119 Gaussian transport, 88 injection pathways and, 127 Design, synthesis, and transistor performance of organic semiconductors, 159–228 N-channel organic semiconductors, 191–202 fullerenes and fullerene derivatives, 192 naphthalene diimide derivatives, 194–195 perylene diimide derivatives, 195–197 phthalocyanines, 192–194 polymeric systems, 201–202 quinoid systems, 197 thiophene based N-channel oligomers, 197–200 trifluoromethylphenyl-based oligomers, 200–201 outlook, 202–203

Organic Field-Effect Transistors p-channel organic semiconductors, 161–191 acenes and non-thiophene-based semiconductor, 162–177 polymers, 182–188 precursor method, 188–191 thiophene-based oligomers, 177–182 table of mobilities, 203–214 DHQT, see Dihexylquinquthiopene Dicyanomethylene groups, 197 Didodecyl -sexithiophene (DD6T), 524 Dielectric(s) alternative, 236 constant, factors affecting, 232 –electrode interface, organic semiconductor growth behavior at, 434 films, potential for pinhole defects, 233 nanoparticle based, 245 polymer, 245 roughness, 233 surface, OTFT, 270 thin phase, 273 Dielectric materials, selection and design of, 229–251 alternative gate dielectric strategies, 240–248 gate dielectrics through anodization of thin-metal films, 240–241 nanocomposite and nanostructured dielectrics, 245–248 self-assembled monolayers/multilayers, 242–245 surface treatment of inorganic materials, 241–242 fundamentals and figures of merit, 231–235 dielectric roughness, 233 factors affecting dielectric constant, 232–233 film morphology, 234 importance of interface between dielectric and semiconductor, 234 processing, 234–235 reliability, 235 thickness, 233 major classes of dielectric materials, 235–240 inorganic dielectrics, 235–237 polymer dielectrics, 237–240 Dielectric-semiconductor interface, 230 charge accumulation at, 234 importance of, 234 metal roughness and, 233 Diels–Alder reactions, photoinduced retro, 410 Differential resistance, probe-electrode separation distance versus, 328 Diffraction peaks, out-of-plane, 267 Diffusion limited aggregation (DLA), 361 Digital-to-analog converter (DAC), 590

8080_IDX.fm Page 599 Tuesday, April 3, 2007 2:27 PM

Index Digital Instruments Multimode AFM, 327 Digital lithography, 421, 422, 426, 472 Dihexylquinquthiopene (DHQT), 240 blending approach, 408 grain structure of, 405 Diindenoperylene (DIP), 363 Dimethyl formamide (DMF), 238 Diode-connected transistors, 496 DIP, see Diindenoperylene Diphenylanthracene, chemical structure of, 414 Dipole barrier, 95 Direct-view displays, cost of, 553 Display, see also Flat-panel displays, organic thinfilm transistors for applications, organic TFT parameters for, 553 backplane fabrication, 427 capacitive coupling, 429 electrical response of, 429 resolution, 428 power savings, 570 random brightness variations across, 564 Dithiophene-tetrathiafulvalene (DT-TTF), 33 Divinyl-tetramethyl-disiloxanebis(benzocyclobutene) (BCB), 239 DLA, see Diffusion limited aggregation DMF, see Dimethyl formamide Docosyltrichlorosilane, use as gate dielectric, 114 DOS, see Density of state Drain leakage current, 558 Dressed charge, polaron and, 80 Drude model, 79 DT-TTF, see Dithiophene-tetrathiafulvalene Durham route, solution processable semiconductors, 189 Dynamic scaling theory, 363

E EFM, see Electric force microscopy Ehrlich–Schwoebel barrier, 353, 359, 360, 362 EL devices, see Electroluminescence devices Electric force microscopy (EFM), 301, 310 applications, 313 basic principle, 311 signals, types of, 311 Electrochemical polymerization, selective, 443 Electrodes, printing, 420 Electroless plating, selective, 441 Electroluminescence, organic crystals and, 74 Electroluminescent (EL) devices, 547 Electron-accumulation mode, OFET, 38 Electron-beam evaporation, metal deposition onto PDMS stamps by, 456

599 Electronic artificial skin (e-skin), 530, see also eskins, flexible, large-area development using active matrix method, 547 fabrication of, 530 sensor cells, 538 stretchable, 535 Electronic coupling(s) Koopman’s theorem, 10 through-space interaction and, 4 Electronic-noses (e-noses), 508, 511 Electronic polarization charge carrier and, 82 dielectric constant and, 232 time, energy gap and, 81 Electronic splittings, Koopman’s theorem and, 11 Electron–lattice interactions, 122 Electron transfer (ET), 3, 5, 83 Electron-vibrational coupling interaction, 4 Electrophoretic displays, photolithography and, 430 Electrophoretic materials, electro-optic behavior of, 569 Energy-minimized monolayer structure, simulated diffraction pattern produced by, 269 e-noses, see Electronic-noses Epoxy stamp, 469 e-skin, see Electronic artificial skin e-skins, flexible, large-area, 529–550 bending tests, 544–547 flexible pressure sensors, 530–533 device manufacturing, 531–532 device performance, 533 flexible thermal sensors, 540–544 device manufacturing, 540 device performance, 540–544 discussion, 544 integrated circuits, 533–535 issues and future prospects, 547–548 stretchable e-skins, 535–539 device manufacturing, 536–537 device performance, 537–539 ET, see Electron transfer Etching, selective, 439 Ethylene, energy scheme of, 78 Euler rotation matrix, 262 European spectral mask, RFID tag, 500 Extrinsic semiconductors, 74

F FCWD, see Franck–Condon weighted density Fermi–Dirac distribution, smoothing out of, 79

8080_IDX.fm Page 600 Tuesday, April 3, 2007 2:27 PM

600 Fermi–Dirac statistics, band transport and, 78–79, 122 Fermi distribution, 88 Fermi energy, 79 Fermi level, 40, 140–141 Fermi’s golden rule, 3 FET, see Field-effect transistor Field-effect mobility(ies) carrier mobility and, 554 expression of, 120 oligomer, 404 table of, 203–214 Field-effect threshold voltage, determination of, 63 x-ray treatment and, 64 Field-effect transistor (FET), 75, 103, 264, 529 accumulation layer of, 120 ambipolar, 115, 116 capacitance, 53 channel material, 509 current-voltage curves of, 76 devices performance of, 46 poly(3-hexylthiophene), 264 dielectrics, 113 field-effect mobilities, 107 gate voltage, 140 inorganic, devices based on carbon nanotubes, 45 mobility, in-plane, 265 operation, 38 pentacene, annealed, 547 performance, TFTs and, 130 P3HT-based, 267 sensor configuration, 518 sensor matrix, 534 single-crystal, transconductance characteristics, 47 solution-processed organic semiconductor used for, 182–183 stressing, recovery performance after, 546 studies on amorphous polymers, 107 transport, 108 unique feature of, 76 Field-induced carriers, transport of, 29 Figures of merit, TFT, 231 Film growth technology, 342–343 thermodynamics, 343 Fixed printing master, 419 Flat-panel displays, organic thin-film transistors for, 551–594 active matrix displays, 552 display technologies, 564–589 active matrix OLED displays, 577–589

Organic Field-Effect Transistors liquid-crystal and electrophoretic displays, 564–577 important parameters for display applications, 553–564 bias-stress instability and hysteresis, 561–563 capacitances, 559–560 contact resistance, 560–561 field-effect mobility, 553–555 leakage currents, 558–559 light sensitivity, 563–564 subthreshold swing, 556–558 TFT nonuniformity, 564 threshold voltage, 555–556 organic electronics for displays, 552–553 using organic TFTs for integrated drivers, 589 Flexible pressure sensor, 530 Flexible thermal sensor network, 540, 541 Fluorene co-polymers, 186 substituted phenyl-thiophene cooligmers, 182 -thiophene derivatives, mobility of, 180 Fluorescence yield (FY), 285 advantage of, 285 detection of molecular orientation, 296 Frame inversion, 569 Franck–Condon weighted density (FCWD), 3, 7 Frank–van der Merwe growth, 344 Frenkel–Poole emission, 233 Fullerenes, 192, 193, 413 Fused aromatic systems, soluble derivatives of, 412 Fused ring compounds, 409 FY, see Fluorescence yield

G GaAs hard stamps, 454 Gases, kinetic theory of, 343 Gate bias, 77, 512–513, 518, 519 Gated four-probe technique, OFET, 149 Gate dielectric(s), 112 /gate combinations, alumina/aluminum, 240 materials, polymers used as, 237, 238 strategies, alternative, 240–248 gate dielectrics through anodization of thin-metal films, 240–241 nanocomposite and nanostructured dielectrics, 245–248 self-assembled monolayers/multilayers, 242–245 surface treatment of inorganic materials, 241–242 Gate leakage current, 558

8080_IDX.fm Page 601 Tuesday, April 3, 2007 2:27 PM

Index Gate voltage, 40 dependent mobility, 88 drain current and, 556 Gauge effect, 37, 65 Gaussian distribution evolution of hole mobility, 22 intermolecular distances, 21 Gaussian transport DOS, 88 GBs, see Grain boundaries Gill’s equation, 86 GIXD, see Grazing incidence x-ray diffraction GIXS, see Grazing incidence x-ray scattering Gradual channel approximation, 77 Grain boundaries (GBs), 326, 329, 330, 380 Grazing incidence angle, poly(3-hexylthiophene), 264 diffraction, horizontal surface, 254, 255 Grazing incidence x-ray diffraction (GIXD), 253–276, 348 AFM combined with, 306 characterization of pentacene monolayer by, 269 examples, 264–272 oligo acene-thiophene, 271–272 pentacene, 269–271 poly(3-hexyl thiophene), 264–269 experiment(s) general approach for, 255 scattered intensity in, 258 schematic diagram of, 260 geometry, top and side view of, 256 interpretation of diffraction data, 258–264 calculation of angle between a- and baxes, 263–264 calculation of structure factor, 261–263 kinematic theory, 255 patterns oligo acene-thiophene, 271, 272 poly(3-octyl thiophene) thin films, 375 possible geometrical setups, 255–257 Bragg peaks, 257 Bragg rod profile, 257–258 scattering geometry, 254 Grazing incidence x-ray scattering (GIXS), 253

H Hall conductivity, rubrene OFETs and, 50–53 Hall effect, mobility determined from, 53 Hall geometry, 37 Hall mobility, effective mobility and, 54 Halogen substitutions, molecular packing and, 163 Hard PDMS, 437

601 Hard stamps, 454 Hartree–Fock INDO, 11 HBC, see Hexabenzocoronene HCDF, see Height difference correlation function Head-to-tail poly(3-hexylthiophene) (HT-PHT) aggregation, model for, 385 –dielectric substrate interface, 392 direct solution deposition of, 390 films, 372, 373, 377 commercially available, 379 crystalline morphologies of, 389, 390, 391 drop-cast films, 377, 381 GIXD analysis of, 385 solubility of, 383 TM-AFM phase images, 381 UV-vis spectra of, 384 nanocrystals, in spin-cast films, 391 Height difference correlation function (HCDF), 363 Heisenberg’s uncertainty principle, 80–81 Heteroarene(s) fused, 166, 169 oligomers, 170 Hexabenzocoronene (HBC), 166 derivative of, 412 liquid-crystalline phases, 111 Hexamethyldisilizane (HMDS), 265, 267, 331 HF tag, see High frequency tag Highest occupied molecular orbital (HOMO), 10, 77 delocalized state of, 115–116 energy level, linearly condensed acenes, 162 –LUMO energy gap, 28, 40, 41 positions, measurement of, 142 splitting(s) calculation of, 12 evolution of, 13, 14 subthreshold swing and, 558 wave function, 12 High frequency (HF) tag, 493 antenna in, 493 asynchronous clock generation in, 499 divider-based clock generation in, 498 rectification strategies, 496 Highly ordered pyrolytic graphite (HOPG), 346 HMDS, see Hexamethyldisilizane Hole-accumulation mode, OFET, 38 Hole injection, energy barrier, 124 Hole mobility evolution of, 21 electric field, 18 Gaussian distribution, 22 intermolecular distance, 19 reorganization energy, 19 stacking axis, 20, 21

8080_IDX.fm Page 602 Tuesday, April 3, 2007 2:27 PM

602 field-effect transistors and, 121 ultrapure pentacene crystals, 56 Holstein-like models, 66 Holstein–Peierls model, 54 HOMO, see Highest occupied molecular orbital HOPG, see Highly ordered pyrolytic graphite Hopping distance, 118 incoherent, 53 models, 86, 120, 122, 502 rate, self-exchange reaction, 5 regime, 2 transport, 85 Hot lift-off technique, 469, 470 HT-PHT, see Head-to-tail poly(3-hexylthiophene) Huang–Rhys factor, 4, 7, 8 Hydrocarbon cracking, 65 Hysteresis, 561

I IC, see Integrated circuits Identification tags, see Radio frequency identification tags IEEE standard, OTFT characterization, 563 Imidazolylquinoline compounds, mobilities of, 168 Imprint lithography, 475 Incomplete condensation, 354 Indium tin oxide (ITO), 440, 570 INDO, see Intermediate neglect of differential overlap In-gap trap states, 62 Inkjet printed liquids, parameters controlling, 420 Inkjet printed organic thin film transistors, 419–432 additive methods, 422–427 display backplane fabrication, 427–431 subtractive methods, 420–422 Inkjet printing materials compatibility and, 426 prepatterned substrate, 423 steps of patterning metal feature, 421 TFT fabrication by, 425 Inorganic dielectrics, 235, 236 Inorganic materials, surface treatment of, 241 In-plane crystal structures, analysis of, 253 Insulating polymers, 160 Insulator capacitance of, 76 /organic interface, Fermi level at, 514

Organic Field-Effect Transistors Insulator–semiconductor interface charge distribution and, 96 charge trapping at, 97 electrical potential at, 91 Integrated circuits (ICs), 533, 548 Intergrain charge transport, 514 Intermediate neglect of differential overlap (INDO), 11, 13 International Technology Roadmap for Semiconductors (ITRS), 434 Intrinsic transport, 46–47 Inverse photoemission spectroscopy (IPES), 142 Inverse UPS, 95 Ionic crystals, 74 Ionization potential (IP), 141, 185 IP, see Ionization potential IPES, see Inverse photoemission spectroscopy Island density, 354, 356, 361 growth, 344 shape, 353 size distribution analysis of, 348 at different submonolayer coverages, 358 dynamic, 355 ITO, see Indium tin oxide ITRS, see International Technology Roadmap for Semiconductors

J j-clusters, 353, 357

K Kardar–Parisi–Zhang (KPZ) numerical estimate, 363 Kelvin probe force microscopy (KFM), 143, 150, 279–280, 301, 311 CPD sensitivity, 316 polythiophene-based devices, 317 setup, schematic of, 318 KFM, see Kelvin probe force microscopy Kickback voltage, 573 Kinetics, rate equations, 350 Koopman’s theorem to estimate electronic couplings, 10 Kossel crystal, 344, 345 KPZ numerical estimate, see Kardar–Parisi–Zhang numerical estimate

8080_IDX.fm Page 603 Tuesday, April 3, 2007 2:27 PM

Index

L Langmuir–Blodgett (LB) technique, 177, 372, 511 Langmuir monolayers, mosaicity of, 258 Laser thermal transfer printing, 475, 476 Lattice polarization, 81 Lattice vector, calculation of, 263 LB technique, see Langmuir–Blodgett technique LCD, see Liquid-crystal display Leakage, worst-case, 558 LEDs, see Light-emitting diodes LEEM, see Low-energy electron microscope LEOFETs, see Light-emitting organic field-effect transistors Light-emitting diodes (LEDs), 103 Light-emitting organic field-effect transistors (LEOFETs), 114, 155 Light piping, 563 Light sensitivity, display applications and, 563 Linear acenes, 162 Liquid crystal(s) display (LCD), 552 response to applied voltage, 572 twisted-nematic, 567 Lithography, see also Soft lithography for fabricating organic thin-film transistors digital, 421, 422, 426, 472 imprint, 475 nanoimprint, 476, 477 soft-contact optical, 473, 474 step-and-flash imprint, 476 Lorentz force, 52, 53 Low-energy electron microscope (LEEM), 348–349, 353 Lowest unoccupied molecular orbital (LUMO), 10, 77 delocalized state of, 115–116 energy gap level, barrier height for tunneling through, 322 positions, measurement of, 142 splittings, evolution of, 13, 14 subthreshold swing and, 558 wave function, 12 LUMO, see Lowest unoccupied molecular orbital

M MAC mode, see Magnetic-AC mode Macrocyclics, 175 Magic angle ambiguity, 288 NEXAFS and, 283

603 Magnetic-AC (MAC) mode, 303 Magnetic force microscopy (MFM), 301 Marcus–Hush electron transfer theory, 122 Marcus model, 83 Marcus theory, 4, 84 MC simulations, see Monte Carlo simulations Melting memory effect, 390 3-Mercaptopropyltrimethoxysilane (MPTMS), 453 Metal–insulator–metal capacitor, cold welding fabrication of, 467 Metal-insulator-semiconductor (MIS), 76 Metal-Insulator-Semiconductor Field Effect Transistor (MISFET), 509–510 Metal–insulator–semiconductor structure, accumulation mode, 231 Metal–organic interface(s) charge transport across, 142 injection barrier at, 144 schematic energy diagram of, 39 Metal–organic semiconductor, electronic structure, 141 Metal-oxide active layer, conductivity of, 509 Metal-oxide semiconductor field-effect transistor (MOSFET), 29, 75, 553 channel capacitance in, 559 mobility degradation in, 95 parameter extraction in, 91 transfer characteristic for, 557 Metal phthalocyanines (MPc), 46 Metal–polymer interface, hole injection at, 124 Metal transfer printing, 468 Methanesulfonic acid (MSA), 201 MFM, see Magnetic force microscopy Microcontact printing, 438–447 attractive feature of, 440 defining gold patterns by, 439 selective chemical or electrochemical polymerization, 443–444 selective electroless plating, 441–443 selective etching, 439–441 submicron channel devices fabricated by, 442 Microcrystalline polymers, 108–110 Micromolding in capillaries (MIMIC), 469 schematic of, 471 variation of, 472 Microwave tags, 491 Miller–Abrahams equation, 86 MIMIC, see Micromolding in capillaries MIS, see Metal-insulator-semiconductor MISFET, see Metal-Insulator-Semiconductor Field Effect Transistor Mobility(ies) degradation, MOSFET, 95 edge, 121

8080_IDX.fm Page 604 Tuesday, April 3, 2007 2:27 PM

604 field-effect, table of, 203–214 Poole–Frenkel dependence of, 118 Model(s) birth–death, 358–362 Bixon and Jortner, 4 charge transport, 79 Drude, 79 head-to-tail poly(3-hexylthiophene) aggregation, 385 Holstein-like, 66 Holstein–Peierls, 54 hopping, 86, 120, 122, 502 Marcus, 83 molecular polaron, 83 Mott–Schottky, 94 multiple trap and release, 46, 87 effect of trapping, 47 RFID tag, 502 vanishing of mobility anisotropy within, 49 polaron, 83, 85 thermionic emission, 127 variable range hopping, 87 Molecular crystals example of, 74 polarization in, 80 Molecular electronics, NEXAFS for, 295 Molecular polarization, 81 Molecular polaron, 82–83 Molecular solids, solution deposition of, 404 Molecular tape structure, quinoid systems, 197 Molecular translations, hole mobility and, 20 Molecular weight (MW), 108 Molecule(s) conjugated, 74 diffusing, activation potential profile for, 359 directly functionalized discotic, 412 nearest-neighbor, 260 next-nearest neighbor, 261 reciprocal disk of, 259 Monosilanylene-oligothienlene alternating polymers, 186 Monte Carlo (MC) simulations, 16 electron-transfer rates and, 16 injected carrier hopping and, 118 off diagonal disorder and, 86 reorganization energy and, 18 MOSFET, see Metal-oxide semiconductor fieldeffect transistor Mott–Hubbard insulating state, 58 Mott–Schottky limit, 124 Mott–Schottky (MS) model, 94 Mott-Schottky rule, 141 Mott–Schottky theory, 125

Organic Field-Effect Transistors MPc, see Metal phthalocyanines MPTMS, see 3-Mercaptopropyltrimethoxysilane MSA, see Methanesulfonic acid MS model, see Mott–Schottky model MTR model, see Multiple trap and release model Multiple trap and release (MTR) model, 46, 87 effect of trapping, 47 RFID tag, 502 vanishing of mobility anisotropy within, 49 Multiwalled carbon nanotubes (MWCNT), 313 Mura, 564 MW, see Molecular weight MWCNT, see Multiwalled carbon nanotubes

N Nanoimprint lithography (NIL), 476 difference between S-FIL and, 478 patterning with, 478 schematic of, 477 thin-film transistor fabrication using, 479 Nanoparticle(s) based dielectrics, 245 composite dielectrics, barium titanate, 246 titanium oxide, 246, 247 Nanoscale laminated transistor, charge-carrier mobility, 463 Nanotransfer printing (nTP), 447 advantage of, 458 gold electrodes printed by, 451 noncovalent, 456, 457 schematic of, 449, 452 success of, 448 Naphthalene, geometry relaxations, 8 Naphthalene diimide derivatives, 194–195 Naphthalene tetracarboxylic dianhydride (NTCDA), 194 Naphthalene tetracarboxylic diimide (NTCDI), 194 National synchrotron light source (NSLS), 289 NC cutting plotter, see Numerically controlled cutting plotter N-channel organic semiconductors, 191–202 fullerenes and fullerene derivatives, 192 naphthalene diimide derivatives, 194–195 perylene diimide derivatives, 195–197 phthalocyanines, 192–194 polymeric systems, 201–202 quinoid systems, 197 thiophene based N-channel oligomers, 197–200 trifluoromethylphenyl-based oligomers, 200–201

8080_IDX.fm Page 605 Tuesday, April 3, 2007 2:27 PM

Index Near-edge x-ray absorption fine structure (NEXAFS) spectroscopy, 277–299 advantage of, 282 Auger electron yield mode, 285 data analysis, 287–289 examples, 289–295 molecular electronics, 295 oriented liquid crystalline polymers, 294–295 pentacene, 289–292 poly(3-hexyl thiophene), 292–294 experimental apparatus, 285 experimental considerations, 284–287 future horizons, 295–296 importance of structure, 277–280 magic angle, 283 measurements, sample charging and, 286 NEXAFS background, 280–282 NEXAFS for organic electronics, 282–284 for organic electronics, 282–284 orientation measurements, 283 partial electron yield mode, 283 resonant excitation, band energy diagram, 281 spectrum, poly(3-hexylthiophene), 280, 282 Nearest-neighbor (NN) molecule, 260 Nearfield scanning optical microscopy (NSOM), 301 NEXAFS spectroscopy, see Near-edge x-ray absorption fine structure spectroscopy Next-nearest neighbor (NNN) molecule, 261 NIL, see Nanoimprint lithography NN molecule, see Nearest-neighbor molecule NNN molecule, see Next-nearest neighbor molecule Noncovalent nanotransfer printing, schematic of, 457 Non-thiophene-based semiconductor, 162–177 fused and extended heteroarenes, 166–172 linear fused rings, 162–166 macrocyclics, 175–177 oligoaryls, 173–175 star-shaped oligomers, 173 NSLS, see National synchrotron light source NSOM, see Nearfield scanning optical microscopy NTCDA, see Naphthalene tetracarboxylic dianhydride NTCDI, see Naphthalene tetracarboxylic diimide nTP, see Nanotransfer printing Numerically controlled (NC) cutting plotter, 537

605

O Octadecyltrichlorosilane (OTS), 265, 350, 351, 423, 444 Octadecyl-trimethoxysilane (OTMS), 241 OFET, see Organic field-effect transistor Off-current, 558 Off diagonal disorder, 86 Ohmic contact, definition of, 140 Ohm’s law, contact resistance and, 93 OLEDs, see Organic light-emitting diodes Oligoacene(s), 162 anthracene-based, 163 intramolecular reorganization energy of, 8 pentacene-based, 164 rubrene-based, 164 tetracene-based, 163 -thiophene, GIXD patterns, 271, 272 Oligoaryls, 173, 175 Oligofluorene thiophene derivatives, AFM images of, 308 Oligomer(s) contact resistance values, 152, 153 field-effect mobility, 404 heteroarene, 170 HOMO and LUMO splittings, 15 solubility of, 179 solution deposition of, 403–418 conjugated oligomers, 404–409 fused ring compounds, 409–414 future prospects, 414 star-shaped, 173 thioacene-based, 167 thiophene, 162, 177 molecular engineering of, 179 thiazole rings in, 171 trifluoromethylphenyl-based, 200 Oligomers, charge transport in, 73–101 conjugated oligomers, 77–86 band transport, 78–80 hopping transport, 85–86 polaron transport, 80–85 operating mode of organic thin-film transistor, 75–77 parameter extraction, 89–97 contact resistance, 91–95 mobility degradation, 95–97 threshold voltage, 91 trap limited transport in organic transistors, 86–89 Oligothiophene(s) alkyl-substituted, 178 Bis-silylated versions of, 179 chemical structure of, 406 conjugation length, 178

8080_IDX.fm Page 606 Tuesday, April 3, 2007 2:27 PM

606 ester thermolysis, 409 reactive mesogens based on, 111 star-shaped, 173 tolyl-substituted, 182 OMCs, see Organic molecular crystals One-dimensional position-sensitive detector (1-D PSD) 1-D PSD, see One-dimensional position-sensitive detector OPV cells, see Organic photovoltaic cells Organic crystals field-effect structures, 35 surface phase on free facets of, 34 Organic electronics, 74, 282–284 Organic field-effect transistor (OFET), 29, 139, 160 access resistance, 147 ambipolar, 43, 155 -based display drivers, prototypes of, 403 bending tests, 544 bottom-contact, 410 channel dimensions, 145 charge injection process, 143 conduction channel, lithography-free patterning of, 59 contact resistance values, 151, 152, 153 crystallization rate, 31 DC characteristics, 532 density of mobile carriers in, 53 device(s) architecture, choices of, 146 contact resistance, 146 electrical characterization, voltage-sensing probes during, 149 fabrication bottom contact configurations, 146 top contact architectures, 146 transistor circuitry, 37 four-probe contact geometry, 40, 42, 44 ideal, 139 industrial interest in using, 104 integrated circuits based on, 533 key steps of operation, 2 low-voltage, solution-processable dielectrics for, 113 mobile charges in, 38 mobility anisotropy observed in, 52 ohmic contacts in, 140 oligomers and, 403 onset voltage, photoinduced shift of, 60 operation, linear regime of, 150 parylene passivation layers, 545 pentacene, 245, 409 performance, 104, 161 polaronic transport in, 59

Organic Field-Effect Transistors reliability of, 106 rubrene Hall conductivity in, 50–53 Hall mobility, 55 intrinsic regime for, 55 red boundary of photoinduced threshold shift in, 59 temperature dependence of mobility in, 51 vacuum-gap, 51 Schottky contact resistance, 40 single-crystal, see also Single-crystal organic field-effect transistors, charge carrier transport in air-gap, 63 carrier density, 58 charge transport, 42 CuPc-based, 177 density of defects in, 63 development of, 57, 67 fabrication of, 30, 36 field-effect mobility and, 56 light-induced effects, 59 mobility of, 162 potential of, 66 rubrene-based, 48 surface defects and, 64 tuning of intermolecular distance, 55 vacuum-gap, 65 solution fabrication of stable, 185 sourcedrain electrodes, 114 surface potential profile, 151 technology, growth of, 203 threshold voltage shifts in, 128 top gate, 105, 146, 148 total device resistance, 145 vacuum-gap, 45 Organic film growth kinetics, 349 Organic light-emitting diodes (OLEDs), 58, 94–95, 155, 464 displays, 105, 580 electro-optic behavior of, 578 incorporation of aromatic amines in, 173 light output, 579 operational lifetime of, 579 use of soft-contact lamination to construct, 464, 465 Organic materials atomic number of, 255 most important feature of, 160 Organic molecular crystals (OMCs), 28, 64 Organic photovoltaic (OPV) cells, 279 Organic RFID tag, 500 Organic semiconductor(s), 74 ambipolar, 112 ceramics and, 245

8080_IDX.fm Page 607 Tuesday, April 3, 2007 2:27 PM

Index characterization, measurement for, 289 charge carrier mobility of, 270 charge injection physics of, 105 defects, 279 development, challenge of, 277 growth behavior, 434 growth mechanisms, 203 in situ crystallization of, 273 ionization potential, 141 mobility of, 547 molecule, structural motif, 278 N-channel, 191–202 fullerenes and fullerene derivatives, 192 naphthalene diimide derivatives, 194–195 perylene diimide derivatives, 195–197 phthalocyanines, 192–194 polymeric systems, 201–202 quinoid systems, 197 thiophene based N-channel oligomers, 197–200 trifluoromethylphenyl-based oligomers, 200–201 negative shift of threshold voltage, 127 orbital orientation, 287 p-channel, 161 pentacene, effects of SAMs on mobility for, 242 sexithiophene crystals, 325 single crystals, soft-contact lamination and, 460 soluble discotic, 413 solution-processable, 103–104, 182–183 thin films, obtaining from solution, 371 with trifluoromethyl endgroups, 201 Organic semiconductors, charge transport in, 1–26 electronic coupling, 10–16 influence of intermolecular separation, 12–13 influence of long- or short-axis displacements, 13–16 electron-transfer theory, 3–5 electron-vibration coupling and reorganization energy, 5–10 intramolecular reorganization energy, 5–8 intramolecular reorganization energy of oligoacenes, 8–10 molecular parameters to carrier mobilities, 16–22 Gaussian disorder, 21–22 influence of electric field, 17 influence of intermolecular distance, 18–19 influence of molecular translations, 20–21 influence of reorganization energy, 18

607 Organic surfaces, transport of field-induced carriers on, 29 Organic thin-film transistor (OTFT), 75, 161, 265, 341, 371, 552 active layers investigated in, 515 AMOLED display, 582, 589 applications, polythiophenes used for, 372 artificial-skin applications, 526 backplanes, fabrication of, 570 bottom-contact, 433, 434, 442 channels, 278 characterization, IEEE standard, 563 charge transport in, 265 classification of, 433 CMOS implementation of, 524 commercial success of, 343 conduction mechanisms, 515 devices electrical performance of, 373 PHT-based, 382 display, first reported, 566 doping effects in, 516 drain-current transient, 562 electrical responses, 517 electrophoretic displays using, 568 film growth, 346 first proposal of, 507 high mobility in, 75 mechanical flexibility, 553 nonuniformity, analysis of, 564 pentacene monolayers and, 355 short-channel, 516 threshold voltage, 574 voltage-dependent mobility, 554, 555 performance pentacene, 165 side chains length and, 179 substrate temperature and, 364 top contacts and, 572 response repeatability, 518, 519 scanning probe techniques and, 302 sensors testing pulse program, one-cycle, 520 small molecules and, 341 source-drain current changes, 507 structure, 511 subthreshold swings, 558 testing of spiro-linked compounds in, 169 three-dimensional view of, 76 threshold voltages in, 556 top-contact, 434 trend observed in, 96 use of pentacene in, 269 use of phosphonate groups in, 180

8080_IDX.fm Page 608 Tuesday, April 3, 2007 2:27 PM

608 Organic transistor chemical sensors, 507–528 applications, 524–526 organic thin-film transistor sensors, 511–524 device structure, 511–512 gate-induced response repeatability and enhancement, 516–520 interface-dependent OTFT responses, 515–516 multiparametric OTFT sensors, 512–515 scaling behavior of sensing responses, 522–524 selectivity of OTFT, 521 overview, 508–511 Oriented liquid crystalline polymers, NEXAFS spectroscopy and, 294–295 OTFT, see Organic thin-film transistor OTMS, see Octadecyl-trimethoxysilane OTS, see Octadecyltrichlorosilane Output characteristics, current-voltage, 89

P Parameter extraction, threshold voltage and, 89 Parasitic resistances, 560 Partial electron yield (PEY) mode, 283, 296 Parylene deposition process, 36 dimers, 35 layer, 545 semiconductor–dielectric interface, 35 Parylene C, dielectric films generated using, 239 Passive matrix LCDs, 552 Passive matrix OLED (PMOLED) displays, 577 advantage of AMOLED over, 581 architecture, 580 Passive tags, 490 P3AT, see Poly(3-alkyl thiophene) PBTTT, see Poly(2,5-bis(3-alkylthiophene-2yl)thioeno[3,2-b]thiophene PCBM, see Phenyl C61-butyric acid methyl ester p-channel organic semiconductors PDDA, see Poly (dimethyldiallylammonium chloride) PDLC, see Polymer-dispersed liquid-crystal PDMS, see Polydimethylsiloxane PDPs, see Plasma display panels PECVD, see Plasma-enhanced chemical vapor deposition PEDOTPSS, see Poly(3,4-ethylene dioxythiophene)poly(styrene sulfonic acid) PEEM, see Photoelectron emission microscopy PEN, see Polyethylene naphthalate

Organic Field-Effect Transistors Pentacene AFM topographic image, 331 antibonding orbital directions of, 291 carrier mobilities, 16 charge distribution, 97, 98 charge injection, 314 crystal(s) hole mobility in, 56 structure, 269 devices channel sheet resistance for, 155 drain current response of, 522 KFM and, 317 Diels-Alder reaction of, 189–190 EFM analysis, 314–315 electric field, 17 evaporated, 241 FETs, annealed, 547 film(s) critical island size for, 359 formation, TM-AFM tracking topographs for, 310 PEEM monitoring of, 350 imaging potentials in, 318 ionization peak of, 8, 9 mobility of, 165 monolayer(s) GIXD pattern, 270 OTFTs and, 355 NEXAFS spectroscopy and, 289–292 OFETs, 245 oligoacenes, 164 organic semiconductors, effects of SAMs on mobility for, 242 OTFT(s) characteristics, looping, 562 performance, 165 short-channel, 516 threshold voltage, 574 voltage-dependent mobility, 554, 555 oxidation, product of, 62 perfluorinated version of, 200 precursor(s) first example of, 409 structure of, 410 reorganization energy, 9, 17 soluble precursor of, 189 –substrate interactions, balance between, 270 substrate surface roughness, 364, 366 substrate temperature, 405 TFTs bottom-contact, 444 cold welding and, 466 mobility of, 238, 243, 244, 247 ScL procedure, 458, 459

8080_IDX.fm Page 609 Tuesday, April 3, 2007 2:27 PM

Index TM-AFM measurements of diffusionmediated growth, 307 transistors, 246, 523, 525 triisopropylsilyl, 165 use in OTFTs, 269 vacuum deposition, 314 Pentathienoacene (PTA), 168 Perfluoropolyethers (PFPEs), 437 Perylene diimide derivatives, 195–197 Perylene tetracarboxylic dianhydride (PTCDA), 195 Perylene tetracarboxylic diimides (PTCDI), 195, 540, 541 PES, see Poly(ethylene succinate) PET, see Polyethylene terephthalate PEY mode, see Partial electron yield mode PF-like equation, see Poole–Frenkel-like equation PFPEs, see Perfluoropolyethers Phenoxazine moieties, polymers containing, 188 Phenyl C61-butyric acid methyl ester (PCBM), 112 Phenylene-thiophene co-oligomers, chemical structures of, 407 Phenyl-thiophene cooligmers, fluorene substituted, 182 Photoconductivity, organic crystals and, 74 Photoelectron emission microscopy (PEEM), 349, 350 Photolithography electrophoretic displays and, 430 organic semiconductor and, 434 Photon absorption spectroscopies, transmission geometry of, 284 Photo-oxidation, self-sensitized, 62 Photosensitive polymers, 160 Photovoltaic effect, organic crystals and, 74 P3HT, see Poly(3-hexylthiophene) Phthalocyanines, 176, 192–194 Physical vapor deposition (PVD), 342, 363 Physical vapor transport (PVT), 30, 31 PI, see Polyimide Pinch-off voltage, 91 Pixel(s) AMLCD, 580 AMOLED, 580, 582, 587 ballasted, 573 capacitance, 431 charging, 573 current-programmed, 586, 588 design, optimal, 430 inversion, 569 leakage, 574 voltage-programmed, 584 Plasma display panels (PDPs), 552 Plasma displays, 564

609 Plasma-enhanced chemical vapor deposition (PECVD), 425–426 Plastic optoelectronics, ambipolar operation in OFETs, 43 PLD, see Pulsed laser deposition PLEDs, see Polymer light-emitting diodes PMMA, see Polymethylmethacrylate PMOLED displays, see Passive matrix OLED displays Polarization energy, polyacenes, 83 factor, NEXAFS and, 287 lattice, 81 molecular, 80, 81 times, residence time and, 82 Polaron(s) binding energy, 10, 117, 122 charge transport and, 28 conduction channel and, 59 dressed charge and, 80 mobility, calculations of, 54 model, 83, 85 molecular, 82–83 poly(3-hexylthiophene), 117, 122 stability, 80 transport, 80–85 Marcus model, 83 molecular polaron, 82–83 polarization in molecular crystals, 80–82 Polaronic conduction, 66 Polyacenes metallic mobility in, 55 mobility, 48 polarization energy in, 83 reorganization energy, 85 Polyalkylidene fluorenes, 188 Poly(3-alkyl thiophene) (P3AT), 278, 372 Poly(benzobisimidazobenzophenanthroline) (BBL), 111 ambipolar behavior in, 202 x-ray scattering from, 201 Poly(2,5-bis(3-alkylthiophene-2-yl)thioeno[3,2b]thiophene (PBTTT), 377 Polycyclopentadithiophenes, 186 Polydiacetylene derivatives, synthesis of, 191 Poly(3,3-didodecylquaterthiophene) (PQT-12), 376, 424, 430 Poly (dimethyldiallylammonium chloride) (PDDA), 248 Polydimethylsiloxane (PDMS), 37, 436, 532 formulation, Dow Corning, 436 hard, 437 masks, 473, 474 model, micromolding in capillaries and, 469 solvent swelling, 437

8080_IDX.fm Page 610 Tuesday, April 3, 2007 2:27 PM

610 stamp(s) air-gap, 45 cold welding and, 466 copper lines printed with, 454 creation of, 436 deposition of metals onto, 456 gold patterns printed with, 453 mechanical distortion of, 440, 441 scanning electron micrographs of, 450 schematic of, 438 second-generation, 437 silver dots printed with, 454 stamping technique, modification of, 37 substrate, sequential deposition, 461 Poly(3,4-ethylene dioxythiophene)poly(styrene sulfonic acid) (PEDOTPSS), 434, 435, 443, 521 Poly(ethylene succinate) (PES), 377 Polyethylene naphthalate (PEN), 435, 442, 531, 536, 565 Polyethylene terephthalate (PET), 104, 440, 564 Poly(3-hexylthiophene) (P3HT), 108, 264, 280 antibonding orbital directions of, 294 chains, self-assembly of, 394 charge-induced absorption spectrum of, 117 charge transfer transition, 122 conductivity, 128 domains, orientations of, 266, 267 electron affinities, 114 films, charge mobility values of, 265 K-edge NEXAFS spectrum for, 280, 282 microstructure, 264 microwires, single-crystal, 394 mobility of, 108, 109 nanowires, TM-AFM images of, 392 NEXAFS spectroscopy and, 292–294 oxidative stability of, 110 photostability, 110 polarons in, 122 regiorandom, 379–380 regioregular, 110 solubility of, 377 Polyimide (PI), 443–444 Polymer(s) alkyl groups, 186 -based TFT, 428, 429 coated substrate, work function of, 124 conductive interaction with organic vapors, 510 sensing circuits, 507, 508 conjugated, 106–110 amorphous polymers, 107 charge transport of polaronic carriers in, 117 microcrystalline polymers, 108–110

Organic Field-Effect Transistors crystal growth mechanisms, 373 dielectric(s) capacitances, 245 properties for, 237 slow polarization in, 562 -dispersed liquid-crystal (PDLC), 565 electrodes, laser thermal transfer printing and, 475 heterointerface, solution-processed, 113 ink, spreading of, 422 insulating, 160 light-emitting diodes (PLEDs), 422 monosilanylene-oligothienlene alternating, 186 neutral, energy diagram of, 117 oriented liquid crystalline, NEXAFS spectroscopy and, 294–295 photosensitive, 160 polyethylenenaphthalate, 531 precursor, 189, 190 regiorandom, 183 regioregular, 185 resist, scanning electron micrograph of, 477 semiconducting, long axis orientation measurement of, 283 semiconductors, long-axis orientation, 279 semicrystalline crystallization behaviors of, 377 melting memory effect, 390 solubility of, 182 solution deposition of, 371–401 contact resistance and, 153 effect of solvent, 382–387 molecular structure, 374–382 processing condition, 387–392 systems n-type, 202 saturation regime, 201 TFT performance and, 242 thiophene-based, 184, 189, 382 transistor stability, 186 ultrathin cross-linked, 243, 244 Polymerization, electrochemical, 443 Polymethylmethacrylate (PMMA), 237, 350, 427, 468 metal transfer printing and, 468 use as gate dielectric materials, 237 Poly(3-octyl thiophene) thin films AFM topographs of, 374 grazing incidence x-ray diffraction patterns of, 375 Poly-para-phenylene-vinylene (PPV), 78, 187 Polysilicon thin-film transistors (poly-Si TFTs), 552, 558 Poly-Si TFTs, see Polysilicon thin-film transistors

8080_IDX.fm Page 611 Tuesday, April 3, 2007 2:27 PM

Index Polystyrene (PS), 237 leakage currents, 243 phosphonate-terminated, 247 ultrahigh-molecular-weight, 413 use as gate dielectric materials, 237 Polythiophene amphiphilic, self-assembly of, 372 -based devices, KFM and, 317 synthesis of, 185 Polytriarylamine (PTAA), 107, 113 field-effect mobility, 120, 121 mobility of, 240 Polyvinyl alcohol (PVA), 237, 435, 572 -co-poly(vinyl acetate)-co-poly(itaconic acid) (PVAIA), 246 etch barrier, mechanical cracking of, 435 loading of titanium oxide nanoparticles in, 246 Polyvinylphenol (PVP), 237, 422 leakage currents, 243 reproducibility, 243 solution-processable, 442 Poole–Frenkel (PF)-like equation, 86, 118 Porphyrins, macrocyclic, 176 Potential energy surfaces, 6 Powder averaging, 258 PPV, see Poly-para-phenylene-vinylene PQT-12, see Poly(3,3-didodecylquaterthiophene) Precursor polymers, 189, 190 route, solution processable semiconductors, 189 Pressure sensor(s) conformable network of, 537 flexible, 530 Printed device technology, RFID tag, 503 Printed organic electronics, circuit implications of, 504 Printing, see also Microcontact printing laser thermal transfer, 475, 476 metal transfer, 468 methods, roll-to-roll, 419 PS, see Polystyrene PTAA, see Polytriarylamine PTCDA, see Perylene tetracarboxylic dianhydride PTCDI, see Perylene tetracarboxylic diimides Pulsed laser deposition (PLD), 342 PVA, see Polyvinyl alcohol PVAIA, see PVA-co-poly(vinyl acetate)-copoly(itaconic acid) PVD, see Physical vapor deposition PVP, see Polyvinylphenol PVT, see Physical vapor transport

611

Q Quarterthiophenes, fabrication of transistors based on, 198 Quinoid derivatives, 198 Quinoid systems, 197

R Radio frequency identification (RFID) tags, 277, 403, 489–505, 529 all-printed RFID tags, 492–500 antenna stage, 493–495 digital section and modulation stage, 497–500 rectifier/power supply and clamp, 495–497 antenna in, 493 archetypal first organic RFID tag, 500–501 archetypal HF passive, 493 asynchronous clock generation in, 499 bias stability, 503 coupling efficiency, 494 digital subcircuit architecture for, 497 disadvantage for, 547 divider-based clock generation in, 498 European spectral mask for, 500 implications of tag architecture on device considerations, 501–504 circuit issues, 503–504 transistor performance and structural implications, 501–503 inductively coupled, 494, 495 rectification strategies, 496 silicon, 491–492 standards and classifications, 490–491 13.56 MHz RFID, 491 135 kHz RFID, 490–491 900 MHz and 2.4 GHz RFID, 491 Rapid roughening, 364 Rate equation(s) elements, 353 microscopic, 352 Real space lattice vector, 263 Reciprocal disk, molecule, 259 Rectification circuit, RFID tag, 495 Reflection high-energy electron diffraction (RHEED), 353 Regiorandom polymer, 183 Regioregular polymers, 185 Relaxation energy, harmonic approximation, 7 Reorganization energy Bixon and Jortner model and, 4 intramolecular, 5

8080_IDX.fm Page 612 Tuesday, April 3, 2007 2:27 PM

612 Monte Carlo simulations and, 18 oligoacenes, 8 pentacene, 9, 17 polaron binding energy and, 10 Resistance chain length plots vs., 324 differential, probe-electrode separation distance versus, 328 grain boundary, 330 low-bias, 323 mapping, 320 parasitic, 560 Return signal, 305 RFID tags, see Radio frequency identification tags RHEED, see Reflection high-energy electron diffraction Rigid-band approximation, 81 Ring opening metathesis polymerization (ROMP), 243 Ring oscillator sensor, recovery of, 525 RMS amplitude value, see Root mean square amplitude value Rocking-curve measurements, crystalline planes, 267, 268 Roll-to-roll printing methods, 419 ROMP, see Ring opening metathesis polymerization Root mean square (RMS) amplitude value, 305 Row inversion, 569 Rubber conductive, 538 pressure-sensitive, response time of, 534 Rubrene chemical structure of, 414 crystal(s) growth, inhibition of, 413 packing of molecules in, 491 electronic coupling, 15 OFET, 41, 48 air-gap, 63 carrier density, 58 field-effect mobility and, 56 Hall conductivity in, 50–53 Hall mobility, 55 intrinsic regime for, 55 red boundary of photoinduced threshold shift in, 59 temperature dependence of mobility in, 51 threshold voltage of, 63 tuning of intermolecular distance, 55 vacuum-gap, 51 oligoacenes, 164 polar plot of mobility, 50 single crystal transistor, current-voltage characteristics, 462

Organic Field-Effect Transistors solubility of, 166 sublimed grade, 33 Rydberg excitations, 280 Rydberg transitions, 280

S SAM, see Self-assembled monolayer Sample charging, NEXAFS and, 286 Saturated compounds, 77 Saturation regime, 89, 90 Scaling laws, 350 Scanning capacitance microscopy (SCM), 301 Scanning electron microscopy (SEM), 348, 522 Scanning Kelvin probe microscopy (SKPM), 125, 126, 316–319 Scanning probe microscopy (SPM), 301 techniques based on AFM, 302, 303 variant of, 319 Scanning probe techniques, 301–339 atomic force microscopy, 303–310 AFM tip shape, 304–305 applications, 306–310 basic principle of intermediate contact mode, 305–306 conducting probe AFM, 319–331 electric force microscopy, 310–316 applications, 313–316 basic principle, 311312 Kelvin probe force microscopy, 316–319 Scanning tunneling microscopy (STM), 34, 301, 347, 379 AFM vs., 303 comparison of CP-AFM with, 319 growth law extraction using, 347 images, HT-PHT crystal, 379, 380 packing motif observed with, 34 tunneling current and, 320 Scattering factor, Bragg rod, 262 Schottky barrier, 39, 45, 66 calculated, 125 contacts behaving as, 95 ScL, see Soft-contact lamination SCLC, see Space charge limited current SCM, see Scanning capacitance microscopy Scotch tape adhesion tests, 448 Secondary ion mass spectrometry (SIMS), 282 Seemann–Bohlin geometry, 257 Selective etching, 439 Self-affine surfaces, HCDF and, 363 Self-assembled monolayer (SAM), 108, 154, 270 alkylamine, 241 alkylthiol, 439 contact pretreatment and, 154

8080_IDX.fm Page 613 Tuesday, April 3, 2007 2:27 PM

Index CP-AFM approach to junction formation and, 321 crystallization motifs and, 288 dielectric data, 244 effect on mobility for pentacene organic semiconductors, 242 formation, thiol-terminated hydrocarbon molecules for, 155 functioning of as monolayer dielectric, 242 HT-PHT–dielectric substrate interface, 392 hydrophobic, 108 NEXAFS spectroscopy and, 295 OTFT dielectric surface and, 270 perylene deposition and, 350 -templated substrates, 446 use as covalent glue, 451, 455 Self-assembled polymers, solution casting and, 374 Self-exchange reaction carrier mobilities and, 2 hopping rate for, 5 Self-sensitized photo-oxidation, 62 SEM, see Scanning electron microscopy Semiconducting materials, electrostatic behavior of, 313 Semiconductor(s) acene fused-thiophene hybrid, 191 amine-based polymer, 129 amorphous polymer, field effect mobility for, 239 complementary metal oxide, 235 –dielectric interface parylene, 35 polymer ordering at, 392 extrinsic, 74 –insulator interface, carrier trapping at, 60 intrinsic, 73 mobility, alkyl chains and, 169 –molecule interfaces, CP-AFM and, 321 organic, 74 ambipolar, 112 ceramics and, 245 characterization, measurement for, 289 charge carrier mobility of, 270 charge injection physics of, 105 defects, 279 development, challenge of, 277 growth behavior, 434 growth mechanisms, 203 –inorganic, 446 in situ crystallization of, 273 ionization potential, 141 mobility of, 547 molecule, structural motif, 278 negative shift of threshold voltage, 127

613 orbital orientation, 287 p-channel, 161 pentacene, effects of SAMs on mobility for, 242 sexithiophene crystals, 325 single crystals, soft-contact lamination and, 460 soluble discotic, 413 solution-processed, 182–183 thin films, obtaining from solution, 371 with trifluoromethyl endgroups, 201 polymer, long-axis orientation, 279 printed, 425 printing, 422 solution processable, precursor route, 189 tertiary diamine structure, 169 Semicrystalline polymers crystallization behaviors of, 377 melting memory effect, 390 Sensistor AB, 509 Sensor cells, e-skin, 538 S-FIL, see Step-and-flash imprint lithography Shadow masking, patterning using, 572 Sheet scanner, 548 Signal-to-noise ratio CHEMFETs and, 510 NEXAFS and, 285 Silicon chips, cost of, 492 radio frequency identification using, 491–492 SIMS, see Secondary ion mass spectrometry Single crystal mobility, 514 Single-crystal organic field-effect transistors, charge carrier transport in, 27–72 charge transport on organic single crystal surface, 38–59 anisotropy of mobility, 48–50 basic FET operation, 38–46 comparison with Holstein–Peierls model, 54–55 longitudinal and Hall conductivity in rubrene OFETs, 50–54 multiple trap-and-release model, 46–48 photoinduced processes in single-crystal OFETs, 58–59 surface versus bulk transport, 56–58 tuning of intermolecular distance, 55–56 defects at organic crystal surface, 59–65 bulk and surface electronic defects in organic crystals, 61–63 density of defects in single-crystal OFETs, 63–64 single-crystal OFETs as tools to study surface defects, 64–65 fabrication of single-crystal OFETs, 30–38

8080_IDX.fm Page 614 Tuesday, April 3, 2007 2:27 PM

614 field effect in small-molecule organic semiconductors, 28–30 Single wall carbon nanotubes (SWNT), 475 SIP, see Surface-initiated polymerization SKPM, see Scanning Kelvin probe microscopy Small molecule precursors, 191 Soft-contact lamination (ScL), 448, 458 building of thin-film transistors, 464 OLED construction using, 464, 465 schematic of, 459 study of charge transport using, 460 Soft lithography for fabricating organic thin-film transistors, 433–488 cold welding, 466–468 device structures and conventional fabrication techniques, 433–435 hot lift-off, 469 imprint lithography, 475–482 laser thermal transfer printing, 475 metal transfer printing, 468 microcontact printing, 438–447 other microcontact printing derivatives, 445–447 selective chemical or electrochemical polymerization, 443–444 selective electroless plating, 441–443 selective etching, 439–441 stamp-and-spin-cast, 444–445 micromolding in capillaries, 469–473 nanotransfer printing, 447–458 soft-contact lamination, 458–466 soft-contact optical lithography, 473–475 stamps for soft lithography, 435–438 Soft x-ray absorbance, NEXAFS measurement of, 284 SOG, see Spin-on glass Solid(s) band transport, 78 categories, electronic space distribution in, 75 molecular, solution deposition of, 404 most important physical distinction in, 73 Soluble fused ring derivatives, structures of, 411 Soluble precursors, fused aromatic systems and, 410 Solution -based processing, unconventional deposition methods, 160 casting, self-assembled polymers and, 374 processable semiconductors, 189 Solution-processed organic field-effect transistors, charge transport physics of, 103–137 charge injection physics, 124–127 charge transport physics, 115–123

Organic Field-Effect Transistors defect states and device degradation mechanisms, 127–130 gate dielectrics, 112–115 n-type organic semiconductors, 111–112 outlook, 130 p-type organic semiconductors, 106–111 conjugated polymers, 106–110 small molecules, 110–111 Solvent(s) solubility of HT-PHT in, 383 swelling, PDMS, 437 Space charge limited current (SCLC) experiments, 38 measurements, 61, 62 Spin-coating, oligothiophenes and, 178 Spin-on glass (SOG), 440 SPM, see Scanning probe microscopy Spreading resistance profiling (SRP), 320 SRP, see Spreading resistance profiling Stamp(s) air-gap transistor, 38 epoxy, 469 hard, 454 soft lithography, 435 Stamp-and-spin-cast technique, 444, 445 Static roughness exponent, 363 Step-and-flash imprint lithography (S-FIL), 476 difference between NIL and, 478 schematic, 481 template quality, 482 STM, see Scanning tunneling microscopy Stranski–Krastanov growth, 344 Substrate gold-coated, 443 inkjet printing onto a prepatterned, 423 PDMS -coated plastic, 459 sequential deposition, 461 polymer coated, work function of, 124 surface roughness of, 364 temperature, pentacene, 405 Subthreshold swing, 556–558 Superconductivity, organic crystals and, 74 Surface-initiated polymerization (SIP), 243 SWNT, see Single wall carbon nanotubes Sylgard PDMS formulation, 436 Symmetric contacts, 155 Synchrotron radiation, 254, 256–257

T Tags-talk-first protocols, 498 Tantalum oxide, permittivity of, 236

8080_IDX.fm Page 615 Tuesday, April 3, 2007 2:27 PM

Index Tapping mode (TM), 303 atomic force microscopy (TM-AFM), 270, 271 intermittent tip-sample contact in, 303 tip geometry, 305 TC, see Transconductance change TCNQ, see 7,7,8,8-Tetracyanoquinodimethane TEM, see Transmission electron microscopy Terthiophene, 406 Tetrabenzoporphyrin precursor, 411 Tetracene ionization peak of, 8 oligoacenes, 163 single crystals grown from vapor phase, 32, 34 Tetrachlorocyclohexadiene, 409 7,7,8,8-Tetracyanoquinodimethane (TCNQ), 31, 37, 197 Tetrahydrofuran (THF), 382 Tetrathiafulvalene (TTF),171, 172 TEY, see Total electron yield TFT, see Thin-film transistor Thermally activated resistivity, interpretation of, 74 Thermal sensor(s) network, 541, 543 stand-alone, 540 Thermionic emission, 39, 127, 141 THF, see Tetrahydrofuran Thiazole oligomers, transistors fabricated based on, 200 Thin film characterization techniques, 346–349 electron-based techniques, 348–349 scanning probe techniques, 347–348 x-ray scattering, 348 Thin films, see Vacuum evaporated thin films Thin-film transistor (TFT), 29, 104, 229, 419, see also Organic thin film transistor achieving high capacitances in, 230 active matrix, 427 amorphous silicon,104 applications for, 419 backplanes, 431, 591 bottom-contact, digital lithography and, 472 bottom gate organic, 230 CuPc-based, 46 data, printed semiconductors, 425 device(s) double gate, 239 fabrication, bottom-gate, 426 oligoaryls used in, 175 encapsulation, 427 environmental stability of, 428 fabrication with patterned gate electrodes, 425 FET performance and, 130 inkjet printing on fabrication of, 420

615 light illumination in, 58 nanoimprint lithography and, 479 nonuniformity, 564 on–off current ratio, 575 organic, 75, 161, 265 pentacene bottom-contact, 444 cold welding and, 466 mobility of, 238, 243, 244, 247 performance demands of AMOLED display on, 581 polymers and, 242 polysilicon, 558 printing approaches used to fabricate, 419 prototypical organic, 229 rubrene, 67 schematic energy band diagram for, 513 soft-contact lamination for building, 464 trend observed for, 57 Thioacene-based oligomers, 167 Thiophene(s) co-oligomerization of, 406, 407 copolymers, 187 core-substituted, 183 derivatives, 181, 199 N-channel oligomers, 197–200 oligomer(s), 162, 177 chemical structure of, 408 molecular engineering of, 179 thiazole rings in, 171 -phenyl cooligomers, 180 polymers, 184, 189, 382 star-shaped, 174 Threshold voltage, 91, 555–556 Time of flight (TOF) experiments, 28, 38, 48 measurements carrier density and, 29 carrier mobility and, 86 intrinsic band-like polaronic conduction, 52 Time-resolved photoluminescence, charge-carrier excitation mechanisms and, 59 TIPS pentacene, see Triisopropylsilyl pentacene TLM, see Transfer line method TM, see Tapping mode TM-AFM, see Tapping mode atomic force microscopy TOF, see Time of flight Tolyl-substituted oligothiophenes, 182 Top gate architectures, OFET, 146, 148 Total electron yield (TEY), 285 TPD, pentacene values and, 9 Transconductance change (TC), 91, 92 Transfer characteristic, current-voltage, 89

8080_IDX.fm Page 616 Tuesday, April 3, 2007 2:27 PM

616 Transfer integral amplitudes, 22 Transfer line method (TLM), 92 alternative method to, 93 drawbacks, 93 Transistor, see also Field-effect transistor circuitry, microfabrication methods, 37 DC characteristics of, 545 diode-connected, 496 performance, see Design, synthesis, and transistor performance of organic semiconductors Transition dipole matrix element, 282 Transmission electron microscopy (TEM), 245 Transmission line plot, 149 Transport states, Gaussian density of, 118 Transverse gate electric field, 59 Trap(s) exponential distribution of, 88 -dominated regime, 52 -dominated transport, 46–47 Trichlorobenzene, spin-casting from, 108, 109 Trifluoromethylbenzenethiol, 43 Trifluoromethylphenyl-based oligomers, 200–201 Triisopropylsilyl (TIPS) pentacene, 165 TTF, see Tetrathiafulvalene

U UHF tags, see Ultrahigh frequency tags UHMW-PS, see Ultrahigh-molecular-weight polystyrene UHP, see Ultrahigh-purity Ultrahigh frequency (UHF) tags, 491 Ultrahigh-molecular-weight polystyrene (UHMW-PS), 413 Ultrahigh-purity (UHP), 32–33 Ultraviolet photoelectron spectroscopy (UPS), 95, 142 Unit cell atomic coordinates in, 261 centered rectangular, 261 laboratory coordinate, 262–263 UPS, see Ultraviolet photoelectron spectroscopy

V Vacuum evaporated thin films, 341–369 describing film growth, 343–345 effects of substrate, 364–365 surface energy, 364 surface roughness, 364–365 film growth technology considerations, 342–343 inorganics versus organics, 345–346

Organic Field-Effect Transistors organic film growth kinetics, 349–364 rate equations (macroscopic), 363–364 rate equations (microscopic), 352–362 thermodynamic driving force, 351–352 outlook, 365–366 thin film characterization techniques, 346–349 electron-based techniques, 348–349 scanning probe techniques, 347–348 x-ray scattering, 348 Vacuum-gap OFET, 45 Vacuum-gap stamps, 37–38 Valence band offset, 141 van der Waals forces, 28, 37 Vapor-Bridgman-grown crystals, 34, 61 Variable range hopping (VRH) model, 87 Vector orbital, 287 Video-rate display, disadvantage for, 547 Vinylene groups, incorporation in thiophene backbone, 174–175 Vissenberg hopping model, 120, 122 Volmer–Weber growth scenario, 344 Voltage(s) bias condition, 522 breakdown, 577 change, ballasted pixel and, 573 -to-current conversion, 584 kickback, 573 -programmed AMOLED displays, 590 -programmed pixels, 584, 585 threshold, 91, 555–556 VRH model, see Variable range hopping model

W Warped data, 568

X Xerographic materials, aromatic amines used as, 173 XPS, see X-ray photoemission spectroscopy X-ray(s), see also Grazing incidence x-ray diffraction absorbance, soft, measurement of, 284 diffraction kinematic theory of, 254 measurements, synchrotron, 266, 267 thin-film, 272 wide-angle, 269 photoemission spectroscopy (XPS), 280 scattering, 348 anti-Bragg configuration, 353, 359, 362 Bragg’s law and, 348

8080_ColorInsrt.fm Page 1 Tuesday, April 3, 2007 2:16 PM

5

Analyte delivers trigger pulse

VG(V)

0 –5

(V) 5 0 Ids (μA) –1 0 300

400 500 600 Time (sec) (a)

700

dDDα6T/1-hexanol, Vds = –5V

Ianalyte (μA)

–1.0

–0.5

0.0

0

10 Time (sec) (b)

Vg = –5V Vg = –4V Vg = –3V Vg = –2V Vg = –1V 20 Vg = 0V

dDDα6T / 1-hexanol, on current 0.05 0.00 –0.05 –0.10 –0.15

Cycle number

60 40 20 0 0

10 Time (sec)

20

(c)

FIGURE 6.2.8 (a) One-cycle OTFT sensors testing pulse program; (b) source-drain transient currents at different gate biases; (c) color-coded response plot at Vg = –5 V for 70 cycles. (From Torsi, L. and Dodabalapur, A., Anal. Chem., 77, 380A, 2005. With ACS permission.)

8080_ColorInsrt.fm Page 2 Tuesday, April 3, 2007 2:16 PM

OTFT Array Map of Measured Values R01T R01B R02T R02B R03T R03B R04T R04B R05T R05B R06T R06B R07T R07B R08T R08B R09T R09B R10T R10B

LOT ID : C01 3.115E+00 2.945E+00 2.726E+00 2.675E+00 1.754E+00 1.803E+00 1.420E+00 1.813E+00 1.614E+00 8.876E-01 1.242E+00 8.747E-01 8.785E-01 7.418E-01 1.030E+00 6.337E-01 7.000E-01 5.079E-01 8.298E-01

042000 C02 2.690E+00 2.326E+00 1.729E+00 1.446E+00 9.689E-01 1.107E+00 1.201E+00 1.207E+00 1.159E+00 9.204E-01 1.324E+00 5.328E-01 6.265E-01 3.324E-01 4.862E-01 4.062E-01 5.638E-01 4.768E-01 6.500E-01

Parameter : C04 1.171E+00 1.229E+00 1.809E+00 1.422E+00 1.467E+00 1.158E+00 1.515E+00 1.070E+00 9.070E-01 8.383E-01 9.629E-01 8.551E-01 8.066E-01 6.897E-01 1.089E+00 9.137E-01 1.157E+00 1.366E+00 7.971E-01 1.011E+00 1.279E+00 1.109E+00 5.148E-01 8.579E-01 8.678E-02 1.304E+00 6.123E-01 5.899E-01 6.476E-01 9.465E-01 5.548E-01 4.201E-02 7.050E-01 4.025E-01 -6.597E-02 1.035E+00 5.002E-01 7.616E-01 C03

Vtx

C05 C06 C07 1.532E+00 9.547E-01 9.728E-01 1.216E+00 9.453E-01 1.310E+00 1.143E+00 1.137E+00 1.774E+00 1.101E+00 9.643E-01 9.992E-01 1.120E+00 7.398E-01 9.365E-01 1.016E+00 6.955E-01 1.022E+00 9.164E-01 6.254E-01 -1.833E-01 1.292E+00 1.439E+00 8.892E-01 1.050E+00 1.638E+00 5.651E-01 1.594E+00 1.107E+00 1.557E+00 1.591E+00 1.021E+00 1.597E+00 8.943E-01 1.157E+00 1.182E+00 2.154E-01 1.626E+00 1.429E+00 5.748E-01 9.824E-01 1.049E+00 7.051E-01 1.359E+00 1.163E+00 1.092E+00 1.388E+00 6.384E-01 7.662E-01 7.677E-01 5.719E-01 3.294E-01 5.039E-01 7.272E-01 4.204E-01 5.779E-01

C08 C09 C10 2.621E-01 7.534E-01 7.870E-01 5.995E-01 5.461E-01 7.376E-01 8.111E-01 1.029E+00 7.470E-01 9.633E-01 1.000E+00 1.046E+00 4.074E-01 5.789E-01 3.806E-01 7.112E-01 4.184E-01 -1.027E+00 1.307E-01 2.541E-01 6.303E-01 -3.332E-01 5.673E-01 1.424E+00 1.386E+00 5.132E-01 1.059E+00 9.657E-01 6.730E-01 9.952E-01 1.082E+00 6.387E-01 9.450E-01 9.474E-01 8.080E-01 7.307E-01 1.074E+00 1.104E+00 5.722E-01 1.059E+00 9.190E-01 8.140E-01 1.064E+00 8.764E-01 3.788E-01 7.248E-01 7.346E-01 6.269E-01 8.566E-01 2.760E-01 1.219E+00 6.684E-01 -2.904E-01 3.418E-01 4.964E-01

Min.

Max.

1

-1.00E+15

-6.766E-01 -1.364E-01

3

-6.766E-01

14

-1.364E-01

4.038E-01

78

4.038E-01

9.441E-01

70

9.441E-01

1.484E+00

15

1.484E+00

2.025E+00

1

2.025E+00

2.565E+00

5

2.565E+00

1.00E+15

Min.

Max.

2

-1.00E+15

2.543E-01

OTFT Array Map of Measured Values R01T R01B R02T R02B R03T R03B R04T R04B R05T R05B R06T R06B R07T R07B R08T R08B R09T R09B R10T R10B

LOT ID : C01 4.949E-01 4.880E-01 5.025E-01 5.358E-01 4.973E-01 4.483E-01 4.271E-01 4.503E-01 4.705E-01 4.168E-01 4.322E-01 4.116E-01 4.312E-01 4.509E-01 4.395E-01 4.601E-01 4.787E-01 4.487E-01 4.647E-01

042000 C02 5.240E-01 5.040E-01 4.560E-01 4.390E-01 4.333E-01 4.291E-01 4.234E-01 4.162E-01 4.417E-01 4.199E-01 4.291E-01 4.251E-01 4.256E-01 4.211E-01 4.384E-01 4.227E-01 4.377E-01 4.114E-01 3.927E-01

Parameter : usat_max C04 C05 5.186E-01 4.936E-01 4.032E-01 4.888E-01 4.344E-01 4.232E-01 4.700E-01 4.543E-01 4.179E-01 4.300E-01 4.086E-01 3.874E-01 4.335E-01 4.289E-01 4.095E-01 3.779E-01 3.572E-01 4.289E-01 3.826E-01 3.798E-01 3.803E-01 3.751E-01 3.734E-01 4.172E-01 3.864E-01 3.722E-01 3.554E-01 3.889E-01 3.467E-01 3.658E-01 3.798E-01 3.619E-01 3.895E-01 3.655E-01 3.208E-01 3.937E-01 3.753E-01 2.787E-01 4.026E-01 3.907E-01 3.449E-01 4.241E-01 3.915E-01 3.726E-01 4.012E-01 3.286E-01 4.045E-01 4.116E-01 3.656E-01 3.989E-01 3.231E-01 4.071E-01 1.864E-01 3.817E-01 4.224E-01 4.212E-01 C03

C06 4.694E-01 4.273E-01 3.812E-01 3.596E-01 3.743E-01 3.504E-01 3.679E-01 3.588E-01 3.578E-01 3.421E-01 3.337E-01 3.268E-01 3.565E-01 3.848E-01 3.779E-01 3.639E-01 4.067E-01 4.196E-01

C07 4.147E-01 4.332E-01 3.850E-01 3.827E-01 3.590E-01 3.559E-01 3.449E-01 3.570E-01 3.276E-01 3.158E-01 3.425E-01 3.583E-01 3.703E-01 3.545E-01 3.857E-01 3.813E-01 3.898E-01 3.730E-01 3.976E-01 4.088E-01

C08 3.999E-01 4.014E-01 3.823E-01 3.948E-01 3.600E-01 3.589E-01 3.402E-01 3.552E-01 3.612E-01 3.506E-01 3.608E-01 3.709E-01 3.794E-01 3.970E-01 3.876E-01 3.901E-01 3.981E-01 4.200E-01

C09 4.264E-01 4.149E-01 3.882E-01 3.693E-01 3.487E-01 3.580E-01 3.541E-01 3.414E-01 3.606E-01 3.642E-01 3.651E-01 3.578E-01 3.623E-01 3.702E-01 3.607E-01 3.897E-01 3.963E-01 4.063E-01 4.002E-01

C10 4.422E-01 4.273E-01 3.940E-01 2.505E-01 3.577E-01 3.552E-01

3.431E-01 3.738E-01 3.960E-01 3.887E-01 3.833E-01 3.793E-01 3.839E-01 4.009E-01 3.900E-01 4.007E-01 3.538E-01 3.900E-01

1

2.543E-01

3.013E-01

15

3.013E-01

3.483E-01

84

3.483E-01

3.953E-01

63

3.953E-01

4.423E-01

14

4.423E-01

4.893E-01

8

4.893E-01

5.364E-01

0

5.364E-01

1.00E+15

FIGURE 6.4.8 Threshold voltage and mobility uniformity maps for a 1- × 1-cm 200-OTFT array using pentacene devices with a SiO2 gate dielectric.

FIGURE 6.4.11 A portion of a 3-in. diagonal 64 × 128 color AMLCD using pentacene OTFTs on glass from ERSO/ITRI. (From Ho, J.-C. et al., SID Int. Symp. Dig. Tech. Papers, 1298, 2004. Copyright Society for Information Display. With permission.)