INtroduction To Thin Film Transistor PDF

S. D.
Brotherton
Introduction
to Thin Film
Transistors
Physics and Technology of TFTs
Introduction to Thin Film Transistors
S. D. Brotherton
Introduction to Thin Film

Transistors
Physics and Technology of TFTs
123
S. D. Brotherton
TFT Consultant
Forest Row, E Sussex
UK
ISBN 978-3-319-00001-5 ISBN 978-3-319-00002-2 (eBook)

DOI 10.1007/978-3-319-00002-2
Springer Cham Heidelberg New York Dordrecht London
Library of Congress Control Number: 2012956303
Ó Springer International Publishing Switzerland 2013

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of
the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or
information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed. Exempted from this legal reservation are brief
excerpts in connection with reviews or scholarly analysis or material supplied specifically for the
purpose of being entered and executed on a computer system, for exclusive use by the purchaser of the
work. Duplication of this publication or parts thereof is permitted only under the provisions of
the Copyright Law of the Publisher’s location, in its current version, and permission for use must always
be obtained from Springer. Permissions for use may be obtained through RightsLink at the Copyright
Clearance Center. Violations are liable to prosecution under the respective Copyright Law.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt
from the relevant protective laws and regulations and therefore free for general use.
While the advice and information in this book are believed to be true and accurate at the date of
publication, neither the authors nor the editors nor the publisher can accept any legal responsibility for
any errors or omissions that may be made. The publisher makes no warranty, express or implied, with
respect to the material contained herein.
Printed on acid-free paper
Springer is part of Springer Science?Business Media (www.springer.com)

Preface
Thin film transistors, TFTs, are now fundamental electronic components in

virtually all consumer and professional display products, from smart phones to
large diagonal, flat panel TVs. The commercial application of TFTs has been
driven by the use of hydrogenated amorphous silicon, a-Si:H, which is the
workhorse of the industry. However, a greater range of TFT materials, such as
polycrystalline silicon, amorphous oxide semiconductors and organic semicon-
ductors, are now commanding interest, and this book has been written as an
accessible introduction to the field. Whilst there are a number excellent books
aimed at the specialist working with these individual materials, this book was
conceived to meet the needs of a less specialist audience. The book was written for
a target audience including undergraduate and postgraduate students taking dis-
play-related courses, postgraduate and new researchers to the field, engineers
supplying equipment to the flat panel display industry, researchers in one specialist
TFT field looking for an overview of another and technical managers within the
FPD industry.
Whilst some familiarity with the background physics of semiconductor device
operation has been assumed, the book is intended to be self-contained, with
introductory chapters covering the device physics concepts needed to follow the
later chapters on the TFTs themselves. These introductory chapters deal with band
bending effects at semiconductor surfaces, electron-hole pair generation and
recombination and the operation of classical MOSFET devices. A further chapter
covers the operation of active matrix displays.
This book has grown principally out of a series of lectures on TFT technologies
given to an Italian research consortium, and which incorporated material from
specialist lectures on active matrix displays and TFTs presented in Masters courses
at Southampton and Dundee Universities.
It is a pleasure to acknowledge the contribution made by the referees with their
insightful and helpful comments, and these have undoubtedly improved the book.
In addition, colleagues and other contacts within the research community have
generously responded to my requests for background information and access to
published material. These have included Dr. G. Fortunato (CNR-IMM, Rome),
v
vi Preface
Prof. G. J. Parker (Southampton University), Dr. J. R. Ayres, Dr. N. D. Young and

Dr. M. J. Trainor (Philips Research), Prof. J. M. Shannon and Dr. R. Sporea
(Sussex University), Dr. Wen-yi Lin (AU Optronics Corp), Prof. R. B. Wehrspohn
(Fraunhofer Institute for Mechanics of Materials), Dr. R. Paetzel (Coherent
GmbH) and Prof. M.-K. Han (Seoul National University). I am also indebted to
Tom Spicer of Springer for his help and guidance during the preparation of the
book.
Contents
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Part I Background
2 Semiconductor Device Physics for TFTs . . . . . . . . . . . . . . . . . . . 9

2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2 Semiconductor Surface Physics . . . . . . . . . . . . . . . . . . . . . . 10
2.2.1 Ideal MIS Capacitor and Surface Band Bending. . . . . 10
2.2.2 Gate Bias and Threshold Voltage . . . . . . . . . . . . . . . 17
2.2.3 Real MIS Structures . . . . . . . . . . . . . . . . . . . . . . . . 17
2.2.3.1 Work Function Differences . . . . . . . . . . . . . 18
2.2.3.2 Oxide Charges and Interface States . . . . . . . 20
2.2.4 Evaluation of Surface Potential . . . . . . . . . . . . . . . . 25
2.3 Electron-Hole Pair Generation and Recombination . . . . . . . . . 28
2.3.1 Thermal Equilibrium. . . . . . . . . . . . . . . . . . . . . . . . 29
2.3.2 Non-equilibrium, Steady State . . . . . . . . . . . . . . . . . 31
2.3.3 Generation Currents . . . . . . . . . . . . . . . . . . . . . . . . 33
2.3.4 Recombination Processes. . . . . . . . . . . . . . . . . . . . . 35
2.3.4.1 Low-Level Injection. . . . . . . . . . . . . . . . . . 36
2.3.4.2 High-Level Injection . . . . . . . . . . . . . . . . . 37
2.4 Current Flow Equations. . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
2.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
Appendix: Summary of Key Equations . . . . . . . . . . . . . . . . . . . . . . 42
A.1 Semiconductor Surface Band Bending. . . . . . . . . . . . 42
A.2 Carrier Recombination and Generation . . . . . . . . . . . 43
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
vii
viii Contents
3 Insulated Gate Field Effect Transistors, IGFETs . . . . . . . . . . . . . 45

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.2 MOSFET Operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
3.3 Current-Voltage Equations. . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.3.1 Simplified Format. . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.3.1.1 Linear Regime . . . . . . . . . . . . . . . . . . . . . 52
3.3.1.2 Saturation Regime . . . . . . . . . . . . . . . . . . . 53
3.3.2 Full MOSFET Equation . . . . . . . . . . . . . . . . . . . . . 54
3.3.2.1 Linear Regime . . . . . . . . . . . . . . . . . . . . . 54
3.3.2.2 Saturation Regime . . . . . . . . . . . . . . . . . . . 55
3.3.3 Non-ideal MOSFET Behaviour . . . . . . . . . . . . . . . . 57
3.4 Sub-Threshold Currents . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
3.5 Thin Film Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . 62
3.5.1 Threshold Voltage . . . . . . . . . . . . . . . . . . . . . . . . . 63
3.5.2 Saturation Voltage, VD(sat) . . . . . . . . . . . . . . . . . . . . 64
3.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
Appendix: Summary of Key Equations . . . . . . . . . . . . . . . . . . . . . . 66
A.1 Simplified MOSFET On-State Analysis. . . . . . . . . . . 66
A.2 Simplified MOSFET Sub-Threshold Analysis . . . . . . 67
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
4 Active Matrix Flat Panel Displays . . . . . . . . . . . . . . . . . . . . . . . . 69

4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
4.2 Liquid Crystal Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
4.2.1 LC Material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
4.2.2 Twisted Nematic LC Cell Structure . . . . . . . . . . . . . 73
4.3 Active Matrix Addressing . . . . . . . . . . . . . . . . . . . . . . . . . . 76
4.4 Pixel Layout Considerations. . . . . . . . . . . . . . . . . . . . . . . . . 83
4.4.1 General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
4.4.2 Performance Artefacts . . . . . . . . . . . . . . . . . . . . . . . 85
4.4.2.1 Voltage Kick-Back . . . . . . . . . . . . . . . . . . 85
4.4.2.2 Vertical Cross-Talk . . . . . . . . . . . . . . . . . . 87
4.4.2.3 Row Resistance Effects . . . . . . . . . . . . . . . 87
4.5 AMLCD Fabrication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
4.5.1 TFT Plate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
4.5.2 Colour Filter, CF, Plate . . . . . . . . . . . . . . . . . . . . . . 91
4.5.3 LC Cell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
4.5.4 Display Module . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
4.6 Other Display Technologies . . . . . . . . . . . . . . . . . . . . . . . . . 93
4.6.1 Active Matrix Electrophoretic Displays . . . . . . . . . . . 93
4.6.2 Active Matrix Organic Light Emitting Diode
Displays . . . . . . . . . . . . . . . . . . . . . . . . . . ...... 96
4.6.2.1 OLED Operation . . . . . . . . . . . . . . ...... 96
4.6.2.2 AMOLED Pixels . . . . . . . . . . . . . . ...... 99
Contents ix
4.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
Part II TFTs
5 Hydrogenated Amorphous Silicon TFT Technology

and Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
5.2 a-Si:H Material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
5.3 a-Si:H TFT Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
5.3.1 Back-Channel-Etched TFT Fabrication . . . . . . . . . . . 114
5.3.2 Etch-Stop TFT Fabrication . . . . . . . . . . . . . . . . . . . 115
5.4 TFT Layout Considerations . . . . . . . . . . . . . . . . . . . . . . . . . 117
5.4.1 Photolithography Process. . . . . . . . . . . . . . . . . . . . . 117
5.4.2 TFT Layout Issues . . . . . . . . . . . . . . . . . . . . . . . . . 118
5.5 Plasma Enhanced Chemical Vapour Deposition, PECVD . . . . 119
5.5.1 Undoped a-Si:H . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
5.5.2 Doped a-Si:H . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
5.5.3 a-SiNx:H Gate Insulator. . . . . . . . . . . . . . . . . . . . . . 130
5.6 Novel a-Si:H TFTs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
5.6.1 Self-Aligned TFTs . . . . . . . . . . . . . . . . . . . . . . . . . 131
5.6.2 Short Channel TFTs . . . . . . . . . . . . . . . . . . . . . . . . 133
5.6.3 Hydrogen-Diluted TFT Depositions . . . . . . . . . . . . . 135
5.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
6 Hydrogenated Amorphous Silicon TFT Performance . . . . . . . . . . 141

6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
6.2 Defect Structure and a-Si:H Density of States . . . . . . . . . . . . 142
6.2.1 Basic Material Properties. . . . . . . . . . . . . . . . . . . . . 142
6.2.2 a-Si:H Density of States, DOS . . . . . . . . . . . . . . . . . 149
6.2.3 Band Bending and Surface Space Charge . . . . . . . . . 150
6.2.4 Field Effect Mobility . . . . . . . . . . . . . . . . . . . . . . . 155
6.2.5 Equilibration in Thin Films . . . . . . . . . . . . . . . . . . . 158
6.2.5.1 Gate Dielectric . . . . . . . . . . . . . . . . . . . . . 160
6.2.5.2 n- and p-channel TFTs . . . . . . . . . . . . . . . . 162
6.3 TFT Characteristics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
6.3.1 On-State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
6.3.2 Off-State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
6.4 Bias-Stress Instability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
6.4.1 Gate Bias. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
6.4.2 Pulsed Bias Stress. . . . . . . . . . . . . . . . . . . . . . . . . . 174
6.4.3 Gate and Drain Bias Stress Effects . . . . . . . . . . . . . . 176
x Contents
6.5 Other Meta-Stability Effects . . . . . . . . . . . . . . . . . . . . . . . . 178

6.5.1 Staebler Wronski Effect . . . . . . . . . . . . . . . . . . . . . 178
6.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
7 Poly-Si TFT Technology and Architecture. . . . . . . . . . . . . . . . . . 185

7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
7.2 Poly-Si Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
7.2.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
7.2.2 Excimer Laser Crystallisation. . . . . . . . . . . . . . . . . . 190
7.2.2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . 190
7.2.2.2 Crystallisation Process . . . . . . . . . . . . . . . . 190
7.2.2.3 TFT Crystallisation . . . . . . . . . . . . . . . . . . 196
7.2.2.4 ELA Process Control Issues . . . . . . . . . . . . 200
7.2.3 Other Laser Techniques. . . . . . . . . . . . . . . . . . . . . . 203
7.2.4 Metal Induced Crystallisation. . . . . . . . . . . . . . . . . . 204
7.2.4.1 Ni Mediated Crystallisation of a-Si . . . . . . . 204
7.2.4.2 SMC Poly-Si TFTs . . . . . . . . . . . . . . . . . . 208
7.3 Gate Dielectrics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
7.3.1 Silicon Dioxide . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
7.3.2 Alternative Dielectrics. . . . . . . . . . . . . . . . . . . . . . . 217
7.4 Poly-Si TFT Architecture and Fabrication . . . . . . . . . . . . . . . 218
7.4.1 Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
7.4.1.1 Self Aligned Source and Drain Doping . . . . 219
7.4.1.2 Drain Field Relief . . . . . . . . . . . . . . . . . . . 222
7.4.1.3 Other TFT Architectures . . . . . . . . . . . . . . 224
7.4.2 Fabrication Process . . . . . . . . . . . . . . . . . . . . . . . . . 224
7.5 Advanced Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
7.5.1 Large Grain Poly-Si . . . . . . . . . . . . . . . . . . . . . . . . 227
7.5.2 Modified Excimer Laser Crystallisation. . . . . . . . . . . 227
7.5.2.1 Sequential Lateral Solidification . . . . . . . . . 230
7.5.3 Green Laser Crystallisation . . . . . . . . . . . . . . . . . . . 234
7.5.3.1 Pulsed Nd:YAG Lasers . . . . . . . . . . . . . . . 234
7.5.3.2 CW Nd:YVO4 Lasers. . . . . . . . . . . . . . . . . 237
7.5.4 Comparison of Large Grain Crystallisation
Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
7.6 Poly-Si Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
7.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246
8 Poly-Si TFT Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253

8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253
8.2 Electrical Conduction in Poly-Si. . . . . . . . . . . . . . . . . . . . . . 254
8.2.1 Analytical Bulk Conduction Model. . . . . . . . . . . . . . 254
Contents xi
8.2.2 Analytical Model of TFT Conduction . . . . . . . . . . . . 260

8.2.3 Limitations of Analytical Model. . . . . . . . . . . . . . . . 261
8.3 Poly-Si Density of States, DOS . . . . . . . . . . . . . . . . . . . . . . 262
8.4 TFT Off-State Currents . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266
8.5 Performance Artefacts and Drain Field . . . . . . . . . . . . . . . . . 272
8.5.1 Electrostatic Drain Field, F . . . . . . . . . . . . . . . . . . . 273
8.5.2 Hot Carrier Damage and LDD . . . . . . . . . . . . . . . . . 274
8.5.3 Field-Enhanced Leakage Currents. . . . . . . . . . . . . . . 280
8.6 Other Bias-Stress Instabilities. . . . . . . . . . . . . . . . . . . . . . . . 282
8.6.1 Gate Bias Stress . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
8.6.1.1 Ionic Instability . . . . . . . . . . . . . . . . . . . . . 283
8.6.1.2 Negative Bias-Temperature Instability . . . . . 283
8.6.2 Combined Gate and Drain Bias Stress. . . . . . . . . . . . 285
8.7 Short Channel Effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 288
8.7.1 Parasitic Resistance Effects . . . . . . . . . . . . . . . . . . . 290
8.7.2 Floating Body Effects . . . . . . . . . . . . . . . . . . . . . . . 292
8.7.2.1 Kink Effect . . . . . . . . . . . . . . . . . . . . . . . . 292
8.7.2.2 Sub-Threshold and Threshold Voltage
Effects . . . . . . . . . . . . . . . . . . . . . . ..... 294
8.8 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 296
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 297
9 Transparent Amorphous Oxide Semiconductor TFTs . . . . . . . . . 301

9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301
9.2 Material Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302
9.3 TFT Architecture and Fabrication. . . . . . . . . . . . . . . . . . . . . 305
9.3.1 Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 306
9.3.2 Fabrication Processes . . . . . . . . . . . . . . . . . . . . . . . 308
9.3.2.1 a-IGZO Layer . . . . . . . . . . . . . . . . . . . . . . 308
9.3.2.2 Gate Dielectric . . . . . . . . . . . . . . . . . . . . . 308
9.3.2.3 Post Deposition Annealing . . . . . . . . . . . . . 310
9.3.2.4 Metallisation . . . . . . . . . . . . . . . . . . . . . . . 311
9.3.2.5 Process Flow. . . . . . . . . . . . . . . . . . . . . . . 311
9.4 a-IGZO TFT Performance . . . . . . . . . . . . . . . . . . . . . . . . . . 312
9.4.1 n-Channel Characteristics . . . . . . . . . . . . . . . . . . . . 312
9.4.2 Conduction Process and Density of States
Distribution, DOS . . . . . . . . . . . . . . . . . . . . . . . . . . 318
9.4.2.1 Conduction Process . . . . . . . . . . . . . . . . . . 318
9.4.2.2 Density of States Distribution, DOS. . . . . . . 320
9.4.3 Bias Stress Instability . . . . . . . . . . . . . . . . . . . . . . . 325
9.4.3.1 Gate Bias Instability . . . . . . . . . . . . . . . . . 326
9.4.3.2 Negative Bias Illumination Stress (NBIS). . . 328
9.4.3.3 Stability Considerations for Active
Matrix Addressing TFTs. . . . . . . . . . . . . .. 330
xii Contents
9.5 AOS TFT Circuits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 332

9.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 334
10 Organic TFTs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 339

10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 339
10.2 Background and Materials . . . . . . . . . . . . . . . . . . . . . . . . . . 342
10.2.1 Conjugated Molecular Systems. . . . . . . . . . . . . . . . . 342
10.2.2 Molecular Bonding . . . . . . . . . . . . . . . . . . . . . . . . . 344
10.2.3 Molecular Organisation . . . . . . . . . . . . . . . . . . . . . . 344
10.2.4 Metal/Organic Contacts . . . . . . . . . . . . . . . . . . . . . . 347
10.2.5 Carrier Transport . . . . . . . . . . . . . . . . . . . . . . . . . . 350
10.3 OTFT Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351
10.4 Materials and Fabrication Processes . . . . . . . . . . . . . . . . . . . 354
10.4.1 Solution Processing Techniques . . . . . . . . . . . . . . . . 355
10.4.1.1 Spin-coating . . . . . . . . . . . . . . . . . . . . . . . 355
10.4.1.2 Drop-casting . . . . . . . . . . . . . . . . . . . . . . . 355
10.4.1.3 Zone-casting . . . . . . . . . . . . . . . . . . . . . . . 356
10.4.1.4 Printing . . . . . . . . . . . . . . . . . . . . . . . . . . 357
10.4.2 Organic Semiconductor Layers for p-Channel
TFTs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...... 358
10.4.2.1 Vacuum Thermal Evaporation. . . . . ...... 358
10.4.2.2 Solution Processing . . . . . . . . . . . . ...... 360
10.4.3 Organic Semiconductor Layers for n-Channel
TFTs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363
10.4.4 Gate Dielectric . . . . . . . . . . . . . . . . . . . . . . . . . . . . 368
10.4.4.1 Inorganic Dielectrics . . . . . . . . . . . . . . . . . 369
10.4.4.2 Organic Dielectrics . . . . . . . . . . . . . . . . . . 370
10.4.4.3 Self-Assembled Monolayers . . . . . . . . . . . . 373
10.4.5 Metals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 373
10.4.6 Process Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 374
10.4.7 Novel Processing . . . . . . . . . . . . . . . . . . . . . . . . . . 376
10.5 OTFT Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 378
10.5.1 General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 378
10.5.2 Contact Effects. . . . . . . . . . . . . . . . . . . . . . . . . . . . 381
10.5.3 Contact Architecture . . . . . . . . . . . . . . . . . . . . . . . . 386
10.6 Instability Effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 389
10.6.1 Air-Instability. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 389
10.6.2 Gate Bias Stress Instability . . . . . . . . . . . . . . . . . . . 390
10.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 395
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 396
Contents xiii
Part III Novel Substrates and Devices
11 TFTs on Flexible Substrates . . . . . . . . . . . . . . . . . . . . . . . . . . . . 407

11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 407
11.2 Substrate Handling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 409
11.3 Substrate Bending. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413
11.4 a-Si:H TFTs on Flexible Substrates . . . . . . . . . . . . . . . . . . . 416
11.4.1 a-Si:H Fabrication Processes . . . . . . . . . . . . . . . . . . 416
11.4.1.1 Direct Processing on Plastic . . . . . . . . . . . . 417
11.4.1.2 Steel Foil Substrates . . . . . . . . . . . . . . . . . 422
11.4.1.3 Carrier Plate Technology . . . . . . . . . . . . . . 423
11.4.1.4 Roll-to-Roll (R2R) Processing. . . . . . . . . . . 426
11.4.2 Uniaxial Strain Effects on a-SiH TFTs . . . . . . . . . . . 428
11.5 Poly-Si TFTs on Flexible Substrates . . . . . . . . . . . . . . . . . . . 430
11.5.1.1 Direct Fabrication on Plastic Substrates . . . . 431
11.5.1.2 Fabrication on Steel Foils . . . . . . . . . . . . . . 433
11.5.1.3 Transfer Processes . . . . . . . . . . . . . . . . . . . 435
11.5.2 Uniaxial Strain Effects on Poly-Si TFTs . . . . . . . . . . 438
11.6 Organic TFTs on Flexible Substrates . . . . . . . . . . . . . . . . . . 440
11.6.1.1 Direct Processing. . . . . . . . . . . . . . . . . . . . 441
11.6.1.2 Carrier Plate Processing . . . . . . . . . . . . . . . 442
11.6.2 Uniaxial Strain Effects on Organic TFTs. . . . . . . . . . 442
11.7 Plastic Substrate Issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . 444
11.8 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 446
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 447
12 Source-Gated Transistors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 453

12.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 453
12.2 Schottky Barrier Diodes . . . . . . . . . . . . . . . . . . . . . . . . . . . 455
12.3 SGT Structure and Operation . . . . . . . . . . . . . . . . . . . . . . . . 459
12.3.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 459
12.3.2 SGT1 Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 460
12.3.3 SGT2 Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 467
12.4 Fabrication Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 468
12.5 Comparison of SGT and FET Characteristics . . . . . . . . . . . . . 470
12.6 Gate/Source-Barrier Interactions in OTFTs . . . . . . . . . . . . . . 473
12.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 478
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 479
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 481
Biographical Information
Dr. Stan D. Brotherton started his research career at the GEC Hirst Research
Laboratory, England, before taking a post-doctoral Research Fellowship at
Southampton University in 1971. From there he moved to the Philips Research
Laboratory, Redhill, England, where he was a Senior Principal Scientist, and he
now works as an independent TFT Consultant.
He has led a wide range of research projects investigating semiconductor devices,
and related materials issues, particularly the influence of deep level defects on the
performance of Si devices. The devices studied have included MOSFETs and
CCDs, power devices, and IR imaging devices. His most recent field of activity
was thin film transistors, within which he initiated the Philips research programme
on poly-Si TFTs. Activity within this field has continued with consultancy
contracts from a number of international organisations.
He has published *120 papers on the physics and technology of silicon devices,
and in 1989 was awarded a DSc by London University for published work on deep
level defects in silicon. Amongst the later publications, *65 have been on poly-Si
TFTs. He also contributed a chapter on this topic to the Springer Handbook of
Visual Display Technology, and was a member of the Editorial Board for that
book. He has presented numerous invited and contributed papers at major
international conferences, and has been a contributor of specialist TFT lectures on
the DisplayMasters course at Dundee University.
xv
Chapter 1
Introduction
This book reviews the operation, technology and application of the main classes of
thin film transistor (TFT), which are of current interest in the field of large area
electronics. The major application area is flat panel, active matrix liquid crystal
displays, AMLCDs, and, with the range of screen diagonals currently available,
this is the most ubiquitous and successful display technology to date. Screen sizes
from 1 inch to more than 100 inches dominate applications in most areas of life,
from small, portable mobile phone displays, through medium-sized tablets, net-
books and lap-tops, to large-screen monitors and HD televisions.
In the 1960s, there was early work into a variety TFT materials and applica-
tions, and an engaging and personal account of that work has been given by Brody
[1, 2], who was a pioneer of both TFTs and active matrix displays. However, the
present interest in thin film transistors can be dated back to the research of Spear
and Le Comber [3, 4] in the mid-1970s. They showed that glow discharge
hydrogenated amorphous silicon, a-Si:H, had a low enough trap state density that it
could be doped, and used in thin film transistor structures. This stimulated world-
wide research and development of the a-Si:H TFT, and led to its application in
active matrix liquid crystal displays.
From the first, small scale AMLCD production facilities in the late 1980s,
processing A4-sized glass plates, to the current 10th generation plant, processing
2.85 9 3.05 m2 glass plates, the panel industry has grown, on average, by more
than 10 % per year to a turn-over of approximately $B86 in late 2011 [5]. Whilst
AMLCDs continue to dominate the current product range in terms of volume,
other display media are now appearing. These are electro-phoretic displays, EPDs,
for e-readers, and organic light emitting diode, OLED, displays for smart phones,
and which are predicted to be of longer-term interest for TV.
To cover this field, and to make it accessible to students and new researchers,
the layout of this book is organised into three sections: Part I—Background, Part
II—TFTs, and Part III—Novel Substrates and Devices. The contents of these
sections progress from basic semiconductor physics to speculative new materials
and device structures. In the introductory Part I, Chaps. 2 and 3 contain the
background device physics to TFT operation, and Chap. 4 describes the operation
S. D. Brotherton, Introduction to Thin Film Transistors, 1

DOI: 10.1007/978-3-319-00002-2_1, Springer International Publishing Switzerland 2013
2 1 Introduction
of active matrix displays. Part II, containing Chaps. 5–10, describes the technol-
ogy and device physics of the major classes of TFT, which are of current interest in
the field of large area electronics. These are:
• Hydrogenated amorphous silicon (a-Si:H), which forms the basis of the AM-
LCD industry,
• Poly-crystalline silicon (poly-Si), which facilitates the integration of electronic
circuits into flat panel displays. Cost considerations limit this application to
small diagonal LCDs, whilst the high drive current of these devices favours their
application to small AMOLED displays,
• Transparent amorphous oxide semiconductors, of which indium-gallium-zinc
oxide is the most studied example. These newer materials are attracting much
interest as they have considerably higher carrier mobility than a-Si:H, but, due
to their amorphous nature, are expected to have comparable excellent unifor-
mity. Their projected applications are to large diagonal, high resolution AM-
LCD TVs [6], and, in the longer term, to AMOLED TVs [6],
• Organic semiconductors. In the preferred implementation, these TFTs contain
organic materials for both the semiconductor and the gate dielectric layers. They
are currently low carrier mobility devices, but can be produced by low tem-
perature solution-processing, including printing, and are seen as a cost-effective
route to flexible electronics on polymer substrates.
The majority of TFT production is based upon the use of rigid, glass substrates,
and, in Part III, Chap. 11 reviews the properties and constraints of polymer sub-
strates, and gives an overview of the strategies used to fabricate TFTs on flexible
materials. Finally, Chap. 12 introduces a new type of TFT, the source-gated
transistor, SGT. This has markedly different characteristics from conventional
TFTs, and may be an interesting vehicle for achieving improved performance from
low-mobility disordered materials, including organic semiconductors.
Irrespective of the display medium (whether it is LCD, EPD or OLED), the
operation of the display itself is dependent upon the presence of an active matrix of
TFTs to address and drive the individual pixels in the display. The TFTs are field
effect transistors, FETs, whose operation is akin to the semiconductor industry
metal–oxide–semiconductor transistor, the MOSFET. Hence, as an introduction to
the TFT field, Chaps. 2 and 3 deal with the fundamentals of FET operation. In
particular, Chap. 2 introduces some basic semiconductor physics topics, which
underpin the operation of TFTs. This includes the bending of the semiconductor
energy bands, and the change in free carrier concentration at the semiconductor
surface, in response to an orthogonally applied electric field. The relationships are
derived analytically for a single crystal substrate, and these expressions are then
used as reference points in later chapters to clarify the changes needed to describe
TFT behaviour. The other topic covered is electron–hole pair recombination and
generation, and these are processes governing non-thermal equilibrium effects in
TFTs, such as photoconductivity and leakage currents. Whilst most practical TFT
analysis makes use of 2-D numerical modelling packages, an analytical approach
is used here in order to illustrate the underlying physics. However, the relevant
1 Introduction 3
equations for device simulation are summarised, so that the reader has a better
appreciation of the background to the many publications reporting 2-D device
simulation results.
Chapter 3 takes a similar analytical approach in describing the operation of the
simple, long-channel MOSFET, with the aim of demonstrating the basic physics
behind its operation. Again, these analytical concepts are used in the later chapters
dealing with TFTs, as the fundamental physics are the same, and the equations are
universally used to extract key performance parameters from all classes of TFT.
The equations have the merit of simplicity and ease of use, and their limitations are
discussed in the individual TFT chapters.
Chapter 4 describes the basic operation of active matrix displays, and, in par-
ticular, the role of the TFT as a switch and storage element within each pixel of the
display. There is also a discussion of pixel layout considerations, and of their
impact upon display performance. Although the emphasis in this chapter is on
liquid crystal displays, an overview of the operation of electro-phoretic and OLED
displays is also included.
The most widely employed transistor throughout the industry, and across the
application range, is the hydrogenated amorphous silicon, a-Si:H, TFT, whose
technology and performance details are described in Chaps. 5 and 6, respectively.
Chapter 5 describes the architecture and fabrication of a-Si:H TFTs, whilst
Chap. 6 deals with some of the performance issues of these devices. Of these, one
extensively studied topic is gate bias stress instability, which is a fundamental
effect related to the meta-stability of the defect concentration within the material.
The chapter also includes a brief discussion of two other meta-stability effects in a-
Si:H, which are associated with impurity doping and optical illumination.
At the time of writing, the only other TFT material in mass production is poly-
Si, and Chaps. 7 and 8 deal with the technology and performance of these devices,
respectively. In Chap. 7, the fabrication processes for poly-Si are reviewed, and
the currently preferred process of laser crystallisation is discussed in more detail.
These devices have a coplanar, self-aligned architecture, which is quite different
from the other TFTs, and is much more similar to single crystal silicon devices.
The characteristics of poly-Si TFTs are discussed in Chap. 8, and, indeed, the
performance of high quality poly-Si devices is also closer to crystalline Si
MOSFETs than to the alternative, contemporary TFTs. This chapter includes a
review of device performance issues, many of which are related to the high
electrostatic field at the drain junction, resulting in field-enhanced leakage currents
and hot carrier instabilities.
Chapters 9 and 10 cover the emerging TFT technologies of transparent amor-
phous oxide semiconductors, and organic semiconductors, respectively. These are
not yet in mass production, although initial products, employing each of these
device types, have been predicted to appear in 2012 [6–8]. Both chapters discuss
the fabrication and performance of devices made in these two types of material,
although the content in Chap. 10, on organic materials, is notably different from
the preceding chapters. Whereas the other chapters have focussed on one material,
there is a plethora of different organic compounds, which have been engineered for
4 1 Introduction
application to both the semiconductor layer and to the gate dielectric layer. In
addition to the variety of materials, in many cases there is also a variety of
alternative deposition procedures, ranging from vacuum deposition to solution
processing, and printing. At the moment, there is not a consensus on the most
favoured organic semiconductor materials, and the situation is further complicated
by a trade-off between device performance and the ease of material deposition.
Chapter 10 contains an overview of the materials of current interest, together with
their different deposition procedures, and reviews the performance issues of these
devices.
Chapter 11 discusses the challenges in changing from a rigid, glass substrate
technology to one utilising a flexible substrate, which, ideally, would be a low cost
polymer material. These materials present many handling issues, including
reduced temperature processing, dimensional instability, and have coefficients of
thermal expansion which are a poor match to the materials used in the inorganic
TFT technologies. However, given the maturity of the a-Si:H and poly-Si fabri-
cation technologies, a variety of routes has been explored in order to integrate
them into a flexible substrate process. These approaches, including the impact of
substrate bending on device characteristics, are reviewed for both the inorganic
and the organic TFT technologies.
A new device structure, the source-gated transistor, SGT, is presented in
Chap. 12. This is a thin-film FET, having an inverted staggered TFT architecture,
but with an injection barrier at the source junction. The height of this barrier is
controlled by the gate and drain biases, and the reported device characteristics are
remarkably different from those of conventional TFTs. In particular, the SGT has a
reduced saturation voltage, higher output impedance, reduced short channel effects,
and the drain current is insensitive to modest variations in the source-drain sepa-
ration. The physics of its operation are presented, and its behaviour is compared
with staggered organic TFTs, which have the closest equivalent architecture. As a
relatively new device, the detailed understanding of its operation was still devel-
oping at the time of writing, and further refinements may emerge in due course.
The TFT field continues to be characterised by burgeoning activities in both
research and product development. New display media, such as OLEDs, are
expected to become more widespread, and the pixels may well be driven by new
TFT technologies, such as amorphous oxide semiconductors [8]. In addition, new
products, such as flexible displays, are forecast to appear in the near-future, and
organic TFTs are expected to play a major role in this area. These developments
will broaden and extend the current technology range, and will further enhance the
ubiquity of flat panel display products.
This introductory book on TFTs is complementary to the established books in
the field, where a particular TFT technology (which is covered by just one or two
chapters here) warrants a book in itself, and is aimed at the specialists in that field.
In contrast, this book is designed to be an accessible introduction to TFTs, and is
aimed at undergraduate teaching courses, post-graduate and new researchers to the
1 Introduction 5
field, engineers supplying equipment to the flat panel display industry, researchers
in one specialist TFT field looking for an overview of another, and technical
managers within the FPD industry.
References
1. Brody TP (1984) The thin film transistor—a late flowering bloom. IEEE Trans Electron Dev
ED-31(11):1614–1628
2. Brody TP (1996) The birth and early childhood of active matrix—a personal memoir. J SID
4(3):113–127
3. Spear WE, Le Comber PG (1972) Investigation of the localised state distribution in amorphous
Si films. J Non-Cryst Solids 8–10:727–738
4. Spear WE, Le Comber PG (1975) Substitutional doping of amorphous silicon. Solid State
Commun 17(9):1193–1196
5. http://www.displaysearch.com/cps/rde/xchg/displaysearch/hs.xsl/resources_paneltrack.asp
(Accessed Jan 2012)
6. Semenza P (2011) Large TFT-LCD panels shift into high resolution. SID Inf Disp 9(11):30–32
7. http://www.plasticlogic.com/news/pr_education_announce_sep122011.php (Accessed Sept
2011)
8. http://www.displaysearch.com/cps/rde/xchg/displaysearch/hs.xsl/
120106_oled_display_technology_moving_to_compete_in_the_tv_market.asp (Accessed Jan
2012)
Part I
Background
Chapter 2
Semiconductor Device Physics for TFTs
Abstract Two device physics topics are discussed in this chapter, namely, surface
band bending and surface charges in the metal–insulator–semiconductor, MIS,
structure, and electron-hole pair recombination/generation processes. The treat-
ment of the MIS structure covers the relationship between the voltage on the metal
gate, the induced surface charge in the semiconductor and the resulting surface
potential. This is treated analytically, using single crystal equations, and the
relationships are fundamental to the understanding of IGFET operation. Equally,
the concepts are widely employed in analysing TFT behaviour. The electron-hole
pair generation process underlies the leakage current behaviour of many semi-
conductor devices, and can be applied to the analysis of TFT off-state behaviour.
The recombination process determines steady state carrier concentrations under
injection conditions, such as optical illumination. Finally, there is a brief discus-
sion of carrier flow in semiconductor devices, including the equations used in
numerical simulation packages.
2.1 Introduction
In this chapter, device physics topics are introduced, which are relevant to TFTs,
and for following the discussion in later chapters. The emphasis is on background
understanding of basic device physics principles, and an analytical approach is
followed, using single crystal semiconductor equations. These concepts and
equations are modified in later chapters for the more complex situation of the non-
single crystal TFT materials.
The first topic presented deals with semiconductor surface physics, focussing on
band bending and surface charge in the metal–insulator–semiconductor, MIS,
system. This underlies the relationship between the voltage on the metal gate and
the induced charge in the semiconductor surface, and introduces the concepts of
the flat band voltage and the threshold voltage for surface inversion. The most
widely studied semiconductor/insulator interface is the c-Si/SiO2 interface, and the
numerical examples given are based upon this system, although the equations

DOI: 10.1007/978-3-319-00002-2_2, Ó Springer International Publishing Switzerland 2013
10 2 Semiconductor Device Physics for TFTs
themselves are quite general. In Chap. 6, these surface space charge equations are
re-formulated to include the U-shaped density of states functions found in a-Si:H
and other TFT materials, and the influence of these extra states is compared to the
simpler situation discussed in this chapter.
The second topic is electron-hole pair generation and recombination. These are
basic processes underlying both leakage current effects, and steady state carrier
densities in devices under injection conditions, such as optical illumination.
Finally, there is a brief discussion of carrier flow and the coupled field/space
charge equations, which need to be solved in order to calculate the current–voltage
characteristics of a device.
Much current research into device behaviour makes use of commercial device
simulators to solve these latter equations. These simulation packages give a deeper
insight into device behaviour by relating its current–voltage characteristics to
internal field and carrier distributions. However, in published work, the funda-
mental equations are rarely listed. In the final Sect. 2.4, of this chapter, those
equations are presented, and they are built up from the material in the preceding
sections of the chapter (but, the numerical techniques used for the solution of the
equations go beyond the scope of this book).
As is apparent, the range of device physics topics covered here is limited, and,
for a broader coverage of this field, there are many excellent books available, such
as Sze and Ng [1]. Examples of device simulation packages can be found in Refs.
[2, 3], as well as a discussion of the numerical solution techniques in [2].
Some of the key simplified equations, from Sects. 2.2 and 2.3, are listed in the
Appendix. These are useful for cross-reference purposes in later chapters, and also
for direct, analytical calculations.
2.2 Semiconductor Surface Physics
2.2.1 Ideal MIS Capacitor and Surface Band Bending
Figure 2.1a shows the band diagram of an ideal MIS capacitor, on a p-type sub-
strate, in which the Fermi levels in the metal and in the semiconductor perfectly
align, such that there is no induced band bending within the structure (In Sect.
2.2.3, a more realistic situation will be discussed, in which there are work function
differences between the metal and the semiconductor, and the semiconductor
interface contains interface trapping states). In the treatment below, the following
conventions will be used: the Fermi potential, VF, will be measured from the bulk
intrinsic level, Ei, and will be taken as positive beneath Ei and negative above it.
Similarly, the band bending, Vs, will be measured from the bulk intrinsic level, and
the polarity convention will be the same as used for VF.
When a positive charge, QG, is placed on the metal gate, it will induce an equal
and opposite negative charge in the semiconductor, Qs, and this negative charge
2.2 Semiconductor Surface Physics 11
will consist of an increase in the electron density and a decrease in the free hole
density, thereby, leaving behind immobile, ionised acceptor centres, Na. In order to
accommodate these changes in free carrier density, the bands within the semi-
conductor will have to bend downwards near its surface, as shown in Fig. 2.1b. It
will also be seen that the positive charge on the gate results from a positive bias
being applied to the gate relative to the semiconductor. Following the convention
discussed above, the Fermi level in the metal is moved downwards in response to
the positive gate bias, VG, and the semiconductor surface potential is +Vs. The
situation shown in this diagram is for a small positive potential on the gate, such
that the surface electron density, ns, is small compared with Na, and the surface is
said to be depleted (of free holes). For a larger positive gate bias, the situation
shown in Fig. 2.1c occurs. In this case, there is a corresponding increase in band
bending, Vs, and the free electron concentration at the surface is larger than Na: the
surface is now said to be inverted. Between these two situations, when the band
bending, Vs = VF, the surface will be intrinsic, and ns = ps = ni. Further positive
Fig. 2.1 Ideal metal–insulator–semiconductor, MIS, band diagrams, showing different surface
charge conditions for a p-type substrate: a flat bands, b hole depletion, c electron inversion, and
d hole accumulation
band bending beyond this point will lead to ns [ ps. Finally, as shown in Fig. 2.1d,
with a negative bias applied to the gate, there is an increase in positive charge
induced in the semiconductor, due to an increase in the free hole density. This is
associated with the bands bending upwards by an amount -Vs. In this case, the
surface is said to be accumulated. With an n-type substrate, the opposite situation
occurs, in that a negative gate bias will cause the surface to be depleted/inverted
rather than accumulated, and a positive bias will cause surface accumulation.
To establish the relationship between Vs, and Qs and VG, it is necessary to solve
Poisson’s equation, where:
d2 V qqðxÞ
2
¼ ð2:1Þ
dx e0 es
e0 is the permittivity of free space, es is the dielectric constant of the semicon-
ductor, and the space charge density, q(x) is:
qðxÞ ¼ pðxÞ nðxÞ Na ð2:2Þ
The free carrier densities are defined by the intrinsic carrier concentration, ni,
and the separation of the Fermi level from the intrinsic level, i.e.:
qð VF V Þ
pðxÞ ¼ ni exp
kT ð2:3Þ
qðVF V Þ
nðxÞ ¼ ni exp
kT
pðxÞ nðxÞ ¼ 2ni shqðVF VÞ=kT ð2:4Þ
Where T is the temperature, and k is Boltzmann’s constant. At the surface,
V = Vs, and in the bulk, where V = 0:
qVF qVF
p0 ¼ ni exp and n0 ¼ ni exp
kT kT
From charge neutrality,1
p0 n0 ¼ Na ¼ 2ni shqVF =kT ð2:5Þ

d2 V 2ni q qðVF VÞ qVF
2
¼ sh sh ð2:6Þ
dx e0 es kT kT
2
d dV 2
Using ddxV2 ¼ 12 dV dx , and integrating Eq. 2.6 from the bulk (V = 0, and
dV/dx = 0) to the surface (V = Vs, and dV/dx = -Fs):
1
VF = (kT/q)ln(Na/ni).
Table 2.1 Polarity relationship between band bending, surface field and space charge in an MIS
structure
Vs Fs Qs Free carrier conditions
+ve +ve -ve Reduction of holes and/or increase of electrons
-ve -ve +ve Reduction of electrons and/or increase of holes
2
dV 4ni q kT qðVF Vs Þ kT qVF qVF
Fs2 ¼ ch ch þ Vs sh ð2:7Þ
dx V¼Vs e0 es q kT q kT kT
From Gauss’ Law, the surface field, Fs, is related to the total areal charge, Qs,
contained within the surface by:
Qs ¼ e0 es Fs ð2:8Þ
Hence, the relationship between Qs and Vs is given by:
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Qs ¼ 4ni qe0 es GðVs ; VF Þ ð2:9Þ
where G(Vs, VF) is given by:

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
kT qðVF Vs Þ kT qVF qVF
GðVs ; VF Þ ¼ ch ch þ Vs sh ð2:10Þ
q kT q kT kT
Given the positive and negative values of Qs in Eq. 2.9, the appropriate signs of
Vs, Fs and Qs are given in Table 2.1, and the correct value of Qs is given by:
Vs
Qs ¼ jQs j ð2:11Þ
j Vs j
Equation 2.9 has been evaluated for a crystalline Si substrate doped with 1015
acceptors/cm3, and the key features in the relationship between Qs and Vs are
shown in Fig. 2.2.2 The three previously discussed regimes of accumulation,
depletion and inversion are indicated, and, for accumulation and inversion, Qs
increases exponentially with Vs. In these two regimes, the main contributors to Qs
are the free carriers, so that the hole and electron densities increase exponentially
with Vs in accumulation and inversion, respectively. In the third regime of
majority carrier depletion, Qs increases with HVs, and extrapolating the inversion
arm of the curve back into this regime shows that the ionised acceptor charge
dominates Qs. Hence, in the three regimes, either free carriers or fixed space
charge constitutes the major part of Qs. As will be seen in Chap. 6, this is not
necessarily the case with TFT materials, in which trapping states in the forbidden
2
Note that in evaluating Eq. 2.10, if the units of Boltzmann’s constant, k, are in eV/K, then q has
the value of unity. If SI units are used, then q has its usual value of 1.602 9 10-19 C.
Fig. 2.2 Calculated variation

of surface charge density with
surface potential (substrate
doping density is
1 9 1015 acceptors/cm3)
band gap can continue to make a major contribution to Qs even when the surface
has enough free carriers to support substantial conduction.
For the calculation in Fig. 2.2, the value of qVF was 0.288 eV below Ei, and the
electron concentration in inversion starts to dominate Qs at Vs [ *2VF. In fact,
the conventional definition of the threshold for strong surface inversion is at
Vs = 2VF, where the volume concentration of free electrons at the surface, ns, is
equal to the volume concentration of acceptors, Na. When Vs = VF, the surface is
intrinsic (ns = ps = ni), and the band bending regime beyond this, and up to strong
inversion, is referred to as weak inversion (ps \ ns \ Na).
In the depletion/inversion regime, the band bending, Vs, is positive, and for Vs,
and VF [ kT/q, i.e., more than kT from the flat band position, Eq. 2.9 can be
simplified to:
0:5
pffiffiffiffiffiffiffiffiffiffiffiffi kT qðVs VF Þ
Qs ffi 2qe0 es ni exp þ Na Vs ð2:12Þ
q kT
The first term in the brackets relates to the free electron concentration, and the
second term to the ionised acceptor space charge density, and the dependence of
those terms on Vs shows the exponential and quadratic forms, respectively, dis-
cussed above. This simplified expression, with the physically obvious terms,
provides a good approximation to the full Eq. 2.9, as seen by the calculations in
Fig. 2.3. The calculations used two different substrate doping levels, and, for the
more heavily doped substrate, with 1016 acceptors/cm3, the value of qVF is
0.348 eV below Ei, and this correspondingly increases the value of Vs at which
strong inversion occurs. Also, the increase in Qs, for the more heavily doped
substrate, at a given value of Vs in depletion, is *3 times more than for the less
heavily doped substrate, as expected from the HNa dependence in Eq. 2.12.
Between flat bands and inversion, when the ionised acceptor space charge
dominates, Eq. 2.12 can be further reduced to:
Fig. 2.3 Comparison of

surface charge density
calculations using the full
(solid line) and approximate
(symbols) expressions from
Eqs. 2.9 and 2.12,
respectively. (Substrate
doping densities of 1 9 1015
acceptors/cm3, and 1 9 1016
acceptors/cm3)
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Qs ffi 2qe0 es Na Vs Qb ð2:13Þ
Where Qb is the areal acceptor space charge density. Equation 2.13 is the same
as directly calculated from Poisson’s equation using the depletion approximation,
i.e. from:
d 2 V qNa
¼ ð2:14Þ
dx2 e0 es
and integrating this with respect to V. Alternatively, if Eq. 2.14 is integrated with
respect to x, then:
Qs ¼ qNa xd ð2:15Þ
where xd is the width of the space charge depletion region at Vs. At inversion,
when Vs = 2VF, further increases in band bending cause such large increases in
Qs, due to the exponentially increasing free electron density, that, to a first
approximation, the fixed space charge can be regarded as having reached a limiting
maximum value, Qbmax. This can be obtained by substituting Vs = 2VF into
Eq. 2.13:
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Qbmax ¼ 2qe0 es Na 2VF ð2:16Þ
and from Eq. 2.15,

Qbmax ¼ qNa xdmax ð2:17Þ
where xdmax is the maximum width of the depletion region, and, from Eqs. 2.16
and 2.17, xdmax is given by:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2e0 es 2VF
xdmax ¼ ð2:18Þ
qNa
Fig. 2.4 Illustration of bulk depletion layer, Qb, and inversion layer, Qn, charge densities in:
a inversion, and b depletion
Qs in Eq. 2.12 can be represented by the sum:

Qs ¼ Qn þ Qb ð2:19Þ
Where Qb is the extended depletion layer charge, and Qn is the areal density of
inversion layer electrons. These will be confined close to the semiconductor sur-
face, by virtue of the electrons being in a parabolic potential well of depth Vs. The
terms in Eq. 2.19 are pictorially represented in Fig. 2.4a for inversion (Vs [ 2VF),
and by Fig. 2.4b for depletion (Vs \ 2VF). The variation of xdmax with Na can be
calculated from Eq. 2.18, and is shown in Fig. 2.5. From a TFT point of view,
assuming a certain equivalence between Na and the TFT trap state density, a curve
of this type can be used to determine whether a thin film is fully depleted or not at
inversion. For instance, the maximum equilibrium space charge width, at inver-
sion, for a substrate doped with 1017 acceptors/cm3 is 100 nm; therefore, a thin
film of 100 nm, with an equivalent volume trapped charge density of \1017 cm-3,
will be fully depleted before Vs = 2VF. As a result of this, the threshold of
inversion will occur at a lower value of band bending than for a thicker film having
the same trap state density. This is discussed further in Chap. 3
Fig. 2.5 Variation of
maximum depletion layer
width at inversion, xdmax,
with substrate doping level
2.2.2 Gate Bias and Threshold Voltage
Referring to Fig. 2.1, the voltage on the gate, VG, is dropped partially across the
dielectric, Vi, and partially across the semiconductor, Vs, so that:
VG ¼ Vi þ Vs ð2:20Þ
And, for charge neutrality, the charge on the gate, QG, equals the charge in the
semiconductor, Qs, and
QG ¼ e0 ei Fi ¼ e0 ei Vi =di ¼ Ci Vi ¼ Qs ð2:21Þ
Where Fi is the field in the gate dielectric, and Ci is the capacitance/unit area of
the gate dielectric. Hence,
VG ¼ Vs Qs =Ci ð2:22Þ
Equation 2.22 can be used to relate the voltage on the gate of an MIS capacitor
to the induced charge density in the semiconductor, Qs, and to the associated band
bending, Vs. It can also be used to calculate the threshold voltage, VT, of the
structure, i.e. the gate voltage necessary to induce band bending of 2VF at the
semiconductor surface:
2qe0 es Na 2VF
VT ¼ 2VF þ ð2:23Þ
Ci
Figure 2.6 shows the calculated variation of VG with Vs for the two substrate
doping levels used in Fig. 2.3, and for a 150 nm thick gate dielectric of SiO2. The
more heavily doped substrate requires a larger gate bias to achieve a given degree
of band bending in depletion, and, equally, has a larger value of VT, as shown by
Eq. 2.23. Once the surface has gone into strong inversion, the free electron den-
sity, Qn, dominates Qs, and Qn is given by:
Qn ¼ Ci ðVG VT Þ ð2:24Þ
As will be seen in Chap. 3, this is a widely used expression for the calculation
of channel currents in MOSFET devices above threshold. It is equally widely used
for similar calculations in TFTs. Hence, the threshold voltage is a key parameter in
determining on-state device characteristics, and the dependence of threshold
voltage on substrate doping level is shown in Fig. 2.7.
2.2.3 Real MIS Structures
In contrast to the idealised MIS structure shown in Fig. 2.1, real structures may
have work function differences between the gate metal and the semiconductor,
Fig. 2.6 Calculated variation of surface potential, Vs, with gate bias, VG, in an ideal MIS
structure, and showing the threshold voltage points. (Dielectric layer of 150 nm of SiO2, and p-
type substrate doping of 1 9 1015 cm-3, and 1 9 1016 cm-3)

of threshold voltage with
substrate doping level, and
thickness of SiO2 gate
dielectric
fixed charges in the oxide, and interface trapping states at the dielectric/semi-
conductor interface. These will change the zero-gate-bias band bending, and affect
the relationship between VG and Vs and Qs.
2.2.3.1 Work Function Differences
In general, the Fermi level position in a free metal will be different from the Fermi
level in a free semiconductor. These differences are usually expressed in terms of a
work function difference, where the work function is the energy required to remove
Fig. 2.8 MIS energy band diagrams in the presence of metal–semiconductor work function
differences: a separated metal–insulator and insulator–semiconductor, b MIS system in thermal
equilibrium
an electron from the Fermi level to the vacuum level. When the two materials are
connected, and in thermal equilibrium, the Fermi levels need to be coincident, and
there will be a flow of electrons from one material to the other, resulting in an
interfacial dipole, which establishes this equilibrium. In the semiconductor, this will
result in surface band bending, the extent of which will be much greater than in the
metal due to the short screening length of the latter’s high electron density.
In an MIS structure, it is conventional to reference the Fermi levels to the
dielectric’s conduction band edge [4] (rather than to the vacuum level), so that the
quoted barriers, UMi and vi, are the energies needed to remove an electron to the
dielectric conduction band (as shown in Fig. 2.8, and v is the electro-negativity of
the semiconductor). For the Si/SiO2 system, these barrier energies were experi-
mentally established by photo-emission measurements from the metal gate or the
semiconductor into the oxide conduction band, and were 0.5–1 eV less than the
respective vacuum values [5]. Figure 2.8a schematically shows the metal and the
semiconductor separated, and Fig. 2.8b shows them connected in thermal equi-
librium. In this case, electrons have flowed from the metal to the semiconductor to
bring the Fermi levels into coincidence. As a result of this, the bands in the
semiconductor have bent downwards by Vs0, with a corresponding voltage drop
across the dielectric of Vi0. Equating the potentials either side of the high point in
the insulator conduction band:
UMi þ Vi0 ¼ vi þ Ei =q þ VF Vs0 ð2:25Þ
The metal–semiconductor work function difference, UMS, is, therefore, given by:
UMS ¼ UMi ðvi þ Ei =q þ VF Þ ¼ ðVi0 þ Vs0 Þ ð2:26Þ
Thus, to remove the zero gate bias band bending in the system, and to restore
flat band conditions, the gate voltage needs to be made more negative to induce
Fig. 2.9 Experimentally

determined MOS work-
function differences, UMS, as
a function of Si-substrate
doping level, for different
gate contact materials.
(Reproduced from [1] with
permission of John Wiley &
Sons, Inc)
more positive charge in the semiconductor surface. The gate voltage at flat bands,
VFB, is given by:
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2qe0 es Na Vs0
VFB ¼ UMS ¼ ðVi0 þ Vs0 Þ ¼ þ Vs0 ð2:27Þ
Ci
Equally, the gate voltage required to achieve a given degree of band bending,
including the threshold for inversion, will be modified by the metal–semiconductor
work function difference, so that the expression for the threshold voltage now
becomes:
2qe0 es Na 2VF
VT ¼ VFB þ þ 2VF ð2:28Þ
Ci
As indicated in Eq. 2.26, the work function difference will be a function both of
the gate metal, and of the doping level in the semiconductor substrate, as shown in
Fig. 2.9 for Al and doped poly-Si gates [1]. Due to its refractory properties, doped
poly-Si has been extensively employed as the gate electrode in MOSFET tech-
nology, and, with appropriate doping, it can be tailored to produce the required
sign of flat band voltage.
2.2.3.2 Oxide Charges and Interface States
Many gate dielectrics, including SiO2, are characterised by charged centres in the
oxide, which induce an opposite charge in the semiconductor surface. If the
charged centres, Qi, are located at a distance x from the dielectric-semiconductor
interface in an MIS capacitor, the charge induced in the semiconductor, Qs, is:
Qs ¼ Qi ðd xÞ=d ð2:29Þ
Hence, for charges at the dielectric/semiconductor interface, the induced charge

is equal to the charge, Qi, in the dielectric, whilst those at the metal interface have
no effect. In many cases, the location of the charges may not be known, and an
effective charge density, Qieff, will be determined, as though it were located at the
dielectric/semiconductor interface.
As with the work function difference, this will result in a non-zero flat band
voltage, given by:
VFB ¼ Qieff =Ci ð2:30Þ
Positive charges are commonly found in SiO2 films on Si, and are referred to as
‘fixed’ charges, as their effect is not dependent upon the surface potential in the
semiconductor. In other words, they have a constant effect, irrespective of the bias
conditions, and present a fixed density.
In real crystalline semiconductor surfaces, the termination of the regular,
periodic bulk potential introduces allowed states into the forbidden band gap,
usually referred to as surface states. These are intrinsic states at the surface, and, in
addition to these, there may be extrinsic states due to impurities. The densities of
intrinsic states can be up to *1015 cm-2 on an atomically clean surface, but, in a
well-passivated surface, such as the Si/SiO2 interface produced by thermal oxi-
dation of Si, these can be reduced to \1–5 9 1012 cm-2. The reduction is due to
the SiO2 network providing oxygen pairing atoms for the Si dangling bonds, and
further passivation with hydrogen typically reduces the overall density to
*1010 cm-2 or less. This combination of thermal oxidation and hydrogen pas-
sivation has generated one of the best-controlled semiconductor interfaces, and
provided the basis for the Si semiconductor device industry.
Surface or interface states are generally distributed in energy across the for-
bidden band gap, and, in the case of the Si/SiO2 interface, the distribution is
U-shaped between the valence and conduction band edges. In contrast to the fixed
charge, the charge in interface states will be determined both by their distribution
across the band gap, and by the position of the Fermi level at the surface. Using the
zero Kelvin approximation, those states beneath the Fermi level can be regarded as
filled with electrons, whilst those above will be empty. Take, for example, a
constant density, Nss(E), across the band gap, with acceptor levels in the upper half
of the band gap, and donor levels in the lower half, and with a neutral level, E0, at
mid-gap. This distribution will have zero net charge in the interface states when
the surface Fermi level is at mid-gap. However, if this distribution were added to
the MIS band diagram used in Fig. 2.1a, where the Fermi level is below mid-gap,
there would be a flow of electrons from the interface states below E0, leading to
positively charged, empty donor states at the surface, and an equal negative charge
density in the semiconductor surface. Hence, the semiconductor bands at the
surface will be bent down by an amount Vs0 to ensure charge neutrality between
the interface and the bulk, as shown in Fig. 2.10. In this example, the block of
donor states between EF and E0 would be empty of electrons, and have a positive
charge, Qss, balanced by the negative semiconductor surface charge, Qs.
In general, the electron occupancy, F(E), of a surface state at energy E is given

by the Fermi–Dirac distribution function:
1
FðEÞ ¼ ð2:31Þ
1 þ expðqVF qVS EÞ=kT
Using the polarity convention of the earlier sections, E will have a negative
value for centres above E0 (which is at mid-gap), and a positive value for those
below. Hence, the total negative charge in the surface states, Q- ss, with band
bending Vs [ VF is:
ZEC E0 þqðV
Z s VF Þ
Q
ss ðVs Þ ¼ qNss ðEÞFðEÞdE qNss ðEÞdE ð2:32Þ
E0 E0
In Eq. 2.32, the full integral expression, which includes the Fermi–Dirac
function, can be reduced to the simpler analytical form by using the zero Kelvin
approximation, in which the acceptor states are negatively charged between the
neutral level and the surface Fermi level, and neutral above the Fermi level. For a
uniform distribution of acceptor states, Nss can be brought outside the integral, and
the density of negatively charged states, when Vs [ VF, is:
Q
ss ðAÞ ¼ qNss qðVs VF Þ ð2:33Þ
Similarly, for Vs \ VF (as shown in Fig. 2.10), the density of positively
charged donor states is:
Fig. 2.10 Zero gate-bias

MIS band diagram with a
uniform distribution of
interface states across the
band gap, and with the cross-
hatched region showing the
charged states. (Donor and
acceptor states in the lower
and upper halves of the band-
gap, respectively, and with
the neutral level at mid-gap)
Qþ
ss ðDÞ ¼ qNss qðVF Vs Þ ð2:34Þ
For a constant distribution of surface states, the zero Kelvin approximation is a
useful simplification. However, for trapping state distributions where the density
varies significantly over an energy interval of kT, it is less accurate, and a
numerical evaluation of the full integral in Eq. 2.32 would be required. This issue
is discussed further in Sect. 6.2.3, where the volume trap density in a-Si:H varies
exponentially with position from the band edge.
As with the work function difference and the fixed oxide charge, a bias, VFB,
will need to be applied to the gate to establish the flat band condition in the
semiconductor, where VFB is given by:
Qss ðDÞðVs ¼ 0Þ qNss qVF
VFB ¼ ¼ ð2:35Þ
Ci Ci
However, in contrast to the fixed oxide charge, the presence of interface states
will give more than just a constant shift in the relationship between VG and Vs and
Qs. In particular, as Vs is increased, not only does the gate bias need to induce the
increasing value of Qs, but it must also supply the increasing charge going into the
interface states. The relationship between VG and Vs from Eq. 2.22 is modified to
be:
Qs Qþ ðDÞ Q ðAÞ
VG ¼ Vs ss ss ð2:36Þ
Ci Ci Ci
and, for the example being used, Qss(A) = 0 for Vs \ VF, and Qss(D) = 0 for
Vs [ VF, and the non-zero values of Qss(A) and Qss(D) are negative and positive,
respectively.
Figure 2.11a compares the relationship, between VG and Vs, for an ideal trap-
free interface (Nss = 0) and an interface containing a range of Nss densities from
1 9 1010 cm-2 eV-1 to 1 9 1013 cm-2 eV-1. The lowest interface state density
of 1010 cm-2 eV-1 is so low that the VG–Vs relationship is indistinguishable from
the trap-free case. However, for the higher densities, increasingly large values of
VG are required to achieve a given amount of band-bending due to the charge
going into the interface states. The curves have a common crossing point at
Vs = 0.288 V, when the surface Fermi level is at mid-gap. This corresponds to the
neutral level of the interface states, so that all the interface state distributions have
the same zero charge in them. The flat band voltages can also be directly read from
the curves by the values of VG corresponding to Vs = 0 V.
The other parameter which can be extracted from these curves is the value of Vs
when VG = 0, which is the equilibrium zero-gate-bias bend banding, Vs0, and these
values, together with the corresponding values of Qss (at Vs0), are shown in
Fig. 2.11b. The ‘S’-shaped curves demonstrate some simple physical principles of
charge trapping over the range of Nss values used. For low Nss values, such as
1010 cm-2 eV-1, the Fermi level position at the surface is dominated by the sub-
strate doping level, and the surface states have almost no effect. In this case, the Nss-
Fig. 2.11 Calculations using the uniform interface state distribution shown in Fig. 2.10
a variation of gate bias with surface potential, Vs, for different values of Nss, and b variation
of zero gate bias band bending, Vs0, and Qss (at Vs0) with Nss. (Substrate acceptor density,
Na = 1015 cm-3, qVF = 0.288 eV, tox = 150 nm, and interface state density, Nss = 1010–
1013 cm-2 eV-1)
induced band bending tends towards zero, and the interface states are ionised almost
down to the position of the bulk Fermi level, and Qss * qNssqVF. Although a
substantial fraction of the states are ionised, the charge in them is so low, that
negligible band bending is required to produce an approximately equal amount of
charge in the substrate. In contrast, for high Nss values, such as 1013 cm-2 eV-1, the
band-bending is dominated by the charge in the surface states, and the Fermi level
tends towards being pinned at the surface state neutral level, i.e. Vs0 ) VF. In this
case, VF-Vs0 = 0.01 V, and the surface state neutral level is just 100 meV above
the Fermi level. However, with the large surface state density, this still corresponds
to a substantial charge, Qss, in the fractionally ionised donor surface states. For
intermediate values of Nss between 1011 and 1012 cm-2 eV-1, there is a progressive
increase in Vs0 with Nss, and Qss is given by qNssq(VF-Vs0).
Equation 2.36 also shows that the charge going into surface states has to be
allowed for in calculating the gate threshold voltage for surface inversion. The
surface potential at which this occurs is still 2VF, but the gate bias, in excess of the
flat band voltage, to achieve this surface potential is increased to:
2qe0 es Na 2VF Qss ðVs ¼ 2VF Þ
VT ¼ VFB þ þ 2VF þ ð2:37Þ
Ci Ci
and, as Qss(D) = 0 for Vs [ VF, only charge in the acceptor states will contribute,
and from Eq. 2.33, Qss(A) at inversion is given by:
Qss ðAÞ ¼ qNss qð2VF VF Þ ¼ qNss qVF ð2:38aÞ
Finally, the effects of the work function difference, the dielectric charge and the
charge in interface states are all independent and additive, so that, in the presence
of all three, the flat band voltage is given by:
VFB ¼ UMS Qieff =Ci Qss ðVs ¼ 0Þ =Ci ð2:38bÞ
2.2.4 Evaluation of Surface Potential
From Eq. 2.9, the surface potential, Vs, uniquely defines the space charge density,
Qs, and Eqs. 2.22 and 2.36 define the gate bias, VG, needed to achieve that potential,
with and without surface states, respectively. Therefore, the measurement of Vs
facilitates a quantitative insight into the state of a semiconductor surface. A common
technique for establishing Vs is through the small signal capacitance-voltage mea-
surement of an MIS diode, where the capacitance is given by:
dQG dQs dVs dQs
C¼ ¼ ¼
dVG dVG dVG dVs
dVG 1 dQs dQs ð2:39Þ
¼1 and ¼ Cs
dVs Ci dVs dVs
) C ¼ 1=ð1=Ci þ 1=Cs Þ
In other words, the capacitance of the MIS diode is the series combination of
the capacitance of the dielectric, Ci, and of the semiconductor surface, Cs, where
Cs is given by the differentiation of Eq. 2.9:
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ni qe0 es ½shqðVs VF Þ=kT þ shqVF =kT
Cs ¼ ð2:40Þ
GðVs ; VF Þ
and G(Vs, VF) is given by Eq. 2.10. The normalised C–V curves resulting from the
evaluation of Eq. 2.40, for a silicon substrate doped with 1015 and 1016 acceptors/
cm3, and a gate oxide thickness of 150 nm, are shown by the solid lines in
Fig. 2.12 Normalised high

and low frequency MIS C–V
curves (calculated for 150 nm
thick SiO2 gate dielectric)
Table 2.2 Relationship between normalised flat band capacitance, minimum C–V capacitance
and substrate doping level (for 150 nm SiO2 gate dielectric)
Na (cm-3) CFB/Ci Cmin(LF)/Ci Cmin(HF)/Ci
1 9 1015 0.77 0.38 0.32
1 9 1016 0.91 0.63 0.59
Fig. 2.12. These curves have a characteristic ‘V’ shape, with the capacitance at
negative VG values tending towards the capacitance of the gate dielectric, due to
the large capacitance of the hole accumulation layer. With increasing positive VG,
the capacitance falls due to the growth of the hole depletion region at the surface,
which progressively reduces the surface capacitance. Finally, the curves start to
rise as the surface becomes strongly inverted, and the large free electron con-
centration screens the underlying depletion layer, with the total capacitance once
again tending towards the value of the gate dielectric.
As these curves have been calculated for an ideal MIS diode, the energy bands
will be flat at VG = 0, and, hence, the normalised capacitance at flat bands can be
directly obtained from the curves. Given the unique relationship between surface
potential and capacitance, the flat band voltage can, in principle, be directly
obtained from any experimental C–V measurement using Eq. 2.40, assuming that
the substrate doping level is known. Moreover, the minimum capacitance is also a
unique function of doping level and dielectric capacitance, so that the doping level
can be extracted from the minimum of the experimental C–V curve. The nor-
malised flat band and minimum capacitance values from Fig. 2.12 are listed in
Table 2.2, and they have also been extensively tabulated by Sze and Ng [1]. Also
marked on the curves in Fig. 2.12 are the capacitance values at which the band
bending gives an intrinsic surface (Vs = VF), and the threshold for an inverted
surface (Vs = 2VF).
The solid line calculations assume thermal equilibrium, and the increase in
capacitance in inversion will only be observed experimentally if the free electrons
in the inversion layer can follow the a.c. measuring signal. In a high quality Si/
SiO2 capacitor, the inversion layer response time can be of the order of seconds, or
more [6], as it depends upon a thermal generation process within the surface space
charge region. Hence, the calculated curves are only likely to be replicated with a
low frequency measurement, and the solid line curves in Fig. 2.12 are often
referred to as low frequency C–V curves. Moreover, in the presence of interface
states, there can be a further complication in the interpretation of the experimental
low frequency curves. This will arise if the occupancy of the interface states at the
Fermi level can follow the measuring signal, and, hence, contribute an interface
state capacitance, Css, in parallel with the semiconductor surface capacitance Cs.
The surface state capacitance, Css, is given by:
dQss
Css ¼ qNss ð2:41Þ
dVs
where, as discussed in the previous section, Qss can be approximated by

qNssq(Vs-VF) for constant (or slowly varying) surface state densities. The pres-
ence of Css in an experimental C–V measurement means that Eq. 2.40 can no
longer be used to give a unique relationship between measured capacitance and
band bending, Vs. By implication, a high frequency measurement can be used to
minimise the capacitive contributions of the surface states, and will also suppress
the free electron capacitance in strong inversion. Hence, to interpret the high
frequency C–V curves, a high frequency version of Eq. 2.40 is required. This
essentially means suppressing the electron response in inversion, and this can be
accomplished by expanding the cosh and sinh terms containing Vs, and setting to
zero those of the form expq(Vs-VF)/kT, since, in inversion, Vs [ VF and those
terms will govern the electron contribution to the surface capacitance. Hence the
high frequency expression, Cs(HF) is given by:
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ni qe0 es ½0:5 exp qðVs VF Þ=kT þ shqVF =kT
Cs ¼ ð2:42Þ
G0 ðVs ; VF Þ
and G0 (Vs,VF) is given by:

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

0 kT qðVF Vs Þ kT qVF qVF
G ðVs ; VF Þ ¼ 0:5 exp ch þ Vs sh ð2:43Þ
q kT q kT kT
The high frequency C–V curve, calculated from Eq. 2.42, for qVs [ 3kT, is
shown by the dotted lines in Fig. 2.12. The inversion layer response has been
suppressed, and the surface capacitance in inversion tends to a constant value,
approximately given by:
e0 es
CsðHFÞ ¼ ð2:44Þ
xdðmaxÞ
Where xd(max) is the limiting maximum thickness of the surface depletion

region, as discussed in Sect. 2.2.1, and is given by Eq. 2.18. As with the minimum
capacitance value of the low frequency curves, the minimum value of the high
frequency curves is also a unique function of doping level, and can be used to
establish the doping level in experimental samples. These values are also shown in
Table 2.2. More importantly, any horizontal displacement, DVG, of an experi-
mental C–V curve from the theoretical C–V curve can be used to establish Qss, as a
function of Vs, from:
Qss ðVs Þ ¼ DVG =Ci ð2:45Þ
and Nss can then be derived from dQss/dVs.

Whilst the emphasis in this section has been on high frequency measurements,
and the use of the measured MISC capacitance to establish surface potential, there
is also a low frequency procedure, called the quasi-static C–V measurement, which
uses the change in capacitance, at a given value of VG, to determine Nss [7, 8].
Finally, to complete this section, one other well-established technique for the
Fig. 2.13 Illustration of a reverse biased n+-p junction, showing generation and diffusion
leakage currents, and b e–h pair thermal generation process
evaluation of surface state densities should be mentioned, and this is the a.c
conductance measurement of MISCs [9]. A detailed overview of these procedures
can be found in Sze and Ng [1].
2.3 Electron-Hole Pair Generation and Recombination
Electron-hole pair generation and recombination are fundamental semiconductor

processes relevant to both off-state leakage currents and to steady-state photo-
currents in TFTs.
Two sources of leakage current in semiconductor devices are by ohmic con-
duction in lightly conducting and/or low generation lifetime material, or by
electron-hole pair generation and/or diffusion current flow in reverse biased
junctions. The latter processes are illustrated in Fig. 2.13a for a reverse biased n+p
junction. Due to the field within the space charge region, the holes generated
within this region are swept into the p-type substrate, where they constitute an
ohmic (drift) hole current to the substrate contact, Idrift. Equally, the electrons
generated in the space charge layer are swept into the n+ region, where they flow to
the n+ contact. For current continuity, there will be equal hole and electron cur-
rents, and these are referred to as generation currents. The electron-hole pair
generation process, from a deep lying centre within the forbidden band gap, is
shown schematically in Fig. 2.13b. In addition to the direct generation process,
there is also a diffusion current, Idiffn, of minority carrier electrons from the neutral
p-type substrate into the adjacent depletion region.
The role of electron-hole pair generation on leakage currents, including the
diffusion current, is discussed in Sect. 2.3.3, and the recombination processes are
presented in Sect. 2.3.4.
2.3 Electron-Hole Pair Generation and Recombination 29
Fig. 2.14 Carrier emission and capture processes from a deep level centre in the forbidden band
gap: a electron emission, b electron capture, c hole emission d hole capture. The horizontal block
arrow shows the state’s occupancy after the transition
In order to develop the expressions for electron-hole pair generation and

recombination, it is necessary to consider the basic processes of carrier emission
and carrier capture by a deep level centre, based upon the Shockley, Read, Hall
(SRH) recombination statistics [10, 11], and the extension of this analysis to p–n
junction characteristics [12].
The term deep level centre refers to intrinsic or extrinsic defects having energy
levels within the forbidden band gap of the semiconductor, and which are usually
further from the band edges than the shallow level dopant impurities. A deep level
centre can behave as a trap, or as a generation centre, or as a recombination centre,
depending upon its environment. For instance, in thermal equilibrium, a deep level
acceptor in n-type material can capture electrons, thereby, reducing the free carrier
density, and is acting as trap.3 In a reverse biased space charge region, it can act as
a generation centre, sequentially emitting electrons and holes, resulting in a
reverse bias leakage current. Or, in material subject to electron-hole pair injection,
it can act as a recombination centre, where the sequential electron and hole capture
process establishes the steady-state free carrier density under the injection con-
ditions. Hence, all three terms are often used interchangeably to refer to the same
deep level centre, and, for brevity, the term trap is used as the generic term for the
centre in this section.
2.3.1 Thermal Equilibrium
For a deep level centre within the semiconductor’s forbidden band gap, four carrier
transitions can be identified, as shown in Fig. 2.14. If the trap is an acceptor level, it
has two charge states, one of which is negatively charged when it is occupied by an
electron, and the other is neutral when it is empty of the electron (These two charge
states can be equivalently described in terms of hole occupancy, which are empty
3
The Fermi level position and the free electron concentration can be obtained from the
þ
numerical solution of the charge neutrality equation: n þ N
T ¼ Nd .
and filled, respectively. The same transitions will occur from a deep donor centre,
the only difference being that it is positively charged when empty of an electron and
neutral when occupied by an electron). Hence, there is a pair of possible carrier
transitions from each of the two charge states, as shown in Fig. 2.14. From the
negatively charged acceptor centre, these are either (a): electron emission to the
conduction band, by a thermally stimulated process, or (d): hole capture from the
valence band, and the release of phonon energy back to the lattice. Both leave behind
the neutral centre. From the neutral centre, the two possible transitions are (b):
electron capture from the conduction band, or (c): thermally stimulated hole emis-
sion to the valence band (which is equivalent to electron emission from the valence
band). Both leave behind the negatively charged centre.
If the density of traps/generation/recombination centres is NT, and if the centre
and the Fermi level are positioned at ET and EF, respectively, above the valence band
edge, then the trap occupancy is defined by the Fermi–Dirac distribution function,
and from this the carrier transition rates can be calculated. These rates are deter-
mined by the number of free carriers and the appropriate occupancy of the centre, for
the capture process, or just by the occupancy of the centre for the emission process
(where it is implicitly assumed that there are sufficient empty states in the bands to
accommodate the emitted carriers). In thermal equilibrium, the fractional electron
occupancy of the trap, FD(ET, EF), is given by the Fermi–Dirac distribution function:
1
FD ¼ ð2:46Þ
1 þ expðET EF Þ=kT
The free carrier densities, n0 and p0, in n-type material doped with Nd donors,
are given by4:
n0 þ NT ¼ Nd ¼ NC exp ðEC EF Þ=kT; p0 ¼ NV exp EF =kT ð2:47Þ
and n0 p0 ¼ n2i ð2:48Þ

The electron emission rate, Ree is:
Ree ¼ en NT ¼ en NT FD ð2:49Þ
where en is the electron emission rate constant. Similarly, the rate of hole emission,
Rhe, is:
Rhe ¼ ep NT0 ¼ ep NT ð1 FD Þ ð2:50Þ
where ep is the hole emission rate constant. The rates of electron capture, Rec, and
hole capture, Rhc, are:
Rec ¼ cn n0 NT0 ¼ rn vth n0 NT ð1 FD Þ ð2:51Þ
4
From 2.47, if NT
Nd, EF ¼ kT lnðNd NV =n2i Þ.
Rhc ¼ cp p0 NT ¼ rp vth p0 NT FD ð2:52Þ
where the electron and hole capture rate constants, cn and cp, are given by the
product of the trap’s capture cross section, rn and rp, respectively, and the carrier’s
thermal velocity, vth. The thermal velocity is H(3kT/qm*), where m* is the
effective mass, so that the thermal velocities for holes and electrons are different
from each other. However, for simplicity in this analysis, they are taken to be the
same.
In thermal equilibrium, the rate of electron capture must equal the rate of
electron emission, and the equivalent equality must also exist for the hole tran-
sitions. Hence,
en NT FD ¼ rn vth n0 NT ð1 FD Þ ð2:53Þ
And substituting for FD and n0, using Eqs. 2.46 and 2.47, respectively:
en ¼ rn vth NC exp ðEC ET Þ=kT ð2:54Þ
Similarly, for the hole transitions:
ep ¼ rp vth NV exp ET =kT ð2:55Þ
Hence, the thermal emission rate constants, en and ep, are fundamental prop-
erties of the trap itself, and not dependent on the local concentrations of carriers.
Moreover, their values are exponentially dependent upon the separation of the trap
from the band edges.
2.3.2 Non-equilibrium, Steady State
Under steady state, non-thermal equilibrium conditions (where np 6¼ n2i ), the trap
will achieve a new occupancy, described by its quasi-Fermi level, and determined by
the equality of the net electron transition rate and the net hole transition rate,5 i.e.
dNT
¼ Rec Ree þ Rhe Rhc ¼ 0 ð2:56Þ
dt
In contrast to the thermal equilibrium situation, the rates of electron emission
and electron capture will not be equal. This is the situation in which there will be a
net recombination or a net generation process, depending upon the divergence of
the carrier concentrations from their thermal equilibrium values. Substituting the
rate constants into Eq. 2.56 gives:
ncn NT0 en NT ¼ pcp NT ep NT0 ð2:57Þ
5
If these rates are not equal, then the trap occupancy will change with time.
From Eq. 2.57, and using NT0 ¼ NT NT , the hole occupancy is:
NT ðen þ pcp Þ
NT0 ¼ ð2:58Þ
ncn þ pcp þ en þ ep
This can be used to calculate the net local electron generation/recombination
rate, RGR, which, in steady state, will also be equal to the net hole recombination/
generation rate, i.e.:
RGR ¼ Ree Rec ¼ NT en ncn NT0 ð2:59Þ

Substituting for the trap occupancy terms:
NT ðen ep npcn cp Þ
RGR ¼ ð2:60Þ
ncn þ pcp þ en þ ep
rn rp vth NT ½fNC :NV exp ðEC ET Þ=kT: exp ET =kTg np
RGR ¼ ð2:61Þ
rn ðn þ NC exp ðEC ET Þ=kTÞ þ rp ðp þ NV exp ET =kTÞ
Eliminating the exponential terms in ET in the numerator, and replacing the
remaining exponential term by n2i ; gives the following expression:
rn rp vth NT ½n2i np

RGR ¼ ð2:62Þ
rn ðn þ NC exp ðEC ET Þ=kTÞ þ rp ðp þ NV exp ET =kTÞ
The numerator shows that the rate of recombination/generation is proportional
to the term ðn2i npÞ, which expresses the deviation of the carrier population from
thermal equilibrium. For np [ n2i , the free carrier excess stimulates a net recom-
bination process, where the carrier excess could result, for example, from optical
illumination of the device, the forward biasing of a p–n junction or avalanche
generation at a reverse biased p-n junction. For np \ n2i , the free carrier deficit
gives rise to net generation, and the free carrier deficit is most commonly
encountered in the depletion region of a reverse biased p–n junction. In thermal
equilibrium, the net rates of recombination or generation are zero.
Equation 2.60 (or its equivalents 2.61 or 2.62) would be typically used in
device simulation packages to calculate the local rates of recombination or gen-
eration, but, for analytical work, a simplified version of this expression is often
used by making the simplifying assumption that rn = rp = r, and:
rvth NT ½n2i np

RGR ¼ ð2:63Þ
n þ p þ 2ni coshðET Ei ÞkT
where the following substitutions have been used for NC and NV in the
denominator:
ni ¼ NC exp ðEC Ei Þ=kT ¼ NV exp Ei =kT ð2:64Þ
2.3.3 Generation Currents
As mentioned above, the general treatment of generation or recombination is fully

specified by the use of Eq. 2.60, and, in this section, an approximate, but more
accessible, analytical treatment is presented for describing generation currents. For
instance, in the space charge layer of a reverse biased junction, as shown in
Fig. 2.13a, the free carrier concentrations can be taken as zero, so that there are no
carrier capture processes, and Eq. 2.60 reduces to an expression for the local,
steady state generation rate, RG:
NT en ep
RG ¼ ð2:65Þ
en þ ep
The steady state leakage current, JR, due electron-hole pair generation, in a
reverse biased p-n junction depletion layer of width W is:
ZW
en ep
JR ¼ q RG dx ¼ qRG W ¼ qWNT ð2:66Þ
en þ ep
0
Where it is assumed that RG is constant across W, and, in analogy with

Eq. 2.18, W is given by:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2e0 es ðVR þ V0 Þ
W¼ ð2:67Þ
qNa
VR is the reverse bias, and V0 is the built-in p–n junction potential.

Returning to Eq. 2.66, the ratio enep/(en ? ep) is a maximum when the
denominator is a minimum, which will occur when en = ep, and the traps are sited
close to mid-gap. As ET moves away from mid-gap either en or ep will increase
exponentially, as will the sum en ? ep, whilst the product enep will be invariant.
Hence, any centres located at mid-gap are likely to be the most efficient generation
centres, and will dominate the device leakage current, JR. For mid-gap centres,
with equal capture cross sections, r:
en ¼ ep ¼ rvth NC exp Ei =kT ¼ rvth ni ð2:68Þ
) JR ¼ qRG W ¼ 0:5qni Wrvth NT ð2:69Þ

The generation lifetime, sg, is a figure of merit for the leakage current, and is
defined by the following relationship:
JR qni W=sg ð2:70Þ
Hence, for mid-gap centres, the generation lifetime is:
sg ¼ 1=ð0:5NT rvth Þ ð2:71aÞ
If the capture cross-sections are not equal:

sg ¼ 1=ðNT rn vth Þ þ 1=ðNT rp vth Þ ð2:71bÞ
Equation 2.71b explicitly shows that the rate of the two-stage generation pro-
cess is determined by the sequential emission of holes and electrons, and, if one
capture cross-section is larger than the other, then the overall generation process
will be rate limited by carrier emission from the charge state of the centre having
the smaller cross-section. In experimental samples, this difference in capture cross-
sections is likely to be found. For example, if the trap is an acceptor, then it will be
neutral before capturing an electron, and negatively charged before capturing a
hole. The Coulombic attraction between the hole and the negatively charged centre
will result in the hole capture cross-section being larger than the neutral cross-
section for electron capture [13]. As an example of this, one of the most widely
studied deep level centres in silicon is gold, which has a near mid-gap acceptor
level at 0.54 eV below the conduction band edge, and its capture cross-sections are
1.5 9 10-14 cm2 and 9 9 10-17 cm2, for holes and electrons, respectively [14]. In
this case, Eq. 2.71b would be more appropriate than 2.71a for calculating the
generation lifetime. In both cases, the generation lifetime is inversely proportional
to the number of generation centres, and, as would be expected, the leakage current
is proportional to this density.
However, if the only deep level centres are off mid-gap by more than a few kT,
then, for those in the upper half of the band gap, en ep (and en
ep if they are
in the lower half). Taking, for example, a centre in the upper half of the band-gap,
Eq. 2.66 reduces to:
JR ¼ qWNT ep ð2:72Þ
and the leakage current is limited by just the smaller of the two emission rate
constants (in this case ep), because, once the hole has been emitted, the following
step of electron emission is so much faster that it has a negligible impact upon the
overall two stage emission process. Expanding ep:
JR ¼ qWNT rp vth ni exp ðET Ei Þ=kT ð2:73Þ
) sg ¼ 1=NT rp vth exp ðET Ei Þ=kT ¼ 1=NTðeff Þ rp vth ð2:74Þ
NTðeffÞ ¼ NT exp ðET Ei Þ=kT, and is the equivalent number of mid-gap

centres, which would give the same leakage current. Hence, the shift in position of
a generation centre from mid-gap increases the generation lifetime by exp(ET-Ei)/
kT, and decreases the leakage current by the same amount.
In summary, the reverse bias p–n junction leakage current is characterised by
the generation lifetime, sg, which is related to the physical characteristics of the
deep level generation centres themselves.
The other contribution to junction leakage current is the flow of minority

carriers, into the reverse biased space region, from the adjacent neutral material. In
an n+p junction, this will be electrons from the p-type substrate, and those adjacent
to the space charge region will be swept through it, and into the n+-region. This
leaves a deficit of electrons at the depletion region edge, which establishes a local
minority carrier concentration profile, thereby, driving a diffusion current, Jdiffn. As
the ohmic contact to the p-type substrate cannot readily supply electrons, they need
to be generated within the material by thermal generation, and the diffusion current
is determined by the generation lifetime, and is given by [4]:
qffiffiffiffiffiffiffiffiffiffiffiffi
Jdiffn ¼ qnp Dn =sg ð2:75Þ
Where np is the equilibrium electron concentration in the p-type substrate

(=p0/n2i ), and Dn is the electron diffusion coefficient (=kTln/q). A significant dif-
ference between the diffusion current and the generation current (Eq. 2.70) is that
the diffusion current is not a function of the reverse bias voltage.
The treatment in this section has been limited to basic thermal emission pro-
cesses, and these are directly used in the discussion of TFT leakage currents in
Sect. 8.4. In Sect. 8.5.3, the treatment is extended to include field-enhanced
emission processes.
2.3.4 Recombination Processes
Recombination will occur in response to a steady state carrier injection process

(such as above band-gap optical illumination), where the injection results in an
increase of the free carrier concentration above the thermal equilibrium values.
The steady state carrier concentrations are given by the equality:
Rinj ¼ RR ð2:76Þ
Where Rinj is the injection rate, and RR is the recombination rate given by
Eq. 2.60 (or the equivalent Eqs. 2.61 and 2.62). As with the generation process, it
is useful to simplify these equations to more physically meaningful forms. For
charge neutrality, the injection process will result in an equality of the change in
trap occupancy, and in the steady state excess densities of both holes and electrons,
given by Dp and Dn, respectively. Where p = p0 ? Dp, and n = n0 ? Dn, and p0
and n0 are the thermal equilibrium carrier densities, and n0p0 = n2i .
As with the discussion in the preceding sections, an n-type substrate is assumed,
and, depending upon the relative values of Dp and n0, two different injection
conditions are identified. When Dp
n0, this is regarded as low-level injection,
and when Dp C n0 this is defined as high-level injection.
2.3.4.1 Low-Level Injection
Consider an n-type substrate, with low-level injection, meaning that Dp

n0, so
that n * n0. The low-level recombination lifetime, sR, is defined by:
RR Dp=sR ð2:77Þ
An examination of the physical recombination process will illustrate how
Eq. 2.60 can be simplified to yield a more tractable expression for the recombi-
nation lifetime. Under low-level injection, the change in the free electron con-
centration is so small that the equilibrium Fermi level can still be used to describe
the electron concentration. Hence, all trap levels beneath the Fermi level are filled
with electrons, and ready to trap a hole as the first stage in the two-stage
recombination process. Having captured a hole, with the rate constant
cp(p0 ? Dp), the subsequent electron capture will proceed with the rate constant
cnn0, which, given that n0 (p0 ? Dp), will be a much faster process, and will
not rate limit the overall recombination process. The only other consideration is
whether the hole could be thermally emitted back into the valence band before the
electron capture process has been completed. For the electron capture rate to
exceed the hole emission rate, we need cnn0 [ ep, i.e.:
cn NC exp ðEC EF Þ=kT [ cp NV exp ET =kT ð2:78Þ
i:e: ET [ ðEC EF Þ þ kT lnðcp NV =cn NC Þ ð2:79Þ

Taking the logarithmic term in Eq. 2.79 to be smaller than EC-EF, Eq. 2.79
simply shows that, for efficient electron capture, the trap level must be further from
the valence band edge than the Fermi level is from the conduction band edge. In
other words, providing the trap level, ET, is positioned somewhere between the
Fermi level and an equivalent distance from the valence band edge, the recom-
bination process will be limited just by hole capture. Replacing n and p in
Eq. 2.62:
rn rp v2th NT n0 Dp
RR ¼ ð2:80Þ
cn n0 þ en þ cp ðp0 þ DpÞ þ ep
From the above discussion, for EF [ ET [ EC-EF, then cnno [ en, ep, and
cp(p0 ? Dp), and Eq. 2.80 reduces to:
RR ¼ rp vth NT Dp Dp=sR ð2:81Þ
and sR ¼ 1=rp vth NT ð2:82Þ

In other words, the recombination lifetime is a function of the trap state density,
NT, and the hole capture cross-section, and is independent of the trap energy level
over a substantial fraction of the band gap (as defined above). Equation 2.82 is
similar to the generation lifetime for a mid-gap generation centre (Eq. 2.71), but,
in contrast to the recombination lifetime, once ET is more than a few kT from mid-
gap, the generation lifetime increases substantially (as shown by Eq. 2.74).
Therefore, except in the special case of a mid-gap centre, the steady state gener-
ation and recombination lifetimes can be substantially different.
When ET lies outside the range prescribed for Eq. 2.81, the recombination
lifetime will increase. If ET [ EF, then the traps will be largely empty of electrons,
and the number of traps available for hole capture will be low. In this situation,
en [ cnn0, and Eq. 2.80 becomes:
rn rp v2th NT n0 Dp
RR ¼ rp vth NT Dp expðEF ET Þ=kT ð2:83Þ
en
and sR ¼ expfðET EF Þ=kTg=rp vth NT ðfor ET [ EF Þ ð2:84Þ
Hence, as ET rises above EF, the lifetime is larger than that given by Eq. 2.82,
and it increases exponentially with increasing values of ET.
Similarly, if the separation of ET from the valence band edge is smaller than the
value of (EC - EF), then ep will exceed cnn0, and, from Eq. 2.80, the recombi-
nation lifetime will be:
NV expðEC EF ET Þ=kT
sR ¼ ðfor ET EC EF Þ ð2:85Þ
NC rn vth NT
In other words, as the trap level gets closer to the valence band edge, and ET
reduces, the lifetime exponentially increases, but is now controlled by the capture
cross section for electrons, rather than for holes, as it has been in the previous
cases.
2.3.4.2 High-Level Injection
As with low-level injection, the recombination process is described by the general

Eq. 2.60 (or its equivalents), and to reduce these to a simpler form, the high-level
injection situation of Dp n0 is considered. In this case, if Dp NT, then charge
neutrality will be determined just by equality in the excess free carrier densities,
i.e. n = p = Dp = Dn, and the equilibrium Fermi level splits into separate quasi-
Fermi levels, EFp and EFn, for holes and electrons, respectively, where the quasi-
Fermi levels are given by:
p ¼ NV exp EFp =kT and n ¼ NC exp ðEC EFn Þ=kT ð2:86Þ
If the recombination centre is sited between the quasi-Fermi levels, such that
EFp \ ET \ EFn, then ncn [ en and pcp [ ep, and the steady state recombination
centre occupancy can be obtained from Eq. 2.58:
NT rp
NT0 ¼ NT ð1 FD Þ ð2:87Þ
rn þ rp
FD is the centre’s Fermi-Dirac function, given by Eq. 2.46, in which the

equilibrium Fermi level, EF, is replaced by the quasi-Fermi level for the recom-
bination centre, EFT. For the special case of equal capture cross sections, FD = 0.5,
and the quasi-Fermi level for the recombination centre is located at ET.
For the recombination centre located between the hole and electron quasi-Fermi
levels, the recombination lifetime, sR, can be calculated from Eqs. 2.60 and 2.77:
NT ðen ep npcn cp Þ NT Dp2 cn cp Dp

RR ¼ ð2:88Þ
ncn þ pcp þ en þ ep Dpcn þ Dpcp sR
and
sR ¼ 1=rp vth NT þ 1=rn vth NT ð2:89Þ
Hence, the difference between the low-injection and high-injection lifetimes is
that, at high-injection, the lifetime explicitly accounts for the two-stage electron
and hole capture process, and the lifetime is correspondingly longer. This is due to
the equal densities of holes and electrons. In contrast, the low-injection lifetime
was determined just by hole capture, due to the more rapid electron capture
process resulting from the much higher majority carrier density of electrons.6
As with low-level injection, once the recombination level is outside the energy
interval bracketed by the carrier quasi-Fermi levels, the lifetime will increase
exponentially with this separation.
2.4 Current Flow Equations
The purpose of this section is to introduce the concepts underlying the modelling
of current flow in semiconductor devices, so that the reader has an appreciation of
the equations used in device simulation packages. As explained in the introduction,
the numerical techniques used to solve these equations are beyond the scope of this
book, and further information on device simulation can be found in Refs. [2] and
[3].
To establish the basic equations, one-dimensional flow will be considered
initially, and then extended to show the form used in 2-D or 3-D simulation
programmes.
For current flow by carrier drift and diffusion, the total current density, JT, is
given by the sum of the electron, Jn, and hole, Jp, drift/diffusion currents [1]:
JT ¼ Jn þ Jp ð2:90Þ
6
The situation described in Sects. 2.3.4.1 and 2.3.4.2 is for n-type substrates, and, for injection
into p-type substrates, the low-level lifetime will be determined by electron capture.
2.4 Current Flow Equations 39
on
Jn ¼ qln nF þ qDn ð2:91Þ
ox
op
Jp ¼ qlp pF qDp ð2:92Þ
ox
Where F is the field, l is the carrier mobility, n and p are the free carrier
densities, and D is the diffusion coefficient. These are essentially low field equa-
tions, because, as discussed below, at high fields, carrier velocity saturation can
occur [15], requiring the lF product to be replaced by the appropriate high field
velocity. In Eqs. 2.91 and 2.92, the first term describes the flow of carriers under
the influence of an internal field, which would result from an externally applied
bias, and the second term is the carrier flow by diffusion resulting from a non-
uniform carrier distribution.
The continuity equation links these equations to local generation and recom-
bination processes by:
on oJn
¼ Gn RGR þ ð2:93Þ
ot qox
op oJp
¼ Gp RGR ð2:94Þ
ot qox
These equations show that the local rate of change of carrier concentration,
within a given volume of material, is determined by the difference in the external
carrier generation rate, G, and the internal recombination rate, RGR, plus the
difference in the carrier flow into and out of the volume. In steady state, the local
carrier concentrations are constant, and Eqs. 2.93 and 2.94 are equal to zero. The
internal recombination/generation rate, RGR, is given by Eq. 2.60 from Sect. 2.3.2,
and the external generation rate, G, could be due to optical absorption of above
band-gap light. In this case, the local optical generation rate, GO is:
Go ðzÞ ¼ aUð1 RÞ exp az ð2:95Þ
Where a is the optical absorption coefficient, U is the incident photon flux, R is
the surface reflectivity, and z is the distance in the material perpendicular to the
illuminated surface.7
Another potential source of carrier generation is high-field generation of
electron-hole pairs by impact ionisation. This is also referred to as avalanche
generation, since the generated carriers will also be accelerated by the field, and
produce further carriers by impact ionisation themselves. This is most likely to
occur with high carrier mobility semiconductors, and in high field regions of
devices, such as reverse biased junctions. Where the carrier mobility is low, due to
short mean free paths and efficient scattering, the carrier is unlikely to gain enough
energy from the field. However, where the mean free path is large, the carriers can
7
For a 1-D treatment, we require a \ z for uniform carrier generation.
be efficiently accelerated and heated by the field. If the carriers attain sufficient
energy, then lattice collisions can lead to electron-hole pair generation. The impact
ionisation generation rate, GII, is given by [1]:
GII ¼ nvn an þ pvp ap ð2:96Þ
Where vn (vp) is the electron (hole) drift velocity, an (ap) is the electron (hole)
ionisation coefficient, and:
qF FC
an;p ¼ exp ð2:97Þ
EI F
FC is the critical field, F is the field parallel to current flow, and EI is the carrier
ionisation energy (for Si this is 3.6 eV for electrons and 5.0 eV for holes). A
simpler expression [16] has also been used in simulations of TFTs [17], where the
ionisation coefficient is:
FC
an;p ¼ a1
n;p exp ð2:98Þ
F
and a? is a constant, which is determined, together with FC, from simulation fits.
The carrier velocity, in Eq. 2.96, at low fields, is given by l0F, where l0 is the
low field mobility. At high fields, velocity saturation occurs, and the following
relationship between velocity and field was found to provide a good fit to the
experimental c-Si data shown in Fig. 2.15 [15]:
l0 F
vn;p ¼ h i1=b ð2:99Þ
1 þ ðl0 F=vS Þb
Where vS is the saturated velocity (*107 cm/s for holes and electrons in Si),
and b is a constant (*2 for electrons and *1 for holes in Si).
Carrier Velocity (cm/sec)
107 I I
electrons
I I I I II II
holes
106
103 104 105 106
Electric Field (V/cm)
Fig. 2.15 Carrier velocity measurements as a function of the electric field in silicon. The solid
line is the empirical fit to Eq. 2.99 (Reprinted from [15] with permission of IEEE)
2.4 Current Flow Equations 41
Equations 2.90–2.94 contain the electrostatic field, F, which is the local gra-
dient of the potential, dV/dx, and this has to be established in a self-consistent
fashion by the simultaneous solution of Poisson’s equation:
d2 V qqðxÞ
2
¼ ð2:100Þ
dx e0 es
þ
and q ¼ p n þ Nd Na þ NTd NTa ð2:101Þ
where Nd and Na are shallow donor and acceptor dopants, and NTa and NTd are
deep acceptor and donor defects, respectively. For a TFT, the deep levels could
equally well refer to a distribution of defect levels across the band-gap.
The above equations have been presented, for simplicity, in a 1-D format, but,
for accurate device simulation, these equations need to be solved in two or three
dimensions. In that case, the coupled equations in steady state are [18]:
r:rV ¼ qq=e0 es
r:Jn ¼ qðGn RGR Þ
r:Jp ¼ qðGp RGR Þ ð2:102Þ
Jn ¼ qnln rV þ qDn rn
Jp ¼ qplp rV qDp rp
These are solved subject to a set of boundary conditions determined by the

biases applied to the device terminals, the field continuity conditions across
internal material boundaries (such as the semiconductor/dielectric interface), and
the conditions at the external boundaries of the device, where, for instance, the
field may be clamped at zero.
In the commercial simulators used for Si-based TFTs, the internal recombi-
nation/generation rate, RGR, includes additional processes to those considered in
Sect. 2.3. In particular, the treatment of carrier emission is extended to include
field-enhanced emission processes from traps, as well as band-to-band tunnelling.
Those mechanisms are discussed further in Sect. 8.5.3.
2.5 Summary
Some analytical device physics concepts, using single crystal equations, have been
introduced as a background topic to later chapters. These simple concepts are
widely used in interpreting TFT behaviour, and they provide a solid basis for
appreciating the assumptions and approximations used in that work.
The topics in this chapter have been restricted to those of most relevance to
TFTs, and have focussed, firstly, on aspects of semiconductor surface physics,
which describe the relationship between gate bias, surface space charge, and band
bending in MIS structures. This topic underpins the understanding of IGFET
behaviour, which will be the subject of Chap. 3. The second topic is electron-hole
pair recombination and generation processes through deep level centres in the
forbidden band gap. These processes play a role in establishing steady-state carrier
concentrations under injection processes, and determine reverse bias leakage
currents from junction depletion regions. Finally, there is a brief overview of the
current flow processes, which are incorporated in device simulation packages.
Appendix: Summary of Key Equations
A number of the simplified equations from the text, which can be used in basic
analytical calculations, are reproduced below. The equation numbers are retained
for quick reference back to the original derivations.
A.1 Semiconductor Surface Band Bending
Relationships between surface potential, Vs, space charge density, Qs, gate volt-
age, VG, and threshold voltage, VT.
(a) Depletion and Inversion Space Charge Density
0:5
q kT
(b) Depletion Space Charge Density
Qs ffi 2qe0 es Na Vs ð2:13Þ
(c) Gate Voltage, Surface Potential and Space Charge Density
(d) Gate Threshold Voltage
2qe0 es Na 2VF
Ci
and, from Eq. 2.5, VF = (kT/q)ln(Na/ni)
2.5 Summary 43
(e) Depletion Layer Thickness at Inversion
2e0 es 2VF
xdmax ¼ ð2:18Þ
qNa
(f) Flat-Band Voltage
VFB ¼ UMS Qieff =Ci Qss =Ci
A.2 Carrier Recombination and Generation
Relationships between trap density, NT, and capture cross section, r, and the
carrier generation/recombination rates.
(a) General Recombination and Generation Rate Expression
rvth NT ½n2i np

RGR ¼ ð2:63Þ
n þ p þ 2ni coshðET Ei ÞkT
(b) Generation Leakage Current
JR qni W=sg ð2:70Þ

(c) Generation Lifetime, sg
sg ¼ 1=ð0:5NT rvth Þ ð2:71aÞ

(d) Carrier Recombination Rate, and Recombination Lifetime, sR
RR ¼ rp vth NT Dp Dp=sR ð2:81Þ

(e) Recombination Lifetime (Low Injection Level)
sR ¼ 1=rp vth NT ð2:82Þ

(f) Recombination Lifetime (High Injection Level)
sR ¼ 1=rp vth NT þ 1=rn vth NT ð2:89Þ

References
1. Sze SM, Ng KK (2007) Physics of semiconductor devices, 3rd edn. Wiley, New York
2. http://www.iue.tuwien.ac.at/phd/klima/node8.html (Accessed Aug 2010)
3. http://www.silvaco.com/products/vwf/atlas/2D/tft/tft_03.pdf (Accessed Aug 2010)
4. Grove AS (1967) Physics and technology of semiconductor devices. Wiley, New York,
pp 278–285
5. Deal BE, Snow EH, Mead CA (1966) Barrier energies in metal–silicon dioxide–silicon
structures. J Phys Chem Solids 27(11–12):1873–1879
6. Brotherton SD, Gill A (1978) Determination of surface and bulk generation currents in low
leakage silicon MOS structures. Appl Phys Letts 33:890–892
7. Berglund CN (1966) Surface states at steam-grown silicon–silicon dioxide interface. IEEE
Trans Electron Dev 13(10):701–705
8. Castagne R, Vapaille A (1971) Description of the Si02–Si interface properties by means of
very low frequency MOS capacitance measurements. Surf Sci 28(1):157–193
9. Nicollian EH, Goetzberger A (1967) The Si–SiO2 interface—electrical properties as
determined by the MIS conductance technique. Bell Syst Tech J 46:1055
10. Hall RN (1952) Electron-hole recombination in germanium. Phys Rev 87(2):387
11. Shockley W, Read WT (1952) Statistics of the recombination of holes and electrons. Phys
Rev 87(5):835–842
12. Sah CT, Noyce RN, Shockley W (1957) Carrier generation and recombination in p–n
junction and p–n junction characteristics. Proc IRE 45(9):1228–1243
13. Lax M (1960) Cascade capture of electrons in solids. Phys Rev 119(5):1502–1523
14. Brotherton SD, Bradley P (1982) Measurement of minority carrier capture cross sections and
application to gold and platinum in silicon. J Appl Phys 53(3):1543–1553
15. Caughey DM, Thomas RE (1967) Carrier mobilities in silicon empirically related to doping
and field. Proc IEEE 55:2192–2193
16. Chynoweth AG (1958) Ionisation rates for holes and electrons in silicon. Phys Rev
109(5):1537–1540
17. Valletta A, Gaucci P, Mariucci L, Pecora A, Cuscunà M, Maiolo L, Fortunato G (2010)
Threshold voltage in short channel polycrystalline silicon thin film transistors: Influence of
drain induced barrier lowering and floating body effects. J Appl Phys 107:074505-1–074505-9
18. Guerrieri G, Ciampolini P, Gnudi A, Rudan M, Baccarani G (1986) Numerical simulation of
polycrystalline-silicon MOSFETs. IEEE Trans ED-33(8):1201–1206
Chapter 3
Insulated Gate Field Effect Transistors,
IGFETs
Abstract The metal-oxide-semiconductor field effect transistor, MOSFET, is the

most widely studied and understood IGFET, and is used here to describe key
aspects of IGFET behaviour. The topic is introduced by a physical description of
MOSFET operation, which identifies the linear and saturation operating regimes.
A simple analytical model is developed from this, yielding expressions for
important parameters such as threshold voltage, saturation voltage and carrier
mobility. These are extensively used in TFT analysis. A more rigorous analysis is
presented to explicitly include the role of substrate doping, and its effect upon
saturation voltage. These descriptions are valid for the on-state regime, in which
the gate bias, VG, is greater than the threshold voltage, VT. A further operating
regime is described for VG \ VT, which is referred to as the sub-threshold regime.
In this regime, the current increases exponentially with gate bias, and is charac-
terised by the sub-threshold slope. This is another concept, and parameter, which is
widely used in the analysis of TFT behaviour. Finally, the role of film thickness, in
thin film devices, in modifying the standard expressions for threshold voltage and
saturation voltage is presented.
3.1 Introduction
This chapter will follow the convention of Chap. 2, and use single crystal equa-
tions to deal analytically with IGFET operation. The analysis will be based upon
the most widely studied IGFET, the c-Si metal-oxide-semiconductor, MOSFET,
whose concepts are widely invoked in analysing TFT behaviour. The basic device
equations, and underlying assumptions, are extensively used to extract parameters,
such as the threshold voltage, VT, the saturation voltage, VD(sat), and the carrier’s
field effect mobility, from TFT measurements.
In Sect. 3.2, the basic MOSFET behaviour is described in terms of the
dependence of the channel current, Id, on gate and drain bias, which leads to the
definition of the linear and saturation current regimes. The three terminal current-

46 3 Insulated Gate Field Effect Transistors, IGFETs
voltage equations are developed in the classical, simplified form in Sect. 3.3.1, and
then extended to take account of substrate doping effects in Sect. 3.3.2. These two
sections deal with the on-state behaviour of the device, when the gate voltage, VG,
is larger than the threshold voltage, VT.
In Sect. 3.4, the current flow equations are considered for the regime where
VG \ VT, which is defined as the sub-threshold regime. Operation in this regime is
characterised by the sub-threshold slope, S, which is given by dVG/dlog(Id).
Finally, in Sect. 3.5, the way in which the MOSFET concepts need to be modified,
to take account of the limited film thickness in thin film devices, is discussed.
Some of the key equations from this chapter are listed in the appendix for future
reference.
3.2 MOSFET Operation
A schematic diagram of an n-channel MOSFET is shown in Fig. 3.1, in which there

are n+ doped source and drain regions, separated by a distance L, which defines the
channel length. In this particular case, the edges of the source and drain regions are
coincident with the gate edges, which is the standard MOSFET architecture. This
architecture is referred to as self-aligned, and is achieved by using the gate electrode
as a mask for doping the source and drain regions by ion implantation. The width of
the channel, in the z-direction (which is perpendicular to the page), is W. Surface
band bending, perpendicular to the Si/SiO2 interface, is in the x-direction.
In the on-state, positive biases VD and VG are applied to the drain and gate
terminals, respectively, with respect to the earthed source terminal. There is a
fourth terminal in a MOSFET, which is the p-type substrate connection, which,
under reverse bias, will increase the transistor’s threshold voltage, but, for the
following discussion, it will be taken as earthed, and will have no direct effect
upon the device’s behaviour. For TFTs, this connection to the device layer will
generally not be available.
The device model to be described is the long/gradual channel model, in which
the field along the channel (driving the on-current) is much smaller than the
vertical field (determining the inversion layer concentration), and the two are
effectively decoupled, so that fundamental 2-D phenomena, such as short channel
effects, are not incorporated. The band bending and energy level conventions are
those used in Sect. 2.2, i.e. all energies and potentials in the semiconductor are
measured from the intrinsic level, and are positive for levels beneath the intrinsic
level (see Fig. 2.1 for further details).
Consider first the situation shown in Fig. 3.2a, in which there is an inversion layer
induced by the gate bias, VG (which is larger than the threshold voltage, VT), and, at
zero drain bias, the inversion layer will be uniform along the channel. As discussed
in Chap. 2 (Eqs. 2.17–2.19), the areal charge densities within an inverted surface can
be represented by the free electron inversion charge density, Qn, and the underlying
ionised acceptor space charge density, Qb, where Qn, Qb, and VT are given by:
3.2 MOSFET Operation 47
Fig. 3.1 Schematic cross section of a MOSFET, with the gate voltage larger than the threshold
voltage, and showing the surface inversion and depletion layers. The channel length is L, and the
width, W, is perpendicular to the page
Fig. 3.2 Cross-section of MOSFET, in the on-state, for different values of drain bias, VD:
a VD = 0, b VD \ VG - VT, c at pinch-off, with VD = VG - VT = VD(sat), and d VD [ VD(sat),
showing channel shortening. (Gate and drain biases are applied with respect to the grounded
source terminal)
Qn ¼ Ci ðVG VT Þ ð3:1aÞ
p
Qb ¼ qNa xdmax ¼ ð2qe0 es Na 2VF Þ ð3:1bÞ
p
VT ¼ 2VF þ ð2qe0 es Na 2VF Þ=Ci ð3:1cÞ
Ci is the capacitance of the gate dielectric, Na is the substrate doping concen-
tration, xdmax is the maximum width of the equilibrium surface depletion layer at
inversion, and qVF is the Fermi level position relative to the intrinsic level in the
semiconductor substrate.
Positive bias, +VD, applied to the drain terminal will reverse bias the n+ drain
region with respect to the p-type substrate, and the current flow between the two is
limited to the leakage current of the junction. In high quality structures, this
leakage current will be negligible. In contrast to this low current, the drain bias
will also promote the flow of an electron drift current, in the electron rich inversion
layer, between the sources and drain terminals. At the drain end of the channel, the
electrons will be swept through the reverse biased drain space charge region into
the drain junction.
When the drain bias, VD, is low (and significantly less than VG - VT), then VD
will be uniformly dropped along the channel inversion layer, resulting in both a
constant field VD/L and the flow of an ohmic electron current, Id, between the
source and drain contacts. (Due to the n+ doping of these regions, it is assumed that
there is negligible voltage drop within these regions themselves). This is defined as
the linear regime of device behaviour.
As VD is increased, the ohmic channel current increases, and, eventually, VD
reaches a value where, although it is still less than VG - VT, it is of the same order
(e.g.[*5–10%). As this happens, the potential at the drain end of the channel needs
to be taken into account when computing the inversion charge density along the
channel. For instance, at the drain end of the channel, the voltage drop across the
oxide between the inversion layer and the gate electrode is now VG - VT - VD,
whereas it remains at VG - VT at the source end of the channel. Hence, there is a
progressive reduction in Qn along the channel, and, to maintain current continuity,
there will be corresponding field redistribution along the channel, with it increasing
more at the drain than at the source as VD is increased. The current now no longer
increases linearly with VD, but becomes sub-linear, and the device behaviour moves
from the linear regime. This situation is illustrated in Fig. 3.2b. Moreover, in order to
maintain charge neutrality between the charge on the gate, QG, and the charge in the
semiconductor, the reduction in Qn will be balanced by an increase in Qb, such that
the thickness of the ionised acceptor space charge layer increases beyond the thermal
equilibrium value of xdmax (Eqs. 3.1b, and 2.18). This is shown qualitatively, in
Fig. 3.2b, by the progressive widening of the space charge region beneath the
channel. As shown by Eqs 3.1b and 2.18, the value of xdmax is given by:
2e0 es 2VF
xdmax ¼ ð3:2Þ
qNa
3.2 MOSFET Operation 49
Hence, for the depletion width to increase beyond this value, the amount of
band bending has to increase beyond 2VF, and it progressively increases along the
channel from 2VF at the source to 2VF ? VD at the drain. The qualitative changes
in the charge distributions and the band bending, at the source and drain ends of
the channel, are shown in Fig. 3.3a–d. As shown in Fig. 3.3d, the bias at the drain
results in a non-thermal equilibrium situation, with the thermal equilibrium Fermi
level splitting into separate hole and electron quasi Fermi levels, EFp and EFn,
respectively, which are separated by VD. Hence, this diagram shows physically
why the band bending, Vs, has to be increased to 2VF ? VD at the drain: it is to
bring the intrinsic level at the surface to qVF below the electron quasi-Fermi level
at the drain [1, 2]. This is the condition required to invert the p-type surface of a
reverse-biased gated n+-p diode [2], and is the direct analogue of the thermal
equilibrium situation (discussed in Sect. 2.2), in which inversion occurs when the
intrinsic level is qVF below the equilibrium Fermi level.
Finally, as shown in Fig. 3.2c, when VD = VG - VT, Qn is reduced to zero at
the drain, and channel is pinched-off. However, as all electrons arriving at the edge
of the drain space charge region are swept through it into the drain, this does not
pinch-off the current, but causes it to saturate, and the device operation moves into
the saturation regime. The voltage at which this happens is the saturation voltage,
VD(sat). The current saturation occurs because, in principle, the maximum possible
potential difference, VD(sat), has been dropped along the channel, and the pinch-off
voltage remains at VD(sat) even when VD is increased. However, in reality, as VD is
increased, the drain space charge region grows, and the pinch-off point moves
towards the source. Hence, even though the total voltage drop along the channel
remains at VD(sat), the effective channel length shortens, and the mean channel field
increases. This is shown in Fig. 3.2d, and this increase in field leads to a corre-
sponding increase in the saturation current. Since pinch-off occurs at VD =
VD(sat) = VG - VT, increasing VG leads to a corresponding increase in VD(sat).
This causes the saturation current to increase as (VG - VT)2, due to the correlation
of VD(sat) with VG. The detailed voltage dependence of the current is discussed in
the following section.
The experimental MOSFET output characteristics [1] in Fig. 3.4 illustrate
the key features described above: in particular, the linear and saturation regimes,
the increasing values of the saturation voltage, VD(sat), with increasing VG, and the
finite output impedance in saturation.
3.3 Current-Voltage Equations
3.3.1 Simplified Format
With reference to Fig. 3.2b–d, the current density, JD, at any point x, y in the
channel is given by:
Fig. 3.3 Inversion, Qn, and depletion layer, Qb, charge densities at different positions in the
channel, for VD less than, but of the same order as, VG - VT: a at the source, b at the drain.
Corresponding band bending diagrams at (c) the source, and (d) drain ends of the channel
Fig. 3.4 Experimental

MOSFET output
characteristics showing the
linear and saturation
operating regimes, as well as
the finite output impedance
(W/L = 25). (Reproduced
from [1] with permission of
John Wiley & Sons, Inc)
JD ¼ rn ðx; yÞFðyÞ ð3:3Þ

where y is the direction of current flow, x is the direction perpendicular to the Si/
SiO2 interface, rn is the local electron conductivity, and F is the local field in the
direction of current flow. The conductivity, rn, is given by:
rn ðx; yÞ ¼ qln nðx; yÞ ð3:4Þ
3.3 Current-Voltage Equations 51
The current Id(y), at position y, is given by integrating the electron density, n,

over the depth of the channel (from the Si/SiO2 interface to the point xi at which
n = ni), and across the width of the channel, W, i.e.
Zxi Zxi
dVðyÞ
Id ðyÞ ¼ W rðx; yÞFðyÞdx ¼ W qln nðx; yÞ dx
dy
0 0
ð3:5Þ
Zxi
dVðyÞ
Id ðyÞ ¼ qln W nðx; yÞdx
dy
0
where it is assumed that the electron mobility is independent of the field in both
the x and y directions, and V(y) is the potential in the channel at point y. As
discussed in Sect. 2.2, for an inverted surface, the areal charge in the inversion
layer can be represented by Qn, where Qn is the integral of the volume electron
concentration through the depth of the channel, i.e.
Zxi
q nðx; yÞdx ¼ Qn ðyÞ ð3:6Þ
0
and Qn is related to the gate bias, VG, by:

Qn ðyÞ ¼ Ci ½VG VT VðyÞ ð3:7Þ
Substituting Eqs. 3.6 and 3.7 into Eq. 3.5 gives:
dV dV
Id ¼ ln W Qn ðyÞ ¼ ln WCi ½VG VT VðyÞ ð3:8Þ
dy dy
Integrating y along the channel from 0 to L, and V(y) from 0 to VD:
ZL ZVD
Id dy ¼ ln WCi ½VG VT VðyÞdv ð3:9Þ
0 0
For current continuity, Id is independent of position within the channel, hence:

ln WCi
Id ¼ ðVG VT ÞVD 0:5VD2 ð3:10Þ
L
This is the classical, simplified MOSFET equation, which is widely used to
interpret TFT behaviour, and, in particular, to extract the carrier mobility, as
discussed below.
3.3.1.1 Linear Regime
For VD VG - VT, Eq 3.10 reduces to:

ln WCi ðVG VT ÞVD
Id ¼ ð3:11Þ
L
Equation 3.11 describes the current-voltage characteristics in the linear regime
(in which the current increases linearly with VD), and the electron mobility is
derived from the slope of an experimentally measured Id - VG transfer charac-
teristic, using:
L dId
ln ¼ ð3:12Þ
WCi VD dVG
An example of a linear regime plot is shown in Fig. 3.5, in which a straight line
has been put through the data points having the maximum slope, from which the
mobility can be obtained. The extrapolation of this line onto the x-axis gives the
threshold voltage, VT, and this procedure is widely used in extracting these two
parameter values from TFT transfer characteristics.
The transconductance of the device, gm, is given by:
dId l WCi VD
gm ¼ n ð3:13Þ
dVG L
and the channel conductance, gd, is:
dId l WCi ðVG VT Þ
gd ¼ n ð3:14Þ
dVD L
The mobility derived from Eq. 3.12 is the field effect mobility, and the above
analysis implicitly assumes that all the charge induced in the semiconductor sur-
face by the gate bias, VG, is composed entirely of free electrons, which contribute
Fig. 3.5 Experimental,

linear regime transfer
characteristic, showing the
best fit straight line (from
which the carrier mobility can
be calculated), and the
extrapolation of the line onto
the x-axis to give VT
to the current as specified by Eq. 3.8. However, using Eq. 2.22, the general
relationship between VG and the charge in the semiconductor surface is:
VG ¼ Vs Qs =Ci ¼ 2VF ðQn þ Qb Þ=Ci ð3:15Þ
and differentiating by VG gives:

dQn dQb
¼ Ci ð3:16Þ
dVG dVG
Hence, if dQb/dVG is zero, then all the induced charge is, indeed, free inversion
layer charge (as assumed above for a MOSFET). But, as will be discussed in the TFT
chapters, the TFTs are usually found to have a distribution of trapping states across the
band gap, such that there may be continuous partitioning of the induced charge
between free carriers and carriers being trapped in the band-gap states. In this case,
dQb/dVG is not zero, and dQn/dVG \ Ci, which means that the field effect mobility
(derived from Eq. 3.12, by assuming that all the induced surface charge is free charge)
will be less than the actual free carrier mobility. This is discussed further in Chap. 6
3.3.1.2 Saturation Regime
For VD = VG - VT = VD(sat), the current saturates at Id(sat), and Eq. 3.10 reduces
to:
2
ln WCi ðVG VT Þ2 ln WCi VDðsatÞ
IdðsatÞ ¼ ð3:17Þ
2L 2L
Hence, in saturation, the current increases quadratically with VG - VT because
the inversion charge increases by this amount, as does the maximum potential drop
along the channel. As in the linear regime, the field effect mobility may be
obtained from experimental Id - VG data in saturation (such as measurements at
VD = 5 V for the data in Fig. 3.4, for example), by taking the slope of the HId -
VG plot, and using the relationship in Eq. 3.17 to give:
pffiffiffiffi2
2L d Id
ln ¼ ð3:18Þ
WCi dVG
The carrier mobility is once again the field effect mobility, and the same
qualifications, as discussed in the linear regime, apply to the mobility extracted in
the saturation regime. Also, as in the linear regime, the intercept with the x-axis of
the line used to calculate the mobility gives the threshold voltage.
The saturation regime transconductance is:
ln WCi ðVG VT Þ
gmðsatÞ ¼ ð3:19Þ
L
3.3.2 Full MOSFET Equation
In Eq. 3.7, the role of the substrate doping density was implicitly ignored in
relating the channel potential to the gate bias, which is equivalent to taking the
substrate doping density as zero. For a realistically doped substrate, the doping
density needs to be taken into account, and Eq. 3.7 should be rewritten as:
Q n ð yÞ ¼ Q s ð yÞ Q b ð yÞ ð3:20Þ
Where, from Eq. 2.13, Qb(y) is
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Qb ðyÞ ¼ 2qe0 es Na Vs ðyÞ ð3:21Þ
From Eq. 2.22, the relationship between the gate voltage and the surface
potential, Vs, is:
VG ¼ Vs ðyÞ Qs ðyÞ=Ci ð3:22Þ
For VG [ VT, at any point in the channel, Vs(y) is given by:
Vs ðyÞ ¼ 2VF þ VðyÞ ð3:23Þ
Substituting Eqs. 3.21–3.23 into Eq. 3.20:

Qn ðyÞ ¼ Ci ðVG f2VF þ VðyÞgÞ ½2qes e0 Na ð2VF þ VðyÞÞ0:5 ð3:24Þ
Substituting this into the channel current Eq. 3.8, and integrating y from 0 to L,
and V(y) from 0 to VD gives:
ZVD n o
Id L ¼ ln W Ci ðVG f2VF þ VðyÞgÞ ½2qes e0 Na ð2VF þ VðyÞÞ0:5 dV
0
ð3:25Þ
(
)
ln WCi VD 2 ½2qes e0 Na 0:5 h 1:5 1:5
i
Id ¼ VG 2VF þ VD ðVD þ 2VF Þ ð2VF Þ
L 2 3 Ci
ð3:26Þ
3.3.2.1 Linear Regime
For VD VG - VT, and expanding (VD ? 2VF)1.5 as a power series, in which

quadratic and higher powers of VD are ignored:
ðVD þ 2VF Þ1:5 ffi ð2VF Þ1:5 þ 3=2VD ð2VF Þ0:5 ð3:27Þ

and Eq. 3.26 reduces to:

ln WCi l WCi
Id ¼ ðVG VT ÞVD 0:5VD2 ffi n ðVG VT ÞVD ð3:28Þ
L L
In other words, in the linear regime, Eq. 3.26 becomes the same as the simple
expression 3.11, and Eq. 3.28 is equally valid for extracting the linear regime field
effect mobility. The commonality of Eqs. 3.11 and 3.28 is not surprising given that
it has been argued that the effects of the substrate doping level only become
significant when VD is comparable to VG - VT, causing the value of Qb to start to
increase along the channel.
3.3.2.2 Saturation Regime
The saturation voltage can be determined from Eq. 3.24 by putting Qn(L) = 0
when V(L) = VD(sat) [1]. This directly gives the following expression for VD(sat):
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
2qes e0 Na VDðsatÞ þ 2VF
VDðsatÞ ¼ VG 2VF ð3:29Þ
Ci
Re-arranging Eq. 3.29, and solving for VD(sat) gives:
VDðsatÞ ¼ VG 2VF þ K 2 1 1 þ 2VG =K 2 ð3:30Þ
where,
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
qes e0 Na
K¼ ð3:31Þ
Ci
Equation 3.29 shows that VD(sat) is smaller than VG - VT due to the inclusion
of VD(sat) itself in the square root term (without this inclusion, VD(sat) would just be
equal to VG - VT). A direct consequence of the reduction in VD(sat) is that the
saturation current, at a given value of VG - VT, will be smaller than the value
given by Eq. 3.17 using the simple analysis.
The physical meaning of Eq. 3.29 can be more easily understood by re-
arranging it in terms of the value of the gate bias, VG(sat), at which the channel will
just be pinched off at a given value of VD (for VG greater than this value, the
channel will not be pinched off) i.e:
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2qes e0 Na ðVD þ 2VF Þ
VGðsatÞ ¼ 2VF þ VD þ ð3:32Þ
Ci
In other words, VG(sat) is the gate threshold voltage for inverting the p-type
surface adjacent to an n+ p junction, which is reverse biased with a potential VD.
This also shows that the larger the value of Na, the greater VG(sat) is for a given
value of VD (and conversely, the smaller VD(sat) is for a given value of VG). This is
just an extension of the thermal equilibrium threshold voltage condition discussed
in Sect. 2.2.2, in which the threshold voltage increased with HNa. What has been
added here is that, in the presence of a reverse biased junction, the band bending
for inversion is no longer just 2VF but 2VF ? VD.
The quantitative impact of the substrate doping level, Na, on VD(sat) is shown in
Fig. 3.6, in which VD(sat) is plotted as a function of VG - VT for different dopant
concentrations (and for a gate oxide thickness of 40 nm). A line of unity slope
(VD(sat) = VG - VT) has been plotted in order to indicate the displacement of
VD(sat) from the simple condition of VG - VT, and only the substrate doped with
1015 acceptors/cm3 is a close fit. In contrast, for a substrate doped with 1017
acceptors/cm3, VD(sat) is only 4.9 V when VG - VT is 8 V; hence, there is a
substantial reduction in the saturation voltage at this substrate doping level, and a
corresponding reduction in the saturation current, as shown in Fig. 3.7. This figure
plots the saturation currents, calculated from Eqs. 3.26 and 3.30, for substrates
doped with 1015 and 1017 acceptors/cm3, in which the more heavily doped sub-
strate shows a smaller saturation current at any given value of VG. Part of this
difference is due to the higher threshold voltage in the 1017 cm-3 doped substrate,
but, even after allowing for this, the current is still 38 % less at VG - VT=10 V.
However, the quadratic dependence of Id(sat) on VG is still maintained, as shown by
the plots of HId(sat), due to the continued scaling of VD(sat) with VG, albeit at a
lower rate as Na increases. This is apparent from the reducing slopes of the
HId(sat) - VG plots in Fig. 3.7, and, if Eq. 3.18 is used to extract the field effect
mobility from these plots, the apparent mobility decreases with increasing sub-
strate doping. (For the curves in Fig. 3.7, a mobility of 500 cm2/Vs was used to
calculate the saturation currents, but the extracted values were 469 and 316 cm2/
Vs for the 1015 and 1017 cm-3 doped substrates, respectively). Hence, the simple
analysis underlying Eq. 3.18 is likely to underestimate the carrier mobility in
saturation.
As shown diagrammatically in Sect. 3.2, the space charge region beneath the
conducting channel significantly widens between the source and the drain ends of
the channel under drain saturation conditions. Equation 3.2 can be modified to
include the influence of the drain saturation voltage on the space charge width,
xdD, at the drain end of the channel, which is given by:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

2e0 es 2VF þ VDðsatÞ
xdD ¼ ð3:31Þ
qNa
Figure 3.8 is a plot of xdD as a function of VD(sat) for different substrate doping
levels (where the zero bias value of xdD is also the space charge width at the source
end of the channel). For Na \ 1017 cm-3, the surface space charge width at the
drain is greater than 100 nm, which is significant for TFTs with film thickness
values less than this, as discussed in Sect. 3.5.
Fig. 3.6 Calculated values of the drain saturation voltage, VD(sat), as a function of VG - VT, for
different substrate doping levels, and for a gate oxide thickness of 40 nm. (The line of slope S=1
illustrates the divergence of VD(sat) from the simple relationship VD(sat) = VG - VT)
Fig. 3.7 Calculated saturation regime Id(sat) - VG characteristics, for different substrate doping
levels, and the associated plots of HId(sat) vs. VG (W/L = 1, tox = 40 nm)
3.3.3 Non-ideal MOSFET Behaviour
The preceding discussions have assumed that the threshold voltage is determined
just by the substrate doping level, so that its numerical value is positive for n-channel
TFTs and negative for p-channel TFTs. That is to say both channel types are in the
off-state at zero gate bias. This type of transistor is referred to as an enhancement
device. However, as discussed in Sect. 2.2.3, real MIS structures will have some, or
Fig. 3.8 Surface depletion

layer thickness, xdD, at the
threshold of surface
inversion, as a function of
substrate doping level, and at
different values of VD(sat)
all, of the following artefacts: work function differences, UMS, between the metal
gate and the semiconductor, fixed charges in the dielectric, Qieff, and charges,
Qss(Vs), in the interface states, Nss(E), at the semiconductor/dielectric interface.
These will produce a non-zero flat band voltage, VFB, given by Eq. 2.38b:
VFB ¼ UMS Qieff =Ci Qss ðVs ¼ 0Þ =Ci ð3:32Þ
which needs to be accounted for in the calculation of the device’s threshold

voltage, given by Eq. 2.28:
2qe0 es Na 2VF
Ci
If VFB is sufficiently negative (due, for instance, to a large positive fixed charge
density in the dielectric), and larger than the other two terms in Eq. 3.33, then the
threshold voltage for an n-channel device, will be negative, and the device will be
in the on-state at zero gate bias. This is referred to as a depletion device. (A
depletion p-channel device would have a positive threshold voltage).
The threshold voltage shift due to the work function difference and the fixed
oxide charge will both simply displace the MOSFET on-state Id - VG charac-
teristic along the gate voltage axis without changing its shape. In addition, for
fixed charge alone, the shift will be the same for both n- and p-channel transistors.
However, for interface states, whose occupancy changes with band bending, and,
hence, with gate bias above VT, their effect will be to reduce the slope of the linear
Id - VG characteristics for both channel types, so that they are displaced away
from each other. The analysis of those curves using Eq. 3.12 will show an apparent
reduction in carrier mobility, due to the violation of the underlying assumption in
this equation that all charge induced by the gate voltage above VT is free charge in
the channel. (In the presence of interface states, the induced charge is partitioned
between the free carriers in the channel and carriers trapped up to the Fermi level
in the interface states).
3.4 Sub-Threshold Currents
The MOSFET equations developed above have been derived for gate voltages in
excess of the threshold voltage, which has been defined to occur at a band bending
value of 2VF (where the volume concentration of minority carriers at the surface is
equal to the volume concentration of the majority carriers in the bulk). In addition,
for band bending greater than 2VF, the surface charge is dominated by the free
carriers, whereas, for band bending less than this, the surface charge is dominated
by the ionised dopant space charge. However, within this latter regime, the
minority free carrier density at the surface is increasing exponentially with band
bending by several orders of magnitude from n2i /Na, at flat bands, to Na at the
threshold voltage (assuming an n-channel device, with the substrate doped with an
acceptor density Na). The sub-threshold regime describes the contribution of these
carriers to current flow in the device for gate voltages less than the threshold
voltage.
To a first order in the gradual channel approximation, the gate-bias-induced
fixed space charge screens the free carriers from the drain field, so that the drift
current is assumed to be zero. However, free electrons adjacent to the positively
biased drain region will be swept through the drain space charge region, and into
the drain. If the drain space charge region is regarded as a perfect electron sink,
then the volume surface electron concentration, ns, at the drain will be approxi-
mately zero, and independent of VD for VD [ 3kT/q [3, 4]. At the source, it will be
n2i /ps, where ps = Naexp - Vs/kT, and Vs is the surface band bending. Hence,
there will be a carrier gradient along the channel, driving a sub-threshold diffusion
current, Id, which, for current continuity, requires the gradient to be constant. The
diffusion current, Id, is given by [3]:
Id ¼ qQsn WDn =L ð3:34Þ
where Qsn is the areal electron concentration at the source end of the device, and
Dn is the electron diffusion coefficient, which is given by the Einstein relationship
as lnkT/q. The areal electron concentration, Qsn, is the integral (in the x-direction)
of the electron concentration through the depletion region, i.e.:
Z xd Z 0
dx
Qsn ¼ nðxÞdx ¼ nðVÞ dV ð3:35Þ
0 Vs dV
Due to the band bending, V(x), within the depletion region, the electron con-
centration rises exponentially from its value n2i /Na at the edge of the space charge
region (x = xd and V = 0) to the following value at any point x in the space
charge region, where the band bending is V(x):
nðVÞ ¼ ðn2i =Na Þ exp qVðxÞ=kT ð3:36Þ

The field, dV/dx, can be derived from Eq. 2.13 as:
rffiffiffiffiffiffiffiffiffiffiffiffiffiffi
dV 2qNa V
¼ ð3:37Þ
dx e0 es
Hence, substituting Eqs. 3.36 and 3.37 into 3.35:
rffiffiffiffiffiffiffiffiffiffi Z 0
n2i e0 es 0:5
Qsn ¼ V exp qV=kT dV ð3:38Þ
Na 2qNa Vs
For 2VF [ Vs [ kT, an approximate analytical solution to Eq. 3.38 has been
obtained by integrating just the dominant exponential term, and evaluating V-0.5
at the integration limits [4], giving:
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
kTn2i e0 es qVs
Qsn exp ð3:39Þ
qNa 2qNa Vs kT
Substituting Eq. 3.39 into 3.34 gives the sub-threshold current:
qln W kT 2 n2i e0 es qVs
Id exp ð3:40Þ
L q Na 2qNa Vs kT
This can be related to the gate bias, VG, by Eqs. 2.13 and 2.22:
2qe0 es Na Vs
VG ¼ V s þ ð3:41Þ
Ci
Equations 3.40 and 3.41 can be evaluated as a function of Vs, and the resulting
Id - VG characteristics rise exponentially with VG [3, 4] as shown in Fig. 3.9. The
results in this figure were calculated for different values of Na, and display two
important features. Firstly, that the currents rise with an approximately constant
slope over several orders of magnitude, and, secondly, that this slope decreases
with increased substrate doping level. Given these features, the sub-threshold
regime is characterised by the inverse slope, S, of the log Id - linear VG plot,
which has the units of volts/decade, and is the number of volts required to change
the current by one decade. This procedure is widely used in the characterisation of
TFTs, and, for a given value of the gate capacitance, Ci, the higher the device
quality, the smaller is S.
The physical significance of this measurement can be seen by differentiating
Eq. 3.40 by VG:
d log Id d log Id dVs
¼ : ð3:42Þ
dVG dVs dVG
3.4 Sub-Threshold Currents 61
where, from Eq. 3.41:

dVG 1 qe0 es Na
¼1þ ð3:43Þ
dVs Ci 2Vs
The capacitance of the depletion layer, Cs, is given by:
e0 es qe0 es Na
Cs ¼ ¼ ð3:44Þ
xd 2Vs
Hence,
dVG Cs
¼1þ ð3:45Þ
dVs Ci
From Eq. 3.40:
rffiffiffiffiffiffiffiffiffiffi
log Id 0:5 log Vs þ log10 e ð3:46Þ
L q Na 2qNa kT

d log Id q 1 Cs 1
: 1þ ð3:47Þ
dVG kT ln 10 Ci
where the logVs term has been ignored in differentiating Eq. 3.46, and log10e has
been replaced by 1/ln10 in Eq. 3.47. The inverse slope, S, is:

dVG kT Cs
S 1þ ln 10 ð3:48Þ
d log Id q Ci
From Eq. 3.48, the inverse slope, S, will increase with Cs, and, hence, with Na,
as seen in Fig. 3.9. It will decrease with increasing Ci, and, hence, with reducing
dielectric thickness, reaching a limiting minimum value of 59.5 mV/dec when
Ci Cs. Conversely, for Ci Cs, such that S 59.5 mV/dec, S can be
approximated by:
Fig. 3.9 Calculated values

of the sub-threshold current,
Id, as a function of gate bias,
for different substrate doping
levels, and for a gate oxide
thickness of 40 nm. The
dashed lines are linear fits
over the plotted data ranges,
from which the S values are
72, 90 and 143 mV/dec for
Na=1015, 1016 and 1017 cm-3,
respectively. (W/L=1)
kT Cs
S ln 10 ð3:49Þ
q Ci
and S will scale with HNa. With high quality MOSFETs, Eq. 3.48 is likely to be the
most appropriate expression, but, with the lower quality TFTs, containing a bulk
distribution of trapping states across the band gap, Eq. 3.49 is a useful approximation.
In this case, the value of Cs itself will be determined by the occupancy of trapping
states within the forbidden band gap, and can be considered analogous to the fixed
space charge density responsible for Cs in Eq. 3.44. However, in contrast to the
situation with the substrate dopant, Na, in MOSFETs (which gives a constant space
charge density through the depletion region), the local density of trapped charge in a
TFT will be a function of potential within the surface space charge region. Because of
this, the trapping state density cannot be as readily extracted from measured values of
S, but, nevertheless, the measured values of S are widely quoted as figures of merit for
the TFT. The smaller S is, the higher the TFT quality is. Needless to say, in making any
such comparison across different devices, the S values will need to be normalised by
any differences in the dielectric capacitance between these devices.
The sub-threshold slope is also affected by interface states. In the sub-threshold
regime, the surface potential Vs is changing from *0 to *2VF, and, in the presence
of interface states, Nss(E), their charge occupancy will also be changing over this
voltage range, as they are moved below the Fermi level. Hence, for a given change in
VG, part of the induced surface charge will be trapped by the interface states, leaving
a reduced amount of free charge, and concomitantly reduced band bending. The
reduced amount of free charge will reduce the sub-threshold current, and S will
increase. This can be seen when the influence of interface states is included in
Eq. 3.48. The charging of interface states was discussed in Sect. 2.2.4, where it was
shown that the interface states contribute a capacitance, Css, in parallel with the
space charge capacitance. This capacitance was given by Eq 2.41:
dQss
dVs
Hence, in the presence of interface states, Eq. 3.48 changes to:

dVG kT Cs þ Css
S 1þ ln 10 ð3:51Þ
d log Id q Ci
This equation shows how S will increase with increasing interface state density,
and, providing the other parameter values are known, Nss can be extracted from S.
3.5 Thin Film Considerations
The MOSFET discussion in the previous sections has implicitly assumed that the
substrate is sufficiently thick that there is no constraint on the thickness of the surface
depletion regions. In this section, we will consider the constraints imposed by the
3.5 Thin Film Considerations 63
finite film thickness in a TFT on the device threshold voltage and saturation voltage.
In using the preceding analytical equations to examine the TFT behaviour, an analogy
will be drawn between the shallow dopant impurity in the MOSFET and the defect
states distributed across the band gap in the TFTs. However, it should be emphasised
that there is not a precise equivalence because the shallow dopant in the MOSFET is
fully ionised throughout the surface space charge region, and the space charge density
is simply equal to the dopant density. In contrast, the space charge density in the TFT
is determined by the position of the Fermi level within the defect state distribution,
and, hence, by the local band bending, V(x). In this case, the space charge density is a
function of depth, x, and cannot be specified by a single value, in contrast to the
dopant density, Na, used with the MOSFET. Nevertheless, the simplicity of the
analytical equations makes them useful in providing a physical insight into the broad
effects of the finite film thickness in TFTs. (A more rigorous discussion of the detailed
nature and impact of the TFT trapping states is presented in Chap. 6).
3.5.1 Threshold Voltage
For the thick substrate devices, the threshold voltage is defined by Eq. 3.1c:
p
VT ¼ 2VF þ ð2qe0 es Na 2VF Þ=Ci ð3:52Þ
Where the equilibrium depletion layer thickness beneath the channel is:
2e0 es 2VF
xdmax ¼ ð3:53Þ
qNa
The dependence of xdmax on Na is shown in Fig. 3.8, by the line marked

VD = 0 V, from which it will be seen that xdmax is greater than 100 nm for substrate
doping densities less than 1 9 1017cm-3. Bearing in mind the qualifications con-
cerning precise parallels between the space charge densities in MOSFETs and TFTs,
some broad conclusions can, nevertheless, be drawn for an effective space charge
density, NTFT (such as could be extracted from the sub-threshold slope of TFTs).
For TFTs made in films thinner than 100 nm, and with effective charge den-
sities, NTFT, less than 1017 cm-3, the film will not be thick enough to accom-
modate the equilibrium space charge thickness defined by Eq. 3.53. Given that the
maximum space charge width cannot exceed the film thickness, the band bending
will be less than that conventionally used to define threshold voltage in Eq. 3.52.
To a first approximation, the threshold voltage will be determined by the gate
voltage required to fully deplete the film, and any further increase in gate bias will
induce free carriers rather than increasing the depth of the space charge region
(although some of these carriers may continue to be trapped). With this assump-
tion, the modified threshold voltage, for a film of thickness Ts, will be given by:
VT ¼ qNTFT T2s =ð2e0 es Þ þ qNTFT Ts =Ci ð3:54Þ

where, for Ts \ xdmax, the first term is the band bending required to fully deplete
the film, and the numerator in the second term is the space charge density in the
fully depleted film. If the film is not fully depleted at inversion, then the normal
threshold voltage expression, 3.52, can be applied.
Hence, for thin enough films, VT is a linear function of NTFT, as opposed to the
HNa dependence found for MOSFETs, and is also a direct function of the film
thickness.
3.5.2 Saturation Voltage, VD(sat)
As discussed in Sect. 3.3.2.2, the depletion layer thickness at the drain end of the
channel increases with VD, as shown in Fig. 3.8. Clearly, if Ts \ xdmax, at the
threshold of inversion, then the space charge layer thickness will be unable to grow
with increasing VD as required by Eq. 3.31 (and as qualitatively shown in
Fig. 3.2). Equation 3.29 can be re-arranged to read:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
2qes e0 Na VDðsatÞ þ 2VF
VG ¼ VDðsatÞ þ 2VF þ ð3:55Þ
Ci
which can be put into the more general form:
Qs
VG ¼ VDðsatÞ þ Vs ðVD ¼ 0Þ þ ð3:56Þ
Ci
However, in the fully depleted film, the maximum areal space charge density,
Qs, is determined by the film thickness, and:
Qs = qNTFTTs, and Vs(VD=0) = qNTFTT2s /(2e0es). Substituting these into
Eq. 3.56 gives:
qNTFT Ts2 qNTFT T

VG ¼ VDðsatÞ þ þ ð3:57Þ
2e0 es Ci
The last two terms on the right hand side of Eq. 3.57 are equal to VT (see
Eq. 3.54), and, therefore:
VG ¼ VDðsatÞ þ VT
or
VDðsatÞ ¼ VG VT ð3:58Þ
3.5 Thin Film Considerations 65
Hence, in TFTs, in which the film thickness is less than the equilibrium
depletion layer thickness for surface inversion, xdmax, the saturation voltage is
equal to the simple, classical MOSFET value of VG - VT. This is quite different
from the detailed analysis of realistically doped MOSFETs (where
VD(sat) \ VG - VT). The simpler expression for TFTs is a consequence of the
limited areal space charge density, which can be developed within the thin film,
even in the presence of a high volume space charge density. Indeed, the near
equality of VD(sat) and VG - VT is widely observed across the range of different
types of TFT, including a-Si:H [5], poly-Si [6], organic TFTs [7] and amorphous
oxide TFTs [8]. The most obvious example of VD(sat) \ VG - VT occurs in
source-gated transistors, SGTs, [9], due to their fundamentally different mode of
operation, and this is discussed in Chap. 12.
3.6 Summary
The published analyses of TFT behaviour frequently make use of simple, ana-
lytical MOSFET concepts, and, in this chapter, those concepts are explained, and
the key equations developed. The physical description of MOSFET operation was
used to introduce the linear and saturation operating regimes. A simple analytical
model was developed from this, and contained expressions for parameters such as
threshold voltage, saturation voltage, and carrier mobility, which are extensively
used in analysing TFT behaviour. This simple analysis ignores the implicit effects
of the substrate doping level on the saturation voltage, and a more rigorous
analysis was presented to quantify this effect. This demonstrated that the saturation
voltage is generally less than the value of VG - VT, which is predicted in the
simple model.
The above models are valid for the on-state regime, in which the gate bias, VG,
is greater than the threshold voltage, VT. A further operating regime is described in
which VG \ VT, and this is known as the sub-threshold regime. This is a low
current regime, but one in which the channel current increases exponentially with
gate bias (in contrast to the linear increase in the high-current, linear regime).
Device operation in this regime is characterised by the sub-threshold slope, S, and
this is commonly used as a figure of merit for TFT operation.
The standard MOSFET analysis implicitly assumes an infinitely thick substrate,
and some of the conclusions from that analysis need to be modified in the presence
of the finite film thicknesses in TFTs. The key parameter determining this is
whether the film is thinner than the expected equilibrium space charge width at the
point of surface inversion. When this occurs, the film thickness modifies the usual
expression for threshold voltage, and the saturation voltage reduces to the simple
MOSFET expression of VG - VT, as widely observed in most TFTs.
Appendix: Summary of Key Equations
A number of the simplified equations from the text, which can be used in basic
analytical calculations, are reproduced below. The equation numbers are retained
for quick reference back to the original derivations.
A.1 Simplified MOSFET On-State Analysis
Relationships between the drain current, and the gate and drain biases.
(a) drain current as a function of terminal biases
ln WCi
L
(b) threshold voltage
p
VT ¼ 2VF þ ð2qe0 es Na 2VF Þ=Ci ð3:1cÞ
(c) linear regime current
ln WCi ðVG VT Þ VD
Id ¼ ð3:11Þ
L
(d) linear regime mobility
L dId
ln ¼ ð3:12Þ
WCi VD dVG
(e) linear regime transconductance
dId l WCi VD
gm ¼ n ð3:13Þ
dVG L
(f) saturation regime current
2
ln WCi ðVG VT Þ2 ln WCi VDðsatÞ
2L 2L
(g) saturation regime mobility
pffiffiffiffi2
2L d Id
ln ¼ ð3:18Þ
WCi dVG
Appendix: Summary of Key Equations 67
(h) saturation regime transconductance

ln WCi ðVG VT Þ
gmðsatÞ ¼ ð3:19Þ
L
A.2 Simplified MOSFET Sub-Threshold Analysis
(a) sub-threshold current

Id exp ð3:40Þ
L q Na 2qNa Vs kT
(b) relationship between surface potential, Vs, and gate bias
2qe0 es Na Vs
VG ¼ V s þ ð3:41Þ
Ci
(c) sub-threshold slope, S

dVG kT Cs
S 1þ ln 10 ð3:48Þ
d log Id q Ci
(d) sub-threshold slope with interface states

dVG kT Cs þ Css
S 1þ ln 10 ð3:51Þ
d log Id q Ci
(e) interface state capacitance
dQss
dVs
References
1. Grove AS (1967) Physics and technology of semiconductor devices. John Wiley and Sons Inc,
New York
2. Grove AS, Fitzgerald DJ (1966) Surface effects on p-n junctions: characteristics of surface
space-charge regions under non-equilibrium conditions. Solid-State Electron 9:783–806
3. Brews JR (1979) Subthreshold behaviour of uniformly and nonuniformly doped long-channel
MOSFET. IEEE Trans ED-26 9:1282–1291
4. Taylor GW (1978) Subthreshold conduction in MOSFETs. IEEE Trans ED-25 3:337–380
5. Powell MJ (1989) The physics of amorphous-silicon thin-film transistors. IEEE Trans ED-
36(12):2753–2763
6. Bonfiglietti A, Valletta A, Gaucci P, Mariucci L, Fortunato G, Brotherton SD (2005) Electrical

characterisation of directionally solidified polycrystalline silicon. J Appl Phys 98(3):033702
7. Klauk H, Zschieschang U, Pflaum J, Halik M (2007) Ultralow-power organic complementary
circuits. Nature 445:745–748
8. Nomura K, Takagi A, Kamiya T, Ohta H, Hirano M, Hosono H (2006) Amorphous oxide
semiconductors for high-performance flexible thin-film transistors. Jpn J Appl Phys
45(5B):4303–4308
9. Shannon JM, Balon F (2007) High-performance thin-film transistors in disordered and poor-
quality semiconductors. IEEE Trans ED-54(2):354–358
Chapter 4
Active Matrix Flat Panel Displays
Abstract The currently dominant flat panel display technology is based upon the
use of liquid crystal materials to produce active matrix liquid crystal displays,
AMLCDs. Other display effects, which are now appearing in commercial products,
are electrophoretic displays, EPDs, in e-readers, and organic light emitting
diodes, OLEDs, in some small diagonal, portable displays. The operation of all
three types of display are described, although prominence is given to discussion of
the operating principles of AMLCDs. The topics covered are LC effects, active
matrix addressing, pixel layout and a brief overview of display performance
artefacts. The treatment of EPDs and OLEDs also covers their operating princi-
ples, aspects of pixel layout and performance issues.
4.1 Introduction
This chapter deals with the operation and addressing of active matrix flat panel
displays, AMFPDs, and will primarily focus on liquid crystal displays. The active
matrix LCD is currently the most commercially significant display technology,
offering products across the size range from *100 portable displays through to 10000
TVs, as illustrated in Fig. 4.1. The annual revenue of the AMLCD industry in late
2011 was *$B86 [1], with a long-term annual growth rate of *10 %. Within this
market, *85 % of the revenue came from larger diagonal displays for the note-
book, monitor and TV markets, and the remaining 15 % from the small and
medium diagonal display market, for portable products such as cell phones, por-
table media players, net-books and tablets [1]. The growth of this industry has been
facilitated by the manufacturability of high quality thin film transistors for display
addressing, and, equally, the development of the industry has stimulated the world-
wide interest and research into TFTs. Among the different TFT technologies,
amorphous Si TFTs dominate the current product range, although,

70 4 Active Matrix Flat Panel Displays
Fig. 4.1 AMLCD product applications grouped by display diagonal (the dotted box for LCTV
illustrates the larger diagonal TVs demonstrated at electronics shows)
poly-Si TFTs are found in some of the small and medium-sized products. Also,
within this latter size range, two other display technologies are emerging, in
particular, electrophoretic displays, EPDs [2, 3], and organic light emitting diode
displays, OLEDs [4]. The EPDs are reflective displays, with low power con-
sumption, and are of particular relevance to e-reader devices. The current
addressing devices are primarily a-Si TFTs [5], although the e-reader has also been
identified as an appropriate vehicle for flexible displays, and for the use of organic
addressing TFTs [6]. The OLED displays demand a higher drive current than
LCDs, and the current manufacture of AMOLEDs, for products like cell-phones,
has focussed on the use poly-Si TFTs for the addressing transistors [7]. There has
also been strong research interest in using amorphous oxide TFTs (such as indium-
gallium-zinc-oxide), rather than poly-Si TFTs, for addressing large diagonal
AMOLEDs [8]. The motivation here has been the recognition that AMOLED
displays are more sensitive to TFT drive current non-uniformities than AMLCDs,
and the amorphous nature of the oxide TFTs is perceived as offering better uni-
formity than the laser crystallised poly-Si TFTs. In view of the emergence of these
alternative display technologies, the operating principles of AMEPDs and
AMOLEDs are also covered in this chapter.
Liquid crystal displays can operate in either the transmissive mode (illuminated
by a back-light), or the reflective mode (relying upon the reflection of ambient
light), or the transflective mode, where the latter is a mixture of the other two. The
transmissive mode is used in the majority of applications, particularly for note-
books, monitors and TVs, whereas the other modes are more frequently employed
in smaller portable displays, where power consumption and battery life are
important issues. The discussion in this chapter is limited to the more wide-spread
4.1 Introduction 71
transmissive mode. The basic operation of the LC cell is described in Sect. 4.2, and
the principles of active matrix addressing (as applied to LCDs) is covered in Sect.
4.3. This is followed by a discussion of pixel layout issues in Sect. 4.4, and the
principle fabrication stages of the AMLCD module are summarised in Sect. 4.5.
Finally, Sect. 4.6 describes the operating principles of EPDs and AMOLEDs.
4.2 Liquid Crystal Cells
4.2.1 LC Material
A liquid crystal material is one in which there is a temperature dependent phase,

which exhibits the structural organisation of a crystal, and the viscosity of a liquid.
The former property results in an anisotropic refractive index and birefringence,
and the latter enables the material to rapidly respond to an external influence, such
as a change in an applied field. Liquid crystal materials come in a variety of forms
and phases, and the type of specific interest for AMLCDs is composed of long-
chain molecules, as shown in Fig. 4.2a, and by the space-filling molecular model
in Fig. 4.2b [9]. This material is 40 -n-pentyl-4-cyano-biphenyl, which has a rigid
biphenyl head, to give orientational order, and a flexible hydrocarbon tail which
promotes the formation of a liquid crystal phase between the solid and liquid
phases [9]. The length and diameter of this rod-like molecule are *2 and 0.5 nm,
respectively, giving an aspect ratio of *4, which is typical for many LC materials.
A large number of different liquid crystal structures have been engineered for
Fig. 4.2 a 40 -n-pentyl-4-cyano-biphenyl (5CB) LC molecule, b space filling model of 5CB,

c 1,2,3 trifluorobenzene derivative molecule developed for TNLCs, and d temperature dependent
phases of rod-like LC molecules (a, b and d Reproduced from [9] with permission of John Wiley
& Sons, Inc, and c Reprinted from [10] with permission of SID)
ALMCD applications [10], and those currently favoured for AMLCDs contain
fluorinated polar end groups, as shown in Fig. 4.2c [10].
Figure 4.2d schematically shows the changes of phase with temperature of rod-
like LC molecules [9], from an isotropic liquid at the highest temperature, down to
a crystalline solid at the lowest temperature. Between these two phase extremes,
intermediate liquid crystal phases are shown, with the nematic phase being the
highest temperature one, and the clearing point of the material is defined as the
temperature at which the isotropic liquid forms. In the nematic phase, the mole-
cules have orientational order, but no positional order. In other words, the long
axis of the molecules is aligned predominantly in the same direction, which is
called the liquid crystal director, but there is no further ordering within this
direction. At a lower temperature, the smectic-A phase can arise, where the ori-
entational ordering of the nematic phase is preserved, but there is now some
positional ordering of the aligned molecules into layers, although there is no
further positional ordering within the layers themselves. For this phase, the liquid
crystal director is perpendicular to the layers. Some materials display further
ordering into the smectic-C phase, in which the director is now at an angle to the
layer. The nematic phase has lower viscosity than the grease-like smectic phases,
and is the preferred material for LCDs, although it should be added that the
nematic materials are synthesized compounds containing a number of constituent
mixtures rather than being a single, simple LC material. Indeed, for a given
application, the constituent parts are used to optimise the properties of the mate-
rial, such as clearing point, speed of response or reduced driving voltages etc. [10].
For the practical application of these materials in AMLCDs, the nematic LC phase
needs to be stable from *-40 C to *100 C, have high resistivity, fast response
times, specific dielectric properties, and excellent chemical stability [10].
The directional alignment of the LC molecules in Fig. 4.2d makes the material
dielectrically and optically anisotropic [9], and, in this case, the optical axis is
parallel to the direction of the LC director. Due to the anisotropy, the material has
different dielectric constants for electric fields parallel and perpendicular to the
director, and the difference in dielectric constants, De, is given by De = e//-e\.
The dielectric anisotropy is necessary for the LC director to change direction with
electric field, which is an essential aspect of the operation of the LC cell. For the
materials of interest in AMLCDs, e// and e\ are of the order *10 and *4,
respectively. Similarly, there are different refractive indices for light polarised
parallel to the optical axis, n//, and perpendicular to it, n\. Material with this
optical property is known as birefringent, and the birefringence, Dn, is given by
Dn = n// - n\ [9]. Depending upon the relative sizes of the two refractive indices,
Dn can be positive or negative, and, for the nematic materials of interest to
AMLCDs, positive Dn materials are more generally used. Birefringent materials
change the polarisation state of polarised beams of light passing through them,
when the angle of incidence is between 0 and 90 with respect to the direction of
the LC director (the optical axis). This change of polarisation is exploited in
AMLCDs, using a particular form of nematic cell structure, known as a twisted
nematic cell [11].
4.2 Liquid Crystal Cells 73
Fig. 4.3 TN LC cell

structure (a) transmitting
state, with zero applied bias
across LC, and (b) black
state, due to LC bias voltage
permission of John Wiley &
Sons, Inc)
4.2.2 Twisted Nematic LC Cell Structure
The structure of a twisted nematic, TN, cell is shown in Fig. 4.3, which consists of
a 4–6 lm thick liquid crystal layer sandwiched between two glass plates. The inner
surfaces of the plates are coated with a transparent metal conductor, such as
indium tin oxide, ITO, so that an electrical bias can be applied across the liquid
crystal. In addition, the outer surface of each ITO layer is coated with an LC
alignment layer, such as polyimide, which is textured by rubbing, in order to
anchor the rod-like nematic LC molecules to the surface in a predetermined
fashion. The outer surfaces of the glass plates are sandwiched between two
crossed, linear polariser plates, and the LC cell is illuminated with white light from
one side of the stack, and viewed through the polariser on the other side. Hence,
the light entering the cell is plane polarised, and, in the absence of the LC material,
there would be no optical transmission through the cell due to the action of the
crossed polarisers. However, as shown in Fig. 4.3a, the alignment layers are
fabricated so that the LC directors of the molecules in immediate contact with the
plates are parallel to the polarisation directions, but, because the two directions are
perpendicular to each other, a 90 twist is induced in the orientation of the LC
director through the cell. This induces a 90 rotation in the plane of polarisation of
the light passing through the cell, so that the light is transmitted by the crossed
polariser on the exiting surface of the cell, and the cell appears bright. Referring to
Fig. 4.3b, when a suitable voltage is applied across the liquid crystal, the LC
directors line up with the field, and the 90 twist is removed from the LC. In this
case, there is no longer a rotation in the plane of polarisation of the light passing
through the cell, and the light is not transmitted by the crossed polariser on the
opposite face: the cell now appears dark. When the voltage is reduced back to zero,
the LC molecules relax back to the 90 twist shown in Fig. 4.3a, and the speed of
this relaxation is a function of the molecular form of the LC and the cell thickness,
d. For a given liquid crystal material, with a birefringence Dn, the required cell
thickness, to give the 90 twist to the light in the unbiased state, is determined by
the Gooch-Tarry condition [9]:
p
dDn ¼ k m2 1=4 ð4:1Þ
where m is an integer, and the thinnest cell is given by m = 1, so that dDn is

*480 nm for k = 550 nm. Thus, for d *5 lm, the LC material needs to be
engineered to have a Dn value of *0.1.
From the above discussion of the bright and dark states, it is clear that the TN
cell acts as a voltage controlled, monochromatic light valve. Moreover, interme-
diate voltages determine the particular value of the LC twist between 0 and 90,
and, thereby, control the detailed optical transmission of the cell, as shown by the
voltage/transmission curve in Fig. 4.4. A cell having this transmission curve is
referred to as normally white (whilst with parallel polarisers, it would be normally
black). In this curve, the threshold voltage is the minimum voltage required to
induce a measurable reduction in bright state transmission, and the saturation
voltage is the minimum voltage to achieve the black state. The contrast ratio is
given by the ratio of the maximum and minimum transmission values, and would
typically be *250 or more. The transmission curve in Fig. 4.4 is shown only for
positive drive voltages, but it is symmetric about the y-axis for both negative and
positive drive voltages. In fact, a.c. drive is always used with these LCs to avoid
the electro-chemical damage to the material, which would arise from d.c. driving.
Given the monochromatic nature of the TN cell, coloured displays are achieved
by passing the light from the cell through red, green and blue (RGB) coloured
filters. The cell structure shown in Fig. 4.3 should be regarded as a single pixel,
and a display can consist of more than a million pixels, each of which need to be
individually addressed to ensure that each has the correct voltage applied to
achieve the required optical transmission. In principle, this can be achieved with
Fig. 4.4 Transmission

versus voltage curve of
normally white TN LC
(Reprinted from [12],
Merck KGaA, Darmstadt,
Germany)
4.2 Liquid Crystal Cells 75
Fig. 4.5 a Illustration of AMLCD cell, b detail of pixel layout, showing TFT, c schematic cross-
section of a-Si:H TFT (b and c reprinted with permission from [19])
‘cross-wire’ addressing, in which a matrix of patterned ITO addressing lines are

used on one side of the display together with an orthogonal set on the other side of
the display, and the combined voltage at any point will determine the optical
transparency at that point. This is known as passive matrix addressing, but is only
used in low resolution and low quality displays, because, as the number of pixels
increases, the addressing time of each pixel decreases, and the image loses
brightness and contrast. All present day high quality displays use active matrix
addressing, in which there is a transistor in each pixel, as illustrated in Fig. 4.5a.
This schematically shows the definition of the ITO on the bottom plate into a
matrix of individual pixel areas, each of which has an addressing transistor. The
ITO on the upper glass plate is unpatterned, and is often referred to as the common
electrode. The coloured filters are also shown on the top plate, where they have
been fabricated in a stripe arrangement (although other arrangements are also
possible). In fact, each of the three pixels associated with the triad of coloured
filters can be thought of as sub-pixels of the display, which is usually shown in the
notation for the number of horizontal pixels in a display as N(93), where N is
related to the resolution of a display. For example, in a coloured VGA display,
with a resolution of 640 9 480 pixels, there is an array of 640(93) horizontal
pixels and 480 vertical pixels. Other common display formats are listed in
Table 4.1. The transistors in each pixel are used as switching devices to set the
voltage across the pixel in a short addressing time, and to store it until the next
refresh signal is obtained, as described in the following section.
Table 4.1 Common display formats, listing the number of rows and columns (for coloured
displays)
Format # Columns # Rows # Pixels R address time (s)
VGA 640(93) 480 921,600 3.47E-05
SVGA 800(93) 600 1,440,000 2.78E-05
XGA 1,024(93) 768 2,359,296 2.17E-05
WXGA 1,280(93) 800 3,072,000 2.08E-05
SXGA 1,280(93) 1024 3,932,160 1.63E-05
UXGA 1,600(93) 1200 5,760,000 1.39E-05
HDTV 1,920(93) 1080 6,220,800 1.54E-05
WUXGA 1,920(93) 1200 6,912,000 1.39E-05
QXGA 2,048(93) 1536 9,437,184 1.09E-05
QSXGA 2,560(93) 2048 15,728,640 8.14E-06
QUXGA 3,200(93) 2400 23,040,000 6.94E-06
4.3 Active Matrix Addressing
Active matrix addressing is shown schematically in Fig. 4.5a, in which the plate is
patterned into an array of pixels, each one of which is addressed by its own
transistor. The most widely used addressing transistor is an a-Si:H TFT: this is
shown in the pixel layout in Fig. 4.5b, and in cross-section in Fig. 4.5c. Although
this is a field effect transistor, like the MOSFET discussed in Chap. 3, its archi-
tecture is quite different, being a bottom-gated inverted staggered structure, rather
than a top-gated coplanar structure. Moreover, the source and drain connections
are not self-aligned to the gate electrode, but overlap it. This overlap results in a
parasitic capacitance, CGD, between the gate and the drain terminals, which can
induce display performance artefacts, as discussed in Sect. 4.4.2.1. Further details
on the design, fabrication and performance of a-Si:H TFTs are contained in
Chaps. 5 and 6.
As illustrated in Fig. 4.6a, the matrix of TFTs is connected in rows and col-
umns, with a common connection to the gate electrodes in each row, and a
common connection to the source terminals in each column. The row lines carry
the TFT addressing signals, and the column lines carry the data signals, Vs. The
drain of each TFT is connected to its ITO pixel electrode, and the timing of the
row and column voltages are shown in Fig. 4.6b and c, respectively. The high
resistivity LC material is represented by a capacitor, which is charged up to the
appropriate signal voltage, Vs, through its TFT. This is achieved by line-at-a-time
addressing, during which the data signal is fed into the display in a serial row-by-
row process, by sequentially switching on each row (VG = VG(on)), whilst holding
the others in the off-state (VG = VG(off)). If the addressing TFTs are made from
a-Si:H, they will be n-channel devices, and the row select voltages, VG(on), will be
positive, whilst the row de-select voltages, VG(off), will be negative. The timing
signals are related to the a.c. mains frequency, such that the display is refreshed
every frame time, Tf, which will be 16.7 ms for a 60 Hz frame rate. If there are M
4.3 Active Matrix Addressing 77
Fig. 4.6 a Active matrix

array layout, b Gate line
timing signal for row m, and
c data signal waveform for
column n, showing positive
and negative frames
rows and N columns in the display, then the time available, Ton, to charge the
TFTs in a given row is given by Tf/M, and this will be a function of the display
format. Table 4.1 lists the row addressing times, which vary from *35 to *7 ls,
across the range of display formats. During this interval, the signal voltages on the
column electrodes of the display are applied in parallel to the common sources of
the TFTs, but will charge only the one row of LC capacitors connected to the row
of TFTs, which has been turned on. The on-current of the TFTs must be large
enough to charge the capacitors in the time available. The maximum charge will
be when the voltage is changing from +black to -black, i.e. +Vsat to –Vsat (or vice
versa), and an approximate value of the required current is given by:
Ion [ 5 2Vsat CLC =Ton ! ðfor example, Ion 3:6 107 A for CLC ¼ 0:25pFÞ
ð4:2Þ
where CLC is the capacitance of the LC, and the factor 5 is to ensure the full
charging of the pixel. However, the charging rates for the positive and negative
charging cycles are not the same, due to the influence of the data signal polarity on
the source-gate bias, VGS, of the TFT. For instance, when the data signal, Vs, is
negative, the effective value of VGS in determining the channel resistance is VG -
VT ? Vs, and this more negative terminal acts as the source of the n-channel TFT
during the charging cycle. Conversely, for the charging cycle to a positive value of
the data signal, the TFT terminal connected to the LC is now the most negative
terminal, and, therefore, acts as the source. In this case, as the pixel capacitance
charges, its voltage, VLC, changes from –Vs to +Vs, which effectively reduces VGS,
so that, at the end of the charging cycle, VGS is VG-VT-Vs. Hence, the positive
charging cycle is slower than the negative charging cycle, because VGS is fixed at
VG-VT ? Vs during the negative charging cycle, but reduces from VG-VT ? Vs
to VG-VT-Vs during the positive charging cycle. The charging of the LC
capacitor is analogous to the charging of an RC circuit, but there are some
important differences, due to the non-linear nature of the TFT Id-VD relationship
(using the MOSFET Eq. 3.10), and the changing channel resistance during the
charging cycle. The charging transient can be calculated from:
dVLC 1 dQLC Id
¼ ¼ ð4:3Þ
dt CLC dt CLC
The charging current, Id, through the TFT is given by modifying Eq. 3.10 to
allow for the changing voltage, Vs-VLC, between the source and drain terminals,
and for the changing voltage, VG-VT-VLC, between the source and gate
terminals:
l WCi h i
Id ¼ n ðVG VT VLC ÞðVs VLC Þ 0:5ðVs VLC Þ2 ð4:4Þ
L
ZVLC Zt
dVLC l WCi
2
¼ n dt ð4:5Þ
ðVG VT VLC ÞðVs VLC Þ 0:5ðVs VLC Þ CLC L
V1 0
where V1 is the value of VLC at t = 0, and the solution of this equation is:
Vs ½V1 þ Vs 2ðVG VT Þ þ ½Vs 2ðVG VT ÞðV1 Vs Þ exp t=s
VLC ¼ ð4:6Þ
V1 þ Vs 2ðVG VT Þ ðV1 Vs Þ exp t=s
where,
CLC L
s¼ ð4:7Þ
ln WCi ðVG VT Vs Þ
Although Eq. 4.6 has more terms than the simple RC charging expression, it
still has a recognisable, effective time constant term, given by Eq. 4.7. This
expression is the product of the LC capacitance, CLC, and the TFT on-resistance at
the end of the charging cycle, when the value of VGS is VG-VT-Vs. As men-
tioned above, for the negative charging cycle, Vs is negative, and VGS is numer-
ically larger than it is for the positive charging cycle; hence, the time constant for
the negative cycle is smaller. This is illustrated in Fig. 4.7a, in which the positive
and negative switching transients are shown for V1 = -3 V and Vs = 3 V, and
V1 = 3 V and Vs = -3 V, respectively, and the slower tail on the positive cycle
is apparent. This is more clearly seen in the logarithmic plot in Fig. 4.7b, in which
both transients start at the same rate, because VGS is initially the same at
VG-VT-Vs for the negative transient (and Vs = -3 V), and is VG-VT-VLC for
Fig. 4.7 Calculated values

of LC voltage, VLC, during
pixel charging following data
signal, Vs, inversion (open
symbols: V1 = 3 V, Vs =
-3 V, and closed symbols:
V1 = -3 V, Vs = 3 V).
a VLC vs time, and
b Logarithmic plot of
absolute values of VLC-Vs vs
time, in which the lines are
best linear fits to the data.
(The illustrative calculations
were for a 15.4’’ WXGA
pixel, and used the following
parameter values: VG-
VT = 8 V, W/L = 5,
l = 0.5 cm2/Vs, ti = 0.3 lm,
and CLC = 0.25 pF)
the positive transient (where, at t = 0, VLC = V1 = -3 V). However, as VLC

becomes more positive, VG-VT-VLC becomes numerically smaller, and the
instantaneous channel resistance increases. The time constants extracted from the
slopes of the lines in Fig. 4.7b agree with the values given by Eq. 4.7, noting that,
for the positive transient, the line has been fitted to the slow tail.
At the end of the row on-time, that row is deselected (by switching the gate bias
to a negative value), and the adjacent row is switched on, and the process
described above is repeated. Hence, the display information is loaded row by row
during a frame time, and, at the end of this period, the whole process is repeated
with the next ‘field’ of information. However, between the loading of the data
signals onto a given row of TFTs, and its updating a frame-time later, it is
important that the original signal voltages are maintained across the LC capacitors.
During this interval, the sources of the de-selected TFTs will experience a variety
of different voltages, which will be different from the voltage, VLC, on the drain
terminals, and, depending upon the TFT leakage current, this potential difference
acts as a driving force to reset the voltage on the drain. To ensure that this effect is
minimised, the maximum tolerable leakage current, Ioff, is constrained to be below

a certain maximum value, to give, for instance, less than 1 % discharge of the pixel
voltage, so that:
Ioff \102 CLC Vmax =Tf ð4:5 1013 A for CLC ¼ 0:25pFÞ ð4:8Þ
Hence, with line-at-a-time addressing, the TFT is acting as a high quality
switch, with a minimum on:off current ratio given by Eqs. 4.2 and 4.8:
Ion =Ioff [ 500M [ 5 105 for a display with 1000 rows ð4:9Þ
The other potential leakage current path is through the LC itself, but its
resistivity is typically *1 9 1012 Xcm, and, although this is not negligible, the
associated leakage will be less than the TFT leakage.
The discussion of the positive and negative biases applied to the LC cell has
implied that the polarity of the signal voltages is changed in a uniform fashion across
the whole field between frames. This is illustrated in Fig. 4.8a, and is known as frame
inversion. From a display performance point of view, this has well documented
drawbacks, such as increased flicker, as discussed in Sect. 4.2.2.1, and increased
vertical cross talk due to different rows in the display experiencing opposite polarity
source and drain voltages in the off-state for different lengths of time. In fact, there are
several alternative inversion schemes, such as row inversion (the polarity of each row
of data alternates within one frame), or column inversion (the polarity of each column
Fig. 4.8 AMLCD data

signal drive inversion
schemes, showing the signal
polarity in consecutive
frames with (a) frame
inversion, and (b) dot
inversion
of data alternates within one frame), or dot inversion, as shown in Fig. 4.8b. This is a
mixture of row and column inversion, in which the polarity of every other pixel
alternates, and gives the lowest values of cross-talk and flicker. In all these schemes,
the polarity pattern within one frame is reversed in the next frame, so that each LC
pixel experiences alternating positive and negative biases.
The foregoing description of active matrix addressing of LCDs is somewhat
simplified and idealised, and there are important behavioural artefacts associated
with parasitic capacitances within the pixel, such as cross talk and pixel off-set
voltages. The magnitude of these effects is determined by the detailed pixel layout,
and these effects are discussed in the next section. It should also be mentioned that
additional capacitance is frequently introduced in parallel with the pixel capaci-
tance, by means of a storage capacitor, which is also discussed in Sect. 4.4.
Finally, the viewing angle dependence of brightness and contrast in the simple
TN cell is limited, and, although it is acceptable in smaller displays up to notebook
size, it was necessary to improve it in the larger monitor and TV displays. In the
TN cell, at mid-grey levels, the LC director is changing direction from horizontal
towards vertical in the middle of the cell, and, at different viewing angles, this
variable direction changes the transmittance of the cell. To improve the viewing
angle, modified cell architectures have been developed, of which the two most
widely used are the in-plane switching mode (IPS) [13], and the multi-domain
vertical alignment mode (VA) [14, 15].
The operation of the IPS cell, with crossed polarisers, is shown in the off and on
states in Fig. 4.9a and b [10], respectively. In both cases, the LC director is parallel
to the plane of the glass plates (and, in Fig. 4.9a, the director is perpendicular to
Fig. 4.9 Illustration of the normally black, IPS mode of LC cell switching. a Zero bias dark-
state. b Biased bright-state (Reprinted from [10] with permission of SID)
Fig. 4.10 Illustration of the normally black, VA mode of LC cell switching. a Zero bias dark-
state. b Biased bright-state (Reprinted from [10] with permission of SID)
the page). The major difference from the simple TN cell, is that the switching
electrodes are on the TFT plate, so that, when a bias is applied between these
electrodes (Fig. 4.9b), the molecules rotate about their short axis, and the director
aligns with the field. This puts a twist into the column of liquid crystal molecules,
and the plane of polarisation of the light is similarly rotated. The improved
viewing angle arises from the long axis of the molecules always remaining parallel
to the plane of the cell. However, the trade-off with IPS is a reduced aperture ratio
due to the inclusion of the switching electrodes within the pixel [13].
The operation of the VA cell, with crossed polarisers, is shown in the off and
on-states in Fig. 4.10a and b [10], respectively. The molecules are vertically
aligned at zero bias, and, as the LC mixture is engineered to have a negative
dielectric anisotropy, when a bias is applied between the top and bottom plates, the
molecules rotate in order to align their short axis with the field. With the short axis
vertically aligned, the long axis of the molecule is free to rotate within the hori-
zontal plane, and, by sub-dividing and texturing the pixel surface [15], the long
axis is constrained to lie in different directions within the sub-divided pixel. Hence,
in a pixel containing four sub-pixel domains, the same molecular orientation is
seen from different angles, which greatly improves the viewing angle performance
of the VA cell compared with the conventional TN cell.
4.4 Pixel Layout Considerations 83
Fig. 4.11 Pixel layout considerations a effect of bus bars and ITO clearance margin on aperture
ratio, b effect of black mask, BM, on aperture ratio, c illustrative a-Si pixel with storage capacitor
(Reprinted from [20] with permission of SID), and d poly-Si pixel with storage capacitor
4.4 Pixel Layout Considerations
4.4.1 General
The pixel layout will consist of the addressing TFT, the ITO pixel pad, the row and
column bus bars, various interconnections, including contact windows through
insulating layers, and possibly a storage capacitor. These will be designed in
compliance with the design rules of the fabrication process, which will specify the
minimum feature sizes, and the gaps between different layers. In a conventionally
designed pixel, these minimum gaps between different layers will result, for
instance, in a gap between the bus bars and the ITO pixel pad, within which the
optical transmission will be un-modulated, as shown in the schematic pixel layout
in Fig. 4.11a (for clarity, the TFT has been omitted from this diagram). If these
regions are left clear, they will permit residual optical transmission when the ITO
pixel area is turned off, and the contrast ratio of the display will be seriously
degraded. In order to ensure a high contrast ratio, these clear areas must be
overlaid with an occluding layer, which is normally a black masking pattern,
fabricated on the upper, colour filter, CF, plate. The design of this feature will be in
accordance with the alignment tolerances between the TFT and CF plates, and the
black mask area, BM, may be shown superimposed on the pixel layout, so that the
aperture ratio can be visualised. The BM pattern is shown in Fig. 4.11b, and the
area within its inner edges will be the final transmitting area of the pixel, AT. This
will determine the aperture ratio of the pixel, which is given by the ratio of this
area to the total pixel area, AP. For example, taking a 15.4’’ WXGA (16:10)
display, having 87 9 259 lm2 pixels (where AP = 2.24 9 10-4 cm2), with 5 lm
bus bars, 3 lm gaps between the ITO pad and the bus bars, and a 5 lm alignment
tolerance between the TFT and CF plates, the value of AT is 1.57 9 10-4 lm2,
giving an aperture ratio of 70 %. Hence, there can be considerable loss of aperture
just due to the bus bars and the black mask, even before the TFT and storage
capacitor areas are subtracted. The process design rules will be independent of the
pixel size, so that the above fractional losses may increase as the pixel size
decreases. Hence, for some high resolution [16] or small pixel geometry displays
[17, 18], where the conventional BM will seriously reduce the aperture ratio, its
inclusion on the TFT plate has been investigated, using either the bus bars
themselves [16, 17], or a separate, earthed metal track [18]. Except for the latter
architecture, the trade-off for a larger aperture is potentially higher cross-talk in the
display, due to the parasitic capacitances between the bus-bars and the overlapping
ITO layer. However, these are usually minimised by using thick inter-layer
dielectrics (ideally, with a low dielectric constant). (Where a separate earthed track
is used [18], this is to minimise the parasitic capacitance between the pixel pad and
the signal lines). For the majority of current displays, however, the BM is con-
ventionally fabricated on the CF plate [19].
As mentioned above, the pixel may include a storage capacitor, Cst, in parallel
with the LC pixel capacitor, which will reduce the effects of parasitic coupling, as
well as increasing the tolerance of the structure to TFT leakage currents between
refresh signals. In this case, Eq. 4.8 is modified to read:
Ioff \102 ðCLC þ Cst ÞVmax =Tf ð4:5 1013 ð1 þ Cst =CLC ÞA for CLC ¼ 0:25pFÞ
ð4:10Þ
The off-state leakage current in poly-Si TFTs is generally larger than in a-Si:H
TFTs (as discussed in Chaps. 6 and 8, respectively), and, in these displays, Cst/
CLC *2–5, whereas, in a-Si:H displays, it is more common to find that Cst/
CLC *1. The two favoured techniques for building a storage capacitor into a pixel
are either to use the previous gate line, or to introduce a separate metal track into
the pixel as one plate of the capacitor, with the ITO or TFT drain pad being the
other plate. An illustrative a-Si:H TFT pixel layout [20] is shown in Fig. 4.11c,
and, in this case, a separate metal track over the ITO pad has been used to form the
storage capacitor. Figure 4.11d shows a poly-Si pixel design [21], in which the
alternative strategy, of using the previous gate line, has been employed. The other
detailed difference between Fig. 4.11c and d is that, in the latter case, an extension
to the doped poly-Si drain has been used as the lower plate of the storage capacitor
rather than the ITO pixel pad itself. (Forming the capacitor with the drain
extension is electrically equivalent to using the ITO pad, as the two are connected
to each other. However, by using the drain extension, a higher capacitance/unit
area can be achieved, due to the thinner gate oxide separating the poly-Si layer
from the gate electrode, whereas thicker interlayer dielectrics would generally
separate the ITO from the gate line).
Whilst Fig. 4.11c and d cannot be quantitatively compared with each other in
terms of aperture ratio, it is generally the case that the storage capacitor formed
from the adjacent gate line is the more area efficient, as the gate bus bar has to be
present in any case. However, as shown in Fig. 4.11d, its width was greater than
the column lines in order to give the required storage capacitance, and to provide
for the alignment tolerances within the process.
4.4.2 Performance Artefacts
This section describes aspects of the layout/processing conditions, which can lead
to performance artefacts in the display. These can result from parasitic capacitive
coupling between the drain and gate terminals, parasitic coupling between the
source and drain terminals, and the effects of the resistance of the gate row lines.
The first leads to a voltage resetting of the pixel voltage when the row select
voltage is changed to deselect, the second gives cross-talk, and the third can lead to
incomplete pixel charging in long rows.
4.4.2.1 Voltage Kick-Back
As shown in Fig. 4.5b and c, the drain terminal overlaps the gate electrode, leading
to a parasitic capacitance, CGD, between the two terminals. This is included,
together with a storage capacitor, Cst, in the pixel equivalent circuit in Fig. 4.12a.
As a result of this coupling, the capacitor, CGD, acts as a voltage divider for
changes in the gate voltage, DVG, leading to changes in the pixel voltage, DVLC,
given by:
DVLC ¼ DVG CGD =ðCGD þ CLC þ Cst Þ ð4:11Þ
This occurs at the end of the pixel charging period, when the gate voltage is
switched from the positive select value, +VG(on), to the negative de-select value, -
VG(off), causing the pixel voltage to be reset by a negative amount, DVLC, as shown
in Fig. 4.12b. This voltage reset is referred to as the kick-back voltage, and, as it is
always negative, it reduces VLC during the positive phase, and increases it during
the negative phase, of the pixel charging sequence. In addition to this parasitic
Fig. 4.12 a Pixel equivalent circuit, including storage capacitor, b row addressing waveform,
column data voltage, Vs, LC voltage, VLC, and kick-back voltage, DVLC, and c effect of DVLC on
LC transmission-voltage curve, where j VLC(-ve) j [ j VLC(+ve) j
capacitive coupling, when the TFT is switched off, the channel charge is ejected
and *50 % leaves by each of the two TFT terminals, leading to extra negative
charge being added to the LC capacitor, given by:
DQLC ¼ 0:5WLCi ðVG VT Vs Þ ð4:12Þ
where W and L are the channel width and length of the TFT, respectively, and Ci is
the gate insulator capacitance/unit area. This charge leads to a further change in
the stored pixel voltage, which varies with the size and polarity of Vs.
These changes in pixel voltage cause the optical transmittance of the pixel to be
different in the negative and positive charging phases (even though the signal
voltage, Vs, is numerically the same, and differs only in its polarity). The differ-
ence in transmission in the positive and negative frames is shown in Fig. 4.12c,
and this can lead to the appearance of flicker in the display. The flicker can be
reduced by using the minimum design rule values of overlap in the pixel design,
and by the presence of the storage capacitor, Cst. In principle, the voltage kick-
back effect is constant, and could be compensated for by making the common
electrode voltage on the CF plate more negative. However, due to the dielectric
anisotropy of the LC, its capacitance is a function of the bias across it, so that the
voltage division is not constant, and full compensation cannot be achieved this
way. Nevertheless, off-setting the common electrode voltage from zero can reduce
the flicker. In addition, the presence of a storage capacitor reduces the dependence
of the total capacitance on pixel voltage, and, therefore, reduces the variability in
kick-back voltage. A further technique to reduce the perception of flicker is to use
dot inversion, rather than frame inversion (see Fig. 4.8), so that adjacent pixels
experience alternating positive and negative biases, which visually averages out
the perception of frame-to-frame changes in overall brightness.
4.4.2.2 Vertical Cross-Talk
Vertical cross talk occurs when the signal voltage, Vs, being applied to a particular
row of TFTs, affects the transmittance of pixels in other rows, and this can result
from a parasitic capacitance, CDS, between the pixel pad and the adjacent column
lines. The capacitance can arise from 2-D coupling of the adjacent lines and pads,
the direct overlap of the ITO pixel electrode, or the overlap of the drain terminal of
the TFT with the source line. A particular example of the latter is shown in the
pixel layout in Fig. 4.11d. The change in pixel voltage due to CDS is given by:
DVLC ¼ Vs CDS =ðCDS þ CLC þ Cst Þ ð4:13Þ
As with the kick-back voltage, the magnitude of this effect is reduced by the
presence of a storage capacitor, and the value of CDS itself is minimised by using
thick dielectric layers, ideally with a low dielectric constant, between the column
lines and the pixel pads.
4.4.2.3 Row Resistance Effects
As AMLCDs increase in size, the length of the addressing lines gets longer, and, in
the absence of any changes in the structure of the lines, their total line resistance
increases. Similarly, as the resolution of displays increases, the number of pixels
increases, and the capacitive loading of the addressing lines increases (due to the
greater number of pixel TFTs and of row/column crossovers). Hence, the RC
charging times of the rows is likely to increase whilst, at the same time, the
increased resolution will reduce the row on-time. As a consequence of this, if the
RC charging time constant is too long, there is a potential problem of inadequate
pixel charging in larger and higher resolution displays. In fact, because the unit
cell resistance and capacitive loading of each pixel is repeated along the line, the
charging of the line is best described as a distributed RC network, in which the
pixel furthest from the point at which the gate switching signal is applied will
receive the most delayed signal. This effect is demonstrated in Fig. 4.13a and b, in
which the gate voltage and pixel charging waveforms have been simulated for the
Fig. 4.13 Pixel charging 20 (a) Front Pixel

waveforms (a) 1st pixel, and 99% rise time = 20 µs
(b) last pixel in row, where 15 Voltage error = –0.268 V
the dotted line is the gate
voltage, the dashed line is the 10
Voltage (V)
data signal, and the solid line
is the pixel voltage 5
(Reprinted from [22] with
permission of SID) 0
–5
–10
0 25 50 75 100 125 150 175 200 225 250 275 300
Time (µs)
20
(b) End Pixel
99% rise time = 27µs
15 Voltage error = –3.385 V
Voltage (V)
10
–5
–10
0 25 50 75 100 125 150 175 200 225 250 275 300
Time (µs)
first and last pixels, respectively, in a row addressed by a line whose resistance is
too high [22]. As this diagram shows, the pixel charging occurs accurately for the
first pixel, but the distortion in the gate voltage waveform at the last pixel leads to
a large error in the pixel voltage, which, in this particular example, was due to the
slow turn-off of the gate pulse. Hence, the resistance of the gate lines has to be low
enough to meet the requirements of the display, in terms of its size and resolution.
The size is usually defined by the display diagonal, D, or its horizontal width, LH
and vertical height, LV. The resolution can be defined by the number of rows, M,
and columns, N, which can also be expressed as the pixel resolution/unit length, r,
given by N/LH.
The total row resistance, RR, is given by:
RR ¼ qLH =tG WR RSG LH =WR ð4:14Þ
where q is the resistivity of the gate material, tG is its thickness, WR is the width of
the row line, and RSG is the sheet resistance of the metal.
The total row capacitance, CR, is given by:
CR ¼ NCGTFT þ NCX þ NCstðparÞ ð4:15Þ
where the TFT gate capacitance, CGTFT = AGTFTe0ei/ti (ti is the thickness of the
gate dielectric), and the other capacitance terms are parasitic ones due to: the row/
column cross-overs, CX = WRWCe0ed/td (td is the thickness of the cross-over
dielectric, and WC is the width of the column line), and the storage capacitor,
Cst(par), assuming that the gate is being used as the storage capacitor for the TFT in
the adjacent row. In this case, the parasitic capacitance will be the series combi-
nation of Cst and CLC through the pixel in the adjacent row, i.e. Cst(par) = CstCLC/
(Cst ? CLC). If the gate-line storage capacitor is absent, this term can be ignored.
Other potential parasitic capacitances would need to be assessed from the detailed
pixel layout, and could include the capacitance of the gate line to the common
electrode on the CF plate.
To ensure full charging of the furthest pixel in the row, the RC time-constant of
the row line will need to be less than a small fraction, a, of the row on-time, Ton
(= Tf/M) [22], i.e.:
RR CR \aTf =M ð4:16Þ

q\aTf tG WR =LH MN CGTFT þ CX þ CstðparÞ ð4:17Þ
Hence, the required resistivity of the row metal will scale inversely with the
size and resolution of the display, but can be traded-off against the thickness and
width of the line. However, this trade-off will be limited by the process design
rules, which may specify a maximum film thickness (to ensure good step coverage
by subsequent layers) or a maximum line width to ensure an acceptable aperture
ratio. Figure 4.14 [22] is an example of the calculated trade-off in resolution
(defined in pixels/mm) and display diagonal, which can be met by metals of
different resistivity. (In this calculation, the aperture ratio was kept constant, and
the metal film thickness, tG, was held at 300 nm). In order to put these values into
the context of specific displays, the dotted lines show different display formats,
Fig. 4.14 Calculated trade- 10

off between a-Si:H AMLCD
resolution and display Clad Al
diagonal, for different display 8
formats (dotted lines) and
different gate metals (solid
Resolution (mm–1)
lines) (Reprinted from [22] 6

Cu+
Al
with permission of SID)
Cu
4
HDTV
Ta SXGA
2 XGA
Mo/Ta
VGA
0
0 10 20 30 40
Diagonal (in.)
defined in terms of the pixel count shown in Table 4.1, and only those displays
which fall to the left of a particular metal can be satisfactorily driven by it. The
number of display formats in this figure is limited, and a subsequent publication
extended the range up to QUXGA [16]. As illustrated by Fig. 4.14, lower resis-
tivity metals are needed to drive displays of increasing size and resolution, and the
evolution in preferred gate metal can be observed in the change from Cr gate lines
(q = 20 lX cm), in the earliest displays, through Mo (q = 13 lX cm), and AlNd
(q = 7 lX cm), to Al (q = 4 lX cm) or Cu (q = 2.5 lX cm) [23] in contem-
porary ones.
The above discussion has focussed on the resistance requirements of the gate
lines, and has included the option of using the TFT gates, in adjacent rows, as
storage capacitors. Where separate, non-gate lines are used as storage capacitors,
the loading of the row lines will be reduced, and the resistance of the storage
capacitor lines will need to be separately considered, to ensure that it is low
enough to charge the storage capacitors during the row on-time. Finally, there are
similar calculations to establish the maximum acceptable resistance for the column
lines, but these usually have lower capacitive loading than the row lines, and,
therefore, the preferred row metal should be suitable for the column lines as well.
4.5 AMLCD Fabrication
This section provides a brief overview of AMLCD fabrication [19], and describes
the major stages of this process, consisting of the TFT and colour filter plate
fabrication, the LC cell construction, and the assembly of the final display module.
4.5.1 TFT Plate
Most commercial AMLCD manufacturers use a-Si:H TFTs, and the fabrication of
these devices is described in Chap. 5. By and large, the basic a-Si:H TFT fabri-
cation schedule is used for the active TFT plate, but with the addition of a final
dielectric layer (by plasma enhanced chemical vapour deposition, PECVD, of
a-SiNx:H for a-Si:H TFTs), as an interlayer dielectric to electrically separate the
ITO film from the source and drain metallisation. Contact windows are photo-
lithographically patterned down to the drain contact pads, and then ITO is sputter
deposited at *200 C, and defined into the individual pixel pads. Figure 4.15a
shows a plan-view and cross-sectional diagram of a pixel, including a storage
capacitor, on the finished plate.
4.5 AMLCD Fabrication 91
Fig. 4.15 a Plan and cross-section of AMLCD pixel, and b cross-section of colour filter plate
(a Reprinted from [52]. Copyright (2005), with permission from Elsevier, and b Reprinted with
permission from [24])
4.5.2 Colour Filter, CF, Plate
The cross section of a completed CF plate is shown in Fig. 4.15b [24], and the first
stage in its fabrication process is the deposition and definition of the black mask
material. This is a resin containing black particles so that the final film, which is
photo-lithographically defined, has a high optical density and low reflectivity. The
next stages deposit and define the three different colour filter layers within the
apertures in the black mask pattern. The colour filter films are spin-on photo-resist
layers containing pigments to give the required red, blue and green colours, and, as
the films are photosensitive, they are directly defined by photo-lithography. Fol-
lowing the formation of the colour filters, the plate is covered by a planarising film,
and then sputter coated with the transparent conductor, ITO, at *200 C, to form
the common electrode. The ITO film thickness is thicker than on the TFT plate, as
it has to have a greater conductivity.
4.5.3 LC Cell
The LC cell is the aligned and sealed combination of the TFT and CF plates, with
the LC material sandwiched between them. The first stage in the cell assembly
process is the coating of the active side of both plates by a polyimide alignment
layer, which is then passed beneath a fine-haired, velvet-coated cylindrical roller in
order to micro-groove the surface. This pattern establishes the subsequent align-
ment direction of the LC molecules [19, 25]. For control of the final cell thickness,
photoresist pillars, *4–6 lm high, are defined on the CF plate, outside the col-
oured areas, so as to avoid a reduction in the display’s aperture ratio.
The LC filling of the cell, and its alignment and sealing, are accomplished in a
vacuum chamber by the ‘one drop fill’, ODF, technique [26]. This is a major
improvement in time and LC material usage compared with the earlier technique
in which the assembled, but empty, cell was evacuated in a vacuum chamber,
lowered into a vessel of LC material, and then the pressure in the chamber was
increased so that the pressure difference, plus capillary action, forced the
LC material through a gap in the seal line into the evacuated space between the
plates. As cell size increased, this became an increasingly inefficient process [26].
With the ODF process, a sealant (or glue) line is dispensed/printed onto one plate,
and a precisely monitored quantity of LC material, which will be sufficient to fill
the cell, is dropped onto the plate. Both plates are then loaded into a vacuum
chamber, where they are aligned, and the chamber pressure increased to disperse
the LC, and to make the initial seal, after which UV exposure and thermal curing
complete the sealing process. At this point in the process, the mother glass plate,
containing several individual displays is still intact, and the final fabrication step is
the scribing and breaking of the plate into the separate LC cells.
4.5.4 Display Module
The display module is the completed working unit, which is housed in a rigid,
protective case, and consists of the LC cell, the polariser plates, the back light and
the external drive circuits. The polariser plates are laminated to the cell, together
with any other films for cell protection or viewing performance enhancement.
The external drive circuits control the data and clocking signals sent to the
display, and, with a-Si:H displays, will need to make contact to each row and
column in the display. This high density of connections is facilitated on the TFT
plate by a metallisation fan-out pattern, which is fabricated during the row and
4.5 AMLCD Fabrication 93
column metallisation stages. The drive circuits themselves are crystalline Si ICs,
designed for generating the clock signals to drive the rows, or to supply the column
data signals, together with their appropriate timing signals. These chips are con-
nected to the metallisation fan-outs at the side and top of the display, respectively,
using either a chip-on-glass, COG, process, or a tape automated bonding, TAB,
process (which is also referred to as chip-on-film, COF) [19]. The COG process
mounts the chips upside-down, directly on the TFT substrate, whereas, with the
TAB process, the chips are mounted on a flexible foil substrate, which is then
bonded to the fan-out on the TFT plate. The TAB material contains a photo-
lithographically defined wiring pattern, with bond pad areas to which the ICs are
fixed, and a second set of bond pad areas, with which connection is made to the
lead-out pattern on the TFT plate. For both COG and TAB processes, the bonding is
accomplished with an anisotropic conducting film, ACF, which is, essentially, an
anisotropic glue. The anisotropy arises because the glue is a non-conducting matrix
containing small metallised spheres, which, when compressed between two verti-
cally aligned bonding pads, makes metallic contact with the pads. The absence of a
continuous, horizontal conducting path in the material avoids lateral short circuits
between adjacent pads on the plates. The COG approach is more commonly used in
smaller displays, as it is a more area-efficient connection scheme.
The final stage of the assembly process is the attachment of the back-light. This is
either a cold cathode fluorescent light, or an LED array [27] (although it should be
emphasised that these inorganic LEDs are quite different from the organic LEDs
used in AMOLEDs, where the OLEDs are used instead of the LCD film). Depending
upon the application, the illumination source can be either side-mounted with an
appropriate light guide and diffuser to give uniform illumination of the display, or
directly mounted at the back of the display with just the diffuser plate. LED back-
lights are now increasingly applied to medium and large sized display products, with
*50 % of AMLCD panels currently incorporating them [28].
4.6 Other Display Technologies
4.6.1 Active Matrix Electrophoretic Displays
Electrophoretic displays [2, 3] are now commercially used in a variety of e-reader/

e-book products. Their primary advantage over AMLCDs is that they are low
power, reflective products, which operate well in bright environments, so that a
power-consuming back-light is not required, and, furthermore, the display medium
can operate with very low refresh rates, and even in a bi-stable mode. In addition,
the reflected image is ‘paper-like’ [2], so that the EPDs are well suited to low
power, portable ‘electronic-paper’ products. Given the e-reader application, which
does not require fast response times and short addressing periods, they can be
addressed by low carrier mobility TFTs. This includes not just the industry
standard a-Si:H TFTs [5], but organic TFTs on flexible substrates [6] as well.
Conversely, their slow response, and, until recently, black/white nature has meant
that they do not compete in the conventional AMLCD product market.
As the name implies, EPDs rely upon electrophoresis, which is the field-driven
movement of charged particles suspended in a fluid medium. In the current EPDs,
there are micro-encapsulated black and white charged particles, and the micro-
capsules, in a polymeric binder, are deposited as a 100 lm thick layer onto an ITO
coated polyester film [3, 29]. The completed electrophoretic foil, attached to a rear
electrode, is illustrated in Fig. 4.16a, together with a close-up of a micro-capsule
[30]. The principle of image formation is shown in Fig. 4.16b, in which the
negatively charged, black particles have been driven to the upper surface with a
negative bias on the rear electrode. Conversely, the positively charged, white,
titanium oxide particles will be pushed to the surface with a positive bias. As the
white particles reflect the ambient light, and the black particles absorb it, a black
Fig. 4.16 a EPD foil attached to a bottom electrode (Reprinted from [30] with permission of
SID), b schematic illustration of micro-encapsulated EP behaviour, and c image of an ‘e-ink’ line
(Reprinted from [3] with permission of SID)
4.6 Other Display Technologies 95
and white image is readily formed, and a magnified image of a well-defined 200
dpi white line, formed from 70 lm micro-capsules, is shown in Fig. 4.16c.
To form a high density image, the EPD foil is laminated onto an active matrix
TFT plate, and Fig. 4.17a shows an a-Si:H TFT pixel layout [31], and Fig. 4.17b
illustrates its cross-section [5]. The pixel has many common features with AM-
LCD pixels, including the ITO pixel pad and a storage capacitor. However, one
particular feature of the EPD pixel is the field-shielded pixel design, in which the
ITO pixel pad extends over the row and column bus bars, and over the TFT, and is
separated from them by a thick, low dielectric constant insulator layer to reduce
parasitic capacitances. This design has two primary objectives: firstly, to give a
large effective optical aperture, by applying the pixel voltage to the largest possible
area of EPD film, and, secondly, to screen the EPD film from the voltages
appearing on the bus bars and the TFT, which could cause local areas of unwanted
contrast switching [31].
The AMEPD is addressed line-at-a-time, with typical pixel voltages of ±15 V
for the saturated black and white states, and row select voltages of +22 V and
-20 V [32, 33]. In view of the larger pixel voltages compared with AMLCDs, two
TFTs are used in series (in the pixel shown in Fig. 4.17) in order to reduce the
leakage current, and the storage capacitor also contributes to minimising the
discharge of the pixel capacitance through the EPD film [31]. Grey scale can be
achieved either by reducing the pixel voltage, or by pulse width modulation [30],
and coloured displays are being realised with coloured filters [34], arranged in a
quartet of sub-pixels of red, green, blue, and B/W [35].
Fig. 4.17 AMEPD pixel

structure, showing TFT,
storage capacitor and ITO
pixel; a pixel layout
(reprinted from [31] with
permission of SID), and
b cross section (reprinted
SID)
4.6.2 Active Matrix Organic Light Emitting Diode Displays
OLEDs are electroluminescent devices, in which the emission of light is due to the
passage of an electric current through the material, promoting radiative electron–
hole pair recombination. Hence, AMOLEDs are emissive displays, and, unlike
AMLCDs, do not need a backlight, nor coloured filters, as they can be fabricated to
yield the required RGB spectrum. This offers the prospect of lower power, thinner,
lighter and cheaper displays, and, as they are emissive (and not dependent upon the
passage of light through a layer of molecularly twisted liquid crystal) they also
have a much wider viewing angle. In addition, the response time is faster than
LCDs, and OLEDs are perceived to be more compatible with flexible substrates, as
there is not the requirement to maintain a well controlled cell gap. However, being
current driven devices, the detailed addressing process differs from both AMLCDs
and AMEPDs (in which the TFTs are essentially voltage switching devices), as the
TFTs in AMOLEDs are operating in an analogue current mode. Therefore, the
uniformity of the TFTs is of greater importance, and this is reflected in a more
complex pixel layout, and also places a premium on the uniformity of the chosen
TFT technology. In addition, OLEDs, and their cathode materials, are very sen-
sitive to contamination by oxygen and water vapour, and the whole structure needs
to be hermetically encapsulated to ensure an acceptable operational lifetime.
In Sect. 4.6.2.1, the basic operation of OLEDs is described, and AMOLED
pixel issues are discussed in Sect. 4.6.2.2.
4.6.2.1 OLED Operation
Two classes of organic LED have been investigated for AMOLED applications,
and these are small molecule organic LEDs (SMOLEDs) and polymer LEDs
(PLEDs). The SMOLEDs are generally prepared by vacuum thermal evaporation,
VTE, whilst the PLEDs can be solution processed and printed [36]. However,
solution processed SMOLEDs have more recently been reported, albeit with
reduced performance compared with the VTE devices [36, 37]. Much of the
current AMOLED activity is focussed on SMOLEDs, and the following discussion
is limited to these materials. In view of this, the shortened term OLEDs will be
used to refer just to SMOLEDs in the rest of this section.
The initial interest in OLEDs was stimulated by the observation of electrolumi-
nescence in a thin double layer structure of 75 nm of diamine covered by 60 nm of 8-
hydroxyquinoline aluminium, Alq3 [38]. A low work function cathode of Mg/Ag was
deposited on the Alq3, which served as the electron injector, and the hole injecting
anode was an ITO layer on glass. The injected electrons and holes formed bound-
pairs, or excitons, on the Alq3 side of the organic materials’ interface, where they
recombined in a fluorescent, radiative decay process, emitting light at 550 nm. The
luminescent efficiency was 1.5 lm/W, and, following this initial work, subsequent
developments in materials and OLED design have increased this efficiency to a peak
value of *70 lm/W at 550 nm [39].
One factor contributing to increased emission efficiency has been an engineered
change in the mechanism of the radiative recombination process itself, so that the
internal quantum efficiency, IQE, was improved [40–42]. The exciton population
is distributed between singlet and triplet states, but, due to spin conservation rules,
only the singlet state, with a spin of zero, participates in the fluorescent decay
process. As the singlet states comprise only *25 % of the exciton population, it
was believed to restrict the IQE to *25 % [40], although later observations of
higher than expected IQE have shown that triplet–triplet annihilation can also give
emission in fluorescent materials [43]. However, by adding guest molecules of a
phosphorescent dye to the host emission material, the 75 % triplet population can
directly participate, via an intersystem crossing process, in a phosphorescent
radiative decay process, and offers the opportunity for 100 % IQE [42]. This
modified class of OLEDs is referred to as phosphorescent OLEDs, or PHOLEDs.
The photon energy from phosphorescent decay will necessarily be less than the
fluorescent energy of the host, and, therefore, the host and dye materials need to be
matched to obtain the required emission spectra from RGB PHOLEDs.
In addition to the improvement in IQE, the OLED construction has been modified,
from the original bi-layer structure of diamine and Alq3, to a multi-layer device.
These structures have been engineered to improve the injection and transport of
carriers to the recombination layer, within which they are confined for radiative
recombination. This has been achieved by appropriately matching the highest
occupied molecular orbitals, HOMOs, of the organic material carrying holes, to the
work function of the anode, and similarly matching the lowest unoccupied molecular
orbitals, LUMOs, of the electron carrying materials, to the work function of the
cathode. Equally, the barriers either side of the recombination layer have been
designed to confine the electrons and holes within this region, and to prevent their
direct flow to the anode and cathode, respectively. Also, to exploit the higher potential
IQE with PHOLEDs, it is also necessary to confine the longer lived excitons to the
phophorescent dyed layer [44]. Hence, the diodes typically consisting of some, or all,
of the following sequentially grown layers: hole injection, HIL, and hole transport,
HTL, layers, an electron blocking layer, EBL, the emissive layer, EL, a hole blocking
layer, HBL, and electron transport, ETL, and injection, EIL, layers. In addition to
these organic layers, the diodes are capped by the anode and cathode materials. An
example structure [44] is shown in Fig. 4.18, in which the schematic energy level
diagram is illustrated in Fig. 4.18a, and the chemical structure of the layers is shown
in Fig. 4.18b. In this particular example, hole and electron injecting layers were
absent, because the small HOMO barrier for hole injection from the anode, and the
small LUMO barrier for electron injection from the cathode, obviated the need for
specifically low-barrier carrier injection layers. Where injection layers are used, they
are to bridge larger gaps between the metal work function and the band edge of the
carrier transport layer. Equally, for carrier blocking layers, if the HOMO of the
electron transport layer is significantly below the HOMO of the recombination layer,
this will block hole transport and avoid the need for a separate hole blocking layer (and
Fig. 4.18 Illustration of a PHOLED stack containing hole and electron blocking layers
(a) HOMO and LUMO equivalent band structure, and (b) material stack. (a-NPD = 4,4’-bis[N-
(1-napthyl)-N-phenyl-amino] biphenyl; Irppz = fac-tris(1-phenylpyrazolato,N,C2’)iridium(III);
CBP = 4,4’-N,N’-dicarbazolebiphenyl; FPt2 = platinum(II)(2-(4,6-difluorophenyl)pyridinato-
N,C2’)b-diketonate complex; BCP = bathocuproine) (Reprinted from [44] Copyright (2003),
with permission from Elsevier)
similar arguments hold for EBLs) [44]. However, this condition was not met with the
layers in Fig. 4.18, and the HBL and EBL were introduced. This discussion has
focussed on carrier blocking, and similar arguments apply to exciton triplet blocking,
so that they too are confined within the doped electroluminescent layer [44], as
schematically shown by the star-burst icons in Fig. 4.18a.
Finally, being current driven devices, the improvement in emission efficiency
has reduced the drive current needed to achieve acceptable levels of brightness in
AMOLEDs, and facilitated the use of low carrier mobility semiconductors, such as
a-Si:H [45] and organic TFTs, to address these displays. In addition, the lower
drive currents have also increased the operating lifetimes of the OLEDs them-
selves [46]. The primary lifetime issue with OLEDs of all types [47], and with blue
emitting structures in particular [37], is a reduction in the luminescent intensity
during device operation. OLEDs are subject to a number of extrinsic and intrinsic
aging processes, where the former include moisture absorption, and contamination
within the materials themselves [48], and the intrinsic processes cover internal
defect centre formation leading to non-radiative recombination pathways [47].
Detailed discussion of OLED instability is beyond the scope of this book, apart
from noting that effective device encapsulation is routinely employed to suppress
moisture and oxygen contamination, and improved materials synthesis is an on-
going activity [48]. In addition, a comprehensive understanding of the nature and
role of internal defect formation continues to be the subject of research. Not-
withstanding the incomplete understanding of the internal degradation processes,
there is considerable empirical activity to identify and use specific OLED layer
compositions, which give enhanced operating lifetimes [49–51].
Fig. 4.19 Schematic cross-section of an AMOLED pixel showing a top-emitting, and b bottom-
emitting architectures (Reprinted from [52]. Copyright (2005), with permission from Elsevier)
4.6.2.2 AMOLED Pixels
The two topics to be addressed in this section are the in-pixel addressing circuits,
and top-side or bottom-side OLED emission. The test structures referred to in the
previous section all had ITO anodes on glass, and were, by implication, bottom-
side emitting structures. Whilst this does not present a problem in basic materials
work, a bottom-side emitter will have a smaller pixel aperture ratio than a top side
emitter, due to the occluding effects of the bus bars and the addressing TFTs, as
shown schematically in Fig. 4.19 [52]. Hence, to build a top-side emitter, either a
(semi-)transparent cathode and reflecting anode is needed [53], as shown, or the
structure needs to be built with the normally transparent anode on top [54]. Also,
to realise the potential benefits of the larger aperture ratio, it is necessary to
optimise the optical coupling within the structure containing the semi-transparent
cathode, so that there is maximum light emission, with minimum colour distortion,
through the top surface [53]. This structure is referred to as a micro-cavity, and its
internal reflectance characteristics need to be carefully controlled [56]. Examples
of engineered semi-transparent cathodes are a thin Mg:Ag layer capped by an
80 nm ITO layer of optimum thickness [55], and, in another case, a 0.5 nm LiF/
1 nm Al/20 nm Ag layer capped by a 40 nm dielectric film of TeO2 [56]. These
materials were optically matched to the rest of the OLED stack, and, whilst there
Fig. 4.20 AMOLED drive circuits a p-channel 2T1C (anode on drain), b n-channel 2T1C
(anode on source), and c n-channel 2T1C (cathode on drain). (Reproduced from [57] with
permission of John Wiley & Sons, Inc)
are numerous publications on top emitter AMOLEDs, details on the choice of

cathode materials and optimisation of the structure generally remain proprietary.
The simplest pixel driving circuit consists of two TFTs and a capacitor, as
shown in Fig. 4.20a [57], and is frequently referred to as a 2T1C pixel. This shows
two p-channel TFTs, where T1 is the addressing transistor, and T2 is the drive TFT,
which is connected to the anode of the OLED. As the anode is biased more
positively than the cathode in the emitting state, this means that the OLED is
connected to the drain of T2. The source of T2 is connected to the OLED power
line, VDD, and the OLED cathode is connected to a common terminal, which may
be earthed. Hence, when T1 is switched on, during the row-select period, the signal
voltage, Vs, on the data line is passed through TI to charge the gate of T2, and the
storage capacitor, Cs (connected between the gate and source of T2) up to Vs. As
with other line-at-a-time addressing schemes, at the end of the row-select time, T1
is switched off, and the combination of T2’s gate capacitance and Cs retains Vs on
the gate of T2 for the frame period. When comparing this pixel addressing scheme
with AMLCD pixels, it will be noted that, in addition to the extra TFT, there is
also an extra bus-bar to carry the OLED power line. These extra features, in this
most basic pixel design, will reduce the pixel aperture ratio, and contributes to the
interest in top-side emitting OLEDs, where the aperture ratio is largely indepen-
dent of the pixel layout.
The p-channel circuit, illustrated in Fig. 4.20a, can be directly implemented
with p-channel TFTs fabricated with poly-Si or organic TFTs. However, with
n-channel drive TFTs (a-Si:H or amorphous oxide semiconductor, AOS, TFTs)
connected to the OLED anode, as shown in Fig. 4.20b, the OLED is now on the
source side of T2, so that the voltage drop across the OLED, VOL, will change the
gate-source bias, VGS, and, hence, change the drive current delivered by this TFT
(for a given value of Vs). An alternative configuration with n-channel TFTs is the
inverted OLED structure, in which the cathode is connected to the drain of T2, as
shown in Fig. 4.20c.
The data signal, Vs, on the gate of T2 determines its on-current, and, hence,
controls the current flowing through the OLED, which then determines the voltage
drop, VOL, across the OLED. Transistor T2 would normally be operated in satura-
tion, so that, using the MOSFET Eq. (3.17), the channel current Id(sat) is given by:
lWCi ðVGS VT Þ2
2L
For p-channel TFTs, VGS is the gate-source bias, which, in this case is equal to
Vs - VDD, but, for n-channel TFTs (using the circuit shown in Fig. 4.20b), VGS is
equal to Vs - VOL, and, if VOL varies across the display or changes due to aging,
this will cause non-uniformities in the display [58, 59]. In spite of this, the sim-
plicity of this circuit is its main attraction, and it continues to be used for a range of
demonstrator AMOLEDs, including those addressed by AOS TFTs [8], and OT-
FTs [60].
However, inspection of the terms in Eq. 4.17 illustrates some of the limitations
in this driving circuit, where the required ±1 % accuracy in setting OLED
brightness may not be met:
• The current is very dependent upon VT, and, to a lesser extent, the carrier
mobility, l, and local non-uniformities in these values will appear as brightness
variations in the display.
• Changes in these values during display operation (due to TFT lifetime effects)
will similarly affect display performance. This is particularly true of the
threshold voltage in a-Si:H TFTs, where, in contrast to the low duty cycle in
AMLCDs, the drive transistor will be on for most of the time. This will usually
increase VT, and, to compensate for this, the signal voltage will need to increase.
(Gate bias instabilities in a-Si:H TFTs are discussed in Chap. 6, and a com-
parison of the gate bias stability of the major TFT technologies for the AM-
OLED application is shown in Fig. 9.25 (in Sect. 9.4.3.3).
• Equation 4.17 assumes infinite output impedance, such that the drain saturation
current is not dependent upon the particular value of drain bias across the TFT.
In reality, this will not be achieved, although some technologies, such as a-Si:H,
have better output characteristics than others, such as poly-Si.
• The sheet resistance of the OLED power lines must be low enough that the
voltage drop along these lines is not large enough to significantly reduce the
value of VDD at any particular pixel.
• Lifetime issues in the OLEDs themselves will ultimately reduce their brightness
at a given current level, leading to the requirement for increased drive currents.
Given the sensitivity of the eye to small local variations in brightness, the often
quoted OLED lifetime to 50 % reduction in brightness overestimates the life of
the display.
In order to address these issues, many different designs of pixel drive circuit
have been explored [58, 59, 61], with TFT numbers as high as ten/pixel [61].
These include two classes of pixel circuit design, which can be classified as
‘compensated’, and ‘current mode’ pixels [61]. In the ‘compensated’ pixels, the
pixel circuit detects and corrects for VT and/or VOL changes [55, 59], whilst, with
the ‘current mode’ pixels, current mirror circuits are used to directly set the desired
OLED current, during the select interval, and this setting is stored as a voltage on
the gate of the driver TFT [61].
An alternative approach to dealing with these non-uniformity effects, particu-
larly in a-Si:H and poly-Si AMOLEDs, has been to use an a-Si:H photo-TFT in the
pixel. This senses the optical output from the OLED, and sets the gate voltage on
the drive TFT to ensure the correct LED luminescence [62]. This 5T2C circuit
automatically compensates for both TFT and OLED instabilities.
Further reference to the AMOLED application can be found in Sects. 7.6 and
9.4.3.3, dealing with poly-Si and AOS TFTs, respectively. However, more detailed
discussion of the various AMOLED pixel circuits is beyond the scope of this book,
but further information on this topic can be gained from the cited references, and
the topic is a regular feature in the annual SID Conference Digests.
4.7 Summary
Active matrix liquid crystal displays, AMLCDs, are today’s most ubiquitous flat
panel display product, and cover the complete range of display sizes and appli-
cations from small portable products, like cell phones, to the largest LCTV dis-
plays. As the name suggests, these are based upon the use of LC materials, and the
basic twisted nematic LC material concepts are covered in sufficient detail to
enable the reader to understand the operating principle of LCDs. In order to be
used in high quality displays, the LC pixels are addressed by a matrix of thin film
transistors, and the principles of active matrix addressing have been presented,
together with the key considerations of pixel design. During the development of
today’s high quality displays, many performance artefacts of AMLCDs have been
identified and rectified, and a number of these have been reviewed, including
viewing angle limitations, effects underlying flicker and cross-talk, and the
importance of bus bar resistances.
Whilst AMLCDs are the dominant display technology, other display products,
based upon different principles, have emerged in recent years. The two most
prominent are electrophoretic displays, EPDs, for e-reader products, and organic
light emitting diodes, OLEDs, for some portable products. As with AMLCDs,
these are also driven by an active matrix of thin film transistors, but their basic
image generation processes are different, and their operating principles have been
described.
References 103
References
1. DisplaySearch PanelTrack December 2011 Results. http://www.displaysearch.com/cps/rde/

xchg/displaysearch/hs.xsl/resources_paneltrack.asp. Accessed Jan 2012
2. Comiskey B, Albert JD, Yoshizawa H, Jacobson J (1998) An electrophoretic ink for all-
printed reflective electronic displays. Nature 394:253–255
3. Ritter J (2001) Active-matrix electrophoretic displays. In: Proceedings Asia Display/IDW’01,
pp 343–346
4. Baldo MA, O’Brien DF, You Y, Shoustikov A, Sibley S, Thompson ME, Forrest SR (1998)
Highly efficient phosphorescent emission from organic electroluminescent devices. Nature
95:151–154
5. French I (2009) Flexible e-books. In: SID’09 Digest, pp 100–103
6. Burns SE (2010) Flexible displays made with plastic electronics. In: SID Inform Display, vol
2. pp 16–19
7. Choi JB, Park C-H, Chung I-D, Lee K-H, Min HK, Kim C-W, Kim SS (2009) Line-scan
sequential lateral solidification process for AMOLED application. In: SID’09 Digest, pp
88–91
8. Kim HD, Jeong JK, Chung H-J, Mo Y-G (2008) Technological challenges for large-size
AMOLED display. In: SID’08 Digest, pp 291–294
9. Yang D-K, Wu S-T (2006) Fundamentals of liquid crystal devices, chapter 1. Wiley, NY
10. Pauluth D, Tarumi K (2005) Optimization of liquid crystals for television. J SID
13(8):693–702
11. Schadt M, Helfrich W (1971) Voltage-dependent optical activity of a twisted nematic liquid
crystal. Appl Phys Lett 18(4):127–128
12. Heckmeier M, Lussem G, Tarumi K, Becker W, Liquid crystals for active matrix displays.
http://www.merck-chemicals.com/united-kingdom/lcd-emerging-technologies/brochures/
c_8Veb.s1Okg4AAAEW3tYW4z8T;sid=acMAFlesq4AAFhtq0rUSQf9susCFCZ-4ruw-
MfhtTPPt50ewahr9Ln6rSDcZUlTkQ80lJrpdusCFCYSv7NwlJrpd. Accessed Oct 2010
13. Kondo K (2003) Recent Progress of IPS-TFT-LCDs. In: Proceeding IDMC’03
14. Ohmuro K, Kataoka S, Sasaki T, Koike Y (1997) Development of super-high-image-quality
vertical-alignment-mode LCD. In: SID’97 Digest, pp 845–848
15. Takeda A, Kataoka S, Sasaki T, Chida H, Tsuda H, Ohmuro K, Sasabayashi T, Koike Y,
Okamoto K (1998) A super-high image quality multi-domain vertical alignment LCD by new
rubbing-less technology. In: SID’98 Digest, pp 1077–1080
16. Ueki T (2000) Requirements for large size and high resolution TFT-LCDs. In: Proceeding
IDMC 2000, pp 177–181
17. Brotherton SD, McCulloch DJ, Gowers JP, Trainor MJ, Edwards MJ, Ayres JR (1996) Film
thickness effects in laser crystallised poly-Si TFTs. In: Proceeding AMLCD’96, pp 21–24,
(Kobe, 1996)
18. Ohori T, Takei M, Takahara K, Okabe M, Abe M (1996) Low-temperature poly-Si SXGA
TFT-LCDs with monolithic drivers. In: SID’96 Digest, pp 673–676
19. Yamamoto Y (2012) LCD processing and testing. In: Chen J, Cranton W, Fihn M (eds)
Handbook of visual display technology, Springer, Berlin
20. Sarma KS (1999) Active matrix liquid crystal displays. Lecture based on SID ‘99 Seminar.
http://hackman.mit.edu/6976/LHandouts/6.976.ST01.Lecture18-3.pdf. Accessed Oct 2010
21. Brotherton SD, Ayres JR, Edwards MJ, Fisher CA, Glaister C, Gowers JP, McCulloch DJ,
Trainor M (1999) Laser crystallised poly-Si TFTs for AMLCDs. Thin Solid Films
337:188–195
22. Howard WE (1995) Limitations and prospects of a-Si:H TFTs. J SID 3(3):127–132
23. Koike J, Hirota K, Naito M, Yun P, Sutou Y (2010) Cu-Mn electrodes for a-Si TFT and its
electrical characteristics. In: SID’10 Digest, pp 1343–1346
24. Lippens P (2012) Indium Tin Oxide (ITO): sputter deposition processes. In: Chen J, Cranton
W, Fihn M (eds) Handbook of visual display technology. Springer, Berlin, p 2012
25. Parry Jones L (2012) Alignment properties of liquid crystals. In: Chen J, Cranton W, Fihn M
(eds) Handbook of visual display technology. Springer, Berlin, p 2012
26. Kamiya H, Tajima K, Toriumi K, Terada K, Inoue H, Yokoue T, Shimizu N, Kobayashi T,
Odahara S, Hougham G, Cai C, Glownia JH, von Gutfeld RJ, John R, Lien S-CA (2001)
Development of one drop fill technology for AM-LCDs. In: SID’01 Digest, pp 1354–1357
27. Schwedler W, Nguyen F (2010) LED backlighting for LCD TVs. In: SID’10 Digest,
pp 1091–1096
28. DisplaySearch Market Update: http://www.displaysearch.com/cps/rde/xchg/displaysearch/
hs.xsl/101109_q3_large_area_tft_shipments_and_revenues_fell_as_supply_chain_sought_
inventory_reductions.asp. Accessed Nov 2010
29. Kazlas P, Au J, Geramita K, Gates H, Steiner M, Honeyman C, Drzaic P, Schleupen K,
Wisnieff B, Horton R, John R (2001) 12.1’’ SVGA microencapsulated electrophoretic active
matrix display for information appliances. In: SID’01 Digest, pp 152–155
30. Amundson K, Sjodin T (2006) Achieving graytone images in a microencapsulated
electrophoretic display. SID’06 Digest 1918–1921
31. Henzen A, van de Kamer J, Nakamura T, Tsuji T, Yasui M, Pitt M, Duthaler G, Amundson
K, Gates H, Zehner R (2003) Development of active matrix electronic ink displays for
handheld devices. SID’03 Digest 176–179
32. Gates H, Zehner R, Doshi H, Au J (2005) A5 sized electronic paper display for document
viewing. In: SID’05 Digest, pp 1214–1217
33. Lu C-M, Chen H-H, Liao J-S, Weng M-C, Chu Y-N, Tseng W-T, Yu H-T (2010) High
integration and functionality controller for active matrix electrophoretic display. In: SID’10
Digest, pp 374–378
34. Duthaler G, Au J, Davis M, Gates H, Hone B, Knaian A, Pratt E, Suzuki K, Yoshida S, Ueda
M, Nakamura T (2002) Active-matrix color displays using electrophoretic ink and color
filters. In: SID’02 Digest, pp 1374–1377
35. E-ink Triton brochure: http://www.eink.com/sell_sheets/eink_brochure.pdf. Accessed Nov
2010
36. Seki S, Uchida M, Sonoyama T, Ito M, Watanabe S, Sakai S, Miyashita S (2009) Current
status of printing OLEDs. In: SID’09 Digest, pp 593–596
37. Xia S, Cheon K-O, Brooks JJ, Rothman M, Ngo T, Hett P, Kwong R C, Inbasekaran M,
Brown JJ, Sonoyama T, Ito M, Seki S, Miyashita S (2009) Printable phosphorescent organic
light-emitting devices. J SID 17(2):167–172
38. Tang CW, VanSlyke SA (1987) Organic electroluminescent diodes. Appl Phys Lett
51(12):913–915
39. Forrest SR (2003) The road to high efficiency organic light emitting devices. Org Electron
4(2–3):45–48
40. Baldo MA, O’Brien DF, You Y, Shoustikov A, Sibley S, Thompson ME, Forrest SR (1998)
Highly efficient phosphorescent emission from organic electroluminescent devices. Nature
395:151–154
41. O’Brien DF, Baldo MA, Thompson ME, Forrest SR (1999) Improved energy transfer in
electrophosphorescent devices. Appl Phys Lett 74(3):442–444
42. Adachi C, Baldo MA, Thompson ME, Forrest SR (2001) Nearly 100% internal
phosphorescence efficiency in an organic light emitting device. J Appl Phys
90(10):5048–5051
43. Kondakov DY (2009) Role of triplet–triplet annihilation in highly efficient fluorescent
devices. J SID 17(2):137–144
44. Adamovich VI, Cordero SR, Djurovich PI, Tamayo A, Thompson ME, Andrade BWD,
Forrest SR (2003) New charge-carrier blocking materials for high efficiency OLEDs. Org
Electron 4(2–3):77–87
45. Lih J-J, Sung C-F, Weaver M S, Hack M, Brown JJ (2003) A phosphorescent active-matrix
OLED display driven by amorphous silicon backplane. In: SID’03 Digest, pp 14–17
References 105
46. Hack M, Chwang A, Tung Y-J, Hewitt R, Brown J, Lu J, Shih C, Ho J, Street RA, Moro L,
Chu X, Krajewski T, Rutherford N, Visser R (2005) Status and opportunities for high
efficiency OLED displays on flexible substrates. Mater Res Soc Symp Proc 870E, H3.1.1
47. Giebink NC, D’Andrade BW, Weaver MS, Mackenzie PB, Brown JJ, Thompson ME, Forrest
SR (2008) Intrinsic luminance loss in phosphorescent small-molecule organic light emitting
devices due to bimolecular annihilation reactions. J Appl Phys 103:044509-1–044509-9
48. Becker H, Bach I, Holbach M, Schwaiger J, Spreitzer H (2010) Purity of OLED-materials
and the implication on device-performance. In: SID’10 Digest, pp 39–42
49. Herron N, Gao W (2010) Advanced materials for printed OLED displays. In: SID’10 Digest,
pp 469–472
50. Scholz S, Meerheim R, Lüssem B, Leo K (2009) Exciton induced chemical reactions in
organic light emitting devices. In: SID’09 Digest, pp 681–684
51. Kondakov DY, Young RH (2009) Role of chemical transformations of hole transport
materials in operational degradation of the current generation of highly efficient fluorescent
OLEDs. In: SID’09 Digest, pp 687–690
52. den Boer W (2005) Active matrix liquid crystal displays. Elsevier, Oxford
53. Sasaoka T, Sekiya M, Yumoto A, Yamada J, Hirano T, Iwase Y, Yamada T, Ishibashi T,
Mori T, Asano M, Tamura S, Urabe T (2001) A 13.0-inch AM-OLED display with top
emitting structure and adaptive current mode programmed pixel circuit (TAC). In: SID’01
Digest, pp 384–387
54. Yoo JS, Lee H, Kanicki J, Kim C-D, Chung I-J (2007) Novel a-Si:H TFT pixel circuit for
electrically stable top-anode light-emitting AMOLEDs. J SID 15(8):545–551
55. Lu M-H, Weaver M S, Zhou TX, Rothman M, Kwong R C, Hack M, Brown JJ (2002) High-
efficiency top-emitting organic light-emitting devices. Appl Phys Lett 81(21):3921–3923
56. Wu C-C, Lin C-L, Hsieh p-Y, Chiang H-H (2004) Methodology for optimizing viewing
characteristics of top-emitting organic light-emitting devices. Appl Phys Lett
84(20):3966–3968
57. Hildner ML (2005) OLED Displays on Plastic. In: Crawford G P (ed) Flexible flat panel
displays. Wiley, England
58. Sanford JL, Libsch FR (2003) TFT AMOLED pixel circuits and driving methods. In: SID’03
Digest, pp 10–13
59. Jung S-H, Lee h-K, Park s-J, Ahn T-J, Lee s-W, Yoo J-S, Yoon S-Y, Kim C-D, Kang IB
(2009) A New AMOLED pixel compensating the combination of n-type TFT and normal
OLED device. In: SID’09 Digest, pp 1–4
60. Katsuhara M, Yagi I, Yumoto A, Noda M, Hirai N, Yasuda R, Moriwaki T, Ushikura S,
Imaoka A, Urabe T, Nomoto K (2010) A flexible OLED display with an OTFT backplane
made by scalable manufacturing process. J SID 18(6):399–404
61. Stewart R G (2010) Active matrix OLED pixel design. In: SID’10 Digest, pp 790–793
62. Fish D, Young N, Deane S, Steer A, George D, Giraldo A, Lifka H, Gielkens O, Oepts W
(2005) Optical feedback for AMOLED display compensation using LTPS and a-Si:H
technologies. In: SID’05 Digest, pp 1340–1343
Part II
TFTs
Chapter 5
Hydrogenated Amorphous Silicon TFT
Technology and Architecture
Abstract Hydrogenated amorphous silicon TFTs are the work-horse of the active
matrix flat panel display industry, and the architecture and fabrication processes of
these devices are described in this chapter. The properties of a-Si:H, including its
meta-stability, are briefly summarised as background to later sections, which
include a description of the inverted staggered device architecture, and a consid-
eration of TFT layout issues. The semiconductor and dielectric layers in the TFT
are deposited by plasma enhanced chemical vapour deposition, PECVD, and the
current implementation and understanding of these processes are presented.
Finally, some novel a-Si:H TFT structures are described, including self-aligned
and short channel TFTs, as well as high stability devices deposited under condi-
tions of enhanced hydrogen dilution of the PECVD reactant gases.
5.1 Introduction
Hydrogenated amorphous silicon TFTs dominate the current flat panel display
industry, particularly for active matrix LCDs and for active matrix electrophoretic
displays. The technology of these devices is presented in this chapter, and their
performance and device physics issues are presented in Chap. 6.
In order to provide some technical background to the later sections, Sect. 5.2
contains a brief overview of the material properties of hydrogenated amorphous
silicon, a-Si:H, and these are discussed in greater detail in Chap. 6. The device
architecture is described in Sect. 5.3, and this is followed by a discussion of TFT
layout issues in Sect. 5.4. These issues apply not only to a-Si:H TFTs, but to the
other TFT technologies as well.
Device quality a-Si:H is an alloy of Si with *10 % hydrogen. The hydrogen
plays a key role in defect passivation within the material, by reducing the dangling
bond density from *1020 cm-3, in H-free a-Si, to *1016 cm-3 in current device
material [1]. The performance characteristics of a-Si:H TFTs are determined by
the deposition conditions of the material itself, and the preferred process for this is
large area, low temperature plasma enhanced chemical vapour deposition,

110 5 Hydrogenated Amorphous Silicon TFT Technology and Architecture
PECVD. The process of RF glow discharge deposition was identified, in the

pioneering work on a-Si [2, 3], as delivering material with a low trap state density,
whose surface potential could be controlled in an insulated gate, field effect
structure, and which could also be doped (although later work showed that the
doping efficiency was very low [4]). In this early work, uniform a-Si:H deposition
was achieved over areas of *10 cm 9 10 cm [5], and subsequent industrial
development has led to the present-day commercial PECVD reactors, which can
handle substrates in excess of 2 m 9 2 m [6]. The present understanding of the
PECVD deposition process is described in Sect. 5.5, and covers the deposition of
both undoped and doped a-Si:H, as well as the gate dielectric a-SiNx:H.
The TFT structures and deposition processes described in Sects. 5.3 and 5.5
deal with the standard a-Si:H TFT of the FPD industry, and Sect. 5.6 reviews some
alternative structures and processes which have been investigated to address some
of the limitations of the current device. However, it should be noted that none of
these alternative processes has yet been industrialised.
5.2 a-Si:H Material
In order to introduce some of the concepts used in Sect. 5.5, a brief overview of the
properties of a-Si:H follows. A comprehensive treatment of the properties of
a-Si:H is contained in Ref. [7].
As an amorphous material, a-Si has short range order, given by the preferred
four-fold coordination of silicon, but no long range order, due to the variety of
bond angles and bond lengths, which can be accommodated in the material.
Compared with crystalline silicon, which has a bond length and tetrahedral
bonding angle of 2.35Å and 109.5°, respectively [7, 8], the rms deviation from
these values in a-Si:H are *2 % and *9–10 %, respectively [1, 8]. This disorder
leads to a corresponding distribution of bond energies, and the occurrence of weak
bonds (WBs) within the material, which play an important role in its defect
structure. In general, the silicon atoms will be four-fold coordinated, with coor-
dination defects occurring where the local inter-atom strain breaks a weak bond.
This is illustrated by the bonding network diagram in Fig. 5.1, showing both
dangling bond (DB) defects, and a hydrogen-passivated dangling bond. Hydrogen
plays a key role in a-Si:H, which is an alloy of Si with *10 % hydrogen, and, in
high quality, hydrogenated amorphous silicon, the DB density is reduced
to *1016 cm-3, due to the presence of hydrogen forming the passivating Si–H
bonds at the DBs.
The dangling bond states are amphoteric, in that the neutral state, containing
one unpaired electron, can trap a further electron and become negatively charged,
or release the unpaired electron and become positively charged. Hence, these
defects have three charge states of positively charged, neutral, or negatively
charged with zero, one or two electrons, respectively. For defects with a positive
correlation energy, such as the dangling bond in a-Si:H [7] (and many deep level
5.2 a-Si:H Material 111
Fig. 5.1 Illustration of the

four-fold-coordinated
amorphous silicon random
network, containing dangling
bond defects (dashed line),
and a hydrogen passivated
dangling bond (solid circle)
defects in c-Si), the doubly occupied centre is above the singly occupied centre in
the band-gap, and the three charge states are shown as a function of the Fermi level
position in Fig. 5.2. The defect is negatively charged when the Fermi level is
above the upper level, positively charged when it is beneath the lower level, and
neutral when it is between the two levels. Moreover, the two electron level does
not exist until the DB centre already has at least one electron in it. As with the
Si–Si bonds themselves, the uncoordinated Si bonds have a range of carrier
binding energies, so that the dangling bonds give rise to a double distribution of
deep-lying centres in the forbidden band gap.
In addition, the lack of a long range regular atomic potential (in contrast to that
found in single crystal Si) results in localised, as well as extended, electron states.
The localised states are characterised by a distribution of band tail states in a-Si:H,
extending from both the conduction and valence bands into the forbidden band
gap, defining a ‘mobility gap’ of 1.85 eV [7] (which is significantly larger than the
room temperature band gap of 1.12 eV for c-Si). In poor quality material, with
high defect densities within the mobility gap, extended state conduction is neg-
ligible, and carrier conduction is by localised inter-trap hopping. For device
quality material, conduction is by excitation of trapped carriers into extended
states, but, as only a small fraction of the total trapped charge is thermally released
from the traps, this contributes to the low carrier mobility values measured in
a-Si:H. In fact, due to the high density of band tail states, it has not been possible
to raise the Fermi level above the mobility edge, and pure extended state
Fig. 5.2 Illustration of the electron occupancy and charge state of an amphoteric dangling bond,
as a function of the Fermi level position (a) unoccupied centre, and positively charged lower
level, Ed1 (b) singly occupied centre, and neutral lower and upper levels, and (c) doubly occupied
centre, and negatively charged upper level, Ed2
Fig. 5.3 Density of states 22

10
distribution of a-Si:H Band States
21
10 Donor-like States Acceptor-like States
Density of States (cm –3 eV–1)

permission of IEEE) 20 Tail States
10
19
10
18
10 Deep States
17
10
16
10
Dangling Bonds
15
10
EV EC
14
10
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
Energy (EF – EV) (eV)
conduction has not been observed. The density of states distribution in a-Si:H is
shown in Fig. 5.3 [9], illustrating both the band tail states and deep states, the
density of which is given by the summation of the dangling bond densities. The
influence of the mobility gap states on carrier mobility in a-Si:H is discussed
further in Sect. 6.2.1.
Hence, the major defects in a-Si:H are dangling bonds and weak bond states,
and hydrogen-mediated equilibration occurs between the larger number of weak
bonds, and the smaller density of dangling bond and hydrogen passivated states
[11]. The details of this interaction are described by the defect pool model [10].
The interaction between hydrogen, the DBs and the WBs after growth determines
the DB density under a given set of conditions, although the a-Si:H deposition
conditions determine the background WB density [11]. The material is defined as
meta-stable as the defect structure is frozen-in, at *500 K, during cooling from
the growth temperature, although the precise freeze-in temperature will depend
upon the cooling rate. Moreover, given enough time at lower temperatures, a new
equilibrium defect density will develop during a thermally activated relaxation
process. This has an activation energy of *1.0–1.5 eV [7, 12], and the temper-
ature dependent equilibrium defect densities have an activation energy
of *0.2 eV [12]. The defect density within the material also equilibrates in
response to other stimuli, such as changes in doping level [4], electron–hole pair
recombination induced by optical illumination (the Staebler-Wronski effect) [13–
15], and changes in the free carrier density in the TFT channel (induced by the gate
bias). This latter dependence underlies the well established bias-stress instability
displayed by a-Si:H TFTs [10, 16]. A fuller discussion of these meta-stability
effects, and of the density of states in the material, is presented in Chap. 6.
5.3 a-Si:H TFT Architecture 113
5.3 a-Si:H TFT Architecture
Figure 5.4a shows the most common architecture of a-Si:H TFTs, which is an
inverted staggered structure, and consists of a bottom metal gate, a non-stoichi-
ometric silicon nitride, a-SiNx:H, gate dielectric layer, and the undoped a-Si:H
transistor body on top of this. The contacts to the TFT are n+ doped a-Si:H source
and drain regions, which, in turn, are contacted by lower resistance metal pads.
The device structure is ‘inverted’ in that the gate is beneath the channel, and it is
‘staggered’ in that the source and drain contact regions are not coplanar with the
electron channel at the a-Si:H/a-SiNx:H interface, but on the opposite side of
the film. Hence, the channel current will need to flow through the thickness of the
a-Si:H layer to give current continuity between the source and drain electrodes.
Whilst the structure is fundamentally that of an IGFET (metal/insulator/semi-
conductor structure), it is superficially quite different from the c-Si MOSFET
architecture (discussed in Chap. 3), which is top-gated, and coplanar (with the
source and drain contact regions on the same level as the channel). However,
the architecture, which has been developed for the a-Si:H TFT, is a reflection of
the optimised structure which can be fabricated with acceptable performance,
within the constraints of a high throughput, low temperature process on large area
glass substrates (\ 350 °C and [ 2 m 9 2 m, respectively). Two other major
differences from MOSFETs (which are used in p- and n-channel complementary,
CMOS, circuits), are that, firstly, only n-channel a-Si:H TFTs are used in com-
mercial applications. This is due to the poorer performance of p-channel devices,
which is discussed further in Chap. 6. Secondly, the source and drain regions are
not self-aligned to the gate edges, but require a photolithographic alignment stage,
giving significant parasitic capacitance between the gate and the source and drain
terminals. This latter effect has been discussed in Chap. 4, and contributes to
AMLCD performance artefacts.
As mentioned above, a key aspect of the a-Si:H TFT process is its high man-
ufacturing throughput. This is due, in part, to the low mask count, of four or less
photolithography stages, needed to fabricate it, and with an extra mask needed for
Fig. 5.4 Cross sectional

diagrams of inverted
staggered a-Si:H TFT
structures: (a) back-channel-
etched, BCE, and (b) etch-
stop, ES, architectures
an AMLCD. (In contrast, a sophisticated c-Si CMOS circuit can use 12 or more
mask stages).
There are two slightly different TFT architectures shown in Fig. 5.4, which
differ by the presence of an a-SiNx:H layer on top of the TFT in Fig. 5.4b. These
two architectures have different fabrication schedules, and the industry standard,
shown in Fig. 5.4a, is referred to as a back-channel-etched, BCE, structure, whilst
that in 5.4b is an etch-stop, ES, structure. The fabrication of the BCE and ES
structures are described below, and, following this, the differences between them
are discussed.
5.3.1 Back-Channel-Etched TFT Fabrication
The fabrication of the a-Si:H TFT consists of a sequence of material deposition

stages, each followed by patterning of the layers into the desired structures.
The process used for film patterning consists of photolithography [17] and etching
[18–20] stages, which are described further in Sect. 5.4.1.
Figure 5.5 shows the main process steps in fabricating an inverted staggered
BCE a-Si:H TFT. The first stage is the sputter deposition, and photolithographic
definition, of the metal gate electrode, by mask stage M1. The gate electrodes are
then coated by a stack of three consecutively deposited films of a-SiNx:H, undoped
a-Si:H, and n+ a-Si:H, in a plasma enhanced chemical vapour deposition reactor
[6] (described in Sect. 5.5). The film thicknesses for commercial TFTs are typi-
cally 300 nm a-SiNx:H, 200 nm a-Si:H, and 50 nm n+ a-Si:H [6]. In the second
patterning stage, M2, the triple stack is defined into the individual TFT islands.
The next two stages consist of the deposition of the source and drain contact
metals, and their definition into tracks and contact pads by mask stage M3. This
metallisation pattern is then used as an etch mask to selectively remove the n+
layer from those areas of the TFT island stack where it is not protected by the
metal. The etching of the n+ region removes what would otherwise be a channel-
shorting layer between the source and drain contacts. The TFT fabrication process
is completed by the deposition of a final capping/passivation layer of a-SiNx:H
(not shown), which is patterned by mask M4 into contact holes down to the source,
drain and gate pads.
A critical stage in this process is the etch removal of the n+ layer between the
source and drain contacts. As the etch selectivity between the n+ layer and the
undoped a-Si:H is minimal [19], a well controlled etching process is required. This
is a precisely timed etch to ensure that the n+ layer is completely removed, whilst
the necessary over-etching into the underlying undoped a-Si:H is well controlled
and reproducible between plates. In order to accommodate the over-etching into
the undoped a-Si:H layer, a relatively thick layer of *200 nm is used. The n+
etching stage defines the channel length, L, of the TFT, which is given by the
separation between the edges of the source and drain n+ regions.
Fig. 5.5 Illustration of the back-channel-etched TFT fabrication steps (stage 6 is not shown)
In summary, the BCE process consists of four mask stages, two PECVD
depositions, and two metal sputter deposition stages.
5.3.2 Etch-Stop TFT Fabrication
Comparison of the BCE and ES architectures in Fig. 5.4 shows that the ES
structure differs mainly by the presence of an a-SiNx:H pad beneath the edges of
the n+ regions. This is to give greater etch selectivity [19] between the etching of
the n+ regions and the underlying a-SiNx:H, and, thereby, avoid the need to use a
critical timed etch. However, the cost of this is a more complex fabrication
schedule, as shown in Fig. 5.6. Following the first mask stage, M1, to define the
gate electrodes, a triple layer deposition of a-SiNx:H, a-Si:H and a-SiNx:H is
carried out, and the a-SiNx:H on top of the stack is defined into the etch stop pads
by mask M2. This is then followed by the separate deposition of the n+ layer, and
the definition of the device islands by mask stage M3. The source and drain contact
metals are deposited, and defined using mask M4, and, as with the BCE process,
the unwanted n+ material is removed using the metal pads as an etch mask.
However, in contrast to the timed BCE etch, the a-SiNx:H pad acts as a more
selective etch stop layer. This is, in principle, a more easily controlled process, and
permits the use of thinner a-Si:H layers (typically down to 100 nm). The final
process is the deposition of a capping layer of a-SiNx:H (not shown), and mask M5
is used to open contact windows down to the source, drain and gate contact pads.
Fig. 5.6 Illustration of the etch-stop TFT fabrication steps (stage 7 is not shown)
In the ES process, the size of the a-SiNx:H etch-stop island determines the
separation of those parts of the n+ source and drain regions which are in contact
with the TFT body, and, hence, defines the TFT channel length, L.
In summary, the ES process involves five mask stages, and three PECVD
depositions, but avoids a critical etching stage, whilst also delivering a potentially
better back-face passivation layer (in the first triple layer deposition) [21]. How-
ever, when comparing the process details with the BCE process in Table 5.1, it is
clear that, providing the back channel etch is well controlled, the BCE process is
simpler, with fewer steps. Hence, it can deliver higher manufacturing throughput
and lower overall plate cost, and is extensively implemented in production facil-
ities. For many research laboratories, the ES process is preferred for basic device
studies, as it is more easily implemented and controlled with less sophisticated
facilities than in production units.
Table 5.1 Comparison of back-channel-etched (BCE), and etch-stop (ES) a-Si:H TFT pro-
cessing stages
Process BCE ES
Photolithography 4-mask 5-mask
PECVD 2 3
n+ a-Si:H removal Timed etch into i a-Si:H Differential etching with
respect to SiNx:H
Final passivation layer Possible sensitivity Screened by 1st SiNx:H layer
In this book, the two a-Si:H TFT architectures are referred to as BCE and ES,
but it should be noted that an alternative naming convention exists in the literature,
in which the BCE and ES architectures are referred to as ‘bi-layer’ and ‘tri-layer’
structures, respectively [19]. In this nomenclature, reference is made to the number
of layers deposited, and no distinction is made between the deposition of the
i-a-Si:H and the n+ a-Si:H layers in the BCE structure, so that the ‘bi-layer’ device
refers to the two types of material deposited: a-SiNx and a-Si:H. In the ES
structure, the sequential deposition of a-SiNx, a-Si:H, and a-SiNx is counted as a
tri(ple)-layer deposition.
5.4 TFT Layout Considerations
This section will discuss some general TFT layout considerations, which are not
restricted to a-Si:H TFTs, but relevant to TFTs in general. However, as the dif-
ferent TFT technologies use different device architectures, the specific diagrams
will not necessarily apply to all technologies, but the broad points will.
5.4.1 Photolithography Process
Photolithography is the process whereby the different layers in the TFT are aligned
and patterned [17] in order to make the final TFT structure. This is accomplished
by optically imaging a pattern from a mask firstly onto a layer of photo-sensitive
material, the photo-resist, and then using this as an etching mask to selectively
remove the unprotected areas of the film by an etching process. The overall
procedure [17] consists of applying a thin layer of photo-resist to the plate, by
centrifugal spinning or by slit coating through a linear nozzle, and then exposing it
to UV light, which has been passed through the mask pattern. For negative photo-
resist, the resist will be hardened where it is exposed to the light, and for positive
resist it will be softened in these areas. Once the photo-resist layer has been
exposed, it is then developed to remove those areas of the film which have not
been hardened. Finally, the plate containing the developed photo-resist pattern is
exposed to an etching process, which can use either wet chemicals [18] or a dry
plasma etchant [19, 20]. Those areas of the film covered by the photo-resist will be
protected from the etchant, and the film will be removed elsewhere. Once
the etching process is complete, the photo-resist pattern will be removed from the
plate using a suitable solvent. In addition to the processing stages listed above, it is
conventional to bake the photo-resist layers at temperatures around 100 °C after
the resist application, and after the development stage to further harden it. The
pattern defined in the film will be a direct image of the dark areas on the mask
when positive photo-resist is used, or a negative image with negative photo-resist.
The performance of a given TFT, such as its on-state and off-state currents, will be
determined both by its material properties and by its channel width and length. These
latter parameters, together with the process layout and design rules, will determine the
overall TFT appearance and size. The design rules will specify the minimum overlaps
between different layers, and also the minimum feature sizes within a layer, including
the contact window dimensions. The overlap dimensions will be determined by both
the photolithographic alignment tolerances, and by the overall critical dimensional
control of the lithography process, including the etching stage. Equally, the final
feature sizes will be determined both by the resolution of the photolithography pro-
cess, and also by the etchant characteristics of the pattern definition stage.
The photolithographic aligners, which are used for TFT fabrication within the
FPD industry, have alignment and resolution specifications of typically ±0.6 lm
and 3.0–3.5 lm, respectively [22]. Based upon these figures, the minimum feature
size will be 3.0–3.5 lm, and, when the alignment accuracy and the tolerances in
the overall photolithographic process, including the etching stages, are taken into
account, the typical gate/drain overlap will be 2.0–3.0 lm.
A detailed description of the photolithography process [17], and of the wet [18],
and dry etching [19, 20] stages for commercial TFT fabrication can be found in the
cited references.
5.4.2 TFT Layout Issues
Figure 5.7a shows the layout of an inverted staggered TFT, such as a-Si:H, where
the source and drain contact pads are used to define the edges of the n+ regions.
Figure 5.7b shows the layout of a different TFT architecture, in this case a top-gated
coplanar structure, such as a non-self-aligned poly-Si TFT (see Chap. 7), in which
the TFT body is covered with a dielectric layer, and contact windows are needed for
the source and drain metals to make contact with the underlying doped regions.
Also, as the gate electrode, and the source and drain electrodes are at the same level
in this example, there will be a minimum clearance gap specified between these
electrodes. As with the layout in Fig. 5.7a, a clear overlap is shown between the
gate metal and the doped source and drain regions. Taking, for example, a TFT with
a 5 lm long channel, the minimum gate length in both layouts will be 11 lm to
ensure 3 lm overlap with the doped regions. Whilst the design rules will specify
minimum dimensions, in many cases the actual dimensions will exceed these
minimum values, due, for instance, to TFT performance requirements demanding a
channel width or length greater than the minimum allowed.
If the minimum layout tolerances are not adhered to, the situation depicted in
Fig. 5.7c can arise, in which the gate does not overlap the doped region, but leaves
a portion of the TFT channel area, which is not controlled by the gate electrode.
This will result in a high channel resistance, giving reduced on-currents, and
possibly asymmetric TFT characteristics, depending upon which electrode is used
as the source or drain terminal.
5.4 TFT Layout Considerations 119
Fig. 5.7 Illustrative TFT layouts showing: (a) an inverted staggered TFT, e.g. a-Si:H (b) a top
gated coplanar TFT, e.g. poly-Si (c) source-gate misalignment, and (d) channel-gate
misalignment
Another common feature to the layouts in Fig. 5.7a and b, is the extension of
the gate electrode beyond the width of the TFT island. This is to ensure that the
designed channel width, W, determines the device characteristics, and that the gate
electrode fully controls the currents flowing through the channel. An example of
misalignment in this direction is shown in Fig. 5.7d, in which the gate does not
completely overlap the TFT island. In this case, the on-current will be less than
expected for a channel width of W, and, in the off-state, the current in the
un-modulated part of the channel may be higher than the minimum current
measured in a correctly processed TFT. (This will be the case for a TFT in which a
non-zero gate voltage is needed to achieve the minimum off-state current). A
comparable situation can also arise if the semiconductor layer itself is not pat-
terned, and extends over an area larger than the gate electrode. In this case, not
only will the off-state currents be larger than expected, but fringing fields from the
gate, source and drain electrodes may also draw a larger than expected on-state
current, for the given values of W and L. This will result in an overestimate of the
carrier’s field effect mobility, if it is calculated using Eqs. 3.12 or 3.18.
5.5 Plasma Enhanced Chemical Vapour

Deposition, PECVD
PECVD is used for the deposition of the key non-metallic layers in the TFT:
namely, the a-SiNx:H gate insulator, the undoped a-Si:H for the transistor body,
and the phosphorus doped n+ a-Si:H for the source and drain regions. These layers
Fig. 5.8 Schematic drawing

of AKT PECVD processing
chamber (Reprinted from [6]
with permission of IEEE)
are deposited as a sequential stack in the BCE structure, without breaking the
system vacuum, in order to minimise the growth of unwanted interfacial films
between layers, and to suppress ambient contamination. Also, to suppress cross-
contamination, these deposition processes are carried out in separate chambers,
with robotic plate transfer between chambers, and the equipment is usually con-
figured as a cluster tool in current manufacturing systems [6]. Apart from
depositing material of the desired quality, the other requirements of the deposition
system are that the depositions can be carried out with a high level of uniformity
over large areas (which are currently in excess of 2 m 9 2 m), and at low enough
temperatures (B350 °C) to be compatible with the use of glass substrates.
A PECVD deposition chamber is illustrated in Fig. 5.8 [6]. This is a parallel
plate reactor, housed in a vacuum chamber, and consists of an earthed, heated
substrate holder (the susceptor), a gas shower head, which has been engineered in
such a way that the reactant gas distribution leads to uniform material growth over
the plate area, and the shower head is powered by a 13.56 MHz RF generator. The
basic reactant gas is silane, SiH4, for the deposition of intrinsic a-Si:H; phosphine
is added to this for the deposition of n+ a-Si:H, and ammonia is added for the
a-SiNx:H deposition. In addition to these basic gases, it is often common to run the
deposition process with dilutant carrier gases such as hydrogen [23–26], helium
[24, 26], argon [24], or nitrogen [26]. However, due to the key role played by
hydrogen both in the a-Si:H material, and in the deposition process itself [11, 27],
hydrogen is the most common dilutant for the deposition of the amorphous silicon
layer.
The optimisation of the deposition process is a complex multi-parameter task,
in which the variables display inter-dependent behaviour, and where the variables
are the substrate temperature, the RF power, the parallel plate spacing, the gas
pressure and mixture, and the gas residence time [28]. Some of the key features
affecting this optimisation are presented in the following subsections.
5.5 Plasma Enhanced Chemical Vapour Deposition, PECVD 121
The final feature shown in Fig. 5.8 is the in-situ plasma cleaning process to
remove deposition products from the chamber walls, and to suppress the build-up
of particulates in the chamber.
5.5.1 Undoped a-Si:H
The processes involved in the RF PECVD deposition of a-Si:H can be broken

down into the gas phase processes within the plasma, which determine the dis-
sociation products of the silane reactant gas, their transport and interaction with the
substrate surface, and their eventual incorporation into the growing film [11, 16,
28]. The plasma consists of the fractionally ionised gas and free electrons, and
collisions between the energetic electrons and the gas molecules dissociate the
silane into reactive radicals such as SiH3, SiH2, and hydrogen, as well as ions, all
of which may be involved in film growth. The radical formation products can be
represented by [7]:
SiH4 ¼) SiH3 þ H ð5:1Þ
SiH4 ¼) SiH2 þ H2 ð5:2Þ

Under high power, or high pressure conditions, larger molecules can also form,
such as [7]:
SiH4 þ SiH2 ¼) Si2 H6 ð5:3Þ
Si2 H6 þ SiH2 ¼) Si3 H8 ð5:4Þ

Although the bulk of the plasma is neutral, containing equal numbers of ions
and electrons, there is depletion of the highly mobile electrons close to the elec-
trodes, leading to the creation of a positive space charge region. This positively
charged region is referred to as the plasma sheath region, and, as a result of this,
the plasma sits at a positive potential with respect to the earthed susceptor. Ions
will be accelerated across the sheath region, and the effect of this acceleration is
determined by the gas pressure and the resulting mean free path. At low gas
pressures, there will be large mean free paths, leading to ion bombardment of the
substrate. For high gas pressures, there will be many particle collisions in the
sheath region, and the energy gained by the ions in the field will be largely
dissipated [16]. These two gas pressure regimes are referred to as the a- and
c-regimes, respectively, and are discussed further below.
The plasma deposition processes are schematically shown in Fig. 5.9 [28],
where the type of gas-phase process in the plasma is determined by the operating
conditions. For low pressures and low RF powers, the primary plasma products are
SiH3 and H radicals and ions. Moreover, the long mean free paths in the low
pressure environment support an appreciable ion flux, as well as the neutral radical
flux, to the substrate surface. However, in this regime, the power is poorly coupled
Fig. 5.9 Illustration of the

Electrical Power
processes involved in the Electron density
PECVD deposition of a-Si:H Energy distribution
permission of the IET)
SiH 4 Flow e–+ SiH 4 Pump
Primary Increasing Secondary
Plasma processes
reactions RF power, reactions
Pressure,
e–+ SiH 4 Geometry, SiHx + SiH 4
Flow
Si n H m
SiH x + H TS Clusters, Polymers
Powder
Ions, Radicals, Photons,...

Bulk Processes
Surface and
Growth zone Surface mobility

Bulk Chemical equilibrium
Substrate Temperature
into the plasma, and this, in combination with low dissociation rates, leads to low
deposition rates. In contrast, at higher pressures and/or higher RF powers, the
power is more efficiently coupled into the plasma, and the silane dissociation rate
is increased, leading to higher deposition rates. At the same time, secondary
plasma products are formed, such as the higher order SinHm molecules and
polymers, and gas phase reactions are stimulated, leading to what are referred to as
‘dusty’ plasmas. Under well controlled conditions, the negatively charged ‘pow-
der’ particles are electro-statically contained within the plasma, and high quality,
faster depositions can be obtained within this regime [28]. Also, the reduced mean
free path at these higher pressures reduces the ion flux to the substrate surface.
The dependence of the film growth rate on the total gas pressure, and the
hydrogen:silane dilution ratio, is shown in Fig. 5.10 [23, 25], and two regimes
have been identified in this data set. The first regime is the a-regime, giving low
growth rates (\100 Å/min), which varied linearly with pressure, and the second
regime is the c-regime giving the higher growth rates, and is associated with
‘dusty’ plasma conditions. The approximate boundary between the two regimes
was at *130 Å/min. The device and material characteristics were found to
depend upon the growth regime, and were most clearly correlated using samples
grown either in the a-regime, or in the saturated c-regime (the transition regime
gave variable results) [23]. The influence of the hydrogen dilution ratio was to
increase the a-c transition point to higher gas pressures, and to reduce the depo-
sition rate. In this context, the hydrogen can be regarded as an ‘etchant’ [29],
Fig. 5.10 Measured +

variation of the a-Si:H +
300
deposition rate, at 240 °C, as
Deposition Rate (Å/min)
+
a function of the deposition No H2 + +
+
+
pressure and the hydrogen: H2:SiH4 2:1
silane dilution ratio 200
++
(Reproduced with permission 4:1 6:1 +
from [23]) +
8:1
100
+
+
+
+
0
30 70 110 150 190
Deposition Pressure (Pa)
which selectively breaks weak Si–Si bonds to form volatile silicon hydrides
[29, 30], and, under conditions of strong hydrogen dilution ([95 %), leads to the
formation of micro-crystalline Si [29]. The a- and c-growth regimes can be
associated with the primary and secondary reaction plasma regimes, respectively
[23].
As mentioned above, the two different deposition regimes influenced the per-
formance of TFTs made in these materials, as shown by the carrier mobility and
threshold voltage stability results in Fig. 5.11a and b, respectively [23]. These
parameters are shown as a function of deposition temperature, with the gas
pressure/deposition regime as the independent variable. The figures showed that
the a-regime gave the higher electron mobility below *300 °C, but the two
(a) (b) 12
40 Pa (α-TFTs)
0.8
Saturated Mobility (cm2 V –1 s–1)
Threshold Voltage Shift (V)
70 Pa (α-TFTs)
40 and 70 Pa 8
(α-TFTs)
0.4
4
130 and 160 Pa

(γ-TFTs) 130 and 160 Pa
(γ-TFTs)
0 0
210 250 290 330 210 250 290 330
Deposition Temperature (°C) Deposition Temperature (°C)
Fig. 5.11 Variation of device properties with deposition temperature and pressure, for
depositions with a H2:SiH4 dilution ratio of 2:1 (a) saturation regime electron mobility, and
(b) change in threshold voltage after a gate bias stress of 30 V for 1,000 s (Reproduced with
regimes were comparable above this temperature. In contrast, the c-regime TFTs
showed better gate-bias stability, although for both regimes device stability
improved with increasing deposition temperature. The optimisation of deposition
conditions for device fabrication is to maximise electron mobility, whilst mini-
mising device instability. However, for commercial devices, priority is usually
given to stability considerations, in view of its role in determining product life-
time. Hence, the c-regime deposition at *300–330 °C is favoured for the indus-
trial fabrication of a-Si:H TFTs. The other constraining factor in commercial
depositions is the plate throughput rate, which places a premium on higher
deposition rates. In order to optimise both device performance and deposition rate,
it is common practice to deposit the a-Si:H film at two different deposition rates
[6]. For instance, for a 200 nm thick intrinsic layer in the standard back-channel-
etched TFT, the first 50 nm (containing the channel region) are deposited at a
slower rate than the final 150 nm [6].
The plots in Fig. 5.11 also illustrate the issues which arise with lower tem-
perature depositions, particularly the reduced gate-bias stability. These effects are
discussed further in Sect. 11.4, dealing with a-Si:H TFT processing on low tem-
perature, plastic substrates.
In addition to the dependence of the key device parameters on the deposition
regime and temperature, the a-Si:H material itself also shows a variation in its
hydrogen bonding pattern. This is usually assessed by infra-red spectroscopy,
where the different silicon-hydrogen bonding states are identified by their char-
acteristic vibrational absorption spectra. An illustration of the different bonds, and
their vibrational modes, are shown in Fig. 5.12 [31]. The ‘SiH’ bond represents the
:Si–H bonding configuration, in which the H is passivating a silicon dangling
bond, and its stretching mode has an absorption peak at a wave-number of
2000 cm-1. The poly-hydride ‘SiH2’ and ‘SiH3’ bonds represent the = Si = H2
and -Si:H3 bonding configurations, respectively, which are more prevalent under
high hydrogen concentrations, and have been associated with void formation in the
a-Si [7].
FTIR absorption spectra of the a- and c-regime depositions over the tempera-
ture range 210–270 °C are shown in Fig. 5.13 [23]. For the a-regime, the spectra
were dominated by the SiH stretching mode at 2,000 cm-1, whereas, in the
c-regime, there were contributions from the stretching modes of both the SiH and
SiH2 bonds at 2,000 and 2,090 cm-1, respectively, plus the 840–880 cm-1 dou-
blet. In addition, the SiH2 bond density decreased with increasing deposition
temperature. This reduction in SiH2 density was accompanied by a reduction in the
total chemical concentration of hydrogen within the c-regime films [23]. For
depositions at 300 °C and above, the FTIR spectra for both deposition regimes
were the same, showing just the SiH peaks, plus a reduced concentration of the
840–880 cm-1 peaks [23]. When comparing the bond configuration results with
the TFT results in Fig. 5.11, a link was established between the deposition regime,
the specific hydrogen bonding structure, and the TFT performance, at least in
terms of electron mobility. However, the device stability appeared to be controlled
by factors other than the type of Si/H bonds identified by the FTIR absorption
Fig. 5.12 Schematic illustration of the SiHn vibrational modes, and their associated frequencies,
for different SiH bonding configurations (Reprinted with permission from [31]. Copyright (1979)
by the American Physical Society)
(a) Single 2000 cm–1 and 860 (b) 2000 and 2090 cm –1
cm –1 absorption peaks stretching peaks
–1
840-880 cm
Absorbance (A.U)
Absorbance (A.U)
Tdep 210°C
doublet
240°C Tdep 210°C
240°C
270°C
270°C
2200 1700 1200 700 2200 1700 1200 700

–1 –1
Wavenumber (cm ) Wavenumber (cm )
Fig. 5.13 FTIR absorption spectra from a-Si:H layers deposited over the temperature range
210–270 °C, and with a H2/SiH4 dilution ratio of 2:1. Depositions in: (a) the a-regime at 40 Pa,
and (b) the c-regime at 130 Pa. (Reproduced with permission from [23])
peaks [23, 25]. For example, the films deposited in the a-regime displayed
increasing levels of intrinsic compressive strain as the deposition temperature
reduced (which was attributed to the ion flux in this regime), and this was asso-
ciated with the increased gate bias instability [25].
In more general terms, a high relative density of SiH bonds has been widely
used as a figure of merit for high quality a-Si:H films [7], and the maximisation of
the SiH:(SiH2 ? SiH) bond ratio has been empirically used to optimise the
deposition conditions, particularly under non-standard conditions, such as at
reduced temperatures [24, 26].
So far as the growth process itself is concerned, one of the objectives in
understanding the mechanism has been to explain the relatively low concentration
of *10 % hydrogen in the final film, compared with the 75 % hydrogen con-
centration in the SiH3 growth radical [32], and how the growth process influences
the final defect density within the film [11]. In order to account for this, the growth
models have made a distinction between the hydrogen-rich surface layer of the a-
Si:H, and the low hydrogen-content of the deeper film, with the mobile SiH3
radical playing a role in both abstracting hydrogen from the surface, as well as
bonding to it [11, 27, 28, 32]. This is illustrated by the atomistic model in
Fig. 5.14, in which the SiH3 radical is shown interacting with the hydrogen-ter-
minated surface in a variety of ways [11]. The primary reaction is physisorption of
the SiH3 radical onto the hydrogenated surface, forming a weak Si–H-Si bond,
which permits the radical to diffuse between adjacent sites until it encounters a
silicon dangling bond, where a more permanent Si–Si bond is formed. Alterna-
tively, the radical can abstract a hydrogen atom, and be released back into the gas
phase as a SiH4 molecule, thereby creating another silicon dangling bond. With
increasing temperature, hydrogen can also be thermally desorbed from the surface
creating more dangling bonds, and, hence, increasing the growth rate. Although
Fig. 5.14 Illustrative

diagram of surface processes
during the PECVD growth of
a-Si:H (the open and solid
circles represent H and Si
atoms, respectively).
(Reproduced with permission
from [11])
desorption of the radical from the surface is also shown, this is assumed to be at a
negligible rate compared to the other processes [11].
The hydrogen reduction mechanism, when comparing its concentration in the
bulk and in the surface layers, occurs by two parallel processes, driven by the
substrate temperature and by the plasma conditions (at least over the temperature
range of 250 to 400 °C) [11]. The thermal process is one in which the hydrogen
concentration is driven down to its equilibrium solubility value of several percent
by an out-diffusion process of atomic hydrogen from a sub-surface layer to the
surface, where it is released as a H2 molecule. Even though this requires the
hydrogen to diffuse up a concentration gradient, it is diffusing down a chemical
potential gradient. The physical explanation of this is that the hydrogen release
involves the breaking of Si–H bonds, and it is argued that Si–Si plus H–H bonds
are more stable than two Si–H bonds [11].
The plasma component of the hydrogen release process invokes the role of H
atoms and ions, particularly in low pressure or hydrogen diluted plasmas. The
atoms can abstract hydrogen from the surface and the subsurface regions to form
H2 molecules which effuse from the surface. The H-ions can break sub-surface
Si–H bonds by ion impact displacement, and again H2 diffuses back up to the
surface [11]. Hence, ions have a beneficial effect on overall film quality, although
they can also contribute to undesirable compressive strain within the films [25], as
mentioned above.
5.5.2 Doped a-Si:H
In order to produce the source and drain contact regions, the deposition of doped
layers is an important stage in the TFT fabrication process. These layers are
produced by introducing phosphine (for n+ regions), or diborane (for p+ regions),
into the silane pre-cursor gas in the same type of PECVD reactor as used for the
undoped layers [33]. Whilst the deposition process itself has not been studied as
closely as the deposition process for the undoped layers, there is, nevertheless,
extensive information available on the effect of the plasma deposition conditions,
as well as the dopant flow rates, on the deposition rate [33]. However, in addition
to the characterisation of the deposition process, considerable attention has also
been focussed on the doping mechanism.
The early work [3] on doped layer deposition showed that substitutional doping
could be observed with both p- and n-type dopants, albeit at a rather low level. For
instance, with 1 % phosphine in the PECVD gas mixture, the resistivity of the
phosphorus doped region was *100 Xcm. Whilst this should not be directly
compared with c-Si with 1 % phosphorus doping, which has a resistivity of
*2 9 10-4 Xcm, the several orders of magnitude difference between the two
materials (which cannot be accounted for by carrier mobility differences alone)
suggests very different doping mechanisms.
However, in one sense, what was surprising about these results was that doping
was observed at all. The incorporation of impurities into amorphous materials was
thought to be governed by Mott’s ‘‘N, 8-N bonding rule’’ [4, 7, 34], which
describes the preferred bonding coordination leading to the optimum number of
covalent bonds. N is the number of valence electrons in the outer shell of the
impurity, and the rule states that for N \ 4, the coordination number will be N, and
for N [ 4, the coordination number will be 8-N. Hence, both group III boron and
group V phosphorus atoms would both be expected to have a coordination number
of 3 in the a-Si network, and these would be non-dopant sites. In contrast, to
achieve the substitutional doping observed in c-Si, they both directly replace
Si atoms, and have a coordination number of 4. In the case of the amorphous
network, the bonding of the impurity into the relatively flexible network is
determined by the valence requirements of the impurity itself, whereas, in the
more rigid crystalline lattice, the bonding arrangements are constrained by the
lattice [4, 7].
However, given that doping was observed in a-Si:H, it was clear that the ‘8-N’
rule needed modification, and a key proposal was that it should be applied to
configurations of both the dopant and host atoms [4]. In particular, the positively
charged P4+ atom in a dopant site would have the same configuration as neutral
four-fold coordinated Si04, and the Si dangling bond state Si3- would have the same
three-fold coordination as the non-bonding P03 state. Hence, the modified ‘8-N’ rule
was applied to the phosphorus/dangling bond defect configuration [4]:
Si04 þ P03 ¼) Si þ

3 þ P4 ð5:5Þ
In which the interaction of a three-fold coordinated P-atom with a weak Si–Si
bond led to the formation of an ionised P-dopant atom, and a compensating
negatively charged dangling bond defect. With increasing P-concentration, the
increased occupancy of the negatively charged dangling bond states shifted the
Fermi level towards the conduction band edge. This increased the electron
occupancy of the tail states, thereby increasing the conductivity of the material
(even though most of the P-donated electrons were trapped in the compensating
dangling bonds). The model was supported by the experimental observation that
the dangling bond density increased with the dopant concentration in the material
[4]. Although phosphorus has been used as an example to illustrate the process, the
same arguments can also be used for other dopant atoms.
The dopant/defect model successfully describes the doping of both a-Si:H and
a-Ge:H with the most commonly used substitutional dopants, as shown by
Fig. 5.15. This demonstrates a unique relationship between the total doping effi-
ciency of the materials and the dopant concentration in the plasma gas phase [4, 7].
The figure shows that high doping efficiency was only achieved at very low gas
dopant ratios. Thereafter, the efficiency decreased with the square root of the gas
dopant concentration, such that, with 1 % dopant in the gas phase, less than 1 % of
the dopant in the material was electrically active. This is quite different from the
doping of c-Si, in which the dopant displays *100 % activity. Moreover,
Fig. 5.15 Total doping 1

efficiency of group III and
group V impurities in a-Si:H
and a-Ge:H, as a function of Slope 1/2
10 –1
Total doping efficiency

the gas phase concentration
of dopants. (Reprinted with
permission from [4].
Copyright (1987) by the 10 –2
American Physical Society)
Host Dopant
Si P
Si B
10 –3 Si As
Ge P
Ge B
10 –4
10 –6 10 –5 10 –4 10 –3 10 –2 10 –1
Dopant gas concentration
the common curve, embracing all data points, was only found when relating the
doping efficiency to the gas phase dopant concentration, rather than to the more
intuitively obvious solid phase dopant concentration [4, 7], as discussed below.
The total doping efficiency, gT, is defined as the product of the solid phase
doping efficiency, gsol, and the incorporation efficiency, ginc, of the dopant from the
gas phase into the material. Where these two terms are given by:
N4
gsol ¼ ð5:6Þ
N3 þ N4
N3 þ N4
ginc ¼ ð5:7Þ
NSi Cg
N3 and N4 are the concentrations of P03 and P+4 , respectively, NSi is the silicon
concentration within the film, and Cg is the ratio of the dopant gas concentration to
the silane concentration in the plasma chamber. Hence, the total doping efficiency,
gT, is given by:
N4
gT gsol ginc ¼ ð5:8Þ
NSi Cg
and N4 was experimentally determined from the number of band tail electrons plus
the number of dangling bond states resulting from the dopant incorporation [4, 7].
The use of the total, or gas phase, doping efficiency, rather than the solid phase
doping efficiency to characterise the doping process, was a reflection of the
importance of the plasma deposition process itself. This determined the film
growth conditions, the incorporation of the dopant, and the density of weak Si–Si
bonds, which participated in the formation of the electrically active doping centres.
5.5.3 a-SiNx:H Gate Insulator
The universally preferred gate dielectric for commercial, inverted staggered a-Si:H
TFTs is a form of silicon nitride represented by the expression a-SiNx:H. This
indicates that it is both H-rich (20–40 %, depending upon deposition conditions
[35, 40]), and does not have the same as stoichiometry as Si3N4, which is typically
deposited at higher temperatures in the semiconductor device industry. The a-
SiNx:H films are deposited by low temperature PECVD, in the same deposition
systems used for the a-Si:H active layers, with silane and ammonia as the reactant
gases (plus the possible addition of a dilutant gas such as hydrogen [35, 36] or
helium [24]). As with the a-Si:H deposition, there is a potentially complex opti-
misation process involving the gas mixture ratio of NH3:SiH4, the gas pressure and
flow rates, RF power, and deposition temperature etc. [35], with the aim of tai-
loring the nitride film’s properties to meet the needs of the TFT. These will include
minimising the trap state density within the nitride to maximise device stability,
controlling its charge density to meet the TFT’s threshold voltage requirements
[43], as well as controlling the intrinsic stress within the film, all of which need to
be achieved in a manner consistent with a commercially acceptable deposition
rate. Indeed, as with the a-Si:H deposition, the commercial deposition of 300 nm
of gate dielectric a-SiNx:H usually consists of fast growth for the first 250 nm at
200 nm/min, followed by slower growth of the final 50 nm (which is adjacent to
the TFT channel) at 100 nm/min [6].
The value of the N:Si ratio, x, in a-SiNx:H films can be established by Auger
electron spectroscopy [36] or by Rutherford back-scattering [37, 38], and some
early results illustrated the variation of electron mobility and threshold voltage
stability with x. The range of x varied from 0.85–1.25, and was achieved by
varying the NH3:SiH4 gas flow ratio from 1:1 to 4:1 [36]. This work reported better
TFT characteristics at x *1.0–1.1 [36], which, when compared with a stoichi-
ometric value of 1.33, would make these films Si-rich. However, it is now more
generally accepted that the preferred a-SiNx:H films are N-rich, with x up
to *1.6–1.7. This is particularly so for films deposited in the aminosilane regime,
using large NH3:SiH4 gas flow ratios of [12:1 [35, 37–41]. The aminosilane
deposition regime is discussed further below.
In much reported work, the N:Si ratio, x, is not quoted, instead, the deposition
conditions and device properties are correlated with alternative material parame-
ters, such as the wet-etch rate [38], the refractive index [38], and the optical band-
gap [35, 36] of the film, and cross-correlated with the internal bonding structure of
the film, as revealed by FTIR absorption peaks due to Si–N, N–H, H–N–H, and Si–
H bonds [24, 35, 38, 39]. From these studies, it is recognised that the best a-SiNx:H
gate dielectric films are not just slightly N-rich, but also dense, due to a low H-
content, and have a low wet-etching rate, a refractive index of *1.9, and a low Si–
H bond density compared with the N–H density.
Whilst a lot of the optimisation of the deposition process was empirical, the
examination of the SiH4/NH3 plasma chemistry [40, 41] has contributed to the
overall understanding of the deposition process, and, by identifying its critical

features, eased the task of transferring a process from one reactor to another [39].
The work identified the tri-aminosilane radical, Si(NH2)3, as the key plasma
species for the deposition of high quality a-SiNx:H films, and, for a given
NH3:SiH4 ratio, empirically specified the ratio of RF power to SiH4 flow rate,
which ensured that it was the dominant radical species [39–41]. Amongst the most
important deposition conditions was an excess of NH3 relative to SiH4 (typically
between 12 and 25 times greater [35, 39–41]), together with an adequate gas
residence time [39]. This ensured that the NHn radical density exceeded the SiHn
radical density, and was able to fully aminate the silane, yielding tetra-aminosi-
lane, Si(NH2)4, and the tri-aminosilane radical, Si(NH2)3, as the dominant pre-
cursor plasma species [40, 41]. These conditions also minimised both the SiH
radical density in the plasma, and the Si–H bond density in the film, where a low
Si–H bond density has been correlated with a low trap density film [42]. Thus, the
well controlled aminosilane plasma was demonstrated to remove Si–H bonds from
the a-SiNx:H films, leaving just Si–N, N–H and H–N–H bonds [39–41], whilst
further work, using hydrogen-dilution, demonstrated that the hydrogen incorpo-
ration in the film could be further controlled with the removal of H–N–H bonds,
resulting in an increase in the film’s density and its optical band gap [35]—all of
which are indicators of improved film quality.
The preference for a-SiNx:H, as the gate dielectric in a-Si:H TFTs, is in contrast
to the use of SiO2 as the preferred dielectric in most other Si-based devices,
especially c-Si MOSFETs and poly-Si TFTs. The issue with the use of SiO2 in a-
Si:H TFTs is discussed further in Chap. 6, where it is shown that the choice of
dielectric is determined not just by the intrinsic properties of the dielectric itself,
but by the impact of those properties on the meta-stable defect density in the a-
Si:H film [43, 44].
5.6 Novel a-Si:H TFTs
5.6.1 Self-Aligned TFTs
In spite of the commercial dominance of the conventional, inverted staggered BCE

structure, a number of investigations have been reported into the fabrication of
self-aligned structures, in which the edges of the source and drain regions were
automatically aligned to the edges of the gate. The principal reasons for interest in
this structure are to remove a critical alignment stage, to reduce the parasitic
overlap capacitance between the gate and the source/drain terminals, and as a route
to shorter channel length TFTs [45–51]. The majority of these structures were
inverted staggered devices [45–49], in which a back-face photolithography stage
used the bottom gate to define a pattern in positive photo-resist, on top of the
device stack, for subsequent alignment of the source and drain regions. Within this
approach, there were significant differences in the way in which the source and
drain regions were actually defined. The other technique used a top-gated structure
[50, 51], and the back-face exposure process was avoided. However, this technique
used ion-doping to form the n+ doped regions, either side of the previously defined
top-gate (similarly to self-aligned poly-Si TFT fabrication—see Chap. 7), and ion-
doping is not a standard stage in a-Si:H TFT fabrication.
One example of the more conventional approach is shown in Fig. 5.16 [48], in
which the back exposure of the gate is used to form an etch-stop, ES, TFT, and this
architecture has been more widely investigated for self-aligned TFTs [46, 48, 49]
than has the self-aligned, back-channel-etched, BCE, architecture [45]. Part of the
preference for the ES architecture is that it can be implemented with a thin a-Si:H
layer, and this is advantageous for the transmission of the UV lithography radia-
tion through the absorbing a-Si film. The particular process shown in Fig. 5.16
consists of the conventional metal gate definition, followed by the deposition of
the a-SiNx:H/a-Si:H/a-SiNx:H triple layer stack. Positive photo-resist was used for
back illumination lithography, because the areas not exposed to the UV light were
retained after the exposure and development of the resist. Hence, the residual resist
pattern on the plate surface replicated the gate area, as shown in Fig. 5.16a, and
was used as an etch mask in defining the etch-stop a-SiNx:H layer. In this example,
a small amount of resist undercutting was implemented in order to ensure a degree
Fig. 5.16 Process steps in (a)

the fabrication of a self-
resist
aligned, etch-stop a-Si:H - Gate deposition
TFT, where ‘BE’ refers to a SiN
& pattern
back-face exposure stage (Mask 1) SiN
(Reprinted from [48] with - Tri-layer Mo a-Si:H
permission of IEEE) deposition Glass
- Define channel
(BE1)
resist
n+
(b) SiN
- n+ deposition SiN
- Data pattern Mo a-Si:H
(BE 2/ Mask 2) Glass
S/D Mask
(c) Mo/Al
SiN n+
- Data metal
deposition
SiN
& liftoff Mo a-Si:H
- n+ etch
Glass
5.6 Novel a-Si:H TFTs 133
of positive overlap of the contacts over the gate. After definition of the etch stop
layer, the photo-resist was removed, and the n+ layer deposited. A second back-
face lithography stage, shown in Fig. 5.16b as ‘BE 2’, used a mask, and defined a
new photo-resist pattern, consisting of a self-aligned resist pad over the gate, plus
pads each side of it. The whole structure was then coated with the final metalli-
sation material, and immersed in photo-resist solvent, which dissolved the resist,
and, in so-doing, removed the metal where it had been directly on top of the resist
pads. This is usually referred to as a ‘float-off’, or a ‘lift-off’ process. The lift-off
process defined the metallisation pattern, which was then used as an etch mask for
the subsequent removal of the n+ a-Si:H layer on the ES pad, and for the removal
of the undoped a-Si:H outside the TFT areas, leaving behind a self-aligned ES
TFT, as shown in Fig. 5.16c [48].
Other variants on this process have used a single back-face exposure to define
the etch-stop pad, and then used an alignment stage to define the edges of the
source and drain metal contacts on top of the ES pad [46, 49], with the metal acting
as the etch mask for the removal of the n+ a-Si:H layer, as in the conventional ES
process. Hence, this process resulted in a greater overlap of the metal contacts and
the ES pad than in the lift-off process, and would be less suitable for short channel
devices.
A lift-off process has also been used to form self-aligned BCE TFTs [45], in
which only one back-face lithography stage was used to define a photo-resist pad
on top of the a-SiNx:H/a-Si:H/n+ a-Si:H stack. The top metal was deposited on top
of this residual photo-resist pattern, and the metal was selectively removed in a
lift-off process. This left a space between the edges of the source and drain metal
pads, which was aligned to the gate. The normal back-channel etch was then used
to clear the n+ a-Si:H from this region.
In spite of the variety of processes demonstrated for self-aligned structures,
none has been adopted for a-Si:H TFT manufacturing. This is because the standard
BCE process is well enough controlled to give the required display performance,
and the self-aligned processes introduce greater processing complexity, and,
hence, higher manufacturing cost.
5.6.2 Short Channel TFTs
There has been relatively limited interest in short channel a-Si:H TFTs, although
device characteristics have been published for channel lengths down to *1 lm
and less [45, 52–56]. These devices have employed some novel architectures,
including the self-aligned BCE structure [45], described in Sect. 5.6.1, to achieve
0.4 lm channel lengths, as well as the vertical TFT structure shown in Fig. 5.17
[52]. With this device, the channel length was determined by the thickness of the
horizontal layer of a-SiNx:H, separating the two vertically stacked n+ source and
drain regions, rather than by a photolithographic process. A conventional BCE
process has been used for gate lengths down to *1 lm [54], and a high resolution
Fig. 5.17 Cross-sectional Gate

diagram of a short channel,
vertical a-Si:H TFT 0.2 μm
(Reprinted from [52] with 0.3 μm Drain
permission of IEEE) Mo
+
n a-Si 0.3 μm
Al SiN 1.0 μm
SiN n+a-Si 0.3 μm
n–a-Si Mo Source
Corning 7059
Fig. 5.18 Variation of 0.8

electron mobility with TFT
channel length (Reprinted
with permission from [54]) Field-effect mobility (cm 2/Vs)
0.6
0.4
0.2
S/D overlap = 1.5μm

S/D overlap = 3.0μm
0.0
0 1 2 3 4 5 6
Channel length (μm)
electron beam lithography process, together with metal lift-off, has been used to
reduce them to 0.2 lm [55, 56]. A common observation with short channel TFTs
was that the on-current did not scale with 1/L [45, 52–55], as shown by the
electron mobility results in Fig. 5.18 [54], and this was attributed to series resis-
tance effects in the doped contact regions. In one instance [55], the change of
contact metal from Cr to Al gave mobility values which were independent of
channel length, and this was attributed to reduced contact resistance due to the
diffusion of Al through the contact regions and into the underlying bulk a-Si:H.
The other feature noted in short channel a-Si:H TFTs was an increase in off-state
current with reducing channel length for a given source-drain bias [45, 52–56]. This
was attributed to space charge limited currents [45], or to field assisted carrier
emission processes [56]. However, when devices were measured at a constant
channel field (fixed Vd/L), the off-state current dependence on channel length
reduced [45, 53, 54], and, in some cases, its value increased with increasing channel
length [53, 54]. The latter effect was ascribed to the influence of the increasingly
large drain bias itself (rather than the channel field), particularly with the longer
channel length TFTs.
5.6.3 Hydrogen-Diluted TFT Depositions
The stimulus for this work [30, 57–59] was the interest in using a-Si:H TFTs for
active matrix OLED addressing. As discussed in Chap. 4, the duty cycle of the
drive TFTs in an AMOLED display is 100 %, whereas, in AMLCDs, the duty
cycle of the addressing TFT is *0.1 % or less. Given that a-Si:H TFTs are subject
to gate bias instabilities [10, 60] (to be discussed in more detail in Chap. 6), the
much greater duty cycle in AMOLED displays requires a more stable TFT than
those used in AMLCDs. Hence, although the standard a-Si:H TFTs fabricated for
AMLCD applications have sufficient stability for that application, an improvement
in stability of up to *1000 times is required for the AMOLED application. This is
to ensure that the overall AMOLED display lifetime is not limited by the TFT
stability, but that the TFT lifetime at least matches the green phosphorescent
OLED lifetime of *250,000 h (which is defined to be the time for a 50 % drop in
brightness) [57, 58].
As mentioned in Sect. 5.5.1, the presence of hydrogen in the a-Si:H deposition
plasma can be regarded as a surface etchant species [29], which selectively breaks
weak Si–Si bonds, and this effect can be enhanced by increasing the H2:SiH4
dilution ratio. Low-field, gate bias instability in a-Si:H TFTs is associated with an
adjustment in the number of weak bonds and dangling bonds in response to a
change in the Fermi level, or free carrier density, within the device [10]. Conse-
quently, it is argued that by reducing the weak bond density during a-Si:H growth,
device stability can be enhanced [30, 57–59]. In this work, the a-Si:H, for ES

normalised TFT channel
currents with stress time, for
‘standard’ and ‘improved’ a-
Si:H TFTs. (The open and
closed symbols are for the
linear and saturation stress
regimes, respectively).
(Reprinted with permission
from [59]. Copyright (2009)
American Institute of
Physics)
TFTs, was deposited at 300 °C with a H2:SiH4 flow rate of 10:1 [57, 58], and the
improvement in device stability is shown by the curve labelled ‘Process A’ in
Fig. 5.19 [59]. These results show the fractional change in channel current with
stress time, and were measured under low gate bias and drain saturation conditions
(VG = 7.5 V, and VD = 15 V) in order to reflect the typical TFT operating
conditions in an AMOLED display [58]. However, TFT instability, due to gate
bias stress, is usually measured in terms of the change in threshold voltage, DVT(t),
with time. In order to normalise the standard TFT instability results to an equiv-
alent change in drain current at a fixed gate bias, the linear and saturation regime
MOSFET device Eqs. 3.11 and 3.17, respectively, were used, i.e.:

ln WCi VG VT0 DVTðlinÞ ðtÞ VD
IdðlinÞ ðtÞ ¼ ð5:9Þ
L
2
ln WCi VG VT0 DVTðsatÞ ðtÞ
IdðsatÞ ðtÞ ¼ ð5:10Þ
2L
where VT0 is the pre-stress value of the threshold voltage.
The extrapolations of the experimental Process A data in Fig. 5.19 were based
upon the measured time dependence of the DVT values, which, in the low gate
field regime, have a power law dependence given by [59]:
DVTðlinÞ ðVG VT0 Þðt=s0 Þb ð5:11Þ
The values of s0 and b were determined by fits to the experimental time

dependence of the DVT(lin) (t) data, and, in saturation, the following relationship
was used:
DVTðsatÞ 2=3DVTðlinÞ ð5:12Þ
where the 2/3 multiplicative factor was introduced to account for the smaller
average free carrier density in the channel of a device operating in saturation [61].
The TFTs labelled ‘Process A’ in Fig. 5.19, had a standard silane/ammonia
a-SiNx:H layer deposited at 300 °C. A further improvement in device stability, to
the curve labelled ‘Process B’, was obtained with H2 dilution of the a-SiNx:H
reactant gases, and deposition at 350 °C [59]. It is generally accepted that nitride-
limited instability occurs at high gate fields [60], and is caused by carrier tun-
nelling into nitride traps. However, it was argued here that the change in a-SiNx:H
properties had affected the low-field stability of the subsequently deposited a-Si:H,
and this was ascribed to an improvement in its quality close to its interface with the
a-SiNx:H layer [30, 59]. The predicted ‘half life’ of these TFTs [59] greatly
exceeded that of the green phosphorescent OLEDs, and a reduced deposition
temperature of 300 °C, for the improved a-SiNx:H layer, gave a predicted TFT
‘half-life’ of 8 9 105 h, which was still in excess of the OLED’s [30].
Hence, the use of enhanced H2-dilution for the deposition of both the a-Si:H
and the a-SiNx:H films produced up to five orders of magnitude improvement in
the time to 50 % current degradation in these TFTs, compared with the ‘standard’
a-Si:H TFTs.
5.7 Summary
The current active matrix flat panel display, AMFPD, industry is built around the
ubiquitous hydrogenated amorphous silicon thin film transistor, and this chapter
contains an overview of the structure and fabrication processes for these TFTs.
Hydrogenated amorphous silicon is an alloy of silicon with *10 % hydrogen,
in which the hydrogen plays a key role both in passivating dangling bond defects,
and in the a-Si:H deposition process itself. The emphasis in this chapter is on the
commonly used structures and processes employed in the AMFPD industry, and
discussion of the device architecture is largely limited to the inverted staggered
TFT. The two variants of this are the back-channel-etched, BCE, TFT, which is
widely used in manufacturing, and the etch-stop, ES, TFT, which is often used in
research. Both structures consist of an a-SiNx:H gate dielectric layer, and n+-doped
a-Si:H contacts to the intrinsic a-Si:H transistor body. The BCE configuration is
preferred for manufacturing because, in spite of it requiring a well controlled
etching procedure to remove the n+ a-Si:H from the channel area, it has a simpler
processing schedule, with fewer deposition and lithography stages. In addition to
the details on device architecture, there is also a brief discussion of the issues
underlying the design and layout of TFTs.
The formation of the principal device layers, namely the intrinsic and doped
a-Si:H, as well as the a-SiNx:H gate dielectric, is by plasma enhanced chemical
vapour deposition, PECVD, in a parallel plate reaction chamber, which is normally
driven by a 13.56 MHz RF generator. Silane gas is used for the deposition of all
layers, with the addition of phosphine and ammonia for the deposition of the n+
a-Si:H and a-SiNx:H layers, respectively. In addition to these primary reactant
gases, hydrogen dilution is also used both to control the deposition process, and to
influence the bonding structure and electrical properties of the layers themselves.
Whilst the focus has been mainly on the inverted staggered TFT, some novel
device structures are briefly reviewed. These include self-aligned and short
channel TFTs, as well as high stability TFTs fabricated using enhanced hydrogen-
dilution of the PECVD reactant gases.
References
1. Flewitt AJ, Milne WI (2004) a-Si:H TFT thin film and substrate materials. In: Kuo Y (ed)
Thin film transistors, materials and processes, Vol. 1. Amorphous silicon thin film transistors.
Kluwer Academic Publishers, Norwell, Massachusetts
2. Spear WE, Le Comber PG (1972) Investigation of the localised state distribution in

amorphous Si films. J Non-Crystalline Solids 8–10:727–738
3. Spear WE, Le Comber PG (1975) Substitutional doping of amorphous silicon. Solid State
Commun 17(9):1193–1196
4. Stutzmann M, Biegelsen DK, Street RA (1987) Detailed investigation of doping in
hydrogenated amorphous silicon and germanium. Phys Rev B 35(11):5666–5701
5. Snell AJ, Mackenzie KD, Spear WE, Le Comber PG, Hughes AJ (1981) Application of
amorphous silicon field effect transistors in addressable liquid crystal displays. Appl Phys
24:357–362
6. Yang Y-T, Won TK, Choi SY, Takehara T, Nishimura Y, White JM (2007) The latest
plasma-enhanced chemical-vapor deposition technology for large-size processing. IEEE J
Disp Tech 3(4):386–391
7. Street RA (1991) Hydrogenated amorphous silicon. Cambridge University Press, Cambridge
8. Morel G, Katiyar RS, Weisz SZ, Jia H, Shinar J, Balberg I (1995) Raman study of the
network disorder in sputtered and glow discharge a-Si:H films. J Appl Phys 78(8):5120–5125
9. Hack M, Shur MS, Shaw JG (1989) Physical models for amorphous-silicon thin-film
transistors and their implementation in a circuit simulation program. Trans IEEE ED-36(12),
2764–2769
10. Powell MJ, Deane SC (1996) Defect-pool model and the hydrogen density of states in
hydrogenated amorphous silicon. Phys Rev B 53(15):10121–10132
11. Robertson J (2000) Growth processes of hydrogenated amorphous silicon. Mat Res Soc Symp
Proc 609:A1.4.1–A1.4.12
12. Street RA (1989) Thermal equilibrium electronic properties of a-Si:H. IEEE Trans ED-
36(12), 2770–2774
13. Staebler DL, Wronski CR (1980) Optically induced conductivity changes in discharge-
produced hydrogenated amorphous silicon. J Appl Phys 51(6):3262–3268
14. Stutzmann M (1997) Microscopic aspects of the Staebler-Wronski effect. Mat Res Soc Symp
Proc 467:37–48
15. Fritzsche H (2001) Development in understanding and controlling the Staebler-Wronski
effect in a-Si:H. Annu Rev Mater Res 31:47–79
16. Flewitt AJ (2012) Hydrogenated Amorphous Silicon Thin Film Transistors (a–Si:H TFTs).
In: Chen J, Cranton W, Fihn M (eds) Handbook of visual display technology. Springer, Berlin
17. Lin W-Y, Wu WB, Cheng KC, Li HH (2012) Photolithography for thin-film-transistor liquid
crystal displays. In: Chen J, Cranton W, Fihn M (eds) Handbook of visual display technology.
Springer, Berlin
18. Cheng H-C (2012) Wet etching. In: Chen J, Cranton W, Fihn M (eds) Handbook of visual
display technology. Springer, Berlin
19. Kuo Y (2004) Plasma etching in a-Si:H TFT array fabrication. In: Kuo Y (ed) Thin film
transistors, materials and processes, Vol. 1. Amorphous silicon thin film transistors. Kluwer
Academic Publishers, Norwell
20. Stamate E, Yeom GY (2012) Dry etching. In: Chen J, Cranton W, Fihn M (eds) Handbook of
visual display technology. Springer, Berlin
36(12), 2753–2763
22. http://www.nikon.com/products/precision/lineup/fx/pdf/FX-Series-GeneralCatalog.pdf
(Accessed Feb, 2011)
23. French ID, Deane SC, Murley DT, Hewett J, Gale IG, Powell MJ (1997) The effect of the
amorphous silicon alpha-gamma transition on thin film transistor performance. Mat Res Soc
Symp Proc 467:875–880
24. He S, Nishiki H, Hartzell J, Nakata Y (2000) Low Temperature PECVD a-Si:H TFT for
plastic substrates. SID’00 Digest, 278–281
25. Wehrspohn RB, Deane SC, French ID, Gale I, Hewett J, Powell MJ, Robertson J (2000)
Relative importance of the Si–Si bond and Si–H bond for the stability of amorphous silicon
thin film transistors. J Appl Phys 87(1):144–154
References 139
26. Sazonov A, Nathan A, Murthy RVR, Chamberlain SG (2000) Fabrication of a-Si:H TFTs at
120 °C on flexible polyimide substrates. Mat Res Soc Symp Proc 558:375–380
27. Robertson J (2000) Deposition mechanism of hydrogenated amorphous silicon. J Appl Phys
87(5):2608–2617
28. Cabarrocas PR (1998) Growth of undoped a-Si:H by PECVD. In: Searle T (ed) Properties of
amorphous silicon and its alloys. Institution of Engineering and Technology
29. Tsai CC, Anderson GB, Thompson R (1990) Growth of amorphous, microcrystalline and
epitaxial silicon in low temperature plasma deposition. Mat Res Soc Symp Proc 192:475–480
30. Sturm JC, Hekmatshoar B, Cherenack K, Wagner S (2009) Enabling mechanisms for a-Si
TFT’s with 100-year lifetimes compatible with clear plastic substrates. Proc 5th Internal TFT
Conf, ITC’09, 9.1
31. Lucovsky G, Nemanich RJ, Knights JC (1979) Structural interpretation of the vibrational
spectra of a-Si:H alloys. Phys Rev B 19(4):2064–2073
32. Kessels WMM, Smets AHM, Marra DC, Aydil ES, Schram DC, van de Sanden MCM (2001)
On the growth mechanism of a-Si:H. Thin Solid Films 383:154–160
33. Cabarrocas PR (1998) Growth of doped a-Si:H by PECVD. In: Searle T (ed) Properties of
amorphous silicon and its alloys. Institution of Engineering and Technology
34. Street RA (1982) Doping and the Fermi energy in amorphous silicon. Phys Rev Lett
49(16):1187–1190
35. Murley D, French I, Deane S, Gibson R (1996) The effect of hydrogen dilution on the
aminosilane plasma regime used to deposit nitrogen-rich amorphous silicon nitride. J Non-
Cryst Solids 198–200:1058–1062
36. Hiranaka K, Yosimura T, Yamaguchi T (1987) Influence of an a-SiNx:H gate insulator on an
amorphous silicon thin film transistor. J Appl Phys 62(5):2129–2135
37. Lustig N, Kanicki J (1989) Gate dielectric and contact effects in hydrogenated amorphous
silicon-silicon nitride thin-film transistors. J Appl Phys 65(10):3951–3957
38. Gleskova H, Wagner S, Gasparik V, Kovac P (2001) Low temperature silicon nitride for thin-
film electronics on polyimide substrates. Appl Surf Sci 175–176:12–16
39. Murley DT, Gibson RAG, Dunnett B, Goodyear A, French ID (1995) Influence of gas
residence time on the deposition of nitrogen-rich amorphous silicon nitride. J Non-Cryst
Solids 187:324–328
40. Smith DL, Alimonda AS, Chen C-C, Ready SE, Wacker B (1990) Mechanism of SiNxHy
deposition from NH3-SiH4 plasma. J Electrochem Soc 137(2):614–623
41. Smith DL (1993) Controlling the plasma chemistry of silicon nitride and oxide deposition
from silane. J Vac Sci Technol A 11(4):1843–1850
42. Smith DL, Alimonda AS, Chen C-C, Tuan HC (1990) Reduction of charge injection into
PECVD SiNxHy by control of deposition chemistry. J Electron Mater 19(1):19–27
43. Deane SC, Clough FJ, Milne WI, Powell MJ (1993) The role of the gate insulator in the
defect pool model for hydrogenated amorphous silicon thin film transistor characteristics.
J Appl Phys 73(6):2895–2901
44. Deane SC, Powell MJ, Hughes JR, French ID (1990) Thermal bias annealing evidence for the
defect pool in amorphous silicon thin-film transistors. Appl Phys Lett 57(14):1416–1418
45. Busta HH, Pogemiller JE, Standley RW, Mackenzie KD (1989) Self-aligned bottom-gate
submicrometer-channel-length a-Si:H thin-film transistors. IEEE Trans ED-36(12),
2883–2888
46. Kuo Y (1992) A self-aligned, tri-layer, a-Si:H thin film transistor prepared from two photo
masks. J Electrochem Soc 139:1199–1204
47. Hirano N, Ikeda N, Hishida S, Kaneko S (1996) A 33 cm-diagonal high-resolution TFT-LCD
with fully self-aligned a-Si TFTs. IEICE Trans on Electron, E79-C(8), 1103–1108
48. Thomasson DB, Jackson TN (1998) Fully self-aligned tri-layer a-Si:H thin-film transistors
with deposited doped contact layer. IEEE Elec Dev Lett 19(4):124–126
49. Cheng I-C, Kattamis AZ, Long K, Sturm JC, Wagner S (2006) Self-aligned amorphous-
silicon TFTs on clear plastic substrates. IEEE Elec Dev Lett 27(3):166–168
50. Kim SK, Choi YJ, Cho SI, Cho kS, Jang J (1998) A novel self-aligned coplanar amorphous
silicon thin film transistor. SID Symp Digest 29:379–382
51. Powell MJ, Glasse C, Green PW, French ID, Stemp IJ (2000) An amorphous silicon thin-film
transistor with fully self-aligned top gate structure. IEEE Electron Dev Lett 21(3):104–106
52. Uchida Y, Nara Y, Matsumura M (1984) Proposed vertical-type amorphous-silicon field-
effect transistors. IEEE Electron Device Lett 5(4):105–107
53. Kim C-D, Matsumura M (1996) Short-channel amorphous-silicon thin-film transistors. IEEE
Trans ED-43(12), 2172–2176
54. Kim SH, Hur JH, Kim KM, Koo JH, Jang J (2006) Short-channel amorphous-silicon TFT for
AMOLED. J Korean Phys Soc 48:S80–S84
55. Fortunato G, Gentili M, Luciani L, Mariucci L, Pecora A, Petrocco G (1990) Source-drain
contact effects in short-channel a-Si:H thin-film transistors. Jpn J Appl Phys 29(12):
L2353–L2356
56. Fortunato G, Gentili M, Luciani L, Mariucci L, Mattacchini A, Pecora A (1991) Short-
channel effects in 0.2 micron channel length a-Si:H thin-film transistors fabricated by
electron beam lithography. J Non-Cryst Solids, 137 and 138, 1225–1228
57. Hekmatshoar B, Cherenack KH, Kattamis AZ, Long K, Wagner S, Sturm JC (2008) Highly
stable amorphous-silicon thin-film transistors on clear plastic. Appl Phys Lett 93(3):
032103-1–032103-3
58. Hekmatshoar B, Cherenack KH, Wagner S, Sturm JC (2008) Amorphous silicon thin-film
transistors with DC saturation current half-life of more than 100 years. IEEE IEDM’08,
89–92
59. Hekmatshoar B, Wagner S, Sturm JC (2009) Trade off regimes of lifetime in amorphous
silicon thin-film transistors and a universal lifetime comparison framework. Appl Phys Lett
95(14):143504-1–143504-3
60. Powell MJ, van Berkel C, French ID, Nicholls DH (1987) Bias dependence of instability
mechanisms in amorphous silicon thin-film transistors. Appl Phys Lett 51(16):1242–1244
61. Karim KS, Nathan A, Hack M, Milne WI (2004) Drain-bias dependence of threshold voltage
stability of amorphous silicon TFTs. IEEE Electron Dev Lett 25(4):188–190
Chapter 6
Hydrogenated Amorphous Silicon TFT
Performance
Abstract As the dominant device for active matrix flat panel displays, a-Si:H
TFTs have been extensively studied, and many of their properties linked to the
meta-stability of the defect structure in the material. The topics covered in this
chapter include a review of the electronic structure and the material properties of
a-Si:H, which are largely determined by the density of states, DOS. The DOS
consists of band tail states, due to weak Si–Si bonds, and deeper dangling bond
states. Their influence on the threshold voltage and field effect mobility of the TFT
is discussed in a semi-analytical fashion, by using an exponential approximation to
the DOS. The factors controlling the on-state and off-state characteristics are
reviewed, which, apart from the field effect mobility and threshold voltage, include
series resistance effects and front channel hole conduction, respectively. Gate bias
stress induces threshold voltage instability, due to both defect formation in the
a-Si:H layer and carrier trapping in the a-SiNx:H gate dielectric. The modification
of these effects by pulsed bias stressing, and combined gate and drain bias stressing
are also reviewed. Finally, a brief overview of defect instability under optical
illumination, known as the Staebler–Wronski effect, is presented.
6.1 Introduction
Chapter 5 focussed on the technology of a-Si:H TFTs, and, in this chapter, their
performance and device physics will be described. Much of their performance
is determined by the material properties and defect structure of a-Si:H, which is
discussed in Sect. 6.2, starting with a review of the basic material properties. This
introduces the dangling bond and band tail state distributions, and focuses on the
meta-stability of the material, which is driven by the hydrogen-mediated equili-
bration processes between these two types of defects. For practical device model-
ling, the defect distributions are frequently approximated to by exponential density
of states, DOS, distributions across the band-gap. These approximations are used, in
a semi-analytical fashion, to examine the impact of the DOS on the relationship
between surface band-bending and the equilibrium surface charge density. This

142 6 Hydrogenated Amorphous Silicon TFT Performance
facilitates a direct comparison with the band bending in c-Si (presented in Chap. 2),
and highlights the role of the a-Si:H DOS in controlling threshold voltage and the
carrier’s field effect mobility. Finally, in Sect. 6.2.5, the role of band bending in
defect equilibration in thin films is discussed, and the resulting DOS explains the
choice of a-SiNx:H, rather than SiO2, as the preferred gate dielectric. It also con-
tributes, together with the wider valence band tail state distribution, to the poorer
performance of p-channel TFTs compared with n-channel TFTs.
Section 6.3 deals with more practical, architectural issues of device performance,
in particular, the sources of series resistance in the device, and their effect upon the
on-state current, and, secondly, the role of the front and back channels on the off-
state leakage currents. Section 6.4 reviews the gate bias instability phenomenon in
a-Si:H TFTs, which is a major performance limitation in the application of these
devices. The intensive study of this effect has led to an improved understanding of
the hydrogen-mediated defect formation processes in the material. Finally, Sect. 6.5
contains a brief overview of another meta-stability effect, known as the Staebler–
Wronski effect, which results from intense, long duration illumination of the
material. This is a more important issue for the long-term performance of a-Si:H
solar cells, rather than for TFTs, but is included in the chapter for completeness.
6.2 Defect Structure and a-Si:H Density of States
6.2.1 Basic Material Properties
As briefly discussed in Chap. 5, a-Si:H can be represented by a band diagram,

which, in spite of the absence of long range order, is similar to that used for
crystalline silicon. This is because with the bond length and bond angle deviations
from crystallinity of *2 % and *10 %, respectively, these distortions in the
network only correspond to an energy disorder of *0.1 eV relative to the Si–Si
bond strength of 2.5 eV [p 62 in 1]. This means that the same molecular orbital
considerations can be used to calculate the energy bands (as for the crystalline Si
phase), and these are shown schematically in Fig. 6.1. The isolated Si atom has
four valence electrons, which are equally divided between the 3s and 3p states,
and, to bond tetrahedrally to four other Si atoms, four hybrid sp3 orbitals are
formed containing these four electrons. Each of these can accept another electron
from the four surrounding atoms, and the bonding interaction between these lowers
the energy of the state, and leads to the splitting of the sp3 level into the bonding
and anti-bonding states, which correspond to the valence and conduction bands,
respectively [1, 2]. As shown in the figure, there are s-like states at the bottom of
the conduction band and p-like states at the top of the valence band. The p-like
states are more affected by bond angle disorder, leading to greater disorder energy,
and to the broader tail state distribution adjacent to the valence band mobility edge
[p 83 in 1].
6.2 Defect Structure and a-Si:H Density of States 143
Fig. 6.1 Schematic molecular orbital model of a-Si:H, and its density of states distribution.
(From [1], Copyright Cambridge University Press 1991, reproduced with permission)
A broader valence band tail is characteristically found in a-Si:H, and its width is
an indicator of the disorder in the material. The band tail width can be extracted
from TFT measurements, but it is also common to measure it as a basic material
parameter using optical absorption [2, 3]. This assessment of the valence band tail
width measures the exponential dependence of the optical absorption coefficient, a,
on the photon energy just below the mobility edge. The data was modelled by the
following expression:
aðEÞ ¼ a0 expðE E1 Þ=E0 ð6:1Þ
which was valid over the absorption coefficient range 2 9 102 \ a \ 5
9 103 cm-1, and where the constants a0 = 1.5 9 106 cm-1, E1 = 2.2 eV were
obtained from data fitting [3]. The characteristic width of the distribution, E0, is the
Urbach energy [3]. At these near band edge photon energies, the electron transi-
tions are from the valence band tail states to the conduction band. Over a broad set
of samples, some of which had been deliberately dehydrogenated to increase the
defect density, the Urbach energy was found to lie in the range 50–100 meV [3].
The Urbach energy is a measure of the disorder in the material [2, 3], as dem-
onstrated in Fig. 6.2 by the correlation between the Urbach energy and the dan-
gling bond density in a-Si:H [4]. This data set spanned a wide range of defect
densities, and good quality device material sits at the lower end of this range with
an Urbach energy of *50 meV [5]. From a device point of view, the TFTs of most
interest are n-channel, whilst the valence band tail states will only affect p-channel
TFTs. However, it has been demonstrated that the conduction band tail width
scales with the Urbach energy, with it increasing from 19 to 34 meV, as the
Urbach energy increased from 45 to 57 meV [6]. Given this correlation, the Ur-
bach energy is widely used as a figure of merit for the material.
Fig. 6.2 Correlation between

the Urbach energy, E0, and
the dangling bond density
measured in various types of
a-Si:H. (Reprinted from [4]
with permission of Taylor and
Francis Ltd (http://
www.tandfonline.com))
Where there are no bonding interactions, such as at a dangling bond, the sp3
level remains un-split, and gives the dangling bond, DB, defect energy level near
the middle of the band gap.
As discussed in Sect. 5.2, the absence of long range periodic order means that
electrons occupying the tail states are localised, and give rise to the mobility gap. The
size of this gap is defined by the mobility edges, where the extended states begin.
However, even in the extended states, the carrier scattering length is of the order of
the inter-atomic spacing, giving band mobilities, l0, for free electrons and holes of
*10–20 cm2/Vs, and 1–10 cm2/Vs, respectively [pp 238–240 in 1, 7]. For carriers
confined predominantly in the localised states, the drift mobility, ld, recorded in
photo-conductivity time-of-flight measurements is defined by [p 73 in 1]:
ld ¼ l0 sfree =ðsfree þ strap Þ ð6:2Þ
Hence, the drift mobility is reduced from the band mobility by the ratio of the
time, which a thermally excited carrier spends in an extended state, sfree, to its time
in a localised state, strap. The time-of-flight measurements have shown that, even for
high quality a-Si:H, this ratio is\0.1 for electrons, and it is even smaller for holes,
due to the broader tail state distribution near the valence band edge. Hence,
sfree\\strap. A full analytical treatment of this effect for the band tail states [pp
238–240 in 1] is beyond the scope of this discussion, but an appreciation of the
physical arguments can be obtained from the carrier capture and thermal emission
rate expressions in Sect. 2.3.1. For carriers in deep traps of density NT, and at an
energy ET below the conduction band edge, the carrier emission time constant (to
release the carrier to the band) is (cnNcexp-ET/kT)-1, and the capture time constant
of the emitted free carrier is (cnN0T)-1. (N0T is the density of empty traps, Nc is the
effective density of states at the conduction band edge, and cn is the product of the
trap capture cross section and the carrier’s thermal velocity). These time constants
can be regarded as indicative of the intervals spent by the carrier in the localised and
extended states, respectively. Hence, the deeper the trap, the greater the emission
time constant, with the residence time of the carrier in the trap increasing expo-
nentially with ET. For a trap density, which is large compared with the free carrier
density, N0T * NT, and the capture time constant is independent of ET. Hence, the
ratio of sfree to strap will decrease exponentially with increasing ET. Similar con-
siderations of the fractional division of carriers between the localised and extended
states also applies to the field effect mobility, and this is discussed in Sect. 6.2.4.
With the 10 % hydrogen concentration in device grade a-Si:H, the majority of it
is contained in Si–H bonds, with two hydrogen atoms being accommodated into the
lattice per broken Si–Si bond [8], giving a structure variously designated by SiHHSi
or SiHSiH or the H*2 complex [9]. The latter designations reflect the location of the
two passivating hydrogen atoms on the bond-centred (BC) and adjacent tetrahedral
(Td) sites [8]. The hydrogen incorporation relieves the strain in the lattice, but the
incorporation itself leaves a residual strain, with the result that, once a minimum
strain has been achieved, there remains a steady state density of weak bonds
of *1019 cm-3 [8, 10]. The residual unpassivated dangling bond density of
*1016 cm-3 is formed by the breaking of these weak bonds in the reaction:
Weak bonds $ Dangling bonds
The dangling bonds affect the electrical performance of TFTs, as discussed
below, and their density can be inferred from measurements of TFT characteris-
tics. In addition, the neutral dangling bond is a paramagnetic centre, with a g-value
of 2.0055, which can be detected by electron spin resonance, ESR, measurements
[11]. These measurements enable the direct monitoring of the DB density during
modification of the material, but are essentially bulk measurements, requiring the
use of a sufficient volume of material to yield measurable ESR signals above the
detection limit of *6 9 1011 DBs at room temperature. This usually requires
stacked films, or films of sufficient thickness, and, in the following work, films in
the range 20–90 lm were used, not merely to yield a large enough ESR signal, but
also to ensure that the bulk signal dominated any surface signal [11]. a-Si:H is
described as meta-stable, as its defect characteristics are frozen-in at
*450–500 K, during cooling from the growth temperature, but, if annealed long
enough at lower temperatures, the material will tend to reach a new equilibrium.
This equilibration is illustrated in Fig. 6.3 [11], in which the normalised concen-
tration of dangling bonds is shown as a function of annealing time and tempera-
ture, and the equilibrium densities are shown as a function of temperature in
Fig. 6.3a, b, respectively. Prior to these anneals, the samples were held at 290 °C
for 10 min, which was long enough to set the defect density to its equilibrium
value at that temperature. These equilibration effects can be represented on the
configuration coordinate diagram in Fig. 6.4a, in which the defect formation
Fig. 6.3 Equilibration of the

dangling bond density in a-
Si:H after quenching at
290 °C: a time dependence at
different anneal temperatures
(40 lm film deposited at
40 W), and b equilibrium
density at different
temperatures (for different
deposition conditions).
by the American Physical
Society)
energy, Ud, is the energy difference between the ground state and the defect state,
and this determines the equilibrium defect density, Nde. In its most generalised
form, the relationship between the equilibrium defect density, Nde, and the equi-
librium weak bond density, NWBe, is given by the law of mass action, where:
Nde ¼ NWBe exp Ud =kT ð6:3Þ
(the sum of these two concentrations equals the total weak bond density, NWBT)
and, from Fig. 6.3b, Ud * 0.18 eV [11]. In this simplified discussion, the neutral
defect formation energy, Ud, is given by the difference in electron energies of the
singly occupied dangling bond state, E0d, at 0.6 eV above the valence band
Fig. 6.4 a Configuration coordinate diagram showing the barrier between two defect equilibrium
states (From [1], Copyright Cambridge University Press 1991, reproduced with permission),
b schematic density of states diagram illustrating the energy required to create a dangling bond, at
E0D, from the distribution of weak bond energies (Reprinted with permission from [10]. Copyright
(1990) by the American Physical Society)
mobility edge, and the energy of the weak bond, EWB, in the valence band tail
state, from which it was formed [11]:
Ud ¼ E0d EWB ð6:4Þ

However, for realistic modelling of the process, the number of defects is cal-
culated by integrating the law of mass action over the range of weak bond energies
in the valence band tail [10, 11], as illustrated in Fig. 6.4b, where the formation
energy at E is:
Ud ðEÞ ¼ DE ¼ E0d E ð6:5Þ
and the weak bond density at E is given by:

NWB ðEÞ ¼ Nv0 exp E=Ev0 ð6:6Þ
where Ev0 is the width of the valence band tail state distribution. Hence, the
equilibrium dangling bond density, Nde, is then given by integrating the law of
mass action over all band tail states [10], and:
Z1
Nv0 expðE=Ev0 ÞdE
Nde ¼ 0:5 ð6:7Þ
1 þ expfðE Ed0 Þ=kTg
0
Referring back to Fig. 6.4a, the reaction rate for the defect creation process is
expected to be thermally activated, with an activation energy given by the defect
formation barrier, EB. The experimental relaxation rate data in Fig. 6.3a showed a
temperature dependent equilibration regime, which had a stretched exponential
time dependence given by:
h i
DN ¼ DN0 exp ðt=sÞb ð6:8Þ
and where the time constant, s, was thermally activated:

s ¼ s0 expEs =kT ð6:9Þ
The activation energy, Es, corresponds to the defect formation barrier, EB,
shown in Fig. 6.4a. In these samples, the value of Es was *1.5 eV, and b was
*0.6 at 200 °C [11]. The use of a stretched exponential, rather than a single
exponential, is indicative of a range of barrier heights, as can be expected from the
disordered material, and the value of 1.5 eV can be interpreted as the mean barrier
height. The equilibration processes are reversible in response to changes in the
annealing temperature. Hence, the formation and removal of the DBs is reversible,
as the defect density achieves a new equilibrium value with respect to the weak
bond density at different temperatures.
From a study of the reaction kinetics and defect densities, various bonding
models have been examined. The simplest is the breaking of a weak bond to yield
two dangling bonds; however, it is argued that these two nearby DBs on the same
weak bond site would not be paramagnetic, and would not give an ESR signal [11].
The alternative would be for the dangling bonds to separate, but such mobile
defects have not been identified [11]. These problems can be resolved by involving
mobile hydrogen in defect formation, and two possible models are shown in
Fig. 6.5a and b, in which the downward and upward arrows describe defect for-
mation and removal, respectively. The reaction equations are [11]:
SiH þ WB $ DH þ DW þ SiH ð6:10Þ
2SiH þ WB $ SiHHSi þ 2DH ð6:11Þ
Fig. 6.5 Illustration of two

hydrogen-mediated weak
bond $ dangling bond
creation and removal
processes. (Reprinted with
Copyright (1989) by the
Both models show the involvement of hydrogen in the dangling bond formation
process. From a thermodynamic point of view, a distinction is made between the
isolated dangling bonds, DH, and those broken bonds on the weak bond site, DW,
which remain adjacent to the nearby hydrogen passivated bond (although these
two types of DB may have the same electrical behaviour) [11]. From a comparison
of the model predictions with the experimental data, the model shown in Fig. 6.5a
was favoured. Clearly, hydrogen is believed to be intimately involved not just in
defect passivation, but in the detailed mechanism of defect formation and
annealing [9–11].
The above discussions are illustrative of some of the physical considerations
employed in analysing the defect formation process. When taking into account
specific defect reactions, such as the hydrogen-mediated bond breaking, this more
detailed treatment of the topic goes beyond the scope of this book. Furthermore, the
above discussion was limited to the formation of the neutral dangling bond, whilst
the formation of the charged dangling bond requires further considerations. These
include the influence of the different charge states on the defect formation energy,
and the changed energy level of the defect itself in the band gap. The latter effect is
a direct consequence of the defect not being a discrete level, but having a distri-
bution of energies [9–11], and this topic is briefly reviewed in Sect. 6.2.5. In spite of
the limited coverage here, it should be noted that there is an extensive literature on
this subject, some examples of which are contained in Refs. [1, 4, 9–11] and the
citations therein.
6.2.2 a-Si:H Density of States, DOS
Figure 6.1 showed a molecular orbital model of the a-Si density of states, and
many authors have confirmed its general features, as shown by the measured DOS
in Fig. 6.6 [12, 13]. The details of the trap distribution were derived by fitting the
model to experimental TFT Id(VG, Vd) data [13]. The figure shows both the deep
state and the tail state distributions, which, from an analytical point of view, can be
adequately represented by exponential functions, decaying from the band edges
with characteristic energy widths [13–15]. The states in the upper half of the band
gap are acceptor-like (which are neutral when empty, and become negatively
charged when occupied by electrons), whilst those in the lower half of the band
gap are donor-like. As discussed previously, and as shown by Fig. 6.6, the dis-
tributions are asymmetric, with a greater density in the lower half of the band gap,
such that the Fermi level in undoped material is in the upper half of the band gap
[14]. This location also defines the neutral level of the trapping state distribution,
where the net charge density is zero. The tail and deep state densities can be
approximated by the following exponential distributions:
Nta ðEÞ ¼ Nt0 exp ðEc EÞ=Eta ð6:12Þ
Fig. 6.6 Approximation to 10 22

Band States
the a-Si:H DOS, using Donor-like States Acceptor-like States
10 21 gd (E) ga (E)
different exponential
Tail States
DOS <cm–3 eV–1>

distributions for the tail states 10 20
and deep states. (Reprinted
1019
with permission from [12].
Copyright (1988) American 1018 Deep States
Institute of Physics)
1017
1016
1015
EV EC
1014
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
Energy (EF – EV) <eV>
Nda ðEÞ ¼ Nd0 exp ðEc EÞ=Eda ð6:13Þ

for the acceptor-like tail state and deep states, respectively. Comparable
expressions exist for the donor-like states. The values of the parameters Nt0, Nd0,
Eta and Eda vary with the sample, but Eta is of the order of 20–30 meV [13, 14],
and Eda is *3 times larger. The values quoted for Nt0 have varied from
2 9 1020 cm-3 eV-1 to 5 9 1022 cm-3 eV-1 [13, 14]. Part of this difference is
due to the way in which the tail state distribution has been terminated at the
conduction band mobility edge, either by using a linear distribution over the
intermediate 0.1 eV between the exponential distribution and the band edge [13],
or by specifying Nt0 directly at the band edge [14]. The deep state density, Nd0, has
been quoted in the range 5 9 1018 – 1 9 1019 cm-3 eV-1 [13, 14]. As an alter-
native to the exponential approximation for the distribution of deep states, some
authors have represented them by a pair of Gaussian distributions [16].
6.2.3 Band Bending and Surface Space Charge
When an a-Si:H TFT is switched on, the surface banding bending, induced by a
positive gate bias, will sweep some (or all) of the deep states and tail states through
the bulk Fermi level. This means that in order to induce a free electron density, to
support on-state conduction, the occupancy of the DOS will change as well. These
changes are normally calculated numerically [12–16], but, with some simplifica-
tions, it is possible to examine the key features at an analytical level, as discussed
below.
For a trapping state distribution given by:
NT ðEÞ ¼ NT0 exp ðEc EÞ=ET ð6:14Þ
the total trap occupancy, with the Fermi level at a position EF in the upper half of
the band gap, is given by:
ZEC
NT0 exp ðEC EÞ=ET
NT ¼ dE ð6:15Þ
1 þ exp ðEF EÞ=kT
0
This equation is not analytically soluble [15], but, by using the zero Kelvin
approximation (in which the states are occupied up to EF and empty above it),
Eq. 6.15 reduces to 6.16. (The limitations in using this approximation are dis-
cussed further below).
ZEF
NT ¼ NT0 exp ðEC EÞ=ET dE ð6:16Þ
0
from which,

EC EF
NT ¼ NT0 ET exp exp 1 ð6:17Þ
ET ET
With appropriate modification, Eq. 6.17 can be used to describe the local trap
occupancy in a surface space charge region, where the band bending is V. This is
shown schematically in the MIS band bending diagram in Fig. 6.7a, b, at flat bands
and at a positive value of gate bias, VG, respectively. For simplicity, a symmetric
exponential DOS is shown in this figure, with its neutral level at mid-gap, which is
where the bulk Fermi level is also located. The DOS is shown only at the a-Si/a-
SiNx:H interface, but it is continuous throughout the bulk of the a-Si, and the
acceptor-like state occupancy changes with position in the region within which the
bands are bent. The same convention is used as in Chap. 2, in which the potentials
are measured from the intrinsic level, so that the local trap occupancy at position x
from the surface, where the band bending is V, is given by:
Fig. 6.7 Schematic MIS

band bending diagram for a-
Si:H, showing trap
occupancy: a at flat bands—
VG = 0 V, and b with
electron accumulation—+ve
VG. (The traps are spatially
distributed in the a-Si:H, but,
for illustration purposes, the
diagram shows them only at
the a-Si:H/a-SiNx:H
interface)

EC qV
NT ðxÞ ¼ NT0 ET exp exp 1 ð6:18Þ
ET ET
and, by referencing all levels to mid-gap, the numerical value of EC in Eq. 6.18
is half the mobility gap, EG/2.
To establish the relationship between the surface potential, Vs, and Qs, it is
necessary to solve Poisson’s equation, where:
d2 V qqðxÞ
2
¼ ð6:19Þ
dx e0 es
e0 is the permittivity of free space, es is the dielectric constant of the semi-
conductor, and the space charge density, q(x), for positive gate bias, is:
qðxÞ ¼ nðxÞ NT ðxÞ ð6:20Þ
qV
nðxÞ ¼ ni exp ð6:21Þ
kT
Substituting Eqs. 6.18 and 6.21 into Poisson’s equation, and using the solution
procedure of Chap. 2 to obtain the surface space charge density, Qs, as a function of
2
d dV 2
the surface potential, Vs {i.e. using ddxV2 ¼ 12 dV dx , and integrating Eq. 6.19 from
the bulk (V = 0, and dV/dx = 0) to the surface (V = Vs, and dV/dx = -Fs)}:

0:5
pffiffiffiffiffiffiffiffiffiffiffiffi kT qVs EC qVs
Qs ffi 2qe0 es ni exp þ NT0 ET exp ET exp 1 Vs
q kT ET ET
ð6:22Þ
for Vs [ kT. For comparison, the corresponding expression from Chap. 2 for the
c-Si substrate is:

0:5
q kT
It will be seen that these two expressions have a similar form:
Qs ¼ A½BðVs Þ þ CðVs Þ0:5 ð6:24Þ

The first term in Eq. 6.24 relates to the free carrier contribution, and it has a
comparable form in both equations, differing only by the non-zero value of VF in
Eq. 6.23 due to the p-type doping of the c-Si substrate. The second term in
Eq. 6.24 relates to the immobile space charge. This is due to the ionised acceptor
density, Na, in the c-Si, whilst, in a-Si:H, it is due to the carriers trapped in the
DOS. However, these terms have a different form in the two materials, with the Na
term being linear in Vs, whereas the DOS term varies exponentially with Vs. As
seen for the c-Si substrate in Chap. 2, the exponential rise of the free carrier term in
Eq. 6.23 ultimately led to the free carrier density dominating the surface space
charge density at large enough values of Vs (see Fig. 2.2). In contrast, the fixed
space charge term in Eq. 6.22, which increases exponentially with Vs, can increase
at a comparable rate to the free carrier term if the value of ET is of the order of kT.
As discussed in Sect. 6.2.2, the width of the tail state distribution beneath the
conduction band edge is *20–30 meV, which is in the same range as the room
temperature value of kT. Hence, due to the tail states, there may not be a clear cut
situation in which the surface space charge density is dominated by the free carrier
density. As will be discussed in Sect. 6.2.4, this has implications for the mea-
surement of the carrier’s field effect mobility.
Some illustrative calculations of Qs as a function of Vs (using Eq. 6.22) are
shown in Fig. 6.8a, b. (In performing these calculations, the mobility gap has been
taken to be 1.7 eV [12], and the associated intrinsic carrier concentration estimated
to be 1.2 9 105 cm-3). In Fig. 6.8a, only a deep level distribution has been used,
with NT0 = Nd0 = 1 9 1018 cm-3 eV-1 and ET = Eda = 0.1 eV, and, in addi-
tion to the data points for Qs, the dashed and dotted lines represent the evaluation
of Eq. 6.22 using just the free charge or the fixed charge terms, respectively. When
one curve or the other is approximately equal to Qs, this is indicative of that term
dominating Qs. As will be seen in Fig. 6.8a, over most of the surface potential
range, Qs is determined by the trapped charge density, except at the largest values
of Vs where the increased rise in Qs is due to the free carrier charge. If the
distribution of deep traps were the only defects in the material, then the device
threshold voltage would be where the free carrier density starts to dominate Qs.
However, in practical devices, there is not just a deep level trap distribution, but
the tail states are present as well, and this is shown in Fig. 6.8b. In this example, a
sum of two exponential trap distributions has been used in Eq. 6.22: the deep level
distribution used in Fig. 6.8a, plus a tail state distribution given by
Nt0 = 2 9 1021 cm-3 eV-1 and Eta = 0.03 eV. The curve in Fig. 6.8b looks
superficially like the one in Fig. 6.8a, but the important difference is shown by the
dashed and dotted curves, which demonstrate that the increase in Qs, at the larger
values of Vs, is mainly due to the charge going into the tail states, and not due to
the free carrier density. Indeed, the free carrier density remains a minor contributor
to Qs. The significance of this on the field effect mobility is discussed further in
Sect. 6.2.4.
As with the c-Si situation, the gate voltage required to achieve a given surface
potential, Vs, is given by Eq. 2.22 from Chap. 2:
Hence, as Qs increases, due to the presence of trapping states, so the value of
VG needed to attain a particular value of surface potential Vs increases
accordingly.
The above discussion, and the use of Eq. 6.22, has been introduced to give an
analytical appreciation of the effect of the exponential distributions of the mobility
gap states on the way in which the surface space charge density is largely deter-
mined by these densities, with the free carrier density playing a minor role.

of surface space charge, Qs,
with surface band bending,
Vs, (using the analytical
approximation in Eq. 6.22):
a deep level defects only:
Nda = 1 9 1018 cm-3 eV-1,
Eda = 0.1 eV, b deep
level ? band tail defects:
Nda, Eda as above;
Nta = 2 9 1021 cm-3 eV-1,
Eta = 0.03 eV. (Qs—points
and solid lines; trapped
charge—dotted line; free
carrier term—dashed line)
However, it should be emphasised that Eq. 6.22 has been derived by using the zero
Kelvin approximation, and, particularly for the narrow distribution of the tail
states, where Eta * kT, this approximation cannot be used for a reliable, quan-
titative calculation of Qs—it merely gives a qualitative indication of the trends. An
accurate calculation requires the use of the Fermi–Dirac function to determine the
occupancy of the trapping states (i.e. Eq. 6.15), and an example of the numerically
calculated free and trapped charge densities (for the double exponential DOS in
Fig. 6.6) is shown in Fig. 6.9 [13]. This is a plot of the free and trapped carrier
densities at the a-Si:H surface as a function of the Fermi level at the surface
(measured from the valence band edge), where the neutral level is at 1.0 eV (see
Fig. 6.6). Hence, the x-axis in Fig. 6.9 is displaced by 1.0 eV compared with the
calculations in Fig. 6.8, but, apart from this, there is a clear qualitative similarity
between these two figures, including the change in slope of the trapped carrier
density, as the Fermi level moved from the deep states into the band tail states. The
filling of the deep states, and the movement of the Fermi level into the tail states,
Fig. 6.9 Numerically 1021

calculated free and trapped
electron densities as a 1020
function of the surface Fermi n
level position, using the DOS nt
1019
in Fig. 6.6. (Reprinted with
n, nt (cm–3)
Copyright (1989) American 1018
1017
1016
1015
1014
1.0 1.2 1.4 1.6 1.8 2.0
Fermi Level (eV)
has been empirically identified as the threshold for on-state behaviour in a TFT
[13, 17, 18], such that the value of the deep state density will determine the TFT
threshold voltage through Eq. 6.25. Figure 6.9 also shows that the free carrier
density remains below the trapped charge density, over much of the band bending
range, and the effect of this on the field effect mobility is discussed in the next
section. Finally, at the larger values of EF in Fig. 6.9, the Fermi level is above the
conduction band edge, leading to a ‘crystalline-like’ regime of operation at very
high gate voltages, where the band mobility for electrons would be attained.
However, this regime has not been achieved in practical devices [19, 20].
Only the surface carrier densities are shown in Fig. 6.9, and similar calculations
to those in the figure have examined the spatial distribution of trapped and free
carriers within the bulk of the surface band bending region [19]. For representative
values of band bending below and above threshold, such as 0.4 and 0.6 eV,
respectively (equivalent to EF = 1.4 and 1.6 eV in Fig. 6.9), the relative densities
in the bulk closely followed the ratios at the surface.
6.2.4 Field Effect Mobility
As discussed in Sect. 3.3.1, the carrier mobility can be extracted from the
experimental relationship between the drain current, Id, and gate bias, VG, in either
the linear or the saturation regimes of the TFT on-state characteristic. The fol-
lowing discussion uses the linear regime characteristics as an example, but the
principle applies to both regimes. From Eq. 3.11 in Sect. 3.3.1, the Id–VG rela-
tionship (for small VD) is given by:
ln WCi ðVG VT ÞVD

Id ¼ ð6:26Þ
L
and the electron field effect mobility, lFE, is derived from the slope of the Id–VG
curve above VT, using:
L dId
lFE ¼ ð6:27Þ
WCi VD dVG
Implicit in this relationship is the assumption that, for VG [ VT, all the charge
induced in the surface by the gate voltage is free charge. In other words, the areal
free charge density, Qn, is given by:
Qn ¼ Ci ðVG VT Þ Qs ð6:28Þ
This means that Qn is approximately equal to the total charge density, Qs. This
was clearly the case for the Qs calculations shown in Fig. 2.2 for Vs [ 2VF, and is
approximately the case for the large Vs values in Fig. 6.8a, but is not the case in
either Figs. 6.8b or 6.9 where a realistic tail state distribution was used in the a-
Si:H DOS. In these latter cases, it is necessary to modify Eq. 6.27 to take account
of the division of induced charge between free carriers and trapped carriers.
Writing Eq. 6.27 as:
L DId
lFE ¼ ð6:29Þ
WCi VD DVG
Where, from Eq. 6.26:
ln WVD DQn
DId ¼ ð6:30Þ
L
DQs
DVG ¼ ð6:31Þ
Ci
DQs ¼ DQn þ DQt ð6:32Þ
and DQt is the fixed space charge in traps. Substituting Eqs. 6.30 and 6.31 into
6.29 establishes the relationship between the field effect mobility and the band
mobility, ln:
DQn
lFE ¼ ln ð6:33Þ
DQs
Hence, when the free carrier density dominates the surface space charge (and
DQn * DQs) lFE * ln (as is the case for c-Si devices). However, for a-Si:H
TFTs with band tail states, DQn \ DQs, and consequently lFE \ ln. For a quan-
titative evaluation of this effect, a numerical integration of the free and trapped
charge through the film is required. This is shown in Fig. 6.10 [12], in which the
calculated ratio of the electron field effect mobility to the band mobility has been
Fig. 6.10 Calculated ratio of 0.10

electron field effect mobility 0.09 Relative
Electron
to electron band mobility, 0.08
T = 350 K Mobilities
using the DOS in Fig. 6.6 0.07
UFET /Uband
(thickness of a-SiNx:H was
0.06
300 nm). (Reprinted with
permission from [12], 0.05 T = 300 K
Copyright (1988) American 0.04
Institute of Physics) 0.03
0.02 T = 250 K
0.01
0.00
0 5 10 15 20 25 30 35 40 45 50 55
Gate Potential <Volts>
evaluated for the DOS in Fig. 6.6. This figure shows that the ratio increases both
with gate bias, due to differential trap and band occupancy, and with temperature
due to increased thermal occupancy of the conduction band. At VG = 15 V, the
room temperature ratio is *0.04, which, for a band mobility of 10–20 cm2/Vs,
gives a field effect mobility of *0.4–0.8 cm2/Vs. This is consistent with the
electron field effect mobility of *0.4–1.0 cm2/Vs measured in a-Si:H TFTs [21],
an example of which is shown in Fig. 6.11. This back-channel-etched TFT had
a-Si:H and a-SiNx:H film thicknesses of 200 and 400 nm, respectively, and a
threshold voltage and field effect mobility of *3 V and 0.60 cm2/Vs, respectively.
Equation 6.33 shows that the field effect mobility is equal to the band mobility
reduced by the ratio of the free carrier density to those trapped in band tail states,
and this is essentially the same as the drift mobility, which is also of the order of
1 cm2/Vs for electrons [7].
Whilst this section has focussed on a-Si:H TFTs, other TFT technologies have
also been characterised with a distribution of trapping states across the band gap,
and similarly show differences between the field effect and the band mobilities.
Therefore, the discussion in this section is broadly applicable, although the
Fig. 6.11 Experimental, 10 –6

linear regime transfer W
= 7.5
characteristics of a back L
1.0
channel etched a-Si:H TFT. d ins = 4000Å
10 –8 VD = IV
permission of IEEE)
I DμA
I D(A)
10 –10
0.5
10 –12
–5 0 5 10 15
VG (V)
particular mobility ratio will depend upon the detailed trapping state distribution
within the material. For instance, the analysis of large grain, high mobility poly-Si
TFTs has identified a lFE:ln ratio of 0.55 [22].
6.2.5 Equilibration in Thin Films
The effects to be discussed in this section are not specific to thin films, but the thin
films used in TFTs make them particularly sensitive to the influence of surface band
bending during device fabrication. The background to the process of dangling bond
equilibration was introduced in Sect. 6.2.1, in which the monitoring of the bulk
equilibration process used ESR measurements on undoped samples up to 90 lm
thick [11]. In the layer thickness range suitable for TFTs (i.e.\0.5 lm), the surfaces
play a relatively much greater role in determining the detailed electronic properties
of the resulting material. These electronic effects are controlled by the Fermi level
position during equilibration, and this ultimately determines the choice of gate
dielectric material, and the preferred channel type, as discussed in the following two
sub-sections. Not only is the equilibrium defect density determined by the Fermi
level position during equilibration, but it also influences the position of the DB states
within the mobility band gap. The reason why these effects are so important in thin
films is because the Fermi level at the surface is determined by the surface band
bending, which in turn is controlled by the charges in the gate dielectric during the
cooling of the TFT stack from its deposition temperature to the freeze-in tempera-
ture. As discussed in Sect. 5.2, the dangling bond is a positive-U amphoteric centre,
with an acceptor level at Ed2, and a donor level at Ed1 (see Fig. 5.2). Hence, if there is
positive charge in the gate dielectric, the bands will be bent down, and the surface
Fermi level will be positioned closer to the conduction band edge, such that, if it is
above the acceptor level of the defect state at Ed2, the state will be negatively
charged. Conversely, if the dielectric layer has a net negative charge, the bands will
be bent upwards, and the dangling bond will be positively charged if the surface
Fermi level is below the donor level at Ed1. The movement of the surface Fermi level
towards either of the band edges has two effects: it changes the defect formation
energy of the charged states compared with the uncharged state, and, as the defect
levels are distributed in energy, the resulting DB energy levels are also shifted from
the levels formed in near-intrinsic material [Chap. 6 in 1, 9–11]. The formation
energies are changed due to the change in electron energies associated with the
charged defect states. In the simplified presentation in Sect. 6.2.1, the defect for-
mation energy for the neutral state, was given by Eq. 6.4:

Ud D0 ¼ Ed1 EWB ð6:34Þ
Continuing with the same simplified formalism for the charged states, the
formation energy for the negatively charged state, at Ed2, becomes [p 195 in 1]:
Ud ðD Þ ¼ Ed1 EWB ðEF Ed2 Þ ð6:35Þ

The formation energy is reduced by EF - Ed2, which is the gain in energy due
to an electron falling from the Fermi level, EF, down to the upper level of the
dangling bond at Ed2. For the positively charged state, with Ed1 above the Fermi
level:
Ud ðDþ Þ ¼ Ed1 EWB ðEd1 EF Þ ð6:36Þ

In this case, the defect itself is empty of electrons, and the formation energy of
the neutral defect is reduced by the electron dropping from the defect level down
to EF. From the law of mass action, these changes in defect formation energies will
change the equilibrium defect densities, compared with the density produced in
intrinsic material.
As mentioned in Sect. 6.2.1, the above analysis is simplified by the assumption
of a single weak bond defect level at EWB, whereas, in reality, the calculation of
the equilibrium number of charged defects will require the law of mass action to be
integrated over the valence band tail state distribution [10, 11].
Finally, as the defect levels are distributed in energy, the defect formation
process will naturally select those states with the lowest formation energy, and, for
a Gaussian distribution of defect energies with a width rD, the change in defect
energy level, DE, for each added electron during the formation process is given by
[p 196 in 1]:
DE ¼ r2D =Ev0 ð6:37Þ
where Ev0 is the width of the valence band tail state distribution. The shifts in the
equilibrated defect energy levels with the position of the Fermi level during DB
formation are illustrated in Fig. 6.12, in which, for r * 0.1 eV, DE * 0.2 eV [p
196 in 1]. It will be seen that, with the Fermi level in the lower half of the band
gap, the positively charged, empty defect centre is formed at a high position in the
band gap. With higher Fermi level positions during defect formation, the addition
of each electron to form either the neutral or negatively charged levels causes
those levels to be formed progressively lower in the band gap. For each Fermi
p-type Undoped n-type

EC EC EC
EF
ΔE
Energy
2ΔE
EF
EF
EV EV E N (E)
V
Fig. 6.12 Illustration of the a-Si:H DOS, showing the dependence of the dangling bond energy
levels on the position of the Fermi level during defect formation. (From [1], Copyright
Cambridge University Press 1991, reproduced with permission)
level position during defect formation, the associated donor and acceptor energy
locations are shown in Fig. 6.12, and it should be emphasised that these energy
levels are frozen-in during cooling from growth. At room temperature, the frozen-
in distribution will not change, but the room temperature Fermi level can be moved
through the distribution, by surface band bending, thereby changing the occupancy
of the frozen-in centre.
As with the discussion of the neutral defect in Sect. 6.2.1, the presentation
above is highly simplified, in order to bring out the underlying physical arguments,
and for a more comprehensive overview and treatment the reader is referred to
Ref. [9].
6.2.5.1 Gate Dielectric
The preferred gate dielectric for a-Si:H TFTs is a-SiNx:H, which has a large
density of positive charge in the dielectric, and this might be expected to lead to
depletion devices with large negative threshold voltages [23]. However, this
intuitive expectation does not happen, as discussed below. On the other hand, SiO2
is preferred in c-Si MOSFETs, and has a much lower charge density, which might
be expected to lead to a threshold voltage closer to zero volts. Again, this does not
happen in practice. To examine the impact of the intrinsic dielectric charge on the
equilibrated defect density in a-Si:H TFTs, devices were annealed at 240 °C for
10 min with different biases on the gate electrode to simulate the effect of different
charge densities in the dielectric during sample preparation. To avoid charge
trapping effects in the normal a-SiNx:H gate dielectric, it was replaced by a
thermally grown layer of SiO2 on a heavily doped Si substrate, which also acted as
the gate electrode [24, 25]. The impact of differently biased equilibration regimes
on n- and p-channel TFT characteristics are shown in Fig. 6.13a, and simulated
results are shown in Fig. 6.13b. These are normalised transfer characteristics,
showing the channel sheet conductance versus the total charge density induced by
the gate measurement bias. Also, by using both p- and n-channel TFTs, the defect
distribution in both halves of the band gap could be assessed [24]. For the n-
channel TFTs in Fig. 6.13a, the devices with the lowest density of states in the
upper half of the band gap had the sharpest sub-threshold slopes, and were
obtained with a positive gate bias anneal (placing the Fermi level close to the
conduction band at the surface). Conversely, those with the highest defect density
had the lowest sub-threshold slope, and were obtained with a negative bias anneal.
The devices annealed at zero bias are representative of the native charge density in
the oxide films, and gave intermediate results in terms of n-channel behaviour.
Given the correlation between the sub-threshold slope and the defect density in the
upper half of the band gap, these results showed that superior n-channel perfor-
mance was obtained when the Fermi level was in the upper half of the band gap
during the equilibration process. Conversely, this equilibration condition degraded
p-channel performance.
(a) VGS(V) (b) VGS(V)

–20 –10 0 10 20 30 –20 –10 0 10 20 30
10
–7
10 –7
Vba= –10V 10–6 Vba= –10V 10–6
10–8 Vba= 0V 10–8 Vba= 0V
Vba= +10V 10–7 Vba= +10V 10–7
10–9 10–9
G (Ω/ )
G (Ω / )
Vba= +20V Vba= +20V
ISD(A)
ISD(A)
10–8 10–8
10–10 10–10
10–9 10–9
10–11 10–11
10–10 10–10
10–12 10–12
10–11 10–11
–20 –10 0 10 20 30 –20 –10 0 10 20 30
2 11 2
11
Q i(10 electrons/cm ) Q i(10 electrons/cm )
Fig. 6.13 Transfer characteristics of n- and p-channel a-Si:H TFTs with an SiO2 gate dielectric,
in which the dangling bond density was equilibrated with different gate biases, Vba: a measured
after equilibration at 240 °C, and b simulations. (Reprinted with permission from [24]. Copyright
(1993) American Institute of Physics)
The simulations in Fig. 6.13b included constant band tail state densities and a
spatially varying DB density, where the densities at the a-Si:H/SiO2 interface, and
in the bulk, are shown in Fig. 6.14. They demonstrate that a positive bias anneal
positions the minimum of the surface defect distribution above mid-gap (leading to
the enhanced n-channel performance), and causes the peak of the defect
distribution to be located below mid-gap (thereby degrading the p-channel per-
formance). The opposite effects are seen with the negative bias anneal. As will be
appreciated, the positioning of the peaks in the defect distributions at the surface
correlate with the surface Fermi level position during bias equilibration, and are
consistent with the Fermi level effects illustrated in Fig. 6.12. A related feature in
Fig. 6.14 is the difference between the surface DB densities and those in the bulk,
illustrating that the defect density is spatially varying, and is due to the spatial
variation in the separation of the Fermi level and the band edges within the band
bending region. Moreover, the defect distribution in the neutral bulk is more
Fig. 6.14 Spatially varying 10 20

DOS used in the defect
equilibration simulations in
1019
g(E) (cm–3 eV )
Fig. 6.13b. (Reprinted with

–1

Copyright (1993) American 1018
17
10
bulk
16 V ba = 0V interface
10
V ba = –10V interface
V ba = +20V interface
15
10
EV 0.4 0.8 1.2 1.6 EC
E (eV)
VGS(V)
(a) –40 –30 20 –10 0 10 (b) 1020
10–7 –5
Vba = 0V 10
10–8 1019
g(E) (cm–3eV )
–6
Vba= –15V
–1
10
10–9
G (Ω / )
–7 18
10 10
I (A)
–10
10 –8
SD
10 1017
10–11 –9
10
1016 Vba = –15V interface
10–12 –10
10 Vba = 0V interface
15
10
–40 –30 20 –10 0 10 Ev 0.4 0.8 1.2 1.6 Ec
Qi (1011electrons/cm 2 ) E (eV)
Fig. 6.15 a Transfer characteristics of n- and p-channel a-Si:H TFTs, with an a-SiNx:H gate
dielectric, in which the dangling bond density was equilibrated with different gate biases, and
b interfacial values of the spatially varying DOS used to simulate the results in (a). (Reprinted
with permission from [24]. Copyright (1993) American Institute of Physics)
evenly distributed across the middle of the band-gap, as expected when the Fermi
level is near mid-gap. In summary, the changes in the density and energy location
of the defects frozen-in after the equilibration anneals are consistent with the
Fermi level effects discussed in the preceding section, and are a key feature of the
defect pool model of defect formation in a-Si:H [24].
In support of the conclusions drawn from Fig. 6.13a, the transfer characteristics
of a conventional a-Si:H TFT, with an a-SiNx:H gate dielectric, containing a fixed
positive charge density of 11.1 9 1011 cm-2 have also been studied [24]. To
reduce the effects of charge injection into the dielectric, the bias annealing con-
ditions were limited to 0 V and -15 V. The measured characteristics are shown in
Fig. 6.15a, and the defect distribution used to simulate them is shown in Fig. 6.15b
[24]. The device annealed at zero bias had near ideal characteristics, and the
simulated defect distribution, in Fig. 6.15b, had the same minimum defect density
in the upper half of the band gap (and a peak in the lower half of the band gap) as
the oxide samples annealed with positive gate bias. Moreover, if the zero bias
annealed SiO2 sample is compared with the zero biased a-SiNx:H sample, it is
clear that the large positive fixed charge density in the latter gave superior TFT
performance, and explains the universal preference for the a-SiNx:H gate dielec-
tric. In addition, the high defect density in the lower half of the band gap reduces
hole conduction at negative gate biases in the off-state of n-channel TFTs [23, 24].
6.2.5.2 n- and p-Channel TFTs
From the density of states distribution shown for the conventional a-Si:H TFT in
Fig. 6.15b, it is clear that the fabrication process which delivers a low dangling
bond defect density in the upper half of the band gap, also delivers a density which
is *100 times greater in the lower half. As discussed in Sect. 6.2.3, this high
Fig. 6.16 Transfer 2.0 100

characteristics measured on
n- and p-channel ambipolar 80
1.5
TFTs. (Reprinted with
ISD (μA)
ISD (μA)
permission from [26]. 60
1.0
Copyright (1987) American
40
0.5
20
0.0 0
–25 –20 0 2.5 5
VG
defect density will have a direct impact upon the threshold voltage of p-channel
TFTs, by making it more negative. This is apparent in the transfer characteristics
shown in Fig. 6.15a, and is more directly shown in the comparison of the n- and p-
channel linear regime characteristics in Fig. 6.16 [26]. These show that the
threshold voltage of the p-channel TFT was *-23 V compared with *2.5 V for
the n-channel TFT. In addition, the greater density of valence band tail states,
compared with the conduction band tail states, will further reduce the hole field
effect mobility from its band value. The field effect mobility is usually calculated
from Eq. 6.27; however, for devices with identical geometries, Eq. 6.26 shows
that the simple ratio of their on-currents is indicative of the ratio of their field effect
mobilities. From an inspection of the current scales in Fig. 6.16, it is clear that the
hole field effect mobility is *50–100 times less than the electron field effect
mobility. Hence, this data set indicates a hole field effect mobility of
*0.005–0.01 cm2/Vs. This is of the same order as the hole drift mobility of
0.003 cm2/Vs [7], and, as discussed in Sect. 6.2.4, these two measurements are
essentially measuring the same quantity, which is the band mobility reduced by the
ratio of carriers in extended states to localised band tail states.
In summary, the large threshold voltage and the very low field effect mobility of
p-channel a-Si:H TFTs explain the universal preference for n-channel TFTs in
commercial applications.
6.3 TFT Characteristics
6.3.1 On-State
The on-state characteristics are determined, to a first order, by the threshold

voltage and the field effect mobility. As discussed in Sects. 6.2.3 and 6.2.4, the
threshold voltage is determined by the deep state density, and the field effect
mobility is determined by the tail state density. Both device parameters are usually
extracted from the experimental transfer characteristics, in either the linear or
saturation regime, using the simple IGFET Eqs. 3.11 and 3.17, respectively, from
Chap. 3. In particular, the field effect mobility is calculated from the slope of
the Id-VG or HId(sat)–VG curves, using Eqs. 3.12 or 3.18, respectively, and the
threshold voltage is obtained from the extrapolated intercept of the curve with the
VG axis at Id = 0, as illustrated in Fig. 6.11.
Although these simple equations are used for basic parameter extraction, for
more accurate representation of the characteristics in compact circuit modelling
packages, such as SPICE, account needs to be taken of detailed aspects of device
behaviour including the distribution of carriers in the band tail states, and the finite
output impedance using the following expression [15, 27]:
lWCia1 n
Id ¼ ½ðVG VT Þa ðVG VT VD Þa ½1 þ kðVD VDsat Þ ð6:38Þ
aL
.
Where a and n are related to the tail state energy width, Et, and the carrier
density trapped in the tail states, Nt, by:
a ¼ 2Et =kT ð6:39Þ
ðqes e0 akTNt Þ1a=2

n¼ ð6:40Þ
a1
For the conduction band tail states, when Et * kT, then a * 2 and n * 1, and
the first part of Eq. 6.38 reduces to the basic IGFET Eq. 3.10:
ln WCi
L
The final term in Eq. 6.38 contains the channel shortening parameter, k, which
allows for the finite output impedance of the Id–VD characteristic in saturation.
A further modification to Eq. 6.38 (and, indeed, to Eqs. 3.10 and 3.26) is
required when there is series resistance in the TFT structure [27, 28]. This can
arise for several reasons, including contact resistance between the metal source
and drain electrodes and the n+ regions, the limited conductivity of the n+ regions
themselves, the thickness (or bulk resistance) of the a-Si:H film through which
carriers flow from the channel to the n+ regions, and the overlap of the gate and the
source/drain contacts [28]. For a fixed ohmic resistance, Rs and Rd, at the source
and drain contacts, there will be voltage drops IdRs and IdRd, respectively, which
will change the effective values of the gate-source voltage, VGS, and source-drain
voltage, VDS, from the externally applied voltages VG and VD, respectively, to:
VGS ¼ VG Id Rs ð6:42Þ
VDS ¼ VD Id ðRs þ Rd Þ ð6:43Þ

Hence, VG and VD in Eq. 6.38 (and in Eqs. 3.10 and 3.26) need to be replaced
by the expressions for VGS and VDS, respectively. However, the series resistance
has to be of the same order as the channel resistance for it to have a significant
6.3 TFT Characteristics 165
Fig. 6.17 2-D simulations of series resistance effects in BCE and FGP TFTs. a The variation of
the TFT on-state conductance with specific contact resistance. The effects on the mobility and
threshold voltage of b the specific contact resistance, c the a-Si:H film thickness, and d the
overlap of the gate and source/drain electrodes. (The square and circular symbols represent the
TFT and FGP data, respectively). (Reprinted from [28]. Copyright (1998), with permission from
Elsevier)
effect. If the series resistance is non-ohmic, then its effective resistance value will
be a function of the potential across it, and the detail of this will depend upon the
current flow mechanism. For instance, the current flow through the thickness of the
a-Si:H film is a space charge limited current [29], in which IµV2.
The calculated effect of the specific contact resistivity is shown in Fig. 6.17a by
the normalised channel conductance plots of a conventional BCE TFT [28], where
the conductance, G, in the linear regime, can be calculated from Eq. 6.26, using:
Id L
G¼ ¼ ln C i ð V G V T Þ ð6:44Þ
VD W
With increasing values of the specific contact resistance, the roll-off in G with
increasing VG was due to the reducing channel resistance becoming increasingly
comparable to the fixed series resistance.
Similar calculations were made for a modified TFT structure, containing two
extra electrodes contacting the channel (labelled as a gated 4-probe, GFP, TFT)
[28]. These two electrodes sense the channel potential without drawing any cur-
rent, and, therefore, were not affected by the contact resistance. The calculated
GFP TFT conductance (not shown) was quite different, since, in the absence of
series resistance effects, its G curve was independent of the contact resistance and
did not roll off with increasing VG values. Figure 6.17b compares the extracted
values of lFE and VT for the two TFT architectures, and, as expected, lFE in the
normal TFT architecture fell with increasing contact resistance, as did VT, whilst
those of the GFP TFT remained constant. (The mobility is evaluated at the point of
maximum slope of the Id – VG curve, and this tangent to the curve is also used to
extract VT, which, to a certain extent, is an artefact of the extraction procedure).
Two-dimensional device simulations were also used to quantify the effects, on
the channel conductance, of the film thickness and of the overlap of the gate and
the source/drain electrodes [28]. Their impact upon the extracted field effect
mobility and threshold voltage are shown in Fig. 6.17c, d, respectively. In the
simulations, unless the specified parameter was the variable, the intrinsic (resis-
tance-free) field effect mobility was 1.78 cm2/Vs, the specific contact resistivity
was 0.25 X cm2, the gate-source/drain overlap was 3 lm, the a-Si:H film thickness
was 200 nm, and its dark resistivity was 3.8 9 1010 Xcm [28, 30]. These values
corresponded to experimental TFT and GFP structures, which displayed charac-
teristics comparable to the simulations [28, 30]. As will be seen in Fig. 6.17, each
parameter reduced both the apparent field effect mobility and the threshold volt-
age. However, the spreading resistance due to the gate-source/drain overlap was
not significant for overlaps of 3 lm or more, which is the typical value for con-
ventional a-Si:H TFTs. Equally, the effect of a 200 nm, or thinner, a-Si:H film was
marginal compared with much thicker films, while the measured specific contact
resistivity of 0.25 X cm2 had the largest overall effect.
The contact resistance effects simulated in Fig. 6.17a were for constant resis-
tances, but bad contacts can also give voltage dependent resistance values. These
can lead to poor output characteristics, displaying low drain bias current crowding
effects, as shown in Fig. 6.18a [31]. For comparison, a TFT with a more ideal
output characteristic, and no current crowding effects, is shown in Fig. 6.18b [30].
High quality TFTs will not show current crowding effects [21], but, where it does
occur, the field effect mobility cannot be meaningfully evaluated in the linear
regime, and, for this reason, it is common practice for the mobility to be evaluated
in the saturation regime. It should also be noted that current crowding effects are
not limited to a-Si:H TFTs, but are observed with other TFT materials as well, and,
for the same reason, the saturation regime mobility is also frequently quoted for
those devices.
Fig. 6.18 Output characteristics from two different TFTs showing a current crowding at low
drain bias (Reprinted from [31] with permission of IEEE), and b good quality contacts (Reprinted
from [30] with permission of IEEE)
6.3.2 Off-State
The off-state characteristics of the n-channel a-Si:H TFT define its leakage current
behaviour under negative gate bias, and several leakage current mechanisms have
been identified. These include leakage through the a-SiNx:H gate dielectric, bulk
conduction through the a-Si:H film itself, back channel electron conduction, and
front channel hole conduction [32–34]. The magnitude of the current is a function
of the following parameters: the applied gate and drain biases, the a-Si:H film
thickness (which will determine its bulk conductance, and the coupling of the gate
to the back channel), the gate/drain overlap, and the overall fabrication process,
which will determine, among other things, the back-face passivation [32, 35]. An
example of an extensively studied set of characteristics is shown in Fig. 6.19a, b,
which were taken from an etch-stop TFT, with 50 and 250 nm thick a-Si:H and
a-SiNx:H layers, respectively [33]. Two of the significant negative gate bias fea-
tures of these curves are the exponential increase in current with gate and drain
bias at large negative gate bias, and the change from inverse channel length
dependence to channel length independence with increasing negative gate bias.
This latter change highlights a difference in the leakage current mechanism, where
the channel length independence is indicative of a current limited by generation
processes at the drain terminal, whereas the channel length dependence is more
consistent with an ohmic-like current flow.
Of the four leakage current processes listed above, the bulk leakage through the
a-Si:H and a-SiNx:H layers are the limiting minimum currents [33], and, in
practical devices, the processes of back channel electron conduction and front
channel hole conduction were found to dominate [34]. The back channel con-
duction is due to a weak electron accumulation layer at the back interface, and is
Fig. 6.19 Experimental TFT transfer characteristics, showing the leakage current as a function
of (a) drain bias (W = 50 lm, L = 20 lm; the solid line is a simulation), and (b) channel length.
(Reprinted from [33] with permission of IEEE)
determined by the top passivation layer. This can be depleted by the front gate, if
the a-Si:H layer is thin enough to permit front gate/back channel coupling. This
mechanism is shown by the fitted curve to the experimental data in the reverse sub-
threshold regime in Fig. 6.19a. The reverse sub-threshold slope in this region is a
function of both the deep trap density at the back of the film [33], and the strength
of the gate coupling, which is diminished by the electrostatic screening of a hole
accumulation layer at the front interface [34]. Hence, the depletion of the back
channel current is strongest in thinner films [32]. This mechanism dominates at
low negative gate voltages.
At large negative gate voltages, a hole accumulation layer is formed at the front
interface, which can support hole current flow, but, given the non-hole-injecting
nature of the n+ contacts, it is limited by the internal hole generation rate. Holes are
thermally generated from the deep traps in a-Si:H by sequential electron-hole
emission, and, in the presence of large gate and drain biases, this process can be
enhanced by the Poole–Frenkel effect [32–34], which lowers the emission barrier
(ET) for carrier generation by DET. This field-enhanced emission process takes
place in the high field region of the device in the vicinity of the gate/drain overlap,
where the Poole–Frenkel generation rate, ePF(F), is given by [36]:
ePF ðFÞ ¼ e0 expðDET =kTÞ ð6:45Þ
e0 is the field-free thermal generation rate (see Sect. 2.3.1 and Eqs. 2.54 and 2.55
for further details), and:
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
DET ¼ ðqF=pes e0 Þ ð6:46Þ
In this process, the generated electrons are swept into the drain, and the holes
flow to the front interface, where they accumulate as a steady state hole accu-
mulation layer. This transports the hole generation current, which flows from the
drain area and is injected into the forward biased source junction. From Eqs. 6.45
and 6.46, the carrier generation rate increases exponentially with the square root of
the field, F, and this model provided a good fit to the large negative gate bias data
in Fig. 6.19a, where the leakage current rose exponentially with gate and drain
bias.
In many respects, this process is similar to the field enhanced leakage currents
observed in poly-Si TFTs [33], with one notable difference being the coplanar
architecture of the latter giving stronger 2-D coupling of the gate and drain fields.
Further details on the basic electron and hole generation processes can be found in
Sect. 2.3, and the application of this process to TFT off-state currents is described
more fully in Sect. 8.4, together with a broader discussion of the field enhancement
mechanisms in Sect. 8.5.3.
6.4 Bias-Stress Instability
6.4.1 Gate Bias
The major instability in a-Si:H TFTs is a shift in threshold voltage after the
application of a gate bias stress, VGst, and this has been extensively studied over
many years. Two principle instability mechanisms have been identified, which are
carrier trapping in the a-SiNx:H gate dielectric, and dangling bond formation in the
a-Si:H itself. These two effects were distinguished in ambipolar [26, 37] and
optically illuminated n-channel [38] a-Si:H TFTs, in which the effects of gate bias
could be observed in both the electron and hole conduction branches of the same
TFT. For carrier trapping in the gate dielectric, both characteristics were expected
to shift in the same direction (e.g., in a positive direction for positive gate bias
giving electron trapping), and in opposite directions for amphoteric deep state
creation, in which electron filling of the acceptor (0/-) level would increase the
n-channel TFT threshold voltage, and hole filling of the donor (0/+) level would
increase the p-channel threshold voltage. These effects were seen in the ambipolar
TFT measurements shown in Fig. 6.20a, b [21], in which low and high gate bias
stresses, of +25 and +55 V, caused the characteristics to either move apart or to
both shift in the same direction, respectively. Hence, the low positive bias stress
caused deep state formation, but which was overwhelmed by the effects of electron
injection into the a-SiNx:H at large gate biases.
With negative bias stress of -25 V, positive charge injection dominated,
although there was also some defect formation in the a-Si:H, but less than under
positive bias stress [26, 37].
The positive stress bias, at which carrier injection starts to dominate, is
determined by the quality of the a-SiNx:H layer, and this is a function of the
deposition conditions, including the SiH4/NH3 gas ratio [39]. The quality of
the a-SiNx:H layer has been correlated with the value of its optical band gap, and
the greater the band gap, the higher was the threshold for the onset of the trapping
instability, with a field of *2MV/cm required for injection into the best quality
Fig. 6.20 The effect of gate

bias stress (broken lines) on
the transfer characteristics of
ambipolar TFTs: a +25 V
stress for 4 h, and b +55 V
stress for 103 s. (Reprinted
from [21], with permission of
IEEE)
nitride layers with a 5.0 eV band gap [39]. In contrast, differently prepared sam-
ples showed a mixture of both instabilities under all positive bias stress conditions
[37]. Hence, depending upon the fabrication procedures, the resulting devices may
display varying relative amounts of carrier injection into the nitride and trap
generation in the a-Si:H, under a given set of bias conditions [26, 37].
In addition to the different stress voltage dependence of the two instability
mechanisms, there was also a different time dependence, as shown in Fig. 6.21a
[21]. In this plot, the bias stress at 100 V caused the carrier injection instability to
dominate, and the change in threshold voltage, DVT, had an approximately log-
arithmic time dependence, given by [40]:
DVT ðtÞ ¼ Alogð1 þ t=t0 Þ ð6:47Þ
where DVT(t) = VT(t)-VT0 (VT0 = VT at t = 0), A is a constant, and t0 is a
characteristic time. The logarithmic time dependence of this instability has been
explained by the exponential dependence of the tunnelling injection current on the
injecting field, and the field decreases with carrier trapping due to the screening
effect of the trapped carriers [40]. The injection instability also had very weak
temperature dependence [40].
The instability caused by the 25 V bias stress, shown in Fig. 6. 21b on a log–log
plot, was due to deep state creation in the a-Si:H, and has been described by a
power law [40]:
6.4 Bias-Stress Instability 171
Fig. 6.21 Time dependence

of gate bias stress instability
(a) log-linear plot, and
(b) log–log plot. (Reprinted
from [21], with permission of
IEEE)
DVT ðtÞ ¼ DVT ð1Þðt=sÞb ð6:48Þ

Other authors have empirically fitted the change in threshold voltage (due to
deep state generation) with a stretched exponential [38], which has been widely
used, and, in some cases, fitted to all the instability results [37, 41], where:
h i
DVT ðtÞ ¼ DVT ð1Þ 1 exp ðt=sÞb ð6:49Þ
in which DVT(?) = VGst-VT0 [37, 40–42], b is weakly temperature dependent,

and the characteristic defect formation time constant, s, is given by:
Fig. 6.22 Time dependent threshold voltage as a function of: a bias temperature for
VG = +20 V, and b different gate bias stress values at 403 K (on the left hand side), and
relaxation data, at VG = 0 V, after gate bias stressing (on the right hand side). (Reprinted with
permission from [43] Copyright (1990) by the American Physical Society)
s ¼ s0 expðEA =kTÞ ð6:50Þ

For times short compared with s, Eq. 6.49 reduces to the power dependent form
given by Eq. 6.48 [37, 40].
Equations 6.49 and 6.50 describe a process which is thermally activated, and
which saturates after several time constants with a threshold voltage shift of
DVt(?). These features are displayed in the time dependent threshold voltage data
in Fig. 6.22a [43], in which the samples were stressed at VGst = 20 V at different
temperatures. Figure 6.22b shows the dependence of the threshold voltage shift on
the stress bias, VGst, and also demonstrates that the shift could be annealed back to
zero after bias removal, if the sample was held for long enough at the stressing
temperature.
The features in Fig. 6.22 are similar to the effects discussed in Sects. 6.2.1 and
6.2.5, on defect generation and equilibration in a-Si:H, and these non-injection
effects of gate bias stress are similarly interpreted in terms of hydrogen-mediated
weak bond breaking and dangling bond formation. The driving force for bond
breaking is the increased carrier density in the band-tail states, and the associated
movement of the Fermi level towards the band edges by the gate bias, which
induces a re-equilibration process in the material. The effects have been interpreted
in terms of specific defect reaction kinetics, using the defect pool concepts [42], to
replace the empirical stretched exponential fit with a more fundamental model. In
particular, the model needed to account for the poor stretched exponential fit for
t [ s [38, 44], and also to incorporate the power dependence of DVT on VGst-VT0
[37, 42], which is shown in Fig. 6.23a [42]. By accounting for the forward and
backward defect formation and annihilation rates, and the dependence of the
formation energy of charged defects on the Fermi level position, as shown in
Fig. 6.23b, a defect reaction model was established which fitted the data better
Fig. 6.23 a Intermediate time (t = 1000 s) threshold voltage shift as a function of stress bias, V0
(V0 = VGst-VT0), for constant current stressing of low and high mobility TFTs, and b charged
defect formation energy as a function of the Fermi level position. (Reprinted with permission
from [42] Copyright (2003) American Institute of Physics)
than the stretched exponential [42]. It also yielded a value of a = 1.5, in agree-
ment with the power dependence of DVT shown in Fig. 6.23a [42]. Whilst these
results were derived from the numerical solution of the reaction rate equations, the
solution for a = 1.5 was well approximated to by a stretched hyperbola of the
form [5, 44]:
h i2
b
DVT ðtÞ / 1 ðt=sÞ þ1 ð6:51Þ
In common with the stretched exponential, the above expression accounts for
the fact that the defect formation rate cannot be represented by a single reaction
time constant, and this is interpreted as being due to the presence of a range of
reaction barrier heights, rather than the single one shown in Fig. 6.4a. For quan-
titative evaluation of defect formation rates between different samples and stress
temperatures, it was useful to replace t in Eq. 6.51 with a thermalisation energy
[42, 44, 45], Eth. This is the reaction barrier height below which all defects will be
formed in a time t, where the relationship between t and Eth is given by:
Eth ¼ kTlnðmtÞ ð6:52Þ
m is the attempt to escape frequency, which was found by curve fitting to be 109 Hz
for defect creation [45]. For a mean barrier height EA, and a barrier height dis-
tribution width of kT0, the stretched hyperbola is reformulated as [45]:
(
2 )
Eth EA
DVT ðEth Þ ¼ ðVGst VT0 Þ 1 exp þ1 ð6:53Þ
kT0
Depending upon the deposition conditions, EA was found to be 0.93–0.98 eV

and kT0 was 54–69 meV for defect creation, whereas, for the defect annihilation, m
was calculated to be 5.7 9 1013 Hz for the Si–H bond stretching mode, EA was
*1.55 eV and kT0 was *41 meV [45]. The defect creation results have been
interpreted in terms of the breaking of a weak Si bond in the vicinity of a doubly

hydrogenated centre, which triggered a rearrangement of the hydrogen atoms to
give two nearby defect complexes, both consisting of a dangling bond adjacent to a
hydrogenated bond [45]:
SiSi þ SiHHSi ! 2SiD þ SiHHSi ! 2SiHD ð6:54Þ
The larger value of EA for defect removal (when m is taken to be the Si–H bond
vibration frequency) is similar to the activation energy for hydrogen diffusion,
suggesting that defect annihilation is limited by the release and movement of
hydrogen to a dangling bond site [45].
6.4.2 Pulsed Bias Stress
The discussion above focussed on DC gate bias stress instability, but, in active
matrix displays, the pixel TFT experiences a low duty cycle of positive gate bias
during the row on-time, followed by a longer period of negative gate bias for the
rest of the frame time. Hence, a pixel TFT experiences alternating positive and
negative pulsed gate bias stress, but with quite different duty cycles, and it has
been demonstrated that the effects of negative pulsed bias stress are quantitatively
different from DC gate bias stress [37]. Figure 6.24 compares the threshold voltage
shift for the DC and pulsed bias stress conditions, for both stress polarities, as a
function of the integrated time during which the pulsed stress was applied. For
positive pulse bias stress, DVT was slightly smaller than DC stress, and this was
attributed to partial relaxation of the deep trap density during the pulse off period.
Hence, by taking into account the pulse bias duty cycle, the DC and pulsed results
were very similar. For negative pulse bias stress, there was a strong pulse width
dependence, in which DVT decreased with reduced pulse width. This was attrib-
uted to the instability process being driven by the hole density in the channel, and,
due to the n+ source and drain contacts, there was not a ready supply of holes. In
other words, the hole density could not respond rapidly to the changes in negative
gate bias, but was limited by the supply process for the holes. The experimental
dependence of DVT on the pulse width was empirically modelled by an RC circuit,
in which the ‘resistance’ element of the time constant represented the supply of
holes to charge the gate dielectric capacitor at the a-Si:H/a-SiNx:H interface [37].
Due to the difference between the DC and pulsed stressing, it was demonstrated
that, in order to estimate the projected lifetime of an AMLCD, the calculation
needed to be based upon the pulsed bias instability results rather than the DC
results, and that the summation of the separate positive and negative pulse bias
stress results accurately reflected an alternating positive/negative pulse bias stress.
Hence, the net effect of alternating stress, DV± T , could be calculated from [37]:

DV þ
T ðtÞ ¼ DVT ðtÞ þ DVT ðtÞ ð6:55Þ

threshold voltage instability
due to DC and pulsed bias
stressing. (Reprinted with
Copyright (1998) The Japan
Society of Applied Physics)
and,
þ aþ
DVþ
T ðtÞ ¼ A VG VT0 ðtTon =Tfr Þbþ ð6:56Þ

a
DV
T ðtÞ ¼ B VT0 VG ðtf1 Ton =Tfr gÞb FPW ð6:57Þ
Where Eqs. 6.56 and 6.57 are based upon the power law Eq. 6.48 (which is an
acceptable approximation for DVT values far from saturation), and they also include
the power dependence of the bias conditions (the first term) to reflect the experi-
mental data shown in Fig. 6.23a. The parameters A, B, a and b were empirically
fitted constants to the DC bias results, the effects of which were modified by the duty
cycle, given by the ratio of the row on-time, Ton, to the frame time, Tfr, and FPW was a
further weighting factor (determined from the RC charging model), to represent the
specific pulse width dependence of the negative bias pulse instability. These
equations were shown to approximate well to experimental alternating pulse data,
and were used to predict the long term DVT values for a pixel TFT in an SXGA
AMLCD. This is shown in Fig. 6.25, and indicates that, for a maximum tolerable
threshold voltage shift of 3 V, the display would have an acceptable operating life of
*10 years at 70 °C [37]. The initial, negative values of DVT were due to the longer
duty cycle of the negative gate bias, but, eventually, the stronger cumulative effect of
the positive bias stress dominated the overall shift. This was due to the larger value of
b+ compared with b- in Eqs. 6.56 and 6.57 [37].
Fig. 6.25 Predicted pixel

TFT lifetime in an SXGA
AMLCD. (Reprinted with
6.4.3 Gate and Drain Bias Stress Effects
In early work studying the combined effects of gate and drain bias stress [46], very
little influence of the drain bias (or the level of channel current) was seen on the
threshold voltage instability, which was determined mainly by the magnitude of
the gate bias. In subsequent work, in both the low (\15 V) [47] and higher gate
bias (C20 V) regimes [48], where the dominant instability processes were defect
state generation in the a-Si:H and charge injection into the a-SiNx:H, respectively,
a drain bias dependence was demonstrated, which, at most, had the effect of
reducing the normal gate bias instability. This is shown in Fig. 6.26a, for various
combinations of gate-source bias, VGS, and source-drain bias, VDS, in which, at
constant VGS = 15 V, the largest values of DVT occurred for VDS = 0 V, and the
smallest values for VDS = 15 and 30 V [47]. The latter results indicate that the
ameliorating effect of the drain bias saturated at the drain saturation voltage. For
defect formation in the a-Si:H, which is driven by the gate induced electron
concentration in the band-tail states, the results were interpreted in terms of the
average charge density, QG, in the channel. In the linear regime at low VDS, this is
given by:
QG0 ¼ Ci WLðVGS VT Þ ð6:58Þ
Fig. 6.26 a Threshold voltage instability measured after different gate and drain bias stress
combinations (solid lines modeled using Eq. 6.60), and b measured and modeled normalized
threshold voltage shift as a function of normalized channel charge. (Reprinted from [47] with
permission of IEEE)
However, when VGS and VDS are of the same order (and VGD = VGS-VDS),
the gate and channel charge are given by [47]:
2 ðVGS VT Þ3 ðVGD VT Þ3
QG ¼ Ci WL ð6:59Þ
3 ðVGS VT Þ2 ðVGD VT Þ2
For small VDS, Eq. 6.59 tends towards 6.58, and, for VGD * VT
(i.e.VDS * VGS–VT) Eq. 6.59 tends towards 2/3QG0. From the power law
Eq. 6.48, together with Eqs. 6.58 and 6.59, DVT can be represented by the nor-
malised channel charge [47]:
DVT ðtÞ ¼ ðQG =QG0 ÞAðVGS VT0 Þtb ð6:60Þ

Using Eq. 6.60, the threshold voltage shift, normalised to the maximum shift at
VDS = 0, is plotted against the normalised channel charge in Fig. 6.26b, and the
solid lines in Fig. 6.26a were also calculated using Eq. 6.60 [47]. Similar argu-
ments, in terms of the reduced gate field at the drain, were also used for the
reduction in carrier injection into the gate dielectric under combined gate and drain
bias stress [48].
6.5 Other Meta-Stability Effects
Meta-stable defect formation induced by gate bias stressing and film doping were
discussed in Sects. 6.4 and 5.5.2, respectively, and a further situation, under which
there is meta-stable defect production, is when a-Si:H is exposed to high, absorbed
light intensities for prolonged periods of time. This is known as the Staebler–
Wronski effect [49], and was first observed as a reduction in the photoconductivity
and dark conductivity of illuminated material. This was attributed to an increased
recombination centre density, and is a major degradation mechanism in a-Si:H
solar cells. It reduces the carrier recombination lifetime, and hence the cell’s
quantum efficiency. Although this is less of an issue for TFTs, for completeness, a
brief overview of this effect is presented below.
6.5.1 Staebler Wronski Effect
The Staebler–Wronski effect has been extensively studied for many years [49–53],
and one outcome of this has been the recognition that there are a number of light
induced defect generation processes. These depend upon the illumination tem-
perature, the growth procedure, and the quality, hydrogen content and doping
content of the material [50, 51, 53]. In view of this, there is still discussion over the
detailed mechanisms and processes [53], and the over-view in this section is
restricted to defect generation in undoped a-Si:H near room temperature, which
illustrates the essential features of the classical Staebler–Wronski effect.
Figure 6.27a shows the change in concentration, with illumination time and
intensity at 310 K, of the a-Si:H g = 2.0055 paramagnetic centre defect, which
has been associated with the neutral Si dangling bond [52]. The photon energy was
1.9 eV, although, after normalising for the carrier generation rates at different
photon energies (determined by the optical absorption coefficients and the incident
photon flux), similar defect generation rates were seen over the energy range
1.2–2.1 eV [52]. The density increased with time, and, for the longer times,
increased with a power time dependence of t1/3, whilst the insert shows that the
dependence on the illumination intensity, I, was I0.6. The process was also very
weakly temperature dependent, with an activation energy of 0.04 eV. The absence
of a linear dependence on time and intensity showed that the defect generation
process was not simply cumulative with illumination time due to direct photo-
dissociation. The data was interpreted as a self-limiting defect generation process,
in which dangling bonds were formed by non-radiative recombination of the
photo-generated electron-hole pairs.
6.5 Other Meta-Stability Effects 179
Fig. 6.27 Time dependent

defect generation due to
optical illumination at
(a) 310 K, and (b) above
90 °C. (Reprinted with
Copyright (1985) by the
For photon energies above band-gap, the photo-generated electrons and holes
were assumed to rapidly thermalise from extended states to shallow band-tail
states before recombining. (This is consistent with photo-luminescence studies,
where the emitted photon energy was *1.2–1.3 eV, and was independent of the
energy of the stimulating photons [52]). For sub-band gap illumination over the
energy range down to 1.2 eV, the carrier excitation was directly from the valence
band-tail states, as in the Urbach edge measurement discussed in Sect. 6.2.1.
However, in all cases, the recombination event leading to defect generation was
assumed to be non-radiative between band-tail states. The dangling bond creation
energy was *1 eV, and, at photon energies greater than *1.2 eV, the subsequent
non-radiative carrier recombination released sufficient phonon energy back to the
lattice to break weak silicon bonds, and form dangling bonds. Once near mid-gap
states, such as the neutral dangling bond, have been created, non-radiative
recombination can proceed through them by sequential electron and hole capture.
The phonon energy released after each carrier capture stage is less than half band
gap energy, and is insufficient to break more weak Si–Si bonds. Hence, the
increasing concentration of dangling bonds provided an increasing number of

alternative, two-stage recombination paths, and this reduced the rate of dangling
bond generation by the direct band-tail to band-tail recombination [52]. This
model contained the essential features of the experimental data, where, in a sim-
plified version, the defect generation rate is proportional to the recombining free
carrier concentrations, and is given by:
dNDB =dt ¼ Anp ð6:61Þ
where NDB is the DB concentration, A is a constant, and n and p are the steady
state electron and hole concentrations produced by the illumination, and are given
by:
n ¼ G=BNDB and p ¼ G=CNDB ð6:62Þ
The electron-hole pair generation rate, G, is proportional to the illumination
intensity, and the denominators are the inverse of recombination lifetime-like
factors, which are proportional to the number of dangling bond recombination
centres (see Chap. 2 for a fuller discussion of the electron-hole pair generation and
recombination processes). Substitution of Eq. 6.62 into 6.61, and the integration of
this equation gives [52]:
N3DB N3DB0 ¼ 3ðA=BCÞG2 t ð6:63Þ

When NDB(t) is more than twice the initial concentration, NDB0, the latter term
can be ignored, and Eq. 6.63 reduces to:
NDB ¼ f3ðA=BCÞg1=3 G2=3 t1=3 ð6:64Þ

Equation 6.64 shows the same dangling bond density dependence on I and t as
displayed in Fig. 6.27a. For temperatures above 90 °C, the defect density deviated
from the t1/3 dependence, and saturated with the temperature dependent density
shown in Fig. 6.27b [52]. Some authors have fitted the saturating curve with a
stretched exponential time dependence [51]. As with the defects formed during
gate bias stress, the light induced defect concentration could be reduced back to its
original value by annealing at elevated temperatures, with a distribution of acti-
vation energies between 0.9 and 1.3 eV [52]. The temperature dependence of the
equilibrium densities, in the insert in Fig. 6.27b, has been attributed to the balance
between defect production and the increased annealing rates at higher temperatures
[51].
The defect generation and annealing features of optically stressed samples share
many qualitative similarities with other metastable defect generation processes in
a-Si:H, and all are thought to involve a hydrogen-mediated weak bond breaking
process. This brief overview of the Staebler–Wronski effect has concentrated on
defect production in intrinsic material near room temperature, and a contemporary
review of further features of this effect can be found in Ref. [53].
6.6 Summary 181
6.6 Summary
As the dominant addressing device in active matrix flat panel displays, a-Si:H
TFTs have been extensively studied, and many of their properties linked to the
defect structure of the basic material. In particular, the electronic properties of
a-Si:H are strongly influenced by the defect states distributed across the mobility
gap. These consist of band-tail states due to weak Si–Si bonds in the disordered
Si network, and dangling bonds, DBs, arising from three-fold coordinated
Si atoms. The presence of high concentrations of hydrogen (*10 %) in the
material reduce the DB density to *1016 cm-3. However, the defect density is
frozen-in at *500 K, during the cooling of the material from its growth tem-
perature, and the density will tend to relax to a new value in response to changes in
external stimuli, such as prolonged exposure to temperatures above room tem-
perature, increases in carrier concentration, and exposure to optical illumination.
The electronic defect state density is described by the density of states, DOS,
distribution, and the inter-relationship between the surface potential and the
resulting surface space charge density was treated in a semi-analytical fashion, by
using an exponential approximation to the DOS. This permitted a direct com-
parison with defect free c-Si, and highlighted the role of the band tail states in
continuously partitioning the induced charge density between free and trapped
electrons, and giving an electron field effect mobility more than ten times smaller
than the band mobility. The other contributor to the space charge density was the
DBs, which determined the TFT threshold voltage.
The position of the DB energy level within the mobility gap is determined by its
charge occupancy during the growth of the material, and the large positive charge
density in the a-SiNx:H gate dielectric layer causes the DB distribution to be
formed in the lower half of the band gap. This minimises its impact upon the sub-
threshold slope of n-channel TFTs, and illustrates why a-SiNx:H is the preferred
gate dielectric for n-channel TFTs; conversely, because of this DB location, it also
gives rise to poor p-channel TFT performance.
The other factors controlling the on-state characteristics, apart from the field
effect mobility and threshold voltage, are series resistance effects due to contact
resistance, the a-Si:H film thickness, and the spreading resistance associated with
the overlap of the gate electrode and the source and drain contacts. For conven-
tional TFT geometries, with a-Si:H film thicknesses less than 200 nm and G-S,D
overlaps of *3 lm, 2-D device simulations showed that the typical specific
contact resistance of 0.25 X cm2 was the major cause of series resistance effects.
Contributory factors to the TFT leakage current characteristics are bulk con-
duction through both the a-Si:H and the a-SiNx:H layers, electron conduction at
the back interface and hole conduction at the front interface. For practical TFTs,
the latter two effects dominate, and, at large drain and large negative gate biases,
field-enhanced electron-hole pair generation rates give exponentially increasing
leakage currents supported by front interface hole conduction.
Gate bias stressing leads to long-term threshold voltage instability in TFTs, due
to carrier trapping in the a-SiNx:H dielectric and DB formation in the a-Si:H layer.
The trapping instability usually dominates with both negative gate bias stress and
with large positive gate bias stress, whilst DB formation is more significant with
smaller positive gate biases. However, the balance between the two is determined
by the quality of the a-SiNx:H layer, and other device processing features. The DB
formation is another aspect of the meta-stability of the a-Si:H defect structure, and
is driven by the increased free carrier concentration in the band tail states. Con-
siderable attention has been given to understanding the bias, time and temperature
dependence of this effect, and it has been similarly attributed to hydrogen-medi-
ated weak bond breaking. In clarifying the instability mechanisms, most attention
has been given to DC gate bias stressing, but, to extend those results to TFT
operation in AMFPDs, the study of pulsed bias stressing has demonstrated that the
negative gate bias instability is very pulse width dependent. This is due to the slow
generation rate of holes in n-channel TFTs. By modifying the DC results to
account for the duty cycle and for pulse width dependences, the effects of active
matrix drive conditions on the TFT stability were well modelled. Combined gate
and drain bias were shown to reduce the magnitude of the gate bias instability,
particularly in saturation, due to the reduction in free carrier density and vertical
field along the channel. Finally, a brief overview of defect instability under optical
illumination, known as the Staebler–Wronski effect, was presented.
References

2. Flewitt AJ, Milne WI (2004) a-Si:H TFT thin film and substrate materials. In: Kuo Y (ed)
Thin film transistors, materials and processes, volume 1: amorphous silicon thin film
transistors. Kluwer Academic Publishers, Norwell
3. Cody GD, Tiedje T, Abeles B, Brooks B, Goldstein Y (1981) Disorder and the optical-
absorption edge of hydrogenated amorphous silicon. Phys Rev Lett 47(20):1480–1483
4. Stutzmann M (1989) The defect density in amorphous silicon. Philos Mag B 60(4):531–546
5. Wehrspohn RB, Deane SC, French ID, Gale IG, Powell MJ, Bruggemann R (1999) Urbach
energy dependence of the stability in amorphous silicon thin-film transistors. Appl Phys Lett
74(22):3374–3376
6. Sherman S, Wagner S, Gottscho RA (1996) Correlation between the valence- and
conduction-band-tail energies in hydrogenated amorphous silicon. Appl Phys Lett
69(21):3242–3244
7. Street RA (2000) Introduction. In: Street RA (ed) Technology and applications of amorphous
silicon. Springer-Verlag, Heidelberg
8. Nickel NH, Jackson WB (1995) Hydrogen-mediated creation and annihilation of strain in
amorphous silicon. Phys Rev B 51(8):4872–4881
9. Powell MJ, Deane SC (1996) Defect-pool model and the hydrogen density of states in
10. Winer K (1990) Defect formation in a-Si:H. Phys Rev B 41(17):12150–12161
11. Street RA, Winer K (1989) Defect equilibria in undoped a-Si:H. Phys Rev B
40(9):6236–6249
References 183
12. Shaw JG, Hack M (1988) An analytical model for calculating trapped charge in amorphous
silicon. J Appl Phys 64(9):4562–4566
13. Shur M, Hack M, Shaw JG (1989) A new analytical model for amorphous silicon thin film
transistors. J Appl Phys 66(7):3371–3380
14. Nathan A, Servati P, Karim KS, Striakhilev D, Sazonov A (2004) Device physics, compact
modelling, and circuit applications of a-Si:H TFTs. In: Kuo Y (ed) Thin film transistors,
materials and processes, volume 1: amorphous silicon thin film transistors. Kluwer Academic
Publishers, Norwell
15. Khakzar K, Lueder EH (1992) Modeling of amorphous-silicon thin-film transistors for circuit
simulations with SPICE. IEEE Trans-ED 39(6):1428–1434
16. Kanicki J, Martin S (2003) Hydrogenated amorphous silicon thin-film transistors. In: Kagan
CR, Andry P (eds) Thin-film transistors. CRC Press, Boca Raton
17. Shur M, Hack M (1984) Physics of amorphous silicon based alloy field-effect transistors.
J Appl Phys 55(10):3831–3842
18. Flewitt AJ (2012) Hydrogenated Amorphous Silicon Thin Film Transistors (a-Si:H TFTs). In:
Chen J, Cranton W, Fihn M (eds) Handbook of visual display technology. Springer-Verlag,
Berlin, p 2012
19. Shur M, Hyun C, Hack M (1986) New high field-effect mobility regimes of amorphous
silicon alloy thin-film transistor operation. J Appl Phys 59(7):2488–2497
20. Shur MS, Jacunski MD, Slade HC, Hack M (1995) Analytical models for amorphous-silicon
and polysilicon thin-film transistors for high-definition-display technology. J SID
3(4):223–236
36(12):2753–2763
22. Valletta A, Bonfiglietti A, Rapisarda M, Mariucci L, Fortunato G, Brotherton SD (2007)
Grain boundary evaluation in sequentially laterally solidified polycrystalline-silicon devices.
J Appl Phys 101(9):094502
23. Deane SC, Clough FJ, Milne WI, Powell MJ (1993) The role of the gate insulator in the
defect pool model for hydrogenated amorphous silicon thin film transistor characteristics.
J Appl Phys 73(6):2895–2901
24. Deane SC, Powell MJ (1993) Field-effect conductance in amorphous silicon thin-film
transistors with a defect pool density of states. J Appl Phys 74(11):6655–6666
25. Deane SC, Powell MJ, Hughes JR, French ID (1990) Thermal bias annealing evidence for the
defect pool in amorphous silicon thin-film transistors. Appl Phys Lett 57(14):1416–1418
26. van Berkel C, Powell MJ (1987) Resolution of amorphous silicon thin-film transistor
instability mechanisms using ambipolar transistors. Appl Phys Lett 51(14):1094–1096
27. Servati P, Striakhilev D, Nathan A (2003) Above-threshold parameter extraction and
modeling for amorphous silicon thin-film transistors. IEEE Trans ED-50(11):2227–2235
28. Chen C-Y, Kanicki J (1998) Origin of series resistances in a-Si:H TFTs. Solid-State Electron
42(5):705–713
29. Powell MJ (1984) Material properties controlling the performance of amorphous silicon thin
film transistors. Mater Res Soc Symp Proc 33:259–273
30. Chen C-Y, Kanicki J (1996) High field-effect-mobility a-Si:H TFT based on high deposition-
rate PECVD materials. IEEE Electron Device Lett 17(9):437–439
31. Tuan HC, Thompson MJ, Johnson NM, Lujan RA (1982) Dual-gate a-Si:H thin film
transistors. IEEE Electron Device Lett 3(12):357–359
32. Servati P, Nathan A, Sazonov A (2000) A physically-based SPICE model for the leakage
current in a-Si:H TFTs accounting for its dependencies on process, geometrical, and bias
conditions. Mater Res Soc Symp Proc 609:A28.3.1–A28.3.6
33. Servati P, Nathan A (2002) Modeling of the reverse characteristics of a-Si:H TFTs. IEEE
Trans ED-49(5):812–819
34. Nathan A, Servati P, Karim KS, Striakhilev D, Sazonov A (2003) Thin film transistor
integration on glass and plastic substrates in amorphous silicon technology. IEE Proc Circuits
Devices Syst 150(4):329–338
35. Yamakawa S, Yabuta S, Ban A, Okamoto M, Katayama M, Ishii Y, Hijikigawa M (1998)

Plasma treatment effect on the off current characteristics of a-Si TFT. SID Symp Digest Tech
Pap 29(1):443–446
36. Vincent G, Chantre A, Bois D (1979) Electric field effect on the thermal emission of traps in
semiconductor junctions. J Appl Phys 50(8):5484–5487
37. Chiang C-S, Kanicki J, Takechi K (1998) Electrical instability of hydrogenated amorphous
silicon thin-film transistors for active-matrix liquid-crystal displays. Jpn J Appl Phys
37(9A):4704–4710
38. Jackson WB, Moyer MD (1987) Creation of near-interface defects in hydrogenated
amorphous silicon–silicon nitride heterojunctions: the role of hydrogen. Phys Rev B
36(11):6217–6220
39. Powell MJ, van Berkel C, French ID, Nicholls DH (1987) Bias dependence of instability
40. Powell MJ, van Berkel C, Hughes JR (1987) Time and temperature dependence of instability
41. Libsch FR, Kanicki J (1993) Bias-stress-induced stretched-exponential time dependence of
charge injection and trapping in amorphous thin-film transistors. Appl Phys Lett
62(11):1286–1288
42. Wehrspohn RB, Powell MJ, Deane SC (2003) Kinetics of defect creation in amorphous
silicon thin film transistors. J Appl Phys 93(9):5780–5788
43. Jackson WB (1990) Role of band-tail carriers in metastable defect formation and annealing in
44. Deane SC, Wehrspohn RB, Powell MJ (1998) Unification of the time and temperature
dependence of dangling-bond-defect creation and removal in amorphous-silicon thin-film
transistors. Phys Rev B 58(19):12625–12628
45. Flewitt AJ, Lin S, Milne WI, Wehrspohn RB, Powell MJ (2006) Mechanisms for defect
creation and removal in hydrogenated and deuterated amorphous silicon studied using thin
film transistors. Mater Res Soc Symp Proc 910:0910–A19-01
46. Kaneko Y, Sasano A, Tsukada T (1991) Characterization of instability in amorphous silicon
thin-film transistors. J Appl Phys 69(10):7301–7305
stability of amorphous silicon TFTs. IEEE Electron Device Lett 25(4):188–190
48. Kuo A, Won TK, Kanicki J (2008) Advanced amorphous silicon thin-film transistors for AM-
OLEDs: electrical performance and stability. IEEE Trans ED-55(7):1621–1629
49. Staebler DL, Wronski CR (1977) Reversible conductivity changes in discharge-produced
amorphous Si. Appl Phys Lett 31:292–294
50. Stutzmann M (1997) Microscopic aspects of the Staebler–Wronski effect. Mater Res Soc
Symp Proc 467:37–48
51. Fritzsche H (2001) Development in understanding and controlling the Staebler–Wronski
effect in a-Si:H. Annu Rev Mater Res 31:47–79
52. Stutzmann M, Jackson WB, Tsai CC (1985) Light-induced metastable defects in
hydrogenated amorphous silicon: a systematic study. Phys Rev B 32(1):23–47
53. Fritzsche H (2010) A new perspective on an old problem: the Staebler–Wronski effect. Mater
Res Soc Symp Proc 1245:A14–01
Chapter 7
Poly-Si TFT Technology and Architecture
Abstract This chapter discusses the fabrication of high quality poly-Si films, by
the industry standard technique of excimer laser crystallisation of a-Si:H precursor
films. Alternative crystallisation procedures are also reviewed, including metal-
induced solid phase crystallisation, as well as advanced procedures for achieving
large grain and high mobility TFTs, using green solid-state lasers in addition to
modified excimer laser techniques. The architecture of poly-Si TFTs is top-gated
with a silicon dioxide gate dielectric, and issues with the implementation of self-
aligned and non-self-aligned architectures are discussed, and illustrative process-
ing schedules are listed. Finally, a simple qualitative cost model is presented,
illustrating why the major commercial application of poly-Si TFTs is in the small-
diagonal, portable display market.
7.1 Introduction
The interest in poly-Si, as an alternative TFT material, started soon after a-Si:H
TFTs became recognised as the most promising technology for the large scale
production of AMLCDs. The low carrier mobility of a-Si:H TFTs (\1 cm2/Vs) is
perfectly adequate for the addressing TFTs within each pixel, but is inadequate for
the faster switching circuits needed for addressing the rows and columns in the
display itself. In contrast to the carrier mobility in a-Si:H, which has remained within
the range 0.5–1.0 cm2/Vs over the last 20 years or so, the electron mobility in poly-
Si has increased, over the same period, from \5 to *120 cm2/Vs with routine
processing, and up to *900 cm2/Vs with innovative techniques yielding quasi-
single-crystal large-grain material [1]. The higher carrier mobility offered by the
more crystallographically-ordered polycrystalline silicon was seen as leading to a
second generation AMLCD technology, with addressing circuitry and other circuit
functions integrated onto the AMLCD plate. Whilst this has been achieved, and

186 7 Poly-Si TFT Technology and Architecture
highly integrated poly-Si displays [2, 3] are now in mass production for small/
medium diagonal, portable active matrix flat panel displays, AMFPDs, a-Si:H
remains the dominant AMLCD technology, particularly for the larger sized note-
book, monitor and TV displays. Hence, the current application of poly-Si is directed
towards small/medium diagonal displays, where it accounts for *36 % of that
segment of the AMLCD market [4]. In addition, it has virtually 100 % of the rapidly
growing small diagonal AMOLED market, particularly for smart phones [5].
The study of poly-crystalline silicon, as a semiconductor device material, has a
long history, and predates the present interest in the low temperature fabrication of
TFTs on glass or flexible substrates. Traditionally, its major application has been
as a heavily doped layer for the gate electrode in MOSFETs [6], because, as a
refractory material, it can withstand high temperature integrated circuit processing.
As MOSFET circuit speeds increased, its conductivity was enhanced by com-
bining it with a refractory metal to form a low resistivity metal-silicide. In addition
to this, poly-Si resistors have also been used in VLSI circuits [7], leading to
detailed studies of conduction in the material [8]. Finally, in the mid-1980s poly-Si
TFTs on quartz substrates were fabricated, using a high temperature VLSI-like
process, for use in projection AMLCDs [9]. Much of this work formed a useful
background to the research and development of low temperature poly-Si TFTs on
glass, which is the subject of this chapter and of Chap. 8. A particular example of
this information infrastructure is the analytical model of conduction in polycrys-
talline silicon [8], which is still used to provide a simple numerical and pictorial
description of carrier flow over grain boundaries. More sophisticated treatments
have now been adopted, but, in view of the continued use of this model, it will be
presented in Chap. 8, together with its analytical extension to carrier flow over
grain boundaries in TFTs [10].
As mentioned above, highly integrated poly-Si AMLCDs are in mass produc-
tion, with the level of integration expected to increase as more circuit functionality
is added to the panel: the objective being a System-on-Glass, SOG [2, 3, 11]. In
order to achieve the higher circuit speeds needed to replace external MOSFET
circuits, the performance of the poly-Si TFTs has been leveraged by way of higher
carrier mobility and shorter channel lengths. Both of which have been achieved
within the constraints of a glass-compatible process.
In Sect. 7.2, the more conventional processing of poly-Si TFTs is presented; in
particular, the formation of poly-Si thin films by laser crystallisation. It will be
shown that the thin film formation technique determines grain size and structure,
and this has a fundamental impact upon carrier mobility and other basic device
parameters, such as threshold voltage and leakage current. These performance
parameters can also be affected by the gate dielectric and the device architecture,
and those topics, together with the process flow for TFT and AMLCD fabrication,
are presented in Sects. 7.3 and 7.4, respectively. Advanced crystallisation pro-
cesses, leading to large grain and high carrier mobility material, are reviewed in
Sect. 7.5. Finally, Sect. 7.6 contains a brief overview of the issues underlying the
application of poly-Si TFTs to AMFPDs, including the factors favouring it for
small/medium sized displays.
7.1 Introduction 187
In summary, this chapter focuses on poly-Si material preparation, and TFT

design and fabrication. The discussion of the performance of these devices is
presented in Chap. 8. Throughout these two chapters, parallels are drawn, where
appropriate, between the physics and technology of poly-Si TFTs and IC MOS-
FETs, from which it will be seen that poly-Si TFTs share more similarities with
these devices than with the other TFT technologies discussed in this book.
7.2 Poly-Si Preparation
As mentioned above, the electron mobility in poly-Si has increased from\5 [12] to
*900 cm2/Vs [1] over the last 20 years or so. These increases are due to several
evolutionary changes in the preparation of the material, which are the subject of this
section and of Sect. 7.5. At the upper end, the mobility is comparable to the low field
value of long-channel SOI MOSFETs [1], although, in poly-Si, these are field effect
mobility values, which are degraded from the band mobility values by the presence
of band gap states near the conduction band edge. The nature and distribution of
these states, together with their impact on carrier mobility is discussed in Chap. 8.
In discussing poly-Si performance, the electron mobility is frequently quoted as
a figure of merit, and this approach is also used in this chapter. However, it should
be appreciated that a large mobility alone is not sufficient for the device to have a
useful application in the display-addressing field: it must also be accompanied by a
low leakage current and a low threshold voltage. Moreover, these parameters need
to be achieved with a glass-compatible process, within which the maximum pro-
cess temperature is, ideally, kept below *450 C. Even this modest temperature
requires a harder glass than that used in the lower temperature a-Si:H process, and
a brief overview of glass plate issues is presented in Sect. 7.2.1. Finally, the overall
complexity of the process needs to borne in mind, as, the more complex the
process, the more expensive the final product will be.
7.2.1 Background
Before proceeding to a detailed discussion of the laser crystallisation process, it is

worth briefly reviewing the preceding technologies from which the laser process
was ultimately identified as the most commercially attractive.
The first generation technology was based upon a well-established LPCVD
process, which was employed in the semiconductor industry for the fabrication of
MOSFET gate layers. This was the direct deposition of a poly-Si layer at
*620 C, but which had a small grain size of *100 nm and yielded an electron
field effect mobility of *5–7 cm2/Vs [12, 13]. This was a demanding process for
glass substrates, as the deposition process approached the substrate’s softening
temperature, and could lead to warping and distortion of the substrate. In addition,
the finished TFTs required exposure to atomic hydrogen, for several hours at
*350 C in an RF plasma, in order to passivate defects in the film. The long
plasma exposure time was because the hydrogen diffused laterally through the gate
oxide before entering the poly-Si. The consequence of this lateral diffusion process
was that TFTs, with increasing channel length, required increasingly long expo-
sure times for the defects in the centre of the channel to be passivated [12, 13].
The direct deposition process was superseded by a second generation process,
using a precursor film of a-Si. This could be deposited either by LPCVD at
550–600 C [13–15], or by PECVD (using an a-Si:H reactor) at 350 C [15], or the
standard, small grain LPCVD films could be amorphised by high dose silicon ion
implantation [16]. In all cases, the objective was to use pre-cursor films, which were
essentially amorphous. These films then underwent a phase transformation into a
polycrystalline form by a process of solid phase crystallisation (SPC) [13–16],
which relied upon a sparse distribution of incipient seeds or random nucleation
within the film, to seed grain growth. For seed-free films, this was a two-stage
process consisting of the growth of stable nuclei, which then seeded grain growth.
Both of these processes are thermally activated, and the overall activation energy of
the SPC process has been found to be within the range 3.1–3.9 eV, depending upon
the film preparation technique [16]. The thermal budget for this process was up to
tens of hours in the temperature range 580–630 C [16], and the resulting films had
large, faulted dendritic grains, up to 1lm long [14, 16]. These larger grains gave
TFTs with electron field effect mobilities up to *40 cm2/Vs [13, 15].
Even more than with the direct deposition process, the longer exposure times to
high temperatures were particularly demanding on the glass substrate, and the
dimensional instability of the glass became an issue. The mechanical properties of
glass are determined by its manufacturing process, and, in particular, its cooling
rate from the melt determines the temperature at which it solidified. Only at this
temperature is the glass in thermal equilibrium, and, at lower temperatures, it will
tend to relax to a new equilibrium state. For the glasses used for TFTs, this will be
such a slow process at room temperature that it can be ignored, and should be
distinguished from reversible dimensional changes governed by its coefficient of
thermal expansion. However, at the temperatures used for SPC, the irreversible
dimensional changes become significant, as shown by the results in Fig. 7.1 [15].
The results in Fig. 7.1a are normalised measurements of shrinkage (usually
referred to as compaction), in a hardened glass substrate, as a function at exposure
time to the range of temperatures used for SPC. It will be seen that the compaction
is a thermally activated process to an equilibrium state, which, as expected, is
temperature dependent. In summary, the higher the annealing temperature, the
smaller was the dimensional change, and the faster it occurred. For photolithog-
raphy, it is essential that the substrate is dimensionally stable through all mask
alignment stages, and the required degree of stability is determined by the mask
alignment tolerances and the size of the substrate. For instance, with a 30 cm
substrate and a 3 lm alignment tolerance, the substrate will need to maintain a
dimensional stability of 10 ppm from the first to the last alignment stage. As seen
from Fig. 7.1a, the overall dimensional stability of glass was far worse than this,
7.2 Poly-Si Preparation 189
(a) 2500 (b) 1000
Compaction rate dc/dt (ppm/hr)

2000 100
Compaction (ppm)
1500 ε (∞)
10
1000
Data: 600°C 1
500 Data: 625°C
Data: 650°C
Modelled curves
0.1
1 10 100 1000 0 20 40 60 80 100
Anneal time (h) Anneal time (h)
Fig. 7.1 a Normalised compaction measurements, on ‘hardened’ glass plates, as a function of

annealing time at 600–650 C (Reprinted from [15] with permission of SID), and b normalised
compaction rate measurements at 625 C
and it was necessary to pre-shrink the glass prior to the first alignment stage.
Furthermore, as the equilibrium compaction was a function of the anneal tem-
perature, the substrate needed to be pre-compacted at the temperature at which the
SPC process was carried out. The easiest way to determine the required pre-
compaction time was from a plot of the differential of the data in Fig. 7.1a, and, as
an example, the compaction rate at 625 C, as a function of the anneal time, is
shown in Fig. 7.1b. Hence, if, for example, a 10 h SPC process was used, then a
compaction rate, prior to processing, of 1 ppm/h was needed to ensure that
\10 ppm compaction occurred during the SPC process, and this required a pre-
compaction anneal of *80 h.
As with the direct deposition process, prolonged exposure to atomic hydrogen
in an RF plasma was required to passivate the defects in the large, but highly
faulted, SPC grains. Nevertheless, demonstration displays, with integrated drive
circuits, were fabricated on glass substrates by this process [13, 15]. However, the
thermal budgets, the issues with glass, and the plasma hydrogenation process did
not make SPC an attractive process for industrialisation. Most of these issues were
resolved with the third generation process, which used excimer laser crystallisa-
tion, and this is now the industry standard process. A more extensive review of the
SPC process can be found in Ref. [17].
The current industry standard crystallisation procedure of excimer laser
annealing, ELA, is discussed in Sect. 7.2.2, and in Sect. 7.2.3 other laser tech-
niques are briefly considered, although a more detailed discussion of them is
contained in Sect. 7.5, which deals with advanced crystallisation techniques for
large grain material. Finally, in Sect. 7.2.4, a variant of the SPC process is pre-
sented, which uses metal silicide-mediated crystallisation to reduce the thermal
budget of the SPC process.
7.2.2 Excimer Laser Crystallisation
7.2.2.1 Introduction
Excimer lasers are gas lasers operating in the ultra-violet wavelength range, from
193 to 351 nm depending upon the gas mixture chosen. For crystallisation of a-Si,
the preferred gas mixture is XeCl, giving a wavelength of 308 nm; similar crys-
tallisation results have been obtained from KrF excimer lasers at 248 nm, but the
308 nm laser is preferred for industrialisation, as the longer wavelength is less
damaging to the optical components in the beam path. These are pulsed lasers,
with a typical pulse duration of *30 ns, a maximum repetition frequency of
600 Hz, and can deliver up to 0.9 J/pulse [18]. The raw pulse shape is semi-
Gaussian, with dimensions of *1 cm 9 1 cm, but, for crystallisation, a pencil
shaped beam is preferred, and beam shaping optics are used to produce a highly
elongated line-beam, whose dimensions can be up to 465 mm in the long axis and
down to 350 lm for the short axis [18]. The steep edges in the short axis profile
have led to the beam shape being referred to as a ‘top-hat’ beam, and all the line-
beam irradiations discussed in this section will be assumed to have this shape,
unless specified otherwise. Measurements of an industrial line-beam profile are
shown in Fig. 7.2a, and a schematic illustration of an ELA crystallisation system is
shown in Fig. 7.2b, where the key components are an attenuator for controlling
beam intensity, the homogeniser and beam shaper to produce the line-beam, and a
condensing lens to focus the beam on the underlying plate. The plate is
mechanically swept through the short axis of the beam at a rate, which delivers a
multi-shot process, of typically 10–30 shots per point for commercial processing,
so that the plate translation distance between shots is typically in the range
12–50 lm for short axis lengths of 350–500 lm.
7.2.2.2 Crystallisation Process
At 248 nm and 308 nm wavelengths, the optical absorption depth in a-Si is 5.7 nm
and 7.6 nm, respectively, so that the incident energy is strongly absorbed in the
silicon film, resulting in intense heating. If the incident energy density is high
enough, this will heat the film to its melting point, Tm, of 1420 K, where the
optical and thermal constants for a-Si [19] and c-Si [20] are shown in Table 7.1.
The required energy density, Eth, to cause surface melting can be calculated from
the solution of the heat diffusion equation, which yields the following approximate
analytical solution [21] (provided the optical absorption depth of the film, 1/a, is
less than the thermal diffusion length, H(Ds), where D = k/qCp):
pffiffiffi pffiffiffiffiffiffi
ðTm T0 Þ pqCp Ds HDz
Eth ¼ þ ð7:1Þ
2ð1 RÞ qð1 RÞ
Fig. 7.2 a Laser intensity profiles from a 464 mm 9 340 lm line-beam excimer laser annealing,
ELA, system (data supplied by Coherent GmbH), b schematic illustration of an ELA system
The first term in Eq. 7.1 is the intensity to bring the surface of the film up to the
melting temperature, and the second term accounts for the latent heat required to
melt a thin film of thickness Dz. (T0 is initial film temperature, q is the density of
a-Si, Cp is its specific heat, D its diffusion coefficient, s is the heating time, H is the
latent heat of melting, and R is reflectivity).
Evaluation of Eq. 7.1, using the optical and thermal coefficients listed in
Table 7.1, predicts a melt threshold energy density of *75 mJ/cm2 for a-Si, which
agrees well both with full numerical simulations giving 70–90 mJ/cm2 [22], and
experimental data in the range 70–100 mJ/cm2 [23, 24]. The thermal diffusion
length in a-Si, for a 30 ns pulse, is 120 nm, so that we can readily expect films up
192
Table 7.1 Optical and thermal constants for amorphous, crystalline, and liquid Si [19, 20, 22]
Material 1/a R 1/a Tm q Cp k H
(at 308 nm)(nm) (at 308 nm) (at 248 nm)(nm) (K) (gcm-3) Jg-1K-1 (Wcm-1K-1) (Jg-1)
a-Si 7.6 0.57 5.7 1418 2.26 1.1 1.3 9 10-2 1282
c-Si 6.9 0.66 5.5 1687 2.31 1.0 0.23 1800
Liquid Si 6.9 0.69 6.4 2.52 1.1 0.51
7 Poly-Si TFT Technology and Architecture
Fig. 7.3 Schematic illustration of a-Si melting regimes during excimer laser irradiation, and the
resulting poly-Si grain structure
to at least this thickness to be melted at high enough incident energy densities.

Indeed, experimental data have shown the melt depth increasing linearly with
energy density up to 145 nm [25], and simulations have predicted the same
dependence for melt depths up to at least 300 nm [26].
In fact, current understanding of the crystallisation mechanism of a-Si, has
shown that the melt depth of the film is a crucial factor in determining the outcome
of the process [27–30]. This is shown schematically in Fig. 7.3, where three
melting regimes are identified. Following partial film melting, the crystallised film
has a vertically stratified appearance, as shown in Fig. 7.4a, with mid-sized grains
(*100 nm) within the previously melted region, and either a fine grain or
amorphous region beneath it. Fig. 7.3b shows the film almost fully melted, with
small solid islands remaining at the back of the film, seeding the growth of large
(*300 nm) high quality grains, as the film cools. These columnar grains extend
through the entire thickness of the film, which no longer shows vertical stratifi-
cation. This process was identified by Im et al. [27, 28], and given the name super-
lateral growth (SLG). As will be shown below, this is the crystallisation regime,
which yields high quality TFTs. Finally, for the fully melted film in Fig. 7.3c, the
seeding centres responsible for the SLG growth are lost, and crystallisation of the
film relies upon random nucleation in a super-cooled melt, resulting in a fine grain
film [27].
From the above scenario, it is apparent that the optimum energy range is that
which results in the SLG regime: energy densities above and below this result in
smaller grain poly-Si. However, as shown by the grain size results in Fig. 7.4b, the
SLG regime occurred over a very limited energy density range of *45 mJ/cm2
Fig. 7.4 a Cross-sectional TEM micrograph of a partially melted 150 nm thick a-Si film,
showing the resulting vertically stratified grain structure (Reprinted from [23] with permission of
IEEE), and b variation of average grain radius with laser energy density for excimer laser
crystallized a-Si films (100 nm thick films capped with 50 nm SiO2). (Reprinted with permission
from [27]. Copyright (1993) American Institute of Physics)
from *195 to *240 mJ/cm2 [27]. These particular results were from single shot
irradiations [27], and qualitatively identical results were obtained from multi-shot
irradiations [30], apart from a growth in SLG grain size with increasing shot
number. As discussed below, the limited size of the SLG window has major
implications for the implementation of the crystallisation process to produce high
performance TFTs.
The essential features of the above process are fully demonstrated in the corre-
lation of TFT behaviour with the energy density used to crystallise the film. This can
be most easily seen with static irradiations of a 40 nm thick a-Si film, using the raw,
semi-Gaussian excimer laser beam, as shown in Fig. 7.5a–c [25]. It should be
emphasised that this mode of irradiation is not the conventional TFT crystallisation
procedure (which involves swept, line-beam irradiations), but was used purely as an
experimental tool. Figure 7.5a shows a profile of electron mobility measurements
made on a line of non-self-aligned TFTs, with a W/L ratio of 50/6 lm, and with a
spatial pitch of 220 lm. The profile through the approximately Gaussian distribution
Fig. 7.5 Electron mobility as a function of position in stationary KrF excimer laser beam,
measured after irradiations at the following peak intensities: a 273 mJ/cm2, b 330 mJ/cm2,
c 330 mJ/cm2 followed by 273 mJ/cm2. (Reproduced with permission from [34])
of energy densities within the beam facilitates a precise mapping of electron

mobility (and other device parameters) against incident energy density [25]. The
maximum beam intensity had been set to *270 mJ/cm2, which was a value in the
SLG regime, and a progressive increase in mobility from zero to *200 cm2/Vs is
seen as the crystallisation energy increased from zero to its maximum. Also of note
are the rapid changes in mobility at the x-axis locations of 0.3 and 0.8 cm, which
corresponded to the positions within the beam at which the threshold energy density
occurred for the SLG regime (*225 mJ/cm2). In Fig. 7.5b, the maximum energy
density was increased to 330 mJ/cm2, which was large enough to induce full melting
of the film over the central portion of the beam. The ensuing fine grain material [27]
resulted in a reduction of electron mobility, within the centre of the irradiated region,
from *200 to *25 cm2/Vs, whilst the edge regions, irradiated at lower energy
densities, retained the appearance seen in Fig. 7.5a. The abrupt decrease in mobility
gave an SLG window size of *45 mJ/cm2, which is much the same as the energy
window in the TEM data of Im et al. [27]. Finally, Fig. 7.5c shows the result of re-
irradiating the material from Fig. 7.5b using lower intensity conditions, which
corresponded to the SLG regime. This converted the previously fine grain material
back into larger grain SLG material, with a consequent recovery in electron
mobility. Hence, the material can be cycled in and out of the SLG regime, depending
upon the final energy density used to crystallise the film. This has particular
implications for the uniformity and process window size of the conventional swept
beam process, as discussed in the following section.
7.2.2.3 TFT Crystallisation
An overview of TFT results from swept beam processing is shown in Fig. 7.6 [25,
31]. This figure shows the dependence of electron mobility, in n-channel TFTs, on
the irradiation energy density, with the a-Si precursor film thickness as an inde-
pendent parameter. Figure 7.6a contains results from LPCVD precursor a-Si,
crystallised by a 248 nm KrF semi-gaussian beam, and the data in Fig. 7.6b are
from the more commonly used PECVD pre-cursor a-Si:H, crystallised by a
308 nm XeCl line beam. The two sets of curves demonstrate the same essential
Fig. 7.6 Electron mobility as

a function of excimer laser
energy density for a-Si
precursor films of different
thickness (a) 248 nm (KrF)
irradiation of LPCVD a-Si,
(b) 308 nm (XeCl) irradiation
of PECVD a-Si:H. (Reprinted
SID)
Fig. 7.7 Cross sectional TEM melt depth (nm)

TEM measurements of the 0 40 80 120 160
melt depth in a 150 nm thick 600 600
film with crystallization
ET v thickness
energy density (+ symbol, top
500 Melt depth v Epk 500
and right axes). Variation of
the threshold laser energy
ET (mJ/cm2 /pulse)
Epk (mJ/cm2 /pulse)

density for high mobility 400 400
TFTs as a function of film
thickness (• symbol, left and
300 300
bottom axes) (from the TFT
measurements in Fig. 7.6a).
(Reprinted with permission 200 200
Physics) 100 100
0 0
0 40 80 120 160
Film thickness (nm)
features of the crystallisation process, which were independent of the laser

wavelength, the type of precursor material, and the beam shape details. The two
most obvious features in these curves are, firstly, that the outcome of the crys-
tallisation process was a strong function of film thickness. In fact, the energy
densities required to achieve maximum carrier mobility scaled approximately
linearly with film thickness, as shown in Fig. 7.7 [25]. Secondly, the crystallisation
process was not a simple monotonic function of energy density. The results in
Fig. 7.6 only show electron mobility, and the other key device parameters of sub-
threshold slope and threshold voltage showed a corresponding dependence on
crystallisation energy density, and qualitatively identical results were also obtained
from p-channel TFTs [25]. Figure 7.8 shows the variation of leakage current with
laser intensity, and this displays more complex behaviour than the mobility results,
insofar as the leakage current initially increased and then decreased. From the
dependence of leakage current on channel length, the initial increase in leakage
current was ascribed to a resistive, drift current, which scaled with increasing
mobility. The current at higher energy density was a generation current, and this
decreased as the improving material quality increased the electron–hole pair
generation lifetime [25]. The detailed analysis of this data is discussed in Chap. 8.
Figures 7.6 and 7.8 have focussed on specific parameter values, and Fig. 7.9
shows the overall transfer characteristics of high quality, non-self-aligned n- and
p-channel TFTs obtained from this process. They illustrate the attainment of low
leakage currents, small sub-threshold slopes, as well as high on-currents.
The crystallisation regimes in Fig. 7.6 can be broken down into four phases,
which are most clearly seen in the thicker films, but can also be discerned by points
of inflection in the thinner films. The four phases were an initial increase in
mobility (I), a plateau region (II), a second rise in mobility (III), and, finally,
Fig. 7.8 TFT leakage 10–11

current as a function of
Transit time
excimer laser energy density limited #539/2
for a-Si precursor films of Generation VD = 5V
different thickness limited W = 50
crystallized by swept beam L 6
10–12
248 nm (KrF) irradiation.
(A/µm)
Joff
Physics)
10–13
29
40 tsi (nm)
80
145
10–14
100 200 300 400 500 600
Epk(mJ/cm2/pulse)
Fig. 7.9 Transfer

characteristics of n- and
p-channel excimer laser
crystallized non-self-aligned
TFTs. (Reprinted with
Copyright (1998), The
Electrochemical Society)
a decrease in mobility (IV). The initial increase in mobility occurred above the
melt threshold energy, as the surface of the film was converted into a poly-
crystalline form. When the thickness of this region was less than the equilibrium
band bending thickness at inversion, the band bending extended into the under-
lying fine grain/amorphous silicon, as shown schematically in Fig. 7.10a (and this
vertical stratification is clearly visible in the TEM micrograph in Fig. 7.4a). As
discussed in Sect. 6.2.4, when there is a high trap density within the band bending
region, the partition of induced carriers between trapped and free states results in a
diminished value of field effect mobility. Moreover, the fewer trapping states there
Fig. 7.10 Schematic illustration of band bending diagrams superimposed upon the grain
structure regime in the laser crystallized films (a) regime I, (b) regime II, and (c) regime III. (For
simplicity, constant values of gate bias and band bending are shown in each diagram)
are in this region, the higher the field effect mobility will be. Figure 7.7 shows that
the melt depth increased with increasing crystallisation energy, and, as the
thickness of this crystallised region increased, the electron field effect mobility
similarly increased (due to the reduced number of carriers going into the fine grain,
high trap density material at the bottom of the space charge region). This mobility
increase continued until the band bending was contained entirely within the
crystallised region, as shown schematically in Fig. 7.10b. When this occurred,
regime II was entered, and further increases in crystallisation depth had a minimal
effect upon the carrier mobility. The regime II plateau was resolved most unam-
biguously in the thicker films in Fig. 7.6, and Fig. 7.11a shows a TEM micrograph
of the mid-sized grain structure in the upper portion of the partially melted film.
Figure 7.6a shows that the threshold energy density for regime II with the 145 nm
film was *300 mJ/cm2, and, from Fig. 7.7, the melt depth at this energy density
was *50 nm. Hence, the depletion depth for surface inversion was comparable to
50 nm at this energy density, and illustrates why regime II was also seen in the
80 nm film, but was not resolved in the 40 nm film.
Ultimately, with increasing incident energy density, the melt depth increased
until the film was almost fully melted, and the SLG regime was initiated. This is
depicted in the band bending diagram shown in Fig. 7.10c, and the SLG grain
structure is shown in Fig. 7.11b. The increase in grain size and quality, compared
with regime II grains (Fig. 7.11a), was responsible for the abrupt increase in
electron mobility in regime III, and Figs. 7.6 and 7.7 demonstrate how the energy
density for this regime scaled with film thickness [25]. Hence, to obtain maximum
carrier mobility, the crystallisation energy density must be matched to the film
thickness.
Finally, at the highest energy densities, the electron mobility started to decrease
in regime IV, due to the incipient formation of fine grain material. However, the
Fig. 7.11 Plan-view TEM micrographs showing the grain structure in ELA poly-Si, crystallised
in (a) regime II (partial melting) and (b) regime III (SLG). (Reprinted with permission from [25].
Copyright (1997) American Institute of Physics)
decrease in mobility was much smaller than seen with the stationary beam process
of Fig. 7.5b. This was because, in the swept beam process, subsequent lower
intensity pulses, in the trailing edge of the beam, can reset the material back into
the SLG regime. As discussed in the next section, the resetting of the material back
into the SLG regime is a function of beam shape and shot number, and has
implications for the overall trade-off between plate throughput and device
parameters.
7.2.2.4 ELA Process Control Issues
As shown by the results in Fig. 7.6, to obtain high carrier mobility devices, the film
needs to be crystallised within the SLG regime, and, ideally, at the energy density,
Em, giving the maximum mobility. But, there is limited accuracy in the precise
setting of the laser energy density, and, moreover, the pulse-to-pulse fluctuations
in energy density [32] mean that samples will occasionally be exposed to higher
intensity irradiations, and, where this causes full melt-through, can lead to a
consequent degradation in device parameters. Comparison of Figs. 7.5b and 7.6b
shows that the magnitude of this degradation is determined by the opportunity,
within the process, to re-irradiate the fine grain material, and to convert it back to
large grain SLG material. The static irradiations in Fig. 7.5b can be regarded as a
zero pulse overlap process (giving gross mobility non-uniformity when full melt-
through occurred), whereas the 100-shot, swept beam process in Fig. 7.6b can be
regarded as a 99 % pulse overlap process (giving greatly improved uniformity,
even after full melt-through). This latter point can be understood by recognising
that when an anomalously high intensity pulse has fully melted the stripe of
material exposed to it, the next pulse will overlap most of this poorly crystallised
region, and the fine grain material will be re-melted, and converted back to the
large grain SLG material. Nevertheless, there will be a thin strip of material (equal
to the plate translation distance), which will not be fully overlapped by the top-hat
region of the next pulse, but, at best, by its trailing edge. The recovery of this thin
strip will then depend upon the relative size of the plate translation step, Dx, and
the spatial width of the SLG window on the trailing edge of the beam, XSLG, as
illustrated in Fig. 7.12a [33]. Only if the SLG width on the beam edge, XSLG, is
greater than the translation step, Dx, will good recovery occur. (Where Dx = W/N,
and W is the beam width, and N is the number of shots in the process).
Hence, the practical issue of plate processing is the trade-off between higher
plate throughput (and reduced pulse overlap) and acceptable uniformity within a
realistic process window.
This trade-off can be best illustrated within a quantified framework: for example,
an acceptable laser process window, DE, could be defined as one within which the
average carrier mobility, lav, varies by \±10 %, or within which the maximum
scatter in mobility (lmax–lmin/lav) is\±10 %. (The smaller of these two windows
Fig. 7.12 Schematic

illustrations of excimer line-
beam pulse shape, showing
the width of the SLG
recovery regions, XSLG, with:
a top hat pulse, and b ramped
pulse (where ESLG is the
threshold energy for the SLG
regime, and Em is full melt-
through energy). (Reprinted

electron mobility values, and
the maximum scatter in those
values, for 100-shot and 10-
shot top-hat ELA line-beam
processing. (Reprinted with
would be the relevant one.) Given that the fundamental issue is the tolerance of the
process to random fluctuations in laser intensity, which cause full melt-through, this
can be examined by deliberately irradiating the sample into this regime, and
examining the size of the process window, DE, as a function of pulse overlap. This
is illustrated in Fig. 7.13, in which crystallisation energy densities above the full-
melt threshold are used to compare a 100-shot process (99 % overlap) with a 10-
shot process (90 % overlap). The 100-shot process yielded an average mobility of
145 ± 15 cm2/Vs, within a broad process window, DE, of 53 mJ/cm2, due to good
recovery of the full-melt material by the trailing edge of the beam [33, 34]. In
addition, the mobility scatter remained well below ± 10 % over this energy range.
The 10-shot process resulted in a lower average mobility of 90 ± 10 cm2/Vs (due
to a smaller SLG grain size [30, 31]), and a much stronger fall-off in average
electron mobility at the higher energies, giving a significantly smaller process
window, DE, of 20 mJ/cm2. In both cases, this window was centred about the
optimum intensity, Eopt, of 310 mJ/cm2, which gave the maximum mobility.
Hence, for the 10-shot process, the size of the energy window was just ± 3.3 % of
the target intensity, Em, so that random variations in pulse intensity greater than this
will cause unacceptable uniformity variations. Associated with the lower average
mobility at higher laser intensities, in the 10-shot sample, was also a sharp increase
in mobility scatter once the full melt through regime was triggered. This was due to
incomplete and variable recovery of the fine grain material by the trailing edge of
the pulse, and was consistent with SEM observations of a periodic grain size
variation, from small grain to large grain, at the pitch of the plate movement
through the laser beam [33]. One way to improve the trailing edge recovery process
is to simply broaden the trailing edge; however, the energy within the pulse can be
used more efficiently by ramping the top of the beam instead, as illustrated in
Fig. 7.12b. With this beam shape, the size of the energy process window can be
increased by a factor of *2 for a 10-shot process [33].
In view of the uniformity issues with a high throughput and low shot number
process, typical plate processing uses a 20-shot process, yielding an electron
mobility of *120 cm2/Vs, within an energy window of *30–40 mJ/cm2. How-
ever, the major contributory factor to improved control and uniformity of the ELA
process for poly-Si AMLCDs has been improvements in peak-to-peak pulse
energy stability [18, 32] to 3 % over ± 3 sigma. Moreover, when this is combined
with a technique to randomise intensity variations along the beam length, the more
stringent uniformity requirements for AMOLED displays may also be met [18].
7.2.3 Other Laser Techniques
Whilst excimer laser annealing is the currently preferred technique for the com-
mercial crystallisation of poly-Si, other lasers have been examined as alternatives
in order to address some of the ELA issues themselves. These include pulse-to-
pulse stability, and the cost of ownership, which includes frequent gas refills,
window cleaning, and general downtime for tube and system maintenance. In
particular, diode pumped solid state lasers have been advocated to address both the
pulse stability and cost of ownership issues, with neodymium-doped yttrium
aluminium garnet, Nd:YAG, lasers being the most commonly studied [35–37].
These can be operated either in continuous wave, CW, mode, or in Q-switched
mode, delivering short duration pulses (C10 ns) of 800 mJ/cm2, with a repetition
frequency of 4 kHz and an output power of 200 W at 532 nm [35]. The funda-
mental output wavelength of these lasers is 1064 nm, which is in the infrared, and
radiation at this wavelength would not be efficiently absorbed in thin silicon films.
Hence, for crystallisation of a-Si, the Nd:YAG lasers are usually operated in the
frequency-doubled mode at a wavelength of 532 nm. This is in the visible radi-
ation band, and these Nd:YAG-based systems are frequently referred to as green
laser annealing systems. For CW use, although Nd:YAG lasers have been used
[38], Nd:YVO4 lasers are preferred for high power applications. These are also
frequency doubled, and can emit 10 W of radiation at 532 nm [39–42].
Both the pulsed and CW approaches can produce material with comparable
properties to the ELA process, but they are also able to produce large grain, high
quality material with electron mobilities up to 250 cm2/Vs [36] and 566 cm2/Vs
[39], respectively. In view of the large grain options with these solid state lasers, it
is more appropriate to discuss them in detail in Sect. 7.5, where they can be
compared with the large grain processing procedures available with modified
excimer laser crystallisation.
7.2.4 Metal Induced Crystallisation
Metal induced crystallisation (MIC) is a solid phase crystallisation (SPC) process,

in which the presence of the metal can enhance the crystallisation rate of a-Si into
poly-Si. This process has been viewed as a potentially simpler, cheaper and more
uniform alternative to ELA. Equally, by accelerating the SPC process, so that it
could be implemented at lower temperatures and with shorter crystallisation times,
it was seen as a more production-worthy alternative to conventional SPC. How-
ever, the resulting TFTs have had persistently high leakage currents, which has
limited the commercial application of the technique, although these can be min-
imised by combining it with ELA.
A large number of different metals have been found to promote MIC [43], some
of which act in the elemental state, such as In, whilst others, including several
transition metals, as well as Pd and Pt, act via the metal silicide. This latter process
is sometimes referred to as silicide-mediated crystallisation (SMC). For the
application of MIC to poly-Si TFTs, the metal of greatest interest has been Ni,
because one of its silicides is a very good lattice match to crystalline Si. The role
of Ni in the enhanced crystallisation rate of a-Si is discussed in Sect. 7.2.4.1 and
the use of this procedure to fabricate poly-Si TFTs is reviewed in Sect. 7.2.4.2
7.2.4.1 Ni Mediated Crystallisation of a-Si
When thin films of Ni, or high concentrations of Ni are introduced into c-Si, a
number of different nickel silicides readily form upon subsequent low temperature
annealing. At *200 C, Ni2Si is formed, this is converted into NiSi at *240–
390 C, and the thermodynamically favoured final phase of nickel disilicide, NiSi2,
is formed at *325–480 C [43, 44]. NiSi2 is a good lattice match to Si, having the
same cubic lattice, and a lattice constant of 5.406 Å, which is within 0.4 % of the Si
lattice constant of 5.430 Å. In Ni-doped a-Si films, crystallisation proceeded via a
three-stage process of NiSi2 precipitate formation, the nucleation of Si on the
precipitates, and the subsequent migration of the precipitates, leaving trails of
crystallised Si behind them [44]. In particular, regular octahedral precipitates of
NiSi2 formed at *400 C, and the good lattice match of NiSi2 to Si made it an
effective seed for the crystallisation of a-Si from the {111} faces of the precipitates
[44]. Using 95 nm thick a-Si films implanted with Ni, both in situ TEM observation
of crystallisation at 660 C, and TEM examination of 500 C furnace crystallised
samples showed that the crystallisation process produced needle-like grains of Si.
These grains propagated in the \ 111 [ directions from the {111} precipitate
faces, and were preferentially orientated in the \ 110 [ direction to the sample
surface. This was the precipitate orientation, which permitted maximum growth
within the plane of the film [44]. A TEM micrograph of the needle-like grains is
shown in Fig. 7.14 [45].
Fig. 7.14 Plan-view TEM

micrograph of needle-like
grains in NiSi2-mediated
crystallisation of a-Si,
induced by 20 h annealing at
500 C. (Reprinted with
The grain propagation took place by the movement of the precipitate through
the a-Si, which consumed a-Si at its head and left the thin c-Si grain in its wake
[44]. This process is shown schematically in Fig. 7.15a-c, in which the precipitate
migrates towards the right, in the direction of the arrow. The overall process was
driven by the reduction in free energy due to the conversion of a-Si to c-Si, which
occurred with the nucleation of c-Si on one face of the NiSi2 precipitate. In
addition, the chemical potential of Ni was lower at the NiSi2/a-Si interface than at
the NiSi2/c-Si interface, and the opposite was true of the Si atoms, i.e. the Si
chemical potential was lowest at the NiSi2/c-Si interface. It was proposed that the
crystallisation process was driven by one of two alternative mechanisms [44]. In
the first model, the NiSi2 dissociated at the NiSi2/c-Si interface, with the Si being
incorporated into the growing grain, and the Ni diffusing down the chemical
potential gradient to the head of the precipitate. On reaching the NiSi2/a-Si
interface, the Ni reacted with a-Si to form NiSi2, and propagated the grain forward.
This was referred to as the dissociative Ni diffusion process. The alternative
mechanism was the non-dissociative diffusion of Si directly from the NiSi2/a-Si
interface to the NiSi2/c-Si interface to sustain the migrating-precipitate/crystalli-
sation process. In both cases, the rate limiting process was atomic diffusion within
the precipitate, and, although the specific process was not positively identified, it
was tentatively ascribed to dissociative Ni diffusion [44].
Given the controlling role played by Ni (or Si) diffusion through the precipitate,
faster crystallisation rates were found with thinner precipitates, and crystallite
Fig. 7.15 Schematic (a)

representation of NiSi2- a-Si
mediated growth of c-Si Nucleation of
grains. (Reprinted with NiSi2 NiSi2
c-Si on NiSi2
Institute of Physics) c-Si
(b) Si a-Si
Rejection of excess Si
NiSi2 c-Si NiSi2
to c-Si/NiSi2 interface
c-Si
(c) Si a-Si
Growth of
c-Si NiSi2 c-Si NiSi2
epitaxial Si
Ni Growth
growth rates of 5 Å/s were measured at *507 C for a 50 Å thick precipitate.

Progressive crystallisation of large areas occurred due to the migration, and ulti-
mately, intersection of numerous crystalline needles from a multiplicity of seeds,
and the complete crystallisation of a film doped with a dose of 5 9 1015 Ni/cm2
occurred within 5 min at 569 C [44]. However, the local crystallisation process
was essentially one-dimensional along the extending length of the needle-like
grains, and with much slower lateral growth of the crystallite width, driven by SPC
alone. A UV reflectance study of the change in crystallinity of a Ni doped, 40 nm
thick a-Si film, following annealing at 550 C, showed a two-stage process, in
which *85 % crystallinity was rapidly achieved in *4–5 h by the SMC process,
but a further 100 h was needed to achieve 99 % crystallinity [46]. This slow,
second stage of the process was attributed to the solid phase crystallisation of
residual a-Si regions, which had remained between the needle-like grains. The
activation energy of this latter process was 3.0 eV, which is consistent with the
SPC process [16]. In contrast to the high activation energy for the SPC process, the
activation energy for the Ni MIC process has been quoted to be 1.5–1.75 eV [43].
A variety of techniques have been used to introduce Ni into a-Si films, and
these have included thin metal film deposition, co-sputtering of Ni and Si, ion
implantation of nickel [46], and spin-coating of a nickel-containing pre-cursor
solution [47], or localised ink-jet printing of Ni containing droplets [48]. By and
large, these samples were then used for low temperature SMC over the tempera-
ture range 500–600 C. However, one feature consistently seen in MIC TFTs was
a high leakage current, which was initially attributed to Ni contamination of the
poly-Si. The qualitative relationship between poly-Si quality and Ni contamination
Fig. 7.16 Illustration of the high good

relationship between Ni areal Quality:
Crystallization Temperature
density, MIC crystallisation Grain size
temperature and the resulting Defect density
Quality of Poly-Si
poly-Si quality. (Reprinted Contamination
Springer SBM)
low bad
12 2
10 atoms/cm 1015 atoms/cm2
Ni Area Density
Fig. 7.17 MILC of a-Si

using ink-jet printed Ni dots
(a) partial inter-dot
crystallisation after 2 h at
580 C, and (b) full
crystallisation after 6 h at
580 C. (Reprinted with
is summarised by the plot in Fig. 7.16, which also illustrates the dependence of the
crystallisation temperature on the concentration of nickel [43].
In view of the issue of Ni contamination in MIC material, there has been con-
siderable interest in metal induced lateral crystallisation, MILC, in which the NiSi2
precipitates migrate from localised Ni-doped areas into the adjacent, undoped
material and crystallise this [47–51]. Optical photographs of this process are shown
in Fig. 7.17a and b, in which a matrix of 80 lm diameter nickel nitrate dots had
been ink-jet printed onto an a-Si film and heated at 580 C for 2 and 6 h, respec-
tively [48]. The lighter area around the Ni dots in Fig. 7.17a shows the extent of the
lateral crystallisation beyond the dots after 2 h, and complete crystallisation after
6 h is shown in Fig. 7.17b. The two situations are illustrated schematically in the
accompanying line drawings. Although lateral crystallisation can be used to reduce
the level of direct Ni contamination in the poly-Si, it does not eliminate it com-
pletely. SIMS measurements have shown a 100-times reduction of the Ni con-
centrations in MILC areas compared with the MIC areas, but only a 10-times
reduction in the precipitate areas at the head of the crystallisation front [50].
7.2.4.2 SMC Poly-Si TFTs
A survey of MIC and MILC poly-Si TFTs, fabricated within the temperature range
500–600 C, showed that the majority of devices had electron field effect mobilities
between *70 and *100 cm2/Vs [43], which is higher than achieved with con-
ventional SPC, but lower than with ELA. Also, the leakage currents of the SMC TFTs
were higher than in good quality ELA material, typically being within the range of
1–100 9 10-12 A/lm (of channel width) for SMC TFTs, but *2 9 10-14 A/lm
for ELA TFTs (when measured at 5–10 V drain bias) [52]. However, direct
comparisons between different publications are not straightforward due to the use of
different measurement biases and different gate oxide thicknesses. Both of these
parameters determine the drain field, and, as discussed in Chap. 8, the TFT leakage
current, at high drain fields, is exponentially dependent upon the electrostatic field
at the drain. Hence, the most useful comparisons are those made in the same
publication. Nevertheless, the figure of *2 9 10-14 A/lm quoted for ELA TFTs
makes a good reference point when assessing SMC TFTs. The role of Ni contami-
nation was highlighted when comparing MIC and MILC TFTs, with values of
8.8 9 10-12 A/lm and 3.6 9 10-13 A/lm, respectively, measured at 1 V drain bias
[49]. Although the MILC device had *20 times lower leakage current than the MIC
device, it was still more than 10 times higher than an ELA device. However, for Ni
concentrations below *2.5 9 1019 cm-3, in MIC devices, the leakage current of
*2 9 10-12 A/lm was independent of the Ni doping level [46], suggesting
contributory factors, other than metal contamination, to the leakage current,
including the crystal structure.
Other authors have clarified the role of the local grain and grain boundary
structures in MILC TFTs, where a continuous grain boundary has been identified
at the MIC/MILC boundary, as shown in Fig. 7.18a, and a further major grain
boundary has been identified when two MILC fronts meet [51]. Where these
boundaries are perpendicular to current flow, and are located in the channel or at
the source or drain junctions, they can be expected to degrade device behaviour.
There are, thus, a variety of device configurations, with respect to these grain
boundaries, determined by their relative locations, as shown in Fig. 7.18b. The
device location represented by ‘111’ encompasses all three boundaries (where the
MIC regions have also been used for the source and drain regions), and ‘000’
encompasses just the MILC material itself. Figure 7.19 shows the device struc-
tures used to engineer these two situations, where, in Fig. 7.19a, the Ni seeding
region was offset from the ultimate location of the source junction, and MILC
growth was unidirectional in the channel area. In Fig. 7.19b, the MIC/MILC
junctions were coincident with the source and drain junctions, and the bi-direc-
tional MILC fronts met in the centre of the channel. The resulting n-channel TFT
characteristics are shown in Fig. 7.20, and, interestingly, the mid-channel con-
tinuous GB had only a minor effect upon the electron mobility, by reducing it from
78 to 70 cm2/Vs, whereas the minimum leakage current, at Vd = 5 V, for the
‘111’ TFT was 3 9 10-11 A/lm compared with 2 9 10-12 A/lm for the ‘000’
TFT [51]. Hence, the continuous GBs at the junction boundaries had a strong
Fig. 7.18 a Plan-view TEM micrograph showing the continuous grain boundary at the MIC/
MILC boundary, and b optical micrograph of MIC and MILC regions, where the numbered areas
show the device locations within the material. (Reprinted from [51] with permission of IEEE)
Fig. 7.19 Illustration of device and grain boundary locations within (a) a unilateral MILC TFT
and (b) a MIC/MILC TFT. (Reprinted from [51] with permission of IEEE)
Fig. 7.20 Transfer

characteristics obtained from
unilateral MILC (MIUC) and
MIC/MILC TFTs (see
Fig. 7.19a and b,
respectively). (Reprinted
IEEE)
degrading effect upon the leakage current, but, even in their absence, the leakage
current still remained high in the ‘000’ device compared with ELA TFTs.
This again points to fundamental aspects of the grain structure, and not just Ni
contamination or continuous grain boundaries, being responsible for high leakage
currents in SMC TFTs. In confirmation of this, the only SMC devices with sub-
stantially lowered leakage currents were those in which the crystal structure was
changed by a further process, such as excimer laser annealing. One example of this
was with a process referred to as L-MIC [53], in which the samples used for post-
MIC ELA had originally been implanted with 5 9 1013 Ni/cm2, and then annealed
at 550 C to achieve [90 % crystallinity. These samples had typical MIC char-
acteristics, with an electron mobility of *85 cm2/Vs, and a minimum leakage
current at Vd = 5 V of 2.5 9 10-12 A/lm. Following a subsequent 7-shot ELA
process, the changes in electron mobility and leakage current are shown in
Fig. 7.21, from which it will be seen that the minimum leakage current density was
reduced to 2.5 9 10-14 A/lm, and the electron mobility increased to *210 cm2/
Vs [53]. In addition, the electron mobility results were obtained within an energy
density window of *70 mJ/cm2, which is much larger than obtained from the
conventional ELA process (see Fig. 7.13 for a comparison with a conventional 10-
shot ELA process), and the low leakage current could be obtained with as few as
two shots with L-MIC process. The explanation for these results was that in the
near-melt-through SLG regime, un-melted MIC grains at the bottom of the film
seeded the re-growth, and this process also crystallised any residual a-Si regions,
which had remained between the MIC grains [53].
Other groups have also employed a process comparable to L-MIC, to produce
high performance TFTs, using a process referred to as continuous-grain-silicon (or
CGS) [54, 55]. This implemented a MIC process in 45 nm thick films of a-Si,
Fig. 7.21 Changes in electron mobility and TFT leakage current as a function of post-MIC ELA,
using a 7-shot process. (Reprinted from [53] with permission of IEEE)
using a spin-on solution of Ni in acetic acid, and a crystallisation stage, which was
carried out below 550 C. Following the Ni-mediated crystallisation, the samples
were passed through an ELA process to improve the crystal quality of the layer,
and self-aligned TFTs were then made using thermally activated boron and
phosphorus ion doping for the source and drain regions of p- and n-channel poly-Si
TFTs, respectively. (A general discussion of poly-Si architecture and processing can
be found in Sect. 7.4). The thermal activation of the phosphorus-doped regions
was also described as a gettering process, which reduced the Ni content in the TFT
channels. These devices were compared with ELA TFTs, in which the MIC stage
had been omitted, and the CGS and ELA TFTs showed electron mobilities of 320
and 115 cm2/Vs, respectively, and identical, low leakage currents of 5 9 10-14 A/lm
(at Vd = 1 V) [55].
These results are comparable to the L-MIC work [53], in which the electron
mobilities were higher than in conventional ELA TFTs. To clarify the differences
between CGS and ELA material, the grain structures were compared by electron
backscatter diffraction pattern (EBSP) analysis and by transmission electron
microscopy [55]. The EBSP measurements for CGS material, within
10 lm 9 20 lm sampling areas, showed that, at a pitch of 200 nm, the point-to-
point changes in mis-orientation angles were just a few degrees. Whereas for a
2 lm 9 2 lm area of ELA material, 30 nm point-to-point measurements gave
abrupt changes of mis-orientation angle by 30–60 between grains, but was nearly
zero within the grains. The absence of these large changes in the CGS material led
to the identification of 7–15 lm sized ‘domains’, within which the misorientations
were \5. No such domains were identified in the ELA material, just individual
grains in the size range 0.2–0.5 lm. In the CGS material, [111] planes dominated
the crystal orientation normal to the sample surface, which was consistent
with the \ 111 [ preferred crystallisation direction of the NiSi2 precipitates.
TEM analysis confirmed the sharp grain boundaries between the ELA grains, but,
for the CGS material, the grains were larger than 1 lm, and were frequently
separated by low angle grain boundaries. These fundamental differences between
the grain structure of CGS and ELA material, with the large CGS ‘domains’
containing only low-angle grain boundaries, suggest reduced grain boundary
scattering, explaining the higher carrier mobilities. It is also clear that the crys-
tallographic nature of this material, even after the ELA treatment, contains a grain
structure strongly influenced by the original Ni-mediated crystallisation process.
In summary, the SMC process was originally investigated as a low cost, and
simpler, alternative to conventional ELA, but, for reasons still not fully clarified,
the resulting poly-Si TFTs suffered from unacceptably high leakage currents.
However, when combined with an ELA process, the leakage current issues were
resolved, and higher performance TFTs, than from ELA alone, were fabricated
[52–55]. Moreover, as shown in some of this work [53], this could be achieved
with a low-shot ELA process and a large process window.
7.3 Gate Dielectrics
The preferred gate dielectric for poly-Si TFTs is silicon dioxide, as it is for single
crystal Si devices, in contrast to the use of silicon nitride, which is preferred for a-
Si:H TFTs. As discussed in Chap. 5, the preference for a-SiNx:H in a-Si:H TFTs is
determined by the meta-stability of this material and the reduced DOS resulting
from the positive charge in the nitride. For poly-Si, these meta-stability consid-
erations do not apply, and the positive charges in the nitride cause unnecessarily
large negative shifts in threshold voltage; in addition, the nitride itself is also
susceptible to gate bias induced trapping instabilities. These reasons also militate
against the use of Si3N4 in MOSFET devices. In the following section, the use of
SiO2 as a gate dielectric is discussed, and there is a brief review of alternative
dielectrics in Sect. 7.3.2.
7.3.1 Silicon Dioxide
The quality of the gate oxide in a TFT is crucial to the performance of the device,
and the oxide needs to have low leakage current, low densities of fixed charges and
interface states, high breakdown field, low pin-hole density, and good bias-stress
stability. Moreover, for the low temperature TFTs of interest in this book, the
oxide deposition process must be implementable below the glass softening point,
and, for practical purposes in commercial processing, this means below *450 C,
and, finally, over large areas with good uniformity. The temperature constraint
rules out the thermally grown oxide, which has underpinned the crystalline silicon
integrated circuit industry. Nevertheless, the properties of thermal oxides have
7.3 Gate Dielectrics 213
been widely used as a benchmark against which to measure TFT oxides, and the
closer these properties have matched the better has been the performance of the
low temperature oxides. The correlation of electrical properties, such as the den-
sity of states at the Si/SiO2 interface, can be readily understood, but even
mechanical properties such as the etch rate in hydrofluoric acid is a good indicator
of the density of the film [56], and this density has been correlated with the films
porosity and its propensity to absorb water [57]. Furthermore, this porosity has
been associated with gate-bias-stress induced threshold voltage instability in the
TFT, due to the motion of adsorbed H+ and OH- ions in the oxide [58]. Hence, a
number of simple measurements have been used to screen potential oxide
candidates.
A powerful technique for the assessment of the porosity of the film, and its
water content, has been thermal desorption spectroscopy (TDS), [57], in which, as
the temperature of the film is steadily raised, the desorbed gases from the film are
analysed in a mass spectrometer, set to atomic mass 18, to detect water. The TDS
plots in Fig. 7.22 were obtained from films deposited by PECVD from TEOS
(tetraethylorthosilicate) and O2 in helium carrier gas, and show three characteristic
water desorption peaks obtained from SiO2 films. The two low temperature peaks
at 100–200 and 150–300 C have been associated with adsorbed water at room
temperature in porous films. This was confirmed by noting that the heights of these
peaks could be increased by prolonged air exposure at room temperature, and
decreased by vacuum annealing at elevated temperatures. Hence, dense, low
porosity films should show minimal desorption peaks at these two temperatures.
The high temperature peak at 350–650 C was associated with isolated silanol, Si–
OH, bonds formed during film growth, and, unlike the lower temperature peaks,
was seen in all high quality oxides. As will also be seen from Fig. 7.22, the overall
peak heights were determined by the TEOS ? He/O2 flow ratios, and the highest
oxygen dilution ratio led to the densest and least porous film.
In addition to the use of TEOS and O2 gas mixtures in PECVD reactors [57, 59],
SiO2 depositions can also be obtained from SiH4 and N2O gas mixtures [42, 60, 61].
Fig. 7.22 Thermal

desorption spectra (TDS)
measured on PECVD TEOS
films of SiO2, deposited with
different levels of
TEOS ? He:O2 dilution.
The Japan Society of Applied
Physics)
These PECVD depositions at 300–400 C are typically carried out in parallel plate
reactors operating at 13.56 MHz, or 27 MHz, and, given the large area capabilities
of these reactors [59], PECVD deposition is the mainstream technique used for the
deposition of high quality gate oxides for poly-Si TFTs. As mentioned above, the
highest quality oxides have been obtained at low growth rates, which were achieved
by a large dilution of the silicon bearing gases by using high volumes of either
oxygen or helium carrier gas [56, 57, 60, 61].
Other oxides, which have been examined as possible gate dielectrics in poly-Si
TFTs, have included deposition by:
• APCVD from SiH4, and O2 [58],
• remote plasma, RPCVD, from SiH4, O2 and He at 300–350 C [62],
• ECR-CVD at 2.45 GHz from SiH4, O2 and He at 25–270 C [63, 64].
Comparative TDS measurements from some of these low temperature oxides
are shown in Fig. 23a [61] and b [65], and it will be seen from Fig. 7.23a that only
the helium-diluted SiH4/N2O films deposited at 250–300 C showed low porosity,
which also correlated well with an etch rate comparable to thermal SiO2 [60].
Hence, the minimum deposition temperature for this oxide is above 250 C, which
is not a problem for TFTs on glass, but is an issue for TFTs on low-temperature
polymer substrates. However, low-deposition-temperature ECR oxides were of
high density and low porosity, and are possible candidates for poly-Si TFTs on
polymer substrates [64]. This topic is discussed further in Chap. 11. The TEOS
oxides in Fig. 7.23b were deposited above 300 C with high O2 dilution ratios, and
compare well, in terms of low porosity, with the helium diluted SiH4/N2O refer-
ence sample, and are widely used in state-of-the-art TFT processing. The APCVD
oxides in Fig. 7.23a show large, low temperature desorption peaks, which, as
mentioned above, have been correlated with gate bias instability in TFTs [58], and
these oxides are not compatible with high quality TFTs.
The other essential treatment in producing a device quality oxide is a post-
deposition anneal in the temperature range 300–400 C to reduce both the fixed
oxide charge in the film and the interface state density [56, 60, 63, 64]. This is the
same as the low temperature hydrogen passivation anneal, used with thermal
oxides in MOSFET technology, to reduce both of the above states. Hydrogen
containing ambients, such as forming gas, or wet-N2, are used, and the treatment is
frequently carried out after final metallisation, and may be referred to as a post-
metal anneal. In the latter case, reducing gate metals, such as aluminium, can react
with hydroxyls in the film releasing hydrogen even in an inert annealing ambient
such as N2. In order to avoid the problem of separating interface trapping states
from trapping states in the poly-Si film itself, the assessment of this treatment is
usually carried out by C–V measurements of the film deposited onto a lightly
doped silicon substrate. Figure 7.24 demonstrates the effectiveness of a forming
gas anneal at 400 C on a dense, helium diluted SiH4/N2O oxide deposited at
350 C, in which the positive charge density in the oxide was reduced from
*8 9 1011 to *2.5 9 1011 cm-2, and the near-mid-gap interface density was
reduced from *1 9 1012 to *4 9 1010 cm-2eV-1 [56].
Fig. 7.23 TDS spectra of

SiO2 films (a) by PECVD
deposition using He-diluted
SiH4/N2O gases at different
temperatures, plus an
APCVD film and a sputtered
film (Reprinted from [61]
with permission of SID), and
(b) various PECVD O2-
diluted TEOS films together
with a reference He-diluted
SiH4/N2O film (unpublished
data from Dr N D Young of
Philips Research)
Whilst the above post-deposition anneal is extensively used, various other post-
deposition annealing regimes have been reported as leading to an improvement in
oxide quality, such as an excimer laser anneal, at an energy below the silicon melt-
threshold, which led to improvements in carrier mobility, oxide charge density and
gate bias and drain bias stress stability [66]. Another procedure giving similar
results was a high-pressure water vapour anneal at 1.3 Mpa and 260 C [67].
However, it is not clear that these, or other techniques, have achieved widespread
Fig. 7.24 MOS high 1.2

frequency and quasi-static Sample #85-H
C–V measurements on a 1.0
350 C PECVD He-diluted
SiH4/N2O SiO2 film, before 0.8
and after a post-metallisation
C/COX
anneal, PMA, at 400 C for 0.6

30 min. (Reprinted with
permission from [56]. 0.4
Copyright (1986) American As deposited (HF)
Institute of Physics) 0.2 After PMA (HF)
After PMA (QS)
0
–15 –10 –5 0 5 10 15
BIAS (V)
application. Moreover, there is always a cost consideration in implementing a

more complex processing schedule, which needs to be balanced against the level
of improvements which might be achieved when applied to the current state-of-
the-art high performance TFT process.
In terms of basic dielectric properties, the dense, helium diluted SiH4/N2O
oxides display a high breakdown field of *8 MV/cm, as shown by the large area
(5.2 9 10-3 cm2) capacitor results obtained from single crystal Si substrates [56],
and by the large area poly-Si TFTs (W = L = 100 lm) in Fig. 25a and b,
respectively. The latter results demonstrate a thickness independent breakdown
field, which is important for short channel TFTs, where the gate oxide thickness
needs to be scaled down with reducing channel length in order to suppress short
channel effects. The use of SiO2 films down to 20 nm thickness is demonstrated in
this data, but at 10 nm thickness premature breakdown was observed, indicating
the current limit of oxide thickness scaling. This probably results from a combi-
nation of the rough poly-Si surface, giving poor film integrity at this thickness, as
well as step coverage problems over the edges of the poly-Si islands. In the
following section, some alternative dielectric strategies are discussed for thin
oxides.
In addition to the gate dielectric application, high quality oxides are also needed
for interlayer dielectrics separating different levels of metallisation, such as the
gate and drain layers, and also for capping the glass substrates. The capping layer
has the double purpose of providing an electronically well-controlled layer at the
back interface of the TFT, as well as acting as a diffusion barrier to alkali ions
Fig. 7.25 Electro-static breakdown measurements on He-diluted SiH4/N2O SiO2 films (a) break-
down field of large area capacitors of 64 nm thick films on c-Si substrates (Reprinted with
permission from [56]. Copyright (1986) American Institute of Physics), and (b) breakdown
voltage measurements as a function of SiO2 film thickness in poly-Si TFTs (Reprinted from [76]
from the substrate. In fact, for the latter purpose, it is common to deposit SiN
directly onto the glass surfaces, as it is a better diffusion barrier than SiO2, and then
to cap this with SiO2. Typical thicknesses for these two layers are 100 nm SiN and
400 nm SiO2.
7.3.2 Alternative Dielectrics
In sub-0.1micron channel length MOSFETs, the scaling of gate oxide thickness

has resulted in ultra thin SiO2 films of\15 Å, where leakage current and reliability
concerns have become important issues. This has led to the study of high-k
dielectrics [68] as a way to use thicker films, whilst still maintaining the same
equivalent oxide thickness, EOT, and, hence, the same gm. There have not been the
same fundamental issues with the SiO2 layers used in poly-Si TFTs, but, as dis-
cussed in the previous section, a limit has been identified in the minimum thick-
ness of SiO2, which can be reliably used, and some investigations have been
reported on the use of higher-k dielectrics for this TFT application. The dielectric
constant of SiO2 is 3.9, and the two alternative dielectrics were PECVD a-SiNx:H
(k = 6.5) [69] and sputtered Ta2O5 (k = 11–25) [70]. (The dielectric constant for
bulk Ta2O5 is 25, but the thin film layers only gave a value of 11.) In both cases, a
dual dielectric approach was followed, in which a thin layer of SiO2 was used as an
interfacial layer between the poly-Si and the high-k dielectric. This process was
adopted in order to maintain the electronic quality of the poly-Si/SiO2 interface.
For the dual dielectric system, the EOT is given by:
3:9t2
EOT ¼ tSiO2 þ
k2
where t2 and k2 are the thickness and dielectric constant of the high-k layer,
respectively.
The p- and n-channel devices made with the a-SiNx:H/SiO2 dielectric contained
either 5 nm or 10 nm thick layers of SiO2, and gave good TFT characteristics
down to an EOT of 9.8 nm. The device parameters such as mobility, breakdown
field and drain bias stability were largely independent of the SiO2 layer thickness
[69], and this dielectric combination is a plausible route to thin oxide TFTs with an
EOT at least down to 10 nm.
The TFTs with the SiO2/Ta2O5 gate dielectric had a minimum EOT of 46 nm,
and needed several hours post-metallisation annealing, at 350 C in mixed gas, to
achieve low threshold voltages and high carrier mobilities [70], indicating the need
for further optimisation of this higher-k system in order to achieve an acceptable
combination of TFT performance and processing schedule.
Fig. 7.26 Schematic cross-sections of n-channel, top-gated poly-Si TFTs (a) non-self-aligned
architecture, and (b) self-aligned
7.4 Poly-Si TFT Architecture and Fabrication
7.4.1 Architecture
Poly-Si TFTs are most commonly implemented as top-gated structures, as shown

in Fig. 7.26a and b for non-self-aligned (NSA) and self-aligned (SA) devices,
respectively. As will be seen, this is a different architecture from the inverted
staggered architecture of a-Si:H TFTs (see Chap. 5), and is much more similar to
the architecture of single crystal silicon-on-insulator (SOI) MOSFETs. Further
points of similarity with the SOI MOSFETs, and distinction from a-Si:H TFTs and
emerging TFT processes in organic or amorphous oxide materials, are that both n-
channel and p-channel poly-Si TFTs, with comparable performance, can be fab-
ricated with a common process. Secondly, the source and drain doping is usually
accomplished using ion shower doping of boron for p-channel TFTs and pho-
phorus for n-channel TFTs. This process has a number of similarities to ion
implantation for MOSFETs, in that both involve the high-dose implantation of
selected ions. In other words, the architecture and a key fabrication stage of poly-
Si TFTs are closer to MOSFETs than to devices fabricated in other thin film
materials. However, this high-dose ion doping process is potentially damaging to
the crystallographic order of the poly-Si film, particularly for the heavier phos-
phorus ion compared with boron. The degree of damage will be determined by the
ion dose, ion energy and substrate temperature, and, it has been shown that, for
high enough doses into single crystal silicon, the damage can be sufficient to cause
complete amorphisation of the implanted layer [71]. This is also an issue in poly-
Si, and is discussed further in Sect. 7.4.1.1.
The essential difference between the two architectures in Fig. 7.26 relates to the
formation of the source and drain regions, and this determines the overlap of the
gate electrode across these regions. For the NSA structure, with significant over-
lap, the source and drain regions can be ion doped prior to the crystallisation of the
film, and the dopants in these regions will then be activated during film crystal-
lisation. This process can be implemented with direct ion doping into bare a-Si,
7.4 Poly-Si TFT Architecture and Fabrication 219
and results in a high level of dopant activation, with minimal residual ion damage.
A doping process employing a phosphorus dose of 1 9 1015 cm2 at 10 keV,
typically gives a doped region sheet resistance of *300 ohms/square. Subsequent
to this stage, the gate dielectric is deposited and defined, followed by gate, drain
and source metal deposition and definition. The overlap between the gate metal
and the doped source and drain regions is determined by the alignment rules for
the process, and is likely to be up to *3lm. This is a simple and robust fabrication
procedure, and is well suited to the basic study of material parameters (as
described in Sect. 7.2), but the architecture suffers from larger parasitic gate-drain
capacitance, which will degrade high frequency transistor performance. In addi-
tion, it is not well suited to the fabrication of shorter channel devices (L \ 3lm)
due to alignment issues, and the possibility of uncontrolled channel shortening due
to lateral diffusion of the source and drain dopant during the laser crystallisation
stage.
The above limitations are removed with the SA architecture, but, in several
respects, it results in a more complex fabrication process, primarily associated with
the source and drain doping stage, which is discussed in the following section.
7.4.1.1 Self Aligned Source and Drain Doping
The self-alignment between the edges of the gate and the edges of the source and
drain regions is achieved by using the gate electrode as an ion doping mask during
the doping stage, which means that the dopants are implanted through the gate
dielectric. In order to penetrate this layer, the doses and energies will be larger than
those used for the NSA TFTs. Furthermore, as the doping occurs after the tran-
sistor channel layer has been crystallised, a further processing stage has to be
introduced to activate the dopant: this can be either a second pass through the laser
or a thermal activation process in a furnace, or by rapid thermal annealing, RTA.
For the furnace process, the constraints imposed by the glass substrate dictate a
low thermal budget, such as a few hours at a maximum temperature of 450 C, and
relies upon the re-growth of the ion-damaged film being seeded by an undamaged
poly-Si layer at the bottom of the film. Hence, to ensure this seeding, full film
amorphisation must be avoided, otherwise the thermal budget will increase to the
values quoted in Sect. 7.2.1 for the SPC process. As mentioned above, amorphi-
sation is a more serious issue with the heavy phosphorus ions, and the effects
discussed below were not seen with boron doping for p-channel TFTs. Figure 7.27
shows the variation of poly-Si sheet resistance as a function of phosphorus ion
dose, and compares laser and furnace activation. With laser activation, the sheet
resistance varied inversely with phosphorus dose, as would be expected, but, with
furnace activation at 450 C, this was only seen at the lower doses. At the higher
doses, the sheet resistance began to increase with dose and, at the highest dose, it
was above the measurement limit of 1 9 106 ohm/square. A broader data set
showed that there was a critical combination of ion energies and doses [72], above
which thermal activation at 450 C was ineffective. This was confirmed to be due
Fig. 7.27 Variation of sheet

resistance of phosphorus
implanted poly-Si as a
function of dose and
activation method (80 keV
implant through 145 nm of
SiO2)
to film amorphisation, since doses, which could not be activated in 40 nm thick

films, could be activated in 80 nm thick films. Finally, it should be noted that even
where furnace activation was achieved, it resulted in a higher sheet resistance than
obtained with laser activation. This is likely to be due to lower dopant solubility at
the lower temperature. Hence, from the standpoint of comparing phosphorus
activation, the 450 C process resulted in higher sheet resistance and a smaller
useable dose range than laser activation. A similar difference in the sheet resis-
tance between laser activated and thermally activated regions was also found when
higher temperature rapid thermal annealing was used instead of a 450 C furnace
process [73].
Further issues in the SA doping process are revealed when the resulting TFT
transfer characteristics, processed with furnace or laser activation, are compared in
Fig. 7.28 [72]. One obvious feature in these curves, and in other publications [74, 75],
is the higher minimum leakage current in the laser-activated n-channel TFTs,
which is due to residual ion-doping damage at the edge of the drain junction [72, 75].
Cross-sectional TEM of a more heavily phosphorus-implanted sample showed a
200 nm wide residual amorphous region, located in the doped region near the gate
edge, and also extending beneath the gate [34] (due to lateral end-of-range damage
not exposed to the laser). Very similar results have been reported in arsenic doped
n-channel TFTs, where the leakage current and TEM-imaged crystallographic
damage in the exposed silicon near the gate edge have been correlated with dif-
fraction of the laser beam by the gate edge [75].
Another feature in Fig. 7.28 is the lower on-current in the furnace activated
n-channel TFT, due to this activation process giving higher sheet resistance in the
source and drain regions, as shown in Fig. 7.27. Finally, the field effect mobility
extracted from laser crystallised n-channel TFTs was less than from identically
crystallised NSA TFTs [72]. As with the leakage current, this has also been seen in
arsenic doped TFTs, and, again, associated with the diffraction-induced residual
damage effects at the gate edge [75]. This damage can also have a significant
Fig. 7.28 Transfer

characteristics of self-aligned
n- and p-channel poly-Si
TFTs, showing the influence
of laser or furnace dopant
activation. (Reprinted with
impact upon the operation of short channel TFTs [75–77], as shown by the field
effect mobility results in Fig. 7.29 [76]. The mobility values in this figure were
corrected for the series resistance in the doped regions themselves, and were used
to calculate the parasitic resistance specifically induced by the phosphorus ion
doping process.
Various techniques have been reported for minimising the general ion doping
damage effects in SA n-channel TFTs, including a careful control of ion dose and
energy [72]. For the laser activation process, in particular, oblique-incidence laser
irradiation has been shown to minimise the gate-edge diffraction [78], and,
thereby, reduce the associated damage effects on leakage current and carrier
mobility. A technique producing comparable results is the use of an off-set gate, in
which the gate length is reduced by lateral etching after ion doping, so that the
diffraction and masking effects at the gate edge no longer interfered with the full
Fig. 7.29 Dependence of

electron field effect mobility
measurements on channel
length, for SA aligned n-
channel poly-Si TFTs.
permission of SID)
melting of the doped region. This can lead to complete damage removal, as shown
in Fig. 7.30 by the TFT characteristics of a device with a 0.5 lm off-set gate [34].
The off-set gate is also one of a number of techniques, to be discussed in Sect.
7.4.1.2, for reducing the field at the drain edge, which are used to improve the
drain-bias-stress stability of poly-Si TFTs. Finally, laser activation, followed by
furnace activation, can be used to reduce the leakage current to the lower value
seen in furnace activated TFTs.
As is apparent from the above discussion, the process for fabricating SA TFTs,
with minimal performance artefacts, is considerably more complex than the NSA
TFT process, and this is the reason for favouring the use of NSA TFTs for basic
poly-Si materials studies, such as those presented in Sect. 7.2. However, the
potentially superior performance of SA TFTs, due to lower parasitic capacitances,
and their better compatibility with sub-micron channel length TFTs makes them
the preferred choice for many poly-Si applications.
7.4.1.2 Drain Field Relief
When poly-Si TFTs are exposed to a drain-bias stress, there can be a critical fall-
off in performance, as illustrated by the results in Fig. 7.31. The exposure of the
device, to a 13 V drain bias for 60 s, led to an increase in off-state current and a
reduction in on-state current, as well as a major change in the shape of the output
characteristics [79]. This phenomenon has been widely ascribed to hot electron
damage [80, 81], and the physics of this process are discussed in Chap. 8, where it
is shown that an alteration in device architecture is needed to suppress this
instability. The architectural change is one in which the field at the drain/channel
Fig. 7.30 Effect of gate

offset in removing the
residual ion doping damage
from laser activated n-
channel SA poly-Si TFTs.
from [34])
Fig. 7.31 Influence of drain bias stress of 13 V for 60 s on NSA TFT transfer characteristic. The
insert shows its effect upon the output characteristic. (Reprinted with permission from [79].
Copyright (1998) The Japan Society of Applied Physics)
junction is reduced, and this is most easily accomplished by the use of lightly
doped drain (LDD) regions, as shown schematically in Fig. 7.32a and b. These two
figures illustrate two LDD variants, where the lightly doped region is either
external to the gate, or is overlapped by the gate (GOLDD). In both cases, an extra
masking and processing step is involved to form this region, and, as with the drain
region itself, the LDD region is formed by phosphorus ion doping, but with a lower
dose in the poly-Si. (It is typically in the range 5 9 1012 – 5 9 1013 cm-2, in
contrast to the fully doped drain region where the dose is likely to be in the range
5 9 1014 – 1 9 1015 cm-2). When the region is defined photo-lithographically,
its size will be determined by the process design rules, and is likely to be of the
order of 1–3 lm.
The qualitative trade-off with dose is that lower doses will give more field
relief, and hence, better stability, but a larger series resistance, and hence, lower
on-current. This trade-off is minimised with GOLDD, where the gate modulates
the conductivity of the GOLDD region, and it also gives better device stability (see
Chap. 8). However, as with the NSA architecture, GOLDD gives greater parasitic
capacitances, and is less compatible with short channel length devices.
To obtain sub-micron LDD regions, a non-lithographic definition procedure is
required, such as the lateral etch-back of the gate, as used to form the off-set gate.
With this technique, the gap can be left undoped, with reliance upon the lateral
diffusion of the drain dopant, during a laser activation stage, to dope the LDD
region. Alternatively, a low dose implant can be used to deliberately dope the
Fig. 7.32 Cross-sectional diagrams of SA poly-Si TFTs showing LDD regions (a) conventional
LDD, and (b) gate overlapped LDD, GOLDD
region [82]. This procedure requires good control of the lateral etching process,
because too large a gap will give increasingly large series resistance in the
on-state.
A potentially more controllable sub-micron process is to use a sidewall spacer
on the gate to define either an undoped off-set gate [83], or a doped LDD region,
which can be a GOLDD region if the spacer itself is conducting [84].
7.4.1.3 Other TFT Architectures
Although the top-gated structures discussed above are the most widely used, bottom
gated NSA TFTs have been implemented [85]. SA bottom gate TFTs have also
been demonstrated by using backside exposure of photoresist on the top of the plate,
where the bottom gate acts as an in situ mask to define the channel length [86].
It is also possible to fabricate NSA top-gated TFTs, without using ion doping,
by sequentially depositing intrinsic a-Si and n+ doped a-Si, and removing the n+
a-Si from the channel region [87]. This is similar to the back-channel-etch process
in a-Si:H TFTs (see Sect. 5.3.1). Following the clearance of the channel region,
laser crystallisation is used in the normal way on both channel and doped regions,
and is followed by gate oxide deposition, contact window opening, and metalli-
sation to complete the TFT fabrication process.
7.4.2 Fabrication Process
Two examples of the poly-Si TFT fabrication process are shown in Figs. 7.33 and
7.34, and these list both the major processing steps and the photolithography
stages. The fabrication of a simple NSA n-channel TFT (whose cross-section is in
Fig. 7.26a) is listed in Fig. 7.33, and it is a 4-mask process involving the definition
of the poly-Si island, the source and drain doping locations, contact window
opening through the gate oxide, and definition of the metal contact areas. The SA
Fig. 7.33 Fabrication stages for non-self-aligned poly-Si n-channel TFTs
n-channel TFT (see Fig. 7.26b) is also a 4-mask process, with island definition,
gate metal definition, contact window opening, and source and drain metal defi-
nition. These 4-mask process flows are the minimum number of mask stages
needed to fabricate a device without field relief. For the fabrication of an AMLCD,
with CMOS drive circuits and stable n-channel TFTs, an increase in the number of
depositions and mask stages is needed, and a 9-mask process is shown in Fig. 7.34.
Compared with the single channel TFTs, the extra stages are separate photoli-
thographic definitions for the source and drain areas for the n- and p-channel TFTs,
definition of the LDD area for the n-channel TFTs, contact windows for the ITO
pixel electrodes, and the ITO patterning itself.
For both architectures, two other stages are frequently necessary, namely, de-
hydrogenation, and low dose B+ doping, in steps 3 and 5, respectively. The a-Si:H
pre-cursor film typically contains *10 % hydrogen, and, if this film is exposed to
a top-hat laser beam, there will be an explosive release of hydrogen, causing
severe blistering of the film. A low temperature thermal anneal at 400–450 C is
used to reduce the hydrogen content to *1 % or less, at which level the film can
be exposed to a top-hat beam without risk of mechanical damage. An alternative to
this process is higher temperature deposition of a-Si, giving a lower H-content, or
multi-shot irradiations at increasing energy density, to progressively heat the film
leading to controlled out-diffusion of the hydrogen [88]. The same effect can also
be achieved with a shaped laser beam, in which the leading edge is much less sharp
than in the top-hat beam [89]. This leads to more gradual heating of the film over
several incrementally higher intensities, giving controlled exo-diffusion of
hydrogen before the film is ultimately exposed to the maximum laser beam
intensity.
The low dose boron ion doping is used to compensate for an intrinsic electron
richness in the crystallised poly-Si film, which is due to a combination of positive
charges in the oxides above and below the film, and/or the neutral level of the
poly-Si band-gap states being above mid-gap. The net effect of these centres is to
shift the threshold voltage of the TFTs in a negative direction. Boron doping is
Fig. 7.34 Fabrication stages for a poly-Si AMLCD, containing complementary self-aligned
TFTs, and with LDD field relief for the n-channel TFTs
used to compensate this shift, and to achieve symmetrically positioned transfer

characteristics of p- and n-channel TFTs either side of VG = 0 V (as seen with the
characteristics in Figs. 7.9 and 7.28).
Much of the processing of poly-Si TFTs and AMLCDs can be accomplished
with the equipment used for the fabrication of a-Si:H AMLCDs, and this would
include plate cleaning, photolithography and etching stages, as well as the depo-
sition of various metals such as aluminium and ITO. Also, the PECVD deposition
of the gate oxide and a-Si:H precursor films can be deposited in standard, large
area PECVD reactors, using the same deposition conditions as for the a-Si:H TFT
films. However, there are also large area PECVD reactors, which can deposit a-Si
at temperatures up to *400–450 C, and incorporate an in-situ de-hydrogenation
anneal, so that the films are ready for laser crystallisation without any further
treatment [90]. The additional items of equipment needed for poly-Si TFTs are a
laser crystallisation system [18], and a large area ion doping system [91]. The ion
doping system will deliver a rectangular (or ribbon-shaped) beam of ions, through
which the substrate plates are scanned in a direction parallel to the short axis of the
beam. The beam will not have the high-resolution mass-separation of a semi-
conductor industry ion implanter, but will have sufficient ion mass filtering to
ensure that hydrogen is excluded from the beam, that phosphorus and boron cross-
contamination is avoided, and that for a phosphorus doping stage, say, only P+ ions
are selected, whilst other species such as P2 + are removed. (The original ion
doping systems lacked these refinements.) The other item of equipment, which
might be employed, is a high-resolution photolithography station, if short channel

TFTs are used for the addressing circuits. These will need higher resolution than
the 3lm typically available with the large area LCD-industry aligners.
7.5 Advanced Processing
7.5.1 Large Grain Poly-Si
As discussed above, the conventional ELA technique produces devices with limited
electron mobility (\200 cm2/Vs), limited throughput and a small laser processing
energy window. Various techniques have been explored to address some of these
problems, with one aim also being larger grain poly-Si, giving higher carrier
mobility. These techniques have included both solid state green lasers, as well as
modified excimer laser crystallisation procedures. The modified excimer laser
procedures include sequential lateral solidification (SLS) [29, 92–94], phase
modulated excimer laser annealing, PMELA [1, 95–97], and a micro-Czochralski
process, l-Cz, for location controlled single grain growth [98, 99]. In all these
cases, a technique has been developed to control and extend lateral grain growth
during the SLG process, rather than relying upon the random process, which occurs
during conventional ELA. These techniques are discussed in Sect. 7.5.2.
As briefly discussed in Sect. 7.2.3, solid state green lasers have been advocated
to overcome some of the technical and cost of ownership issues with ELA, as well
as being able to produce large grain material [35–42], and these approaches are
presented in Sect. 7.5.3. Finally, a comparison is made between the excimer and
green laser approaches in Sect. 7.5.4.
7.5.2 Modified Excimer Laser Crystallisation
The SLS and PMELA techniques are conceptually similar in that both procedures
expose the substrate to a spatial intensity fluctuation, varying from a below-melt-
threshold intensity to a value high enough to fully melt the silicon film. At the
boundary between these two regions, there will be a portion of material which will
experience near-melt-through conditions, where the normal SLG process will
occur. Those grains seed lateral growth into the adjacent fully molten regimes,
giving grain growth of a few microns or more. SLS directly achieves the intensity
fluctuation by passing the laser beam through a projection mask, containing
alternate clear and occluded regions. With PMELA, the laser beam is passed
through a phase shifting mask, so that optical interference in the emergent beam
produces the required intensity fluctuation. However, the implementation is dif-
ferent, with SLS using a multi-shot scanning procedure to promote grain
Fig. 7.35 SEM micrograph of location-controlled large grain arrays grown by PMELA.
(Reprinted with permission from [96]. Copyright (2008) The Japan Society of Applied Physics)
propagation, whereas PMELA uses a single shot technique to produce an array of

location controlled large-grain (or single crystal) areas, typically 5 lm square, as
shown in Fig. 7.35 [96]. The SEM micrograph shows a combination of large grain
areas, lateral growth areas plus the nucleation area. To achieve this well defined,
location-controlled grain structure, with a robust fabrication process, the phase
shift mask has evolved from its original, simple form, of etched depressions in a
quartz plate [95], to a complex matrix of variably sized pits and humps on the
quartz plate, which is referred to as a 2-dimensional bipolar phase modulator plate,
2-D BPM [96, 97].
The large grains from the SLS and PMELA processes yielded high electron
mobility values of up to *600 cm2/Vs [93] and *900 cm2/Vs [1], respectively.
For the PMELA assessment, TFTs with W = 2 lm and L = 1 lm were used, so
that they were entirely contained within the 5 lm 9 5 lm large-grain areas. In
*10 % of these grains, both p- and n-channel PMELA TFTs matched the per-
formance of single crystal SOI MOSFETs, which were identically processed (apart
from the crystallisation stage). These high performance TFTs were located in
regions of {100} pseudo-single-crystal (PSX) material; however, as the PSX
regions accounted for only 10 % of the crystallised areas, techniques still need to
be established to improve uniformity by generating this orientation in all crys-
tallised regions [1].
In contrast, to the above two techniques, the l-Cz technique [98] requires the Si
film to be patterned rather than modulating the incident laser beam. This is
achieved photo-lithographically, by etching shallow, small diameter (\100 nm)
holes into a layer of SiO2 which is subsequently coated with a-Si. The a-Si film
7.5 Advanced Processing 229
fills the holes, so that the Si film thickness at these points is greater than on the
surrounding surface, and, when exposed to a laser intensity sufficient to melt the
surrounding film, the film remains solid at the bottom of the hole. Hence, there will
be a transition region in the hole, within which the normal ELA process will form
some crystallites, and these will seed lateral grain growth into the surrounding
molten Si, as shown schematically in Fig. 7.36a [99]. By controlling the diameter
of this hole, it can act as a grain filter and reduce the number of seeding crystallites
to unity, with the objective of producing single crystal areas. As with the PMELA
technique, this procedure has been implemented to produce an array of square,
abutting grains, whose sides are *6 lm in length, with the seeding region sited in
the centre of these grains. The grain structure is shown in Fig. 7.36b, and electrical
assessment of the material was carried out with small channel TFTs (W and L
*2 lm), which were able to fit within the individual grains [99]. Figure 7.36b
illustrates different TFT positioning within the grains, and those sited directly over
the grain filter, such as ‘C’, had the worst characteristics. For those TFTs hori-
zontally or vertically displaced from this location, such as ‘X’ or ‘Y’, the device
Fig. 7.36 a Cross-sectional

diagram of l-Cz process and
c-Si TFT channel sited within
a single grain, and b SEM
micrograph of an array of
location controlled grains
grown by the l-Cz technique.
Superimposed on this image
are TFT locations with
respect to the grain filter
positions. (Reprinted from
[99] with permission of
IEEE)
characteristics were much better, with electron mobilities up to *600 cm2/Vs.

However, the standard deviation in these values was ±100 cm2/Vs [99], and it is
likely that uniformity improvements will be needed for routine implementation of
this process.
Both the PMELA and l-Cz processes have been implemented with single shot
crystallisation, whilst the minimum shot number in SLS is two, and all processes
use beam intensities sufficient to cause full melt-through. Hence, these techniques
offer the potential for higher throughput and a larger energy density processing
window than conventional ELA (in addition to the improved TFT performance
through higher carrier mobility). At the moment, the SLS technique appears to be
the closest to commercial implementation [100], and it will be described in more
detail in the following section.
7.5.2.1 Sequential Lateral Solidification
As mentioned above, SLS uses a spatially varying beam intensity, which consists
of a number of beamlets, just a few micrometers wide in one direction and several
millimetres in the other. The beamlets are produced by projecting the homogenised
excimer laser beam through a mask having alternate clear and occluded stripes [29,
32], and the ensuing crystallisation process is shown schematically in Fig. 7.37a–
c. Within the high intensity beamlet areas, full film melting takes place and, at the
edge of these regions, there will be an SLG region, which seeds the lateral growth
into the molten region. Following single shot irradiation, two scenarios are shown,
depending upon the relative widths of the beam, WB, and of the seeded lateral
growth distances, Llat. For Llat \ WB/2, the centre of the molten region undergoes
random nucleation before the lateral growth front reaches this region, and results
in fine grain material in the centre of each exposed stripe. To avoid this happening,
the beamlet size needs to be \*5 lm (in contrast to the conventional line beam
ELA, in which the beam is *350–500 lm wide), resulting in the situation shown
in Fig. 7.37c. In this case, the two lateral growth fronts meet in the centre of the
previously melted region, and produce a collision-front grain boundary. In the
orthogonal direction, there are low angle sub-grain boundaries running parallel to
the lateral growth direction.
Following this first shot, the material consists of alternating stripes of a-Si and
poly-Si, and further irradiations are required to complete the crystallisation pro-
cess, as shown schematically in Fig. 7.38a and b. In Fig. 7.38a, the plate is moved
a distance Dx prior to the next irradiation, where Dx \ WB/2, such that the original
melt-front-collision GB is melted by the second pulse, and the remaining poly-
crystalline material seeds further lateral growth, propagating the grain to the right.
With an appropriate projection mask design, this process can be indefinitely
repeated to produce grains tens or hundreds of micrometres long. The main con-
straint with this implementation of the process is the throughput, which will
inversely scale with the number of grain-propagating laser shots.
Fig. 7.37 Schematic illustration of the SLS process: a micro-beamlets of width WB, separated
by gap Wd, b lateral grain growth, and random nucleation when WB is too large, and c complete
crystallisation of the melted area when WB is less than the ateral growth limit
Fig. 7.38 Illustration of the SLS grain propagation process: a generation of extended grain
material, where the beamlet stepping interval, Dx \ WB/2, and b repeating array of poly-Si
grains, where Dx [ WB/2
When Dx [ WB/2, the situation shown in Fig. 7.38b is produced, and the
original collision-front grain boundary is retained, and a similar grain structure is
generated by the second irradiation. With an appropriate mask design, this process
can be implemented as a 2-shot process, giving the highest throughput obtainable
with SLS. The grain structures resulting from these two SLS implementations are
shown in Fig. 7.39a and b, respectively [100]. Figure 7.39a is an SEM micrograph
of the multi-shot grain propagation process, giving grains [15 lm long, with
closely spaced sub-grain boundaries running parallel to the propagation direction.
The spacing of these sub-grain boundaries increases with increasing film thickness
[101], and, with increasing grain propagation, there is grain filtering and a ten-
dency for the low-angle sub-grain boundaries to become high-angle [102]. The
micrograph in Fig. 7.39b is of the 2-shot SLS process, showing the regularly
spaced collision-front grain boundaries, as well as the low-angle sub-grain
boundaries.
Fig. 7.39 SEM micrographs

of SLS Si: a extended grain
material, and b ‘2-shot’ SLS
poly-Si. (Reprinted from
ITC’07)

electron mobility with pulse
translation distance, Dx, in
SLS material, for carrier flow
parallel and perpendicular to
the low angle grain
boundaries. The demarcation
line between extended grain
and ‘2-shot’ SLS poly-Si is
shown at Dx = WB/2
(WB = 2.5 lm,
tSi = 100 nm)
Some of the characteristic features of SLS material are shown by the TFT
results in Fig. 7.40 [103]. The electron mobility is plotted as function of the
substrate translation distance between pulses, Dx, and results are shown for carrier
flow parallel and perpendicular to the sub-grain boundaries. These results were
taken from 100 nm thick poly-Si films, with a sub-GB spacing of 0.37 lm. The
two major features in this plot are, firstly, the reduction in mobility for carrier flow
perpendicular to the sub-grain boundaries, due to the GB anisotropy giving greater
carrier scattering in this direction of current flow. Secondly, for parallel flow, the
reduction in electron mobility from *300–350 cm2/Vs to *240 cm2/Vs, with
increasing plate translation distance, is due to the change in material from the
extended grain form (Dx \ WB/2) to ‘2-shot SLS’ material containing regularly
spaced collision-front grain boundaries (Dx [ WB/2). Qualitatively identical
results have also been obtained for the hole mobility in p-channel TFTs. Within the
two specific types of material, extended grain and ‘2-shot SLS’, the process was
able to yield good mobility uniformity of ±6–7 %, and ±2–3 %, respectively
[103]. The better uniformity of the ‘2-shot’ material was most likely due to better
averaging of grain properties over those much smaller grains. In other work,
optimisation of the high throughput, ‘2-shot SLS’ process has given electron
mobility values up to 350 cm2/Vs [94].
The above discussion has focussed on the formation of ‘directional’ material by
scanning line-beamlets. The technique can also be implemented to give localised
single crystal regions by the use of chevron shaped beamlets [29], or by imple-
menting the 2-shot process in two orthogonal directions [104]. However, ‘single
crystal’ SLS material has not been electrically characterised as intensively as the
PMELA and l-Cz materials.
7.5.3 Green Laser Crystallisation
Diode pumped solid state lasers have been advocated to address both the pulse
stability and cost of ownership issues associated with ELA, and Q-switched
Nd:YAG and CW Nd:YVO4 lasers have been the most widely used solid state lasers
[35–42]. The fundamental output wavelength from both lasers is 1064 nm, and, in
order to achieve efficient absorption into thin a-Si films, the second harmonic phase
at 532 nm has been utilised, where the absorption depth in a-Si and poly-Si is
100 nm and 125 nm, respectively [24]. This is *10 times greater than the
absorption depth of 308 nm radiation, so that the coupling efficiency is smaller,
although the authors of the green laser work argue that the greater absorption depth
gives a more uniform temperature profile in the silicon, which is discussed further
below.
7.5.3.1 Pulsed Nd:YAG Lasers
In the Q-switched mode at 532 nm, Nd:YAG lasers can deliver short duration pulses
(C10 ns) of 800 mJ/cm2, with a repetition frequency of 4 kHz and an output power
of 200 W [35]. The basic beam shape from a Nd:YAG laser is circular, with a
Gaussian distribution of energies across its diameter, and an a-Si crystallisation
system has been developed in which this shape has been optically converted into a
line-beam [35]. In this system, the length of the line beam, which had a top-hat
profile, was 105 mm, whilst, in the short direction, the Gaussian shape was retained,
with a full width at half maximum of 40 lm, as shown in Fig. 7.41. For crystalli-
sation, the plate scanning direction was parallel to the short axis, with typical
translation stage movements of 1–5 lm between pulses [35–37], giving a multi-shot
process of 40–8 shots, respectively. The pre-cursor a-Si films were typically in the
thickness range 50–100 nm, and, when crystallised into poly-Si, three energy
Fig. 7.41 Beam profile measured in 532 nm, Q-switched Nd-YAG laser crystallisation system.
(Reprinted from [35], with permission of Ulvac)
Fig. 7.42 Dependence of TFT characteristics on Q-switched 532 nm Nd:YAG laser power
(a) electron mobility, and (b) threshold voltage and sub-threshold slope. (Reproduced with
density regimes were identified [36]: the first, at low energies (85–95 W laser
power), was referred to as vertical crystallisation, and gave grains of *100 nm, or
more, in partially melted films. The second regime (105–120 W) was named SLG,
which occurred when the film had fully melted and yielded larger grains, and the
third regime (above 130 W) yielded fine grain material. When the TFT device
parameters were measured as a function of laser power, as shown in Fig. 7.42, they
showed a good correlation with the power-dependent grain structure, and gave a
maximum electron mobility of 250 cm2/Vs over the optimum laser power range of
*105–120 W for a 20 shot process [36]. As will be apparent, these regimes, and the
associated device properties, are similar to the regimes identified with ELA. How-
ever, the authors argue that there is a fundamental difference due to the *10 times
Fig. 7.43 Relationship

between Q-switched 532 nm
Nd:YAG laser beam profile
and local poly-Si grain
morphology. (Reprinted from
[37] with permission of SID)
larger absorption depth of the green laser [35–37], which produces a more uniform
temperature distribution through the depth of the film. As a result of this, the SLG
process does not start from the bottom interface, but from the edges of the melted
area [35–37]. In support of this argument, the micrographs in Fig. 7.43 show fine
grain silicon in an area fully melted by the Nd:YAG beam, with large grain silicon
appearing at a specific energy density on the edge of the irradiated area. Moreover, it
was shown that this laterally seeded large grain region could be laterally propagated
as the plate is stepped through the beam, leading to a larger processing window and
larger grains than with the standard ELA process. The process comparison, in terms
of the range of energy densities over which high carrier mobility TFTs can be
produced, is shown in Fig. 7.44 [37].
However, this Nd:YAG process appears to be similar to the SLS process [29],
discussed in Sect. 7.5.2.1, which also operates in full melt-through, relying upon
seeded crystallisation from the edges of the melted areas, and uses small plate
stepping intervals of 2–3 lm to propagate large sized grains. In view of these
similarities between Nd;YAG irradiations and excimer SLS, the features of the
Nd:YAG process cannot be uniquely attributed to the greater optical absorption
depth of 532 nm radiation in a-Si. Indeed, as discussed above, in Sect. 7.2.2.2, the

electron mobility and
processing window sizes for
ELA and Q-switched Nd-
YAG crystallisations.
permission of SID)
large thermal diffusion length in a-Si leads to full melting of excimer laser irra-
diated thin films up to at least *200 nm.
In summary, pulsed Nd:YAG crystallisation can lead to a larger process win-
dow, and potentially higher performance devices than conventional ELA, together
with good pulse-to-pulse stability and lower cost of ownership. However, SLS also
offers a larger processing window, and better device performance than conven-
tional ELA. From a manufacturing perspective, system throughput and complexity
are also important considerations, and these are compared in Sect. 7.5.4.
7.5.3.2 CW Nd:YVO4 Lasers
For CW crystallisation, Nd:YVO4 lasers are preferred to Nd:YAG lasers, as they

operate with higher power and can deliver 10 W at 532 nm, with a power stability
of better than 1 %. However, the conditions for successful crystallisation, using the
high power CW laser, were more restrictive than for pulsed lasers, with the major
problems being de-lamination of the a-Si film and damage to the glass substrate.
To avoid these problems, the a-Si film was pre-patterned into discrete islands,
rather than left undefined, and the preferred patterning was into close-packed
50 lm 9 200 lm lozenge-shaped areas, with an inter-island spacing of 5 lm
[39, 40]. These areas were large enough to accommodate 1–2 TFTs with
W = 3 lm and L = 5 lm. Secondly, crystallisation of films thicker than 300 nm
damaged the glass substrate, and, to avoid this, the a-Si thickness range was
restricted to 50–150 nm, and, thirdly, high speed scanning at more than 20 cm/s
was required. All these measures served to limit both the amount of energy
absorbed by the film and coupled into the substrate, although clearly the power
setting of the laser played a role as well.
The laser spot size was 200 lm 9 20 lm, which was scanned in a direction
parallel to the short axis along the length of the pre-defined a-Si islands. The
resulting crystal structure displayed a strong interdependence on scan speed and
Fig. 7.45 CW Nd:YVO4 laser crystallisation of poly-Si at 532 nm: a dependence of grain size
and structure on laser power and scan speed, and b optical micrograph of pre-defined Si island,
showing the Secco-etched large grain poly-Si in the centre of the island. (Reprinted with
permission from [40]. Copyright (2002) The Japan Society of Applied Physics)
laser power setting, as shown by the results for 150 nm thick films in Fig. 7.45a
[40]. Three crystallisation regimes were identified: at low power or high scanning
speeds, small grain silicon was formed, whose appearance (as shown by the insert),
at 10 W and 100 cm/s, was comparable to the SLG grains in ELA processing,
whilst at 200 cm/s the grain structure was more consistent with solid phase
crystallisation [39]. At higher powers and/or slower scan speeds, large laterally
propagated grains were formed (see insert), whose dimensions were
*3 lm 9 20 lm, with the grain boundaries running parallel to the beam scanning
direction. At the slowest scan speeds and highest powers the film and substrate
were damaged.
The large grains were formed preferentially in the central regions of the pre-
defined Si islands, because faster cooling rates along the edges of the islands
resulted in smaller grains there, as shown in Fig. 7.45b. The transfer characteristics
from high performance p- and n-channel TFTs, formed in a 150 nm thick film
crystallised with 6 W, at a scan speed of 20 cm/s, are shown in Fig. 7.46. The high
carrier mobility was attributed both to the large grain size and to the (100) surface
orientation of these grains (although 100 nm thick films had a mixture of (110) and
(111) oriented grains). The more general dependence of electron mobility on plate
scan speed is shown in Fig. 7.47, together with the associated grain structures. The
TFT transfer characteristics in Fig. 7.46 were for carrier flow parallel to the grain
boundaries, and the data in Fig. 7.47 are for carrier flow both parallel and per-
pendicular to the grain boundaries. For perpendicular flow in the large grain
material, the mobility was considerably lower than for parallel flow, which is
similar to the situation with SLS silicon, due to the scattering effect of the grain
boundaries. For the smaller grain, ELA-like material, the grains were equi-axed
and there was no anisotropy in electron mobility.
Fig. 7.46 Transfer

characteristics of n- and p-
channel large grain TFTs
crystallised by CW Nd:YVO4
laser at 532 nm. (Reprinted
The uniformity of device characteristics has not been published [39, 40].
However, in view of the spatial variation in grain structure across the width of the
crystallised islands (shown in Fig. 7.45b), there must be some concern, although
earlier work on unpatterned films showed better uniformity than ELA [41].
There is clearly a trade-off in scan speed and material quality with this process,
and this issue is compared with SLS and pulsed Nd:YAG crystallisation in the next
section.
Fig. 7.47 Electron field effect mobility (for parallel and perpendicular current flow) and grain
structure as a function CW Nd:YVO4 laser scan speed. (Reprinted with permission from [39].
7.5.4 Comparison of Large Grain Crystallisation Systems
All the crystallisation systems considered in this section can produce large grain,
high quality TFT material, as well as being revealing tools with which to inves-
tigate the crystallisation process itself. Where a direct comparison can be made,
they also offer a solution to some of the technical issues with conventional ELA,
such as the size of the processing window and the associated pulse stability
problem, and, in the case of the solid state lasers, they offer lower cost of own-
ership. These latter points are largely manufacturing issues, and a key consider-
ation in this industrial context is plate throughput. To a large extent, throughput
will scale with both the power of the system, which will determine the beam area
and shot frequency, and with the efficiency of the usage of this power, which will
also embrace the number of shots in the process. Taking these simple consider-
ations into account, the crystallisation rate, R, can be calculated using: R ¼ ANb f for
pulsed lasers, and R ¼ LAminbv
for the CW laser. Where Ab is the beam area on the
plate, f is the pulse frequency, N is the number of shots in the process, v is the plate
scan velocity, and Lmin is the small axis beam dimension.
The key features of the ELA [18], SLS [32, 94], pulsed Nd:YAG [35] and CW
Nd:YVO4 [39] systems are compared in Table 7.2, where a first order estimate of
throughput is provided by the crystallisation rate, R. It should be emphasised that
this is only a first order estimate as it does not account for the deceleration period
of the plate translation stage at the end of each scan, and its movement and
acceleration intervals for the next scan. In addition, the smaller the beam
dimensions, the more scans will be needed to crystallise a whole plate, and the
more significant the scan interruptions will be.
From this table, it is clear that the throughput of the Nd:YAG system is sig-
nificantly lower than the ELA system, although it offers a larger process window,
better pulse stability and lower cost of ownership. To what extent these beneficial
features offset its lower throughput would need to be determined in a manufac-
turing environment. However, the overall system costs will play a role, as these
determine whether it is economic to have a larger number of Nd:YAG systems as a
way of maintaining the required plate processing capacity. It is also interesting to
Table 7.2 Comparison of crystallization rates in different laser annealing systems

Laser system Power Beam size Freq Shot Step Scan Crystn
(W) (cm 9 cm) (Hz) no. size vel rate
(lm) (cm/s) (cm2/s)
ELA [18] 600 46.5 9 3.5 9 10-2 600 20 17.5 48.8
SLS [32, 315 2.5 9 0.15 300 2 2.5 56.3
100]
Nd:YAG 200 10.5 9 4.0 9 10-3 4000 20 2 8.4
[35]
Nd:YVO4 10 4.0 9 10-2 9 2.0 9 10-3 CW 50 2
[39]
note that the crystallisation rate figures for these two systems do not simply scale
with system power (for the same shot number), as there is a further discrepancy by
a factor of two when normalising R by power. Whilst, the overall optical efficiency
of the two beam shaping systems will play a role here, there is also reduced
coupling efficiency of the 532 nm radiation into the thin film because of its larger
absorption depth. Hence, if a given absorbed energy density is required to melt a
film of a given thickness, a higher incident density will be required with the
532 nm laser. For a given maximum output pulse energy, the incident energy
density will be determined by the final beam dimensions, Ab, on the plate, and this
directly controls R.
Given the sub-optimal absorption from green lasers, the use of a solid state,
445 nm blue laser has been reported [105]. The radiation at this wavelength will be
more strongly absorbed, but this is an undeveloped process, with no device results
reported to date.
The CW Nd:YVO4 system produces large grain material comparable to the
2-shot SLS process, and avoids the SLS system complexity of projection masks
and the high resolution optical system required to produce 4–5 lm beamlets.
However, its crystallisation rate is *25 times lower than the SLS system. In
recognition of this low basic crystallisation rate, a high throughput system has
been designed [41], in which a number of laser sub-beams were used to simul-
taneously, and selectively, irradiate just the TFT areas of a display plate. This is
shown in Fig. 7.48, and a crystallisation rate of 48 cm2/s was calculated using
sixteen 30–50 lm sub-beams, scanned at 2 m/s, in order to crystallise pixel TFTs
at a pitch of 148 lm [41]. At this high scan speed, the electron mobility was
*150 cm2/Vs (see Fig. 7.47), which is perfectly adequate for pixel TFTs. The
peripheral driver TFTs occupy a much smaller area than the display area, but a
higher carrier mobility is beneficial. For these areas, four 150 lm sub-beams were
scanned at 50 cm/s, delivering a crystallisation rate of 2.4 cm2/s, and produced
TFTs with an electron mobility of 400–600 cm2/Vs.
At the time of writing, in spite of these solid-state laser alternatives, ELA
continues to be the dominant crystallisation process for poly-Si TFTs.
Fig. 7.48 Illustration of

multiple sub-beam CW
Nd:YVO4 laser system for
high throughput
crystallisation. (Reprinted
SID)
Fig. 7.49 Illustration of active matrix display cells showing drive circuits: a a-Si:H TFT display
with externally mounted c-Si ICs, and b poly-Si TFT display with monolithically integrated poly-
Si drive circuits
7.6 Poly-Si Applications
In order to deliver the appropriate signals to the pixels in an AMLCD (or AM-
OLED display), using line-at-a-time addressing (see Chap. 4 for a detailed dis-
cussion of active matrix display operation) it is necessary to apply appropriate
drive signals to the rows and columns of a display. For an a-Si:H display, where
the switching speed of the a-Si:H TFTs is too slow to perform these functions, it is
necessary to mount external silicon ICs around the edge of the display, as shown
schematically in Fig. 7.49a. These circuits may be directly attached to the plate, or
connected to the plate by flexible foils, upon which the ICs are mounted (as
discussed in Sect. 4.5.4). The number of these ICs will be determined by the
resolution of the display, i.e. the number of rows and columns to be addressed, and
this will feed through into the cost of the display module in terms of the direct chip
costs plus their mounting costs. The higher carrier mobility in poly-Si enables the
direct fabrication of these addressing circuits on the active plate, as illustrated in
Fig. 7.49b, and thereby reduce the module cost.
However, the cost saving due to the integration of the drive circuitry needs to be
balanced against the increased cost of manufacturing the poly-Si active matrix
plate. This is due to the greater number of masking and deposition stages (typically
9-mask stages, as shown in Sect. 7.4.2), as opposed to the 5-mask process for a-
Si:H AMLCDs (see Sect. 5.3). In addition, there is the specific extra capital
equipment needed to make poly-Si TFTs, principally the laser crystallisation and
the ion doping systems (as discussed in Sect. 7.4.2), as well as extra aligners and
deposition systems to give a balanced manufacturing process with the higher mask
and deposition process count. Finally, there is a possible difference in yield in the
more complex poly-Si process, where yield modelling is handled by assuming that
the yield, Y, is given by a Poisson distribution of random defects, of density D, in a
defect-sensitive area, A [106]:
Y ¼ exp DA
and D is given by Nd, where d is the defect density per photolithographic step, and
N is the number of steps. Hence, more mask stages can be expected to reduce
process yield and add to unit costs.
7.6 Poly-Si Applications 243

qualitative cost models, for a-
Si:H displays with external
drive circuits, and poly-Si
displays with integrated drive
circuits
Whilst detailed a detailed cost model, based on the evaluation of the above
factors, is beyond the scope of this book, a qualitative model can give insight into
the essential features of the price comparison, and this is shown in Fig. 7.50. In
this figure, the cost of an individual display is plotted as function of the display
diagonal, where it is assumed that there is a fixed cost for processing a glass plate
of a given size, using either the poly-Si or the a-Si:H process. As shown, the
smaller the display diagonal, the more displays there will be per glass plate, and,
hence, the cost per display scales with its size. The slopes of the lines represent the
different plate manufacturing costs for the low mask-count a-Si:H process and the
higher mask-count poly-Si process. The intercept of these lines on the y-axis is
determined by the number and cost of the externally added ICs, which is governed
by the display resolution rather than by the display diagonal. Hence, these circuits
represent a higher relative fixed cost as the display diagonal reduces, and these
considerations demonstrate that poly-Si AMLCDs are most cost effective for small
display diagonals. A cost cross-over point at 8-10 inch diagonals is shown as an
approximate representation of the current situation, but this will vary with future
IC costs and changes in the economics of plate manufacturing.
Thus, the major current application for poly-Si AMLCDs is in the high volume,
‘mobile’ display market (mobile/smart phones, digital camera view-finders, per-
sonal media players etc.), where circuit integration offers further advantages over
and above the cost benefit. For instance, these portable displays need to be rugged,
light-weight, and compact, and the poly-Si integrated circuits eliminate the risks of
IC chip de-lamination due to mechanical shock. They also reduce the size of the
glass module itself by reducing the space reserved for chip bonding. Finally, for
higher resolution, small diagonal displays, the reduced pixel pitch can become a
limiting factor for aligning and bonding the external ICs, but can be more easily
met with the integrated drivers.
The poly-Si TFT CMOS circuits currently integrated into poly-Si displays
include the row and column shift registers, level shifting circuits, 6-bit D/A
converters for the column drivers, charge pumps and control logic [2, 11, 107,
108]. In contrast to these n- and p-channel CMOS circuits, which require the high
mask count discussed above, it has been suggested that a lower level of circuit
integration could be achieved at a lower processing cost, by using just p-channel
TFTs [109]. This would reduce the mask count by three masks, by eliminating the
separate masking stages for the n- and p-channel dopant locations, since the gate
electrode alone is sufficient to define the source and drain regions in single channel
SA TFTs. In addition, the LDD mask can be removed as the superior drain bias
stability of p-channel TFTs obviates the need for drain field relief. These process
simplifications can lead to a more cost competitive technology compared with a-
Si:H for large diagonal displays, albeit at the expense of reduced circuit perfor-
mance. This is because single channel inverters conventionally have a greater
power dissipation, smaller dynamic range and slower switching speed than CMOS
inverters, but various single channel circuit alternatives have been proposed to
reduce these performance limitations [110].
The other emerging application for poly-Si TFTs is in small/medium diagonal
AMOLED displays, particularly for smart phones. At the moment, poly-Si is the
preferred pixel transistor in these commercial displays, as it can provide the higher
drive currents needed for OLEDs. Moreover, this application requires a high duty
cycle on-current, and, under these conditions, poly-Si TFTs also have better bias
stability than a-Si:H TFTs. However, amorphous oxide semiconductor, AOS,
TFTs, using, for instance, a-indium-gallium-zinc-oxide, have also been identified
as potential candidates for commercial AMOLED products [111], and a compar-
ison of the stability of the major TFT technologies for this application is shown in
Fig. 9.25 (in Sect. 9.4.3.3). For a fuller discussion of the background issues
associated with AMOLED pixel design, see Sect. 4.6.2.2.
7.7 Summary
Poly-Si TFTs are now in mass production, particularly for small/medium diagonal,
portable displays, containing integrated drive circuits. The circuit integration is
facilitated by the high carrier mobility, and this has been achieved through a
number of technology evolutions. The current, commercial fabrication process
relies upon excimer laser crystallisation of a-Si:H pre-cursor layers, where the
preferred excimer laser wavelength is 308 nm, which is obtained from a XeCl gas
mixture. These are pulsed lasers, with a pulse duration of *30 ns, and the com-
bination of UV wavelength and short pulse duration enables the a-Si film to be
melted without undue heat transmission into the underlying glass substrate. The
resulting grain structure displays a complex dependence upon the fractional
melting of the film, where the optimum grain structure, and device performance,
results from almost full melting of the a-Si films. Under these conditions, high
quality *300 nm sized grains are formed, and the resulting electron field effect
mobility lies between *100 cm2/Vs and *250 cm2/Vs, depending upon the
number of laser shots in the process. The most uniform films, giving the highest
performance devices, result from high shot number processing, but this gives the
lowest throughput. Hence there is a trade-off between uniformity, performance and
throughput, with typical manufacturing processes employing 20–30 shot
processing.
7.7 Summary 245
In view of these well understood issues, alternative lasers and crystallisation

procedures have been reported, including green solid-state lasers, as well as
modified excimer laser technologies. These have all been able to produce larger
grain material, with increased electron mobility values, which, in some cases,
approach single crystal values. In addition to these laser-based techniques, there
has also been interest in enhanced SPC procedures, using a metal catalyst to
increase the speed of this inherently slow process. This is fundamentally a
homogeneous large area process, which offers improved large area uniformity in
contrast to the laser techniques. With a laser, the beam area itself is always far
smaller than the substrate plate, and, therefore, beam uniformity, pulse-to-pulse
energy fluctuations and pulse overlap effects all need to be carefully controlled to
produce acceptable uniformity. The favoured metal for the enhanced SPC process
is nickel, because it has a silicide phase, which is well lattice-matched to Si. This
seeds the crystallisation of a-Si by a process variously referred to as silicide-
mediated crystallisation (SMC) or metal-induced crystallisation (MIC). TFTs
produced using this process have tended to have high off-state leakage currents,
which have limited their practical application. However, it has been established
that a subsequent excimer laser crystallisation stage can resolve that problem, and
produce high performance TFTs. Nevertheless, in spite of these widely investi-
gated alternatives, the current industrial production of poly-Si TFTs is still largely
based upon excimer laser crystallisation.
The prevailing architecture of poly-Si TFTs is top-gated, and self-aligned (SA),
with an SiO2 gate dielectric, all of which are more similar to the semiconductor
industry’s silicon-on-insulator MOSFETs, rather than to the AMLCD industry’s
inverted staggered a-Si:H TFTs. In addition, this structure is implemented using
ion doping for the source and drain dopants. However, the use of high-energy
phosphorus ions, for the n-channel TFTs, is potentially damaging to the lattice
within the doped regions. This damage needs to be carefully controlled to mini-
mise device performance artefacts, namely increased leakage current and reduced
on-state currents. For some purposes, particularly for basic materials investigation,
a non-self-aligned structure (which is also top-gated and ion doped) offers a
simpler process, and avoids the ion damage issues of the SA process. However, the
SA process is necessary for short channel TFTs, and is the standard industry
architecture. The other architectural feature specific to n-channel poly-Si TFTs is a
lightly n--doped region adjacent to the drain junction, which is needed to reduce
the drain field, and, thereby, suppress drain-bias-stress induced hot carrier insta-
bilities. It also diminishes other performance artefacts associated with a high field
at the drain, in particular field enhanced leakage currents and poor current satu-
ration of the output characteristics. These latter device performance issues are
discussed in detail in Chap. 8.
Single channel TFTs of either architecture can be produced using a four-mask
process. However, for a poly-Si AMLCD, with integrated complementary TFT
drive circuits, a higher mask count than for a-Si:H AMLCDs is necessary, and a
nine-mask process is illustrated. This higher mask count, and a more complex
process, influences the commercial application of poly-Si TFTs, and a simple,
qualitative cost model illustrates why the current application is limited to small/
medium diagonal displays.
In summary, poly-Si TFTs are now in mass-production, particularly for small/
medium diagonal displays, where the integration of drive circuitry can give both
cost and performance benefits. The circuit integration is facilitated by the high
carrier mobility in poly-Si, and this can be achieved in both p- and n-channel
TFTs, which enables complementary TFT circuit design. Within the small/medium
display market, poly-Si accounts for *40 % of the AMLCD product revenues,
and dominates the current smart phone AMOLED market.
References
1. Mitani M, Endo T, Taniguchi Y, Katou T, Shimoto S, Ohno T, Tsuboi S, Okada T, Azuma

K, Kawachi G, Matsumura M (2008) Ultrahigh performance polycrystalline silicon thin-
film transistors on excimer-laser-processed pseudo-single-crystal films. Jpn J Appl Phys
47(12):8707–8713
2. Nishibe T, Nakamura T (2004) Realization of new-concept Input Display by p-Si SOG
technology. Proc AMLCD’04:85–88
3. Harada K, Kimura H, Miyatake M, Kataoka S, Tsunashima T, Motai T, Kawamura T (2009)
A novel low-power-consumption all-digital system-on-glass display with serial interface.
SID’09 Digest, pp 383–386
4. Q3’09 Quarterly small/medium shipment and forecast report. Displaysearch, Oct 2009
5. http://www.displaysearch.com/cps/rde/xchg/displaysearch/hs.xsl/
120105_amoled_display_revenue_to_surge_in_2011_from_smart_phone_adoption.asp
(Accessed Feb 2012)
6. Parrillo LC (1983) VLSI process integration. In: Sze SM (ed) VLSI technology. McGraw-
Hill, New York
7. Lu NC-C, Lu C-Y (1984) I-V characteristics in polysilicon resistors at high electric field and
the non-uniform conduction mechanism. Solid State Electron 27(8/9):797–805
8. Seto JYW (1975) The electrical properties of polycrystalline silicon films. J Appl Phys
46(12):5247–5254
9. Morozumi S, Oguchi K, Misawa T, Araki R, Ohshima H (1984) 4-25-in. and 1.51-in. B/W
and Full-Color LC Video displays addressed by Poly-Si TFTs. SID’84 Digest, pp 316–319
10. Levinson J, Shepherd FR, Scanlon PJ, Westwood WD, Este G, Rider M (1982) Conductivity
behaviour in polycrystalline semiconductor thin film transistors. J Appl Phys 53(2):1193–
1202
11. Moon K-C, Kim C-H, Kim C-M, Park K-C, Joo S-Y, Kim I-G, Park T-H, Kang J-K, Chung
E-J, Kim C-W (2004) 2-inch qVGA SOG-LCD employing TS-SLS. Proc AMLCD’04, pp
21–24
12. Brotherton SD, Ayres JR, Young ND (1991) Characterisation of low temperature poly-Si
thin film transistors. Solid-State Electron 34(7):671–679
13. Mimura A, Konishi N, Ono K, Ohwada J-I, Hosokawa Y, Ono YA, Suzuki T, Miyata K,
Kawakami H (1989) High performance low-temperature poly-Si n-channel TFTs for LCD.
IEEE Trans ED-36(2):351–359
14. Hatalis MK, Greve DW (1988) Large grain polycrystalline silicon by low-temperature
annealing of low pressure chemical vapour deposited amorphous silicon films. J Appl Phys
63(7):2260–2266
References 247
15. Brotherton SD, Young ND, Edwards MJ, Gill A, Trainor MJ, Ayres JR, Clarence IR, Bunn
RM, Gowers JP (1994) Low temperature furnace processed poly-Si AMLCDs. Proc SID
IDRC, pp 130–133, (Monterey, Ca, 1994)
16. Iverson RB, Reif R (1987) Recrystallisation of amorphised polycrystalline films on SiO2:
temperature dependence of the crystallisation parameters. J Appl Phys 62(5):1675–1681
17. Brotherton SD (1995) Polycrystalline silicon thin film transistors. Semicond Sci Technol
10:721–738
18. Paetzl R, Brune J, Herbst L, Simon F, Turk BA (2009) Advanced laser crystallisation for
active-matrix display manufacturing. Proceeding of 5th international TFT conference
ITC’09, 10.2, France
19. Forouhi AR (1985) Optical functions of a-Si and a-Si:H. In: Properties of amorphous silicon
(EMIS data reviews No. 1). IEE Inspec, London
20. Aspnes DE (1988) Optical functions of intrinsic Si. In: Properties of silicon (EMIS data
reviews No. 4). IEE Inspec, London
21. Baeri P, Campisano SU (1982) Chapter 4. In: Poate JM, Mayer JW (eds) Laser annealing of
semiconductors, Academic Press, New York
22. De Unamuno S, Fogarassy E (1989) A thermal description of the melting of c- and a-silicon
under pulsed excimer lasers. Appl Surf Sci 36:1–11
23. Brotherton SD, McCulloch DJ, Clegg JB, Gowers JG (1993) Excimer-laser-annealed poly-
Si thin film transistors. IEEE Trans ED-40(2):407–413
24. Fogarassy E, Prevot B, De Unamano S, Elliq M, Pattyn H, Mathe EL, Naudon A (1993)
Pulsed laser crystallisation of hydrogen-free a-Si thin films for high-mobility poly-Si TFT
fabrication. Appl Phys A56:365–373
25. Brotherton SD, McCulloch DJ, Gowers JP, Ayres JR, Trainor M (1997) Influence of melt
depth in laser crystallised poly-Si thin film transistors. J Appl Phys 82(8):4086–4094
26. Leonard JP, Bessette MA, Gupta VV, Im JS (1997) The effect of film thickness and pulse
duration variation in excimer laser crystallisation of thin Si films. Mat Res Symp Soc
452:947–952
27. Im JS, Kim HJ, Thompson MO (1993) Phase transformation mechanisms in excimer laser
crystallisation of amorphous silicon films. Appl Phys Lett 63(14):1969–1971
28. Im JS, Kim HJ (1994) On the super lateral growth phenomenon observed in excimer laser-
induced crystallisation of thin Si films. Appl Phys Lett 64(17):2303–2305
29. Im JS, Crowder MA, Sposili RS, Leonard JP, Kim HJ, Yoon JH, Gupta VV, Jin Song H,
Cho HS (1998) Controlled super-lateral growth of Si films for microstructural manipulation
and optimisation. Phys Stat Sol (a) 166(2):603–617
30. Kim HJ, Im JS (1994) Multiple pulse irradiations in excimer laser-induced crystallisation of
amorphous Si films. Mat Res Soc Symp Proc 321:665–670
31. Brotherton SD, McCulloch DJ, Gowers JP, French ID, Gale I (2000) Issues in laser
crystallisation of poly-Si. Proc IDMC’2000, Seoul, pp 65–69
32. Herbst L, Simon F, Rebhan U, Osmanov R, Fechner B (2004) New technology for creation
of LTPS with excimer laser annealing. Proc Asia Display/IMID’04, Korea, pp 319–321
33. Brotherton SD, McCulloch DJ, Gowers JP (2004) Influence of excimer laser beam shape on
poly-Si crystallisation. Jpn J Appl Phys 43(8):5114–5121
34. Brotherton SD, McCulloch DJ, Gowers JP, Ayres JR, Fisher CA, Rohlfing FW (2000)
Excimer laser crystallisation of poly-Si TFTs for AMLCDs. Mat Res Soc Symp Proc
621:Q7.1.1–Q7.1.12
35. Tamagawa K, Ikeda H, Oonishi T, Yamamoto Y, Kikuchi M, Morimura T, Hashimoto M,
Asari S (2006) Green laser annealing system for manufacturing LTPS TFTs. Ulvac Tech J
64E:32–36
36. Zan HW, Huang CY, Saito K, Tamagawa K, Chen J, Wu TJ (2006) Crystallization
mechanism of Poly-Si thin film using high-power Nd:YAG Laser with Gaussian Beam
profile. Mater Res Soc Symp Proc 910:A14–03
37. Morikawa K, Okamoto T, Kojima T, Yura S, Nishimae J, Sato Y, Tanaka M, Inoue M

(2004) Comparison of Poly-Si TFT characteristics crystallized by a YAG2x laser and an
excimer laser. SID 04 Digest, pp 1088–1091
38. Helen Y, Gautier G, Mohammed-Brahim T, Bonnaud O, Andra G, Bergmann J, Falk F
(2001) High field effect mobility polysilicon thin film transistors involving frequency
doubled cw-Nd:YAG-laser crystallisation. Proc AMLCD-01, pp 187–188
39. Hara A, Takei M, Takeuchi F, Suga K, Yoshino K, Chida M, Kakehi T, Ebiko Y, Sano Y,
Sasaki N (2004) High performance low temperature polycrystalline silicon thin film
transistors on non-alkaline glass produced using diode pumped solid state continuous wave
laser lateral crystallization. Jpn J Appl Phys 43(4A):1269–1276
40. Hara A, Takeuchi F, Takei M, Suga K, Yoshino K, Chida M, Sano Y, Sasaki N (2002)
High-performance polycrystalline silicon thin film transistors on non-alkali glass produced
using continuous wave laser lateral crystallization. Jpn J Appl Phys 41(3B):L311–L313
41. Sasaki N, Hara A, Takeuchi F, Mishima Y, Kakehi T, Yoshino K, Takei M (2002) High
throughput CW-laser lateral crystallization for low-temperature poly-Si TFTs and
fabrication of 16 bit SRAMs and 270 MHz shift registers. SID 02 Digest, pp 154–157
42. Hara A, Takeuchi F, Takei M, Yoshino K, Suga K, Sasaki N (2001) Ultra-high performance
poly-Si TFTs on a glass by a stable scanning CW laser lateral crystallization. Proc
AMLCD’01, pp 227–230
43. Jang J (2004) Poly-Si TFTs by non-laser crystallization methods. In: Kuo Y (ed) Thin film
transistors, materials and processes: polycrystalline silicon thin film transistors, vol 2.
Kluwer Academic publishers, Dordrecht
44. Hayzelden C, Batstone JL (1993) Silicide formation and silicide-mediated crystallization of
nickel-implanted amorphous silicon thin films. J Appl Phys 73(12):8279–8289
45. Yoon SY, Kim KH, Kim CO, Oh JY, Jang J (1997) Low temperature metal induced
crystallization of amorphous silicon using a Ni solution. J Appl Phys 82(11):5865–5867
46. Van der Zaag PJ, Verheijen m A, Yoon SY, Young ND (2002) Explanation for the leakage
current in polycrystalline-silicon thin-film transistors made by Ni-silicide mediated
crystallization. Appl Phys Lett 81(18):3404–3406
47. Meng Z, Zhao S, Wu C, Zhang B, Wong M, Kwok H–S (2006) Polycrystalline silicon films
and thin-film transistors using solution-based metal-induced crystallization. IEEE J Disp
Technol 2(3):265–273
48. Lee J-S, Kim M-S, Kim D, Kim Y-M, Moon J, Joo S-K (2009) Fabrication and
characterization of low temperature polycrystalline silicon thin film transistors by ink-jet
printed nickel-mediated lateral crystallization. Appl Phys Lett 94(12):122105-1–122105-3
49. Lee S-W, Joo S-K (1996) Low temperature poly-Si thin- film transistor fabrication by
metal-induced lateral crystallization. IEEE Electron Dev Lett 17(4):160–162
50. Wong M, Jin Z, Bhat GA, Wong PC, Kwok HS (2000) Characterization of the MIC/MILC
interface and its effects on the performance of MILC thin-film transistors. IEEE Trans
Electron Dev 47(5):1061–1067
51. Meng Z, Wang M, Wong M (2000) High performance low temperature metal-induced
unilaterally crystallized polycrystalline silicon thin film transistors for System-on-Panel
applications. IEEE Trans Electron Dev 47(2):404–409
52. Murley D, Young N, Trainor M, McCulloch D (2001) An investigation of laser annealed
and metal-induced crystallized polycrystalline silicon thin-film transistors. IEEE Trans
Electron Dev 48(6):1145–1151
53. Yoon SY, Young N, van der Zaag PJ, McCulloch D (2003) High-performance poly-Si TFTs
made by Ni-mediated crystallization through low-shot laser annealing. IEEE Electron Dev
Lett 24(1):22–24
54. Sakamoto H, Makita N, Hijikigawa M, Osame M, Tanada Y, Yamazaki S (2000) 2.6 inch
HDTV panel using CG Silicon. SID 00 Digest, pp 1190–1193
55. Mizuki T, Matsuda JS, Nakamura Y, Takagi J, Yoshida T (2004) Large domains of
continuous grain silicon on glass substrate for high-performance TFTs. IEEE Trans Electron
Dev 51(2):204–211
References 249
56. Batey J, Tierney E (1986) Low-temperature deposition of high-quality silicon dioxide by

plasma-enhanced chemical vapor deposition. J Appl Phys 60(9):3136–3145
57. Hirashita N, Tokitoh S, Uchida H (1993) Thermal desorption and infrared studies of
plasma-enhanced chemical vapor deposited SiO films with tetraethylorthosilicate. Jpn J
Appl Phys 32(4):1787–1793
58. Young ND, Gill A (1992) Water-related instability in TFTs formed using deposited gate
oxides. Semicond Sci Technol 7:1103–1108
59. Asari S, Kurata T, Kikuchi T, Hashimoto M, Saito K (2002) Evaluation of gate insulation
film for large-scale substrate in low temperature poly-Si TFTs. Proc IDMC’02
60. Young ND, Harkin G, Bunn RM, McCulloch DJ, French ID (1996) The fabrication and
characterisation of EEPROM arrays on glass using a low-temperature poly-Si TFT process.
IEEE Trans ED-43(11):1930–1936
61. Young ND, McCulloch DJ, Bunn RM, French ID, Gale IG (1998) Low temperature poly-Si
on glass and polymer substrates. Proc Asia Display’98, pp 83–93
62. Okumura F, Yuda K (1998) High-quality low-temperature gate oxide for poly-Si TFTs.
Electrochem Soc Proc 98–22:133–142
63. Higashi S, Abe D, Hiroshima Y, Miyashita K, Kawamura T, Inoue S, Shimoda T (2002)
High-quality SiO2/Si interface formation and its application to fabrication of low-
temperature-processed polycrystalline Si thin-film transistor. Jpn J Appl Phys
41(6a):3646–3650
64. Maiolo L, Pecora A, Fortunato G, Young ND (2006) Low-temperature electron cyclotron
resonance plasma-enhanced chemical-vapor deposition silicon dioxide as gate insulator for
polycrystalline silicon thin-film transistors. J Vac Sci Technol A 24(2):280–285
65. Young ND Private communication, unpublished results
66. Lee J-H, Shin M-Y, Moon KC, Han M-K (2004) Reliable poly-Si TFT employing XeCl
excimer laser irradiation on gate oxide. Phys Scr T114:199–201
67. Miyashita M, Sugawara Y, Uraoka Y, Yano H, Hatayama T, Fuyuki T, Sameshima T (2005)
Improvement of reliability in low-temperature Poly-Si thin film transistors by water vapor
annealing. Proc AM-LCD’05, pp 277–280
68. Wilk GD, Wallace RM, Anthony JM (2001) High-k gate dielectrics: current status and
materials properties considerations. J Appl Phys 89(10):5243–5275
69. O’Sullivan MP, Young ND, French ID, Wilks RW (2003) Low threshold voltage poly-Si
TFTs formed using silicon dioxide/silicon nitride dual layer dielectric. Proc IDW’03, pp
443–445
70. O’Sullivan MP, Young ND, Glass C, Wilks RW (2003) Low threshold voltage poly-Si TFTs
formed using Ta2O5 gate dielectric. Proc IDW’03, pp 335–338
71. Ryssel H, Ruge I (1986) Ion implantation, chapters 2 and 9. Wiley, New York
72. Brotherton SD, Ayres JR, Fisher CA, Glaister C, Gowers JP, McCulloch DJ, Trainor MJ
(1998) The technology and application of laser crystallised poly-Si TFTs. In: Kuo Y (ed)
Thin film transistor technologies IV, proceedings, vol 98–22. Electrochemical Society, New
Jersey, pp 25–42
73. Yoneda K (1997) State-of-the-art low temperature processed poly-Si TFT technology. Proc
SID-IDRC’97, pp M40–M47
74. Peng D-Z, Chang T-C, Zan H-W, Huang T-Y, Chang C-Y, Liu P-T (2002) Reliability of
laser-activated low-temperature polycrystalline silicon thin-film transistors. Appl Phys Lett
80(25):4780–4782
75. Park K-C, Nam W-J, Kang S-H, Han M-K (2004) Incomplete laser annealing of ion doping
damage at source/drain junctions of poly-Si thin-film transistors. Electrochem Solid-State
Lett 7(6):G116–G118
76. Brotherton SD, Lee S-G, Glasse C, Ayres JR, Glaister C (2002) Short channel poly-Si TFTs.
Proc IDW’02, pp 283–286
77. Valletta A, Mariucci L, Fortunato G, Brotherton SD (2003) Surface scattering effects in
polycrystalline silicon thin film transistors. Appl Phys Lett 82(18):3119–3121
78. Nam W-J, Park K-C, Jung S-H, Park S-J, Han M-K (2003) Observation and annealing of
incomplete recrystallized junction defects due to the excimer laser beam diffraction at the
gate edge in poly-Si TFT. Mat Res Soc Symp Proc 762:A17.8
79. Ayres JR, Brotherton SD, McCulloch DJ, Trainor M (1998) Analysis of drain field and hot
carrier stability of poly-Si TFTs. Jpn J Appl Phys 37(4a):1801–1808
80. Young ND, Gill A, Edwards MJ (1992) Hot carrier degradation in low temperature
processed polycrystalline silicon thin film transistors. Semicond Sci Technol 7(9):1183–
1188
81. Young ND (1996) The formation and annealing of hot-carrier-induced degradation in poly-
Si TFTs, MOSFETs, and SOI devices, and similarities to state-creation in a-Si:H. IEEE
Trans ED-43(3):450–456
82. Rohlfing FW, Ayres JR, Brotherton SD, Fisher CA, McCulloch DJ (2000) Fabrication and
characterisation of poly-Si TFTs with self-aligned lightly-doped drain. Proc 20th SID-
IDRC, pp 119–122
83. Yoshinouchi A, Morita T, Itoh M, Yoneda H, Yamane Y, Yamamoto Y, Tsucimoto S,
Funada F, Awane K (1996) Process technologies for monolithic low-temperature poly-Si
TFT-LCDs. Proc SID EuroDisplay’96, pp 29–32
84. Glasse C, Brotherton SD, French ID, Green PW, Rowe C (2003) Short channel TFTs made
with sidewall spacer technology. Proc AMLCD’03, pp 317–320
85. Gosain DP, Westwater J, Usui S (1995) High performance bottom gate TFTs by excimer
laser crystallisation and post hydrogenation. J Appl Phys 34(1):937–941
86. Tsai C-C, Chen H-H, Chen B-T, Cheng H-C (2007) High-performance self-aligned bottom-
gate low-temperature poly-silicon thin-film transistors with excimer laser crystallization.
IEEE Electron Device Lett 28(7):599–602
87. Mariucci L, Fortunato G, Bonfiglietti A, Cuscuná M, Pecora A, Valletta A (2004)
Polysilicon TFT structures for kink-effect suppression. IEEE Trans ED-51(7):1135–1142
88. Mei P, Boyce JB, Hack M, Lujan RA, Johnson RI, Anderson GB, Fork DK, Ready SE
(1994) Laser dehydrogenation/crystallization of plasma-enhanced chemical vapor deposited
amorphous silicon for hybrid thin film transistors. Appl Phys Lett 64(9):1132–1134
89. Brotherton SD, McCulloch DJ, Edwards MJ (1994) Beam shape effects with excimer laser
crystallisation of PECVD and LPCVD a-Si. Solid State Phenom 37–38:299–304
90. Kikuchi M (2006) Plasma CVD system for manufacturing LTPS TFTs. Ulvac Tech J
64E:30–31
91. Suzuki T (2006) Flat panel displays for ubiquitous product applications and related impurity
doping technologies. J Appl Phys 99:111101-1–111101-15
92. Sposili RS, Im JS (1996) Sequential lateral solidification of thin silicon films on SiO2. Appl
Phys Lett 69(19):2864–2866
93. Voutsas AT (2003) Assessment of the performance of laser-based lateral-crystallisation
technology via analysis and modelling of polysilicon thin film transistor technology. IEEE
Trans Ed-50(6):1494–1500
94. Park J-Y, Park H-H, Lee K-Y, Chung H-K (2004) Design of sequential lateral solidification
crystallisation method for low temperature poly-Si thin film transistors. Jpn J Appl Phys
43(4a):1280–1286
95. Oh C-H, Matsumura M (1998) Preparation of location-controlled crystal-silicon islands by
means of excimer-laser annealing. Proc AMLCD’98, pp 13–16
96. Endo T, Taniguchi Y, Katou T, Shimoto S, Ohno T, Azuma K, Matsumura M (2008)
Pseudo-single-nucleus lateral crystallization of Si thin films. Jap J Appl Phys 47(3):1862–
1866
97. Taniguchi Y, Katou T, Hiramatsu M, Matsumura M (2008) New phase modulators for next-
generation low-temperature crystallization method of Si films. Jap J Appl Phys
45(10A):7753–7760
98. Van der Wilt PC, van Dijk BD, Bertens GJ, Ishihara R, Beenakker CIM (2001) Formation
of location-controlled crystalline islands using substrate-embedded seeds in excimer-laser
crystallization of silicon films. Appl Phys Lett 79(12):1819–1821
References 251
99. Rana V, Ishihara R, Hiroshima Y, Abe D, Inoue S, Shimoda T, Metselaar W, Beenakker K

(2005) Dependence of single-crystalline Si TFT characteristics on the channel position
inside a location-controlled grain. IEEE Trans ED-52(12):2622–2628
100. Turk BA, Simon F, Pätzel R (2007) 2D-projection-based method for forming uniform and
high-quality poly-Si films via sequential lateral solidification. Proc 3rd international TFT
conference, ITC’07, Rome, pp 78–81
101. Crowder MA, Limanov AB Im JS (2000) Sub-grain boundary spacing in directionally
crystallized Si films obtained via sequential lateral solidification. Proc Mat Res Soc Symp
621E:Q9.6.1–Q9.6
102. Chung UJ, Limanov AB, van der Wilt, PC, Chitu AM, Im JS (2006) Microstructural
analysis of directionally solidified materials obtained via line-scan SLS of Si films. Proc
IMID/IDMC’06, pp 1744–1749
103. Brotherton SD, Crowder MA, Limanov AB, Turk B, Im JS (2001) Characterisation of poly-
Si TFTs in directionally crystallized SLS Si. Proc Asia Display/IDW’01, pp 387–390
104. Crowder MA, Voutsas AT, Droes SR, Moriguchi M, Mitani Y (2004) Sequential lateral
solidification processing for polycrystalline Si TFTs. IEEE Trans ED-51(4):560–568
105. Shirai K, de Dieu Mugiraneza J, Suzuki T, Okada T, Noguchi T, Matsushima H, Hashimoto
T, Ogino Y, Sahota E (2011) Crystallization behavior of sputtered amorphous silicon films
by blue-multi-laser-diode annealing. Jpn J Appl Phys 50:021402
106. Young BE, Young R (1998) Cost and yield model comparison for a-Si and poly-Si displays.
SID’98 Digest, pp 1193–1196
107. Ayres JR, Edwards MJ, Young ND, Pook R, Pearson AD, Henzen AV (2002) A highly
integrated AMLCD with low voltage operation. Proc 2nd international display
manufacturing conference. IDMC 2002:19–22
108. Hashimoto K, Yamashita K, Matsuki F, Watsuda H, Ayres JR, Edwards MJ (2005) A high
resolution LTPS AMLCD with integrated DACs. Proc IDW’05, pp 1031–1032
109. Ha Y-M (2000) P-type technology for large size low temperature poly-Si TFT-LCDs. SID
symposium digest of technical papers 31(1):1116–1119
110. Hong S, Kim J, Choi B-D (2009) A high-speed and low-power inverter circuit using
p-channel metal oxide semiconductor low-temperature polycrystalline silicon thin film
transistors. Jpn J Appl Phys 48(3):03B010_1–03B010_5
111. http://www.displaysearch.com/cps/rde/xchg/displaysearch/hs.xsl/120106_oled_display_
technology_moving_to_compete_in_the_tv_market.asp (Accessed Jan 2012)
Chapter 8
Poly-Si TFT Performance
Abstract This chapter deals with the electrical characterisation, and related
performance issues, of poly-Si TFTs. The subjects covered include an analytical
model for conduction in poly-crystalline material, issues with the measurement of
the poly-Si DOS, low field and high field leakage current behaviour, and an
overview of the bias stress instability mechanisms in these devices. Finally, there
is a discussion of short channel effects, including the role of parasitic bipolar
transistor action caused by floating body effects.
8.1 Introduction
Chapter 7 dealt with the preparation of poly-Si material, as well as describing the
architecture and fabrication process of poly-Si TFTs. In this chapter, the charac-
terisation and performance of these devices is dealt with in greater detail. Poly-Si
is a polycrystalline material, consisting of small grains separated by grain
boundaries, and the understanding of this composite material, which is spatially
inhomogeneous, is important both in establishing a basic understanding of device
behaviour, and in recognising the limitations of some of the assumptions made in
analysing its behaviour. Hence, Sect. 8.2 introduces a classical model of con-
duction in poly-Si [1]. This could be regarded as a model covering polycrystalline
material in general, within which grain boundary states play an important role in
trapping free carriers. This is a simple analytical model, which establishes some
basic physics and concepts in a readily accessible fashion, even though the sim-
plifications in this model have been superseded by more complex numerical
models. This point is developed in Sect. 8.3, which discusses the measurement of
the density of states, DOS, in poly-Si, and highlights the issues of distinguishing
trapping states within the grains from those in the grain boundary regions.

254 8 Poly-Si TFT Performance
Section 8.4 describes the low field leakage current behaviour in TFTs from the
standpoint of a classical carrier flow and electron–hole pair generation rate model,
and relates this to the quality of the poly-Si.
Section 8.5 discusses other performance issues with poly-Si TFTs, in particular,
the effects associated with a high electrostatic field at the drain junction. Firstly, it
is demonstrated that this is a two dimensional field, which is controlled by both the
voltage on the drain and the gate field. The high field is responsible for hot carrier
instabilities, and field enhanced leakage currents, in high quality poly-Si TFTs.
The high fields are essentially a device architecture issue, rather than a material
quality issue, and the solution to this is found in the relief of the drain field by
lightly doped drain regions (LDD), as discussed in Sect. 7.4.1.2. Although the hot
carrier damage, discussed in Sect. 8.5.2, is a major reliability issue in n-channel
TFTs, there are other instability mechanisms in poly-Si TFTs, which are displayed
in other biasing regimes, and these are briefly discussed in Sect. 8.6.
Finally, the current understanding of the performance of short channel TFTs is
reviewed in Sect. 8.7, which also deals with the issue of floating body effects in the
TFTs.
The understanding of the issues discussed in this chapter has been crucial to
their resolution, and, as discussed in Chap. 7, this has facilitated the mass-pro-
duction of stable, high quality poly-Si TFTs for flat panel display applications.
8.2 Electrical Conduction in Poly-Si
8.2.1 Analytical Bulk Conduction Model
As is readily apparent from electron microscopy images of poly-Si, it is a poly-

crystalline material, consisting of discrete grains, of variable crystallographic
orientation, separated by inter-grain regions. These regions need to accommodate
the changes in crystallographic orientation in going from one grain to the next,
and, rather like the surfaces of crystalline regions, the interruption to the periodic
lattice structure of the grain can be expected to introduce defects states into the
forbidden band gap at its surface. In addition, there may also be impurity defect
states here. Hence, the grain boundary regions will be regions of structural dis-
order, and will contain deep level trapping states within the forbidden band-gap.
Indeed, many aspects of the performance of poly-Si have been ascribed to carrier
trapping effects at the grain boundaries, and the resistivity of doped poly-Si is a
good example of this.
Figure 8.1a shows the variation of resistivity with boron concentration in p-type
poly-Si, for two different grain sizes, and compares it with single crystal Si [2].
There a number of striking differences between the two types of material. Firstly,
in single crystal silicon, the resistivity varies almost inversely with boron con-
centration, and this is only seen at the higher doping levels in poly-Si, where the
8.2 Electrical Conduction in Poly-Si 255
(b)1000
Experimental
(a) 10 6 measured
(un-doped) (25°C) { +o REF. [4]
This work
Theoretical
Single Crystalline
10 5 Theory - This work Silicon
Hole Mobility in cm2 /V-sec

10 4
100
Resistivity ( –cm)
10 3
Lcal 230Å
10 2
Lcal 1220Å
10 1
10 0 10
Single
10 –1 Crystal Si
10 –2
10 –3 1
10 15 10 16 10 17 10 18 10 19 10 20 10 15 10 16 10 17 10 18 10 19 10 20
Doping concentration (cm–3) Doping concentration (cm–3)
Fig. 8.1 Comparison of parameter dependence on boron doping concentration in poly-Si and
c-Si a resistivity for two different poly-Si grain sizes (Reprinted from [2] with permission of
IEEE), and b hole mobility (Reprinted with permission from [1]. Copyright (1975) American
resistivity approaches that of c-Si. Secondly, at very low concentrations, the poly-
Si resistivity is several orders of magnitude higher than c-Si, and, thirdly, over a
narrow range of doping levels, the poly-Si resistivity drops very abruptly. There
are also clear differences in the behaviour of the Hall mobility in the two material
types, as shown in Fig. 8.1b [1]. Whereas the hole mobility in crystalline Si shows
a gradual decrease with increasing doping level, due to increased charged impurity
scattering, the Hall-effect hole mobility in poly-Si shows a completely different
behaviour, with the mobility falling to a very low value at moderate doping levels,
before rising back to a value comparable to c-Si at higher doping levels.
The characteristic behaviour of poly-Si, as shown in Fig. 8.1, has been attrib-
uted to carrier trapping in the grain boundaries, GBs, and has been quantitatively
described using a simple, analytical model [1]. In this, the material is represented
by a combination of trap-free single crystal grain regions, doped with a boron
acceptor density N, and separated by planar grain boundary regions containing an
areal density, QT, of deep donor trapping levels at ET-EV, as illustrated in Fig. 8.2.
Figure 8.2a shows a single crystal grain of length L, which is separated from the
adjacent grains by the grain boundaries. Figure 8.2b shows the charge distribution
between the GBs, where the GBs have trapped free holes from the grain. For
charge neutrality, the positive trapped charge in the GBs will be balanced by the
negative space charge in the regions of length l within the grains, where the ionised
boron acceptors have lost their holes to the GBs. Hence, to maintain the overall
space charge neutrality, the positive charge in the GBs, Q+T, will be equal to the
Fig. 8.2 Analytical model

for grain boundary effects in
poly-Si a GB model, b space
charge distribution, and
c band diagram, showing
band bending (Reprinted with
negative charge, Nl, in the two space charge regions either side of the GB. The
band bending, in Fig. 8.2c, is shown for a particular situation, in which the Fermi
level is close to the valence band edge, and beneath the GB trapping level, ET. This
represents a heavily doped grain, and fully occupied hole traps. There is a potential
barrier at the GB due to the charge QT on the traps, and the compensating
depletion region extending partially into the grain by the distance l. The band
bending in this region can be calculated from Poisson’s equation:
d2 V qN
2
¼ ð8:1Þ
dx e0 es
Integrating this twice, and assuming that dV/dx = 0 at l, the potential distri-
bution in the grain is:
qNx2
V¼ ð8:2Þ
2e0 es
and the potential barrier height at the GB is:

qNl2
Vb ¼ ð8:3Þ
2e0 es
For charge neutrality (and using the depletion approximation and zero Kelvin
Fermi–Dirac statistics):
Nl ¼ 0:5QT ; and, therefore, the barrier height, Vb, can be expressed as:
qQ2T
Vb ¼ ð8:4Þ
8e0 es N
In other words, for partially depleted grains, and fully occupied traps, where
Nl = 0.5QT, the GB barrier height, Vb, diminishes as the grain doping level
increases.
However, in the alternative situation, where Nl \ 0.5QT, all the holes are
trapped in the partially occupied GB traps, and the grain is fully depleted, giving
rise to the high resistivities seen at low doping levels in Fig. 8.1a. In this fully
depleted case, l = L/2, the Fermi level will be above the trapping level, ET, and
Eq. 8.3 becomes:
qNL2
Vb ¼ ð8:5Þ
8e0 es
and, for the fully depleted grains, the GB barrier height increases with grain doping
level. For very lightly doped grains, Vb is close to zero, and it increases linearly
with doping concentration up to a maximum value, Vb(max), of:
qN L2
VbðmaxÞ ¼ ð8:6Þ
8e0 es
at the critical doping concentration, N*, where N*L = QT. Hence, Vb(max) is a
function of both grain size and GB trap state density, and an example of the Vb and
Vb(max) dependences on N and QT is shown, for mid-gap traps, in Fig. 8.3 [3]. It
will be appreciated that increments in doping concentration, above the critical
dopant level, N*, now start to add an almost equal number of free carriers within
the grain, and this, combined with the reducing barrier height, leads to the steep
reduction in resistivity, seen in Fig. 8.1a.
An analytical expression for the conductivity of the material [1] was developed
by assuming that, when a bias was applied to the poly-Si material, current flow was
limited by the motion of carriers over the potential boundary at the GB, rather than
by current flow within the grain itself. The maximum carrier flow over the
potential barrier is given by thermionic emission, which is independent of the
shape of the potential barrier and is determined only by its height. For simplicity,
this mechanism was assumed, where the current density, Jth, is given by the
number of carriers moving towards the GB with energies sufficient to overcome
the potential barrier [4], and a general expression for the thermionic emission
current over a barrier of height Ub is:
Fig. 8.3 Calculated

variation, with doping
concentration, of the potential
barrier height at a grain
boundary. The calculations
are for different
concentrations, Nt, of a
discrete grain boundary trap
at mid-gap (Reprinted with

qUb qVa
Jth ¼ qnv exp exp 1 ð8:7Þ
kT kT
where, v is the carrier’s thermal collection velocity (= {kT/2pm*}0.5) [5], n is the

carrier concentration at the bottom of the barrier, m* is its effective mass, and Va is
the potential drop across the barrier.
In using this expression, the carrier density employed by Seto [1] was the
average number, Pa, in the grain, which was calculated by integrating the free hole
distribution, p(x), across the grain, and dividing it by L. For a fully depleted grain,
p(x) is given by:

pðxÞ ¼ NV exp qV ð xÞ Ef =kT ð8:8Þ
and Pa is:
" 1=2 #
ni 2pes e0 kT 1=2 Eb þ Ef qL N
Pa ¼ exp erf ð8:9Þ
Lq N kT 2 2es e0 kT
where Eb = qVb, and ni is the intrinsic carrier concentration. The Fermi level, Ef,
is determined by the following equality:
Q
NL ¼ T ð8:10Þ
1 þ 2 exp ET Ef =kT
For partially depleted grains, the carrier concentration, pb, in the undepleted
region is approximately equal to N in non-degenerate samples, and is related to the
Fermi level by:
pb ¼ NV exp ðEf =kTÞ ð8:11Þ
The average concentration, Pa, of the un-depleted and fully depleted regions is
[1]:
( " 1=2 #)
QT 1 2pes e0 kT 1=2 qQT 1
Pa ¼ pb 1 þ erf ð8:12Þ
LN qL N 2 2es e0 kTN
From Eq. 8.7, the thermionic emission current, Jth, for a voltage Va applied
across a grain boundary, and where Vb [ kT, is [1]:
1=2

kT qVb qVa
Jth ¼ qPa exp exp 1 ð8:13Þ
2pm kT kT
and, if qVa \\kT, Eq. 8.13 reduces to:

1=2
2 1 qVb
Jth ¼ q Pa exp Va ð8:14Þ
2pm kT kT
The conductivity, r, of a material is related to the current density, J, and the
electric field, E, by J = rE, so, the poly-Si conductivity can be evaluated from
Eq. 8.14 together with Eqs. 8.9 or 8.12:
1=2
2 1 qVb
r ¼ Lq Pa exp ð8:15Þ
2pm kT kT
and from r = qpl, the effective carrier mobility, leff, is given by:
1=2
1 qVb
leff ¼ Lq exp ð8:16Þ
2pm kT kT
Hence, as Vb goes through a maximum with doping level, the effective mobility
goes through a minimum, as shown in Fig. 8.1b. A generic version of this
equation,

qVb
leff ¼ l0 exp ð8:17Þ
kT
is also often used to relate the poly-Si field effect mobility to an assumed GB
barrier height in the TFT channel [5, 6].
Using Eqs. 8.15 and 8.16 (together with some scaling parameters), the model
was successfully fitted to the resistivity and Hall-effect mobility data by Seto [1].
Examples of these fits by Lu et al. [2] and Seto [1] are shown by the solid lines in
Fig. 8.1a and b, respectively. (The scaling parameters are discussed in Sect. 8.2.3).
Hence, this simple analytical model, based upon local carrier trapping at grain
boundaries, and the associated potential barriers in the valence band, encompassed
the essential features of the experimental data. It offered an explanation for the
high resistivity of lightly doped poly-Si, the sharp fall in resistivity at a critical
doping level, the asymptotic approach to c-Si data at high doping levels, and the
anomalous behaviour of the Hall-effect hole mobility.
The above analysis has been based upon p-type doping of poly-Si, together with
GB-associated deep donor traps, and the induced potential barriers in the valence
band edge. The essential features of the model have been shown to apply equally
well to n-type doped poly-Si, in which potential barriers arise in the conduction
band edge due to electron trapping at deep acceptor levels in the upper half of the
band gap [3].
8.2.2 Analytical Model of TFT Conduction
The key features of the bulk conduction model have also been used to produce an
analytical model for conduction in poly-Si TFTs, in which the dependence of the
barrier height on doping level has been replaced by an equivalent dependence
upon the carrier density induced in the channel by the gate voltage [5]. In the same
way as in the Seto model, this model also assumed discrete trapping levels for the
GBs, and thermionic emission over the potential barriers at the band edges. An
expression was derived for the channel current as a function of gate voltage, in
which the gate voltage controlled the carrier density in the channel, and, hence, the
height of the potential barriers at the GBs [5]:

l0 WCox VG Vd q3 Q2T ti
Id ¼ exp ð8:18Þ
L 8es e0 kTCox VG
W, L are TFT channel width and length, respectively, VG and Vd are gate and
drain biases, respectively, Cox is the gate oxide capacitance, and ti is the inversion
layer thickness (giving the equivalent volume charge density in the channel from
CoxVG/ti). The barrier height term in Eq. 8.18 corresponds to Eq. 8.4, which
described the potential barrier at a GB in a partially depleted grain. Also, in this
expression, the mobility term, l0, was given by qLv/kT (where v was the carrier’s
thermal collection velocity), and this retained the thermionic emission criterion.
Equation 8.18 can be used to analyse experimental TFT transfer characteristics,
by plotting ln(Id/VG) versus 1/VG, as shown by the inset in Fig. 8.4, from which,
the slope gives QT, and the intercept gives l0. This plot is from a 2-D simulation of
TFT transfer characteristics, with 0.5 lm long grains, and mid-gap traps at the
GBs [7], and similar experimental data plots have been published by Levinson
et al. [5], and by many other authors. The range of VG values (within which the
plot was linear) was from *4.5 to 7 V, which corresponded to the sub-threshold/
near-threshold current regime, where the increase in channel current was ascribed,
by the model, to the reduction in GB-barrier height. For this particular simulation,
QT could be correctly calculated to be 2 9 1012 cm-2, by taking the inversion
layer thickness, ti, to be 34 nm [7]. However, this value will generally not be
Fig. 8.4 Calculated transfer

characteristic of a TFT
containing 0.5 lm long
grains, with a mid-gap trap at
the GB. The inset is a
Levinson plot of ln(Id/VG)
versus 1/VG. (Reprinted from
[7])
known experimentally, and this introduces an unavoidable numerical uncertainty

in using this technique.
8.2.3 Limitations of Analytical Model
As will be shown in Sect. 8.3, contemporary evaluations of the density of states,

DOS, in poly-Si TFTs show a continuous distribution of trapping states across the
band gap. Hence, the assumption of just a discrete trapping level is a major
simplification, which fails to reflect the true nature of poly-Si, and the value of QT
extracted from the analytical TFT analysis [5] will be an artefact of this
assumption. However, as shown by Fig. 8.5, one of the key features of the model is
retained with a distributed DOS [3], namely, the initial increase in GB barrier
height with increasing doping level, and then a reduction in height once a critical
doping level was exceeded. In this case, the critical doping density, N*, occurred,
Fig. 8.5 Calculated 0.7

variation, with doping
L =10 –5 cm
concentration, of the potential 0.6
barrier height at a grain
Barrier height [eV]
boundary. The calculations 0.5

NSS= 1014cm–2eV–1
are for different
concentrations of GB traps 0.4
uniformly distributed across EF N SS= 5×1013
the band-gap. (Reprinted with 0.3
NSS= 2×1013
0.2
Copyright (1978) American NSS = 1013
0.1
12
NSS = 5×10
0
1016 1017 1018 1019
Impurity concentration (cm–3)
for a uniform trap distribution, NGB(E), when the total space charge in the grain,
N*L, was equal to the integral of the trapped charge in the distributed GB states,
and this equality could be expressed as [3]:
1
NGB q2 NGB L N
N ¼ 1þ kT ln ð8:19Þ
L 8es e0 ni
which needs to be solved iteratively for N*.

The other simplifications have been, firstly, the assumption of thermionic
emission, as opposed to drift/diffusion carrier transport, and, secondly, the nature
of the GB region itself, and whether it is adequately represented by a planar
boundary [8]. These issues were raised, in part, by the scaling factors used to fit
Eq. 8.15 to the experimental data [1, 8]. The added scaling factors were a pre-
exponential term, f, and an ideality factor, n, giving the modified conductivity
equation:
1=2
1 qVb
r ¼ fLq2 Pa exp ð8:20Þ
2pm kT nkT
where f was 0.12 and n was 6.49. It was shown in this later work [8] that, by
treating the GB region as an a-Si-like region of *1–2.5 nm width, with band-tail
and deep states, there was trap-limited conduction in this region. Hence, the
material’s conductivity was not merely limited by transport both within the grains
and across the potential barriers at the GBs, but account also had to be taken of
transport within finite width grain boundary regions themselves. By adding these
further considerations, including drift/diffusion flow over the potential barriers, it
was possible to fit the experimental data in Fig. 8.1a, without the arbitrary fitting
factors, f and n, used in the earlier work [8].
In spite of these detailed qualifications to the bulk conduction model of Seto [1],
its TFT equivalent, Eq. 8.18 [5], has continued to be widely used in many publi-
cations to analyse the poly-Si TFT characteristics, and to extract a GB trapping state
density, QT. In view of the above limitations in the basic model, and the particular
problem of knowing the inversion layer thickness, ti, in Eq. 8.18, the extracted QT
values should be taken more as a figure of merit for the TFT, rather than as an
accurate estimate of the trapping state density associated with the GB region.
8.3 Poly-Si Density of States, DOS
Section 8.2 introduced the role of GB trapping states in explaining the behaviour
of bulk poly-Si resistors, and briefly discussed the extension of a simple analytical
model [1] to analysing TFT behaviour [5]. This section continues that discussion
by looking, more broadly, at the evaluation of the poly-Si DOS, but without the
limitations imposed by the simplifications in the analytical model. In general
8.3 Poly-Si Density of States, DOS 263
Fig. 8.6 Small grain poly-Si ID (A)

TFT transfer characteristics 10–5
W 50
measured after different L
=
6
exposures times to a VD = 5V
hydrogen plasma: (1) 0 h, (2) 1 10–7
0.5 h, (3) 4 h. The increasing
exposure times produced 2
10–9
reducing densities of states 3
across the band-gap.
(Reprinted from [9] 10–11
Copyright (1991), with
permission from Elsevier) –30 –20 –10 0 10 20 30
VG(V)
terms, the overall performance of poly-Si TFTs is affected by states within the
forbidden band-gap: those near the centre contributing to leakage current, whilst
those close to the band-edge will reduce the field effect mobility from the band
value, and states in between will determine the sub-threshold slope and threshold
voltage. These dependences are illustrated in Fig. 8.6, for a set of poly-Si TFTs,
whose different DOS levels were determined by different exposure times to an
atomic hydrogen passivation process [9]. Comparing the TFT with zero exposure
time (and high DOS) to the device with 4 h exposure (and low DOS), the latter had
a higher on-current, lower off-current, lower threshold voltage and smaller sub-
threshold slope. The Levinson analysis [5] showed a reduction in effective GB
trapping state density from 2.7 91012 cm-2 to 1.0 9 1012 cm-2, although, as
discussed in Sect. 8.2.2, these absolute values are subject to some numerical
uncertainty, but the relative change is a more reliable indicator of the overall
change in DOS.
Given the limitations of the Levinson analysis, many authors have used a
variety of more sophisticated techniques to analyse the TFT characteristics, in
order to extract the distribution of trapping states [10–15]. In these analyses, it has
also been recognised that certain forms of poly-Si, such as the large faulted grain
material produced by the SPC process, are likely to have defects states within the
grains themselves, and not merely at the grain boundaries [10, 14]. In addition, the
assumption of mono-energetic trapping states was relaxed, and the DOS was
evaluated across the back-gap. The techniques have included 1-D analysis of the
field-effect conductance, obtained from the TFT transfer characteristic [10], the
measurement of the temperature dependent emission of electrons from the traps,
using deep level transient spectroscopy [11], or, more commonly, 2-D numerical
simulations to reproduce the transfer characteristic [12–15]. From this work, there
is a general consensus that there is a continuous distribution of trapping states
across the band-gap [10–15], as shown in Fig. 8.7 [12], and this is frequently
represented by a double exponential distribution of states in each half of the band-
gap:
Fig. 8.7 Calculated poly-Si

DOS, obtained from a 2-D
simulator fit to experimental
transfer characteristics
Physics)
E E
NðEÞ ¼ NT exp þ ND exp ð8:21Þ
ET ED
where NT and ND are the band-edge concentrations of tail and deep states,
respectively, and ET and ED are the characteristic energy widths of these states.
The same expression is used for both the acceptor states in the upper half of the
band-gap and the donor states in the lower half of the band-gap, although the
individual parameter values are usually different in the two halves of the band-gap.
The total densities, integrated across the band-gap, are given by NTET and NDED,
for the tail-states and deep states, respectively.
However, a persistent problem in uniquely determining the DOS in poly-Si has
been the difficulty in distinguishing between trapping states within the grains and
trapping states at the grain boundaries. This is because the conventional poly-Si
grain size is smaller than the typical TFT channel dimensions, and device mea-
surements always yield an effective trapping state density per unit gate area, which
can be attributed to an equivalent density of charges trapped in the grains and/or in
the GBs. Some authors have attempted to partition the distribution between these
two sites [14, 15], whilst others have used a spatially uniform effective density of
states [12, 13] (which is also more computationally efficient, as it requires fewer
mesh lines within the model). In all these cases, justification for the assumed
distributions was based upon demonstrating a good fit of the model to experi-
mental transfer characteristic data. However, with the large number of adjustable
parameters in these simulations, and the different underlying assumptions, the
question of an unambiguous characterisation of inter-grain and intra-grain defects
remained unresolved.
As with the poly-Si resistor analysis, a further problem is the appropriate
representation of the grain boundary itself, either as a planar, 2-D distribution of
defects, or as a thin disordered, 3-D region of defects between the grains.
8.3 Poly-Si Density of States, DOS 265
Fig. 8.8 a Experimental (symbols) and fitted (lines) data to SLS poly-Si transfer characteristics,
for carrier flow parallel and perpendicular to the sub-GBs in the SLS material, b in-grain and GB
DOS extracted from the SLS material (Reprinted with permission from [16]. Copyright (2007)
American Institute of Physics)
Recent work using long-grain SLS poly-Si has addressed these problems, by
exploiting the anisotropic nature of this material (as discussed in Sect. 7.5.2.1), in
which the grain size in one direction exceeded the channel length [16]. In that
work, the carrier flow parallel to the sub-GBs (ln = 218 cm2/Vs) was assumed to
be controlled only by states within the grains, whereas carrier flow orthogonal to
the sub-GBs (ln = 76 cm2/Vs) was assumed to be controlled by both the in-grain
states as well as by the GB states. Hence, by using data from orthogonal TFTs, the
in-grain DOS could be separated from the GB DOS. The temperature dependent
transfer characteristics were measured in these two directions, and 2-D simulations
of the transfer characteristics and their activation energies were used to establish
the DOS for the intra-grain and inter-grain states [16]. The fitted characteristics,
measured at 324 K, are shown in Fig. 8.8a, and, to obtain a consistent fit across the
range of measurement temperatures, the GB states could not be treated as a 2-D
planar distribution of defects, but had to be treated as a 2 nm wide, thin disordered
region. This was because, in spite of the mobility differences for parallel and
perpendicular flow remaining constant into strong inversion, the activation energy
for perpendicular flow was close to zero. Such a small activation energy was not
consistent with a GB barrier height large enough to produce the observed mobility
difference (Eq. 8.17 predicted a barrier height of 27 meV at 295 K). Within the
narrow GB region, the fitted DOS determined the carrier density, and, hence, the
resistive current flowing through that region. This resistive current flow was the
limiting transport mechanism in the TFT, and was also characterised by a near-
zero activation energy, in agreement with the experimental data. The resulting
DOS distributions for the in-grain and the GB states are shown in Fig. 8.8b. As
with the earlier evaluations, such as in Fig. 8.7, the distribution of both types of
states was continuous across the band-gap, with an exponential-like distribution
near the band-edges. The results in Fig. 8.8b also demonstrated that there was a
much higher integrated volume density of traps within the GB region than in the
grains themselves: for the band-tail states, in the upper half of the band gap, the
densities were 7.7 9 1018 cm-3 and 1.7 9 1018 cm-3 respectively, and the deep
state densities were *50 times larger in the GBs. However, the 2 nm width of the
GB region, LGB, compared with the 300 nm grain size, LG, meant that the total
density of GB states per unit surface area of the film (3.4 9 1011 cm-2) was only
*5 % of the equivalent density within the grains (7.2 9 1012 cm-2), where the
density per unit surface area, NSA, of GB states, NGB, is given by:
NGB tf LGB
NSA ¼ ð8:22Þ
LGB þ LG
and tf is the film thickness.
The detailed analysis of the sub-GB defects in the SLS poly-Si TFTs was
consistent with the analysis of conduction in poly-Si resistors [8], which avoided
the use of fitting factors, and demonstrated that the results could be most satis-
factorily explained by assuming that the GB region was a narrow disordered
region, rather than a 2-D plane of defects. Thus, the transport of carriers within this
region determined the overall conductivity of the material, rather than the emission
of carriers over the potential barrier at the GBs.
In some TFT simulation work [16, 17], the temperature and field dependence of
the carrier mobility has been incorporated into the 2-D models by using functions
developed for c-Si simulations such as the Canali [18] and Lombardi [19] for-
mulations. In particular instances, such as the simulation of long channel length
TFTs [17], the incorporation of the transverse field dependence of the mobility has
been essential for the accurate reproduction of the mobility variation with gate bias
at large gate biases.
8.4 TFT Off-State Currents
The basic electron–hole pair generation process has been discussed in Chap. 2, and
this process, plus the transport of the generated carriers to the contacts, will be
used to examine the leakage current behaviour observed in poly-Si TFTs.
Experimentally, the leakage current is found to be very dependent upon the
magnitude of the field at the drain junction, and low field and high field regimes
can be readily distinguished. The former is governed by the field-independent
generation processes described in Chap. 2, whilst the latter is determined by field-
enhanced emission processes. The low-field generation process will be described
in this section, using a simple analytical model to give physical insight into the
process, and the high field process will be covered in Sect. 8.5.3, which groups
together other drain field dependent processes.
The transfer characteristics, from a range of different channel length TFTs, are
shown in Fig. 8.9a–c for different DOS levels. From a leakage current perspective,
8.4 TFT Off-State Currents 267
Fig. 8.9 Channel length

dependent transfer
characteristics for poly-Si
TFTs with different DOS
levels, after varying degrees
of plasma hydrogenation
a zero, b partial, and c full
hydrogenation (Reprinted
from [20], with permission
from IOP Publishing Ltd)
the major difference between curves Fig. 8.9a and c, apart from the absolute values
of current, is that the leakage current was channel length dependent in the high
DOS case, whereas, in the low DOS case, the leakage current was independent of
channel length [20]. The corresponding ID-VD leakage current curves, for different
values of VG, are shown in Fig. 8.10a–c, for 10lm or 6lm channel length TFTs.
For the high DOS devices in Fig. 8.10a, the minimum current varied inversely
with channel length [9], and the ID-VD curve was linear, suggesting a resistive, or a
drift-limited, current flowing through a non-rectifying contact. For the medium
DOS device in Fig. 8.10b there was resistive current flow at low VD, which
saturated into a quadratic current at higher VD, and the applied voltage, at which
Fig. 8.10 Gate and drain (a) VG (V)

bias dependent leakage 10–9 –14
current characteristics for W 50
poly-Si TFTs with different =
L 10 –7
DOS levels, after varying 10 –10
0
ID (A)
degrees of plasma
hydrogenation a zero,
b partial, and c full 10 –11
hydrogenation. (Reprinted Line of
from [20], with permission unity slope
from IOP Publishing Ltd) 10 –12
(b) 10 –9
W = 50
L 10
ID (A) 10–10
VG (V)
10 –11 –7
–6
–5
10–12 –4
0.1 1.0 10.0
VD (V)
(c) 10 –9 VG (V)
W = 50 –15
10 –10 L 6
–11
ID (A)
–7
10 –11
10 –12
10 –13
0 5 10
VD (V)
this occurred, reduced with increasing negative gate bias. Finally, for the lowest
DOS device in Figs. 8.9c and 8.10c, the channel length independence and the
exponential increase in current with both VG and VD indicates a field enhanced
generation current. The leakage current mechanism, displayed in Fig. 8.10a and b,
is described below by a simple analytical n+p junction model [9], and the field
enhanced current is analysed in Sect. 8.5.3.
The model used for the low-field currents in Fig. 8.10a and b is shown in
Fig. 8.11, in which there is electron–hole pair generation in the drain space charge
region, supported by the voltage dropped across this region, Vgen, and resistive
hole flow to the source junction, causing a voltage drop, VR, across the neutral
Fig. 8.11 Cross sectional diagram of TFT, showing carrier generation, voltage distribution, and
leakage current paths
portion of the transistor body. From Chap. 2, the generation current, Jgen, is given
by:
qni xD
Jgen ¼ ð8:23Þ
s
and the drift current, JR, is given by:
JR ¼ qðnln þ plp ÞF qapi lp F ð8:24Þ
assuming that p = api = ani and that a [ 1 for negative gate biases, where there
will be surface accumulation of holes (ni and pi are the intrinsic carrier concen-
trations). The field, F, in the transistor body is given by:
VR
F¼ ð8:25Þ
L xD
where, from Chap. 2, the space charge width, xD, is:
2es e0 Vgen
xD ¼ ð8:26Þ
qNT
VD ¼ Vgen þ VR ð8:27Þ
Jgen ¼ JR ¼ JD ð8:28Þ
Therefore,
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
N al s2
L þ L2 þ 4qni alp VD 2eT s e0pni qns i
JD ¼ ð8:29Þ
N al s2
2 2eT s e0pni qns i
Whilst Eq. 8.29 has been fitted to experimental data [9], it is more instructive to
look at the predictions of this equation, and to compare it to the general form of the
Fig. 8.12 Calculated leakage

current curves, JD-VD, using
Eq. 8.29 a low generation
lifetime case, s1 = 10-11s,
and b higher lifetime,
s = 10s1 (Reprinted from [9]
permission from Elsevier)
data in Fig. 8.10a and b. Some representative calculations are shown in Fig. 8.12a
and b, using the parameter values shown in the figure caption. The curves in
Fig. 8.12a have been calculated with a low value of the carrier generation lifetime,
corresponding to a high DOS near mid-gap, and for two different values of the hole
concentration, corresponding to two different degrees of hole accumulation. Under
these conditions, the ID-VD characteristics are linear, and increasing the hole
concentration by a factor of ten increases the current by the same amount. These
curves are similar to the experimental ones in Fig. 8.10a. In this case, the junction
is not rectifying because the potential electron–hole pair generation rate is so high,
that the current flow through the device is not limited by this generation process,
but by the transport of holes to the source junction (where it is implicitly assumed
that they recombine with electrons). Hence, there is drift-limited conduction, and
the resistive current is simply determined by the hole concentration in the film.
In contrast, the curves in Fig. 8.12b were calculated with a ten times higher
generation lifetime, and showed the same general form as the experimental data in
Fig. 8.10b. The curves had a linear ID-VD characteristic only at low values of VD,
and, eventually, the slope reduced to 0.5, when the current became generation rate
limited. The slope of 0.5 occurred because the increase in size of the space charge
region, within which generation took place, increased only with the square root of
VD, as shown in Eq. 8.26. In this case, the higher lifetime reduced the carrier
generation rate, and, as VD increased, the holes were swept away at an increasing
rate, until the generation rate at the drain could not sustain this current flow. When
this happened, the current became limited by the generation rate at the drain. It
will also be seen that when the hole concentration increased by a factor of ten, the
transition from a drift-limited to a generation-limited current occurred at a lower
value of VD. This is because, with the higher hole concentration, there was initially
a 10 times higher drift current, which had to be sustained by the generation rate at
the drain; but, as VD increased, the generation rate quickly became insufficient to
supply this increasingly large current, and the process became generation rate
limited at the lower value of VD.
As seen above, the simple analytical model in Eq. 8.29 provides physical
insight into the low-field leakage current behaviour of poly-Si TFTs, and it can be
used to explain the bell-shaped leakage current curves of ELA poly-Si TFTs, as
discussed in Sect. 7.2.2.3 and shown in Fig. 7.8. The leakage current curves, as a
function of ELA energy density, for L = 6 lm and L = 60 lm TFTs, from a
40 nm thick film [21], are shown in Fig. 8.13. These display the bell-shape already
referred to, and, at low ELA energies, the currents scaled inversely with channel
length, but were almost independent of channel length at high laser energies. This
indicates that, at low energies, and high DOS, the leakage currents were drift-
limited, and at high energies, and low DOS, became generation rate limited. In
order to model this transition with Eq. 8.29, it is necessary to use the changing
Fig. 8.13 Experimental and

modelled (Eq. 8.29) leakage
current characteristics, as a
function of TFT channel
length and ELA energy
density, using the generation
lifetime, s, as the adjustable
parameter (Reprinted from
[21] Copyright (1999), with
values of measured hole mobility as the ELA energy density increased, and this
was done in the fitted curve in Fig. 8.13. In fitting this curve to the data, the
generation lifetime, s, was an adjustable parameter, and the values plotted against
the right hand axis show, as expected, a monotonically increasing lifetime with
increasing ELA energy density. Hence, the initial rise in leakage current with
energy density was due to the increasing carrier mobility in the drift-limited arm of
the curve, but, at the same time, the generation lifetime was increasing with energy
density, leading eventually to a transition to generation-limited currents. Once the
current became generation rate limited, it fell with increasing energy density as the
lifetime increased due to the reduced DOS in the high quality, optimally crys-
tallised material.
8.5 Performance Artefacts and Drain Field
Three major performance artefacts have been identified even in low DOS, high
quality poly-Si TFTs, and these have been associated with high drain fields in the
devices. Firstly, the off-state current increases exponentially with both gate and
drain bias, as seen in Figs. 8.9c and 8.10c, respectively. Secondly, n-channel TFTs
are subject to drain bias stress instability, as shown in Fig. 8.14, and, finally, the
output characteristics show poor current saturation, as seen by the insert in
Fig. 8.14 Influence of drain bias stress of 13 V for 60 s on NSA TFT transfer characteristic. The
insert shows its effect upon the output characteristic (Reprinted with permission from [24].
8.5 Performance Artefacts and Drain Field 273
Fig. 8.15 Simulations of the maximum field at the drain/channel junction as a function of a gate
bias and poly-Si DOS (56 nm lateral dopant spread), and b lateral spread of drain dopant. (The
DOS, N(E), gave a TFT with a sub-threshold slope, S, of 1 V/dec). (Reprinted with permission
Fig. 8.14. The details of the instability process and the leakage current effects are
discussed further in Sects. 8.5.2 and 8.5.3, respectively, whilst the electrostatic
drain field itself is discussed in Sect. 8.5.1.
8.5.1 Electrostatic Drain Field, F
The 2-D electrostatic field in the space charge region of a semiconductor junction
is given by Poisson’s equation, which relates the space charge density, Nsc, and
field, F:
oFx oFy qNsc
þ ¼ ð8:30Þ
ox oy es e0
Using a 2-D simulator, evaluation of this expression, for poly-Si TFTs, showed
that the differential terms were opposite in sign, and both were much larger than
the space charge term on the right side, such that their difference was only
*10–20 % of their absolute values [22]. Thus, the internal electrostatic field was
not controlled by the poly-Si density of states, DOS, as it would be in a 1-D
analysis. This is shown in Fig. 8.15a, in which the maximum drain field was
calculated as a function of gate bias, with the poly-Si DOS as a scaled independent
parameter [22]. In this diagram, the different DOS values produced sub-threshold
slopes of 3.0, 1.0 and 0.3 V/dec in the simulated TFT transfer characteristics (for a
gate oxide thickness, tox, of 0.15lm). To put those slopes into context, a high
quality poly-Si TFT would have a sub-threshold slope of 0.4 V/decade for this
oxide thickness. Therefore, for TFTs of practical interest, with a sub-threshold
slope of \ 1 V/dec, the DOS would have no significant effect upon the drain field
in the off-state and sub-threshold regime, although there was a weak dependence
of drain field on trap state density at large positive VG, when the device was turned
on [24]. However, as shown in the next section, the on-state hot carrier degradation
was, nevertheless, greater in the low DOS device.
The simulations showed that the drain field was a 2-D effect, and was deter-
mined primarily by the combined effects of the drain and gate biases, the oxide
thickness, and the lateral distribution of the drain dopant at its edge [22, 24]. This
latter point is shown in Fig. 8.15b, demonstrating that the more abrupt the junction
is, the greater is the peak field. The 2-D dependence upon gate bias and oxide
thickness meant that, as the transverse gate field increased, so did the lateral drain
field, so that, for example, reducing the oxide thickness will increase the drain
field.
Because of the significance of the lateral dopant distribution, the drain field in
NSA TFTs decreases with increasing ELA shots (due to lateral diffusion of the
drain dopant) [23], and, equally, the drain field in SA TFTs is likely to be larger
than in NSA TFTs (due to reduced lateral diffusion in the former).
As the poly-Si quality itself had little effect upon the magnitude of the drain
field, a change in architecture was required to produce a greater lateral dopant
spread, and this is most readily accomplished through the use of lightly doped drain
(LDD) regions, the fabrication of which was discussed in Sect. 7.4.1.2. It is worth
noting that similar high drain field effects occur in MOSFETs, and these also
routinely incorporate LDD regions for field relief.
8.5.2 Hot Carrier Damage and LDD
The high drain field, combined with the high electron mobility, leads to carrier
heating and the generation of additional electron–hole pairs by impact ionisation.
These additional carriers contribute to the poor saturation of the output charac-
teristics of poly-Si TFTs, and to the so-called ‘kink’ effect when the impact ion-
isation-induced avalanche currents ultimately dominate the current flow [25, 26].
The impact ionisation, underlying the ‘kink’ effect, is enhanced by parasitic
bipolar transistor (PBT) action in the floating body of the TFT [25], and this effect
increases with reducing channel length [27]. Further discussion of PBT action is
presented in Sect. 8.7, dealing with short channel effects.
The impact of hot carrier damage (HCD) on n-channel TFT transfer charac-
teristics has already been shown in Fig. 8.14, and the changes in both on-state and
off-state currents are attributed to the creation of localised electron trapping cen-
tres near the drain junction. Also shown by the inset is the change in output
characteristics as a result of the HCD: at low drain bias, there is reduced drain
current (compared with the unstressed sample) due to the high resistance of the
channel near the drain, and, at large drain bias, there is a higher drain current due
to the negatively charged trapping states increasing the drain field, and, hence,
Fig. 8.16 Hot carrier instability effects in NSA TFTs a inter-dependence of gate and drain biases
for a 30 % reduction in on-current after 60 s stress (L = 6 lm and tox = 150 nm), b variation of
on-current degradation with drain bias and stress time (L = 4 lm and tox = 40 nm), c device
lifetime (for 30 % on-current loss) versus 1/VD, and d device lifetime versus VD. (a Reprinted
with permission from [24]. Copyright (1998) The Japan Society of Applied Physics)
increasing the avalanche generation rate [28]. Some typical examples of the
dependence of the hot carrier instability on gate and drain bias, and stress time, are
shown in Fig. 8.16a–d. The ‘spoon-shaped’ gate and drain bias dependence in
Fig. 8.16a is characteristic of the hot-carrier instability, and is discussed further
below. The change in on-current in Fig. 8.16b can be used to empirically deter-
mine device lifetime, where unacceptable performance can be defined to occur at a
given reduction in on-current, such as 30 % [26]. As will be seen in this figure, the
stress time needed for a given loss of on-current reduced logarithmically as the
drain bias increased. This strong dependence essentially occurred once the drain
bias caused the device to go into weak avalanching, and, within this regime, there
was an approximately exponential relationship between device lifetime and drain
bias. Indeed, from Eq. 8.33 below, the logarithm of this time should scale with
reciprocal drain bias, as seen in Fig. 8.16c, and the plot can be extrapolated to
determine the maximum drain bias to achieve a given device lifetime. An
approximately linear plot can also be obtained directly against drain bias [26], as
shown in Fig. 8.16d, and the 0.78 V/dec slope of the fitted line is an empirical
measure of the relationship between lifetime and drain bias. This indicated that for
each 0.78 V increment in drain bias, the device lifetime decreased by one decade.
The value is specific to this NSA device (L = 4 lm, tox = 50 nm), and it will
change with device geometry and architecture, so that, for instance, SA devices,
with more abrupt junctions, will typically have greater slopes, demonstrating a
stronger dependence of lifetime on drain bias. For example, 1lm channel length
SA TFTs, with 50 nm thick gate oxides, had a drain bias dependent lifetime of
0.32 V/dec.
Hot carrier damage has been extensively studied in c-MOSFETs, and the
damage centres were identified as acceptor states at the Si/SiO2 interface, as well
as positively charged states in the gate oxide itself. From the dependence of these
densities upon the gate bias, during drain bias stress, it was concluded that both
hot-electron and hot-hole injection into the oxide had to take place, and that the
generation of interface traps resulted from a two-stage process of hole capture at
oxide traps near the interface and subsequent recombination with injected elec-
trons [29]. It is likely that the recombination energy released by electron capture
could break weak interface bonds. There is injection of both carrier types because,
under the normal bias-stress conditions of VD [ VG, there is field reversal in the
oxide near the drain junction, which promotes hole injection. In spite of this field
reversal, significant electron injection can also take place against the field if the hot
electron is within a scattering distance of the Si/SiO2 interface [29].
In analogy with these HCD effects in MOSFETs, the poly-Si damage centres
have been widely attributed to states at the poly-Si/SiO2 interface [24, 26, 28, 30],
or, in analogy with defect creation in a-SiH TFTs, attributed to bulk states in the
poly-Si [31]. In both cases, good fits were obtained between simulation models and
the experimental characteristics, but the latter conclusion was based partly upon
the observation of damage at the back of the film. However, subsequent simulation
work showed that, even with top gated devices, damage centres are produced at
both the front and back interfaces [28, 32].
As mentioned in Sect. 7.4.1.2, the problem of hot carrier damage can be reduced
by the use of field relief regions, such as lightly doped drain, LDD, and gate-
overlapped lightly doped drain, GOLDD (shown in cross-section in Fig. 7.32),
and, by and large, lower doping levels give better field relief. However, dose
optimisation for field relief using LDD structures involves a compromise with
series resistance and reduced on-current [32]. In contrast, optimisation of dopant
concentration in GOLDD structures involves almost no compromise with series
resistance in the on-state, although there will be some current reduction due to
the increased channel length to accommodate the GOLDD region. Less obviously,
the GOLDD structure also gave significantly better field relief, as shown by the
comparison of the bias-stress results in Fig. 8.17a–d. In the GOLDD device
(Fig. 8.17c), there was almost no reduction in the drain current after a 30 V drain
bias stress, compared with the 10–100 times reduction in the LDD device after 10 V
stress. There was also no degradation in the output characteristic at low Vd, and the
characteristic second saturation regime in the GOLDD TFTs was reduced by the
stress [32].
These differences have been explained by applying the MOSFET HCD model
to poly-Si TFTs [32], in which the simulated carrier densities and fields in the TFT
Fig. 8.17 Experimental measurements before and after drain bias stress—arrows indicate
direction of changes a transfer characteristic, and b output characteristics of LDD TFTs after
drain bias stress at 8, 10 and 12 V for 5,000 s (LDD dose = 1x1013 P/cm2), c linear transfer
characteristics, and d output characteristics of GOLDD TFTs after drain bias stress at 30 V for
75, 1,270 and 10,000 s (GOLDD dose = 5 9 1012 P/cm2). (Reprinted from [32] with permission
of IEEE)
were used to model the hot-carrier injection currents into both the top and bottom
oxide layers by using the ‘lucky electron’ model [33]. The hot-electron injection
current is given by [32]:
Z
Je ðxÞ ¼ A nðx; yÞPe ðyÞdy ð8:31Þ
and the electron injection probability, Pe(y), is given by [33]:

y Ub
Pe ðyÞ / exp : exp ð8:32Þ
k Fx k
(The x direction is parallel to the semiconductor/insulator interface, and y is
perpendicular to it). Ub is the zero field oxide barrier height, and k is the electron
mean free path. The first term in Eq. 8.32 describes the probability of an electron,
at depth y in the poly-Si, travelling to the interface without an inelastic collision,
and the second term is the probability of the hot-electron having enough energy to
surmount the barrier Ub. A similar expression holds for hole injection, and the
barrier height for electrons (holes) is 3.1 eV (4.7 eV), and the electron (hole) mean
free path is 6.5 nm (4.7 nm).
The ‘lucky electron’ model has also been used to empirically predict device
lifetime, s, [34], by arguing that the device degradation will be determined by the
flux of injected hot carriers, which will be proportional to the normalised channel
current, ID/W and the hot carrier injection probability. The expression for s was
given by [34]:
W Ub
s¼C exp ð8:33Þ
ID Fx k
where C is a constant. Using device simulation, this expression has been evaluated
to investigate the gate bias and trap state density dependence of the device lifetime
in poly-Si TFTs [24]. The calculated lifetime, as a function of gate bias (with
VD = 15 V), is shown in Fig. 8.18, for two values of trap state density. It will be
seen that this expression predicts the ‘spoon-shaped’ dependence on gate bias, as
seen in the experimental data in Fig. 8.16a, where the worst-case stress condition
is approximately at VG = VT. The simple physical explanation for this effect is
that, up to this gate voltage, the electron surface density (and current) has been
increasing exponentially with gate bias, which more than compensates for the
reducing drain field. For gate bias values greater than VT, the electron concen-
tration only increases linearly with gate bias in inversion, and the corresponding,
linearly increasing current is not sufficient to compensate for the reducing drain
field due to the increasing value of VD(sat). Secondly, it will be seen that the worst-
case lifetime is slightly worse in the low DOS device, further reinforcing the
earlier argument that improvements in basic poly-Si material quality will not
reduce HCD effects.
Using Eq. 8.32, within a 2-D simulator, the device characteristics in Fig. 8.17
have been reproduced [32]. From the calculated hot-carrier injection currents, and
Fig. 8.18 Calculated

dependence of relative hot-
carrier device lifetime on gate
bias, using a 2-D simulator to
evaluate Eq. 8.33 (drain bias
was 15 V). (Reprinted with
Fig. 8.19 Simulated characteristics before and after drain bias stress—arrows show direction of
changes from un-stressed states a transfer characteristic, and b output characteristics of LDD
TFTs following stress at 8 V, 10 V and 12 V for 5,000 s (LDD dose = 1 9 1013 P/cm2),
c output characteristics of GOLDD TFT following 23 V stress for 323 and 10,000 s (GOLDD
dose = 5 9 1012 P/cm2). (Reprinted from [32] with permission of IEEE)
the ensuing two stage carrier recombination process within the oxide, the positive
charge density (due to trapped holes in the oxide) and the negative charge in
interface states was calculated iteratively with time, so that, as these charges built
up, their effect upon the field and current distribution in the poly-Si was
re-calculated, which then altered the trap generation rates [32].
Figures 8.19a and b show the simulated Id-VG and Id-Vds plots for the LDD
devices, and Fig. 8.19c shows the Id-Vds plot for the GOLDD device. In all cases,
these results accurately reflect the experimental data in Fig. 8.17. For the LDD
structure, acceptor trap generation started at the channel/LDD interface, where the
field was at a maximum. With increasing time and Vds-stress, the trap generation
spread into the adjacent channel and LDD regions, at both the top and bottom
poly-Si/SiO2 interfaces. The positive trapped charge started in the same location,
but finished up as a bi-modal distribution either side of the high field point. The
negative charge in the interface states dominated, and constricted the electron
channel current near the drain into the centre of the film. This formed a resistive
bottleneck, which was responsible for the loss of on-current [32]. For the GOLDD
structure, the generated interface state densities were lower than in the LDD
structure, and the positive trapped charge was largely generated near the top
Si/SiO2 interface. This partially compensated the negatively charged interface
states charges at this interface, and resulted in the uncompensated negatively

charged traps at the back interface having the greatest net effect. As a consequence
of this distribution, the effect of the hot carrier damage on the low Vds on-current
near the top interface was minimal. At large Vds, the current was deflected away
from the top interface, but the large negative charge density at the back interface
maintained the current flow more centrally within the film [32], and this reduced
the impact ionisation rate in the second plateau region of the GOLDD TFTs, as
seen in Figs. 17d and 19c [32, 35].
As shown, the GOLDD architecture gives significantly better hot carrier sta-
bility than the simpler LDD region, but, as with the non-self-aligned architecture,
it increases the parasitic capacitance between gate and drain, and the choice of
LDD architecture will be dictated by the trade-off in speed and high voltage
stability requirements. As discussed in Sect. 7.4.1.2, a sub-micron GOLDD region
would be needed for short channel devices, and this has been demonstrated with a
conducting spacer technology [36].
P-channel TFTs, in common with p-channel MOSFETs, are far less susceptible
to HCD [24, 26], and both MOSFET [37] and TFT [38] results have been suc-
cessfully modelled with only electron injection. In view of this, there does not
seem to be a unified HCD model using the same injection conditions for both
n- and p-channel TFTs.
Other instability effects in poly-Si TFTs, over and above HCD, are discussed in
Sect. 8.6.
8.5.3 Field-Enhanced Leakage Currents
High quality poly-Si TFTs, when biased in the off-state, show a leakage current
characteristic, which is channel length independent, as shown in Fig. 8.9c, and
which increases exponentially with both gate and drain bias, as seen in Figs. 8.9c
and 8.10c, respectively [20]. The channel length independence indicates a current,
which is limited by electron–hole pair generation processes at the drain, rather by
resistive flow through the body of the device (which would scale inversely with
channel length). However, normal thermal generation processes would produce
currents, which have only a square root dependence on drain bias, and an even
weaker dependence on gate bias. Hence, these anomalously large currents are
described as field-enhanced currents, and Fig. 8.20 schematically shows the range
of electron emission processes, which can be expected in the space charge region
of a reverse biased junction. As discussed in Chap. 2, the basic leakage current
mechanism is controlled by sequential electron and hole emission from deep traps
near the centre of the band-gap, and, for simplicity, Fig. 8.20 shows only the
electron emission processes to the conduction band. There will be equivalent hole
emission processes to the valence band.
At low fields and high temperatures, simple thermal emission (with no field
dependence) will dominate, giving drain current activation energies of *EG/2. At
Fig. 8.20 Schematic

illustration of electron
emission processes into the
conduction band from a trap
at depth ET. The trap is
located within a reverse
biased space charge region
high fields and low temperatures, tunnelling will dominate giving near-zero acti-
vation energies. Between these two extremes, two further processes are shown,
which are the field-induced lowering of the emission barrier for a Coulombic
centre (the Poole–Frenkel effect), and thermal emission to a virtual state followed
by tunnelling (phonon assisted tunnelling), both of which are field dependent.
Experimental results, near room temperature, showing thermal activation ener-
gies decreasing from *EG/2 with increasing VD [22] and VG [39] rule out pure
tunnelling, and the best fits to the data have been with phonon assisted tunnelling [7,
22, 39–41], although, at the highest biases, band-to-band tunnelling is also signifi-
cant [40, 41]. The phonon assisted tunnelling can be represented as a field dependent
enhancement, en(F), of the low-field electron thermal emission rate, en0, by:
en ðFÞ ¼ en0 cðFÞ ð8:34Þ
The field enhancement factor, c(F), for a discrete Coulomb centre at ET, is
given by [42]:
ZET ( )( )
DET q E 4E1:5 ð2m qÞ0:5 DET 5=3
cðFÞ ¼ exp þ exp 1 dE
kT kT kT 3
hF E
DET
ð8:35Þ
where, DET is the emission barrier lowering due to the Poole–Frenkel effect, and
this will be zero for a Dirac centre.
The simplified expression for a Dirac centre can be approximated by [42]:
8 !2 9
2p0:5 qhF < 1 qhF =
cðFÞ exp ð8:36Þ
ðkTÞ1:5 2ð2m Þ0:5 :3ðkTÞ3 2ð2m Þ0:5 ;
The exponential term in the above expression is quadratic in F, and the

depletion approximation relates the maximum field to the drain bias by Fmax /
Vd0:5 ; so that this simplified expression correctly predicts the observed exponential
dependence of leakage current on drain bias.
The field-free thermal emission rate, en0, is given in Chap. 2 by:
ET
en0 ¼ vrn NC exp ð8:37Þ
kT
where r is the electron capture cross section, v is the electron thermal velocity and
Nc is the effective density of conduction band states.
Equivalent expressions for ep(F) and ep0 can also be formulated for the hole
emission rates, and the leakage current density is given by integration across the
drain space charge region, of width d:
Zd
en ep
J ¼ qNT dx ð8:38Þ
en þ ep
0
Although the preceding expressions are for a discrete trap, and the poly-Si DOS
is distributed across the band-gap, current generation processes, via electron–hole
pair emission, are localised on near-mid-gap centres. This is in order to minimise
the sum en ? ep, and this process can be reasonably accurately represented by
discrete near-mid-gap states. In numerical simulations of the leakage current
process, the full trap distribution was used [40, 41], and the results remain con-
sistent with the above qualitative considerations, although it has been shown that,
at high fields, the most efficient generation centres can be up to 50 meV off mid-
gap [43] due to the Poole–Frenkel effect. An interesting aspect of those calcula-
tions was the observation that, for distributed traps, the current can be limited by
emission from the Coulomb state of the centre, located slightly deeper than EG/2,
whereas, for a discrete trap at EG/2, the current would normally be limited by the
slower emission from the Dirac state of the centre.
As with the hot carrier instability, the field enhanced leakage currents are a
direct consequence of the high drain fields in poly-Si TFTs, and field relief at the
drain, using LDD or GOLDD regions, will also reduce these field-enhanced cur-
rents [24, 41]. Although the DOS values in high quality TFTs will not influence the
drain field itself, the magnitude of the currents will scale with the trap density near
mid-gap, and, hence, with the DOS values near mid-gap. Also, in common with
the hot carrier instability, field enhanced leakage currents are observed in MOS-
FETs [44], and are not specific to poly-Si TFTs.
8.6 Other Bias-Stress Instabilities
Section 8.5.2 dealt with hot carrier induced instability, which is at a maximum
when high drain-bias-stress is combined with a low gate-bias setting approxi-
mately equal to the threshold voltage [23, 24]. Other important bias-stress
8.6 Other Bias-Stress Instabilities 283
combinations are high gate bias without drain bias, and combined high gate and
high drain bias.
8.6.1 Gate Bias Stress
8.6.1.1 Ionic Instability
In common with MOSFET technology, alkali ion contamination, particularly

sodium, has to be avoided, as, under the influence of positive gate bias-temperature
stress, negative threshold voltage shifts occur in poly-Si TFTs. The specific ion
contaminant can be identified from its characteristic ion current signature observed
during a gate bias sweep at 250 °C [45].
Gate bias stress can also result in threshold voltage shifts, due to charged ion
movement, in porous, poor quality gate oxides [46]. As discussed in Sect. 7.3.1,
the absorption of water in these films has been identified by characteristic thermal
desorption spectroscopy peaks at 100–200° C [47], and the presence of H+ and
OH- ions in the oxide has led to both negative and positive threshold voltage shifts
after positive and negative bias-stressing, respectively [46]. However, current
state-of-the-art gate oxide films, formed by PECVD deposition from either oxy-
gen-diluted tetraethylorthosilicate, TEOS [47], or helium-diluted silane and nitrous
oxide [48, 49] have deliberately employed dilution conditions to reduce the oxide
deposition rate, and to deposit dense, low porosity oxides, which are free from
water-induced instabilities.
8.6.1.2 Negative Bias-Temperature Instability
NBTI has been widely studied in MOSFET devices, where it is particularly seen in
p-channel transistors stressed with negative gate bias (\ 6MV/cm) at elevated
temperatures of 100–250 °C [50]. The stress causes an increase in the absolute
magnitude of threshold voltage, as well as degradation of sub-threshold slope and
gm. These changes are negligible under comparable positive bias-stress conditions,
and also negligible with both positive and negative gate bias-stress in n-channel
MOSFETs. The changes in device characteristics have been attributed to a com-
bined increase in trapping states at the Si/SiO2 interface, as well as positive charge
in the gate oxide. The physical model for the trapping state change is the de-
hydrogenation of previously passivated Si dangling bonds at the interface,
although the detailed process whereby this occurs, together with the increase in
oxide charge, is still a matter of discussion, with different models invoking the role
of mobile hydrogen ions or holes [50].
Similar instabilities have been reported in a wide range of p-channel poly-Si
TFTs, made from directly deposited LPCVD poly-Si, solid phase crystallised [51,
52] and excimer laser crystallised material [53–55]. In most of this work [51–53],
Fig. 8.21 Threshold voltage

instability of p-channel poly-
Si TFTs with negative gate
bias temperature stress
(NBTI) a time dependence at
different gate biases, b gate
bias dependence at different
temperatures (Reprinted from
IEEE)
the increase in threshold voltage was thermally activated over the temperature
range 297–430 K, with a power dependence on stress time, and an exponential
dependence on gate bias stress, as shown in Fig. 8.21 [53]. This dependence was
empirically represented by:
DVT / tn expðEa =kT Þ expðC jVG jÞ ð8:39Þ
where n was 0.28–0.34, Ea was *0.14 eV, and C was a fitting parameter whose
value was *0.1–0.13. The power time dependence was qualitatively similar to
that reported in MOSFETs [50], although the gate bias dependence differed
between authors. In the MOSFET work, there was a power dependence of E1.5 ox
(where Eox = {VG-VFB-Ws}/tox), rather than an exponential dependence. In other
TFT work, there was a linear dependence of DVT on Eox above a critical oxide
field of 0.5MV/cm [54, 55].
Notwithstanding some of the differences with the MOSFET data, the TFT
results have been broadly interpreted in the same way: namely, by an increase in
positive oxide charge, and an increase in trapping state density [55]. In one case
[53], the change in sub-threshold slope of the Id-VG characteristic was attributed to
interface state generation, and the near-threshold part of the characteristic was
analysed using the Levinson method [5], and identified an increase in the GB
trapping state density. However, as discussed in Sect. 8.3, all trapping states can be
represented by an effective density per unit surface area, and identifying the
specific physical location of the traps can be difficult. In other work on NSA TFTs
[54], numerical simulation of the characteristics after NBTI could satisfactorily
account for the NBTI changes by the introduction of an exponential distribution of
Si/SiO2 interface states, Nss(E), without any increase in positive oxide charge. In
contrast, SA TFTs showed linear increases in both Nss and oxide charge with Eox
[55].
In summary, NBTI is a well documented instability process in p-channel poly-
Si TFTs, which has been interpreted in terms of an increase in trapping states in the
material, where the state location has been attributed to both Si/SiO2 interface
states as well GB states, and the change in oxide charge seems to be sample
dependent.
8.6.2 Combined Gate and Drain Bias Stress
The combination of high gate and high drain bias-stress results in a large current
flow through the device, and this can lead to self-heating induced instabilities [55–
61], in which the modulus of the threshold voltage of both n-channel and
p-channel TFTs increased with increasing bias stress. In addition to the increase in
threshold voltage, a decrease in on-current, an increase in off-current, and a
degradation in sub-threshold slope have also been observed [56].
As might be expected, the self-heating effects, induced by different combina-
tions of gate bias and drain bias, were best represented by the stress-power
dissipation (= Id 9 Vd) within the device, and, for a given device geometry, the
bias-stress threshold voltage shifts could be uniquely correlated with this power
dissipation, as shown in Fig. 8.22 [56]. The self-heating effects have been con-
firmed by both direct infra-red detector measurements of the surface temperature
of the device [56, 58], as well as by numerical simulations of Joule heating within
the TFT [56, 57, 59, 61]. The measurements showed surface temperatures from
*100 °C up to *250 °C, depending upon the specific device geometry and bias
conditions [56, 58], and the magnitude of the bias-stress instability has been
Fig. 8.22 Dependence of

threshold voltage shift on
stress power, where Vd was
varied between 5 and 30 V,
and VG determined the drain
current. Stress time was 10 s
Physics)
correlated with the measured temperature of the device [56, 58], as shown in
Fig. 8.23 [58]. The TFTs used for this data set had different grain sizes, resulting in
increasing electron mobility with grain size. The mobility differences gave dif-
ferent on-currents for the same bias conditions, and the large grain devices showed
the largest threshold voltage shifts after stress. However, when the results were
normalised with respect to the measured device temperatures, the results, for
different device geometries as well as grain size, fell upon the common curve
shown in Fig. 8.23.
The device geometry was identified as a key parameter in determining the
internal temperature distribution. This was because the low thermal conductivity
of the glass substrate (*1.3 W/m/K) made it an inefficient route for the dissipation
of heat from the TFT channel, resulting in the device edges and its top surface
becoming significant channels for heat dissipation. Hence, the size of the device,
and particularly the width of the channel (for a given channel length), played an
important role in determining its temperature, with wider devices displaying both
higher temperatures and greater stress-induced threshold voltage shifts [56, 58,
61], as shown in Fig. 8.24 [56].
In agreement with the experimental measurements, the simulations also pre-
dicted channel temperatures of *120 to *300 °C [56, 57, 59, 61], depending
Fig. 8.23 Correlation of

self-heating threshold voltage
shift with device temperature
for TFTs of different grain
size and different channel
widths (Reprinted with
Fig. 8.24 Influence of

channel width on self-heating
threshold voltage shifts
Physics)
upon the biasing conditions and the device geometry. In addition, the simulations
demonstrated that, even when the temperature rise was not sufficient to cause bias-
stress instability, it was sufficient to induce a negative output conductance in high
mobility TFTs, due to increased phonon scattering of the channel electrons [59].
The threshold voltage instability has been correlated with an increase in the
trapping state density either near the middle of the poly-Si band-gap [56], or at the
Si/SiO2 interface [61], and attributed to thermally-induced Si–H bond breaking
either in the poly-Si grain boundaries [56] or at the interface [61]. Other workers
have suggested electron injection into the gate oxide [57, 58], in analogy with hot
carrier defect generation [57]. However, some of the results have been obtained at
drain bias values less than the saturation voltage [56], which would keep the drain
field low, and militate against hot carrier injection in those cases.
For p-channel TFTs, there is a similarity with the NBTI effects described in the
preceding section, in which the drain-bias-induced self-heating plays the role of
the deliberate temperature increase used to observe NBTI. Some authors have
correlated the two effects, and argued that the self-heating instability is a mani-
festation of NBTI, in which the threshold voltage shift was found to be propor-
tional to the oxide field, Eox [54, 55] (see the discussion of Eq. 8.39). Hence, the
only difference between the two effects was that, with the self-heating instability,
the varying channel potential, due to the drain bias, caused a variation of oxide
field, Eox, along the channel, resulting in spatially non-uniform interface state and
positive oxide charge creation. The device characteristics after self-heating stress
were successfully simulated using the calculated values of the oxide field, Eox, to
fit a spatially varying interface state density and oxide charge density to the
experimental characteristics [55]. The experimental and simulated transfer char-
acteristics, before and after self-heating bias-stress, are shown in Fig. 8.25a, and,
for the 10,000s stressed sample, the calculated interface state density at the source
end of the channel was five times higher than at the drain end, due to the larger
value of Eox at the source. This spatial asymmetry in the interface state density also

(symbols) and simulated
(lines) results of self-heating
bias stress a transfer
characteristics before and
after stressing at VG = -
15 V and Vd = -9 V, and
b output characteristics after
10,000 s stress with normal
and reversed source/drain
contacts (Reprinted from [55]
successfully explained the change in output characteristics when the source and
drain terminals were reversed, as shown in Fig. 8.25b. When the source terminal,
with the higher interface state density was used as the drain, this higher density
increased the drain field, causing a stronger avalanche current, as seen in the
figure.
Whilst self-heating effects have been well established as a reliability issue in
poly-Si TFTs on glass, they are a bigger issue with TFTs on flexible, polymer
substrates [60, 61], where the thermal conductivity is even lower at *0.2 W/m/K.
Some common solutions have been suggested for both types of substrate, such as
multi-fingered devices to avoid large W values, as well as overall device scaling to
reduce both device area and drain biases, whilst retaining the required levels of
channel current at lower power levels [60]. Alternatively, reduced self-heating
effects have been demonstrated with thermally conducting flexible substrates, such
as copper [57] or stainless steel [61] foils. The use of thin foils is discussed further
in Chap. 11, which reviews the field of flexible substrate technologies for TFTs.
8.7 Short Channel Effects
Short channel effects (SCEs) have been well established, and widely studied, in
c-Si MOSFETs [62], and they refer to a range of phenomena for which the
8.7 Short Channel Effects 289
classical long channel model, described in Chap. 3, no longer applies. As channel

length, L, is reduced, examples of this behaviour are a drain current, which rises
more strongly than 1/L, threshold voltage, which reduces with channel length, sub-
threshold currents which increase with drain bias, Vds, and poor saturation in the
output characteristics. These are a consequence of a number of inter-related
phenomena, such as the size of the source and drain space charge regions
becoming comparable to the channel length. This reduces the amount of charge
needed on the gate to invert the channel surface, and decreases the threshold
voltage with reducing channel length. Also, when the drain space charge region is
comparable to channel length, such that the source and drain space charge regions
start to overlap, increased drain bias can reduce the potential barrier between the
source and the channel (drain-induced barrier lowering, DIBL), giving drain-bias-
dependent sub-threshold currents and poor current saturation in the output char-
acteristic. Strategies have been developed to minimise these effects, from simply
reducing the gate oxide thickness and/or increasing the substrate doping level.
Both of these changes increase the field at the drain (for a given drain bias), as
discussed in Sect. 8.5.1, and reduce the spread of the space charge regions along
the channel. More sophisticated changes include modifications in device design to
tailor the doping level to optimise overall device performance, through to the use
of field relief structures [62].
Whilst the gate length in deep sub-micron MOSFETs has been progressively
shrunk to the current production values of *50 nm [63], channel length reduction
in poly-Si TFTs has been more modest, with the minimum reported dimension
being *0.5 lm [64–66]. Nevertheless, a variety of short channel effects have been
seen in poly-Si TFTs having a channel length of 2–3 lm or less, with the effects
becoming stronger as the channel length reduced. The results in the following sub-
sections have been taken from SA n-channel, laser crystallised TFTs made in
40 nm thick layers of poly-Si. Some of the effects seen are classical short channel
effects, such as a gate length dependent threshold voltage [67], and poor output
impedance, as shown in Fig. 8.26a and b, respectively, and both can be mitigated
by using a reduced oxide thickness. The threshold voltage measurements in
Fig. 8.26a were made at a low drain bias of 0.1 V, and at normalised channel
currents of W/Lx10-7A, both of which were chosen to minimise other short
channel effects to be discussed further below. The solid lines are simulation results
derived from a double exponential DOS whose parameters were adjusted in order
to give the best fit to the Id-VG characteristics across the range of channel lengths
[67]. Both the experimental data and the simulations confirm the reduction in
SCEs due to a reduced gate oxide thickness. The comparison of the Id-VD char-
acteristics in Fig. 8.26b were made using different gate voltages for the two oxide
thicknesses in order to normalise the measurements to common values of low field
channel conductance. The TFTs with the thicker gate oxide showed very poor
current saturation, compared with the thinner gate oxide samples. To quantify
the effect of the thinner gate oxide, the current in saturation can be represented
by [68]:
Fig. 8.26 Classical short channel effects in 1lm channel length SA poly-Si TFTs a variation of
threshold voltage with L and gate oxide thickness (Reprinted from [67] Copyright (2005), with
permission from Elsevier), and b comparison of output characteristics for two different gate oxide
thicknesses
Id ¼ Ido ð1 þ kVd Þ ð8:40Þ

for Vd [ Vd(sat), where k is commonly used in circuit modelling as an output
characteristic quality factor, whose value is zero for devices with perfect saturation
characteristics. The values for k are shown in Fig. 8.26b, and they display an
8-fold improvement with the thinner gate oxide.
In addition to these classical effects, a number of other performance artefacts
have been reported in short channel poly-Si TFTs, which have included the
presence of parasitic contact resistance [65, 67, 69], the ‘kink’ effect in the output
characteristics at high drain bias [27], and drain bias dependent sub-threshold
characteristics [72]. The latter two effects are related to floating body effects in the
TFTs, and will be discussed separately below. As will be apparent in that section,
the floating body related SCEs have features, which are also observed in floating
body SOI MOSFETs.
8.7.1 Parasitic Resistance Effects
When the linear regime channel current, in variable length TFTs, is normalised by
channel length, a common curve would be expected as a function of gate bias. In
contrast, the normalised currents, measured in SA n-channel TFTs, decreased with
decreasing channel length, as shown in Fig. 8.27a [69]. This figure demonstrates
that only the longest channel device had an approximately linear dependence of
current on gate bias, and that, with reducing channel length, the expected linearity
deteriorated as the currents failed to scale with 1/L. These results, together with the
Fig. 8.27 TFT transfer

characteristics, normalised by
channel length
a experimental data, and
b simulations (also shown in
the insert is the location of
the ion-damaged regions in
the simulation model)
permission of IEEE)
almost saturated nature of the L = 0.5 lm characteristic, are indicative of a par-

asitic resistance, Rp, in the channel. The total resistance of the device, Rmeas, will
be given by the sum of the channel resistance/unit length, Rch, the doped source
and drain resistance, Rn, as well as Rp, i.e.
Rmeas ¼ Vd =Id ¼ Rch L þ Rn þ Rp ð8:41Þ
and, from a plots of 1/Id versus L, at different values of VG, the channel resistance
can be calculated from the slope, and the average value of Rp across the different
channel lengths can be extracted from the intercept (where Rn can be calculated
from the sheet resistance of the source/drain dopant and the geometry of these
doped regions). The carrier mobility can be derived from the normalised channel
resistance, using the linear regime expression for TFT channel current,
Rch ¼ 1=flWCox ðVG Vt Þg ð8:42Þ
For the TFTs in Fig. 8.27a, this analysis yielded a value of Rp which was
*600ohms [69]. For gate voltages [ 6.5 V, this was more than the channel
resistance of TFTs with channel lengths \ 1 lm, and was responsible for the
almost VG–independent channel currents seen in the L = 0.5 lm TFTs in
Fig. 8.27a.
It is significant that this effect was seen only with SA n-channel TFTs, and not
with NSA n-channel TFTs nor with SA p-channel TFTs, and, as discussed in
Sect. 7.4.1.1, it has been physically associated with residual phosphorus ion doping
damage in the poly-Si near the edges of the gate. As also discussed in that section,
the magnitude of this damage was determined by the process details, and, in
modelling this effect [69], for both high and low values of Rp, it has been found
appropriate to represent this damaged area by two adjacent high DOS regions, one,
*50 nm wide, immediately beneath the gate, and the other, *200 nm wide,
extending from the gate edge into the doped regions. Using this model, with DOS
Fig. 8.28 Output

characteristics for different
channel length TFTs
(normalised by channel
length) a experimental data,
and b simulations, with and
without impact ionisation
switched on (Reprinted with
values in region II *10 times higher than in the channel, it was successfully fitted
to the experimental data as shown in Fig. 8.27b.
As is apparent, this SCE is quite different from the classical effects seen in
MOSFETs, and is a consequence of the particular processing schedule favoured
for SA n-channel poly-Si TFTs.
8.7.2 Floating Body Effects
8.7.2.1 Kink Effect
Figure 8.28a shows measurements of the output characteristics for a range of TFT
channel lengths from L = 20 lm to L = 1 lm [27], where the characteristics
have been measured at a low enough value of gate bias to avoid the parasitic
resistance effects discussed in the preceding section. In this case, the linear regime
currents showed good scaling with L, but the saturation currents showed a strong L
dependence, with increasingly large non-saturating currents appearing at succes-
sively lower values of drain bias. In analogy with SOI MOSFETs, it is these
anomalous currents, which have been referred to as the kink effect. However, in
non-fully depleted SOI devices, the kink is seen as a step in the saturation current
in the output characteristic [70], rather than the continuously increasing current
seen in Fig. 8.28a. These latter characteristics are much more similar to the
behaviour of fully-depleted SOI devices [71], in which there is not so much a step
in the saturation current, but more of a continuous increase, which has been
compared to the open-base breakdown voltage in a bipolar transistor [73].
2-D simulations were used to analyse the results in Fig. 8.28a, where a double
exponential DOS was established by fitting to the low Vds transfer characteristics
of a long channel length device, and optimised impact ionisation parameters were
used to establish a fit to the high Vds output characteristics. The impact ionisation
generation rate, GII, was given by [72]:
GII ¼ nvn an þ pvp ap ð8:43Þ
where vn (vp) is the electron (hole) velocity, and an (ap) is the electron (hole)
ionisation coefficient, which is given by the Chynoweth law:

1 FC
an;p ¼ an;p exp ð8:44Þ
Fll
and FC is the critical electric field and Fll is the field parallel to the current flow.
The simulated characteristics are shown in Fig. 8.28b, where FC = 2.9 9 106
V/cm and an ¼ 5 105 cm1 [27].
The model was used to examine the relative roles of DIBL and avalanching in
the kink effect by comparing simulations with and without the impact ionisation
effect switched on. As seen in Fig. 8.28b, impact ionisation was needed to simulate
the results. During impact ionisation, electron-hole pairs are produced in the high
field region near the drain junction, and the electrons are swept into the drain by
this field. In the floating body of the TFT, the holes have to flow through the poly-
Si layer towards the source junction, and a steady state concentration of holes will
be stored in the layer. This forward biases the junction, permitting a hole injection
current into the source. The forward biased source junction will also inject further
electrons into the channel, thereby generating parasitic bipolar transistor, PBT,
action. The PBT contribution to the overall current, Ids, can be analysed in the
following way [27]:
Ids ¼ MðIch þ Ie Þ ¼ Ich þ Ie þ Ib ð8:45Þ
Where M is a multiplication factor, Ich is the channel current in the absence of
avalanching, and Ie is the emitter current due to the PBT action. Ib is the base
current, given by the flow of holes generated by impact ionisation at the drain
(which was calculated from the integral of the impact ionisation rate). The emitter
current gain, b, is given by:
b ¼ ðIe Ib Þ=Ib ð8:46Þ
and the excess device current, DI, due to impact ionisation and the PBT is:
DI ¼ Ids Ich ¼ Ie þ Ib ð8:47Þ
DI=Ids ¼ ðb þ 2ÞðM 1Þ=M ð8:48Þ

Fig. 8.29 Simulated results

showing the emitter current
gain, b, as a function of drain
bias, for different channel
length TFTs (Reprinted with
Hence, the normalised excess current is a function of M and b. The calculated

values of M-1 varied with drain bias from *10-8 to 10-2, and the corresponding b
values are shown in Fig. 8.29. It will be seen that b is greatest, and displays the
strongest L dependence, at low values of Vds, where the hole injection is weakest,
and the junction injection efficiency is greatest. With increasing Vds, the avalanche
rate and Ib increase, which causes a progressive lowering of the source barrier,
such that each increment in hole current produced a reduced relative increment in
electron current due to the lower barrier height. Nevertheless, the values of b range
from [105, at low Vds, to [10, at large Vds, and demonstrate that the basic ava-
lanche current, Ib, is significantly magnified by the PBT effect. In other words,
whilst the kink effect is initiated by impact ionisation in the high drain field, the
majority of the current flowing through the device is the result of current ampli-
fication due to parasitic transistor action in the floating ‘base’ of the device. This is
analogous to the amplification of the collector–base leakage current in an open-
base bipolar transistor [73]. The contribution of the injected current, Ie, as a
function of channel length, L, (at Vds = 4 V, where b [ 100), is illustrated by the
current component comparison in Fig. 8.30 [27]. This figure shows the low values
of Ib compared to Ids at all values of L, but, at L = 1lm, the major component of
the total current was due to Ie and the PBT effect. In contrast, for L = 20 lm, the
PBT effect was much weaker, and the current was essentially a pre-kink, basic
channel current.
These TFT results, including the dependence of b and M-1 on Vds, are similar
to those reported in fully depleted SOI MOSFET devices [71].
8.7.2.2 Sub-Threshold and Threshold Voltage Effects
This section deals with the drain bias dependence of both sub-threshold currents
and of threshold voltage. Whilst these are essentially classical MOSFET short
channel effects, the involvement of floating body effects in poly-Si TFTs [72]
warrants further discussion.
Figure 8.31 compares the drain bias, Vds, dependent characteristics of short
(L = 0.4 lm) and long (L = 20 lm) channel TFTs, in which the L = 20 lm TFT
displays classical long channel behaviour, with the sub-threshold currents being
Fig. 8.30 Simulated results

showing the different current
components of Ids as a
function of channel length (at
Vds = 4 V) (Reprinted with
independent of Vds, while the L = 0.4 lm TFT displays a strong Vds dependence
[72]. This dependence will also manifest itself as a Vds-dependent threshold
voltage, as shown in Fig. 8.32, where the threshold voltage was measured at
Id = 10-7W/L A to minimise series resistance effects. Also shown by the solid
lines in this figure are fitted simulation results obtained from optimised DOS and
impact ionisation parameters [72], as discussed in the previous section. As will be
seen, the influence of drain bias on VT was much stronger in the short channel
TFTs, and is ostensibly similar to a classical short channel DIBL effect.
However, given the presence of impact ionisation in the simulated results, the
relative contributions of the direct DIBL and the PBT effects, which will also
lower the source-channel barrier, have been clarified [72]. Figure 8.33 shows the
simulations of VT, with and without drain avalanching turned on. At low Vds, the
DIBL was primarily responsible for the barrier lowering, but the effect of hole
accumulation, via the PBT effect, reduced the barrier at higher values of Vds. As
would be expected from Fig. 8.31, both effects increased as the channel length
decreased. These barrier-lowering effects were also directly seen in simulations of
Fig. 8.31 TFT transfer

characteristics measured as a
function of drain bias
a L = 0.4 lm, and
b L = 20 lm (Reprinted with

measured (symbols) and
simulated (lines) threshold
voltage as a function of drain
bias for different channel
length TFTs (Reprinted with
Fig. 8.33 Simulated values

of threshold voltage as a
function of drain bias, with
(solid lines) and without
(broken lines) impact
ionisation switched on
Physics)
the source barrier height itself [72], with the influence of PBT lowering being
much smaller in the longer channel devices.
As can be appreciated, the kink effect discussed in the previous section, and the
VT effects in this section, are related phenomena. Both are caused by the gener-
ation of impact ionisation holes at the drain junction, leading to PBT action in the
floating body of the TFT. The kink effect discussion introduced these concepts,
and focussed on the current gain within the device, and its impact on the total
current flow. This section has looked in more detail at the relative contributions of
DIBL and the PBT effect to the source barrier height in short channel TFTs.
8.8 Summary
In this chapter, various aspects of the characterisation and performance of poly-Si

TFTs have been considered. Whilst much current analysis of poly-Si TFTs is based
upon the use of 2-D numerical simulators, a simple analytical model of grain
boundary conduction was introduced in Sect. 8.2, as this embodies some key
physical considerations of carrier transport through poly-crystalline material. The
limitations of this model were discussed, and, in Sect. 8.3, the DOS in poly-Si was
shown to be continuous across the band-gap, rather than based upon a discrete
trapping level (as assumed in the analytical model). It was found that the DOS
8.8 Summary 297
could be well represented by a pair exponentials, which could be related to band

edge states and deep states.
In Sect. 8.4, the low-field TFT leakage currents were characterised by an
analytical carrier generation rate and transport model, which encompassed all the
key features of the drain voltage and channel length dependences of the leakage
current in TFTs containing both high and low DOS values. In addition to this
classical low field phenomenon, poly-Si TFTs display a number of undesirable
artefacts, such as low output impedance, drain bias stress instability and field
enhanced leakage currents, which are associated with a large electrostatic field at
the drain junction. The drain field, in high quality TFTs, is essentially determined
by the 2-D architecture of the device, rather than by the DOS level, and the
mechanisms of hot carrier instability and field enhanced leakage currents were
described in Sect. 8.5. Other bias instability processes in poly-Si TFTs, including
ionic drift under gate bias stress, threshold voltage instability under elevated
temperature negative gate stress, and self-heating instabilities under combined gate
and drain bias stress were discussed in Sect. 8.6.
Short channel effects, SCEs, were presented in Sect. 8.7, where it was shown
that some of these effects were comparable to the well-established SCEs in
MOSFETs. However, others, such as parasitic channel resistance effects, are
specific to certain types of n-channel TFT technology. A further class of SCEs was
enhanced by parasitic bipolar transistor action due to floating body effects in the
devices, and are comparable to similar phenomena in silicon-on-insulator
MOSFETs.
The extensive investigation of poly-Si TFTs has identified the performance
artefacts discussed in this chapter, and the understanding and control of these
effects has enabled the mass production of stable, high performance poly-Si TFTs
for a variety of flat-panel display applications. As discussed in Chap. 7, these have
focussed on the small/medium diagonal display market, and include both AM-
LCDs and AMOLEDs.
References
1. Seto JYW (1975) The electrical properties of polycrystalline silicon films. J Appl Phys
46(12):5247–5254
2. Lu NC-C, Gerzberg L, Lu C-Y, Meindl JD (1981) Modeling and optimization of monolithic
polycrystalline silicon resistors. IEEE Trans ED 28(7):818–830
3. Baccarani G, Ricco B, Spadini G (1978) Transport properties of polycrystalline silicon films.
J Appl Phys 49(11):5565–5570
4. Sze SM (1981) Physics of semiconductor devices, 2nd edn. Wiley, New York, p 255
5. Levinson J, Shepherd FR, Scanlon PJ, Westwood WD, Este G, Rider M (1982) Conductivity
behaviour in polycrystalline semiconductor thin film transistors. J Appl Phys
53(2):1193–1202
6. Voutsas AT (2003) Assessment of the performance of laser-based lateral-crystallisation
technology via analysis and modelling of polysilicon thin film transistor technology. IEEE
Trans ED 50(6):1494–1500
7. Ayres JR, Brotherton SD, Young ND (1992) Low temperature poly-Si for liquid crystal
display addressing. Optoelectron Devices Technol 7(2):301–320
8. Kim DM, Khondker AN, Ahmed SS, Shah RR (1984) Theory of conduction in polysilicon:
drift-diffusion approach I crystalline-amorphous-crystalline semiconductor system—part I:
small signal theory. IEEE Trans ED 31(4):480–493
9. Brotherton SD, Ayres JR, Young ND (1991) Characterisation of low-temperature poly-Si thin
film transistors. Solid-State Electron 34(7):671–679
10. Fortunato G, Migliorato P (1986) Determination of gap state density in polycrystalline silicon
by field-effect conductance. Appl Phys Lett 49(16):1025–1027
11. Ayres JR (1993) Characterisation of trapping states in polycrystalline-silicon thin film
transistors by deep level transient spectroscopy. J Appl Phys 74(3):1787–1792)
12. Hack M, Shaw JG, LeComber P, Willums W (1990) Numerical simulations of amorphous
and polycrystalline silicon thin-film transistors. Jpn J Appl Phys 29(12):L2360–L2362
13. Mariucci L, Giacometti F, Pecora A, Massussi F, Fortunato G, Valdinoci M, Colalongo L
(1998) Numerical analysis of electrical characteristics of polysilicon thin film transistors
fabricated by excimer laser crystallisation. Electron Lett 34(9):924–926
14. Kong H-S, Lee C (1995) Two-dimensional simulation study of field-effect operation in
undoped poly-Si thin-film transistors. J Appl Phys 78(10):6122–6131
15. Chou T-KA, Kanicki J (1999) Two-dimensional numerical simulation of solid-phase-
crystallised polysilicon thin-film transistor characteristics. Jpn J Appl Phys 38(4B):
2251–2255
16. Valletta A, Bonfiglietti A, Rapisarda M, Mariucci L, Fortunato G, Brotherton SD (2007)
Grain boundary evaluation in sequentially laterally solidified polycrystalline-silicon devices.
J Appl Phys 101(9):94502
17. Valletta A, Mariucci L, Fortunato G, Brotherton SD (2003) Surface-scattering effects in
polycrystalline silicon thin-film transistors. Appl Phys Lett 82(18):3119–3121
18. Canali C, Majni G, Minder R, Ottaviani G (1975) Electron and hole drift velocity
measurements in silicon and their empirical relation to electric field and temperature. IEEE
Trans ED 22:1045–1047
19. Lombardi C, Manzini S, Saporito A, Vanzi M (1988) A physically based mobility model for
numerical simulation of nonplanar devices. IEEE Trans CAD 7(11):1164-1171
20. Brotherton SD (1995) Polycrystalline silicon thin film transistors. Semicond Sci Technol
10:721–738
21. Brotherton SD, Ayres JR, Edwards MJ, Fisher CA, Glaister C, Gowers JP, McCulloch DJ,
Trainor M (1999) Laser crystallised poly-Si TFTs for AMLCDs. Thin Solid Films
337:188–195
22. Brotherton SD, Ayres JR, Trainor MJ (1996) Control and analysis of leakage currents in
poly-Si TFTs. J Appl Phys 79(2):895–904
23. Ayres JR, Brotherton SD (1996) Influence of drain field on poly-Si TFT behaviour. In:
Proceeding SID EuroDisplay’96, pp 33–36
24. Ayres JR, Brotherton SD, McCulloch DJ, Trainor M (1998) Analysis of drain field and hot
carrier stability of poly-Si TFTs. Jpn J Appl Phys 37(4a):1801–1808
25. Valdinoci M, Colalongo L, Baccarani G, Fortunato G, Pecora A, Policicchio I (1997)
Floating body effects in polysilicon thin-film transistors. IEEE Trans ED 44(12):2234–2241
26. Young ND, Gill A, Edwards MJ (1992) Hot carrier degradation in low temperature processed
polycrystalline silicon thin film transistors. Semicond Sci Technol 7(9):1183–1188
27. Valletta A, Gaucci P, Mariucci L, Fortunato G, Brotherton SD (2004) Kink effect in short
channel polycrystalline silicon thin film transistors. Appl Phys Lett 85(15):3113–3115
28. Mariucci L, Fortunato G, Carluccio R, Pecora A, Giovannini S, Massussi F, Colalongo L,
Valdinoci M (1998) Determination of hot-carrier induced interface state density in
polycrystalline silicon thin-film transistors. J Appl Phys 84(4):2341–2348
29. Hofmann KR, Werner C, Weber W, Dorda G (1985) hot-electron and hole-emission effects in
short n-channel MOSFETs. IEEE Trans ED 32(3):691–699
References 299
30. Young ND (1996) Formation and annealing of hot-carrier-induced degradation in poly-Si

TFTs, MOSFETs and SOI devices, and similarities to state-creation in a-Si:H. IEEE Trans
ED 43(3):450–456
31. Brown TM, Migliorato P (2000) Determination of the concentration of hot-carrier-induced
bulk defects in laser-recrystallized polysilicon thin film transistors. Appl Phys Lett
76(8):1024–1026
32. Valletta A, Mariucci L, Fortunato G (2006) Hot-carrier-induced degradation of LDD
polysilicon TFTs. IEEE Trans ED 53(1):43–50
33. Tam S, Ko P, Hu C (1984) Lucky-electron model of channel hot-electron injection in
MOSFETs. IEEE Trans ED 31(9):1116–1125
34. Hu C, Tam SC, Hsu F-C, Ko P-K, Chan T-Y, Terrill KW (1985) Hot-electron-induced
MOSFET degradation—model, monitor, and improvement. IEEE Trans ED 32(2):375–385
35. Valletta A, Mariucci L, Fortunato G, Brotherton SD, Ayres JR (2002) Hot carrier-induced
degradation of gate overlapped lightly doped drain (GOLDD) polysilicon TFTs. IEEE Trans
ED 49(4):636–642
36. Glasse C, Brotherton SD, French ID, Green PW, Rowe C (2003) Short channel TFTs made
with sidewall spacer technology. In: Proceeding AMLCD’03, pp 317–320
37. Schwerin A, Hansch W, Weber W (1987) The relationship between oxide charge and device
degradation: a comparative study of n- and p-channel MOSFETs. IEEE Trans ED
34(12):2493–2500
38. Mariucci L, Gaucci P, Valletta A, Templier F, Fortunato G (2007) Hot carrier effects in p-
channel polysilicon thin film transistors fabricated on flexible substrates. Jpn J Appl Phys
46(3B):1299–1302
39. Migliorato P, Reita C, Tallarida G, Quinn M, Fortunato G (1995) Anomalous off-current
mechanisms in n-channel poly-si thin film transistors. Solid State Electron 38(12):2075–2079
40. Colalongo L, Valdinoci M, Baccarani G, Migliorato P, Tallarida G, Reita C (1997)
Numerical analysis of poly-TFTs under off conditions. Solid State Electron 41(4):627–633
41. Bonfiglietti A, Cuscuna M, Valletta A, Mariucci L, Pecora A, Fortunato G, Brotherton S D,
Ayres JR (2003) Analysis of electrical characteristics of GOLDD poly-Si TFTs with different
LDD doping concentration. IEEE Trans ED 50(2):2425–2433
42. Vincent G, Chantre A, Bois D (1979) Electric field effect on the thermal emission of traps in
semiconductor junctions. J Appl Phys 50(8):5484–5487
43. Lui OKB, Migliorato P (1997) A new generation-recombination model for device simulation
including the Poole-Frenkel effect and phonon-assisted tunneling. Solid State Electron
41(4):575–583
44. Nathan V, Das NC (1993) Gate-induced drain leakage current in MOS devices. IEEE Trans
ED 40(10):1888–1890
45. Young ND, Gill A, Clarence IR (1989) Mobile ion effects in low-temperature silicon oxides.
J Appl Phys 66(1):187–190
46. Young ND, Gill A (1992) Water-related instability in TFTs formed using deposited gate
oxides. Semicond Sci Technol 7:1103–1108
47. Hirashita N, Tokitoh S, Uchida H (1993) Thermal desorption and infrared studies of plama-
enhanced chemical vapor deposited SiO films with tetraethylorthosilicate. Jpn J Appl Phys
32(4):1787–1793
48. Batey J, Tierney E (1986) Low-temperature deposition of high-quality silicon dioxide by
plasma-enhanced chemical vapor deposition. J Appl Phys 60(9):3136–3145
on glass and polymer substrates. In: Proceeding Asia Display’98, pp 83–93
50. Schroder DK, Babcock JA (2003) Negative bias temperature instability: Road to cross in
deep submicron silicon semiconductor manufacturing. J Appl Phys 94(1):1–18
51. Maeda S, Maegawa S, Ipposhi T, Nishimura H, Ichiki T, Mitsuhashi J, Ashida M, Muragishi
T, Inoue Y, Nishimura T (1994) Mechanism of negative-bias temperature instability in
polycrystalline silicon thin film transistors. J Appl Phys 76(12):8160–8166
52. Maeda S, Maegawa S, Ipposhi T, Kuriyama H, Ashida M, Inoue Y, Miyoshi h, and Yasuoka
A (1998) An analytical method of evaluating variation of the threshold voltage shift caused
by the negative-bias temperature stress in poly-Si TFTs. IEEE Trans ED 45(1):165–172
53. Chen C-Y, Lee J-W, Wang S-D, Shieh M-S, Lee P-H, Chen W-C, Lin H-Y, Yeh K-L, Lei T-
F (2006) Negative bias temperature instability in low-temperature polycrystalline silicon
thin-film transistors. IEEE Trans ED 53(12):2993–3000
54. Mariucci L, Gaucci P, Valletta A, Cuscunà M, Maiolo L, Pecora A, Fortunato G (2009)
Negative bias–temperature stress in non-self-aligned p-channel polysilicon TFTs. Thin Solid
Films 517:6379–6382
55. Gaucci P, Valletta A, Mariucci L, Pecora A, Maiolo L, Fortunato G (2010) Analysis of self-
heating related instability in self-aligned p-channel polycrystalline silicon thin film
transistors. IEEE Electron Dev Lett 31(8):830–832
56. Inoue S, Ohshima H, and Shimoda T (2002) Analysis of degradation phenomenon caused by
self-heating in low-temperature-processed polycrystalline silicon thin film transistors. Jpn J
Appl Phys 41(11a):6313–6319
57. Takechi K, Nakata M, Kanoh H, Otsuki S, Kaneko S (2006) Dependence of self-heating
effects on operation conditions and device structures for polycrystalline silicon TFTs. IEEE
Trans 53(2):251–257
58. Uraoka Y, Kitajima K, Kirimura H, Yano H, Hatayama T, Fuyuki T (2005) Degradation of
low temperature poly-Si thin film transistors depending on grain boundaries. Jpn J Appl Phys
44(5A):2895–2901
59. Valletta A, Moroni A, Mariucci L, Bonfiglietti A, Fortunato G (2006) Self-heating effects in
polycrystalline silicon TFTs. Appl Phys Lett 89:093509-1–093509-3
60. Miyasaka M, Hara H, Karaki N, Inoue S, Kawai H, Nebashi S (2008) Technical obstacles to
thin film transistor circuits on plastic. Jpn J Appl Phys 47(6):4430–4435
61. Fortunato G, Cuscuna M, Gaucci P, Maiolo L, Mariucci L, Pecora A, Valletta A, Templier F
(2009) Self-heating effects in p-channel polysilicon TFTs fabricated on different substrates.
J Korean Physical Society 54(1):455–462
62. Sze SM, Ng KK (2007) Physics of semiconductor devices, 3rd edn. Wiley, New York,
pp. 328–347
63. International Technology RoadMap for Semiconductors, 2009 edn. (http://www.itrs.net/
Links/2009ITRS/2009Chapters_2009Tables/2009_ExecSum.pdf. Accessed July 2010
64. Iriguchi C, Higashi S, Inoue S, Shimoda T, Nobari AR, Iba T (2003) 0.5 lm fine-gate TFT
fabrication on 300 9 300 mm2 glass substrate using holographic microlithography. SID 03
Digest 1092–1095
65. Brotherton SD, Glasse C, Glaister C, Green P, Rohlfing F, Ayres JR (2004) High-speed,
short-channel polycrystalline silicon thin-film transistors. Appl Phys Lett 84(2):293–295
66. Miyano S, Kitagishi Y, Okada T, Kawachi G (2009) High-frequency thin-film transistor
circuits for system displays. Jpn J Appl Phys 48:03B013-1–03B013-5
67. Fortunato G, Valletta A, Gaucci P, Mariucci L, Brotherton SD (2005) Short channel effects in
polysilicon thin film transistors. Thin Solid Films 487:221–226
68. Shichman H, Hodges DA (1968) Modeling and simulation of insulated-gate field-effect
transistor switching circuits. IEEE J Solid-State Circs 3(3):285–289
69. Gaucci P, Valletta A, Mariucci L, Fortunato G, Brotherton SD (2006) Numerical simulation
of parasitic resistance effects in polycrystalline silicon TFTs. IEEE Trans ED 53(3):573–577
70. Kato K, Wada T, Taniguchi K (1985) Analysis of kink characteristics in silicon-on-insulator
MOSFETs using two-carrier modeling. IEEE Trans ED 32(2):458–462
71. Choi J-Y, Fossum JG (1991) Analysis and control of floating-body bipolar effects in fully
depleted submicrometer SOI MOSFETs. IEEE Trans ED 38(6):1384–1391
72. Valletta A, Gaucci P, Mariucci L, Pecora A, Cuscunà M, Maiolo L, Fortunato G (2010)
Threshold voltage in short channel polycrystalline silicon thin film transistors: Influence of
drain induced barrier lowering and floating body effects. J Appl Phys 107:074505-1–074505-9
Chapter 9
Transparent Amorphous Oxide
Semiconductor TFTs
Abstract Transparent amorphous oxide semiconductors, especially a-InGaZnO,

are the most recent TFT materials of interest for driving active matrix displays.
The high level of activity on these devices is because, even though the material is
amorphous, it can offer a carrier mobility of 10 cm2/Vs, or more. This is 10–20
times greater than a-Si:H, and the higher mobility is advantageous for driving
organic light emitting diode, OLED, displays. Also, due to the amorphous nature
of the material, it may have better uniformity than poly-Si, which is the currently
preferred TFT technology for commercial, hand-held AMOLED displays. In this
chapter, the material and electronic properties of AOS materials are reviewed,
paying particular attention to a-InGaZnO TFTs. Other topics include device
architecture and fabrication, the DOS and conduction mechanisms within the
material, overall device performance, and bias stability issues.
9.1 Introduction
The current interest in transparent amorphous oxide semiconductor, AOS, TFTs

was stimulated by the publication, in 2004, of the operation of amorphous indium-
gallium-zinc oxide, a-IGZO, TFTs on low temperature, flexible, polymer sub-
strates [1]. Since that time, the work in the AOS area has rapidly developed, with
a-IGZO TFT addressed active matrix displays being demonstrated for LCDs [2],
OLEDs [3], and electrophoretic e-readers [4]. This interest is due to the particular
properties of AOS TFTs, which make them well suited to the FPD application.
Firstly, this amorphous material has Hall-effect and field-effect electron mobilities
of *10 cm2/Vs, or more, which is [10 times higher than a-Si:H [5]. This makes
the material more appropriate than a-Si:H for driving AMOLED displays, and it is
even relevant for addressing 120 Hz, high resolution AMLCDs, where a carrier
mobility of *2 cm2/Vs is required for display diagonals [90 inches [5, 6].
Indeed, the mass-production of large size, high resolution AMLCD TVs with AOS
pixel TFTs has been forecast to begin in 2012 [7]. In addition, its amorphous
nature should mean that it will have better uniformity than poly-crystalline

302 9 Transparent Amorphous Oxide Semiconductor TFTs
materials [5]. Hence, there is interest in AOS TFTs for large diagonal AMOLED
TVs, where poly-Si (which is presently used in small diagonal AMOLED displays)
is perceived to be too non-uniform for this application [3]. However, this argument
is not yet proven, and, as discussed in Sect. 7.2.2.4, advanced crystallisation
techniques have improved the uniformity of laser crystallised poly-Si [8].
In addition to the higher electron mobility compared with a-Si:H, a-IGZO TFTs
also have a better overall performance, with state-of-the-art devices having a sub-
threshold slope of 0.12 V/dec [10], compared with 0.4 V/dec in a high quality
a-Si:H TFT [11]. Both these mobility and sub-threshold slope differences are
indicative of a lower density of trapping states near the conduction band edge in
a-IGZO [5].
From a device fabrication point of view, the material can be deposited by
sputtering at temperatures down to room temperature, and a further attractive
feature of a-IGZO TFTs is that they can be fabricated with a simple contact
metallisation scheme, without doped contacts, and a wide range of gate dielectrics
have given acceptable performance [5].
These features of AOS TFTs are discussed in the following sections, starting
with the background material properties in Sect. 9.2, and the device architecture
and fabrication processes in Sect. 9.3. Device performance, including the DOS
assessment, series resistance, and bias stability effects, are discussed in Sect. 9.4.
Finally, the issues of complementary p-channel TFTs, and driver circuit integra-
tion in demonstrator displays are briefly reviewed in Sect. 9.5.
9.2 Material Properties
One of the key features of the AOS materials is their large carrier mobility, even in
the amorphous state, where the mobility may be 50 % or more of the crystalline
mobility. This is very different from the a-Si:H/c-Si comparison, and the key to
understanding this difference lies in the difference between the covalent bonding of
Si and the ionic bonding of the AOS materials [5]. As discussed in Sect. 6.2, the
conduction and valence bands of Si are formed from hybridised sp3 states, and, in
the amorphous Si network, strained and broken Si–Si bonds lead to localised band
tail states and deep gap states, respectively. The band tail states reduce the field
effect mobility by [10 times, compared with the band mobility, which itself is
very low due to the carrier scattering distance in the extended states being of the
order of the inter-atomic spacing. In the ionic bonding configuration of AOS
materials, due to charge exchange between the metal cation and the oxide anion
(in, for instance, ZnO), the outer s-states of the metal ion are empty, and the outer
p-states of the oxygen ion are filled. The charge exchange results in a Madelung
potential, which separates the metal and oxygen ion orbitals, with the empty
s-states of the metal cation predominantly forming the conduction band minimum,
and the filled p-states of the oxygen anion mainly forming the valence band
9.2 Material Properties 303
Fig. 9.1 Schematic diagram

of (a) charge transfer and
(b) band gap formation
process in an ionic oxide
semiconductor. Molecular
orbital representation of the
conduction band bottom,
showing carrier conduction
pathways in (c) crystalline
and (d) amorphous oxide
semiconductor (a and b:
Reprinted by permission from
Macmillan Publishers Ltd:
NPG Asia Materials [5],
copyright (2010); c and d:
Reprinted from [12] with
permission of IEEE)
maximum, as shown in Fig. 9.1a, b [5, 12]. The separation between these band
edges was *3 eV in the original pulsed laser deposited a-IGZO material [1]
(giving these amorphous oxide materials their optical transparency), although
more recent, sputter deposited a-IGZO layers have given Tauc optical gaps up to
3.7 eV [13]. For metal oxides, the spatial spread of the spherically symmetric
metal s-state orbital is determined by the principal quantum number, n, of the
metal ion [14]. For post-transition metals with n C 5, such as In, Ga and Sn, it is
sufficiently large that there is an overlap between adjacent cations, as shown in
Fig. 9.1c [12]. This leads to small electron effective masses and high electron
mobilities, as found, for instance, in the transparent conducting oxide ITO. The
illustration in Fig. 9.1c shows the s-state overlap for a crystalline lattice, and this
overlap is still maintained in the disordered amorphous structure shown in
Fig. 9.1d. Hence, these spatially extensive s-state orbitals, in the post-transition
metal oxides, explain the relatively high carrier mobilities found in the transparent
amorphous oxide semiconductors [5, 12].
The AOS material can be doped, but, due to the flexibility of the ionic bond, this is
not achieved by introducing different valence atoms (as is standard with group IV
semiconductors), but most commonly occurs via the oxygen vacancy. The vacancy
results in a non-bonded metal cation producing a shallow donor level [5]. The
vacancy concentration is determined by the oxygen partial pressure during depo-
sition, and, for a low partial pressure, the interrelationship between the composition
of the ternary structure In2O3-Ga2O3-ZnO, the Hall mobility and the as-grown
doping level is shown in Fig. 9.2a [5]. Over most of this compositional space, the
material is amorphous, although it tends to be crystalline within *10 % of either
pure ZnO or pure In2O3. However, as will be seen along the base of the triangle, this
structural difference had little effect upon the Hall mobility, whereas increasing the
Fig. 9.2 a Compositional dependence of Hall effect mobility and free carrier density (in
parentheses, with units of 1018 cm-3) in the In2O3-Ga2O3-ZnO system, and b dependence of free
carrier density on the O2 partial pressure during deposition of a-IGZO and a-IZO films.
Ga content reduced both the mobility and the free carrier density. A low free carrier
density is required in TFT material to ensure a low off-current, and Ga plays an
important role in this respect, because its strong Ga-O bond reduces the O-vacancy
density [5]. In addition, the free carrier concentration can be further reduced by
increasing the oxygen partial pressure during deposition, as shown in Fig. 9.2b. This
also compares a-IGZO with IZO, and demonstrates the importance of Ga incorpo-
ration in the material [12]. For this reason, InGaZnO4 (IGZO) is the preferred
composition for TFTs, even though it is not the highest mobility material [12].
The oxygen vacancy has been referred to as forming a shallow donor level in
IGZO material; however, the vacancy is a negative U-centre, with the fully occu-
pied, neutral level in the lower half of the band-gap [5, 15]. This level occurs within a
relatively un-relaxed, ‘stoichiometric’ lattice, whereas the formation of the charged
centre is accompanied by an inward relaxation of the metal ions around the oxygen
vacancy location [15, 16]. Given the lattice relaxation necessary to form the ionised
donor level, it is more likely to be formed during film growth (under low oxygen
partial pressure) [15, 16], although an increased free carrier density has also been
found in material subjected to prolonged vacuum exposure [17]. This was attributed
to donor formation by effusion of oxygen from the material, and the process was
reversible by re-exposing the sample to an oxygen ambient. Both types of oxygen
vacancy may be found in a-IGZO films, and the instability caused by negative bias
illumination stress, NBIS, discussed in Sect. 9.4.3.2, is believed to be due to photo-
ionisation of the deep lying neutral oxygen vacancy [18]. More work is currently
required for a fuller understanding of NBIS, and of the conditions determining the
relative concentrations of neutral and ionised oxygen vacancies in as-prepared thin
films of a-IGZO.
Low temperature annealing in a H2-N2 gas mixture has also been found to
increase the free electron concentration in a-IGZO [12], and calculations of the
electronic properties of various point defects in c-IGZO have shown that
9.2 Material Properties 305
interstitial hydrogen bonds to an oxygen ion, forming a donor level above the
conduction band edge [19]. A similar interaction has been found from bonding
calculations in a-IGZO [12], with H always forming an O–H bond. It is suggested
that the donor activity results from the following reaction [12]:
H0 þ O2 ) O H þ e
This reaction is believed to underlie the susceptibility of unpassivated a-IGZO
TFTs to ambient moisture effects, in which water immersion [20] shifted the TFT
threshold voltage in a negative direction. As with the vacuum/oxygen exposure
results discussed above, the effect of water immersion was also reversible, in this
case by baking the sample in vacuum to break the O–H bonds and to effuse the
water molecules [20].
Finally, the Hall mobility was found to increase with increasing carrier
density, and was attributed to percolation limited transport around potential
barriers within the conduction band [12, 21]. A schematic illustration of the
a-IGZO density of states, DOS, is shown in Fig. 9.3, and includes the percolation
barriers, which are indicated by their average height, Ecenter, and distribution, DE.
Also shown in this figure is the location of the shallow donor levels, and a low
density band-tail at the conduction band edge. As the valence band edge is
composed of p-type states, these do not have the overlapping orbitals of the
conduction band states, and there is a large density of localised states, as well as
deep states above the valence band mobility edge [12], which inhibit useful
p-channel TFT behaviour. The a-IGZO DOS and electron conduction mecha-
nisms are discussed further in Sect. 9.4.2.
9.3 TFT Architecture and Fabrication
Unlike a-Si:H and poly-Si TFTs, AOS TFT technology is still developing, and
there is not an established process around which a consensus has emerged,
Fig. 9.3 Illustrative a-IGZO

density of states distribution,
DOS, showing band edges,
band-gap states and
distribution of percolation
potentials in the conduction
band. (Reprinted from [12]
leading to a well-defined manufacturing technology. However, there are clearly

emerging trends, and these are reviewed below, with a particular focus on the
TFT architecture and processes used to fabricate high quality demonstration
displays [2, 3, 6, 13].
9.3.1 Architecture
Although the original demonstration a-IGZO TFT was a top-gated, coplanar

structure [1], the presently preferred configuration for demonstration displays is
bottom gated, with either staggered [3, 13, 22] or coplanar [2, 23] source and drain
electrodes, as illustrated in Fig. 9.4a and b, respectively [3, 23]. The inverted
staggered structure is more widely used, but, predominantly, with the etch–stop
configuration shown in Fig. 9.4a [3]. The two principal inverted staggered TFT
architectures, of back-channel etched (BCE) and etch-stop (ES), were discussed in
Sect. 5.3, in the context of a-Si:H TFTs, where it was shown that the BCE
architecture was preferred for mass production. However, for the a-IGZO TFTs,
the preference is for the ES architecture because of the sensitivity of the back
channel to ambient effects, and also to damage and donor formation during the
deposition and etching of the source and drain metals [24]. This was demonstrated
in a direct comparison of BCE and ES TFTs, in which the ES TFTs had an SiOx
etch-stop layer deposited by PECVD, and both device types had sputter deposited
and dry etched MoW source and drain metals. With a 200 nm thick PECVD SiNx
gate dielectric, the sub-threshold slopes and electron mobilities of the BCE and ES
TFTs were 3.5 V/dec and 0.59 V/dec, and 5.0 cm2/Vs and 35.8 cm2/Vs, respec-
tively [24]. Hence, the damage to the unprotected back surface of the a-IGZO film
in the BCE TFT resulted in almost an order of magnitude degradation in device
performance. The choice of ES layer was also crucial in obtaining high perfor-
mance. For a-Si:H TFTs, a-SiNx:H is usually used in the ES structure, but, when
PECVD SiNx was used in the a-IGZO TFTs, the gate modulation of the channel
current was lost, and the devices became simple resistors due to a large free carrier
concentration of *491019 cm-3 in the material [24]. This was attributed to
hydrogen entering the a-IGZO film from the SiNx layer, and either acting as a
donor [25], or reducing the a-IGZO and introducing oxygen donor vacancies. To
avoid these effects, SiOx is the preferred ES/passivation layer in many publications
[2, 3, 13, 22–24], with some groups evaluating both PECVD and sputter deposi-
tion, given the latter’s lower H-content [26]. The SiOx has also been used as the
uppermost film in a double layer passivation process [13, 22], with an alternative
material placed in direct contact with the back face of the a-IGZO. For instance, a
DC sputter deposited layer of Al2O3 gave excellent bias stability results, as well as
good passivation [22]. In another case, the lower film of Ti, from a sputtered
double layer of Ti and Mo for the source/drain metallisation, was exposed to an
oxygen plasma to convert it to an insulating TiOx passivation layer, as depicted in
Fig. 9.5 [13]. After the definition of the metallisation pattern, a final capping layer
9.3 TFT Architecture and Fabrication 307
Fig. 9.4 Cross-sections of

a-IGZO TFT structures
a inverted staggered etch-stop
permission of SID), and
b inverted coplanar
Physics)
of PECVD SiOx was added to this structure. This approach had the additional
benefit of a reduced number of deposition stages, as the passivation layer was
directly formed from the source/drain metallisation.
As discussed in Sect. 5.3, the inverted staggered, rather than the inverted
coplanar structure, is exclusively used with a-Si:H TFTs in order to position the n+
layer on the top of the device stack, so that it could be selectively etched to form
the source and drain contact regions during final patterning. As doped contact
regions have not been routinely implemented with the a-IGZO TFTs, this con-
straint is removed, and the source and drain metal regions, in the coplanar
structure, can be deposited, and defined, before the deposition of the a-IGZO layer
itself, as illustrated in Fig. 9.4b [23]. This architecture also has the advantage that
the a-IGZO layer is not exposed to the deposition and etching processes required
for the source/drain metallisation, and requires one less mask than the ES process
[6], as discussed further in Sect. 9.3.2.4. However, the structure still incorporated a
SiOx passivation film on top of the completed device structure [6, 23], in order to
protect it against moisture in the ambient, and from damage during subsequent
display processing.
In addition to these conventional architectures, there is interest in self-aligned
structures to reduce the overlap capacitance in pixel TFTs, and also as a route to
future channel length reduction [27, 28]. A bottom gated self-aligned structure has
been demonstrated, in which the transparency of the a-IGZO layer has been
exploited to use the bottom gate as a UV exposure mask for the definition of the
etch-stop layer by back-face illumination [27]. This process reduced the total mask
Fig. 9.5 Cross section of

inverted staggered etch-stop
TFT, showing the local
conversion of Ti source/drain
metal to a TiOx passivation/
ES layer. (Reprinted from
Fig. 9.6 Cross-section of a

self-aligned top-gated TFT.
permission of SID)
count for the ES process by one, as well as reducing the gate/drain overlap to 2lm.
A more novel process has been used to produce the self-aligned top-gated TFT,
shown in Fig. 9.6 [28], in which the total overlap of the gate and source/drain
regions was reduced to 0.8 lm or less. This structure incorporated low resistance
contacts to the a-IGZO layer, which were self-aligned to the gate. In order to form
them, after the definition of the gate electrode and the removal of the gate
dielectric around it, the structure was coated in a thin film of Al, and annealed at
300 °C in oxygen. During this process, the lower surface of the Al reacted with the
a-IGZO to form a low resistance contact, whilst the upper surface of the Al was
oxidised to form an insulating layer of Al2O3. The self-aligned, low resistance
source and drain regions were contacted by the top metallisation pattern through
windows in the interlayer dielectric. TFTs with a minimum channel length of 4 lm
were fabricated by this process, and high quality transfer characteristics, as well as
good gate bias stability, were reported [28].
9.3.2 Fabrication Processes
9.3.2.1 a-IGZO Layer
The active a-IGZO layer is typically *30-60 nm thick in the FPD demonstrator
TFTs, and is usually deposited at room temperature by DC sputtering [2, 3, 6, 13,
26]. The sputter targets had a compositional ratio of In2O3:Ga2O3:ZnO in the range
1:1:0.5 to 1:1:1, and the sputtering gas was an Ar/O2 mixture. Whilst other
techniques, including pulsed laser deposition [1] and RF sputtering [24], have also
been used, the advantages of DC magnetron sputtering are that it is widely used for
ITO deposition in AMLCDs [29], and the equipment is readily available for large
area depositions.
9.3.2.2 Gate Dielectric
In the demonstration displays, the gate dielectric was 200 nm thick, and was most
commonly SiOx deposited by PECVD [3, 6, 22], although in some cases PECVD
SiNx [30] and SiOxNy [23] were also used. Where SiNx was used, there was no
apparent effect on threshold voltage, in spite of the high density of positive charge
which is customarily found in a-SiNx:H layers. A fuller discussion of the PECVD
deposition of a-SiNx:H and SiOx layers can be found in Sects. 5.5.3 and 7.3.1,
respectively.
In view of the bottom gate structure, and the absence of Fermi level dependent
meta-stability effects (such as seen in a-Si:H), the particular gate dielectric and its
deposition procedure have, in principle, little impact upon the subsequently
deposited a-IGZO layers, thereby affording a broad choice of gate dielectric. This
is apparent from the successful use of a-SiNx:H, SiOx and SiOxNy as cited above;
however, these high quality PECVD dielectrics require a deposition temperature
of *300 °C, or higher, and are more suited to glass than flexible, polymer sub-
strates. Indeed, the original paper reporting a-IGZO TFTs on low temperature,
flexible PET films used room temperature deposition of Y2O3 as the gate dielectric
[1]. Hence, there is continuing interest in alternative gate dielectrics, and in
reduced deposition temperatures compared with conventional PECVD depositions.
As part of this broad investigation, a range of low temperature, high-k dielectrics
has been examined, in which it is agued that the high dielectric constant can
compensate for a poorer interface compared with the higher quality PECVD films
[31]. Amongst the dielectrics examined have been Ta2O5 (k = 29) [32], ZrO2
(k = 25) [33] and HfLaO (k = 25) [34], all of which were deposited by room
temperature sputtering, with a film thickness in the range 200-300 nm (i.e.
comparable to the PECVD dielectrics). However, only HfLaO gave a notably low
value of sub-threshold slope, S, of 0.076 V/dec, whereas Ta2O5 and ZrO2
were *0.6 V/dec (compared with 0.29 V/dec for a PECVD SiOx dielectric [3]),
suggesting that these two high-k dielectrics were, indeed, producing a poor quality
interface. In addition, the TFT with a Ta205 dielectric had a high gate leakage
current, and a channel current on:off ratio of 105, compared with [ 108 in the SiOx
device [3]. The other two dielectrics had on:off ratios of 1-59107, and the par-
ticularly low value for Ta2O5 [31, 32] has been attributed to its small band gap, and
small conduction band offset with respect to the a-IGZO layer [31, 35]. A com-
promise approach has been a room temperature sputtered composite high-k and
SiO2 dielectric, such as Ta2O5–SiO2, and HfO2-SiO2, or even a tri-layer stack of
SiO2/HfO2-SiO2/SiO2, leading, in the first case, to an improved on:off ratio of
39106, and 29107 for the triple stack [31].
From the above discussion, it is clear that there is on-going research to identify
the optimum gate dielectric for AOS TFT fabrication on low temperature, flexible
substrates. In addition to the impact of the gate dielectric layer on the static TFT
characteristics, it also needs to be evaluated in terms of the bias stability of the
device, and this topic is reviewed in Sect. 9.4.3.
9.3.2.3 Post Deposition Annealing
Post-deposition annealing of the a-IGZO TFT structure is commonly used, and it

has been shown to lead to a major improvement in device performance and uni-
formity [10]. In that study, the effects of 400 °C annealing, in dry and wet oxygen,
were investigated using inverted staggered TFTs fabricated on thermally oxidised
n+ silicon substrates, and the a-IGZO layers were produced by pulsed laser
deposition at two different oxygen partial pressures. This difference in deposition
conditions led to threshold voltage differences of 6 V, as shown by the transfer
characteristics in Fig. 9.7a. However, the dry annealing removed this difference,
and gave a common threshold voltage of *-1 V, and a reduction in S values
from 0.4–0.5 V/dec to a common value of *0.25 V/dec [10]. Wet oxygen
annealing had a similar effect, in terms of removing the built-in differences due to
different oxygen partial pressures during deposition, and the overall effect was a
function of the water partial pressure, as shown in Fig. 9.7b. At the optimum water
partial pressure of *20 %, the wet annealing was more effective than the dry
annealing, and normalised the threshold voltage to *1 V and S to *0.12 V/dec
[10]. In addition, both dry and wet annealing improved the overall uniformity of a
given set of TFTs, and, once again, the wet annealing was the most effective in this
respect. Hence, the post-deposition annealing of a-IGZO, especially in a wet
Fig. 9.7 a The influence of deposition partial pressure of O2, and post-deposition annealing on
TFT transfer characteristics, and b effect of water partial pressure during post-deposition
annealing on TFT parameters {open (closed) circles 6.2 Pa (6.9 Pa) O2 partial pressure during
deposition}. (Reprinted with permission from [10]. Copyright (2008) American Institute of
Physics)
oxygen ambient, improved device performance, and uniformity, and reduced the
dependence of the device characteristics on the initial layer deposition conditions.
To stabilise the a-IGZO film, post-deposition annealing is routinely applied to
the TFTs used for demonstrator displays [3, 6, 13, 22, 23], and is typically 1–2 h in
the temperature range 250–350 °C, although the ambient is rarely specified.
9.3.2.4 Metallisation
The choice of gate and source/drain metals is determined by considerations of line

resistance, processability (to include deposition by sputtering), and experience
within the LCD industry. Accordingly, in some instances, Mo [3, 22] or Ti/Mo
[13] has been used for the metallisation in inverted staggered devices, and, as
discussed in Sect. 9.3.1, the Mo was etched away over the channel region, in the
latter case, to selectively expose Ti to an oxygen plasma, which formed a dual-
purpose TiOx passivation and etch-stop layer [13].
A combined Ti/Al/Ti layer was used in some inverted coplanar TFT structures
[23], and, as a consequence of the a-IGZO post-deposition anneal, the Ti reduced
the a-IGZO adjacent to it to produce a layer of electron-rich, high conductivity
material, thereby reducing the contact resistance between the metallisation and the
channel [23].
9.3.2.5 Process Flow
The 4-mask process flow for the inverted coplanar TFT is shown in Fig. 9.8 [6], in
which the mask definition stages are the definition of the gate metal (M1), the
Fig. 9.8 Inverted coplanar

TFT process flow. (Reprinted
source/drain metal (M2), the a-IGZO layer (M3) and the contact holes (M4)
through the top passivation layer (not shown). For the inverted staggered etch-stop
TFT, shown in Fig. 9.4a, the comparable stages are deposition and definition (M1)
of the gate metal, deposition of the gate dielectric, deposition of the a-IGZO and
etch-stop layers, definition of the etch-stop pad (M2), definition of the a-IGZO
layer (M3), deposition and definition of the source/drain metallisation (M4), and
deposition of the final capping/passivation layer and contact window opening
(M5). Hence, the inverted staggered ES architecture requires one more mask than
the inverted coplanar process.
9.4 a-IGZO TFT Performance
9.4.1 n-Channel Characteristics
The transfer characteristics of a high quality, inverted staggered etch-stop a-IGZO

TFT, with W = 25 lm and L = 10 lm, are shown in Fig. 9.9a. This device, with a
200 nm thick SiOx PECVD gate dielectric, had a field effect mobility of 21 cm2/Vs,
a sub-threshold slope of 0.29 V/dec and an on:off ratio of [108 [3]. Typical output
characteristics [12], of an a-IGZO TFT with a saturation mobility of 11.8 cm2/Vs,
are shown in Fig. 9.9b, and the device displayed good current saturation at large
VD, and an absence of current crowding at low VD.
However, even though current crowding is not an issue in high quality a-IGZO
TFTs, the absence of a doped contact to the material has led to the investigation of
series resistance effects in these devices [36–38], and its impact upon the field
effect mobility as channel length is reduced [24]. If an ohmic resistance, Rs and Rd,
Fig. 9.9 High quality a-IGZO TFT characteristics (a) transfer characteristic of ES TFT
(Reprinted from [3] with permission of SID, and (b) output characteristics (Reprinted from [12]
9.4 a-IGZO TFT Performance 313
Fig. 9.10 a Total TFT resistance measured as a function of L and VG, and b variation of channel
resistance, Rch, and series resistance, RSD, with VG. (Reprinted from [36] with permission of
IEEE)
is present at the source and drain terminals, respectively, then the total measured
device resistance, Rm, is
Rm ¼ Rch þ Rs þ Rd ¼ Rch þ RSD ð9:1Þ
where RSD = Rs ? Rd, and the channel resistance, Rch, in the linear regime, can
be obtained from Eq. 3.11 as:
L
Rch ¼ ð9:2Þ
ln WCi ðVG VT Þ
Hence, for gate-bias-independent RSD, a plot of Rm vs L, for any value of
VG-VT, will give RSD where the line intersects the Rm axis, and its slope will give
the channel resistance per unit length at that gate voltage. An example of this
measurement is shown for BCE inverted staggered a-IGZO TFTs in Fig. 9.10a,
and the two extracted resistance parameters are shown in Fig. 9.10b [36]. There
was not a common intersection point of the data sets, and the measurement yielded
an approximate evaluation of the series resistance, RSD, which was gate-bias-
dependent. The background, and limitations, to this measurement procedure are
briefly discussed below.
This resistance technique was originally developed for MOSFETs, and, with
different values of VG-VT, it was found that the individual lines may have a
common intersection point at a positive value of L, DL, rather than at L = 0 [39].
This was interpreted as channel shortening during processing, such that the
effective channel length, Leff, differed from the mask value by DL, and the value of
RSD was taken at this common intersection point (and not on the Rm axis). The
technique has also been used with a-Si:H inverted staggered ES TFTs [40, 41],
where a common intersection point was found at negative values of L, indicating
an increase in the effective channel length due to current collection spreading
along the n+ doped regions. Using the above procedure, the identification of the
common intersection point, and the related evaluation of the effective channel
length, relies upon the series resistance being independent of gate bias, whereas,
Fig. 9.11 Gate bias dependence of (a) change in effective channel length, LT, and (b) specific
resistivity of source and drain contacts (Reprinted from [36] with permission of IEEE)
for the a-IGZO TFTs in Fig. 9.10 this was not the case. A modified version of the
common intersection measurement procedure was developed to handle this situ-
ation for MOSFETs, and, for each value of gate bias, VG, a common intersection
point was measured at closely paired values of VG at VG ± DVG, where DVG may
be *0.25 V. The technique was referred to as the ‘paired VG method’ [42].
However, this procedure was not used with the above a-IGZO data set; instead, the
intersection of each line with the x-axis in Fig. 9.10a was used as an expedient
measure of the change in channel length, DL. This is shown as ‘LT’ in Fig. 9.11a
[36], demonstrating that the effective channel length was weakly gate bias
dependent, and *0.4 lm longer than the mask length. This was attributed to
current spreading beneath the metal source and drain contacts.
Figure 9.10b showed that the total series resistance and the normalised channel
resistance were comparable, and, therefore, for a device, with a channel length of
5 lm, the series resistance would be *20 % of the channel resistance, and have an
appreciable effect upon the on-current (although its impact would diminish as the
channel length increased).
The gate bias dependent values of RSD were attributed to the influence of the gate
bias on the semiconductor/Mo contact on the top of the a-IGZO film. The values of
RSD in Fig. 9.10b were specific to the technology and geometry of the particular
TFTs used, and these values may be corrected to remove the geometric effects
either by normalising for the channel width (i.e. RSDW Xcm), where this makes no
assumption about the current distribution in the contact, or by normalising by
contact area, Ac, to give the specific contact resistivity, Rc (i.e. RSDAc Xcm2). This
requires knowledge of the contact area, and the specific contact resistivity is usually
measured in simple vertical contact structures of known area [43], rather than in
TFTs. In the measurements here, it was assumed that the channel length extension
of *0.4 lm was an effective measure of the contact length, from which the specific
contact resistance at each end of the channel was given by [36]:
WRSD DL
Rc ¼ ð9:3Þ
2 2
Fig. 9.12 Variation of field

effect mobility with gate bias
and channel length.
Physics)
The two factors of 2 in the denominator account for the measurements of RSD
and DL being the sum their effects at both ends of the channel. The dependence of
Rc on the gate bias is shown in Fig. 9.11b, and, as with RSD, it reduced with
increasing gate bias.
In this study [36], the electron mobility was calculated from the slope of the
inverse channel resistance/unit length against gate bias, and the intercept with the
x-axis gave VT. Thus, this procedure did not explicitly use the channel length to
calculate mobility, but, in contrast, the more widely used field effect mobility
calculation does, and, to give an accurate value of mobility, that would require
both the use of the effective channel length, and the series resistance. However, by
using long channel length TFTs to measure the field effect mobility, both of these
corrections are minimised.
It can be assumed that the series resistance effects noted in this paper [36] are
common, although the specific values will vary with the detailed processing con-
ditions. A common example of a series resistance artefact is the reduction in field
effect mobility with increasing gate bias and reducing channel length, as shown in
Fig. 9.12, for an ES inverted staggered a-IGZO TFT [24]. From plots of total
resistance as a function of gate length, the series resistance was again shown to be
gate bias dependent. The gate width normalised resistance, WRSD, at VG = 20 V
was *250 Xcm, whereas the comparable figure for the data shown in Fig. 9.10a
was *12.5 Xcm, illustrating the variability between differently prepared samples.
Indeed, some very short channel length TFTs, with L = 50 nm, have been reported
with a field effect mobility of 8.2 cm2/Vs, and good output characteristics, up to a
drain bias of 10 V, as seen in Fig. 9.13 [44]. These were BCE inverted staggered
TFTs, with a PECVD SiNx gate dielectric, which was *40 nm thick, as was the
a-IGZO layer. The device displayed very few short channel effects, including
minimal drain induced barrier lowering, as is apparent from the tightly clustered
transfer characteristics in the sub-threshold regime, and from the high impedance of
the output characteristics. No doubt the thin SiNx layer contributed to the suppres-
sion of short channel effects, but the device still had remarkably good characteristics
given its channel length. (Short channel effects are discussed in greater detail in
Sect. 8.7).
The detailed channel resistance measurements discussed above were for an
inverted staggered BCE TFT [36], and a comparable study has been made of ES
Fig. 9.13 I–V characteristics of short channel TFTs with dimensions W/L = 200/50 nm
(a) transfer, and (b) output characteristics. (Reprinted from [44] with permission of IEEE)
TFTs [37], in which the defined channel length, determined by the width of the ES
layer, had less control over the channel current than the separation of the source
and drain contacts on top of the ES layer. The device structure is shown in
Fig. 9.14a, and the conventional channel length, defined by the width of the etch
stop layer, ESL, is given by Lcnt, and the separation of the source and drain
contacts on ESL is given by Lmask. In this structure, the metals were MoW, the
200 nm SiNx gate dielectric and the SiOx etch-stop layer were both deposited by
PECVD, and the 50 nm thick a-IGZO layer was deposited by RF sputtering [24,
37]. Figure 9.14b shows that the channel current was independent of the width of
ESL, Lcnt, over the range 10–16 lm, whereas Fig. 9.14c shows that the current
varied with the magnitude of Lmask over the range 4–10 lm [37]. In broad terms,
these results suggest a channel current dominated by contact resistance effects, in
which the contact resistance was determined by the magnitude of the overlap of
the source and drain pads on ESL. Using the source/drain pad separation, Lmask, as
the reference channel length, the dependence of Rm and DL on VG were evaluated
using the ‘paired VG method’ [42], and the results are shown in Fig. 9.15. This
demonstrated that, as VG increased, the effective channel length, Leff, increased
beneath the overlap region towards the metal contacts at the edge of ESL (i.e. Leff
tended towards Lcnt) [37]. At the same time, the total series resistance, Rext,
decreased with VG, where, from Fig. 9.14a:
Rext ¼ RS þ RD þ RovS þ RovD ¼ RSD þ Rovlp ð9:4Þ
Further analysis of this data, including the separation of RSD and Rovlp, showed that
both resistances reduced with increasing VG, but Rovlp was *10 times less than RSD,
which itself was *10 times less than the channel resistance. The width normalised
series resistance was *30 Xcm at VG-VT = 10 V [37], which is of the same order as
Fig. 9.14 a TFT structure, and definition of layout dimensions Lcnt and Lmask, b variation of
output characteristics with channel length, Lcnt, for Lmask = 4lm, and c variation of output
characteristics with Lmask, with constant channel length, Lcnt = 16lm. (Reprinted from [37] with
permission of IEEE)
the value of 12.5 Xcm measured at 20 V in the BCE device, shown in Fig. 9.10b [36].
A full explanation of these anomalous ES TFT results [37], which are not consistent
with conventional MOSFET theory, has not been given, beyond noting that it
appeared to be related to the accumulation of electrons beneath the overlap region as
the device was turned-on more strongly. However, if this effect is common in a-IGZO
TFTs with the ES architecture, then the conventional use of Lcnt to calculate the field
effect mobility will overestimate its value (because Lcnt overestimates the effective
channel length).
Fig. 9.15 Gate bias

dependence of the change in
effective channel length, DL,
and the contact resistance,
Rext, for different values of
the source/drain metallisation
pad overlap on the etch stop
layer. (Reprinted from [37]
9.4.2 Conduction Process and Density of States

Distribution, DOS
9.4.2.1 Conduction Process
Some of the characteristic features of the conduction process in a-IGZO films have
been obtained from Hall effect measurements of doped films [12, 45, 46], such as
those shown in Fig. 9.16. Among the key features was the increase in the Hall
mobility with the doping density, in Fig. 9.16a. This is opposite to the behaviour
seen in c-Si, where the mobility reduces due to increased charged impurity and
electron–electron scattering. In Fig. 9.16b, the carrier density was temperature
independent for densities above 1017cm-3, indicating that the material was
degenerate, with the Fermi level above the conduction band edge [12]. This is also
quite different from the behaviour of a-Si:H, where the band tail density was
sufficiently great that the Fermi level never entered the conduction band. At the
other end of the dopant range in Fig. 9.16b, the lightly doped material displayed a
low temperature activation energy of 0.11 eV, as the material ‘froze-out’ with the
electrons trapped on the donor level. Figure 9.16c showed that, in contrast to the
temperature independent carrier density at moderate doping levels, the mobility
continued to be thermally activated, until much higher carrier concentrations in
excess of 291018cm-3, when it also became temperature independent [12].
Finally, Fig. 9.16d shows the electrical conductivity, r, which is the product of the
carrier density and mobility, and this was thermally activated at doping levels
below 391019 cm-3, but, only at the lowest carrier concentration was there a
unique activation energy. At the intermediate doping levels, where there was not a
simple Arrhenius plot, ln(r) was shown to vary as T-1/4 [46]. These results have
been interpreted both in terms of a low DOS near the conduction band edge, which
permitted the material to become degenerate, and a distribution of potential bar-
riers within the conduction band, which are schematically illustrated in Fig. 9.17a
[45, 46]. As indicated, electron transport around these barriers was by percolation,
where there was assumed to be a Gaussian distribution of barrier heights, g(E),
given by [46]:
gðEÞ ¼ exp½ðE /0 Þ2 =ð2r2/ Þ ð9:5Þ
and /0 is the average barrier height, and r/ is the energy distribution width, as
shown in the simplified diagram in Fig. 9.17b. In this work, the carrier mobility, l,
was related to the band mobility, l0, by the following percolation model expres-
sion [47]:
" 2
#
q/0 ðqr/ Þ
l ¼ l0 exp þ ð9:6Þ
kT 2ðkTÞ2
Fig. 9.16 Hall effect data from doped a-IGZO samples (a) variation of mobility with electron
concentration. Temperature dependence of (b) electron density, (c) electron Hall mobility, and
(d) conductivity. (a and c: Reprinted from [45] Copyright (2006), with permission from Elsevier;
b and d: Reprinted from [12] with permission of IEEE)
The percolation process gave a direct physical explanation for the thermally
activated mobility, and its eventual temperature independence at high enough
carrier concentrations, when the Fermi level was above the maximum barrier
height. The model also gave a good fit to the detailed experimental carrier con-
centration and conductivity data, as shown by the solid lines in Fig. 9.16b, d
respectively [12, 46]. It also followed the T-1/4 temperature dependence of the
conductivity data (not shown) [46]. In establishing the data fits in Fig. 9.16b, d the
potential barrier height distribution was found to be dependent upon the back-
ground doping level, as shown by Fig. 9.17c [46], in which the average barrier
height, /0, was 40–120 meV above the conduction band mobility edge, Em, and
the distribution width, r/, was 20–30 meV.
9.4.2.2 Density of States Distribution, DOS
Given the dependence of the Hall mobility on the electron density, some caution is
required in using device simulation to establish the DOS in TFTs, particularly
when it is done by fitting a calculated transfer characteristic to a measured char-
acteristic [48]. This is because, if the carrier-dependent mobility is ignored, then a
field-effect mobility extracted at large VG will overestimate its value at lower VG
values, with a corresponding overestimate of the DOS [49] (as discussed further
below). Equally, the gate-voltage-dependent field-effect mobility itself cannot be
Fig. 9.17 a Schematic illustration of potential barriers in the conduction band, and percolation
transport of electrons, b idealised potential barrier model used in calculations, and c variation of
fitted barrier parameters with doping density. (a: Reprinted from [45] Copyright (2006), with
permission from Elsevier; b and c: Reprinted with permission from [46]. Copyright (2010)
American Institute of Physics)
used in the simulation, as this parameter is an artefact of carrier trapping in the

DOS, as well as being affected by percolation effects. Indeed, the separation of
carrier trapping and percolation phenomena on the field effect mobility should be
done by explicitly analysing it for these two effects [21], as discussed below.
In view of these concerns, some authors have used C–V measurements on TFTs
[48–50] (rather than Id-VG measurements) to determine the trap occupancy as a
function of gate bias, because the electron trapping effects will be independent of
the electron mobility [48–50]. The DOS thus established was then used to simulate
the Id-VG characteristics, with the mobility as an adjustable parameter [50]. This
yielded an average mobility at each value of gate bias, whereas a more detailed
analysis extracted the mobility as a function of free carrier density [49], which
could be directly compared with the carrier-dependent Hall-effect mobility.
The C–V measurements [51, 52] made use of the capacitor formed by the gate
and the source/drain contact regions in BCE TFTs [48, 49], and the resulting C-VG
and Id-VG curves are shown in Fig. 9.18. The C-VG curve was measured at a low
enough frequency that the electrons could follow the frequency of the measuring
signal. Using the band bending notation from Chap. 2, the surface potential, Vs,
was extracted from the measured C-VG curve by using [48, 51]:
Qs ¼ Ci ðVG Vs Þ ð9:7Þ
dVs =dVG ¼ 1 ðCi Þ1 dQs =dVG ¼ 1 Cm =Ci ð9:8Þ

where Ci is the dielectric capacitance (measured at large positive gate bias), Cm is
the measured capacitance, and Vs is calculated as a function of VG by integrating
Eq. 9.8:
Z VG
Vs ¼ ð1 Cm =Ci ÞdVG ð9:9Þ
VFB
From Gauss Law, the surface field, Fs, is related to the total surface charge, Qs,
by:
Qs ¼ e0 es Fs ð9:10Þ
Hence, the surface field can be calculated using Eqs. 9.7 and 9.10, together with
the numerically calculated values of Vs from Eq. 9.9. The surface field is also
related to the space charge density, q, by Poisson’s equation:
d2 V q
¼ ð9:11Þ
dx2 e0 es
and, for space charge densities dominated by trapped carriers:
Z V
qðxÞ ¼ q NdV ð9:12Þ
0
Fig. 9.18 C–V and I-V

measurements on a BCE
a-IGZO TFT. (Reprinted with
The trap density, N, was used iteratively, as an adjustable parameter, in

Eqs. 9.11 and 9.12 to ensure that the values of Vs and Fs from those equations
matched those established in Eqs. 9.9 and 9.10. The DOS measured by this pro-
cedure is shown by the solid lines in Fig. 9.19a [49], and illustrates the reduction
in DOS produced by the post-deposition annealing in wet and dry oxygen, as
discussed in Sect. 9.3.2.3. The decrease in the DOS was attributed to the reduction
in oxygen deficiencies and structural relaxation as a result of the oxygen annealing
procedure [49].
Having established the DOS and the relationship between Vs and VG, the
electron concentration could be calculated through the film at any given value of
VG. This carrier profile was then used in a computational procedure which strat-
ified the a-IGZO film parallel to the dielectric interface, and progressively built up
a mobility profile such that the integral of the vertical mobility and carrier con-
centration profiles accurately simulated the transfer characteristic [49], with the
resulting dependence of mobility on electron concentration shown in Fig. 9.19b.
This figure showed that the post-deposition annealing treatments primarily affected
the DOS rather than the mobility, and also demonstrated that, for a given carrier
concentration in the TFT, the electron mobility was lower than the Hall mobility in
the doped samples. As indicated, the percolation model was fitted to the field-effect
mobility data, and the mobility difference from the Hall samples was attributed to
the doping reducing the mean percolation barrier height in the latter, as is apparent
in Fig. 9.17c. Hence, this procedure separated the components of electron con-
duction in the a-IGZO TFTs due to trap limited conduction, determined by the
DOS, and percolation conduction in the extended states.
The DOS values represented by the broken lines in Fig. 9.19a were obtained by
simply fitting simulated transfer characteristics to measured ones [53], in which
the electron mobility, in the simulations, was fixed at the conventionally measured
field effect mobility at large gate bias. The overestimate in the DOS, over a
substantial fraction of the band-gap, was attributed to the field effect mobility
overestimating the correct, carrier-dependent mobility [49].
An alternative C-VG technique was based upon the use of optical illumination
during the measurement to increase the frequency response of the trapped carriers
Fig. 9.19 a a-IGZO DOS measured by C-V technique (solid lines), and DOS measured by
conventional TFT simulation (broken lines), and b extracted electron mobility as a function of the
free electron concentration in TFTs (solid lines), average mobility (broken line), percolation
model (squares), and Hall mobility in doped films (open circles). (Reprinted with permission
[50], and resulted in a DOS, which was qualitatively similar to that shown in
Fig. 9.19a, but with a lower band edge density. The distribution, N(E), was rep-
resented by a double exponential of tail and deep states given by:
NðEÞ ¼ 1:2 1018 exp ðEc EÞ=0:125 þ 9:5 1016 exp ðEc EÞ=1:4
ð9:13Þ
in which the first term represented the tail state density [50].
As mentioned above, the field effect mobility itself has also been analysed in
order to separate the DOS responsible for carrier trapping from the potential
barrier distribution in the conduction band, controlling the percolation effects [21].
This was done by fitting the data to a model containing percolation-modulated
trapping effects when the Fermi level was below the conduction band edge, and
percolation transport alone when it was within the band. To account for the effect
of the DOS on the mobility, the modified MOSFET equation was used [21, 54].
This was introduced in Chap. 6 (Eq. 6.27) in the discussion of on-state currents in
a-Si:H TFTs, and the drain current expression is:
l0 WCia1 n
Id ¼ ½ðVG VT Þa ðVG VT Vd Þa ð9:14Þ
aL
where l0 is the band mobility, and a and n are related to the tail state energy width,
Et, and the carrier density trapped in the tail states, Nt, by [54]:
a ¼ 2Et =kT ð9:15Þ
ðqes e0 akTNt Þ1a=2

n¼ ð9:16Þ
a1
A Taylor expansion of Eq. 9.14 in the linear regime gives [54]:
l0 WCia1 n h i
Id ¼ ðVG VT Þa1 Vd ð9:17Þ
L
And from Eq. 9.17, the field effect mobility is given by:
lFE ¼ l0 Cia2 nða 1ÞðVG VT Þa2 ¼ l0 AðVG VT Þ2ðEt=kT 1Þ ð9:18Þ

However, the band mobility, l0, must be modified to include the influence of
percolation, so that, using Eq. 9.6, Eq. 9.18 becomes:
" #
q/0 ðqru Þ2
lFE ¼ l0 exp þ AðVG VT Þ2ðEt=kT 1Þ ð9:19Þ
kT 2ðkTÞ2
In summary, Eq. 9.19 describes the situation in which the Fermi level is below
the conduction band edge, and transport is limited by both carrier trapping and
percolation effects.
At a gate bias in excess of VP, which is the threshold voltage for percolation, and
occurs when the Fermi level is positioned at the conduction band edge, the mobility
is controlled only by percolation, and the field effect mobility is given by [21]:
" #
q/0 ðqr/ Þ2 qDVs
lFE ¼ l0 exp þ 2
exp ð9:20Þ
kT 2ðkTÞ kT
Equation 9.20 is valid for Fermi level positions between the band edge and the
maximum barrier height, /M, and DVs corresponds to the change in Fermi level
position within the band due to a change in band bending DVs. By developing an
approximate analytical expression for the free carrier density as a function of Vs, a
relationship was established between DVs and DVG, from which the following
expression was derived [21]:
" #
q/0 ðqr/ Þ2
lFE ¼ l0 exp þ 2
BðVG VP Þ4S ð9:21Þ
kT 2ðkTÞ
where S is the ratio, (DB-WB)/DB, of the percolation barrier width, WB, and the
inter-barrier spacing, DB. Hence, both equations 9.19 and 9.21 show that the field
effect mobility has a power law dependence on VG of the form:
lFE ¼ KðVG VT;P Þc ð9:22Þ

The values of K and c were extracted from the solid line fits of the model to the
experimental field effect mobility data shown in Fig. 9.20a (for a bi-layer a-IZO/
a-IGZO TFT). These fits yielded a band mobility of 70 cm2/Vs, values of /0, and
r/ of 50 meV and 45 meV, respectively, and a band tail DOS given by:
Fig. 9.20 a Measured values (data points) of Ids, and lFE in an a-IZO/a-IGZO TFT, and the fitted
expressions (solid lines) from equations 9.19 and 9.21, and b calculated variation of free and
trapped electron densities with Fermi level position. (Reprinted with permission from [21].
NðEÞ ¼ 4 1019 exp ðEc EÞ=0:035 ð9:23Þ

Using the extracted DOS parameters, Fig. 9.20b shows the calculated free and
trapped carrier densities as a function of Fermi level position, and identifies the
threshold voltage, VP, for percolation limited conduction at 13 V [21]. This figure
also illustrates the greater concentration of trapped charge, compared with free
charge, when the Fermi level was more than *0.05 eV below the conduction
band edge, and confirms the trap limited conduction within this gate voltage
range [21].
It is apparent that accurate DOS and percolation parameter assessment in AOS
TFTs present a number of challenges, and different techniques are being explored.
Also, whilst the currently published DOS values show some variability with
device processing and layer composition, it is equally clear that the field effect
mobility and tail state DOS values are consistently superior to those found in
a-Si:H TFTs.
9.4.3 Bias Stress Instability
Due to its different bonding configuration, the meta-stable weak bond breaking
observed in a-Si:H is predicted not to occur in AOS devices [55], but, even without
this mechanism, both positive [53, 56–58] and negative [57, 58] gate bias instabilities
have been widely reported in a-IGZO TFTs. The instability is predominantly a
positive threshold voltage shift with positive gate bias stress, and is accompanied by
minimal changes in mobility or sub-threshold slope. The shift is indicative of elec-
tron trapping at/near the AOS/dielectric interface, and is driven by the electron
accumulation within the film. For negative bias stress, the threshold voltage shifts
negatively, and is consistent with hole trapping. In addition, optical illumination
during bias stress results in a further instability process, which greatly increases the
magnitude of the negative bias instability [57, 59], and reduces the positive bias stress
shift [60], or even produces a net negative shift [61]. The optically enhanced insta-
bility has been attributed to photo-ionisation of the oxygen vacancy [18, 59, 61–63],
and has also been correlated with persistent photoconductivity in the a-IGZO layer
after the removal of the illumination [61–63]. Similar gate bias instability effects
have also been reported in a wide range of other AOS TFT materials (including
HfInZnO, AlSnInZnO, ZnInSnO, ZnInO, ZnSnO and ZnO) [62, 64, 65], and the
a-IGZO results reviewed below can be taken to be broadly representative of the
instabilities observed in other AOS materials of interest.
9.4.3.1 Gate Bias Instability
Typical examples, and artefacts, of positive gate bias instabilities are shown in
Fig. 9.21a–d [53, 56]. These results were obtained after constant current stress (with
positive gate and drain bias), from differently prepared inverted staggered TFTs.
Figure 9.21a and b compare the shift in the transfer characteristics of an unannealed
device with one given a post-deposition wet anneal at 400 °C [53]. As discussed in
Sect. 9.3.2.3, anneals of this type are essential to improve the basic device charac-
teristics, and they similarly reduce the magnitude of the gate bias instability,
resulting in a small parallel shift of the transfer characteristics. From simulations
[53], and a-IGZO film thickness dependent stability results [57], it was concluded
that the defects were electron traps at the a-IGZO/dielectric interface in the annealed
TFTs. A further aspect of device processing was identified by the results in
Fig. 9.21c and d, in which an organic photoacryl and an inorganic, PECVD SiOx
passivation layer were compared in etch-stop TFTs [56]. The SiOx layer was found
to result in a smaller and more parallel shift of the transfer characteristics, and,
indeed, the magnitude of the instability was found to be critically dependent upon
the type of passivation layer used. The best results were obtained with a dense SiOx
layer, which suppressed the gate bias instability, and this was attributed to it pro-
viding an impervious barrier to the ingress of ambient oxygen and moisture into the
top surface of the a-IGZO film, where these impurities led to trap generation within
the film. Further work has directly correlated the instability of unpassivated, bottom
gate devices with the relative humidity of the ambient during bias stressing [66], and
also to water vapour exposure prior to the stress measurement [58]. A high quality
passivation layer is, therefore, needed to minimise this contribution to positive gate
bias instability, where the residual instability is associated with trapping at the
dielectric/semiconductor interface.
A high quality dielectric layer is also necessary to reduce trapping within the
layer itself [58, 67], and, for top-gate TFTs, the dielectric deposition process itself
needs careful control to avoid damage to the underlying AOS layer [64]. For
bottom gate TFTs, superior stability was obtained with a PECVD SiOx gate
dielectric compared with PECVD SiNx [67], and the difference was attributed to a
higher hydrogen content in the latter giving a larger trap state density.
Fig. 9.21 Positive gate bias

instabilities induced by
constant current stress in
a-IGZO TFTs (a) unannealed,
and (b) with post-deposition
wet anneal of a-IGZO at
400 °C (50 h stress at 5lA)
Physics). Time dependent
stress at 10lA with
(c) organic photo-acryl, and
(d) inorganic PECVD SiOx
passivation layers. (Reprinted
The time dependence of the positive gate bias instability was thermally acti-
vated, and followed a stretched exponential [57, 58, 60, 65] of the form:
t b
DVT ¼ ðVGSt VT0 Þ 1 exp ð9:24Þ
s
where VGSt is the stress bias, VT0 is the pre-stress threshold voltage, b = T/T0, and
the thermally activated time constant, s, is given by:

EA
s ¼ m exp ð9:25Þ
kT
This process describes time dependent trapping into an exponential distribution of
trapping states, with a characteristic width T0. It is apparent that the kinetics of the
instability process can be empirically modelled using the same formalism as used
to describe gate bias instabilities in a-Si:H TFTs (see Sect. 6.4.1), although the
physical mechanisms underlying the instabilities are quite different. Depending
upon the details of sample fabrication, and the measurement conditions, activation
energies, EA, have been reported over the range 0.4 eV [60] to 0.7 eV [58].
The bias stress induced threshold voltage shifts were reversible after the
removal of the gate bias, and the recovery characteristics also displayed a stretched
exponential time dependence, but with a different activation energy compared with
the initial instability results [58, 60].
As is apparent from the above discussion, careful device processing is required
to minimise the positive gate bias instability, where important details include the
choice of gate dielectric, defect removal in the AOS film by post-deposition
annealing, and the implementation of a suitable passivation process over the
device structure. With such processing, the negative gate bias instability is usually
smaller than any residual positive gate bias instability. However, if negative gate
bias stressing occurs under conditions of optical illumination, large negative
threshold voltage shifts can occur [57, 59], even for devices showing good neg-
ative bias stability in the dark, as discussed in the following section.
9.4.3.2 Negative Bias Illumination Stress (NBIS)
Due to the transparency of the AOS materials, and the frequent use of transparent
gate metals, there may be no natural light screening of the TFT when it is used for
active matrix addressing of LCDs and OLEDs. As this will be a photon rich
environment, the optical response of the TFTs, and the impact of optical illumi-
nation on their bias stability, have been widely studied [18, 57, 59, 61–63]. The
NBIS instability is a negative shift in threshold voltage, and this occurs for both
band-gap illumination (E [ 3.0 eV), as well as for sub-band-gap illumination, as
shown by the representative results in Fig. 9.22 [57]. These results were obtained
from wet-annealed inverted staggered a-IGZO TFTs, and similar results have been
obtained from top-gated TFTs [59], as well as from bottom-gated a-HfInZnO [62],
and bi-layer a-InZnO/a-IGZO [61] TFTs passivated with SiOx. The results in
Fig. 9.22a showed that the devices were stable with negative gate bias stress in the
Fig. 9.22 Time dependent negative bias stress effects (VG = -20 V) on TFT transfer charac-
teristics (a) dark, (b) illumination at E = 3.4 eV, and (c) illumination at E = 2.7 eV. (Reprinted
from [57] with permission of SID)
–5
10 4
accumulated photo-
Ln(τ)
T decreasing 3 generated electrons
–6
from 350 to 293 K
10 2 EA = 0.33eV V0++
0.032 0.036 0.04
VGS<0V
I DS (A)
1/kT (1/meV)
–7
10
trapped
V0+
hole
VDS = 0.5 V
–8 photo-
10 VGS = Open
generated
Light Dark hole in EV V0
0 1000 2000 3000 4000 5000 6000 V0+++2e V 0
reduced
time (s)
Fig. 9.23 a Time and temperature dependent dark relaxation of channel current after
illumination (Reprinted with permission from [62]. Copyright (2010) American Institute of
Physics), and b TFT band bending diagram during NBIS, showing shallow donor formation at the
front of the film, and electron accumulation at the back of the film. (Reprinted from [61] with
permission of SID)
dark, but, with both 3.4 eV (365 nm) band-gap illumination and with 2.7 eV
(460 nm) sub-band visible illumination (in Fig. 9.22b and c, respectively), the
threshold voltage was negatively shifted, and there was an associated time-
dependent deterioration in the sub-threshold slope and the mobility values as well
[57]. The shift in the transfer characteristics, and the increase in sub-threshold
slope, are indicative of NBIS-induced positive charge trapping and trap generation
in the upper half of the band-gap, respectively. Although the effects were greatest
with band-gap light, more attention has been paid to the sub-band light, as this is
the wavelength range experienced by the addressing TFTs in active matrix dis-
plays. Even without a negative gate bias, optical illumination alone shifted the
threshold voltage negatively, as shown by the zero stress time results in Fig. 9.22b,
c.
Following the termination of the NBIS tests, and the removal of the illumination
and gate bias, there was a slow and thermally activated relaxation of the charac-
teristics over a period of many hours, irrespective of whether bias had been applied
or not. An example of the relaxation of the channel current (at VG = 0 V), fol-
lowing illumination alone, is shown in Fig. 9.23a, where the data has been fitted by
a stretched exponential, and the insert shows the relaxation activation energy. The
illumination-induced threshold voltage shift (both without and without bias [61]),
leading to the large, dark channel currents, is referred to as persistent photocon-
ductivity [61, 62].
The above effects have been attributed to photo-ionisation of the neutral oxygen
vacancy, V0 [18, 59, 61, 62]. This vacancy level is in the lower half of the band
gap, and, by virtue of its usual neutral charge state and its position below the Fermi
level, it does not usually influence the operation of n-channel TFTs. However,
under negative gate bias it is raised towards the Fermi level at the surface, and sub-
band white light can photo-ionise it by exciting electrons to the conduction band,
producing positively charged V0+ and V0++ centres. As the oxygen vacancy is a
negative-U centre [18, 62], the ionised levels are above the neutral level, in the
upper half of the band gap, and the V0++ level is expected to be close to the
conduction band edge [18]. For the centre to have this negative-U behaviour, it is
necessary for the lattice to relax, by the movement of adjacent metal ions, when
the centre is ionised, and this results in a potential barrier between the ionised and
neutral states, which must be overcome for the centre to relax back to its neutral
state when the illumination is switched off [62]. This is consistent with the ther-
mally activated relaxation process, shown in Fig. 9.23a, where the measured
activation energy is of the same order as calculations of the thermal barrier height
[62]. The effects described above are summarised in the illustrative band diagram
in Fig. 9.23b [61] of the TFT under NBIS. This shows the upward bending of the
bands at the surface, and the localised presence of the V0++ centres, whilst, at the
back of the film, the bands are bent downwards to accommodate the photo-gen-
erated electrons, and, in this region, the oxygen vacancy is occupied by electrons
and is neutral. The band bending will separate the photo-generated holes and
electrons, with the field sweeping the holes to the front surface of the film, where
they are trapped in the gate dielectric or in interface states.
Although the role of the oxygen vacancy has been widely invoked in the NBIS
instability mechanism, there are also results linking the instability to inadequate
passivation of the back surface of the film, and AlOx [68] and Y2O3 [69] pas-
sivated structures have been shown to have much better stability than unpassivated
ones. From the a-IGZO film thickness dependence of the instability, it was also
shown that the photo-ionisable defects in the lower half of the band-gap, which
underlie the instability in the unpassivated films, were generated on the top
(exposed) surface of the film [69]. The greater instability in the unpassivated films
has been attributed to the introduction of water molecules or the desorption of O-
ions [68], or possibly due to proton movement from absorbed hydrogen [69].
Whilst improved control of the NBIS instability has been demonstrated, the
precise mechanism, as the above review indicates, is still open to debate, and it
continues to be the subject of widespread research.
9.4.3.3 Stability Considerations for Active Matrix Addressing TFTs
For active matrix applications, the addressing TFT will experience an alternating
combination of positive and negative gate biases, and, in assessing the long-term
lifetime of the TFT, it is useful to determine whether this pulsed AC stress pro-
duces the same instability as the DC stress discussed above. Figure 9.24 compares
the effect of DC stress with both single polarity and bipolarity pulsed bias stress (in
an inverted staggered a-IGZO TFT passivated with an SiOx layer) [70]. In these
particular devices, the threshold voltage shifts were comparable for both negative
and positive bias DC stresses, but they were reduced almost to zero for pulsed
negative bias stress, and halved for pulsed positive bias stress. These pulsed results
were for a 50 % duty cycle with a 5 ms pulse width. For this duty cycle, the
negative pulse bias results were independent of the absolute pulse width (with DVT
Fig. 9.24 Comparison of the

threshold voltage shifts after
pulsed bias and DC bias
stress of a-IGZO TFTs.
permission of SID)
remaining close to zero), whilst, for positive pulse bias, DVT increased with
increasing pulse width, and tended towards the DC values. The small DVT noted
with the negative pulse bias stress was attributed to the slow hole accumulation
processes within the material, and the consequent reduction in hole trapping at the
surface compared with the DC results [70]. Due to the minimal effect of the
negative pulse stressing, the bipolar pulse stressing in Fig. 9.24 (open square
symbols) was almost entirely dominated by the positive pulse bias results, and
were well approximated by the summation of the negative and positive pulse bias
results. In contrast, the summation of the DC results did not predict the effect of
bipolar pulse stressing. This investigation did not include negative pulsed bias
stress under illumination, which, for DC stressing, produces a much larger DVT.
However, this effect has been examined, in a-HfInZnO TFTs [71], as a function of
duty cycle and frequency, with a square wave pulse oscillating between +20 V and
-20 V. In agreement with the dark, negative pulse bias results of Fig. 9.24, the
negative threshold voltage shift with illumination was suppressed for duty cycles
greater than *1 % (positive bias), and the net shift was determined almost
entirely by the positive bias stress cycle. Hence, these results indicate that, under
pulsed operation conditions, the NBIS instability may be considerably less than
observed with DC measurements.
Notwithstanding the above comments, all instability processes should be
reduced as much as possible for TFTs used in commercial applications. The
interaction of processing procedures and instability effects were discussed in
Sect. 9.4.3.1, and improved stability can be achieved by post-deposition annealing
of the a-IGZO, the use of a high quality gate dielectric, and effective passivation
of the back surface to prevent moisture and oxygen ingress into the TFT. The
sequential layer depositions must also be carried out in a way which minimises
Fig. 9.25 Constant current

stress instability of Al2O3 and
SiOx/SiNx passivated a-IGZO
TFTs compared with a-Si:H,
poly-Si, and micro-crystalline
Si TFTs. (Reprinted from
damage to the underlying films, whilst also being consistent with large area, high-
throughput manufacturing. As indicated in Sect. 9.1, there are many examples of
demonstrator displays being fabricated with a-IGZO addressing TFTs [2–4, 6, 13],
several of which report high stability devices. One example of this [72, 73] was an
inverted staggered etch-stop TFT, employing a dual layer gate dielectric of
PECVD SiNx and SiOx (with the latter uppermost and adjacent to the DC-sput-
tered a-IGZO), an SiOx etch-stop layer, Mo source/drain contacts covering the a-
IGZO island edges, and a final passivation layer of DC-sputtered Al2O3. From
positive gate bias stress measurements at 50 °C, an operating life of 10 years was
predicted with an extrapolated threshold voltage shift of \1 V. In addition, after
104s NBIS at -20 V, using a white light source emitting in the range 380-720 nm,
the threshold voltage shift was only -0.2 V, and this was attributed to excellent
back-face passivation by the DC-sputtered Al2O3 film [72]. With the above
measures, the long-term stability of these passivated a-IGZO TFTs for AMOLED
displays is expected to exceed that of a-Si:H TFTs, be comparable to micro-
crystalline Si, and to approach poly-Si, as shown by the constant current stress
data in Fig. 9.25 [73].
9.5 AOS TFT Circuits
In view of the use of AOS material for the pixel driving TFT in demonstrator FPDs,
there is widespread interest in exploiting the high carrier mobility to integrate driver
circuits into these displays. From considerations of power consumption in digital
logic circuits, it is preferable to use a complementary (p- and n-channel) device
9.5 AOS TFT Circuits 333
technology. As discussed in Sect. 9.2, the AOS materials, including a-IGZO, used for
the high electron mobility devices, have a large density of states in the lower half of
the band-gap, giving very poor p-channel behaviour [12]. However, by careful
choice of the constituent materials, good p-channel TFT operation has been achieved
in SnO, due to the valence band edge being made up of spatially spread s-orbitals
from the Sn2+ ion [74]. The hole mobility was 1.3 cm2/Vs, but the n-channel TFTs in
this material had a field effect mobility of only 1.5910-4 cm2/Vs [75]. Hence, as yet,
there is not a monolithic AOS technology capable of delivering well-matched
complementary pairs of TFTs in the same material, and most of the activity in this
area has focussed on circuit implementation using the single channel n-type a-IGZO
TFTs.
Given these limitations, alternative inverter circuit architectures (compared with
the conventional two complementary TFTs) have been investigated to obtain good
voltage switching, ideally with low power consumption. One option is to use a
depletion device as the load TFT, and this has been engineered by using a double-
gated load TFT to give a more negative threshold voltage than the drive TFT [76].
This gave a large output voltage swing, but power consumption was high. An
alternative approach used a cross-coupled inverter design, containing six TFTs, to
achieve a high output voltage swing, and *70 % lower power consumption than a
conventional 2-stage, single channel inverter [77]. In addition to these specific
circuit studies, a range of driver circuits have already been integrated into a-IGZO
demonstrator displays, including row driver circuits [78, 79] for AMOLEDs, row
and column drivers for an LCD [23], as well as an LCD driver circuit [80] incor-
porating a common electrode inversion circuit to reduce the overall power con-
sumption (compared with the more common dot inversion technique [81]).
(AMLCD drive inversion schemes are briefly reviewed in Sect. 4.3).
9.6 Summary
Transparent amorphous oxide semiconductor materials, and in particular a-InG-

aZnO (a-IGZO), are attracting increasing attention as an alternative TFT material
to a-Si:H and poly-Si, especially for large sized AMOLED displays. This is
because the material has an electron mobility *10 times higher than a-Si:H, and
is, therefore, more easily able to deliver the larger drive currents needed for
OLEDs. Although the mobility is still an order less than poly-Si, it may have better
uniformity by being an amorphous material, and promises a simpler processing
schedule. However, the current TFTs are only n-channel, and integrated logic
circuits made in this material will be slower and consume more power than the
complementary pair circuits available with poly-Si TFTs. Nevertheless, there are
now examples of driver circuits integrated into a-IGZO demonstration displays,
and attention is being given to the design of low power, single channel circuits.
The high electron mobility in this amorphous material is a consequence of its
ionic bonding structure, which also delivers a low density of trapping states near the
conduction band edge. Hence, the TFTs have a sharper turn-on than a-Si:H TFTs,
and state-of-the-art devices also display better gate bias stability, which is another
important issue for the AMOLED application. However, there still are stability
issues with AOS TFTS, especially with negative bias stress under illumination
conditions. This instability has been associated with photo-ionisation of the oxygen
vacancy, and this is an on-going research topic. From an application point of view,
the instability is reduced under pulsed bias conditions, and good passivation of the
device has been demonstrated to significantly diminish the effect.
Due to the low DOS, the Fermi level can be moved into the conduction band,
where the conduction process in a-IGZO is controlled by percolation. This results
in a carrier density dependent mobility, which has to be taken into account when
measuring the DOS. In particular, the conventional iterative fit of a device model
to the transfer characteristics will over-estimate the DOS if the on-state field-effect
mobility is simply used. To avoid these problems, C-V techniques have been
established for the DOS measurement, and, by using these DOS values, the carrier-
dependent mobility can be extracted from the transfer characteristics.
The device structure is typically either inverted staggered etch-stop or inverted
coplanar, which can be made with a 5 or 4 mask process, respectively. In principle,
this is a low temperature process, including the deposition of a-IGZO by DC
sputtering at room temperature, although post-deposition anneals are required to
stabilise the material. Nevertheless, the potential for low temperature processing
also makes it attractive for flexible substrate applications. Many current devices
are fabricated with metal source and drain contacts directly on the a-IGZO, and
attention is being directed towards series resistance issues, with some processing
schedules incorporating a metal sinter stage to reduce the contact resistance
between the metal and the semiconductor. Device performance is sensitive to
ambient oxygen and water, and effective passivation of the structure is essential for
the fabrication of high quality TFTs with stable long-term operation.
In summary, AOS TFTs are commanding widespread application interest, with
numerous demonstrations of a-IGZO addressed displays. This is supported by a
substantial research activity into the details of device behaviour, and further
improvements in device physics and TFT performance can be anticipated. In
addition, commercial applications to AMOLED displays, and to large, high
resolution AMLCDs have been forecast.
References
1. Nomura K, Ohta H, Takagi A, Kamiya T, Hirano M, Hosono H (2004) Room-temperature

fabrication of transparent flexible thin-film transistors using amorphous oxide
semiconductors. Nature 432:488–492
2. Lu H–H, Ting H-C, Shih T-H, Chen C-Y, Chuang C-S, Lin Y (2010) 32-inch LCD Panel
Using Amorphous Indium-Gallium-Zinc-Oxide TFTs. SID’10 Technical Digest 1136–1138
3. Mo YG, Kim M, Kang CK, Jeong JH, Park YS, Choi CG, Kim HD, Kim SS (2011)
Amorphous-oxide TFT backplane for large-sized AMOLED TVs. J SID 19(1):16–20
References 335
4. Ito M, Miyazaki C, Ishizaki M, Kon M, Ikeda N, Okubo T, Matsubara R, Hatta K, Ugajin Y,

Sekine N (2008) Application of amorphous oxide TFT to electrophoretic display. J Non-Cryst
Solids 354:2777–2782
5. Kamiya T, Hosono H (2010) Material characteristics and applications of transparent
amorphous oxide semiconductors. NPG Asia Mater 2(1):15–22
6. Hung M-C, Hsiao H-T, Lin W-T, Tu C-H, Chang J–J, Chen P-L (2011) Employ present five
masks amorphous silicon thin-film transistor design and process flow to realize 5-in.
InGaZnO active-matrix liquid crystal display with improved stress stability. Jpn J Appl Phys
50:03CB07
7. Semenza P (2011) Large TFT-LCD panels shift into high resolution. Information Display
27(9):30–32
8. Paetzl R, Brune J, Herbst L, Simon F, Turk B A (2009) Advanced laser crystallisation for
active-matrix display manufacturing. Proc 5th International TFT Conf, ITC’09, 10.2, France
9. Hosono H, Nomura K, Ogo Y, Uruga T, Kamiya T (2008) Factors controlling electron
transport properties in transparent amorphous oxide semiconductors. J Non-Cryst Solids
354:2796–2800
10. Nomura K, Kamiya T, Ohta H, Hirano M, Hosono H (2008) Defect passivation and
homogenization of amorphous oxide thin-film transistor by wet O2 annealing. Appl Phys Lett
93:192107
11. Kuo A, Won TK, Kanicki J (2008) Advanced multilayer amorphous silicon thin-film
transistor structure: film thickness effect on its electrical performance and contact resistance.
Jpn J Appl Phys 47(5):3362–3367
12. Kamiya T, Nomura K, Hosono H (2009) Origins of high mobility and low operation voltage
of amorphous oxide TFTs: Electronic structure, electron transport, defects and doping.
J Display Technology 5(7):273–288
13. Seo H-S, Bae J-U, Kim D-W, Ryoo C I, Kang I-K, Min S-Y, Kim Y–Y, Han J-S, Kim C-D,
Hwang Y-K, Chung I-J (2010) Development of highly stable a-IGZO TFT with TiOx as a
passivation layer for active-matrix display. SID’10 Digest 1132:1135
14. Narushima S, Orita M, Hirano M, Hosono H (2002) Electronic structure and transport
properties in the transparent amorphous oxide semiconductor 2 CdO.GeO2. Phys Rev B
66:035203
15. Kamiya T, Nomura K, Hirano M, Hosono H (2008) Electronic structure of oxygen deficient
amorphous oxide semiconductor a-InGaZnO4–x: Optical analyses and first-principle
calculations. Phys Stat Sol C 5(9):3098–3100
16. Kamiya T, Nomura K, Hosono H (2009) Electronic structure of the amorphous oxide
semiconductor a-InGaZnO4–x: Tauc–Lorentz optical model and origins of subgap states. Phys
Stat Sol A 206(5):860–867
17. Kang D, Lim H, Kim C, Song I, Park J, Park Y, Chung J (2007) Amorphous gallium indium
zinc oxide thin film transistors: Sensitive to oxygen molecules. Appl Phys Lett 90:192101
18. Ryu B, Noh H-K, Choi E-A, Chang KJ (2010) O-vacancy as the origin of negative bias
illumination stress instability in amorphous In–Ga–Zn–O thin film transistors. Appl Phys Lett
97:022108
19. Omura H, Kumomi H, Nomura K, Hirano M, Kamiya T, Hosono H (2009) First-principles
study of native point defects in crystalline indium gallium zinc oxide. J Appl Phys
105:093712
20. Park J-S, Jeong JK, Chung H-J, Mo Y-G, Kim HD (2008) Electronic transport properties of
amorphous indium-gallium-zinc oxide semiconductor upon exposure to water. Appl Phys
Lett 92:072104
21. Lee S, Ghaffarzadeh K, Nathan A, Robertson J, Jeon S, Kim C, Song I-H, Chung U-I (2011)
Trap-limited and percolation conduction mechanisms in amorphous oxide semiconductor thin
film transistors. Appl Phys Lett 98:203508
22. Arai T, Morosawa N, Tokunaga K, Terai Y, Fukumoto E, Fujimori T, Sasaoka T (2011)
Highly reliable oxide-semiconductor TFT for AMOLED displays. J SID 19(2):205–211
23. Osada T, Akimoto K, Sato T, Ikeda M, Tsubuku M, Sakata J, Koyama J, Serikawa T, and
Yamazaki S (2010) Development of liquid crystal display panel integrated with drivers using
amorphous In–Ga–Zn-Oxide thin film transistors. Jpn J Appl Phys 49: 03CC02
24. Kim M, Jeong JH, Lee HJ, Ahn TK, Shin HS, Park J-S, Jeong JK, Mo Y-G, Kim HD (2007)
High mobility bottom gate InGaZnO thin film transistors with SiOx etch stopper. Appl Phys
Lett 90:212114
25. Godo H, Kawae D, Yoshitomi S, Sasaki T, Ito S, Ohara H, Kishida H, Takahashi M,
Miyanaga A, and Yamazaki S (2010) Temperature dependence of transistor characteristics
and electronic structure for amorphous In–Ga–Zn-Oxide thin film transistor. Jpn J Appl Phys
49:3CB04
26. Ohara H, Sasaki T, Noda K, Ito S, Sasaki M, Toyosumi Y, Endo Y, Yoshitomi S, Sakata J,
Serikawa T, Yamazaki S (2009) 4.0 in. QVGA AMOLED display using In-Ga-Zn-Oxide
TFTs with a novel passivation layer. SID’09 Digest 284–287
27. Geng D, Kang DH, Jang J (2011) High-performance amorphous indium–gallium–zinc–oxide
thin-film transistor with a self-aligned etch stopper patterned by back-side UV exposure.
IEEE Electron Dev Lett 32(6):758–760
28. Morosawa N, Ohshima Y, Morooka M, Arai T, Sasaoka T (2011) A novel self-aligned top-
gate oxide TFT for AM-OLED displays. SID’11 Tech Digest 479–482
29. Lippens P (2012) Indium tin oxide (ITO): Sputter deposition processes. In: Chen J, Cranton
W, Fihn M (eds) Handbook of Visual Display Technology. Springer-Verlag, Berlin
Heidelberg, p 2012
30. Hsieh H–H, Tsai T–T, Chang C-Y, Wang H–H, Huang J-Y, Hsu S-F, Wu Y-C, Tsai T-C,
Chuang C-S, Chang L-H, Lin Y-H (2010) A 2.4-in. AMOLED with IGZO TFTs and inverted
OLED devices. SID’10 Digest 140–143
31. Barquinha P, Pereira L, Gonçalves G, Kuscer D, Kosec M, Vilà A, Olziersky A, Morante JR,
Martins R, Fortunato E (2010) Low-temperature sputtered mixtures of high-j and high
bandgap dielectrics for GIZO TFTs. J SID 18(10):762–772
32. Chiu CJ, Chang SP, Chang SJ (2010) (2010) High-performance a-IGZO thin-film transistor
using Ta2O5 gate dielectric. IEEE Electron Dev Lett 31(11):1245–1247
33. Lee JS, Chang S, Koo S-M, Lee SY (2010) High-performance a-IGZO TFT with ZrO2 gate
dielectric fabricated at room temperature. IEEE Electron Dev Lett 31(3):225–227
34. Su NC, Wang SJ, Chin A (2009) High-performance InGaZnO thin-film transistors using
HfLaO gate dielectric. IEEE Electron Dev Lett 30(12):1317–1319
35. Robertson J (2002) Electronic structure and band offsets of high-dielectric-constant gate
oxides. MRS Bull 27:217–221
36. Park J, Kim C, Kim S, Song I, Kim S, Kang D, Lim H, Yin H, Jung R, Lee E, Lee J, Kwon K-
W, Park Y (2008) Source/drain series-resistance effects in amorphous gallium–indium zinc-
oxide thin film transistors. IEEE Electron Dev Lett 29(8):879–881
37. Kondratyuk RI, Im K, Stryakhilev D, Choi CG, Kim M-G, Yang H, Park H, Mo YG, Kim
HD, Kim SS (2011) A study of parasitic series resistance components in In–Ga–Zn–Oxide (a-
IGZO) thin-film transistors. IEEE Electron Dev Lett 32(4):503–505
38. Bae H, Kim S, Bae M, Shin JS, Kong D, Jung H, Jang J, Lee J, Kim DH, Kim DM (2011)
Extraction of separated source and drain resistances in amorphous indium–gallium–zinc
oxide TFTs through C-V characterization. IEEE Electron Dev Lett 32(6):761–763
39. Chern JGJ, Chang P, Motta RF, Godinho N (1980) A New Method To determine MOSFET
channel length. IEEE Electron Dev Lett 1(9):170–173
40. Kanicki J, Libsch FR, Griffith J, Polastre R (1991) Performance of thin hydrogenated
amorphous silicon thin-film transistors. J Appl Phys 69(4):2339–2345
41. Chen C-Y, Kanicki J (1998) Origin of series resistances in a-Si:H TFTs. Solid-State
Electronics 31(4):694–602
42. Hu G J, Chang C, and Chia Y-T (1987) Gate-voltage-dependent effective channel length and
series resistance of LDD MOSFETs. IEEE Trans Electron Dev ED-34(12):2469–2475
43. Kanicki J (1988) Contact resistance to undoped and phosphorus-doped hydrogenated
amorphous silicon films. Appl Phys Lett 53(20):1943–1945
References 337
44. Song I, Kim S, Yin H, Kim CJ, Park J, Kim S, Choi HS, Lee E, Park Y (2008) Short channel
characteristics of gallium–indium–zinc–oxide thin film transistors for three-dimensional
stacking memory. IEEE Electron Dev Lett 29(6):549–552
45. Hosono H (2006) Ionic amorphous oxide semiconductors: Material design, carrier transport,
and device application. J Non-Crystalline Solids 352:851–858
46. Kamiya T, Nomura K, Hosono H (2010) Origin of definite Hall voltage and positive slope in
mobility-donor density relation in disordered oxide semiconductors. Appl Phys Lett
96:122103
47. Kamiya T, Nomura K, Hosono H (2009) Electronic structures above mobility edges in
crystalline and amorphous In-Ga-Zn-O: Percolation conduction examined by analytical
model. J Display Technology 5(12):462–467
48. Kimura M, Nakanishi T, Nomura K, Kamiya T, Hosono H (2008) Trap densities in
amorphous-InGaZnO4 thin-film transistors. Appl Phys Lett 92:133512
49. Kimura M, Kamiya T, Nakanishi T, Nomura K, Hosono H (2010) Intrinsic carrier mobility in
amorphous In–Ga–Zn–O thin-film transistors determined by combined field-effect technique.
Appl Phys Lett 96:262105
50. Jeon K, Kim C, Song I, Park J, Kim S, Kim S, Park Y, Park J-H, Lee S, Kim DM, Kim DH
(2008) Modeling of amorphous InGaZnO thin-film transistors based on the density of states
extracted from the optical response of capacitance-voltage characteristics. Appl Phys Lett
93:182102
51. Kimura M, Nozawa R, Inoue S, Shimoda T, Lui BO-K, Tam SW-B, Migliorato P (2001)
Extraction of trap states at the oxide-silicon interface and grain boundary for polycrystalline
silicon thin-film transistors. Jpn J Appl Phys 40:5227–5236
52. Kimura M (2010) Extraction technique of trap densities in thin films and at insulator
interfaces of thin-film transistors. IEEE Electron Dev Lett 31(6):570–572
53. Nomura K, Kamiya T, Hirano M, Hosono H (2009) Origins of threshold voltage shifts in
room-temperature deposited and annealed a-In–Ga–Zn–O thin-film transistors. Appl Phys
Lett 95:013502
54. Servati P, Striakhilev D, and Nathan A (2003) Above-threshold parameter extraction and
modeling for amorphous silicon thin-film transistors. IEEE Trans ED-50(11):2227–2235
55. Robertson J (2008) Disorder and instability processes in amorphous conducting oxides. Phys
Stat Sol (b) 245(6):1026–1032
56. Jeong JK, Yang HW, Jeong JH, Mo Y-G, Kim HD (2008) Origin of threshold voltage
instability in indium-gallium-zinc oxide thin film transistors. Appl Phys Lett 93:123508
57. Nomura K, Kamiya T, Hosono H (2010) Interface and bulk effects for bias–light-illumination
instability in amorphous-In–Ga–Zn–O thin-film transistors. J SID 18(10):789–795
58. Lopes ME, Gomes HL, Medeiros MCR, Barquinha P, Pereira L, Fortunato E, Martins R,
Ferreira I (2009) Gate-bias stress in amorphous oxide semiconductors thin-film transistors.
59. Oh H, Yoon S-M, Ryu m K, Hwang C-S, Yang S, Park S-HK (2010) Photon-accelerated
negative bias instability involving subgap states creation in amorphous In–Ga–Zn–O thin film
transistor. Appl Phys Lett 97:183502
60. Chen T-C, Chang T-C, Tsai C-T, Hsieh T-Y, Chen S-C, Lin C-S, Hung M-C, Tu C-H, Chang
J–J, Chen P-L (2010) Behaviors of InGaZnO thin film transistor under illuminated positive
gate-bias stress. Appl Phys Lett 97:112104
61. Ghaffarzadeh K, Lee S, Nathan A, Robertson J, Jeon S, Kim S, Kim C, Chung U-I, Lee J-H
(2011) Light-bias induced instability and persistent photoconductivity in In-Zn-O/Ga-In-Zn-
O thin film transistors. SID’11 Technical Digest 1154–1157
62. Ghaffarzadeh K, Nathan A, Robertson J, Kim S, Jeon S, Kim C, Chung U-I, Lee J-H (2010)
Persistent photoconductivity in Hf–In–Zn–O thin film transistors. Appl Phys Lett 97:143510
63. Chowdhury MDH, Migliorato P, Jang J (2010) Light induced instabilities in amorphous
indium–gallium–zinc–oxide thin-film transistors. Appl Phys Lett 97:173506
64. Park S-HK, Ryu M-K, Yoon S-M, Yang S, Hwang C-S, Jeon J-H (2010) Device reliability
under electrical stress and photo response of oxide TFTs. J SID 18(10):779–788
65. Jackson WB, Hoffman R, Yeh B, Emery T, Koch T, McConica C, Kwon O (2010)
Metastability in multicomponent oxide transistors. Phys Stat Sol A 207(3):695–699
66. Lee K-H, Jung JS, Son KS, Park JS, Kim TS, Choi R, Jeong JK, Kwon J-Y, Koo B, Lee S
(2009) The effect of moisture on the photon-enhanced negative bias thermal instability in
Ga–In–Zn–O thin film transistors. Appl Phys Lett 95:232106
67. Lee J, Park J-S, Pyo YS, Lee DB, Kim EH, Stryakhilev D, Kim TW, Jin DU, Mo Y-G (2009)
The influence of the gate dielectrics on threshold voltage instability in amorphous indium-
gallium-zinc oxide thin film transistors. Appl Phys Lett 95:123502
68. Yang S, Cho D-H, Ryu m K, Park S-HK, Hwang C-S, Jang J, Jeong JK (2010) Improvement
in the photon-induced bias stability of Al–Sn–Zn–In–O thin film transistors by adopting AlOx
passivation layer. Appl Phys Lett 96:213511
69. Nomura K, Kamiya T, Hosono H (2011) Highly stable amorphous In-Ga-Zn-O thin-film
transistors produced by eliminating deep subgap defects. Appl Phys Lett 99:053505
70. Fung T-C, Abe K, Kumomi H, and Kanicki J (2009) DC/AC electrical instability of R.F.
sputter amorphous In-Ga-Zn-O TFTs. SID’09 Tech Digest 1117–1120
71. Son K-S, Kim H-S, Maeng W-J, Jung J-S, Lee k-H, Kim T-S, Park JS, Kwon J-Y, Koo B, Lee
S-Y (2011) The Effect of Dynamic Bias Stress on the Photon-Enhanced Threshold Voltage
Instability of Amorphous HfInZnO Thin-Film Transistors. IEEE Electron Dev Lett
32(2):164–166
72. Arai T, Morosawa N, Tokunaga K, Terai Y, Fukumoto E, Fujimori T, Sasaoka T (2011)
Highly reliable oxide-semiconductor TFT for AMOLED displays. J SID 19(2):205–211
73. Arai T and Sasaoka T (2011) Emergent oxide TFT technologies for next-generation AM-
OLED displays. SID’11 Tech Digest 710–713
74. Ogo Y, Hiramatsu H, Nomura K, Yanagi H, Kamiya T, Hirano M, Hosono H (2010)
P-channel thin-film transistor using p-type oxide semiconductor, SnO. Appl Phys Lett
93:032113
75. Hosono H, Nomura K, and Kamiya T (2011) An Ambipolar Oxide TFT. SID’11 Tech Digest
486–487
76. Seok MJ, Choi MH, Mativenga M, Geng D, Kim DY, Jang J (2011) A Full-swing a-IGZO
TFT-based inverter with a top-gate-bias-induced depletion load. IEEE Electron Dev Lett
32(8):1089–1091
77. Hwang T-H, Yang I-S, Kwon O-K, Ryu M-K, Byun C-W, Hwang C-S, Park S-H K (2011)
Inverters using only n-type indium gallium zinc oxide thin film transistors for flat panel
display applications. Jpn J Appl Phys 50, 03CB06
78. Kang C-K, Park Y-S, Park S-I, Mo Y-G, Kim B-H, Kim S S (2011) Integrated scan driver
with oxide tfts using floating gate method. SID’11 Tech Digest 25–27
79. Miura K, Ueda T, Nakano S, Saito N, Hara Y, Sugi K, Sakano T, Yamaguchi H, Amemiya I,
Akimoto K, Kameoka H, and Tonotani J (2011) Low-temperature-processed IGZO TFTs for
flexible AMOLED with integrated gate driver circuits. SID’11 Tech Digest 21–24
80. Jeong H, Mativenga M, Jang J, Lee S G, and Ha Y M (2011) Design of a low power
consumption a-IGZO TFT-based Vcom driver circuit with long-term reliability. SID’11 Tech
Digest 338–341
81. Kudo Y, Akai A, Furuhashi T, Matsudo T, Yokota Y (2003) Low-power and high-integration
driver IC for small-sized TFT-LCDs. SID’03 Tech Digest 1244–1247
Chapter 10
Organic TFTs
Abstract A large number of organic TFT materials now display carrier mobility
values approaching, or exceeding, those of a-Si:H, and this has stimulated
considerable interest in using them for low cost, flexible substrate applications,
such as RFID tags, and e-reader displays. This chapter reviews work in this field,
and starts by introducing some of the key organic chemistry concepts, which
underpin the semiconducting behaviour of these materials. Other topics include
molecular ordering, carrier transport, TFT architecture, and the deposition of both
semiconductor and dielectric materials, together with their properties. These are
linked to device behaviour, with a focus on the key material and design issues,
which currently limit TFT performance, such as metal/organic contacts, and
ambient and bias stability effects.
10.1 Introduction
The commercial interest in organic TFTs, OTFTs, for display applications started
in the early-1990s, when the carrier mobility in small molecule material, partic-
ularly pentacene, exceeded 0.1 cm2/Vs, and began to approach the values of
0.5–1.0 cm2/Vs obtained in a-Si:H. Since that time, the performance of organic
materials has continued to improve, and, when viewed on a 20 year timescale,
increases in carrier mobility of several orders of magnitude have been achieved,
with the highest values now in excess of 10 cm2/Vs, as shown in Fig. 10.1 [1].
Four classes of material are labelled in this diagram, but these include sub-
divisions of the two main families, which are small molecule (sometimes referred
to as oligomer) and long molecule (normally referred to as polymer) conjugated
semiconductors. The further sub-division is amongst the small molecule types,
where they are specified as vacuum deposited, solution processed and single
crystal. (The interest in single crystal organic semiconductors has largely been to

340 10 Organic TFTs
Fig. 10.1 Changes in carrier

mobility in organic TFTs
between 1986 and 2006.
(Reproduced from [1]
with permission of John
Wiley & Sons, Inc)
identify ultimate performance limits. This type of material has been recently
reviewed [2], and will not be discussed further here). The polymer samples are
usually solution processed, which means that their deposition is from a liquid
phase by printing, casting or spin-coating. This is expected to be the most cost-
effective approach to film formation, and underlies the interest in polymers in spite
of their lower mobility, as seen in Fig. 10.1.
Although the attainment of a useful mobility is essential for the currently
envisaged commercial applications, the mobility alone is not sufficient for organic
materials to be seen as potential replacements for a-Si:H. Indeed, there is no
likelihood of them challenging the mainstream technology of a-Si:H TFTs on glass
in the near future. Rather, the interest in them lies in their compatibility with a
flexible substrate technology, using inexpensive polymer substrates, in order to
produce light-weight, portable displays, which can be contoured, folded or rolled.
Commonly used plastic substrates in this work are polyethylene naphthalate
(PEN), and polyethylene terephthalate (PET), which, in their heat stabilised forms,
have maximum handling temperatures of 200 °C and 150 °C, respectively. Further
properties of these, and other polymer substrates, are shown in Table 11.1. In
addition, the challenges in using flexible substrates are discussed in Sect. 11.2, in
which it is shown that the issues of maximum processing temperature, the mis-
match in coefficients of thermal expansion between the substrate and the layers on
them, and general substrate flexing are more acute with the inorganic TFT tech-
nologies. Thus, these topics are expected to be more easily manageable with a TFT
technology which uses the less rigid, and better physically matched, organic TFT
materials. For example, suitably capped OTFT samples, which placed the TFTs in
the neutral strain plane, have been reported with a minimum bending radius of
0.5 mm [3], and working displays have been demonstrated with a bending radius
of 4 mm [4], as shown in Fig. 10.2.
Fig. 10.2 Photograph of a

flexible AMOLED display,
driven by OTFTs, rolled
around a 4 mm radius
cylinder. (Reprinted from [4]
There are now plentiful examples of OTFT research demonstrator displays on

flexible substrates, using LCD [5], electro-phoretic, EPD [6–8], and OLED [4, 9, 10]
media, as well as demonstration logic circuits [11–13]. The most obvious first
application of OTFTs is in low cost RFID tags, and in AMEPDs (due to the reduced
pixel drive current requirements). Some e-reader proto-products have been pre-
sented [6, 8], in which the ultimate aim is a compact device containing a rollable/
foldable screen [8], and the first product, a rigid-framed e-reader for educational use,
was announced in late 2011 [14].
In common with other TFT technologies, a mature technology should deliver
not merely basic device performance, with an acceptable carrier mobility, leakage
current and threshold voltage, but also good device uniformity and reliability.
These properties should be delivered in a cost-effective, reproducible process, and
offer the prospect of increased device/function integration. Of all the TFT tech-
nologies, the OTFT technology is currently the most immature in these respects,
with no clear consensus on the preferred organic material, gate dielectric and
process flow, but it is driven by the prospect of a low-cost, flexible substrate
technology, with laboratory demonstrators and the first product highlighting the
potential of this approach.
In another important respect, the activity covered in this chapter is different
from the preceding TFT chapters, in that they dealt with just one semiconductor
material for the TFT channel. In contrast, the research programmes into OTFTs
have investigated an extensive range of different organic materials, both for the
TFT channel, and, to a lesser extent, for the gate dielectric and contact metals.
Moreover, continued development of new and improved molecular structures is an
on-going activity, and recent reviews [15–17] have given a detailed overview of
the range of materials examined. Similar specialist coverage of different materials
is less appropriate for this book. Instead, the following sections of this chapter will
focus on the broader background concepts and issues in using organic materials to
fabricate TFTs, and with examples drawn from the polymer and oligomer mate-
rials of current interest.
Sections 10.2.1 and 10.2.2 are an elementary introduction to some of the
background organic chemistry concepts routinely employed in the study of organic
materials for TFTs, and cover the role of the molecular and orbital structure of the
material in determining its semiconductor behaviour. Section 10.2.3 reviews
the role of structural organisation in optimising TFT performance, including the
342 10 Organic TFTs
growth of ordered domains and poly-crystalline regions. Organic semiconductors

are usually prepared undoped, and n- and p-channel behaviour is partially deter-
mined by the particular metal used for the contacts, and, specifically, by the
alignment of the metal work function and the carrier bands in the semiconductor.
Metal/organic contacts have been widely studied, and an overview of the key
issues for TFTs is presented in Sect. 10.2.4. Finally, Sect. 10.2.5 reviews the
current understanding of the carrier transport mechanisms in these materials.
Section 10.3 discusses the architecture of OTFTs, and Sect. 10.4 deals with the
materials and processing procedures used in TFT fabrication. The materials
coverage includes sub-sections on the semiconductors, the gate dielectrics and the
metallisation options. There is also a sub-section on novel processing schemes,
which exploit the self-organising properties of these materials. These techniques
differ radically from the more conventional TFT fabrication process of layer-
by-layer definition by photo-lithography and etching.
The electrical characteristics of OTFTs are discussed in Sect. 10.5, with an
overview of the performance parameters from a range of common semiconductors,
but with a more detailed focus on contact and series resistance effects, which play
a major role in determining overall TFT behaviour. Finally, the ambient and bias
stability issues of OTFTs are reviewed in Sect. 10.6.
10.2 Background and Materials
10.2.1 Conjugated Molecular Systems
The materials used in OTFTs are conjugated molecular systems, which means that
they have alternating single and double carbon–carbon bonds, and their molecular
p-bonding orbitals are responsible for the ability of the material to transport
charge. A simple example of a conjugated molecule is the benzene ring shown in
Fig. 10.3a [18], which is a building block for a number of important OTFT
compounds. The figure shows the chemical structure of benzene, C6H6, which is a
planar ring with the alternating double bonds between the carbon atoms. Also
shown are the two commonly employed symbols, which are used to represent the
benzene ring in more complex molecules, such as those shown in Fig. 10.3b, c.
Figure 10.3b is an acene structure, which consists of a number of linearly fused
benzene rings, giving a rod-like molecular structure, and the subscript n, in the
diagram, refers to the number of repeating units of the bracketed ring structure.
A specific example of this structure is the pentacene molecule, consists of five
rings, as shown in Fig. 10.3c, and this is the most widely studied small molecule
material, yielding high mobility p-channel TFTs.
A similar planar ring structure molecule is thiophene, C4H4S, which is shown in
Fig. 10.3d, together with its symbolic representation. Thiophene is a building
block of another group of OTFT materials, consisting of both small molecule
10.2 Background and Materials 343
Fig. 10.3 Chemical structure of a benzene ring, and its symbolic representations, b acene
molecule, c pentacene, d thiophene, and its symbolic representation, e polythiophene, f regio-
regular poly(3-hexylthiophene), P3HT, g poly(3,300 dialkylquaterthiophene), PQT-12, and
h hexane linear chain molecule, and its symbolic representation. (f Reproduced from [22] with
permission of John Wiley & Sons, Inc, and g Reprinted with permission from [21]. Copyright
(2004) American Chemical Society)
oligothiophenes [19], as well as the polymers polythiopene and its derivatives

poly(3-hexylthiophene), P3HT [20], and poly(3,3000 dialkylquaterthiophene),
PQT-12 [21], shown in Fig. 10.3e–g, respectively. The substituted side chains, R,
on the basic thiophene molecule are alkyls, CnH2n+1, which are radicals of the
alkane group, CnH2n+2.These are single-bond chains of carbon and hydrogen, and
the six carbon molecule, hexane, C6H14, is illustrated in Fig. 10.3h. (This figure
also shows its conventional symbolic representation, although a simple straight
line, plus the symbol R, is also widely used for this and other substituents, as in
Fig. 10.3f, g). For solution processing of the polymers, alkanes with six or more
carbon atoms are favoured for the substituting alkyl groups on polythiophene
[21, 22], as seen by the use of hexyl, C6H13, in P3HT, and dodecyl, C12H25, in
PQT-12.
The long-range organisation of the molecular structures is essential to give
good TFT performance, and this is discussed in Sect. 10.2.3.
344 10 Organic TFTs
Fig. 10.4 a pi molecular

orbital formation in an
ethene molecule (Reproduced
by permission of Hodder
Education from [18],
copyright (2003)), and
b schematic illustration of
the extended pi-orbitals in a
conjugated molecular system
Physics)
10.2.2 Molecular Bonding
The isolated carbon atom has two electrons in the 2s orbital and two distributed in
the three 2p orbitals, and, when carbon atoms are brought together to form a
molecule containing a double bond, these four atomic orbitals are converted into
three hybrid sp2 orbitals, leaving one remaining 2p orbital. The three hybrid sp2
orbitals form covalent bonds with the adjacent atoms, defined by the molecular r
bonding-orbital in the plane of the atoms, and with the 2p atomic orbitals
perpendicular to this plane. The overlapping 2p-orbitals, on adjacent atoms in the
molecule, form a molecular p-orbital [18], as shown for the simple double bond
molecule, ethene, C2H4, in Fig. 10.4a. The p-orbital only forms where there is a
r-orbital, and double bonds are composed of r and p orbitals. For a conjugated
molecule, the p-orbital delocalises its electrons within the molecule, or along a
conjugated polymer backbone, as shown schematically in Fig. 10.4b [23]. Hence,
the p-orbitals support the flow of charge carriers within the material, and, in a
solid, the filled p-bonding orbitals form the highest occupied molecular orbital,
HOMO, and the empty p-anti-bonding orbitals form the lowest unoccupied
molecular orbital, LUMO [23]. These can be thought of as corresponding to the
valence and conduction bands in conventional inorganic semiconductors, and are
used in the same way to represent the band off-sets between the different
constituents of a TFT structure [24].
10.2.3 Molecular Organisation
It has been widely demonstrated that there needs to be a high degree of structural
order within the material, together with strong electronic coupling between
adjacent molecules, to achieve effective charge transport and high performance
TFTs. Hence, as discussed below, the favoured materials show regions of, at least,
micro-crystallinity if not poly-crystallinity, in contrast with less favoured, low
mobility amorphous materials [28]. Whilst this structural distinction is widely

observed, it would appear not to be universal, as high performance n-channel TFTs
have been reported in amorphous material [25]. Although there is a strong theo-
retical framework underpinning molecular engineering and materials development
[26], it is apparent that a full understanding of the correlation between material
structure and device performance has not yet been achieved [2]. In the remainder
of this section, examples are given of the structural order engineered into both
oligomer and polymer materials.
High performance P3HT OTFTs [27, 28] have been achieved with regio-regular
molecules, in which there is predominantly head-tail, HT, coupling of the adjacent
thiophene rings (at the 2 and 5 positions), with the hexyl substituent, R, in the
number 3 position on the ring [22]. This is illustrated in Fig. 10.3f. The HT
coupling produces molecules with a torsional twist of \10°, indicating good
co-planarity, which is required for efficient p-orbital coupling between adjacent
molecules. Alternative couplings, such as head-head (the 2, 2 positions), can lead
to severe twisting of the molecule by up to 40°, and degraded p-orbital conduction
[22]. Detailed discussion of the chemical synthesis of thiophene polymers can be
found in Ref. [22], but the topic itself is beyond the scope of this book.
The regio-regular P3HT molecules, with a high HT content, self-organise into
micro-crystalline domains containing close packed lamellae, which are vertically
aligned to the substrate surface, as shown in Fig. 10.5a [28]. In contrast, in films
with a low concentration of HT coupling, the lamellae are oriented parallel to the
substrate surface. When these structures were used in OTFTs, the field effect
mobility was up to 103 times higher in the high HT-content films, as illustrated, in
Fig. 10.5b, by the dependence of the field effect mobility on the percentage of HT
coupling [27]. This work also demonstrated the large anisotropy in mobility within
these films, with the most effective carrier transport being by inter-chain, p–p
coupling in these p-stacked materials. Hence, not only is molecular ordering
within the film important, but as significant is the orientation of the molecular
stacking with respect to the direction of channel current flow. It has also been
demonstrated that the formation of vertically aligned molecules can be enhanced
by the prior deposition of self-assembled mono-layers, SAMs, onto the substrate
surface [29]. As will be seen in Sect. 10.4, the use of SAMs is a widely employed
technique for controlling the orientation of organic semiconductor molecules, and
can be regarded as comparable to the use of alignment layers in LCD cells to
control the alignment of the LC molecules [30].
Another thiophene based molecule, regio-regular PQT-12 (see Fig. 10.3g), has
also been demonstrated to self-assemble in an ordered vertical lamellar structure,
which is similar to the P3HT discussed above. The stacking is shown schemati-
cally in Fig. 10.5c [21], and, for TFT samples, the gate oxide surface was coated
with a film of octyltrichlorosilane, OTS, before spin coating with the polymer.
X-ray diffraction of the as-spun films showed a loosely packed lamellar structure,
but, after annealing at 135 °C, a more highly ordered lamellar stack was observed,
oriented perpendicular to the substrate surface. TEM examination of these films
showed an inter-stack spacing of 3.7 Å [21]. The hole mobility was 0.14 cm2/Vs,
346 10 Organic TFTs
Fig. 10.5 a Lamellar molecular ordering of regio-regular P3HT (Reproduced from [28] with
permission of John Wiley & Sons, Inc). b variation of hole mobility with fractional head–tail
coupling in P3HT (Reprinted by permission from Macmillan Publishers Ltd: Nature [27],
copyright (1999)), and c lamellar molecular ordering of PQT-12 (Reprinted with permission
from [21]. Copyright (2004) American Chemical Society)
which is comparable to well ordered P3HT films, but the PQT-12 samples showed
much less hysteresis during measurements, and much better long-term ambient
stability. This was attributed to the larger ionisation potential of PQT-12, which
reduced its susceptibility to oxidation. Device performance issues are discussed in
greater detail in Sects. 10.4, 10.5 and 10.6.
The most widely studied small molecule material is pentacene [26], which,
under well controlled vacuum evaporation conditions, forms a polycrystalline film
on the substrate surface. The grain structure is triclinic, with the (001) plane
parallel to the substrate surface [15], and the molecular alignment follows an edge-
to-face (‘herringbone’), intermolecular stacking pattern. This is one of the
characteristic structures for acenes [26], and is shown in Fig. 10.6 [31].
Figure 10.6a shows that the rigid pentacene molecules are arranged in vertically
stacked layers, with a pitch of *15.4 Å, and, within each layer, the molecules are
at an angle of *84 ± 0.5° to the substrate surface, depending upon the termi-
nation of the surface [32]. The inter-molecular ‘herringbone’ pattern, within each
layer, is more clearly seen in the image in Fig. 10.6b, which is viewed parallel to
the long-axis of the molecule. Given the near vertical alignment of the molecules
on the substrate surface, the structure is often schematically illustrated in TFT
cross-sections, as shown by Fig. 10.6c. Although single crystal pentacene has been
studied, the material is usually poly-crystalline in its thin film form, and an
example of the role of grain size on carrier mobility is shown in Fig. 10.6d [33].
Fig. 10.6 Crystalline structure of pentacene, viewed a along, and b perpendicular to the plane of
the substrate surface. c Schematic TFT representation showing the vertically stacked planes of
near-vertically aligned molecules (Reprinted with permission from [31]. Copyright (2003) by the
American Physical Society), and d variation of field effect mobility with pentacene grain size,
where the dashed line is from a 2-D device model with grain-boundary trapping. (Reprinted with
permission from [33]. Copyright (2005) American Institute of Physics)
In his work, the grain size was controlled by varying the pentacene evaporation
rate, and was assessed by atomic force microscopy. The TFT geometry was
inverted staggered with an SiO2 gate dielectric layer, and the hole field effect
mobility was found to increase with increasing grain size (up to a maximum value
of 0.35 cm2/Vs). The influence of grain size was attributed to grain boundary
trapping effects, as shown by the GB modelling of the data [33]. As with the
polythiophenes, the carrier transport process in pentacene films is also essentially
two-dimensional, by virtue of the better electronic coupling between molecules
within a layer, rather than between layers [34].
10.2.4 Metal/Organic Contacts
The organic channel materials are undoped, and the distinction between n-channel
and p-channel devices is determined by which carrier is injected from the source
348 10 Organic TFTs
contact. In other words, if the work function of the source metal is close to the
LUMO level, electrons will be more easily injected (giving n-channel operation),
and, conversely, if it is closer to the HOMO level, holes will be preferentially
injected, giving p-channel behaviour. Hence, a key consideration in matching the
metal to the organic semiconductor, to achieve n- or p-channel operation, is the
relative displacement of the metal work function from the HOMO or the LUMO
levels. However, it should be noted that whilst a good injecting contact is
necessary for both p- and n-channel device operation, it alone is not sufficient to
produce high performance, stable devices, where impurities and carrier trapping
will be additional determinants of device behaviour [35]. Indeed, the fabrication of
high quality n-channel devices has proven to be more difficult to achieve than for
p-channel devices [25, 35], particularly due to electron trapping effects when SiO2
was used as the gate dielectric [36]. Because of this, well matched complementary
p- and n-channel TFTs, for low-power digital circuits, have not been available
until relatively recently [25, 37]. The issues with the development of high
performance n-channel TFTs are discussed further in Sect. 10.4.3.
In view of the importance of the metal/organic semiconductor contact, it has
been extensively studied [24, 38–40]. The HOMO/LUMO energy band repre-
sentation of the material is particularly useful in considering the contact, as
illustrated in Fig. 10.7a, b [39]. Figure 10.7a shows the simple alignment of the
metal and the semiconductor vacuum levels in the Schottky-Mott limit. The hole
injection barrier, /h, is then given by the difference between the metal work
function, UM, and the semiconductor ionisation energy, IE, and the electron
injection barrier, /e, by the difference between UM and the electron affinity, EA.
With knowledge of these values, an appropriate choice of metal may, in principle,
be made. Unfortunately, practical systems rarely accord with this simple diagram,
irrespective of whether the materials are prepared in UHV [39], or under more
realistic conditions, such as air exposure of the metal prior to the deposition of the
organic material [24, 38]. Usually, a dipole barrier, D, is found at the metal/
organic, MO, interface, as shown in Fig. 10.7b, and this may be either positive or
negative, resulting in an offset of the work function from the location predicted by
the direct alignment of the vacuum levels. In the example shown, the effect of the
dipole barrier is to change the metal from a hole injector to an electron injector.
Moreover, the metal work function itself is sensitive to its environment, and care is
needed in using the text book values. For example, gold is widely used as a contact
metal in OTFTs, and its work function is conventionally quoted as *5.1–5.2 eV
[41], and this value is confirmed in the UHV cited work here [39]. However,
exposing a gold film to air, or rinsing it in an organic solvent, produced a dipole
barrier of 0.7 eV, giving an effective Au work function (the energy gap between
the gold Fermi level and the polymer vacuum level) of *4.5 eV, whereas ozone
exposure increased it back to 5.2–5.5 eV [24]. The effective work function of
*4.5 eV was also found following both the deposition of several different
p-conjugated polymers onto gold, and UHV gold deposition onto pre-deposited
films of the same polymers [24]. The dipole barrier found in this work, of
Fig. 10.7 HOMO and LUMO band diagram of an organic semiconductor showing the band
offset with respect to a metal contact a in the Schottky–Mott limit, and b with the formation of a
surface dipole layer. (Reproduced from [39] with permission of John Wiley & Sons, Inc)
*0.7 eV, is well established [38], although values as high as 1.08 eV have been
quoted for the UHV deposition of pentacene onto gold [39].
The work function, ionisation energy and dipole barrier values quoted above
were derived from ultraviolet photoelectron spectroscopy, UPS, measurements
[24, 38, 39]. Most metals have been found to have a non-zero dipole barrier when
in contact with organic materials, where the values recorded may be a function of
the preparation and handling of the material [24]. Hence, the band diagram shown
in Fig. 10.7b is the more appropriate representation of the contact than Fig. 10.7a.
Several models have been invoked to explain the dipole barriers [24, 38–40],
including Fermi level pinning at interface states, charge exchange between the
materials, and molecular mixing effects. The gold results, cited above, were
explained by a more fundamental ‘push-back’ effect [24], in which the electron
density from a clean metal surface decreases exponentially for a few angstroms
beyond the surface, and charge neutrality is maintained by the corresponding deficit
of negative charge within the material close to the surface. The electron tail
produces a surface dipole, and a corresponding dipole potential energy, with which
the work function scales. The larger the dipole potential energy (and the work
function), the more polarisable the surface is, and the greater the effect of
physisorbed layers pushing the electron tail back, and reducing the surface dipole
(and the associated work function). Thus, metals like gold, with a large work
function, are more sensitive to adsorbed films than low work function metals like
Ca [24].
350 10 Organic TFTs
10.2.5 Carrier Transport
Carrier transport in OTFTs, as measured by the field effect mobility, is a function

of several parameters [34]. The intrinsic factors include the molecular structure of
the material, its ordering to promote good inter-molecular p–p coupling, and the
molecular alignment on the substrate surface so that the high conduction direction
is parallel to the plane of the TFT channel. The extrinsic factors include some or
all of the following: the sample temperature, the lateral electric field, the free
carrier density, and the carrier injection efficiency of the source and drain contact
materials. The precise details of the charge transfer process, and the resultant
mobility values, are still the subject of on-going theoretical study [34, 42].
However, the experimentally measured field effect mobility values are usually
described as being limited by an inter-trap hopping process, in highly disordered
material [43], or by multiple trapping and release, MTR, from localised defect
states, in more ordered material [34, 42] (a process which was discussed in detail
in Sect. 6.2.4), or by trapping effects at grain boundaries [33].
Hopping and MTR are thermally activated processes, and the field effect
mobility, lFE, can be described by the generic equation:
lFE ¼ l0 A exp ðEA =kTÞ ð10:1Þ
where l0 is the trap-free mobility for hopping, and, for MTR, it is the band
mobility, with A proportional to the fractional division of carriers between those in
traps and those in the band. As seen for inorganic semiconductors, such as a-Si:H
and IGZO (Chaps. 6 and 9, respectively), the increase in trap filling with
increasing gate bias leads to a greater relative increase in the free carrier density,
and, thereby, increases the field effect mobility. The same has been observed with
organic materials, whereby the mobility frequently displays a dependence on the
carrier density, and, hence, on the gate bias above threshold. This has been
empirically characterised as [44]:

VG VT c
lðVG Þ ¼ l0 ð10:2Þ
VAA
where the mobility enhancement factor, c ([0), and VAA were extracted from data
fits. A similar power dependence on gate bias has also been theoretically
demonstrated for both the MTR model and for hopping in more disordered
materials [45]. The influence of this gate-bias-dependent mobility on the TFT
current-voltage equations is discussed further in Sect. 10.5.1.
In addition, under lateral fields greater than *104 V/cm, the mobility may be
enhanced by the lateral field, and is attributed to emission barrier height lowering
by the Poole-Frenkel effect [34, 46]:
pffiffiffiffi
lðFÞ / lF0 ðTÞ exp bðTÞ F ð10:3Þ
where F is the lateral electric field, lF0 is the zero field mobility, and the parameter
b is proportional to 1/T. The mechanism of field enhancement is also temperature
dependent, and, at cryogenic temperatures, it has been found to change from the
thermally activated Poole-Frenkel process to a temperature-independent field
emission process [46].
Some, or all, of these effects are seen in many OTFTs. For instance, Eq. 10.1
describes an increase in mobility with increasing temperature, which has been
widely reported. Reference [42], for example, reviewed and confirmed this
behaviour in more than a dozen oligomer and polymer materials, including both p-
and n-channel TFTs, with room temperature carrier mobilities ranging from 0.02
to 0.42 cm2/Vs. The thermal activation energy, EA, showed an inverse correlation
with the measured mobility, and was 20–30 meV and 40–70 meV for the higher
and lower mobility devices, respectively [42]. When using the EA values to extract
l0, the l0 values were found to be quite similar, within the range 0.2–0.8 cm2/Vs,
and much closer than the wider spread in measured lFE values, which supported
the MTR model of conduction. This thermal excitation of carriers from trapping
states is a typical feature of most OTFTs, and distinguishes them from single
crystal organic devices. For example, the mobility in crystalline rubrene TFTs had
the opposite temperature dependence, and was found to decrease from *30 to
20 cm2/Vs as the temperature was increased from 150 K to 300 K, due to
increased phonon scattering [47].
In contrast to the widely observed thermal activation of the mobility in OTFTs,
its lateral electric field enhancement appears to be less frequently reported, and this
is most likely because the strength of the lateral field dependence decreases with the
amount of order in the material [34]. For instance, the Poole-Frenkel effect has been
reported at room temperature in small-grain pentacene TFTs (with a mobility of
3.3 9 10-3 cm2/Vs) [48], in P3HT TFTs (with a mobility of 4.6 9 10-2 cm2/Vs)
[47], and in TIPS-pentacene TFTs (with a mobility of 1.1 9 10-4 cm2/Vs) [47].
However, in other samples of TIPS-pentacene (with a room temperature mobility of
*1.2 cm2/Vs), the mobility was independent of the lateral field at temperatures
above *200 K [49], and, at high fields, the mobility increased with reducing
temperature. Hence, these high mobility TFTs displayed quite different temperature
and field dependent behaviour compared with their more disordered and lower
mobility counterparts.
10.3 OTFT Architecture
The device architectures which have been used in OTFTs are all non-self-aligned,
and are broadly the same as those used for a-Si:H and AOS TFTs, although there is
no sign of one particular architecture being identified as the optimum choice for
OTFTs. Whereas the a-Si:H TFTs are almost always inverted staggered, both top
and bottom gate architectures have been used in OTFTs. The three widely used
device configurations are shown in Fig. 10.8 [15], and the nomenclature used is
352 10 Organic TFTs
Fig. 10.8 OTFT architecture a bottom gate, top contact, BG/TC. b bottom gate, bottom contact,
BG/BC, and c top gate, bottom contact TG/BC. (Reprinted with permission from [15])
slightly different from the inorganic TFTs, in that they are more often labelled
bottom gate/top contact (BG/TC) rather than inverted staggered etc.
For much of the simple investigative work into the organic semiconductor layer
itself, widespread use has been made of the bottom gated structure formed by
using a doped silicon substrate as the gate electrode, and with its thermal oxide
layer as the gate dielectric. The choice of the thermal oxide was implicitly
assumed to provide a stable, high quality dielectric, such that all performance
artefacts in the TFTs could be ascribed to the organic semiconductor itself.
However, as will be seen in Sect. 10.4.3, SiO2 films caused their own artefacts
with respect to the performance of n-channel TFTs. Nevertheless, with the simple
Si/SiO2 structure, the main architectural choice is between top or bottom source/
drain contacts, together with the issues related to the deposition and patterning of
the source/drain contacts themselves. With bottom contacts, the metal can be
deposited, and then defined using standard lithography techniques. This has the
benefit that small channel lengths can be readily achieved, but the presence of the
metal can interfere with the molecular alignment of the organic channel layer in
the boundary region between the SiO2 gate dielectric layer and the metal. More-
over, due to the large difference in conductivity of the OTFT channel and the metal
contacts, the current flow from the channel to the contacts will be preferentially
through the edge of the metal electrode in these BC structures [50]. In addition,
molecular orientation discontinuities at this edge lead to contact resistance prob-
lems, as observed in pentacene TFTs with gold contacts [50, 51]. This effect is
shown schematically in Fig. 10.9a [1], where the near vertically aligned molecules
on the polar SiO2 layer lose their order on the non-polar metal surface [50], and is
responsible for the bright line at the SiO2/Au boundary in the AFM image in
Fig. 10.9b. The use of self-assembled mono-layers, SAMs, has been demonstrated
to be effective in changing surface polarity, and increasing the overall polarity of
the coated metal surface. This helps to maintain the same molecular ordering on
the metal as on the SiO2, giving the improved contrast uniformity at the SiO2/Au
10.3 OTFT Architecture 353
Fig. 10.9 a Illustration of

pentacene misalignment at
SiO2/Au edges in BC TFTs,
and AFM images of
pentacene on SiO2 and at the
edge of the Au electrodes on
b untreated surfaces, and
c surfaces coated with a
SAM. (Reproduced from [1]
with permission of John
Wiley & Sons, Inc)
edge in the AFM micrograph in Fig. 10.9c. Similar effects have been reported with
8 nm thick PMMA buffer layers [51]. The use of these layers has also been
demonstrated to reduce contact resistance effects in BC TFTs [50, 51], and,
because of the self-limiting thickness of SAMs, their direct contribution to series
resistance is small [50].
For the simple process top-contact (BG/TC) OTFTs, a particular issue is
potential damage to the organic layer both during the metal deposition, and in
patterning it using conventional lithography, where the chemicals (photoresist,
developer, etchant and solvent) may attack the organic layer [52]. To avoid these
problems, it is common practice to use shadow-mask evaporation for the top
contacts, thereby avoiding on-plate patterning. The major limitation with this
approach is the minimum achievable channel length, which is likely to be of the
order of *10–20 lm. Hence, this approach is not appropriate for devices with a
channel length of just a few microns. Without the use of SAMs in BC devices, TC
devices have usually given higher performance [50], but this difference has been
substantially removed with SAMs in BC structures. One of the residual differences
is then the additional series resistance in the BG/TC device due to the flow of
current through the vertical thickness of the film from the channel at the bottom of
the film to the contacts on the top of the film [53]. The impact on device
performance of these architectural differences in BC and TC devices is discussed
in greater detail in Sect. 10.5.3.
Another issue with the simplified device technology, irrespective of the contact
position, is the patterning of the organic film itself. The patterning is to ensure that
the channel width is sufficiently well defined to avoid the TFT performance
artefacts discussed in Sect. 5.4.2, such as high leakage currents and erroneously
large field effect mobility values. Given the problems associated with the lithog-
raphy chemicals, active device area definition has been achieved, in some
354 10 Organic TFTs
instances, by simple scratch-isolation to form a trench around the device [19], or

by shadow mask depositions [54]. However, more precise techniques have been
implemented, which use lithography, but protect the organic film from direct
chemical exposure. One example of this approach is to coat the structure with an
organic solvent barrier, such as parylene [54], define it photolithographically, and
use this film as a dry etch mask to pattern the organic layer beneath it. (This
process is discussed further in Sect. 10.4.6). It has also been argued that some
polymer semiconductors, such as poly(9,9-dioctylfluorene co-bithiophene), F8T2,
do not need patterning, as its inherent conductivity is very low [55], but, further
published work showed that island definition was used in high-quality demon-
strator displays in order to obtain low-enough off-currents in the pixel TFTs [57].
The above non-lithographic techniques, using doped silicon substrates as a BG,
have the merit of simplicity, and do not require extensive fabrication facilities, but
are clearly only suitable for basic test devices, and are not appropriate for
sophisticated demonstrator displays, nor for manufacturing. Nevertheless, these
investigations have provided the background understanding of many materials,
which have been incorporated into the range of contemporary demonstration
displays, and which employ both top and bottom gates. An important consideration
when comparing the processing of top and bottom gate structures, particularly with
solution processing, is the solubility of the first deposited layer in the second, and
this needs to be avoided by using orthogonal solvents for the different layers.
Finally, a potential advantage of a TG structure, compared with a BG, is the
automatic encapsulation of the active layer by the gate dielectric in the TG
structures, and this can be used to suppress the ambient instability shown by many
organic materials [56]. With BG structures, an encapsulant will need to be
incorporated into the process if the channel material is susceptible to air-
instabilities.
All three principal architectures have been used in demonstrators, as illustrated
by the following examples: for oligomer-based displays, one group has imple-
mented the BG/TC architecture for flexible AMOLED [4] and AMEPD [7]
displays, whilst another group has used BG/BC for an AMOLED display [10]. For
polymer materials, AMEPDs have been made with a TG/BC architecture [57], and
high mobility, n-channel TFTs have been reported with the same TG/BC archi-
tecture [25, 58]. The processing conditions used to fabricate these devices are
discussed in Sect. 10.4.6. The one architecture which has been little used is the
TG/TC, almost certainly due to carrier injection problems.
10.4 Materials and Fabrication Processes
Sections 10.4.2–10.4.5 contain details of the materials used for the different device
layers in OTFTs, in particular, the organic semiconductor, the gate dielectric and
the metallisation layers, with the focus on the more widely studied materials, and
those of current interest. The details include an overview of the deposition
10.4 Materials and Fabrication Processes 355
procedures, and, as background to this, Sect. 10.4.1 briefly reviews solution

processing techniques.
The most widely studied OS materials give p-channel TFT behaviour, and these
are presented in Sect. 10.4.2, and, for clarity, the n-channel materials are discussed
separately in Sect. 10.4.3. Examples of illustrative process flows for TFTs used in
display demonstrators are shown in Sect. 10.4.6, and less conventional printing-
based fabrication processes are reviewed in Sect. 10.4.7.
10.4.1 Solution Processing Techniques
One of the attractive features of OTFTs is the potential for low cost manufacturing
by solution processing. Several techniques have been reported to implement this,
including spin-coating, drop-casting, zone-casting, and printing, which are briefly
reviewed below.
10.4.1.1 Spin-Coating
This process is commonplace for photo-resist application in conventional photo-

lithography, and the controlling parameters to achieve a film of the required
thickness are the volume of liquid deposited on the plate, the spin speed and the
spin duration. The substrate temperature, or the post-spin anneal temperature, are
also important for organic semiconductor films in order to control the solvent
evaporation rate and the crystallisation of the film. In some cases, the poly-
crystalline film is formed directly after spin-coating on SAM coated substrates
[59], whilst for other materials post-deposition baking [60] or solvent annealing
was required [61]. Hence, the details of the required preparation procedure are
very much material dependent.
10.4.1.2 Drop-Casting
Drop casting has been demonstrated for the deposition of triisopropylsilylethynyl

pentacene (TIPS-pentacene) [62], which is a solution processable derivative of
pentacene. The drop casting technique is illustrated in Fig. 10.10a [63], in which a
solution of TIPS-pentacene in toluene was dropped from a capillary needle onto
the TFT substrate tilted at 3.5° in a sealed jar, and allowed to evaporate. The
process can also be implemented with horizontal substrates [64], but, with the
tilted substrate, the lower drop line spread down the substrate, whilst its upper
contact line remained pinned. Nucleation started at this upper line, and the
crystallite stacking axis grew down the substrate, giving ribbon-shaped grains
200–800 nm thick, 20–80 lm wide, and 200 lm–5 mm long [63].
356 10 Organic TFTs
Fig. 10.10 Semiconductor layer deposition by a drop-casting (Reprinted with permission from
[63]. Copyright (2007) American Institute of Physics). b Zone-casting (Reprinted with permission
from [65]. Copyright (2008) American Chemical Society), and (c) inkjet printing (Reprinted from
[71] with permission of IEEE)
10.4.1.3 Zone-Casting
Zone casting is a deposition technique in which the solubilized organic material is

delivered to a moving substrate through a slit-like nozzle, as illustrated in
Fig. 10.10b [65], and has been demonstrated with pentacene films derived from a
solution of pentacene in an anhydrous chlorinated solvent. Grain size and quality
was controlled by the temperature of both the liquid and the substrate, as well as by
the substrate scanning speed and the flow rate of the liquid [65]. Under conditions
optimised to obtain stationary film solidification, long grains were produced, with
a width of 10–100 lm, and a length extending to millimetres in the substrate
scanning direction. With this particular implementation using pentacene, the
choice of substrate surface (which would be the gate dielectric in a BG TFT) was
critical, and the best results were obtained with a BCB (divinylsiloxane-bis-
benzocyclobutene resin) buffer layer on top of SiO2. This gave extended grain
growth, whereas bare SiO2 resulted in cracks across the width of the grains [65].
10.4.1.4 Printing
Film deposition by printing has the major benefit of in-built pattern definition, as it
delivers material only to those locations where it is needed. Hence, conventional
lithography can be avoided, and, moreover, some of printing’s unique features,
such as sensitivity to substrate surface energy, can be exploited to deliver novel
processing schemes [66], as discussed in Sect. 10.4.7. However, printing has a
number of specific material requirements, such as the availability of appropriate
inks for the printing of conductors, semiconductors, dielectrics, and of solvents for
via hole opening. In addition, the inks need to have the correct viscosities and
evaporation rates. The printing technique also has its own inherent limitations with
respect to feature size, resolution, film thickness and plate throughput. Several
different printing processes are available, including screen, offset, gravure and
inkjet, which have quite different characteristics in terms of these parameters, and
are reviewed in references [67, 68]. Although there are examples of all-printed
TFTs and circuits using offset [69], and screen printing [70], more attention has
been directed towards inkjet printing. Compared with the other printing tech-
niques, inkjet can achieve relatively high resolution and alignment accuracy,
although throughput is lower [67]. The preferred implementation of this technique
is by piezo-electric drop-on-demand (rather than by continuous droplet delivery),
and a schematic diagram of the print head is shown in Fig. 10.10c. A signal to the
piezo crystal causes it to expand against a membrane, thereby increasing the
pressure in the ink chamber, and ejecting an ink droplet [71]. Large area systems,
with piezo-electric control of the printing nozzles, have been designed to release
droplet volumes in the range 1–10 pL, with a positional accuracy of ±5 lm [72].
Controlling the droplet size is crucial to line width control and positioning, and this
will determine one of the most critical TFT dimensions, which is its channel
length. For both top and bottom contact TFTs, this will be determined by the
separation of the source and drain contacts. For example, 1 pL nozzles have been
demonstrated to give 25 lm wide metallic tracks, using a silver ink [73], although
repeated printing was necessary to produce lines of sufficient thickness to give
an acceptable line resistance of 0.36 X/h. All other layers were also inkjet printed
in this work, including the silver gate, the PVP gate dielectric, and the TIPS-
pentacene active layer. As the dimensions of these layers were greater than those
of the source/drain contacts, a 10 pL nozzle was used for faster printing. The
finished device was a BG/BC structure with W/L = 150/11 lm, and had a field
effect mobility of 0.05 cm2/Vs [73]. For smaller channel lengths, down to 1 lm, a
sub-femtolitre inkjet printing process has been demonstrated, using an ink
containing silver nano-particles for the critically dimensioned source/drain
contacts. In this case, n- and p-channel BG/TC OTFTs were fabricated, with the
slow, but high-resolution, sub-femtolitre inkjet printing process used just for the
source/drain contacts [74]. An alternative approach has been to use surface energy
conditioning of the substrate to control its wetting properties, and to confine the
droplet within these appropriately conditioned regions [55, 66, 67]. This process is
discussed further in Sect. 10.4.7.
358 10 Organic TFTs
10.4.2 Organic Semiconductor Layers for p-Channel TFTs
10.4.2.1 Vacuum Thermal Evaporation
Pentacene is one of the most widely studied small molecule materials, largely
because of its high performance, with mobility values equalling or exceeding those
of a-Si:H. It has been extensively prepared from a purified source by vacuum
evaporation at 10-4–10-5 Pa, where the deposition rate has a strong influence on
the grain size. This is seen in Fig. 10.11a–c by the AFM images from 60 nm thick
films grown at different deposition rates on a SiO2-coated substrate. All films
showed a granular surface structure, where the largest grains were in films grown
at 0.12 nm/min, and the smallest at 12 nm/min [75]. Very similar results have also
been obtained for depositions onto PVP and PMMA coated substrates. The hole
mobility directly correlated with increasing grain size, as shown in Fig. 10.11d,
with the maximum mobility value approaching 2 cm2/Vs for the largest grains
[75]. In view of this result (and similar ones in Fig. 10.6d), evaporated pentacene
is usually deposited at a rate of *0.1 nm/min at room temperature. Higher sub-
strate temperatures may be used, but there is a tendency for the evaporated film to
contain an increasing fraction of the bulk pentacene crystalline form, rather than
the ‘thin-film’ form, which would lead to reduced TFT performance.
In spite of its large mobility, one of the limitations of pentacene is its sensitivity
to oxidising reactions, which, in unencapsulated devices, leads to device instability
during ageing under ambient conditions [13, 76]. The instability is a reduction in
hole mobility, which has been attributed to shallow state formation near the
HOMO level [77]. Oxidation is a chemical reaction which involves electron
transfer from pentacene to the oxidising agent, and the susceptibility of material to
oxidation can be reduced by changing the molecular structure to one with a greater
ionisation energy, and, hence, a deeper HOMO level [13, 76]. In principle, this
would have the effect of increasing the injection barrier height for preferred
contacts, such as gold, and, thereby, increasing the contact resistance [76].
However, whilst there is evidence for this effect [76], it does not appear to be
universally present, and examples exist in which the molecular structure has been
changed from pentacene to alternative molecules [13, 78], with a HOMO *0.4 eV
deeper, in which there was improved ambient stability, and only a small increase
in the measured contact resistance [78]. The explanation for this variability in
contact resistance between different samples is most likely due to the influence of
different interfacial dipole layers. These more stable materials, shown in
Fig. 10.12a, b, were di(phenylvinyl)anthracene (DVAnt) [78], and dinaphtho-[2,
3-b:20 ,30 -f]thieno[3,2-b]thiophene (DNTT) [13], with ionisation energies of 5.4 eV
(compared with 5.0 eV for pentacene). Both were vacuum evaporated onto sub-
strates held at 60 °C, and had similar TFT characteristics to pentacene, with hole
mobilities of 0.3 and 0.6 cm2/Vs, respectively. There is greater contemporary
interest in DNTT due to its better air stability and mobility [15].
Fig. 10.11 AFM images of pentacene grains grown at deposition rates of a 0.12 nm/min,
b 1.2 nm/min, and c 12 nm/min. The AFM image size is 5 lm 9 5 lm. d variation of mobility
with pentacene grain size (Reprinted from [75], with permission from IOP Publishing Ltd)
Although DVAnt and DNTT showed improved air-stability as a result of

increased ionisation energies (which were achieved without a significant impact
upon the effective height of the source injection barrier), there is a limit to the
extent to which the ionisation energy can be increased without injection becoming
a problem. An alternative approach has been to engineer the molecule so that it is
inherently less reactive with oxygen, without increasing its HOMO, by adding
substituents on the periphery of the molecule [79]. An example of this is vacuum-
deposited 3,9-diphenyl-peri-xanthenoxanthene (Ph-PXX) (see Fig. 10.12c, with
the H substituent). This is a derivative of peri-xanthenoxanthene, PXX [4, 79], and
has an ionisation energy of 5.1 eV (i.e. similar to pentacene), but it displayed
much better air-stability. The molecular packing structure was face-to-face, as
opposed to the herringbone arrangement of pentacene, although the long molecular
axis was similarly stacked perpendicular to the substrate surface. This material has
been used in a 4.1in wide rollable AMOLED display, where the measured hole
mobility of *0.4 cm2/Vs was *4 times larger than pentacene prepared under
identical conditions [4].
360 10 Organic TFTs
Fig. 10.12 Vacuum deposited small molecule semiconductors a di(phenylvinyl)anthracene,

(DVAnt) (Reproduced from [78] with permission of John Wiley & Sons, Inc). b Dinaphtho-[2,
3-b:2’,3’-f]thieno[3,2-b]thiophene, (DNTT) (Reproduced from [13] with permission of John Wiley
& Sons, Inc), and c peri-xanthenoxanthene derivatives, Ph-PXX and PrPh-PXX (Reprinted with
permission from [79]. Copyright (2009) American Chemical Society). Solution processed materials
d triisopropylsilylethynyl pentacene, (TIPS-pentacene) (Reprinted with permission from [63].
Copyright (2007) American Institute of Physics). e poly(4,8-dialkyl-2,6-bis(3-alkylthiophen-
2-yl)benzo[1,2-b:4,5-b’]dithiophene), R = hexyl (Reprinted with permission from [87]. Copyright
(2007) American Chemical Society), and f poly(2,5-bis(3-alkylthiophen-2-yl)thieno[3,2-b]
thiophene) (PBTTT) (Reprinted by permission from Macmillan Publishers Ltd: Nature Materials
[60], copyright (2006))
10.4.2.2 Solution Processing
As discussed previously, solution processing is regarded as a more cost effective

deposition technique than vacuum thermal evaporation, and has traditionally been
used with polymer materials. Although much of the early work on pentacene used
vacuum evaporation, there are now solubilised examples of pentacene and its
derivatives, as well as of the PXX molecule discussed above. The oligomer and
polymer materials are reviewed in the following two sub-sections.
Small Molecule Materials
Soluble pentacene pre-cursors have been used to deposit films of pentacene from
solution, by, for instance, spin-coating a soluble derivative in chloroform solution,
and then converting it into pentacene by annealing between 130 and 200 °C for 25 or
1.5 min, respectively [80]. A more extensive investigation compared spin-coating,
drop-casting and zone-casting from a solution of pentacene in an anhydrous chlo-
rinated solvent [65]. Under optimised deposition conditions, zone-casting was found
to give the best, and most continuous, long range grain structure, and the resulting
TFTs had mobilities within the range 0.4–0.7 cm2/Vs (which compared well with
vacuum deposited pentacene films) [65]. Pentacene TFTs have also been printed, but
with a non-conventional printing technique, using organic vapour jet printing, which
was solvent free, and has been compared to thermal evaporation [81]. As with other
solution based processing procedures, the performance of the printed TFTs
compared well with TFTs made using vacuum deposited films.
Another solution processable material is triisopropylsilylethynyl pentacene,
TIPS-pentacene, which is a functionalised derivative of pentacene with tri-iso-
propylsilylethynyl substituents (the molecular structure is shown in Fig. 10.12d). It
self-assembles in the solid-state with co-facial p–p stacking, rather than the
herringbone stacking of pentacene, leading to closer packing of the pentacene
molecules, and is also expected to be more oxidative stable than pentacene [83]. It
has been widely investigated for solution processing using drop-casting [63], spin-
coating [82], and, in another publication, optimised spin, dip, and drop-casting
were compared [83]. In that work, when comparing both the structural organisa-
tion of the films, and the hole mobility in the TFTs, the drop-casting was found to
give the best results, and spin-coating the worst. The spin and drop-cast films had
mobilities of 0.05–0.2 and 0.2–1.8 cm2/Vs, respectively. The mobility differences
correlated with the differences in structural organisation, and were attributed to
slower solvent release and slower film growth in the drop-cast films [83].
Compared with these high mobility TFTs, ink-jet printed TIPS-pentacene TFTs
only had a mobility of 0.05 cm2/Vs, which was associated with contact resistance
effects to the ink-jet printed silver source/drain electrodes [73].
Another derivative of the molecule peri-xanthenoxanthene, PXX (which was
discussed in Sect. 10.4.2.1 as having improved air-stability compared to pentacene)
has been developed for solution processing. This is 3,9-bis(p-propylphenyl)-peri-
xanthenoxanthene (PrPh-PXX) [7, 79], which differs from Ph-PXX by the
replacement of hydrogen by the propanyl substituent on the phenyl end groups (see
Fig. 10.12c). It has a HOMO level comparable to Ph-PXX. Films of PrPh-PXX were
prepared by spin-coating a solution of it onto a PVP gate dielectric, which was then
cured in air at 120 °C to remove the solvent and to crystallise the film. The resulting
BG/TC TFTs had a hole mobility of *0.5 cm2/Vs, and were used in a 13.3in UXGA
flexible AMEPD demonstrator [7]. The same material was also used to make a
smaller, ink-jet printed 4.8in VGA AMEPD [84].
As is apparent from the above overview, high-performance solution-processed
small molecule TFTs have been obtained using a number of different molecular
materials and preparation techniques, but these procedures mainly produced
undefined, large area films, which then needed patterning for TFT applications.
Hence, while there are some examples of ink-jet printing of small molecule
materials, it would appear that there is a need for the printing technologies to be
further developed for widespread application to these materials.
362 10 Organic TFTs
Polymer Materials
Two extensively studied regio-regular polymers are P3HT [27, 85, 86] and PQT-12
[21] (see Fig. 10.3f, g), where the PQT-12 molecule has been engineered to give
better air-stability then P3HT [21]. Solution processing has dispensed the purified
material in a suitable solvent using spin-coating, dip-coating or drop-casting. For
P3HT dissolved in chloroform, dip-coating gave the best results, and further
improvement was obtained after annealing at 160 °C for 3 min in N2, giving a
mobility of 0.11 cm2/Vs [86]. Spin-coating has, however, been successfully used
with the solvent 1,2,4-trichlorobenzene [85], which has a higher boiling point than
chloroform. This led to much slower drying of the films (compared with the
chloroform solvent), and, following vacuum annealing at 100 °C, resulted in better
controlled development of the characteristic crystalline domain structure discussed
in Sect. 10.2.3, and shown in Fig. 10.5a. The hole mobility in these devices was
0.12 cm2/Vs, which was 100 times greater than films spin-coated using a chloro-
form solvent. Similarly, PQT-12 has been applied by spin-coating, and post-
deposition annealing at 120–140 °C improved both the lamellar packing and the
hole mobility from 0.02–0.05 to 0.07–0.12 cm2/Vs [21].
Other solution processable polymers have been reported, which have improved
behaviour compared with P3HT and PQT-12, such as poly(4,8-dialkyl-2,6-bis
(3-alkylthiophen-2-yl)benzo[1,2-b:4,5-b0 ]dithiophene) [87] (see Fig. 10.12e for its
molecular formula). This was spin-coated from a 1,2-dichlorobenzene solution
onto an OTS coated substrate, where it directly formed a well organised lamellar
structure without post-deposition annealing. It solidified into *1 9 1 lm
domains, giving a hole mobility of *0.2 cm2/Vs, and displayed improved air-
stability compared with P3HT and PQT-12. Another polymer was poly(2,5-bis
(3-alkylthiophen-2-yl)thieno[3,2-b]thiophene) (PBTTT) (see Fig. 10.12f), which
was a higher mobility polymer, with better air-stability [60]. The films were
formed by spin-coating a solution of PBTTT in 1,2-dichlorobenzene, and
annealing at 120–160 °C for 10–15 min. With the C14 alkyl substituents, this gave
200 nm grains, and a hole mobility of 0.6 cm2/Vs. This mobility is larger than
usually found with polymers, and is comparable to the better small molecule
materials. PBTTT has a liquid crystal phase, which was established during the
post-deposition annealing, and the high mobility was attributed to the organised
assembly of the molecular chains in the LC phase, and their retention, as the
structure crystallised [60].
As discussed in the previous section, the preferred deposition technique for low
cost TFT fabrication is by printing. Amongst the polymers, P3HT has been deposited
by inkjet printing, and TG/BC TFTs prepared by this process compared well with
spin-coated TFTs, where the hole mobilities were *0.08 and *0.09 cm2/Vs,
respectively [88]. An inkjet printed proprietary p-channel polymer had better TFT
characteristics, with a mobility of *0.25 cm2/Vs, whilst the mobility of its
spin-coated control was 0.6 cm2/Vs [88]. Although PQT-12 has also been ink-jet
printed, its performance was poor, with a mobility of 1 9 10-4 cm2/Vs, which was
attributed to non-optimised preparation conditions [89]. The polymer poly(9,
9-dioctylfluoreneco-bithiophene), F8T2 [55], and other polyfluorene-based mate-

rials [57, 90], have been inkjet printed in demonstration displays. These materials
had low mobilities of *0.03 cm2/Vs [57], but were, nevertheless, good enough for
AMEPDs.
10.4.3 Organic Semiconductor Layers for n-Channel TFTs
The development of high-performance OTFTs has traditionally been much easier

with p-channel TFTs than with n-channel TFTs. A 2004 review paper [35] iden-
tified several issues with the operation of n-channel TFTs, including the high
injection barrier with large work function metals and the ambient instability of low
work function metals, low electron mobilities (in many cases 0.1 cm2/Vs), the
instability of these devices due to electron trap formation in the presence of
oxygen and water vapour, and high threshold voltages with some high mobility
materials [35]. Indeed, the question was raised whether there were fundamental
reasons preventing the carrier mobility in n-channel TFTs from matching that in
p-channel TFTs [36]. As mentioned above, a simple practical issue was the
mismatch of the work function of the commonly used gold source/drain material
with the LUMO level of the organic semiconductor. To resolve this injection
problem, the LUMO had to be deeper, or a lower work function metal, such as Ca,
had to be used. Neither approach solved the problem of poor n-channel perfor-
mance. However, as discussed in Sect. 10.3, much of the basic materials work on
OTFTs used a thermally oxidised Si substrate as a combined gate dielectric and
gate electrode, and one of the key observations with n-channel operation was that
hydroxyl molecules on the surface of the SiO2 gate dielectric formed silanol,
SiOH, electron traps at the dielectric/channel interface [36]. The trapping mech-
anism is shown in Fig. 10.13a, and the trapped electrons result in negatively
charged defects, which compensate the gate field, and can result in very high
threshold voltages. Although the silanol coverage of high quality SiO2 surfaces is
very small, at \10 % of a monolayer, this equates to a potential areal trap density
of 3 to 7 9 1013 cm-2, which is more than an order of magnitude greater than the
electron densities induced by the gate bias. Hence, with an SiO2 gate dielectric, a
large fraction of the induced electrons were trapped at interface defects, and
similar trapping effects were also observed with some other commonly used
dielectrics including poly(vinyl phenol), PVP, and polyimide, both of which
contain hydroxyl groups [36]. Using the poly-fluorene based semiconductor
poly(9,9-dioctylfluorene-alt-benzothiadiazole), F8BT, the influence of the gate
dielectric is shown in Fig. 10.13b. There was no TFT behaviour with the F8BT
directly deposited onto a bare SiO2 layer, but, after spin-coating the SiO2 with a
layer of polyethylene, good TFT operation was displayed, and the on-current
increased by several orders of magnitude. Intermediate results were obtained with
several SAMs, none of which was able to completely eliminate electron trapping at
SiOH groups. These devices showed initial n-channel behaviour, but this was
364 10 Organic TFTs
Fig. 10.13 a Silanol electron

trapping mechanism at the
semiconductor/SiO2
interface, and b transfer
characteristics of F8BT
n-channel TFTs with different
gate dielectrics and SAMs
(OTS, DTS and HMDS).
(Reprinted by permission
from Macmillan Publishers
Ltd: Nature [36], copyright
(2005))
quenched by strong air-instability effects [36]. It was also noted that the sensitivity
to silanol trapping reduced with increasing depth of the LUMO level.
Although the use of an appropriate dielectric layer can suppress electron traps
associated with silanol, this will not necessarily solve the problem of air-instability.
The material’s susceptibility to oxidation reactions is enhanced by the presence of
channel electrons at the relatively shallow LUMO level, and these are able to react
with oxygen and moisture in the ambient. Ambient instability in n-channel TFTs was
reviewed in a 2007 paper [91], and, to address this instability, there is a requirement
for either a dense molecular structure to suppress oxidant ingress [35, 91], or a deep
enough LUMO to suppress the oxidation reactions [91, 92]. For the former,
fluorinated side-group substituents may act as a kinetic barrier to the diffusion of
oxidising species [35], and perfluorinated copper phthalocyanine, F16CuPc, (see
Fig. 10.14a) is a widely studied, air-stable n-channel material [35, 91, 93]. For
LUMO-controlled instability, a minimum LUMO value of 4 eV has been identified
[92], and this is consistent with other reports of air-stable n-channel TFTs using
materials with LUMO values of 4.8 eV and 5.05 eV [93, 94]. It is worth noting that
F16CuPc, with a LUMO of 4.8 eV, possesses both attributes, but has a modest
mobility of *0.08 cm2/Vs [93].
Fig. 10.14 a F16CuPc

American Chemical Society).
b P(NDI2OD-T2). c transfer
characteristics of P(NDI2OD-
T2) TG/BC TFTs (W/
L = 1000/50) with different
gate dielectrics (Cytop–
(poly(perfluoroalkenylvinyl
ether), k = 2.1), PTBS—
(poly(t-butylstyrene),
k = 2.4), PS—(polystyrene,
k = 2.5), ActivInk D2200—
(polyolefin-polyacrylate,
k = 3.2) and PMMA—
(poly(methylmethacrylate),
k = 3.6), and d output
characteristics of P(NDI2OD-
T2) TFT with a PMMA gate
dielectric. (Reprinted by
permission from Macmillan
Publishers Ltd: Nature [25],
copyright (2009))
Many of the air-stable materials, with electron mobility values in the range
*0.1–0.6 cm2/Vs, are oligomers, and are most easily deposited by vacuum
evaporation [91, 93, 94], whereas an ideal n-channel material would be solution
processable [37]. Such materials, which are air-stable, and have a mobility
C0.1 cm2/Vs, appear to be few in number [95], but, a promising candidate is the
polymer poly{[N,N9-bis(2-octyldodecyl)-naphthalene-1,4,5,8-bis(dicarboximide)-
2,6-diyl]-alt-5,59-(2,29-bithiophene)}, (P(NDI2OD-T2). This has a LUMO level
of *4.0 eV, an optical band-gap of 1.45 eV, and its structure is shown in
Fig. 10.14b [25]. The material is soluble in xylene and dichlorobenzene, and can
be deposited by spin-coating, gravure printing or inkjet printing. It has been
demonstrated in TG/BC structures, with a range of spin-coated polymer
dielectrics, giving air-stable, n-channel TFTs with mobilities in the range
0.1–0.85 cm2/Vs [25]. The highest mobilities of 0.65–0.85 cm2/Vs were with a
proprietary polyolefin-polyacrylate dielectric, D2200, but a PMMA dielectric gave
the next highest mobility values in the range 0.2–0.45 cm2/Vs. The transfer
characteristics of devices made with P(NDI2OD-T2), and with a range of different
spin-coated organic gate dielectrics, are shown in Fig. 10.14c. These have broadly
similar characteristics, yielding high mobilities, as well as low threshold voltages,
and on:off current ratios of 6–7 orders of magnitude [25]. In addition, the TFT
output characteristics in Fig. 10.14d displayed good operation in both the linear
and saturation regimes. Further commercial developments of P(NDI2OD-T2) have
yielded a proprietary material for solution processable, air-stable n-channel TFTs
366 10 Organic TFTs
with a mobility of *3.0 cm2/Vs [96]. This has been demonstrated in spin-coated
and inkjet printed TG/BC TFTs, using a proprietary gate dielectric, and fabricated
on PET and PEN substrates. These devices displayed minimal degradation during
1400 h of accelerated bias-stress stability measurements, and this was assessed to
be equivalent to an operating life of [10,000 h under typical AMOLED operating
conditions [96].
One of the reasons for interest in n-channel TFTs is to enable the fabrication of
low-power-dissipation complementary logic circuitry [25, 35, 37], and this was
demonstrated with p- and n-channel TFT inverter circuits using P3HT and
P(NDI2OD-T2) layers, respectively. These devices, with a common PMMA gate
dielectric layer and Au contacts, showed inverter switching gains of 25 and 60 at
supply voltages of 20 and 40 V, respectively [25]. Whilst there are many
demonstrations of complementary inverter circuitry (for instance [11, 70, 88, 89,
94]), it has been argued that a simplification to the fabrication process may be
achieved with a single ambipolar material, rather than with two separate materials
for n-channel and p-channel operation. An important proviso for the ambipolar
material is that it can deliver high performance, well-matched p- and n-channel
operation using the same injecting electrodes [97]. (This necessarily requires small
bandgap material, so that the injection barriers are low enough for each carrier
type, and, equally, the material needs good transport properties for both holes and
electrons). Ambipolar material would also ease potential compatibility issues with
the dielectrics and contact materials for the two device types, as well as simpli-
fying the definition stages for the device layers [97]. As with the more direct
complementary devices, there are many reports of ambipolar material (for instance
[42, 92, 95, 97–99]).
An example of ambipolar complementary device behaviour is shown in
Fig. 10.15a–d [97]. The material for these devices was PSeDPPBT, which is a low
bandgap polymer based upon a diketopyrrolopyrrole (DPP) core flanked with two
selenophene rings (thiophene with the S replaced by Se) and a benzothiadiazole
(BT) monomer. The device architecture was BC/TG with Au contacts. Both the
PSeDPPBT, and the PMMA gate dielectric, were deposited by spin-coating,
and, following annealing at 200 °C, saturation hole and electron mobilities of
0.46 cm2/Vs and 0.84 cm2/Vs, respectively, were obtained [97]. Although the
material had a bandgap of 1.05 eV, current crowding, due to injection problems
from the Au electrodes, is visible in the output characteristics at low drain biases.
Figures 10.15a, c show the p-channel behaviour (at negative gate and drain biases),
and the complementary n-channel behaviour with positive biases is seen in
Fig. 10.15b, d. However, conventional p-channel and n-channel behaviour is only
seen over certain voltage ranges (e.g. large negative gate and drain biases for p-
channel, and the opposite conditions for n-channel). Outside these ranges, and in
contrast to the usual unipolar devices, there are anomalous features in the transfer
and output characteristics. For instance, for the ‘p-channel’ operation, at large
negative drain bias and small positive gate bias, there is an increasing minimum off-
current with increasing drain bias, which is characteristic of ambipolar device
behaviour [98]. Over this voltage range, the surface adjacent to the negatively
biased ‘drain’ terminal becomes electron accumulated, with the terminal acting as
the ‘source’ for n-channel conduction. Hence, the anomalous appearance of the
characteristics occurs when the device is ceasing to act as a unipolar device, and is,
instead, supporting ambipolar conduction. This is also responsible for the absence
of saturation in the output characteristics at low gate biases in Fig. 10.15c, d. The
potentially poor off-currents adversely affect the power dissipation in inverter
circuits (compared with conventional complementary TFT inverters), and ambi-
polar inverters are also unable to fully switch the output voltage between the supply
voltage and ground [97, 98]. Hence, the trade-off with ambipolar inverters is
potentially simpler device processing, but with reduced inverter performance.
Fig. 10.15 Transfer and ouput characteristics obtained from a PSeDPPBT ambipolar device.
a and c p-channel, and b and d n-channel behaviour at negative and positive gate and drain
biases, respectively. (Reproduced from [97] with permission of John Wiley & Sons, Inc)
368 10 Organic TFTs
Nevertheless, there is continuing interest in this approach, and PSeDPPBT inverters

(using pairs of self-aligned ambipolar TFTs) had gains of [30 and [40 at supply
voltages of 10 and 20 V, respectively, and 3-stage ring oscillators operated at
182 kHz, with a small delay/stage of 0.91 ls at a supply voltage of 50 V [97].
10.4.4 Gate Dielectric
As is clear from the preceding sections, much of the simple investigative materials
work has been carried out using thermally grown SiO2 (dielectric constant,
k = 3.9) as the gate dielectric, but this material is not relevant to low cost glass
and polymer substrates, which is where the main application of OTFTs is expected
to be. Moreover, as discussed in Sect. 10.4.3, SiO2 also gave unacceptably poor
n-channel TFT performance [36]. Hence, there was a clear need for alternative
dielectrics to SiO2. Nevertheless, the basic materials research using this dielectric
identified the critical importance of the semiconductor/dielectric interface [100],
and the key role of SAMs in tailoring the interface to promote the required
ordering of the semiconductor on both the dielectric [101], and on the metal
contacts in BC structures.
The dielectric material requirements for high performance OTFTs are similar to
those of other TFTs, namely: a very good dielectric/semiconductor interface
(which generally requires a smooth surface), low leakage currents and low pin-
hole densities, and a high dielectric breakdown field in excess of several MV/cm
[102]. Three broad classes of dielectric have been investigated: (a) inorganic
layers, such as SiNx and SiOx, as well as some other inorganic materials chosen for
their large dielectric constants (high-k materials), (b) organic layers, many of
which have dielectric constants, k, in the range 2.0–4.5 [101], and (c) self-
assembled monolayers [101–103]. As mentioned previously, SAMs are also
extensively used in combination with many other dielectric layers, and two of the
more common SAMs are octadecyltrichorosilane (OTS) and hexamethyldisilazane
(HMDS).
There are other desirable TFT attributes, which can be enhanced by the
appropriate choice of gate dielectric, such as low voltage operation, and solution
processing. However, the particular material choices may be constrained in order
to mitigate their effect upon other parameters. For instance, low voltage operation
would, in principle, require a thin dielectric layer, but this may be incompatible
with a low leakage current and low pin-hole density, and, indeed, the thickness
range of organic dielectrics is *300 nm, or more, to minimise gate leakage [102].
On the other hand, SAMs are especially thin, and have been demonstrated as gate
dielectrics of potential interest [102–104]. An alternative approach to low voltage
operation is the use of high-k materials, which would favour inorganic dielectrics,
but these would not necessarily be compatible with solution processing. In addi-
tion, the higher dielectric constant insulators are more polar, and have been found
to degrade TFT performance due to increased carrier trapping in disordered
interfacial regions [101, 105, 106]. For solution processed materials, it is also
essential to use orthogonal solvents with the different layers, so that the deposition
of the organic semiconductor does not dissolve the already-deposited dielectric
layer in BG structures, and vice versa in TG structures. The three dielectric options
listed above are briefly reviewed in the following sub-sections.
10.4.4.1 Inorganic Dielectrics
High performance OTFTs, using both solution-processed PQT-12 and pentacene,

have been demonstrated using PECVD SiOx (k * 4.0) and SiNx (k = 4.5–7.2) as
the gate dielectric in combination with the SAMs OTS or b-phenethyltrichloros-
ilane (PTS) [107, 108]. It was argued that the extensive experience of these
dielectrics within the inorganic TFT industry, including the availability of large
area deposition equipment, makes them a low-risk route to manufacturing.
However, the well-established dielectric deposition processes are at *300 °C, and
pentacene TFTs with a 300 °C SiNx dielectric had comparable performance to
devices with a thermal SiO2 gate dielectric [108]. In contrast, PQT-12 TFTs
processed on plastic substrates, using SiNx deposited at 150 °C and SiOx at
180 °C, showed a smaller carrier mobility compared with higher temperature
depositions of these materials [107]. (Details of the PECVD deposition processes
for SiNx and SiOx can be found in Sects. 5.5.3 and 7.3.1, respectively).
Other inorganic insulators have been grown by oxidation of the gate metals Al
and Ta, forming AlOx [109] and Ta2O5 [10, 110], respectively. The AlOx was
grown by plasma oxidation of the Al gate, and coated with a SAM, to produce
complementary TFT inverter circuits (using pentacene and F16CuPc) on a poly-
imide substrate [109]. The Ta2O5 (k = 24) was formed by anodization of the Ta
gate, and was then coated with an HMDS SAM before vacuum deposition of
pentacene. These TFTs were used as drive transistors in an AMOLED display on a
PEN substrate [110].
Some other high-k dielectrics, which have been used in low-voltage pentacene
TFTs, were HfO2 (k = 14.9) [111], and HfLaO (k = 10.4) [112]. The 50 nm thick
HfO2 films were deposited by atomic layer deposition at 200 °C, and its surface
was modified with phosphonic acid to passivate the polar surface groups, and to
improve the pentacene morphology. The resulting TFTs displayed low voltage
operation with a sub-threshold slope of 120 mV/dec, a threshold voltage of
-0.4 V, and a carrier mobility of 0.4 cm2/Vs [111]. In contrast, 40 nm thick
HfLaO films, deposited by RF sputtering, required annealing in NH3 at 400 °C,
and did not display such low voltage operation, with the sub-threshold slope and
threshold voltage being 0.26 V/dec and -2.6 V, respectively [112]. It is clear that
the differences in the TFT parameter values between these two dielectrics cannot
be simply ascribed to differences in film thickness and dielectric constant. The
HfLaO films did not have any surface modification beyond the NH3 exposure, and
it is possible that the polar dielectric was degrading the interface, as has been
reported with other polar dielectric films [105, 106]. Indeed, by deliberately
370 10 Organic TFTs
Fig. 10.16 Field effect

mobility in P3HT TFTs as a
function of the dielectric
constant of the tri-layer
sol-gel silica gate dielectric.
Physics)
changing the dielectric constant, k, over the range 7–10, in tri-layer sol-gel silica
gate dielectric films, the carrier mobility in both pentacene and P3HT BG/TC
TFTs was found to reduce as the dielectric constant was increased, as shown in
Fig. 10.16 [106]. In these samples, the dielectric constant was modified by
exposing the first and second surfaces in the tri-layer films to an argon or air
plasma, before the deposition of the next sol-gel layer. This changed the
concentration of polar OH groups in the films, and it was concluded that the
polarisability of the film influenced the carrier flow in the TFT channels [106].
Parallels were drawn between this result and comparable k-dependent mobility
values seen with organic dielectrics [105], as discussed in the next section.
10.4.4.2 Organic Dielectrics
The interest in organic insulators stems, in part, from their solution processability,
and the characteristics of a wide range of organic dielectric films have been
extensively reviewed [101, 102, 105]. Some of the more widely investigated
materials (with dielectric constants comparable to, or greater than, SiO2) are PVP
(polyvinylphenol, k = 4.5), PMMA (polymethylmethacrylate, k = 3.5), and PVA
(polyvinylalcohol, k = 7.8), and amongst the lower-k materials are Cytop
(k = 2.1), and PPCB (polypropylene-co-butene, k = 2.3). The chemical structures
of these materials are shown in Fig. 10.17a–e. They are usually solution processed,
by, for instance, spin-coating, and then annealed to remove the solvent and also to
cross-link the film. When the low-k dielectrics were used with several different
TFT materials, including P3HT, F8T2, and the low-mobility amorphous polytri-
arylamine semiconductor (PTAA), they were found to result in higher carrier
mobilities than observed with PVP, PMMA, PVA and other high-k dielectrics
[101, 105]. This is illustrated for PTAA by the dependence of mobility on
dielectric constant, in Fig. 10.18a, and by the temperature dependence of the
mobilities in Fig. 10.18b [105]. In Fig. 10.18b, the field effect mobility for
different dielectrics is compared with the bulk mobility (evaluated by time of
flight, TOF, measurements), and the reducing absolute values of field effect
Fig. 10.17 Chemical structures of common organic gate dielectrics a PVP, b PMMA, c PVA,
d Cytop, and e PPCB. (Reprinted with permission from [101]. Copyright (2004) American
Chemical Society)
mobility, and their increasing activation energy, with increasing dielectric constant
were attributed to increased carrier localisation at the semiconductor/dielectric
interface. The model for this is shown in Fig. 10.18c, with a Gaussian DOS
distribution in the semiconductor. The effect of the random orientation of dipoles
in the high-k, polar dielectrics was to increase the energy fluctuations, and, hence,
the disorder, in the semiconductor surface. This was characterised by a broadening
of the DOS at the interface, leading to increased carrier trapping. It was also
suggested that the use of SAMs not only improved the packing order of the organic
semiconductor molecules in BG structures, but also partially screened the effects
of the polar dielectric, thereby making a further contribution to increased carrier
mobility [105].
The influence of the organic dielectric on pentacene film growth, and TFT
performance, has been studied with the dielectrics PVP, PVA, poly(2-vinylnaph-
thalene) (PVN), polystyrene (PS), and poly(4-methylstyrene) (PMS) [100]. The
dielectrics were selected for their differing glass transition temperatures, TG, and
surface contact angles. Polymer dielectric thicknesses of 320–360 nm were used in
BG/TC structures by spin-coating them onto thermally oxidised silicon, and then
vacuum baking them at 80 °C. (Uncoated SiO2 layers, referred to as ‘bare’, were
used as control samples). For pentacene evaporated onto substrates held near room
temperature, the gate dielectric was found to influence the pentacene grain size, as
shown in Fig. 10.19, but this varying grain size had little impact upon device
performance and carrier mobility. The largest grains were on the PVP and ‘bare’
surfaces, which had the lowest contact angles, and were the most hydrophilic. This
permitted greater movement of the pentacene molecules on the dielectric surface,
resulting in the larger grains. The worst samples were grown on the PVA film,
372 10 Organic TFTs
Fig. 10.18 a Carrier mobility measured in PTAA TFTs as a function of the dielectric constant of
the gate dielectric. b Temperature dependence of the mobility for different gate dielectrics, and
c schematic illustration of the effects of disorder induced in the DOS at the semiconductor surface
by a polar dielectric. (Reproduced from [105] with permission of John Wiley & Sons, Inc)
which has a bulk TG of 49 °C, whereas the others were [100 °C [100]. The small
grain size with PVA was attributed to a surface TG smaller than the bulk value,
which perturbed the film during pentacene deposition, resulting in a rough inter-
face. For the other films, it was concluded that for grain sizes above *0.8 lm,
grain boundary trapping had only a secondary effect on charge transport [100].
It was also noted that, in these samples, the dielectric constant had a smaller
impact upon the carrier mobility than reported in other work. Indeed, there are
examples in the literature, in which the trends with k were the opposite from those
discussed above [101, 102]. Hence, the detailed understanding of the role of the
dielectric in OTFTs is still incomplete.
The largest grains in Fig. 10.19 were grown on a PVP dielectric, and this
material, which can be inkjet printed [73], as well as spin-coated, is widely used as
Fig. 10.19 Influence of

organic gate dielectric on
pentacene grain size and
carrier mobility. (Reproduced
from [100] with permission
of John Wiley & Sons, Inc)
a polymer dielectric in TFT studies. It has also been used in demonstrator displays
on flexible substrates, in which it was mixed with OTS to reduce the cross-linking
anneal temperature [4, 7].
10.4.4.3 Self-Assembled Monolayers
The use of SAMs as gate dielectrics has been reviewed in Ref. [102], in which
layers *3 nm thick have displayed good dielectric properties. For instance,
2.8 nm thick OTS SAMs gave low leakage currents of 10-8 A/cm2 at 5 MV/cm,
and breakdown fields of 9–12 MV/cm. Improved TFT performance was achieved
by adding a phenoxy end group to OTS, forming PhO-OTS (k = 2.5), which
displayed closer molecular packing than OTS, and facilitated the fabrication of
BG/TC pentacene TFTs [113]. (The OTS SAM alone had not functioned as an
effective gate insulator with pentacene due to their intermixing). With the 2.5 nm
PhO-OTS SAM insulator, the pentacene TFTs had a sub-threshold slope of
100 mV/dec, a threshold voltage of -1.3 V, and a mobility of 1 cm2/Vs. Other
SAM insulators of interest have been alkylphosphonic acid molecules, in which
the length of the alkyl chain CnH2n+1 was optimised for minimum gate leakage
currents and good pentacene TFT performance, by using n values of 14–16 [104].
10.4.5 Metals
The metals used in TFTs are for the gate, and the source/drain contacts, and for the
bus bars in displays and circuits, and, although the latter are usually the same metals
as used in the TFTs, they can be different. Where the structure has a BG, it is often
convenient to use evaporation or sputtering to deposit the gate metal, and lithog-
raphy to define it. For TG structures, with a polymer gate dielectric, this may be less
practical, and solution processing is one option. For the source/drain contacts, gold
374 10 Organic TFTs
is widely used in both experimental TFTs, as well as in demonstrator devices, but its
deposition is limited to vacuum evaporation. For full solution-processing of
devices, other materials have been investigated for both the TG metal and source/
drain contacts. These have included inkjet printing of dispersions of inorganic
metallic nano-particles [90], such as silver, which had a resistivity 8 lXcm, and
repeat printing was used to build up 200 nm thick layers, with a sheet resistance of
0.4 X/h [70]. However, although Ag has very good conductivity, its work function
is quite low at 4.7 eV [114], and contact resistance effects have been noted in
printed TIPS-pentacene TFTs with Ag contacts [70, 114]. The use of thiol-based
SAMs has been demonstrated to reduce the injection barrier at the Ag/TIPS-
pentacene interface, and this improved the mobility from 0.01–0.04 cm2/Vs to
0.04–0.17 cm2/Vs for bare and treated Ag electrodes, respectively [114]. Source/
drain contacts have also been inkjet printed using poly(ethylenedioxythiophene)
doped with poly(styrene sulfonic acid) (PEDOT/PSS) [90]. This has low contact
resistance into many TFT materials, but its large resistivity of 0.01 Xcm makes it
unsuitable for bus bars and gate lines [90].
10.4.6 Process Flow
Given the variety of materials used for both the organic semiconductor and for the
gate dielectric, plus the various deposition options, there is not, at this moment, a
consensus on the preferred choice of components, and, hence, there is not a typical
device processing schedule. In view of this, the following two examples are simply
illustrative of processes which have been implemented with small molecule and
polymer semiconductors, and should not be regarded as representative of the
industry as a whole.
The first example is of a flexible AMOLED display, addressed by BG/BC
pentacene TFTs on a PEN substrate [10, 110]. Some of the key process stages in
the process are illustrated in Fig. 10.20, and are listed below:
(i) RF sputter deposition of SiO2 and SiNx films onto the PEN substrate as barrier
and adhesion layers,
(ii) RF sputter deposition of Ta, and its definition into gate electrodes by reactive
ion etching,
(iii) Anodisation of Ta gate to form 170–200 nm thick Ta2O5 gate dielectric,
(iv) Deposition and definition of Cr/Au source/drain contacts, and ITO pixel
electrode,
(v) Deposition of HMDS SAM,
(vi) Vacuum deposition of 100 nm of pentacene,
(vii) Room temperature CVD deposition of parylene (2 lm) and sputter depo-
sition of SiO2 (50 nm),
(viii) Photoresist application and photolithographic patterning,
Fig. 10.20 Fabrication

stages of pentacene TFTs for
an AMOLED display.
permission of SID)
(ix) CF4 plasma etching of SiO2, which then acted as the etch mask for oxygen
plasma etching of the parylene and pentacene layers.
(x) Final encapsulation with parylene and a photosensitive polymer.
These TFTs had a mobility of 0.05–0.1 cm2/Vs, an on:off ratio of [106, and a
threshold voltage of 12 V.
The second example is of a flexible AMEPD using TG/BC unpatterned polymer
TFTs, and the pixel/device cross section is shown in Fig. 10.21 [115]. The
processing steps were based upon direct-write solution processing and laser
patterning, without the need for mask alignment. The main fabrication stages were:
(i) PET substrate planarization,
(ii) Source/drain metal deposition (unspecified procedure),
(iii) Unpatterned polyfluorene-based semiconductor deposition (unspecified
procedure),
(iv) Spin-coat polymer dielectric layer,
(v) Inkjet print Ag gate electrode,
(vi) Deposition of organic dielectric passivation layer (10 lm),
(vii) Via hole opening by laser ablation,
(viii) Inkjet print PEDOT/PSS pixel electrode.
376 10 Organic TFTs
Fig. 10.21 AMEPD pixel/

TFT cross section. (Reprinted
from [115] with permission
of SID)
These TFTs had a mobility of 0.01 cm2/Vs, an on:off ratio of 104, and a
threshold voltage of -5 to -10 V. The poor on:off ratio was attributed to the
undefined channel layer, and, in later work, this ratio was improved to 105–106 by
patterning the semiconductor [57].
10.4.7 Novel Processing
One of the challenges with inkjet printing is achieving fine lines and short channel
lengths, and surface energy conditioning procedures have been demonstrated to
address these issues [90], and even to facilitate sub-100 nm channel lengths [66].
The key feature in this approach was to change the wetting-angle of the sample
surface, so that hydrophobic and hydrophilic regions were defined, which were
then used to localise water-based conducting polymer ink drops. This procedure is
shown in Fig. 10.22a, in which the surface of a hydrophobic layer was locally
exposed to a laser beam to create hydrophilic regions. Because of the high reso-
lution and positioning accuracy of the laser, gaps of a few microns were defined
between the exposed regions, which became the TFT channel regions. When the
water-based PEDOT/PSS inks were printed over these areas, the inks were rejected
from the hydrophobic regions and retained within the hydrophilic regions. These
regions formed the source and drain contacts, with a separation down to 10 lm.
Then a polymer semiconductor, such as F8T2, was inkjet printed over these
contacts, followed by solution deposition of a polymer gate dielectric, and inkjet
printing of a 45 lm wide silver gate electrode, with a positional accuracy of
±20 lm [90]. The devices with L = 10 lm had a mobility of 8 9 10-3 cm2/Vs
and an on:off ratio of 105, which was comparable to more conventionally fabri-
cated devices with photolithographically defined gold electrodes. Alternative
surface conditioning procedures were also described, such as the use of an electron
Fig. 10.22 a TG/BC TFT printing, using substrate surface patterning to confine the water-based
droplets of PEDOT-PSS forming the source and drain regions (Reprinted from [90]. Copyright Ó
Materials Research Society 2003), and b sub-100 nm channel length TFTs using self-separating
source/drain regions, formed by surface modification of the first droplet. (Reproduced from [66]
with permission of John Wiley & Sons, Inc)
beam to write a fine exposure pattern in a PMMA photoresist film, which had been
deposited on a SiO2 layer. These PMMA patterned areas were ultimately used as a
template to localise the subsequently printed PEDOT/PSS source/drain regions,
giving sub-micron channel lengths [90].
A further extension of this procedure is shown in Fig. 10.22b, and this avoided
the direct surface energy patterning of the substrate itself [66]. In this case, the
surface of the first droplet was modified to make it hydrophobic, such that, when
the second droplet was released onto it, it slid off leaving a small gap between the
two. This self-alignment of the source and drain areas did not require precise
alignment of the second droplet. The surface of the first PEDOT/PPS ink droplet
was made hydrophobic by either exposing it to a CF4 plasma, which fluorinated it,
and, at the same time, the exposed glass substrate surface was made hydrophilic by
the etching action of the plasma. The alternative procedure was to have a suitable
surfactant within the ink, which segregated to the surface on drying. Device
processing was completed with the addition of solution processed layers of F8T2, a
120–130 nm thick PMMA dielectric, and the inkjet printing of the top gate. The
channel length in the published device was estimated to be *60 nm, with an
on:off ratio of 104 for Vd \ -5 V [66]. The devices had significant contact
resistance, and displayed other short channel effects, which is understandable in a
device with such a short channel and thick gate dielectric. (More detailed infor-
mation on short channel effects, and their amelioration, can be found in Sect. 8.7).
378 10 Organic TFTs
10.5 OTFT Characteristics
10.5.1 General
A representative set of transfer and output characteristics for a Ph-PXX p-channel

TFT, with L = 50 lm, is shown in Fig. 10.23a, b, respectively [79]. Similar
curves have been published for other high quality p-channel OTFTs, and
Fig. 10.14c, d show the equivalent set of characteristics for a P(NDI2OD-T2)
n-channel TFT [25]. The Ph-PXX transfer characteristics measured in saturation,
at a drain bias of -40 V, are shown in Fig. 10.23a, and the right hand axis is a plot
of the square root of drain current versus gate bias. This shows reasonable line-
arity, and, as with inorganic TFTs, the simple MOSFET equations developed in
Chap. 3 were used to extract the performance parameters of OTFTs from their
transfer characteristics. In the saturation regime, the field effect mobility is
calculated from the slope of the HId - VG plot using Eq. 3.18, i.e.
pffiffiffiffi2
2L d Id
lFE ¼ ð10:4Þ
WCi dVG
and the threshold voltage is given by the extrapolated intersection of this curve
with the VG axis. In the example shown, the mobility was *0.42 cm2/Vs, and the
threshold voltage was *0 V [79]. The output characteristics in Fig. 10.23b show
good saturation and negligible current crowding at zero drain bias, which is
indicative of good injecting contacts. With these contacts, it would be possible to
extract the mobility at low drain bias in the linear regime, however, as injection
problems are frequently encountered in OTFTs, the mobility is more commonly
extracted in the saturation regime.
Examples of good and bad injecting contacts in TIPS-pentacene TFTs, with Ag
source/drain contacts, are shown in Fig. 10.23c, d, respectively [114]. The
difference between the two TFTs was in the surface treatment of the Ag electrodes
by different thiophenol-based SAMs. These changed the effective work function of
the Ag, which, in the good case, was a much closer match to the HOMO of the
TIPS-pentacene, and gave the linear Id - Vd curves at low drain bias in
Fig. 10.23c. The key characteristic of the bad contact, in Fig. 10.23d, was the non-
linearity of the curves near zero drain bias, which is referred to as current
crowding, and would preclude an assessment of field effect mobility in the linear
regime. The series resistance associated with this injection barrier was also
responsible for the order of magnitude reduction in the current in the saturation
regime compared with the good sample. It is worth noting that the impact of series
resistance at the injecting contact is also a function of its magnitude relative to the
channel resistance, and a good injecting contact, at low gate bias and long channel
length, can become a poor one as the channel resistance is decreased by either
reducing the channel length and/or increasing the gate bias [53, 117]. Contact
effects are discussed further in Sect. 10.5.2.
10.5 OTFT Characteristics 379
Fig. 10.23 a Transfer characteristics, and b output characteristics of a Ph-PXX TFT (Reprinted
with permission from [79]. Copyright (2009) American Chemical Society). Output characteristics
of TIPS-pentacene TFTs, with Ag S/D contacts covered with different thiophenol-based SAMs,
showing c good, and d poor injecting contacts. (Reprinted with permission from [114]. Copyright
(2008) American Institute of Physics)
Table 10.1 summarises the parameter values which have been extracted from a
number of TFT materials discussed in the preceding sections. The mobility values
and on:off current ratios are usually quoted in publications as simple figures of
merit, and, where available, the threshold voltage, VT, and sub-threshold slope, S,
values have also been listed. However, these latter parameter values scale inver-
sely with the gate dielectric capacitance, and this needs to be taken into account
when comparing VT and S values between different samples. Unfortunately, the
dielectric film thicknesses were not always published, so it is difficult to identify
meaningful differences in these parameters across the whole sample set, although
the smallest sub-threshold slope of 0.1 V/dec clearly correlates with a very thin
gate dielectric of just 5.3 nm [13]. Where the dielectric thicknesses have been
quoted, comparisons can be made between the normalised sub-threshold slope
values, SN, in which the measured values have been normalised to a fixed
dielectric capacitance of 3.45 9 10-8 Fcm-2 (which is equivalent to 100 nm of
SiO2). The smallest SN value is for the TIPS-pentacene TFT, and is indicative of a
lower overall density of trapping states in that material. Comparable, but slightly
larger values are seen for the PXX and N3000 TFTs as well, and are also indicative
of modest trap state densities, compared with the polymer p-channel materials,
380
Table 10.1 Summary of basic TFT performance parameters from a range of organic semiconductors
Material Channel Gate dielectric Dielectric Mobility Threshold On:off ratio S SN Ref #
thickness (nm) (cm2/Vs) voltage (V) (log10) (V/dec) (V/dec)
P3HT p SiO2 ? HMDS 230 0.05–0.1 – [6 1–1.5 0.44–0.65 116
PQT-12 p SiO2 ? OTS 100 0.14 – [7 1.5 1.5 21
Pentacene p SiO2 ? OTS – 1.00 -5 7 0.5 – 78
DVAnt p SiO2 ? OTS – 1.30 -16 7 0.5 – 78
TIPS-pentacene p SiO2 ? HMDS 370 0.65 3.4 8 0.5 0.14 83
PXX p PVP/OTS 400 0.40 – 7 0.6 0.17 4, 79
Pentacene p AlOx ? SAM – 0.50 – [5 – – 109
DNTT p AlOx ? SAM 5.3 0.60 – 6 0.1 – 13
F16CuPc n AlOx ? SAM – 0.01 – [4 – – 109
P(NDI2OD-T2) n PMMA 600–900 0.2–0.45 5–10 6–7 3–5 0.6–0.4 25
N3000 n D2000 400–700 3.00 – 7 1.0 0.24–0.14 58
SN is the normalised value of S, using a fixed dielectric capacitance of 3.45 9 10-8 Fcm-2
10
Organic TFTs
P3HT and PQT-12. The other general points to note from this table are that the
mobility values for the majority of materials are *0.4 cm2/Vs or greater, and that
the on:off current ratios are 6–7 orders of magnitude, which are sufficient for active
matrix addressing.
Although there is near-universal use of the basic MOSFET equations for the
simple extraction of carrier mobility and threshold voltage from OTFTs, a compact
model, representing a more accurate description of device behaviour, is needed for
circuit simulation [45, 118]. As with the inorganic TFTs (see Sects. 6.3.1 and
9.4.2.2), this needs to take account of non-ideal effects such as series resistance,
channel length shortening and the dependence of mobility on gate bias. For
example, incorporating the effects of series resistance, and the mobility
enhancement factor, c, from Eq. 10.2, the following dependence of channel current
on gate and drain bias has been derived [45]:
Wl0 Ci h i
Id ¼ ðVG VT VS Þcþ2 ðVG VT VD Þcþ2 ð10:5Þ
Lðc þ 2Þ
where VS and VD are the channel potentials adjacent to the source and drain
contacts, and reflect both series resistance effects plus any voltage drops across the
Schottky barrier contacts themselves. Equation 10.5 is similar to the a-Si:H TFT
Eq. 6.38, in which the exponent c ? 2 is replaced by a, and, in both cases, this
represents the influence of the band gap DOS in partitioning the gate-induced
charge between free and trapped states. The evaluation of c, VS and VD from
experimental OTFT characteristics, and the good fit of the model to the data, is
presented in Ref. [118].
10.5.2 Contact Effects
The source and drain metal contacts frequently have a voltage offset between the
energy level of their work function and the HOMO (p-channel) or the LUMO
(n-channel) levels of the semiconductor channel material. As discussed in
Sect. 10.2.4, due to the presence of surface dipole layers, these offsets are often
different from expectations based upon the published values of the work functions
of clean metals [41]. The source junction is effectively a reverse biased Schottky
barrier, and, if the barrier height is too large, severe injection problems will result.
For a given choice of semiconductor and dielectric materials, if this injection
barrier is too high it can dominate device performance, particularly as channel
length reduces [117, 120, 121]. The contact effect may be characterised in terms of
an effective series resistance, and channel width-normalised series resistance
values spanning the range 10 kXcm–10 MXcm have been reported [117]. (To put
these values into context, performance artefacts have been noted in poly-Si, AOS
and a-Si:H TFTs with series resistance values of 3 Xcm, 30–250 Xcm, and
7.8 kXcm, respectively—for further information on these values see Sects. 8.7.1,
382 10 Organic TFTs
Fig. 10.24 SKPM potentiometry measurements (a) measurement setup using an AFM probe
head. Surface potential measurements on a P3HT TFT with Au contacts (b) in the linear regime
with different drain biases, at a fixed gate bias, and c different gate biases at a fixed drain bias.
(Reprinted with permission from [119]. Copyright (2002) American Institute of Physics)
9.4.1 and 6.3.1, respectively). In view of the large values observed in OTFTs,
considerable attention has been paid to the choice of contact materials, and to the
evaluation of the injection barriers. This has been by direct barrier measurement in
operating TFT structures, using scanning Kelvin probe microscopy [119], as well
as by UV photo-electron spectroscopy measurements of dipole layer effects in
simpler test structures [24, 38, 39]. These measurements are key contributors to the
detailed understanding of the observed series resistance effects in OTFTs. In these
studies, several issues were addressed, namely the effect of dipole layers on barrier
heights, the injection process at the source [40] and how this translates into a
measured resistance, and the influence of device architecture (TC vs BC) on the
total series resistance. For instance, in the TC structure, there is vertical current
flow through the body of the semiconductor, whilst in BC structures the current
injection may be limited to the edge of the source contact adjacent to the channel.
An overview of investigations into contact and parasitic resistance effects in
OTFTs is presented below.
Non-contact, scanning Kelvin probe microscopy, SKPM, which is based upon
the use of a scanning force microscope, has been used to measure the semicon-
ductor voltage along the surface of an operating BG/BC TFT [119], and 2-D
device simulations confirmed that the surface potential closely followed the
channel potential [122]. The experimental set-up is shown in Fig. 10.24a, and the
voltage on the conducting tip of the probe is proportional to the channel potential.
Figures 10.24b, c show the potential measurements made along the surface of a
P3HT TFT with gold contacts and 5.2 lm channel length [119]. In the first case,
with a gate voltage of -20 V, the device continued to operate in the linear regime
as drain bias increased from -1 to -8 V, and in (c) the device went from the
saturation regime to the linear regime when the gate bias increased from -2 to
-30 V, with a fixed drain bias of -8 V. These profiles confirmed the applicability
of the simple, first order MOSFET model, with good linear potential drops along
the channel in the linear regime, and a progressive transfer of voltage to the drain
end of the channel as the device moved into saturation. The steep voltage drops,
DV, at either end of the channel in the linear regime were due to the contact
resistances, Rc, which can be simply evaluated as DV/Id, where Id is the channel
current. In this sample, DV was the same at both ends of the channel, and the total
series resistance was 50 kXcm. As will be seen below, the situation in which
DVs = DVd is indicative of a low injection barrier, and there are many examples
in which the inappropriate choice of contact metal led to significantly larger values
of DVs at the source end of the channel. In other measurements, the voltage drop at
the drain was not observed [123], and 2-D simulations have demonstrated that this
voltage drop can be associated with low mobility, disordered material adjacent to
the contacts [124]. The disorder had the effect of restricting current flow at the
vertical edges of the contacts, and forcing it along the horizontal surface of the
contacts, where the carrier density was much lower than in the channel.
The field effect mobility, and its field dependence, was also extracted from the
potential profiles using:
Id ¼ WlFE ðVG ; VðxÞÞCi ½VG VðxÞdVðxÞ=dx ð10:6Þ
The SKPM procedure has been applied to a variety of materials and contact
metals, including P3HT [117, 119], F8T2 [117] and pentacene [123].
Figures 10.25a, b show further measurements on P3HT with Cr-Au contacts (where
Cr was used as an adhesion layer on the SiO2 gate dielectric) and Cr contacts,
respectively [117]. These measurements were made in the linear regime at VG = -
40 V and Vd =
-8 V, and the Au contact results, in Fig. 10.25a, were consistent with the compa-
rable measurements in Fig. 10.24b, c. In particular, the values of DVs and DVd were
very similar, whilst, for the Cr contact, DVs [ DVd, and this was taken to be
indicative of a substantial injection barrier at the Cr source contact. The value of DVs
at the source end of the channel is a direct measure of the reverse bias developed
across the Cr Schottky barrier contact at the source, and it reduces both the effective
gate-source and source-drain biases to VG - DVs, and Vd - DVs, respectively. The
consequence of this is a self-consistent reduction in channel current, for the given
terminal biases VG and Vd, as is required for current continuity through the large
source barrier and the channel.
Taking the ionisation potential, IP, of P3HT to be 5.0 eV, the difference in
injection barriers in Fig. 10.25a, b correlated with the different work functions, UM,
of Au (5.2 eV) and Cr (4.7 eV). (In this work, the dipole barrier was not measured,
but it was assumed that the usual reduction in effective work function would be
comparable for Au and Cr, and that Au would, therefore, give the smaller injection
barrier [117], whilst the barrier would be[0.3 eV with Cr). From the values of DVs
384 10 Organic TFTs
Fig. 10.25 SKPM

measurements on P3HT and
F8T2 TFTs with Au or Cr
contact metals a P3HT with
Cr-Au, b P3HT with Cr, and
c F8T2 with Au. (Reprinted
and DVd, the total series resistance was estimated to be 5400 kXcm for Cr and
22 kXcm for Au. The more usual way of determining series resistance is from the
channel length dependence of the drain current at fixed gate voltages, and the SPKM
measurements agreed well with those measurements [117]. (For further information
on the conventional series resistance measurement, the background to it was pre-
sented in Sect. 9.4.1). Finally, the difference in injection barriers identified by
SKPM correlated with low-voltage current crowding in the output characteristics of
the TFTs with Cr contacts, whilst it was absent with the Au contacts.
Figure 10.25c shows SKPM measurements on an F8T2 TFT with Au contacts.
In contrast to the measurements of P3HT TFTs with Au contacts, F8T2, which has
an IP of 5.5 eV, displayed a substantial injection barrier at the source contact,
which correlated with a value of UM - IP [ 0.3 eV [117].
Carrier flow over the barrier is often assumed to be due to thermionic emission
in 2-D modelling work on OTFTs [121], but analysis of the temperature depen-
dence of both the barrier height and the carrier mobility argued against this simple
model, and favoured a thermally assisted tunnelling process [117].
As is apparent from Fig. 10.25, the forward-biased drain junction was not found
to present a significant barrier to current flow in any sample. However, the small
barrier at this junction did represent an element of series resistance, which was also
present at the source of the Cr-Au P3HT TFT. This resistance was found to scale
inversely with the carrier mobility of the material, and was attributed to carrier
flow through a narrow depletion region at the edge of the drain junction. This was
also assumed to be the case at the source junction in the Cr-Au P3HT TFT, where
the injection barrier was estimated to be close to zero [117].
The above results were for polymer semiconductors, and broadly similar effects
have been seen with pentacene TFTs. For example, measurements of SKPM and of
the output characteristic of BC TFTs, with the dual-metal contacts of Pd and Pt,
gave the results shown in Fig. 10.26a, b, respectively [123]. The SKPM mea-
surement showed a near-zero injection barrier when Pt was used as the source
contact (and Pd as the drain), but a substantially larger one with the contacts
reversed, with Pd as the source. This correlated with the greater current crowding in
the output characteristics when Pd was used as the source contact. Whilst the dipole
layers on the metals were not directly measured, the injection barrier differences
were consistent with the different work functions of *5.8 and 5.1 eV for Pt and Pd,
respectively [41]. However, in measurements of TFTs with Pd and Ni dual-metal
contacts, which have similar work functions, the use of Ni as the source contact
showed a much larger injection barrier than Pd, and also worse current crowding.
This was attributed to oxidation of the Ni surface during UV-ozone cleaning of the
substrates prior to the pentacene deposition. The dipole layers were not measured
on these samples, but these results indicate the varying sensitivities of different
metal surfaces to chemical exposure, as has also been reported for Au [24]. The
contact resistance values were directly evaluated from measurements of the channel
length dependence of the total device resistance, as a function of gate bias, in
Pd-contacted TFTs [121]. These characteristics were de-convolved (as discussed in
Sect. 9.4.1) to separate the channel and parasitic resistances, and are shown in
Fig. 10.26c. The resistance curves demonstrate that the parasitic resistance was
comparable to, and about 3 times less than, the channel resistances in 10 lm and
30 lm channel length TFTs, respectively. Hence, the injection barrier associated
with the Pd contacts had an increasing impact upon the operation of TFTs as the
channel length reduced from 30 lm, and the series resistance associated with Au
contacts was even higher in this work [121]. The width normalised contact resis-
tances at VG = -20 V were 6.9 kXcm and 51 kXcm for Pd and Au, respectively.
It is apparent from these results that the contact resistance in OTFTs consists of
two separate elements: a carrier transport resistance between the channel and the
source and drain contacts (which is comparable at both contacts), and, secondly,
where there is a large offset between the carrier band (HOMO or LUMO) and the
metal work function, an injection barrier resistance at the source electrode, which
can exceed the transport resistance. Hence, the overall series resistance is sensitive
to the choice of metal, and its chemical exposure, and to the choice of semicon-
ductor, particularly its carrier mobility. In addition, the resistance is also affected
by the use of SAM layers in BC TFTs to improve the molecular organisation of the
386 10 Organic TFTs
Fig. 10.26 Measurements of pentacene TFTs with reversible Pt and Pd dual-metal source/drain
contacts. a SKPM measurements. b output characteristics (Reprinted with permission from [123].
Copyright (2003) American Institute of Physics). c Channel and series resistance values as
function of VG on Pd-contacted BG/BC pentacene TFTs. (Reprinted with permission from [121].
semiconductor material on the electrode itself. Improved understanding and

optimisation of contact resistance is an on-going activity [40, 122, 125], and is
essential for the meaningful reduction of channel length, L, in order to improve
OTFT circuit speed. Indeed, speed will only scale as 1/L2, when the series
resistance is significantly less than channel resistance.
10.5.3 Contact Architecture
Given the importance of the contact effects discussed in the preceding section, there
has been much interest in whether top or bottom contact devices display different
contact limitations in their performance. For instance, with staggered BG/TC

(or TG/BC) devices there is a requirement for vertical current flow through the bulk
of the device, from the channel to the contacts, which is film thickness dependent
[53, 126]. On the other hand, there is more opportunity in the staggered device for
the channel current to be distributed along the width of the contacts, due to the
overlap of the gate and contact electrodes [126], whereas, in a coplanar BG/BC
structure, the current flow is limited to the contact electrode edges.
So far as TFT measurements are concerned, there are varied reports on the
relative values of contact resistance in BG/BC and BG/TC TFTs. For pentacene
with Au contacts, in some studies the BC TFTs were observed to have the higher
contact resistance [121, 127]. In other work, whilst the lowest resistance was found
with 20 nm thick pentacene films in TC structures, the resistance in the TC
structures increased with film thickness, and lower contact resistance was found in
the BC TFTs with 40-80 nm thick pentacene films [53]. For Pd-contacted penta-
cene films, the contact resistance was lowest in BC structures at low gate bias, and
it then became lower in TC devices as the gate bias increased [128]. These
different experimental results are no doubt related to subtle differences in device
technology and processing.
A more fundamental investigation of contact effects can be obtained from 2-D
device simulation, and, in these studies, TC devices have been consistently found
to have lower values of contact resistance, and better performance, than BC
structures [121, 122, 129]. This has been partially explained by the carrier tail from
the channel extending into the substrate, and reducing the bulk resistance, and also
by the spreading of the carriers along the contacts themselves. An example of the
computed difference between TC and BC structures is shown in Fig. 10.27, in
which the output characteristics were calculated for three different injection barrier
heights of 0, 0.2 and 0.3 eV [121]. For the 0 eV barrier, the two sets of curves
were identical, indicating that, in these calculations, the differences in transport
resistance were negligible between the two architectures. However, as the barrier
height increased, the device operation was modulated by the size of the injection
barrier, and this effect was far greater in the BC TFT. In explaining these, and
Fig. 10.27 2-D simulation of

TC (filled symbols) and BC
(open symbols) TFTs, with
source injection barriers of
0.0, 0.2, and 0.3 eV.
Physics)
388 10 Organic TFTs
other TC simulations, one topic which was not reported was the capacitive
coupling of the gate to the Schottky barrier source contact, and the influence of the
gate bias on the injection barrier height at the TC. It has also been noted that
injection barriers of \0.3 eV are needed for the efficient operation of OLEDs,
whereas the operation of OTFTs seems to be tolerant of apparently much larger
barriers [131]. An obvious difference between the OLED and the TFT structures is
the presence of the gate in the latter, and, in subsequent work [130–132], there has
been increasing recognition of the role of the gate bias in modulating the height of
the reverse biased source barrier.
In one case [130], dealing with poly-crystalline material in staggered OTFTs, it
was reported that the hole accumulation layer, induced by the gate bias, increased
the field at the source and lowered the source barrier by the Schottky effect [133].
In another publication [131], 2-D simulation was used to compare gate-bias-
induced barrier lowering in coplanar and staggered TFTs. In the coplanar TFTs,
the direct Schottky effect was identified between the edge of the source contact and
the adjacent hole accumulation layer, with a continuous reduction in barrier height
with increased gate bias. The Schottky effect was also invoked for the staggered
structure, but it was shown that the effect saturated at large gate biases. The
saturation was attributed to the hole accumulation layer, on the opposite side of
the film from the source, screening the source contact from further increases in the
gate-induced space charge field. However, these evaluations were carried out at
low values of reverse bias on the source, and did not allow for the increased field at
the source once its reverse bias was increased, and started to deplete the hole
accumulation layer [134]. Indeed, in a comprehensive 2-D numerical analysis of
staggered OTFTs, it was shown that the field at the source was due to the
combination of both the gate bias and the source reverse bias, and that gate bias
alone (at low values of VD) had only a very small impact upon the source barrier
height [132]. This work demonstrated significant barrier lowering at the source
contact due to the combined effects of the gate and source biases. Whilst these
papers have identified the involvement of the gate in barrier height control in
OTFTs, there is presently not a consensus on the precise details, and further
analysis can be anticipated on the broad topic of gate-bias-controlled series
resistance effects in OTFTs.
In this context, it is worth mentioning a novel form of TFT, called the Source-
Gated Transistor, SGT, which has both a BG/TC structure, and a Schottky barrier
source. In the SGT, it is argued that, in saturation, the height of the reverse biased
source barrier is controlled by the electrostatic coupling of the gate potential, and
this determines the source-drain current in injection-limited structures [134–136].
Moreover, 2-D simulations of these structures have indicated that they can
enhance the performance of low mobility organic materials compared with their
use in conventional TFTs [136]. The physics of SGT operation is presented in
Chap. 12, where the current understanding of SGT behaviour is compared with
recent analyses of source barrier effects in OTFTs.
10.6 Instability Effects 389
10.6 Instability Effects
A variety of instability effects have been observed in OTFTs, and, to a certain

extent, these reflect the different materials, transistor designs, and preparation
techniques used with these devices [137]. Nevertheless, it is possible to define
some of the more common and representative instabilities, including ambient
instability, hysteresis during sample measurement, and longer-term bias stress
instability. It should be noted that the specific details are likely to be a function of
the particular organic semiconductor, its gate dielectric, and the conditions under
which the instability is assessed. For instance, depending upon the materials, the
ambient has been found to have an important effect, as demonstrated by differ-
ences observed between devices measured in vacuum, dry gases, air, and high
humidity environments.
10.6.1 Air-Instability
Air-instability is related to the material’s sensitivity to the presence of oxygen and

water, which are able to diffuse through many semiconductor materials, and to
cause an oxidising reaction. Oxidation is a process involving electron transfer from
the host material to the oxidant [18], and the more tightly bound the electrons are
in the host, the lower its susceptibility to this process. Hence, for a p-channel
organic semiconductor, the greater its ionisation potential (or HOMO level), the
more air-stable it should be. The same argument applies with respect to the LUMO
level of a semiconductor under electron accumulation (i.e. for an n-channel device
in the on-state), and the deeper the LUMO level is the more air-stable it should be.
As mentioned in Sect. 10.4.2, the air-instability effect most commonly observed
was a reduction in carrier mobility, due to trap creation by the oxidation reactions,
but changes in threshold voltage and off-current have been noted as well [76].
Amongst the more widely studied p-channel TFT materials, pentacene and P3HT
were shown to be affected by air exposure, and more stable materials, such as
DNTT [13], DVAnt [78], have been developed with deeper HOMO levels. An
alternative approach was to change the molecular structure, such as with Ph-PXX
[4, 79], in which substituents were added to the periphery of the molecule to
passivate reactive sites, and thereby reduce its sensitivity to oxygen. A comparison
of the air stability of pentacene and Ph-PXX is shown in Fig. 10.28a, and illus-
trates the much better ambient stability of Ph-PXX. Similarly, PQT-12 has a
modified molecular structure compared with P3HT, which improved its oxidative
stability, even though its HOMO was only 0.1 eV deeper [21].
Early n-channel materials, which were used in TFTs were made on SiO2 gate
dielectrics, were very sensitive to oxidation by silanols on the insulator surface.
This reaction resulted in the formation of electron traps, which suppressed electron
conduction, and also precluded ambipolar behaviour in good p-channel materials
390 10 Organic TFTs
Fig. 10.28 a Comparison of the air-stability of pentacene and Ph-PXX TFTs (Reprinted with
permission from [79]. Copyright (2009) American Chemical Society). Stability of electron
mobility in n-channel P(NDI2OD-T2) TFTs with different gate dielectrics {PMMA, PS
(polystyrene), and commercial D2200 (polyolefin-polyacrylate)}: b During ambient storage at
relative humidity levels of 20–60 %, and c after 24 h storage at different humidity levels.
(Reprinted by permission from Macmillan Publishers Ltd: Nature [25], copyright (2009))
[36]. Hydroxyl-free insulators, such as BCB, polyethylene, PMMA and parylene

were demonstrated to remove/reduce this electron trapping problem, and gave
good n-channel behaviour [36]. However, n-channel materials were still sensitive
to air-degradation, due to the diffusion of oxygen and water through the film, and it
was predicted that the LUMO level had to be deeper than *4 eV to suppress the
oxidation reaction [92]. Good air-stability was subsequently demonstrated with the
n-channel materials F16CuPc [93], and FPTBBT [94], having LUMO levels of 4.8
eV and 5.05 eV, respectively. An alternative approach was to use a dense
molecular structure to impede the ingress of oxidants, and high performance, air-
stable n-channel TFTs were obtained with P(NDI2OD-T2) [25], although its
LUMO level was only *4.0 eV. The excellent air-stability of the electron
mobility in P(NDI2OD-T2) TFTs, with different gate dielectrics, is shown in
Fig. 10.28b, and these devices also displayed good stability at relative humidity
levels up to *60 %, as seen in Fig. 10.28c [25].
10.6.2 Gate Bias Stress Instability
Given the earlier development of high quality p-channel TFTs, a substantial body of
work has been published on the bias-stress instability of these devices. By and large,
the most significant instability has been under gate bias stress, with the drain bias
having a relatively minor effect (as with a-Si:H TFTs). As bias stress instability is
such a fundamental aspect of device performance, most semiconductor and gate

dielectric materials of interest have been routinely evaluated under gate bias stress.
The most common instability was an increase in threshold voltage towards the
value of the stress bias. In other words, there was a negative shift in threshold
voltage with negative gate bias stress, and a positive shift with positive bias stress.
This has been particularly true with the SiO2 gate dielectric, which, in many studies,
was deliberately chosen as a highly stable material, in order to focus on the
instability within the organic semiconductor itself. However, it should also be noted
that some TFTs with organic dielectrics, such as PVP [138, 139], have given the
opposite polarity shifts, where the TFT behaviour has been dominated by insta-
bilities within the organic dielectric material.
The typical features of gate bias stress instability include the polarity effect
described above, in which the saturated shift in threshold voltage tended towards
the value of the gate bias stress voltage. This direction of threshold voltage shift is
also equivalent to a reduction in the on-current at a fixed value of gate bias, as can
be seen in Fig. 10.29a. This figure shows the time dependent change in the transfer
characteristics, due to the application of a gate bias stress of -20 V to a PTAA TFT
with a thermally grown SiO2 gate dielectric [140]. The change, which is charac-
teristic of this type of instability, was initially rapid (and could be observed on a
time scale of seconds), but it progressively slowed down, and continued over
periods of hours or more. In view of the extensive time scale of the effect, it is
frequently represented on a logarithmic time axis, as shown in Fig. 10.29c.
Moreover, given the initially fast rate of change, the threshold instability may also
be observed as hysteresis in a transfer characteristic between the outward and return
gate voltage sweeps. The time dependence can be empirically fitted with a stretched
exponential [137, 140–142], as shown by the dashed line in Fig. 10.29c [140]. The
trends shown for the PTAA TFT in Fig. 10.29 have also been found in similarly
prepared devices containing the following alternative channel materials: pentacene,
P3HT, F8T2, PTV (polythienylene-vinylene), and 3-BuT5 (3-butyl-quinquethi-
ophene) [140, 141]. Similar results have also been measured in single crystal
rubrene, TIPS-pentacene, and tetracene TFTs on parylene dielectric layers [142].
The stretched exponential fitted to the data in Fig. 10.29c has the form:
DVT ¼ V0 f1 exp ðt=sÞb g ð10:7Þ

where V0 * VGstress, the exponent b was weakly temperature dependent, with a
value in the range 0–1, and the characteristic relaxation time, s, was thermally
activated, and given by:
s ¼ s0 expðEA =kTÞ ð10:8Þ
For the device shown in Fig. 10.29c, which was measured in air at 30 °C, the
fitted parameter values were V0 = 19 V, b = 0.43, s = 104 s, and the activation
energy, EA, was *0.6 eV [140]. However, when stressing in vacuum, the insta-
bility was much slower, with s = 106s, although the other parameters remained
much the same. Similar behaviour was seen with the semiconductors pentacene,
392 10 Organic TFTs
Fig. 10.29 Transfer curves measured on a PTAA TFT as a function of time in air a with -20 V
gate bias stress, b recovery after removal of gate bias, c gate bias stress data (symbols), and fitted
stretched exponential (dashed line), and d recovery data (symbols), and fitted stretched
exponential (dashed line) (Reprinted with permission from [140]. Copyright (2010) by the
P3HT, F8T2, PTV, and 3-BuT5, and, in particular, they all had comparable
activation energies of *0.6 eV, indicating the same underlying instability
mechanism. Where they differed was in the values of their relaxation time, s,
which scaled inversely with the HOMO level of the semiconductor, as shown by
the results in Fig. 10.30 (taken at 25 °C, in vacuum) [141]. In this plot, the
materials with the deepest HOMO levels were the least stable. A qualitatively
Fig. 10.30 Variation of the

characteristic stress
relaxation time, s (at 25 °C in
vacuum) with the HOMO
level of the semiconductor.
Physics)
similar dependence of relaxation time on the HOMO level was also seen with the
single crystal samples of rubrene, TIPS-pentacene and tetracene, albeit with a
smaller slope than the data in Fig. 10.30 [142], indicating that these broad trends
are quite common.
The other common features to the instability process are that it is reversible
once the gate bias is removed, as shown in Fig. 10.29b, and that the time
dependence could also be fitted with a stretched exponential, as shown by the
dashed lines in Fig. 10.29d [140]. There are three sets of recovery data in this
figure, which started at different points in the stress cycle (as indicated by the
arrows from Fig. 10.29c), and one of the key features of the recovery process was
that the fitted recovery time, s, increased with the duration of the initial stress, ts.
The s values were 7.7 9 104 s, 6.5 9 103 s, and 2 9 103 s for curves A
(ts = 105 s), B (ts = 104 s), and C (ts = 2 9 103 s), respectively. Hence, the
characteristic recovery time was more than 100 times longer for the stress of 105 s
(curve A) than it was the 2 9 103 s stress (curve C). The correlation between the
stress duration and the recovery speed has been reported previously in PQT-12
[143], and, in some cases, authors have noted that full recovery had not been
achieved within a given recovery time [142, 143].
The type of gate bias instability seen in Fig. 10.29a has been widely observed,
and attributed to carrier trapping either in the semiconductor, in the dielectric or at
the interface [137]. However, in interpreting the instability effects in Fig. 10.29,
account was taken of the fact that humidity has been found to strongly influence
gate bias stability, and, in the cited results, there were large differences in the
relaxation times between vacuum and air stressing, and that gate bias instability
was greatly reduced with the hydrophobic gate insulator Cytop [140]. This indi-
cated that water played a role in the instability, and the instability was attributed to
proton generation from water at the semiconductor/SiO2 interface, and its sub-
sequent diffusion into the gate dielectric. One of the reasons for proposing proton
participation was because its activation energy for diffusion in SiO2 was 0.5 eV,
and this was comparable to the instability activation energy of *0.6 eV [140,
394 10 Organic TFTs
144]. The detailed microscopic model involved an electrolytic reaction between

free holes and water molecules at the semiconductor/SiO2 interface, which pro-
duced protons by:
2H2 O þ 4OSþ ) 4Hþ þ O2 þ 4OS ð10:9Þ
and the reverse reaction of protons back to holes:
2Hþ þ 2OS ) H2 þ 2OSþ ð10:10Þ
where OS+ and OS represent holes and neutral sites in the semiconductor,
respectively. Hence, the instability was triggered by the conversion of free holes
into protons, and the establishment of equilibrium densities of protons across the
interface. This was assumed to be a fast process, and the dynamics of the insta-
bility were governed by the diffusion of the protons away from the interface, into
the gate oxide. The role played by the individual semiconductor HOMO level was
in stimulating the reaction shown in Eq. 10.9, because a free hole density
(equivalent to an absence of electrons) in the semiconductor can be regarded as
representing the oxidised state of the material, and, the deeper the HOMO is, the
less stable the oxidised state, OS+, is [140, 141]. Hence, this carrier driven
instability is the opposite of the ambient instability, in which the most unstable
samples are those with the shallowest HOMO levels.
The model also accounted for the recovery behaviour of the material, since
removing the gate bias removed the hole accumulation layer, and this disturbed the
equilibrium between holes and protons at the interface. The absence of holes
stimulated the back reaction at the interface (Eq. 10.10), and produced a reverse
diffusion gradient of protons back to the interface. Hence, the longer the bias-stress
period, the further the protons had diffused into the oxide, and the longer it took
them to return to the interface, where they were neutralised by conversion into
holes, and were then swept into the contacts. (The solid lines in Fig. 10.29 were
based upon the detailed kinetics of the proton diffusion controlled instability model
[140], although the generation of protons has not been confirmed by direct mea-
surements [144]). The conclusion drawn from this work was that water had to be
suppressed from the material in order to achieve good bias stress stability.
As mentioned above, the single crystal samples of rubrene, TIPS-pentacene and
tetracene displayed qualitatively similar results, but a different microscopic model
was advanced. In this model, the holes were captured in a distribution of tail states
above the HOMO level in the parylene dielectric, and the deeper the semicon-
ductor HOMO level, the greater the density of tail states available to it in the
parylene. Once captured in the tail states of the dielectric, the holes then drifted
towards the gate electrode under the influence of the gate field [142]. With the
parylene and tetracene HOMO levels at *7.0, and 6.2 eV, respectively, tetracene
showed much faster instability than rubrene, with its shallower HOMO at 4.2 eV.
The detailed kinetics of this instability model, based upon the drift of holes in the
dielectric, gave a stretched hyperbolic function, which was fitted to the data [142].
There are evidently qualitatively similar gate bias stress results across a range
of organic semiconductors, with the semiconductor’s HOMO level playing an
important role, but no current agreement on a common microscopic model.
Finally, it is worth noting the role of the dielectric layer in qualitatively
changing the type of results discussed above. In particular, those instances in
which the polarity of the instability has been reversed, whereby a negative gate
bias produced a positive threshold voltage shift. This is also related to ambient
effects, in that it has been observed for PVP stressed in air [138, 139] and also for
PVA [145], but not for PVP stressed in vacuum, where the normal polarity of
threshold voltage shift was displayed [139]. The anomalous threshold voltage shift
[146] has been variously attributed to electron injection from the gate electrode
[138, 145], the movement of ionic impurities, or the polarization of water mole-
cules in the dielectric, given its sensitivity to moisture in the atmosphere [147]. In
most of those studies the dielectric was thermally cross-linked, and it was shown
that deep UV curing of the material led to more complete cross-linking, and
suppressed the instability [146]. For the above two dielectrics, ambient moisture
played a role in the bias instability, as was also cited for the proton diffusion model
in SiO2 [140, 141], and better device stability has been reported with a moisture-
excluding, hydrophobic dielectric, such as Cytop [148].
In summary, whilst devices with good bias-stability (comparable to a-Si:H
TFTs) have been demonstrated [96, 137], there is a complex range of phenomena
present in OTFTs, involving both the semiconductor and dielectric materials, as
well as their sensitivity to ambient effects. In addition, there are different micro-
scopic models for seemingly similar phenomena, and it is presently unclear
whether there are, indeed, quite distinct instability mechanisms taking place in
different material combinations.
10.7 Summary
A large number of organic TFT materials now display carrier mobility values
approaching, or exceeding, those of the bench-mark TFT for the flat panel display
industry, namely a-Si:H. This has stimulated considerable interest in using OTFTs
for low cost, flexible substrate applications, such as RFID tags, and e-reader
displays. Indeed, many research demonstrator displays have been presented, not
just for AMEPDs, but for AMLCDs and AMOLEDs as well.
The current organic semiconductors of choice have been engineered with the
appropriate molecular structure and ordering to give efficient charge transport in a
TFT structure, and a wide range of both small molecule and polymer materials
have been identified. Generally, better performance is obtained from vacuum-
deposited small molecule materials, but polymer materials more readily lend
themselves to lower-cost, solution processing. However, a range of solution-
processable small molecule materials have now also been developed. Initial
materials research used oxidised Si as a combined gate electrode and gate
396 10 Organic TFTs
dielectric, but this is not practical for commercial applications, and, in tandem with
the semiconductor developments, considerable attention has been paid to alter-
native dielectrics. These have included both low temperature inorganic dielectrics,
as well as a range of organic insulators, and there is continuing research to opti-
mise the use of these materials.
The typical device architecture is a non-self-aligned structure, similar to those
used for a-Si:H and amorphous oxide TFTs, although, unlike those technologies,
there is not one clearly preferred architecture. Equally, the industry has not iden-
tified a particular combination of semiconductor and dielectric layers as optimum,
with a range of materials continuing to be used in research demonstrator displays.
As with other TFT materials, OTFTs display various instabilities, including
ambient and gate bias stress instability. The ambient instabilities have been
addressed both by modifying the molecular structure of the material to reduce its
sensitivity to oxygen and water, or by decreasing its permeability to these mole-
cules. Gate bias instability most commonly results in a threshold voltage shift of
the same polarity as the stress bias, and, in a number of cases, the instability time
constant has been related to the depth of the semiconductor’s HOMO level.
However, there is not a single, accepted microscopic model for the underlying the
instabilities. At a practical level, gate bias instabilities in well engineered struc-
tures have been demonstrated to match those of a-Si:H TFTs, and the appearance
of a first AMEPD product has been slated for 2012.
References
1. Kelley T (2006) High-performance pentacene transistors. In: Klauk H (ed) Organic

electronics: materials, manufacturing and applications. Wiley, Weinheim
2. Mas-Torrent M, Rovira C (2011) Role of molecular order and solid-state structure in
organic field-effect transistors. Chem Rev 111:4833–4856
3. Sekitani T, Iba S, Kato Y, Noguchi Y, Someya T, Sakurai T (2005) Ultraflexible organic
field-effect transistors embedded at a neutral strain position. Appl Phys Lett 87:173502-
1–173502-3
4. Nomoto K, Noda M, Kobayashi N, Katsuhara M, Yumoto A, Ushikura S-i, Yasuda R-i,
Hirai N, Yukawa G and Yagi I (2011) Rollable OLED display driven by organic TFTs.
SID’11 tech digest, pp 488–491
5. Sheraw CD, Zhou L, Huang JR, Gundlach DJ, Jackson TN, Kane MG, Hill IG, Hammond
MS, Campi J, Greening BK, Francl J, West J (2002) Organic thin-film transistor-driven
polymer-dispersed liquid crystal displays on flexible polymeric substrates. Appl Phys Lett
80(6):1088–1090
6. Burns S (2010) QUE: an e-reader built using flexible display technology. SID’10 tech
digest, pp 477–479
7. Yoneya N, Ono H, Ishii Y, Himori K, Hirai N, Abe H, Yumoto A, Kobayashi N, Nomoto K,
Urabe T (2011) Flexible electrophoretic display driven by solution-processed OTFTs.
SID’11 tech digest, pp 288–291
8. McGoldrick K (2011) Flexible displays ready for lift off. eBook & tablet market evolution
conference, SID Display Week. http://www.polymervision.com/files/Karl_McGoldrick_
Polymer_Vision_%20SID%202011.pdf. Accessed Aug 2011
References 397
9. Mizukami M, Hirohata N, Iseki T, Ohtawara K, Tada T, Yagyu S, Abe T, Suzuki T, Fujisaki

Y, Inoue Y, Tokito S, Kurita T (2006) Flexible AM OLED panel driven by bottom-contact
OTFTs. IEEE Electron Dev Lett 27(4):249–251
10. Nakajima Y, Takei T, Tsuzuki T, Suzuki M, Fukagawa H, Yamamoto T, Tokito S (2009)
Fabrication of 5.8-in. OTFT-driven flexible color AMOLED display using dual protection
scheme for organic semiconductor patterning. J SID 17(8):629–634
11. Klauk H, Zschieschang U, Pflaum J, Halik M (2007) Ultralow-power organic
complementary circuits. Nature 445:745–748
12. Jeon J, Murmann B, Bao Z (2010) Full-swing and high-gain pentacene logic circuits on
plastic substrate. IEEE Electron Dev Lett 31(12):1488–1490
13. Zschieschang U, Ante F, Yamamoto T, Takimiya K, Kuwabara H, Ikeda M, Sekitani T,
Someya T, Kern K, Klauk H (2010) Flexible low-voltage organic transistors and circuits
based on a high-mobility organic semiconductor with good air stability. Adv Mater
22:982–985
14. http://www.plasticlogic.com/news/pr_education_announce_sep122011.php. Accessed Sept
2011
15. Klauk H (2012) Organic thin film transistors based on vacuum-deposited small-molecule
semiconductors. In: Chen J, Cranton W, Fihn M (eds) Handbook of visual display
technology. Springer, Berlin
16. Redinger D, Payne M (2012) Organic TFTs: solution-processable small-molecule
semiconductors. In: Chen J, Cranton W, Fihn M (eds) Handbook of visual display
technology. Springer, Berlin
17. Liu F, Ferdous S, Briseno AL (2012) Polymer thin film transistors. In: Chen J, Cranton W,
Fihn M (eds) Handbook of visual display technology. Springer, Berlin
18. Hunt A (2007) Complete A-Z chemistry handbook, 3rd edn. Hodder Arnold, London
19. Halik M, Klauk H, Zschieschang U, Schmid G, Ponomarenko S, Kirchmeyer S, Weber W
(2003) Relationship between molecular structure and electrical performance of
oligothiophene organic thin film transistors. Adv Mater 15(11):917–922
20. Bao Z, Dodabalapur A, Lovinger AJ (1996) Soluble and processable regioregular poly(3-
hexylthiophene) for thin film field-effect transistor applications with high mobility. Appl
Phys Lett 69(26):4108–4110
21. Ong BS, Wu Y, Liu P, Gardner S (2004) High-performance semiconducting polythiophenes
for organic thin-film transistors. J Am Chem Soc 126:3378–3379
22. McCullough RD (1998) The chemistry of conducting polythiophenes. Adv Mater
10(2):93–116
23. Malliaras G, Friend R (2005) An organic electronics primer. Phys Today 58(5):53–58
24. Braun S, Salaneck WR, Fahlman M (2009) Energy-level alignment at organic/metal and
organic/organic interfaces. Adv Mater 21:1450–1472
25. Yan H, Chen Z, Zheng Y, Newman C, Quinn JR, Dotz F, Kastler M, Facchetti A (2009) A
high-mobility electron-transporting polymer for printed transistors. Nature 457:679–687
26. Anthony JE (2006) Functionalized acenes and heteroacenes for organic electronics. Chem
Rev 106:5028–5048
27. Sirringhaus H, Brown PJ, Friend RH, Nielsen MM, Bechgaard K, Langeveld-Voss BMW,
Spiering AJH, Janssen RAJ, Meijer EW, Herwig P, de Leeuw DM (1999) Two-dimensional
charge transport in self-organized, high-mobility conjugated polymers. Nature 401:685–688
28. Sirringhaus H (2005) Device physics of solution-processed organic field-effect transistors.
Adv Mater 17:2411–2425
29. Kim DH, Park YD, Jang Y, Yang H, Kim YH, Han JI, Moon DG, Park T, Chang C, Joo M,
Ryu CY, Cho K (2005) Enhancement of field-effect mobility due to surface-mediated
ordering in regioregular polythiophene thin film transistors. Adv Func Mater 15(1):77–82
30. Swiggers ML, Xia G, Slinker JD, Gorodetsky AA, Malliarasa GG, Headrick RL, Weslowski
BT, Shashidhar RN, Dulceya CS (2001) Orientation of pentacene films using surface
alignment layers and its influence on thin-film transistor characteristics. Appl Phys Lett
79(9):1300–1302
398 10 Organic TFTs
31. Ruiz R, Nickel B, Koch N, Feldman LC, Haglund RF, Kahn A, Scoles G (2003) Pentacene
ultrathin film formation on reduced and oxidized Si surfaces. Phys Rev B 67:125406
32. Schiefer S, Huth M, Dobrinevski A, Nickel B (2007) Determination of the crystal structure
of substrate-induced pentacene polymorphs in fiber structured thin films. J Am Chem Soc
129:10316–10317
33. Di Carlo A, Piacenza F, Bolognesi A, Stadlober B, Maresch H (2005) Influence of grain
sizes on the mobility of organic thin-film transistors. Appl Phys Lett 86:263501
34. Coropceanu V, Cornil J, da Silva Filho DA, Olivier Y, Silbey R, Bredas JL (2007) Charge
transport in organic semiconductors. Chem Rev 107:926–952
35. Newman CR, Frisbie CD, da Silva Filho DA, Bre0 das J-L, Ewbank PC, Mann KR (2004)
Introduction to organic thin film transistors and design of n-channel organic
semiconductors. Chem Mater 16:4436–4451
36. Chua L-L, Zaumseil J, Chang J-F, Ou EC-W, Ho PK-H, Sirringhaus H, Friend RH (2005)
General observation of n-type field-effect behaviour in organic semiconductors. Nature
434(7030):194–199
37. Sirringhaus H (2009) Electrons in the fast lane. Nature 457:667–668
38. Ishii H, Sugiyama K, Ito E, Seki K (1999) Energy level alignment and interfacial electronic
structures at organic/metal and organic/organic interfaces. Adv Mater 11(8):605–625
39. Kahn A, Koch N, Gao W (2003) Electronic structure and electrical properties of interfaces
between metals and p-conjugated molecular films. J Polym Sci Part B: Polym Phys
41:2529–2548
40. Natali D, Caironi M (2012) Charge injection in solution-processed organic field-effect
transistors: physics, models and characterization methods. Adv Mater 24:1357–1387
42. Letizia JA, Rivnay J, Facchetti A, Ratner MA, Marks TJ (2010) Variable temperature
mobility analysis of n-channel, p-channel, and ambipolar organic field-effect transistors.
Adv Funct Mater 20:50–58
43. Fishchuk II, Kadashchuk AK, Vakhnin A, Korosko Y, Bässler H, Souharce B, Scherf U
(2006) Transition from trap-controlled to trap-to-trap hopping transport in disordered
organic semiconductors. Phys Rev B 73:115210
44. Necliudov PV, Shur MS, Gundlach DJ, Jackson TN (2000) Modeling of organic thin film
transistors of different designs. J Appl Phys 88(11):6594–6597
45. Marinov O, Deen MJ, Zschieschang U, Klauk H (2009) Organic thin-film transistors: part
I—compact DC modeling. IEEE Trans Electron Dev 56(12):2952–2961
46. Worne JH, Anthony JE, Natelson D (2010) Transport in organic semiconductors in large
electric fields: from thermal activation to field emission. Appl Phys Lett 96:053308
47. Podzorov V, Menard E, Borissov A, Kiryukhin V, Rogers JA, Gershenson ME (2004)
Intrinsic charge transport on the surface of organic semiconductors. Phys Rev Lett
93(8):086602
48. Bolognesi A, Berliocchi M, Manenti M, Di Carlo A, Lugli P, Lmimouni K, Dufour C (2004)
Effects of grain boundaries, field-dependent mobility, and interface trap states on the
electrical characteristics of pentacene TFT. IEEE Trans ED-51(12):1997–2003
49. Sakanoue T, Sirringhaus H (2010) Band-like temperature dependence of mobility in a
solution-processed organic semiconductor. Nat Mater 9:736–740
50. Kymissis I, Dimitrakopoulos CD, Purushothaman S (2001) High-performance bottom
electrode organic thin-film transistors. IEEE ED-48(6):1060–1064
51. De Angelis F, Cipolloni S, Mariucci L, Fortunato G (2005) High-field-effect-mobility
pentacene thin-film transistors with polymethylmethacrylate buffer layer. Appl Phys Lett
86:203505
52. Gundlach DJ, Jackson TN, Schlom DG, Nelson SF (1999) Solvent-induced phase transition
in thermally evaporated pentacene films. Appl Phys Lett 74(22):3302–3304
53. Mariucci L, Simeone D, Cipolloni S, Maiolo L, Pecora A, Fortunato G, Brotherton S (2008)
Effect of active layer thickness on electrical characteristics of pentacene TFTs with PMMA
buffer layer. Solid-State Electron 52:412–416
References 399
54. Kymissis I, Dimitrakopoulos CD, Purushothaman S (2002) Patterning pentacene organic

thin film transistors. J Vac Sci Technol B 20(3):956–959
55. Sirringhaus H, Kawase T, Friend RH, Shimoda T, Inbasekaran M, Wu W, Woo EP (2000)
High-resolution inkjet printing of all-polymer transistor circuits. Science 290:2123–2126
56. Arias AC, Endicott F, Street RA (2006) Surface-induced self-encapsulation of polymer thin-
film transistors. Adv Mater 18:2900–2904
57. Burns SE, Reeves W, Pui BH, Jacobs K, Siddique S, Reynolds K, Banach M, Barclay D,
Chalmers K, Cousins N, Cain P, Dassas L, Etchells M, Hayton C, Markham S, Menon A,
Too P, Ramsdale C, Herod J, Saynor K, Watts J, von Werne T, Mills J, Curling CJ,
Sirringhaus H, Amundson K, McCreary MD (2006) A flexible plastic SVGA e-paper
display. SID’06 tech digest, pp 74–76
58. Yan H, Zheng Y, Xia Y, Newman C, Quinn J, Lu S, Facchetti A (2010) Through-breaking
organic TFT materials for active matrix display backplane application. SID’10 tech digest,
1679–1681
59. Park SK, Mourey DA, Subramanian S, Anthony JE, Jackson TN (2008) High-mobility spin-
cast organic thin film transistors. Appl Phys Lett 93:043301
60. Mcculloch I, Heeney M, Bailey C, Genevicius K, Macdonald I, Shkunov M, Sparrowe D,
Tierney S, Wagner R, Zhang W, Chabinyc ML, Kline RJ, Mcgehee MD, Toney MF (2006)
Liquid-crystalline semiconducting polymers with high charge-carrier mobility. Nat Mater
5:328–333
61. Dickey KC, Anthony JE, Loo Y-L (2006) Improving organic thin-film transistor
performance through solvent-vapor annealing of solution-processable triethylsilylethynyl
anthradithiophene. Adv Mater 18:1721–1726
62. Anthony JE, Brooks JS, Eaton DL, Parkin SR (2001) Functionalized pentacene: improved
electronic properties from control of solid-state order. J Am Chem Soc 123:9482–9483
63. Lee WH, Kim DH, Jang Y, Cho JH, Hwang M, Park YD, Kim YH, Han JI, Cho K (2007)
Solution-processable pentacene microcrystal arrays for high performance organic field-
effect transistors. Appl Phys Lett 90:132106
64. Kim DH, Lee DY, Lee HS, Lee WH, Kim YH, Han JI, Cho K (2007) High-mobility organic
transistors based on single-crystalline microribbons of triisopropylsilylethynyl pentacene
via solution-phase self-assembly. Adv Mater 19:678–682
65. Duffy CM, Andreasen JW, Breiby DW, Nielsen MM, Ando M, Minakata T, Sirringhaus H
(2008) High-mobility aligned pentacene films grown by zone-casting. Chem Mater
20:7252–7259
66. Sele CW, von Werne T, Friend RH, Sirringhaus H (2005) Lithography-free, self-aligned
inkjet printing with sub-hundred-nanometer resolution. Adv Mater 17:997–1001
67. Sirringhaus H, Sele C W, von Werne T, Ramsdale C (2006) Manufacturing of organic
transistor circuits by solution-based printing. In Klauk H (ed) Organic electronics: materials,
manufacturing and applications. Wiley, Weinheim
68. Menard E, Meitl MA, Sun Y, Park J-U, Shir DJ-L, Nam Y-S, Jeon S, Rogers JA (2007)
Micro- and nanopatterning techniques for organic electronic and optoelectronic systems.
Chem Rev 107:1117–1160
69. Yun C, Kim M, Lee SW, Moon H, Park S, Koo JB, Kim JW, You I-K, Yoo S (2011) High-
performance pentacene thin-film transistors fabricated by printing technology. IEEE
Electron Dev Lett 32(10):1454–1456
70. Guerin M, Daami A, Jacob S, Bergeret E, Bènevent E, Pannier P, Coppard R (2011) High-
gain fully printed organic complementary circuits on flexible plastic foils. IEEE Trans
Electron Dev 58(10):3587–3593
71. Parashkov R, Becker E, Riedl T, Johannes H-H, Kowalsky W (2005) Large area electronics
using printing methods. Proc IEEE 93(7):1321–1329
72. http://www.dimatix.com/news-events/press-releases.asp?display=detail&id=81. Accessed
Oct 2011
73. Hong Y, Chung S (2010) Technical issues towards all inkjet-printed organic thin-film
transistors. SID’10 tech digest, pp 1147–1150
400 10 Organic TFTs
74. Sekitani T, Noguchi Y, Zschieschang U, Klauk H, Someya T (2008) Organic transistors

manufactured using inkjet technology with subfemtoliter accuracy. Proc Natl Acad Sci U S
A 105(13):4976–4980
75. Kitamura M, Arakawa Y (2008) Pentacene-based organic field-effect transistors. J Phys:
Condens Matter 20:184011
76. Locklin J, Ling MM, Sung A, Roberts ME, Bao Z (2006) High-performance organic
semiconductors based on fluorene–phenylene oligomers with high ionization potentials.
Adv Mater 18:2989–2992
77. De Angelis F, Cipolloni S, Mariucci L, Fortunato G (2006) Aging effects in pentacene thin-
film transistors: analysis of the density of states modification. Appl Phys Lett 88:193508
78. Klauk H, Zschieschang U, Weitz RT, Meng H, Sun F, Nunes G, Keys DE, Fincher CR,
Xiang Z (2007) Organic transistors based on di(phenylvinyl)anthracene: performance and
stability. Adv Mater 19:3882–3887
79. Kobayashi N, Sasaki M, Nomoto K (2009) Stable peri-xanthenoxanthene thin-film
transistors with efficient carrier injection. Chem Mater 21:552–556
80. Afzali A, Dimitrakopoulos CD, Breen TL (2002) High-performance, solution-processed
organic thin film transistors from a novel pentacene precursor. J Am Chem Soc
124:8812–8813
81. Yun C, Kim M, Lee SW, Moon H, Park S, Koo JB, Kim JW, You I-K, Yoo S (2011) High-
performance pentacene thin-film transistors fabricated by printing technology. IEEE
82. Whiting GL, Arias AC (2009) Chemically modified ink-jet printed silver electrodes for
organic field-effect transistors. Appl Phys Lett 95:253302
83. Park SK, Jackson TN, Anthony JE, Mourey DA (2007) High mobility solution processed
6,13-bis(triisopropyl-silylethynyl) pentacene organic thin film transistors. Appl Phys Lett
91:063514
84. Kawashima N, Kobayashi N, Yoneya N, Ono H, Fukuda T, Ohe T, Ishii Y, Nomoto A,
Sasaki M, Nomoto K (2009) A high resolution flexible electrophoretic display driven by
OTFTs with inkjet-printed organic semiconductor. SID’09 tech digest, pp 25–27
85. Chang J-F, Sun B, Breiby DW, Nielsen MM, Solling TI, Giles M, McCulloch I, Sirringhaus
H (2004) Enhanced mobility of poly(3-hexylthiophene) transistors by spin-coating from
high-boiling-point solvents. Chem Mater 16:4772–4776
86. Wang G, Swensen J, Moses D, Heeger AJ (2003) Increased mobility from regioregular
poly(3-hexylthiophene) field-effect transistors. J Appl Phys 93(10):6137–6141
87. Pan H, Li Y, Wu Y, Liu P, Ong BS, Zhu S, Xu G (2007) Low-temperature, solution-
processed, high-mobility polymer semiconductors for thin-film transistors. J Am Chem Soc
129:4112–4113
88. Baeg K-J, Khim D, Kim D-Y, Jung S-W, Koo JB, You I-K, Yan H, Facchetti A, Noh Y-Y
(2011) High speeds complementary integrated circuits fabricated with all-printed polymeric
semiconductors. J Polym Sci Part B Polym Phys 49:62–67
89. Tan HS, Wang BC, Kamath S, Chua J, Shojaei-Baghini M, Rao VR, Mathews N,
Mhaisalkar SG (2010) Complementary organic circuits using evaporated F16CuPc and
inkjet printing of PQT. IEEE Electron Dev Lett 31(11):1311–1313
90. Burns SE, Cain P, Mills J, Wang J, Sirringhaus H (2003) Inkjet printing of polymer thin-
film transistor circuits. MRS Bull 28(11):829–834
91. Jones BA, Facchetti A, Wasielewski MR, Marks TJ (2007) Tuning orbital energetics in
arylene diimide semiconductors. Materials design for ambient stability of n-type charge
transport. J Am Chem Soc 129:15259–15278
92. Anthopoulos TD, Anyfantis GC, Papavassiliou GC, de Leeuw DM (2007) Air-stable
ambipolar organic transistors. Appl Phys Lett 90:122105
93. Singh SP, Sellinger A, Dodabalapur A (2010) Electron transport in copper phthalocyanines.
J Appl Phys 107:044509
References 401
94. Fujisaki Y, Nakajima Y, Kumaki D, Yamamoto T, Tokito S, Kono T, Nishida J-i,

Yamashita Y (2010) Air-stable n-type organic thin-film transistor array and high gain
complementary inverter on flexible substrate. Appl Phys Lett 97:133303
95. Usta H, Risko C, Wang Z, Huang H, Deliomeroglu MK, Zhukhovitskiy A, Facchetti A,
Marks TJ (2009) Design, Synthesis, and characterization of ladder-type molecules and
polymers. Air-stable, solution-processable n-channel and ambipolar semiconductors for
thin-film transistors via experiment and theory. J Am Chem Soc 131:5586–5608
96. Yan H, Zheng Y, Xia Y, Newman C, Quinn J, Lu S, Facchetti A (2010) Through-breaking
organic TFT materials for active matrix display backplane application. SID’10 tech digest,
pp 1679–1681
97. Kronemeijer AJ, Gili E, Shahid M, Rivnay J, Salleo A, Heeney M, Sirringhaus H (2012) A
selenophene-based low-bandgap donor–acceptor polymer leading to fast ambipolar logic.
Adv Mater 24:1558–1565
98. Meijer EJ, De Leeuw DM, Setayesh S, Van Veenendaal E, Huisma B-H, Blom PWM,
Hummelen JC, Scherf U, Klapwijk TM (2003) Solution-processed ambipolar organic field-
effect transistors and inverters. Nat Mater 2:678–682
99. Fan J, Yuen JD, Wang M, Seifter J, Seo J-H, Mohebbi AR, Zakhidov D, Heeger A, Wudl F
(2012) High-performance ambipolar transistors and inverters from an ultralow bandgap
polymer. Adv Mater 24:2186–2190
100. Kim C, Facchetti A, Marks TJ (2007) Gate dielectric microstructural control of pentacene
film growth mode and field-effect transistor performance. Adv Mater 19:2561–2566
101. Veres J, Ogier S, Lloyd G, de Leeuw D (2004) Gate insulators in organic field-effect
transistors. Chem Mater 16:4543–4555
102. Facchetti A, Yoon M-H, Marks TJ (2005) Gate dielectrics for organic field-effect
transistors: new opportunities for organic electronics. Adv Mater 17:1705–1725
103. Halik M (2006) Gate dielectrics. In: Klauk H (ed) Organic electronics: materials,
manufacturing and applications. Wiley, Weinheim
104. Fukuda K, Hamamoto T, Yokota T, Sekitani T, Zschieschang U, Klauk H, Someya T (2009)
Effects of the alkyl chain length in phosphonic acid self-assembled monolayer gate
dielectrics on the performance and stability of low-voltage organic thin-film transistors.
105. Veres J, Ogier SD, Leeming SW, Cupertino DC, Mohialdin-Khaffaf S (2003) Low-k
insulators as the choice of dielectrics in organic field-effect transistors. Adv Funct Mater
13:199–204
106. Tan HS, Mathews N, Cahyadi T, Zhu FR, Mhaisalkar SG (2009) The effect of dielectric
constant on device mobilities of high-performance, flexible organic field effect transistors.
107. Li FM, Nathan A, Wu Y, Ong BS (2008) A comparative study of plasma-enhanced
chemical vapor gate dielectrics for solution-processed polymer thin-film transistor circuit
integration. J Appl Phys 104:124504
108. Stott J, Kumatani A, Minari T, Tsukagoshi K, Heutz S, Aeppli G, Nathan A (2011) Bottom-
contact pentacene thin-film transistors on silicon nitride. IEEE Electron Device Lett ED-
32(9):1305–1307
109. Sekitani T, Someya T (2011) Stretchable and foldable displays using organic transistors
with high mechanical stability. SID’11 tech digest, pp 276–279
110. Suzuki M, Fukagawa H, Nakajima Y, Tsuzuki T, Takei T, Yamamoto T, Tokito S (2008)
5.8-inch phosphorescent color AM-OLED display driven by OTFTs on plastic substrate. In:
Proceedings of the IDW’08, pp 1515–1518
111. Zhang X-H, Tiwari SP, Kim S-J, Kippelen B (2009) Low-voltage pentacene organic field-
effect transistors with high-k HfO2 gate dielectrics and high stability under bias stress. Appl
Phys Lett 95:223302
112. Deng LF, Lai PT, Chen WB, Xu JP, Liu YR, Choi HW, Che CM (2011) Effects of different
annealing gases on pentacene OTFT with HfLaO gate dielectric. IEEE Electron Dev Lett
32(1):93–95
402 10 Organic TFTs
113. Halik M, Klauk H, Zschieschang U, Schmid G, Dehm C, Schutz M, Maisch S, Effenberger

F, Brunnbauer M, Stellacci F (2004) Low-voltage organic transistors with an amorphous
molecular gate dielectric. Nature 431:963–966
114. Hong J-P, Park A-Y, Lee S, Kang J, Shin N, Yoon DY (2008) Tuning of Ag work functions
by self-assembled monolayers of aromatic thiols for an efficient hole injection for solution
processed triisopropylsilylethynyl pentacene organic thin film transistors. Appl Phys Lett
92:143311
115. Burns SE, Reynolds K, Reeves W, Banach M, Brown T, Chalmers K, Cousins N, Etchells
M, Hayton C, Jacobs K, Menon A, Siddique S, Too P, Ramsdale C, Watts J, Cain P, von
Werne T, Mills J, Curling C, Sirringhaus H, Amundson K, McCreary MD (2005) A scalable
manufacturing process for flexible active-matrix e-paper displays. J SID 13(7):583–586
116. Sirringhaus H, Tessler N, Friend RH (1998) Integrated optoelectronic devices based on
conjugated polymers. Science 280:1741–1744
117. Burgi L, Richards TJ, Friend RH, Sirringhaus H (2003) Close look at charge carrier
injection in polymer field-effect transistors. J Appl Phys 94(9):6129–6137
118. Deen MJ, Marinov O, Zschieschang U, Klauk H (2009) Organic thin-film transistors: part
II—parameter extraction. IEEE Trans Electron Dev 56(12):2962–2968
119. Burgi L, Sirringhaus H, Friend RH (2002) Noncontact potentiometry of polymer field-effect
transistors. Appl Phys Lett 80(16):2913–2915
120. Klauk H, Schmid G, Radlik W, Weber W, Zhou L, Sheraw CD, Nichols JA, Jackson TN
(2003) Contact resistance in organic thin film transistors. Solid-State Electron 47:297–301
121. Gundlach DJ, Zhou L, Nichols JA, Jackson TN, Necliudov PV, Shur MS (2006) An
experimental study of contact effects in organic thin film transistors. J Appl Phys
100:024509
122. Kim CH, Bonnassieux Y, Horowitz G (2011) Fundamental benefits of the staggered
geometry for organic field-effect transistors. IEEE Electron Dev Lett 32(9):1302–1304
123. Nichols JA, Gundlach DJ, Jackson TN (2003) Potential imaging of pentacene organic thin-
film transistors. Appl Phys Lett 83(12):2366–2368
124. Li T, Ruden PP, Campbell IH, Smith DL (2003) Investigation of bottom-contact organic
field effect transistors by two-dimensional device modelling. J Appl Phys 93(7):4017–4022
125. Wang SD, Yan Y, Tsukagoshi K (2010) Understanding contact behavior in organic thin film
transistors. Appl Phys Lett 97:063307
126. Richards TJ, Sirringhaus H (2007) Analysis of the contact resistance in staggered, top-gate
organic field-effect transistors. J Appl Phys 102:094510
127. Pesavento PV, Puntambekar KP, Frisbie CD, McKeen JC, Ruden PP (2006) 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
128. Necliudov PV, Shur MS, Gundlach DJ, Jackson TN (2003) Contact resistance extraction in
pentacene thin film transistors. Solid-State Electron 47:259–262
129. Hill IG (2005) Numerical simulations of contact resistance in organic thin-film transistors.
130. Vinciguerra V, La Rosa M, Nicolosi D, Sicurella G, Occhipinti L (2009) Modeling the gate
bias dependence of contact resistance in staggered polycrystalline organic thin film
transistors. Org Electron 10:1074–1081
131. Brondijk JJ, Torricelli F, Smits ECP, Blom PWM, de Leeuw DM (2012) Gate-bias assisted
charge injection in organic field-effect transistors. Org Electron 13:1526–1531
132. Rapisarda M, Valletta A, Daami A, Jacob S, Benwadih M, Coppard R, Fortunato G,
Mariucci L (2012) Analysis of contact effects in fully printed p-channel organic thin film
133. Sze SM, Ng KK (2007) Metal-semiconductor contacts. In: Physics of semiconductor
devices, 3rd edn. Wiley, New York
quality semiconductors. IEEE Trans Electron Dev ED-54(2):354–358
References 403
135. Balon F, Shannon JM (2006) Analysis of Schottky barrier source-gated transistors in a-Si:H.
Solid-State Electron 50:378–383
136. Guo X, Balon F, Hatton RA, Shannon JM (2008) High performance transistors in low
mobility organic semiconductors for analog and high-frequency applications. In: IEEE
flexible electronics and displays conference and exhibition. Phoenix, pp 1–5
137. Sirringhaus H (2009) Reliability of organic field-effect transistors. Adv Mater
21:3859–3873
138. Lee CA, Park D-W, Jung K-D, Kim B-j, Kim YC, Lee JD, Park B-G (2006) Hysteresis
mechanism in pentacene thin-film transistors with poly(4-vinyl phenol) gate insulator. Appl
Phys Lett 89:262120
139. Kim TH, Song CK, Park JS, Suh MC (2007) Constant bias stress effects on threshold
voltage of pentacene thin-film transistors employing polyvinylphenol gate dielectric. IEEE
140. Sharma A, Mathijssen SGJ, Smits ECP, Kemerink M, de Leeuw DM, Bobbert PA (2010)
Proton migration mechanism for operational instabilities in organic field-effect transistors.
Phys Rev B 82:075322
141. Sharma A, Mathijssen SGJ, Bobbert PA, de Leeuw DM (2011) Influence of the
semiconductor oxidation potential on the operational stability of organic field-effect
transistors. Appl Phys Lett 99:103302
142. Lee B, Wan A, Mastrogiovanni D, Anthony JE, Garfunkel E, Podzorov V (2010) Origin of
the bias stress instability in single-crystal organic field-effect transistors. Phys Rev B
82:085302
143. Salleo A, Endicott F, Street RA (2005) Reversible and irreversible trapping at room
temperature in poly(thiophene) thin-film transistors. Appl Phys Lett 86:263505
144. Bobbert PA, Sharma A, Mathijssen SGJ, Kemerink M, de Leeuw DM (2012) Operational
stability of organic field-effect transistors. Adv Mater 24:1146–1158
145. Lee CA, Park DW, Jin SH, Park IH, Lee JD, Park B-G (2006) Hysteresis mechanism and
reduction method in the bottom-contact pentacene thin-film transistors with cross-linked
poly(vinyl alcohol) gate insulator. Appl Phys Lett 88:252102
146. Choi S-J, Kim J-H, Lee HH (2009) Deep-UV Curing of Poly(4-Vinyl Phenol) Gate
dielectric for hysteresis-free organic thin-film transistors. IEEE Electron Dev Lett
30(5):454–456
147. Jung T, Dodabalapur A, Wenz R, Mohapatra S (2005) Moisture induced surface
polarization in a poly(4-vinyl phenol) dielectric in an organic thin-film transistor. Appl
Phys Lett 87:182109
148. Kalb WL, Mathis T, Haas S, Stassen AF, Batlogg B (2007) Organic small molecule field-
effect transistors with CytopTM gate dielectric: eliminating gate bias stress effects. Appl
Phys Lett 90:092104
Part III
Novel Substrates and Devices
Chapter 11
TFTs on Flexible Substrates
Abstract The flat panel display industry has been built around TFTs on rigid
glass substrates, but there are well-identified applications requiring flexible sub-
strates. This chapter summarises the properties of common plastic substrate
materials, and discusses the issues of implementing TFT fabrication schedules on
them. It looks in detail at the strategies which have been developed in order to
fabricate a-Si:H, poly-Si and organic TFTs on flexible substrates. These have
included direct fabrication on the plastic substrates at reduced temperatures, as
well as carrier plate and transfer plate processing. For carrier plate processing, the
plastic substrates are temporarily bonded to glass carrier plates during processing,
and, at its completion, the plastic substrate, plus its TFT layers, are detached from
the glass. For the transfer process, the TFT layers themselves are detached from
the glass, and bonded to a separate plastic substrate. A third approach has been to
use an alternative flexible substrate, which is easier to handle, such as thin foils of
stainless steel. In addition to these technological considerations, the mechanics of
bending and strain in flexible substrates is summarised.
11.1 Introduction
Chapter 5 has discussed the processing of a-Si:H TFTs on glass substrates, and this
is the current, dominant active matrix display technology, irrespective of whether
it is for LCD, or electrophoretic e-reader displays. However, there are obvious
mechanical limitations in displays made on glass, namely their rigidity and
potential fragility. This is particularly true of small portable displays, where
mechanical robustness is essential, and, for many of these displays, it is also
desirable to minimise their weight and size. The weight reduction may be achieved
through the use of thinner, lower density substrates, such as plastic, and the
robustness and reduced size can be achieved with increased substrate flexibility, so

408 11 TFTs on Flexible Substrates
that, for instance, the display may be rolled or folded, when not in use. There are
also other applications where the display needs to be conformable to fit, for
instance, the contours of a car’s dashboard, or to be contoured to be worn on a
wrist. These requirements can, in principle, be met with flexible substrates, pro-
viding these substrates can be used to make displays, which match the quality of
the glass-based displays, and, ideally, at the same or at a reduced cost.
The same arguments can also be applied to poly-Si TFTs, particularly as their
major application is in small/medium diagonal portable displays, including both
AMLCDs and AMOLEDs, where the need for reduced weight, flexibility and
robustness is greatest.
In view of the enormous investment in a-Si:H TFT technology, and its wide-
spread application as the dominant active matrix display technology, it is under-
standable that there are strong technical and economic reasons to extend this
technology, where possible, to new applications. Equally, these arguments can be
applied to poly-Si technology, as it currently commands a large fraction of the
small diagonal, portable display market. However, as shown in the following
sections, the application of the relatively high temperature glass-based processes,
to lower temperature plastic substrates, presents many challenges, which are less
severe with organic TFTs. Nevertheless, the greater background understanding of
the inorganic TFT technologies means that many of today’s more sophisticated
demonstrator displays and circuits, on flexible substrates, have been made using
these devices. Hence, the emphasis in this chapter is on the implementation of
these technologies on plastic substrates, with a briefer overview of the work on
organic TFTs.
In this chapter, the properties and handling issues of plastic substrates are
covered, before looking in detail at the implementation of a-Si:H, poly-Si and
organic TFT technologies on flexible substrates. In addition, the impact of sub-
strate flexing on the electrical performance of these TFTs is presented.
To a large extent, the current technological developments have been an
extension of the glass-based technology to flexible substrates, and many of the
more successful processes have used a hybrid glass/plastic approach in delivering
a final plastic-based demonstrator. However, for future low-cost fabrication of
flexible devices [1] it is likely that quite different processing technologies will be
required, particularly where they can exploit the unique features of flexible sub-
strates, such as roll-to-roll (R2R) processing [1–3]. Whilst aspects of R2R have
been demonstrated with a-Si:H TFTs [2], it is better suited to a solution-based
organic TFT technology, as this is more compatible with an all-printed process [4].
However, the implementation of an all-printed R2R technology still presents many
challenges.
Section 11.2 summarises the properties of the commonly used plastic sub-
strates, and gives an overview of the issues associated with handling them. Sec-
tion 11.3 discusses the effects of strain on the deposited thin films when the
flexible substrate is bent. The strategies for processing a-Si:H, poly-Si and organic
TFTs on flexible substrates are described in Sects. 11.4, 11.5, and 11.6, respec-
tively. These sections also summarise the effects of uniaxial strain on the device
characteristics. Finally, the issues of power dissipation in the TFT, leading to its
self-heating, and to the subsequent heating of the plastic substrate, are briefly
reviewed in Sect. 11.7.
11.2 Substrate Handling
In terms of flexibility and material cost, plastic substrates look very attractive, and
have been the subject of considerable research, which has identified and addressed
many of the problems involved in shifting from glass to plastic substrates. Some of
the issues are related to the differences in thermal and mechanical properties
between TFT materials, glass and plastics, and these are summarised in
Table 11.1. This table lists key properties of many of the most widely studied
plastics, and the corresponding values are also listed for the TFT materials, as well
as for glass and steel foil substrates. The plastics are polyimide (PI), aromatic
fluorine-containing polyarylates (PAR), polyethersulphone (PES), polyethylene
naphthalate (PEN), and polyethylene terephthalate (PET) [5]. The second column
lists the maximum processing temperatures for these materials, where the glass
transition temperature has been used for the plastics [5], and it will be seen that, for
most of the plastics, the maximum permitted processing temperature is well below
that for glass substrates. This has the direct consequence that the a-Si :H and poly-
Si TFT processing schedules, developed for glass substrates, cannot be directly
implemented on plastic substrates, and, therefore, new lower temperature sched-
ules were needed. The two possible exceptions to this are PI and PAR; however, PI
is not transparent, and would not be suitable for a back-lit AMLCD display.
Nevertheless, it could be used with reflective or emissive displays, and PI sub-
strates are currently used in the development of flexible reflective LC, top-emitting
AMOLED and electrophoretic e-reader displays. The issue with PAR is its
Table 11.1 Properties of TFT substrates and layers (*—Glass strain temperature; HS—heat
stabilised; many of the figures are illustrative, and will show some variation with substrate
preparation technique)
Material Tmax CTE, a Y m Transparency WVTR
( C) (ppm/K) (GPa) (%) (g.m-2day-1)
Glass 666* 3.2 71 0.23 90 0
PI (Kapton) 350 17 2.5 0.32 30–60 (yellow)
PAR 340 53 2.9 90
PES 223 54 2.2 90 80
PEN (HS) 150 (200) 13 6.1 [85 2
PET (HS) 80 (150) 15 5.3 [85 9
Steel [1,000 14 200 0 0
SiN [1,000 2.7 210 0.25 0
SiO2 [1,000 0.5 70 0
Si [1,000 2.6 130 0.28 0
coefficient of thermal expansion, CTE, and this general issue is discussed in the
next paragraph.
The third column in the table lists the coefficients of thermal expansion, a, and,
for the plastics, these are all much larger than the values for glass and for the
inorganic TFT materials. Following the deposition of a TFT layer at a temperature
DT above room temperature, this mismatch between the CTEs of the TFT layers
and the plastic substrates will induce room temperature strain, eM, in the system,
given by [6]:
eM ¼ ðaf as ÞDT ð11:1Þ
where the sub-scripts ‘s’ and ‘f’ refer to the parameter values for the substrate and
TFT film, respectively.
The strain in the TFT layers is given by [7]:
!
ðaf as ÞDT
ef ¼ ð11:2Þ
Y f df
Y s ds þ 1
(where Y is Young’s modulus, and positive values of ef correspond to tensile strain
in the film, and negative values to compressive strain). Unless Ysds Yfdf, the
denominator is larger than unity, and the strain in the system is not concentrated
just in the TFT film, but is shared with the substrate as well. This is typical for a
compliant substrate with a small value of Ys. So, for the deposition of a thin SiN
layer on one side of a plastic substrate, given the CTE differences in Table 11.1, at
room temperature the substrate will be constrained from shrinking as much as a
free substrate would, and it will be under tensile stress, while the SiN film will be
shrunk more than a free film, and it will be under compressive stress. These CTE-
induced stresses will cause the structure to distort into a cylindrical shape, with the
compressive film on the outer surface of the cylinder. (In contrast, with
mechanically induced stress, where the structure is wrapped around a cylinder, the
outer surface will be under tension). The detailed relationship between process-
induced strain and the radius of the resulting cylinder is discussed further below.
For a rigid substrate, such as glass, with Ysds [ Yfdf, the strain is largely con-
tained within the TFT film, and equation (11.2) reduces to:
ef ¼ ðaf as ÞDT ð11:3Þ
The strain in equations (11.2) or (11.3) will have two possible consequences.
Firstly, if the strain in the film is too high it will crack, and a rule of thumb,
suggesting that the strain should be kept below 0.3 %, can further limit DT, and,
thereby, limit the maximum processing temperature [8]. This is shown in
Fig. 11.1, in which the maximum thickness of silicon nitride, before it cracks, is
plotted against the nitride deposition temperature for two engineered, 300 C
plastics having CTE values of 45 ppm/K and \ 10 ppm/K. As will be seen, the
maximum nitride thickness, at a deposition temperature of 250 C, was only
200 nm for the nitride on the high CTE substrate [9]. As this may be thinner than
11.2 Substrate Handling 411
Fig. 11.1 Thickness of

cracked (square) and crack-
free (filled square) SiNx films
deposited over a range of
temperatures on engineered
polymer substrates
(CTE = 45 ppm/K, and
\10 ppm/K for film B).
permission of SID)
needed for either substrate capping or for the gate insulator layer, the maximum
processing temperature must be kept below this temperature if thicker films are
required. PAR, with a CTE of 53 ppm/K, will be similarly constrained, and using
the rule of thumb of 0.3 % maximum strain, it was recommended that the maxi-
mum processing temperature should be kept below 220 C, which is significantly
lower than its glass transition temperature of 340 C [10].
As mentioned above, with a compliant plastic substrate, having a small
Young’s modulus compared with the thin film’s value, the substrate will buckle
into a cylindrical shape, with a radius of curvature, R, given by [6]:
" #
ds ð1 jg2 Þ2 þ 4jgð1 þ gÞ2
R¼ ð11:4Þ
6ð1 þ mÞeM jg 1þg

where, g ¼ df ds ; j ¼ Yf Ys and d and Y are the thickness and Young’s modulus
values, respectively, of the thin film and the substrate, and m is Poisson’s ratio
(which is assumed to be the same for the film and the substrate). For a rigid
substrate, such as glass, where Yf & Ys, and for g \ 0.1, the above equation
simplifies to the Stoney formula, giving spherical rather than cylindrical distortion
[6]:

ds
R¼ ð11:5Þ
6eM jg
The normalised radius of curvature is plotted against the ratio of film thickness
to substrate thickness in Fig. 11.2, for the two situations of a compliant substrate
(j = 100), and a closely matched film and substrate (j = 1) [6]. (These two cases
were chosen to represent a TFT film on plastic or on a steel substrate, respectively.
As will be seen below, the use of thin steel foils is one strategy chosen for the
fabrication of TFTs on flexible substrates [6]). Figure 11.2 shows that for a typical
case of 1 lm thick TFT layers on 100 lm thick flexible substrates, the plastic
substrate develops more than ten times greater curvature than the steel substrate.
Also, if the j = 1 curve is used as an approximate description of a 1 lm thick TFT
Fig. 11.2 Normalized

radius of curvature calculated
as a function of the film/
substrate thickness ratio,
using eqs. (11.4) (solid lines)
and (11.5) (broken lines).
Physics)
layer on 500 lm thick glass substrate, the curvature of the thinner plastic substrate
is *100 times greater. The curvature in the plastic substrates will need to be
handled during processing either by bonding the substrates to a rigid carrier or by
using frames to flatten them [10]; either way, this will add extra complication to
the processing schedule. Whilst the curvature can be minimised by coating
identical capping layers on both sides of the flexible substrate, and, thereby,
equalising the surface strains, it will not be practical, nor cost effective, to
duplicate all TFT layers on both sides of the substrate. Hence, there will always be
CTE-induced bending strain due, for instance, to the SiNx gate dielectric layer in
the TFT. In addition to the unbalanced CTE strains, there may also be residual
bending strains in the as-delivered plastic substrates, as well as built-in strain in the
films themselves due to the deposition processes [14], all leading to substrate
handling issues due to its curvature.
To reduce the CTE mismatch problem, some proprietary, transparent plastic
substrates have been developed, in which the CTE values have been engineered to
be a closer match to those of the TFT layers, by being less than 10 ppm/K in one
case [9], and *14 ppm/K in another [11]. The first of these substrates also has a
maximum processing temperature of C300 C, so that it is also a better match to
standard a-Si:H TFT processing schedules [9].
The final column in Table 11.1 lists the water vapour transmission rates,
WVTR, in g/m2/day for the plastic substrates [11], and the high values present
another handling issue. This is because water absorption, as well as oxygen
absorption, leads to a swelling of the plastic, which compromises its dimensional
stability. For photolithography, good dimensional stability is essential for the
correct registration of successive mask patterns. In addition, low WVTR values of
10-1–10-2 g/m2/day are needed for LCDs and considerably lower values are
needed for encapsulation of AMOLEDs [11]. This problem can be addressed either
through impermeable, hard surface coatings deposited by the substrate manufac-
turer, or by low temperature deposition of SiNx and/or SiO2 layers using either
sputtering or PECVD. For instance, a PES film coated with a UV-cured acrylate
film plus a sputter deposited SiOx film reduced its WTVR from *50 to 0.01 g/m2/
day [11]. The coatings also serve to protect the substrates from the potentially
11.2 Substrate Handling 413
Fig. 11.3 Shrinkage rate

measurements of various
plastic substrates at 150 or
200 C. (Reprinted from [12]
harmful chemicals, which would be used during the device patterning processes.
Given the crucial role played by these encapsulation films, they need to be of high
quality and pin-hole free.
Finally, the as-delivered plastic films show a tendency to shrink at typical
processing temperatures. To maintain dimensional stability throughout the TFT
fabrication process, they need to be pre-shrunk, prior to device processing, by
annealing them at the intended processing temperature [10, 12]. Shrinkage rate
data for many of the common plastics are shown in Fig. 11.3 [12], and this rate
parameter can be readily used to match the in-process shrinkage of the film to the
required dimensional stability of the substrate. For instance, if a 30 9 30 cm
polymer substrate is being processed, and the required alignment tolerance is
3 lm, then the shrinkage of the plate between first and last photolithography stage
must be less than 3 lm, or less than 10 ppm. If the total high temperature cycle
time after the first lithography stage is 2 h, then the substrate needs to be annealed
until its shrinkage rate is \5 ppm/h. From the data in Fig. 11.3, those samples
needed to be pre-shrunk for *10–100 h to achieve this degree of stability.
11.3 Substrate Bending
For displays having an application dependent upon their flexibility, it is important

to establish the effect of strain on the performance of the TFT, and, equally, to
determine the critical strain at which the device fails. This is usually done by
measuring device performance as a function of tensile and compressive strain, by
bending the processed substrates around preformed cylinders of defined radius, as
shown in Fig. 11.4 [13]. By varying the radius of curvature, R, the strain can be
varied, where the relationship between radius and bending-induced strain at the top
of the film is given by [6]:
Fig. 11.4 Photographs of steel foils wrapped around cylinders to measure the effects of uniaxial
strain on TFT behaviour: (a) tensile, and (b) compressive strain. (Reprinted with permission from
[13]. Copyright (2009) American Institute of Physics)

df þ ds 1 þ 2g þ jg2
ef ¼ ð11:6Þ
2R ð1 þ gÞð1 þ jgÞ
For externally applied bending strains, if the film is the outer surface of the
cylindrical shape it will be under tensile strain.
The normalised surface strain is plotted in Fig. 11.5 as a function of the relative
thickness of the film and substrate, for two different ratios of Young’s modulus
corresponding to well matched elastic properties of the film and substrate (j = 1)
and compliant (j = 100) substrates. For the former, the unity value of the nor-
malised strain corresponds to:

df þ ds
ef ¼ ð11:7Þ
2R
and this describes the situation in which the neutral plane is in the centre of the
film-substrate bilayer, and the strain in the film increases linearly with the
Fig. 11.5 Normalized strain

in a film calculated as a
function of the film/substrate
thickness ratio. Two
substrates are illustrated: steel
(Yf/Ys & 1) and plastic
(Yf/Ys & 100). (Reprinted
11.3 Substrate Bending 415
thickness of both materials, and scales inversely with the radius of curvature. This
simple situation, of the neutral plane being in the centre of the bilayer, is only
approached with the compliant substrate when there is a gross mismatch in the
thickness of the film and the substrate, such that g \ 10-3 or g [ 10, and, in these
two cases, one material or the other dominates the bending induced strain. For the
more typical case of g * 10-2, the neutral plane moves from the centre of the
compliant substrate towards the stiffer TFT film, thereby reducing the strain in the
film by approximately a factor of two compared with the j = 1 case.
From the mechanical stressing of a-Si:H TFTs into tension and compression,
the safe strain range for reversible elastic distortion was established, together with
the threshold strain beyond which plastic failure occurred. This is shown in
Fig. 11.6 [14], and failure under tensile strain occurred at 0.5 % due to crack
propagation, whereas, under compressive strain, the failure mode was buckling
and de-lamination at the higher strain of 2 %. The failure under tension has been
attributed to the propagation of a crack originating at a flaw, where the propagation
occurs when the driving force, G, is greater than the surface energy, C, of the
newly formed crack [15]:
ð1 m2f Þr2f df
G¼b [C ð11:8Þ
Yf
rf is the stress in the film, and b is a dimensionless constant dependent upon the
elastic properties of the film and substrate, and, for a compliant substrate, b is
much larger than one, leading to easy crack propagation [15]. De-lamination
occurs when a defective region at the film/substrate interface results in an unb-
onded area, which is large enough to permit the film to buckle. Once that happens,
the internal stresses can propagate the unbonded area like a crack, and the critical
unbonded length for buckling, lc, is given by [15]:

pdf Yf 0:5
lc ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð11:9Þ
3ð1 m2f Þ rf
Fig. 11.6 Overview of the response of a-Si:H TFTs to mechanical strain. (Reproduced from
[14] with permission of Wiley, Inc)
(where rf has a negative value in compression). Given that the compressive failure
strain, ef, in the film was [2 % [14, 15], the corresponding stress can be deter-
mined, giving lc \ 10df * 10 lm. This is an unusually large defective area in a
high quality structure, explaining why the films were so stable under compressive
strain [15].
Further details of the response of a-Si:H TFTs to mechanical strain are pre-
sented in Sect. 11.4.1.1, in which the replacement of the brittle SiNx by a more
resilient hybrid layer of silicon dioxide and silicone polymer (for both substrate
capping and the gate dielectric) improved the tensile and compressive failure
strains to 5 and 2.5 %, respectively [16].
This discussion has focussed on the direct assessment of strain effects in TFTs
on a free-standing polymer substrate. However, in a finished display, the substrate
will be attached to another layer to form the display cell. If the elastic properties of
this top-plane are a good match to the substrate, the TFT layer would be located
near the neutral plane, and would suffer little strain when the whole structure is
flexed. Whilst this will be difficult to readily achieve with the inorganic layers used
in a-Si:H and poly-Si TFTs, it should be easier to achieve with organic TFTs, and
it was shown that the minimum bending radius for pentacene TFTs on PI could be
reduced from 5 mm, for the uncoated substrate, to 0.5 mm when it was coated
with a thickness-matched parylene capping layer [17].
11.4 a-Si:H TFTs on Flexible Substrates
11.4.1 a-Si:H Fabrication Processes
Given the above discussion on the issues of using and handling the compliant
plastic substrates for TFT fabrication, three different strategies have emerged to
deal with these problems. The first has been to take the constraints imposed by the
plastic substrates, and to develop both lower temperature TFT processing, and
substrate handling procedures, in order to implement direct fabrication techniques
on these substrates. The second approach is to use an alternative flexible substrate
to circumvent the issues with handling plastic substrates, and the material of
choice has been polished steel foils. The third strategy has been to maintain as
much as possible of the well-established processing techniques used with glass
substrates, and, to minimise the handling issues of plastic substrates by bonding
them to glass carrier plates during processing, and then separating them from these
plates at the end of processing. These three approaches are discussed in the fol-
lowing three sub-sections, and an embryonic, but more innovative, roll-to-roll
technology is presented in Sect. 11.4.1.4.
11.4 a-Si:H TFTs on Flexible Substrates 417
11.4.1.1 Direct Processing on Plastic
In the direct fabrication of a-Si:H TFTs on plastic substrates, the major objectives
have been stress control in the films (where the stress arises from both CTE
mismatch, as well as intrinsic stress within the film), and optimisation of device
performance at the reduced deposition temperatures needed when using the low
temperature substrates, such as PES or PEN.
These objectives have been met by modifying the deposition procedures for the
a-Si:H and the SiNx layers. Figure 11.7 shows the variation of strain in SiNx layers
deposited at 150 C onto 50 lm thick Kapton PI substrates, as a function of
deposition power density [18]. In this diagram, the total strain in the film, eM, has
been split into its constituent components due to CTE mismatch, eth, intrinsic
deposition strain, e0, and humidity strain, ech. As might be expected, only the
intrinsic strain varied with deposition power, and it could be changed from tensile
to compressive. Hence, by suitably controlling the deposition power density, the
strain can be tuned to the processing requirements of the substrate. The deposition
power has also been correlated with the gate bias instability in a-Si:H TFTs [19], in
which increasing power density led to improved device stability, and the use of
this control procedure is discussed below in the context of a specifically optimised
process schedule [8].
As stated above, the deposition power used for the SiNx layer can affect the gate
bias stability of the TFT, and, hence, influences the electronic quality of the layer.
The other deposition parameter having an influence upon the quality of the SiNx
layer, as well as upon the a-Si:H layer, is the gas dilution of the reactant gases. The
deposition gases are SiH4 for a-Si:H, and SiH4 plus NH3 for SiNx. For a-Si:H, the
electronic quality of the film has been correlated with a high ratio of SiH bonds to
SiH2 bonds, and the SiH bond density can be increased by diluting the SiH4 with
hydrogen [20, 21]. This has been explained mechanistically by arguing that the
excess hydrogen acts as an etchant during deposition, and preferentially breaks
weak Si–Si bonds to form volatile hydrides. As these weak bonds can form defect
states, their removal should increase the quality and improve the bias-stress
Fig. 11.7 Components of

the total mismatch strain, eM,
in 300 nm thick PECVD
SiNx films deposited at
different RF powers onto
50 lm thick Kapton 200E PI
substrates at 150 C.
permission of SID)
Fig. 11.8 Ratio of

SiH:(SiH ? SiH2) bond
densities in PECVD a-Si:H
films, deposited at different
temperatures, as a function of
the hydrogen dilution of the
silane deposition gas.
permission of SID)
stability of the films [22]. Moreover, as shown by Fig. 11.8, the optimum dilution
ratio increased as the deposition temperature reduced [21]; hence, the a-Si:H
deposition conditions established for use on glass substrates needed to be re-
optimised for use on lower temperature plastic substrates. Similarly, the relative
density of NH and SiH bonds needed to be controlled in lower temperature SiNx
films, to prevent the SiNx becoming either too nitrogen-rich or too silicon-rich, and
this was accomplished by dilution of the reactant gases with either hydrogen,
helium of argon [21]. An example of this is shown in Fig. 11.9, in which the NH/
SiH ratio is plotted as a function of the dilution conditions for the three different
carrier gases. Also shown on this figure is the ratio, labelled HT, achieved by
conventional SiNx deposition at 300 C, and the data point, labelled LT, showing
that the ratio was too nitrogen-rich if the same gas composition was used at
200 C. At this lower temperature, increased gas dilution, with either argon or
hydrogen, was needed to restore the film’s compositional balance. Suitable
Fig. 11.9 Ratio of NH:SiH

bond densities in 200 C
PECVD SiNx films as a
function of the dilution of the
reactant gases SiH4 and NH3
by He, Ar or H2. (Reprinted
SID)
dilution ratios were identified for the deposition of the a-Si:H and SiNx films at
200 C to produce TFTs with electron mobility, threshold voltage, and gate bias
stability comparable to those of the 300 C control TFTs [21].
In the context of the above background considerations, it is apparent that dif-
ferent groups have selectively used elements of those techniques to minimise strain
and to optimise the performance of a-Si:H TFTs on plastic substrates. In many
cases, it has been common practice to deposit SiNx on both sides of the substrate to
equalise bending stresses, although the films themselves remained under stress.
However, the combined CTE mismatch and intrinsic stresses are greatest in un-
patterned films, and can lead to film de-lamination from the substrate [23].
Therefore, one approach has been to pattern the SiNx gate insulator films into the
small areas needed just for the TFTs themselves [23], but this cannot be done to a
layer of SiNx, which is also being used as a barrier film to prevent moisture
absorption into the substrate. Hence, a process has been implemented to circum-
vent these problems [23, 25], by replacing the SiNx barrier films with 1,200 nm
thick layers of a vapour phase polymer consisting of SiOxCyHx, and replacing
most of the gate dielectric of SiNx with a 200 nm thick layer of benzocyclobutene,
BCB. As the CTE of BCB is 43 ppm/K, it is a good match to that of PES, which
was the substrate used in this work. However, to ensure good electronic properties
of the a-Si:H TFT, a thin 50 nm layer of SiNx was retained as the first layer of a
conventional triple stack of SiNx, i a-Si:H (150 nm), and n+ a-Si:H (50 nm) on top
of the thicker BCB gate insulator, and all three layers in the stack were defined
down to individual device islands, as shown by Fig. 11.10. The definition of the
thin SiNx gate insulator layer into individual islands led to a significant relaxation
of the strain within the structure, as shown by the substrate distortion measure-
ments in Fig. 11.11 [24]. In many respects, the TFT fabrication process was
similar to the conventional back-channel-etch process, apart from the extra mask
stage needed to define the SiNx film at the bottom of the triple stack, and the use of
aluminium for metal tracks due to its good ductility [25]. In addition, as the PES
substrate limited processing temperatures to \220 C, the SiNx and a-Si:H
depositions at 150 C had to be optimised to maintain good TFT properties, and
this was done by increasing the dilution of the reactant gases with helium [25].
These TFTs fabricated at 150 C on PES compared well with conventional TFTs
Fig. 11.10 Cross-sectional

view of a back-channel-
etched a-Si:H TFT with a
combined SiNx/BCB gate
insulator, within which the
SiNx had been locally defined
into an island. (Reprinted
IEEE)
Fig. 11.11 Elongation of

the substrate after each
process step for an a-Si:H
TFT on PES with a combined
BCB/SiNx gate insulator.
permission of Wiley, Inc)
fabricated at 300 C on glass, and had an electron mobility of 0.4 cm2/Vs, a

threshold voltage of 0.7 V and an insulator leakage current of \10-13 A. The
transfer characteristics are shown in Fig. 11.12.
Other groups have also looked to replace the stiff SiNx layers, and Han et al.
[16] used a homogeneous hybrid layer of SiO2 and a silicone polymer, which was
deposited at *23 C, for both the gate dielectric and for capping the PI substrate.
As this hybrid layer was far more compliant than the usual SiNx layers used for
these purposes, the finished devices were less susceptible to cracking under tensile
strain, and were able to tolerate *5 % tensile strain, which was ten times more
than the standard SiNx-based TFTs shown in Fig. 11.6.
An alternative approach has been to use the PECVD deposition power density
to engineer the strain throughout the structure, particularly within the SiNx layers
used to coat the clear plastic substrate, whilst deliberately using a high power
deposition for the gate insulator SiNx, in order to maintain good gate bias stress
stability in the finished TFTs [8]. Figure 11.13 shows a cross-section of an etch-
stop TFT, in which the back-face and top-face layers of SiNx, used to cap the
75 lm thick plastic substrate, were deposited with compressive stress (at high
Fig. 11.12 Transfer

characteristics of an a-Si:H
TFT fabricated on a PES
substrate at 150 C, with a
BCB/SiNx gate insulator.
permission of SID)
Fig. 11.13 Schematic cross-sectional view of an etch-stop a-Si:H TFT, showing the engineered
stresses within each of the layers. (Reprinted from [8] with permission of IEEE)
power density) and tensile stress (at low power density), respectively. In addition,
the high power density gate insulator was necessarily deposited with compressive
stress, and the etch stop layer, which was deposited at lower power, was under
tensile stress (see Fig. 11.7 for the relationship between strain and deposition
power). The engineered substrate used for this work had a working temperature of
C300 C, a CTE of \10 ppm/K, and the depositions were carried out at 280 C,
with the substrate held flat, but not clamped, by a rigid frame. At the end of
processing, the samples had a final anneal in air for 30 min [8].
The process delivered high quality TFTs, with an electron mobility of 0.96 cm2/
Vs, a threshold voltage of *3.5 V and an on/off current ratio of [107, which was
almost identical to TFTs processed on glass at the same temperature [8]. Of
particular concern to the authors was the gate bias instability of these TFTs, as they
had previously demonstrated that with reducing deposition temperature there was
increased instability. A comparison of the threshold voltage shifts under gate bias
stress is shown in Fig. 11.14, for TFT processing at 150, 250 and 300 C on plastic
and glass substrates [8]. Both the a-Si:H and SiNx films deposited at 150 C had
hydrogen diluted reactant gases, whereas the higher temperature films did not [19].
As expected, the higher deposition temperatures improved device stability, but was
not good enough for using these devices as pixel driver TFTs in AMOLEDs [22].
In subsequent work, the authors improved device stability by using 10:1 hydrogen
dilution during the a-Si:H deposition to reduce the gate-bias-induced defect gen-
eration in the a-Si:H, and further optimised the SiNx deposition, so that the gate
bias stability obtained with 300 C processing on the low CTE, clear plastic
substrate exceeded the industry standard of a-Si:H TFTs on glass [22]. These TFTs
had an electron mobility of 0.8–1 cm2/Vs, a threshold voltage of 1.5–2.5 V, and an
off-state current of *10-13A, all of which were similar to control devices on
glass. Figure 11.15 shows the fractional change in TFT drain current as a function
of time, under combined gate (7.5 V) and drain bias (12 V) stress, with the
Fig. 11.14 Gate-bias

stability measurements on ES
a-Si:H TFTs fabricated at
different temperatures on
glass and plastic substrates.
permission of IEEE)

bias-stress stability
measurements, for different
a-Si:H TFT technologies,
showing the reduction in
normalised saturation current
as a function of stress time,
after stressing at Vg = 7.5 V
and Vd = 12 V. (P.U. is
Princeton University).
(Reprinted from [22])
improved TFT lifetime exceeding that of the industry standard process, and being
suitable for use in AMOLED displays [22].
11.4.1.2 Steel Foil Substrates
Steel substrates have good tolerance to high temperature processing, and, by being
impermeable and stiff, they are less susceptible to the dimensional changes
induced in plastic substrates by moisture absorption and the deposition of strained
TFT layers with a large CTE mismatch. However, the as-delivered steel foil
surface can be very rough, with greater than 1 lm peak-to-peak surface roughness
reported [26], which needs to be reduced, before device processing, either by
surface polishing and/or by surface planarisation. There is also a need for a thick
capping layer on the steel, in order to reduce the capacitive coupling between the
TFTs and the conducting substrate [27]. Hence, the surface capping layer has a
double purpose, and, in the use of steel foils to investigate pixel design for flexible
AMOLEDs, 75 lm thick, 5 cm 9 5 cm steel foils were coated with 1.6 lm of a
spin-on glass followed by 600 nm of PECVD SiNx deposited at 280 C [26]. After
deposition and patterning of Cr gates, a conventional triple stack of SiNx, i a-Si:H,
and n+ a-Si:H was deposited at 280 C (SiNx) and 230 C (a-Si:H). These layers
were processed into back-channel-etched TFTs, yielding an electron mobility of
0.5 cm2/Vs, a threshold voltage of 2 V, and an on:off current ratio of 107. The bias
stress stability of these TFTs was not specified.
Other work on steel substrates has also sought to make flexible displays, but
minimised the handling problems of flexible substrates by temporarily bonding
them to rigid carriers during processing [28, 29]. In both cases, the temporary
adhesive limited the processing temperature to 180 C or lower. In one case [28],
100 lm thick steel foils were bonded to rigid carriers, and spin-coated with a 2 lm
thick planarising film, followed by 300 nm of SiNx deposited at 180 C by
PECVD. After Mo gate deposition and patterning, a triple stack of SiNx (300 nm),
i a-Si:H (80 nm) and SiNx (100 nm) were deposited as the first deposition stage in
an etch-stop TFT process. The finished TFTs had an electron mobility of 0.7 cm2/
Vs and a threshold voltage of 1.6 V, but the stability of these low temperature
TFTs was not reported.
In the other process [29], 76 lm thick steel foils were attached to a glass carrier
plate with a weak adhesive, and chemically polished before planarising them with
a 3 lm thick layer of polymer resin, which was then capped with 0.4 lm of
150 C PECVD SiNx. The TFTs were processed through a conventional back-etch
process, albeit with the PECVD depositions at 150 C, and the resulting TFTs had
an electron mobility of 0.35 cm2/Vs and a threshold voltage of 1.47 V. Gate bias
stressing at 30 V for 1000 s at 60 C, gave a threshold voltage shift of 2.6 V,
which is qualitatively consistent with the conclusions drawn in other work about
the increasing degree of gate bias instability as the PECVD deposition temperature
is reduced [8]. However, the authors argued that in applying these devices on steel
to flexible AMOLEDs, the TFT stability problems could be mitigated by using
phosphorescent OLEDs, whose increased efficiency would permit the TFTs to
operate at lower values of VG [29].
11.4.1.3 Carrier Plate Technology
As has been discussed in the preceding two sections, there are considerable
problems to be overcome in the direct fabrication of TFTs on flexible substrates,
which has included not just the substrate handling issues, but also the impact of
reduced deposition temperatures on the bias-stress stability of the TFTs them-
selves. The handling issues can be minimised if all processing is carried out on the
glass substrates familiar to the AMLCD industry, and device performance issues
can also be diminished if conventional processing temperatures are used. One such
approach offering all these options is the EPLaR process (electronics on plastic by
laser release) [30, 31].
The EPLaR process is implemented on an industry standard glass substrate, but,
before processing, a 10 lm thick polyimide film is spun onto the glass and cured at
350 C to fully polymerise the film. The PI is then coated with a PECVD layer of
SiNx to protect it against the chemicals used in the subsequent TFT process. The
cured PI film can withstand processing temperatures up to 350 C, and can,
therefore, be passed through the standard a-Si:H TFT process [30]. At the end of
processing, laser exposure, through the back-face of the glass plate, is used to
release the PI film, and its TFT layers, from the substrate, yielding an active matrix
of TFTs on a flexible plastic substrate. Moreover, as the PI film only increased the
weight of the glass substrate by \1 % and its thickness by *1.4 % [31], the
standard robotic handling tools used in an a-Si:H TFT production facility can be
used with the EPLaR substrates. This means that the EPLaR process can be
introduced into a standard line with a minimum of disturbance, which sets it apart
from the other technologies discussed in this and the previous sections. However,
the use of non-transparent PI, as the final display substrate, prevents its use in
transmissive displays, and the current application of EPLaR is for electrophoretic
e-reader displays [31]. A cross-section of the back-channel-etched TFTs made
with this process, prior to release from the glass substrate, is shown in Fig. 11.16,
and an overview of the process flow is shown in Fig. 11.17 [30]. It has been
demonstrated that the performance of the TFTs is unaffected by the laser release
from the glass carrier substrate, and 9.7 inch electrophoretic displays have been
made with this technology [31].
In contrast to the spin-on PI layer used in EPLaR, more conventional plastic
substrates have also been used, and bonded to rigid substrates to avoid the prob-
lems of directly processing flexible substrates. Where low temperature plastics
have been chosen, the maximum processing temperature is still limited by the Tmax
Fig. 11.16 Schematic cross-section of an EPLaR a-Si:H TFT array, shown whilst the PI film is
still anchored to the glass substrate. (Reprinted from [30] with permission of SID)
Fig. 11.17 EPLaR process sequence for making flexible electrophoretic displays. (Reprinted
from [30] with permission of SID)
of the plastic substrate, or by the temperature at which unacceptable stresses build

up in both the films and the adhesive layer during the processing stages [32]. One
example of this approach has used pre-shrunk and SiNx barrier-coated PES [32,
33] and PEN [32, 34] substrates bonded to glass, and processed at 130 and 100 C,
respectively. The bonding, processing and plastic substrate release stages are
shown schematically in Fig. 11.18, for both the TFT substrate and a colour filter
top-plate on plastic for either an LC display or an electrophoretic e-reader display
[35]. For the PES substrate processing, 200 lm thick layers were bonded to glass
with an adhesive, which could be released at the end of processing, either ther-
mally or by UV exposure. After bonding, a further SiNx film was deposited at
130 C, before the conventional triple stack was deposited, also at 130 C, for a
back-channel-etch TFT process. In order to minimise compressive strain, the
layers were deposited at low plasma power [32] (in agreement with other work
[8]), and this low power was also found to improve device performance, whereas
other work reported that high deposition power was necessary to improve bias
stress stability of the SiNx [8]. The resulting 130 C TFTs had an electron mobility
of 0.5 cm2/Vs, a leakage current of 10-13 A, and an on:off current ratio of 107.
Gate bias stressing with Vg = 20 V and Vd = 10 V led to a threshold voltage shift
of *5 V, and *50 % loss of on-current in 5,000 s, which is comparable to the
results presented for 200 C TFTs in Fig. 11.15. A similar process schedule was
used for the 100 C TFTs on PEN substrates, although the a-Si:H deposition was
made at a reduced total gas pressure to improve the device characteristics, and a
further anneal at 100 C was made at the end of processing to reduce the contact
resistance of the n+ layer [32, 34]. These 100 C TFTs also showed reasonable
room temperature characteristics, with an electron mobility of 0.4 cm2/Vs and an
on:off current ration of 107. However, bias stress measurements showed a 5 V shift
in threshold voltage after 3,960 s, and it is clear that these instability effects need
further attention in low temperature TFTs on plastic [32]. However, in spite of the
Fig. 11.18 Process sequence for flexible display fabrication using a carrier plate technology.
device instabilities, 7 inch transmissive AMLCDs, and 14.3 inch electrophoretic

displays were demonstrated on the PES and PEN substrates, respectively.
At the moment, this carrier plate technology, including EPLaR, is the favoured
route for developing flexible backplanes of a-Si:H TFTs, using readily available
plastic substrates, with a number of groups, in addition to those cited above,
demonstrating LCD, electrophoretic and OLED displays [36].
11.4.1.4 Roll-to-Roll (R2R) Processing
Roll-to-roll processing is frequently viewed as the ultimate, low cost route to

producing displays on flexible substrates [3]. The ideal R2R process would involve
continuous movement of the flexible substrate from its host reel through the dif-
ferent layer deposition and definition stages until the final product is gathered on
the collection reel [37]. A detailed discussion of the equipment required, and other
R2R issues, is contained in Ref. [37], although no example is given of a particular
R2R process being implemented. In this section, a recent application of these
principles to the fabrication of a-Si:H TFTs is described [2, 38], using a self-
aligned imprint lithography (SAIL) process. Given the previous discussion on the
dimensional instability of plastic substrates, the authors designed a self-aligned
process to overcome one of the major hurdles associated with handling plastics.
The process imprinted the etch mask on the substrate, so that any deformation in
the plastic, causing potential misalignment between one pattern and the next,
would also deform the mask by the same amount, thereby cancelling out the
pattern shift. The process employed sequential deposition of all layers in the
device structure [38], starting with an adhesion layer, the gate metal, the triple
stack of SiNx, i a-Si:H and n+ micro-crystalline Si to the final metallisation layer
without any intermediate pattern definition. Hence, the deposition process could be
achieved without breaking the vacuum, leading to reduced contamination prob-
lems, and a smaller equipment footprint. Following this, a three-dimensional, UV-
transparent elastomer etch mask, which had been wrapped around a quartz roller,
was imprinted onto the surface of the device layer stack by rolling the quartz
cylinder across the substrate. Whilst the roller was in contact with the sample
surface, UV radiation passed through the cylinder and polymerised the imprinted
material on the substrate surface. The roller plus imprint stamp is shown in
Fig. 11.19a, and an SEM image of a high-aspect ratio and high-resolution
imprinted pattern is shown in Fig. 11.19b. The multiple step heights in this pattern
were used to define the individual layers in the device stack. In other words, the
variable step heights represented the individual mask layers, which would be used
in a conventional photolithography process. The use of the imprint pattern to
define the device structure is illustrated in Fig. 11.20, and Fig. 11.20a shows the
imprinted pattern on top of the device layer stack prior to any etching. Selective
etching was used for the removal of individual layers in the stack, as well as for
etching and removal of the polymer mask itself. The first etch stage is shown in
Fig. 11.20b, in which the total device stack was defined, and Fig. 11.20c shows
lateral undercutting of the gate metal, which removed it from narrow ‘fuse’ areas,
such as the source and drain contact lines, assuring good isolation between the
source/drain and gate metallisation layers. In Fig. 11.20d, the thinnest mask region
had been completely etched away, exposing the stack of layers over the gate
contact pad, and Fig. 11.20e shows the removal of those layers, down to the
Fig. 11.19 a UV-transparent quartz roller with an elastomeric imprint stamp wrapped around it,
and b SEM image of imprinted polymeric mask pattern used in the roll-to-roll SAIL process.
Fig. 11.20 Illustration of SAIL imprint mask and etching sequence to fabricate back-channel
etched a-Si:H TFTs. (Reprinted from [38] with permission of SID)
contact pad itself. The next thinnest portion of the etch mask covered the TFT
channel region, and, as shown by Figs. 11.20f and g, removal of that masking
material permitted the n+ mc-Si, and the source/drain metallisation, to be removed
from the channel region, yielding a conventional back-channel-etched TFT.
Figure 11.20h shows the removal of the last remaining layer of masking material,
and the consequent exposure of the source/drain contact pads.
The process was implemented on a 13 inch wide and 50 lm thick web of
polyimide material, with layer depositions at a maximum temperature of 250 C,
and used dry etching for the removal of the elastomer mask material and the Si-
based layers, and wet etching for the metal regions.
The transfer characteristics, normalised by channel size, are shown in Fig. 11.21,
and reasonable current scaling was obtained from devices with channel lengths from
100 lm down to 1 lm. The electron mobility in the short channel devices was
0.34 cm2/Vs, and the on:off current ratio was 107. Small electrophoretic demon-
stration displays were also made using these roll-to-roll processed substrates [38].
11.4.2 Uniaxial Strain Effects on a-SiH TFTs
The effect of uniaxial strain on a-Si:H TFTs was measured by bending the samples
around cylinders of different diameter, with the strain parallel to the direction of
current flow in the channel. The strain was computed from the bending radius
Fig. 11.21 Channel-

dimension-normalised
transfer characteristics of
different channel length a-
Si:H TFTs, fabricated by the
SAIL process. (Reprinted
SID)
using equation 11.6, and the samples were under tensile strain when the TFT layer
was on the outer surface of a hollow cylinder, and under compressive strain when
it was on the inner surface (see Fig. 11.4).
Consistent effects of strain have been reported on differently prepared samples.
In one case it was measured on etch-stop TFTs prepared at 150 C on 50 lm thick
polyimide foils, which had used PECVD SiNx layers for substrate capping and for
the gate insulator [39]. In the other case, back-channel-etched TFTs were fabri-
cated on 200 lm thick PES substrates, with an organic layer as the major portion
of the gate insulator, topped by a thinner layer of SiNx, which was defined into
small device island areas in order to relieve the overall process-induced strain in
the substrates [40]. The variation of electron mobility with strain is shown in
Fig. 11.22 [39], where the applied strain values were established to lie within the
limits of reversible elastic deformation. Beyond these limits, the samples were
found to display irreversible changes in their characteristics, due to the formation
of stress-induced cracks. Figure 11.22 shows a linear relationship between
mobility and strain, in which compressive strain reduced the mobility and tensile
strain increased it. The constant of proportionality in this data set was 26, and, in
the other work, nearly identical results were found with a constant of propor-
tionality of 28 [40]. The mobility changes occurred within the time taken to begin
the TFT measurements, and there was no dependence of the changes upon the
duration of the bending, up to a maximum measured duration of 40 h; and, once
Fig. 11.22 Normalised

electron mobility changes in
a-Si:H TFTs as a function of
uniaxial bending strain.
Physics)
the strain was released, the mobility reverted to its pre-strain values [39]. Uniaxial
strain has been demonstrated to cause electron mobility changes in crystalline
silicon [41] (which is discussed further in Sect. 11.5.2), but it was argued that those
mechanisms could not apply to a-Si:H TFTs [39]. A tentative physical explanation
for the a-Si:H results was given by noting that the measured electron mobility is
governed by the width of the band-tail states, and that these are determined by the
structural disorder in the material, which will change with strain [39, 40].
For both sets of samples, the off-state current was found to be independent of
strain [39, 40], whilst the threshold voltage had no discernable dependence on
strain, and the sub-threshold slope showed a small increase with compressive
strain [39].
11.5 Poly-Si TFTs on Flexible Substrates
The introductory section on flexible substrates applies equally to poly-Si TFTs, as

to a-Si:H TFTs, and, as discussed, plastic substrates present a number of handling
issues in terms of their low maximum processing temperature, dimensional
instability and propensity to adopt a cylindrical shape in response to accumulated
processing strains. In common with a-Si:H processing, broadly the same three
strategies have been adopted in developing a flexible substrate technology for
poly-Si TFTs. These are:
• A low temperature process matched to the properties of the plastic substrate,
• A higher temperature process, which could be implemented on thin steel foils,
• A transfer process, as distinct from a carrier plate process, whereby the func-
tional TFT layer was removed from a glass substrate at the end of a conventional
fabrication schedule, and transferred to a plastic substrate. This is the easiest
technical option in the sense that the poly-Si on glass technology can be directly
used, but is more expensive due to the additional process stages for the transfer,
plus the extra substrate cost if the substrate cannot be reused.
Each of these three processing strategies is described below, followed by a
further section on the response of poly-Si TFTs to bending strains. In Sect. 11.7,
there is a discussion of some of the broader issues of device and circuit operation
on plastic substrates, particularly for poly-Si TFTs, over and above the previously
discussed handling issues.
11.5 Poly-Si TFTs on Flexible Substrates 431
11.5.1.1 Direct Fabrication on Plastic Substrates
Two approaches can be seen in the poly-Si work: one being to develop an ultra-
low temperature process at \200 C for use on the transparent, lower temperature
substrates such as PEN, PET and PES [42, 46, 47] listed in Table 11.1, and the
other to modify the established poly-Si process for the higher temperature sub-
strates such as PI and PAR [10, 42]. In all cases, substrate preparation by drying,
capping and pre-shrinking (as discussed in Sect. 11.2) was necessary to ensure
dimensional stability of the substrates through successive photo-lithography
stages.
The low temperature tolerance of the plastic substrates meant that only the
excimer laser annealing, ELA, process was suitable for crystallisation of the
precursor film, and the deposition temperature of this film, and of the gate oxide
film, had to be reduced to meet the temperature constraints of the substrate.
Similarly, the thermal budget for the post-metallisation anneal of the TFT, which
is important for the reduction of charges and interface states, as discussed in Sect.
7.3.1, also had to be reduced.
For the precursor a-Si film, both conventional PECVD was used at 180 C or
200 C for PES and PI substrates, respectively [10], and sputter deposition in He
was used for PES substrates [47]. The PECVD films were passed through the
standard top-gated, non-self aligned, NSA, process, described in Sect. 7.3.2.
The NSA architecture was chosen in preference to the SA architecture, because the
latter needs a second laser pass to activate the source and drain dopants, and this
could lead to potentially damaging, direct laser exposure of the plastic substrate
(in regions where the poly-Si had been removed during the poly-Si island defi-
nition stage) [42]. With the NSA architecture, the source and drain dopants were
implanted before crystallisation, and the unpatterned a-Si film protected the
underlying plastic from direct exposure to the laser beam. Prior to ELA, the excess
hydrogen had to be removed from the films, and this was done with a low intensity
(230 mJ/cm2) laser beam, which had a deliberately ramped leading edge for
controlled release of the hydrogen. The film was then crystallised in the normal
way with a second laser pass at 320 mJ/cm2. After TFT island definition, the gate
oxide layers were deposited at 180 C and 250 C for PES and PI substrates,
respectively, but as these deposition temperatures were below the 300 C normally
used, the oxide quality was poorer and the interface state densities were higher
than normal [10, 42]. The thermal desorption spectra, TDS, of these low tem-
perature oxides [42] have already been discussed in Sect. 7.3.1, and, as shown in
Fig. 7.23a, for depositions below 200 C, the oxides showed high values of the
undesired low temperature peaks. Figures 11.23a and b show the changes in
n-channel sub-threshold slope and electron mobility, respectively, as a function of
the post-metallisation anneal time and temperature [10, 12]. Even the higher
quality oxides, deposited at 250 C, required 1,000 min annealing at 250 C for
the electron mobility to reach 80 % of the standard-process mobility. Whereas, for
the oxide deposited at 200 C, and annealed at this temperature, only *22 % of
the standard-process mobility was reached in this time. This situation was also
Fig. 11.23 Changes in

(a) sub-threshold slope and
(b) normalised electron
mobility in poly-Si TFTs with
post-metallisation annealing
at the SiO2 deposition
temperature. (Reprinted from
mirrored by the sub-threshold slope results, and it was concluded that acceptable
TFT performance could be achieved from a 250 C process, with the transfer
characteristics shown in Fig. 11.24. However, even at this temperature, the p–
channel TFTs had larger sub-threshold slopes than the n-channel TFTs, and, at
lower annealing temperatures, acceptable device performance was not obtained
from p-channel TFTs. Only n-channel TFTS were obtained from a 200 C process
[10, 42, 46], and, hence, the CMOS poly-Si process was only compatible with the
higher temperature PAR or PI substrates, and not with the PES substrates.
The reduced temperature oxide deposition process was identified as the pro-
cessing stage limiting device performance, and, as discussed in Sect. 7.3, ECR
Fig. 11.24 Transfer

characteristics of poly-Si
TFTs fabricated on PI
substrates at a maximum
temperature of 250 C.
from [10])
oxides, with good TDS spectra, have been deposited down to room temperature
[44], but they have still needed post-deposition annealing above 350 C to achieve
low interface densities and useful TFT characteristics [43, 44].
For the He-sputtered precursor a-Si films on PES substrates [47], it was reported
that He removal was easier than hydrogen removal, and this was accomplished
with the leading edge of the crystallisation beam [47]. No details were given of the
beam shape, but it must be assumed that it was closer to the raw semi-gaussian
beam than to a top-hat beam. The overall processing of these samples was quite
different from the work described above, and a schematic of the process flow is
shown in Fig. 11.25. The gate oxide was deposited by PECVD at 110 C, and,
after gate metal definition, anisotropic plasma etching was used to remove the gate
oxide from those areas not beneath the gate electrode. This enabled the exposed
source and drain areas to be doped using a plasma-doping procedure in PH3 or
B2H6-containing gases, giving SA n- and p-channel TFTs, respectively. The do-
pants were subsequently activated by ELA, and it was argued that this process also
led to heating of the Si/SiO2 interface, and a consequent reduction of the interface
state density. This 110 C process on PES yielded high quality devices with an
electron mobility of 250 cm2/Vs and a sub-threshold slope of 0.16 V/decade [47].
However, many aspects of this process have not been fully revealed, such as the
very low temperature PECVD oxide deposition process, and the overall quality of
the resulting oxide. Nor have the details of the ELA activation process been
discussed, including its possible impact upon the exposed PES substrate and its
role in improving the Si/SiO2 interface. Without these details, it is difficult to
assess whether this process is readily implementable on PES, or on even lower
temperature substrates, such as PEN.
11.5.1.2 Fabrication on Steel Foils
As discussed in Sect. 11.4.1.2, when compared with plastic substrates, steel foils
are dimensionally stable, can be exposed to elevated temperatures, and are less
susceptible to attack by a range of chemicals. Nevertheless, as-delivered steel
substrates are too rough for direct use and need to be polished to achieve an
acceptable level of surface roughness of a couple of nanometers, they also need to
be thoroughly cleaned before use and capped with a thick enough dielectric layer
to minimise parasitic capacitive coupling between the conducting substrate and the
overlying TFT circuits [27, 48]. In the work quoted below, 125 lm thick polished
foils of 304-type steel were used, with an average surface roughness of 3 nm. Both
surfaces were coated with 3 lm of PECVD SiO2 as both a planarising layer, and as
a protective barrier to prevent metal ion diffusion into the overlying poly-Si TFTs
[48]. This thickness was also necessary to minimise parasitic capacitive coupling
between the conducting substrate and the TFTs [27].
The TFTs were fabricated using a SA process, similar to that described in Sect.
7.4.2 and shown in Fig. 7.34. The a-Si precursor layer was 50 nm thick, and was
crystallised using an SLS process; then a 50 nm thick PECVD gate oxide was
Fig. 11.25 Processing

sequence of SA poly-Si TFTs
on a PES plastic substrate,
showing the crystallisation
and plasma doping stages.
Physics)
deposited from SiH4 and N2O, and a doped poly-Si layer was used as the gate
electrode. The SA source and drain regions were ion doped and activated by
thermal annealing at 650 C. Following silicidation of the doped regions at
400 C, the device was exposed to a hydrogen plasma at 300 C, capped with a
300 nm SiO2 interlayer dielectric, and, after source and drain metallisation, a final
post-metallisation anneal at 300 C completed the process [48]. Low leakage, high
performance, short channel (L = 1 lm) n- and p-channel TFTs, were obtained,
with sub-threshold slopes of 0.22 and 0.16 V/dec, and carrier mobilities of *300
and *150 cm2/Vs, respectively.
As will be appreciated, the high processing temperatures tolerated by the steel
foils permitted the fabrication of TFTs with performance comparable to conven-
tional glass-based processing, and considerably better than has been achieved
directly on plastic substrates. In addition to the TFTs, a variety of demonstration

circuits were fabricated on the foils, including ring oscillators, static shift registers
with buffer output stages, decoders, sample-and-hold circuits and 5-bit digital-to-
analogue converters [48]. In general, satisfactory performance was obtained from
these circuits, although there was a gate-line resistance problem with the ring
oscillators associated with the doped poly-Si gates. This would be resolved with
lower resistance metal gates—but common metals, like aluminium, could not be
used with a 650 C dopant activation process. However, this is an issue with the
specific process implemented on steel substrates in this work, rather than being a
limitation imposed by the substrate choice itself. The ring oscillator circuits were
fabricated with 1 lm and 2 lm channel length TFTs, and, the delay per stage was
0.88 ns and 1.0 ns, respectively. The speed should scale as 1/L2, and, this poor
scaling with channel length suggested that the high frequency performance was
being limited by parasitic effects [48]. It could be that thicker or lower-k dielectric
layers may be needed for substrate capping to reduce the coupling.
11.5.1.3 Transfer Processes
With transfer processes, the basic glass-based processing schedule was retained,
and, at the end of it, the TFT layers were detached from the glass and bonded to a
plastic substrate. Two examples of this are, firstly, SUFTLA (surface free tech-
nology by laser annealing) [49, 50], in which a hydrogen-rich a-Si:H release layer
was interposed between the glass substrate and the TFT layer, and, secondly, a
glass etching procedure, in which the glass substrate was thinned to several tens of
microns [51, 52] before being bonded to a plastic substrate. A third procedure used
a different technique, whereby a polymer layer, which ultimately became the
plastic substrate, was spun onto the glass substrate before TFT processing, and, at
the end of processing, the polymer layer was released from the substrate with the
TFT layers attached [53, 54]. This is conceptually similar to the EPLaR process
[30, 55], described in Sect. 11.4.1.3, for the fabrication of a-Si:H TFTs on plastic
substrates.
The SUFTLA process is shown schematically in Fig. 11.26a–e [50], and, in
Fig. 11.26a, the fully processed TFT plate is shown with the sacrificial layer of
hydrogen-rich a-Si:H, which had been deposited on the glass substrate prior to
TFT processing. Before the removal of the glass substrate, the completed TFTs
were mechanically protected from damage during the removal process by tem-
porarily bonding a glass substrate to the top surface with water-soluble adhesive
(Fig. 11.26b). The back surface of the original substrate was then exposed to an
excimer laser, which released hydrogen from the a-Si:H layer, and detached the
TFT layer from the glass substrate (Fig. 11.26c). The TFT layer was then per-
manently glued to a 400 lm thick plastic substrate (Fig. 11.26d) [49], after which
the temporary protective substrate on top of the TFT layer was removed by dis-
solving the adhesive in water (Fig. 11.26e). The TFT process [49] was a con-
ventional self-aligned fabrication schedule, using a 50 nm pre-cursor a-Si layer,
Fig. 11.26 SUFTLA

process flow at the end of
poly-Si device processing,
showing the transfer stages
from the glass substrate to the
final plastic substrate.
permission of SID)
which was crystallised by ELA, and capped by a 120 nm gate oxide. After defi-
nition of Ta gate electrodes, the source and drain regions were ion doped, and
activated at 300 C. TEOS layers were used as interlayer dielectrics to separate the
gate metallisation level from the source/drain metallisation, and also for final
capping of the finished devices. Small changes in device characteristics were noted
before and after the transfer process, such as electron mobility and threshold
voltage changes of 119–124 cm2/Vs and 3.3–3.9 V, respectively [49]. However,
the transfer process was described as high yield, and many demonstration devices
were produced with this process, including a 0.7 inch AMLCD [49], a 7.1 inch
electrophoretic display, both with integrated drive circuits, and a range of other
electronic circuits including an 8-bit microprocessor [50].
The glass etch transfer process was optimised for application to large
300 mm 9 350 mm substrates, and the final bonding was to an engineered fibrous
Fig. 11.27 Transfer sequence, at the end of poly-Si TFT processing, showing the thinning of the
glass substrate by etching, and the subsequent adhesive-bonding to a plastic substrate. (Reprinted
from [52] with permission of IEEE)
glass-reinforced plastic, FRP, substrate [11]. The substrate was designed to have a
CTE of *14 ppm/K, which is lower than most of the plastics listed in Table 11.1,
and is a better match to the TFT layers. The adhesive attaching the TFT layers to
the FRP substrate was also optimised to reduce de-lamination of the substrate from
the TFT layer under bending strain. The transfer process is shown schematically in
Fig. 11.27a–d [52], and, in common with the SUFTLA process, the fully processed
TFT plate was capped with a protective film, secured by a heat-releasable adhe-
sive, to prevent damage to the TFT layer during the transfer process (Fig. 11.27a).
The glass substrate was then jet etched (Fig. 11.27b) to reduce its thickness to a
few tens of microns in a uniform fashion, and this left behind a smooth etched
surface. The thinned glass was then permanently bonded, using the optimised
adhesive, to a 220 lm thick FRP substrate (Fig 11.27c). Finally, the top protective
film was removed, by heating the structure to 110 C for 90 s to release the
temporary adhesive, and the finished TFT layer on the plastic substrate was
annealed at 150 C to complete the process. Very good reproducibility of the TFT
characteristics was found before and after the transfer process, as shown by the
results in Fig. 11.28.
A potentially simpler process option is one in which the TFTs are fabricated
directly on a thick spun-on layer, of a suitable polymer, on a rigid substrate, and
the double layer of TFTs and polymer film removed from the rigid substrate at the
end of processing. In this case, the resulting TFT layer is automatically on a
flexible substrate, and a separate substrate attachment stage can be avoided. The
implementation of this process [53, 54] used an 8 lm thick spin-on layer of
polyimide on an oxidised silicon substrate. Following thermal curing of the PI
layer at 350 C, it had a CTE of 3 ppm/K, which was much lower than the bulk
PI film in Table 11.1, and was a good match to the silicon substrate. The cured PI
layer was coated with a barrier stack of SiNx and SiO2, prior to the growth of a
70 nm thick a-Si pre-cursor layer, capped by a 25 nm thick heavily phosphorus
doped a-Si layer. A channel-etched, non-self-aligned TFT process was imple-
mented with these layers [45], but, due to the thermal constraints imposed by the
underlying PI layer, the normal 450 C thermal dehydrogenation process could not
be used on the a-Si film. Instead, it was partially dehydrogenated for 18 h at
Fig. 11.28 Poly-Si TFT

transfer characteristics before
(solid line) and after (broken
line) the glass etching transfer
process. (Reprinted from [52]
350 C, before being crystallised with a modified excimer laser beam, having a
ramped leading edge designed for the controlled release of hydrogen from the film
[53]. The gate dielectric was a 150 nm thick layer of ECR-PECVD SiO2 [44]
deposited at room temperature, and, after deposition and definition of the alu-
minium source, drain and gate electrodes, there was a post-metallisation anneal at
350 C. At the end of this process, the poly-Si devices on the 8 lm thick PI layer
were mechanically released from the silicon substrate, although the details of the
release process have not been revealed. A number of demonstration poly-Si cir-
cuits were fabricated on these flexible substrates, including logic gates, ring
oscillators, and amplifiers, as well as an integrated humidity sensor [54].
EPLaR has also been used to make poly-Si TFTs [30], but with a conventional
self-aligned p-channel process, which included laser activation of the source and
drain dopants. However, the PI intermediate layer on the glass substrate was
damaged by this direct exposure to the laser, and an unspecified shielding layer
was interposed between the TFT layers and the PI to protect it. As with the a-Si
TFT implementation of EPLaR, the poly-Si TFT layer on PI was released from the
glass substrate by excimer laser exposure of the PI layer through the glass sub-
strate. The TFT characteristics remained unchanged by the release process.
11.5.2 Uniaxial Strain Effects on Poly-Si TFTs
The effect of strain on poly-Si TFTs has been studied with SLS-crystallised devices,
made on flexible steel substrates, using the process described in Sect. 11.5.1.2
[27, 48]. Small 20 mm 9 35 mm sections of the substrates were cut from the
processed foils and bent around hollow cylinders of different radius [13]. The TFTs
were under tensile strain when bent around the outside of the cylinders, and under
compressive strain when bent around the inner surface, and the dependence of
strain on radius of curvature is given by Eq. 11.6.
Fig. 11.29 Effect of uniaxial tensile strain on n-channel poly-Si TFTs. (Reprinted with
By measuring the TFT characteristics as a function of the bending radius, the

dependence of the various device parameters on uniaxial strain was established
[13]. In all measurements, the strain was parallel to the direction of current flow,
which itself was parallel to the SLS-induced sub-grain boundaries, and the
n-channel characteristics under tensile strain are shown in Fig. 11.29. There was
decreased off-current and increased on-current with increasing strain, where the
increased on-current was caused by an increase in electron mobility. The opposite
trend was found for compressive strain, and the summarised carrier mobility
results are shown in Fig. 11.30. This figure also includes the hole mobility results
derived from p-channel TFTs; as will be seen, there were opposite trends between
hole and electron mobilities. In contrast, the leakage current behaved identically in
both channel types: decreasing with tensile strain and increasing with compressive
strain. In single crystal silicon p-n junction diodes, the leakage current was also
found to increase with compressive stress, which was attributed to stress-induced
decrease of the Si band-gap at the rate of *0.1 meV/MPa [56]; similarly,
MOSFET junction leakage current was found to increase with compressive stress
and to decrease with tensile stress [57]. The leakage-current in poly-Si TFTs
clearly behaves in the same way as in crystalline silicon, and has been similarly
attributed to stress-induced changes in the band-gap [13].
The carrier mobility changes have also been correlated [13] with similar stress-
induced mobility changes in crystalline Si MOSFETs. For p-channel MOSFETs,
compressive stress increased the curvature of the lowest energy valence band sub-
band, reducing the hole effective mass [58], and, hence, increased the hole
mobility. For n-channel MOSFETs, tensile stress led to splitting and repopulation
of degenerate conduction band sub-bands, such that the population in the sub-band
having the low transverse effective mass was enhanced at the expense of the
Fig. 11.30 Effect of

uniaxial tensile (+) and
compressive (-) strain on
normalised hole and electron
mobilities in poly-Si TFTs.
Physics)
sub-band having the higher longitudinal effective mass, thereby increasing the
overall electron mobility [59].
The TFT mobility changes, discussed above, were observed in long grain SLS
poly-Si, and similar changes have also been reported in metal-induced crystallised
poly-Si [60]. In more conventional ELA poly-Si, on thin plastic substrates, no
discernable mobility changes were detected down to a bending radius of 1.3 cm
[53]. However, using Eq. 11.6, the strain in the TFT layer on the 8 lm thick
compliant PI substrate is calculated to be only 0.012 %, which is considerably less
than the values in Fig. 11.30, and this may be why no mobility changes were
observed.
For the other device parameters, the sub-threshold slope in both channel types
increased with tensile stress and decreased with compressive stress, whilst
threshold voltage showed no dependence upon compressive stress, but decreased
with tensile stress [13]. In contrast to the leakage current and carrier mobility
changes, no simple physical models have been advanced to explain these sub-
threshold slope and threshold voltage changes.
11.6 Organic TFTs on Flexible Substrates
In principle, the fabrication of organic TFTs on plastic substrates should present

fewer issues than is the case with inorganic TFTs, particularly in terms of the
temperature constraints imposed by the substrates. As has been discussed above,
these constraints set an upper processing temperature limit, determined by the
substrate’s glass transition temperature, but the actual maximum temperature can
be reduced below this by CTE mismatch considerations. For organic TFTs, there
should be a better CTE match, and device processing itself is conventionally
performed close to room temperature. However, although the upper temperature
constraint is less of an issue, other attributes of plastic substrates, such as their
11.6 Organic TFTs on Flexible Substrates 441
dimensional instability, sensitivity to chemical attack etc., still have to be con-

trolled during processing. To deal with the dimensional instability problem, most
groups pre-shrink the substrate before processing; however, details on subsequent
substrate encapsulation are left unspecified in a number of publications. Also,
some groups have adopted a rigid carrier plate process to reduce the handling
problems of flexible substrates.
As is apparent from Chap. 10, there are a large number of organic transistor
materials, device geometries, and fabrication processes, which have already been
discussed, and those details will not be repeated below. Only those aspects of the
process, specific to fabrication on plastic substrates, will be considered, and the
reader is referred to Chap. 10 for further information.
In the following two sub-sections, the direct fabrication of OTFTs on plastic
substrates, and rigid carrier plate processing are reviewed, and, finally, the impact
of substrate bending on TFT performance is considered.
11.6.1.1 Direct Processing
Where discrete TFTs have been fabricated on plastic substrates, particularly to

examine bending effects, both PI [17] and PEN [61] substrates have been used,
with evaporated gold bottom-gate and top source and drain contact layers. The
gate dielectric was a spin-coated PVP (polyvinylphenol) layer, and shadow mask
evaporation was used for the pentacene TFT body layer. No details were given of
the handling of the substrates. In another example of pentacene TFTs on PEN
substrates [62], the substrate handling was more clearly defined, and followed the
procedures familiar from the discussion of a-Si and poly-Si TFT processing on
plastic substrates. In particular, the substrate was initially annealed at 180 C to
degas and pre-shrink it, and it was then encapsulated with 100 nm thick SiN
encapsulation layers. The bottom-gated and bottom-contacted TFTs had a spin-
coated PVP gate insulator, which was cured at 180 C, and the pentacene layer
was capped with parylene and Cr/Au in order to photolithographically define it. In
order to protect the pentacene from the effects of ambient water vapour and
oxygen, the whole structure was encapsulated with a 1 lm thick spin-coated layer
of photacryl, which was cured at 150 C after photo-patterning.
A full-colour 2.5 inch AMOLED demonstrator was fabricated on a 200 lm
thick PES substrate, which had been annealed at 180 C to shrink it prior to
processing, and an alignment accuracy of a few microns was obtained over the
4 inch substrate [63]. The evaporated pentacene TFTs were bottom-gated and
bottom-contacted, using gold for both, and a 400 nm thick spin-coated organic
PVP-OTS (octadecyltrichlorosilane) layer was used for the gate dielectric, which
was cross-linked at 130 C. The TFT structure was then encapsulated in organic
passivation and planarisation layers prior to the evaporation of the OLED struc-
ture. The all-organic composition of the display contributed to its satisfactory
operation at a bending radius of 20 mm [63].
11.6.1.2 Carrier Plate Processing
A number of groups have used rigid carriers to reduce the handling problems of
plastic substrates, where the carriers have either been glass plates (carrying PET
[64] and PEN [65] films) or silicon wafers used with PI films [66]. Generally, the
plastic substrates were pre-shrunk by baking for 2 h at 120 C and at 150 C for
the 125 lm thick films of PEN and PET, respectively. Inorganic barrier films of
SiN were used in both cases to prevent moisture and oxygen absorption, and,
thereby, improve dimensional stability. The plastic foils were then bonded to the
carrier plates using temporary adhesives. In all cases, a bottom-gated and bottom-
contacted device architecture was used, with pentacene as the semiconducting TFT
material. This process was used to make flexible AMOLED displays [64, 65], and
electrophoretic display and shift register [66] demonstrators.
In contrast to the bottom-gated pentacene TFTs, a production facility has been
established for flexible electrophoretic displays based upon a printing technology,
using top-gated and bottom-contacted polyfluorene-based polymer TFTs [67, 68].
The facility can produce 10.7 inch displays by processing Gen 3.5-sized substrates
(*0.62 9 0.72 m2) of 125 lm thick PET. A major advantage of the printing
process is that it can produce devices without the conventional alignment stages,
and that the position of the printing head can be locally adjusted to compensate for
substrate distortion [67, 68], but the substrates were, nevertheless, temporarily
attached to glass carrier plates during the patterning stages [68]. To minimise
dimensional instability, the substrates were pre-annealed, and a planarising layer
was deposited before the metal source and drain contacts were formed. The device
processing was at, or near, room temperature, using a solution-deposited polymer
dielectric for both the gate insulator and the interlayer insulator (between the gate
electrode and the pixel electrode).
11.6.2 Uniaxial Strain Effects on Organic TFTs
As with the inorganic TFTs, strain effects were assessed by measuring changes in
the TFT characteristics as function of the externally applied bending radius, and
this was then converted into a strain value using equation 11.6. Bending strain
measurements made on pentacene TFTs, fabricated on 125 lm thick PEN sub-
strates, are shown in Fig. 11.31a–d [61]. These were bottom-gated and top-con-
tacted TFTs, with a 900 nm thick PI gate dielectric. The on-current results in
Fig. 11.31a and b were obtained with the strain parallel to current flow, but the
hole mobility results in Fig. 11.31c and d were obtained with the strain both
11.6 Organic TFTs on Flexible Substrates 443
Fig. 11.31 Effect of bending strain on pentacene TFT transfer characteristics under
(a) compression (b) tension, and the effect of bending strain on the hole mobility under
(c) tension and (d) compression (filled and open circles were for current flow parallel and
perpendicular to the strain). (Reprinted with permission from [61]. Copyright (2005) American
parallel and perpendicular to current flow. As will be seen, the on-currents and
hole mobility values increased with compressive strain and decreased with tensile
strain, irrespective of the strain direction. The strain effects were explained in
terms of hopping transport in the TFTs, where the hopping barrier height was
determined by the molecule spacing in the pentacene film, and this decreased
under compressive strain and increased under tensile strain [61]. The strain isot-
ropy was attributed to an inter-grain current flow path in which coupled grains
were randomly oriented.
The authors also monitored the changes in capacitance, C, of the 900 nm thick
PI gate dielectric, and found that it increased by *3 % with 1.5 % tensile strain,
and decreased by *2.2 % with 1.5 % compressive strain (both of which corre-
sponded to a 4 mm bending radius) [61]. This was explained by changes in the
thickness of the gate dielectric caused by the Poisson effect (rather than by changes
in dielectric permittivity), where those orthogonal changes are related to the
bending radius, R, by:
Fig. 11.32 Changes in the on-current in pentacene OTFTs, on 13 lm thick PI substrates, as a

function of the bending radius. The TFTs were capped with parylene passivation layers with
thicknesses of: (a) 13 lm (b) 10 lm and (c) 0 lm. (Reprinted with permission from [17].
1 C=C0 ¼ D=2ð1 mÞR ð11:10Þ

C0 is the unstrained capacitance, D is the substrate thickness, and m is Poisson’s
ratio, which is 0.4 for the PI film [61]. It was argued that these changes, and other
dimensional changes in the TFT geometry, should be taken into account when
calculating the strain-dependent carrier mobility values.
As seen in Fig. 11.31d, device failure occurred at a compressive bending radius
of *5 mm, which generated a compressive strain [1.5 %, and led to buckling of
the gold electrodes [61]. These measurements were made on un-laminated sub-
strates, in which the TFTs on the substrate surface experienced maximum strain
for a given bending radius. In a practical application of TFTs on a flexible sub-
strate, the substrate will be capped by another film, such as an electrophoretic film
or OLED layer etc. This will move the TFTs towards the neutral plane, and,
thereby, reduce the strain for a given bending radius. The beneficial effect of this is
shown in Fig. 11.32, in which the same type of pentacene TFTs, as shown in
Fig. 11.31, were fabricated on 13 lm thick PI substrates, and laminated with
parylene layers of 13 lm and 10 lm thickness, and compared with an uncapped
control sample (curves A, B and C, respectively) [17]. The un-laminated sample C
showed the same failure, as previously, at a bending radius of 5 mm, but the well-
matched sample A, with the same substrate and top-layer thicknesses of 13 lm,
could be bent down to a radius of 0.5 mm before failure.
11.7 Plastic Substrate Issues
Over and above the previously discussed issues of handling and processing TFTs
on plastic substrates, other concerns have been identified in their use with high-
speed poly-Si devices. A particular issue has been power dissipation in the TFT,
leading to self-heating, and subsequent heating of the plastic substrate [7]. Where
11.7 Plastic Substrate Issues 445
this is sufficient to cause a temperature rise above the glass transition temperature of
the substrate, thermal deformation of the plastic can occur leading to mechanical
damage to the TFT. Self-heating is also a potential problem on glass substrates [69]
(where the thermal conductivity is 1.38 W/m/K), leading to threshold voltage
instability under on-state stress conditions of combined gate and drain bias (see
Sect. 8.6.2). Self-heating is a greater problem on plastic substrates due to their
lower thermal conductivity of *0.2 W/m/K, and, as this means that radiation from
the device edges becomes the limiting mechanism for heat dissipation, the device
design becomes important. In particular, devices with large values of W and L are
vulnerable to self-heating, and the authors [7] recommend replacing large W
devices with a number of parallel devices each having a smaller W. Also, the
overall power dissipation can be reduced by reducing the drive voltages, and this
can be accomplished by scaling down the TFT geometry [7], especially the channel
length and the gate oxide thickness, as well as the drain bias, by a scaling factor k.
This should reduce power dissipation per unit area, at fixed channel current, by the
factor k, as well as increasing the operating frequency by k.
A further recommendation is to improve the energy efficiency of the circuits
themselves, by using asynchronous circuits [7]. Conventional MOSFET and TFT
circuits are usually termed synchronous, and global clock signals are applied to all
logic blocks throughout the system. In asynchronous circuits, the clock signals are
only sent to particular logic blocks when they are needed, and poly-Si asynchro-
nous microprocessors have been found to dissipate 73 % less energy than syn-
chronous circuits performing the same function [7].
Hence, both changes in device design and circuit design are advocated for
optimising the performance of poly-Si circuits on low-temperature plastic sub-
strates, even where the handling issues of plastic substrates have been circum-
vented by using a transfer process.
Self heating has also been reported in a-Si:H TFTs on PI substrates [70], in
which the width-normalised drain current of TFTs with W/L = 80 lm/8 lm and
10 lm/8 lm agreed at low drain bias, but, at high gate and drain bias, the large W
TFT showed a higher current than the small W TFT. In the large W TFT, the
current also failed to saturate, as shown in Fig. 11.33a. Hence, the same geometric
factors leading to self-heating have been identified in a-Si:H TFTs as in poly-Si
TFTs. For self-heating to occur, both gate and drain bias must be applied in order
to drive current through the TFT, as is the case in Fig. 11.33a. When threshold
voltage stability was investigated under saturation drain bias conditions
(VD = VG-VT) at large VD, the threshold voltage shift was found to increase with
increasing channel width, as shown in Fig. 11.33b, and this shift was larger than
with gate bias stress alone [70]. Device stability has been studied under combined
gate and drain bias stress in conventional a-Si:H TFTs on glass [71], where the
change in threshold voltage was found to be less than under gate bias stress alone.
This was due to the reduced overall carrier density in a pinched-off inversion layer
(see Sect. 6.4.3). Hence, the opposite effect was seen when comparing combined
bias-stress effects in TFTs on glass and plastic substrates, and the larger shifts seen
on the plastic substrates have been attributed to self-heating in large W TFTs [70].
Fig. 11.33 a Normalised drain current in a-Si:H TFTs for W = 80 lm, and W = 8 lm,
showing higher currents for W = 80 lm at large values of VG and VD, and b dependence of
threshold voltage instability on channel width, W, under saturation bias-stress conditions.
In addition, the self-heating-induced threshold voltage instability was further

increased by bending the substrate down to a 15 mm radius of curvature [70]. Self-
heating has been previously reported in short channel a-Si:H TFTs on glass [72],
but neither channel width nor device stability effects were investigated in that
work. In common with poly-Si TFTs, it is clear that large W TFTs can cause self
heating, which, in the case of a-Si:H, exacerbates bias-stress instability effects, and
attention to device design is required to minimise these effects.
11.8 Summary
The flat panel display industry has been built around a-Si:H TFTs on rigid glass
substrates, but there are clearly identified applications requiring flexible substrates.
This has stimulated much research investigating how the established a-Si:H, and
also poly-Si, TFT technologies could be implemented on inexpensive plastic
substrates. The major issues in using these readily available substrates have been
their low maximum processing temperatures, thermal mismatch to the conven-
tional device layers such as SiNx:H, SiO2 and Si, and also the dimensional
instability of the plastic itself. In view of these problems, a number of processing
strategies have emerged, including reduced temperature TFT processing directly
onto the plastic substrates, and the use of non-standard plastic substrates, which
have been engineered to better match the thermal properties of the TFT layers.
Secondly, the use of alternative flexible substrates, principally stainless steel
foils, which circumvent many of the handling problems of plastic. Thirdly, and
11.8 Summary 447
principally for a-Si:H TFTs, a carrier plate technology, whereby the plastic sub-
strate is temporarily bonded to a glass substrate during processing, or a plastic film
is spun onto the glass, which, at the end of processing, is released from the glass
carrier plate. For poly-Si, a variant of this has emerged, which is a transfer process,
whereby the TFTs and circuits are processed in the conventional way on a glass
substrate, and, at the end of processing, they are removed from the glass substrate
and transferred to a plastic substrate. The most sophisticated demonstrators so far
have emerged from either the carrier process for a-Si:H TFTs or the transfer
process for poly-Si TFTs. How economic these processes are will emerge in the
longer term, but, for the moment, they are, at least, successful expedients which
are able to exploit the well established glass plate technologies. The long-term
aspiration in this area is for a truly plastic-based, roll-to-roll technology, which
does not try to impose conventional photolithography on dimensionally unstable
substrates, and some early examples of this approach have been reported for a-
Si:H TFTs.
The alternative to the conventional inorganic TFTs, are organic TFTs, which
should have a better match to the plastic substrate constraints of reduced pro-
cessing temperatures and high coefficients of thermal expansion, although the
dimensional instability of the plastic still represents a problem for pattern defini-
tion. As with the inorganic TFTs, similar strategies have been adopted, including
direct substrate processing, carrier plate processing, plus a printing process, which
is less sensitive to the dimensional instability of the substrate. With organic TFTs,
the most sophisticated demonstrators, so far, have used the carrier plate process or
the print-based process.
In addition to the discussion of device processing, the mechanics of bending
and strain in flexible substrates is summarised.
References
1. Sarma KR (2012) Flexible displays: attributes, technologies compatible with flexible

substrates and applications. In: Chen J, Cranton W, Fihn M (eds) Handbook of visual display
technology. Springer, Berlin, p 2012
2. Jackson WB, Almanza-Workman M, Chaiken A, Garcia RA, Jeans A, Kwon O, Luo H, Mei
P, Perlov C, Taussig C, Braymen S, Jeffrey F, Hauschildt J (2008) Active-matrix backplanes
produced by roll-to-roll self-aligned imprint lithography (SAIL). SID Symp Digest
39:322–325
3. Yeo J-S, Elder R, Jackson W, Hoffman R, Henze D, Koch T (2011) Paper-like electronic
media: the case for R2R-processed full-color reflective displays. SID Inf Disp 27(1):18–23
4. Sarma KR (2012) Flexible displays: TFT technology: substrate options and TFT processing
strategies. In: Chen J, Cranton W, Fihn M (eds) Handbook of visual display technology.
Springer, Berlin, p 2012
5. MacDonald BA, Rollins K, MacKerron D, Rakos K, Eveson R, Hashimoto K, Rustin B
(2005) Engineered films for display technologies. In: Crawford GP (ed) Flexible flat panel
displays, Wiley ltd., Chichester, England
6. Suo Z, Ma EY, Gleskova H, Wagner S (1999) Mechanics of rollable and foldable film-on-foil
electronics. Appl Phys Letts 74(8):1177–1179
7. Miyasaka M, Hara H, Karaki N, Inoue S, Kawai H, Nebashi S (2008) Technical obstacles to
thin film transistor circuits on plastic. Jpn J Appl Phys 47(6):4430–4435
8. Cherenak KH, Kattamis AZ, Hekmatshoar B, Sturm JC, Wagner S (2007) Amorphous-silicon
thin film transistors at 300 C on a free-standing foil substrate of clear plastic. IEEE Electron
Device Lett 28(11):1004–1006
9. Long K, Cheng I-C, Kattamis A, Gleskova H, Wagner S, Sturm JC (2007) Amorphous-
silicon thin-film transistors made at 280 C on clear-plastic substrates by interfacial stress
engineering. J SID 15(3):167–176
10. Young ND, Trainor MJ, Yoon S-Y, McCulloch DJ, Wilks RW, Pearson A, Godfrey S, Green
PW, Roosendaal S, Hallworth E (2003) Low temperature poly-Si on flexible polymer
substrates for active matrix displays and other applications. Mat Res Soc Symp Proc
769:17–28
11. Ito H, Oka W, Goto H, Umeda H (2006) Plastic substrates for flexible displays. Jpn J Appl
Phys, 45(5B), 4325–4329
12. Young ND, Bunn RM, Wilks RW, McCulloch DJ, Deane SC, Edwards MJ, Harkin G,
Pearson AD (1997) Thin-film-transistor and diode-addressed AMLCDs on polymer
substrates. J SID 5(3):275–281
13. Kuo P-C, Jamshidi-Roudbari A, Hitalis M (2009) Electrical characteristics and mechanical
limitation of polycrystalline silicon thin film transistor on steel foil under strain. J Appl Phys
106:114502-1–114502-5
14. Wagner S, Gleskova H, Cheng I-C, Sturm JC, Suo Z (2005) Mechanics of TFT technology on
flexible substrates. In: Crawford GP (ed) Flexible flat panel displays, Wiley Ltd., Chichester,
England
15. Gleskova H, Wagner S, Suo Z (1999) Failure resistance of amorphous silicon transistors
under extreme in-plane strain. Appl Phys Letts 75(19):3011–3013
16. Han L, Song K, Mandlik P, Wagner S (2010) Ultra flexible amorphous silicon transistors
made with a resilient insulator. Appl Phys Letts 96:042111-1–042111-3
17. Sekitani T, Iba S, Kato Y, Noguchi Y, Someya T, Sakurai T (2005) Ultra flexible organic
field-effect transistors embedded at a neutral strain position. Appl Phys Letts 87:173502-
1–173502-3
18. Cheng I-C, Kattamis A, Long K, Sturm JC, Wagner S (2005) Stress control for overlay
registration in a-Si:H TFTs on flexible organic-polymer-foil substrates. J SID 13(7):563–568
19. Kattamis AZ, Cherenack KH, Hekmatshoar B, Cheng I-C, Gleskova H, Sturm JC, Wagner S
(2007) Effect of SiNx gate dielectric deposition power and temperature on a-Si:H TFT
stability. IEEE Electron Device Lett 28(7):606–608
20. Sazonov A, Nathan A, Murthy RVR, Chamberlain SG (2000) Fabrication of a-Si:H TFTs at
120 C on flexible polyimide substrates. Mat Res Soc Symp Proc 558:375–380
21. He S, Nishiki H, Hartzell J, Nakata Y (2000) Low temperature PECVD a-Si:H TFT for
plastic substrates. SID‘00 Digest, 278–281
22. Sturm JC, Hekmatshoar B, Cherenack K, Wagner S (2009) Enabling mechanisms for a-Si
TFT’s with 100 year lifetimes compatible with clear plastic substrates. Proc 5th Internal TFT
Conf, ITC’09, 9.1
23. Won SH, Hur JH, Lee CB, Nam HC, Chung JK, Jang J (2004) Hydrogenated amorphous
silicon transistor on plastic with an organic insulator. IEEE Electron Device Lett
25(3):132–134
24. Jang J, Won SH, Kim BS, Hong MP, Chung KH (2005) High-resolution full-color flexible
TFT LCDs based on amorphous silicon. In: Crawford GP (ed) Flexible flat panel displays,
Wiley, Ltd., Chichester, England
25. Won SH, Lee CB, Nam HC, Jang J, Chung JK, Hong M, Kim BS, Lee YU, Yang SH, Huh
JM, Chung K (2003) A high-resolution full color TFT-LCD on transparent plastic. SID 03
Digest, 992–995
References 449
26. Kattamis AZ, Cheng I-C, Hong Y, Wagner S (2006) Amorphous silicon 2-TFT pixel circuits
on stainless steel foils. Mater Res Soc Symp, 910
27. Afentakis T, Hatalis M, Voutsas AT, Hartzell J (2006) Design and fabrication of high-
performance polycrystalline silicon thin-film transistor circuits on flexible steel foils. IEEE
Trans ED-53(4), 815–822
28. O’Rourke SM, Venugopal SM, Raupp GB, Allee DR, Ageno S, Bawolek EJ, Loy DE,
Kaminski JP, Moyer C, O’Brien B, Long K, Marrs M, Bottesch D, Dailey J, Trujillo J,
Cordova R, Richards M, Toy D, Colaneri N (2008) Active matrix electrophoretic displays on
temporary bonded stainless steel substrates with 180 C a-Si:H TFTs. SID 08 Digest,
422–424
29. Ma R, Rajan K, Silvernail J, Urbanik K, Paynter J, Mandlik P, Hack M, Brown JJ, Yoo JS,
Kim Y-C, Kim I-H, Byun S-C, Jung S-H, Kim J-M, Yoon S-Y, Kim C-D, Kang I-B, Tognoni
K, Anderson R, Huffman D (2009) Wearable 4-inch QVGA full color video flexible
AMOLEDs for rugged applications. SID 09 Digest, 96–99
30. French I, George D, Kretz T, Templier F, Lifka H (2007) Flexible displays and electronics
made in AM-LCD facilities by the EPLaR process. SID‘07 Digest, 1680–1683
31. French I (2009) Flexible e-book displays produced in standard TFT and module factories.
SID Inf Disp 25(12):8–11
32. Nikulin, IV, Hwang TH, Jeon HI, Kim SI, Roh NS, Shin SS (2007) Near 100 C low
temperature a-Si TFT array fabrication on flexible PES and PEN substrates. Proc 3rd Internal
TFT Conf, ITC’07, 64–67
33. Lee W, Hong M, Hwang T, Kim S, Hong WS, Lee SU, Jeon HI, Kim SI, Baek SJ, Kim M,
Nikulin I, Chung K (2006) Transmissive 7 VGA a-Si TFT plastic LCD using low temperature
process and holding spacer. SID Digest 06:1362–1364
34. Hwang TH, Lee W, Hong WS, Kim SJ, Kim SI, Roh NS, Nikulin I, Choi JY, Jeon HI, Hong
SJ, Lee JK, Han MJ, Baek SJ, Kim M, Lee SU, Shin SS (2007) 14.3 inch active matrix-based
plastic electrophoretic display using low temperature processes. SID Digest 07:1684–1685
35. Souk JH, Roh NS (2009) Processing flexible displays. SID 09 Digest, 622–624
36. Ma R, Hack M, Brown JJ (2010) Flexible AMOLEDs for low-power, rugged applications.
SID Inf Disp 26(2):8–14
37. Gregg A, York L, Strnad M (2005) Roll-to-roll manufacturing of flexible displays. In:
Crawford GP (ed) Flexible flat panel displays, Wiley Ltd., Chichester, England
38. Kim H-J, Almanza-Workman M, Garcia B, Kwon O, Jeffrey F, Braymen S, Hauschildt J,
Junge K, Larson D, Stieler D, Chaiken A, Cobene B, Elder R, Jackson W, Jam M, Jeans A,
Luo H, Mei P, Perlov C, Taussig C (2009) Roll-to-roll manufacturing of electronics on
flexible substrates using self-aligned imprint lithography (SAIL). J SID 17(11):963–970
39. Gleskova H, Wagner S, Soboyejo W, Suo Z (2002) Electrical response of amorphous silicon
thin-film transistors under mechanical strain. J Appl Phys 92(10):6224–6229
40. Won SH, Chung JK, Lee CB, Nam HC, Hur JH, Jang J (2004) Effect of mechanical and
electrical stresses on the performance of an a-Si:H TFT on plastic substrate. J Electrochem
Soc 151(3):G167–G170
41. Thompson SE, Sun G, Choi YS, Nishida T (2006) Uniaxial-process-induced strained-Si:
extending the CMOS roadmap. IEEE Trans ED-53(5), 1010–1020
on glass and polymer substrates. Proc Asia Display’98, 83–93
43. Higashi S, Abe D, Hiroshima Y, Miyashita K, Kawamura T, Inoue S, Shimoda T (2002)
High-quality SiO2/Si interface formation and its application to fabrication of low-
temperature-processed polycrystalline Si thin-film transistor. Jpn J Appl Phys, 41(6a),
3646–3650
44. Maiolo L, Pecora A, Fortunato G, Young ND (2006) Low-temperature electron cyclotron
resonance plasma-enhanced chemical-vapor deposition silicon dioxide as gate insulator for
polycrystalline silicon thin-film transistors. J Vac Sci Technol, A 24(2):280–285
45. Mariucci L, Fortunato G, Bonfiglietti A, Cuscuná M, Pecora A, Valletta A (2004) Polysilicon
TFT structures for kink-effect suppression. IEEE Trans ED-51(7), 1135–1142
46. Young ND, French ID, Trainor MJ, Murley DT, McCulloch DJ, Wilks RW (1999) LTPS for
AMLCD on glass and polymer substrates. Proc IDW’99, 219–222
47. Gosain DP, Noguchi T Usui S (2000) High mobility thin film transistors on a plastic substrate
at a processing temperature of 110 C. Jpn J Appl Phys, 39(3A/B), L179–L181
48. Troccoli MN, Roudbari AJ, Chuang T-K, Hatalis MK (2006) Polysilicon TFT circuits on
flexible stainless steel foils. Solid-State Electron 50:1080–1087
49. Inoue S, Utsunomiya S, Saeki T, Shimoda T (2002) Surface-free technology by laser
annealing (SUFTLA) and its application to poly-Si TFT-LCDs on plastic film with integrated
drivers. IEEE Trans ED-49(8), 1353–1360
50. Miyasaka M (2007) Flexible microelectronics becoming a reality with SUFTLA transfer
technology. J SID 15(7):479–484
51. Takechi K, Eguchi T, Kanoh H, Ito T, Otsuki S (2005) High-rate glass etching for
transferring polycrystalline silicon thin-film transistors to flexible substrates. IEEE Trans
Semicond Manuf 18(3):384–389
52. Takechi K, Eguchi T, Kanoh H, Kanemasa K, Otsuki S (2007) Transfer process for thermally
stable large-size TFT flexible substrates. IEEE Trans Semicond Manuf 20(1):20–25
53. Pecora A, Maiolo L, Cuscuna M, Simeone D, Minotti A, Mariucci L, Fortunato G (2008)
Low-temperature polysilicon thin film transistors on polyimide substrates for electronics on
plastic. Solid-State Electron 52:348–352
54. Fortunato G, Cuscunà M, Maiolo L, Maita F, Mariucci L, Minotti A, Pecora A, Simeone D,
Valletta A, Bearzotti A, Macagnano A, Pantalei S, Zampetti E (2009) Flexible electronics
based on polysilicon thin film transistor. Proc IMID 2009 Digest, 258–261
55. French I, McCulloch D, Boerefijn I, Kooyman N (2005) Thin plastic electrophoretic displays
fabricated by a novel process. Proc SID 05 Digest, 1634–1637
56. Smeys P, Griffin PB, Rek ZU, De Wolf I, Saraswat KC (1999) Influence of process-induced
stress on device characteristics and its impact on scaled device performance. Trans IEEE ED-
46(6), 1245–1252
57. Wang T-J, Ko C-H, Chang S-J, Wu S-L, Kuan T-M, Lee W-C (2008) The effects of
mechanical uniaxial stress on junction leakage in nanoscale MOSFETs. Trans IEEE ED-
55(2), 572–577
58. Sun Y, Thompson SE, Nishida T (2007) Physics of strain effects in semiconductors and
metal-oxide-semiconductor field-effect transistors. J Appl Phys, 101, 104503-1–104503-22
59. Thompson SE, Sun G, Choi YS, Nishida T (2006) Uniaxial-process-induced strained-Si:
extending the CMOS roadmap. IEEE Trans ED-53(5), 1010–1020
60. Cheon JH, Bae JH, Jang J (2008) Mechanical stability of poly-Si TFT on metal foil. Solid-
State Electron 52:473–477
61. Sekitani T, Kato Y, Iba S, Shinaoka H, Someya T, Sakurai T, Takagi S (2005) Bending
experiment on pentacene field-effect transistors on plastic films. Appl Phys Letts, 86, 073511-
1–073511-3
62. Han SH, Kim YH, Lee SH, Choi MH, Jang J, Choo DJ (2008) Stable organic thin-film
transistor in a pixel for plastic electronics. Org Electron 9(6):1040–1043
63. Yagi I, Hirai N, Miyamoto Y, Noda M, Imaoka A, Yoneya N, Nomoto K, Kasahara J,
Yumoto A, Urabe T (2008) A flexible full-color AMOLED display driven by OTFTs. J SID
1691), 15–20
64. Zhou L, Wanga A, Wu S-C, Sun J, Park S, Jackson TN (2006) All-organic active matrix
flexible display. Appl Phys Lett, 88, 083502-1–083502-3
65. Nakajima Y, Takei T, Tsuzuki T, Suzuki M, Fukagawa H, Yamamoto T, Tokito S (2009)
Fabrication of 5.8-in. OTFT-driven flexible color AMOLED display using dual protection
scheme for organic semiconductor patterning. J SID, 17(8), 629–614
66. Gelinck GH, Huitema HEA, Van Veenendaal E, Cantatore E, Schrijnemakers L, Van Der
Putten JB, Geuns TCT, Beenhakkers M, Giesbers JB, Huisman B-H, Meijer EJ, Benito EM,
Touwslager FJ, Marsman AW, Van Rens BJ, De Leeuw DM (2004) Flexible active-matrix
displays and shift registers based on solution-processed organic transistors. Nat Mater
3:106–110
References 451
67. Burns SE, Reeves W, Pui BH, Jacobs K, Siddique S, Reynolds K, Banach M, Barclay D,
Chalmers K, Cousins N, Cain P, Dassas L, Etchells M, Hayton C, Markham S, Menon A, Too
P, Ramsdale C, Herod J, Saynor K, Watts J, von Werne T, Mills J, Curling CJ, Sirringhaus H,
Amundson K, McCreary MD (2006) A flexible plastic SVGA e-paper display. SID 06 Digest,
74–76
68. Burns SE (2010) Flexible displays made with plastic electronics. SID Inf Disp 2:16–19
69. Inoue S, Ohshima H, Shimoda T (2002) Analysis of degradation phenomenon caused by self-
heating in low-temperature-processed polycrystalline silicon thin film transistors. Jpn J Appl
Phys, 41(11a), 6313–6319
70. Kao S-C, Zan H-W, Huang J–J, Kung B-C (2010) Self-heating effect on bias-stressed
reliability for low-temperature a-Si:H TFT on flexible substrate. IEEE Trans ED-57(3),
588–593
stability of amorphous silicon TFTs. IEEE Electron Dev Letts 25(4):188–190
72. Wang L, Fjeldly TA, Iniguez B, Slade HC, Shur M (2000) Self-heating and kink effects in a-
Si:H thin film transistors. Trans IEEE, ED-47(2), 387–397
Chapter 12
Source-Gated Transistors
Abstract The source-gated transistor, SGT, is an inverted staggered thin film

transistor, which has been engineered to have a barrier contact at the source, rather
than the more conventional ohmic contact. Furthermore, the device geometry is
designed so that current flow is barrier limited, and, in saturation, the height of this
barrier is controlled by the gate voltage. As a result of these design features, the
device has quite different behaviour from conventional TFTs. For example, it has a
lower saturation voltage, higher output impedance, reduced short channel effects,
and its current is insensitive to modest variations in the source-drain separation.
The most readily formed barrier is with a Schottky contact, and these contacts have
been used in SGTs fabricated in both a-Si:H and poly-Si. They have also been
incorporated into the 2-D simulations of device operation, and, as a prelude to the
discussion of the SGT, the operation of Schottky barrier diodes is reviewed. In
addition to sections over-viewing the structure, operation and fabrication of SGTs,
there is also a discussion of the differences in SGT and TFT characteristics.
Finally, there is increasing recognition of the influence of the gate bias on the
source barrier height in staggered organic TFTs, and analyses of the gate-source
interaction in these devices are compared with the SGT analysis.
12.1 Introduction
The source-gated transistor, SGT, is an inverted staggered thin film transistor, in

which the current is determined by the height of an injecting barrier at the source
junction [1, 2]. Moreover, in saturation, the height of this barrier is controlled by
the voltage on an insulated gate electrode. This is in contrast to the conventional
TFT, which is also controlled by the voltage on the gate electrode, but, in that case,
the gate bias determines the conductivity of a carrier-rich channel layer. The
inverted staggered SGT and TFT structures are superficially similar in appearance,
as seen in Fig. 12.1 [3]. However, they are engineered to have different properties.
For instance, the source junction in a conventional TFT will usually have the
lowest possible resistance, so that its value is less than the channel resistance.
In a-Si:H and poly-Si TFTs this is achieved with a doped ohmic contact. In organic

454 12 Source-Gated Transistors
Fig. 12.1 Cross-sectional diagrams of (a) an inverted staggered TFT, and (b) an SGT.
and AOS TFTs, doped contacts are not feasible, and metal Schottky barrier con-
tacts are used, where the barrier heights will be minimised by appropriate choice of
metal. In high quality devices, the contact resistance will be designed to be less
than the channel resistance. Nevertheless, as discussed in Chapter 10, barrier
injection problems have been identified in organic TFTs, and aspects of their
behaviour may be regarded as having some qualitative commonality with SGTs.
The source junction in the SGT will deliberately be a barrier structure, such that
on-state operation is determined by carrier emission over and/or through this
barrier, and the resistance associated with the source will be engineered to be
greater than the channel resistance. This has been achieved by using short source-
drain separations (channel lengths) in the SGT. Further essential features of the
SGT architecture are that the semiconductor film is thin enough for it to be readily
depleted by modest reverse bias values on the source electrode, and that there is
adequate source-gate overlap to facilitate electrostatic coupling between these
electrodes. These differences in architecture, compared with conventional inverted
staggered TFTs, lead to remarkable changes in transistor behaviour, in that the
SGT has a much lower saturation voltage, shows minimal short channel effects,
and can be insensitive to the separation of the source/drain contacts. Moreover, in
the case of a-Si: H, the SGT displays much improved gate bias stability compared
with standard TFTs.
To date, operation of this device structure has been demonstrated using a-Si:H
[1, 2, 4], and poly-Si [5, 6], and 2-D simulations have shown the same features as
the experimental data. Simulations have also indicated that the SGT should have
significant benefits when used with low mobility materials, such as disordered
organic semiconductors [3, 7, 8].
Whilst most of the reported work has used a Schottky barrier diode, it is worth
noting that device operation is not limited to this structure. In principle, any barrier
structure could be used, providing the carrier transmission through the barrier is
controllable by the local field. Other suggested barrier structures include bulk
unipolar, metal-insulator-semiconductor, MIS, and space charge limited diodes [9].
Given the extensive use of Schottky barrier diodes in the published work on
SGTs, Sect. 12.2 reviews the operation of these diodes, as background to the
detailed discussion on SGTs. Following this, the SGT structure and its operation
are presented in Sect. 12.3, and Sect. 12.4 contains details of device fabrication,
including the control of the zero-bias barrier height at the Schottky source junction
in a-Si:H SGTs. Section 12.5 presents a comparison of the electrical characteris-
tics of SGTs and conventional TFTs, and Sect. 12.6 reviews other analyses of gate/
source-barrier interactions, which have been published in the organic TFT field,
and contrasts them with the SGT analysis.
12.2 Schottky Barrier Diodes
The operation of the SGT relies upon the presence of an injection barrier, such as a
Schottky diode, at the source junction, and the operation of reverse biased
Schottky diodes is summarised below. This review also includes the use of surface
doping to control the zero-bias barrier height, which has been used with the
devices made to date [1, 2, 5, 6].
The band diagrams of a Schottky barrier diode, on an n-type semiconductor, at
zero bias, and at a reverse bias, VR, are shown in Fig. 12.2a, b, respectively. At
zero bias, there are equal and opposite electron currents over the barrier, /B, from
the metal to the semiconductor, and vice versa, giving a zero net current. Due to
the built-in bias, Vbi, the semiconductor bands are bent upwards at the surface, and
there is a small depletion region at the surface. At a reverse bias, VR, this voltage is
dropped across an increased width of semiconductor depletion region, so that the
barrier to electron flow from the semiconductor is greatly increased, and the
predominant current flow is just from the metal to the semiconductor. In this
simplified diagram, the injection barrier from the metal is shown as unchanged by
the reverse bias, but in reality there is an image-force field, the Schottky effect,
which lowers the barrier in the presence of an applied electric field, E, as shown in
Fig. 12.2c. The change in barrier height, D/, in this example, is given by [10]:
rffiffiffiffiffiffiffiffiffi
qE
D/ ¼ ð12:1Þ
4pe0
The diagram shows the Schottky effect on the metal work function in the
presence of a constant field, E, in a vacuum. The same effect occurs when the
metal contacts a semiconductor surface, as shown in Fig. 12.2d. The barrier
lowering, D/, is given by Eq. 12.1, except that the permittivity of free space, e0, is
multiplied by the dielectric constant of the semiconductor, es, and the relevant
value of field is that at the metal surface, EB, i.e.
rffiffiffiffiffiffiffiffiffiffiffiffiffi
qEB
D/ ¼ ð12:2Þ
4pe0 es
In a reverse biased metal-semiconductor Schottky diode, because of the pres-
ence of the semiconductor space charge in the depletion region, the internal field
Fig. 12.2 Band diagrams for a Schottky barrier diode on an n-type semiconductor at (a) zero
bias, (b) reverse bias, VR, (c) reverse bias image force lowering of the injection barrier in
vacuum, and (d) barrier lowering in a metal-semiconductor Schottky diode. [(c) and (d) repro-
duced from [10] with permission of John Wiley & Sons, Inc]
varies from zero at the edge of the depletion region to the maximum, EB, at the
surface. Hence, to calculate the Schottky barrier lowering, the field needs to be
evaluated at the semiconductor surface.
The process of electron injection at the barrier is also field-sensitive, as shown
in Fig. 12.3. The different current flow processes are qualitatively similar to those
for carrier emission from a deep trap, as discussed in Sect. 8.5.3. Using the
conventional terminology for Schottky barriers [10], these processes are:
• Thermionic emission, TE, of electrons directly over the top of the barrier
(including barrier lowering due to the Schottky effect), which is likely to occur
at high temperatures and low fields,
• Field emission, FE, by direct tunnelling of electrons at the Fermi level through
the barrier, and is most likely to occur at low temperatures and high fields,
• Thermionic field emission, TFE, which is a mixture of TE and FE in that it
involves the thermal excitation of electrons to a virtual state, from which they
tunnel through a reduced barrier width.
Thus, by increasing the field, it is possible to change the injection mechanism
from simple thermal emission, TE, to the field-enhanced process, TFE, which is
the mechanism engineered into the experimental SGTs. In a doped semiconductor,
the field, EB, at the reverse biased barrier is given by:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2qNd ðVR þ Vbi Þ
EB ¼ ð12:3Þ
es e0
12.2 Schottky Barrier Diodes 457
Fig. 12.3 Illustration of the current flow processes in a reverse biased Schottky diode
Equation 12.3 has been derived from Poisson’s equation, using the depletion
approximation (see Eq. 2.14 for more details), and it shows that the field increases
with both the reverse bias, VR, and the semiconductor doping level, Nd.
For TE, the reverse bias current density, JR, is given by:

/ D/ qVR
JR ¼ A T 2 exp q B 1 exp ð12:4Þ
kT kT
where /B is the barrier height, and A* is Richardson’s constant, which is a function

of the carrier mass, and is equal to 120 A/cm2/K2 for free electrons. (A correction
may be applied to A* to allow for the specific carrier effective mass of the
semiconductor, and, whilst these values have been tabulated for common single
crystal semiconductors [10], the basic value of 120 A/cm2/K2 is usually assumed
for other materials). For VR [ 3 kT, the second parenthetic term can be ignored,
although VR still has an important influence on the current via the barrier lowering
term, D/ (Eqs. 12.2 and 12.3).
For high-field TFE in a-Si:H, it has been shown that for a triangular tunnelling
barrier, with image-force correction, in which the current rises exponentially with
field, and VR [ 3 kT, the current is given by [4, 11, 12]:

2 /B aEB
JR ¼ A T exp q ð12:5Þ
kT
where a is an effective tunnelling constant of *4 nm [11].

A further issue, particularly for a-Si:H SGTs made with the easily processable
metals in common use in a-Si:H technology, is control of the zero-bias barrier
height. For example, Cr on a-Si:H has a barrier of 0.91 eV, which is too high for
useful operation of an SGT, and a lower barrier is needed. In single crystal
Si technology, Schottky barrier heights can be readily modified by using shallow
ion implantation to locally dope the Si surface [10, 13, 14], and the principles are
Fig. 12.4 Schottky barrier

height control by shallow
surface doping (a) idealised
doping profile, and (b) field
distribution with reverse bias,
VR
illustrated in Fig. 12.4. Figure 12.4a shows the approximate donor distribution for
a shallow dose D (given by Nd2d), and its effect on the field distribution is shown
in Fig. 12.4b. If the surface doping, Nd2, is much higher than the background
doping, Nd1, then the latter can be ignored, and the maximum field, EB, will be
given by the integral of the dopant distribution within the fully depleted surface:
qNd2 d qD
EB ð12:6Þ
es e0 es e0
which, from Eq. 12.2, will reduce the barrier by [10]:
rffiffiffiffiffiffi
q D
D/ ð12:7Þ
es e0 4p
However, if the dose is large enough, the reduction in barrier height is
accompanied by a reduction in barrier thickness, and this will enhance the tun-
nelling-based TFE process. Thus, the surface doping changes both the zero-bias
barrier height, and its tunnelling transparency. For a-Si:H ion implanted with
group V impurities, it was found that the empirical relationship between the ion
dose and the effective reduction in Cr Schottky barrier height (calculated from the
reverse bias currents) was [15]:
aqDc
D/ ð12:8Þ
es e0
where a is the effective tunnelling constant, and c was the fractional activity of the
dopant, which reduced with increasing dose (in qualitative agreement with the gas
phase doping of a-Si:H discussed in Sect. 5.5.2). It will be noted that Eq. 12.8
differs from Eq. 12.7, and this is partly because Eq. 12.7 gives just the direct
12.2 Schottky Barrier Diodes 459
barrier height reduction. Whereas, the reverse bias current data, used to generate
Eq. 12.8, gave an effective barrier height reduction, which implicitly included the
contribution from enhanced TFE due to the barrier narrowing [15]. The practical
control of barrier height in a-Si:H SGTs is discussed further in Sect. 12.4.
12.3 SGT Structure and Operation
12.3.1 Background
As is apparent from the chapters on conventional TFTs, contact resistance effects

have been identified in all those structures, although their impact has been sig-
nificantly larger in organic TFTs, where the contacts are Schottky diodes. Also, as
mentioned in those chapters, a series resistance in the source and drain contacts
influences the way in which the drain bias, Vds, is distributed through the transistor
structure. The voltage is divided between the channel and the contacts in a self-
consistent way to ensure current continuity through these different regions. One
direct consequence of this is that with a voltage drop, Vs, across the source contact,
the effective gate bias at the source end of the channel is reduced to VG-Vs. The
potential drop along the channel is also reduced accordingly. In the case of the
SGT, and organic TFTs with Schottky barrier contacts, the voltage drop Vs is
developed as a reverse bias on the source barrier. Similarly, if there is a voltage
drop, Vd, across the drain contact, this further reduces the potential drop along the
channel, with its net value reduced to Vds-Vs-Vd, for a drain potential Vds.
Before considering the detailed analysis of SGT operation, it is useful to identify
its most characteristic differences from conventional TFTs (which will simply be
referred to as FETs in the rest of this chapter), and one of these is its lower saturation
voltage. This is shown in Fig. 12.5 [1, 16] by the comparison of the measured output
characteristics of an a-Si:H SGT and an a-Si:H FET (having the same dielectric and
semiconductor film thicknesses of 300 nm and 100 nm, respectively). The FET had
a threshold voltage of *4 V and a saturation voltage, VD(sat), of *6 V at
VG = 10 V (i.e. the usual dependence on VG-VT), whereas, at VG = 12 V, the SGT
had a saturation voltage of *1.5 V. Detailed analysis of these characteristics also
showed that the SGT had a higher output resistance. A further difference was the
lower current in the SGT, due to its barrier limited operation. In addition, the SGT
had a gm value, in saturation, which progressively increased with VG. However, the
saturation value of gm displays a more varied dependence on VG than indicated by
this figure, and this is shown in Fig. 12.6a [4]. In the on-state, the saturation drain
current increased exponentially with VG, and this was reflected in the comparably
increasing values of gm. However, for gate biases between *5 V and *10 V, there
was a small hump in gm, and its overall value was almost constant. A schematic
version of the gm curve is shown in Fig. 12.6b, and it has been labelled to identify the
different modes of operation of the SGT [17]. In particular, SGT1 represents the gate

output characteristics for a-
Si:H devices (a) SGT, and
(b) FET. [(a) Reprinted with
Institute of Physics, and
(b) Reprinted from [1] with
permission of IEEE]
voltage range where gm is steeply rising, and SGT2 denotes the region with the small
gm peak, around which gm is almost constant. As will be discussed below, these
represent two different modes of operation, with the SGT1 characteristics being due
to gate-controlled barrier lowering of the source diode, whereas the SGT2 mode
entails no barrier lowering [17]. These two operating regimes are reviewed in the
following two sub-sections, starting with SGT1, which has been the focus of most of
the published analyses.
12.3.2 SGT1 Mode
As stated above, this is the regime in which gate controlled lowering of the source
barrier has been invoked to explain device behaviour [1–4]. The proposed mech-
anism is based upon a simple physical principle of using a reverse biased diode to
12.3 SGT Structure and Operation 461
Fig. 12.6 a Gate bias dependence of the saturation values of drain current and gm, measured in
an a-Si:H SGT (Reprinted from [4]. Copyright (2006), with permission from Elsevier), and
b schematic illustration of the gm variation with gate bias. (Reprinted with permission from [17].
Copyright (2012), The Electrochemical Society)
deplete a thin layer, or sheet, of charge, Qd, at a distance, t, from the diode. As a
result of this depletion, the field at the diode increases in direct proportion to the
density of depleted charge, Qd. The most direct illustration of this is with a thin,
highly doped layer in more lightly doped background material, similar to the sit-
uation shown in Fig. 12.4a, except that the highly doped region is not adjacent to
the diode, but separated from it by a distance, t. Using Poisson’s equation, and the
depletion approximation, the reverse bias on the diode, VR(t), required to deplete
just the background material doped with Nd donors to a depth t, is:
qNd t2
VR ðtÞ ¼ ð12:9Þ
2es e0
And the electric field, EB(t), at the diode is:
qNd t
EB ðtÞ ¼ ð12:10Þ
es e0
To deplete the sheet of dopant, Qd, at t, the increased reverse bias is:
qNd t2 qQd t
VR ðt; Qd Þ ¼ þ ð12:11Þ
2es e0 es e0
And the increased electric field at the diode is:
qNd t qQd
EB ðt; Qd Þ ¼ þ ð12:13Þ
es e0 es e0
Hence, for an increment in areal dopant concentration, DQd, the increase in
reverse bias needed to deplete it, and the resulting increase in field at the diode, are
given by the following equations:
qDQd t
DVR ðt; DQd Þ ¼ ð12:14Þ
es e0
qDQd
DEB ðt; DQd Þ ¼ ¼ DVR ðt; DQd Þ=t ð12:15Þ
es e0
In the case of the SGT, the sheet of charge to be depleted is not due to a local
spike in doping level, but is due to a carrier accumulation layer induced by the gate
on the opposite side of the film from the diode. Hence, the operating principle of
the SGT is that, for a given voltage on the gate, a particular reverse bias will be
required on the source in order to deplete the accumulation layer induced by the
gate. This will increase the field at the Schottky source contact, and, if the barrier
height of the diode is field dependent, it will be reduced by the field, and the
current over the barrier will increase. Once the drain bias is large enough for the
voltage on the source to fully deplete both the film and the carrier accumulation
layer beneath it, no further increase in reverse bias can be accommodated beneath
the source. Hence, the device current will saturate at Vds=Vds1, and this has been
described as a ‘pinch-off’ mechanism at the source [1–4].
Further increases in drain bias, beyond the source pinch-off voltage, Vds1, will
increase the film potential adjacent to the source, thereby reducing both VG-Vs and
the local carrier density in the FET channel. The reduced carrier density and higher
lateral field self-consistently maintain an unchanged channel current [18]. The
higher lateral field is a 2-D effect at the edge of the source adjacent to the channel,
and the SGT operation relies upon the diode emission current not increasing with
this lateral field. (This is discussed further below, and also in Sect. 12.6, in which
the results of 2-D numerical simulations of SGTs and OTFTs are compared).
Eventually, there will also be the conventional pinch-off of the FET channel at the
drain, at Vds=Vds2=VG-VT, but this is barely visible in many output characteristics,
such as those in Fig. 12.5a.
The following analysis extends the above qualitative discussion of SGT oper-
ation, and develops the relationship between the gate bias and the device saturation
voltage at that gate bias [2, 4]. In this analytical treatment, the depletion
approximation is used, and it is implicitly assumed that the channel resistance, for
current flow between the source and the drain, is much lower than the source
resistance, such that current flow is entirely controlled by the emission current over
the reverse biased source diode.
The band bending, and the field distributions, below the source electrode in an
a-Si:H SGT are shown in Fig. 12.7a–d, for the two situations where the drain bias
is less than, and equal to the source pinch-off voltage, Vds1, [1], respectively. In the
former case (Fig. 12.7a, c), the centre of the film is field-free, and there is positive
space charge at the top of the film, and negative space charge, plus an electron
accumulation layer, at the bottom of the film. In this condition, at the insulator/
semiconductor boundary:
e0 ei Ei1 ¼ e0 es Es1 þ Qn ð12:16Þ
where Qn is the electron accumulation layer density, Ei1 is the field in the gate
dielectric, and Es1 is the field in the semiconductor at the semiconductor/dielectric
Fig. 12.7 Band bending in the SGT at (a) low and (b) saturation drain biases (Reprinted from [1]
with permission of IEEE). Electrostatic field distribution in the SGT at (c) low and (d) saturation
drain biases (Ein is the field in the dielectric, and EBn, and Esn are the fields in the semiconductor
at the Schottky and dielectric interfaces, respectively)
interface. When the electron accumulation layer is fully depleted by the source
bias, the field-free region and the negative charge at the back surface are removed,
and the condition at the insulator/semiconductor boundary becomes:
e0 ei Ei2 ¼ e0 es Es2 ð12:17Þ
If the source is just pinched-off, and the gate bias is increased, further electrons
are induced at the semiconductor/insulator interface, and the semiconductor film is
no longer fully depleted. A further increase in drain bias is then needed to increase
the reverse bias at the source, and to fully deplete the film. This establishes a new
saturation voltage at the increased gate voltage.
With the assumption of a fully depleted film, the relationship between the gate
bias and the saturation voltage can be calculated. However, the SGT situation is
subtly different from that described by Eq. 12.11, in which the depletion of the free
carriers left behind the ionised donor charge of density Qd. For the SGT, the
depletion of the accumulation layer (whose original density could be described by
VGCi) means that the gate voltage is no longer dropped just across the dielectric,
but is dropped across the series combination of the dielectric and the depleted
semiconductor. Hence, by treating the series combination of the insulator and
semiconductor capacitances (Ci and Cs, respectively) as a voltage divider [1, 2, 4],
the change in potential across the Schottky barrier depletion layer, DVs, for a
change in gate bias, DVG, is:
DVs ¼ DVG Ci =ðCi þ Cs Þ ð12:18Þ
In order to keep the film fully depleted, the increase in reverse bias, DVs, across
the diode is equal to the change in saturation voltage, DVD(sat)=DVds1=DVs,
required to deplete the charge induced by the gate. Thus, the relationship between
gate bias and saturation voltage is [2, 4]:
DVds1 =DVG ¼ Ci =ðCi þ Cs Þ ¼ RV ð12:19Þ
This analysis is referred to as the ‘dielectric model’ [2].
The ratio, RV, is a function of the insulator and semiconductor film thicknesses,
and has a maximum value of unity when Ci Cs, such that most of the voltage is
dropped across the semiconductor film. In general, when Cs [ Ci, RV will be less
than one, and can be reduced well below one by using a thin semiconductor layer
compared with the insulator thickness (which is usually the case with a-Si:H
TFTs). Indeed, for Cs Ci, such that negligible voltage is dropped across the very
thin and depleted semiconductor film, Eq. 12.19 becomes directly equivalent to
Eq. 12.14 (and CiDVG = qDQd). In contrast to the values of Rv \ 1 for the SGT,
RV is always *1 for an FET. The measured values of RV for two SGTs, with
different a-Si:H layer thicknesses, are shown by the data points in Fig. 12.8 [2],
and the lines are based upon Eq. 12.19 (with a SiNx dielectric constant of 6.5). As
shown, with a 100 nm thick a-Si:H film, a ratio of *0.16 was achieved. These
devices had a source-drain separation of \10 lm, and, as the results were con-
trolled by the source barrier, rather than the channel length, the dependence on the
source-drain separation within this range was minimal. However, as the source-
drain separation is increased, the parasitic channel resistance increases, and the
device becomes more FET-like. This is discussed further in Sect. 12.5.
From Eqs. 12.15 and 12.18, and the linear field distribution shown in
Fig. 12.7d, the change in field at the source, DEB, for a change in gate bias, DVG,
is given by [2]:
DEB ¼ DVs =ts ¼ DVG Ci =ts ðCi þ Cs Þ DVds1 =ts ð12:20Þ
where ts is the semiconductor film thickness. Hence, for a fully depleted film, the
surface field increases linearly with gate bias, and the value of EB, at any value of
gate bias, can be obtained directly from the measurement of the saturation voltage,
Vds1. From the current Eq. 12.5, the barrier height lowering, D/, due to the gate
bias can be calculated as [4]:
D/ ¼ aEB ¼ aVds1 =ts ð12:21Þ
From Eqs. 12.5, 12.20 and 12.21, the SGT saturation current should increase
exponentially with VG due to the depletion of the charge at the semiconductor/
dielectric interface by the reverse bias on the source diode. This strong dependence
on VG is shown in Fig. 12.5a. The effective tunnelling constant, a, in Eq. 12.21,
will depend upon the internal field distribution, which will be determined by the
details of the device fabrication process. Hence, a needs to be calculated for any
given process, and it can be obtained from the slope of a plot of gm against IR,
using the transfer characteristic measured in saturation [4]:
Fig. 12.8 Ratio of the

differential change in SGT
saturation voltage with gate
voltage as a function of
a-Si:H film thickness (data
points). The lines are ratios
calculated using Eq. 12.19.
(Reprinted from [2].
dIR aqIR Cs Ci
gm ¼ ¼ ð12:22Þ
dVG e0 es kTðCs þ Ci Þ
By inserting the calculated values of D/ in Eq. 12.5, the zero-bias barrier height,
/B, can be extracted from the effective barrier height, /0 B (given by /B-aEB) at
any value of gate bias. An a value of *2.2 nm was extracted from a set of a-Si:H
SGT output characteristics, and the calculated values of /B and /0 B are shown in
Fig. 12.9a for a high barrier diode [4]. This figure shows a constant zero-bias
barrier height of *0.61–0.62 eV (which was determined by the fabrication pro-
cess, and is discussed in Sect. 12.4), and the effective barrier height, /0 B, varied
linearly with gate bias, as expected from Eqs. 12.19 and 12.21.
The above analysis is based upon a 1-D analytical treatment, and would suggest
that the current is uniform along the length of the source electrode. However,
experimental measurements have shown that the current saturates with increasing
source length, and 2-D simulations have demonstrated that this is because the current
is localised towards the channel end of the source electrode [18]. The other 2-D
effect, as mentioned above, is the increased lateral voltage drop at the source junc-
tion, as Vds is increased above the SGT source pinch-off voltage, Vds1. This continues
until the conventional pinch-off occurs at the drain (at Vds = VG-VT = Vds2), after
which the lateral voltage drop across the source also saturates. In spite of the
increasing lateral field at the source between the Vds1 and Vds2, the reported mea-
surements and simulations have not shown an increase in transistor current. This is in
contrast to the OTFT measurements and simulations discussed in Sect. 12.6, and the
relative insensitivity of the SGT current to the field at the edge of the source is the key
difference between OTFTs and SGTs. However, it should be noted that field relief
architectures at the diode edge have been advocated [1], and their implementation in
Fig. 12.9 a Variation of zero-bias barrier height, /B, and effective barrier height, /0 B, with gate
bias, calculated from experimental a-Si:H SGT characteristics, and b experimental a-Si:H SGT2
mode output characteristics (d = 2 lm, and 4 9 1014 P31/cm2 barrier lowering implant).
(Reprinted from [4]. Copyright (2006), with permission from Elsevier)
poly-Si SGTs has been demonstrated to have a major effect upon the SGT charac-
teristics [19]. In particular, in both experimental measurements and 2-D simulations,
the field relief architecture was essential to obtain the characteristic low saturation
voltage of the SGT. Without this field relief, SGT-like behaviour was lost, and the
saturation voltage was at the conventional TFT value of VG-VT [19]. However, it is
not evident that explicit field relief has been implemented in any of the measure-
ments or simulations of a-Si:H SGTs discussed in this section. On the other hand, the
doping of this material will produce different net dopant and DOS densities across
the source/channel boundary. In particular, there is the shallow P+ doping beneath
the source electrode, but also B+ doping in the channel region to compensate the local
P+ doping [2] (see Fig. 12.10 and Sect. 12.4).
As discussed in Sect. 5.5.2, the doping efficiency of heavily doped a-Si:H is
low, and the active phosphorus sites result from the formation of dopant-dangling
bond pairs, P+-D- [21]. In contrast, compensated material will have a higher
doping efficiency, with the formation of P+-B- pairs, rather than P+-D- pairs,
giving a lower defect density, but broader band-tails [20, 21]. The preferential
formation of the P+-B- pairs also leads to very accurate compensation, even when
the total dopant densities are not accurately matched [21]. Precisely how these
differential dopant and defect densities either side of the source/channel boundary
would affect the field has not been evaluated, but the lower defect density, and the
accurate compensation of electrically active dopants in the channel area, should
give a lower space charge density, and hence a lower field, on the channel side of
the source electrode. The experimental results would indicate that the net effect is
field relief on the channel side of the junction in a-Si:H SGTs. Although this
argument can be made for implicit field relief in the experimental a-Si:H SGTs, it
is not clear why there is not continued barrier lowering, for Vds values greater than
Vds1, in the a-Si:H SGT simulations, where the material is treated as spatially
uniform. This remains an unresolved discrepancy with respect to simulations of
staggered OFETs, and is discussed further in Sect. 12.6.
Fig. 12.10 Cross-section of

an a-Si:H SGT. (Reprinted
from [24])
12.3.3 SGT2 Mode
In contrast to the SGT1 mode, there is no gate-induced barrier lowering in the

SGT2 mode, and the gm value in saturation is approximately constant [4, 17]. This
latter feature is a consequence of the saturation current varying linearly with gate
bias, as shown in Fig. 12.9b [4]. Operation in the SGT2 mode is shown sche-
matically in Fig. 12.6b [17], in which, at low gate bias in region A, the device
operation is initially limited by the parasitic FET channel. In this regime, there is a
small device current, which can be supplied by the Schottky diode operating just
above zero reverse bias. However, as the gate bias increases, the FET current
would linearly increase, and more current would need to be drawn from the
Schottky source diode (in regime B). Thus, its reverse bias would increase by a
few kT until the maximum, low-bias saturation current is drawn. The current will
flow across the film and along the carrier accumulation layer at the semiconductor-
dielectric interface beneath the source. This will layer act like a JFET, with sat-
uration occurring when the layer is pinched-off at the channel edge of the source,
where there is the greatest reverse bias [17]. In the absence of any further effect,
the current through the structure would be limited by this low-bias diode current,
and gm would progressively fall to zero, producing the peak in the gm curve.
However, with increasing gate bias, the charge at the interface becomes sufficient
to start lowering the source barrier, and the transition to SGT1 occurs [17].
The other point to note is that the absolute magnitude of the low-bias diode
current is also a function of the source length, and of the diode’s zero-bias barrier
height. Hence, the longer the source, or the lower the barrier, the greater will be the
gate voltage range over which the SGT2 mode is observed [17]. This mode has
been seen experimentally, particularly where ion implantation has been used to
produce low barrier devices, as shown in Fig. 12.9b [4].
The operation of constant gm SGTs has been independently reported in other 2-
D simulations, and these accurately modelled experimental characteristics similar
to those shown in Fig. 12.9b [22]. The simulated carrier and potential distributions
also confirmed that for drain biases above the saturation voltage, the carrier
depletion was greatest towards the edge of the source contact, where the channel
potential also continued to rise. As barrier lowering was not employed in these
particular simulations, it was concluded that the SGT effect was simply due to a
large gate-controlled series resistance in the depletion region beneath the source
contact [22]. However, as this analysis was carried out before the distinction
between SGT1 and SGT2 modes was identified, the conclusion of simple gate-
controlled series resistance applies, at best, to the SGT2 mode only. Moreover,
recent analysis has concluded that Schottky source barriers, in any TFT structure,
cannot be represented as a simple resistor, but should be represented by a more
complex resistor/diode network [23], and this is discussed further in Sect. 12.6.
12.4 Fabrication Process
SGTs have been made with both a-Si:H and poly-Si, and one of the key features of
the process, namely, the control of the zero-bias barrier height, was the same. In
view of this, and the fact that the SGT process is of greater interest with low
mobility materials [3], the process will be illustrated with a-Si:H. (Details of the
conventional a-Si:H TFT technology can be found in Chap. 5).
A cross-section of an a-Si:H SGT is shown in Fig. 12.10 [24]. This is an
inverted staggered structure, and its main features are the overlap length, S, of the
gate and source electrodes, the separation, d, of the source and drain electrodes, the
source barrier metal, and the phosphorus, P31, implant used to control the zero-bias
source barrier height. As discussed in Sect. 12.2, surface ion doping is a well-
established technique for controlling the zero-bias barrier height of Schottky
diodes, and has been used with both the a-Si:H [2, 15] and poly-Si [5, 6] SGTs.
Measurements of the effective, zero-bias barrier height as a function of phosphorus
dose, D, in a-Si:H with a Cr Schottky contact, are shown in Fig. 12.11 [2]. The low
dose results (circles) were taken directly from Schottky diodes [15], and the higher
dose results (triangles) were extracted from SGT measurements, using the pro-
cedure discussed in Sect. 12.3.2 [2]. These are referred to as effective barrier
heights, as they have been derived from reverse-bias current measurements, and
implicitly contain the effects of barrier thinning, as well as height reduction.
In other words, although the current flow is assumed to be by TFE, the effective
barrier height is the value which would be required to give a TE current of the
same magnitude. Over the extensive dose range of the measurements, the barrier
height was found to scale with log D, and this weak dependence facilitated
accurate dose control of the barrier height [2].
The fabrication stages, for the structure shown in Fig. 12.10, initially followed a
standard a-Si:H TFT process, and were [2]:
• Cr gate deposition and patterning (M1)
• PECVD deposition of SiNx (300 nm) and a-Si:H (100–150 nm)
• Patterning of a-Si:H, and contact window opening to gate (M2, M3)
• Unmasked P31 ion implantation at 10 keV
• Cr source and drain deposition, and patterning (M4)
• BF2+ implant at 12 keV between S and D metal to compensate P31
• Anneal at 250 C for 30 min
12.4 Fabrication Process 469
Fig. 12.11 Variation of the

zero-bias barrier height, in
Cr/a-Si:H Schottky contacts,
with low energy P31 ion dose.
This was a 4-mask process, and, in this example, the drain was a Schottky
barrier diode, which was forward biased during device operation. The P31 implant
was unmasked, and applied to the whole plate. To prevent this highly doped region
from shorting the source and drain contacts, a compensating dose of BF2+ was
implanted between them, using the Cr electrodes as an implant mask. A combined
dopant activation and damage anneal stage was optimised at 250 C [15]. The
finished devices had source/drain gaps, d, of 2–10 lm, the gate/source overlap, S,
was 5 lm, and the source width was 12–600 lm.
The poly-Si SGTs [5, 6] had a similar processing schedule, with one key
difference being the implementation of the barrier lowering implant. For the poly-
Si SGTs, the low energy P31 implant was through a window in a passivating SiO2
layer on top of the device island, where the window defined the source area. As the
implant was localised in the source area, this obviated the need for a compensating
BF2 implant. The source metal was subsequently defined over this window, and
the size of the source metal pad was designed to ensure that it overlapped the edge
of the implant window. As discussed in Sect. 12.3.2, this architecture was shown to
give field relief around the edge of the source electrode [19].
From the above processing details, it is evident that SGT fabrication can be
implemented within standard TFT processing facilities, providing the semicon-
ductor film is thin enough to be depleted by modest reverse biases on the source
electrode. In addition to this fundamental requirement, the other issues are the
possible use of a barrier lowering implant, and a photo-lithography process
capable of yielding small source-drain separations, d. With the appropriate
selection of the source metal, barrier lowering implantation would not be neces-
sary, but ion doping is, nevertheless, a routine stage in poly-Si TFT processing
[25]. However, even in the absence of an ion doping stage, it is suggested that the
source electrode should be overlapped onto an adjacent dielectric layer, as this
would achieve the field relief architecture demonstrated with the poly-Si SGTs
[19]. Higher resolution lithography is also available for the smaller plate sizes used
in poly-Si TFT fabrication [26]. Although poly-Si has been mentioned in the
context of ion doping and higher resolution lithography, it should be noted that this
was solely to demonstrate that equipment for these processes is readily available,
and not to imply that poly-Si would be the preferred material for SGTs. Indeed, it
has been argued that low mobility organic materials could benefit the most from
the SGT structure [3, 28].
12.5 Comparison of SGT and FET Characteristics
Section 12.3 highlighted the differences in saturation voltage between FETs and
SGTs, and, in this section, the wider differences and similarities are looked at more
generally, and are illustrated with experimental examples taken from a-Si:H. In the
preceding sections, it has been emphasised that SGT behaviour is observed when
the injection resistance is the controlling resistance, and the parasitic channel
resistance between the source and drain contacts was kept low by keeping the gap,
d, between 2–10 lm. However, as the separation increases, the parasitic channel
resistance increases, and this will begin to limit SGT behaviour. Ultimately, when
the gap is large enough, the channel resistance will control the drain current, and the
device will display FET behaviour. This transition from SGT to FET behaviour can
be readily observed in the dependence of the ratio RV (dVD(sat)/dVG) on the source-
drain separation, d (where RV is\1 for the SGT, and equal to one for the FET). This
is seen in the variation of the measured values of RV with d in Fig. 12.12a, with it
increasing from the SGT value of *0.2, at small channel lengths, to conventional
FET behaviour (RV = 1) at channel lengths of 100 lm [27]. The same behaviour is
shown by the 2-D device simulations in Fig. 12.12b, in which RV increased from
the SGT value of *0.1 (in good agreement with the capacitor ratio in the dielectric
model), to the FET value over a comparable range of d. (In the simulations, the
a-Si:H thickness was 60 nm, compared with 100 nm in experimental devices,
which is why the capacitor ratio was lower in the simulations). Hence, as the
channel resistance between the source and drain terminals increased, there was a
transition from SGT to FET behaviour, at the end point of which the drain current
was controlled solely by the channel resistance. Conversely, to maintain SGT
behaviour, the source-drain separation must be kept small.
There are two direct implications for device operation in the SGT mode, which
are that, for a given source-drain separation, the SGT current is, by definition, less
than the FET current, and that, at small separations, the SGT current will be
largely independent of that separation. In contrast, the current in an FET should
scale inversely with channel length, and, therefore, control of channel length is a
critical aspect of the FET process for ensuring good device uniformity. Both the
measurements and simulations in Fig. 12.12a, b, respectively, confirmed the weak
dependence of device behaviour on small values of d, and this is seen more clearly
in the simulated SGT output characteristics in Fig. 12.12c [27]. The values of d
were varied from 0.5 lm to 2.0 lm, and there were minimal differences in the
12.5 Comparison of SGT and FET Characteristics 471
Fig. 12.12 Variation of the ratio RV with source-drain separation: a experimental a-Si:H data
(ts = 100 nm, ti = 300 nm), b 2-D simulations (ts = 60 nm, ti = 300 nm), and c simulated SGT
output characteristics for different values of the source-drain separation, d. (Reprinted with
characteristics, including the absence of short channel effects at the smallest d

value. This will not occur in an FET, due to channel shortening, and eventual
drain-induced barrier lowering. These differences are illustrated in Fig. 12.13,
showing simulations of the output characteristics of short channel SGTs and FETs,
with d = L = 250 nm [3]. The SGT continued to show good current saturation,
which was entirely absent from the FET due to strong short-channel effects. The
short-channel effects in the FET, in Fig. 12.13b, were, in part, related to the thick
gate dielectric of 120 nm. As discussed in Sect. 8.7, short channel effects can be
reduced by scaling the dielectric layer thickness with the channel length. However,
what is notable about the SGT is that this scaling is not required, and thick
dielectric layers can be successfully used with small source-drain separations [3].
This can be particularly useful with low quality dielectrics, where thick layers may
be needed to suppress leakage currents and pin-holes [28].
Equations 12.4 and 12.5 show that the SGT current increases exponentially
with decreasing barrier height, and, to ensure adequate drain currents, the reduc-
tion of the zero-bias barrier height by ion implantation is an important part of the
device design. To compare the differences in SGT and FET currents, Fig. 12.14
shows the scaling of SGT currents with ion dose, and the scaling of FET currents
with channel length [2]. This indicates that the saturation current from an a-Si:H
SGT, with a barrier lowering implant of 5 9 1014 P/cm2, is the same as from an
Fig. 12.13 Simulated output characteristics of short-channel a-Si:H devices (a) SGT, d = 250 nm
(ts = 40 nm, ti = 120 nm, and /B = 0.35 eV), and (b) FET, L = 250 nm. (Reprinted from [3]
a-Si:H FET with a 15 lm channel length. Hence, moderately high-current SGTs

can be fabricated, although, as discussed in Sect. 12.3.3, these will be low zero-
bias barrier devices, which will operate in the SGT2, constant gm, mode over a
broad range of VG values.
Finally, due to the reduced free carrier concentration in the channel region
beneath the source in SGTs, there should be improved gate-bias stability in those
devices in which the instability is driven by Fermi level dependent trap generation,
such as a-Si:H. (See Sect. 6.4.1 for a discussion of bias stability effects in a-Si:H
TFTs). As already discussed, the SGT saturation voltage is determined by carrier
depletion at the source, but, at drain biases greater than VG-VT, there is also carrier
depletion at the drain. This is not only another factor improving the output
impedance of the SGT [1], but it also substantially depletes the channel of carriers
as well. The differences in carrier concentrations along the channel, in SGTs and
FETs, are shown by the simulations in Fig. 12.15a, b, and the field distributions are
shown in Fig. 12.15c, d [3]. At drain biases [7.5 V, the carrier concentration
along the SGT channel is lower than in the FET, and particularly so at the source,
where the concentration is highest in the FET. These differences result in superior
gate-bias stability in a-Si:H SGTs compared with FETs, as shown by the bias-
stress results in Fig. 12.16 [9]. For each of these measurements, the gate bias was
adjusted to give the pre-stress drain current shown on the x-axis, and the drain bias
was also large enough to ensure saturation. The change in drain current was then
measured after 24 h combined gate and drain bias stressing. As seen in the figure,
the reduction in drain current was just a few percent in the SGT, even at the
highest current (and carrier densities), compared with [20 % in the FET [9].
In summary, the SGT displays markedly different behaviour from a conven-
tional FET, providing the source-drain gap is small enough for the channel
resistance to be significantly smaller than the injection resistance. However, as the
channel resistance increases, the device becomes increasingly FET-like in its
behaviour. When the SGT mode of operation is maintained, the device shows
some clear advantages over an FET, namely:
12.5 Comparison of SGT and FET Characteristics 473

drain currents measured in
a-Si:H SGTs and FETs as a
function of barrier lowering
implant dose and channel
length, respectively,
(VG = 10 V, W = 600 lm).
• Lower saturation voltage, giving lower voltage operation, and reduced power
consumption in circuits.
• Higher output impedance, giving higher gm/gd ratios
• Reduced short channel effects
• Improved gate bias stability
When these features are considered as a package, it has been argued that they
address many of the problems experienced with a-Si:H FETs driving AMOLED
pixels, and that the a-Si:H SGT is better suited to this application [29]. A further
proposed application is in low power, high gain analogue circuits, which would
benefit from the high voltage gain offered by the large output impedance [19].
Finally, in disordered materials, such as organic semiconductors, in which carrier
transport is by hopping, and the carrier mobility is a function of the lateral field
[30, 31], the SGT has been predicted to offer improved on-currents and frequency
response compared with FETs [3, 28].
12.6 Gate/Source-Barrier Interactions in OTFTs
As all the published work on the SGT has come from one research group, it is
useful to consider how the properties and operating principles of these devices
compare with analyses of other TFT structures. The nearest equivalent device
structures are organic TFTs implemented with a short channel, staggered TFT
architecture, and with the normal Schottky barrier contacts. Given the injection
Fig. 12.15 Simulated carrier concentrations, between the source-drain regions, in (a) the FET,
and (b) the SGT, and the lateral field distributions in (c) the FET, and (d) the SGT. (S = 2 lm,
the source edge is at x = 0.5 lm, L = d = 1.5 lm, and /B = 0.45 eV). (Reprinted from [3]

gate bias stress instability
measurements in a-Si:H
SGTs and FETs
(d = L = 2 lm). (Reprinted
from [9]. Copyright (2008),
with permission from
Elsevier)
problems identified with these structures, particularly as the channel length

reduces, limited attention has been paid to short channel devices, and no examples
of unambiguous SGT-like behaviour have been reported. Indeed, the emphasis in
much of the OTFT work has been to keep the injection barrier as low as possible,
so that classical TFT behaviour is observed, and barrier limited behaviour has been
12.6 Gate/Source-Barrier Interactions in OTFTs 475
avoided in the pursuit of high performance TFT structures. However, as mentioned

in Sect. 10.5.3, there is now increasing recognition of the interaction between the
gate bias and the barrier height in OTFTs [23, 32–35], which is the essential
feature of SGT operation. All these publications have modelled the influence of the
gate bias on the source barrier height, but, from the SGT perspective, the most
relevant is Ref. [34] as it directly evaluated the change in injection barrier height at
drain bias values large enough to ensure that the device was in saturation. In the
other publications [33, 35], the drain bias was low, and many of the calculations
were in the linear regime.
In order to clarify source contact effects, the characteristics of TG/BC OTFTs,
with 60 nm thick, printed TIPS-pentacene films and Au contacts, were studied
over the channel length range 10–200 lm [34]. The transfer characteristics of the
longest channel devices showed perfect agreement with the standard MOSFET
equation, whilst the shortest channel devices did not fit this simple model. This
was attributed to barrier controlled operation in the L = 10 lm devices, and their
non-standard output characteristics are shown by the data points in Fig. 12.17a.
These characteristics are neither those of a classical TFT nor an SGT, and have
been attributed to a device in which both the channel resistance and the barrier
injection resistance play a strong role in determining the overall device behaviour.
In order to interpret this behaviour, 2-D simulation was used, with a model which
linked the lowering of the source barrier to the local electrostatic field, and
included the Schottky effect, as well as trap and phonon assisted tunnelling [34].
The solid lines in Fig. 12.17a were the best fit to the data using a 0.46 eV Schottky
barrier source, with field-induced barrier lowering. The importance of barrier
lowering is shown by the broken lines, which were calculated with the barrier
lowering switched off, and demonstrated that only at low drain bias could the
experimental characteristics be modelled without barrier lowering. Indeed, for
drain bias values above *2 V, the field-induced source barrier lowering was a key
feature of the data fit. The drain bias dependence of the potential through the
thickness of the TIPS-pentacene film (above the source edge, at VG = -44 V), is
shown in Fig. 12.17b (where the inset shows the location of the cut line). This data
was also used to compute the reduction in source barrier height, which continu-
ously fell with drain bias, with a maximum reduction of 150 meV at Vd C -30 V.
In contrast, the barrier lowering at a drain bias of 0.1 V was only *2 meV,
demonstrating that barrier lowering was due to the combined effects of gate and
drain bias. The saturation in barrier lowering occurred when the channel at the
drain pinched-off (at Vds = VG-VT = -30 V), such that further increases in
drain bias were localised at the drain junction, and the reverse bias on the source
saturated. The simulations at VG = -44 V also showed that the hole density was
depleted from the semiconductor/dielectric interface above the source electrode,
and that current injection was predominantly from the source edge in the vicinity
of the cut-line in the insert in Fig. 12.17b [34].
The data shown in Fig. 12.17b has been interpreted in terms of a distributed
diode/resistor network at the source terminal, where the resistor values were
determined by the gate bias [23]. At low drain bias, the increase in TFT current
Fig. 12.17 a Measured (circles) and simulated (lines) output characteristics of a TIPS-pentacene
TFT with L=10 lm (solid and dashed lines are with and without barrier lowering, respectively),
and b cut lines of the electrostatic potential through the pentacene film near the edge of the source
contact (as shown by the inset), and the calculated change of the source barrier height with drain
bias, Vds. (Reprinted from [34]. Copyright (2012), with permission from Elsevier)
with increasing gate bias (the broken lines in Fig. 12.17b) has been attributed to a
spreading resistance effect along the source electrode, in which the increasing gate
bias reduced the channel resistance above the source, and, thereby, extended the
effective emitting length of the source diode. Hence, the gate-bias-dependent
saturation currents (in the absence of barrier lowering) were due to the increasing
effective source area. With barrier lowering switched on, the current increased
above these saturation levels, but, under all bias conditions, the greatest potential
across the diode network was at the channel end of the source electrode, and, with
increasing drain bias, this point pinched-off first at Vds1, giving the inflection point
in the output characteristics. The pinch-off also defined the effective diode area,
which did not increase with further drain bias. Following pinch-off at the source,
further increases in drain bias increased the floating potential at the channel end of
the source, producing a larger electrostatic field at this point. This resulted in
progressive field-induced lowering of the source barrier, as shown in Fig. 12.17b,
and gave the increased device current shown by the solid lines in Fig. 12.17a [23].
Once the channel pinched-off at the drain, at Vds2, the reverse bias on the source
saturated, and the total device current also saturated.
The OTFT data summarised above shows some features in common with the
SGT model, such as the low voltage pinch-off at the edge of the source (due to
reverse bias depletion of the hole accumulation layer), the field induced barrier
lowering due to the combined effects of the gate and drain biases, and the satu-
ration of the reverse bias at the source once pinch-off occurred at the drain.
However, there are detailed aspects of the inverted OTFT behaviour which are
quite different from the SGT data. The most important difference is the continued
reduction in source barrier height for drain biases in excess of the source pinch-off
voltage, Vds1. It will be appreciated that for drain biases in the range
Vds1 \ Vds \ Vds2, the voltage for both the OTFT and the SGT will be divided
between the channel and the source diode in order to self-consistently give current
12.6 Gate/Source-Barrier Interactions in OTFTs 477
continuity through these two regions. The main difference between the two
structures is that with the lower relative channel resistance in the SGT, a larger
fraction of the drain voltage in excess of Vds1 will be dropped across its source.
Hence, for both the OTFT and the SGT, the field at the channel edge of the source
will continuously increase beyond Vds1 until the channel is pinched-off at the
drain. In spite of this similarity, the current in the SGT appears to saturate at Vds1,
whereas it saturates at Vds2 in the OTFT.
Another 2-D numerical simulation of staggered OTFTs [35] showed similar
results to those discussed above [34], although an effective ‘contact’ resistance,
rather than a barrier height, was evaluated. In this case, the simulations were of a
BG/TC pentacene TFT, containing a 30 nm thick film, a 5 lm channel length, a
source barrier of 0.6 eV, and field-induced barrier lowering due to both the
Schottky effect and direct tunnelling [35]. The simulations showed TFT-like
characteristics, and a ‘contact’ resistance, which was a function of both the gate
and drain biases, and which continuously reduced with drain bias up to the drain
saturation voltage (due to the increased reverse bias at the edge of the source). The
emission current was also concentrated at the channel edge of the source electrode,
where the field and barrier lowering were greatest. Once again, saturation in source
‘contact’ resistance and device current only occurred after pinch-off at the drain.
The low saturation voltage, Vds1, of the SGT is a key feature of its behaviour,
and, in the published work on barrier-limited-operation of staggered OTFTs,
discussed above, the saturation voltage is at the conventional TFT saturation
voltage of VG-VT (i.e. Vds2). Nevertheless, in both the SGT and the OTFT sim-
ulations, field-induced lowering of the source barrier (due to the combined action
of VG and Vds) has been demonstrated, and the field at the edge of the source
(adjacent to the channel) has also been shown to increase with drain bias up to the
drain saturation voltage, Vds2. However, barrier lowering for drain biases between
Vds1 and Vds2 has only been found in the OTFT, and not in the SGT. This key
difference in barrier lowering at the source edge between the SGT and the OTFT is
best explained in terms of field relief around the edge of the source electrode in the
SGT. Considering the experimental results for a-Si:H [2, 4] and poly-Si SGTs [19],
a field relief architecture for the source electrode was explicitly identified in the
poly-Si devices as essential for SGT operation. Without this field relief, there was
continued reduction of the source barrier height at the channel edge for Vds values
between Vs1 and Vs2 [19]. This is in good agreement with the OTFT simulations
[23, 34, 35]. For the a-Si:H SGTs, there was no direct field relief structure, but the
process technology, involving a compensating implant in the channel, introduced a
different DOS, as well as a net difference in dopant density, across the source/
channel boundary, and is very likely giving implicit field relief. On the strength of
these experimental SGT results, it is apparent that the most reliable route for
obtaining SGT behaviour in any material, including organic semiconductors, is by
the incorporation of a field-relief architecture into the device structure. This is not
implemented in conventional organic TFTs, and is the most likely explanation for
the absence of reported SGT-like behaviour in these structures.
So far as the SGT simulations are concerned, as mentioned above, the poly-Si
simulations (without field relief) show good agreement with the relevant OTFT
simulations. However, neither the a-Si:H [18] nor the organic SGT [28] simula-
tions appear to incorporate explicit field relief, and do not show barrier lowering
for Vds values beyond Vs1. In both cases, an appropriate zero-bias barrier height
was chosen, with a uniform DOS through the material and across the source/
channel boundary. The a-Si:H simulations [18] appear to be in the SGT2 mode,
where gate bias dependent barrier lowering is absent [17], but it is not clear why
the device should be insensitive to the field at the edge of the source induced by the
drain bias. Hence, whilst the experimental results for both a-Si:H and poly-Si
SGTs, as well as the simulations of poly-Si SGTs, can be reconciled with the
OTFT simulations, there remains an unexplained difference between the simula-
tions of OTFTs and the simulations of both a-Si:H and organic SGTs.
In summary, it has been argued that, when the influence of field relief at the
edge of the source electrode in experimental poly-Si and a-Si:H SGTs is taken into
account, the behaviour of OTFTs and SGTs can be reconciled. Hence, by impli-
cation, in order to realise the particular features associated with the SGT-mode of
behaviour, the explicit incorporation of field relief into organic semiconductor
device structures appears to be necessary.
12.7 Summary
The source-gated transistor, SGT, is a type of staggered thin film transistor, which
has a barrier contact at the source, rather than the more conventional ohmic contact
of TFTs. The height of this barrier is controlled by the reverse bias on the source
diode depleting the accumulation layer induced by the gate voltage. As a conse-
quence of this dependence on the gate and source biases, the device has markedly
different behaviour from conventional TFTs: it has a lower saturation voltage,
higher output impedance, reduced short channel effects, and its current is insensi-
tive to modest variations in the source-drain separation. The experimental imple-
mentation of the structure has been with a-Si:H and poly-Si, using Schottky barrier
contacts, and the device behaviour has been investigated using 2-D simulations.
The SGT architecture is an inverted staggered structure, and, in the barrier-
lowering, SGT1, mode, the analysis of device behaviour demonstrated that the
overlap of the gate and source electrodes enabled the source barrier height to be
controlled by the gate voltage. This meant that the device current saturated due to
pinch-off at the source, when the carrier accumulation layer beneath it was fully
depleted by the reverse bias on the source electrode. Current flow from the source
to drain terminals was facilitated by the conducting channel between them, also
induced by the gate voltage (as in a conventional TFT). However, for the obser-
vation of SGT behaviour, this parasitic channel resistance has to be less than the
effective injection resistance. This places a premium on small source-drain sepa-
rations, but, providing the channel resistance is small, the precise size of this gap is
12.7 Summary 479
unimportant. On the other hand, with an increasing gap, the device eventually
tends towards conventional TFT behaviour, as the channel resistance approaches
and/or exceeds the injection resistance.
The architecture of the SGT is much the same as a staggered organic TFT, which
customarily has a Schottky barrier source contact. In spite of the extensive inves-
tigation of OTFTs, it should be noted that SGT-like behaviour has not been reported
in these structures. Although gate-bias-dependent source barrier lowering has been
identified in OTFTs, there is an important difference in the way in which the barrier
lowering continues to increase with drain bias after the source has pinched-off in the
OTFT. On the other hand, in the SGT, the barrier lowering saturates at source
pinch-off. It should also be noted that the increase in lateral field at the source is
qualitatively identical in both structures, and only saturates when the there is pinch-
off at the drain. This apparent contradiction in behaviour between these two classes
of similar device has been explained by the presence of field relief around the edge
of the source electrode in experimental SGT structures. This is explicitly present in
poly-Si SGTs, and it is argued that it implicitly arises in the ion doping technology
used for fabricating a-Si:H SGTs. One conclusion that can be drawn from this is that
SGT-like behaviour, in devices using organic semiconductors, is more likely to be
seen if field relief is also incorporated into their source electrodes. As a corollary to
this, it is suggested that the absence of such structures in conventional OTFTs is
why SGT behaviour has not been reported in them.
References
1. Shannon JM, Gerstner EG (2003) Source-gated thin-film transistors. IEEE Electron Dev Lett
24(6):405–407
2. Shannon JM, Gerstner EG (2004) Source-gated transistors in hydrogenated amorphous
silicon. Solid-State Electron 48:1155–1161
quality semiconductors. IEEE Trans Electron Dev ED-54(2):354–358
4. Balon F, Shannon JM (2006) Analysis of Schottky barrier source-gated transistors in a-Si:H.
Solid-State Electron 50:378–383
5. Shannon JM, Dovinos D, Balon F, Glasse C, Brotherton SD (2005) Source-gated transistors
in poly-silicon. IEEE Electron Dev Lett 26(10):734–736
6. Sporea RA, Trainor MJ, Young ND, Guo X, Shannon JM, Silva SRP (2011) Performance
trade-offs in polysilicon source-gated transistors. Solid-State Electron 65–66:246–249
7. Guo X, Shannon JM (2009) Current-mode logic in organic semiconductor based on source-
gated transistors. IEEE Electron Device Lett 30(4):365–367
8. Sporea RA, Guo X, Shannon JM, Silva SRP (2011) Source-gated transistors for versatile
large area electronic circuit design and fabrication. ECS Trans 37(1):57–63
9. Shannon JM, Balon F (2008) Source-gated thin-film transistors. Solid-State Electronics
52:449–454
10. Sze SM, Ng KK (2007) Metal-semiconductor contacts. In: Physics of Semiconductor
Devices, 3rd edn, Wiley, Inc
11. Shannon JM, Nieuwesteeg KJBM (1993) Tunneling effective mass in hydrogenated
amorphous silicon. Appl Phys Lett 62(15):1815–1817
12. Nieuwesteeg KJBM, van der Veen M, Vink TJ, Shannon JM (1993) On the current
mechanism in reverse-biased amorphous-silicon Schottky contacts. II. Reverse-bias current
mechanism. J Appl Phys 74(4):2581–2589
13. Shannon JM (1974) Reducing the effective height of a Schottky barrier using low-energy ion
implantation. Appl Phys Lett 24(8):369–371
14. Shannon JM (1974) Increasing the effective height of a Schottky barrier using low-energy ion
implantation. Appl Phys Lett 25(1):75–77
15. Chai MK, Shannon JM, Sealy BJ (1998) Barrier height changes in amorphous silicon
Schottky diodes following dopant implantation. Electron Lett 34(9):919–921
16. Shannon JM (2004) Stable transistors in hydrogenated amorphous silicon. Appl Phys Lett
85(2):326–328
17. Sporea RA, Shannon JM, Silva SR (2012) Modes of operation and optimum design for
application of source-gated transistors. In: Proceedings ECS PRiME Conference, Hawaii, Oct
2012
18. Balon F, Shannon JM (2005) Modeling of source-gated transistors in amorphous silicon.
J Electrochem Soc 152(8):G674–G677
19. Sporea RA, Trainor MJ, Young ND, Shannon JM, Silva SRP (2012) Field plate optimization
in low-power high-gain source-gated transistors. IEEE Trans Electron Dev 59(8):2180–2186
20. Street RA, Biegelsen DK, Knights JC (1981) Defect states in doped and compensated a-Si:H.
Phys Rev B 24(2):969–984
22. Lindner T, Paasch G, Scheinert S (2005) Simulated operation and properties of source-gated
thin-film transistors. IEEE Trans Electron Dev 52(1):47–55
23. Mariucci L, Rapisarda M, Valletta A, Daami A, Jacob S, Benwadih M, Coppard R, Fortunato
G (2013) Current spreading effects in fully printed p-channel organic thin film transistors
with Schottky source-drain contacts. Org Electron 14:86–93
24. Shannon JM, Balon F (2004) New transistors for large area electronics. Proceedings
AMLCD’ 04, 321–324
25. Suzuki T (2006) Flat panel displays for ubiquitous product applications and related impurity
doping technologies. J Appl Phys 99:111101-1–111101-15
26. http://www.nikon.com/products/precision/lineup/fx/fx-903n/index.htm (Accessed Sept 2012)
27. Balon F, Shannon JM, Sealy BJ (2005) Modeling of high-current source-gated transistors in
amorphous silicon. Appl Phys Lett 86:073503
28. Guo X, Balon F, Hatton RA, Shannon JM (2008) High performance transistors in low
mobility organic semiconductors for analog and high-frequency applications. IEEE flex
electron displays conference and exhibition, Phoenix, 1–5
29. Balon F, Shannon JM (2005) Source-gated transistors in amorphous silicon for active matrix
displays. SID’ 05 Tech Digest, 1262–1265
30. Bolognesi A, Berliocchi M, Manenti M, Di Carlo A, Lugli P, Lmimouni K, Dufour C (2004)
Effects of grain boundaries, field-dependent mobility, and interface trap states on the
electrical characteristics of pentacene TFT. IEEE Trans ED 51(12):1997–2003
31. Wang L, Fine D, Basu D, Dodabalapur A (2007) Electric-field-dependent charge transport in
organic thin-film transistors. J Appl Phys 101:054515
32. Vinciguerra V, La Rosa M, Nicolosi D, Sicurella G, Occhipinti L (2009) Modeling the gate
bias dependence of contact resistance in staggered polycrystalline organic thin film
33. Brondijk JJ, Torricelli F, Smits ECP, Blom PWM, de Leeuw DM (2012) Gate-bias assisted
charge injection in organic field-effect transistors. Org Electron 13:1526–1531
34. Rapisarda M, Valletta A, Daami A, Jacob S, Benwadih M, Coppard R, Fortunato G, Mariucci
L (2012) Analysis of contact effects in fully printed p-channel organic thin film transistors.
Org Electron 13:2017–2027
35. Gruber M, Schurrer F, Zojer K (2012) Relation between injection barrier and contact
resistance in top-contact organic thin-film transistors. Org Electron 13:1887–1899
Index
0-9 Anodisation, 374

2-shot SLS’ material, 233 AOS TFT circuits, 332
3-BuT5 (3-butyl-quinquethiophene), 391 AOS TFT materials, 326
APCVD, 214
Aperture ratio, 82, 84, 89, 92, 99
A a-Si:H material, 110
a-regime, 122 a-Si:H release layer, 435
Accumulation, 12, 26 a-Si:H TFTs on flexible substrates, 416
Acene, 342, 346 a-SiNx:H gate insulator, 130
Active matrix addressing, 71, 76, 81, 328 Asynchronous circuits, 445
Active matrix flat panel displays, 69, 141, 186 Avalanche generation, 32, 39, 275
Active matrix LCD, 69, 109
Active matrix OLED addressing, 135
Addressing transistors, 70 B
Adhesive, 425, 435, 437 Back channel electron conduction, 167
Aging processes, 98 Back-channel-etched TFT, 114, 157, 423, 428
a-HfInZnO, 328, 331 Back-face passivation, 116, 167, 330
a-IGZO, 302, 312, 317, 325, 333 Back-light, 70, 93
a-InZnO/a-IGZO, 328 Band-gap states, 53, 225, 306
Air-instability, 364, 389 Band mobility, 144, 156
Air-stability, 359, 390 Band tail states, 111, 143, 157, 164, 266, 430
Air-stable n-channel TFTs, 365 Band-to-band tunnelling, 281
Alignment layer, 73, 92 Barrier films, 442
Alkali ion contamination, 283 Barrier lowering, 281, 295, 388, 455, 466, 475
Alkyl, 343 Barrier thinning, 468
Ambient instability, 354 Beamlets, 230
Ambient stability, 346 Beam shape, 190, 200, 234
Ambipolar inverters, 367 Bending-induced strain, 413
Ambipolar material, 169, 366 Bending radius, 340, 416, 439, 444
Aminosilane regime, 130 Benzene, 342
AMOLED displays, 2, 135, 203, 244, 301, Benzocyclobutene (BCB), 419
332, 422, 442 Bias-stress instability, 112, 169, 285, 390, 446
AMOLED pixels, 99 Bi-layer, 117
Amorphisation, 218 Birefringence, 71
Amorphised, 188 Black mask, 84, 91
Amphoteric, 110, 158, 169 Bond angle disorder, 142
Anisotropic plasma etching, 433 Bond angles, 110

DOI: 10.1007/978-3-319-00002-2, Ó Springer International Publishing Switzerland 2013
482 Index
B (cont.) Contact resistance, 134, 164, 182, 290, 311,

Bond lengths, 110 352, 383
Breakdown field, 216, 368, 373 Continuity equation, 39
Built-in strain , 412 Continuous grain boundary, 208
Bulk resistance, 164, 387 Continuous-grain-silicon, 210
Continuous wave (CW) mode, 203
Contrast ratio, 74, 84
C Coplanar, 113
Capacitance–voltage, 25 Cost model, 243
Capture cross section, 31, 34, 38 Coulomb centre, 281
Carbon–carbon bonds, 342 Cr contacts, 383
Carrier generation lifetime, 270 Cr-Au contacts, 383
Carrier injection instability, 170 Cross-talk, 81, 84, 87
Carrier plate processing, 442 Crystallisation rate, 240
Carrier plate technology, 423 c-Si MOSFETs, 288
Carrier transport resistance, 385 CTE mismatch, 412, 422
Carrier transport, 350 Current crowding, 166, 167, 378, 385
Carrier trapping in the gate dielectric, 169 C–V curves, 25
Carrier velocity, 39 C–V measurements, 321
Carrier-dependent mobility, 320, 350 Cytop, 370
CGS material, 211
Channel conductance, 52, 165
Channel shortening, 164, 219, 313 D
Chip-on-film (COF), 93 Dangling bond, 110, 144
Chip-on-glass (COG), 93 Dangling bond formation, 149, 169, 172
Chynoweth law, 293 DB energy levels, 158
Circuit integration, 243 DC magnetron sputtering, 308
Circuit simulation, 381 DC-sputtered Al2O3, 332
Close packed lamellae, 345 Deep level centre, 29
Cluster tool, 120 Deep states, 112
CMOS circuits, 243 Defect formation barrier, 147
Coefficients of thermal expansion, 410 Defect formation energy, 145, 146, 158
Cold cathode fluorescent light, 93 Defect generation, 178
Collision-front grain boundary, 230 Defect passivation, 109
Coloured filters, 74 Defect pool, 112, 162, 172
Colour filter(CF), 91 Defect structure, 142
Columnar grains, 193 De-hydrogenation, 225
Column inversion, 80 De-lamination, 415
Common electrode, 92 Demonstration circuits, 435
Compact circuit modelling, 164 Dendritic grains, 188
Compaction, 188 Density of states (DOS), 112, 149, 262, 273,
Compaction rate, 189 305, 318
Compensating implant, 477 Density of states distribution, 320
Complementary inverter, 366 Depletion, 13
Complementary pairs, 333 Depletion approximation, 15
Compressive strain, 126, 415, 429, 439, 443 Depletion device, 58
Conduction band tail, 143, 163 Depletion region, 15
Conduction in poly-Si, 254 Deposition power density, 417
Conduction process, 318 Deposition rate, 358
Conjugated molecular systems, 342 Design rules, 83, 118, 223
Constant gm SGTs, 467 Device architecture, 137, 222, 254, 302, 351,
Contact architecture, 386 382, 442
Contact effects, 381 Device instability, 358
Index 483
Device lifetime, 275 Electron affinity, 348

Device simulation, 38 Electron capture, 30
Dielectric charge, 160 Electron emission, 30
Dielectric layer, 326 Electron mean free path, 277
Dielectric model, 464 Electron thermal emission rate, 281
Diffusion current, 35 Electron traps, 326, 363, 389
Dilution ratio, 122, 135, 213, 418 Electro-negativity, 19
Dimensional stability, 188, 412, 431, 442 Electron-hole pair generation, 28, 268
Diode pumped solid state lasers, 203, 234 Electronic-paper, 93
Dipole barrier, 348 Electrophoresis, 95
Dirac centre, 281 Electrophoretic displays, 93
Direct processing on plastic, 417 Emissive displays, 96
Discrete trapping level, 261 Emitter current gain, 293
Display module, 92 Enhancement device, 57
Dissociative Ni diffusion, 205 EPLAR process (Electronics on plastic by
Distributed diode/resistor network, 475 laser release), 424
Distributed DOS, 261 Equilibrium defect density, 112, 146, 158
DNTT, 358 Equivalent oxide thickness (EOT), 217
Donor formation, 306, 329 E-reader devices, 70
Dopant-dangling bond pairs, 466 Etch selectivity, 114
Dopant/defect model, 128 Etch-stop TFT, 115, 168, 307, 326, 332, 423
Doped a-Si:H, 127 Excimer laser annealing (ELA), 189, 210,
Doping efficiency, 128 227, 431
Dot inversion, 81, 87 Excimer laser crystallisation, 190, 227
Double exponential distribution Excitons, 96
of states, 263 External drive circuits, 92, 242
Drain field, 273
Drain field relief, 222, 244
Drain-bias stress, 222 F
Drain-induced barrier lowering Fermi-Dirac distribution function, 22
(DIBL), 289 Field effect mobility, 45, 53, 119, 153, 155,
Drift current, 48, 197, 269 163, 198, 259, 320, 350,
Drift mobility, 144, 157, 163 378, 383
Drift/diffusion currents, 38 Field emission (FE), 351, 456
Drift-limited, 267 Field enhanced leakage currents, 169, 280
Drift-limited conduction, 270 Field relief architectures, 465, 469, 478
Driver circuits, 333 Field relief regions, 276
Drop-casting, 355 Field-induced barrier lowering, 475
Dual-metal contacts, 385 Field-shielded pixel, 95
DVAnt, 358, 380, 389 Figure of merit, 262
Fine grain material, 195
Finite width grain boundary, 262
E Fixed oxide charge, 58
ECR-CVD, 214 Flat band conditions, 19
ECR oxide, 433 Flat band voltage, 21, 26, 58
Effective barrier height, 465 Flexible substrate, 340
Effective channel length, 313 Flicker, 86
Effective density of states, 264 Floating body effects, 292
Effective gate bias, 459 FPTBBT, 390
Effective source area, 476 Frame inversion, 80
Effective tunnelling constant, 458 Frame time, 76
Effective work function, 348 Free carrier mobility, 53
Elastomer etch mask, 427 Front channel hole conduction, 167
Electroluminescence, 96 FTIR absorption spectra, 124
484 Index
G Humidity strain, 417

Gate and drain bias stress, 176, 285, 472 Hydrogen-mediated weak bond breaking, 172
Gate bias instability, 126, 135, 214, 326, Hydrogen-mediated equilibration, 112
393, 417 Hydrogen passivation, 21, 214, 263
Gate bias stress, 169, 283, 390, 421 Hydrophobic gate insulator, 393
Gate dielectric, 160, 212, 308, 368 Hydroxyl-free insulators, 390
Gate-bias stability, 472 Hysteresis, 389
Gate-controlled barrier lowering, 460
Gated n+-p diode, 49
Gate-overlapped lightly doped drain I
(GOLDD), 223, 276 Illumination-induced threshold
Gate-source/drain overlap, 166 voltage shift, 329
GB anisotropy, 233 Impact ionisation, 39, 274
GB barrier height, 257 Impact ionisation generation rate, 293
GB DOS, 265 In-grain DOS, 265
Generation current, 28, 197, 269 Injecting barrier, 348, 453
Generation lifetime, 28, 33 Inkjet printing, 357, 362, 374, 376
Generation-limited current, 271 In-plane switching, 81
Gettering process, 211 Integrated drive circuits, 189, 436
Glass etch transfer process, 436 Interface state capacitance, 26
Glass transition temperature, 409 Interface states, 21, 62
Glass-compatible process, 187 Interface trapping states, 18
Gold, 34 Interlayer dielectrics, 216
Gold contacts, 382 Internal quantum efficiency, 97
Gradual channel model, 46 Interstitial hydrogen, 305
Grain boundary, 208, 230, 255, 262, 347, 392 Inter-trap hopping, 350
Grain filter, 229 Intrinsic deposition strain, 417
Grain size, 187, 193, 238, 257, 346, 356, 371 Intrinsic surface, 26
Green laser annealing systems, 203 Inverse slope, S, 61
Grey scale, 95 Inversion, 13
Inversion layer, 46
Inversion layer thickness, 260
H Inverted staggered, 76, 113, 306, 332,
Hall mobility, 255, 303 351, 453
Hard surface coatings, 412 Inverted surface, 26
Heavily doped a-Si:H, 466 Inverter circuit architectures, 333
Heavily doped grain, 256 Ion doping damage, 291
Helium-diluted SiH4/N2O films, 214 Ion doping system, 226
Herringbone, 346 Ion implantation, 46
Hexamethyldisilazane (HMDS), 368 Ion shower doping, 218
Highest occupied molecular orbital Ion-doping, 132
(HOMO), 344 Ionic bonding, 302
High-k dielectrics, 217, 309 Ionic instability, 283
High-level injection, 35 Ionisation energy, 348
Hole capture, 30
Hole drift mobility, 163
Hole effective mass, 439 K
Hole emission, 30 Kink effect, 274, 292
Hopping transport, 443
Hot carrier damage (HCD), 274
Hot carrier injection probability, 278 L
Hot-electron, 276 Large Grain Poly-Si, 227
Hot electron damage, 222 Lateral diffusion, 188, 219, 274
Hot-hole, 276 Lateral dopant distribution, 274
Index 485
Lateral field, 350, 462, 473 Mobility and strain, 429

Lateral growth, 194, 206, 227 Mobility edge, 111, 144, 305
Lateral growth front, 230 Mobility gap, 111, 153
Law of mass action, 147, 159 Molecular alignment, 352
LC cell, 92 Molecular organisation, 344
Leakage current, 28, 33, 80, 167, 181, 208, Molecular stacking, 345
266, 353, 439 MOSFET, 45
Led backlights, 93 MOSFET HCD model, 276
Lifetime of an AMLCD, 174 Mott’s ‘N, 8-N bonding rule’, 128
Lift-off process, 133 Multi-domain vertical alignment, 81
Light valve, 74 Multi-fingered devices, 288
Lightly doped drain (LDD), 223, 274 Multiple trapping and release (MTR), 350
Lightly doped grains, 257
Linear polariser, 73
Line-at-a-time addressing, 76 N
Line-beam, 190 Nd:YVO4 lasers, 234
Liquid crystal director, 72 Near-mid-gap centres, 282
Liquid crystal material, 71 Needle-like grains, 204
L-MIC, 210 Negative Bias Illumination Stress (NBIS), 328
Location controlled large-grain, 228 Negative Bias-Temperature Instability
Longitudinal effective mass, 440 (NBTI), 283
Low angle sub-grain boundaries, 230 Negative photo-resist, 117
Low field mobility, 40 Negative threshold voltage shifts, 283
Low resistance contact, 308 Negative U-centre, 304
Low temperature processing, 334 Nematic phase, 72
Lowest unoccupied molecular orbitals Neodymium-doped yttrium aluminium
(LUMOs), 344 garnet, Nd:YAG, lasers, 203
Low-field generation process, 266 Neutral dangling bond, 145
Low-k dielectrics, 370 Neutral level, 23, 149
Low-level injection, 35 Neutral plane, 415
‘Lucky electron’ model, 278 Ni contamination, 206
Ni mediated crystallisation, 204
Nickel disilicide (NiSi2), 204
M Non-radiative recombination, 178
Maximum mobility, 200 Non-rectifying contact, 267
Mean free path, 39 Non-self-aligned (NSA), 218
Mechanical robustness, 407 Normalised sub-threshold slope, 379
Melt depth, 193
Melt threshold energy, 198
Metal cation, 302 O
Metal induced crystallisation (MIC), 204 Oblique-incidence laser irradiation, 221
Metal induced lateral crystallisation Octadecyltrichorosilane (OTS), 368
(MILC), 207 Off-set gate, 221
Metal work function, 97, 342, 348, 385 Oligomer, 339
Metal/organic contact, 347 Oligothiophenes, 343
Metallic nano-particles, 374 One drop fill, 92
Metallisation, 114, 306, 311 Optical absorption depth, 190
Meta-stable defect, 131, 178 Optical band gap, 169, 365
Micro-cavity, 99 Optical generation rate, 39
Micro-Czochralski process (l-Cz), 227 Optically enhanced instability, 326
Mid-gap centres, 33 Optically stressed samples, 180
MIS capacitor, 10 Organic dielectrics, 370
Mobile hydrogen, 148, 283 Organic light emitting diode, 96
Mobility, 302 Organic solvent barrier, 354
486 Index
O (cont.) Pixel TFTs, 241, 242

Organic TFTs on Flexible Substrates, 440 Planarising layer, 433
Orthogonal solvents, 354 Plasma Enhanced Chemical Vapour
Orthogonal TFTs, 265 Deposition (PECVD), 119
Output characteristic quality factor, 290 Plasma etching, 375
Output impedance, 101, 164, 473 Plasma hydrogenation, 189
Oxidant ingress, 364 Plasma sheath region, 121
Oxidative stable, 361 Plasma-doping, 433
Oxide charge creation, 287 Plastic substrates, 409
Oxide charges, 20 Plate throughput, 124, 200, 240, 357
Oxidising reactions, 358, 389 PMMA (polymethylmethacrylate), 370
Oxygen anion, 302 p–n junction, 34
Oxygen partial pressure, 304, 310 Poisson’s equation, 12, 152
Oxygen vacancy, 303, 326, 330 Polar dielectric films, 369
Poly(3-hexylthiophene) (P3HT), 343, 362,
370, 380, 391
P Poly(9,9-dioctylfluoreneco-bithiophene)
p-bonding, 342 (F8T2), 354, 363, 376, 384, 391
p-orbital, 344 Polyarylates (PAR), 409
P(NDI2OD-T2), 365, 380, 390 Polyethersulphone (PES), 409
Paired VG method, 314 Polyethylene naphthalate (PEN), 340, 409
Parasitic bipolar transistor (PBT), 274, 293 Polyethylene terephthalate (PET), 340, 409
Parasitic channel resistance, 470 Polyimide (PI), 409
Parasitic gate-drain capacitance, 219 Polymer materials, 362
Parasitic resistance effects, 290 Polymer, 339
Parylene, 374 Poly-Si DOS, 262
Passivation layer, 326 Poly-Si resistors, 186
Passive matrix addressing, 75 Poly-Si TFTs on Flexible Substrate, 430
PECVD SiNx, 308 Poly(vinyl phenol) (PVP), 363, 370, 380, 391,
PEDOT/PSS, 374 395, 441
Pentacene, 339, 343, 380 Poole–Frenkel effect, 168, 281, 350
Percolation limited transport, 305 Porosity, 213
Percolation, 318 Portable displays, 70, 243, 340, 407
Percolation-modulated trapping effects, 323 Positive photo-resist, 117
Perfluorinated copper phthalocyanine Post-deposition anneal, 214, 310, 322, 344,
(F16CuPc), 364, 380, 390 362, 433
Peripheral driver TFTs, 241, 242 Power dissipation, 244, 285, 366, 444
Peri-xanthenoxanthene (PXX), 359, 380 PPCB (polypropylene-co-butene, 370
Persistent photoconductivity, 326 PQT-12, 345, 362, 369, 380, 389
Phase modulated excimer laser annealing Precursor film, 188
(PMELA), 227 Pre-shrink, 441
Phonon assisted tunnelling, 281, 475 Printing, 206, 357, 362, 376, 442
Phosphorescent OLEDs, 97 Process Flow, 225, 311, 374, 424, 433
Photo-generated electron-hole pairs, 178 Projection AMLCDs, 186
Photo-ionisable defects, 330 Proton diffusion-controlled instability, 394
Photolithographic aligners, 118 Proton generation, 393
Photolithography, 117 PrPh-PXX, 361
Photo-TFT, 102 PSeDPPBT, 366
Ph-PXX, 359 Pseudo-single-crystal (PSX) material, 228
Pinch-off voltage, 49, 462 Pulse overlap, 200
Pin-hole density, 368 Pulsed Bias Stress, 174, 330
Pixel, 74, 76, 83, 88, 243, 375 Pulsed bias stress under illumination, 331
Pixel electrode, 76 Pulsed laser deposition, 308
Pixel Layout, 83 Pulsed Nd:YAG lasers, 234
Index 487
Pulse-to-pulse fluctuations, 200 Self-heating induced instabilities, 285, 444

PVA (polyvinylalcohol), 370, 395 Self-organise, 345
Sequential lateral solidification (SLS), 227
Series resistance, 164, 221, 276, 312, 378,
Q 381, 459
Q-switched mode, 203 SGT simulations, 478
Quasi-Fermi level, 31 Shadow-mask evaporation, 353
Quasi-static C–V, 27 Shallow ion implantation, 457
Shockley, Read, Hall (SRH) recombination, 29
Short Channel Effects (SCEs), 288, 315
R Short channel TFTs, 133, 216, 221, 290, 316
Radius of curvature, 411 Shot number, 200, 230
Rate of recombination/generation, 32 Shrinkage rate, 413
Recombination lifetime, 36 Si/SiO2 interface, 21, 213, 276, 285
Recovery process, 202, 393 Sidewall spacer, 224
Reflective mode, 70 SiH3 radical, 126
Regio-regular molecules, 345 Silane, 120–130, 283, 418
Remote plasma, RPCVD, 214 Silanol, SiOH, electron traps, 363
Residual ion damage, 219 Silicide-mediated crystallisation (SMC), 204
Resistive hole flow, 268 Silicon dioxide, 212, 416
Resistivity of doped poly-Si, 254 Silicon-hydrogen bonding states, 124
Resolution of a display, 75 Silicon-on-insulator (SOI) MOSFETs, 218
Reverse biased source barrier, 388 Single channel inverter, 244, 333
Reversible elastic deformation, 429 Single crystal rubrene, 391
RFID tags, 341 SiOx, 307
Ribbon-shaped grains, 355 SLS poly-Si, 232, 265
Richardson’s constant, 457 Small molecule material, 339, 358, 360
Rigid carrier , 412, 423, 441 Small molecule organic LEDs, 96
Roll-to-Roll (R2R), 426 Smectic-A phase, 72
Room temperature strain, 410 Smectic-C phase, 72
Row inversion, 80 Solid phase crystallisation (SPC), 188,
Row resistance, 87 204, 219
Solution of pentacene, 360
Solution processing, 339, 343, 355, 360, 373
S Source barrier height, 296, 475
r bonding-orbital, 344 Source pinch-off voltage, 462
Safe strain range, 415 Source-Gated Transistor (SGT), 65, 388, 453
Saturation regime, 49 Specific contact resistivity, 165, 314
Saturation voltage, 45, 49, 64 Spin-coating, 355
Scaling factor, 445 S-state orbitals, 303
Scanning Kelvin probe microscopy, 382 Staebler–Wronski effect, 178
Schottky barrier, 381 Staggered OTFTs, 388, 477
Schottky barrier diode, 455 Steel foils, 411, 416, 423, 433
Schottky effect, 388, 455 Steel substrate, 411
Schottky–Mott limit, 348 Stoney formula, 411
Scratch-isolation, 354 Storage capacitor, 81, 95, 100
Self-aligned (SA), 46, 76, 131, 218, 226, 307, Stress-induced changes in the band-gap, 439
427, 435 Stressing in vacuum, 391
Self-aligned imprint lithography (SAIL), 427 Stretched exponential, 147, 171, 180, 327, 391
Self-alignment, 377 Stretched hyperbola, 173
Self-assembled mono-layers (SAMs), 345, Stretched hyperbolic function, 394
353, 363, 368, 373 Sub-band-gap illumination, 328
Self-heating, 286, 444 Substrate doping, 54
488 Index
S (cont.) Transfer process, 435

Sub-threshold currents, 59 Transmissive mode, 70
Sub-threshold slope, 46, 62, 160, 197, 273, Transparent metal conductor, 73, 92
302, 380 Trap limited conduction, 262, 322
SUFTLA (Surface free technology by laser Trapped charge density, 153
annealing), 435 Twisted Nematic LC Cell, 73
Super-lateral growth (SLG), 193
Surface band bending, 19
Surface dipole, 349 U
Surface doping, 455 Ultraviolet photoelectron spectroscopy
Surface electron density, 11 (UPS), 349
Surface energy conditioning, 376 Uniaxial Strain, 414, 428 , 438, 442
Surface melting, 190 Uniformity, 96, 120, 196, 203, 233, 244, 301
Surface polarity, 352 Urbach energy, 143
Surface space charge region, 62, 151
Surface states, 21–27
System-on-glass (SOG), 186 V
Vacuum thermal evaporation, 358
Valence band tail, 143, 147, 163
T Velocity saturation, 40
Tail state energy width, 164, 323 Vertical cross-talk, 87
Tape automated bonding, TAB, 93 Vertical crystallisation, 235
Tauc optical gap, 303 Viewing angle, 81
TEOS (Tetraethylorthosilicate), 213 Voltage Kick-Back, 85
TFT architecture, 113, 218, 224, 305, 351
TFT fabrication, 114, 224, 309, 362, 419
TFT layout, 118 W
TFT lifetime, 101, 135, 176, 422 Water vapour transmission rates (WTVR), 412
TFT off-state currents, 266 Weak bonds, 110, 145
Thermal deformation, 445 Weak inversion, 14
Thermal desorption spectroscopy (TDS), 213 Wet oxygen annealing, 310
Thermal oxide, 352 Wetting-angle, 376
Thermalisation energy, 173 Width normalised series resistance, 316, 381
Thermally conducting flexible substrates, 288 Work function differences, 17, 58
Thermionic emission (TE), 257, 385, 456
Thermionic field emission (TFE), 456
Thiophene, 342 Z
Threshold voltage, 17, 52, 55, 58, 63, 153 Zero Kelvin approximation, 21, 151
TIPS-pentacene, 351, 355, 361, 378, 391 Zero-bias barrier height, 455, 465, 468
Top-hat’ beam, 190 Zone-casting, 356
Trailing edge of the beam, 201
Transconductance, 52

INtroduction To Thin Film Transistor PDF

Uploaded by

Copyright:

Available Formats

INtroduction To Thin Film Transistor PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

INtroduction To Thin Film Transistor PDF

Uploaded by

Copyright:

Available Formats

What is the book about?

What is the book about?

What materials are discussed for TFTs?

What materials are discussed for TFTs?

S. D.

Introduction to Thin Film

ISBN 978-3-319-00001-5 ISBN 978-3-319-00002-2 (eBook)

Library of Congress Control Number: 2012956303

Ó Springer International Publishing Switzerland 2013

Printed on acid-free paper

Springer is part of Springer Science?Business Media (www.springer.com)

Thin film transistors, TFTs, are now fundamental electronic components in

Prof. G. J. Parker (Southampton University), Dr. J. R. Ayres, Dr. N. D. Young and

2 Semiconductor Device Physics for TFTs . . . . . . . . . . . . . . . . . . . 9

3 Insulated Gate Field Effect Transistors, IGFETs . . . . . . . . . . . . . 45

4 Active Matrix Flat Panel Displays . . . . . . . . . . . . . . . . . . . . . . . . 69

4.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102

5 Hydrogenated Amorphous Silicon TFT Technology

6 Hydrogenated Amorphous Silicon TFT Performance . . . . . . . . . . 141

6.5 Other Meta-Stability Effects . . . . . . . . . . . . . . . . . . . . . . . . 178

7 Poly-Si TFT Technology and Architecture. . . . . . . . . . . . . . . . . . 185

8 Poly-Si TFT Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253

8.2.2 Analytical Model of TFT Conduction . . . . . . . . . . . . 260

9 Transparent Amorphous Oxide Semiconductor TFTs . . . . . . . . . 301

9.5 AOS TFT Circuits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 332

10 Organic TFTs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 339

Part III Novel Substrates and Devices

11 TFTs on Flexible Substrates . . . . . . . . . . . . . . . . . . . . . . . . . . . . 407

12 Source-Gated Transistors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 453

S. D. Brotherton, Introduction to Thin Film Transistors, 1

S. D. Brotherton, Introduction to Thin Film Transistors, 9

2.2 Semiconductor Surface Physics

2.2.1 Ideal MIS Capacitor and Surface Band Bending

where G(Vs, VF) is given by:

Fig. 2.2 Calculated variation

Fig. 2.3 Comparison of

and from Eq. 2.15,

Qs in Eq. 2.12 can be represented by the sum:

2.2.2 Gate Bias and Threshold Voltage

2.2.3 Real MIS Structures

Fig. 2.7 Calculated variation

2.2.3.1 Work Function Differences

Fig. 2.9 Experimentally

2.2.3.2 Oxide Charges and Interface States

Hence, for charges at the dielectric/semiconductor interface, the induced charge

In general, the electron occupancy, F(E), of a surface state at energy E is given

Fig. 2.10 Zero gate-bias

VFB ¼ UMS Qieff =Ci Qss ðVs ¼ 0Þ =Ci ð2:38bÞ

2.2.4 Evaluation of Surface Potential

Fig. 2.12 Normalised high

where, as discussed in the previous section, Qss can be approximated by

and G0 (Vs,VF) is given by:

Where xd(max) is the limiting maximum thickness of the surface depletion

and Nss can then be derived from dQss/dVs.

2.3 Electron-Hole Pair Generation and Recombination

Electron-hole pair generation and recombination are fundamental semiconductor

In order to develop the expressions for electron-hole pair generation and

2.3.1 Thermal Equilibrium

and n0 p0 ¼ n2i ð2:48Þ

rn rp vth NT ½n2i np

rvth NT ½n2i np

rvth NT ½n2i np