Introduction to Flat Panel Displays
Wiley-SID Series in Display Technology Series Editor: Anthony C. Lowe Consultant ...
648 downloads
2384 Views
13MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
Introduction to Flat Panel Displays
Wiley-SID Series in Display Technology Series Editor: Anthony C. Lowe Consultant Editor: Michael A. Kriss
Display Systems: Design and Applications Lindsay W. MacDonald and Anthony C. Lowe (Eds) Electronic Display Measurement: Concepts, Techniques, and Instrumentation Peter A. Keller Projection Displays Edward H. Stupp and Matthew S. Brennesholtz Liquid Crystal Displays: Addressing Schemes and Electro-Optical Effects Ernst Lueder Reflective Liquid Crystal Displays Shin-Tson Wu and Deng-Ke Yang Colour Engineering: Achieving Device Independent Colour Phil Green and Lindsay MacDonald (Eds) Display Interfaces: Fundamentals and Standards Robert L. Myers Digital Image Display: Algorithms and Implementation Gheorghe Berbecel Flexible Flat Panel Displays Gregory Crawford (Ed.) Polarization Engineering for LCD Projection Michael G. Robinson, Jianmin Chen, and Gary D. Sharp Fundamentals of Liquid Crystal Devices Deng-Ke Yang and Shin-Tson Wu Introduction to Microdisplays David Armitage, Ian Underwood, and Shin-Tson Wu Mobile Displays: Technology and Applications Achintya K. Bhowmik, Zili Li, and Philip Bos (Eds) Photoalignment of Liquid Crystalline Materials: Physics and Applications Vladimir G. Chigrinov, Vladimir M. Kozenkov and Hoi-Sing Kwok Projection Displays, Second Edition Matthew S. Brennesholtz and Edward H. Stupp Introduction to Flat Panel Displays Jiun-Haw Lee, David N. Liu and Shin-Tson Wu
Introduction to Flat Panel Displays By Jiun-Haw Lee National Taiwan University, Taiwan
David N. Liu Industrial Technology Research Institute, Taiwan
Shin-Tson Wu University of Central Florida, USA
This edition first published 2008 © 2008 John Wiley & Sons Ltd. Registered office John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester, West Sussex, PO19 8SQ, United Kingdom For details of our global editorial offices, for customer services and for information about how to apply for permission to reuse the copyright material in this book please see our website at www.wiley.com. The right of the author to be identified as the author of this work has been asserted in accordance with the Copyright, Designs and Patents Act 1988. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, except as permitted by the UK Copyright, Designs and Patents Act 1988, without the prior permission of the publisher. Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may not be available in electronic books. Designations used by companies to distinguish their products are often claimed as trademarks. All brand names and product names used in this book are trade names, service marks, trademarks or registered trademarks of their respective owners. The publisher is not associated with any product or vendor mentioned in this book. This publication is designed to provide accurate and authoritative information in regard to the subject matter covered. It is sold on the understanding that the publisher is not engaged in rendering professional services. If professional advice or other expert assistance is required, the services of a competent professional should be sought. Library of Congress Cataloging-in-Publication Data Lee, Jiun-Haw. Introduction to flat panel displays / by Jiun-Haw Lee, David N. Liu, and Shin-Tson Wu. p. cm. Includes bibliographical references and index. ISBN 978-0-470-51693-5 (cloth) 1. Flat panel displays. I. Liu, David N. II. Wu, Shin-Tson. III. Title. TK7882.I6L436 2008 621.3815 422—dc22 2008032204 A catalogue record for this book is available from the British Library. ISBN: 978-0-470-51693-5 Set in 9/11pt Times by Integra Software Services Pvt. Ltd, Pondicherry, India Printed in Great Britain by Antony Rowe Ltd, Chippenham, Wiltshire
Contents
Series Editor’s Foreword About the authors Preface Acknowledgements
xi xiii xv xvii
1 Introduction 1.1 Flat panel displays 1.2 Emissive and nonemissive displays 1.3 Display specifications 1.3.1 Physical parameters 1.3.2 Brightness and color 1.3.3 Contrast ratio 1.3.4 Spatial and temporal characteristics 1.3.5 Efficiency and power consumption 1.3.6 Flexible displays 1.4 Applications of flat panel displays 1.4.1 Liquid crystal displays 1.4.2 Light-emitting diodes 1.4.3 Plasma display panels 1.4.4 Organic light-emitting devices 1.4.5 Field emission displays References
1 1 3 3 3 5 5 5 6 6 6 7 7 8 8 9 9
2 Color science and engineering 2.1 Introduction 2.2 The eye 2.3 Colorimetry 2.3.1 Trichromatic space 2.3.2 CIE 1931 colorimetric observations 2.3.3 CIE 1976 uniform color system 2.3.4 Color saturation and color gamut 2.3.5 Light sources
11 11 12 15 15 16 19 21 22
vi
Contents
2.3.5.1 Sunlight and blackbody radiators 2.3.5.2 Backlights of transmissive displays 2.3.5.3 Color rendering index 2.3.6 Photometry 2.4 Production and reproduction of colors Homework problems References
22 23 24 25 27 28 28
3 Thin-film transistors 3.1 Introduction 3.2 Basic concepts of crystallized semiconductor materials 3.2.1 Band structure of crystallized semiconductors 3.2.2 Intrinsic and extrinsic semiconductors 3.3 Disordered semiconductors 3.3.1 Amorphous silicon 3.3.2 Polycrystalline silicon 3.4 Thin-film transistor characteristics 3.5 Passive matrix and active matrix driving schemes 3.6 Non-silicon-based thin-film transistors Homework problems References
31 31 31 32 36 38 39 41 43 47 53 55 56
4 Liquid crystal displays 4.1 Introduction 4.2 Transmissive thin-film transistor liquid crystal displays 4.3 Liquid crystal materials 4.3.1 Phase transition temperatures 4.3.2 Eutectic mixtures 4.3.3 Dielectric constants 4.3.4 Elastic constants 4.3.5 Rotational viscosity 4.3.6 Optical properties 4.3.7 Refractive indices 4.3.7.1 Wavelength effect 4.3.7.2 Temperature effect 4.4 Liquid crystal alignment 4.5 Homogeneous cell 4.5.1 Phase retardation effect 4.5.2 Voltage-dependent transmittance 4.6 Twisted nematic 4.6.1 Optical transmittance 4.6.2 Viewing angle 4.6.3 Film-compensated TN cells 4.7 In-plane switching 4.7.1 Device structure 4.7.2 Voltage-dependent transmittance 4.7.3 Viewing angle 4.7.4 Phase compensation films 4.8 Fringe field switching
57 57 58 60 60 61 62 65 65 66 67 67 68 70 71 72 73 73 74 75 76 78 78 79 79 80 81
Contents
vii
4.9 Vertical alignment 4.9.1 Voltage-dependent transmittance 4.9.2 Response time 4.9.3 Overdrive and undershoot voltage method 4.9.4 Multidomain vertical alignment 4.10 Optically compensated bend cell 4.10.1 Voltage-dependent transmittance 4.10.2 Compensation films for OCB 4.10.3 No-bias bend cell 4.11 Transflective liquid crystal displays 4.11.1 Introduction 4.11.2 Dual cell gap transflective LCDs 4.11.3 Single cell gap transflective LCDs 4.12 Future directions Homework problems References
83 83 83 85 86 88 88 89 91 91 91 93 95 101 101 103
5 Plasma display panels 5.1 Introduction 5.2 Physics of gas discharge 5.2.1 I–V characteristics 5.2.2 Penning reaction and Paschen curve 5.2.3 Priming mechanism 5.3 Plasma display panels 5.3.1 DC PDP 5.3.2 AC PDP 5.3.3 Panel processes 5.4 Front plate techniques 5.4.1 Substrate 5.4.2 Sustain electrode 5.4.3 Dielectric 5.4.4 Protection layer 5.5 Rear plate techniques 5.5.1 Substrate 5.5.2 Address electrode 5.5.3 Dielectric 5.5.4 Barrier rib 5.5.5 Phosphor 5.6 Assembly and aging techniques 5.6.1 Sealing layer formation and panel alignment 5.6.2 Sealing, gas purging and display gas filling 5.6.3 Aging 5.7 System techniques 5.7.1 Cell operation mechanism 5.7.2 Driving 5.7.3 Energy saving 5.7.4 PDP issues Homework problems References
109 109 109 110 111 112 112 112 113 115 117 118 118 119 119 120 121 121 121 122 124 126 126 127 128 128 129 130 130 132 132 132
viii
Contents
6 Light-emitting diodes 6.1 Introduction 6.2 Material systems 6.2.1 AlGaAs and AlGaInP material systems for red and yellow LEDs 6.2.2 GaN-based systems for green, blue and UV LEDs 6.2.3 White LEDs 6.3 Diode characteristics 6.3.1 The p-layer and n-layer 6.3.2 Depletion region 6.3.3 J–V characteristics 6.3.4 Heterojunction structures 6.3.5 Quantum well, quantum wire and quantum dot structures 6.4 Light-emitting characteristics 6.4.1 Recombination model 6.4.2 L–J characteristics 6.4.3 Spectral characteristics 6.5 Device fabrication 6.5.1 Epitaxy 6.5.2 Process flow and device structure design 6.5.3 Extraction efficiency improvement 6.5.4 Package 6.6 Applications 6.6.1 Traffic signals, electronic signage and huge displays 6.6.2 LCD backlight 6.6.3 General lighting Homework problems References
137 137 140 142 143 145 147 148 149 152 153 154 155 156 157 158 161 161 164 165 167 168 169 169 172 173 174
7 Organic light-emitting devices 7.1 Introduction 7.2 Energy states in organic materials 7.3 Photophysical processes 7.3.1 Franck–Condon principle 7.3.2 Fluorescence and phosphorescence 7.3.3 Jablonski diagram 7.3.4 Intermolecular processes 7.3.4.1 Energy transfer process 7.3.4.2 Excimer and exciplex formation 7.3.4.3 Quenching process 7.3.5 Quantum yield calculation 7.4 Carrier injection, transport and recombination 7.4.1 Richardson–Schottky thermionic emission 7.4.2 SCLC, TCLC and PF mobility 7.4.3 Charge recombination 7.4.4 Electromagnetic wave radiation 7.5 Structure, fabrication and characterization 7.5.1 Device structure 7.5.1.1 Two-layer OLED 7.5.1.2 Dopant in the matrix as the EML
177 177 178 179 180 182 183 184 184 185 187 187 189 190 192 193 193 195 196 197 198
Contents
ix
7.5.1.3 HIL, EIL and p–i–n structure 7.5.1.4 Top-emission and transparent OLEDs 7.5.2 Polymer OLEDs 7.5.3 Device fabrication 7.5.3.1 Thin-film formation 7.5.3.2 Encapsulation and passivation 7.5.3.3 Device structures for AM driving 7.5.4 Electrical and optical characteristics 7.5.5 Degradation mechanisms 7.6 Improvement of internal quantum efficiency 7.6.1 Phosphorescent OLEDs 7.6.2 Tandem structure 7.6.3 White OLEDs 7.7 Improvement of extraction efficiency Homework problems References
200 203 204 205 206 209 210 211 213 218 218 220 222 224 225 226
8 Field emission displays 8.1 Introduction 8.2 Physics of field emission 8.2.1 Work function and field enhancement 8.2.2 Vacuum mechanism 8.3 FED structure and display mechanism 8.4 Emitter 8.4.1 Spindt emitter 8.4.2 CNT emitter 8.4.3 Surface conduction emitter 8.5 Panel process 8.6 Field emission array plate techniques 8.7 Phosphor plate techniques 8.8 Assembly and aging techniques 8.8.1 Spacer 8.8.2 Sealing layer formation and panel alignment 8.8.3 Sealing 8.8.4 Evacuation and sealing off 8.8.5 Aging 8.9 System techniques Homework problems References
233 233 233 233 236 237 238 239 240 243 244 247 248 249 251 251 252 252 253 253 254 254
Index
259
Series Editor’s Foreword Article 2 of the bylaws of the Society for Information Display begins “1. The purpose of SID shall be: a) To encourage the scientific, literary and educational advancement of information display and its allied arts and sciences. . .’’. This book series was begun eleven years ago with the express object of extending that encouragement, which in the printed form amounted to publishing conference proceedings and a Journal of peer refereed papers, to the provision of a series of books which would satisfy the needs of scientists and engineers working in the wide and complex field of displays. More recently in 2006, we published “Fundamentals of Liquid Crystal Devices’’ by Deng-Ke Yang and Shin-Tson Wu (who – not coincidentally – is a co-author of this book). That book extended the readership because it was written primarily as a post graduate textbook. This latest volume in the series extends that educational scope still further by describing the operating principles and the methods of fabrication of technologies used or of potential use in flat panel displays, their methods of addressing, systems aspects and the underpinning science. Although general books on flat panel displays have been published in the past, this is the first comprehensive flat panel display textbook to have been written at this academic level. Its readership and its use will extend far beyond post graduate courses as it offers in a single volume material of great value to practising industrial engineers and scientists across the whole range of flat panel technologies. In my foreword, I usually provide a précis of the contents of a book, but the authors have done this so comprehensively that such an effort on my part would be superfluous. It merely remains for me to thank them for the great effort they have put into writing this book and wholeheartedly to commend it to our present and expanding readership. Anthony C Lowe Series Editor Braishfield, UK.
About the authors Jiun-Haw Lee Jiun-Haw Lee received BSEE, MSEE and PhD degrees in electrical engineering in 1994, 1995 and 2000, respectively, all from National Taiwan University, Taipei, Taiwan. From 2000 to 2003 he was with the RiTdisplay Corporation as the director. In 2003 he joined the faculty of National Taiwan University in the Graduate Institute of Photonics and Optoelectronics and the Department of Electrical Engineering, where he is currently an associate professor. His research interests include organic light-emitting devices, display technologies and solid-state lighting. Dr Lee is a member of the IEEE, OSA, MRS and SPIE. He received the Exploration Research Award of Pan Wen Yuan Foundation and Lam Research Award in both 2005 and 2006. He has published over 40 journal papers, 100 conference papers and 20 issued patents. David N. Liu David N. Liu has been the director of the Strategic Planning Division in the Display Technology Center (DTC) of the Industrial Technology Research Institute (ITRI) since 2006. He worked on IC and field emission displays at ERSO (Electronics Research and Service Organization)/ITRI and Bellcore (Bell Communication Research) from 1983 to 1996. He started his research and development work on plasma display panels at Acer Peripheral Inc. and AUO from 1996 to 2002. After his service at AUO, he was in charge of the flat panel display technology division in ERSO/ITRI until 2006. Dr Liu received his PhD degree in electrical engineering from New Jersey Institute of Technology in 1992. He has over 45 issued patents, 18 published papers and a contributed chapter of the Semiconductor Manufacturing Handbook (McGraw-Hill, 2005). He also successfully developed field emission displays, plasma display panels and flat panel displays followed by the receipt of many awards from ITRI, Photonics Industry and Technology DevelopmentAssociation,Administration Bureau of Science Base Industry Park and the Ministry of Economic Affairs (MOEA). He was also a recipient of the Outstanding Project Leader Award from MOEA in 2006. Shin-Tson Wu Shin-Tson Wu is a PREP professor at the College of Optics and Photonics, University of Central Florida (UCF). Prior to joining UCF in 2001, Dr Wu worked at Hughes Research Laboratories (Malibu, California) for 18 years. He received his PhD in physics from the University of Southern California (Los Angeles) and BS in physics from National Taiwan University (Taipei). Prof. Wu has co-authored four books: Fundamentals of Liquid Crystal Devices (Wiley, 2006), Introduction to Microdisplays (Wiley, 2006), Reflective Liquid Crystal Displays (Wiley, 2001) and Optics and Nonlinear Optics of Liquid Crystals (World Scientific, 1993), six book chapters, over 300 journal publications and 75 issued and pending patents.
xiv
About the authors
Prof. Wu is a fellow of the IEEE, OSA, SID and SPIE. He is a recipient of the SPIE G.G. Stokes award, SID Jan Rajchman Prize, SID Special Recognition Award, SID Distinguished Paper Award, Hughes team achievement award, Hughes Research Laboratories outstanding paper award, UCF Distinguished Researcher Award and UCF Research Incentive Award. He was the founding editor-in-chief of the IEEE/OSA Journal of Display Technology.
Preface Flat panel displays (FPDs) are everywhere in our daily lives: mobile phones, notebooks, monitors, TVs, traffic signals and electronic signage are a few examples. Several FPD technologies, such as liquid crystal displays (LCDs), plasma display panels (PDPs), light-emitting diodes (LEDs), organic light-emitting devices (OLEDs) and field emission displays (FEDs), have been developed. They coexist because each technology has its own unique properties and applications. However, due to the diversity of display materials and operating mechanisms, there has not been a textbook covering the fundamental physics of such a wide spectrum of display technologies. There are books dedicated to a specific display technology or book chapters covering different display technologies. This book is intended as a textbook for senior undergraduate and graduate students with a wide variety of backgrounds, such as electrical engineering, electronics, material science, applied physics and optical engineering. It can also be used as a reference book for engineers and scientists working in display industries. Parts of the material in this book and its organization follow the course ‘Introduction to display technologies’, which has been taught by Jiun-Haw Lee in the Graduate Institute of Photonics and Optoelectronics (GIPO) and Department of Electrical Engineering, National Taiwan University (NTU), Taipei, Taiwan, since 2003. This book introduces basic operation principles and underlying physics for thin-film transistors (TFTs) LCDs, PDPs, LEDs, OLEDs and FEDs in each chapter. The LCD is a nonemissive display. From the electrical viewpoint, each pixel is a light switch driven by a TFT. To reduce leakage current of the capacitor, the liquid crystal material should have a high resistivity. Moreover, to achieve a high contrast ratio, most direct-view TFT LCDs require two absorption-type sheet polarizers. These polarizers not only reduce the light efficiency but also limit the LCD’s viewing angle. Therefore, phase compensation films are required for wide-view LCDs. In contrast, the PDP is an emissive display. It can be considered as consisting of millions of miniature fluorescent lamps on a single panel. LEDs and OLEDs are electroluminescent devices with crystallized semiconductors and amorphous organic materials, respectively. Compared with liquid crystal materials which are also organic compounds, OLED materials should exhibit a low resistivity to reduce ohmic losses. A FED is a type of flat cathode ray tube, which has all the advantages of this mature technology. In this book, both basic physics and practical issues (such as material requirements, device configurations, fabrication methods and driving techniques) of different display technologies are addressed. Each display technology is at a different development stage; some are more mature than others. Generally speaking, they are still advancing so rapidly that it is difficult to keep up with the technological advancements. Thus, in this introductory book we have decided to emphasize the fundamental science and only highlight the key technological advancements of each technology. Another objective of this book is to provide background knowledge for readers from interdisciplinary fields to stimulate new ideas. Since display technologies cover very broad scientific spectra, any breakthrough from any aspect may result in substantial progress in this industry. Sometimes there is not only competition but also cooperation among different display technologies. For example, LCDs and LEDs are distinct technologies for different display applications. However, LEDs can be also used as
xvi
Preface
backlights for LCDs. As a result, the color gamut is widened, the dynamic contrast ratio is enhanced and power consumption is reduced. After reading this book, one may expect to have a whole picture of display technologies from scientific, technical and engineering viewpoints. There are different kinds of technologies suitable for different sizes (ranging from smaller than an inch to more than a hundred inches in diagonal measurement) and applications (such as outdoor, indoor and mobile displays). Furthermore, this book may serve as a stepping stone to more advanced research and development. The organization of this book is as follows. Chapter 1 introduces the classifications and specifications of display technologies, which are guidelines for developing a display and judging performance. Applications suitable for different technologies (LCD, PDP, LED, OLED and FED) are also illustrated. Displays are used to produce or reproduce color images. In Chapter 2 we introduce the stages of color formation from a scientific viewpoint. Then, the chromaticity diagram is used to quantitatively describe colors. Finally, one can use the background of color science to engineer the color performance of a display. Chapter 3 describes the TFTs based on semiconductor material, which are used to drive LCDs and OLEDs. Since this is an introductory textbook, some basic semiconductor physics are first introduced, which is also useful knowledge for Chapter 6. Material aspects of amorphous silicon and polycrystalline silicon are discussed. Then, device structures and their performances are introduced. Finally, driving techniques and circuits for LCDs and OLEDs are demonstrated. Emerging TFT technologies, such as organic and oxide TFTs, are briefly discussed. In Chapter 4 we begin with basic liquid crystal compound structures, mixture formulations and their physical properties, and then extend the discussion to device structures and display characteristics. Three major LCDs are introduced: transmissive, reflective and transflective. Most modern LCDs are of the transmissive type. However, these displays might be washed out by direct sunlight. In contrast, reflective displays work well under sunlight but are not readable in dark ambient. To retain the good images of a transmissive display while keeping good sunlight readability, transflective LCDs have been developed. Chapter 5 gives an overview of PDP fundamentals. We begin with a discussion of the physics of a gas discharge, covering the reactions of gas discharges and I–V characteristics. DC PDP and AC PDP panels as well as surface discharge and vertical discharge approaches are introduced. The panel process technologies and useful process approaches are also described. Finally, we discuss system techniques with cell operation and driving mechanism. Semiconductor LEDs are discussed in Chapter 6. We start from the material system because this determines the emission wavelength. Electrical properties of LEDs, typically p–n junctions, and corresponding optical characteristics are then discussed. The fabrication process is introduced, which highlights the practical electrical, optical and thermal issues. Finally, applications of LEDs for displays are described. Chapter 7 describes OLEDs, with fabrication processes and operation principles similar to LCDs and LEDs, respectively. The chapter starts from the material aspect. Opto-physical processes in an organic material are introduced. Electrical injection and transport in organic materials are then described. Device structures and fabrication are then discussed. One serious disadvantage of an OLED is its short lifetime; this issue is also addressed. In Chapter 8 an overview of FED fundamentals is provided. We begin by discussing the physics of field emission, covering the field enhancement and vacuum mechanism. FED structure, display mechanism and various emitters are introduced. The advantages and disadvantages of using low- and high-voltage phosphor are compared. The panel process technology and useful process approaches are also described. Finally, system techniques are discussed. Jiun-Haw Lee, Taiwan David N. Liu, Taiwan Shin-Tson Wu, Florida, USA
Acknowledgements Jiun-Haw Lee would like to thank his colleagues Profs. I-Chun Cheng, Chih-I Wu, Jian-Jang Huang, Yuh-Renn Wu, Hoang-Yan Lin and Ding-Wei Huang of GIPO, NTU, for many helpful discussions. Mr Jia-Xing Lin of ITRI is gratefully acknowledged for kindly providing useful information about TFT technologies. Dr Lee is also grateful to his students in NTU and Dr Zhibing Ge of the University of Central Florida, who helped to prepare drawings, references, homework problems and examples, together with providing valuable remarks and comments from a reader’s perspective. David N. Liu is grateful to his colleagues in ITRI and AUO for useful discussions, and Ted Knoy for his professional proofreading. In particular, he would like to express his gratitude to his wife Janice for her patience and support during the period of writing the book. Shin-Tson Wu is deeply indebted to his present and former group members at the University of Central Florida for their numerous technical contributions, and to Chi-Mei Optoelectronics for the funding support. He is grateful to his wife Cho-Yan for spiritual support during the writing of the book.
1 Introduction 1.1 Flat panel displays A display is an interface containing information which stimulates human vision. Information may be pictures, animation, movies and articles. One can say that the functions of a display are to produce or reproduce colors and images. Using ink to write, draw or print on paper is a traditional display, like a painting or a book. However, the content of such a traditional display is motionless and typically inerasable. In addition, a light source, synthetic or natural, is needed for reading a book or seeing a picture. There are lots of electronic displays that use an electronic signal to create images on a panel and stimulate the human eye. Typically, they can be classified as emissive and nonemissive. Emissive displays emit light from each pixel which constitutes an image on the panel. In contrast, nonemissive displays modulate light, by means of absorption, reflection, refraction and scattering, to display colors and images. For a nonemissive display, a light source is needed. Hence, these can be classified into transmissive and reflective displays. One of the most successful display technologies for home entertainment is the cathode ray tube (CRT), which is in widespread use in televisions (TVs). CRT is already a mature technology which has the advantages of self-emission, wide viewing angle, fast response, good color saturation, long lifetime and good image quality. However, a major disadvantage is its bulky size. The depth of a CRT is roughly equal to the length and width of the panel. For example, a monitor’s depth is about 40 cm for a 19-inch (38.6 cm × 30.0 cm) CRT with an aspect ratio of 4:3. Hence, it is not very portable. The bulky size and heavy weight limit its applications. In this book, we introduce various types of flat panel displays (FPDs). As the name implies, these displays have a relatively thin profile, i.e. several centimeters or less. For instance, the liquid crystal display (LCD) is presently the dominant FPD technology with diagonal sizes ranging from less than 1 inch (microdisplay) to over 100 inches. Such a display is usually driven by thin-film transistors (TFTs). A liquid crystal (LC) is a light modulator because it does not emit light. Hence, a backlight module is required for a transmissive LCD. In most LCDs, two crossed polarizers are employed in order to obtain a high contrast ratio. The use of two polarizers limits the maximum transmittance to about 35–40 %, unless a polarization conversion scheme is implemented. Moreover, the optical axes of two crossed polarizers are no longer perpendicular to each other when viewed at oblique angles. A LC is a birefringent medium which means its electro-optic effects are dependent on the incident light direction. Therefore, the viewing angle of a LCD is an important issue. Most wide-view LCDs require multiple optical phase compensation films; one for compensating the crossed polarizer and another for the birefringent LC. Film-compensated transmissive LCDs exhibit a high contrast ratio, high resolution, crisp image, good color saturation and wide viewing angle. However, the displayed images can be washed out under
Introduction to Flat Panel Displays c 2008 John Wiley & Sons, Ltd
J.-H. Lee, D.N. Liu and S.-T. Wu
2
Introduction to Flat Panel Displays
direct sunlight. For example, if we use a notebook computer at outdoor ambient, the images may not be readable. This is because the reflected sunlight from the LCD surface is much brighter than that transmitted from the backlight so that the signal-to-noise ratio is low. A broadband antireflection coating will definitely help to improve the sunlight readability. Another way to improve sunlight readability is to use reflective LCDs.1 A reflective LCD uses ambient light to produce the displayed images. It does not carry a backlight; thus, its weight is reduced. A wristwatch is such an example. Most reflective LCDs have inferior performances compared to the transmissive ones in contrast ratio, color saturation and viewing angle. Moreover, at dark ambient a reflective LCD is not readable. As a result, its application is rather limited. To overcome the sunlight readability issue while maintaining high image quality, a hybrid display called a transflective liquid crystal display (TR-LCD) has been developed.2 In a TR-LCD, each pixel is divided into two subpixels: transmissive (T) and reflective (R). The area ratio between T and R can be adjusted depending on the application. For example, if the display is mostly used outdoors, then one can design to have 80 % reflective area and 20 % transmissive area. In contrast, if the display is mostly used indoors, then one can have 80 % transmissive area and 20 % reflective area. Within this TR-LCD family, there are still some varieties: double cell gap versus single cell gap, and double TFTs versus single TFT. These approaches are trying to solve the optical path length disparity between the T and R subpixels. In the transmissive mode the light from the backlight unit passes through the LC layer once, but in the reflective mode the ambient light traverses the LC medium twice. To balance the optical path length, we could make the cell gap of the T subpixels twice as thick as that of the R subpixels. This is the so-called dual cell gap approach. The single cell gap approach has a uniform cell gap throughout the T and R regions. To balance the different optical path lengths, several approaches have been developed, e.g. dual TFTs, dual fields (stronger field for T region and weaker field for R region) and dual alignments. Presently, the majority of TR-LCDs adopt the double cell gap approach for two reasons: (1) both T and R modes can achieve maximum light efficiency, and (2) the gamma curve matching between the voltage-dependent transmittance (VT) and reflectance (VR) is almost perfect. However, the double cell gap approach has two shortcomings: first, the T region has a slower response time than the R region because its cell gap is about twice as thick as that of the R region; second, the viewing angle is relatively narrow, especially when homogeneous cells are employed. To widen the viewing angle, a special rod-like LC polymeric compensation film has to be used. Chapter 4 gives detailed descriptions of various types of LCDs. A plasma display panel (PDP) is an emissive display which can be thought of as very many miniature fluorescent lamps on a panel. As an emissive display it typically has a better display performance, such as good color saturation and wide viewing angle. Due to the limitation of fabrication, the pixel size of a PDP cannot be too small. For a finite pixel size, the video content is increased by enlarging the panel size. PDPs are suitable for large-screen applications. In 2008, Panasonic demonstrated a 150-inch PDP TV with 4096 × 2160 pixels. This resolution is four times higher than that of the present full high-definition television (HDTV). Light-emitting diodes (LEDs) and organic light-emitting devices (OLEDs) are electroluminescent devices with semiconductor and organic materials, respectively. Electrons and holes recombine within the emissive materials, where the bandgap of the materials determines the emission wavelength. A field emission display (FED) uses sharp emitters to generate electrons. These electrons bombard the phosphors that are present to emit red (R), green (G) and blue (B) light. A FED is like a ‘flat’ CRT. Due to the mature technologies developed in CRTs, FEDs exhibit all the advantages of CRTs plus the smaller panel thickness. Compared to conventional displays (such as books, magazines and newspapers), electronic displays (such as TVs, mobile phones and monitors) are rigid because they are typically fabricated on glass substrates. Flexible FPDs are emerging. Several approaches have been developed, such as electrophoretic displays and polymer-stabilized cholesteric displays. Flexible displays are thin, robust and lightweight. In the remainder of this chapter, we first introduce FPD classifications in terms of emissive and nonemissive displays, where nonemissive displays include transmissive and reflective displays. Specifications
Introduction
3
of FPDs are then outlined. Finally, the FPD technologies described in the later chapters of this book are briefly introduced.
1.2 Emissive and nonemissive displays Both emissive and nonemissive FPDs have been developed. For emissive displays, each pixel emits light with different intensity and color which stimulate the human eye directly. CRTs, PDPs, LEDs, OLEDs and FEDs are emissive displays. An emitter is called Lambertian when the luminances from different viewing directions are the same. Most emissive displays are Lambertian emitters which results in a wide viewing angle performance. Also, due to the self-emissive characteristics, they can be used even under very low ambient light. When such displays are turned off, they are completely dark (ignoring the ambient reflection). Hence, display contrast ratios (see also Section 1.3.3) are high. Displays that do not emit light themselves are called nonemissive displays. A LCD is a nonemissive display in which the LC molecules in each pixel work as an independent light switch. The external voltage reorients the LC directors which causes phase retardation. As a result, the incident light from the backlight unit or ambient is modulated. Most high-contrast LCDs use two crossed polarizers. The applied voltage controls the transmittance of the light through the polarizers. If the light source is behind the display panel, the display is called a transmissive display. It is also possible to use ambient light as the light source. This resembles the concept of a conventional display, such as reading a book, which is called a reflective display. Since no backlight is needed in a reflective display, its power consumption is relatively low. In a very bright environment, images of emissive displays and transmissive LCDs can be washed out. In contrast, reflective displays exhibit an even higher luminance as the ambient light increases. However, they cannot be used in a dim environment. Hence, transflective LCDs have been developed, which are described in Chapter 4.
1.3 Display specifications In this section, we introduce some specifications which are generally used to describe and judge FPDs from the viewpoints of mechanical, electrical and optical characteristics. FPDs can be smaller than 1 inch for projection displays, 2–4 inches for mobile phones and personal digital assistants, 7–9 inches for car navigation systems, 8–18 inches for notebook computers, 10–25 inches for desktop computers and more than 100 inches for direct-view TVs. For different FPDs, their requirements for pixel resolutions also differ. Luminance and color are two important characteristics which directly affect the display performances. Dependences of these two parameters to viewing angles, uniformity, lifetime and response time should be addressed when describing the performances of an FPD. Contrast ratio is another important parameter, which changes with different ambient environments.
1.3.1 Physical parameters The basic physical parameters of an FPD include display size, aspect ratio, resolution and pixel format. The size of a display is typically described by diagonal length, in units of inches. For example, a 15-inch display means the diagonal of the viewable area of this display is 38.1 cm. There are three kinds of display format: landscape, equal and portrait, corresponding to the display width being larger than, equal to and smaller than its length. Most monitors and TVs use landscape format with a width-to-length ratio, which is called the ‘aspect ratio’, of 4:3, 16:9 or 16:10, typically. An FPD typically consists of a ‘dot matrix’ which can display images and characters. To increase resolution, one may use more dots in a display. Table 1.1 lists some standard resolutions of FPDs. For
4
Introduction to Flat Panel Displays
Table 1.1
Resolution of FPDs.
Abbreviation
Full name
Resolution
VGA SVGA XGA SXGA UXGA WXGA WSXGA WUXGA
Video graphics array Super video graphics array Extended graphics array Super extended graphics array Ultra extended graphics array Wide extended graphics array Wide super extended graphics array Wide ultra extended graphics array
640 × 480 800 × 600 1024 × 768 1280 × 1024 1600 × 1200 1366 × 768 1680 × 1050 1920 × 1200
example, VGA means the display is 640 dots in width and 480 dots in length. Higher resolution typically (but not necessarily) means better image quality. There are some resolutions listed in Table 1.1 starting with the letter ‘W’, which means wide screen with an aspect ratio larger than 4:3. Once the resolution, display size and aspect ratio are known, one may obtain the pitch of the pixels. For example, a 19-inch display with aspect ratio of 4:3 and resolution of UXGA has a pitch of 190.5 m. Note that not all of the pixel area contributes to the display. One can define the ‘fill factor’ or ‘aperture ratio’ as the ratio of the display area in a pixel over the whole pixel size, with its maximum value of 100 %. Besides, for a full-color display, at least three primary colors are needed to compose a color pixel. Hence, each color pixel is divided into three subpixels (RGB) sharing the area. For example, let us assume a color pixel has size of 240 m × 240 m; then the dimension of each subpixel is 80 m × 240 m. If the fill factor is 81 % which actually contributes to light emission or transmission, then the usable pixel area is reduced to 72 × 216 m2 . There are different layouts for RGB subpixels, as shown in Figure 1.1. For the stripe configuration, it is straightforward and easy for fabrication and driving circuit design. However, it has a poor color mixing performance for the same display area and resolution. For mosaic and delta configurations, their fabrication and/or driving circuit are more complicated but their image quality is better because of better color mixing capability. Also, displays with mosaic and delta configurations exhibit faster response times since the moving distance between the pixels is shorter. Actually, as the resolution gets high enough the subpixel arrangement becomes less critical. For medium and large displays, the stripe configuration is typically used. In contrast, for a small-size display which requires high resolution, e.g. video cameras, one may use the mosaic or delta configuration.
* R G B (a) Figure 1.1
(b)
(c)
Subpixel layout of an FPD: (a) stripe, (b) mosaic and (c) delta configurations.
Introduction
5
1.3.2 Brightness and color Luminance and color are two important optical characteristics of an FPD. A display with high luminance looks dazzling in a dark room. On the other hand, a display with insufficient brightness appears washed out under high ambient. Typically, the luminance of an FPD should be as bright as (or slightly brighter than) the real object. Under an indoor lighting environment, a monitor has a luminance of 200–300 cd m−2 (Section 2.3.6). For a large-screen TV, a higher luminance (500–1000 cd m−2 ) may be needed. An FPD is used to produce or reproduce colors; hence, how many colors of an FPD and how real the color is (color fidelity) between an FPD and a real object are two important characteristics of an FPD. Since the color of an FPD is mixed by (at least) three primary colors, i.e. RGB, more ‘pure’ (saturated) primaries results in a broader range of the possibly displayed colors, which is called ‘color gamut’ (Section 2.3.4). One can equally divide the stimuli to the eyes from dark to bright with 2, 4, 8 or more spacings, which is called ‘gray level’ or ‘gray scale’ (Section 2.3.3). For example, an FPD can display 16 million colors (28 × 28 × 28 ≈ 16.8 million) when each RGB subpixel is divided into 8 gray scales.
1.3.3 Contrast ratio The device contrast ratio (CR) of an FPD is defined as CR =
Lw , Lb
(1.1)
where L w and L b are the luminance at white and black states, respectively. Higher CR means higher on/off ratio and hence better image quality and higher color saturation. When CR is equal to or less than 1, the human eye cannot distinguish the on and off colors so that the information content of an FPD is lost or distorted. For most emissive displays, the off-state luminance is zero. Hence, the contrast ratio is infinity in a perfectly dark room. However, due to the surface reflection from the ambient, Equation (1.1) should be modified to Lw + Lar A-CR = , (1.2) Lb + Lar where A-CR is the ambient contrast ratio and L ar is the luminance from ambient reflection. A-CR is used to specify the ambient contrast ratio, to distinguish from the intrinsic ‘device’ contrast ratio as described in Equation (1.1). From Equation (1.2), as the ambient reflection increases, A-CR decreases sharply. To keep a good ambient contrast, one can: (1) increase the on-state luminance, and (2) reduce the reflectivity of the display surface. However, for a very strong ambient, e.g. in sunshine outdoors, luminance from the direct sun is four orders of magnitude higher than that of an FPD, which severely washes out the information content of the FPD. Sunlight readability is an important issue especially for mobile displays. In contrast, an adequate ambient light is required for conventional displays, such as books or newspapers. A similar situation applies to reflective displays, such as reflective LCDs.
1.3.4 Spatial and temporal characteristics Uniformity of an FPD means the luminance and color change over a display area. Human eyes are sensitive to luminance and color differences. For example, a 5 % luminance difference is noticeable between two adjacent pixels. For a gradual change, human eyes can tolerate up to 20 % luminance change over the whole display. Optical characteristics (luminance and colors) may also change at different viewing angles. For Lambertian emitters, such as CRTs, PDPs and FEDs, viewing angle performances are quite good. The emission profile of LEDs and OLEDs can be engineered by packaging and layer structure. However, the viewing angle of LCDs is one of the major issues because LC material is birefringent and crossed
6
Introduction to Flat Panel Displays
polarizers are no longer crossed when viewed at oblique angles. There are several ways to define the viewing angle of an FPD. For example, to find the viewing cone with: (1) a luminance threshold; (2) minimum contrast ratio, say 10:1; or (3) maximum value of color shift. For some cases that contrast ratio is smaller than 1; this is called ‘gray level inversion’. Response time is another important metric. If an FPD has a slow response time, one may see blurred images for fast moving objects. By switching the pixel from ‘off’ to ‘on’ and from ‘on’ to ‘off’, and calculating the time required from 10 to 90 % and 90 to 10 % luminance levels, one can obtain rise and fall time, respectively. One may also define the response time from one gray level to another, which is called the ‘gray-to-gray’ (GTG) response time. Most display scenes contain rich grayscales. Therefore, GTG response time is more meaningful. For LCDs, this GTG response time can be much longer than the black-to-white rise and fall time.3 A TFT is a holding type of active matrix. It is different from the CRT’s impulse type. Therefore, a motion picture response time4 is commonly used to define the response time of a TFT LCD. After long-term operation, the luminance of an FPD (especially an emissive display) decays. In an emissive display, if a fixed pattern is lit on for a long period of time before all the pixels are turned on for the full white screen, one can see nonuniformity of the fixed pattern with a lower brightness, which is called the ‘residual image’. As mentioned before, the human eye can detect less than 5 % nonuniformity between two adjacent pixels. Hence the lifetime of an FPD is crucial for static images. An alternative solution is to use moving pictures, rather than static images, for information display. Then the luminances of all pixels decay uniformly, since the average on time for all pixels is the same.
1.3.5 Efficiency and power consumption Power consumption is a key parameter, especially for mobile displays, as it affects battery life. For displays with wall-plug electrical input, lower power consumption implies lower heat generation, which means heat dissipation is less serious. Typically, one uses the unit lm W−1 to describe power efficiency of an FPD (Section 2.3.6). Lumen (lm) and watt (W) are units for describing light output and electrical input. A portable display with lower power consumption leads to a longer battery life. For notebooks and TVs, high optical efficiency also translates into less heat dissipation and a lower electricity bill. Thermal management in a small-chassis notebook is an important issue.
1.3.6 Flexible displays An FPD is usually fabricated on thin glass plates. Glass is a kind of rigid substrate. In contrast, conventional displays are printed on paper, which is flexible. An interesting research topic is to fabricate FPDs on flexible substrates, as a ‘paper-like’ display.5 Compared to the glass-based FPDs, flexible displays are thin and lightweight. Also, flexible displays can be fabricated by the roll-to-roll process, which is potentially of low cost. Substrate selection of flexible FPDs includes ultrathin glass, plastic and stainless steel. Bendable ultrathin glass substrate is achievable, but the cost is high. Plastic substrate is suitable for flexible displays, but the highest durable temperature is typically lower than 200 ◦ C. Stainless steel substrate is bendable, and durable for high temperature; however, it is opaque hence not suitable for transmissive displays. There are many technical bottlenecks for flexible FPDs, such as material selection, fabrication processes, device configurations, display package and measurement.
1.4 Applications of flat panel displays The following subsections briefly outline the applications of each technology. Detailed mechanisms are described in the related chapters.
Introduction
7
1.4.1 Liquid crystal displays Although LC materials were discovered more than a century ago,6,7 their useful electro-optic effects and stability were developed only in the late 1960s and 1970s. In the early stage, passive matrix LCDs were found useful in electronic calculators and wristwatches.8 With the advance of TFTs,9 color filters10 and low-voltage LC effects,11 active matrix LCDs have gradually penetrated into the market of notebook computers, desktop monitors and TVs. Today, LCDs have found widespread uses in everyday life, including (1) mobile applications, such as mobile phones, personal digital assistants, navigation systems, notebook personal computers; (2) office applications, such as desktop computers and video projectors; and (3) home applications, such as large-screen TVs.12 To satisfy these wide-spectrum applications, three types of LCDs have been developed: transmissive, reflective and transflective. Transmissive LCDs can be further separated into projection and direct-view. In a small-size, high-resolution LCD, the pixel size is around 40 m × 40 m. Here, the aperture ratio becomes particularly important because it affects the light throughput.13 To enlarge the aperture ratio, poly-silicon (p-Si) TFTs are commonly used because their electron mobility is about two orders of magnitude higher than that of amorphous silicon (a-Si). High mobility allows a smaller TFT to be used which, in turn, enlarges the aperture ratio. For the detailed structure of a TFT LCD, see Figure 4.1. For direct-view transmissive TFT LCDs, the pixel size (∼300 m × 300 m) is much larger than that of a microdisplay. Thus, a-Si is adequate although its electron mobility is relatively low. Amorphous silicon is easy to fabricate and has good uniformity. Thus, a-Si TFTs dominate the large-screen (>10 inches) LCD panel market. Similarly, reflective LCDs can also be divided into projection and direct-view displays. In projection displays using liquid-crystal-on-silicon (LCoS),14 the pixel size can be as small as ∼10 m × 10 m because of the high electron mobility of crystalline silicon (c-Si). In a LCoS device, the electronic driving circuits are hidden beneath the metallic reflector. Therefore, the aperture ratio can reach 90 % and the displayed picture is film-like. In contrast, most reflective direct-view LCDs use a-Si TFTs and a circular polarizer. Their sunlight readability is excellent, but they are not readable in dark ambient. Therefore, the application of reflective direct-view LCDs is rather limited. To maintain high-quality transmissive display and good sunlight readability, a hybrid TR-LCD has been developed. In a TR-LCD, each pixel is divided into two subpixels: one for transmissive and another for reflective display.15 In dark to normal ambient, the backlight is on and the TR-LCD works as a transmissive display. Under direct sunlight, the TR-LCD works in reflective mode. Therefore, its dynamic range is wide and its functionality does not depend on the ambient lighting conditions. TR-LCDs have been widely adopted in portable devices, such as mobile phones. For a detailed discussion of TR-LCDs, see Chapter 4.
1.4.2 Light-emitting diodes A LED is an electroluminescent device based on crystalline semiconductors.16 To convert electrical to optical power, one has to inject carriers into the LED through electrodes, and then they recombine to give light. The emission wavelength is mainly determined by the semiconductor materials, and can be fine tuned by device design. Since it is difficult to grow large-size single crystals, the wafer diameter of LEDs is limited to about 8 inches. After device processing, LEDs are diced from the wafer followed by the package process. The dimension of a single LED is typically several millimeters, which means the ‘pixel size’ of the LED is large. Hence, it is difficult to use a LED as a small display or it will have a very low resolution. An exception is to dice LED arrays from a wafer and use as a microdisplay with a size less than 1 to 2 inches. Due to their self-emissive characteristic, LEDs are commonly used for large displays, such as outdoor signages (single color, multicolor and full color), traffic signals and general lighting to replace light bulbs. Compared to conventional displays enabled by light bulbs, LED displays exhibit the advantages of lower
8
Introduction to Flat Panel Displays
power consumption, greater robustness, longer lifetime and lower driving voltage (so safer). There are also lots of outdoor screens with diagonals of over 100 inches which consist of millions of LED pixels. Rather than a display itself, a LED can also be used as the light source, such as the backlight module for a LCD, and general lighting. Compared to a conventional cold cathode fluorescent lamp (CCFL), which resembles a thin fluorescent tube, a LED exhibits a better color performance, longer lifetime and faster response. Another important driving force to the use of LEDs as LCD backlights is that the mercury in CCFLs is harmful to the environment. When using LEDs for general lighting applications, a broad spectrum is preferred to simulate natural light, such as sunlight, for obtaining a high color rendering of reflective objects (Section 2.3.5). This is quite different from the requirements for LED displays and LCD backlights, which usually need a narrow spectrum.
1.4.3 Plasma display panels The typical structure and operation principle for PDPs are similar to those of a fluorescent lamp. In the structure of a fluorescent lamp, two filament electrodes are formed in two ends of an inner glass tube. The wall of the inner glass tube is coated with phosphor. The cavity of the glass tube is filled with a gas mixture of argon and mercury. When a certain voltage is applied to the electrodes, plasma is generated from a gas discharge. Due to the energy level system of the plasma, ultraviolet (UV) radiation is generated with peak wavelength at λ = 254 nm. The phosphor of the fluorescent lamp is excited by the UV radiation which, in turn, emits light. PDPs use a similar operation mechanism to fluorescent lamps but the gases commonly used in PDPs are neon and xenon instead of the argon and mercury used in fluorescent lamps. Neon and xenon gases generate peak wavelengths at 147 and 173 nm which belong to the vacuum ultraviolet (VUV) region. VUV radiation can only propagate in a vacuum because it is strongly absorbed by air. Although the PDP structure is similar to a fluorescent lamp which is composed of two electrodes, phosphor and gases, an additional barrier rib structure is needed in PDPs to sustain the space between upper plate and lower plate.17 Because of the structure of the barrier rib, the unit cell size of PDPs cannot be made too small. In addition, PDP operation voltage is high because a typical plasma generation is needed. The high operation voltage demands a high voltage driver integrated circuit (IC) and results in a high cost of the electronics. However, PDPs exhibit a wider view angle, faster response time and wider temperature range than LCDs. In other words, PDPs remain good candidates for large-panel displays spanning from static pictures to motion pictures, from cold ambient to hot ambient and from personal use to public use. In addition to these performance advantages, PDPs can be fabricated with a low-cost and simple manufacturing process. For these reasons, many different PDP structures intended for a wide spectrum of applications have been developed.18−20
1.4.4 Organic light-emitting devices An OLED is also an electroluminescent device, like a LED, except its materials are organic thin films with amorphous structures.21 Amorphous organic material has a much lower mobility (typically five order of magnitude lower) than crystalline semiconductors, which results in a higher driving voltage of OLEDs. Also, the operation lifetime of OLEDs is one order of magnitude shorter than semiconductor LEDs. However, due to the amorphous characteristics, fabrication with large size (>40 inches) is possible. Since the conductivity of amorphous organic materials is very low, very thin organic films (100– 200 nm in total) are required to reduce the driving voltage to a reasonable value (i.e. <10 V). This is quite a challenge in thin-film formation, especially for large-size substrates. There are several fabrication technologies proposed, such as physical vapor deposition, spin coating, ink-jet printing and laser-assisted patterning. Prototypes of OLED panels of 40 inches have been demonstrated, and OLEDs of 11 inches
Introduction
9
(or less) are also commercially available.22,23 Another challenge for large displays is the reliability. A TV must have a longer lifetime than a mobile phone. Recently, commercial OLED products for mobile displays and monitors have emerged. However, for TV applications, OLED panel lifetime still falls short. Two advantages of OLEDs are: (1) low process temperature, and (2) nonselective to the substrate material, which is suitable for flexible displays. One of the strategies for OLED development is to improve device performance (especially driving voltage and lifetime) so as to be as good as (or not too much worse than) LEDs. Also, due to the possibility of large-size fabrication, the potential manufacture cost of OLEDs is lower than that of LEDs. Because OLEDs have some advantages in performance and fabrication cost over LEDs, they have a chance to replace LEDs in some applications since they are both electroluminescent devices with similar operation principles. Besides, in comparison with LEDs, OLEDs have two unique advantages: larger panel size and higher resolution.
1.4.5 Field emission displays A FED is a display using electrons generated by field emission to excite phosphors and generate luminance. There are several different approaches to generate electron emission, such as thermionic emission, photoemission and field emission.24,25 Thermionic emission electrons are thermally excited over the potential energy barrier while photoemission electrons are excited over the potential energy barrier by incoming photons. In field emission, the electrons tunnel though the surface potential energy barrier, which has been thinned and shaped by the influence of a strong electric field. The field emitter plays an important role in the electron emission of FEDs. The structure of the field emitter can be in the shape of a cone, a wedge, a cylinder or a tube.26 The emitting region is a tip for a conic-shape emitter while the emitting region is an edge for wedge, cylinder and tube shapes. There are also many types of emitters such as the Spindt emitter, carbon nanotube (CNT) emitter and surface conduction emitter (SCE).27 A Spindt emitter uses sharp conic material as an emitter while a CNT emitter uses a carbon tube of nanometric diameter as an emitter. A SCE uses a material named PdO as an emitter with a nano-gap structure to generate surface electrons. These types of emitters can be undesirably damaged by the ions generated from residual gas. This undesirable damage usually results in a short lifetime of operation. Therefore, less residual gas and lower operating voltage are strongly demanded in FEDs. In order to have a lower operating voltage, an emitter material of low work function with a sharp structure is desired. In addition, vacuum is also required so that residual gas can be eliminated. Both FEDs and CRTs use phosphors to generate visible light which demands a vacuum to ensure long life of electron emission. The structure of a FED consists of an emitter, electrode and phosphor which is similar to that of a CRT. The display performance is also similar to a CRT. However, a spacer structure is needed in a FED to maintain the space between phosphor plate and field emission plate. The emission uniformity which is caused by the process of emitter formation is the major challenge for FEDs. Above all, FED structure is simple because it does not require backlight, color filters, polarizers or other optical films which are needed in LCDs. Furthermore, FEDs have higher luminance efficiency, faster response time, a wider view angle and greater temperature range than LCDs.28 FEDs can be widely applied from static pictures to motion pictures, from cold ambient to hot ambient and from personal use to public use.29
References 1. Wu, S.T. and Yang, D.K. (2001) Reflective Liquid Crystal Displays, John Wiley & Sons, Ltd, Chichester. 2. Okamoto, M., Hiraki, H. and Mitsui, S. (2001) US Patent 6,281,952. 3. Wang, H., Wu, T.X., Zhu, X. and Wu, S.T. (2004) Correlations between liquid crystal director reorientation and optical response time of a homeotropic cell. J. Appl. Phys., 95, 5502. 4. Song, W., Li, X., Zhang, Y. et al. (2008) Motion-blur characterization on liquid-crystal displays. J. SID, 16, 587. 5. Crawford, G.P. (2005) Flexible Flat Panel Displays, John Wiley & Sons, Ltd, Chichester.
10
Introduction to Flat Panel Displays
6. 7. 8. 9.
Reinitzer, F. (1888) Monatsh. Chem., 9, 421. Lehmann, O. (1889) Z. Phys. Chem., 4, 462. Ishii, Y. (2007) The world of liquid crystal display TVs: past, present and future. J. Display Technol., 3, 351. Lechner, B.J., Marlowe, F.J., Nester, E.O. and Tults, J. (1971) Liquid crystal matrix displays. Proc. IEEE, 59, 1566. Fischer, A.G. et al. (1972) Design of a liquid crystal color TV panel. Proceedings of the IEEE Conference on Display Devices, p. 64. Schadt, M. and Helfrich, W. (1971) Voltage-dependent optical activity of a twisted nematic liquid crystal. Appl. Phys. Lett., 18, 127. Liu, C.T. (2007) Revolution of the TFT LCD technology. J. Display Technol., 3, 342. Stupp, E.H. and Brennesholtz, M. (1998) Projection Displays, John Wiley & Sons, Inc., New York. Armitage, D., Underwood, I. and Wu, S.T. (2006) Introduction to Microdisplays, John Wiley & Sons, Ltd, Chichester. Zhu, X., Ge, Z., Wu, T.X. and Wu, S.T. (2005) J. Display Technol., 1, 15. Round, H.J. (1907) A note on carborundum. Electrical World, 19, 309. Fischer-Cripps, A.C., Collins, R.E., Turner, G.M. and Bezzel, E. (1995) Stress and fracture probability in evacuated glazing. Building Environ., 30, 41. Oversluizen, G. and Dekker, T. (2006) High efficacy PDP design. SID Symp. Dig., 37, 1110. Hirakawa, H., Shinohe, K., Tokai, A. et al. (2004) Dynamic driving characteristics of plasma tubes array. SID Symp. Dig., 35, 810. Sano, Y., Nakamura, T., Numomura, K. et al. (1998) High-contrast 50-in color ac plasma display with 1365 × 768 pixels. SID Symp. Dig., 29, 275. Tang, C.W. and Vanslyke, S.A. (1987) Organic electroluminescent diodes. Appl. Phys. Lett., 51, 913. Iino, S. and Miyashita, S. (2006) Printable OLEDs promise for future TV market. SID Symp. Dig., 37, 1463. Hirano, T., Matsuo, K., Kohinata, K. et al. (2007) Novel laser transfer technology for manufacturing large-sized OLED displays. SID Symp. Dig., 38, 1592. Gomer, R. (1961) Theory of Field Emission: Field Emission and Field Ionization, Harvard University Press, Cambridge, MA. Dyke, W.P. and Dolan, W.W. (1956) Field emission, in Advances in Electronics and Electron Physics, Vol. 8 (ed. L. Marton), Academic Press, New York. Liu, D., Ravi, T.S., Gmitter, T. et al. (1991) Fabrication of wedge-shaped silicon field emitters with nm scale radii. Appl. Phys. Lett., 58, 1042. Okuda, M., Matsutani, S., Asai, A. et al. (1998) Electron trajectory analysis of surface conduction electron emitter displays. SID Symp. Dig., 29, 185. Utsumi, T. (1991) Keynote address vacuum microelectronics: what’s new and exciting. IEEE Trans. Electron Dev., ED-38, 2276. Itoh, S. et al. (2007) Development of field-emission display. SID Symp. Dig., 38, 1297.
10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29.
2 Color science and engineering 2.1 Introduction Display systems are used to produce and reproduce color images, which makes the topic of ‘color science and engineering’ very important for evaluating display system performance. Typically, the formation of colors can be described by a four-stage process: (1) the light source, either artificial or natural; (2) light–object interaction, such as reflection, absorption and transmission; (3) stimulation of the eyes; and (4) recognition by the brain. As illustrated in Figure 2.1(a), the human eye sees the color of an object under sunlight, which is a ‘white’ light source, because its spectral bandwidth covers the entire visible range and is of relatively uniform intensity across the range. If there were no light source, there would be no photons to stimulate the human eye and, therefore, no color would be observed. When illuminated, an object (e.g. the paper in Figure 2.1(a)) absorbs some of the incident photons and reflects the rest. As shown in Figure 2.1(b), there are yellow and green inks on the white paper. When the incident white light illuminates the yellow ink, the blue component of the white light is absorbed. The reflected red and green components result in a yellow light. Similarly, the green ink absorbs red and blue components. Wherever there is no ink, the white paper reflects all the incident light, so it appears white. From the above discussion, we can deduce that the color of an object is also dependent on the spectral distribution of the incident light. For example, if the light source is red, then yellow ink will appear red. After the light–object interaction, the reflected photons are received by the detector; in this case a human eye. The human eye can distinguish light intensity and color, but not the polarization state and phase of incident light. Variation of the intensity of incident light gives the observer a perception of bright and dark. Photons of different wavelengths stimulate differently the photosensitive cells (cone and rod cells, which are discussed later) of the eye and this creates the perception of different colors. There are three different cone cells in the human eye, with different spectral sensitivities which make it possible to use three primary colors (red, green and blue) to generate all colors (trichromatic space) and to describe colors quantitatively.1 In 1931 the Commission Internationale de l’Eclairage (CIE) suggested the (X, Y, Z) colorimetric system, which can specify all the colors by their distinct coordinates.2 It is a convenient system for describing colors. However, the 1931 CIE system is not suitable for discussing the magnitude of the perceived difference between two colors. To solve this problem, uniform color spaces are proposed (e.g. the 1976 CIE (L∗ u∗ v∗ )- and (L∗ a∗ b∗ )-spaces).2 In such a system, the perceived color difference for two mismatched colors is nearly identical at different positions of the CIE chart, for equal changes in the values of the coordinates. Since the trichromatic space can be quantitatively described by CIE colorimetric systems, all colors can be produced or reproduced in a display device by mixing three
Introduction to Flat Panel Displays c 2008 John Wiley & Sons, Ltd
J.-H. Lee, D.N. Liu and S.-T. Wu
12
Introduction to Flat Panel Displays
white light (λR + λG + λB)
yellow light (λR + λG) white light
white light (λR + λG +λB)
(λG)
white light
yellow ink
PAPER
green light
green ink
white paper
(b)
(a) Figure 2.1 Formation of colors.
primary emitters. Although the reflection spectrum of a real object may be different from that appearing on the display, the colors look the same to human eyes; this is called ‘metamerism’. In this chapter, we first describe the structure of the human eye and its functionalities. This is followed by the formulation of colorimetry which includes the CIE standards, light sources and photometry. Lastly, we discuss metamerism.
2.2 The eye Figure 2.2(a) shows a schematic diagram of a human eye.3 Incoming light passes through the cornea, the aqueous humor, the eye lens and the vitreous humor, and is received by the retina. The eye lens, with a higher refractive index (n = 1.42) than the cornea, the aqueous humor and vitreous humor (n = 1.33 − 1.37), functions to focus a clear image at the retina, as shown in Figures 2.2(b) and (c).4 The shape of the eye lens can be adjusted by the ciliary muscle around it. Such a system can be approximately described by the Gaussian lens formula:5 1 1 1 (2.1) + = , d1 d2 f
aqueous humor
vitreous humor
lens
cornea
(b)
retina ciliary muscle
optic nerve
blind spot (a) Figure 2.2
(c)
(a) Cross-section of the eye.3 (b, c) Formation of images in the human eye.
Color science and engineering
13
where d 1 is the distance from the object to the eye lens, d 2 is the distance from the eye lens to the retina (which is 17 mm typically) and f is the focal length. The image on the retina is totally reversed (upside down and right-side left). However, after interpretation by the brain, we can recognize images at their normal orientation in real space. When the object is farther away, the eye lens becomes flatter, as shown in Figure 2.2(b). When viewing a closer object, the ciliary muscle will contract the eye lens to increase its curvature (Figure 2.2(c)). The retina receives the incoming photons and transforms them into bio-potential signals. These signals are then transmitted through the optic nerve to the brain giving rise to the perception of vision. The retina is a multilayer structure which can be divided into three parts: (1) photoreceptor; (2) connecting nerve tissue (including outer and inner plexiform and nuclear layers, and ganglion cell layers); and (3) optic nerve.6 There are two kinds of photoreceptor cells in the eye, which are called rod and cone cells, named according to their shapes.7 The dimensions and quantities of the cone and rod cells are listed in Table 2.1. Rods are more sensitive than cones and, thus, are more easily saturated when the ambient illumination is high (e.g. indoor lighting). Also, rods can sense light intensity but not colors. In contrast, the cone cells function well under brighter ambient conditions and can distinguish colors. This explains why people can only see monochrome rather than color images under dark ambient conditions (e.g. a moonless night). Figures 2.3(a) and (b) show the spatial distribution of the photoreceptors. Solid and dashed lines show the mean value and one standard deviation away from the mean value of different specimens studied.8 The symbols show the results obtained from previous reports.9 We can see that the cones have a maximum distribution near the visual axis of the eye. Also, there are almost no rods at the visual axis, because this area is occupied by the cones. The number of rod cells increases and reaches a maximum away from the visual axis. As shown in Figure 2.2(a), there is a blind spot in the human eye where the optic nerve joins the eye and hence there are no cone and rod cells at this point. Typically, in bright ambient the eye is most sensitive within a 10◦ viewing cone. Outside this region, colors are almost indistinguishable. At the blind spot we cannot receive optical signals. However, in actual life, we ‘feel’ that we can see quite a large viewing angle and there is no blind spot. This is because our eyeballs can move and rotate, and because of interpretation by the brain. Figure 2.4(a) shows the normalized spectral sensitivity of human eyes at ‘scotopic’ and ‘photopic’ regions (which mean low- and high-level ambient), in terms of V (λ) and V (λ), respectively. It is not completely true to regard the scotopic and photopic vision regimes as single contributions from the rod and cone cells, respectively. Actually, there is an overlap of the intensity responsivity between rod and cone cells. When the ambient light is brighter than full moonlight and dimmer than indoor lighting, both cones and rods can sense light. Higher intensity saturates the rod cells, but a lower intensity cannot stimulate the cone cells. At low and high ambient, the eye is most sensitive at λ = 507 and 555 nm, respectively. Note that there are three different kinds of cone cells, which have different spectral responses, as shown in Figure 2.4(b). The S-, M- and L-cones are sensitive to short, medium and long wavelengths, respectively. The bio-potential signals from the photoreceptors (rod cells and three kinds of cone cells) are processed to obtain the signal intensities, as shown in Figure 2.4(a). Table 2.1 The dimensions and quantities of cone and rod cells. Cell type Rod Cone
Diameter (m)
Length (m)
Quantity
2 2.5–7.5
40–60 28–58
100 000 000 6 500 000
14
Introduction to Flat Panel Displays
(a)
Cones/mm2 (×1000)
25 20
Blind Spot
15 10 5 0
(b)
0
2
4
6
8
10 12 14 16 18 20 22 24
0
2
4
6
8
10 12 14 16 18 20 22 24 Eccentricity, mm
160
Rods/mm2 (×1000)
140 120 100 80 60 40 20 0
Nasal direction Figure 2.3 Spatial distributions of (a) cone and (b) rod cells.8
1 0.8
V' (λ) V (λ)
0.6 0.4 0.2 0.0
Absorbance
Normalized Sensitivity
1.0
L-cone
0.1
M-cone
0.1
S-cone 0.001 0.0001
400
500 600 700 Wavelength (nm) (a)
400
500 600 Wavelength (nm)
700
(b) Figure 2.4 Spectral responses of (a) photopic and scotopic regions,1 and (b) L-, M- and S-cones.10
Color science and engineering
15
2.3 Colorimetry 2.3.1 Trichromatic space In human eyes, the recognition of color comes from the combination of the stimuli of different cones. Since there are three kinds of cones, any stimulus can be represented by a vector in three-dimensional vector space, which is called the trichromatic space, as shown in Figure 2.5. Here, R, G and B represent the unit vectors of the stimuli of red, green and blue, respectively. For an arbitrary stimulus Q, it is always possible to find unique tristimulus values R, G and B to obtain the following equation: Q = RR + GG + BB.
(2.2)
For example, as shown in Figure 2.5, G is larger than its unit vector while R and B are smaller. This means that stimulus Q stimulates green more than red and blue. The length of the stimulus vector Q represents the intensity. The greater the length, the higher is the intensity. Also, two or more stimuli can be added linearly to form a new stimulus. For example: Q1 = R1 R + G1 G + B1 B,
(2.3)
Q2 = R2 R + G2 G + B2 B.
(2.4)
Q = Q1 + Q2 = (R1 + R2 )R + (G1 + G2 )G + (B1 + B2 )B.
(2.5)
Then, The vector Q intersects the plane formed by three points (R, G, B) = (1, 0, 0), (0, 1, 0) and (0, 0, 1), which represents pure red, green and blue, respectively, as shown in Figure 2.5. We can also observe that this point is closer to (R, G, B) = (0, 1, 0), which means the color of Q contains more green component than red and blue components. On this plane, we define r=
R , R+G+B
(2.6)
g=
G , R+G+B
(2.7)
b=
B , R+G+B
(2.8)
r + g + b = 1.
(2.9)
G G
Q
(0,1,0)
Q r+g+b=1
R
R B B
(1,0,0)
(0,0,1)
Figure 2.5 Trichromatic space of (R, G, B) primary colors (redrawn from Wyszecki and Stiles1 ).
16
Introduction to Flat Panel Displays
Each point in the triangle represents a distinct color. The line linking the blue and red represents the color mixing between these two colors. The color inside the triangle is the mixing of these three colors.
2.3.2 CIE 1931 colorimetric observations Considering the trichromatic color space, it is possible to match an arbitrary color by mixing the red, green and blue primary colors. Figure 2.6 shows the setup for color-matching experiments. An arbitrary light of the color under investigation illuminates the lower half of the white screen which produces a stimulus to the human eye through the black shadow. A hole in the shadow delimits a small viewing angle for light passing through, which stimulates only a certain region of the photoreceptors of the human eye. Red, green and blue lights illuminate the upper half of the white screen. The intensities of the red, green and blue lights are adjusted to match the color of the lower light. According to the trichromatic color space, it is possible to find a set of R, G and B to fit the color to be matched. Mathematically, a spectrum can be expressed by superposition of monochromatic units. Hence, broadband light can be viewed as a mixing of many monochromatic components, as shown in Equations (2.3)–(2.5). To obtain the coordinates of each color, the first step is to obtain the RGB stimulus of the monochromatic light. In 1931 the CIE used three primary colors with wavelengths of 700, 546.1 and 435.8 nm to match all visible monochromatic lights. The intensity of the matched monochromatic light is kept constant in terms of radiometry units (W sr−1 m−2 ), which is called the equal energy condition. This means R(λ) = r(λ)E(λ),
(2.10)
G(λ) = g(λ)E(λ),
(2.11)
B(λ) = b(λ)E(λ),
(2.12)
where E(λ) represents the optical intensity in terms of watts and is a constant for every wavelength. As shown in Figure 2.7(a), with the normalization of r(λ) + g(λ) + b(λ) by equations (2.6) to (2.9) at different wavelengths, we can obtain the CIE 1931 (R, G, B) chromaticity diagram, shown in Figure 2.7(b). Due to the linear summation of different color stimuli, a spectrum with many different wavelengths can be divided into monochromatic lights and be viewed as the color mixing of each monochromatic
Black shadow
White screen Red light
Green light Blue light Black shadow
Arbitraty light Figure 2.6 Experimental setup for color-matching experiments.
Color science and engineering
17
0.4 Tristimulus Values
515
r(λ)
b(λ)
2.0
505
525
0.3 0.2
1.5
g
1.0
545
535
g(λ)
495
555 565
0.1
485
0.5
575
475
0.0 –0.1
400
500
600
700 –1.5
–1.0
–0.5
0.0
–0.5
Wavelength (nm)
(a) Figure 2.7
0.0
700
380
0.5 r
1.0
(b)
(a) Tristimulus values for different wavelengths and (b) CIE 1931 (R, G, B) chromaticity diagram.1
component. Hence, any color can be represented by the horseshoe-shaped region, formed by the monochromatic light locus and the connection between 380 and 780 nm. Note that in this color system, the r(λ) values are negative when wavelengths are between 435.8 and 546.1 nm, as shown in Figure 2.7(a). Sometimes, in color-matching experiments as shown in Figure 2.6, one cannot obtain the matched color no matter how one adjusts the (R, G, B) intensities. However, it is possible to move one of the primaries, for example the red one, from the opposite to the same side as the color to be matched. This can be expressed as Q1 + RR = GG + BB,
(2.13)
Q = −RR + GG + BB.
(2.14)
So, in Figure 2.7(b), we have to use negative r values when describing some colors, which is not entirely satisfactory. To improve on this, the CIE 1931 (X, Y, Z) system was proposed. Linear transformations from the CIE 1931 (R, G, B) system are carried out using the following equations: x=
0.49000r + 0.31000g + 0.20000b , 0.66697r + 1.13240g + 1.20063b
(2.15)
y=
0.17697r + 0.81240g + 0.01063b , 0.66697r + 1.13240g + 1.20063b
(2.16)
z=
0.00000r + 0.01000g + 0.99000b . 0.66697r + 1.13240g + 1.20063b
(2.17)
We can then obtain the CIE 1931 (X, Y, Z) chromaticity diagram, as shown in Figure 2.8. In this system, the horseshoe shape lies in the first quadrant which means all colors can be described by positive values. One important feature of the CIE 1931 (X, Y, Z) color system is that the Y value is set as the luminance of the stimulus, in terms of lm sr−1 m−2 or cd m−2 (Section 2.3.6): x X = V, y
Y = V,
z Z = V, y
(2.18)
18
Introduction to Flat Panel Displays
0.9
520nm
0.8
540
0.7
yellowish - green
0.6
yellow green
green
0.5 y
560 580 yellow
0.4
orange publish - green
0.3 0.2
480
0.1 0.0 0.0
0.1
620 770
red
publish - red
blue
460
pink
publish-pink
grrenish - blue
600
raddish-orange
white
puple
publish - puple
380
0.2
0.3
0.4 x
0.5
0.6
0.7
0.8
Figure 2.8 CIE 1931 (X, Y, Z) chromaticity diagram.
where V is the luminance of the stimulus. Then, it is straightforward to understand that V (λ) ≡ y(λ).
(2.19)
To obtain the CIE 1931 (X, Y, Z) color coordinates from a spectrum, we can use the following equations based on the linear summation of each wavelength: X = k P(λ)x(λ) d(λ), (2.20) λ
Y =k
P(λ)y(λ) d(λ)
(2.21)
P(λ)z(λ) d(λ),
(2.22)
λ
Z =k λ
x=
X , X +Y +Z
(2.23)
y=
Y , X +Y +Z
(2.24)
where k = 683 lm W−1 , which represents the transformation from radiometry units (W) to photometry units (lm), and P(λ) is the spectral distribution of the stimulus in terms of W sr−1 m−2 .
Example 2.1 Find the luminance and color coordinates of the mixture of the two color stimuli listed below:
Stimulus 1 Stimulus 2
(x, y)
Luminance (cd m−2 )
(0.6, 0.3) (0.2, 0.7)
30 21
Color science and engineering
19
Answer. From Equations (2.23) and (2.24), X/x = Y /y = Z/z = X + Y + Z: X1 /0.6 = 30/0.3 = Z1 /0.1 X2 /0.2 = 21/0.7 = Z2 /0.1 X1 = 60, Z1 = 10 X2 = 6, Z2 = 3 X = X1 + X2 = 66 Y = Y1 + Y2 = 51 Z = Z1 + Z2 = 13 x = X/(X + Y + Z) = 66/130 = 0.51 y = 51/130 = 0.39 Luminance = Y = 51(cd m−2 ).
2.3.3 CIE 1976 uniform color system Although the CIE 1931 (X, Y, Z) color system can describe a color exactly, there is a problem when dealing with color difference and tolerance. Figure 2.9 shows the famous MacAdam ellipses in the CIE 1931 (X, Y, Z) chromaticity diagram.11 Color differences cannot be discerned by the human eye within the ellipses in this figure. Note that the ellipses are magnified 10 times for clarity, so the ellipses are very small. We can see that ellipses in the blue color region are much smaller than those in the green and red ones, which means that a small shift in color coordinates of the CIE 1931 (X, Y, Z) color system in the blue region results in a serious difference perceived by the human eye. In other words, when the difference of the color coordinate, i.e. (x, y), between two color stimuli is kept constant, the differences of perception by the human eye are largest in the blue region. To better illustrate the ‘color difference’ between two stimuli, e.g. a real object and an image from a display, it is necessary to have a uniform color system. 0.8 0.7 0.6 y
0.5 0.4 0.3 0.2 0.1 0.0 0.0
0.1
0.2
0.3
0.4 x
0.5
0.6
0.7
0.8
Figure 2.9 MacAdam ellipses in the CIE 1931 (X, Y, Z) chromaticity diagram.1
20
Introduction to Flat Panel Displays
520 530 540 550 560 570 510nm
0.6
580
590 610
640 680nm
0.5 490nm
v'
0.4 0.3 480nm
0.2 470nm
0.1 0.0 0.0
Figure 2.10
420nm
0.1
0.2
0.3 u'
0.4
0.5
0.6
CIE 1976 (u , v ) chromaticity diagram and MacAdam ellipses.
One such system is the CIE 1976 (L∗ u∗ v∗ ) color system (Figure 2.10). In this color space, we can see the sizes of the ellipses vary less in different regions of the color space than those in the CIE 1931 (X, Y, Z) color system, although the area occupied by an ellipse in terms of color discrimination is exactly the same in both systems. Equations (2.25)–(2.32) govern the coordinate transform from the CIE 1931 (X, Y, Z) to the 1976 (L∗ u∗ v∗ ) color system; since this is a linear transformation, equations for color mixing are still valid in this color space with some modifications: 1/3 Y Y L ∗ = 116 − 16 for > 0.01, (2.25) Yn Yn u∗ = 13L ∗ (u − un ), ∗
∗
n
v = 13L (v − v ), Lm∗ = 903.3
Y Yn
for
Y ≤ 0.008856, Yn
(2.26) (2.27) (2.28)
where u , v and un , vn are calculated from u =
4X , X + 15Y + 3Z
(2.29)
v =
9Y , X + 15Y + 3Z
(2.30)
un =
4Xn , Xn + 15Yn + 3Zn
(2.31)
vn =
9Yn , Xn + 15Yn + 3Zn
(2.32)
where X n , Y n , Z n are tristimulus values of reference light sources, e.g. daylight or blackbody radiation, when discussing the interaction between the light source and reflective objects. Here, we can see that the color coordinates in CIE 1976 (L∗ u∗ v∗ ) color system is related to the light source. For describing a color without a specified light source, X n , Y n , and Z n values are set to zero.
Color science and engineering
21
Compared to the CIE 1931 (X, Y, Z) color system, not only does the color difference become uniform, the brightness difference becomes uniform by nonlinear transformation between the Y and L ∗ values in the CIE 1931 and 1976 color systems, respectively. Luminance (or Y value, in terms of cd m−2 ) is proportional to the optical intensity in terms of radiometric units (W sr−1 m−2 ). However, due to the nonlinear response of the human eye, it is not appropriate to use the Y value to describe the intensity difference between two stimuli of the same color but different intensities. The L value is more uniform in photometric terms than the Y value. As is evident from Equation (2.25), L is proportional to Y to the power of 3, which means that a certain intensity difference distinguished by the human eye requires larger and smaller physical intensity (i.e. watts) changes under brighter and darker stimuli, respectively. In other words, the human eye can distinguish a smaller change when the luminance is low. This means that, when the stimulus is weak, only a small difference in luminance value (Y ) can result in a large difference of perception by the eye (L ∗ ). However, when the stimulus is bright, a larger difference in luminance is required to obtain the same difference of perception by the eye. In a display, we use the term ‘grayscale’ to determine the brightness difference. The differences of perception by the eye between two adjacent gray levels are the same. Figures 2.11(a) and (b) show the grayscale levels versus the luminance using linear and logarithmic scales, respectively. The slope of the plot in Figure 2.11(b) is called the γ value, which is 2.157 in this case, i.e. greater than 1 and less than 3, as shown in Equation (2.25). For example, a display has grayscales from 0 (black) to 255 (white). Grayscales 1 to 254 represent different ‘grays’ from dark to light. The luminance difference between grayscale 0 and 1 is smaller than that between 254 and 255 since the γ value is not equal to 1.
2.3.4 Color saturation and color gamut As shown in Figures 2.8 and 2.10, the boundary of the horseshoe shape of the chromaticity diagram is formed by the monochromatic line and the purple line, the connection between the shortest and longest wavelengths. Any spectrum can be divided into monochromatic components. It is straightforward to understand that the color coordinates of a stimulus with a broader bandwidth lie nearer to the boundary of the chromaticity diagram. The center of the 1931 (X, Y, Z) chromaticity diagram, i.e. (0.33, 0.33), shows the equal-energy spectrum with a white ‘color’. Color saturation is used to describe the ‘colorfulness’ of a stimulus. Color saturation increases towards the boundary of the chromaticity diagram. A monochromatic stimulus exhibits the highest color saturation. When mixing monochromatic light, the color saturation decreases. The white stimulus can be regarded as the ‘achromatic’ color. In a display, three primary colors are used to generate all color images, based on the trichromatic color space theory. By mixing the three primaries, the colors within the triangle shown in Figure 2.12 can 100
80
log (Luminance)
Luminance (cd/m2)
100
60 40 20 0
0
50
100 150 200 250 300 Gray-scale (a)
Figure 2.11
10
50
100 150 200 250 300 log (Gray-scale) (b)
Luminance versus grayscale level using (a) linear and (b) logarithmic scales.
22
Introduction to Flat Panel Displays
0.8 Green
0.7 0.6 y
0.5 0.4
Red
0.3 0.2
Blue
0.1 0.0 0.0
0.1
0.2
0.3
0.4 x
0.5
0.6
0.7
0.8
Figure 2.12 Color gamut of the NTSC.
be obtained. When the triangle is larger, which means the three primaries are closer to monochromatic light, such a display can exhibit more colors. Such a triangle is called the ‘color gamut’ of a display. The triangle shown in Figure 2.12 is based on the definition of the National Television Standard Committee (NTSC). Color coordinates of the red, green and blue stimuli are (0.67, 0.33), (0.21, 0.71) and (0.14, 0.08), respectively.
2.3.5 Light sources As mentioned before, a light source is essential for generating color stimuli. The same object may appear as different colors under different ambient illumination. The issue of light sources is very important for reflective displays since they show different colors under different illumination. One of the purposes of displays is to reproduce color images as they would appear in real life, when they are illuminated by natural or artificial ‘white lights’. Hence, there are some ‘white standards’ in displays in order to obtain the same image stimuli from the display as from the real object. For example, a display typically has a function enabling it to be switched among different white standards depending on the choice of the user. Two common white standards are D65 and D93, which mean that ‘white’ looks like daylight with color temperatures of 6500 and 9300 K (Section 2.3.5.1), respectively. Also, for non-emissive and transmissive displays, a backlight module is needed since such displays acts as light valves, rather than emitters (this is introduced in Section 2.3.5.1).
2.3.5.1 Sunlight and blackbody radiators One important source of natural light is sunlight, which is typical ‘blackbody radiation’, having a spectral power distribution that depends on the temperature of the light source alone. Typically, the spectral density of the radiant power of blackbody radiation can be described as12 c Me = ueλ = c1 λ−5 (ec2 /T λ − 1)−1 4
(W m−3 ),
(2.33)
where c = 2.99792458 × 108 m s−1 (velocity of light), h = 6.626176 × 10−34 J s (Planck constant), k = 1.380662 × 10−23 J K−1 (Boltzmann constant), c1 = (c/4)8 πhc = 2 πhc2 = 3.741832 × 10−16 W m2 and c2 = hc/k = 1.438786 × 10−2 m K.
Color science and engineering
23
Radiant Excitance (W/m3)
1015 10
1013
3200K, Tungsten Halogen Lamp peak@ 910 nm
1012
2856K, Tungsten Lamp peak@ 1010 nm
1011 1010
Figure 2.13
6500K peak@ 440 nm
14
400
600 800 1000 Wavelength (nm)
1200
1400
Power intensity spectrum of a blackbody radiator at different temperatures.
On increasing the temperature of the blackbody radiator, the radiant power increases and its spectral peak shows a blue shift, as shown in Figure 2.13. A ‘blackbody locus’ can be defined on the CIE 1931 (x, y) chromaticity diagram, which shows the decreasing trends of x and y coordinates with increasing temperature. When the chromaticity coordinates of a stimulus lie on the blackbody locus, we can define the ‘color temperature’ of the color stimulus, which is the temperature of the blackbody radiator at these chromaticity coordinates. When the chromatic coordinates of a stimulus do not lie on the blackbody radiator locus, but pass through a line perpendicular to the tangent of the blackbody locus of the CIE 1976 (L∗ u∗ v∗ ) color system at a certain temperature, this temperature is called the ‘correlated color temperature’of the color stimulus.Atungsten lamp is a typical blackbody radiator with a color temperature of 2856 K, which looks yellowish white and is called the ‘illuminant A’. The correlated color temperature of sunlight at the surface of the Earth ranges from 4000 to 25 000 K, depending on the observation time and place, due to the absorption of the atmosphere. The air mass (m) is defined as the ratio of the length within the atmosphere that light from the sun travels to the observation point on the Earth to the thickness of the atmosphere measured perpendicular to the Earth’s surface. For example, for sunlight observed from an altitude of 30◦ at the equator, the corresponding air mass is 2. Since the atmosphere absorbs light with short wavelengths more strongly than that with long wavelengths, the correlated color temperature is lower for a greater air mass. This explains why sunlight is redder during sunrise and sunset, but looks white at noon. There is a widely used white standard for displays, D65, which corresponds to sunlight with a correlated color temperature of 6500 K.
2.3.5.2 Backlights of transmissive displays Unlike continuous blackbody radiators, one of the criteria of backlights for displays is to provide sharps peaks for RGB primaries, with high efficiency to reduce power consumption. Typically, there are several kinds of backlight technologies such as cold cathode fluorescent lamps (CCFLs),13 external electrode fluorescent lamps (EEFLs),14 hot cathode fluorescent lamps (HCFLs),15 flat fluorescent lamps (FFLs),16 light-emitting diodes (LEDs; Chapter 6) and field emission displays (FEDs; Chapter 8), for different kinds of transmissive displays, as summarized in Table 2.2. Nowadays, CCFLs are the main backlight technology used, especially for liquid crystal displays (LCDs). The operation principle of CCFLs is based on the excitation of mercury and a rare gas mixture under low pressure by a high voltage provided at two electrodes inside the ends of a tube, which is called a glow discharge. In contrast to conventional bulbs, CCFLs provide ‘cold’ light without using
24
Introduction to Flat Panel Displays
Table 2.2 Acronyms of backlight technologies. Acronym
Full name
CCFL EEFL HCFL FFL LED FED
Cold cathode fluorescent lamp External electrode fluorescent lamp Hot cathode fluorescent lamp Flat fluorescent lamp Light-emitting diode Field emission display
a hot filament. Typically, CCFLs can provide a stable white backlight with a power efficiency of over 60 lm W−1 . In EEFLs, the end electrodes are located outside, rather than inside, the tube, which avoids the electrons and mobile ions striking directly onto the electrodes resulting in a shortened operation lifetime.Adielectric barrier discharges with wall charges accumulating on the inner glass having external electrodes, which means the electrode, the glass tube thickness and the glass material have to be designed very carefully. With an optimized design, EEFLs with a power efficiency of 58 lm W−1 and lifetimes of 60 000 hours have been demonstrated. However, because of the complex tube design, the stability of EEFLs, due to dielectric breakdown and pinholes, remains a technical challenge. Recently, HCFLs have attracted much attention due to their advantages of higher luminance and efficiency. The operation principle of HCFLs is similar to that of CCFLs, except that hot electrons with a high density are emitted from the cathode, which results in a higher current and lower driving voltage, as compared to CCFLs. This phenomenon is called arc discharge. Due to the strong collisions between the electrons and the gases, the ‘gas temperature’ is high, and so the emission is called ‘hot’ cathode fluorescence. Typically, the power efficiency of HCFLs can be as high as 90–100 lm W−1 . However, due to the high current characteristics, positive ions bombard the cathode continuously which results in a shorter operation lifetime. Because of their high efficiency, high luminance and short lifetime, HCFLs can be used scanning mode LCD backlights, which greatly improves the image quality. In contrast to the one-dimensional tube structure of CCFLs, HCFLs and EEFLs, two-dimensional illumination using FFLs has been proposed to provide a more uniform light source. Gas is filled between glass substrates where the electrodes are formed. This structure is like a plasma display panel without patterning. However, there are still some technical issues that remain to be solved concerning FFLs, e.g. low efficiency, poor uniformity, long warm-up time, large leakage current and pinhole formation. In addition to their use as displays, LEDs and FEDs can also be used as backlights. The operating principles of LEDs and FEDs are discussed in subsequent chapters.
2.3.5.3 Color rendering index For a reflective display, an external light source is needed to produce visible images. Hence, the color and the brightness of a reflective display strongly depend on the optical characteristics (such as spectrum and intensity) of the light source. For example, an apple which looks green in sunlight looks yellow when viewed under yellow streetlights (this yellow light is produced by high-pressure sodium lamps). To quantitatively describe the color difference between a reflective object illuminated by a light source under test and by light from a blackbody source (such as sunlight), the color rendering index (CRI) is used. The maximum value of CRI is 100, which means the color is exactly the same when viewed with illumination from a light source under test and that from a blackbody source. In contrast, if the CRI value is low, this means the color difference of the same reflective object illuminated by the two different light
Color science and engineering
25
sources is large. Here, color difference (E ∗ ) is defined as the absolute distance between two points in a uniform color system (UCS), such as the 1976 (L∗ , u∗ , v∗ ) color system: E = [(L ∗ )2 + (u∗ )2 + (v∗ )2 ]0.5 ,
(2.34)
where L ∗ , u∗ and v∗ represent the coordinate differences between the two colors. By choosing eight different objects which appear (1) light grayish red, (2) dark grayish yellow, (3) strong yellow-green, (4) moderate yellowish green, (5) light bluish green, (6) light blue, (7) light violet and (8) light reddish purple, under daylight, the CRI can be defined as CRI = 100 −
8
E1∗ ,
(2.35)
i=1
where i are the different objects. To achieve a high CRI, the spectrum of the light source must be broad enough to reflect light from different wavelengths. Light sources with a continuous spectrum, such as an incandescent bulb, exhibit a high CRI value. In contrast, for a fluorescent lamp with discrete emission wavelengths, the CRI value is low (∼50). For monochromatic light, the CRI value can be negative.
2.3.6 Photometry Due to the spectral sensitivity of the human eye, we perceive radiation with the same input power (in terms of watts) to be brighter or dimmer according to its wavelength. The photometric unit, the lumen (lm), is defined as the luminous flux (F) from a monochromatic light at 555 nm emitting an optical power of 1/683 W. As discussed in Section 2.2, the spectral sensitivity of the human eye can be represented as V (λ) for the photopic region and is most sensitive at 555 nm. For example, V (λ) is 0.1 at 650 nm, which means the sensitivity is 10 times less than that at 555 nm. So, a power of 1/68.3 W is needed for monochromatic light at 650 nm to reach 1 lm. Actually, the principle photometric unit is not the lumen, but the candela (cd), which is defined as lumens per unit solid angle (lm sr−1 ), and is called luminous intensity (I). Solid angle (sr) is defined as: sr = A/r2
(2.36)
where A is the area on the sphere centered at the light source with radius r. The initial definition of 1 cd is the luminous intensity of a standardized plumber’s candle. As shown in Figure 2.14, the candle emits light in all directions, where we use lumens to describe the radiant flux. It is the total optical power emitted by the candle in photometric unit. When a human eye looks at the candle directly, only part of the light emitted by the candle was received. In other words, only the light ray passing though area A was detected by the human eye. For an isotropic light source such as the candle in this example, the total surface of the sphere is 4π r2 , and the total solid angle is 4π. Hence, one can obtain: Luminous intensity = luminous flux/4π
(2.37)
However, one has to note that this is not the case for the directional light sources. A candle can be used as the light source to illuminate an object. We can then define the ‘illuminance’ (E) of the light source in terms of lux, the flux emitted per unit area of the source (lm m−2 ). After light–object interaction, the light is modulated (reflected, transmitted, scattered or absorbed) by the object and can be regarded as being re-emitted from the object, which is defined as the luminous exitance (M), also with units of lux. When the object illuminated by the light source is viewed, the human eye receives light from only a certain of angles, so the luminance (L) of the object can be defined in units of lm sr−1 m−2 , or cd m−2 . The definitions of photometric units are summarized in Table 2.3.
26
Introduction to Flat Panel Displays
Luminous flux (lm) Luminous intensity (lm/sr) Eye Luminance (lm/sr*m2)
Light source Illuminance (lm/m2) Object
Figure 2.14 Schematic summarizing photometric units.
Table 2.3 Definitions of photometric units. Photometric term
Symbol
Unit
Definition
Luminous flux Luminous intensity Illuminance Luminous exitance Luminance
F I M E L
lm cd lux lux cd m−2 (or nit)
lm lm sr−1 lm m−2 lm m−2 lm sr−1 m−2
Example 2.2 A perfect diffuse surface requires that its luminance remains constant at all angles of observation. It is also called a Lambertian surface. For example, rough paper closely resembles a Lambertian surface. For a Lambertian surface (of size A) illuminated by a light source with illuminance E, find its expression in terms of luminance (L). Assume this surface can perfectly reflect all the light, i.e. the luminous flux of the incident beam on the surface is equal to that of the light leaving the surface. Answer. From Table 2.3, luminance (L, in units of lm sr−1 m−2 ) can also be regarded as the luminous intensity (I, in units of lm sr−1 ) per unit area (A): L=
dI . dA
(2.38)
When viewing from a larger angle, the area looks smaller as compared to when viewing along the surface normal direction, and a cos θ term is introduced to account for this. θ is the angle between the viewing direction and the surface normal. That is, dA = dA0 cos θ ,
(2.39)
where A0 is the area as viewed from the surface normal direction. Because the luminance of a Lambertian surface is the same for any viewing direction, one can obtain the luminous intensity as I = I0 cos θ ,
(2.40)
Color science and engineering
27
where I 0 is the luminous intensity from the normal direction of the surface. The incident flux on the Lambertian surface can be represented as Fin = EA.
(2.41)
The total luminous flux radiating from the surface is Fout = I dω = I(θ) dω = I(θ)dϕ sin θdθ
2π
π/2
= I0
π/2
cos θ sin θ dθ dϕ = 2πI0 0
0
cos θ sin θ dθ = πI0 = πLA.
(2.42)
0
Since the luminous flux of the incident beam on the surface is equal to that of the light leaving the surface (Fin = Fout ), one can obtain E = πL. (2.43) Typically, power efficiency (in terms of lm W−1 ) is used to describe the efficiency of a display system. For example, if the total input electrical power (wall-plug power) of a display is 10 W and the total radiated flux is 20 lm, the power efficiency of the display is 2 lm W−1 . The power efficiency describes how much optical power (in lm) is received by the human eye produced by a given electrical input power (in W). For electroluminescent (EL) devices such as LEDs, current efficiency is also defined in terms of cd A−1 . The input current creates a number of electron–hole pairs in the display. Electron–hole pairs recombine and generate photons, which are detected by the human eye (cd). For example, if the display described above is a Lambertian EL display with a current of 300 mA, the current efficiency of the display is 21.22 cd A−1 .
2.4 Production and reproduction of colors A display is used to produce and reproduce color images. Typically, three primary colors, red, green and blue, in a trichromatic color space, are needed for generating all the colors recognized by the human eye.
Display
Object
0.4
Intensity (a.u.)
Intensity (a.u.)
1.0
0.2
0.0
400
500 600 700 Wavelength (nm) (a)
Figure 2.15
0.8 0.6 0.4 0.2 0.0 400
500 600 Wavelength (nm) (b)
Spectra of a basketball from (a) the real object and (b) a display.
700
28
Introduction to Flat Panel Displays
In other words, three primary colors are used to generate the same stimuli of the L-, M- and S-cones as a real object, although the spectra of the real object and the display are quite different. Figure 2.15(a) shows the measured spectrum of a basketball illuminated with sunlight, which shows a continuous curve. We can also take a picture of the basketball, display it using an LCD monitor and measure the spectrum again. As shown in Figure 2.15(b), the color of the basketball from the display consists of red, green and blue emissions, which is quite different from the spectrum of the real object. However, both the real object and display should appear to be the same. Using different spectra to obtain the same appearance is called ‘metamerism’.
Homework problems 2.1 What are the rod and cone cells in a human eye? What are the differences between them? 2.2 Is color printing a kind of additive or subtractive mixing? Why? 2.3 The CIE 1931 coordinates of a white light source of illuminance E are (0.33, 0.33). Find the R:G:B luminance ratio (in photometric units) for such a light source. (Hint: the wavelengths for the R, G and B colors are 700, 546.1 and 435.8 nm.)
λ x y z V (λ)
R
G
B
700 nm 0.735 0.265 0 0.0041
546.1 nm 0.273 0.718 0.01 0.9841
435.8 nm 0.166 0.008 0.826 0.018
2.4 Find the R:G:B radiance ratio to obtain illuminance E.
References 1. Wyszecki, G. and Stiles, W.S. (2000) Color Science: Concepts and Methods, Quantitative Data and Formulae, 2nd edn, John Wiley & Sons, Inc., New York. 2. CIE (1995) Colorimetry, 2nd edn, CIE No. 15:2004, Commission Internationale de l’Éclairage, Vienna. 3. Dowling, J.E. (1987) The Retina: An Approachable Part of the Brain, Belknap Press of Harvard University Press. 4. Smith, G. and Atchison, D.A. (1997) The Eye and Visual Optical Instruments, Cambridge University Press. 5. Hecht, E. (2002) Optics, 4th edn, Baker & Taylor Books. 6. Pocock, G. and Richards, C.D. (2006) Human Physiology: The Basis of Medicine, 3rd edn, Oxford University Press. 7. Wandell, B.A. (1995) Foundation of Vision, Sinauer Associates. 8. Curcio, C., Sloan, K.R., Kalina, R.E. and Hendrickson, A.E. (1990) Human photoreceptor topography. J. Comp. Neurol., 292, 497. 9. Osterberg, G.A. (1935) Topography of the layer of rods and cones in the human retina. Acta Ophthalmol., 13(Suppl. 6), 1. 10. Stockman, A., MacLeod, D.I.A. and Johnson, N.E. (1993) Spectral sensitivities of the human cones. J. Opt. Soc. Am. A, 10, 2491. 11. MacAdam, D.L. (1942) Visual sensitivities to color differences in daylight. J. Opt. Soc. Am., 32, 247. 12. Beiser, A. (2003) Concepts of Modern Physics, 6th edn, McGraw-Hill. 13. Nishihara, T. and Takeda, Y. (2000) Improvement of lumen maintenance in cold cathode fluorescent lamp. International Display Workshops, p. 379.
Color science and engineering
29
14. Shiga, T., Hashimoto, K., Baba, Y. et al. (2000) A 13.56 MHz-drive electric-field coupled discharge lamp for LCD backlighting. 6th Asian Symposium on Information Display and Exhibition, Xian, China, 2000, p. 165. 15. Gielen, H.J.G. (2005) Life aspects of HCFL light sources in LCD backlighting application. SID Tech. Dig., 36, 1001. 16. Hinotani, K., Kishimoto, S. and Terada, K. (1988) Flat fluorescent lamp for LCD backlight. Conference Record of the 1988 International Display Research Conference, San Diego, CA, 1988.
3 Thin-film transistors 3.1 Introduction Thin-film transistors (TFTs) are widely used as electronic switches to turn the pixels of active matrix liquid crystal displays (LCDs) and organic light-emitting devices (OLEDs) on and off.1, 2 Typically, the siliconbased materials of TFTs are mainly in the amorphous or polycrystalline rather than the crystalline silicon phase. The reasons for using disordered silicon are: first, the diagonal size of a crystalline silicon wafer is limited to 12 inches, which makes fabrication of large-size displays using silicon wafers impossible; and second, the silicon substrate absorbs visible light, making it unsuitable as a substrate for transmissive LCDs. Also, it is technically difficult to epitaxially grow single-crystal silicon on glass substrates, since the melting point of glass is below 600 ◦ C, far below that of silicon (1200 ◦ C). In contrast, amorphous silicon, grown by low-temperature plasma-enhanced chemical vapor deposition (PECVD), can meet the basic requirements for driving LCDs, and it can provide a uniform thin film over a large area (ex: 2160 mm × 2460 mm). Polycrystalline silicon (poly-Si), made by excimer laser melting and recrystallization of an amorphous silicon thin film, has an increased grain size and hence a higher carrier mobility with the advantages of: (1) an increased aperture ratio due to the higher carrier mobility and (2) a suitability for use in system-on-panel fabrications due to higher performance transistors.3 But nonuniformity created during the recrystallization process and high leakage current resulting from the higher surface roughness are two major technological issues for poly-Si. Therefore, at present, most TFTs for LCD applications are based on amorphous silicon and related fabrication processes. In this chapter, we first introduce some basic semiconductor concepts, which are essential background knowledge not only for TFTs but also for light-emitting diodes (LEDs; Chapter 6). Then, we discuss the material aspects of amorphous and poly-TFTs, followed by their electrical characteristics and the methods to drive a display. Finally, TFTs made from other materials such as organic materials and metal oxides are discussed.
3.2 Basic concepts of crystallized semiconductor materials Before describing the characteristics of amorphous semiconductors with disordered structure, which are typically used in conventional TFTs, we introduce some basic properties of semiconductors with a singlecrystal structure. A semiconductor is a solid-state material with a conductivity between that of an insulator (e.g. glass and quartz) and a conductor (e.g. silver, aluminum and gold).4 Typical semiconductor conductivities range from 10−8 to 103 S cm−1 . In an insulator, the conductivity is too low for carrier transport.
Introduction to Flat Panel Displays c 2008 John Wiley & Sons, Ltd
J.-H. Lee, D.N. Liu and S.-T. Wu
32
Introduction to Flat Panel Displays
In a conductor, by contrast, carriers (electrons and holes) can propagate with little obstruction. The conductivity of a semiconductor can be tuned by means of electric fields and impurity concentrations, which makes it possible for them to be used in various electronic applications. For example, the conductivity of a TFT is modulated by an electric field and hence it can be used as an electronic switch. When doped with different atoms, a p- or n-type semiconductor can conduct by means of holes or electrons, respectively. When p- and n-type semiconductors are connected to each other, a p–n junction is formed that allows unidirectional carrier transport. This is the basic structure of the LED, as discussed in Chapter 6. A semiconductor consisting of only one kind of atom, such as silicon and germanium which are located in group IV of the periodic table, is called an ‘elementary semiconductor’. In this chapter, we mainly focus on the widespread use of silicon in TFT switches due to the advantages of high performance, mature fabrication techniques and comparatively low cost. For LED applications, however, due to its indirect bandgap characteristics (Section 3.2.1), bulk single-crystal silicon cannot emit light efficiently. Hence, a compound semiconductor that combines two or more elements (e.g. gallium from group III and arsenic from group V, forming GaAs) is employed (as discussed in Chapter 6).
3.2.1 Band structure of crystallized semiconductors A semiconductor with a single-crystal structure has long-range order, which means atoms in a crystallized semiconductor arranged with regular and repeated three-dimensional pattern. A group IV atom has four valence electrons (which will be described later) in its outer shell. The optimal number of covalent bonds it can form is 8 − 4 = 4, which arises from the well-known ‘8 – N rule’. Hence, each atom in a silicon crystal is connected to four neighboring atoms equally distant from it but in different directions by covalent bonds, i.e. each atom shares four electrons with four neighboring atoms. Since the atoms are identical in an elementary semiconductor, it is straightforward to understand that the four connected neighboring atoms form an equilateral tetrahedron. Figure 3.1(a) shows the three-dimensional structure of single-crystal silicon. Each sphere represents a silicon atom that is connected to four other atoms by covalent bonds. For a compound semiconductor such as GaAs, the bonding configuration is similar to that of silicon, as shown in Figure 3.1(b). Larger black and smaller white spheres represent gallium and arsenic atoms, respectively. However, since different atoms attract valence electrons to different degrees, the bonding contains some ionic character, and is not purely covalent. The periodic structure of a crystallized semiconductor is usually called a lattice structure. Also, as shown in Figures 3.1, one can define a unit cell of the lattice, which is the smallest repeat unit required to generate the whole lattice. Also, the lattice constant represents the edge length of the unit cell, which is critical for epitaxial growth in LEDs, as discussed in Chapter 6. Considering a single isolated atom, the nucleus at the center attracts electrons, which creates a Coulomb potential that is proportional to the inverse of the distance to the nucleus. There are many discrete energy states inside the potential well, as shown in Figure 3.2(a). To determine the electronic configuration in a semiconductor, the Pauli exclusion principle is applied, which states that each electron must exhibit an identical quantum state. For an energy state in Figure 3.2(a), two electrons can fill one energy state since they exhibit different spin angular momentum vectors (orientation specified by spin-up and spindown). Lower energy states are filled first with the inner shell electrons. Outermost electrons, partially or completely filling the high energy state, are called valence electrons, as mentioned above. When two atoms approach closely and form a bond, each energy level splits into two discrete states, as shown in Figure 3.2(b). Electrons from the two atoms are reallocated again from low to high energy states in this new system. Again, in a crystal, there are lots of atoms, typically with a concentration of 1023 cm−3 . Hence, each energy level in an isolated atom splits into very many closely packed states, which is called an energy band, as shown in Figure 3.2(c).5, 6 Also in this system, inner electrons fill the bands sequentially from the lowest to the highest energy state. Finally, valence electrons completely fill an energy band,
Thin-film transistors
33
a
(a)
Ga As
a
(b) Figure 3.1 Three-dimensional crystalline structure of (a) silicon and (b) GaAs.4
namely the valence band that represents the highest energy state occupied by electrons. This happens at 0 K without thermal effects. Sometimes a valence electron is also called a bonding electron, since it is shared by neighboring atoms. Because the valence electrons are the outmost electrons of an atom, they experience a much smaller influence from the nuclei due to the shielding effect of the inner shell electrons. When considering the temperature effect (T> 0), these valence electrons can obtain thermal energy and escape from the covalent bond, and can then be treated as ‘free electrons’ with the introduction of effect mass concept, which will be described later in this chapter. In the energy band diagram described above, this means the valence electron jumps to the next higher energy state, which is called the conduction band, as shown in Figure 3.3. The energies of the top of the valence band and the bottom of the conduction band are denoted E v and E c , respectively. The energy difference between the conduction and valence bands is called the bandgap (E g ); no energy level is allowed between these two bands. The higher the temperature the more electrons are promoted to the conduction band. At the same time, this leaves ‘holes’ with positive charge in the
34
Introduction to Flat Panel Displays
Discrete energy state (a)
(b)
Energy bands (c) Figure 3.2 Energy levels in (a) an isolated atom, (b) two atoms and (c) a crystal structure. Solid and dashed lines depict the occupied and unoccupied energy states (bands).
– Conduction band (Ec) bandgap (Eg) + – – –
Valence band (Ev) + : hole – : electron
Figure 3.3 Energy band structure.
lattice. An electron from another covalent bond neutralizing a hole creates another hole, and so on. Hence it appears that holes are transported in a crystal, in a direction opposite to that of the electron movement. At 0 K, all the electrons are in the valence band. As the temperature increases, some electrons obtain enough thermal energy to move up to the conduction band. In a semiconductor, the statistical distribution of electrons at different energies is typically described by the concept of Fermi energy (or Fermi level). Under thermal equilibrium at a certain temperature, the probability (F) of finding an electron at energy level E is expressed as 1 F(E) = , (3.1) 1 + exp[(E − EF )/(kT /q)] where E F is the Fermi level (in eV) and k and T are the Boltzmann constant (1.38 × 10−23 J K−1 ) and absolute temperature (in K), respectively. Equation (3.1) is derived from quantum mechanics and is called the Fermi–Dirac distribution. This can be applied to the statistical distribution of any particle obeying the Pauli exclusion principle. From the above equation, it is evident that when E = E F , F(E) = 0.5, which means this energy level is occupied by an electron with 50% probability. Figure 3.4 shows plots of F(E) curves versus energy at different temperatures. When T is close to 0 K, F(E) is a step function, i.e. F(E) is 1 and 0 when E is lower and higher than E F , respectively. On increasing the temperature, the F(E) curves tend to decrease and increase below and above E F , which means the probability of an electron occupying a higher energy increases. In other words, the electron concentration in the conduction band increases with increasing temperature.
Thin-film transistors
35
1.0 0K 0.8
150K
F(E)
300K 0.6
450K
0.4 0.2 0.0 –0.10
–0.05
0.00 E-EF(eV)
0.05
0.10
Figure 3.4 F(E) curves at different temperatures.
Because of the wave properties of particles, an object exhibits a de Broglie wavelength (λ) given by h p
(3.2)
2π , k
(3.3)
λ= or λ=
where h = 6.626 × 10−34 J s is the Planck constant and p and k are the momentum and wavenumber of the ‘particle wave’, respectively. The kinetic energy (E) of this object can be represented as p2 2m
(3.4)
(hk/2π)2 2m
(3.5)
E= or E= or
d 2 E (h/2π)2 , (3.6) = dk2 m where m is the mass of the object. According to Equation (3.2), the de Broglie wavelength is inversely proportional to momentum. For example, a baseball with a mass of 0.145 kg has a de Broglie wavelength of 1.03 × 10−34 m when moving at 160 km h−1 . This wavelength is too small to be measured. Hence, it is difficult to observe the wave characteristics of objects in daily life since they are too heavy. By contrast, the mass of an electron in free space (m0 ) is very small, i.e. 9.1 × 10−31 kg. When an electron moves in a semiconductor, due to the interaction with the atoms, its effective mass (discussed later) is even smaller, i.e. 0.19m0 in silicon for a certain transport direction. For a typical velocity of 107 m s−1 , its de Broglie wavelength is about 3.84 × 10−10 m (or 3.84 Å). This value is comparable to the lattice constant of crystalline structures (5.43 and 5.65 Å for silicon and GaAs, respectively), which means the electron wave will experience diffraction by the lattice. Hence, the E–k relationship is no longer a parabolic form as shown in Equation (3.4). The detailed E–k band structure can be calculated from quantum mechanical theory. Figure 3.5 shows the E–k band structures of silicon and GaAs, where E = 0 corresponds to the maximum of the valance band. One can see that there are bandgaps between the maximum of the valence
36
Introduction to Flat Panel Displays
Γ2′
L3 L1
ENERGY (eV)
2 0
Γ15 Si
Γ25′
X1 X4
–6
–10
4
Γ25′
2
bandgap
–4
–8
Γ15
L3′
–2
X1
L1
0
GaAs
Λ
Γ
Δ
X U,K
Σ
Γ7
X7
Γ6 Γ8
X6
Γ8 Γ7
L6
Γ7
X7 X6
–4
bandgap
–6 L6
–10
Γ1
Γ1
–12
Γ7
Γ6
L6 L4,5
–2
Γ8
Γ8
L6
–8
L2′
L
L4,5
6
Γ2′
4
ENERGY (eV)
6
X6
L6
X6 Γ6
–12
Γ
L
→
Λ
Γ6
Γ
Δ
X U,K
Σ
Γ
→
WAVE VECTOR k (a)
WAVE VECTOR k (b)
Figure 3.5 Calculated E–k band structures of (a) silicon and (b) GaAs.7 The solid and dashed lines in the silicon band structure show results obtained from two different kinds of calculations. Source: J.R. Chelikowsky, and M.L. Cohen, Nonlocal pseudopotential calculations for the electronic structure of eleven diamond and zinc-blende semi-conductors, Phys. Rev. B, 14, 556 (1976). Reprinted with permission from the American Physical Society.
band and the minimum of the conduction band. Excited electrons accumulate near the minimum of the conduction band since this is energetically favorable. The corresponding holes are located near the maximum of the valence band for the same reason. The vertical axis shows different k vectors which can be regarded as viewing the lattice from different angles. Since the atoms are arranged in the ordering shown in Figure 3.1, electrons moving in different directions experience different periods and diffraction effects from the nuclei. For silicon, the two extremes have different k vectors, where the top of the valence band and the bottom of the conduction band are at the and X points on the k-axis, respectively; the material is called an indirect bandgap semiconductor. In contrast, for GaAs, the two extremes coincide at the point and it is called a direct bandgap semiconductor. The difference in band structure between direct and indirect bandgap semiconductors results in different optical characteristics, discussed further in Chapter 6. We note that near the minimum of the conduction band and the maximum of the valence band, the E– k curve has a nearly parabolic shape, where most carriers accumulate. Hence, one can slightly modify Equations (3.5) and (3.6) to describe the motion of an electron in the conduction band of a semiconductor: E=
(hk/2π)2 2m∗
(3.7)
and d 2 E (h/2π)2 = , dk2 m∗
(3.8)
where m∗ is called the effective mass, which can be obtained provided that the E–k relationship is known.
3.2.2 Intrinsic and extrinsic semiconductors A semiconductor without other impurity atoms is called an intrinsic semiconductor. In such a material, electrons and holes in the conduction and valence bands (free electrons and holes) are generated simultaneously, typically from thermal energy. Hence, the concentrations of the free electrons and holes can be described as n = p = ni , (3.9)
Thin-film transistors
37
Si
Si
Si
Si
Si
Si
+
– Si
As
Si
Si
B
Si
Si
Si
Si
Si
Si
Si
(a)
(b)
Figure 3.6 Schematic of bonding of silicon with (a) arsenic and (b) boron impurities.
where n and p are electron and hole concentrations, respectively, and ni is the intrinsic carrier density, which is dependent on the temperature and bandgap energy. Obviously, for a higher temperature or a smaller bandgap, it is easier to generate more electron–hole pairs, corresponding to a larger ni value. One might intentionally add some impurities into a semiconductor to vary the conductivity. This is called an extrinsic semiconductor. As shown in Figure 3.6, incorporating group V atoms such as arsenic and group III atoms such as boron into silicon generates one more electron and one more hole, respectively, which effectively increases the carrier concentrations. This is called n- and p-type doping since the conducting carriers are electrons and holes, respectively. Here, group V and III dopants are typically called donors and acceptors, since they respectively ‘donate to’ and ‘accept from’ the silicon atom an electron. Typically, in a crystalline semiconductor, these extra electrons or holes can be treated as nearly free, which is called the complete ionization condition. This is not the case for some III–V semiconductors, especially for the wide bandgap materials described in Chapter 6. For an n-type semiconductor, since there are free electrons released from the donors, the electron concentration is higher than that of holes. This means it is easier to find a free electron in the conduction band than in the intrinsic semiconductor case, and the E F level shifts toward the conduction band level E c . In contrast, for a p-type semiconductor, E F moves down toward the valence band level E v . For an intrinsic semiconductor, E F is typically very close to the center between E c and E v . Hence, one can determine whether a semiconductor is of n- or p-type by observing whether E F lies at the upper or lower half of the bandgap, respectively. Theoretically, electron concentration (n) can be obtained from n=
N(E)F(E) dE,
(3.10)
where N(E) is the density of states, which indicates how many carriers this energy state allows. In the bulk of a crystalline material, N(E) is proportional to E 0.5 , and the value is zero for E < E c . This means there is no allowable state within the bandgap. Increasing the energy above E c leads to more and more allowable states. In contrast, the probability of finding an electron decreases exponentially as the energy increases, as shown by Equation (3.1). Although Equation (3.10) has no analytical solution, we can still
38
Introduction to Flat Panel Displays
obtain the following equations under the approximation of a low carrier concentration condition, i.e. <1018 cm−3 : Ei − EF p = ni exp , (3.11) kT /q EF − Ei n = ni exp , (3.12) kT /q pn = ni2 ,
(3.13)
where E i is the center of the bandgap (in eV). Note that these three equations are valid for both intrinsic and extrinsic conditions. When the Fermi level is shifted from E i toward the valence band, the hole concentration increases; when E F moves up from E i toward E c , the electron concentration increases. The product of Equations (3.11) and (3.12) yields Equation (3.13), which shows that the product of electron and hole concentrations is a constant that is the same for intrinsic and extrinsic semiconductors. This means that, for example, it is possible to dope donors to increase electron concentrations. However, at the same time, hole concentrations decrease, and vice versa. For an extrinsic semiconductor, under complete ionization condition, one can also write p = NA , (3.14) n = ND ,
(3.15)
where N A and N D are acceptor and donor concentrations, respectively.
Example 3.1 Find the carrier concentrations in silicon doped with boron at a concentration of 1016 atoms cm−3 at 300 K. Also find the energy difference (in terms of eV) between the Fermi level and the top of the valence band (ni = 1.45 ×1010 cm−3 and E g = 1.12 eV at this temperature). Answer. Assuming complete ionization, from Equations (3.4) and (3.13), p = NA = 1016 cm−3 , n= From Equation (3.11), Ei − EF =
ni2 (1.45 × 1010 )2 = = 2.1 × 104 cm−3 . p 1016
p 1.38 × 10−23 × 300 kT 1016 ln = = 0.35 eV ln q ni 1.6 × 10−19 1.45 × 1010 EF − Ev =
1.12 − 0.35 = 0.21 eV. 2
3.3 Disordered semiconductors Amorphous silicon (a-Si) consists of silicon atoms without a crystallized structure (no long-range order). However, a-Si retains short-range order, which means that a silicon atom forms covalent bonds with four neighboring atoms but with variations in bonding length and angle.8 Due to the nonperiodic structure, it is not possible to draw E–k band diagrams. Instead, the concept of density of states is used to describe the
Thin-film transistors
39
electrical and optical characteristics of a-Si. Since atoms in the a-Si network exhibit statistical differences in bonding conditions, there are some allowed states extending from the conduction and valence bands, which are called tail states. Due to the imperfect crystalline structure, there are also lots of dangling bonds, which can be viewed as defects, from the viewpoint of crystalline silicon. Those defects can be described as the deep state with energy inside the bandgap. Carriers with lower energy will be trapped or localized by these states which become immobile. When the number of injected carriers is large enough, high energy carriers can hop within the random network, giving a finite mobility value much lower than that in crystalline silicon. Typically, we can classify silicon in terms of crystallization (grain size), as shown in Figure 3.7.9 In a-Si material, which is viewed as the most disordered system, the mobility is as low as 1 and 0.01 cm2 V−1 s−1 for electrons and holes, respectively. When silicon atoms form a crystalline structure of the order of several to several tens of nanometers, the silicon is called nanocrystalline silicon (nc-Si), with higher mobility values. On increasing the grain size to the micrometer range, the classifications are called microcrystalline silicon (c-Si) and polycrystalline silicon (poly-Si). The increase of grain size from a-Si to nc-Si and even c-Si can be achieved by optimization of fabrication parameters during thin-film deposition. However, to obtain poly-Si, melting and recrystallization processes are typically needed, which require a temperature (1200 ◦ C) that is much higher than the softening temperature of glass substrates, i.e. 600 ◦ C. But local heating by laser irradiation makes it possible to form high-quality polySi thin film on glass or even plastic substrates. In this section, we describe the material characteristics, preparation and carrier transport of a-Si and poly-Si, which are the two most commonly used materials in display technologies.
3.3.1 Amorphous silicon Figure 3.8 shows schematically the structure and density of states of a-Si. One can see that a-Si has a random structure without long-range order.10 Hence, it is not possible to obtain the E–k band diagram as in Figure 3.5. However, one can describe the carrier concentrations at different energies with the concept of density of states, i.e. how many states are allowed at a certain energy level, and then determine the electrical bandgap and optical bandgap of a-Si. The larger and smaller spheres in Figure 3.8(a) represent silicon and hydrogen atoms, respectively. Hydrogen atoms are introduced during thin-film formation by
Figure 3.7 Classification of silicon by grain size, together with the mobility values and process conditions.9 (Reprinted from Wagner, S., Gleskova, H., Cheng, I.C. and Wu, M. (2003) Silicon for thin-film transistors. Thin Solid Films 430, 15. Copyright (2003), with permission from Elsevier)
40
Introduction to Flat Panel Displays
Conduction band Extended states Ec EA EF
Localized states Band tail Mobility Eg gap
Optical opt gap Eg Deep states
Energy
EB Ev
Band tail
Valence band Density of states (a)
(b)
Figure 3.8 (a) Three-dimensional structure of a-Si:H.8 (Street, R.A. (ed.) (1999) Technology and Applications of Amorphous Silicon. With kind permission of Springer Science and Business Media.) (b) Density of states of a-Si:H.10 (Source: Tsukada, T. (2003) TFT/LCD: Liquid-Crystal Displays Addressed by Thin-Film Transistors. Permission of Taylor & Francis.)
PECVD, which will be described later. Hydrogen atoms effectively reduce the number of dangling bonds in a-Si with Si–H bonding. Typically, a-Si is written as a-Si:H, which means the a-Si is hydrogenated. Although hydrogen atoms can passivate a large number of dangling bonds in the a-Si:H network, there are still lots of bonding defects, which are the origins of the deep states in Figure 3.8(b) within the bandgap. As shown in Figure 3.8(a), a-Si can be viewed as a random network connected by covalent bonds without long-range order. Although each silicon atom is still fourfold coordinated, the bonding length and angle are different, which results in the extension of the density of states inside the forbidden gap. These are the tail states in Figure 3.8(b). In crystalline silicon, electrons are transported within the ordered network; this is similar to the free electron case and can be described by the effective mass concept. However, in a-Si material, electrons are confined within a grain, which is called the localized state, and can hop to other grains when they obtain enough kinetic energy, which is called the extended state. When electrical current is injected into a-Si, the excess carriers fill the deep levels first, and then the localized states. At this time, those carriers contribute little to electrical conduction. With a further increase in the injected current, carriers fill the extended states, which leads to a much lower mobility value compared to that of crystallized silicon. Due to the carrier trapping and escaping process, the operating speed of a-Si-based devices is limited. The low mobility value results in high resistivity according to ρ=
1 , nqμ
(3.16)
where ρ is the resistivity (in cm), q is the carrier charge and μ is the carrier mobility. Typically, for undoped or n-type a-Si, conduction from holes can be ignored since the hole mobility is much lower than the electron mobility. One can define the mobility gap (electrical bandgap) from the difference in
Thin-film transistors
41
the localized state energies, which is about 1.9 eV. This bandgap is larger than the bandgap of crystalline silicon (1.1 eV). The high ‘bandgap’ value can be understood from the fact that a-Si exhibits nonzero mobility when the localized states are filled up, rather than the onset of the tail states. As mentioned above, it is not possible to obtain the E–k band diagram for a-Si due to its random structure. Hence, the selection rule for direct and indirect transitions is relaxed in this material. It also results in a-Si having a higher absorption coefficient for photons, compared to crystalline silicon with indirect bandgap characteristics. Consequently, one of the important applications for a-Si is in photovoltaic devices. Due to the complex structure of a-Si, there is no clear cutoff for the absorption edge. Typically, one can define an optical bandgap with a value of about 1.7 eV, which represents the onset energy of optical absorption. As regards a-Si fabrication, PECVD is a common technique for producing a-Si:H thin film, at low temperature, of high quality and large area.9 Figure 3.9 shows schematically a PECVD system. Here, SiH4 is used as the source for silicon deposition. Hydrogen gas is added to further passivate the dangling bonds of the silicon. Plasma generated by radiofrequency (rf) power decomposes the SiH4 into Si and H radical ions, which then deposit on the substrate. The standard substrate temperature is about 250 ◦ C which allows the radical atoms to migrate and form chemical bonds. To obtain high-quality a-Si:H with high throughput at a certain temperature limited by a glass (<600 ◦ C) or flexible (<200 ◦ C) substrate, fabrication parameters such as flow rate, rf power, chamber pressure, substrate temperature, SiH4 /H2 ratio and gas source (NH3 , SiCl2 H2 , Si2 H6 , etc.) should be taken into consideration and optimized. Also, under some process conditions, such as smaller SiH4 /H2 ratio, higher plasma power density and thicker thin film, one tends to obtain nc-Si or even c-Si, rather than a-Si.
3.3.2 Polycrystalline silicon There are many advantages in the use of poly-Si rather than a-Si, since it exhibits a larger grain size and hence has better electrical properties, as shown in Figure 3.7. As mentioned above, a TFT is opaque and sensitive to visible light, so shielding is needed to block illumination of the device, which means the TFT area does not contribute to the information of the display. Better TFT electrical properties can give rise to the possibility of reducing the TFT dimensions, which can further: (1) enhance the display area for each pixel (called the aperture ratio), giving a higher luminance output, and (2) increase the display resolution. Also, due to the increase in the hole mobility, it is possible to fabricate not only n-type but also p-type TFTs. The performance (especially mobility and speed) of p- and n-type poly-Si TFTs is good enough for them also to be used in the peripheral circuits on glass substrates (rather than only for the switch element in pixels of a-Si TFT-based displays), which could greatly reduce the cost of the driving integrated circuits (ICs) and the complexity of manufacture.
rf power generator
Electrode Plasma SiH4 + H2 + N2
Substrate Electrode
Figure 3.9 Schematic of a PECVD system.
42
Introduction to Flat Panel Displays
A feasible way to obtain high-quality poly-Si is to melt a-Si with no (or little) heating of the substrate. During the process of cooling silicon from the liquid to the solid state, the silicon recrystallizes from some silicon seeds, and forms grains of larger size. Typically, a laser with radiation of a suitable wavelength, which can be effectively absorbed by the a-Si but to which the glass substrate is transparent, is selected for melting the a-Si and transforming it into poly-Si.11 As mentioned above, a-Si contains many hydrogen atoms formed during thin-film growth by PECVD. Under high-power laser irradiation, the Si–H bonds break quickly to form a large amount of H2 gas that escapes out of the poly-Si, leading to film ablation. Hence, a dehydrogenation process is needed before recrystallization; this is achieved by heating the substrate or irradiating with a laser beam of lower power density. However, after forming grains of larger size, there are still many traps at the grain boundaries; another hydrogenation process is required to passivate these dangling bonds. Typically, the defect density decreases by one order of magnitude after this passivation process, compared to the as-grown poly-Si. As regards choosing a suitable laser to melt silicon, a shorter wavelength is preferred since the associated absorption coefficient is larger. However, if the laser wavelength is too short the laser radiation will be absorbed by the glass substrate. Typically, a XeCl excimer laser with an emission wavelength of 308 nm is used for silicon recrystallization; the process is called excimer laser crystallization. Although xenon is an inert gas, it reacts with chlorine and forms Xe+ Cl− when promoted from the ground to the excited state. It then releases the energy to return to the ground state, emitting light in the process, and finally splits into Xe and Cl atoms. The XeCl laser is a pulsed laser with a pulse duration of several tens of nanoseconds. During this short period, the silicon film will be heated up above the melting point. It then cools and forms larger grain boundaries. Also, because of the short pulse width, there is not too much heat generated, and it can dissipate with little or no heating of the substrate. Glass has a low thermal conductivity and the heat may melt only the surface of the glass substrate. However, the heat can quickly dissipate to the air from the other side of the glass. The laser crystallization process converts optical energy into the thermal energy used to melt silicon. Since the light propagates from the surface of the silicon thin film toward the glass substrate, the silicon at the top melts first. When the laser fluence increases, more silicon becomes liquid and finally the whole thin film is completely melted. Figure 3.10 shows the mobility and grain size as functions of laser fluence. There is a clear linear dependence between these two parameters. Microscope images of poly-Si thin films are also shown. One can see that the grain size and TFT mobility increase at first and then decrease as the laser fluence increases. Under exposure to the low energy laser beam, the film is uniform and of a-Si phase without being transformed into poly-Si; the grain size and TFT mobility are kept low. When the pulse energy increases, a-Si starts to melt and poly-Si is formed at the top surface of the thin film.
Figure 3.10 Grain size, TFT mobility and photographs of thin films under laser irradiation of different fluences.12 (Reprinted from Voutsas, A.T. (2003) A new era of crystallization: advances in polysilicon crystallization and crystal engineering. Appl. Surf. Sci., 208–209, 250. Copyright (2003), with permission from Elsevier.)
Thin-film transistors
43
0.5
Grain Size (μm)
0.4
NCM
0.3
PM 0.2
0.1
0 0
20
40 60 Number of Shots
80
100
Figure 3.11 Grain size as a function of number of shots (NCM, near complete melting; PM, partial melting).12 (Reprinted from Voutsas, A.T. (2003) A new era of crystallization: advances in polysilicon crystallization and crystal engineering. Appl. Surf. Sci., 208–209, 250. Copyright (2003), with permission from Elsevier.)
At this stage, the melt depth is less than the film thickness, and poly-Si solidifies vertically from the seeds provided by the underlying a-Si, which is called the partial-melting region. As the laser fluence further increases, the thin film is melted through. However, there is still some silicon remaining in the solid phase at the thin film–glass interface. This silicon forms discrete islands, rather than a continuous film, which provide seeds for the recrystallization process. When the pulse energy is further increased, the number of nucleation sites decreases, meaning that the grain size increases. The growth is not only vertical in direction but also lateral. During this stage, the grain size and TFT mobility increase with increasing laser fluence, and they reach a maximum at about 325 mJ cm−2 , as shown in Figure 3.10. This is called the near-complete-melting region. When the laser energy is very high, the whole thin film and the nucleation sites are molten. Nucleation starts somewhere in the liquid silicon, and hence the crystallization is nearly homogeneous without preferred directions, and the grain sizes decreases again. This is called the complete-melting region. One can see that, although it is possible to obtain poly-Si with a large grain size and high mobility, the process window is quite small (<50 mJ cm−2 ), which results in difficulty in manufacturing stable TFTs of large size. Since the irradiation area of the excimer laser beam is typically much smaller than the glass substrate, a scanning stage is used for the process which requires a uniform and stable laser beam. To expand the experimental tolerance and further improve device performance, one can use the multishot line beam scanning technique. As shown in Figure 3.11, with increasing numbers of shots, the grain size significantly increases and the shot-to-shot variation decreases.
3.4 Thin-film transistor characteristics A TFT is a device with three terminals. One is called the gate (G), which acts as a switch to open or close the other two terminals, called the source (S) and drain (D), as shown in Figure 3.12. In an a-Si TFT, the conducting carriers are electrons. Electrons drift from the source electrode to the drain, which results in
44
Introduction to Flat Panel Displays
D
current flow
G
S
Figure 3.12 Schematic representation of and notation for a TFT.
current flow from drain to source. The switching structure is based on the metal–insulator–semiconductor (MIS) structure, which is also used widely in the IC industry with crystalline silicon. In such a structure, the metal can attract and repel carriers at the semiconductor side, which changes the carrier concentration of the semiconductor. It effectively engineers the conductance and hence can be used to switch on and off with high and low conductance, respectively. Although operation modes for crystalline silicon, a-Si and poly-Si are different, the basic physics is similar. Here, we focus on the electrical characteristics of a-Si and poly-Si. Figure 3.13(a) shows the energy band diagram of an ideal MIS structure when V G = 0. Here, we first consider the semiconductor as crystalline silicon with a perfect lattice and no traps. Figure 3.13 implies that the work function of the metal (φ m , which is the energy needed to promote an electron from the metal to a vacuum) is aligned with the Fermi level of the semiconductor, which is not always true with different metal and semiconductor materials. The insulator has a very large bandgap and there is no carrier inside this layer. When applying a positive voltage on the metal side, negative charges are attracted at the surface of the semiconductor, as shown in Figure 3.13(b). Those carriers pile up at the semiconductor–oxide interface which effectively increases the electron density at this boundary. From Equation (3.12), we can see that the increase in electron density means the Fermi level and E c become closer, which results in the band bending of the conduction and valence bands, as shown in Figure 3.13(b). In a uniform semiconductor layer (same material and doping concentrations), the Fermi level must be kept flat. If not, electrons will flow from the high to low E F position until it becomes flat since the Fermi energy describes the probability of an electron occupying an energy state. In Figure 3.13(a) all the bands are flat; this is usually called the flat band condition, which is not necessary the case at V G = 0 for nonideal cases.
– – –
Ec
φm
EF Ei
EF Ei
φm
Ev Metal
Semiconductor Insulator (a)
Figure 3.13
Ec
Ev Metal
Semiconductor Insulator (b)
Band diagram of an ideal MIS structure at (a) V G = 0 and (b) V G > 0.
Thin-film transistors
45
Drain
Source a-Si:H
Glass substrate
Gate
Insulator
Figure 3.14 Cross-section of an a-Si TFT.
When using a-Si as the semiconductor material, the band bending behavior is similar to that of crystalline silicon, as shown in Figure 3.13, with some small modifications. The flat band condition at V G = 0 is typically true for an a-Si TFT.13 However, when applying a positive voltage to the metal electrode (small V G ), due to the trap states in a-Si, electrons accumulate at the oxide–semiconductor interface, first filling deep states and tail states with zero mobility. At this time, these carriers are localized and immobile. On further increasing the voltage to V G , say over a threshold voltage (V T ), a thin layer forms with mobile carriers at the oxide–semiconductor interface. Hence, one can use V G to create and control an electron path at this interface, which is called the channel. Figure 3.14 shows a cross-section of an a-Si TFT fabricated on a glass substrate. The gate electrode is deposited first, which is typically chromium, tantalum or aluminum. An insulating layer is then formed uniformly (using PECVD) covering the gate electrode, which can be silicon nitride (SiNx ) or silicon oxide (SiOx ). Then, a-Si is deposited on the top of the insulator as an MIS structure. On the right and left of the a-Si layer, source and drain electrodes are then formed, which typically consist of multiple layers, i.e. heavily doped n-type a-Si (n+ a-Si) followed by chromium/aluminum, and n+ a-Si forms an ohmic contact with the channel layer (a-Si) for better electron injection. The gate voltage affects the channel conductance and controls the on/off switching of this conductance. Current through drain to source (I D ) is modulated by the gate, which is also dependent on the drain voltage (V D ). Typically, the source is grounded (V s = 0). The I–V characteristics of an a-Si TFT can be described by the following equations, which are similar to those for crystalline silicon devices: 1 W [2(VG − VT )VD − VD2 ], 0 ≤ VD ≤ VG − VT , (3.17) ID = (μn Ci ) 2 L 1 W ID = (μn Ci ) (VG − VT )2 , VD > VG − VT , (3.18) 2 L where μn is the electron mobility (cm2 V−1 s−1 ), C i is the capacitance per unit area of the insulator layer (F cm−2 ), W is the channel width, L is the channel length (distance between the source and drain) and V T is the threshold voltage. Figure 3.15 shows typical electrical characteristics of an a-Si TFT with channel width and length of 200 and 27 m, respectively. Figure 3.15(a) shows the I D –V D curves for different V G values, which are typically called output characteristic curves. For a positive and sufficiently large V G , there are mobile electrons which make the channel conductive. Then, with increasing V D , I D continues increasing until V D = V G − V T , which is called the linear region. Since the channel region is somewhat like a resistance, it is reasonable that higher V D results in higher current. On further increasing the voltage, the channel cannot support a higher current density, leading to a saturation region. The saturation current increases with increasing gate voltage since it attracts more electrons accumulating at the insulator–semiconductor interface. Figure 3.15(b) shows the log(I D )−V G curves at V D = 0.1 and 10 V, which are also called the transfer characteristic curves. When V D = 10 V, one can see that, with increasing V G , I D is very low (about 10−12 A) below the V G < −1V range, sharply increases at 0 to 5 V (from 10−12 to 10−6 A) and saturates at
46
Introduction to Flat Panel Displays
1.5 × 10–5 VG = 20V Linear region
9.0 × 10–6
16V Saturation region 12V
6.0 × 10–6 3.0 × 10–6 0.0
ID(A)
ID(A)
1.2 × 10–5
8V 4V 0
5
10 VD(V)
15
20
10–4 10–5 10–6 10–7 10–8 10–9 10–10 10–11 10–12 10–13 –10
Ion VD = 10V
Subthreshold swing = (slope)–1
Ioff –5
0
5 10 VG(V)
(a)
15
20
25
(b)
2.0 × 10–7
ID(A)
1.5 × 10–7 1.0 × 10–7
VD = 0.1V
5.0 × 10–8 VT 0.0 –5
0
5 VG(V)
10
15
(c) Figure 3.15 (a) Output characteristic, (b) transfer characteristic and (c) I D versus V G at a small V D value of an a-Si TFT. (Data provided by Prof. I-Chun Cheng, National Taiwan University.)
about 10−5 A. From the transfer characteristic, one can obtain the on/off ratio of a TFT, which is the ratio of the maximum to minimum measurable current. Typically, values of 106 to 108 are achievable for an a-Si TFT. A higher on/off ratio is preferred since a higher ‘on’ current can provide better driving capability and lower ‘off’ current means lower leakage current. As is evident from Equations (3.17) and (3.18), increasing the mobility value effectively helps to improve the ‘on’ current, such as is achieved by using poly-Si rather than a-Si. However, due to the increase of the grain size, the surface roughness of poly-Si is greater than that of a-Si, which also increases the ‘off’ current. Hence, one may not achieve a higher on/off ratio when using poly-Si rather than a-Si, although poly-Si exhibits a higher carrier mobility. To quantitatively determine the V T value, it is easier to plot the I D –V G curve using a linear scale. Figure 3.15(c) shows the I D –V G curve for a small value of V D of 0.1 V in the low I D region. Here, the TFT operates in the linear region due to the small V D value. Equation (3.17) can be written approximately as W (VG − VT )VD , (3.19) ID ∼ (μn Ci ) L and we can assume that the V 2D term in equation (3.17) is very small and can be neglected. We can find V T from the x-intercept since I D = 0 at V G = V T ; this is the voltage required to switch on this TFT. Once V T is obtained, one can fit the mobility value for such a device from Equations (3.17) and (3.18). Above V T , there would be a significant current passing through the channel. In contrast, one can define a
Thin-film transistors
47
Staggered
Inverted staggered
Coplanar
Inverted coplanar
electrodes
semiconductor
insulator
Figure 3.16 Different configurations of TFTs.13 (Source: Powell, M.J. (1989) The physics of amorphous-silicon c 1989 IEEE.) thin-film transistors. IEEE Trans. Electron Dev., 36, 2753.
subthreshold swing from Figure 3.15(b). This is the inverse of the slope of log(I D ) versus V G , in terms of V/decade. This parameter indicates how much voltage is needed to switch the TFT on and off. A smaller subthreshold swing is preferred since this results in higher speed and lower power consumption. Typical TFTs can be categorized as normal or inverted structures, depending on whether their gate is on the top or bottom, respectively. Another classification depends on whether the drain/source and gate are on the same or opposite sides; these are called coplanar or staggered structures, respectively. Hence, there are basically four different kinds of TFT structures, which are shown in Figure 3.16. Different structures may be the result of different fabrication procedures and may result in different device performance. Typically, a-Si TFTs use the inverted staggered structure configuration because it makes them easy to fabricate and gives them better electrical characteristics. The a-Si channel is sensitive to photons of the external ambient that mainly illuminates from the bottom side in a display system: an inverted structure can shield the device from this light by virtue of the gate metal. Also, because of the laser irradiation recrystallization process to convert a-Si to poly-Si, a flat and continuous silicon thin film without other layers beneath is a better choice for low-temperature poly-Si (LTPS) TFTs. Hence, these devices typically make use of normal planar structures.
3.5 Passive matrix and active matrix driving schemes To drive an LCD or OLED display, passive matrix (PM) and active matrix (AM) driving techniques are commonly used.14, 15 For a PM LCD, due to the capacitor characteristics of the liquid crystal (LC) layer, a voltage applied to a selected pixel inevitably affects the neighboring pixels and degrades the contrast of the display, which is called crosstalk.16 Each row of a PM LCD or OLED display is sequentially active with a high frequency, and the human eye regards it as a planar frame rather than line scanning due to the persistence of vision. Since only one row can be selected and displayed at a time, the display time for a single row decreases with increasing number of rows, which limits the display size and resolution. In contrast, by using TFTs in an LCD or OLED pixel, the turn-on/turn-off and different grayscales can be controlled independently: this is called AM driving. Figure 3.17 shows the operation of a PM LCD. Electrodes in a stripe configuration are placed on two glass substrates. Stripes on the bottom and top glass substrates are aligned perpendicularly to work as column and row electrodes of the PM LCD, and each crossed section is a pixel region. LC material is filled between the two glass substrates, thus together they can be regarded as a capacitor. To display a
48
Introduction to Flat Panel Displays
Vs 0 0 0
–Vd –Vd –Vd –Vd (b)
(a) 0
0
Vs
0
0
Vs
0
0
–Vd –Vd –Vd –Vd
+Vd –Vd –Vd +Vd
(c)
(d) 0 0 0 Vs
+Vd –Vd –Vd +Vd (e) Figure 3.17 (a) Image displayed on a PM LCD, and the voltage at (b) first, (c) second, (d) third and (e) fourth time slots at each row and column.
‘T’ in a 4 × 4 LCD, the first row is first selected with a pulse voltage V s for the first time slot, then the driving voltages from columns 1 to 4 are applied with −V d , resulting in a voltage difference of V s + V d , and the left row lines 2 to 4 are applied with 0 V in root-mean-square (RMS) average (Vrms ), so their voltage differences in the first time slot are V d . Contrary to the V D and V S in Eq. 3.17 and 3.18, V d and V s here do not mean voltages at sources and drain, respectively, because it is PM driving without TFT. On the other hand, it represents voltages at data and scan lines. Then in the second time slot, the voltage on the row 1 electrode is reset to 0 Vrms , and the scanning voltage V s is applied on row 2, and driving voltages for columns 1 to 4 are set at −V d ; hence only those pixels in row 2 have a voltage of V s + V d and other pixels still experience a voltage difference of V d . For the third time slot, scanning voltages for row 3 are V s , but driving voltages for columns 2 and 3 are −V d , and for columns 1 and 4 are +V d ; thus the voltage differences on pixels (3, 2) and (3, 3) are V s + V d , and voltages V s −V d (V s + V d for on, V s − Vd for off) are applied on pixels (3, 1) and (3, 4). In this slot, other pixels have voltages of either
Thin-film transistors
49
−V d or + V d . A similar driving scheme is then applied for row 4 in the next time slot. Generally, the LCs will respond to the RMS value of the voltages they experience. For a PM LCD with N row lines, if one row is selected, a voltage of V s is applied on that row line to select all the pixels there, and V d or −V d from the column line is then used to turn each pixel on or off. But neighboring pixels in the same column will also be affected by the voltage from the same column line. As a result, for a whole frame scan, the on pixel has a voltage of V s + V d only for 1/N frame time, and has a voltage of V d or −V d for the remaining (N −1)/N frame time. Similarly, for a whole frame, the off pixel has a voltage of V s − V d only for 1/N frame time, and V d or −V d for the remaining (N − 1)/N frame time. Since LCs respond to the RMS value of the receiving voltage, the RMS value of on or off pixels can be calculated from the above schemes: 2 Von =
1 N −1 2 1 N −1 2 2 (Vs + Vd )2 + Vd and Voff Vd . = (Vs − Vd )2 + N N N N
So, on average, different pixels of a display can be made to achieve on or off states by setting appropriate V s and V d values. Also, the difference between V on and V off of this display is related to the number of rows N. For a practical LCD with a set voltage–transmittance curve, the difference between V on and V off cannot be zero; thus the maximum row number of the display is limited. Figure 3.18 shows the operation of a PM OLED. Here, we take a diode to illustrate the OLED display. As in the LCD case, to display a ‘T’ in a 4 × 4 OLED display, the first row is selected with a low voltage. The four OLEDs in the first row are biased at high voltage when the column lines are turned on. Then, for the next time slot, the second row is biased with low voltage, and so forth. The advantages of the PM OLED include ease of fabrication and low cost. However, since each row is L
H
H
L
H
H
H
H
H (a)
H
H
H
H
H
H
H
L
H
H
H
H
(c)
(b)
H
H
L L
H
H (d)
L
L
H
H
L
(e)
Figure 3.18 (a) Image displayed on a PM OLED, and the voltage at (b) first, (c) second, (d) third and (e) fourth time slots in each row and column.
50
Introduction to Flat Panel Displays
selected at a short time slot, the average luminance is the peak luminance over the number of rows. When the number of rows increases, the peak luminance needs to be very high, which shortens the device lifetime.17 There are some differences between OLEDs and LCDs as regards PM driving. For an OLED, each individual pixel has a fast turn-on time that is close to or faster than each slot time; thus the luminance for each pixel reaches a peak from this turn-on slot, and is then averaged over all the rows (or scanning time). In contrast, for an LCD, the LC response time is usually slower than the turn-on slot time; thus, for a single scanning slot with high voltage, there might not be enough time to respond to reach peak transmittance before the voltage is reset at the next time slot. But LCs will respond to the average RMS voltage value and output a transmittance according to this RMS value. Each pixel in a single frame of a PM LCD and PM OLED has on and off states, which means it has two gray levels, 0 and 1. To realize more gray levels for PM displays, pulse width modulation (PWM) is usually applied as one of the common driving schemes. That is, the pulse width on a pixel is controlled to obtain more gray levels. Due to the persistence of vision, the human eye receives the time-average brightness. As shown in Figure 3.19, a frame with two gray levels is divided into three time intervals which are displayed sequentially to obtain four gray levels. When three time intervals are all dark, the resulting screen is also dark (gray level 0). However, if one of the three is bright, then the screen looks dark gray (gray level 1). On increasing the number of bright time intervals, one can perceive that the ultimate color becomes light gray (gray level 2) and then white (gray level 3). Hence, there are four gray levels now controlled by the pulse width. For PM LCDs, due to the slow response of LCs, blurring of the image may occur when displaying moving pictures. For a display with a greater number of rows (higher resolution or larger size), the AM driving technique using TFTs is typically used. Figure 3.20 shows the equivalent circuit of an AM LCD. Only one scan line is selected at a time. Once the scan line is selected, the TFT is switched on; then voltage from the data line can be fed to charge both the LC and the storage capacitor. When the scan line is not selected and the TFT is switched off, the LC pixel can still be biased, since the voltage is held by the storage capacitor. This is also called a hold-type display. Usually an extra capacitor is connected in parallel to the LC to improve the retention characteristics. This is called a storage capacitor. Unlike the PM LCD, where the driving voltage is related to the number of rows, the AM LCD is a hold-type display and crosstalk is much weaker. The gray levels can be controlled independently for each pixel; this is achieved by applying a different V D directly from the data line.
Gray level 0 1
2 3 1
2
3
Time
Figure 3.19 Four gray levels achieved by PWM.
Thin-film transistors
51
Scan line
Data line
VG VS VD
Storage capacitor
LC
Figure 3.20 Equivalent circuit of an AM LCD.
Example 3.2 For the equivalent circuit shown in Figure 3.20, find the charging time of the LC, i.e. V S , from 0 to 4.5 Vrms with V D = 5 and 9 Vrms , respectively. V D is always kept at 5 and 9 Vrms , while V G is a step function which jumps from 0 to 20 Vrms at t = 0. There is no charge stored at LC (V S = 0 Vrms ) for t < 0. Timing diagrams of V G and V D are shown in Figure 3.21. When V G = 0 Vrms , the leakage current can be neglected (open circuit, Roff = ∞). When V G = 20 Vrms , the TFT can be replaced by a resistor with Ron = 5 M The capacitance of the LC layer (including storage capacitor) is C LC = 3 pF. VG 20V t=0
t
VD 5 or 9 V t Figure 3.21 Timing diagram of V G and V D of Example 3.2.
Answer. Figure 3.22 shows the equivalent circuit when V G = 0 and 20 Vrms . When the gate is open at t = 0: I VD
Roff VS
VD
Ron
CLC VG = 0
VS CLC
VG = 20
Figure 3.22 Equivalent circuit for (a) V G = 0 and (b) V G = 20 V.
(continued)
52
Introduction to Flat Panel Displays
I=
VD − VS , Ron
(3.20)
dVS . dt
(3.21)
I = CLC Equating Equations (3.20) and (3.21) yields
VD − VS dVS = CLC dt Ron
(3.22)
or Ron CLC
dVS + VS = VD . dt
(3.23)
Solving this differential equation, we obtain VS = VD [1 − exp(−t/τ )],
(3.24)
τ = Ron × CLC .
(3.25)
where
Hence, for V S , from 0 to 4.5 Vrms with V D = 5 and 9 Vrms , the charging times are 34.5 and 10.4 s, respectively. In reality, the gate signal is only applied for a very short time in a TFT LCD. For example, for a panel with resolution of 1024 × 768 at a frame rate of 60 Hz, the V G duration is only 1/(60 × 768) = 21.7 s. Hence, a suitable V D value is needed to achieving the desired V S at such a short time slot. As shown in Example 3.2, if V D = 5 Vrms , the charging time from 0 to 4.5 Vrms is even longer than the duration of V G , which results in insufficient charging of the LC cell.
The AM driving method of an OLED is similar to that of an LCD. However, since an OLED is a currentdriven device, rather than a capacitor as in the case of the LCD, a more complicated circuit is needed for an AM OLED display. Figure 3.23 shows the equivalent circuit of an AM OLED pixel. Typically, at least two TFTs are required, which are marked as the address and drive TFTs. When the electrical pulse is scanned to a certain row, the address TFT is switched on. To light on a particular pixel in this row, data line is selected, and the current provided by the power line passes through the drive TFT onto the OLED. Note that there is a capacitor, which can hold the voltage when the scan line is deselected, which means that the OLED keeps emitting light when the row is not selected. Hence, the peak luminance of AM OLEDs Power line Scan line
Data line
Cs
Id Drive TFT
Address TFT IOLED OLED Vc
Figure 3.23 Equivalent circuit of an AM OLED with a two-transistor and one-capacitor configuration.15
Thin-film transistors
53
does not need to be as high as that of PM OLEDs to achieve the same average brightness for each frame, which also extends the operation lifetime and makes a larger size display possible. This kind of pixel design is typically called a two-transistor and one-capacitor configuration. Here, the requirements for the address and drive TFT are quite different. The address TFT is used as a switch, and hence the on/off ratio is important. The drive TFT is used to provide current to the OLED, and hence a high current density capability is required. For the drive TFT to provide an adequate current, a high mobility is typically required. From this viewpoint, the LTPS technique is more suitable for driving the OLED. However, the uniformity achieved using the LTPS technique is not as good due to the laser annealing process. Since the J–V characteristics of OLEDs follow space charge limited current or trap charge limited current, described in Chapter 7, the current density varies greatly, even with a small driving voltage difference. Therefore, more than two TFTs are sometimes used to compensate for the nonuniformity of the LTPS TFTs and ensure a stable current density. To achieve different gray levels of the AM OLED, voltage modulation from the data line or PWM can be applied. The voltage modulation scheme in AM OLEDs is similar to that in AM LCDs. However, due to the steep luminance–voltage curve of OLEDs, it is not easy to control the gray levels precisely and uniformly. In the case of PWM driving, due to the higher frame rate, the power consumption is higher and the control system is more complicated.
3.6 Non-silicon-based thin-film transistors In this section, we will introduce two kinds of TFTs where the active layer consists of organic molecules and ZnO-based materials. Organic TFT (OTFT) can be fabricated with a low temperature deposition and a solution process, which is less complex compared with conventional Si-based technologies. In addition, due to the mechanical flexibility of organic material, OTFT can be used as the backbone of the flexible display, and this has the benefit of being rugged, thin and lightweight.18 Also, it is potentially of low cost due to the possibility of using the roll-to-roll process. When fabricating a flexible display on a plastic rather than a glass substrate, one important issue arises concerning the durability of the substrate. Several methods can be used to form the organic thin film. For organic materials with lower molecular weight which can be evaporated or sublimated, one can heat up materials in a vacuum chamber to deposit the organic thin film on the substrate. Pentacene is one of the most common low molecular weight organic materials used for p-channel TFTs. The molecular structure of pentacene is shown in Figure 3.24(a). The mobility value of OTFTs is varied by varying many parameters, such as molecule stacking configuration, surface morphology and grain size. By controlling the fabrication parameters such as deposition rate and substrate temperature, one may obtain pentacene with a larger grain size. The mobility of pentacene-based TFTs can be as high as 0.7 cm2 V−1 s−1 , which is comparable to that of a-Si TFTs.19 By growing a singlecrystal organic thin film such as rubrene (the molecular structure of which is shown in Figure 3.24(b)), the mobility value of an OTFT can exceed 20 cm2 V−1 s−1 .20 Also, one can dissolve the organic materials in certain solvents. This makes it possible to spin-coat, inkjet-print, stamp or imprint the solution on the
(a)
(b) Figure 3.24 Molecule structures of (a) pentacene and (b) rubrene.
54
Introduction to Flat Panel Displays
Figure 3.25
substrate
substrate
: electrode
: electrode
: insulator
: insulator
: active layer
: active layer
(a)
(b)
Device configurations of OTFTs: (a) bottom contact and (b) top contact.
substrate.21 After vaporization of the solvent, the organic material is left as a thin-film structure. Details of the fabrication of organic thin films are discussed in Section 7.5.3. Many OTFTs are sensitive to environmental species, such as water and oxygen.22 This means that it can be difficult to obtain a stable device for display applications. During device fabrication, formation of the organic thin film is typically directly followed by a passivation process, which is used to avoid attack by environmental species. Figure 3.25 shows two common device configurations.23 Both are of the inverted structure type, since the insulator must be formed before the organic thin film. For a bottom contact device, the organic layer is deposited at the top of the device. Drain, source, gate and insulators can be defined by conventional photolithography which provides a high resolution (less than 1 m). For a top contact device, the organic material is formed first, followed by drain and source electrode evaporation through a shadow mask. The resolution is limited to several tens of micrometers in this configuration. As regards device performance, OTFTs with a top contact typically exhibit a superior performance compared to bottom contact types due to the larger contact area and lower contact resistance. Typically, a molecular thin film exhibits a much lower mobility value than semiconductor materials with covalent bonds, which means the electric current provided by OTFTs is typically smaller than that of a-Si and poly-Si based TFTs. It is also difficult to change the carrier concentrations of the molecules by intentionally doping, which limits the flexibility of device design. Also, the long-term stability under operation and storage is still an issue for OTFTs.24 Due to the light-sensitive characteristics, a light shield is needed for the TFT region, which limits the aperture ratio of a display. By using high bandgap semiconductors, such as zinc oxide (ZnO), there is little absorption in the visible region. It is also possible to fabricate a light-insensitive TFT, called a transparent TFT (TTFT), which can greatly increase the aperture ratio.25 Figure 3.26 shows the device structure and transmission spectrum of a ZnO-based TFT. Here, all the electrodes (gate, drain and source) are of transparent indium tin oxide (ITO). The insulator is made of aluminum titanium oxide (ATO), which is also transparent. As shown in Figure 3.26(b), the average transmission over the visible range at the channel and source/drain region is about 75%. Another advantage is the high mobility (up to 20 cm2 V−1 s−1 ), which is much higher than that of a-Si. Fabrication techniques for ZnO thin films include sputtering or pulsed laser deposition. These are physical vapor deposition (PVD) techniques, and different from the CVD system for a-Si growth.26 Unlike a-Si, the amorphous phase is preferred for TTFTs since crystallization results in a high roughness and large leakage current, which reduces the on/off ratio of TTFTs. It is interesting that the mobility value does not differ much for ZnO-based materials in the crystalline and amorphous phases. In such materials, electrons mainly propagate along the large metal atom, which has a good wavefunction overlap in both crystalline and amorphous phases. To obtain good performance, it is also possible to incorporate other metal oxides into ZnO, such as In2 O3 and Ga2 O3 (IGZO).27 There are some requirements for material
Thin-film transistors
55
85 Drain (ITO, 300 nm) Transmission (%)
Source (ITO, 300 nm)
Channel (i-ZnO, 100 nm) Gate insulator (ATO, 220 nm) Gate (ITO, 200 nm)
Channel
75
65
55
Substrate (glass)
Source/drain 45 400
500
600
700
Wavelength (nm)
(a)
(b)
Figure 3.26 (a) Device structure and (b) transmission spectra of a TTFT.25 (Reprinted with permission from Hoffman, R.L., Norris, B.J. and Wager, J.F. (2003) ZnO-based transparent thin-film transistors. Appl. Phys. Lett., 82, 733. Copyright (2003), American Institute of Physics.)
selection. The materials should (1) tend to form an amorphous phase during PVD, (2) have a high carrier mobility and (3) exhibit low carrier concentration for reducing the off current. By reducing the process temperature, TTFTs can also be fabricated on flexible substrates.
Homework problems 3.1 A Si wafer is doped with phosphorus at 1017 cm−3 . The wafer thickness is 350 m. Assuming the shape of the Si wafer is a complete circle with four inches in diameter. Also assume the electron current is much higher than the whole one with electron mobility μn = 700 cm2 V−1 s−1 . Find the current when applying 10 V at both sides of this wafer. 3.2 GaAs and silicon have electron effective masses of 0.07m0 and 0.19m0 , respectively, at the bottom of the conduction band. What is the physical meaning of the fact that the ‘effective’ mass of an electron in a semiconductor is less than the ‘real’ mass of a free electron? Which material exhibits a steeper E–k curve at the bottom of the conduction band? Why? 3.3 Band diagram of an MIS device is shown below. Is V m (voltage applied to the metal) positive, negative, or zero under flat band condition (V FB )? Draw the band diagram under a short circuit condition (V = 0).
φm
EC Ei EF EV
Metal
Semiconductor Insulator
3.4 For a TFT with V T = 1 V, W = 50 m and L = 10 m, μn = 0.5 cm2 V−1 s−1 and C i = 15 nF cm−2 . Calculate the drain current at V G = 10 V and V D = 0.1 V. Repeat the calculation for V G = 20 V and V D = 20 V.
56
Introduction to Flat Panel Displays
References 1. Le Comber, P.G., Spear, W.E. and Ghaith, A. (1979) Amorphous-silicon field-effect device and possible application. Electron. Lett., 15, 179. 2. Kimura, M., Yudasaka, I., Kanbe, S. et al. (1999) Low temperature polysilicon thin film transistor driving with integrated driver for high resolution light emitting polymer display. IEEE Trans. Electron Dev., 46, 2282. 3. Mimura, A., Konishi, N., Ono, K. et al. (1989) High performance low-temperature poly-Si n-channel TFTs for LCD. IEEE Trans. Electron Dev., 36, 351. 4. Sze, S.M. (2001) Semiconductor Devices: Physics and Technology, 2nd edn, John Wiley & Sons, Ltd. 5. Streetman, B.G. and Banerjee, S.K. (2005) Solid State Electronics Devices, 6th edn, Prentice Hall. 6. Peyghambarian, N., Koch, S.W. and Mysyrowicz, A. (1993) Introduction to Semiconductor Optics, Prentice Hall. 7. Chelikowsky, J.R. and Cohen, M.L. (1976) Nonlocal pseudopotential calculations for the electronic structure of eleven diamond and zinc-blende semiconductors. Phys. Rev. B, 14, 556. 8. Tsukada, T. (2003) TFT/LCD: Liquid-Crystal Displays Addressed by Thin-Film Transistors, Taylor & Francis. 9. Wagner, S., Gleskova, H., Cheng, I.C. and Wu, M. (2003) Silicon for thin-film transistors. Thin Solid Films, 430, 15. 10. Street, R.A. (ed.) (1999) Technology and Applications of Amorphous Silicon, Springer-Verlag, Berlin/Heidelberg. 11. Shih, A., Meng, C.Y., Lee, S.C. and Chern, M.Y. (2000) Mechanism for pillar-shaped surface morphology of polysilicon prepared by excimer laser annealing. J. Appl. Phys., 88, 3725. 12. Voutsas, A.T. (2003) A new era of crystallization: advances in polysilicon crystallization and crystal engineering. Appl. Surf. Sci., 208–209, 250. 13. Powell, M.J. (1989) The physics of amorphous-silicon thin-film transistors. IEEE Trans. Electron Dev., 36, 2753. 14. den Boer, W. (2005) Active Matrix Liquid Crystal Displays: Fundamentals and Applications, Newnes. 15. Meng, Z. and Wong, M. (2002) Active-matrix organic light emitting diode displays realized using metal-induced unilaterally crystallized polycrystalline silicon thin-film transistors. IEEE Trans. Electron Dev., 49, 991. 16. Yeh, P. and Gu, C. (1999) Optics of Liquid Crystal Displays, John Wiley & Sons, Ltd, p. 248. 17. Kijima, Y., Asai, N., Kishii, N. and Tamura, S. (1997) RGB luminescence from passive-matrix organic LED’s. IEEE Trans. Electron Dev., 44, 1222. 18. Rogers, J.A. and Bao, Z. (2002) Printed plastic electronics and paperlike displays. J. Polym. Sci. Polym. Chem., 40, 3227. 19. Lin, Y.Y., Gundlach, D.J., Nelson, S.F. and Jackson, T.N. (1997) Pentacene-based organic thin-film transistors. IEEE Trans. Electron Dev., 44, 1325. 20. Briseno, A.L., Tseng, R.J., Ling, M.M. et al. (2006) High-performance organic single-crystal transistors on flexible substrates. Adv. Mater., 18, 2320. 21. Ling, M.M. and Bao, Z. (2004) Thin film deposition, patterning, and printing in organic thin film transistors. Chem. Mater., 16, 4824. 22. Zhu, Z.T., Mason, J.T., Dieckmann, R. and Malliaras, G.G. (2002) Humidity sensors based on pentacene thin-film transistors. Appl. Phys. Lett., 81, 4643. 23. Dimitrakopoulos, C.D. and Mascaro, D.J. (2001) Organic thin-film transistors: a review of recent advances. IBM J. Res. Develop., 45, 11. 24. Benor, A., Hoppe, A., Wagner, V. and Knipp, D. (2007) Electrical stability of pentacene thin film transistors. Org. Electron., 8, 749. 25. Hoffman, R.L., Norris, B.J. and Wager, J.F. (2003) ZnO-based transparent thin-film transistors. Appl. Phys. Lett., 82, 733. 26. Nomura, K., Ohta, H., Takagi, A. et al. (2004) Room-temperature fabrication of transparent flexible thin-film transistors using amorphous oxide semiconductors. Nature, 432, 488. 27. Nomura, K., Takagi, A., Kamiya, T. et al. (2006) Amorphous oxide semiconductors for high-performance flexible thin-film transistors. Jpn. J. Appl. Phys., 45, 4303.
4 Liquid crystal displays 4.1 Introduction Three types of liquid crystal display (LCD) have been developed: (1) transmissive, (2) reflective and (3) transflective. A transmissive LCD uses a backlight for illuminating the LCD panel which results in high brightness (300–500 nits) and high contrast ratio (>2000:1). Some transmissive LCDs do not use phase compensation films or multidomain structures so that their viewing angle is limited and they are more suitable for single-viewer applications, such as personal digital assistants and notebook computers. With phase compensation films and multidomain structures, the direct-view transmissive LCDs exhibit a wide viewing angle and have been used extensively for desktop computers and TVs. Transmissive microdisplays are commonly used in projection displays, such as data projectors.1 For these, high-power arc lamps or light-emitting diode (LED) arrays are used as light sources. Using a projection lens, the displayed image is magnified by more than 50 times. To reduce the size of the optics and cost, the LCD panel is usually made small (less than 25 mm in diagonal) and each pixel size is ∼20–40 m. Thus, poly-silicon-based thin-film transistor (TFT) LCDs are the common choice. Similarly, reflective LCDs can be subdivided into direct-view and projection displays. A direct-view reflective LCD, e.g. cholesteric liquid crystal display (Ch-LCD)2 and bistable nematic LCD,3 uses ambient light to produce the displayed images. A Ch-LCD has a helical structure which reflects color so that the display does not require color filters or polarizers. Thus, the reflectance for a given color band which depends on the pitch length and refractive index of the liquid crystal (LC) employed is relatively high (∼30 %). Moreover, it does not require a backlight so that its weight is low and the total device thickness can be small (<200 m). Therefore, it is a strong contender for color flexible displays. A Ch-LCD is a bistable device so that its power consumption is low, provided that the device is not refreshed too frequently. A major drawback of reflective direct-view LCDs is their poor readability under low ambient light. Another type of reflective LCD is designed for projection TVs using liquid-crystal-on-silicon (LCoS) microdisplay panels. Unlike a transmissive microdisplay, an LCoS display is a reflective device. Here the reflector employed is an aluminum metallic mirror. Crystalline silicon has high mobility so that the pixel size can be made small (<10 m) and aperture ratio greater than 90 %. Therefore, the image not only has high resolution but also is seamless. By contrast, a transmissive microdisplay’s aperture ratio is about 65 %. The light blocked by the black matrices shows up in the screen as dark patterns (also known as screen door effect). The viewing angle of an LCD is less critical in projection than direct-view displays because in a projection display the polarizing beam splitter has a narrower acceptance angle than the employed LCD.
Introduction to Flat Panel Displays c 2008 John Wiley & Sons, Ltd
J.-H. Lee, D.N. Liu and S.-T. Wu
58
Introduction to Flat Panel Displays
For outdoor applications, the displayed images of a transmissive LCD can be washed out by sunlight. A reflective LCD would be a better choice. However, such a reflective display is unreadable in dark ambient. Therefore, a transflective LCD which combines the features of a transmissive display and a reflective display seems an ideal choice. In dark ambient the backlight is on and the display works as a transmissive one; in bright ambient the backlight is off and only the reflective mode is operational. Several monographs have been dedicated to advanced projection displays,4 reflective displays5 and flexible displays.6 Therefore, here we will focus on the mainstream TFT-addressed wide-view transmissive LCDs. We will start by introducing twisted nematic (TN), in-plane switching (IPS) and fringe field switching (FFS) displays, and then multidomain vertical alignment (MVA). Phase compensation methods for achieving wide viewing angles are also addressed.
4.2 Transmissive thin-film transistor liquid crystal displays Figure 4.1 shows the device structure of a TFT LCD using amorphous silicon (a-Si) transistors. An LCD is a nonemissive display, i.e. it does not emit light; instead, it functions as a two-dimensional spatial light modulator. Thus, a backlight is needed. A diffuser is used to homogenize the backlight in order to avoid hot spots. Some optical films are stacked to steer the Lambertian backlight to the middle ±40◦ for improving display brightness. Since most LCDs require linearly polarized light in order to exhibit a high contrast ratio, two sheets of stretched dichroic polarizers are used for large-size direct-view displays. The first glass substrate contains TFT arrays, which serve as independent light switches. Each display pixel is controlled by a TFT. Since TFTs are sensitive and should be shielded from backlight illumination, the actual aperture ratio (the transparent indium tin oxide (ITO) electrode area) is reduced to ∼80 %, depending on the pixel density. As the pixel density increases, the aperture ratio decreases. The LC layer is sandwiched between two ITO substrates whose inner surface is coated with a thin (80–100 nm) polyimide layer. Some LCDs (TN, IPS and FFS) require rubbing but some (MVA and patterned vertical alignment)
a-Si TFTs
Gate or row electrode Common electrode Color filter
ITO
TFT substrate Polarizer
240 µm
Backlight Diffuser
R 80 µm
G
Data or column electrode
B
Gate or row electrode LC
Polarizer Common substrate Figure 4.1
Device structure of a color pixel of a transmissive TFT LCD.
Liquid crystal displays
59
1.0 Intensity (a.u.)
B
G
R
0.8 0.6 0.4 0.2 0.0 350
450
550 650 Wavelength (nm)
750
Figure 4.2 Transmission spectra of RGB color filters (thick gray lines), and emission spectra of CCFL backlight (thin black lines) and LEDs (dashed lines). Source: R. Lu et al., Journal of Display Technology, vol. 1, pp. 3–14 © 2005 IEEE.
do not. The cell gap is usually controlled at ∼3.5–4.0 m for a transmissive LCD. The performance of the display, such as light throughput, response time and viewing angle, is influenced by the LC configuration employed. For direct-view LCDs, compact size, low weight and low power consumption are equally important as viewing angle, color and contrast ratio. For direct-view LCDs, color filters are embedded in the inner side of the top (second) substrate. Three subpixels (red (R), green (G) and blue (B)) form a color pixel. The size of each subpixel is ∼80 m × 240 m. Each subpixel transmits only one color; the other colors are absorbed. Figure 4.2 depicts the emission spectra of a backlight (cold cathode fluorescent lamp, CCFL) and RGB LEDs, and the transmission spectra of RGB color filters. From Figure 4.2, it can be seen that the transmission spectra of RGB color filters are relatively broad. The advantage is to transmit more light however its color purity is degraded. The peak transmission of the RGB color filters is ∼70, 80 and 90 %, respectively. Roughly speaking, each color filter only transmits ∼25 % of the incident white light. The rest, ∼75 %, is absorbed by the color pigments. Moreover, the CCFL emits two unwanted lines: blue-green (∼480 nm) and orange (∼580 nm). The blue-green light will transmit through the blue and green color filters simultaneously. Similarly, the orange light will transmit through the green and red color filters simultaneously. These leaked lights will downgrade the color purity (or color saturation) of the display. Therefore, the color gamut of a typical transmissive TFT LCD is ∼75 % of NTSC (National Television System Committee) standard. With improved CCFL spectra, the color gamut can reach ∼92 %. LEDs have narrower emission spectra that also match better with the transmission spectra of the color filters; thus, the color gamut reaches ∼120 %.7 For a display device, a wider color gamut is not necessarily better; natural colors are also important. After all, display is an art where perception plays an important role. After taking into account the optical losses from polarizers, color filters and TFT aperture ratio, the overall system optical efficiency is about 6–7 % for a direct-view LCD. If wide-view technology is included,8 the total light efficiency is decreased to ∼5 %. Low optical efficiency implies high power consumption and more heat generation inside the display chassis. For a thin LCD, thermal dissipation is a critical issue. For portable displays, low power consumption is desirable because it lengthens the battery operating time. Several approaches have been developed to reduce power consumption, e.g. polarization conversion of backlight9 and two-dimensional LED backlights with local dimming capability.10 12 The use of LED backlights offers several additional advantages such as wide color gamut, high dynamic contrast ratio (>50 000:1), about two times reduction in power consumption and fast turn-on and off times (∼10 ns) for reducing motion picture image blurs.13 Some technological concerns are color and power drifting as the junction temperature changes, and cost.
60
Introduction to Flat Panel Displays
4.3 Liquid crystal materials An LC possesses physical properties that are intermediate between those of a crystalline solid and an isotropic liquid. They are fluid-like, yet the arrangement of molecules exhibits structural orders. Three types of liquid crystals have been discovered so far: (1) thermotropic, (2) polymeric and (3) lyotropic. Thermotropic liquid crystals have been studied extensively and their applications are widespread. Most TFT LCDs use thermotropic (nematic) liquid crystals. Polymeric liquid crystals have found interesting applications in optical films, electronic devices and ultrahigh-strength materials. Lyotropic liquid crystals are receiving increasing scientific and technological attention because of the way they reflect the unique properties of their constituent molecules. Although the LC material only occupies a small portion in a display device, it makes crucial contributions to the device performances. For instance, the device contrast ratio, response time, viewing angle and operating voltage are all related to the LC material employed and its alignment. The refractive indices and cell gap determine the phase retardation of the LC device. The dielectric constants and elastic constants jointly determine the threshold voltage. The viscosity, cell gap, driving voltage and temperature determine the response time. Within the thermotropic LC family, there are three distinct molecular structures: smectic, nematic and cholesteric. Among the smectic branch, ferroelectric liquid crystals (FLCs) exhibit many fascinating properties, such as bistability, layered structure, in-plane molecular reorientation and microsecond response time.14 FLCs have been used in near-the-eye microdisplays15 where a typical panel size is about 2 cm by 2 cm. Thus, a uniform submicrometer cell gap is easier to maintain. For large panels, the alignment uniformity and mechanical stability are both challenging. Thus, the FLC industry is still waiting to take off. Cholesteric liquid crystals (CLCs) exhibit a helical structure. This structure reflects color without using color filters and polarizers if the Bragg reflection is controlled to be in the visible spectral region. The contrast ratio of a polarizer-free CLC is ∼30:1 which is inadequate for high-end displays, such as computers and TVs. A 30:1 contrast ratio implies ∼3 % unwanted light leakage at each pixel which will degrade the color purity. However, for electronic newspaper applications, a 30:1 contrast ratio is sufficient. A typical newspaper has ∼8:1 contrast ratio and ∼50–60 % reflectivity. White printing paper has ∼80 % reflectivity and ∼15:1 contrast ratio. Therefore, a CLC is better suited for such applications. The encapsulated CLC display16 is thin and flexible. It is an emerging technology for flexible LCDs and electronic books. In the following subsections, we focus on the basic molecular structures and physical properties of nematic LCs for mainstream TFT LCDs.
4.3.1 Phase transition temperatures Only a few LC compounds exhibit mesogenic phase at room temperature (∼23 ◦ C). The following compound, 4 -pentyl-4-cyanobiphenyl (best known as 5CB),17 is such an example:
C5H11
CN
(I) Its nematic range is actually from 24 to 35.3 ◦ C, but due to supercooling effect 5CB remains liquid at room temperature. Therefore, many of its properties can be studied conveniently at room temperature without a bulky heating device. However, for display applications a wide nematic range (from −40 to
Liquid crystal displays
61
90 ◦ C) is highly desirable. To widen the nematic range, eutectic mixtures are commonly used. To obtain such a wide nematic range, a commercial mixture consisting of 10–15 components is not uncommon.
4.3.2 Eutectic mixtures Consider a binary mixture as an example to illustrate the working principles. Figure 4.3 shows the phase diagram of a binary mixture. The mesogenic ranges of components 1 and 2 are shown on the right and left vertical axes, where T mp1,2 represents the melting temperature and T c1,2 the clearing temperature of compounds 1 and 2, respectively. The horizontal axis represents the molar concentration (X 2 ) of component 2. As the concentration of compound 2 increases, the melting point of the mixture gradually decreases.At an appropriate molar concentration, the melting point of the mixture reaches a minimum. The mixture formulated according to these compositions is called a eutectic mixture. As the concentration of the second compound exceeds this eutectic point, the mixture’s melting temperature increases gradually. Meanwhile, the clearing point of the mixture is a linear superposition of the two components. That is to say, at the eutectic point the mixture has the lowest melting point and possibly the widest mesogenic range. The optimal mixing ratio of a eutectic mixture is described by the Schröder–Van Laar equation, first suggested more than a century ago:18, 19 Tmp =
Hi , (Hi /Ti ) − R ln(Xi )
(4.1)
where T mp is the mixture’s melting temperature (in K), T i the melting point, H i the heat fusion enthalpy (in cal mol−1 ) and X i the mole concentration of component i, and R is the gas constant (1.98 cal mol−1 K−1 ). In the Schröder–Van Laar equation, the following assumptions are made in order to form an ideal eutectic mixture: (1) the two components crystallize in pure form and do not form mixed crystals, (2) the liquid phase is an ideal mixed phase and (3) the differences in the heat capacities of the pure components in the molten and crystallized form are small. If the compound structures are less like each other, then the calculated melting point from Equation (4.1) will be closer to the experimental data.
TC1
Isotropic TC2
Tmp1 Liquid Tmp2
Solid 0
100
Concentration (X2), % Figure 4.3
Phase diagram showing eutectic formulation of a binary mixture.
62
Introduction to Flat Panel Displays
The clearing temperature (T c ) of the eutectic mixture can be calculated from the clearing point (T ci ) of the individual components as Xi Tci . (4.2) Tc = i
As shown in Figure 4.3, the final clearing temperature of the binary mixture is a linear superposition of that of the individual compounds. LC compounds with a high T ci are helpful for enhancing T c of the mixture. However, its melting point and H value need to be taken into consideration.
Example 4.1: E7 mixture The commercial Merck E7 mixture consists of four components: 5CB, 7CB (4-cyano4 -n-heptylbiphenyl), 8OCB (4-cyano-4 -n-oxyoctylbiphenyl) and 5CT (4-cyano-4 -n-pentylp-terphenyl).20 Their phase transition temperatures and associated heat fusion enthalpies are listed in Table 4.1. Table 4.1
Compositions of E7.
Compound
Phase transitionsa
5CB 7CB 8OCB 5CT
K 24 N 35.3 I K 30 N 42.8 I K 54.5 SA 67.5 N 81 I K 131 N 240 I
H (cal mol−1 )
Content (mol%)
4100 6200 5900 4100
49.60 28.13 14.38 7.89
Content (wt%)
45.53 28.74 16.28 9.46
The unit of phase transition temperature is ◦ C. K, crystalline phase; N, nematic; SA , smectic-A; I, isotropic.
a
As listed in Table 4.1, compound 8OCB has smectic-A phase below nematic. For mixture formulations, this could create some uncertainties about which temperature and heat fusion enthalpy to use in Equation (4.1). According to experimental observation, a nematic mixture containing a small percentage of smectic component will still exhibit a pure nematic phase as long as the smectic component is less than ∼20 %. The calculated melting temperature of E7 is −3 ◦ C and clearing point is 60 ◦ C. In calculations we use mole concentration, but in real mixture formulations we prefer to use weight percent (wt%). To convert mole concentration to weight percent, we use the following formula: Xi Mi (wt%)i = , (4.3) Xj Mj j
where M i is the molecular weight for each individual component. Table 4.1 lists both mole concentration and weight percent of the four components used in E7.
4.3.3 Dielectric constants The dielectric constants of a liquid crystal affect the operation voltage, resistivity and response time. For example, in a vertical alignment (VA) cell the threshold voltage (V th ) is related to dielectric anisotropy √ (ε = ε − ε⊥ ) and bend elastic constant (K 33 ) as Vth = π K33 /ε0 ε.21 Thus, a smaller K 33 and larger ε would lower the threshold voltage. However, a large ε could lead to a high viscosity because of a strong polar group or more polar groups being involved.
Liquid crystal displays
63
From Maier and Meier mean field theory,22 dielectric anisotropy of an LC is mainly determined by the dipole moment (μ), its orientation angle (θ) with respect to the principal molecular axis and order parameter (S) as 2 Fμ 2 ε = NhF ( α// − α⊥ ) − (1 − 3 cos θ)S . (4.4) 2kT Here, N is the molecular packing density, h = 3ε/(2ε + 1) is the cavity field factor, ε = (ε + 2ε⊥ )/3 is the averaged dielectric constant, F is the Onsager reaction field and <α⊥ > and < α⊥> are the principal elements of the molecular polarizability tensor. From Equation (4.4), for a nonpolar compound, μ ≈ 0 and its dielectric anisotropy is very small (ε < 0.5). In this case, ε is determined mainly by the differential molecular polarizability, i.e. the first term in Equation (4.4). For a polar compound, the dielectric anisotropy depends on the dipole moment, angle θ, temperature (T ) and applied frequency. If an LC has more than one dipole, then the resultant dipole moment is their vector summation. In a phenyl ring, the position of the dipole is defined as 2
3
1
4 6
5
From Equation (4.4), if a polar compound has an effective dipole at θ < 55◦ , then its ε is positive. On the other hand, ε becomes negative if θ > 55◦ . Fluoro (F),23 cyano (CN)24 and isothiocyanato (NCS)25 are the three commonly employed polar groups. Among them, the fluoro group possesses a modest dipole moment (μ ≈ 1.5 debye), high resistivity and low viscosity. However, its strong negativity compresses the electron clouds and, subsequently, lowers the compound’s birefringence. For direct-view LCDs, the required birefringence is around 0.1, depending on the LC alignment and cell gap (d) employed. On the other hand, the cyano and isothiocyanato groups not only exhibit a large dipole moment (μ ≈ 3.9 debye for C N and ≈3.7 debye for N C S) but also contribute to lengthen the -electron conjugation. As a result, their birefringence is much higher than their fluorinated counterpart. High birefringence is favorable for long wavelength, such as infrared, applications in order to use a thin cell gap to achieve the same phase change while keeping a fast response time. Under strong anchoring condition, the LC response time is proportional to d 2 . However, the CN compounds are more viscous than the corresponding NCS and fluoro compounds. Therefore, their major applications are in low-end displays such as wristwatches and calculators where response time is not crucial.
Example 4.2: positive ε LCs Positive ε LCs have been used in TN26 and IPS27, 28 displays, although IPS displays can also contain negative ε LCs. For TFT LCDs, the LC material employed must also possess a high resistivity (>1013 cm) in order to steadily hold charges and avoid image flickering.29 The resistivity of an LC mixture depends heavily on the impurity contents, e.g. ions. The purification process plays an important role in removing the ions for achieving high resistivity. Fluorinated compounds exhibit a high resistivity and are the natural choices for TFT LCDs.30, 31 (continued)
64
Introduction to Flat Panel Displays
A typical fluorinated LC structure is shown below: (F) R1
F (F) (II)
Most LC compounds discovered so far possess at least two rings, cyclohexane–cyclohexane, cyclohexane–phenyl or phenyl–phenyl, and a flexible alkyl or alkoxy chain. The compound shown in structure (II) has two cyclohexane and one phenyl rings. The R1 group represents a terminal alkyl chain, and a single or multiple fluoro substitutions take place in the phenyl ring. For multiple dipoles, the net dipole moment can be calculated from the vector sum. From Equation (4.4), to obtain the largest ε for a given dipole, the best position for the fluoro substitution is along the principal molecular axis, i.e. in the 4-position. The single fluoro compound should have ε ≈ 5. To further increase ε, more fluoro groups can be added. For example, compound (II) has two more fluoro groups in the 3- and 5-positions.32 Its ε is about 10, but its birefringence would slightly decrease (because of the lower molecular packing density) and viscosity increases substantially (because of the higher moment of inertia). The birefringence of compound (II) is around 0.07. If a higher birefringence is needed, the middle cyclohexane ring can be replaced by a phenyl ring. The elongated electron cloud will enhance the birefringence to ∼0.12 without increasing the viscosity noticeably. The phase transition temperatures of an LC compound are difficult to predict before the compound is synthesized. In general, the lateral fluoro substitution lowers the melting temperature of the parent compound because the increased intermolecular separation leads to a weaker molecular association. Thus, a smaller thermal energy is able to separate the molecules which implies to a lower melting point. A drawback of the lateral substitution is the increased viscosity.
Example 4.3: negative ε LCs From Equation (4.4), in order to obtain a negative dielectric anisotropy, the dipoles should be in the lateral (2, 3) positions. For the interest of obtaining high resistivity, a lateral difluoro group is a favorable choice. The negative ε LCs are useful for VA.33 The VA cell exhibits an unprecedented contrast ratio when viewed in the normal direction between two crossed linear polarizers. However, a single-domain VA cell has a relatively narrow viewing angle and is only useful for projection displays. For wide-view LCDs, a MVA (four domains) cell is required. The following structure is an example of a negative ε LC:34 F C3H7
F OC2H5
(III)
Liquid crystal displays
65
Compound (III) has two lateral fluoro groups in the (2, 3) positions so that their dipoles in the horizontal components are perfectly cancelled whereas the vertical components add up. Thus, the net ε is negative. A typical ε of lateral difluoro compounds is −4. The neighboring alkoxy group also has a dipole in the vertical direction. Therefore, it contributes to enlarge the dielectric anisotropy (ε ≈ −6). However, the alkoxy group has a higher viscosity than its corresponding alkyl group. To further increase ε, we could substitute more fluoro groups in the middle phenyl ring or replace one of the fluoro groups with a stronger polar group. Both approaches would cause the rotational viscosity to increase.
4.3.4 Elastic constants There are three basic elastic constants (splay K 11 , twist K 22 and bend K 33 ) involved in the electro-optics of an LC cell depending on the molecular alignment.35 Elastic constants affect an LC device through threshold voltage and response time.Asmaller elastic constant leads to a lower threshold voltage; however, the response time, which is proportional to the viscoelastic coefficient (γ1 /K ii ), is increased. Therefore, a proper balance between threshold voltage and response time should be taken into consideration. Several molecular theories have been developed for correlating the Frank elastic constants with molecular constituents. The commonly employed one is mean field theory.36, 37 In mean field theory, the three elastic constants are expressed as Kii = aS 2 ,
(4.5)
where a is a proportionality constant and S is the order parameter. For crystalline materials S = 1 and for isotropic materials S = 0. A nematic LC has S ≈ 0.6. As the temperature increases, S decreases. For many LC compounds and mixtures, the magnitude of elastic constants has the following order: K 33 > K 11 > K 22 . Therefore, LC alignment also plays an influential role in response time. For example, a VA cell (K 33 effect) should have a faster response time than an IPS or FFS cell (K 22 effect) owing to the elastic constant effect, provided that all the other parameters such as cell gap and viscosity remain the same. Usually, lateral difluoro substitutions increase the rotational viscosity because of the increased molecular moment of inertia.
4.3.5 Rotational viscosity Viscosity, especially rotational viscosity (γ1 ), plays a crucial role in the LC response time. The response time of a nematic LC device is linearly proportional to γ1 .38 The rotational viscosity of an aligned LC depends on the detailed molecular constituents, structure, intermolecular association and temperature. A bulkier and heavier compound tends to have a higher viscosity. As the temperature increases, viscosity decreases rapidly. Several theories, both rigorous or semiempirical, have been developed in an attempt to account for the origin of the LC viscosity.39, 40 However, owing to the complicated anisotropic attractive and steric repulsive interactions among LC molecules, these theoretical results are not completely satisfactory.41, 42 A general temperature-dependent rotational viscosity can be expressed as E , (4.6) γ1 = bS exp kT where b is a proportionality constant which takes into account the molecular shape, dimension and moment of inertia, S is the order parameter, E is the activation energy of molecular rotation, k is the
66
Introduction to Flat Panel Displays
Boltzmann constant and T is the operating temperature. When the temperature is not too close to the clearing point (T c ), the order parameter can be approximated as43 S = (1 − T /Tc )α ,
(4.7)
where α is a material parameter. Generally speaking, rotational viscosity is a complicated function of molecular shape, moment of inertia, activation energy and temperature. Among these factors, activation energy and temperature are the most crucial ones.44 The activation energy depends on the detailed intermolecular interactions. An empirical rule is that for every 10–15 degrees of temperature rise, the rotational viscosity decreases by about two times. From the molecular structure viewpoint, a linear LC molecule is more likely to have a low viscosity.45 However, all other properties also need to be taken into account. For instance, a linear structure may lack flexibility and lead to a higher melting point. Within the same homologues, a longer alkyl chain will in general (except for the even–odd effect) have a lower melting temperature. However, its moment of inertia is increased. As a result, a homologue with a longer chain length is likely to exhibit a higher viscosity.
4.3.6 Optical properties The major absorption of an LC compound occurs in the ultraviolet (UV) and infrared (IR) regions. The → ∗ electronic transitions take place in the vacuum UV (100–180 nm) region whereas the → ∗ electronic transitions occur in the UV (180–400 nm) region. Figure 4.4 shows the measured polarized UV absorption spectra of 5CB.46 The λ1 band which is centered at ∼200 nm consists of two closely overlapped bands. The λ2 band shifts to ∼282 nm. The λ0 band should occur in the vacuum UV region (λ0 ≈ 120 nm) which is not shown in Figure 4.4.
2.0 5CB
OPTICAL DENSITY
1.6 II 1.2
0.8
0.4
0
200
250
300
350
WAVELENGTH (nm) Figure 4.4 Measured polarized absorption spectra of 5CB. The middle trace is for unpolarized light. λ1 ≈ 200 nm and λ2 ≈ 282 nm. (Reprinted with permission from S.T. Wu, E. Ramos and U. Finkenzeller, J. Appl. Phys., 68, 78–85. Copyright (1990), American Institute of Physics)
Liquid crystal displays
67
4.3.7 Refractive indices Refractive index has great impact on LC devices. Almost every electro-optic effect of LC modulators, no matter amplitude or phase modulation, involves refractive index change. An aligned LC exhibits anisotropic properties, including dielectric, elastic and optical anisotropies. Consider a homogeneous alignment as an example.47 Assume linearly polarized light is incident on the LC cell in the normal direction. If the polarization axis is parallel to the LC alignment axis (i.e. LC director which represents an average molecular distribution axis), then the light experiences the extraordinary refractive index ne . If the polarization is perpendicular to the LC directors, then the light experiences the ordinary refractive index no . The difference between ne and no is called birefringence, defined as n = ne − no . Refractive indices are dependent on the wavelength and temperature. For a full-color LCD, RGB color filters are employed. Thus, the refractive indices at these wavelengths need to be known in order to optimize the device performance. Moreover, about 50 % of the backlight is absorbed by the polarizer. The absorbed light turns into heat and causes the LCD panel’s temperature to increase. As the temperature increases, refractive indices decrease gradually. The following discussion describes how wavelength and temperature affect LC refractive indices.
4.3.7.1 Wavelength effect Based on the electronic absorption, a three-band model which takes one → ∗ transition (the λ0 band) and two → ∗ transitions (the λ1 and λ2 bands) into consideration has been developed. In the three-band model, the refractive indices (ne and no ) are expressed as follows:48, 49 λ2 λ2 λ2 λ2 λ2 λ2 ne,o ∼ = 1 + g0e,o 2 0 2 + g1e,o 2 1 2 + g2e,o 2 2 2 . λ − λ0 λ − λ1 λ − λ2
(4.8)
The three-band model clearly describes the origins of refractive indices of LC compounds. However, a commercial mixture usually consists of several compounds with different molecular structures in order to obtain a wide nematic range. The individual λi values are therefore different. Under such a circumstance, Equation (4.8) would have too many unknowns to quantitatively describe the refractive indices of an LC mixture. In the off-resonance region, the rightmost three terms in Equation (4.8) can be expanded by a power series to the λ−4 terms to form the extended Cauchy equations for describing the wavelength-dependent refractive indices of anisotropic LCs:50 Be,o Ce,o ne,o ∼ = Ae,o + 2 + 4 , λ λ
(4.9)
where Ae,o , Be,o and C e,o are three Cauchy coefficients. Although Equation (4.9) is derived based on an LC compound, it can be extended easily to include eutectic mixtures by taking the superposition of each compound. From Equation (4.9), if we measure the refractive indices at three wavelengths, the three Cauchy coefficients (Ae,o , Be,o and C e,o ) can be obtained by fitting the experimental results. Once these coefficients are determined, the refractive indices at any wavelength can be calculated. From Equation (4.9), both refractive indices and birefringence decrease as the wavelength increases. In the long-wavelength (IR and millimeter wave) region, ne and no are reduced to Ae and Ao , respectively. The coefficients Ae and Ao are constants; they are independent of wavelength, but dependent on temperature. This means that, in the IR region, the refractive indices are insensitive to wavelength, except for the resonance enhancement effect near the local molecular vibration bands. This prediction is consistent with much experimental evidence.51 Figure 4.5 depicts the wavelength-dependent refractive indices of E7 at T = 25 ◦ C. The extended Cauchy model is extrapolated to the near- and far-IR. The extrapolated lines almost pass through the center of the experimental data measured at λ = 1.55 and 10.6 m. The largest difference between the extrapolated and experimental data is only 0.4 %.
68
Introduction to Flat Panel Displays
1.80
E7
25 °C
Refractive Indices
1.75 ne
1.70 1.65 1.60 1.55
no
1.50 0
2
4
6 8 Wavelength (m)
10
12
Figure 4.5 Wavelength-dependent refractive indices of E7 at T = 25 ◦ C. Open squares and circles are ne and no measured in the visible spectrum. Solid lines are fittings to the experimental data measured in the visible spectrum by using the extended Cauchy equation (Equation (4.9)). The fitting parameters are as follows: [Ae = 1.6933, Be = 0.0078 m2 , C e = 0.0028 m4 ] and [Ao = 1.4994, Bo = 0.0070 m2 , C o = 0.004 m4 ]. The down and up triangles are ne and no measured at T = 25 ◦ C and λ = 1.55 and 10.6 m, respectively. (Reprinted with permission from J. Li and S.T. Wu, J. Appl. Phys., 97, 073501. Copyright (2005), American Institute of Physics)
Equation (4.9) applies equally well to both high- and low-birefringence LC materials in the offresonance region. For low-birefringence (n < 0.12) LC mixtures, the λ−4 terms are insignificant and can be omitted, and the extended Cauchy equations are simplified as52 Be,o ne,o ∼ = Ae,o + 2 . λ
(4.10)
Thus, ne and no each has only two fitting parameters. By measuring the refractive indices at two wavelengths, we can determine Ae,o and Be,o . Once these two parameters are determined, ne and no can be calculated at any wavelength of interest. Because most TFT LC mixtures have n ≈ 0.1, the two-coefficient Cauchy model is adequate to describe the refractive index dispersions. Although the extended Cauchy equation fits experimental data well,53 its physical origin is not clear. A better physical meaning can be obtained by the three-band model which takes three major electronic transition bands into consideration.
4.3.7.2 Temperature effect The temperature effect is particularly important for projection displays.54 Due to the thermal effect of the lamp, the temperature of the display panel can reach 50 ◦ C. It is important to know the LC properties at the anticipated operating temperature beforehand. Birefringence n is defined as the difference between the extraordinary and ordinary refractive indices, n = ne − no , and the average refractive index is defined as = (ne + 2no )/3. Based on these two definitions, ne and no can be rewritten as 2 ne = n + n, 3 1 no = n − n 3
(4.11) (4.12)
Liquid crystal displays
69
To describe the temperature-dependent birefringence, the Haller approximation (Equation (4.7)) can be employed when the temperature is not too close to the clearing point: n(T ) = (n)o
T 1− Tc
α ,
(4.13)
where (n)o is the LC birefringence in the crystalline state (or T = 0 K), the exponent α is a material constant and T c is the clearing temperature of the LC material under investigation. The average refractive index decreases linearly with increasing temperature as55 n = A − BT ,
(4.14)
because the LC density decreases with increasing temperature. By substituting Equations (4.14) and (4.13) back into Equations (4.11) and (4.12), the four-parameter model for describing the temperature dependence of the LC refractive indices is given as56 α T 2(n)o 1− , 3 Tc α T (n)o 1− no (T ) ≈ A − BT + 3 Tc ne (T ) ≈ A − BT +
(4.15) (4.16)
Refractive Indices
The parameters [A, B] and [(n)o , α] can be obtained separately by two-stage fittings. To obtain [A, B], one can fit the average refractive index = (ne + 2no )/3 as a function of temperature using Equation (4.14). To find [(n)o , α], one can fit the birefringence data as a function of temperature using Equation (4.13). Therefore, these two sets of parameters can be obtained separately from the same set of refractive indices but at different forms. Figure 4.6 shows a plot of the temperature-dependent refractive indices of 5CB at λ = 546, 589 and 633 nm. As the temperature increases, ne decreases, but no gradually increases. In the isotropic state, ne = no and the refractive index decreases linearly as the temperature increases. This correlates with the density effect.
1.76 1.74 1.72 1.70 1.68 1.66 1.64 1.62 1.60 1.58 1.56 1.54 1.52 280
ne
5CB 546 nm 589 nm 633 nm
no 290
300 310 320 Temperature (K)
330
Figure 4.6 Temperature-dependent refractive indices of 5CB at λ = 546, 589 and 633 nm. (Reprinted with permission from J. Li and S.T. Wu, J. Appl. Phys., 96, 19–24. Copyright (2004), American Institute of Physics)
70
Introduction to Flat Panel Displays
4.4 Liquid crystal alignment The LC mixtures developed for displays are anisotropic liquids at room temperature. In order to obtain useful electro-optic properties, they need to be confined and aligned between two substrates, glass or plastic. The substrates are usually coated with a thin (∼80–100 nm) alignment layer in order to align the LC molecules. Three commonly employed alignment methods are mechanically buffed polyimide (PI), ion beam etched PI, and evaporated SiOx .57 PI has been commonly used in large-screen displays because of its fabrication simplicity. The inorganic SiOx layer is widely used in projection displays because it is robust and can withstand high-intensity illumination from a high-power lamp. Figure 4.7 shows rubbing- and ion beam etching-induced LC alignment. In mechanical buffing, a nylon cloth is used. The LC directors tilt up along the rubbing direction, as shown in Figure 4.7(a).58 The pretilt angle depends on the PI material and rubbing strength. For a typical rubbing process, the pretilt angle is around 3–5◦ . In contrast, in ion beam etching the LC directors tilt up in the opposite direction to the etching beam, as shown in Figure 4.7(b). An important function of the pretilt angle is to guide the LC directors to have uniform reorientation corresponding to an external field.59 Without this pretilt angle, the LC molecules may not be reoriented in the same direction, especially during relaxation processes, which causes light scattering and slow response. Several LC cell geometries have been developed depending on the rubbing direction and pretilt angle of each substrate. When two substrates are rubbed in antiparallel directions so that their pretilt angles (typically ∼3–5◦ ) are in opposite directions, a homogeneous cell is formed. Similar to a homogeneous cell, if the PI has a large pretilt angle, say 85–90◦ , a homeotropic (also known as vertical alignment) cell is formed.60 When two substrates are rubbed in orthogonal directions, a TN cell is formed.61 If a chiral dopant is added to the LC material and the twist angle is larger than 90◦ , then the cell is called a super twisted nematic (STN) cell.62 Some polymer-dispersed LCs do not need surface treatment. Due to phase separation, they form LC droplets.63 Four types of LC alignments are widely used in the display industry. They are 90◦ TN, homogeneous cell manifested in IPS or FFS, MVA and bend cell. Besides LC alignment, the electrode configuration also
z y Rubbing Direction
Pretilt
θp x
(a) Ion Beam
z
a y Pretilt
θp
x
(b) Figure 4.7 (a) Rubbing- and (b) ion beam etching-induced pretilt angle on a PI-coated substrate.
Liquid crystal displays
71
affects the performance of an LCD, especially its viewing angle. In the following sections, we describe these four widely employed LC alignments.
4.5 Homogeneous cell Homogeneous alignment has been used in IPS, FFS, dual cell gap transflective LCDs and phase-only modulations, depending on the electric field direction. In IPS and FFS cells, the electric field is in the transverse direction, whereas in transflective LCDs and phase-only modulators, the electric field is in the longitudinal direction. The in-plane molecular reorientation leads to a wide viewing angle and is suitable for direct-view displays. In contrast, the out-of-plane reorientation induced by the longitudinal field has a narrow viewing angle. It has been used in phase-only modulations, such as spatial light modulators and laser beam steering.64 To widen the viewing angle for direct-view displays, a special compensation film needs to be used. In this section, we describe the basic electro-optics of a homogeneous cell using a longitudinal electric field. We will defer discussion of the wide-view approach (IPS and FFS cells) to later sections. In a homogeneous cell (also known as an electrically controlled birefringence (ECB) cell), the top and bottom substrates are rubbed in antiparallel directions (x and −x) to generate opposite pretilt angles θ p , as depicted in Figure 4.8. If the rubbings are in parallel and pretilt angles are in the same direction, then a -cell is formed.65 We discuss the electro-optics of -cells in Section 4.10. √ In Figure 4.8(b), when the applied voltage exceeds the Freederisckz transition threshold66 (Vth = π K11 /ε0 ε), the LC directors will be reoriented along the electric field direction, provided that the LC employed has a positive dielectric anisotropy and the anchoring energy is strong. The existence of this threshold voltage is because the electric field-induced torque has to overcome the restoring elastic torque in order to reorient the LC directors. If the anchoring energy is weak, then the threshold voltage will decrease and the overall response time (rise + decay) becomes slower.67 As shown in Figure 4.8(b), the applied electric field is in the longitudinal direction (z-axis), and therefore the LC directors tilt out of the plane. The monolayers are virtually anchored by the alignment layers on the substrate surfaces. The pretilt angle (θ p ≈ 3◦ ) gives a predetermined direction for the LC directors to follow. Without this pretilt angle, the LC directors do not ‘know’ which initial direction to rotate and will cause light scattering and slow response time. If the LC has a negative ε, then the electric field cannot reorient the LC molecules. To drive the negative ε LC, a transverse electric field (or fringe field) has to be used. This mechanism is discussed in Sections 4.7 and 4.8. For a given voltage, from Figure 4.8(b), the LC director distribution inside the cell is actually not uniform, but is symmetric along the middle layer of the cell, assuming the pretilt angles on both surfaces are equal. Figure 4.9 shows the distribution of the LC directors at different voltages, normalized to V th . –X
–X
θp
θp
E
θp
θp
X
X
(a) Figure 4.8
(b)
LC director profile in a homogeneous cell: (a) V = 0; (b) V > V th .
72
Introduction to Flat Panel Displays
90
1 0 Vth
80
3 Vth
Tilt Angle, Deg.
70 60
4 Vth
2 Vth
50 40 30
1 Vth
20 10 0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 Z/d Figure 4.9
LC director distribution profile of a homogeneous cell under different voltages.
Below V th , the LC molecules are not reoriented. As the voltage increases to 3–4 times the threshold, the majority of the LCs have been reoriented. The middle layers (z/d = 0.5) have already been rotated by more than 80◦ . Further increasing the applied voltage to 10V th only reorients the surface LC layers; the bulk layers have already been completely reoriented.
4.5.1 Phase retardation effect When a plane wave is incident normally to a uniaxial LC sandwiched between two polarizers, the outgoing beam will experience a phase retardation (δ) due to the different propagation speed of the extraordinary and ordinary waves inside the media: δ=
2π dn 2πd (ne − no ) = , λ λ
(4.17)
where d is the cell gap, n is the birefringence and λ is the wavelength. When a homogeneous cell is sandwiched between two polarizers, the normalized light transmittance is governed by the equation T = cos2 χ − sin 2β sin 2(β − χ) sin2 (δ/2),
(4.18)
where χ is the angle between polarizer and analyzer, β is the angle between polarizer and LC directors and δ = 2dn/λ is the phase retardation. For the simplest case where β = 45◦ and the two polarizers are crossed (χ = 90◦ ), the normalized light transmittance is simplified to T⊥ = sin2 (δ/2).
(4.19)
For a homogeneous cell, the effective phase retardation depends on the wavelength, temperature and applied voltage. As the voltage increases, the effective birefringence and the phase retardation decrease. In a high-voltage regime where V V th , virtually all the LC directors in the bulk are aligned nearly normal to the substrates except for the boundary layers. The remaining phase retardation is small owing to the vanishing birefringence. However, these boundary layers are quite difficult to completely reorient using the external field. This implies that a good dark state of a homogeneous cell is difficult to achieve without a phase compensation film.
Liquid crystal displays
73
100
Transmittance, T⊥ (%)
Transmittance, T⊥ (%)
100 80 R
G B
60 40 20 0
0
1
2
3 4 Voltage, Vrms (a)
5
6
7
80 R
G B
60 40 20 0
0
1
2
3 4 Voltage, Vrms (b)
5
6
7
Figure 4.10 (a) V –T curve of a homogeneous cell with dn = 275 nm, and (b) a uniaxial film-compensated homogeneous cell (cell dn = 368 nm and film dn = 96 nm but oriented to be orthogonal to the LC directors).
4.5.2 Voltage-dependent transmittance Figure 4.10 depicts the voltage-dependent light transmittance of a homogeneous cell under crossed polarizers. Here the optical losses from polarizers, substrate surfaces and ITO are all neglected. The LC cell parameters are as follows: LC mixture, MLC-6297-000; elastic constants K 11 = 13.4 pN, K 22 = 6.0 pN, K 33 = 19.0 pN; dielectric constants ε = 10.5 and ε = 6.9; cell gap d ≈ 4.3 m; n = 0.125, 0.127 and 0.129 for R = 650 nm, G = 550 nm and B = 450 nm; pretilt angle θ p ≈ 2◦ . From Figure 4.10(a), the bright state intensity variation among RGB colors is around 10 %. Due to the small dn value (275 nm which is ∼λ/2), the transmittance in the high-voltage region is decreased monotonically, but it is difficult to get a common dark state for the RGB colors. This is because the residual phase retardation of the boundary layers as shown in Figure 4.9 is additive (parallel) rather than subtractive (orthogonal). One method to obtain a common dark state for a homogeneous cell is to add a uniaxial compensation film (discussed in Section 4.7.4) whose optical axis is orthogonal to that of the LC directors. Figure 4.10(b) shows the results for a uniaxial film-compensated homogeneous cell whose dn = 368 nm and compensation film’s dn = −96 nm. The dark state is reduced to ∼4 Vrms . This dark state voltage is dependent on the dn values of the LC cell and compensation film employed. For a smaller dn, the dark state voltage will be higher and its width (V for maintaining a good dark state) will be broader.
4.6 Twisted nematic The 90◦ TN cell has been used extensively for small-size displays and notebook computers where viewing angle is not too critical. Figure 4.11 shows the LC director configurations of the normally white (NW) TN cell in the voltage-off and voltage-on states. The top LC alignment is parallel to the optical axis of the top polarizer while the bottom LC directors are rotated 90◦ and parallel to the optical axis of the bottom analyzer. When dn of the LC layer satisfies the Gooch–Tarry first minimum condition,68 the incoming linearly polarized light will follow adiabatically the molecular twist and will transmit through the crossed analyzer. In the voltage-on (∼5 Vrms ) state, the LC directors are reoriented perpendicular to the substrates, except the boundary layers. The incoming light experiences little phase change and is absorbed by the analyzer, resulting in a dark state. The beauty of the TN cell is that the boundary layers are orthogonal so that their residual phase in the high-voltage state compensates for each other. As a result, the dark state occurs at a relatively low voltage. If the twist angle departs from 90◦ , the dark state will be degraded and operation voltage will increase accordingly.
74
Introduction to Flat Panel Displays
POLARIZER
GLASS LC
GLASS
POLARIZER Figure 4.11 LC and polarizer configurations of a 90◦ TN cell. Left: V = 0 (voltage-off state); right: V V th (voltage-on state).
4.6.1 Optical transmittance To compare different operating modes, let us focus on the normalized transmittance by ignoring the optical losses from polarizers, ITO (n ≈ 1.8) layers and the interface reflections from substrates. The normalized transmittance (T ⊥ without any voltage) of a TN cell can be described by the following Jones matrices as T ⊥ = |M|2 :69 M = cos β
cos φ sin β sin φ
cos X − i sin X − sin φ 2 X cos φ sin X −φ X
sin X − sin β X . sin X cos β cos X + i 2 X φ
(4.20)
Here β is the angle between the polarization axis and the front LC director, φ is the twist angle, X = φ 2 + (/2)2 , Γ = 2dn/λ and d is the cell gap. By simple algebraic calculations, the following analytical expression for |M|2 is derived: |M|2 = T⊥ =
φ cos φ sin X − sin φ cos X X
2
+
sin X 2 X
2 sin2 (φ − 2β).
(4.21)
Equation (4.21) is a general formula describing the light transmittance of a TN cell (without voltage) as a function of twist angle, beta angle and dn/λ. For a 90◦ TN cell, φ = /2 and Equation (4.21) is simplified to 2 T⊥ = cos2 X + cos 2β sin2 X. (4.22) 2X Equation (4.22) has a special solution which is cos2 X = 1. When cos2 X = ±1 (i.e. X = mπ, where m is an integer), then sin X = 0 and the second term in Equation (4.22) vanishes. Therefore, T ⊥ = 1, independent of β. By setting X = mπ and knowing that Γ = 2πdn/λ, the Gooch–Tarry condition is found as
dn 1 = m2 − . (4.23) λ 4
Liquid crystal displays
75
√ For the lowest order m = 1, dn/λ = 3/2. This √ is the Gooch–Tarry first minimum condition for the 90◦ TN cell. For the second order m = 2, dn/λ = 15/2. The second minimum condition is used only for low-end displays such as wristwatches and calculators because a large cell gap is easier to fabricate and the cyanobiphenyl LCs are less expensive. For notebook TFT LCDs, the first minimum is preferred because it requires a thinner cell gap which leads to a faster response time and wider viewing angle. Figure 4.12 depicts the normalized voltage-dependent light transmittance (T ⊥ ) of a 90◦ TN cell at three primary wavelengths: R = 650 nm, G = 550 nm and B = 450 nm. Since the human eye has the greatest sensitivity at green, we normally optimize the cell design at λ ≈ 550 nm. From Equation (4.22), the first T ⊥ = 1 occurs at dn ≈ 480 nm. For a 5 m cell gap, the required birefringence is n ≈ 0.096. From Figure 4.12, the wavelength effect on the transmittance at V = 0 is within 8 %. Therefore, the TN cell can be treated as an ‘achromatic’ half-wave plate. The response time of a TN LCD depends on the cell gap and γ1 /K 22 for the LC mixture employed. For a 5 m cell gap, the optical response time is ∼30–40 ms. At V = 5 Vrms , the contrast ratio (CR) reaches ∼500:1. These performances, although not perfect, are acceptable for notebook computers. A major drawback of the TN cell is its narrow viewing angle and grayscale inversion originating from the LC directors tilting out of the plane. Because of this molecular tilt, the viewing angle in the vertical direction is narrow and asymmetric, and has grayscale inversion.70
4.6.2 Viewing angle In a TN cell, the applied electric field is in the longitudinal direction. As a result, the bulk LC directors tilt along the electric field direction, as shown in Figure 4.13(a). This out-of-plane orientation leads to an asymmetric viewing angle in the vertical viewing direction, as illustrated in Figure 4.13(b). The cross-section of the refractive index ellipsoid viewed from this top-down direction is quite asymmetric. Therefore, its light transmittance through crossed polarizers would be different leading to an asymmetric viewing angle. In contrast, the viewing angle in the horizontal direction is wider and more symmetric. Figure 4.14 shows the viewing angle characteristics of a TN LCD. The middle photo represents the normal view; the right and left are horizontal, and top and bottom are vertical views. The horizontal view is indeed fairly symmetric, but the vertical view is much worse. A severe grayscale inversion occurs in the bottom-up direction (bottom photo). To extend TN LCDs to large-screen monitors or TVs, the viewing angle has to be improved. A convenient way to widen the viewing angle of a TN LCD is to use phase compensation films instead 100 450 nm 550 nm 650 nm
Transmittance (%)
80 60 40 20 0 0
1
2 3 Voltage (Vrms )
4
5
Figure 4.12 Voltage-dependent transmittance of a normally white 90◦ TN cell (dn = 480 nm).
76
Introduction to Flat Panel Displays
TN Cell
Vertical view
(a)
(b)
Figure 4.13 (a) LC tilt angles of a TN cell in a voltage-on state, and (b) viewing disparity from vertical direction.
Figure 4.14 Viewing angle of a TN LCD. The middle image is a normal view. Right and left: horizontal view. Top and bottom: vertical view. Grayscale inversion occurs at the bottom-up viewing direction. (Courtesy of Dr Y. Saitoh of Fuji Film)
of implementing multidomain structures.71 The films can be laminated to the inner side of the polarizers. However, due to the asymmetric view of the TN cell, the required compensation film also needs asymmetric phase retardation in the vertical direction.
4.6.3 Film-compensated TN cells In the on-state of a TN cell, the LC directors in the upper half are reoriented along the rubbing direction with almost no twist and the lower half has a similar structure with the director plane orthogonal to that of the upper half. Thus, a uniform phase compensation film, such as a uniaxial A-plate, cannot
Liquid crystal displays
77
Air surface
PDM layer (hybrid alignment) High Degree of Randomness. This structure differs from the ordinary optic materials.
Alignment layer
TAC substrate
OR
OR
OR
OR OR
OR
O
R=
OC
O
M
(M: cross-linkable group) Figure 4.15 Structure of a wide-view film and the discotic compound employed. PDM represents polymerized discotic material.
compensate the upper and lower parts simultaneously. Instead, a pair of wide-view films needs to be used separately on both sides of the TN LC cell in order to compensate each of the half layers. Fuji Photo has developed discotic LC films for widening the viewing angle of TN cells.72 The molecular structures of the wide-view discotic material are shown in Figure 4.15. A discotic material (triphenylene derivatives) is coated on an alignment layer on a triacetyl cellulose (TAC) substrate. The discotic material has a hybrid alignment structure and three important features. (1) It has -electrons spread in a disc-like shape, which gives a high birefringence. (2) It takes on discotic nematic (ND ) phase at lower temperature than the temperature at which the TAC substrate starts to deform. This feature enables a uniform and monodomain film in a wide area without defects. (3) It has cross-linkable groups at all six side chains to make the obtained film durable. When heated, the discotic material takes on the ND phase. The discotic material right next to the alignment layer has a high degree of randomness. And in the vicinity of the alignment layer, the discotic molecules tend to align with the molecular plane almost parallel to the alignment layer surface and have pretilt angle of a few degrees in the rubbing direction of the alignment layer surface. In the vicinity of the air surface the discotic molecules tend to align with the molecular plane almost perpendicular to the air surface. With the pinned alignment on both sides, the discotic material exhibits a hybrid alignment structure in the ND phase. When cured by UV light, the discotic material is polymerized and the hybrid alignment structure of the polymerized discotic material (PDM) layer is fixed even after it is cooled to room temperature. Each film has a hybrid alignment structure in which the director continuously changes in the PDM layer
78
Introduction to Flat Panel Displays
thickness direction without twist while the direction of each discotic molecule fluctuates. This hybrid alignment structure consists of splay and bend deformations. The wide-view Fuji film markedly improves the viewing angle of TN LCDs without losing any light transmittance or degrading image quality. A TN LCD with ∼80◦ viewing cone at CR > 10:1 has been demonstrated.73 A significant advantage of film-compensated TN LCDs is that no change in the panel process is required because the conventional polarizer is simply replaced by a new polarizer laminated with the compensation film. The discotic film is also cost effective. These features enable TN to penetrate into the larger size LCD market segment, say 20- to 26-inch diagonal. However, grayscale inversion can still be observed in large-panel film-compensated TN LCDs which ultimately limits their competitiveness with IPS and MVA.
4.7 In-plane switching To overcome the narrow viewing angle of TN LCDs, IPS was proposed in the 1970s74, 75 and implemented in TFT LCDs in the 1990s.76, 77 The interdigitated electrodes are in the same substrate and the top substrate has no electrode such that the generated fringing fields are in the transverse plane.As shown in Figure 4.16, the LC directors are rotated in the plane which results in a wide viewing angle.
4.7.1 Device structure In an IPS system, the interdigitated electrodes are fabricated on the same substrate and LC directors are aligned homogeneously with a rubbing angle of ∼10◦ with respect to the striped electrodes. The transmission axis of the polarizer can be set to be parallel (e-mode) or perpendicular (o-mode) to the LC directors while the analyzer is crossed to the polarizer. The in-plane electric fields induced by the electrodes twist the LC directors, as shown in Figure 4.17. The incoming linearly polarized light from the polarizer experiences a phase retardation so that its polarization state is changed, resulting in light transmission through the crossed analyzer. However, due to the strong vertical electric field existing above the electrode surface, the LC directors in these regions mainly tilt rather than twist. As a result, the transmittance above the electrodes is greatly reduced. Overall, the conventional IPS mode has a light efficiency ∼75 % of that of a TN LCD, when a positive ε LC material is used. Although using a negative ε LC in the IPS mode could enhance the light efficiency to ∼85 %, the required on-state driving voltage is too high (>10 Vrms ).78 This is because negative ε LC materials tend to have a smaller dielectric anisotropy. As a result, both threshold and on-state voltages are increased. For low power consumption, the preferred operating voltage is less than 5 Vrms . IPS
LC Director E
Electrode (a)
(b)
Figure 4.16 (a) IPS electrodes, electric fields and LC director orientations in a voltage-on state; (b) viewing angle characteristics of in-plane rotation of LC molecules.
Liquid crystal displays
79
G = 8 μm
W= 4 μm
Figure 4.17 LC director distribution, electric field profile (dashed lines) and corresponding light transmission of an IPS cell at V = 5 Vrms . LC, MLC-6686; ε = 10; electrode width W = 4 m; gap G = 8 m; cell gap d = 3.6 m.
4.7.2 Voltage-dependent transmittance Figure 4.18 depicts the voltage-dependent light transmittance at RGB wavelengths corresponding to the IPS device structure shown in Figure 4.17. The threshold is ∼1.5 Vrms and maximum transmittance occurs at ∼5 Vrms for all three wavelengths. Due to absorption, the maximum transmittance of the two polarizers (without LC cell) is 35.4, 33.7 and 31.4 % for R, G and B wavelengths, respectively.
4.7.3 Viewing angle Figure 4.19 shows the calculated iso-contrast contours of an IPS LCD without any compensation films. The 10:1 iso-contrast extends beyond a 60◦ viewing cone. For mobile displays, this viewing angle is adequate because the display is mainly viewed by a single user. However, for large-size displays such as
Transmittance
0.3
450 nm 550 nm 650 nm
0.2
0.1
0.0
0
1
2
3 4 Voltage (Vrms)
5
6
Figure 4.18 Voltage-dependent light transmittance of an IPS LCD. The device structure is shown in Figure 4.17.
80
Introduction to Flat Panel Displays
90 60
120 80 60
10
150
40
0
10
30 10
20 100
180
0
10
0 100
10
210
330 10
240
300 270
Figure 4.19
Simulated iso-contrast contours of an IPS LCD without any compensation films.
TVs this viewing angle is still inadequate and compensation films are required. A properly compensated IPS LCD can have omni-contrast ratio >50:1 beyond an 80◦ viewing cone.
4.7.4 Phase compensation films In an IPS mode, the LC directors are aligned parallel (or crossed) to the optical axis of the entrance polarizer. At normal incidence, the LC layer in the voltage-off state does not modulate the polarization state of the incident linearly polarized light from the entrance polarizer. As a result, a good dark state is achieved because the outgoing linearly polarized light is absorbed by the crossed analyzer. However, at oblique angles the incident light leaks through the crossed polarizers, especially at the bisectors. This light leakage arises from two factors. First, the absorption axes of the crossed polarizers are no longer orthogonal to each other under off-axis oblique view; thus, the extinction ratio of the crossed polarizers decreases and light leakage occurs. Second, the obliquely incident linearly polarized light sees the birefringence of the LC layer and becomes elliptically polarized after traversing through the cell. Consequently, the analyzer cannot completely absorb the elliptically polarized light leading to light leakage off-axis. This light leakage in the dark state degrades the contrast ratio and thereby degrades the viewing angle. To suppress the light leakage at oblique angles and further widen the viewing angle, several phase compensation schemes using uniaxial films79 81 and biaxial films82 84 have been proposed. Table 4.2 lists some commercially available compensation films, classified by their refractive indices. Different LC modes need different types of compensation films in order to obtain a satisfactory compensation effect. For example, IPS mode needs a biaxial film with nx > nz > ny ,85 while VA and optically compensated bend modes need a compensation film with nx > ny > nz .86 A uniaxial film is an anisotropic birefringent film with only one optical axis. From the viewpoint of optical axis orientation, uniaxial films can be classified into A-films and C-films. The optical axis of an A-film resides in its surface, i.e. XY plane, while a C-film’s optical axis is perpendicular (Z-axis) to the film surface. The commonly used λ/2 and λ/4 plates are examples of A-plates. They are made of stretched polymers. Their optical axis is mechanically stretched along either X- or Y -axis. In contrast, the C-plate is isotropic in the XY plane but its optical anisotropy occurs in the Z-axis.
Liquid crystal displays
Table 4.2
81
Different types of compensation films used for wide-view LCDs.
A-plate
nx < ny =nz (Nz = 0)
C-plate
nx = ny > nz
A-plate
nx > ny = nz (Nz = 1)
C-plate
nz > nx = ny
X-Z optical axis
nx > ny > nz (Nz > 1)
X-Y optical axis
nx > nz > ny (0 < Nz < 1)
Negative Uni-axial Positive Anisotropic Negative Bi-axial nx = ny = nz z
Positive
nz > nx > ny (Nz < 0)
y x
Y-Z optical axis
Negative Oblique
NWF Positive
Nz =
nx – nz nx – ny
c 2005 IEEE. Source: R. Lu et al., J. Display Technol., 1, 3–14.
4.8 Fringe field switching To overcome the dead zones that occur above IPS electrodes, FFS has been developed.87 The basic structure of FFS is similar to IPS except for the much smaller electrode gap (G ≈ 0–1 m). In the IPS mode, the gap between electrodes is larger than the cell gap (d). The horizontal component of the electric field is dominant between electrodes. However, in the FFS mode where G < d, a fringing field exists above the electrodes. The fringing fields are able to twist the LC directors above the electrodes. Therefore, light transmittance is improved. However, to fabricate electrode gaps of around 1 m is technically challenging. Current photolithographic precision is about 3–4 m. Therefore, a modified FFS electrode structure, as shown in Figure 4.20, has been developed. The modified FFS structure has two layers of electrodes: interdigitated striped pixel electrodes and a planar common ITO electrode.88, 89 Strong electric fields generated between pixel electrodes and the common electrode twist the LC directors. These electric fields are strong even on top of the pixel electrodes. From Figure 4.20, it can be seen that there is no transmission dip above the electrode areas. Similar to IPS, in FFS mode both positive and negative ε LCs can be used.90 A FFS mode using a negative ε material can achieve a transmittance >90 % of that of a TN cell, depending on the electrode width and gap. The idea of using a positive ε LC material in the FFS mode for achieving high transmittance (>85 % of TN mode) has also been attempted. A positive LC usually exhibits a larger ε and lower viscosity than the corresponding negative ε LC because its polar group is along the principal molecular axis. Figure 4.21 depicts the voltage-dependent light transmittance at RGB wavelengths corresponding to the FFS device structure shown in Figure 4.20. The threshold is ∼1 Vrms and maximum transmittance occurs at ∼4.5 Vrms for all three wavelengths. Due to absorption, the maximum transmittance of the two polarizers (without LC cell) is 35.4, 33.7 and 31.4 % for R, G and B wavelengths, respectively. The normalized transmittance for green is ∼96 %.
82
Introduction to Flat Panel Displays
Transmittance
Top substrate
LC
E
Pixel Common electrode Figure 4.20 Device structure, simulated on-state LC director distribution, electric field directions (dashed lines) and corresponding light transmittance (top) of a FFS cell. Electrode width w = 3 m and electrode gap G = 4.5 m. LC, MLC-6608; ε = −4.2; = 0.083; cell gap d = 4 m.
0.4
450 nm 550 nm 650 nm
Transmittance
0.3
0.2
0.1
0.0 0
1
2
3 4 Voltage (Vrms)
5
6
Figure 4.21 Voltage-dependent light transmittance of the FFS LCD shown in Figure 4.20.
Liquid crystal displays
83
4.9 Vertical alignment Vertical alignment (VA), also called homeotropic alignment,91 is also commonly used in direct-view transmissive displays and reflective projection displays. VA exhibits the highest contrast ratio among all the LC modes developed; moreover, its contrast ratio is insensitive to the incident light wavelength, LC layer thickness and operating temperature. Both projection92, 93 and direct-view displays using VA cells94, 95 have been demonstrated. For projection displays, single-domain VA is employed in reflective LCoS. But for direct-view LCDs, MVA structures have to be used in order to obtain a wide viewing angle. Among all the wide-view LCDs, VA is the only one which does not require a rubbing process; IPS, FFS and TN all require mechanical buffing.
4.9.1 Voltage-dependent transmittance Figure 4.22 shows the voltage-dependent optical transmittance of a VA cell with dn = 350 nm between crossed polarizers. Here, a single-domain VA cell employing Merck high-resistivity MLC-6608 LC mixture is considered. Some physical properties of MLC-6608 are as follows: ne = 1.562, no = 1.479 (at λ = 546 nm and T = 20 ◦ C);96 clearing temperature T c = 90 ◦ C; dielectric anisotropy ε = −4.2; and rotational viscosity γ 1 = 186 mPa s at 20 ◦ C. In principle, to obtain 100 % transmittance for a transmissive VA cell only requires dn ≈ λ/2. Since the human eye is most sensitive in green (λ ≈ 550 nm), the required dn is around 275 nm. However, this is the minimum dn value required because under such conditions the 100 % transmittance would occur at V V th . Due to the finite voltage swing from TFT (preferred to be below 6 Vrms ), the required dn should be increased to ∼0.6λ, i.e. dn ≈ 330 nm. From Figure 4.22, it can be seen that an excellent dark state is obtained at normal incidence. As the applied voltage exceeds the Freederisckz threshold voltage (V th ≈ 2.1 Vrms ), LC directors are reoriented by the applied electric field resulting in light transmission from the crossed analyzer. From the figure, it can be seen that RGB wavelengths reach their peak at different voltages: blue at ∼4 Vrms and green at ∼6 Vrms . The on-state dispersion is more than that of the dark state. A small light leakage in the dark state would degrade the contrast ratio significantly, but less noticeable in the bright state.
4.9.2 Response time Figure 4.23 shows a single-domain VA LC layer sandwiched between two parallel substrates where z = −d/2 and +d/2 represent the bottom and top substrates, respectively. The z-axis is normal to the
Transmittance (%)
100 80
B
G
R
60 40 20 0
0
1
2
3 4 Voltage, Vrms
5
6
7
Figure 4.22 Voltage-dependent normalized transmittance of a VA cell. LC, MLC-6608; dn = 350 nm; R = 650 nm, G = 550 nm and B = 450 nm.
84
Introduction to Flat Panel Displays
Top substrate Z = d/2 E z
θ
LC
x
Z = –d/2 Bottom substrate Figure 4.23
Schematic of a VA LC cell with a pretilt angle and boundary conditions.
plane of the substrates, and the electric field E is along the z-axis. When the backflow and inertial effects are ignored, the Erickson–Leslie equation for describing the dynamics of LC directors has the following form:97, 98 2 ∂ 2θ ∂θ ∂θ 2 2 (K11 sin θ + K33 cos θ ) 2 + (K33 − K11 ) sin θ cos θ + εo εE 2 sin θ cos θ = −γ1 , (4.24) ∂z ∂z ∂t where γ1 is the LC rotational viscosity, K 11 and K 33 are the splay and bend elastic constants, respectively, ε o εE 2 is the electric field energy density, ε is the LC dielectric anisotropy and θ is the tilt angle defined as the angle between the z-axis and the LC directors. In general, Equation (4.24) can only be solved numerically. However, when the tilt angle is small (sin θ ≈ θ ; small angle approximation)99 and K 33 ≈ K 11 (single elastic constant approximation), the Erickson–Leslie equation can be simplified as K33
d2 θ ∂θ + εo εE 2 θ = −γ1 . ∂z2 ∂t
(4.25)
Equation (4.25) has following general solution: θ = [θs sin(αz) + θm cos(αz)] exp(−t/τ ).
(4.26)
If the VA cell has the same top and bottom substrate treatments, then θs is found to be 0. At a given voltage, θm represents the maximum tilt angle in the center of the LC cell ( θ|z=0 = θm ). If the top and bottom substrates have different alignment conditions, then θs = 0 and both terms in Equation (4.26) have to be considered. Throughout this section, for simplicity we assume the pretilt angles on both substrates are symmetric so that θs = 0. When the pretilt angle θ p is zero and the anchoring energy is strong, the following boundary conditions hold: θz=−d/2,d/2 = θp = 0.
(4.27)
Equations (4.26) and (4.27) lead to the following analytical solutions for the decay (d ) and rise (r ) times: τd = τo = τr =
γ1 d 2 , K33 π 2
γ1 τo = . |εo |ε| E 2 − (π 2 /d 2 )K33 | |(V /Vth )2 − 1|
(4.28)
(4.29)
Liquid crystal displays
85
In Equation (4.28), o is called the free relaxation time, i.e. during the decay process there is no bias √ voltage, and in Equation (4.29) the threshold voltage is defined as Vth = π εo K33 /ε. If the pretilt angle deviates from zero, then we have θz=−d/2,d/2 = θp = 0.
(4.30)
Equation (4.26) should satisfy the boundary conditions described by Equation (4.30) at z = −d/2 and d/2. From Equations (4.24) and (4.30), we find the parameter α has the following form: 2 θp α = cos−1 . (4.31) d θm Based on Equation (4.25), the modified response times that take the pretilt angle effect into consideration are derived as100 γ1 τd∗ = τo∗ = 2 , (4.32) α K33 γ1 ∗ τr = . (4.33) |εo |ε| E 2 − α 2 K33 | In most cases, the maximum tilt angle is much larger than the pretilt angle, i.e. θm θp . Under such a condition, the cos−1 ( ) term in Equation (4.31) can be approximated as θp θp π cos−1 (4.34) ≈ − , θm 2 θm and the rise and decay times are τd∗ = τo∗ =
γ1 γ1 d 2 = , α K33 4K33 [(π/2) − (θp /θm )]2 2
γ1 . τr∗ = εo |ε| E 2 − (4K33 /d 2 )[(π/2) − (θp /θm )]2
(4.35)
(4.36)
Strictly speaking, the LC threshold voltage V th no longer exists if the pretilt angle is nonzero, although the threshold-like behavior in the voltage-dependent transmittance still appears. For simplicity, let us assume the threshold voltage still exists. Under such a condition, Equation (4.36) can be simplified as τo∗ . τr∗ = [V /(1 − 2θp /πθm )Vth ]2 − 1
(4.37)
As expected, Equations (4.35) and (4.37) are reduced to Equations (4.28) and (4.29) when the pretilt angle is zero. Equations (4.35) and (4.37) suggest that the LC response time is also dependent on the ratio of θ p /θ m , where θ m originates from the applied voltage. For the case of small pretilt angle, this term is negligible because θ p /θ m → 0.
4.9.3 Overdrive and undershoot voltage method From Equation (4.29), the rise time depends on the applied voltage (V ), especially near the threshold region. Let us use a normally black VA cell as an example. Typically, the cell is biased at a voltage (V b ) which is slightly below V th in order to reduce the delay time incurred during the rising period and to keep a high contrast ratio. For some intermediate gray levels, the applied voltage is only slightly above V th . For such a circumstance, the rise time would be very slow. To overcome the slow rise time, we can apply a high voltage for a short period and then hold the transmittance at the desired gray level, as shown in
Introduction to Flat Panel Displays
Transmittance
86
1 0.8 0.6 0.4 0.2 0
Voltage
6 4 2 0
0
20
40
60
80 100 120 140 160 180 200 Time, ms
Figure 4.24 Overdrive and undershoot voltage method for speeding up LC rise and decay times. Top: optical response; bottom: corresponding voltage waveforms. The dashed lines represent a normal driving and the solid lines are with overdrive and undershoot voltages.
Figure 4.24. This is the so-called overdrive voltage method.101 Meanwhile, during the decay period, the voltage is turned off for a short period and then a small holding voltage is applied to keep the LC at the desired gray level. This is the undershoot effect.102 With voltage overdrive and undershoot, the LC response time can be reduced by about two to three times, depending on the applied voltage.
4.9.4 Multidomain vertical alignment Single-domain VA has a narrow viewing angle and can only be used in projection displays. For direct-view displays, four domains are required in order to eliminate grayscale inversion and widen the viewing angle. Assume each domain is located in each quadrant, as shown in Figure 4.25. Here, P and A represent the optical axes of the polarizer and analyzer, respectively. In order to obtain maximum transmittance, the LC directors in each domain should be oriented at 45◦ with respect to the polarizer’s axis. Fujitsu has developed protrusion-type MVA103, 104 and Samsung has developed patterned vertical alignment (PVA)105, 106 using slits to generate fringing fields. The operating mechanisms are alike, but PVA does not require any physical protrusions so that its contrast ratio is higher. Figure 4.26 shows a schematic of the structure of PVA. In the voltage-off state, as shown in Figure 4.26(a), the LC directors are aligned perpendicular to the substrates. Because there is no physical protrusion, a very good dark state can be obtained. In a voltage-on state, as depicted in Figure 4.26(b), the fringe fields from the top and bottom P 1
2 A
3 Figure 4.25
4
LC orientation in a four-domain structure. P, polarizer; A, analyzer.
Liquid crystal displays
87
(a)
(b)
Figure 4.26 (a) LC directors of PVA at V = 0, and (b) LC directors of PVA at a voltage-on state. The fringe fields generated by the top and bottom slits create two opposite domains in this cross-section. When zigzag electrodes are used, four domains are generated.
Normalized CR
MVA
IPS
Polar angle, Deg. Figure 4.27 Comparison of normalized contrast ratio of MVA and IPS.
slits create two opposite domains as highlighted in the dashed circles. By using zigzag electrodes with 90◦ tilt angle, four-domain VA can be formed. With the combination of A-plate and C-plate, both MVA and PVA can achieve a contrast ratio of more than 50:1 over an 85◦ viewing cone. As shown in Figure 4.26, PVA has no pretilt angle. The four domains are induced by the fringe electric fields. The response time, especially rise time, is relatively slow. In MVA, physical protrusions are used to provide an initial pretilt direction for forming four domains. The protrusions not only reduce the aperture ratio but also cause light leakage in the dark state because the LCs on the edges of the protrusions are tilted so that they exhibit birefringence. It would be desirable to eliminate protrusions for MVA and create a pretilt angle in each domain for both MVA and PVA to guide the LC reorientation direction. Based on this concept, the surface polymer sustained alignment (PSA) technique has been developed.107 A very small percentage (∼0.2 wt%) of reactive mesogen monomer and photoinitiator are mixed in a negative ε LC host and injected into an LCD panel. While a voltage is applied to generate four domains, UV light is used to cure the monomers. As a result, the monomers are adsorbed onto the surfaces. These cured polymers, although in low density, will provide a pretilt angle within each domain to guide the LC reorientation. Thus, the rise time is reduced by nearly two times while the decay time remains more or less unchanged.108 Generally speaking, PVA (and MVA) shows a higher contrast ratio than IPS (and FFS), but only within the 20◦ viewing cone, as shown in Figure 4.27. Beyond this region, the contrast ratios decrease more quickly than that of IPS. Meanwhile, IPS has a smaller color shift (the color change at oblique angle as compared to the normal) than PVA.109 PVA does not require rubbing, but IPS does. Therefore, each technology has its own advantages and shortcomings.
88
Introduction to Flat Panel Displays
4.10 Optically compensated bend cell The optically compensated bend (OCB) mode utilizes a voltage-biased -cell compensated with phase retardation films. Its major advantages are two fold: (1) fast response time and (2) symmetric and wide view angle. In a -cell,110 the pretilt angle in the alignment surfaces is in the opposite direction, as shown in Figure 4.28. The opposite pretilt angle exhibits two special features: (1) its viewing angle is symmetric and (2) its bend director profile eliminates the backflow effect and, therefore, results in a fast response time.
4.10.1 Voltage-dependent transmittance Figure 4.29 shows the voltage-dependent transmittance curves of a uniaxial film-compensated bend cell. To make the splay-to-bend transition, a critical voltage (V c ≈ 1.0–1.5 V) is biased to the -cell. Typically, the cell gap is around 6 m and pretilt angle is 7–10◦ .111 By adjusting the dn value of the compensation film, both normally white and normally black modes can be achieved.112 Figure 4.29 shows the V –T curves of a normally white OCB cell at three primary wavelengths (R = 650, G = 550, B = 450 nm). The following parameters are used for simulations: LC dn = 436 nm, ε = 10, uniaxial A-film dn = 53.3 nm and pretilt angle = 7◦ . In reality, the uniaxial A-film should be replaced by a biaxial film in order to widen the viewing angle. From Figure 4.29, a common dark state for RGB wavelengths appears at ∼4.5 Vrms . Wavelength dispersion is a serious concern for any birefringence mode. To solve this problem, multiple cell gaps have to be used, i.e. the dn/λ value for all three primary wavelengths should be equal. For example, if d = 6 m is used for the green pixels, then the gaps for red and blue pixels should be
θp π cell Figure 4.28
Splay
Bend
LC director transitions from -cell to splay and bend cells. Pretilt angle θp ≈ 7◦ .
Normalized Transmittance
1.0 450 nm 550 nm 650 nm
0.8 0.6 0.4 0.2 0.0 0
1
2 3 4 Voltage, Vrms
5
6
Figure 4.29 Voltage-dependent transmittance curves of a -cell. dn = 436 nm, uniaxial film dn = 53.3 nm and its optical axis is perpendicular to that of the LC cell.
Liquid crystal displays
89
7.1 and 4.9 m, respectively. Here, the wavelength dispersion of the LC material is neglected.113 Once these conditions are satisfied, the V –T curves for R and B will overlap with that of G (dark line). The fast response time of the OCB cell originates from three factors: bias voltage effect (also known as surface mode),114 flow effect and half-cell switching. The switching time between gray levels is less than 3 ms. A fast response time is particularly important for LCD TV applications, especially at cold ambient. For other LCD modes, such as TN, MVA and IPS, flow in the LC layer slows the rotational relaxation process of directors when the applied voltage is changed. For the -cell, in contrast, there is no conflict of torque exerted by flow and relaxation process of directors. The intrinsic wide viewing angle is due to the self-compensating structure. The retardation value stays almost the same even when the incident angle is changed in the director plane. However, retardation is not self-compensated for incidence out of the director plane. In addition, the on-axis CR of the -cell is low because of residual retardation even at a high applied voltage. To obtain a high on-axis CR and a wide viewing angle, an optical compensation film is required. To obtain the bend alignment structure of the -cell, a voltage above the splay-to-bend transition voltage must be applied. The transition from splay to bend takes time, typically of the order of tens of seconds. The transition should be made faster than, say, one second.
4.10.2 Compensation films for OCB To obtain a comparable viewing angle with VA and IPS, OCB requires more sophisticated optical compensation based on a discotic material.115 Figure 4.30 shows the compensation schemes for a normally white OCB mode. The fundamental idea is similar to that for TN. The retardation matching
Transmission axis Slow axis
Polarizer Biaxial film OCB-WV DLC
π cell
DLC OCB-WV Slow axis Transmission axis
Biaxial film Polarizer
Figure 4.30 Simplified model of optical compensation for a -cell combined with Fuji OCB films. (Source: H. Mori, c 2005 IEEE) Journal of Display Technology, 1, 179–186.
90
Introduction to Flat Panel Displays
between the cell and the optical compensation film is especially important for the OCB mode partially because the black-state of the normally white OCB cell has a finite residual retardation value that must be compensated by an optical film. For example, any retardation fluctuation of the cell or the film is easily noticeable. The OCB system requires a high level of uniformity. Also, the cell parameters, as well as the film parameters, should be optimized in order to maximize the optical performance. Using the Fuji film, OCB has a comparable viewing angle performance to MVA and IPS. In addition to faster response time, OCB has another advantage of less color shift at gray levels. In particular, human skin looks good even at oblique incidence.
OCB-LCD
4 3.5 3 2.5 Response Time (ms)
2 1.5 1
8 7 6
0.5 0 8
7
2
6
5 4 Start Level
3
2
5 4 3 End Level
1 1
NBB-LCD
4 3.5 3 2.5 Response Time (ms)
2 1.5 1 0.5 0 8
Figure 4.31
7
6
5 4 Start Level
3
1 2
2
8 7 6 5 4 3 End Level
1
Gray-to-gray response time of OCB (top) and NBB (bottom) cells. (Courtesy of Dr H.S. Kwok)
Liquid crystal displays
91
4.10.3 No-bias bend cell The OCB cell discussed above needs a critical voltage in order to transfer the cell from splay to bend mode. The transition process from splay to a stable bend structure will take a few seconds. Each TFT will need a constant bias voltage. Thus, any fluctuation in the manufacturing process or TFT biased voltage will cause intensity variation in the displayed images. Studies show that if the pretilt angle is larger than ∼48◦ then the -cell is already in the bend mode and no biased voltage is required.116, 117 This pretilt angle can be obtained by blending a homogeneous PI and homeotropic PI together.118 By doing so, a variable pretilt angle can be generated by controlling the composition of each PI. This technique is particularly interesting for generating the medium to large pretilt angle, as required by a no-bias bend (NBB) cell. The NBB cell exhibits faster response time than the OCB cell, as shown in Figure 4.31.119 This fast response time is particularly attractive for color-sequential LCDs in order to minimize color break-up. A drawback of the NBB cell is that its maximum transmittance is decreased to ∼70 % at λ = 550 nm, as compared to ∼85 % for the OCB cell. At such a high pretilt angle, the remaining phase retardation is dramatically decreased.
4.11 Transflective liquid crystal displays 4.11.1 Introduction A transmissive LCD uses a backlight to read out the displayed images. Its performances in resolution, viewing angle, contrast ratio and color gamut are all very good. However, under direct sunlight, the image can be washed out. Figure 4.32 shows an LCD under ambient lighting condition. Here the noise may come from stray ambient light that reaches the eye from scattering elsewhere. Meanwhile, not all the surface reflections of ambient light are within the acceptance angle of the eye. For simplicity, let us only consider the noise that comes to the eye from the LCD surface reflection, i.e. N = AR. Under such circumstances, the ambient contrast ratio (ACR) is defined as ACR =
Ton + N S + AR = , Toff + N (S/CR) + AR
(4.38)
where T on and T off are the bright and dark state transmittance of the LCD, respectively. The device contrast ratio CR is equal to T on /T off . To give an example, let us assume an LCD has CR = 500 and R = 0.5 % which is independent of incident angle. Figure 4.33 shows the ambient contrast ratio as a function of A/S.
Eye N
A
A
R
S
LCD
Backlight Figure 4.32 Consideration of sunlight readability of an LCD. A, ambient light; R, surface reflection of the LCD panel; S, transmitted image signal; N, noise.
92
Introduction to Flat Panel Displays
500
ACR
400 300 200 100 0 0.0
1.0
2.0
3.0
4.0
5.0
A/S Figure 4.33 Ambient contrast ratio versus A/S for an LCD with CR = 500:1 and R = 0.5 %.
In dark ambient where A = 0, the ambient contrast ratio is the same as the LCD’s device contrast ratio. As ambient light increases, the ambient contrast ratio decreases dramatically. At A = S, the perceived contrast ratio declines to ∼140:1. At A > 50S, the ambient contrast ratio drops to below 5:1 and the LCD image is washed out. To improve sunlight readability, we could increase the display brightness (S), reduce the surface reflection (R) or increase the intrinsic device contrast ratio. Judging from Equation (4.38), increasing display brightness and reducing surface reflection are more effective than boosting the device’s intrinsic contrast ratio. To increase display brightness, we could improve the panel’s optical efficiency or increase the backlight brightness, but the latter would increase the power consumption substantially. Moreover, keeping on increasing display brightness would cause discomfort at dark ambient. Therefore, an adaptive brightness control is necessary. This means the display will get brighter at bright ambient and darker at dark ambient. Another approach to overcome sunlight readability is to use a reflective display, as shown in Figure 4.34. The top linear polarizer and a broadband quarter-wave film form an equivalent crossed polarizer for the incident and exit beams. This is because the LC modes work better under crossed-polarizer condition. The bumpy reflector not only reflects but also diffuses the ambient light to the observer in order to avoid specular reflection and widen the viewing angle. This is a critical part for reflective LCDs. The TFT is hidden beneath the bumpy reflector; thus, the reflective LCD can have a large aperture ratio (∼90 %). The light blocking layer (LBL) is used to absorb the scattered light from neighboring pixels. Two popular LCD modes have been widely used for reflective LCDs: (1) VA cell and (2) mixed-mode twisted nematic (MTN) cell. The VA cell utilizes the phase retardation effect while the MTN cell uses a combination of polarization rotation and birefringence effects.120 In a reflective LCD, there is no built-in backlight unit; instead, it utilizes ambient light for reading out the displayed images.121 In comparison to transmissive LCDs, reflective LCDs have advantages of lower power consumption, lower weight and better sunlight readability. However, a reflective LCD is not suitable for use under low or dark ambient conditions. To make a display useable for both dark and bright ambient, transflective liquid crystal displays (TR-LCDs) have been developed.122 A TR-LCD can display images in both transmissive (T) mode and reflective (R) mode simultaneously or independently. Under bright ambient, the backlight can be turned off to save power and therefore the TR-LCD operates in the R mode only. Under dark ambient, the backlight is on and the TR-LCD works in the T mode. In low to medium ambient conditions, the TR-LCD runs in both T and R modes simultaneously. Therefore, a TR-LCD can accommodate a large dynamic range. At present, TR-LCDs are mainly targeted for use in mobile displays, such as mobile phones, digital cameras, personal digital assistants and global positioning systems. These displays are often used outdoors.
Liquid crystal displays
93
Ambient Light
Polarizer λ/4 film
Glass Color filter
LC
Bumpy Reflector
LBL
Source
TFT
Drain
Gate
Glass
Figure 4.34 Device structure of a direct-view reflective LCD.
The major technical challenges for a TR-LCD are to balance the phase retardation and color saturation, to obtain high transmittance and reflectance and matched gamma curves, and to achieve the same response time for the transmissive and reflective light. In a TR-LCD, the ambient light passes the LC layer twice in the reflective region, but the backlight only passes the LC layer once in the transmissive region. Therefore, we need to balance this optical path length disparity in order to obtain high transmittance and reflectance simultaneously. Moreover, in the reflective region, the viewing angle is governed by the reflector and contrast ratio by the surface reflection. To achieve a wide viewing angle for the R mode, a Lambertian type of bumpy reflector, e.g., paper, is preferred. However, this would reduce the optical efficiency. Because a TR-LCD is most likely to be used in palm-size mobile displays, the optimal viewing direction can be easily controlled. Thus, a diffusive reflector that can steer the reflected light to the vicinity of normal view (0–30◦ ) is highly desirable. In the following, we briefly introduce the basic operation principles of a TR-LCD.
4.11.2 Dual cell gap transflective LCDs To balance the optical path difference between the T and R regions for a TR-LCD, the dual cell gap device concept is introduced.123 The basic philosophy for a TR-LCD is to achieve equal phase retardation between the T and R modes, which is expressed as dT (n)T = 2dR (n)R .
(4.39)
If the T and R modes have the same effective birefringence, then the cell gap should be different. This is the so-called dual cell gap approach. On the other hand, if the cell gap is uniform (single cell gap approach), then we should find ways to make (n)T = 2(n)R . Let us discuss the dual cell gap approach first. Figure 4.35(a) shows a schematic of the device configuration of a dual cell gap TR-LCD. Each pixel is divided into a reflective region with cell gap d R and a transmissive region with cell gap d T . The LC employed could be homogeneous alignment (also known as ECB) or VA, as long as it is a phase
94
Introduction to Flat Panel Displays
Ambient Light P λ/4 film CF LC (λ/2) R region
d
T region
2d
λ/4 film P
Back Light (a) 1.0
0.8 T mode R mode
T/R
0.6
0.4
0.2
0.0 0
1
2
3 4 Voltage (Vrms) (b)
5
6
Figure 4.35 (a) Schematic device configuration of a dual cell gap TR-LCD. (b) Simulated VT and VR curves using VA (or MVA) cells. LC, MLC-6608; d T = 4.5 m and d R = 2.25 m; λ = 550 nm.
retardation type. To balance the phase retardation between the single and double pass of the T and R parts, we could set d T = 2d R . Moreover, to balance the color saturation due to single- and double-pass discrepancy, we could use thinner or holed color filters in the R part. The top quarter-wave plate is needed mainly for the reflective mode to obtain a high contrast ratio. Therefore, in the T region the optical axis of the bottom quarter-wave plate should be aligned perpendicular to that of the top one so that their phase retardations are canceled. For a thin homogeneous cell it is difficult to find a good common dark state for RGB wavelengths without a compensation film.124 The compensation film can be designed into the top quarter-wave film shown in Figure 4.35(a) to form a single film. Here, let us take a dual cell gap TR-LCD using VA (or MVA for wide-view) and MLC-6608 (ε = −4.2, n = 0.083) as an example. We set d R = 2.25 m in the R region and d T = 4.5 m in the T region. Figure 4.35(b) depicts the voltage-dependent transmittance (VT) and reflectance (VR) curves at normal incidence. As expected, both VT and VR curves perfectly overlap with each other. Here d R n = 186.8 nm and d T n = 373.5 nm are intentionally designed to be larger than λ/4 (137.5 nm) and λ/2 (275 nm), respectively, in order to reduce the on-state voltage to ∼5 Vrms .
Liquid crystal displays
95
In a TR-LCD using MVA cells, appropriate phase compensation films have to be used in order to obtain a wide viewing angle. If ECB cells are used, then a special compensation film is needed. For ECB mode it is harder to achieve a wide viewing angle and high contrast ratio. Three problems of dual cell gap TR-LCDs are apparent. (1) Due to the cell gap difference, the LC alignment is distorted near the T and R boundaries. The distorted LCs will cause light scattering and degrade the device contrast ratio. Therefore, these regions should be covered by black matrices in order to retain a good contrast ratio. (2) The thicker cell gap in the T region results in a slower response time than the R region. Fortunately, the dynamic response requirement in mobile displays is not as strict as that for video applications. This response time difference, although not perfect, is still tolerable. (3) The view angle of the single-domain ECB mode is relatively narrow because the LC directors are tilted out of the plane by the longitudinal electric field. To improve view angle, a biaxial film125 or a hybrid aligned nematic polymeric film126 is needed. Because the manufacturing process is compatible with LCD fabrication lines, dual cell gap TR-LCDs are widely used in commercial products, such as the iPhone. Wide-view TR-LCDs are emerging because of the data and video convergence in portable displays. Several wide-view technologies described earlier in this chapter can be considered. For instance, dual cell gap IPS,127 FFS,128 MVA129 and two-domain PVA130 have all been proposed. Some of these approaches require a patterned in-cell phase retarder in order to obtain a good dark state for the R mode and some require dual LC alignments in the T and R regions. Sophisticated manufacturing processes imply an increased cost.
4.11.3 Single cell gap transflective LCDs As its name implies, the single cell gap TR-LCD has a uniform cell gap in the T and R regions. Therefore, from Equation (4.39), we need to find methods or device concepts to achieve (n)T = 2(n)R . Several approaches have been proposed to solve this problem. In this section, we discuss four examples: (1) the dual-TFT method in which one TFT is used to drive the T mode and another TFT to drive the R mode at a lower voltage;131 (2) the divided-voltage method,132 with multiple R parts and the superimposed VR curve matches the VT curve; (3) partial switching effect133 in which the R part has a weaker electric field so that the LC directors are only partially switched; and (4) the dual-alignment method,134 where T and R subpixels have different molecular alignments in order to match the VT and VR curves.
Example 4.4: dual-TFT method Figure 4.36(a) shows the device structure of a TR-LCD using two TFTs to separately control the gamma curves of the T and R parts. Here, TFT-1 is connected to the bumpy reflector and TFT-2 is connected to the ITO of the transmissive part. Because of the double passes, the VR curve has a sharper slope than the VT curve and it reaches the peak reflectance at a lower voltage, as shown in Figure 4.36(b). Let us use a 4.5 m vertically aligned LC layer with 88◦ pretilt angle as an example. The LC mixture employed is Merck MLC-6608 and the wavelength is λ = 550 nm. From Figure 4.36(b), the peak reflectance occurs at 3 Vrms and transmittance at 5.5 Vrms . Thus, the maximum voltage of TFT-1 should be set at 5.5 V and TFT-2 at 3 V. This driving scheme is also called the double-gamma method.135 The major advantage of this dual-TFT approach is its simplicity. However, each TFT takes up some space so that the aperture ratio for the T mode is reduced. (continued)
96
Introduction to Flat Panel Displays
R
T P λ/4 S
VA (λ/2)
ITO TFT 2
TFT 1
S λ /4 P
Backlight (a) 1.0
0.8
T mode R mode
T/R
0.6
0.4
0.2
0.0 0
1
2
3 4 Voltage (Vrms) (b)
5
6
Figure 4.36 (a) Device structure of a dual-TFT TR-LCD; (b) simulated VT and VR curves using a 4.5 m MLC-6608 LC layer.
For a TR-LCD, the T mode should have priority over the R mode. The major function of the R mode is to preserve sunlight readability. In general, the viewing angle, color saturation and contrast ratio of the R mode are all inferior to the T mode. In most lighting conditions, except under direct sunlight, the T mode is still the primary display.
Example 4.5: divided-voltage method Figure 4.37(a) shows the device structure of a TR-LCD using the divided-voltage method.136 The R region consists of two subregions: R-I and R-II. Between R-II and the bottom ITO, there is a passivation layer to weaken the electric field in the R-II region. As plotted in Figure 4.37(b), the VR-II curve in the R-II region has a higher threshold voltage than the VT curve due to this voltage shielding
Liquid crystal displays
97
effect. To better match the VT curve, a small area in the T region is also used for R-I. The bumpy reflector in the R-I region is connected to the bottom ITO through a channeled electrode. Because of the double passes of ambient light, the VR-I curve is sharper than the VT curve. By properly choosing the R-I and R-II areas, we can match the VT and VR curves well, as shown in Figure 4.37(b). P ITO LC R Pa ITO Via P R-II
R-I
T
Light Intensity (normalized)
(a) Transmittance
1.0
Reflectance of Reflective Part I Reflectance of Reflective Part II
0.8
Reflectance of Reflec. Part I and II
0.6 0.4 0.2 0.0
0
1
2 3 Voltage, Vrms
4
5
(b) Figure 4.37 TR-LCD using the divided-voltage approach. (a) Device structure and (b) VT and VR curves at different regions. P, polarizer; R, bumpy reflector; Pa, passivation layer.
Example 4.6: partial switching effect Another approach to balance the disparity of optical path length is to partially switch the R part of the LC molecules such that the effective birefringence is only one-half of the T part, i.e. (n)R = (n)T /2. To achieve this goal, the electric field in the R part should be weaker than in the T part. This method is termed partial switching. Figure 4.38 shows a TR-LCD using the partial switching mechanism. The initial LC alignment is vertical with a negative ε material.As indicated in the figure, the bumpy reflectors are fabricated on top of the bottom ITO with a passivation layer. These conductive reflectors are connected together with a common bias voltage, say V b = 0 or 1.5 Vrms . When a voltage is applied between top (continued)
98
Introduction to Flat Panel Displays
and bottom ITO electrodes, the electric fields above the reflectors are partially shielded. As a result, the LC directors rotate less in the R region than in the T region which, in turn, leads to smaller phase retardation in the R region. By optimizing the cell and compensation films, a reasonably good overlap between the VT and VR curves can be obtained.
R
T P λ/4 film Compensation Film ITO
z
x
VA LC Reflector ITO λ/4 film P
Figure 4.38 Device structure of a TR-LCD using the partial switching effect.
Example 4.7: dual-alignment method Dual alignment provides another degree of freedom to overcome the unequal path length between the T and R modes of a TR-LCD. The tradeoff is in the increased fabrication complexity. Figure 4.39 shows a FFS-based TR-LCD using this concept. The LC employed has a negative ε. In the top left, the pixel electrode in the T region is a transparent plate electrode and is further connected to a metallic reflector electrode in the R region. To work as a diffusive slant reflector, the reflector has embossed circle-shape patterns. A chevron-shaped common electrode with many stripes is formed above the pixel electrode. Here, w is the stripe width of the common electrode and is the electrode gap between stripes in the common electrode. In this pixel–common inversion electrode structure, the black matrix for the data and gate bus lines in a conventional structure (pixel electrodes are in stripes) can be removed because the stripe common electrode above the pixel electrode can remove the noise field from the data line at the dark state. From Figure 4.39(a), the LC directors are aligned homogeneous with their optical axis parallel to the transmission axis of the top linear polarizer. The bottom polarizer is crossed to the top one. In the R part, to achieve a good dark state, a hybrid alignment LC cell with a phase retardation value dn ≈ λ/4 (where d is the LC layer thickness, n the LC birefringence and λ the wavelength) is formed by photoalignment.137 Its optical axis on the bottom substrate is 45◦ with respect to the LC optical axis of the T part, which is also 45◦ away from the top polarizer’s transmission axis. Thus linearly polarized light from the top polarizer becomes circularly polarized light after the reflective λ/4 LC
Liquid crystal displays
W
99
l’
Data Bus Line Top Polarizer R
R
T Black Matrix
Color Filter Over Coater Liquid Crystals 45°
135°
Black Matrix 7° Common Electrode Pixel with Reflector
T Bottom Polarizer 112.5°
Backlight (b) Gate Bus Line (a)
Figure 4.39 Schematic of the pixel structure of a FFS-based high-brightness TR-LCD with a negative dielectric anisotropic LC.
layer. Upon reflection, the reflected light traverses the LC layer again and becomes linearly polarized except that its axis is rotated by 90◦ and is blocked by the analyzer, resulting in a good dark state. In the T mode, linearly polarized light from the bottom polarizer passes through the LC layer without changing its polarization state because the LC layer there is aligned parallel to the transmission axis of the bottom linear polarizers. Thus, it is blocked by the top polarizer, resulting in a common dark state, similar to the R mode. In other words, this TR-LCD is a normally black display in the null voltage state. When a voltage is applied to the electrodes, the generated electric fields will rotate the LC directors by ∼45◦ . In the T region, the linearly polarized light from the bottom polarizer propagates along the LC layer with polarization state changed by λ/2, so the polarization direction of this polarized light coincides with the top polarizer’s transmission axis, resulting in a bright state. In the R part, the linearly polarized light passes through the LC layers twice without changing its polarization, also leading to a bright state. Figure 4.40 shows the simulated VT and VR curves of the TR-LCD using optimized cell parameters: the single cell gap is 3.77 m in T and R parts and [w, ] of the T and R common electrode are [3 mm, 5 mm] and [2 mm, 3 mm], respectively. The physical properties of the LC employed are as follows: ne = 1.5512 and no = 1.4742 (at λ = 589 nm); dielectric anisotropy ε = −4.0; rotational viscosity γ 1 = 136 mPa s; and elastic constants K 11 = 13.5 pN, K 22 = 7 pN, K 33 = 15.1 pN. The boundary conditions are as follows: in the T part the surface tilt angle is 2◦ and initial rubbing angle is 7◦ with respect to the direction that is perpendicular to the common electrode stripes as shown in Figure 4.39(a). In the R part, the surface pretilt angle in the bottom array substrate is 2◦ and the initial rubbing angle is ±45◦ with respect to the LC rubbing direction in the T part. The pretilt angle of the LC directors on the top color filter substrate is 90◦ . (continued)
100
Introduction to Flat Panel Displays
Transmittance & Reflectance (%)
100 80 60 40 20 0
0
1
2
3 4 Voltage (V)
5
6
Figure 4.40 Simulated VT (circles) and VR (triangles) curves of the dual-alignment TR-LCD shown in Figure 4.39.
Under such conditions, the threshold voltage is V th ≈ 1.5 Vrms and the on-state voltage is V on ≈ 4.4 Vrms , while the maximum transmittance T max ≈ 80 % (normalized to the maximum transmittance of two parallel linear polarizers) in the T mode. In the R mode, V th ≈ 1.5 Vrms , V on ≈ 4.3 Vrms and Rmax ≈ 90 %. Although a negative ε LC is used, the pixel–common inversion electrode structure gives rise to a low operation voltage (∼4.4 Vrms ) similar to that of a TN display.138 More importantly, the VT and VR curves match very well, which allows the driving of both T and R modes with a single gamma curve. Figure 4.41 shows the iso-contrast contour plots for the T and R modes of the FFS TR-LCD. Without using any compensation film, the T mode shows a 10:1 contrast ratio over a 60◦ viewing cone and the R mode is over 45◦ . These viewing angles are adequate for mobile displays using a small-sized LCDs. Compared to a TR-LCD using ECB mode, the FFS mode is basically free from grayscale inversion. T-mode
R-mode
90
90
500
80 180
0
80
100
40 0
30 180
40 0
0
10 5 0 270
270
Figure 4.41 Iso-contrast contour plots for the transmissive and reflective modes of the FFS-based highbrightness TR-LCD
Liquid crystal displays
101
4.12 Future directions The first TFT LCD was introduced in the mid-1980s.139 In the past two decades, several key scientific and technological issues related to TFT LCDs, e.g. viewing angle, contrast ratio, cell gap control, motion picture image blurring and manufacturing cost, have been solved step by step. Their widespread application has just started. Replacing the existing cathode ray tube market in desktop monitors and TVs is taking off. Recently, the use of LED backlights has pushed this relatively young technology to the next level. LED backlights greatly widen the color gamut (>120 %), enhance the dynamic contrast ratio (>50 000:1) and reduce the power consumption by about two times. Moreover, through sequential colors using RGB LEDs, color filters can be eliminated and optical efficiency can be enhanced by 3–4 times. To minimize color break-up, LC response times need to be reduced to ∼1–2 ms. Low-viscosity LCs and methods for achieving fast response time are urgently needed. Besides OCB, a new rubbing-free, wide-view operation mode with submillisecond gray-to-gray response time and high contrast ratio has been recently demonstrated.140 However, its operating voltage is still too high and needs to be decreased. Sunlight readability is a general concern, not just for mobile phones, but also for notebook computers. The transflective approach is viable, but it sacrifices the optical efficiency of the transmissive mode and color saturation and contrast ratio of the reflective mode. A better approach than the present transflective LCDs remains to be developed. Flexible displays, although not addressed here, are emerging. Several technologies are advancing rapidly, such as electrophoretic displays and organic light-emitting devices. Within the LCD family, flexible and rollable cholesteric displays have been demonstrated but are still in their infancy stage.
Homework problems 4.1 A eutectic mixture is comprised of the following two compounds:
(1)
C3H7
NCS
K 39 N 41.3 I; H = 4300 cal mol−1
(2)
NCS
C3H7
K 66 N 190 I; H = 3000 cal mol−1 (a) Calculate its nematic range (in ◦ C). (b) Calculate the molecular weight of compounds (1) and (2). (c) If we want to prepare 10 g of this mixture, how many grams of each compound should be used? 4.2 In a fast-response LC phase modulator, we have to consider birefringence and viscoelastic coefficient together. A figure-of-merit (FoM) is defined as FoM =
λ(n2 ) γ1
where K 11 is the splay elastic constant, n is the birefringence and γ1 is the rotational viscosity. Assume K 11 ≈ S 2 ; n ≈ S; γ1 ≈ S exp(E/kT ); and S = (1 − T /T c )α . (continued)
102
Introduction to Flat Panel Displays
(a) Prove that FoM exhibits a maximum at the optimal operating temperature (T op ). (b) Derive the analytical expression for T op . Estimate T op if activation energy E = 0.35 eV, α = 0.18, T c = 100 ◦ C and Boltzmann constant k = 0.0861 meV K−1 . (c) Explain why FoM has a maximum at T op . 4.3 At T = 20 ◦ C, an LC mixture has refractive indices as follows: (ne , no ) = (1.5733, 1.4859) at λ = 450 nm and (1.5565, 1.4751) at λ = 633 nm. What are the extrapolated (ne , no ) at λ = 1550 nm? 4.4 A homogeneous cell is useful as a tunable phase retardation plate. The following chart plots the voltage-dependent transmittance of a homogeneous LC cell at λ = 633 nm. The polarizers are crossed and the angle between the front polarizer and the LC rubbing direction is β = 45◦ . 1 Normalized Transmittance
0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0
0.5
1
1.5
2
2.5 3 3.5 Voltage (Vrms)
4
4.5
5
5.5
6
(a) If the cell gap is d = 5 m, what is the birefringence of the LC? (b) At what voltages is the output beam (before the analyzer) circularly polarized? (c) If we want to switch the outgoing beam from circular to linear polarization, which voltages do we use in order to obtain the fastest response time? (d) Plot the VT curve using the above chart if β = 0. 4.5 A student prepared three cells, VA, 90◦ TN and homogeneous cell, but forgot to put labels on them. Can you help the student to identify which cell is which by using a white light table and two linear polarizers? 4.6 IPS and MVA are the two major approaches for wide-view LCD TVs. Compare their pros and cons. 4.7 Similar to a homogeneous cell, a VA cell can be used for phase-only modulation. From an LC molecular structure viewpoint, explain why a homogeneous cell is a favored choice. 4.8 A transflective LCD uses double cell gap TN/MTN cells. The LC parameters are: n = 0.1 (λ = 550 nm), ε = 10, K 11 = 10 pN, K 22 = 6 pN, K 33 = 20 pN and γ 1 = 0.1 Pa s. (a) Sketch the device configuration. (b) What are the required cell gaps for the R and T regions? (c) Draw the expected VT and VR curves.
Liquid crystal displays
103
References 1. Armitage, D., Underwood, I. and Wu, S.T. (2006) Introduction to Microdisplays, John John Wiley & Sons, Ltd, Chichester. 2. Khan, A., Schneider, T., Montbach, E. et al. (2007) Recent progress in color flexible reflective cholesteric displays. SID Symp. Dig., 38, 54. 3. Bae, J.H., Jang, S.J., Choi, Y.S. et al. (2007) The stabilized bistable LC mode for flexible display. SID Symp. Dig., 38, 649. 4. Stupp, E.H. and Brennesholtz, M. (1998) Projection Displays, John Wiley & Sons, Inc., New York. 5. Wu, S.T. and Yang, D.K. (2001) Reflective Liquid Crystal Displays, John Wiley & Sons, Inc., New York. 6. Crawford, G.P. (2005) Flexible Flat Panel Displays, John Wiley & Sons, Ltd, Chichester. 7. Lu, R., Hong, Q., Ge, Z. and Wu, S.T. (2006) Color shift reduction of a multi-domain IPS-LCD using RGB-LED backlight. Opt. Express 14, 6243. 8. Yang, D.K. and Wu, S.T. (2006) Fundamentals of Liquid Crystal Devices, John Wiley & Sons, Inc., New York. 9. Jonza, J.M., Weber, M.F., Ouderkirk, A.J. and Stover, C.A. (1999) Polarizing beam-splitting optical component. US Patent 5,962,114. 10. de Greef. P. and Hulze, H.G. (2007) Adaptive dimming and boosting backlight for LCD-TV systems. SID Symp. Dig., 38, 1332. 11. Chen, H., Sung, J., Ha, T. and Park, Y. (2007) Locally pixel-compensated backlight dimming for improving static contrast on LED backlit LCDs. SID Symp. Dig., 38, 1339. 12. Lin, F.C., Liao, C.Y., Liao, L.Y. et al. (2007) Inverse of mapping function method for image quality enhancement of high dynamic range LCD TVs. SID Symp. Dig., 38, 1343. 13. Anandan, M. (2008) Progress of LED backlights for LCDs. J. SID, 16, 287. 14. Goodby, J.W. Ferroelectricity Liquid Crystals: Principles, Properties and Applications, Routledge. 15. Wand, M., Thurmes, W.N., Vohra, R.T. and More, K.M. (1997) Advances in ferroelectric liquid crystals for microdisplay applications. SID Symp. Dig., 27, 157. 16. Yang, D.K., Lu, Z.J., Chien, L.C. and Doane, J.W. (2003) Bistable polymer dispersed cholesteric reflective display. SID Symp. Dig., 34, 959. 17. Gray, G. Harrison, K.J. and Nash, J.A. (1973) New family of nematic liquid crystals for displays. Electron. Lett., 9, 130. 18. Schroder, L. (1893) Z. Phys. Chem., 11, 449. 19. Van Laar, J.J. (1908) Z. Phys. Chem., 63, 216. 20. Bedjaoui, L., Gogibus, N., Ewen, B. et al. (2004) Preferential solvation of the eutectic mixture of liquid crystals E7 in a polysiloxane. Polymer, 45, 6555. 21. Deuling, H.J. (1978) Liquid Crystals, Solid State Physics Supplement 14 (ed. L. Liebert), Academic Press, New York. 22. Maier, W. and Meier, G. (1961) A simple theory of the dielectric characteristics of homogeneous oriented crystalline-liquid phases of the nematic type. Z. Naturforsch. A, 16, 262. 23. Schadt, M. (1992) Field-effect liquid-crystal displays and liquid-crystal materials: key technologies of the 1990s. Displays, 13, 11. 24. Gray, G., Harrison, K.J. and Nash, J.A. (1973) New family of nematic liquid crystals for displays. Electron. Lett., 9, 130. 25. Dabrowski, R. (1990) Isothiocyanates and their mixtures with a broad range of nematic phase. Mol. Cryst. Liq. Cryst., 191, 17. 26. Schadt, M. and Helfrich, W. (1971) Voltage-dependent optical activity of a twisted nematic liquid crystal. Appl. Phys. Lett., 18, 127. 27. Soref, R.A. (1973) Transverse field effect in nematic liquid crystals. Appl. Phys. Lett., 22, 165. 28. Oh-e, M. and Kondo, K. (1995) Electro-optical characteristics and switching behavior of the in-plane switching mode. Appl. Phys. Lett., 67, 3895. 29. Nakazono, Y., Ichinose, H., Sawada, A. et al. (1997) International Display Research Conference, p. 65. 30. Tarao, R., Saito, H., Sawada, S. and Goto, Y. (1994) Advances in liquid crystals for TFT displays. SID Tech. Dig., 25, 233. 31. Geelhaar, T., Tarumi, K. and Hirschmann, H. (1996) Trends in LC materials. SID Tech. Dig., 27, 167. 32. Goto, Y., Ogawa, T., Sawada, S. and Sugimori, S. (1991) Fluorinated liquid crystals for active matrix displays. Mol. Cryst. Liq. Cryst., 209, 1.
104
Introduction to Flat Panel Displays
33. Schiekel, M.F. and Fahrenschon, F. (1971) Deformation of nematic liquid crystals with vertical orientation in electric fields. Appl. Phys. Lett., 19, 391. 34. Eidenschink, R. and Pohl, L. (1983) US Patent 4,415,470. 35. de Gennes, P.G. and Prost, J. (1993) The Physics of Liquid Crystals, 2nd edn, Oxford University Press, Oxford. 36. Maier, W. and Saupe, A. (1960) A simple molecular statistical theory for nematic liquid crystal phase: II. Z. Naturforsh. A, 15, 287. 37. Gruler, H. (1975) The elastic constants of a nematic liquid crystal. Z. Naturforsh. A, 30, 230. 38. Jakeman, E. and Raynes, E.P. (1972) Electro-optic response times of liquid crystals. Phys. Lett. A, 39, 69. 39. Imura, H. and Okano, K. (1972) Temperature dependence of the viscosity coefficients of liquid crystals. Jpn. J. Appl. Phys., 11, 1440. 40. Diogo, A.C. and Martins, A.F. (1981) Thermal behavior of the twist viscosity in a series of homologous nematic liquid crystals. Mol. Cryst. Liq. Cryst., 66, 133. 41. Belyaev, V.V., Ivanov, S. and Grebenkin, M.F. (1985) Temperature dependence of rotational viscosity of nematic liquid crystals. Sov. Phys. Crystallogr., 30, 674. 42. Wu, S.T. and Wu, C.S. (1990) Rotational viscosity of nematic liquid crystals. Liq. Cryst., 8, 171. 43. Haller, I. (1975) Thermodynamic and static properties of liquid crystals. Prog. Solid State Chem., 10, 103. 44. Osipov, M.A. and Terentjev, E.M. (1989) Rotational diffusion and rheological properties of liquid crystals. Z. Naturforsch. A, 44, 785. 45. Wu, S.T. and Wu, C.S. (1990) Experimental confirmation of Osipov–Terentjev theory on the viscosity of liquid crystals. Phys. Rev. A, 42, 2219. 46. Wu, S.T., Ramos, E. and Finkenzeller, U. (1990) Polarized UV spectroscopy of conjugated liquid crystals. J. Appl. Phys., 68, 78. 47. Wu, S.T., Efron, U. and Hess, L.D. (1984) Birefringence measurement of liquid crystals. Appl. Opt., 23, 3911. 48. Wu, S.T. (1991) A semi-empirical model for liquid-crystal refractive index dispersions. J. Appl. Phys., 69, 2080. 49. Wu, S.T., Wu, C.S., Warenghem, M. and Ismaili, M. (1993) Refractive index dispersions of liquid crystals. Opt. Eng., 32, 1775. 50. Li, J. and Wu, S.T. (2004) Extended Cauchy equations for the refractive indices of liquid crystals. J. Appl. Phys., 95, 896. 51. Wu, S.T., Efron, U. and Hess, L.D. (1984) Infrared birefringence of liquid crystals. Appl. Phys. Lett., 44, 1033. 52. Li, J. and Wu, S.T. (2004) Two-coefficient Cauchy model for low birefringence liquid crystals. J. Appl. Phys., 96, 170. 53. Mada, H. and Kobayashi, S. (1976) Wavelength and voltage dependences of refractive indices of nematic liquid crystals. Mol. Cryst. Liq. Cryst., 33, 47. 54. Stupp, E.H. and Brennesholtz, M.S. (1998) Projection Displays, John Wiley & Sons, Inc., New York. 55. Li, J., Gauza, S. and Wu, S.T. (2004) High temperature-gradient refractive index liquid crystals. Opt. Express, 12, 2002. 56. Li, J. and Wu, S.T. (2004) Temperature effect on liquid crystal refractive indices. J. Appl. Phys., 96, 19. 57. Cognard, J. (1982) Alignment of nematic liquid crystals and their mixtures. Mol. Cryst. Liq. Cryst., 1(Suppl.), 1. 58. Stöhr, J., Samant, M.G., Lüning, J. et al. (2001) Liquid crystal alignment on carbonaceous surfaces with orientational order. Science 292, 2299. 59. Scheffer, T.J. and Nehring, J. (1977) Accurate determination of liquid-crystal tilt bias angles. J. Appl. Phys., 48, 1783. 60. Schiekel, M.F. and Fahrenschon, K. (1971) Deformation of nematic liquid crystals with vertical orientation in electric fields. Appl. Phys. Lett., 19, 391. 61. Schadt, M. and Helfrich, W. (1971) Voltage-dependent optical activity of a twisted nematic liquid crystal. Appl. Phys. Lett., 18, 127. 62. Scheffer, T. and Nehring, J. (1990) Twisted nematic and super-twisted nematic mode LCDs, in Liquid Crystals Applications and Uses (ed. B. Bahadur), World Scientific, Singapore, Vol. 1, Ch. 10. 63. Drzaic, P.S. (1995) Liquid Crystal Dispersions, World Scientific, Singapore. 64. McManamon, P.F., Dorschner, T.A., Corkum, D.L. et al. (1996) Optical phased array technology. Proc. IEEE, 84, 268. 65. Bos, P.J. and Koechler/Beran, K.R. (1984) The pi-cell: a fast liquid crystal optical switching device. Mol. Cryst. Liq. Cryst., 113, 329. 66. Freedericksz, V. and Zolina, V. (1933) Forces causing the orientation of an anisotropic liquid. Trans. Faraday Soc., 29, 919.
Liquid crystal displays
105
67. Jiao, M., Ge, Z., Song, Q. and Wu, S.T. (2008) Alignment layer effects on thin liquid crystal cells. Appl. Phys. Lett., 92, 061102. 68. Gooch, C.H. and Tarry, H.A. (1975) The optical properties of twisted nematic liquid crystal structures with twisted angles ≤90◦ . J. Phys. D, 8, 1575. 69. Wu, S.T. and Wu, C.S. (1999) Mixed-mode twisted-nematic cell for transmissive liquid crystal display. Displays, 20, 231. 70. Mori, H., Itoh, Y., Nishiura, Y. et al. (1997) Jpn. J. Appl. Phys., 36, 143. 71. Yang, K.H. (1991) International Display Research Conference, p. 68. 72. Mori, H., Nagai, M., Nakayama, H. et al. (2003) Novel optical compensation method based upon a discotic optical compensation film for wide-viewing-angle LCDs. SID, 34, 1058. 73. Mori, H. (2005) The wide view film for enhancing the field of view of LCDs. J. Display Technol., 1, 179. 74. Soref, R.A. (1973) Transverse field effect in nematic liquid crystals. Appl. Phys. Lett., 22, 165. 75. Soref, R.A. (1974) Field effects in nematic liquid crystals obtained with interdigital electrodes. J. Appl. Phys., 45, 5466. 76. Kiefer, R., Weber, B., Windscheid, F. and Baur, G. (1992) In-plane switching of nematic liquid crystals. Jpn. Displays, 92, 547. 77. Oh-e, M., Ohta, M., Arantani, S. and Kondo, K. (1995) Principles and characteristics of electro-optical behavior with in-plane switching mode. Asia Display ’95, p. 577. 78. Ge, Z., Zhu, X., Wu, T.X. and Wu, S.T. (2006) High-transmittance in-plane-switching liquid-crystal displays using a positive-dielectric-anisotropy liquid crystal. J. SID, 14, 1031. 79. Chen, J., Kim, K.H., Jyu, J.J. et al. (1998) Optimum film compensation modes for TN and VA LCDs. SID Tech. Dig., 29, 315. 80. Anderson, J.E. and Bos, P.J. (2000) Methods and concerns of compensating in-plane switching liquid crystal displays. Jpn. J. Appl. Phys. Part 1, 39, 6388. 81. Hong, Q., Wu, T.X., Zhu, X. et al. (2005) Extraordinarily high-contrast and wide-view liquid-crystal displays. Appl. Phys. Lett., 86, 121107. 82. Saitoh, Y., Kimura, S., Kusafuka, K. and Shimizu, H. (1998) Optimum film compensation of viewing angle of contrast in in-plane-switching-mode liquid crystal display. Jpn. J. Appl. Phys. Part 1, 37, 4822. 83. Ishinabe, T., Miyashita, T., Uchida, T. and Fujimura, Y. (2001) A wide viewing angle polarizer and a quarterwave plate with a wide wavelength range for extremely high quality LCDs. Proceedings of the 21st International Display Research Conference (Asia Display/IDW’01), p. 485. 84. Ishinabe, T., Miyashita, T. and Uchida, T. (2002) Wide-viewing-angle polarizer with a large wavelength range. Jpn. J. Appl. Phys. Part 1, 41, 4553. 85. Pasqual, F.D., Deng, H., Fernandez, F.A. et al. (1999) Theoretical and experimental study of nematic liquid crystal display cells using the in-plane-switching mode. IEEE Trans. Electron Dev., 46, 661. 86. Ohmuro, K., Kataoka, S., Sasaki, T. and Koite, Y. (1997) Development of super-high-image-quality vertical alignment mode LC. SID Tech. Dig., 26, 845. 87. Lee, S.H., Lee, S.L. and Kim, H.Y. (1998) Electro-optic characteristics and switching principle of a nematic liquid crystal cell controlled by fringe-field switching. Appl. Phys. Lett., 73, 2881. 88. Lee, S.H., Lee, S.L., Kim, H.Y. and Eom, T.Y. (1999) A novel wide-viewing-angle technology: ultra-trans view. SID Tech. Dig., 30, 202. 89. Wu, I.W., Ting, D.L. and Chang, C.C. (1999) Advancement in wide-viewing-angle LCDs. 6th International Display Workshops, p. 383. 90. Jeon, Y.M., Song, I.S., Lee, S.H. et al. (2005) Optimized electrode design to improve transmittance in the fringe-field switching liquid crystal cell. SID Tech. Dig., 36, 328. 91. Schiekel, M.F. and Fahrenschon, K. (1971) Deformation of nematic liquid crystals with vertical orientation in electric fields. Appl. Phys. Lett., 19, 391. 92. Grinberg, J., Bleha, W.P., Jacobson, A.D. et al. (1975) Photoactivated birefringence liquid crystal light valve for color symbology display. IEEE Trans. Electron Dev., ED-22, 775. 93. Sterling, R.D. and Bleha, W.P. (2000) D-ILA technology for electronic cinema. SID Tech. Dig., 31, 310. 94. Takeda, A., Kataoka, S., Sasaki, T. et al. (1997) A super-high-image-quality multi-domain vertical alignment LCD by new rubbing-less technology. SID Tech. Dig., 29, 1077. 95. Oh-e, M., Yoneya, M. and Kondo, K. (1997) Switching of negative and positive dielectric anisotropic liquid crystals by in-plane electric fields. J. Appl. Phys., 82, 528.
106
Introduction to Flat Panel Displays
96. Li, J., Wen, C.H., Gauza, S. et al. (2005) Refractive indices of liquid crystals for display applications. J. Display Technol., 1, 51. 97. Erickson, J.L. (1961) Conservation laws for liquid crystals. Trans. Soc. Rheol., 5, 23. 98. Leslie, F.M. (1968) Some constitutive equations for liquid crystals. Arch. Ration. Mech. Anal., 28, 265. 99. Jakeman, E. and Raynes, E.P. (1972) Electro-optic response times in liquid crystals. Phys. Lett., 39A, 69. 100. Nie, X., Xianyu, H., Lu, R. et al. (2007) Pretilt angle effects on liquid crystal response time. J. Display Technol., 3, 280. 101. Wu, S.T. and Wu, C.S. (1988) Small angle relaxation of highly deformed nematic liquid crystals. Appl. Phys. Lett., 53, 1794. 102. Wu, S.T. (1990) A nematic liquid crystal modulator with response time less than 100 s at room temperature. Appl. Phys. Lett., 57, 986. 103. Ohmuro, K., Kataoka, S., Sasaki, T. and Koike, Y. (1997) Development of super-high-image-quality vertical alignment-mode LCD. SID Tech. Dig., 28, 845. 104. Takeda, A., Kataoka, S., Sasaki, T. et al. (1998) A super high image quality multi-domain vertical alignment LCD by new rubbing-less technology. SID Tech. Dig., 29, 1077. 105. Kwag, J.O., Shin, K.C., Kim, J.S. et al. (2000) Implementation of new wide viewing angle mode for TFT-LCDs. SID Tech. Dig., 31, 256. 106. Kim, S.S. (2005) The world’s largest (82-in) TFT LCD. SID Tech. Dig., 36, 1842. 107. Hanaoka, K., Nakanishi, Y., Inoue, Y. et al. (2004) A new MVA-LCD by polymer sustained alignment technology. SID Tech. Dig., 35, 1200. 108. Kim, S.G. et al. (2007) Stabilization of the liquid crystal director in the patterned vertical alignment mode through formation of pretilt angle by reactive mesogen. Appl. Phys. Lett., 90, 261910. 109. Hong, H.K., Shin, H.H. and Chung, I.J. (2007) In-plane switching technology for liquid crystal display television. J. Display Technol., 3, 361. 110. Bos, P.J. and Koehler/Beran, K.R. (1984) The -cell: a fast liquid crystal optical switching device. Mol. Cryst. Liq. Cryst., 113, 329. 111. Uchida, T. (1998) Field sequential full color LCD without color filter by using fast response LC cell. 5th International Display Workshops, p. 151. 112. Yamaguchi, Y., Miyashita, T. and Uchida, T. (1993) Wide-viewing-angle display mode for the active-matrix LCD using bend-alignment liquid crystal cell. SID Tech. Dig., 24, 277. 113. Wu, S.T. (1986) Birefringence dispersion of liquid crystals. Phys. Rev. A, 33, 1270. 114. Fergason, J.L. (1983) Liquid crystal display with improved angle of view and response time. US Patent 4,385,806. 115. Ito, Y., Matsubara, R., Nakamura, R. et al. (2005) OCB-WV film for fast-response-time and wide viewing angle LCD-TVs. SID Tech. Dig., 36, 986. 116. Xu, M., Yang, D.K. and Bos, P. (1998) SID Int. Symp. Dig. Tech. Papers, 29, 139. 117. Yeung, F.S., Li, Y.W. and Kwok, H.S. (2006) Pi-cell liquid crystal displays at arbitrary pretilt angles. Appl. Phys. Lett., 88, 041108. 118. Yeung, F.S., Ho, J.Y., Li, Y.W. et al. (2006) Variable liquid crystal pretilt angles by nanostructured surfaces. Appl. Phys. Lett., 88, 051910. 119. Yeung, F.S., Li, Y.W. and Kowk, H.S. (2005) Fast response time no-bias bend LCD. International Display Workshops, p. 41. 120. Wu, S.T. and Wu, C.S. (1996) Mixed-mode twisted nematic liquid crystal cells for reflective displays. Appl. Phys. Lett., 68, 1455. 121. Kmetz, A.R. (1980) A single-polarizer twisted nematic display. Proc. SID, 21, 63. 122. Zhu, X., Ge, Z., Wu, T.X. and Wu, S.T. (2005) Transflective liquid crystal displays. J. Display Technol., 1, 15. 123. Shimizu, M., Itoh, Y. and Kubo, M. (2002) Liquid crystal display device. US Patent 6,341,002. 124. Wu, S.T. and Wu, C.S. (1998) A biaxial film-compensated thin homogeneous cell for reflective liquid crystal display. J. Appl. Phys., 83, 4096. 125. Shibazaki, M., Ukawa, Y., Takahashi, S. et al. (2003) Transflective LCD with low driving voltage and wide viewing angle. SID Tech. Dig., 34, 90. 126. Uesaka, T., Ikeda, S., Nishimura, S. and Mazaki, H. (2007) Viewing-angle compensation of TN- and ECB-LCD modes by using a rod-like liquid crystalline polymer film. SID Tech. Dig., 28, 1555. 127. Imayama, H., Tanno, J., Igeta, K. et al. (2007) Novel pixel design for a transflective IPS-LCD with an in-cell retarder. SID Tech. Dig., 38, 1651.
Liquid crystal displays
107
128. Kim, H.Y., Ge, Z., Wu, S.T. and Lee, S.H. (2007) Wide-view transflective liquid crystal display for mobile applications. Appl. Phys. Lett., 91, 231108. 129. Lin, C.H., Chen, Y.R., Hsu, S.C. et al. (2008) A novel advanced wide-view transflective display. J. Display Technol., 4, 120. 130. Sohn, J., Lyu, J., Cho, S.A. et al. (2007) Cost-effective pixel structures for high performance mobile PVA LCDs. SID Tech. Dig., 38, 1659. 131. Liu, K.H. et al. (2003) International Display Manufacturing Conference, p. 215. 132. Tsai, C.Y. et al. (2007) A new wide view transflective display. Proceedings of Asia Display, p. 24. 133. Choi, W.K. and Wu, S.T. (2006) US Patent 7,015,897. 134. Lee, S.H., Park, K.H., Gwag, J.S. et al. (2003) A multimode-type transflective liquid crystal display using the hybrid-aligned nematic and parallel-rubbed vertically aligned modes. Jpn. J. Appl. Phys., 42, 5127. 135. Sheu, C.R. et al. (2003) A novel LTPS transflective TFT LCD driving by double gamma method. SID Tech. Dig., 34, 653. 136. Yang, Y.C. et al. (2006) Single cell gap transflective mode for vertically aligned negative nematic liquid crystals. SID Tech. Dig., 37, 829. 137. Schadt, M., Schmitt, K., Kozenkov, V. and Chigrinov, V.G. (1992) Jpn. J. Appl. Phys. Part 1, 31, 2155. 138. Kim, H.Y., Hong, S.H., Rhee, J.M. and Lee, S.H. (2003) Liq. Cryst., 30, 1285. 139. Ishii, Y. (2007) The world of the TFT-LCD technology. J. Display Technol., 3, 351. 140. Jiao, M., Ge, Z., Wu, S.T. and Choi, W.K. (2008) Submillisecond response nematic liquid crystal modulators using dual fringe field switching in a vertically aligned cell. Appl. Phys. Lett., 92, 111101.
5 Plasma display panels 5.1 Introduction The operating principle of plasma display panels (PDPs) is similar to that of fluorescent lamps. Figure 5.1 shows the structure of a typical fluorescent lamp. In this structure, two filament electrodes are installed at two ends of an inner glass tube. The inner wall of the glass tube is coated with phosphor. The cavity of the glass tube is filled with a gaseous mixture of argon and mercury. When a particular voltage is applied to the electrodes, plasma is generated by gas discharge. The energy leveling mechanism of the plasma results in ultraviolet (UV) emission with a peak wavelength of 254 nm. The UV light excites the phosphor in the fluorescent lamp which, in turn, emits light. Based on the gas discharge mechanism mentioned above, PDPs have been fabricated and used in display applications for decades. UV generation in PDPs is similar to that of a fluorescent lamp. However, the gases most commonly used in PDPs are neon and xenon rather than argon and mercury. The peak wavelengths in PDPs are 147 and 173 nm which belong to the vacuum ultraviolet (VUV) region. VUV can only propagate in a vacuum because it is strongly absorbed by air. Although the unit cell size in a PDP cannot be too small and its operating voltage is high, PDPs have a wider view angle, faster response time and wider temperature range than liquid crystal displays (LCDs). The relatively large cell size is due to the limitation of the barrier rib size. The high operating voltage is due to the requirement of plasma generation. This high operating voltage requires a high-voltage driver integrated circuit (IC), so the cost of the electronics is high. PDPs are good candidates for large-panel displays because they are effective for static pictures and motion pictures, from cold ambient to hot ambient and for personal and public use. Additionally, PDPs can be fabricated at low cost and with a simple manufacturing process. Therefore, many PDP structures have been proposed and used in the past few years.1 4
5.2 Physics of gas discharge Gas discharge involves complex reactions. The four major reactions in gas discharge physics are excitation, metastable generation, ionization and Penning ionization. When a voltage is applied to a gas that is insufficient to ionize individual atoms, the atomic energy is excited to higher levels. This reaction is called excitation. For a helium (He) atom, the equation of its excitation behavior is given by He + e− → He∗ + e− ,
Introduction to Flat Panel Displays c 2008 John Wiley & Sons, Ltd
J.-H. Lee, D.N. Liu and S.-T. Wu
(5.1)
110
Introduction to Flat Panel Displays
Light
Ar Filament Phosphor Electrode
Mercury
UV
Glass tube Light Figure 5.1 Typical structure of a fluorescent lamp.
where He∗ represents the helium atom in the excited state. Meanwhile, this excited atom can decay to a metastable state by the emission of a wavelength ν. This metastable atom is an excited atom with a longer life than that of a typically excited atom. This reaction is called metastable generation and can be written as He∗ → Hem + ν, (5.2) where Hem is a metastable state of the helium atom. If an appropriate voltage is applied to the gas, then the atom is ionized. This reaction is called ionization and can be expressed as He + e− → He+ + 2e− ,
(5.3)
where He+ is the ionization state of a helium atom. To reduce the excitation voltage, a gas mixture is typically used in a PDP. A gaseous mixture can provide more ionization at the applied voltage5 because a metastable atom is likely to be ionized into another species of gas atom. This gas mixture is called a Penning mixture and the mechanism is called a Penning reaction. The Penning mechanism uses energetic particles to accelerate the discharge and to reduce the firing voltage. For helium and xenon gases, the Penning reaction is written as Hem + Xe → He + Xe+ + e− (5.4) where Xe+ is the ionization state of the xenon atom. Figure 5.2 presents these four major reactions of gas discharge for a helium–xenon gas mixture.
5.2.1 I–V characteristics Many gas discharges have similar characteristics. Figure 5.3 plots a typical I–V characteristic of gas discharge.6 In the low-voltage regime, the current is small and increases slightly with voltage. As the applied voltage reaches a particular value, gas discharge is initiated. This voltage is called the firing voltage. As the applied voltage continues to increase, the PDP panel voltage remains constant until glowing occurs. However, the current increases markedly with the applied voltage. This stage is called Townsend discharge. When glow has occurred in the glow stage, the voltage of the PDP panel begins to drop in the subnormal glow stage, stabilizes in the normal glow stage and increases in the abnormal stage, until the arc stage. Among these stages, the normal glow stage is that in which the PDP operates. When the voltage is less than a particular value, the glowing stops. This voltage is called the sustain voltage. Despite the applied voltage being lower than the firing voltage, the gas discharge persists as long as the applied voltage is higher than the sustain voltage.
Plasma display panels
111
e– Electrode
e–
He
He
e–
He*
Hem
He+
Collision e–
ν
e– Xe
He
Xe+ Electrode
Excitation: He + e– → He* + e– Metastable generation: He* → Hem + ν Ionization: He + e– → He+ + 2e– Penning reaction: Hem + Xe → He + Xe+ + e– Figure 5.2 Typical gas reaction in a helium–xenon gas mixture. The bold arrows indicate the collision mechanism in the reactions.
I Arc
Abnormal glow Normal glow Subnormal glow
Townsend discharge
Sustain voltage
Firing voltage
V
Figure 5.3 Typical I–V characteristics of a discharge.
5.2.2 Penning reaction and Paschen curve As briefly discussed above, the Penning reaction is useful in reducing the firing voltage. The firing voltage is related to the product of gas pressure and the distance between the cathode and the anode. The Paschen curve captures this relationship and is plotted in Figure 5.4,6 which indicates that the firing voltage is associated not only with the gas composition but also with the pressure p of the gas and the distance d
112
Introduction to Flat Panel Displays
Firing voltage (volts) 160
Ne Ne + Xe (0.2%)
120
Ne + Ar (0.2%)
80 0
2
4
6
8 pd (torr · cm)
Figure 5.4 Typical Paschen curve.7 (Adapted from Electronic Display Devices, John Wiley & Sons, Inc., 1990)
between the two electrodes. The optimum product p × d is associated with the minimum firing voltage. Furthermore, the additional ionization of the Penning reaction generally lowers the firing voltage of a gas mixture. Neon gas with a low percentage of xenon or argon, as in Figure 5.4, has a lower minimum firing voltage than pure neon gas. This behavior, the mechanism of the Penning reaction and the Paschen curve are particularly important in reducing the operating voltage.
5.2.3 Priming mechanism Gas discharge depends on energetic particles. The mechanism of the generation of energetic particles is called priming: energetic particles are generated and help to accelerate the discharge and reduce the firing voltage. A priming period with a priming function is typically designed at the beginning of the driving waveform so that it can cause gas discharge and generate plasma before the writing period. Although the discharge stops after the priming period, residual ions remain present in the panel. These residual ions reduce the required duration of, and voltage used in, the writing period. This mechanism offers a great advantage in high-speed addressing operations of PDPs.
5.3 Plasma display panels The direct current (DC) and alternating current (AC) structures are the two main PDP structures. DC PDP provides the advantage of simplicity while AC PDP provides the advantage of a longer operating lifetime.
5.3.1 DC PDP DC PDP uses two electrodes that are directly exposed to the display gas. Accordingly, the structure of DC PDP is simpler than that of AC PDP. Figure 5.5 presents a typical DC PDP structure. Since the two electrodes are directly exposed to the gas, plasma is generated by the application of a DC voltage. In order to expose the electrodes to the gas, a part of the area of phosphor is required not to cover the electrodes. When plasma is generated, UV radiation is generated. The UV radiation excites the phosphor which, in turn, emits light. However, the plasma directly bombards the electrodes and phosphor, reducing the lifetime of the display. Many improved approaches have been proposed to reduce
Plasma display panels
113
Light Plasma
Electrode Substrate
Light
UV
Light
UV
Barrier rib Phosphor Substrate
Electrode Figure 5.5 Typical structure of DC PDP.
the bombardment. However, the process is becoming increasingly complex in most of these approaches. Since DC PDP suffers from a relatively short operating lifetime and the improved approaches are not completely effective, AC PDP is the most popular type of PDP application.
5.3.2 AC PDP Since AC PDP uses an alternating current, a dielectric layer is needed. Additionally, a protection layer can be deposited on it. Such a protection layer in AC PDP increases the operating lifetime since this layer protects against plasma bombardment. The two major types of AC PDP are surface discharge and vertical discharge, which are presented in Figure 5.6.8 Electrode
Electrode Substrate Dielectric
Plasma
Plasma
Protection layer UV
UV
Barrier rib
UV
UV
Phosphor Dielectric Substrate Electrode
Electrode
Surface-discharge
Vertical-discharge
Figure 5.6 AC PDP: surface discharge and vertical discharge.
114
Introduction to Flat Panel Displays
In these two types of AC PDP, the protection layer protects against plasma bombardment on the upper plate (front plate). However, in both types, the phosphor in the lower plate (rear plate) undergoes direct plasma bombardment. To reduce the plasma bombardment of the phosphor, surface-discharge AC PDP allows plasma close to the surface of the upper plate. The UV irradiates the phosphor effectively although the plasma is close to the upper plate. Since the plasma is far from the phosphor, the surfacedischarge approach reduces damage to the phosphor caused by the plasma. Accordingly, the lifetime of the phosphor is increased. The major advantage of the vertical-discharge approach is that it can provide a higher resolution, but the operating lifetime is shorter than the surface-discharge type. Hence, surface discharge is the most popular AC PDP structure. Figure 5.7 presents the structure for surface-discharge AC PDP. Plasma is generated by alternating current. Phosphors of R (red), G (green) and B (blue) are adopted to produce bright R, G and B colors. The rib serves not only as a spacer between the upper and the lower plates but also to isolate R, G and B luminance against crosstalk. The rib height is typically hundreds of micrometers. A magnesium oxide (MgO) film is commonly used as a protective layer because it offers high resistance against ion bombardment and higher secondary electron emission. Additionally, MgO also has a high refractory temperature. These characteristics of MgO increase the lifetime of the PDP and reduce the operating voltage. The surface-discharge mechanism is commonly adopted and successfully maintains the plasma on the surface of the upper plate, such that it does not damage the phosphor that is in the lower plate. Accordingly, the display life is extended. A dielectric layer is required for AC operation and provides capacitance while the electrodes provide the energy to discharge the gas. In surface-discharge PDP, the arrangement of the electrode is critical. The electrode in the front plate acts as a sustain electrode in the sustain period of driving operation.9 It should be as transparent as possible because most of the area of the front plate is occupied by electrodes. Indium tin oxide (ITO) is commonly used as a transparent conductive material. However, the conductivity of ITO is not as high as that of a typical metal. A metal electrode with a relatively small line width is affixed to the ITO electrode so the electrode conductivity is increased but the open ratio of the front plate is not markedly reduced. The metal electrode on the ITO electrode is called the auxiliary electrode or the bus electrode. The materials commonly used are chromium (Cr) and copper (Cu) with a Cr/Cu/Cr structure. Sustain auxiliary electrode (Cr/Cu/Cr)
Sustain electrode (ITO) Light
Substrate Front plate
Dielectric
Plasma
Protection layer (MgO) Light UV
UV
Rear plate
Barrier rib Phosphor Dielectric Substrate
Address electrode (Ag) Figure 5.7 Typical structure of surface-discharge AC PDP.
Plasma display panels
115
In this structure of auxiliary electrode, copper is the main material which increases the conductivity, while chromium is the material that performs an adhesion function to the copper layer. The electrode in the rear plate acts as the address electrode in the address period of driving operation. Silver is commonly used for this electrode. Most importantly, AC PDP uses alternating current to produce plasma while a dielectric layer that provides capacitance is also required. The protection layer is adopted to increase the endurance of plasma bombardment and to minimize phosphor damage caused by plasma bombardment. The surface-discharge structure provides the mechanism by which the plasma moves away from the phosphor while UV light remains able to excite the phosphor. Hence, the protection layer and the mechanism of surface discharge significantly and effectively increase the display lifetime.
5.3.3 Panel processes A PDP comprises front and rear plates. The panel process involves the front plate and rear plate, and includes the assembly and aging processes. Figure 5.8 presents a typical process for AC PDP. The major process in the front plate involves the protection layer, and the two major processes in the rear plate are the barrier rib and phosphor processes. The assembly and aging processes involve panel alignment, vacuum-tight sealing and discharge stabilization. Screen printing,10 photolithography and sandblasting are the three major approaches for forming these layers in the PDP process. Screen printing is the major approach. This approach can be adopted for the layers of electrode, dielectric, barrier rib and phosphors. Figure 5.9 presents the side view of a typical screen-printing process. The screen mask, paste and printing machine (including squeezee) are the three major components for screen printing.11 The perfect paste can easily pass through a screen mask when a shear force is applied to the paste. However, the paste of the pattern becomes solid and remains still when no shear force is applied and so does not diffuse and the original size of the pattern is maintained. After the paste has been deposited, drying and firing processes are required. Drying removes the solvent: the process temperature is typically under 150 ◦ C. The critical aspect of the drying process is the uniformity of drying from the outer surface to the inner core and from the edge to the center of the paste layer. The firing process is adopted to remove the binder from the paste and to melt the particles. The process temperature typically exceeds 300 ◦ C in the debinding process and 500 ◦ C in the firing process. Figure 5.10 presents a typical process temperature profile as a function of the process time. The critical part of the firing process is to remove the binder completely without causing stress in the cooling stage. The temperature and time profile are critical for complete debinding. Permanent
Front plate process
Rear plate process
Assembly and aging processes
Figure 5.8 Typical panel processes for AC PDP.
116
Introduction to Flat Panel Displays
Squeezee moving Screen mask
Screen frame
Squeezee Paste
Space
Printed pattern
Substrate
Figure 5.9 Side view of screen printing.
Temp (°C) 600 500 400
Step Cooling
Firing Debinding
300 Continuous Cooling
200 100 0 0.0
0.5
1.0
1.5
2.0 Time (hr.)
Figure 5.10 Typical temperature profile in the firing process.
deformation of the substrate occurs during the cooling stage. Hence, step cooling in the cooling stage releases the stress at the beginning of the cooling stage, markedly reducing substrate deformation. Like screen printing, photolithography is used commonly in display and semiconductor processes. This process uses a photoresist to produce a pattern. This photoresist pattern is used to etch the desired material which is deposited before photoresist formation. After the material is etched, the photoresist is stripped and a desired pattern of a material layer is formed. Sandblasting is an important approach and adopted especially in PDP barrier rib formation. The control of sandblasting particle uniformity is a critical issue, and markedly affects the sandblasting quality. Accordingly, the sandblasting apparatus must have not only a sandblasting function but also the ability to separate small and large particles to form particles of regular size. Figure 5.11 shows a typical sandblasting apparatus. Sand is supplied and blasted to the substrate through the nozzles. After the substrate has been blasted, the sand is returned through the sand separator. Larger particles are collected in the separator whereas the small particles are collected in the cyclone apparatus. Following the separation, this sand is purified and sent for blasting.
Plasma display panels
117
Speed measurement Cyclone
Nozzles
Small particle
Particle filter Input sand Substrate
Return sand
Sand separator
Sand supplier
Figure 5.11 Typical sandblasting apparatus.
5.4 Front plate techniques The front plate comprises substrate, electrode, dielectric and protection layers. Figure 5.12 shows a typical front plate process flow. The major functions are to have a gas discharge and to display an image. An effective electrode structure is a major factor in increasing a high open ratio while a long operation lifetime depends mainly on the performance of the protection layer. Front glass Front glass
Sustain electrode Sustain electrode Sustain auxiliary electrode Sustain auxiliary electrode Dielectric layer Dielectric layer Protection layer
Protection layer
Figure 5.12 Typical front plate process flow.
118
Introduction to Flat Panel Displays
5.4.1 Substrate Sodalime glass is a commonly used PDP substrate.12 The thickness of such a substrate is typically 2.8 mm. The major issue concerning a PDP substrate is its thermal expansion, which is associated with the high temperature of the process.13 The thermal expansion may cause an error in the image position.14 The high strain point of sodalime glass can be used to eliminate the thermal expansion of the substrate and control the error of the image position. Table 5.1 compares high-strain-point sodalime glass with normal sodalime glass. In this table, the strain point of the high-strain-point sodalime glass is 570 ◦ C, which is 64 ◦ C above the strain point of the normal sodalime glass. Due to the characteristics of high-strain-point sodalime glass, the thermal expansion can still be controlled when higher process temperature is used. The floating process of sodalime, presented in Figure 5.13,15 is commonly adopted. At the beginning of this process, raw material is placed in a melting furnace. Molten glass is produced and floated on the surface of molten metal, such that the molten glass has a smooth surface. Glass is formed after annealing. The floating process is a traditional process in glass formation and is cost-effective.
5.4.2 Sustain electrode The electrode in the front panel provides the energy to discharge the gas and to sustain the discharge. Accordingly, this electrode is called the discharge electrode or the sustain electrode. For a surfacedischarge PDP, the structure of this electrode is relatively critical. This electrode should be transparent so as not to block out emitted light. ITO and tin oxide (SnO) are the materials that are commonly adopted for transparent electrodes. Although SnO can withstand heat better than ITO, patterning SnO by chemical etching is difficult. Accordingly, ITO is the most popular material used in PDP transparent electrodes. A thickness of typically 150 nm yields a typical conductivity of 20 sq.−1 and a transmittance of over 80 %. The electrode conductivity required in PDPs is high, especially in large display applications. Since the conductivity of ITO is not as high as that of a typical metal, a conductive metal is used as the bus electrode to increase ITO conductivity. This conductive metal is designed with a relatively small line width and is Table 5.1
Comparison of high-strain-point sodalime glass with normal sodalime glass.
Glass
Normal High strain point a
Density (g cm−3 ) CTEa (× 10−7 Strain point (◦ C) Annealing point (◦ C) Softening point (◦ C) ◦ −1 C ) 2.49 2.76
85 83
506 570
545 620
Coefficient of thermal expansion.
Input raw materials
Melting furnace Floating bath
Molten glass
Molten metal
Annealing
Glass
Figure 5.13 Typical floating process for glass formation.
726 830
Plasma display panels
119
affixed to the ITO electrode to increase the electrode conductivity without markedly increasing the light open aperture of the front plate. Copper is commonly used as the conductive metal. Since copper does not easily adhere, a chromium layer is used to promote the adhesion before and after the copper layer is formed. This metal electrode (Cr/Cu/Cr) on the ITO electrode is called the auxiliary electrode or the bus electrode.16 In practical cases, the ITO electrode width is about 300 m and the discharge gap (space between ITO electrodes) is about 100 m. The Cr/Cu/Cr electrode width is about 100 m. ITO and Cr/Cu/Cr are typically formed by sputtering. Both ITO and Cr/Cu/Cr are patterned by photolithography. Figure 5.14 shows a typical electrode pattern formed photolithographically.
5.4.3 Dielectric The dielectric layer used in the front plate provides capacitance when AC power is applied. This layer must have a smooth surface and be bubble free so that the film surface during the MgO process is smooth. Additionally, this layer must have high transparency and be an effective insulator so that the panel can output more light with a lower leakage current.17 This dielectric layer is typically formed by screen printing with a thickness of about 30 m.
5.4.4 Protection layer This layer must be robust since it is in a high-temperature environment and under ion bombardment. It must also be associated with high secondary electron emission to reduce the discharge voltage.18 Film
Film deposition
Photoresist
Photoresist coating
Photo mask UV exposure
Photoresist pattern
Photoresist developing
Film pattern
Film etching
Photoresist removing
Figure 5.14 Typical electrode process using photolithography.
120
Introduction to Flat Panel Displays
Furthermore, the high transparency of this layer is important for ensuring high light output. Various materials such as CeO2 , La2 O3 , MgO and others are candidates for this layer.19 However, MgO is the most commonly used as the protection layer because it is not only a high-temperature refractory material with high secondary electron emission and high transparency but also can endure ion bombardment. These characteristics of MgO help increase the operating lifetime of PDPs and reduce the operating voltage. The thickness of the MgO layer is typically 600 nm. Although many approaches for MgO formation, such as reactive sputtering, ion plating, electrophoresis deposition and e-beam evaporation,20 23 are available, e-beam evaporation is the most commonly adopted because a film obtained by e-beam evaporation has a generally acceptable quality. A longer life and lower operating voltage can be achieved when an MgO film of higher quality is used. The shortcoming of the e-beam approach is the relative expense of the evaporation source and the need for apparatus with high e-beam voltage. Hence, many further actions such as the co-evaporation or the use of a reactive gaseous dopant to improve the quality of MgO films have also proposed in the last few years. However, the improvements are not completely effective.24, 25
5.5 Rear plate techniques The rear plate comprises the substrate, electrode, dielectric, barrier rib and phosphor layers. Figure 5.15 shows a typical rear plate process. The major function of the rear plate is to provide discharge and generate light. The barrier rib serves not only to maintain the space between the front plate and the rear plate but also to prevent crosstalk among cells. Its formation is a critical process in PDP fabrication. Effective formations of the barrier rib and phosphors are the most important factors in achieving high optical efficiency and long operating lifetime.
Rear glass Rear glass Address electrodes Address electrodes Dielectric layer Dielectric layer Barrier rib Barrier rib
Phosphor layer Phosphor layer
Figure 5.15 Typical rear plate process flow.
Plasma display panels
121
5.5.1 Substrate The substrate in the rear plate is similar to that in the front plate. However, the substrate in the rear plate typically requires an additional exhaustion hole for pumping out normal air from the panel and pumping discharge gas into the panel.
5.5.2 Address electrode The electrode in the rear plate provides energy to discharge the gas. Accordingly, this electrode is called the discharge electrode or the address electrode. Silver paste is commonly used in this electrode with a thickness of 10 m. The address electrode is traditionally processed by screen printing, as shown in Figure 5.16. The silver paste is screen-printed through a patterned screen mask. The pattern is formed onto the substrate and then dried and fired. A firm, solid pattern of the electrode is therefore obtained. Since the resolution of screen printing is limited, an alternative approach that uses photosensitive paste is adopted. This approach deposits a photosensitive silver paste on the substrate and then uses photolithography to produce a pattern. Figure 5.17 shows this approach. In the photosensitive paste approach, a line width of as small as 20 m can be achieved. In terms of higher resolution, it is a great advantage compared with 50 m which is obtained using the screen-printing approach. However, screen printing remains a common approach for forming address electrodes because of its lower material cost and simpler process steps.26
5.5.3 Dielectric The dielectric layer in the rear plate provides capacitance when AC power is applied. Unlike the dielectric in the front plate, this dielectric layer is not required to be transparent. However, a highly insulating layer is required, to reduce the leakage current. In practical use, the dielectric layer is formed typically by screen printing to a thickness of 20 m.
Screen mask
Screen printing
Drying
Firing
Figure 5.16 Typical screen-printing process used in the formation of the address electrode.
122
Introduction to Flat Panel Displays
Film
Film deposition
Photoresist
Photoresist coating
Photo mask
UV exposure
Photoresist pattern
Photoresist developing
Film pattern
Film etching
Photoresist removing
Figure 5.17 Typical photosensitive paste process used in the formation of the address electrode.
5.5.4 Barrier rib The barrier rib process is one of the most important processes. The barrier rib serves mostly as a spacer between the upper and lower plates.27 Moreover, the side walls of the barrier rib provide additional surfaces on which phosphor deposits, increasing the area of phosphor. The brightness increases with the additional area of phosphor. To reduce the difference between the gas pressure inside and outside the panel, the gas pressure of the panel does not exceed 1 atm (760 torr) and is typically set to 500 torr. At this gas pressure, the rib height is about 150 m for a typical gas mixture of neon and xenon. The barrier rib should be as thin as possible to enable a larger aperture ratio or higher resolution of the display cell.28 However, the technology used has limitations as regards reducing the width. In practical cases, the barrier rib is formed by screen printing and the width is about 60 m.29 Figure 5.18 shows a typical screen-printing process for barrier rib formation. Barrier rib formation typically depends on multiple screen printings to yield the required height. The critical purpose of the first layer is to ensure adhesion between the first layer and the substrate. As the final layer, it must have a smooth top surface to prevent cracking when the front plate and rear plate are sealed.30 Since screen printing must be performed many times and the printing approaches are complex, an alternative approach to forming the rib, sandblasting, is sometimes adopted.31 Sandblasting provides a higher pitch resolution and rib height than screen printing. The rib shape can be more easily controlled and the process is simpler than screen printing. Therefore, ribs are commonly formed by sandblasting, as shown in Figure 5.19.
Plasma display panels
123
Screen mask
1st time screen printing
Drying Screen mask 2nd time screen printing
Firing
Figure 5.18 Typical screen-printing process for forming the barrier rib.
Paste
Photoresist
Paste and photoresist deposition Photo mask
UV exposure
Photoresist developing
Sandblasting
Photoresist removing
Figure 5.19 Typical rib formation by sandblasting.
124
Introduction to Flat Panel Displays
Paste Paste deposition Press
Mold Mold pressing
Mold removing
Figure 5.20 Typical rib formation using a press mold.
Although sandblasting is effective at forming a barrier rib, the process is relatively complex and the sand may pollute the environment. Therefore, a press mold is sometimes used to as a simpler and cleaner process than sandblasting. Figure 5.20 shows this process. The press mold approach uses a patterned mold to press on the deposited paste. After the mold has been removed, a pattern of the paste is formed on the substrate. The main shortcoming of this approach is the poor positional accuracy of the pattern. Additionally, a residue can be produced when the paste is pressed away. This residue may influence the discharge performance and uniformity. Table 5.2 compares these approaches. According to this table, sandblasting offers the best positional accuracy of the pattern but suffers from high production cost. The press mold approach has the shortest required process time but poorer positional accuracy and higher equipment cost. The screen printing approach has the lowest equipment cost but suffers from a long process time.
5.5.5 Phosphor R, G and B phosphors are used to generate bright R, G and B colors.32 Unlike in cathode ray tubes, which use electron-excited phosphors, in PDPs the phosphors are UV-excited.33 PDP phosphors are excited by 147 nm UV light, which is emitted by xenon gas. The compositions of R, G and B phosphors are Y0.65 Gd0.35 BO3 :Eu, BaAl12 O19 :Mn and BaMgAl14 O23 :Eu, respectively.34 The decay times of these R, G Table 5.2
Comparison of approaches for barrier rib formation.
Process Screen printing Sandblasting Press molding
Position accuracy
Process time
Equipment cost
Production cost
• ◦
◦ •
• ◦
◦ ◦
Key: •, very good; ◦, good; , average.
Plasma display panels
Table 5.3
125
Characteristics of R, G and B phosphors.
Phosphor Red Green Blue
Brightness
Color purity
Less degradation
Less decay
• • ◦ ◦
◦ • • •
• • ◦
◦ ◦ ◦
Y0.65 Gd0.35 BO3 :Eu Zn2 SiO4 :Mn BaAl12 O19 :Mn BaMgAl10 O17 :Eu
Key: •, very good; ◦, good; , average.
and B phosphors are 9, 17 and <1 ms, respectively. Since the decay time of G phosphor, BaAl12 O19 :Mn, is a little longer than those of R and B phosphors, a new G phosphor, Zn2 SiO4 :Mn, with a decay time of less than 14 ms has been developed. Table 5.3 compares the characteristics of these phosphors. The color purity of the R phosphor, the decay time of the G phosphors and the degradation of the B phosphor must be further improved.35, 36 The degradation of the B phosphor is caused by oxidation of europium during the panel process.37 39 Phosphor brightness will be slightly reduced when a PDP is operating because ions of plasma sputter the phosphor. Outgas from the barrier rib and the other layers of the device may contaminate the phosphor and result in further phosphor degradation. When the phosphor is degraded, the brightness is reduced. More outgas undesirably originates from the barrier rib because of the debinding process of the barrier rib. The PDP phosphor layer is typically formed by screen printing and the thickness of each color in the RGB is typically 30 m. Figure 5.21 presents the typical screen-printing process of the phosphor layer. In this process, phosphor paste is deposited and initially fills the cavity of the barrier ribs. The next process of the phosphor plate is followed by drying and firing. Drying removes solvent material and firing removes binder material. The final shape of the phosphor layer must provide a large discharge space with a
Phosphor
Fill in phosphor
Solvent Remove solvent
Binder Remove binder
Form phosphor layer Figure 5.21 Typical screen-printing process for forming a phosphor layer.
126
Introduction to Flat Panel Displays
sufficiently thick phosphor layer. However, the phosphor layer must be uniformly distributed along the wall of the barrier rib and the dielectric area. Therefore, paste preparation and the process conditions are critical.
5.6 Assembly and aging techniques Assembly involves sealing layer formation, panel alignment, sealing and gas filling. This process binds the front plate and rear plate into a display panel with good alignment, vacuum-tight sealing and a cleaned cell.40 Following assembly, an aging process is required. The purpose of this aging process is to expose defects and stabilize the quality of the display.41 Figure 5.22 shows a typical assembly process. At the beginning of this process, a sealing layer is deposited onto the surrounding area of the rear plate. After the sealing layer is deposited, the front and rear plates are aligned and then sealing and aging are performed.
5.6.1 Sealing layer formation and panel alignment A material for the sealing layer used in PDPs is required to achieve a vacuum-tight seal and be outgas free after the sealing process. Additionally, the material also requires a low melting point, which is compatible with a glass substrate. Typical materials are glass frits or glass powder.42 The use of epoxy material, commonly used in LCDs, as a sealing layer is not appropriate in PDP processing since outgas usually occurs from the epoxy material after the sealing process. This outgas can contaminate the display cell and degrade the performance of the display. The first step of the sealing formation process is to dispense a sealing layer onto the area that surrounds the front plate or the rear plate. Sealing layer
Sealing layer
Panel alignment
Light Plasma
Sealing and aging
Figure 5.22 Typical assembly and aging process.
Plasma display panels
127
The next step is to align the front plate and the rear plate. This alignment accuracy is critical and affects the display quality. The major challenge in the alignment is position shift during high-temperature sealing. The front plate and rear plate are typically clamped and fixed using clippers so that they do not shift during sealing. After the front and rear plates have been aligned and clipped, an additional sealing paste is deposited onto the area around the small opening of the rear plate. Then, an exhausting tube is placed onto the deposited area of the small opening hole in the rear plate. An alternative approach does not use an exhausting tube: this approach is called tubeless sealing. The benefits of the tubeless sealing approach are faster evacuation and a thinner panel.43 The evacuation is faster because the panel vacuum is obtained from a ready vacuum of a whole chamber and this chamber vacuum can pump down before the panel process. The panel can be thinner when a tip-off tail of the exhaust tube is absent. However, the gas given off by the sealing paste using tubeless sealing can be sealed inside of the panel and then contaminate the panel. Tubeless sealing demands extra work to clean the contamination inside of the panel.44
5.6.2 Sealing, gas purging and display gas filling The sealing and gas processes follow panel alignment. Figure 5.23 presents typical sealing and gas processes.45 The sealing process melts the material of the sealing layer to bind the front plate to the rear plate permanently.46 The gas given off during the sealing process can contaminate the surface and protection layers of the display cell. Because of the thermal characteristics of the glass frits and glass powder, sealing is typically performed at high temperature. The typical sealing process temperature is 450 ◦ C. Clipper selection is important since the clipper may lose binding strength during sealing. With improper selection Temp(°C)
Annealing
Sealing
500 400 300 200 100 0
0
2
4
6
8
10
12
14
16
Evacuation
Gas pressure (torr)
18 20 Time (hr.)
Evacuation and aging gas in 100
1 × 10–7 Discharge Purge gas in 0
2
4
6
8
10
Display gas in and tip off 12
14
Figure 5.23 Typical sealing and gas processes.
16
18 20 Time (hr.)
128
Introduction to Flat Panel Displays
of clipper, the panel alignment may be shifted during sealing. Moreover, the glass substrate may crack during sealing at high temperature. Since the display cell must have a clean environment, a vacuum process to evacuate all gas from the display cell must be performed before display gas filling.47, 48 This vacuum process is to remove impurities such as H2 , O2 , N2 , CO2 and CO from air, and the phosphor and other layers of the panel. These impurities can influence the operating voltage and the brightness of the panel. In addition to the physical vacuum pumping, chemical getting is adopted in PDP processing to assist the vacuum pump and adsorb these impurities. The getter is the main material in a chemical pump.49 This getter is a material that can absorb gas to yield a vacuum.50 Radiofrequency heating is commonly adopted to activate and heat the getter but not the glass substrate. Interestingly, improper heating of the getter can crack the substrate. The additional function of the getter is to absorb impurities which sometimes are poison gases. Since the poison gases may be emitted into the display cell of the sealed panel, this poison gas must be removed from the sealed panel. The getter can act as a poison gas absorber and maintain a clean gas environment in the display cell. Getters are of two types: an evaporation getter (EG) and nonevaporation getter (NEG). The EG type has a larger effective area for reaction with gas but the evaporated material may evaporate onto an undesired area and contaminate the device. A vacuum level of typically 10−7 torr is required. After this vacuum level is reached, the display cell is filled with the purge gas, which is evacuated so that the display cell is cleaned. After the purging has been completed, the display cell is filled with display gas and the exhausting tube is tipped off. Tipping off is the process of cutting off and isolating the inside and outside of the panel using a tip end. Neon and xenon are the gases that are typically used as display gases. The amount of display gas is very important since it determines the UV intensity, the required discharge voltage and the discharge efficiency.51 53 The tipping-off of the exhausting tube must be performed carefully so that gas given off during the tip-off process can be evacuated out and not left inside of the panel. Therefore, a preliminary tip-off process is typically needed to evacuate the giving-off gas so that only very little given-off gas is left inside the panel. In order to perform the tip-off process, electric heating or gas heating of the exhausting tube is commonly used. Electric heating uses electricity to heat a ceramic and the heated ceramic is used to tip off the exhausting tube, while gas heating uses gas directly to tip off the exhausting tube. Most importantly, electrical heating needs more apparatus than gas heating although electrical heating is relatively controllable.
5.6.3 Aging The purpose of aging is to expose defects. Defects or contamination on the MgO surface, dielectric and electrode are revealed during the aging process.54 A defect in an electrode can be open, short or anything in between. Additionally, aging can stabilize and reduce the operating voltage since it can polish or smoothen the MgO surface, remove surface contamination from MgO at the discharge site, and emit some gases from the MgO surface.55 57
5.7 System techniques AC PDP circuits provide power supply, signal process, scan/data driving and energy saving functions.58 Figure 5.24 shows a system block diagram. In this diagram, the plasma display panel is driven by a driver circuit while the signal processing circuit provides a video signal to the panel. The circuit of the energy saver is adopted to collect and reuse the energy. This mechanism is particularly useful for AC PDP devices. Since the PDP must be operated at a few tens or a few hundreds of volts, a high-voltage driver IC must be used. The operating mechanism of the display cell that drives an AC PDP is described below.
Plasma display panels
129
Power supply
Signal Process
Driver
Energy saver
Plasma display
Figure 5.24 Typical AC PDP system block diagram.
5.7.1 Cell operation mechanism An AC PDP typically requires firing to begin the first discharge in each frame. When the first applied voltage of each frame is less than the firing voltage (V f ) the first time, the cell cannot discharge, even when the applied voltage exceeds the sustaining voltage (V s ), as shown in Figure 5.25(a). When a voltage above V f is applied, the gas in the cell is discharged into positive ions and negative ions. These positive ions and negative ions are recombined to generate UV radiation, as shown in Figure 5.25(b). Once the Electrode
+
+
+
_
_ _
+
+
Dielectric
(a) V < Vf Positive ion
+
+
+
UV _
_ _
_ _ _
_
(b) V > Vf _ Negative ion
(c) V > Vs +
+ + _
(d) V < Vs + Figure 5.25 Typical operating mechanism of an AC PDP cell.
130
Introduction to Flat Panel Displays
gas is discharged, the applied voltage need not exceed the firing voltage (V f ) but need equal only the sustaining voltage (V s ). However, the electrical polarity must alternate so that the discharge can continue, as shown in Figure 5.25(c). To proceed to the end stage of each frame, the cell must erase the discharge using a low voltage with opposite polarity to that of the existing voltage. The discharge is gradually reduced. Eventually, the discharge in the cell is completely erased, as shown in Figure 5.25(d).59
5.7.2 Driving Amplitude and time modulation are two popular approaches for driving displays. Since a plasma device has a memory characteristic, the use of time modulation in a PDP device has an advantage. However, many approaches for time modulation such as ADS (address display separation) and AWD (address with display) can be used.60 Since the driving waveform and circuit for ADS are relatively simple, AC PDPs commonly use ADS time modulation. Each field has many subfields. The desired gray levels govern the numbers of subfields. The driving waveform of each subfield comprises an address period and a sustain period, as shown in Figure 5.26.61 The address periods in all subfields are the same while the sustain periods vary among the subfields. A typical address period comprises erasing, priming, erasing and writing subperiods.62 The writing subperiod is determined by the time required for each scan line and the number of scan lines in the display panel. The typical address period required for each scan line is microseconds and the number of scan lines is 480 for a display panel of the VGA format. Since the gray levels are determined by the sustain period, various sustain periods in each subfield generate various gray levels. The sustain periods in the subfields are typically arranged as 1, 2, 4, 8, 16, 32, 64 and 128 units of time. Figure 5.27 shows a typicalADS driving modulation with 256 gray levels (8 Bits). Modulating these subfields yields gray levels from 0 to 255. An image of 256 grey levels is therefore obtained, as shown in Figure 5.28. This image comprises eight subfields. These subfields have the subfield gray levels 1, 2, 4, 8, 16, 32, 64 and 128, respectively.
5.7.3 Energy saving As well as the driving mechanism and circuit, an energy recovery circuit is typically adopted in AC PDPs, as shown in Figure 5.29.63 In this circuit, a capacitor is used to store the energy while an inductor is used to conduct and protect the circuit during AC operation. When this circuit is used, energy can be effectively collected and reused. Sub-field (SF) Address period Priming Erase Erase
Sustain period
Write
Address electrode Address pulse Sustain electrode 1
Scan pulse Sustain pulse
Sustain electrode 2 Figure 5.26 Typical driving waveform of AC PDP.
Plasma display panels
131
1 Field (16.6 ms)
SF1 SF2
SF3
SF4
SF5
SF6
SF7
SF8
Sustain period Address period 1
2
4
8
16
32
64
128
(Gray level 1)
(Gray level 128)
Figure 5.27 Typical ADS driving modulation with 256 gray levels.
Gray level 1 Gray level 2 Gray level 4 Gray level 8 Gray level 16 Gray level 32 Gray level 64 Gray level 128
Image with 256 gray levels
Figure 5.28 Image of 256 gray levels.
Transistor
Power supply
Diode Capacitor
Inductor PDP panel
Figure 5.29 Typical energy recovery circuit.
132
Introduction to Flat Panel Displays
5.7.4 PDP issues The false contour issue is a traditional one concerning picture quality in PDPs.64 67 This false contour possibly arises between adjacent pixels when a moving picture is displayed. Many approaches have been proposed to reduce the false contour.68 72 General approaches for reducing the false contour are to divide the most significant bits into subfields and to rearrange the subfield order from most significant bit (MSB) to least significant bit (LSB). Image sticking is another issue as regards PDP quality.73, 74 Image sticking can occur when a static image is displayed for up to tens of minutes. This phenomenon is an important issue that affects PDP image quality. Many approaches have been proposed over the last few years to eliminate the effect.75 In addition to false contours and image sticking, PDPs are also associated with issues of electromagnetic interference (EMI) and infrared (IR) emission. EMI is associated with PDP electronic circuits and PDP panels while IR emission is associated mainly with PDP panels. To reduce these effects, filter glass is required to provide EMI shielding and IR blocking.76 IR is typically generated from the R phosphor of the PDP panel. This IR emission possibly interferes with other appliance remote controls when a PDP is used. ITO flood film, metal mesh film and conducting wire mesh can be used in separate methods for making filter glass. The ITO flood film offers the best transparency. However, its wave shielding is the poorest.
Homework problems 5.1 What is the minimum firing voltage for a gaseous mixture of neon + 0.2 % argon? If the discharge distance is 100 m and the gas pressure is 200 torr, what is the estimated firing voltage? 5.2 Calculate the expansion of normal sodalime glass when a substrate of sodalime glass has dimensions 370 mm × 470 mm and is processed at 400 ◦ C. 5.3 For VGA video format (640 × 480) with a grey level of 256, calculate the required writing period if the writing pulse width is 2 s. 5.4 Given a barrier rib width of 80 m and a luminance area width of 120 m for a red pixel, calculate the open ratio and the pitch size of the red pixel. 5.5 If the brightness of a PDP panel is 1000 cd m−2 and the background glow of the PDP panel is reduced from 5.0 to 0.5 cd m−2 , describe the effect on the performance of the PDP panel. Calculate the dark room contrast ratio for background glows of 5.0 and 0.5 cd m−2 .
References 1. Kuriyama, H., Kanada, H., Chiaki, Y. et al. (2004) High-performance 55-inch diagonal WXGA PDP with extended ALIS technology. SID Dig., 1026. 2. Oversluizen, G. and Dekker, T. (2006) High efficacy PDP design. SID Dig., 1110. 3. Hirakawa, H., Shinohe, K., Tokai, A. et al. (2004) Dynamic driving characteristics of plasma tubes array. SID Dig., 810. 4. Sano, Y., Nakamura, T., Numomura, K. et al. (1998) High-contrast 50-in color AC plasma display with 1365 × 768 pixels. SID Dig., 275. 5. Kruithof, A.A. and Penning, F.M. (1937) Determination of the Townsend ionization coefficient α for mixtures of neon and argon. Physica, 4, 430. 6. Weber, L.F. (1985) Plasma displays, in Flat-Panel Displays and CRTs (ed. L.E. Tannas Jr), Van Nostrand Reinhold, New York, p. 332. 7. Matsumoto, S. (1990) Plasma display panels, in Electronic Display Devices (ed. S. Matsumoto), John Wiley & Sons, Inc., New York, p. 131.
Plasma display panels
133
8. Amano, Y., Endo, J. and Velayudhan, B.K. (1998) A new driving method for vertical discharge PDP. International Display Workshops ’98, p, 551. 9. Choi, K.C., Cho, K.H., Lee, S.M. et al. (2007) High efficient discharge mode in an AC PDP with an auxiliary electrode. SID Dig., 1530. 10. Reimer, D.E. (1988) Analytical engineering model of the screen printing process: I. Solid State Technol., August, 107. 11. Sakamoto, S. and Ogawa, Y. (1995) Screen printing for fabrication of PDPs. International Display Workshops ’95, p. 41. 12. Dumbaugh, W.H. (1992) Status and future directions of flat-panel-display substrates. SID Dig., 805. 13. Maeda, K., Nishizawa, M., Nakashima, T. and Nakao, Y. (1997) Thermal compaction of PDP glass substrates. SID Dig., 544. 14. Ford, P.W. and Veyhl, E.W. (1993) Image position errors due to plate bending. SID Dig., 983. 15. Summer, R.C. (1995) Measurements of fusion and float-glass plate flatness using a Fizeau interferometer. SID Dig., 442. 16. Tachibana, K. (2006) Design and performance of AC-PDP cells with auxiliary electrode structures. SID Dig., 1205. 17. Yokoe, M., Ohno, S., Shenda, S. and Nakayama, K. (2001) Firing of dielectric layer in vacuum for high transmittance. Asia Display ’01, p. 805. 18. Andoh, S., Murase, K. and Umeda, S. (1976) Discharge-time lag in a plasma display-selection of protection layer (γ surface). IEEE Trans. Electron Dev., 23, 319. 19. Urade, T., Iemori, T., Osawa, M. et al. (1976) A protecting layer for the dielectric in AC plasma panels. IEEE Trans. Electron Dev., 23, 313. 20. Park, K.C. and Weltzman, E.J. (1978) E-beam evaporated glass and MgO layers for gas panel fabrication. IBM J. Res. Develop., 22, 607. 21. Hakamori, M., Hibino, Y., Matsuzaki, T. et al. (1997) High rate deposition of MgO film for AC-PDPs by activated reactive evaporation using hollow cathode discharge (HCD-ARE). International Display Workshops ’97, p. 551. 22. Ushio, Y., Banno, T., Matuda, N. et al. (1988) Secondary electron emission studies on MgO films. Thin Solid Films, 167, 299. 23. Ko, M.S. and Kim, Y.S. (2007) Characteristics of MgO nano-sized powder layer via electrophoresis deposition as electron emission layer for ac-PDPs. SID Dig., 1438. 24. Yoshida, K., Uchiike, H., Zhang, S. et al. (1997) Improvement of performance and aging characteristics of color plasma displays by using co-evaporated protection layer. International Display Workshops ’97, p. 559. 25. Ahearn, W.E. and Sahni, O. (1978) Effect of reactive gas dopants on the MgO surface in AC plasma display panels. IBM J. Res. Develop., 22, 622. 26. Hayakawa, H., Sakuda, K., Kuwada, R. et al. (1997) Photoimageable thick film black conductor system for FPD. International Display Workshops ’97, p. 547. 27. Whang, K.W., Bae, H.S., Lee, K.H. and Kim, T.J. (2005) The effect of cell geometry and plasma loss on the luminous efficiency in ac plasma display panel. SID Dig., 1130. 28. Ryu, S.M., Han, M., Yang, D.Y. et al. (2007) Ultra-slim barrier ribs for plasma display panel by X-ray lithography process. SID Dig., 1205. 29. Sakamoto, S. and Kato, K. (1994) A screen-printing process for the fabrication of plasma display panels. Display Manufacturing Technology Conference, p. 127. 30. Fischer-Cripps,A.C., Collins, R.E, Turner, G.M. and Bezzel, E. (1995) Stress and fracture probability in evacuated glazing. Building Environ., 30, 41. 31. Fujii, H., Tanabe, H., Ishiga, H. et al. (1992) A sandblasting process for fabrication of color PDP phosphor screens. SID Dig., 728. 32. Yacobi, B.G. and Holt, D.B. (1994) Luminescence Phenomena. Cathodoluminescence Microscopy of Inorganic Solids, Plenum Press, New York/London, p. 21. 33. Vecht, A. (1994) Advances in phosphor materials for display application. IDRC ’94, p. 86. 34. Shionoya, S. (1995) Luminescence mechanism of phosphors for displays. International Display Workshops ’95, p. 63. 35. Tanner, H., Vecht, A. Smith, D.W. et al. (1995) High-resolution phosphors: characterization and assessment. SID Dig., 623. 36. Greer, J.A., Vanhook, H.J., Nguyen, H.Q. et al. (1994) Thin-film phosphors prepared by pulsed-laser deposition. SID Dig., 827.
134
Introduction to Flat Panel Displays
37. Kajiyama, H., Tanno, H., Shinoda, T. et al. (2007) Lifetime improvement of Eu-doped BAM by plasma treatment. SID Dig., 1321. 38. Ushirozawa, M. (2000) Luminance degradation of blue phosphor BaMgAl10 O17 :Eu for PDP. SID Dig., 224. 39. Zhang, S. (2005) Recent development of blue phosphors for PDP application. SID Dig., 1142. 40. Reisman, A. (1978) Single-cycle gas panel assembly. IBM J. Res. Develop., 22, 596. 41. Moine, B. and Bizarri, G. (2007) Aging processes of the blue phosphor in plasma display panels . SID Dig., 1317. 42. Roth, A. (1966) Permanent seals, in Vacuum Sealing Techniques, Pergamon Press, Oxford, p. 23. 43. Shimosato, Y., Seki, T., Yoshimura, Y. and Nagoshi, T. (1998) Cart pusher in-line exhausting system for PDPs. International Display Workshops ’98, p. 559. 44. Shimosato, Y., Seki, T. and Yoshimura, Y. (1997) Vacuum process for plasma display panel without exhaust pipe. International Display Workshops ’97, p. 539. 45. Park, C.S., Tae, H.S., Kwon, Y.K. et al. (2007) Discharge characteristics of 42-in AC plsama display panel fabricated by vacuum sealing method. SID Dig., 1434. 46. Alpha, J.W. (1976) Glass sealing technology for displays. Opt. Laser Technol., December, 259. 47. Zeng, S.Q., Hunt, A. and Greif, R. (1995) Mean free path and apparent thermal conductivity of a gas in a porous medium. Trans. ASME, 117, 758. 48. Clugston, D.A. and Collins, R.E. (1994) Pump down of evacuated glazing. J. Vac. Sci. Technol. A, 12, 241. 49. Kohl, W.H. (1980) Getter materials, in Handbook of Materials and Techniques for Vacuum Devices, Reinhold Publishing, New York, p. 545. 50. Caloi, R.M., Carretti, C. and Amiotti, M. (1997) Gettering technology in plasma display panels. International Display Workshops ’97, p. 543. 51. Jang, S.K., Tae, H.S., Jung, E.Y. et al. (2007) Influence of He contents on reset and address discharge characteristics under variable panel temperature in ac PDPs. SID Dig., 1629. 52. Lee, D.K., Choi, J.H., Cho, Y.S. et al. (2005) Influences of gas mixing ratio on the characteristics of plasma display panel in He–Ne–Xe gas system. SID Dig., 619. 53. Oversluizen, G., de Zwart, S., Dekker, T. and Gillies, M.F. (202) The route towards a high efficacy PDP: influence of driving condition, Xe partial pressure, and cell design. SID Dig., 848. 54. Park, M.S., Park, D.H., Kim, B.H. et al. (2006) Effect of aging discharge on the MgO protective layer of AC-plasma display panel. SID Dig., 1399. 55. Pleshko, P. (1981) AC plasma display aging model and lifetime calculations. IEEE Trans. Electron Dev., ED-28, 654. 56. Byrum Jr, B.W. (1975) Surface aging mechanisms of AC plasma display panels. IEEE Trans. Electron Dev., ED-22, 685. 57. Aboelfotoh, M.O. (1981) Aging characteristics of AC plasma display panels. IEEE Trans. Electron Dev., ED-28, 645. 58. Mas, C., Troussel, G. and Benoit, E. (2000) A new IC for generating AC power supply. SID Dig., 216. 59. Hirakawa, H., Katayama, T., Kuroki, S. et al. (1998) Cell structure and driving method of a 25-in. (64-cm) diagonal high-resolution color AC plasma display. SID Dig., 279. 60. Uchidoi, M., Saegusa, N., Sato, Y and Okano, T. (1996) Panel design and driving method for 40-in. diagonal AC plasma displays. International Display Workshops ’96, p. 291. 61. Kojima, T., Toyonaga, R., Sakai, T. et al. (1979) Sixteen-inch gas-discharge display panel with 2-lines-at-a-time driving. Proc. SID, 20, 153. 62. Mikoshiba, S. (1999) Advancements in plasma panels. Information Display, 2/99, 28. 63. Choi, J.P., Kim, T.H., Kim, H.Y. et al. (2001) Development of new energy recovery driving method for column PDP addressing. SID Dig., 1232. 64. Mikoshiba, S. (1995) Picture quality issues for color plasma displays. International Display Workshops ’95, p. 57. 65. Koura, T., Yamamoto, T., Ishii, K. et al. (1998) Evaluation of moving-picture quality on 42-in PDP. SID Dig., 620. 66. Yamaguchi, T., Matsuda, T., Kohgami, A and Mikoshiba, S. (1996) Degradation of moving quality in PDPs: dynamic false contours. J. SID, 4, 263. 67. Mikoshiba, S. (1996) Dynamic false contours on PDPs: fatal or curable? International Display Workshops ’96, p. 251.
Plasma display panels
135
68. Makino, T., Mochizuki, A., Tajima, M. et al. (1995) Improvement of video image quality in AC-plasma display panels by suppressing the unfavorable coloration effect with sufficient gray shades capacity. Asia Display ’95, p. 381. 69. Yamamoto, T., Takano, Y., Ishii, K. et al. (1997) Improvement of moving-picture quality on a 42-in. diagonal PDP for HDTV. SID Dig., 217. 70. Kawahara, I. and Wani, K. (1997) Simulation and reduction of motion picture disturbance for 42-in.-diagonal AC-PDP. International Display Workshops ’97, p. 503. 71. Zhu, Y.W., Toda, K., Yamaguchi, T. et al. (1997) A method-dependent equalizing-pulse technique for reducing gray-scale disturbances on PDPs. SID Dig., 221. 72. Ryeom, J., Kim, S.W., Roh, Y.B. and Park, C.B. (1998) An image data rearranged sub-field method for reducing dynamic false contours in PDPs. International Display Workshops ’98, p. 547. 73. Tae, H.S., Park, C.S., Kwon, Y.K. et al. (2007) Solution to boundary image sticking in AC plasma display panel. SID Dig., 1617. 74. Park, C.S., Tae, H.S., Kwon, Y.K. et al. (2006) Experimental study on halo-type boundary image sticking in 42-in AC plasma display panel. SID Dig., 1213. 75. Lee, H.J., Kim, D.H., Kim, Y.R. et al. (2004) Analysis of temporal image sticking in AC-PDP and the methods to reduce it. SID Dig., 214. 76. Zagdoun, G., Heitz, T. and Talpaert, X. (2004) New type of optical filter for PDP TV with improved durability. SID Dig., 918.
6 Light-emitting diodes
6.1 Introduction The operation principle of light-emitting diodes (LEDs) is to inject electrons and holes into a singlecrystal semiconductor where carriers recombine and generate photons. The emission wavelength of the LEDs depends on the material selection, active-layer design and device structure, and can cover the whole visible range. Although electroluminescence (EL) in a semiconductor (SiC) was first observed in the early twentieth century,1 the commercial realization of visible LEDs started around 60 years later. For higher recombination rate, a direct bandgap material (i.e. the minimum of the conduction band and the maximum of the valence band coincide at the k-space) is preferred. In the 1960s, visible LEDs emitting a red color were realized by using a GaAsP active layer on a GaAs substrate.2 One of the major problems of this system arises from the lattice mismatch between the GaAs (lattice constant = 5.65 Å) and GaAsP (lattice constant = 5.45 Å) which results in misfit dislocation and, in turn, to an increase in the nonradiative recombination rate. Emission wavelengths of GaAsP LEDs can be blue-shifted by increasing the percentage of phosphorus. However, a direct–indirect transition occurs when the phosphorus mole fraction increases to 50 %, which results in an efficiency decrease. By incorporating some isoelectronic impurities (N and Zn:O) into indirect bandgap materials (such as GaP or GaAsP), which form a trap level inside the forbidden bandgap, a shorter emission wavelength (orange, yellow or even green) of LEDs was achieved in the 1970s.3 Another material system is AlGaAs on GaAs substrate (1980), which has the advantage of low lattice mismatch since the lattice constants are nearly the same for AlAs (5.66 Å) and GaAs (5.65 Å).4 Hence, a high-quality crystalline epitaxial film can be obtained for reducing the nonradiative recombination rate. Also, the layer structure can be engineered (such as heterojunction and quantum well) easily for better carrier confinement and fine tuning of the emission wavelength due to the lattice-matched condition. On increasing the aluminum concentration, the emission spectrum shifts towards shorter wavelengths. However, a direct–indirect transition occurs at 621 nm with Al = 45 %, which limits the AlGaAs/GaAs LED to operation only at the red emission region. Then, in 1990, the quaternary alloy InGaAsP was introduced with the advantages of: (1) lattice matched to GaAs by tuning the Al/Ga ratio and (2) shorter direct–indirect transition wavelength (555 nm; yellowish green) than GaAsP and AlGaAs systems, which is currently the mainstream of the ‘long-wavelength’ region (red, orange and yellow) of visible LEDs.5 GaN, InGaN and AlGaN (III nitrides) are other material systems, which cover the shorter wavelengths of the visible spectrum from ultraviolet (UV) and blue to green. Two major problems of these material
Introduction to Flat Panel Displays c 2008 John Wiley & Sons, Ltd
J.-H. Lee, D.N. Liu and S.-T. Wu
138
Introduction to Flat Panel Displays
systems arise from the lack of a lattice-matched substrate and the difficulty in achieving p-type doping. By pregrowth of a buffer layer (e.g. AlN or GaN) at low temperature, it is possible to form an epitaxial nitride film on the lattice-mismatched substrate sapphire (Al2 O3 ).6 Acceptors (which should act as p-type dopants) in nitride material are easily passivated by the hydrogen atoms introduced during epitaxial film growth. Hence, a high-temperature annealing under nitrogen ambient is needed to activate the acceptors and achieve an ‘effective’ p-type nitride material.7 For generating white light, two- or three-color mixing is needed. Typical approaches include: (1) all-semiconductor method by using different color LEDs and (2) down-conversion materials (such as phosphors) optically pumped by short-wavelength LEDs (UV or blue). Figure 6.1 shows the evolution of LED performances over time. We can see that the power efficiencies of LEDs have made a huge progress over the last several decades and are already higher than those of incandescent and fluorescent lamps.8 The requirements for achieving high performance from a LED,9 are: (1) inject and transport the carriers into the active layer for recombination without injection barriers and with little ohmic loss, (2) have a much higher radiative than nonradiative recombination rate and (3) couple the light out of the device efficiently. A high resistance of the semiconductor results in higher voltage and lower efficiency. Ohmic contact, rather than Schottky barrier, is needed for efficient carrier injection. Hence, a p–n diode structure with low resistance is preferred for LED applications. Since the radiative recombination process in LEDs is of Langevin type, the recombination rate distribution is proportional to the product of the electron and hole concentrations in the spatial domain. Hence, it is needed to confine carriers in the active region by using heterojunction, electron blocking layer, quantum well, quantum wire and quantum dot structures. Once electron–hole pairs recombine, the energy may relax radiatively by generation of photons or nonradiatively by generation of heat. A single-crystalline semiconductor material is needed for LED applications since defects in semiconductors typically act as nonradiative centers which result in the decrease of the radiative recombination rate. The bandgap of the semiconductor material at the active layer determines the LED emission wavelength. Active layer design also plays an important role in emission spectra. Quantum well, quantum wire and quantum dot structures shift the emission wavelength from the bandgap to the quantized states which allows some flexibility in emission wavelength tuning. In addition, the densities of states of the low-dimensional structures (quantum well, wire and dot structures)
Figure 6.1 Power efficiencies of LEDs with time for long-wavelength (GaAsP, AlGaAs and AlInGaP) and shortwavelength (nitride) material systems.8
Light-emitting diodes
139
are different from that of the bulk material, which results in a change in the full-widths at half-maximum (FWHMs) and different temperature dependences. Once photons are generated in the active layers of LEDs, the next stage is to radiate the photons out of the semiconductor material efficiently and be received by the human eye. Due to the spontaneous emission characteristics, the photon emission is isotropic, and may experience absorption (by defect states), reabsorption (band-to-band or free carrier) or scattering (by defects or interfaces) when propagating in semiconductor materials. If the substrate of the LED has a smaller bandgap (e.g. GaAs with E g = 1.424 eV , corresponding to 870 nm in the near-infrared region) than visible light, the absorption from the substrate (theoretically 50 %) needs to be taken into consideration. When the photons propagate out of the semiconductor to the epoxy layer (which is used for packaging the LEDs), total internal reflection may occur due to the refractive index difference (n = 3.5 for semiconductor and 1.5 for epoxy). From a simple calculation using Snell’s law, light with an incident angle from the semiconductor greater than a critical angle, about 25◦ in this case, cannot radiate to the epoxy layer. Those photons will reflect back into the semiconductor which may be absorbed by the defects or substrate, or propagate inside the semiconductor as waveguiding mode. To overcome these problems, many LED structures have been proposed: (1) surface texturing and reflectors can be used for destroying the waveguide effect and reflecting the light out; (2) a transparent substrate (e.g. GaP) can be employed to decrease the substrate absorption; and (3) the reflective metal electrode can be replaced by a transparent electrode (e.g. indium tin oxide, ITO). Crystalline semiconductor layers with p–n structures and active regions of several micrometers are grown on single-crystal substrates, which is called the epitaxial process.10 Liquid phase epitaxy (LPE) and vapor phase epitaxy (VPE) are common techniques. One of the VPE systems with organometallic material sources, called metalorganic chemical vapor deposition (MOCVD), is one of the most popular methods for epitaxial growth. Many important material characteristics are determined in the epitaxial process, such as doping concentrations, doping profile, defect density and alloy compositions. Other criteria for LEDs with high power efficiency rely on suitable device design and fabrication techniques. Contact resistance between the electrode and the material should be as low as possible, which can be achieved by thermal annealing. Also, the layout of the contact should be optimized between the uniform current spreading and the high outcoupling efficiency. An etching process is needed to confine the current flow and increase the extraction efficiency. However, the etched face results in damage of the periodic lattice and increases nonradiatve recombination. Suitable thermal annealing can effectively passivate dangling bonds. After the LED device is fabricated, suitable packaging is essential in real applications to: (1) protect the LED chips mechanically and environmentally, (2) improve the extraction efficiency and (3) provide a thermal path for high-power applications. LEDs have many applications in the display industry. For traffic signals applications, LEDs have higher power efficiency and hence less power consumption than conventional filtered incandescent lamps. Longer lifetime is another important advantage which reduces maintenance costs. Electronic signage and huge displays over several hundred inches are another niche market for LED displays. Cold cathode fluorescent lamps (CCFLs) are conventional backlights for liquid crystal displays (LCDs); these lamps contain mercury and are harmful to the environment. The continuous spectrum also limits the color gamut of the LCD. White LEDs are promising candidates to replace CCFLs as the backlighting source of LCDs, due to their environmentally friendly characteristic, high color gamut, long lifetime and fast response. In addition to backlight units for LCDs, LEDs can also be used for general lighting due to their high power efficiency and long lifetime. Also, they are rugged due to their all-solid-state nature. The operation voltage is quite low (less than 5 V) as compared to other lighting technologies, which means LEDs are easier to drive and safer. In this chapter, we first describe the material systems for visible LEDs. Then the electrical (diode) and optical (light-emitting) characteristics are introduced, which is followed by device fabrication. Finally, LED applications in displays are discussed.
140
Introduction to Flat Panel Displays
6.2 Material systems Material systems of LEDs determine the emission wavelength and radiative recombination efficiency. To emit photons with a desired wavelength, the active layer of an epilayer (which may be a binary, ternary or quaternary compound semiconductor) with a certain bandgap should be selected (a direct bandgap is much preferred). Then, a suitable substrate, which has to be lattice-matched to the epitaxial layer, is needed to reduce the nonradiative recombination rate. Elementary semiconductors of group VI such as germanium, silicon and carbon (diamond) are indirect bandgap which is typically not suitable for LED applications. Figure 6.2 shows the curves of the bandgap and emission wavelength versus the lattice constant for III–V compound semiconductors. The points represent binary materials and the curves connecting the points represent ternary materials. Alloying two ternaries with one different atom (e.g. InGaP and InAlP) creates a quaternary material (e.g. InGaAlP), which is represented by the line ending at two ternary curves. We can see a general trend (although not universal) from Figure 6.2 that, on increasing the lattice constant, the bandgap decreases and the emission wavelength increases. Since the binding energy (which determines the bandgap) is greater for smaller atomic number, compound semiconductors consisting of ‘smaller’ atoms exhibit a larger bandgap. So, for example, the bandgap of GaAs is less that that of GaP which in turn is less than that of GaN since the atomic size is ordered as As > P > N. This is also generally true when replacing the group III material. For example, the bandgap of InN is less than that of GaN which in turn is less than that of AlN since the atom size is ordered as Al < Ga < In. Arsenic- and phosphorus-based compound semiconductors (e.g. GaP, GaAsP, AlGaAs, GaAs, InGaAsP) can be operated in the red, orange, yellow and even green region. To push the emission wavelength towards shorter wavelength, a smaller group V atom, nitrogen, is used. The lattice constant between the substrate and epitaxial layer should be kept as low as possible. ‘Lattice mismatch’ is defined as the difference in lattice constant relative to the substrate, a/a0 , and should be typically as low as 0.1 % for obtaining a high-quality epitaxial layer. When the lattice constant between the epitaxial layer and the substrate is not matched, the residual strain results in deformation of the epitaxial layer, since this layer is much thinner than the substrate. If the accumulated strain is too high, dislocation occurs which increases the nonradiative recombination rate. Hence, lattice-match condition should be addressed carefully when selecting the material system. Common substrate materials are GaAs, InP and GaP. From Figure 6.2, one can see that GaP and InP exhibit extremely high and low bandgap conditions, respectively. At these two conditions, no suitable ternary and quaternary arsenic- and phosphorus-based
6.5
GaP GaAsP
0.6 0.7 0.8 0.9 1.0 1.5
AlAs GaAs AlGaAs InP
1 InGaAsP/InP InAs 0 5.3
5.5
5.7 5.9 Lattice constant, Å (a)
6.1
Energy gap, eV
2
AIN
0.2
0.5 Wavelength, µm
Energy gap, eV
AIP AIGaInP/GaAs
AIN Wurtzite
4.5
Zinc blende 0.3
GaN SiC–6H 2.5 (W) ZnO 2.8
3.2
GaN InN
0.4
SiC–3C InN Al2O3 MgO
3.6 4.0 4.4 Lattice constant, Å
4.8
Wavelength, µm
3
0.5 0.6 0.7 5.2
(b)
Figure 6.2 Bandgap and emission wavelength versus lattice constant for III–V semiconductors: (a) AlInGa–AsP and (b) AlInGa–N material systems.11 (Reprinted from Mueller, G. (ed.) Electroluminescence I, Semiconductor and Semimetals, Vol. 64. Copyright (2000), with permission from Elsevier)
Light-emitting diodes
141
materials can be found to lattice-match to the substrate. GaAs is a suitable choice since it is latticematched to AlGaAs and (InGa)0.5As0.5 P. However, it should be noted that the bandgap of GaAs is 1.424 eV, corresponding to 870 nm in the infrared region, which means it absorbs visible light and, in turn, reduces the efficiency. For the nitride-based materials, no suitable III–V semiconductor materials could be found as the substrate, which delayed the development of short-wavelength LEDs until the 1990s. However, by pregrowth of low-temperature GaN with wurtzite structure on the lattice-mismatched substrate Al2 O3 , high-efficiency green and blue LEDs can be fabricated. As discussed in Chapter 3, semiconductors can be divided into direct and indirect bandgap materials, which are shown as the solid and dash lines, respectively, in Figure 6.2. Figure 6.3 shows schematic representations of the band structures of direct and indirect bandgap semiconductor materials. Direct bandgap means the minimum of the conduction band coincides with the maximum valence band at the momentum space. Since the momentum of photons is so small compared to electrons and holes, the optical transition is typically vertical in the band diagram shown in Figure 6.3. For the indirect bandgap condition, as shown in Figure 6.3(b), electrons accumulate at the lowest energy valley of the conduction band, which does not coincide with the maximum of the valence band. Hence, the momentum cannot conserve to emit a photon. ‘Phonons’ with low energy but large momentum should participate in the process for efficient light emission. Typically, direct bandgap materials exhibit much higher radiative recombination rates than indirect ones. On increasing the bandgap of As- and P-based materials, such as increasing the Al ratio of AlGaAs and (InGa)0.5Al0.5 P, direct–indirect bandgap transition occurs, which results in an efficiency reduction and limits the EL emission at longer wavelengths. Note that the transition of direct and indirect bandgaps is not an abrupt change in efficiency. Since the electrons tend to reside on
Direct minimum
Indirect minimum
Energy
– – –
+ +
Phonon assisted transition
+
+
+
+
–
– – Trap state
Valence band maximum
momentum
(a)
(b)
–
+
+
– –
+
(c) Figure 6.3 Energy versus momentum curves for semiconductors with (a) direct, (b) indirect and (c) near direct– indirect bandgaps.
142
Introduction to Flat Panel Displays
the lowest energy determined by the Boltzmann distribution, some of the electrons occupy the states in the indirect valley when the energies between the direct and indirect bandgaps are close enough, although the semiconductor still has a ‘direct’ bandgap. Hence, as shown in Figure 6.2(a), (InGa)0.5Al0.5 P systems are latticed-matched to GaAs and are direct bandgap systems from red to amber regions, which they are suitable for visible LEDs in the longwavelength region. The absorbing GaAs substrate can be removed and replaced by transparent GaP substrate after the epitaxial growth for further improving the external quantum efficiency. Moving from short to long wavelengths, InGaN with increasing In concentrations shows a high efficiency in the blue and green region. Hence, the blue-green III–N and amber-red III–P LEDs can cover the whole visible spectrum. However, although the luminous efficiency of the green InGaN LED is high, the internal quantum efficiency is low since the incorporation of In results in In clusters which results in the increase of nonradiative recombination. Since the III–N and III–P materials are typically grown on different substrates (i.e. sapphire and GaAs, respectively), it is difficult to obtain emission of broad-band white light from a single LED chip. There are several ways for obtaining white light emission, such as (1) by assembling multiple LEDs in a module or (2) using phosphors which absorb the LED blue light and emit yellow light.12, 13 By using multiple LEDs, electrical and optical characteristics (such as efficiency, driving voltage and emission wavelength) of each device can be optimized individually. However, the cost is high and the device is bulky since more than a single LED is used. Also, after long-term operation, there may be color shift problems due to the different lifetime performances of different LEDs. For a phosphor-coated white LED, the efficiency is theoretically lower than that using the multi-LED technique since the emission wavelength of the phosphor is shorter than the absorption wavelength. Advantages of this technique include low cost, compact size and stable CIE coordinates. As discussed in Chapter 2, ‘white’ color means specific CIE coordinates located nearly at the center of the chromaticity diagram. However, as the light source of reflective displays, not only should the color of the LEDs be white, but also a high CRI value is important to reproduce the color as it is under the illumination of a blackbody radiator.
6.2.1 AlGaAs and AlGaInP material systems for red and yellow LEDs As shown in Figure 6.2, GaAs is almost lattice-matched to AlAs, which means the ternary AlGaAs can be epitaxially grown on conventional GaAs substrate with low defect density. This also provides a large flexibility in designing epilayer structures, such as heterojunction and quantum well structures (Sections 6.3.4 and 6.3.5), for obtaining higher efficiency and desired emission wavelength.14 The energy gap of Alx Ga1−xAs can be represented as10 Eg (eV) = 1.424 + 1.247x; Eg (eV) = 1.9 + 0.125x + 0.143x 2 ;
x < 0.45 (direct bandgap), 0.45 < x < 1 (indirect bandgap).
(6.1) (6.2)
From the above equations, one can obtain that the crossover point of the direct–indirect bandgap happens at 1.985 eV, which corresponds to red emission of 624 nm. This limits AlGaAs in the red to infrared emission region although the internal quantum efficiency of the AlGaAs/GaAs system can be as high as 99 %. For obtaining shorter emission wavelength, III–P material systems are used in place of the III–As ones. To fit the lattice constant of the conventional GaAs substrate, a quaternary ((Alx Ga1−x )0.5 In0.5 P) rather than ternary material is needed. The energy bandgap can be tuned from red to amber efficiently by varying the value of x, as shown in Figure 6.2. A more detailed expression of the lattice match condition for (Alx Ga1−x )y In1−y P on GaAs is 0.616 y= . (6.3) 1 − 0.027x
Light-emitting diodes
143
The value of y ranges from 0.516 to 0.525 and a value of approximate 0.5 is typically used. The bandgap of AlGaInP can be represented as15 Eg (eV) = 1.900 + 0.61x; Eg (eV) = 2.204 + 0.085x;
x < 0.58 (direct bandgap),
(6.4)
0.58 < x < 1 (indirect bandgap).
(6.5)
As we can see from Figure 6.2, the wide bandgap materials of GaP, AlP and their ternaries are indirect semiconductors. Hence, there is a crossover point from direct to indirect when x increase to 0.65, which corresponds to 2.3 eV or 540 nm (green emission). However, an efficiency decrease is observed when the wavelength is shorter than 590 nm due to the electrons accumulating at the indirect valley. So, AlGaInP material systems can provide an efficient emission range from red to amber. The refractive indexes of AlGaAs and AlGaInP systems are around 3.2 to 3.6, depending on the composition and wavelength. Typically, a smaller bandgap corresponds to a larger refractive index, which can be derived from the Kramers–Kronig relation.16
6.2.2 GaN-based systems for green, blue and UV LEDs Typically, a single-crystal substrate of semiconductor is sliced from an ingot which is obtained by the melting and recrystallization process. For III–V compounds, group V materials exhibit much higher vapor pressure than group III for the same temperature, which increases the difficulties in fabricating substrates of III–V compound semiconductors compared with group IV silicon ingot fabrication. Also, in contrast to arsenic and phosphorus which still exist as solid phase at room temperature and 1 atm, nitrogen N is in the gas phase which means the vapor pressure of nitride is much higher, making it very difficult to obtain a nitride-based substrate. Due to the lack of a conventional semiconductor substrate that is lattice-matched to group III nitrides, insulating sapphire (Al2 O3 ) is commonly used, although the lattice mismatch is as high as 16 % with GaN. Direct epitaxy of nitrides on sapphire substrate results in island (three-dimensional, 3-D) rather than planar (two-dimensional, 2-D) growth, combined with a high dislocation density, as shown in Figure 6.4. By introducing an AlN or GaN buffer layer grown at low temperature, a high-quality nitride epilayer can be obtained since it: (1) provides nucleation sites for epitaxial growth at the same crystal orientation which reduces the dislocation defect densities and (2) decreases the interfacial free energy between the substrate and the epitaxial layer which makes 2-D growth preferable.6 The bandgap of Inx Ga1−x N ternary alloy is roughly given by Eg (eV) = xEg , InN + (1 − x)Eg , GaN − (1 − x)xb, Sapphire LT buffer
GaN
(a) Figure 6.4
(b)
Growth of III–nitride (a) with and (b) without low-temperature buffer layers.
(6.6)
144
Introduction to Flat Panel Displays
where E g ,InN and E g ,GaN are the bandgaps of InN and GaN, respectively. Reference bandgap values are 1.95 and 3.40 eV for InN and GaN, depending on the strain. Parameter b is called the bowing factor with a typical value of 1.00 eV.17 One can note that III–nitride semiconductors exhibit direct bandgaps for the whole range of compositions, from Equation (6.6). By tuning the x value from 0.15 to 0.45, UV to green LEDs can be realized. In nitride LEDs, indium atoms tend to form clusters in the epitaxial layer which results in some inevitable composition fluctuation.18 This can be observed from the spectral broadening of the photo- or electroluminescence which is called ‘alloy broadening’. With increasing indium concentration, the metallic indium clusters act as quenchers which results in low internal quantum efficiency of nitride LEDs for long wavelengths (green, yellow and red). Lattice mismatch occurs not only between the substrate and the epitaxial layer, but also between different epitaxial layers. This results in strain between different layers (e.g. quantum well and barrier; Section 6.3.5) and hence a built-in potential inside the LED along the crystal growth direction. Such a piezoelectric field in nitride LEDs results in the spatial misalignment of the electron and hole wave functions.19 As shown in Figure 6.5, the emission spectra of the LED blue-shifts on increasing the pumping (optical or electrical) due the screening of the built-in electric field by the excess carriers. Similar to the III–P cases, it is also possible to apply quaternary AlInGaN to control the strain and fine tune the emission characteristics.20 Magnesium is typically used for p-type doping of nitrides. Since NH3 is used as the nitrogen source, the formation of high electronic conductivity Mg–H during epitaxial layer formation results in low hole concentrations. By using high-temperature annealing (>700◦ C) for dehydrogenation, it is possible to achieve a reasonable hole concentration and resistivity for LED (or even laser diode) applications.21
Example 6.1 Determine the composition of Inx Ga1−x N in the active layer with emission wavelengths at 550 and 450 nm for green and blue LEDs, respectively. Answer. For green and blue with emissions at 550 and 450 nm: Eg (eV) =
1240 = 2.2545; 550(nm)
Eg (eV) =
1240 = 2.7556. 650(nm)
The x values of ternary alloy Inx Ga1−x N can be calculated from Equation (6.6): Inx Ga1−x N(550nm): Eg (eV) =
1240 = 1.95x + 3.4(1 − x) − x(1 − x); 550(nm)
x = 0.6291
Inx Ga1−x N(450nm): Eg (eV) =
1240 = 1.95x + 3.4(1 − x) − x(1 − x); 450(nm)
x = 0.2997.
Conduction band –
+
–
–
–
–
+
+
+
+
–
+
Increase excitation Valence band
Figure 6.5 Band structure shift of III–nitride under excitation.
Light-emitting diodes
145
6.2.3 White LEDs To generate white light from LEDs, there are at least two different techniques, as summarized in Table 6.1. The first is by using multiple LED chips. As mentioned before, since the integration of different substrates is difficult, the physical size of this white-light module cannot be too small. The second technique is to use the LED (blue or UV) to excite phosphors. This can be also distinguished from the ‘composition’ of the white light, also shown in Table 6.1. Due to the lack of broad-band emission from a single LED, white light consists of two, three or even more colors. From colorimetry, white light from a mixture of two colors exhibits the highest maximum attainable luminous efficiency (in terms of lm W−1 ). However, for obtaining a high CRI value in light source applications, three or more colors that provide a broad-band emission are needed. It is straightforward to mix emissions of two or more LEDs with different wavelengths to obtain white light, following the colorimetric principles described in Chapter 2. The efficiency of this technique is typically higher than the LED pumping phosphors method since wavelength conversion is not needed which wastes energy. However, more LED chips means a higher cost and larger size. Due to the nonlinearity of the light–current relation of different LEDs, the color of this white light is not stable over different current injection. With increasing temperature, LED intensity decreases and spectrally red-shifts which also results in a color shift of the white light. After long-term operation, the different operation lifetimes of the LEDs also result in color change. For the two-color mixing technique, LEDs of two complementary colors (typically blue and yellow) are used. Since the sensitivity of the human eye is highest at 555 nm and decreases on both sides at the photopic region, it is reasonable that there exists an optimized value for the maximum attainable luminous efficiency of two complementary colors with different wavelengths, which can be derived from color science and engineering, as described in Chapter 2. With longer or shorter wavelength combinations, there is a decrease of the maximum attainable luminous efficiency, which implies two-color mixing with narrow emission spectra is the most efficient way to generate white light. Under the assumptions of (1) a combination of two monochromatic complementary colors and (2) 100 % electrical to optical energy transfer, the upper theoretical limit of the white light is over 400 lm W−1 . Hence, such a white light source is very suitable for ‘white’ display applications such as pedestrian traffic signals due to its low power consumption. However, it is not suitable for light source applications since the CRI value is typically low, due to the narrow spectra and two-color combination. To obtain a high CRI value, a broad-band emission is usually needed, which also implies a tradeoff between the CRI value and the efficiency. So, it can be expected that the maximum attainable luminous efficiency for three-color mixing is somewhat lower but with higher CRI value, which is more suitable for lighting applications. It is also possible to increase the number of LED colors (four or five) to obtain larger gamut values, but the tradeoff is the low efficiency and high cost. Another important application for three-color LEDs is in LCD backlights. For this application, LEDs with narrower emission spectra are preferred for larger color gamut values. Another common way to generate white light is by using a blue or UV LED to pump phosphors. Such a method is relatively simple and of low cost since only a single LED is needed. Besides, the phosphors can be coated during the encapsulation process inside the LED package. Hence, small physical dimensions can also be achieved. Since the lifetime of conventional phosphors is long, the color stability of this type of white LED is also very good. However, compared with the multi-LEDs scheme, this technique has an
Table 6.1 White light generation techniques.
Two-color mixing Three-color mixing
Multiple LEDs
LED pumps phosphor
Y + B LEDs R + G + B LEDs
B LED pumps Y phosphors UV LED pumps white phosphors (RGB mixture)
Introduction to Flat Panel Displays
Normalized EL intensity (a.u.)
146
0.5
λ p ~400 nm
λ p ~600 nm
0.4 0.3
λ p ~500 nm
0.2
λ p ~450 nm
0.1 0.0 350 400 450 500 550 600 650 700 Wavelength (nm)
Figure 6.6 Emission spectra of UV radiation (λp ≈ 400 nm) pumping (a) red (λp ≈ 600 nm), (b) green (λp ≈ 500 nm) and (c) blue (λp ≈ 450 nm) phosphors.22 (Source: Sheu, J.K. et al. White-light emission from near UV InGaN–GaN c 2003 IEEE) LED chip precoated with blue/green/red phosphors. IEEE Photon Technol. Lett., 15, 18.
inevitable disadvantage of theoretically lower efficiency. As shown in Figure 6.6, the phosphor absorbs photons with higher energy in the UV region with a peak of 400 nm and emits lower energy photons at 600, 500 and 450 nm, for red, green and blue emissions, respectively.22 Typically, the internal quantum efficiency of the phosphor can be as high as 90 %, which means 90 photons are emitted per 100 photons absorbed. However, there is an energy loss between the photons absorbed and emitted, which is called the Stokes shift. In this case, the energy loss originating from the Stokes shift results in an efficiency decrease of 33.3, 20 and 11.1 % for red, green and blue emissions, respectively. Another possible energy loss arises from the light scattered by the phosphor which is reabsorbed by the LED chip, package and reflectors. There are lots of ‘down-conversion’materials which can absorb blue and UV light and generate photons with longer wavelength, such as organic dyes,23 semiconductors24 and nanocrystals.25 However, there are some disadvantages for these materials, such as reliability issues and/or high cost. Typically, inorganic phosphors are commercially used as wavelength converters due to the advantages of high stability, defectfree structure, easy fabrication and high efficiency.13 Figure 6.7 shows the device structure of a blue LED pumping yellow phosphors and the corresponding emission spectrum. The yellow phosphor was coated on the LED chip which can be fabricated during the packaging process. Some of the blue emission is absorbed by the phosphor which re-emits the yellow light. Other blue light comes out of the device and white light can be obtained. The blue emission from the LED is relatively narrow, with FWHM of 50 nm.
phosphor
Blue LED
Figure 6.7 White LED structure consisting of yellow phosphor pumped by a blue LED and its emission spectrum.13 (Source: Mueller-Mach, R., Mueller, G.O., Krames, M.R. and Trottier, T., High-power phosphor-converted lightc 2002 IEEE) emitting diodes based on III-nitrides. IEEE J. Sel. Top. Quantum Electron., 8, 339.
Light-emitting diodes
147
1
Intensity, a.u.
0.8
YAG emission
0.6 0.4 0.2 4 3 2 1 0 450
550 650 Wavelength, nm
750
Figure 6.8 Emission spectra of YAG phosphors with different compositions: ‘1 → 3’ with decreasing gadolinium concentrations and ‘4’ with high gallium concentrations.13 (Source: Mueller-Mach, R., Mueller, G.O., Krames, M.R. and Trottier, T., High-power phosphor-converted light-emitting diodes based on III-nitrides. IEEE J. Sel. Top. Quantum c 2002 IEEE) Electron., 8, 339.
In contrast, the phosphor emission is quite broad, which is determined by the material and can be fine tuned. One of the most successful yellow phosphor materials is yttrium aluminum garnet (YAG) doped with Ce3+ ions (Y3Al5 O12 :Ce3+ ).13 As shown in Figure 6.8, with the incorporation of gadolinium and gallium, the emission peak of the phosphor (Y1−x Gdx )3 (Al1−y Gay )5 O12 :Ce3+ can be red- and blue-shifted, respectively. The absorption peak of the phosphor is around 460 nm, which is suitable for blue LED pumping. The thickness of the phosphor layer is varied to adjust the relative intensity between blue and yellow emission. Peak wavelengths of these two complementary colors are also close to the upper theoretical limit of white light generation. In this system, the energy loss from the Stokes shift is not so serious since the blue part is directly emitted from the LED, rather than from the phosphor. However, the CRI value is typically less than 80 using this two-color mixing. UV LEDs pumping RGB phosphors can effectively increase the CRI value since the emission characteristic is determined by the broad-band emission from the phosphors. Besides, if a deep-UV LED is used (200 to 320 nm), the mature phosphor technology that is typically used in fluorescent lamps can be applied. However, the large Stokes shift is a problem for this technology.
6.3 Diode characteristics When a p-layer and an n-layer come into contact, a p–n junction forms and shows diode characteristics, ideally which means the device behaves like a short and open circuit under forward and reverse bias condition, respectively. There are more holes (electrons) than electrons (holes) in the p- (n-) layer. Hence, the holes (electrons) diffuse from p- (n-) to n- (p-) side near the p–n junction and recombine with opposite carriers, as shown in Figure 6.9(a).26 After carrier annihilation, negative and positive space charge is they left behind which creates an internal electric field from n- to p-side. This internal field results in the band diagram bending, as shown in Figure 6.9(b). By using the Poisson equation, which is derived from the boundary condition of Maxwell’s equation, one can obtain the relation between the donor and acceptor concentrations and the built-in voltage. This space charge region is nearly free of carriers under thermal equilibrium and is called the ‘depletion region’. Holes and electrons experience the built-in voltage and then drift toward p- and n-side, respectively. Such drift current balances the diffusion current and no net current flows under thermal equilibrium, as shown in Figure 6.9. Under this condition, one can also note the Fermi level of the p- and n-side is aligned due to the band bending. When a forward bias is applied (a voltage from p- to n-side), it cancels the built-in voltage. Once the forward bias exceeds the
148
Introduction to Flat Panel Displays
E-field p-type
n-type
Electron diffusion – –
+ + Hole diffusion + : hole – : electron (a) Figure 6.9
EC EF Ei EV
– – + – + – + +
qVbi
+ : positive space charge – : negative space charge
EC EF Ei EV
(b)
Band diagram of a p–n junction (a) before and (b) after contact under thermal equilibrium.
built-in voltage, the current density shows an exponential increase with applied voltage, which is called the threshold voltage. The value of threshold voltage of a LED is approximately E g volts (which is the bandgap energy in terms of eV) under high carrier concentrations. Layers of p- and n-type here are used to transport carriers into the depletion region for recombination. Hence, the resistance of the p- and n-layer should be as small as possible to reduce the ohmic loss in these regions. Also, this decides the width of the depletion zone, which in turn affects the recombination zone and recombination efficiency. For this kind of homojunction structure, however, there is still some carrier leakage, i.e. holes and electrons propagating to the cathode and anode, respectively, without recombination due to the inefficient carrier confinement. Heterojunction and electron blocking layer are used to prevent carrier leakage to the opposite electrodes. Furthermore, quantum structures (well, wire and dot) are used to confine carriers in a specified recombination region. Energy levels in the quantum structures are discrete, which allows one to fine-tune the emission wavelength, and increase the flexibility during epitaxial growth.
6.3.1 The p-layer and n-layer A LED is basically formed by a p–n junction. Holes and electrons transport through the p- and n-layers, respectively, and then recombine in the active region. To reduce ohmic loss at the p- and n-layers, the resistance of these materials should be as small as possible, which can be achieved by increasing the carrier concentration (and also the carrier mobility). Carrier concentration in these layers also determines the width of the depletion region, which affects the recombination zone in a homojunction LED. By incorporating group IV (e.g. Ge and Si) and VI (e.g. S, Se and Te) materials into III–V compounds to replace the group III and V atoms, it is possible to achieve n-type doping.11 Figure 6.10(a) shows a schematic of Se doped into GaAs and replacing the As site. Since Se has one more valence electron than As, it ‘donates’ one electron to the crystal. This extra electron can be considered as nearly free, since no bond is formed with the crystal. However, it is still attracted by the positively charged Se atom which is similar to the binding between the electron and nucleus of a hydrogen atom. Hence, one can define a binding energy of the donor (E D ) slightly below the conduction band in the forbidden gap, and treat it as a trap state. Since it is a shallow trap, the thermal energy is enough to excite the electrons to the conduction band as free electrons. Similarly, p-type doping can be achieved by using Cd, Zn, Mg and Be (group II) and C and Si (group IV) atoms to replace the group III and V atoms, respectively. Figure 6.10(b) shows an example of Zn doped in GaAs and replacing the Ga atom. Acceptors create an acceptor energy (E A ) above the valence band within the forbidden gap. Electrons in the valence band can be thermally excited to the acceptor state, which looks like a free ‘hole’ created in the valence band. Typically, the effective mass of holes is much greater than that of electrons, which results in E A being greater than E D . Hence, the thermal energy required for free-hole creation is much larger than for electrons,
Light-emitting diodes
149
Ga
As
Ga
Ga
Se
Ga
As
Ga
As
As
As
Ga
As
Zn
As
Ga
As
Ga
+
–
(a) Figure 6.10
(b)
Schematic of bonding of GaAs with (a) Se and (b) Zn impurities.
which means the ‘ionization’ of holes is sometimes not complete and limits the p-doping concentration, especially for wide-gap nitride semiconductors. For example, E A of Mg-doped GaN is 0.2 eV and much higher than the thermal energy (∼25 meV at room temperature) which means only a small portion of acceptors are ionized and contribute to hole carrier concentrations. Sometimes, the dopant does not exactly replace the original atoms, which results in low carrier concentration. For example, Mg doped in GaN does not occupy the Ga site as grown and leads to a low hole concentration. After low-energy electron beam irradiation (LEEBI), the Mg atoms move to the exact Ga sites and contribute to hole concentration. Unintentional introduction of impurities during epitaxy and processing results in defect trap levels which compensate the doping level; for example, oxygen contamination for aluminum-containing compounds and hydrogen passivation for nitride compounds. A suitable post-annealing process is required to remove these impurities. Another way to reduce the resistance of the p- and n-layers is to increase the mobility value. Typically, hole mobility in III–V compounds is much less than that of electrons due to their larger effective mass. For example, n and p are 5000 and 300 cm2 V−1 s−1 in n- and p-type GaAs with doping concentration of 1017 cm−3 at room temperature. There exists a tradeoff between the carrier concentration and carrier mobility. With increasing dopant concentration, the impurities scatter the carriers and reduce the carrier mobility. Typically, it is more effective to reduce the resistance by increasing the carrier concentration.
6.3.2 Depletion region In this section, we quantitatively describe two important parameters of the depletion region, built-in voltage and depletion width. As shown in Figure 6.9, the built-in voltage comes from band bending in the depletion region. In contrast, in the p- and n-layers, the band is still flat. As discussed in Chapter 3, assuming the acceptors and donors are fully ionized, then hole and electron concentrations can be represented as26 Ei − EF = NA , p = ni exp (6.7) (kT /q) n = ni exp
EF − Ei = ND , (kT /q)
(6.8)
where p and n are hole and electron concentrations far away from the depletion region, respectively, ni is the intrinsic concentration and E i and E F are the Fermi energies of intrinsic and doped layer in terms of
150
Introduction to Flat Panel Displays
eV, respectively. Note that although the Fermi level looks flat throughout the whole p–n junction as shown in Figure 6.9, the Fermi energies with respect to the conduction band and valence bands are different in p- and n-layers, far away from the depletion region. K is the Boltzmann constant, T is the temperature and N A and N D are the acceptor and donor concentrations, respectively. Hence, one can obtain ψ p , ψ n and built-in voltage V bi = (ψ n − ψ p ) as shown in Figure 6.9: NA kT ψp = −(Ei − EF ) = − ln , (6.9) q ni ND kT ln , (6.10) ψn = −(Ei − EF ) = q ni NA ND kT ln . (6.11) Vbi = ψn − ψp = q ni2 One can note that with increasing doping concentrations, V bi increases. In LED applications, for reducing device resistance and hence less power consumption, heavily pn-doped layers are needed and the Fermi levels of the p- and n-layers are close to the valence and conduction bands, respectively. Hence, one can note that V bi is close to the bandgap of the semiconductor. When the applied voltage is higher than V bi , then the band bends to the reverse side to the equilibrium case. Under this bias condition, holes and electrons transport easily from p- and n-type layers to the opposite side, which results in an exponential increase in current density as described in the next section. From electromagnetic wave theory, Gauss’s law can be represented as ∇·(εE) = ρ,
(6.12)
∇ 2 V = −ρ/ε,
(6.13)
or can be written as Poisson’s equation: where E is the electric field, ε is the electric permittivity, ρ is the charge density and V is the voltage. Assuming the free carriers are fully ionized in the depletion region, then the distributions of space charge, electric field and voltage are as shown in Figure 6.11. The electric field distribution can be obtained from E(x) = ρ(x) dx ρ(x) =
−qNA ; −xp < x < 0 , +qND ; 0 < x < xn
(6.14)
where x p and x n are the boundary of the depletion region at p and n side, respectively. Hence, the maximum electric field (E max ), which occurs at the p–n boundary, is given by q q (6.15) Emax = ND xn = NA xp . ε ε One has to note that xn + xp = W ,
(6.16)
where W is the width of the depletion region. Hence, the depletion region under thermal equilibrium (V bi ) can be represented as 1 Vbi = Emax W (6.17) 2 or 2ε NA + ND (6.18) Vbi . W= q NA ND
Light-emitting diodes
151
Depletion width p-type
–xp
– – – – – –
+ + +
n-type
ρ (x) xn E(x)
x
x Emax
V(x)
Vbi x
Figure 6.11
Charge density, electric field and voltage distribution of a LED under thermal equilibrium.
when a forward bias V is applied, V bi in Equation (6.18) is replaced by (V bi – V ), which results in a reduction in the depletion width. In contrast, the depletion region increases under reverse bias. With a narrower depletion region, it is easier for carriers to transport across through to the opposite site, leading to higher current density. In contrast, the wider depletion region under reverse bias impedes carrier transport, which explains the rectification effect of the p–n diode.
Example 6.2 Consider a GaAs p–n diode at room temperature. The intrinsic carrier concentration ni is 1.79 × 106 cm−3 and the doping levels are N A = 5 × 1019 cm−3 in the p-side and N D = 1016 cm−3 in the n-side. Find the built-in voltage, depletion width and maximum electric field. (Note that εr = 13.18 for GaAs.) Answer. From Equations (6.9) and (6.10), NA kT 5 × 1019 300 × 1.381 × 10−23 ln ln =− = −0.8013V, ψp = − q ni 1.6 × 10−19 1.79 × 106 ND kT 1016 300 × 1.381 × 10−23 ψn = − ln ln =− = 0.5808V. q ni 1.6 × 10−19 1.79 × 106 The built-in voltage V bi from Equation (6.11) is Vbi = ψn − ψp = 1.3821V. From Equations (6.17) and (6.18), 2ε NA + ND 2 × 13.18 × 8.85 × 10−14 5 × 1019 + 1016 W= Vbi = × 1.3821 q NA ND 1.6 × 10−19 5 × 1019 × 1016 ∼ = 4.48 × 10−5 cm, Emax =
2Vbi ∼ = 61 700V cm−1 W
152
Introduction to Flat Panel Displays
6.3.3 J–V characteristics Carrier transport in a semiconductor can be divided into carrier diffusion and drift. Diffusion means carriers diffuse from high to low concentrations. Drift current means carrier transport effected by an electric field. Hence, carrier transport, taking hole current density, J p , as an example (total current density is the sum of the electron and hole current densities), can be represented as26 Jp = Jp,diff + Jp,drift , Jp,diff = qDp
(6.19)
dp , dx
(6.20)
Jp,drift = pqμp ,
(6.21)
where J p,diff and J p,drift are hole diffusion and drift current density, respectively, Dp is the diffusivity and μp is the hole mobility, defined as the velocity (in cm s−1 ) over the electric field (V cm−1 ). Under thermal equilibrium, holes and electrons diffuse from p- and n-layers to the opposite side, which is balanced by the drift current from n- to p-side. As shown in Fig. 6.12, since drift current depends on the electric field, there is no drift current outside the depletion region, where the band is flat. When a forward bias is applied, built-in voltage decreases and current density increases. Since the built-in voltage decreases with increasing forward bias, the barrier for carrier diffusion decreases which results in a net current from the p- to n-side under forward bias. Although there is no explicit form to illustrate the carrier distribution in a LED, which is essential to derive the J–V characteristics, we can still qualitatively describe it as shown in Figure 6.13. Parameters pp0 , nn0 , pn0 and np0 are the hole concentration in p-layer, electron concentration in n-layer, hole concentration in n-layer and electron concentration in p-layer, respectively, before p–n junction formation. One can note that near the depletion boundary, the minority carrier concentrations pn and np are higher than pn0 and np0 , respectively, which results in a diffusion current for holes and electrons in n- and p-layers. Typically, the J–V characteristics can be approximately represented as
Dp Dn q(V − Vbi ) J =q , (6.22) NA + ND exp τp τn kT where τn and τp are the minority carrier lifetimes, respectively. One can note that once the voltage is over V bi (approximately the value of E g in terms of eV) an exponential increase in current density is observed. This ‘threshold’ voltage is reasonable. Since the LED emits a photon with wavelength λ =hc/E g , one has to provide the driving voltage of at least V = E g /q. Hence, it is straightforward that the driving voltage of shorter wavelength (blue LED) is higher than that of longer wavelength (red LED). One of the disadvantages of this structure, which is called a ‘homojunction’ structure, is the carrier leakage problem.
p
q(Vbi-V)
qVbi n Jdiff Jdrift
Figure 6.12
Jdiff Jdrift
Band diagram of a LED under (a) thermal equilibrium and (b) forward bias.
Light-emitting diodes
153
Carrier Conc. (log)
p
Depletion region
n
p(x) n(x)
pp0
nn0 pn0
np0
x Figure 6.13 Carrier distribution under forward bias.
Since the holes and electrons diffuse into the n- and p-side for recombination, with their carrier lifetimes of τp and τn , it is possible for carriers to leak to the electrode before recombination if the time constants are long enough, which results in a reduction of the efficiency. Such a drawback can be overcome by careful design of the device structure, as discussed in the following sections.
6.3.4 Heterojunction structures Since the bandgap of III–V semiconductors can be adjusted by varying their compositions, materials with different bandgaps can be applied to different layers of a single device during the epitaxial growth, which is called a heterojunction structure.27 A LED with single composition (bandgap) described in the previous section is referred to as a ‘homojunction’ device. Heterojunction structures can be used to confine the carriers in the active region to increase the recombination rate. Figure 6.14(a) shows a single-heterojunction (SH) structure. In this type of LED, p- and n-type materials are different. The p-layer exhibits a smaller bandgap than the n-layer. E c and E v denote the energy difference of the conduction and valence bands between two materials, respectively. The sum of E c and E v represents the bandgap difference between p- and n-type materials. For a hole transporting from p- to n-type material, it encounters an extra barrier height E v which decreases the hole current into the n-region. Electron trapped – –
Carrier confinement (trap+block)
Δ Ec
Eg,p p
– –
+
p
+
Hole blocked Δ Ev
Eg,p < Eg,n (a)
i
n Eg,n
Eg,p
Eg,i + +
n Eg,n
Eg,i < Eg,p and Eg,n (b)
Figure 6.14 Band structures of (a) single- and (b) double-heterojunction LEDs.
154
Introduction to Flat Panel Displays
On the other hand, the electron is energetically preferred to inject into the p-side, and possibly be trapped in the ‘energy well’ formed by E c . Hence, many electron–hole pairs accumulate at the p-side near the SH interface which increases the recombination rate and the internal quantum efficiency. Typically, in a heterostructure the small-bandgap material, the p-side in this case, is designed as the emitting layer. In contrast to the homojunction devices, the recombination zone of SH-LEDs is mainly determined by the energy difference and layer structure. Hence, the p- and n-doping concentrations can be optimized without considering the depletion region width which affects the recombination zone of homojunction LEDs, as described in previous sections. Furthermore, one can apply the heterojunction at both the p- and n-type materials at the same time, which is called a ‘double heterojunction’ (DH) structure, as shown in Figure 6.14(b). An active layer with a smaller bandgap is sandwiched by two wide-bandgap materials (typically called cladding or confinement layer), which ensures the carriers injected into the active layer are confined in this region for a higher recombination rate. Due to the imbalance of carrier mobilities (μn μp ), electrons penetrate easily into the p-side. Hence, an electron blocking layer (EBL) is inserted between the active layer and p-type region to stop electrons with a large bandgap and E c , as shown in Figure 6.15.28
6.3.5 Quantum well, quantum wire and quantum dot structures By inserting a thin layer with a lower bandgap in the active region of a heterostructure, the emission zone can be spatially specified and hence the recombination rate can be further increased, which is proportional to the product of the electron and hole concentrations. Reabsorption probability also decreases since the emission zone is very thin and exhibits a small bandgap. Such a thin layer is typically less than 20 nm, which is so small that the matter wave characteristic of the electron is obvious (Section 3.2.1). That is, the energy level becomes quantized, called a ‘quantum well’ (QW) structure, as shown in Figure 6.16.27 The quantized energy levels can be adjusted by changing the composition and width of the QW, which allows other parameters for fine-tuning the emission wavelength. Another advantage of the QW structure comes from its ultrathin width (compared with the 1 μm thickness of the active layer of the heterostructure) which enables the possibility of lattice mismatch growth to be applied before the strain relaxation. Note that the strain also affects the quantized levels which results in some emission wavelength shift. A QW structure is an ultrathin layer, with quantized level along the epitaxy direction. The composition in the 2-D plane perpendicular to the epitaxial growth is continuous. As shown in Figure 6.17, when a
– – – p
p
i
EBL
Eg,i
Eg,p Eg,EBL
+ +
n
Eg,n
Eg,i < Eg,p and Eg,n < Eg,EBL Figure 6.15 Band structures of a LED with EBL.
Light-emitting diodes
155
QW p
i
n
Figure 6.16 Band diagram of a QW structure.
Quantum Well
Quantum Wire
(a) Figure 6.17
(b)
Quantum Dot
(c)
Structures of (a) quantum well, (b) quantum wires and (c) quantum dots.
low-bandgap material can be fabricated like wires and dots, then carriers are continuous only in one and zero dimensions, respectively, which are called ‘quantum wire’ and ‘quantum dot’ (QD) structures.29 In contrast to the precise control during epitaxy, fabrication of quantum wires and dots is more difficult since high-resolution pattern definition and epitaxial regrowth are needed. One special case is the nitride semiconductor. Since indium clusters form during epitaxy, which results in some composition fluctuation, some quantum dot structures are grown unintentionally, which also results in a higher recombination efficiency.
6.4 Light-emitting characteristics Once electron–hole pairs recombine, the energy may relax radiatively or nonradiateively. To decrease the nonradiative rate, one of the most important things is to obtain a high-quality epitaxial layer. Since the bandgap originates from the crystalline structure, defects act as quenchers which results in an efficiency decrease. Such a periodic structure may be destroyed during the fabrication process (e.g. a cleaved facet and etched surface). Some post-annealing processes are needed to reduce the nonradiative recombination rate. Radiatve recombination in LEDs is of Langevin type, which is proportional to the product of the electron and hole concentrations. Under some approximations, a linear relation between the injection current and light output can be obtained. The emission spectrum from a LED depends on the density of states and the carrier distributions. Densities of states of the bulk, QW, quantum wire and QD structures are different which results in a shift of the peak wavelength and change of the FWHM. Spectral shift can be observed under different temperature and injection levels. Once the photons are generated, the next step is for them to radiate out of the semiconductor. Due to the large refractive index (typically over 3) of semiconductors, total internal reflection limits the extraction efficiency. Also, for the III–P LEDs, the generated visible photons may be absorbed by the absorbing GaAs substrate.
156
Introduction to Flat Panel Displays
6.4.1 Recombination model Radiative recombination in LEDs involves holes and electrons. Once these two opposite carriers meet in the spatial domain, the recombination process happens which is a bimolecular process and can be described in Langevin form: Rrad = rrad np,
(6.23)
where Rrad is the radiative recombination rate, n and p are electron and hole concentrations, respectively, and r rad is the bimolecular recombination coefficient. When the semiconductor is excited, electrons are promoted from the valence band to the conduction band, which create the same number of holes in the valence band. This means the numbers of ‘excess’ holes and electrons are the same, so n = p,
(6.24)
where n and p are excess electron and hole concentrations, respectively. For a doped semiconductor, i.e. p-type material, the hole concentration is much higher than the electron concentration under thermal equilibrium, and hence one can obtain p0 n0 ,
(6.25)
where p0 and n0 are the hole and electron concentrations under thermal equilibrium, respectively. Under low-level excitation (excess carriers are much fewer than majority carriers under thermal equilibrium; p p0 ), one can write Rrad = rrad np = rrad n(p0 + p) ≈ rrad np0 =
n τrad
(6.26)
and τrad =
1 , rrad p0
(6.27)
where τrad is called the radiative lifetime. We note that the radiative lifetime in doped semiconductors under low-level injection is inversely proportional to the doping concentration. That means recombination is ‘faster’ with higher doping concentrations since there are more majority carriers for participating in the radiative recombination, according to the bimolecular recombination process. Energy relaxation not by photons but by phonons (lattice vibration) is called nonradiatve recombination. In a crystal semiconductor, the main cause for nonradiative recombination arises from the defects of the periodic structure, e.g. lattice structure interrupted at interfaces (surface recombination), dislocations and impurity contamination during ingot growth and epitaxy, and process-induced defects (lattice damage after ion implantation and bombardment, etching of surfaces and cleaving of facets). These defect structures can be quantitatively described by the ‘trap states’ which possibly exist inside the forbidden bandgap. Electrons (holes) jump between the conduction (valence) band and the trap states via phonon and photon intermediation. Although the gap states can help photon emission in some indirect bandgap semiconductors (such as doping nitrogen in GaP), typically they act as luminescence quenchers for most direct bandgap materials. Quantitative analysis of recombination via traps was accomplished by Shockley and Read30 and Hall.31 A schematic illustration is shown in Figure 6.18. Typically, one trap can capture and emit one electron (hole) to and from the conduction (valence) band. Once an electron and a hole are simultaneously captured by a trap before they emit, recombination occurs. When the trap is near the conduction band (i.e. shallowtrap condition), we can imagine that the electrons are easily captured and emitted by the trap, but it is difficult for the holes. Only when the trap is located near the mid-gap (i.e. deep-trap condition) is the recombination rate high. For a doped semiconductor (e.g. p-type material) with deep-trap distribution, although the equation derivation is relative complex, we can still understand that: (1) since there are
Light-emitting diodes
157
Ec Hole Capture
Hole Emission
Ei Electron Capture
Electron Emission
Ev Figure 6.18 Recombination (typically nonradiative) via deep-level traps.
many more holes than electrons, recombination takes place once an electron is captured which means the lifetime of nonradiative recombination is determined by the minority carrier concentration and (2) the traps are typically filled with holes because the traps are distributed near the mid-gap, hence the lifetime is independent of hole (majority carrier) concentration. So, one obtains Rnonrad =
n , τnonrad
(6.28)
where Rnonrad and τnonrad are the nonradiative recombination rate and the nonradiative lifetime, respectively. So, the total recombination rate can be described as 1 1 n R = Rrad + Rnonrad = (6.29) + n= , τrad τnonrad τ where τ is called the minority carrier lifetime.
6.4.2 L–J characteristics Typically, a continuity equation is used to describe the carrier concentration change under electrical and optical excitation. For the electron concentration in a p-type semiconductor under low-level injection, one can write ∂n 1 ∂Je n = + G0 + Gext − , ∂t q ∂x τ
(6.30)
where q is the magnitude of electron charge, G0 is the generation rate under thermal equilibrium which is equal to the recombination rate under the same condition (R0 = n0 /τ ) and Gext is the excess generation rate by optical excitation. Equation (6.30) can be rewritten as ∂n 1 ∂Je n = + Gext − . ∂t q ∂x τ
(6.31)
For a LED with electrical pumping under steady state, one can write 0=
1 ∂Je n − . q ∂x τ
(6.32)
The output optical power (Pout ) from a LED can be represented as n hνA τ ∂Je dx = ηext dx Pout = ηext Pgenerated = ηext hνA τrad q τrad ∂x = ηext
hνA τ [Je (in) − Je (out)], q τrad
(6.33)
158
Introduction to Flat Panel Displays
where ηext is the photon extraction efficiency, Pgenerated is the power generated inside the LED, hν is the photon energy, A is the device cross-sectional area, and J e (in) and J e (out) are the electron current supplied in and flowing out of the device, respectively. By using a wide active layer to broaden the recombination region or by designing suitable layer structures (e.g. heterojunction and low-dimensional structures) to confine the carriers, the carrier leakage can be effectively reduced. If the electrons supplied into the devices completely recombine with the same amount of holes, then Equation (6.33) becomes Pout = ηext
hν τ I, q τrad
(6.34)
where I (= JA) is the electrical current. One can see that the output optical power is proportional to the supply current. To improve the output optical power, one has to increase the extraction efficiency (higher ηext ), use a wide-bandgap material (higher hν) and reduce the radiative recombination lifetime (smaller τrad ). For a planar LED structure, the extraction efficiency is limited by the total internal reflection between the interface of the semiconductor and the air governed by Snell’s law, as shown in Figure 6.19. To improve the efficiency, corrugated structures can be designed to destroy the planar structure for coupling out the waveguiding light inside the device. Die shaping can be helpful to redirect the light out of the LED. Packaged by an epoxy dome and reflector not only protects the device but also increases the extraction efficiency. For III–P material systems, since the GaAs substrate is absorptive, this can be replaced with a transparent substrate, also increasing the light outcoupling. Details of the enhancement of extraction efficiency are discussed in Sections 6.5.3 and 6.5.4.
6.4.3 Spectral characteristics Emission wavelength of a LED is determined by the product of density of states (DOS) and the number of carriers in a certain energy state. DOS means how many states are available for recombination for a certain energy. As shown in Figure 6.3, the band diagram shows a parabolic shape both for conduction and valence bands. So, DOS increases from zero at the band edge with increasing energy. Carrier distributions can be obtained from the Boltzmann distribution which decreases exponentially with increasing energy. A typical emission wavelength from a LED is shown in Figure 6.20, where the emission peak wavelength is a little shorter than the bandgap wavelength. Also, the emission spectrum is asymmetric. There is no allowable state within the energy bandgap. Hence, the long-wavelength edge is sharp. In contrast, one can see a smooth curve at the short wavelength region which arises from the exponential tail of the carrier distribution. For QW and QD structures, the energy is quantized and hence the DOS is different from that in a bulk semiconductor. Figure 6.21 shows the DOS at different energy for bulk and quantum structures. For QW and QD structures, the DOS are a step and delta function, respectively, due to the energy confinement by the quantized level, which means the emission spectra of the quantum structures have a narrower FWHM and larger color gamut for display applications.
Air
Semiconductor Figure 6.19
Light escape cone of a LED due to the total internal reflection.
Light-emitting diodes
159
Emission Intensity (a.u.)
100 80 60 40 20 0 550
600 650 Wavelength (nm)
700
Figure 6.20 Emission spectrum from a LED.
DOS
bulk QW
QD Eg
Energy
Figure 6.21 DOS of bulk, QW and QD structures.
When increasing the temperature near room temperature, the emission intensity of LEDs decreases and the spectrum red-shifts. Figure 6.22 shows the experimental results of intensity decrease and spectral shift for R, G and B LEDs for different temperatures. The physical mechanism of the intensity decrease mainly comes from the increase of nonradiative recombination. Due to the increase of phonon intensity with increasing ambient temperature, the nonradiative recombination rate increases sharply which, in turn, decreases the intensity. Temperature increase also results in bandgap shift towards lower energy, which results in the spectral red-shifts. These two phenomena are more serious in long-wavelength LEDs, such as shown in Figure 6.22. Such a temperature dependence of a LED results in serious problems for display applications. For example, a LED traffic signal may have different colors and lower intensity at noon than at night, which results in safety issues. For LEDs as LCD backlights, a lot of heat is generated which 8
blue 6
0.9
green
4
red
0.8
blue
δλ p (nm)
IR, IG & IB (a.u.)
1.0
2
0.7
green 0.6
0
10
20
30 40 50 Temperature (°C)
60
70
Figure 6.22 Temperature dependence of RGB LEDs.
160
Introduction to Flat Panel Displays
results in color shift and color gamut change. Hence, suitable temperature control and compensation circuits are needed to optimize the display performances over a wide temperature range.
Example 6.3 White light is generated by red, green and blue LEDs with their primary wavelengths at 650, 550 and 450 nm, respectively, at room temperature (25◦ C). However, after operating for some time (i.e. 30 min), due to the generated heat, the module temperature increases and stabilizes at 85◦ C. Assume these three LEDs emit monochromatic light. For the red LED, the peak wavelength red-shifts 5 nm and emission intensity decays 20 % with a 30◦ C increase in temperature. For the green and blue LEDs, the wavelength shift and intensity decay are 2.5 nm and 5 % per 30◦ C increase. Find the initial RGB luminance ratio (at 25◦ C) of a white light source at CIE coordinates of (0.33, 0.33) which stably operates at 85◦ C.
Red λ(nm) x y V (λ)
650 0.725 0.275 0.107
Green 660 0.730 0.270 0.032
Blue
550 0.302 0.692 0.995
555 0.337 0.659 1.000
450 0.157 0.018 0.038
455 0.151 0.023 0.048
Answer. After a long time of operation, the emission wavelength of the red LED shifts and stabilizes at 660 nm: x=
Xr Yr Zr = 0.73; y = = 0.27; z = = 0 → Xr = 2.7Yr ; Zr = 0. Xr + Yr + Zr Xr + Yr + Zr Xr + Yr + Zr
Similarly, we can find the relation between X, Y and Z for the green (555 nm) and blue (455 nm) LEDs: X g = 0.5114Y g , Z g = 0.006Y g ; X b = 6.5652Y b , Z b = 35.913Y b . The white light consists of RGB LEDs; the X, Y and Z for white light is Xw = Xr + Xg + Xb = 2.7Yr + 0.5114Yg + 6.5652Yb Yw = Yr + Yg + Yb Zw = Zr + Zg + Zb = 0.006Yg + 35.913Yb For a white-light source at (0.33, 0.33) Yw Xw = 0.33; = 0.33; Xw + Yw + Zw Xw + Yw + Zw
and
Zw = 1 − 0.33 − 0.33. Xw + Yw + Zw
Hence, one can obtain Y r :Y g :Y b = 5.02:28.857:1 at 85◦ C. In order to eliminate the thermal effect causing decay of the emission intensity, the initial luminance (25◦ C) ratio of RGB LEDs would be Yr Yg Yb : : = 7.53 : 28.857 : 1. (1 − 0.4) (1 − 0.1) (1 − 0.1)
Light-emitting diodes
161
6.5 Device fabrication Fabrication of a LED starts from epitaxial growth on a crystalline substrate. There are several epitaxy techniques, such as liquid phase epitaxy (LPE), vapor phase epitaxy (VPE) and metalorganic chemical vapor deposition (MOCVD). LPE is the simplest and most economic technique for obtaining a thick epilayer under thermal dynamic equilibrium conditions, which also limits the semiconductor compositions near the lattice-match condition. The thickness of the epilayer by LPE is controlled by the contact time of the III–V solution and the solid substrate. The fast growth rate results in difficulty in obtaining thin layers, such as QW structures. VPE provides greater flexibility to tune the alloy composition by controlling the gas flow of group III and V materials independently in the vapor phase. Distinguished by their vapor sources, there are chloride, hydride and organometallic VPE methods. Typically, VPE with organometallic sources is typically called MOCVD which is one of the most common ways for obtaining complex layer thicknesses, compositions and doping profiles. Once a high-performance epilayer is obtained, the next step is to fabricate a p–n junction. LED fabrication is relative simple, compared to other III–V devices such as laser diodes and transistors. Suitable electrode materials should be selected and post-annealing may be needed for obtaining an ohmic contact and low serial resistance. Current spreading is used to obtaining uniform emission from the chip. Electrode layout should be well designed since the reflective metal blocks light outcoupling. For improving the extraction efficiency, the epitaxial layer should be well designed to fully open the ‘escape cone’ from the bottom, top and the side walls. Chip shaping and surface engineering are helpful to redirect the light and destroy the waveguiding effects for radiating out from the semiconductor to the epoxy encapsulation layer. The refractive index of the epoxy is as high as possible to increase the critical angle between the semiconductor–epoxy interfaces. And the nonplanar shape of the epoxy is designed to improve the light extraction from the epoxy to the air. Package process is important since it not only protects the LED chips and improves the extraction efficiency, but also improves the thermal dissipation which is an important issue for high-power applications.
6.5.1 Epitaxy LPE is a useful epitaxial technique for LED mass production, due to its simple process, low cost and high growth rate. A schematic illustration LPE is shown in Figure 6.23.32 Different solutions containing saturated solutions of molten mixtures (e.g. Ga, As, Al) are placed in different source holders (solution bins). GaAs substrate is put on a holder which can contact different melts for obtaining different layer Graphite cover
Solution bins
Slider Push rod (quartz)
Substrate in recess Substrate holder Figure 6.23 Schematic diagram of LPE system.32 (Source: Kupha, E. (1991) Liquid phase epitaxy. Appl. Phys. A, 52, 380 (1991). With kind permission of Springer Science and Business Media)
162
Introduction to Flat Panel Displays
compositions and doping profiles by moving the push rod. The temperature decreases when the sources and the GaAs substrates contact, and the atoms from the melt in the liquid phase epitaxially grow on the solid substrate under thermal equilibrium condition. The thickness of each epilayer is controlled by the temperature and the contact time. Since the growth temperature is high (about 700 to 900◦ C), transfer time among different melts should be as short as possible to avoid oxidation reaction occurring which forms defects at the layer interfaces. LPE has a very high growth rate (up to 2 m min−1 ) which is suitable for mass production. However, this also causes difficulties in epitaxial growth of thin layers, such QW structures. Since different atoms (e.g. Al, In, Ga and P) exhibit different distribution coefficients (which is defined as the ratio of concentrations in the solid to the liquid under thermal equilibrium condition), it is difficult to grow AlGaInP layers over GaAs using this technique. It is also difficult to use LPE for III–N systems since nitride is in the gas phase, rather than in the liquid phase. However, LPE works well for the AlGaAs system for red LEDs. VPE uses gaseous sources, rather than the liquid sources of LPE. Hence, it is easier to control the epitaxial thickness and doping profile, and to grow multilayer structures by adjusting the source gas flow. There are different kinds of VPE systems which can be distinguished by their sources, such as chloride and hydride VPE, and MOCVD, as shown in Table 6.2. Due to the low melting points of the vapor sources, it is possible to achieve ‘vapor phase’ growth even at the temperature range of 600–800◦ C. Also, it is obvious that some chemical reaction happens due to the use of these vapor sources. Chloride VPE uses solid and gas sources for group III and V, respectively. Group V chloride first reacts with H2 and forms group V molecules and HCl in the gas phase: 4VCl3 + 6H2 → V4 + 12HCl,
(6.35)
where V means group V materials (e.g. As, P and N). The resulting HCl reacts with the atom or binary compound to obtain the chloride group III sources. In he binary compound case, one can write 4IIIV + 4HCl → 4IIICl + V4 + 2H2 ,
(6.36)
where III means group III materials (e.g. In or Ga). Note that this reaction is reversible. When the temperature is high, the reaction tends to shift to the right (which is at the binary compound source region), so the solid sources of group III are etched and form the group III gas sources. Then, at low temperature (where the substrate is placed), the reaction tends to the left forming the epitaxial layer upon the crystalline substrate. Figure 6.24(a) shows a schematic diagram of a chloride VPE system.33 Note that there are successive reactions in chloride VPE, which means the amounts of group III sources are determined by the group V source. In hydride VPE, the group V sources are replaced by AsH3 and PH3 , which are then cracked at high temperature to form As4 and P4 .34 Also, HCl gas is injected to react with group III metal for obtaining the group III gas source: 2III + 6HCl → 2IIICl3 + 3H2 . Table 6.2
Sources used for different VPE systems.
Chloride VPE
Hydride VPE MOCVD
(6.37)
Group III
Group V
Gas flow
Atoms (In, Ga, etc.) or binary sources (GaAs, InP, InAs, etc.) Atoms (Ga, In, etc.)
Chloride (AsCl3 , PCl3 , etc.)
H2
Hydride (AsH3 , PH3 , NH3 , etc.) Hydride (AsH3 , PH3 , NH3 , etc.)
HCl and H2
Organometallic materials (TMGa, TMAl, TMIn, etc.)
H2 , N2
Light-emitting diodes
163
TWO TEMPERATURE ZONE FURNACE THERMOCOUPLE WELL 790°C
THERMOCOUPLE WELL
950°C
ASH3 + H2 PH3 + H2 H2Se + H2
QUARTZ REACTOR TO TUBE EXHAUST
SEED
Ga or GaAs
MIXING CHAMBER H2 Zn + H2
H2Se + H2
SUBSTRATE
AsCI3
PCI3 HCl + H2
Ga
GALLIUM ZONE 775°C
REACTION ZONE 850°C
DEPOSITION ZONE 750°C
CONTINUES TO EXHAUST AND STOPCOCK
PURIFIED HYDROGEN
(a)
(b)
Figure 6.24 Schematic diagram of (a) chloride33 (reproduced from: Finch, W.F. et al. (1964) J. Electrochem. Soc., 111, 814) and (b) hydride (reproduced from: Tietjen, J.J. et al. (1966) J. Electrochem. Soc., 113, 724) VPE systems.34
Hence, hydride VPE (shown in Figure 6.24(b)) can independently control the amount of group III and V sources and is more flexible for epitaxial growth compared with chloride VPE. Deposition reaction for hydride VPE is the same as for chloride VPE. One of the disadvantages of hydride VPE is the use of the toxic sources of arsine (AsH3 ) and phosphine (PH3 ). By using metal alkyls which are liquid or waxy solids at room temperature, the melting points of the group III sources can be greatly reduced. The epitaxial growth technique with metal alkyls and hydrides for group III and V, respectively, is called MOCVD, or organometallic VPE (OMVPE), as shown in Figure 6.25. By using the carrier gas, such as H2 and N2 , to pass through the group III alkyls and bring the vapor into the growth chamber, the flow rate of group III sources can be controlled precisely. Typical group III sources are: trimethylaluminum, Al(CH3 )3 (TMAl); trimethylgallium, Ga(CH3 )3 (TMGa); and trimethylindium, In(CH3 )3 . The MOCVD reaction can be represented as xAl(CH3 )3 + yGa(CH3 )3 + zIn(CH3 )3 + VH3 → Alx Gay Inz V + 4CH4 ,
(6.38)
where V means group V materials (e.g. As, P and N). Note that the reaction is irreversible which means MOCVD is a kind of nonequilibrium growth. The advantages of this growth technique include: (1) possibility for abrupt junctions and thin epilayers and (2) large flexibility in varying the compound Substrate Susceptor TMGa + TMAl + H2
NH3 + H2 Reactor
RF coil
Figure 6.25 Schematic diagram of a MOCVD system.35 (Source: Hirosawa, K., Hiramatsu, K., Sawaki, N. and Akasaki, I. (1993) Growth of single crystal AlxGa1-xN films on Si substrates by metalorganic vapor phase epitaxy. Jpn. J. Appl. Phys., 32, L1039. Reproduced by permission of the Institute of Pure and Applied Physics)
164
Introduction to Flat Panel Displays
composition. These features are very important in LED applications when quantum structures are applied and different emission wavelengths are required.35 The precise rate control implies a low growth rate (0.1–10 m h−1 ) and can be viewed as a disadvantage of MOCVD.
6.5.2 Process flow and device structure design Figures 6.26(a) and (b) show typical device structures of III–P and III–N LEDs, respectively. Carriers inject into the LED from the electrode to the p- and n-layers, and then recombine in the active region. Photons are generated and radiate out of the LED structures from the top and both sides for III–P and III–N LEDs, respectively. There are several requirements for the electrode design. First of all, good ohmic contacts between the electrodes and p- and n-layers are needed for good carrier injection. Suitable metal materials and annealing processes can increase the conductivity and reduce the power consumption. Then, the current should pass uniformly through the active layer, which means the resistance of the p- and n-layers should be as small as possible. However, this is not always the case, especially for the wide-bandgap p-type doping. By suitably designing the layout of the electrodes, the carriers can distribute uniformly which results in a dramatic increase in luminous efficiency. A concern for the layout design of the electrodes is that the area of the metal electrode should be as small as possible since it impedes light extraction. Due to the use of insulating sapphire substrate, electrodes of nitride LEDs are fabricated on one side of the substrate, which means mesa etching is required. Nitride semiconductors are rigid which means a strong solution for wet etching or reactive ion etching technique are needed for this step. Then, the LED substrate is diced into individual chips for the package process, which is discussed in Section 6.5.4. Prior to the dicing process, sometimes a wafer thinning process is used to improve the production yield. Theoretically, direct contact of a metal and semiconductor results in a Schottky diode, rather than an ohmic contact. The barrier height between the metal work function and the carrier transport band (conduction and valence bands for electron and hole, respectively) limits the current injected into the LED. By mixing dopant metal into the electrode sources and annealing the sample, the dopant metal is driven into the semiconductor and reacts with the cladding layer to form a thin heavily doped layer. Hence, an ohmic contact between electrode and semiconductor can be achieved.14 For AlInGaP and AlGaAs systems, Au–Ge and Au–Zn (or Au–Be) alloys are typically used for the n- and p-type electrodes with dopant concentrations (Ge, Zn, Be) of about 1–15 %, respectively. For p-type contacts, heavy
Top view
P-electrode
P-electrode
N-electrode InGaN epilayers
AlGaInP epilayers Side view N-Substrate
Sapphire Substrate
N-electrode (a)
(b)
Figure 6.26 Device structures of (a) III–P and (b) III–N LEDs.36
Light-emitting diodes
165
doping can be applied during epitaxial growth (e.g. p+ -GaAs, p+ -GaP and p+ -AlGaAs) to improve the conductivity. However, for the nitride LEDs, heavily doped layers are difficult to achieve. Hence, another approach is to reduce the barrier height of the Schottky junction by using a metal with suitable work function. Ti/Al bilayer and Ni/Au are typical metals used for n- and p-type metallization, respectively. Ti reacts with nitride and forms TiN which is conductive. Ni and Au exhibit high work functions which reduce the barrier between the electrode and the semiconductor. Typically, the dimensions of LED chips are much larger than the thickness of the p-cladding layer. Hence, holes inject from the anode and recombine with electrons mostly under the opaque metal region, which results in low efficiency. Also, since the recombination rate is proportional to the product of electron and hole concentrations, the nonuniformity of hole distribution results in lower recombination efficiency. One simple method is to increase the thickness of the p-cladding layer, to improve the ‘lateral’ current flow. This thick layer also serves as the window layer for high extraction efficiency, which is discussed in the next section. Although this technique works well with the AlGaAs system, the high resistance of p-type doping of wider bandgap materials (e.g. AlInGaP and III–N systems) results in higher serial resistance with increasing layer thickness. Another possible solution is to pattern the electrode. By using some ‘finger’ structures to assist the current spreading, the carrier distribution uniformity can be much improved. However, the area occupied by the opaque electrode should be as small as possible since the metal impedes light extraction. For nitride LEDs, since an insulating substrate is used, the anode and the cathode are on the same side of the substrate, which results in even worse current nonuniformity. Hence, both the p- and n-electrode layout should be taken into consideration. Transparent or semitransparent electrodes such as ITO and thin metals (e.g. Ni/Au or Pt) can be used on the whole area of the devices to improve the current spreading in the p-type region. On the other hand, for III–P LEDs, the ITO can form an ohmic contact only with GaAs cladding layer. Since it is absorptive within the visible region, the GaAs layer should as thin as possible to reduce the absorption loss.
6.5.3 Extraction efficiency improvement As mentioned in Section 6.4.2, the extraction efficiency of LEDs is limited by the internal reflection when the light radiates out of the semiconductor planar structure. To improve this, there are typically two methods used: (1) redirecting the light by shaping the whole device and (2) using a nonplanar structure to couple out the waveguiding modes. Combining with the package process (discussed in Section 6.5.4) which increases the critical angle and redirects the light, the extraction efficiency can be effectively increased. Besides, for III–P devices, the photons will be absorbed by the GaAs substrate. It is possible to use a distributed Bragg reflector (DBR) structure between the substrate and the epilayer to prevent light absorption by the substrate, which can act as wavelength selector at the same time. However, this alternative structure increases the complexity and cost in epitaxial growth. There are also some methods to fabricate transparent substrates, instead of absorbing ones. Another issue arises from the metal electrodes which may impede light output. Suitable layout design and device structure are needed to obtain a high extraction efficiency. From Snell’s law, one can obtain the critical angle (θ ≈ 25.38◦ ) between the semiconductor (typically nsemi ≈ 3.5) and the outside medium (typically epoxy in package process, nepo ≈ 1.5) and hence define the ‘escape cone’ shown in Figure 6.19. Assuming the LED is a perfect rectangular solid and light emission is purely isotropic, the six full escape cones are the maximum attainable light extraction. For the absorbing substrate case, obviously the downward escape cone does not exist. Physical dimensions of LEDs are around several hundred micrometers in width and length. The thickness of the active layer is around 1 m. If the thicknesses of the layers on both sides of the active layer (typically called the cladding layers) are not thick enough (>30 m) or they are not transparent, the four side cones will be truncated. However, to grow such thick epilayers needs a long time which is not feasible when using the MOCVD technique with a growth rate of ∼1 m h−1 .
166
Introduction to Flat Panel Displays
For the GaAs/AlGaAs system, the thick cladding layer of large bandgap can be grown by LPE or VPE techniques. By growing a AlGaAs layer of hundreds of micrometers between the active layer and the absorbing substrate, then removing the original GaAs substrate by etching process, a transparent substrate of AlGaAs can be obtained. For the III–P/GaAs system, an upper cladding layer and substrate of transparent GaP is typically used, although there is still a 3.6 % lattice mismatch to the GaAs. By using MOCVD, GaP cladding can be only as high as 15 m which is not sufficient to open the complete side cones. However, following the MOCVD growth of the active layer, VPE growth of an upper cladding layer of 50 m has been demonstrated. GaAs substrate can also be removed by etching process, then bonded to the GaP substrate under high-temperature annealing.37 The resulting device structure is shown in Figure 6.27. A transparent substrate is an intrinsic property for the III–nitride devices. However, due to the difficulties in doping, it is not easy to have a thick cladding layer. Besides, a thick cladding may not be very helpful in extraction efficiency because of the high internal absorption loss in the active layer. The light is absorbed by the active layer before it propagates to the edge. Hence, no thick cladding layer is needed. However, as said before, due to the poor conductivity (especially for p-doping), a metal ‘mesh’ electrode is typically used to improve the luminance uniformity which impedes light emission. Hence, a flip-chip structure can be used, as shown in Figure 6.28.38 With the upside-down structure which emits the light from the transparent sapphire, good conductivity and high extraction efficiency can be obtained simultaneously. The discussions above relate to extracting as much light as possible from a rectangular cubic shape. Actually, it is found that the side cones can extract more light than expected which arises from the nonperfectly planar structure at the edge sides. Besides, the shape of the die can further improve the extraction efficiency by redirecting light to the top-side emission. For example, a trapezoid-shaped structure can effectively reflect the sidewall emission to the top which increases the extraction efficiency. Figure 6.29 shows the evolution of the improvement of light extraction efficiency from a conventional absorbing substrate, a transparent substrate, a ‘finger’ electrode and trapezoid die structures. Much effort has been made to eliminate the total internal reflection condition in the top side of LEDs. A straightforward method is to roughen the top surface which results in light scattering, rather than reflection. Some periodic nanostructures, such as surface plasmonic coupling and photonic structures, are proven which can effectively diffract out the waveguiding mode. However, there are still some techniques issues in terms of fabrication process, cost and device layout.
P-electrode
P-electrode Thin p-cladding
Thick p-cladding
Active layer Active layer n-type GaAs substrate (Absorbing)
n-type GaP substrate (Transparent)
N-electrode N-electrode (a)
(b)
Figure 6.27 (a) Absorbing substrate with thin cladding layer which has one full (upward) plus four partial (sidewall) escape cones, and (b) transparent substrate with thick cladding layer which has six large escape cones.
Light-emitting diodes
167
Light
Silver reflector
GaN Wire bond
Sapphire
Metalic currentLight spreading layer
p-GaN n-GaN Sapphire
Wire bond Metal reflector (a)
Silicon submount
Integrated ESD-protection circuit (b)
Figure 6.28 (a) Mesh electrode (metallic current spreading layer) and (b) flip-chip III–nitride LEDs.38 (Source: Steigerwald, D.A., Bhat, J.C., Collins, D. et al., Illumination with solid state lighting technology. IEEE J. Sel. Top. c 2002 IEEE) Quantum Electron., 8, 310.
Figure 6.29 Images of LEDs under operation with different geometries: (a) absorbing substrate, (b) transparent substrate, (c) finger electrode arrangement and (d) die shaping structure.38 (Source: Steigerwald, D.A., Bhat, J.C., Collins, D. et al., Illumination with solid state lighting technology. IEEE J. Sel. Top. Quantum Electron., 8, 310. c 2002 IEEE)
6.5.4 Package The package process involved in typical semiconductor fabrication has the functions of: (1) protecting the LED and electrode from mechanical and environmental attacks, (2) improving the optical extraction efficiency and (3) helping to dissipate the heat which is especially important for high-power applications. Figure 6.30 shows a packaged LED, a semiconductor chip with dimensions of several hundreds of micrometers mounted on a reflector cup, which redirects the light to the top. Bonding wire is used to
168
Introduction to Flat Panel Displays
Epoxy
Wire Bonding
Cathode Anode
Reflector
LED chip
Figure 6.30 Schematic of a packaged LED.
connect the semiconductor and the external electrodes for providing the electrical current. Here, a LED with semiconducting substrate (e.g. AlGaInP on GaAs substrate) is used as an example. Epoxy seals the semiconductor, reflector and bonding wire inside for protection. The shape of the epoxy is typically designed as a hemisphere for enhancing the extraction efficiency at the epoxy–air interface. At the same time, for improving the light extraction at the semiconductor–epoxy interface, the refractive index of epoxy should be as high as possible with low absorption within the visible range to enlarge the escape cone. Typical values of the refractive index of epoxy are around 1.5 to 1.8. To increase LED optical power, more current should be injected into the device; however, this increases temperature, and, in turn, decreases the optical intensity and red-shifts the emission wavelength. Hence, for high-power LED applications, thermal resistance has to be taken into consideration in designing the package.38 Typically, the LED chip is directly mounted on a heatsink which dissipates the heat from the ‘bottom’ side. Also, silicone rather than epoxy is often used due to its better thermal stability.
6.6 Applications In earlier years (1960s to 1980s), the applications of LEDs were limited to indicators, alphanumeric displays and simple dot matrix displays for consumer products or industrial use. Also, the colors of LEDs were limited to the long wavelengths of the visible range (such as red) due to the lack of suitable semiconductor materials. Recently, because of the fast developments in epitaxial growth of semiconductor materials and device efficiency improvements, high-performances LEDs have become available which can be used for many different applications. For direct-view applications, LEDs can be used for traffic lights, electronic signage and very large displays. The main concerns for the direct-view displays are the high efficiency (for low power consumption) and correct CIE coordinates. Also, LEDs can be used as the backlight of LCD displays, due to the advantages of their being mercury free, of high color gamut and having a long lifetime and fast response, compared with conventional CCFL backlights. For small LCD displays (such as mobile phones), color gamut is not a main issue. Hence, a white LED (blue + phosphors) can be enough as the backlight source to reduce the cost and minimize the module size. For medium- or large-size LCDs, multicolor LEDs are needed to increase the power efficiency and color gamut. For these applications, the FWHM of the LEDs should be as small as possible to increase the color gamut value. Optical characteristics of LEDs change (intensity and spectrum) due to the ambient temperature and operation time, which are also important for the display qualities. The third category of LED application is general lighting, which can be considered as the light source of reflective displays. For this application, not only should the CIE coordinates of the white LEDs be considered, but a high CRI value is also important which provides objects with a similar look under LED and natural light. However, as discussed in Section 6.2.3, there is a tradeoff between the power efficiency and CRI value.
Light-emitting diodes
169
6.6.1 Traffic signals, electronic signage and huge displays Definition of colors, luminous intensity and luminance of traffic signals are important in safety issues. Different countries (e.g. the USA, Europe and Japan) have similar but different standards. For example, in the USA, the specifications of traffic signals are regulated by the Institute of Transportation Engineers (ITE), which defines the acceptable color ranges of five distinct colors for traffic signal lighting: red, yellow, blue-green, Portland orange and lunar white. The first three are used for the standard red, yellow and green lights, which mean ‘stop’, ‘caution’ and ‘go’. The last two colors are used for pedestrian control signals, which are ‘don’t walk’ and ‘walk’, respectively. Using LEDs as traffic signals to replace conventional filtered incandescent lamp has two obvious advantages: lower power consumption and longer lifetime. The power efficiency of a typical incandescent source is about 14 lm W−1 . To obtain a red traffic light, the green and blue portions are filtered out which results in a power efficiency of 3–5 lm W−1 . For the AlInGaP red LEDs, the power efficiency can be as high as 30 lm W−1 , which means a tenfold reduction in power consumption. Compared to the 2- to 3-year lifetime of incandescent lamps, the 10-year operation lifetime of LEDs can further save on maintenance fees. The basic structure of a filtered incandescent lamp is a bulb with a color filter. In contrast, LED traffic signals consist of hundreds of LED pixels (∼200 to 700 pixels). Failure of one or several LEDs would not affect much the functionality of the traffic signal. Once the bulb is burned out, the incandescent traffic signal fails and replacement is needed. Although the unit ‘luminous’ price of LEDs is higher than that of the incandescent lamps, the energy and maintenance savings result in total cost savings by using LED technology as traffic signals. LEDs with AlGaInP are typically used for red, orange and yellow traffic lights. InGaN LEDs are used for green and white (blue or UV + phosphor) traffic lights. There are two disadvantages of LED traffic signals are: (1) the light output variation at different ambient temperatures and (2) the driving circuit complexity. As mentioned in Section 6.4.4, the output intensity decreases and spectrum red-shifts with increasing temperature. Over the range of typical ambient temperature (−40 to 55 ◦ C), the CIE shift is not obvious, which will not exceed the range with proper device design. In contrast, the output intensity variation is obvious, which can be compensated by suitable circuit design. Typically, an incandescent lamp is driven by 120 V AC power provided by a traffic signal controller. To use the same controller, an AC–DC converter is needed and a string of LEDs are series connected. A voltage regulator, limiting resistor or constant current sources are needed to provide a uniform light output. To provide some redundancy, which means the random failure of a LED pixel would not affect the functionality of the traffic signal, some parallel connections between different series string are also needed. Electronic signage and huge displays are other two important applications for LEDs. Due to the fast improvement in LED technology, the colors of electronic signage have improved from monochromatic (typically red), to multiple colors (red, yellow and green) and then to full color (red, green and blue). Compared with conventional electronic signal technologies using small filtered incandescent bulbs, LEDs exhibit the advantages of low power consumption and long lifetime. Also, the resolution can be higher using LEDs due to the small module size. Since the LED can be well passivated by the package process, another advantage is the ruggedness, without the glass envelopes and filaments associated with conventional bulbs. Huge display applications consist of millions of pixels. There are three or four LEDs (red, green and blue) in one single pixel. To achieve maximum white light efficiency, sometimes two red LEDs are needed in one single pixel. Due to the high brightness of the LEDs, huge displays can be viewed even in sunlight.
6.6.2 LCD backlight Using LEDs to replace conventional CCFLs as the backlight of LCDs has several obvious advantages of low power consumption, long lifetime, small size and ruggedness, as described above. Compared to the several hundred to thousand volts driving CCFLs, LEDs have a much lower driving voltage
170
Introduction to Flat Panel Displays
(<5 V) which is more suitable for mobile, monitor and TV applications in driving-circuit design and safety issues. Due to the fast response of LEDs, it is very easy to switch the LED backlights on and off to insert the ‘black frames’ to eliminate motion blur behaviors in the hold-type CCFL-LCD due to its impulse-type characteristics. Also, one of the important advantages for LEDs compared to CCFLs is that they are mercury free and environmentally friendly. However, optical requirements of LEDs for LCD backlights is not only the correct CIE coordinates (like the direct-view application described in the previous section), but also: (1) a suitable and narrow emission spectrum to fit the liquid crystal and color filter and (2) luminance uniformity over the whole panel. Figure 6.31 shows the emission spectra of a CCFL and RGB LED, and the transmittance spectra of RGB color filters. The emission spectrum of the CCFL has multiple peaks, although the shark peaks at 434, 542 and 610 nm correspond to red, green and blue colors, respectively. The peaks at 486 and 585 nm near the overlap of blue–green and green–red color filters result in a decrease of color saturation. Typically, a CCFL LCD has a color gamut of 72 % NTSC. To enlarge the color gamut, one has to decrease the FWHM of the transmission spectra of the color filter. This requires thicker color filter films which results in a decrease in transmission and hence lower efficiency. In contrast, a RGB LED consists of a three-color light source with their FWHMs (typically <50 nm) even smaller than the color filters which results in a high color saturation (105 % NTSC).39 Different from CCFLs, red, green and blue LEDs can be driven individually which makes it possible to light different colors in series to achieve a filter-less LCD. This is called the color sequential or field sequential technique, as shown in Figure 6.32.40 In this example, the pixel was intended to display a yellow color, which is a mixture of red and green subpixels. In a conventional CCFL backlight LCD, the liquid crystal lets the red and green subpixel light pass through and blocks the blue light. After the color filter, one can see the yellow color due to the simultaneously lit red and green subpixels. On the other hand, in color sequential driving, blue, green and red LEDs are lit on one-by-one at different time frames. Hence, no subpixel is needed. In this case, the liquid crystal switch is closed only when the blue light is turned on, which makes the red and green rays pass though. Displayed colors come from the color mixtures of LEDs, rather than the color filters. So the color filter is not needed in this scheme. Without the color filter, the fabrication process is simpler and panel cost is reduced. Also, since the color filter absorbs two-thirds and allows only one-third of the light passing through, the color sequential technique can effectively decrease the power consumption. However, since different colors are lit on in series, then the response of liquid crystal materials must be fast enough. Typically, the methods discussed above are suitable for medium and large LCDs, such as for monitor and TV applications. For some small-size LCDs (especially for mobile use), module size is more important than color performances and hence LEDs with phosphor technology will be applied.
Relative Intensity (%)
80
CCFL B-CF G-CF
60 40 20 B-LED
0 400
Figure 6.31
R-CF
450
G-LED
500 550 600 Wavelength (nm)
R-LED
650
Color Filter Transmittance (%)
100
700
Emission spectra of CCFL and RGB LED, and transmittance spectra of RGB color filters.
Light-emitting diodes
171
B
G
R CF-less CF
LC
LC
Red LED Green LED Blue LED
CCFL (a)
(b)
Figure 6.32 (a) Conventional driving and (b) color sequential technique. 2.5 2
2
1.5
1
0.5
0
1.5
1.5 1
1
0.5
0.5
0
0
–0.5
–0.5
–1
–1
–1.5
–1.5
–2
–2
–2.5
–2.5 –3 2.5
–0.5 –1 –1.5 –2 –2.5 2
2
1.5
1
0.5
0 (a)
Figure 6.33
–3 –0.5 –1 –1.5 –2 –2.5
2.5 2
2
1.5
1
0.5
0
–0.5 –1 –1.5 –2 –2.5 2
1.5
1.5 1
1 0.5
0.5 0
0 –0.5
–0.5
–1
–1
–1.5
–1.5
–2
–2
–2.5 –3 2.5
–2.5 2
1.5
1
0.5
0
–3 –0.5 –1 –1.5 –2 –2.5
(b)
Simulation results of luminance distributions of (a) CCFL and (b) LED backlights.
However, since a LED is a point-type light source, optical design is more complex than for a CCFL for obtaining a uniform luminance over a panel. Figure 6.33 shows the simulation results of luminance distributions of the backlight from a conventional linear-type CCFL and a point-type LED. The light source is hidden in the edge of the top side. For the CCFL, the luminance decreases gradually from the top to the bottom sides. On the other hand, obvious high luminance spots can be observed from the LED chips, which results in a serious nonuniformity. Such a phenomenon is even more serious for the ‘direct’ backlight module, which is an essential technique for large LCDs. Figure 6.34 shows an example of a LED direct backlight module which uses a printed dot pattern to prevent the bright spots directly emitting from the LEDs. Also, the epoxy dome is carefully designed to couple the light through the light guide and diffuser, rather than emitting towards the liquid crystal layer. After these optical designs, a uniform luminance distribution from LED backlight system can be
172
Introduction to Flat Panel Displays
Luminance Uniformity Level (without dot dot printing pattern)
Leaked light Diverter Plate Dot printing
Luminance Uniformity Level (with dot printing pattern)
Reflected light
Reflector
Expanded side view of this photograph Lumileds lighting, Luxeon side emitting Figure 6.34 LED backlight modules.39 (Source: Kakinuma, K. (2006) Technology of wide color gamut backlight with light-emitting diode for liquid crystal display television. Jpn. J. Appl. Phys., 45, 4330. Reproduced by permission of the Institute of Pure and Applied Physics)
obtained. Also, the thermal stability and operation lifetime problems should be taken into consideration. LCD temperature increases during operation which results in luminance decay and wavelength red-shifts. Since the operation lifetimes of RGB LEDs are different, differential aging also results in luminance decay, color shift and panel nonuniformity. Some detectors can be implemented on the panel to compensate the brightness loss and correct the color performances.
6.6.3 General lighting The requirements for general lighting include high power efficiency and high CRI. Hence, a light source with broad-band spectrum is needed. Artificial light sources began with fire, including torches, candles and gas lighting. However, the efficiency is low (<1 lm W−1 ) which can be understood from the large heat dissipation accompanying the light emission, which explains why people feel warm under this kind of yellowish bright environment. Typically, this lighting technology involves transfer from chemical to optical energy. Incandescent bulbs and fluorescent lamps are two common lighting sources using electrical input. The input of the bulb is electric power which heats up a tungsten filament in a vacuum without burning it as a torch or a candle. The temperature of the tungsten (about 2856 K) wire determines its color due to its blackbody radiator characteristics (so it is a broad-band source with very good CRI). Actually, the peak wavelength of the bulb is in the infrared rather than the visible region, which results in the low efficiency (so it is hot) and yellowish color (so it is a ‘warm’ light) due to lower color temperature. The lifetime of a bulb is typically several thousands of hours. During operation, the tungsten will be evaporated by the high temperature to the inner side of the glass and the tungsten wire becomes thinner and thinner, and then burns down. Lower temperature results in longer lifetime. However, it shifts the emission peak to longer wavelength and results in lower power efficiency, typically less than 15 lm W−1 .
Light-emitting diodes
173
The operation principle of fluorescent lamps is based on the energy relaxation during gas discharge under low pressure. UV emission of the gas (e.g. mercury vapor) from the excited state to the ground state pumps the phosphors to generate visible light. The efficiency of the fluorescent lamp can be several times higher than that of an incandescent lamp, which can be observed in that the temperature of the fluorescent tube is much lower than the bulb. The lifetime of a fluorescent lamp can be as high as several tens of thousands of hours. However, due to the sharp characteristics peaks, the CRI value of a fluorescent lamp ranges typically from 50 to 80. Because of the fast development in the efficiency of LEDs, they are becoming a promising technology for general lighting applications. For obtaining a good CRI, blue and UV nitride LEDs with phosphors are typically used to broaden the spectrum. Multi-LEDs can be used for higher efficiency. However, since the FWHM from a single LED is about 50 nm, more than three LEDs or suitable layer structure may be needed to broaden the spectrum. Long lifetime (predicted >100 000 hours) and good resistance to the environment are two obvious advantages of LED lighting, compared with bulbs and fluorescent tubes, as discussed in Section 6.6.1. Also, LED lighting has the advantage of being environmentally friendly. Compared with incandescent bulbs, LEDs have a much higher efficiency which results in lower power consumption. Hence, less electric power is needed and carbon oxide emission is reduced. Power saving for the lighting industry is very important since a quarter of all electrical power is used for lighting purposes (including domestic, industry and outdoors). Although the efficiency of fluorescent lamps is comparable to (or slightly less than) that of LEDs, the mercury vapor (used in fluorescent lamps) results in serious environmental problems.
Homework problems 6.1 Derive the following equation: Cdep ≡
ε dQ = dV W
Here, Cdep is the capacitance per unit area of depletion region. Q and V are charge per unit area and voltage, respectively. Assume the electric field change (dE) is uniform within depletion region. 6.2 A p-type semiconductor is uniformly photo-excited at t < 0. Then, the excitation is turned off at t = 0+ , please find the time-dependent minority carrier concentration np (t).
Hint: continuity equation:
∂np (x, t) 1 ∂Je (x, t) np (x, t) − = gext − ∂t q ∂x τmin
where np (x, t) is the excess concentration of minority carrier, J e (x, t) is the electron current density, τmin is the electron carrier lifetime, gext is the net generation rate of the excited source, and np0 is the electron concentration under thermal equilibrium. 6.3 An LED biased at 3.2 V with the current 20 mA. Luminous intensity of this LED is 0.3 cd (or 300 mcd). Assume this LED is a Lambertian emitter with the single emission wavelength at 470 nm. What are the current efficiency (cd/A), power efficiency (lm/W), and photon number per second, of this LED? Given V(λ) = 0.09 at 470 nm. 6.4 What is the maximum attainable luminous efficiency of D65 (x = 0.313, y = 0.329) light source by mixing LEDs (assuming each LED emits at a single wavelength)? Is it a di- or tri-chromatic white light?
174
Introduction to Flat Panel Displays
References 1. Round, H.J. (1907) A note on carborundum. Electrical World, 19, 309. 2. Holonyak Jr, N., and Bevacqua, S.F. (1962) Coherent (visible) light emission from Ga(As1−x Px ) junctions. Appl. Phys. Lett., 1, 82. 3. Groves, W.O., Herzog, A.H. and Craford, M.G. (1971) The effect of nitrogen doping on GaAs1−x Px electroluminescent diodes. Appl. Phys. Lett., 19, 184. 4. Rupprecht, H., Woodall, J.M. and Petit, G.D. (1967) Efficient visible electroluminescence at 300 K from Ga1−xAlxAs p–n junctions grown by liquid-phase epitaxy. Appl. Phys. Lett., 11, 81. 5. Kuo, C., Fletcher, R., Osentowski, T. et al. (1990) High performance AlGaInP visible light-emitting diodes. Appl. Phys. Lett., 57, 2937. 6. Amano, H., Sawaki, N., Akasaki, I. and Toyoda, Y. (1986) Metalorganic vapor phase epitaxial growth of a high quality GaN film using an AlN buffer layer. Appl. Phys. Lett., 48, 353. 7. Nakamura, S., Senoh, M. and Mukai, T. (1991) Highly P-typed Mg-doped GaN films grown with GaN buffer layers. Jpn. J. Appl. Phys., 30, L1708. 8. Steranka, F.M., Bhat, J., Collins, D. et al. (2002) High power LEDs: technology status and market applications. Phys. Stat. Sol. (a), 194, 380. 9. Schubert, E.F. (2006) Light-Emitting Diodes, 2nd edn, Cambridge University Press. 10. Zukauskas, A., Shur, M.S. and Gaska, R. (2002) Introduction to Solid-State Lighting, John Wiley & Sons, Ltd. 11. Mueller, G. (ed.) (2000) Electroluminescence I, Semiconductor and Semimetals, Vol. 64, Academic Press. 12. Muthu, S., Schuurmans, F.J.P. and Pashley, M.D. (2002) Red, green, and blue LEDs for white light illumination. IEEE J. Sel. Top. Quantum Electron., 8, 333. 13. Mueller-Mach, R., Mueller, G.O., Krames, M.R. and Trottier, T. (2002) High-power phosphor-converted lightemitting diodes based on III-nitrides. IEEE J. Sel. Top. Quantum Electron., 8, 339. 14. Stringfellow, G.B. and Craford, M.G. (eds) (1997) High Brightness Light Emitting Diodes, Semiconductor and Semimetals, Vol. 48, Academic Press. 15. Streubel, K., Linder, N., Wirth, R. and Jaeger, A. (2002) High brightness AlGaInP light-emitting diodes. IEEE J. Sel. Top. Quantum Electron., 8, 321. 16. Schubert, M., Woollam, J.A., Leibiger, G. et al. (1999) Isotropic dielectric functions of highly disordered Alx Ga1−x InP (0≤ x ≤ 1) lattice matched to GaAs. J. Appl. Phys., 86, 2025. 17. Nakamura, S., Mukai, T., Senoh, M. et al. (1993) Inx Ga(1−x) N/Iny Ga(1−y) N superlattices grown on GaN films. J. Appl. Phys., 74, 3911. 18. Lin, Y.S., Ma, K.J., Hsu, C. et al. (2000) Dependence of composition fluctuation on indium content in InGaN/GaN multiple quantum wells. Appl. Phys. Lett., 77, 2988. 19. Feng, S.W., Cheng, Y.C., Chung, Y.Y et al. (2002) Impact of localized states on the recombination dynamics in InGaN/GaN quantum well structures. J. Appl. Phys., 92, 4441. 20. Khan, M.A., Yang, J.W., Simin, G. et al. (2000) Lattice and energy band engineering in AlInGaN/GaN heterostructures. Appl. Phys. Lett., 76, 1161. 21. Nakamura, S., Fasol, G. and Pearton, S.J. (2000) The Blue Laser Diode: The Complete Story, Springer. 22. Sheu, J.K., Chang, S.J., Kuo, C.H. et al. (2003) White-light emission from near UV InGaN–GaN LED chip precoated with blue/green/red phosphors. IEEE Photon. Technol. Lett., 15, 18. 23. Xiang, H.F., Yu, S.C., Che, M. and Lai, P.T. (2003) Efficient white and red light emission from GaN/tris(8-hydroxyquinolato) aluminum/platinum(II) meso-tetrakis(pentafluorophenyl) porphyrin hybrid light-emitting diodes. Appl. Phys. Lett., 83, 1518. 24. Guo, X., Graff, J. and Schubert, E.F. (1999) Photon recycling semiconductor light emitting diode. IEDM Tech. Dig., 600. 25. Chen, H.S., Yeh, D.M., Lu, C.F. et al. (2006) White light generation with CdSe–ZnS nanocrystals coated on an InGaN–GaN quantum-well blue/green two-wavelength light-emitting diode. IEEE Photon. Technol. Lett., 18, 1430. 26. Sze, S.M. (2001) Semiconductor Devices: Physics and Technology, 2nd edn, John Wiley & Sons, Ltd. 27. Chuang, S.L. (1995) Physics of Optoelectronic Devices, John Wiley & Sons, Ltd. 28. Kim, K.C., Choi, Y.C., Kim, D.H. et al. (2004) Influence of electron tunneling barriers on the performance of InGaN–GaN ultraviolet light-emitting diodes. Phys. Stat. Sol. (a), 201, 2663. 29. Peyghambarian, N., Koch, S.W. and Mysyrowicz,A. (1993) Introduction to Semiconductor Optics, Prentice-Hall. 30. Shockley, W. and Read Jr, W.T. (1952) Statistics of the recombinations of holes and electrons. Phys. Rev., 87, 835.
Light-emitting diodes
175
31. Hall, R.N. (1952) Electron–hole recombination in germanium. Phys. Rev., 87, 387. 32. Kupha, E. (1991) Liquid phase epitaxy. Appl. Phys. A, 52, 380. 33. Finch, W.F. and Mehal, E.W. (1964) Preparation of GaAsx P1−x by vapor phase reaction. J. Electrochem. Soc., 111, 814. 34. Tietjen, J.J. and Amick, J.A. (1966) The preparation and properties of vapor-deposited epitaxial GaAs1−x Px using arsine and phosphine. J. Electrochem. Soc., 113, 724. 35. Hirosawa, K., Hiramatsu, K., Sawaki, N. and Akasaki, I. (1993) Growth of single crystal Alx Ga1−x N films on Si substrates by metalorganic vapor phase epitaxy. Jpn. J. Appl. Phys., 32, L1039. 36. http://www.epistar.com.tw/ 37. Kish, F.A., Steranka, F.M., DeFevere, D.C. et al. (1994) Very high-efficiency semiconductor wafer-bonded transparent-substrate (Alx Ga1−x )0.5 In0.5 P/GaP light-emitting diodes. Appl. Phys. Lett., 64, 2839. 38. Steigerwald, D.A., Bhat, J.C., Collins, D. et al. (2002) Illumination with solid state lighting technology. IEEE J. Sel. Top. Quantum Electron., 8, 310. 39. Kakinuma, K. (2006) Technology of wide color gamut backlight with light-emitting diode for liquid crystal display television. Jpn. J. Appl. Phys., 45, 4330. 40. Takahashi, T., Furue, H., Shikada, M. et al. (1999) Preliminary study of field sequential fullcolor liquid crystal display using polymer stabilized ferroelectric liquid crystal display. Jpn. J. Appl. Phys., 38, L534.
7 Organic light-emitting devices 7.1 Introduction The operation principles of organic light-emitting devices (OLEDs) are similar to those of semiconductor light-emitting diodes (LEDs), as described in Chapter 6, except the materials are organic rather than semiconductor materials.1 When a voltage is applied to such a device, holes and electrons are injected from the anode and cathode, respectively, and then recombine to emit light. Energy states in organic materials can be described by molecular orbitals which are combinations of all the atomic orbitals. The electrons fill the molecular orbitals from the lowest energy states so that the highest occupied molecular orbital (HOMO) can be defined. Once the system is excited, the electrons can be promoted to higher molecular orbitals. The first excited state is called the lowest unoccupied molecular orbital (LUMO). The HOMO and LUMO in organic materials are similar to the valence and conduction band in semiconductor materials, in some respects. After the carrier is excited, optically or electrically, to higher energy levels, it relaxes its energy radiatively or nonradiatively. Due to the difference in spin momentum, the singlet and triplet excited states have degenerate energy levels. Typically, the triplet state has slightly lower energy than the singlet state. Radiative recombination from the singlet and triplet states is called fluorescent and phosphorescent emission. Typically, the time constant of radiative recombination for a triplet exciton is much longer than that for a singlet one which results in the low phosphorescence efficiency at room temperature. Charge carrier injection in organic devices is limited by the energy barriers at the metal–organic and organic–organic interfaces which are typically modeled by Richardson–Schottky (RS) thermionic emission.2 Organic molecules stacked together in an amorphous form without a well-defined band structure results in a quite low mobility value, typically less than 10−3 cm2 V−1 s−1 . Carriers are ‘hopping’ among organic molecules in a disordered structure, which means the mobility value increases with applied electric field. In organic materials, since the free carrier density under thermal equilibrium is quite low, carrier transport in an organic thin film is usually described by trap-charge limited conduction (TCLC) and trap-free space-charge limited conduction (SCLC), which is the highest sustainable current in a perfect insulator. By applying an electric field to an anthracene single crystal, Pope et al.3 in 1965 observed blue electroluminescence (EL). However, hundreds of volts of driving voltage were required due to the thick organic layer. The quantum efficiency (in terms of emitted photon numbers per injected carrier) was also very low due to the imbalanced charge injection and transport. In 1987, a two-layer device structure was first introduced by Tang et al. who utilized the thermal evaporation technique in vacuum for organic thin-film deposition in amorphous phase.4 Figure 7.1 shows the device structure. Since the
Introduction to Flat Panel Displays c 2008 John Wiley & Sons, Ltd
J.-H. Lee, D.N. Liu and S.-T. Wu
178
Introduction to Flat Panel Displays
–
+ anode (transparent)
p-type organic material
n-type organic material
cathode (reflective)
Figure 7.1 Two-layer OLED.
organic film is as thin as several tens of nanometers, the driving voltage can be less than 10 V. The two organic layers are: (1) aromatic diamine, which is used as the hole-transporting layer (HTL) material; and (2) aluminum chelate, which is used as the emitting layer (EML) and electron-transporting layer (ETL) material. Electron and hole transport in ETL and HTL is somehow similar to that in n- and p-type materials in semiconductor LEDs, with relatively high electron and hole mobilities, which recombine near the organic–organic interface. Carrier injection from the electrode to organic layer is improved by adjusting the work function of the metal electrodes. A high work function anode, indium tin oxide (ITO), and low work function cathode, Mg:Ag alloy, are hence chosen for better carrier injection. Also, since ITO is transparent to visible light, the emitted photons radiate through the anode and glass substrate, to the far field. To reduce the driving voltage, improve the quantum efficiency and extend the operation lifetime, a multilayer structure was proposed, such as a hole-injection layer (HIL), hole-blocking layer (HBL) and electron-injection layer (EIL). Emission wavelength from an OLED can be adjusted by the EML material and device structures. In 1990, Burroughes et al. demonstrated EL from a conjugated polymer, in a device called a polymeric light-emitting device (PLED).5 The operation principles of PLEDs and OLEDs are basically identical. The only difference comes from the molecular weight and fabrication technology. Conjugated polymers cannot be sublimed under vacuum due to their large molecule weight. In this chapter, we first describe the energy diagram, the photophysical processes and the electrical properties in organic materials. Then, device structures are introduced to obtain OLEDs with better electrical and optical characteristics.
7.2 Energy states in organic materials When two (or more) atoms come close and form a molecule, the resulting molecular orbitals and their energy states can be represented by the combination of distinct atomic orbitals. The energy diagram is shown in Figure 7.2. The lower and higher energies are called the bonding and antibonding orbitals, respectively. The bonding-state energy is less than that of the individual atoms so that the atoms tend to form a molecule. Once the molecule is formed, the electrons fill up the molecular orbitals in order from low- to high-energy states, as shown in Figure 7.3. According to Pauli’s exclusion principle, each state can accommodate two electrons with opposite spins (‘up’ and ‘down’). Hence, we can define the HOMO,
Organic light-emitting devices
179
Energy
atomic orbital (1)
anti-bonding molecular orbital
atomic orbital (2)
bonding molecular orbital
Figure 7.2 Two atomic energy states forming molecular states.6
Energy
LUMO
HOMO
Figure 7.3 HOMO and LUMO states.
which is the ground state under thermal equilibrium. It is clear that the attractive force from the nuclei is weakest for the HOMO electrons. Hence, when there is an excitation (electrical or optical), it is possible to promote those electrons to the next higher energy state, which is the LUMO. In general, the wave function and eigenenergy of a molecular orbital can be divided into a nuclear part and electronic part, according to the Born–Oppenheimer approximation. Since the nuclei are much heavier than the electrons, electrons can move within the framework constructed by the nuclei. Also, electrons move much faster than nuclei so that the electrons can respond instantaneously to any configuration change of the nuclei. This means that, for a given electronic state, a certain potential energy of the nuclei can be established with a distinct configuration. For a different electronic state, a different potential energy can be formed. For a more general case, i.e. three- or many-atom molecules, a potential energy ‘surface’ is needed.
7.3 Photophysical processes Photophysical processes in organic materials include optical absorption and energy relaxation, which follow the Franck–Condon principle. Light absorption means a molecule accepts energy from an electromagnetic field. Under thermal equilibrium, the energy may be relaxed radiatively through light emission (electronic state transition), or nonradiatively through heat dissipation (such as vibrational and rotational state transition). The radiative process is very important for OLEDs and can be divided into fluorescence and phosphorescence, originating from singlet and triplet exciton relaxation. Phosphorescence emission
180
Introduction to Flat Panel Displays
exhibits a spin inversion of the electron with a long carrier lifetime, and hence typically is not observable at room temperature. Unfortunately, upon electrical excitation, the number of triplet excitons is three times more than that of the singlet ones, according to quantum statistics. This means that 75 % of the excitons do not contribute to light emission in a conventional OLED. A Jablonski diagram is typically used to illustrate the whole picture of the photophysical processes, such as absorption, electronic state transition and energy relaxations, in a single molecule. When two (or more) molecules exist in a system, energy transfer may occur between the molecules. When the distance of the organic materials becomes smaller, new molecular orbitals may be created, called the excimer and exciplex, which are formed by the same or different molecules.
7.3.1 Franck–Condon principle Photon energy absorbed by a molecule may excite the electronic, vibrational and rotational modes. Light absorption and emission in organic materials is mainly between the first excited electronic state (LUMO) and the ground electronic state (HOMO) in the ultraviolet (UV) and visible range, i.e. several electron volts. Vibrational states, i.e. several tenths of an electron volt, lie within an electronic state and there are some selection rules for energy transitions which can be distinguished from the absorption and emission spectra. Typically, rotational states are too close, i.e. several hundredths of an electron volt, to be resolved at room temperature. Figure 7.4 shows potential curves of electronic states with vibrational energy levels and wave functions. At room temperature, since energy levels for vibrational and rotational modes are very close, molecules in the ground electronic state are excited to different rotational and vibrational levels by the heat during thermal equilibrium. Since the frequencies (∼1015 s−1 ) for electronic state excitation are much greater than nuclei vibrational motion (∼1013 s−1 ), the electronic transition must be a vertical transition in the potential curves, without any configuration change of the molecule. Also, it can be concluded that the wave function for the nuclei should be the same before and after an electronic transition. This means the vibrational wave function between two selected electronic transitions should be the same. More generally, the transition (including absorption and emission) probability of two vibrational modes in distinct electronic states is proportional to the wave function overlap of these two vibrational modes, which is called the Franck–Condon principle. For example, in Figure 7.4, when a molecule is in the ground electronic state and ground vibrational
LUMO
E v’= 2 v’= 1 v’= 0 HOMO
v=0 R Figure 7.4
Potential curves of electronic states with vibrational energy levels and wave functions.
Organic light-emitting devices
181
state (v = 0), it is difficult to excite it to the v’ = 0 level since the wave function between these two vibrational modes is small. The transition v = 0 → v’ = 1 is possible due to partial overlap of the wave functions. The transition v = 0 → v’ = 2 is preferred since the wave function overlap reaches a maximum in this case. When a molecule absorbs light an electron can be promoted to higher electronic and vibrational states; the electron can then lose some energy through vibration and rotation motions back to the ground vibrational state in the excited electronic state (v’= 0). Light emission can then take place, which must also follow the Frank–Condon principle. Hence, the absorption has a higher energy and a shorter wavelength than the emission, as shown in Figure 7.5(a). The difference between the absorption and emission peak is called the Stokes shift (Figure 7.5(b)). The shape of the absorption spectrum is a ‘mirror image’ of the emission one since the vibrational modes are often similar. Figure 7.5(c) shows the fluorescence (left) and absorption (right) spectra of oligophenylenevinylenes (nPVs) with different chain lengths in dioxane at room temperature. The structures of nPVs are also shown.
E LUMO
Stokes shift v’= n
A
Intensity
v’= 0 E HOMO v=m
Absorption
v=0
Emission
R
(a)
λ
(b)
(c) –
Figure 7.5 (a) Energy levels of the Stokes shift. (b) Spectra of absorption and emission. (c) Mirror image between absorption and emission spectra of nPVs.7 (Reprinted with permission from Gierschner, J., Mack, H.G., Luer, L. and Oelkrug, D. Fluorescence and absorption spectra of oligophenylenevinylenes: vibronic coupling, band shapes, and solvatochromism. J. Chem. Phys., 116, 8596. Copyright (2002), American Institute of Physics)
182
Introduction to Flat Panel Displays
7.3.2 Fluorescence and phosphorescence According to the Pauli exclusion principle, two electrons in the same orbital must have opposite spins. Those two electrons are called paired electrons and the state is called the singlet ground state, denoted by S0 in Figure 7.6. Once the molecule absorbs light, the two electrons can still have different spins in the different electronic states; this is called the singlet excited state (S1 ). Since two electrons in the excited states are in different orbitals, Pauli’s exclusion principle no longer applies, which means that these two electrons can have the same spin and are in the triplet state (T1 ). The energy of the singlet state is higher than that of the triplet state calculated from quantum mechanics, which means energy is preferably transferred from S1 to T1 , together with a spin inversion. This process is called ‘intersystem crossing’. Note that the triplet state must be the excited state or it conflicts with Pauli’s exclusion principle. So, T1 cannot relax back to S0 if there is no other spin inversion. In other words, according to the Frank–Condon principle, T1 → S0 is ‘forbidden’ since the wave function overlap is zero if considering the spin term and only singlet excitons can contribute to light emission. The terms spin-up and spin-down electrons actually refer to the spin angular momentum with positive or negative components in the z-direction. The spin angular momentum is nonzero in the x–y plane, as shown in Figure 7.7(a). When there are two electrons in separate electronic states, they can be destructive or constructive, corresponding to singlet or triplet states. As shown in Figure 7.7(b), there are three possible arrangements for triplet and only one for singlet state from quantum statistics. Under electrical excitation, molecules are promoted to excited states and excitons are created. According to the discussion above, the number of triplet states is three times that of singlet states when a molecule is excited. If emission comes from singlet excitons only, the maximum internal quantum efficiency (IQE), i.e. in terms of photons per injected electron–hole pair, is only 25 %.
Singlet ground state
Singlet excited state
Triplet excited state
S0
S1
T1
Figure 7.6 Singlet ground, singlet excited and triplet states.
Z
Z
Z
Spin-up electron
Spin-down electron (a)
Singlet 1/4
Triplet 3/4
(b)
Figure 7.7 (a) Spin angular moment and (b) coupling of spin angular momenta between two electrons in different orbitals.8
Organic light-emitting devices
183
However, the ‘allowed’ and ‘forbidden’ transitions are not so absolute in real situations since some perturbation terms are added into the total Hamiltonian which creates a mixing of spin multiplicity. Spin– orbital coupling, i.e. interaction between the spin and orbital angular momentum of an electron, is one of the most important terms which results in the mixing of singlet and triplet states. After considering the spin–orbital interaction, there are no pure singlet and triplets states. However, remember that spin–orbital coupling is only a perturbation term. This means the transition from T1 → S0 is nonzero but still a small contribution, compared with S1 → S0 . Radiative recombinations from singlet and triplet excited states to the singlet ground state are called fluorescence and phosphorescence. Phosphorescence means the carrier lifetime is longer, because spin inversion is involved, and typically exhibits low efficiency. The strength of the spin–orbital coupling is proportional to atomic number. For an atom with a large atomic number in the molecule, spin–orbital coupling can be enhanced and the phosphorescence emission efficiency can be improved. This is called the heavy atom effect.
7.3.3 Jablonski diagram Figure 7.8 shows a Jablonski diagram which is used to describe the energy transitions among electronic states. Figure 7.8 only shows the singlet ground state (S0 ), two singlet excited states (S1 and S2 ) and two triplet excited states (T1 and T2 ). More excited states can also be included in this diagram. As discussed above, the electron undergoes a vertical transition without changing the configuration of the nuclei due to its fast speed, and also its low weight. Absorption takes place from the singlet ground state (S0 ) to the singlet excited states (S1 and S2 ). Photon energy determines the final vibrational state in the excited state and the Franck–Condon principle governs the transition probability. Then, electrons in the singlet excited state may relax their energy by release of heat through vibration and rotation. It is also possible to jump from the vibrational mode of S2 to S1 . Typically, energy relaxation (especially by thermal dissipation) without involving spin inversion is called internal conversion. After electrons relax their energy to the ground vibrational and rotational state of the singlet excited state S1 , fluorescence emission takes place from S1 to S0 through photon emission, which has to obey the Franck–Condon principle as well. Obviously, the emission wavelength is longer than the absorption one, i.e. the Stokes shift. It is also possible to relax energy from S1 and S2 to S0 by vibrational and rotational states which contributes to nonradiative recombination. Electrons in the singlet excited state can transfer to triplet
E
S2 ISC Abs.
T2
IC IC
S1
T1
ISC Abs.
F + IC
Abs.
Ph + ISC
S0
Figure 7.8 Jablonski diagram. (Abs., absorption; IC, internal conversion; ISC, intersystem crossing; F, fluorescence; Ph, phosphorescence.)
184
Introduction to Flat Panel Displays
states (T1 and T2 ) via intersystem crossing. Internal conversion may happen from T2 to T1 . Absorption is possible between T1 and T2 since the two states exhibit the same spin angular momentum, which is called triplet–triplet absorption. Relaxation from T1 and T2 may be radiative or nonradiative. Photon emission from T1 to S0 is called phosphorescence, which exhibits a longer wavelength than fluorescence emission. Coupling from higher vibrational modes of T1 (or T2 ) to S1 (or S2 ) is also possible via intersystem crossing which results in delayed fluorescence emission from S1 (or S2 ) to S0 .
7.3.4 Intermolecular processes Not only does energy relaxation occur via intramolecular processes, as described by the Jablonski diagram, but energy transfer and relaxation also occur between molecules which are called intermolecular processes. One of the most important processes is the energy transfer from donor to acceptor materials which is most useful in OLEDs for adjusting the emission wavelength, improving the efficiency and extending the operation lifetime. When two molecules come close, new orbitals form and the emission wavelength shows a red-shift, which is called excimer and exciplex, consisting of the same or different molecules. Quenching is a kind of intermolecular energy relaxation via nonradiative recombination, which may be caused by high concentrations, metal materials or impurities.
7.3.4.1 Energy transfer process A molecule in an excited electronic state, which is called a ‘donor’, may transfer its energy to another molecule, which is called an ‘acceptor’. After this process, the donor molecule returns to its ground electronic state and the acceptor molecule is promoted to a higher state,9 which can be represented as D∗ + A → D + A∗ ,
(7.1)
where D and A are the donor and acceptor. The asterisk denotes the excited state. Energy transfer can be a two-step process without direct interaction between the donor and acceptor molecules as follows: D∗ → D + hv,
(7.2)
hv + A → A∗ .
(7.3)
Here, photons from the radiative recombination of donor are absorbed by the acceptor and promote the acceptor to an excited state. This is called radiative energy transfer since photons are involved in this process. The strength of this energy transfer only depends on the emission efficiency of the donor and the absorption efficiency of the acceptor at this wavelength. A single process without the intermediation of photons, described by Equation (7.1) , is also possible when donor and acceptor molecules are close (less than 10 nm), which is called nonradiative energy transfer, through the energetic resonance of two molecules. As shown in Figure 7.9, this process is isoenergetic and the transfer probability is proportional to the spectral overlap (J) between donor emission (I D (v)) and acceptor absorption ( A (v)) spectra, as follows: ∞ (7.4) J = ID (v)εA (v) dv. 0
Two possible mechanisms account for nonradiative energy transfer, which are dipole–dipole interaction and electron exchange, called Förster and Dexter energy transfer. The rate constant (k ET(Coulomb) ) of the Förster energy transfer, which originates from dipole resonance from donor to acceptor molecules, can be represented as fD fD kET(Coulomb) ≈ 6 2 J, (7.5) RDA v˜
Organic light-emitting devices
185
D*
A*
D
A acceptor absorption spectrum
donor emission spectrum Int.
ν J
Figure 7.9 Donor emission and acceptor absorption spectra and their spectral integral.
where f D and f A are the transition probabilities for donor emission and acceptor absorption, following the Franck–Condon principle, and RDA is the distance between donor and acceptor molecules. We can also see that the interaction strength decreases fast with increasing donor–acceptor separation. Typically, Förster energy transfer is efficient within 10 nm. Considering the spin angular momentum, the following energy transfer is allowed: 1
D∗ +1 A →1 D +1 A∗ .
(7.6)
However, due to the characteristics of dipole–dipole interaction, spin inversion is impossible according to the selection rule. So the following energy transfer is forbidden: 3
D∗ +1 A →1 D +3 A∗ .
(7.7)
The rate constant (k ET(exchange) ) of the Dexter energy transfer can be written as
−2RDA kET(exchange) ≈ exp L
J,
(7.8)
which decreases exponentially with increasing donor–acceptor distance. Since it involves two electrons jumping between donor and acceptor molecules, Dexter energy transfer is a short-range process which is only effective within 1 nm. This characteristic also makes the two energy transfers allowed since the total spins are conserved:10 D∗ +1 A →1 D +1 A∗ ,
(7.9)
D∗ +1 A →1 D +3 A∗ .
(7.10)
1
3
7.3.4.2 Excimer and exciplex formation When two identical molecules (denoted as ‘M’ in Figure 7.10(a)) – one in its ground electronic state and the other in its excited state – come close, new electronic states with lower energy may form which can be viewed as a complex, as shown in Figure 7.10(a). It is called an ‘excimer’ (excited dimer).
186
Introduction to Flat Panel Displays
M*
(MM)* excimer
M
Energy
Energy
excimer emission
Energy
exciplex emission
M*
+
(M–Q+)*
Q
(a)
(b) Figure 7.10 (a) Excimer and (b) exciplex formation.
10000 1000
FPt1 Flrpic
100
Film 1
10
CBP Film 2 FPt1 Film 3 FPt1 excimer Film 4 FPt1+Flrpic
200
300
400 500 Wavelength (cm)
600
Extinction Coefficient (M–1cm–1)
Photoluminescence Intensity (a.u.)
Two different molecules can also form a complex (exciplex) by charge transfer process, as shown in Figure 7.10(b), where ‘M’ and ‘Q’ denote two different molecules.11 Excimer and exciplex may relax their energy radiatively, back to the ground state. Note that excimers and exciplexs are bonded complexes formed only in the excited state and dissociated in the ground state. It is straightforward to see that the emission wavelength from this excited state of the complex is greater than the original molecules due to the smaller bandgap. Figure 7.11 shows an example of excimer fluorescence.12 Film 1 is a neat 4,4 -N,N -dicarbazole-biphenyl (CBP) film. The absorption and photoluminescence (PL) emission peaks are at 350 and 390 nm, respectively. Since the hosts are all CBP for Films 1 to 4, the absorption spectra are nearly identical for these four films. In Film 2, with incorporation of <1 wt% of platinum(II) (2-(4 ,6 -difluorophenyl)pyridinato-N, C2 )(2,4-pentanedionato) (FPt1) into CBP, the PL spectral peak shifts to 470 and 500 nm due to dopant emission. There is a small hump at 380 nm due to insufficient energy transfer. In Film 3, with further increase in FPt1 to 7 wt%, one can see a broad-band PL emission in the long-wavelength region (i.e. 570 nm) which arises from excimer emission. On mixing two dopant materials into CBP, i.e. 6 wt% iridium bis(4,6-difluorophenyl-pyridinato-N, C2)-picolinate (FIrpic) and 6 wt% FPt1, a white PL emission can be achieved from Film 4.
700
Figure 7.11 Absorption and fluorescence spectra of four films. Film 1, neat CBP; Film 2, CBP + <1 wt% FPt1; Film 3, CBP + 7 wt% FPt1; Film 4, CBP + 6 wt% FPt1 + 6 wt% FIrpic.12
Organic light-emitting devices
187
7.3.4.3 Quenching process Fluorescence quenching as a photophysical process can be described by Q
M∗ −→ M,
(7.11)
where Q is called the quencher. The quenching process is an intermolecular process in which the excited molecules relax energy through the quencher, which can be the same as or different from M, called selfquenching and concentration quenching. Excimer and exciplex relaxation may be nonradiative which is a kind of quenching process. Also, some unwanted chemicals in an organic material provide a similar energy relaxation route which is called impurity quenching. In OLEDs, generated excitons may transfer energy to the metal electrode, which is called the electrode quenching process.
7.3.5 Quantum yield calculation From a macroscopic viewpoint, light decreases after passing through an absorptive medium. We can define the absorption coefficient (α) as dI = −αIdx,
(7.12)
where I and x represent the light intensity and the absorption thickness, respectively. In this equation, the intensity decrease is proportional to: (1) the absorption coefficient, (2) the incident intensity and (3) the absorption thickness. After integrating both sides over the whole sample thickness d, the resulting equation is I = I0 e−αd , where I 0 is the initial light intensity. One can write I0 1 A = log10 = αd = εcd, I 2.303
(7.13)
(7.14)
where A is the absorbance (or optical density) of the sample, c is the concentration (in mol l−1 ) and ε is the molar absorption coefficient (in l mol−1 cm−1 ). This is called the Beer–Lambert law.
Example 7.1 Time-of-flight (TOF) is a technique which is commonly used to determine the carrier mobility in organic materials. The basic concept of TOF is to calculate the carrier drift time through a material at a given electric field across it. Carriers are generated by optical pumping. Therefore, the region of carrier generation requires a much narrower width as compared with that of carrier transport. Assume an organic material has an absorption coefficient α = 5 × 105 cm−1 at an excitation wavelength of 355 nm. Find the penetration depth of light, which can be considered the width of the carrier generation region in TOF measurement. Answer. The penetration depth, L, of light impinging on a material is defined as the distance at which the incident intensity drops by a factor of e−1 . According to this definition, L = 1/α = 1/(5 × 105 cm−1 ) = 20 nm. Therefore, the thickness of the materials used in TOF measurements must usually be of the order of several micrometers to make sure that the carrier drift length approximates to the thickness of the deposited material.
188
Introduction to Flat Panel Displays
In a system with many particles, when carriers are promoted (optically or electrically) to the excited electronic state, the rate equation is used to describe the dynamics of relaxation back to the ground state or to other excited electronic states, as follows: ∂n n = G(t) − , ∂t τ0
(7.15)
where G(t) is the carrier generation rate which is used to describe the electrical or optical pumping, n is the carrier density in the excited state and τ0 is the carrier lifetime of the excited state. This can also be written as τ0 = 1/k 0 , where k 0 is the rate constant. So the decay rate of carriers in the excited state is proportional to the carrier density and the rate constant. Hence, the carrier density in the excited state (n) decays with time after being promoted at t = 0: n(t) = n0 exp(k0 t),
(7.16)
which can be used for both radiative or nonradiative relaxation. The quantum yield of a certain relaxation process is defined as the ratio of carrier numbers involved in the process to the total carrier numbers in the excited state. From the Jablonski diagram, carriers relax through different pathways, such as fluorescence to the ground state, internal conversion and intersystem crossing, with rate constants k F , k ic and k isc , respectively. The sum of the quantum yields for all processes is equal to 1. The rate equation for the first singlet excited state can be written as ∂S1 = G(t) − (kF + kic + kisc )S1 . ∂t
(7.17)
And the quantum yield of the fluorescence (φF ) is given by φF =
kF . kF + kic + kisc
(7.18)
Example 7.2 Triplet–triplet exciton annihilation is a well-known phenomenon in phosphorescent OLEDs. The KTT
process is T1 + T1 −→ Tn + S0 or Sn + S0 . Here, T (S) donates triplet (singlet) exciton concentration, k TT is the rate constant and subscripts 0 and n denote ground and nth excited state. This annihilation process is governed by ∂ 2T T ∂T = D 2 − − kTT T 2 , ∂t ∂x τ where D is the diffusion coefficient and τ is the triplet exciton lifetime. Assume k TT = 1.8 × 10−14 cm3 s−1 ; τ = 10 ms; and T = 5 × 1016 cm−3 . (1) Find the exciton lifetime for annihilation process. (2) Find the exciton lifetime for the entire system. Answer. (1) Exciton lifetime is τTT = 1/kTT T . Obviously, the higher the triplet exciton concentration, the faster is the annihilation process. This leads to efficiency roll-off under high injection current, since the number of radiative excitons is reduced: τTT =
1 = 1.1 ms. 1.8 × 10−14 cm3 s−1 × 5 × 1016 cm−3
(2) Exciton lifetime in the entire system is τtotal =
1 1 = = 0.99 ms. 1/τ + 1/τTT (1/10 ms) + (1/1.1 ms)
Organic light-emitting devices
189
7.4 Carrier injection, transport and recombination Electroluminescence (EL) of organic solids comes from the recombination of injected carriers, which is similar to the emission mechanisms of semiconductor LEDs, as described in Chapter 6. Typically, an OLED is fabricated on an indium tin oxide (ITO)-coated substrate. Thin (100–200 nm) amorphous organic layers are deposited on the ITO anode having different functions such as HIL, HTL, EML, ETL and EIL. Finally, a cathode metal is then evaporated. In an OLED, charges injected from electrodes transport through the organic layers, where they then recombine and form excitons which emit light after a transition from an excited state to the ground or an intermediate state.13 Fowler–Nordheim (FN) tunneling and Richardson–Shottky14 (RS) thermionic emission can be used to treat carrier injection in semiconductor devices. However, in an OLED, these can only partially describe the behavior between organic materials and organic–metal interface because of: (1) the chemical reactions that may take place between the interfaces which result in the formation of interfacial layers and (2) the disordered nature of the organic materials. Since OLED (or PLED) thin films are amorphous without a crystalline structure, the charge transport cannot be well described by the band theory developed for semiconductors or molecular crystals. In contrast, in organic thin films, the charge carriers ‘hop’ among different localized states, which explains the low carrier mobility of organic thin films (typically <10−3 cm2 V−1 s−1 ).15 Such localization of the excited states also results in the large binding energy of excitons. For an OLED display, emission wavelengths are within the visible range. One can understand from the Stokes shift, as described in Section 7.3.1, that the absorption should be in the UV region with bandgap of ∼2–4 eV, which results in the low carrier concentrations in organic materials (typically <1010 cm−3 ) since it is difficult to generate ‘free’ carriers thermally. From the simple equation σ = nqμ,
(7.19)
where σ is the conductivity (in ( cm)−1 ), n is the carrier concentration, q is electronic charge (in C) and μ is the carrier mobility, one can obtain the conductivity of an organic thin film as being of the order of 10−12 ( cm)−1 . When the conductivity is between 10−8 and 102 ( cm)−1 , the material is called a semiconductor. Higher and lower conductivities correspond to conductors and insulators. For example, the conductivity of glass is between 10−11 and 10−10 ( cm)−1 , higher even than many organic thin films. From this viewpoint, organic materials for OLED applications can be regarded as more like insulators, rather than semiconductors. For practical use, the organic layer thicknesses must be very small (hundreds of nanometers) provided the driving voltage is not too high (<10 V). The carriers in organic thin films are called ‘space charges’ since free carriers are negligible, and all the carriers are injected from the electrodes. The molecules are positively or negatively charged, when the carriers are injected, which are called cation or anion, respectively. The carriers are ‘hop’16 among different molecules under the influence of an electric field. So, it is easy to understand that the mobility in an organic thin film increases with increasing electric field, which is called the Poole–Frenkel (PF)17 model. To describe carrier transport in organic thin films, time-dependent continuity equations, drift– diffusion current equation and Poisson’s equation are still valid, as described in Chapter 6. However, some modifications are needed for organic thin films. Under certain approximations, the current density– voltage relation can be described by the SCLC and TCLC. Figure 7.12 shows a typical current density (J)–voltage (V ) characteristic of an OLED using a double-logarithm scale. Ohmic current, Johmic ∝ V , dominates in region I which arises from drift of the limited free carriers inside the organic materials. With increasing the driving voltage, carriers are injected from the electrodes and transport inside the organic materials with traps which is TCLC, JTCLC ∝ V l+1 . This is the case of region II where the plot has a slope of l + 1. When the traps are filled up, it becomes SCLC, JSCLC ∝ V 2 . Region III describes this phenomenon, where the slope of the J–V curve is exactly 2 for the double-logarithm scale. The recombination rate in an OLED depends on the spatial overlap of the hole and electron densities. Since the mobility values of the organic materials are so low, as compared to the recombination time, electrons and holes recombine and form excitons as soon as they meet in the space domain, which follows
190
Introduction to Flat Panel Displays
10
I
II
III
Current Density (a.u.)
1 0.1 0.01 1E-3 1E-4 1E-5 1E-6 1E-7 1E-8 0.1
1 10 Applied Voltage (V)
100
Figure 7.12 Typical J–V characteristics of an OLED.
so-called Langevin recombination, as for the LED case described in Chapter 6. Once the excitons relax their energy through radiative recombination and generate light, the electromagnetic wave propagates in the thin-film structure with thickness comparable to the emitting wavelength. The strong interference has great influence on the optical characteristics such as output intensity, spectrum and device efficiency.
7.4.1 Richardson–Schottky thermionic emission In a one-dimensional model, one can assume an organic material with a thickness L with anode and cathode at x = 0 and x = L, respectively. The hole current injected from the anode (x = 0) to the organic layer can be represented by Jp (0) = Jth − Jir + Jptu ,
(7.20)
where J th is the thermionic emission current, J ir is the backflowing interface recombination current and J ptu is the tunneling current. The thermionic emission current has the well-known form18 −Eb ∗ 2 , (7.21) Jth = A T exp ηkT where A∗ is Richardson’s constant, E b is the interfacial energy barrier and η is the ideality factor, the value of which is typically between 1 and 2 for semiconductors.1 Because of the image force, E b depends on the electric field at the interface: q |F(0)| Eb = φB − , (7.22) 4πε where φ B is the Schottky energy barrier at zero field and |F(0)| is the electric field at the interface. In Figure 7.13(a), the image charge and the electric field line at the metal–organic interface is shown, and the barrier energy level change due to the image force is also shown in Figure 7.13(b). The interface recombination current is proportional to the hole density at the interface:18 Jir = qvp(0)
(7.23)
Organic light-emitting devices
191
+
–
x X=0 (a) E(x)
xm x
Δφ eF eφ Bo
(b) Figure 7.13 (a) Image charge and the electric field line at the interface.19 (b) Energy level change due to image force.
The kinetic coefficient v is determined by detailed balance between thermionic and interface recombination:18 A∗ T 2 v= . (7.24) qn0 Parameter n0 is the densities of states; we take the densities of states as 1.0 × 1021 cm−3 for organic materials.18, 20 The FN injection for the tunneling contact is21 q3 8π 2m∗ Eb 3 2 |F| exp − , (7.25) Jptu = 8πhEb 3qh |F| where m∗ is the effective mass of the carrier and h is Planck’s constant. Equations similar to (7.24) and (7.25) can be written for the electron current density. Typically, in an OLED, thermionic emission and FN tunneling (field emission)22 govern the injection mechanism at high and low field conditions.
192
Introduction to Flat Panel Displays
7.4.2 SCLC, TCLC and PF mobility The transport of electrons and holes in an OLED is described by time-dependent continuity equations, with a drift–diffusion current, coupled to Poisson’s equation.18 Considering electron transport in a unipolar organic material without generation and recombination, the continuity equations can be written as ∂n 1 ∂Jn = , ∂t q ∂x
(7.26)
where J n is the electron current density and n is the electron density, as described in Chapter 6. Under steady state, ∂/∂t = 0, Jn = const. (7.27) As described above, carrier density in organic materials is negligible under thermal equilibrium, which means the diffusion current can be neglected. So, only the drift term appears in the current density equation: Jn = nqμE
(7.28)
Jn , qμE
(7.29)
or n=
where E is the electric field, μn is the electron mobility and q is the electronic charge. Coupled with the Poisson equation: ∂E q =− n ∂x ε
(7.30)
∂E q Jn =− , ∂x ε qμE
(7.31)
or
where ε is the static dielectric constant. Assuming μ is a constant, one can obtain the relation between current and electric field, and applied voltage since V = E dx, which is the well-known Mott–Gurney equation to describe the SCLC: 9 V2 JSCLC = εε0 μ 3 , (7.32) 8 d where d is the layer thickness. The physical meaning of SCLC is that it represents the maximum possible unipolar current that a perfect insulator without trap, i.e. organic materials, can sustain at a given potential difference. In the presence of traps, the current will increase faster than quadratic until all traps are filled. In this regard, TCLC deals with the carrier transport in the bulk, which is governed by23 JTCLC ∝
V l+1 , d 2l+1
where l represents the effects of traps. For example, an increase of (l + 1) directly indicates increasing of the trap depth and density. In general, the TCLC model is well used in the low electric field condition, whereas the SCLC is simply sufficient for high electric field. This observation implies that at lower fields, charge carrier traps are involved, which are filled up at higher field. Because of the nature of localization of -electrons in organic molecules, carrier transport in a solid-state organic layer can be seen as ‘hopping’ between molecular sites. The ‘hopping’ process is essentially a one-electron oxidation–reduction process between neutral molecules and their charged derivatives. The charge hopping rate is known to be not only influenced by the effects of disorder, but also strongly dependent on applied electric field. This
Organic light-emitting devices
193
phenomenon directly leads to field-dependent carrier mobility, which is governed by a well-known rule called the PF model.24 The PF mobility, which is very frequently observed in amorphous molecular materials, is √ μ(F) = μ0 exp(β F). (7.33) where μ0 is the zero-field mobility and β is a material related parameter.
7.4.3 Charge recombination Recombination radiation from organic solids is defined as light emission following the fusion of oppositely charged carriers into an electrically neutral state. Following recombination of opposite carriers, excitons are created, which undergo one-dimensional diffusion towards the electrodes. As described above, carrier mobility values in organic materials are much slower than the recombination rate, which means electrons and holes recombine simultaneously when they meet in the space domain. Hence, the recombination process in an OLED is typically described as a bimolecular process, following the Langevin theory, which is governed by R = γ np,
(7.34)
where R is the carrier recombination rate, n (p) is the electron (hole) density and γ is the recombination coefficient of the organic materials. The exciton migration is not affected by applied voltage since it is neutral. The exciton diffusion is forbidden in the material with larger exciton energy. The exciton diffusion length (L) with a typical value of several nanometers, which roughly defines the width of the recombination zone, is related to the exciton diffusion coefficient (D) and its lifetime (τ )25 by L=
√ Dτ .
(7.35)
As discussed in Section 7.3.2, only singlet excitons are generated using optical excitation because of the spin conservation. For electrical excitation, the injection of electrons and holes is followed by the formation of free polarons, which is the deformation of the molecule due to injected carriers. When they experience opposite coulombic attraction, a neutral bound polaron pair, or charge transfer (CT) states are created.26 With electric excitation, the rate of singlet CT (1 CT) and triplet CT (3 CT) state generation obeys simple statistics, thus leading to a ratio of 1:3. The transfer between both types of exciton is governed by intersystem crossing (ISC). The CT states will sequentially transfer to their respective lowest excited Frenkel excitons. The relaxation of singlet excitons to the ground state via radiative decay is called fluorescence, whereas that of triplet excitons is called phosphorescence. In fluorescent organic materials, only singlet excitons, electrically generated excitons of 25 %, are exploited for light emission and the remaining 75 % are relaxed though nonradiative paths.27 To effectively release all available excitons via radiation relaxation, phosphorescent materials are employed to develop highly efficient OLEDs.28 31
7.4.4 Electromagnetic wave radiation As in the case of LEDs in Chapter 6, from classical viewpoints, the emission from organic thin films can be considered as a point source and it has a radiation cone defined by Snell’s law which only takes account of the refractive index of the organic thin films and air.32 Considering the unit sphere shown in Figure 7.14, the shaded region enclosed by the critical angle at the interface of organic material and glass indicates the exit through which the generated photons can escape.
194
Introduction to Flat Panel Displays
θ
1
Figure 7.14 Escape cone from an OLED.
For an organic layer with refractive index of nEML , the fraction of generated light escaping from the substrate (which is defined as the external quantum efficiency) can be calculated from33 θ θEML-air 2 EML-air 1 × dθ × 2π × 1 × sin θ ηc = 0 = sin θ dθ = 1 − cos θEML-air 4π × 12 0
2 1 1 nEML −1 =1− =1− 1− 2 ∼ (7.36) = 2 , nEML nEML 2nEML where ηc is the extraction efficiency and cos θ EML-air is the organic–air critical angle. For example, nEML is typically 1.6, which means the extraction efficiency is only about 20 % and 80 % of photons are trapped inside the OLED. However, a multilayer OLED cannot be described by the ray optics model since the layers have thicknesses less than the wavelength of the emitted light. The ray optics description cannot completely explain the optical characteristics in an OLED. An example is the dependence of far-field emission pattern on the thickness of the organic layer, which cannot be explained by classical theory. By using a transfer matrix method, the optical interference effect in thin-film structures can be well described. The photon emission from organic materials originates from the dipole oscillations. Typically, an OLED consists of a transparent anode with ITO, organic layers and a reflective cathode on the glass substrate, as shown in Figure 7.15. Photons generated in the organic layers propagate out of the OLED through the ITO anode and the glass substrate. Such a structure can be regarded as a Fabry–Perot cavity and the emission spectrum in the normal direction can be described by the equation34 √ up 2 2 R1 cos(4π x/λ)] Eout (λ) = |Ein (λ)|2 × T2 [1 + R1 + √ , (7.37) 1 + R1 R2 − 2 R1 R2 cos(4π L/λ)
Cathode EIL ETL EML HTL HTL light emission
ITO (anode) Glass substrate
Figure 7.15 Typical OLED structure.
Organic light-emitting devices
195
θ2
Metal cathode
θ2
Metal cathode
Organic materials and ITO
θ2
Glass
θ1
Organic materials and ITO
Glass
θ1 (a)
Figure 7.16
θ1 (b)
Schematic of (a) wide-angle and (b) multiple-beam interference.
where the term on the left-hand side is the output intensity, the first term on the right-hand side is the free-space EL intensity, x is the effective distance of a dipole from the reflective cathode and R1 is the reflectivity of the cathode. Also, R2 and T 2 are the effective reflectivity and transmittivity of the ITO anode side, respectively. L is the total optical thickness of the cavity, given by ϕan λ ϕca λ + L = n L + (7.38) j j 4π . 4π j Here, nj and L j are the refractive index and thickness of the jth layer between the two electrodes, ϕ ca is the phase shift corresponding to the effective reflectivity R1 and ϕ an is the phase shift corresponding to the effective reflectivity R2 at the interface between the organic materials and the anode. Equation (7.37) can be rewritten as up 2 Eout (λ) = |Ein (λ)|2 × Tr × [1 + R1 + 2 R1 cos(4π x/λ)] (7.39) with Tr =
T2 . √ 1 + R1 R2 − 2 R1 R2 cos(4π L/λ)
(7.40)
√ The term [1 + R1 + 2 R1 cos(4πx/λ)] represents the antinode enhancement factor of the wide-angle interference, as shown in Figure 7.16(a).35 It has a maximum value when the emitting dipoles are located exactly at the antinode of the standing wave within the microcavity. The term Tr accounts for the effect of multibeam reflective interference, as shown in Figure 7.16(b).36 We can see that the effect of the wide-angle interference is determined only by the dipole position and the cathode reflectivity, while that of the multibeam reflective interference is controlled by the cavity optical length and anode and cathode reflectivities.
7.5 Structure, fabrication and characterization Considerations for device design in a small-molecule or polymer OLED include the electrical and optical optimization. For reducing the driving voltage, carrier injection should be efficient, which means careful energy level alignment is needed. Also, the mobility value should be as high as possible, and the free carrier density should be improved. For improving the recombination rate, suitable energy barriers are needed in the device to confine the carriers, which resemble heterojunction structures in a semiconductor LED. Sometimes, it is difficult to find an organic material with good electrical properties and high efficiency. Organic mixtures, or guest–host (dopant–matrix) systems, are introduced to improve the IQE. Since the planar geometry of an OLED of hundreds of nanometers in thickness is a microcavity
196
Introduction to Flat Panel Displays
structure, optical interference has to be taken into consideration for maximizing the external quantum efficiency. In a small-molecule OLED, precise control of the thin-film deposition enables multilayer structures to be possible. However, in a polymer OLED, a multilayer structure is difficult to accomplish by solution process. The thin-film process is the only difference in fabrication for the small-molecule and polymer OLED. The patterned process before and the encapsulation process after thin-film deposition are nearly identical for the small-molecule and polymer OLED. To drive a display, passive matrix (PM) and active matrix (AM) techniques are typically used, as for liquid crystal displays (LCDs). However, there are still some differences since the OLED is a current-driven device, rather than a capacitor as in the case of a LCD. For example, at least two transistors and one capacitor are needed in a pixel for AM driving. Lifetime is one of the main issues for OLED applications, which can be classified as extrinsic and intrinsic degradation. With a proper encapsulation (or passivation) process and careful environmental control during fabrication, extrinsic degradation can be greatly reduced. However, intrinsic degradation from organic material degradation dominates the ultimate operation lifetime of an OLED.
7.5.1 Device structure The basic structure of an OLED is a layer structure in which the organic layer(s) is (are) sandwiched by two electrodes. Typically, ITO with a high transparency and work function is used as the anode. Then, one or more organic thin films are formed (by evaporation, spin-coating, ink-jet printing, etc., which is discussed in Section 7.5.3.1) followed by a metal cathode. The simplest structure is to use a single organic layer between the electrodes, as shown in Figure 7.17.37 In such a structure, the work functions of the anode and cathode should be matched to the HOMO and LUMO of the organic material, for better carrier injection. The organic material should exhibit ambipolar transport characteristics and emission capability. Typically, it is difficult to find a bipolar organic material with high emission efficiency. Even if an organic material does exhibit such good electrical and optical characteristics, the recombination zone in such a device is not easy to control for highly efficient emission. For example, if recombination takes place near the electrode, which means the dipole oscillation is at the node, rather than the antinode position (Section 7.4.4), then this leads to destructive interference and decreases the device efficiency, called electrode quench. To improve this, two- or multilayer structures are introduced with their functions of carrier injection, transport, and emission.
level Vacuum
Work function of anode Anode
LUMO
Work function of cathode Cathode
h HOMO
Organic Layer Figure 7.17 Device structure of a single-layer OLED.
Organic light-emitting devices
197
7.5.1.1 Two-layer OLED Figure 7.18 shows the device structure of a two-layer OLED, first introduced by Tang et al.4 The two organic layers are aromatic diamine and aluminum chelate (tris(8-hydroxyquinoline) aluminum, Alq3 ) which are the HTL and ETL, respectively. Different from the bipolar characteristics in the single-layer device, HTL and ETL only transport unipolar carriers, holes and electrons. This provides some flexibility in selection of organic materials. Also, this structure helps to improve the device efficiency due to better carrier confinement. In this two-layer structure, carrier recombination occurs near the HTL– ETL interface. Since the HTL and ETL are different organic materials, such a structure is similar to the heterojunction in semiconductor LEDs. Under forward bias, electrons from the cathode will drift from the ETL and be blocked at the HTL–ETL interface since the LUMO value of the HTL is typically lower than that of the ETL. The HOMO value of the ETL is slightly larger than that of the HTL so that holes can readily enter into the ETL. The low hole mobility in the ETL results in high hole concentration inside the ETL and near the HTL–ETL interface. This enhances the collision capture process and recombination occurs in the ETL. So this is the EML at the same time. The interface between these two materials confines the carriers near the interface when applying an electrical field and increases recombination probability. The recombination zone is in the ETL within 10 nm of this interface.38 Typically, the electron mobility in the ETL is at least one order of magnitude lower than the hole mobility in the HTL (μe,ETL μh,HTL ), which means there are many holes waiting in the HTL–EML interface for recombination. The recombination probability (PR ) is given by −1 τrec , (7.41) PR = 1 + τt where τrec is the carrier recombination time and τt is carrier transit time of the electrons from the cathode to the HTL–EML interface. Equation (7.41) can be rewritten as −1 w PR = 1 + , (7.42) de
N O
– +
Al
N
O
N
Alq3
Mg Ag Alq3 Diamine ITO Glass
O
H3C
CH3
CH3
N
N
Diamine
CH3
S
Figure 7.18 Layer structure and organic materials of the first two-layer OLED.4 (Reprinted with permission from Tang, C.W. and Vanslyke, S.A. Organic electroluminescent diodes. Appl. Phys. Lett., 51, 913. Copyright (1987), American Institute of Physics)
198
Introduction to Flat Panel Displays
where w and d e are width of recombination and the ETL thickness, respectively, due to the barrier confinement.39 From this equation, we note that a narrow recombination width is preferred for preventing leakage of holes and electrons toward electrodes, which leads to low efficiency. Also, we also note that transparent ITO was used as the anode and reflective Mg:Ag alloy was used as the cathode. Work functions of anode and cathode should be high and low, for better hole and electron injection. The transparency of the ITO should be high since the photons radiate out through the anode and glass. Figure 7.19(a) shows the curves of current density and EL intensity versus voltage. The driving voltage is about 8 V with a current density of 100 mA cm−2 . Compared with a conventional semiconductor LED, the current density reaches 200 mA cm−2 only at 4 V.4 In this OLED, the HTL and ETL thicknesses are 75 and 60 nm, respectively, which are much thinner than the epitaxy layer (several micrometers) of a semiconductor LED. Although the device thickness is one order of magnitude greater than that of semiconductor LEDs, the driving voltage is more than two times higher due to the high resistivity of the organic materials. Figure 7.19(b) shows the EL spectrum of this device with a full width at half-maximum of about 100 nm which is two times broader than that of semiconductor devices due to more vibrational and rotational modes in the organic materials.
7.5.1.2 Dopant in the matrix as the EML
101
101
100
00
10–1
10–1
10–2
10–3
10–3
10–3
0
5 Bias vollage (V)
10–4 10
EL intensity (arb units)
02
EL intensity (mW cm–2)
Current density (mA cm–2)
In the two-layer structure, recombination occurs at the ETL, which means the ETL is also the EML. So the functions of this layer include carrier transport and emission. However, sometimes it is difficult, although possible, to find an organic material (high emission efficiency and high carrier mobility) for the whole visible range. A common way to solve this problem is to introduce an additional material into the EML as a guest-host system, which is a high-efficiency organic material doped into a transporting material as the EML.9 The criteria of the transporting material in the EML, which is usually called the matrix (or the host), include that it: (1) exhibits good carrier transport characteristics and (2) has suitable energy levels for carrier trapping and/or good overlap between host PL and dopant absorption spectra for efficient energy transfer, as described in Section 7.3.4. For a dopant (or the guest) material, high efficiency is a basic requirement. Not only does this improve the electrical and optical characteristics of an OLED, but the guest–host system also effectively prolongs the stability of a device by transferring recombined excitons to the emissive and stable dopant site thus minimizing nonradiative decay.40 Figure 7.20(a) shows the device structure proposed by Shi and Tang and the component molecular structures.40 Copper phthalocyanine (CuPC), N,N -diphenyl-N,N -bis(1-naphthyl)-1,1 -biphenyl-4,4 -diamine
400
400
400 400 Wavelength (nm)
400
Figure 7.19 (a) J–L–V curves and (b) EL spectrum of the two-layer OLED.4 (Reprinted with permission from Tang, C.W. and Vanslyke, S.A. Organic electroluminescent diodes. Appl. Phys. Lett., 51, 913. Copyright (1987), American Institute of Physics)
Organic light-emitting devices
199
MgAg Alq3 Alq3: DMQA NPB CuPc ITO (a) CH3
O
N N
N
N
O
CH3
DMQA
NPB (b)
Figure 7.20 (a) Device structure and (b) molecular structures of a guest–host system.40 (Reprinted with permission from Shi, J. and Tang, C.W. Doped organic electroluminescent devices with improved stability. Appl. Phys. Lett., 70, 1665. Copyright (1997), American Institute of Physics)
(NPB) and Alq3 are the HIL, HTL and ETL, respectively. A detailed description of the HIL is given in Section 7.5.1.3. In this device, the EML consists of a mixture of Alq3 and N,N-dimethylquinacridone (DMQA), which exhibits high emission efficiency. DMQA concentration should be low enough to keep the molecules apart from each other to prevent excimer quenching (Section 7.3.4). On the other hand, when the dopant concentration is not high enough, the intermolecular energy transfer is not efficient since the distance between the host and dopant is too great. Table 7.1 shows the device performances with different DMQA concentrations. It can be seen that the luminance increases then decreases with increasing the dopant concentration. The first increase results from the higher efficiency of DMQA than Alq3 . And the decrease comes from excimer quenching, also called concentration quenching. We also note that the operation lifetime increases with the incorporation of the dopant since higher radiative emission means lower heat generation, which is beneficial to lifetime extension (Section 7.5.5). Emission wavelength and the CIE coordinates of the doped and undoped materials are similar, which is within the green region. One of the most well-known green dopants Table 7.1
Device performances with different dopant concentrations.40
DMQA (%) Lum. output (cd m−2 ) Efficiency (cd A−1 ) CIEx CIEy EL peak (nm) T 1/2 (h)
0.00
0.26
0.40
0.80
1.40
2.50
518 2.59 0.3872 0.5469 544 4200
1147 5.74 0.3876 0.5858 540 7335
1322 6.61 0.3785 0.5995 540 7500
1462 7.31 0.3922 0.5901 544 7340
1287 6.44 0.4046 0.5799 544 5450
1027 5.14 0.4095 0.5742 544 3650
Adapted with permission from Shi, J. and Tang, C.W. Doped organic electroluminescent devices with improved stability. Appl. Phys. Lett., 70, 1665. Copyright (1997), American Institute of Physics.
200
Introduction to Flat Panel Displays
recently used in OLEDs is 10-(2-benzothiazolyl)-1,1,7,7-tetramethyl-2,3,6,7-tetrahydro-1 H,5H11H[l]benzopyrano[6,7,8-ij]quinolizin-11-one (C545T), which is a derivative of a class of highly fluorescent coumarin laser dyes.41 In addition to improvement of efficiency and lifetime, color tuning is also an important function of guest–dopant systems. By suitable molecular design, it is possible to emit different colors of light with different dopant materials even with the same host. For example, by doping the laser dye molecule DCM1 in Alq3 , a quantum efficiency of about 2.3 % can be obtained with the EL peak at 600 nm.9 With suitable molecular modification, another red dopant material, 4-(dicyanomethylene)-2-t-butyl-6-(1,1,7,7tetramethyljulolidyl-9-enyl)-4H-pyran, known as DCJTB, has been synthesized.42 The efficiency of the red OLED based on DCJTB:Alq3 EML can reach 2.0 cd A−1 with luminance of 400 cd m−2 at a current density of 20 mA cm−2 . One of the criteria for efficient energy transfer from the host to the dopant materials is good spectral overlap between the host PL and the dopant absorption, which implies the dopant should have a larger bandgap than the dopant. This means that it is not applicable to use Alq3 as the host for blue OLEDs. Since OLEDs emit at shorter wavelength and hence higher photon energy than red and green devices, a widebandgap material is required to provide such a high energy. For better energy transfer, a larger bandgap of the host material as compared to that of the dopant is needed. The host material in the EML also has to match the HOMO/LUMO levels for effective energy transfer and charge injection from HTL and ETL. The details of several combinations of guest–host systems have been published. Examples are using distyrylarylene (DSA) derivative as the emitting host material and DSA amine styrylamine 4,4 -bis(2-(9ethyl-9H-carbazol-3-yl)vinyl)biphenyl (BCzVBi) as the emitting dopant43 and doping 2,5,8,11-tetra(tbutyl)-perylene (TBP) into the diphenylanthracene derivatives 9,10-di(2-naphthyl)anthracene (ADN). Related researches are ongoing.
7.5.1.3 HIL, EIL and p–i–n structure As described above, HTL and ETL materials have high carrier mobilities and can transport holes and electrons, for recombination in the EML. Organic materials with high fluorescence quantum efficiency are used as the EML, which determines the emission color. To further improve the hole and electron injection capability from anode and cathode to the organic layers, HIL and EIL are incorporated in OLEDs. A HIL having HOMO level between that of ITO and HTL is used to help the holes inject from the ITO anode into the HTL. Likewise, EIL having LOMO level close to ETL is used to help the electrons inject from the cathode into the ETL. Figure 7.21 shows an example of the insertion of a HIL, 4,4 ,4 -tris(3-methylphenylphe nylaminojtriphenylamine) (m-MTDATA), between the HTL 4,4 -bis(3-methylphenylphenylamino)biphenyl (TPD) and an ITO anode.44 From Equations (7.21) and (7.22), we can see that the current density from thermionic emission decreases exponentially with increasing barrier height. The HOMO value of m-MTDATA is 0.1 eV higher than the work function of ITO and 0.4 eV lower than the HOMO of TPD, which forms a ‘ladder-like’ structure for better hole injection from the ITO to the HTL. Molecular structures and energy diagram are shown in Figures 7.21(a) and (b). From Figure 7.21(d), we note that the luminance at a constant current density is highest in the three-layer structure (HIL/HTL/ETL) of device C, due to better hole injection. When the conventional HTL was replaced by m-MTDATA (device A), the efficiency is lowest since the barrier for hole injection to the EML is highest. Another function of the HIL is to prevent the oxygen from the ITO attacking the organic layer (Section 7.5.5) which extends the lifetime. A copper phthalocyanine (CuPc) layer is used for this purpose. And this improves balances in hole and electron recombination which is a very important parameter for enhancing quantum efficiency in OLEDs. From another viewpoint, the work function of the electrode could also be engineered to improve the injection capability. Figure 7.22 shows surface treatment at the ITO surface. If the anode is treated with acid (Figure 7.22(a)), it will attract the negative charges which form a surface dipole towards the organic
Organic light-emitting devices
N
LUMO 1.9
CH3
3
N
Emergy Level/eV
CH
201
N
N N
N
CH
H C 3
3
H C 3
3.1 HOMO HOMO 5.5 5.8 Alq3 ITO m-MTDATA TPD 5.0
5.1
(a)
ITO coated glass
MgAg
(b)
(b)
MgAg Alq3 TPD
ITO coated glass
(c)
MgAg Alq3
m-MTDATA
TPD
Luminance/cd m–2
10
MgAg Alq3 m-MTDATA
3.7
HOMO
TPD
m-MTDATA
(a)
LUMO 2.4 LUMO
5
104
10
3
102 101
ITO coated glass 100 –1 10
100
101
102
103
Current Density/mA cm–2 (c)
(d)
Figure 7.21 (a) Molecular structures, (b) energy level diagram, (c) device structures and (d) luminance versus current density curves.44 (Reprinted with permission from Shirota, Y. et al., Multilayered organic electroluminescent device using a novel starburst molecule, 4,4 ,4 -tris(3-methylphenylphenylamino) triphenyamine, as a hole transport material. Appl. Phys. Lett., 65, 807. Copyright (1994), American Institute of Physics)
layer, which accelerates the holes from the ITO into the organic layers. Typically, such a surface dipole is represented as the shift of the vacuum levels or the increase of the work function of the anode surface. Due to the vacuum level misalignment, the effective barrier between the ITO and the HOMO of the organic layer is reduced. On the other hand, as shown in Figure 7.22(b), the effective barrier increases which impedes the hole injection by base modification. Dipping the ITO substrate into acid solution is a method to improve the hole injection capability. Another common way is to use an oxygen plasma or UV–ozone
Figure 7.22 (a) Acid and (b) base treatment at the ITO surface.47 (Reprinted with permission from Nuesch, F., Rothberg, L.J., Forsythe, E.W et al., A photoelectron spectroscopy study on the indium tin oxide treatment by acids and bases. Appl. Phys. Lett., 74, 880. Copyright (1999), American Institute of Physics)
202
Introduction to Flat Panel Displays
treatment before depositing the organic layers.45, 46 The oxygen plasma can not only effectively increase the work function of ITO, but also remove hydrocarbon contaminants on the as-grown ITO substrate. The work function of the cathode should be as low as possible for better electron injection. Figure 7.23 shows the current density versus voltage of OLEDs with different cathode structures. Devices with aluminum cathodes exhibit the highest driving voltage (17 V) at 100 mAcm−2 . By replacing the Mg0.9Ag0.1 alloy as the cathode material, a 4 V reduction in driving voltage is achieved due to the lower work function. In this alloy, a small amount of silver is added to improve the stability against the environment and form better adhesion to the organic layers.4 Although low work function materials, such as alkali metals and alkaline earth metals, are preferred for the cathode, they are reactive under typical environments containing oxygen and water, which increases the difficulties in device processing and material handling. An alternative is to deposit an ultrathin and insulating layer, such as LiF, between the air-stable metal, such as aluminum, and the organic layers.48, 49 Details of the physical mechanism are not clear yet. One of the most popular explanations is that the LiF decomposes into lithium and fluorine upon aluminum evaporation. Then lithium penetrates into the organic layer which forms radical anions and effectively increases the electron concentration in the ETL. Also, it forms dipole layers in the organic–cathode interface which facilitate electron injection. However, due to the insulating characteristics, thinner or thicker LiF layers result in voltage increase, as shown in Figure 7.23. In addition to the energy mismatch and the low mobility, the high driving voltage of OLEDs also arises from the low free carrier concentrations of organic materials. The p- and n-type doping technology, with the concept similar to that in semiconductor LEDs, is introduced. For the p-type doping, a HTL material is usually doped with a Lewis acid, which attracts electrons and free holes are generated. This shifts the Fermi level in the organic materials from the center of the bandgap to close to the HOMO, which improves the hole transport capability. Also, due to the increase of the carrier concentration, the Fermi level alignment between the electrode and the p-doped organic layer results in an ohmic-like injection 1000
AI/LiF Current density (mA/cm2)
100
Mg0.0 Ag0.1
AI
LIF (0.5 nm) 10V
13V
17V
LIF (1.0 nm) 10
0
0.1 0
5
10
15
20
Drive voltage (V) Figure 7.23 Plots of log J versus V curves of OLEDs with different cathode materials.50 Reprinted with permission from L.S. Hung, C.W. Tang, and M.G. Mason, Enhanced electron injection in organic electroluminescence devices using an Al/LiF electrode. Appl. Phys. Lett., 70, 152. Copyright (1997), American Institute of Physics.
Organic light-emitting devices
203
property at the interface, which facilitates hole injection. Some effective candidates of guest–host systems for p-type organic materials include doping tetracyanoquinodimethane (TCNQ) derivatives,51 V2 O5 or FeCl3 into zinc phthalocyanine (ZnPC) or arylamine derivatives.52 In contrast to p-type doping, n-type molecular doping is more difficult due to the requirement of the HOMO value of the dopant being smaller than the LUMO of the host for providing the free electrons, which makes such materials unstable against oxygen.53 An alternative approach is the use of alkali metals like lithium, sodium or cesium doped into organic materials, which is called the metal dopant (MD) technique.54 These metals easily release a free electron to the organic host for conducting. With such a MD technique, not only can carrier transport characteristics be much improved by n-type doping, but the interface between the electrode and organic material can also be modified from Schottky barrier to ohmic contact which helps carrier injection and voltage reduction. By applying the p- and n-type doping techniques, p–i–n OLEDs can be fabricated with a low driving voltage and higher power efficiency.55 However, the strong acid and base characteristics of the p- and n-type doping sometimes lead to difficulties in fabrication. Also, the operation lifetime of MD technology is still an issue since it is easy for the metal to diffuse into the EML as an exciton quencher during operation.56
7.5.1.4 Top-emission and transparent OLEDs Typically, an OLED has a transparent ITO anode and a reflective cathode, with high and low work functions. For improving the flexibility in the design of the driving circuit (Section 7.5.3.3) and external quantum efficiency (Section 7.7), top-emission and transparent OLEDs have been fabricated. By replacing the reflective cathode with a transparent conductive layer (e.g. ITO), a see-through display can be achieved. However, since the work function of ITO is high (∼4.7 eV), which is not suitable for electron injection, a suitable buffer layer between the ETL and ITO ‘cathode’ is needed for better carrier injection capability. Such an opaque buffer layer must be thin enough to provide high transmittance. Mg:Ag and LiF/Al bi-layer of several nanometers thickness is suitable for this application due to the good electron injection capability. Another function of the buffer layer is to protect the organic layers during the fabrication of the ITO ‘cathode’, where a sputter process is needed. Although there is a buffer layer in such a device, the radio-frequency power of the sputtered ITO cathode has to be controlled very carefully in order to decrease the energetic ion damages to the underlying organic layer. Under this process condition, the deposition rate is usually as low as several angstroms per second. The optical transmittance of the buffer layers is one of the major concerns in such an architecture. A thicker buffer layer absorbs more light and hence decreases the OLED luminance and transmittance. On the other hand, a thinner buffer layer may not be enough to protect against sputtering damage. However, a thick ITO cathode, several hundred nanometers, is necessary for reducing the ohmic loss due to the low conductivity of ITO, i.e. around 104 S cm−1 . Such a conductivity value is one to two orders of magnitude less than those of metal materials. Hence, a thin and semitransparent metal material formed by thermal evaporation can be used as the transparent or semitransparent cathode. Such a thin cathode of 20–30 nm thick can be samarium,57 Ca/Mg58 and LiF/Al/Ag.59 For a top-emission OLED, which means light comes out from the top side rather from the substrate side, a reflective anode is also needed. Deposition of a reflective metal, such as silver, underneath the ITO is straightforward.60 Some high work function metals, such as gold and titanium, are good candidates, in terms of injection capability. A high reflectivity over the whole visible range is also needed for better optical characteristics. Oxidation of silver produces AgOx with a high work function which forms an effective HIL, while the solver provides good electrical characteristics at the same time.59 In a topemission OLED, the reflective anode and the semitransparent cathode form a microcavity which exhibits a strong multibeam interference (Section 7.4.4). Riel and coworkers fabricated a top-emission OLED with a dielectric layer of ZnSe capped on the semitransparent cathode, as shown in Figure 7.24(a).61 They demonstrated that the EL intensity and spectral characteristics are a function of dielectric thickness, as
204
Introduction to Flat Panel Displays
6 Simulation Experiment
5 Intensity [a.u.]
4
dielectric layer cathode Alq3 BCP CBP + Ir(ppy)3 NPB CuPc metal anode substrate
3 2 1 0
0
20
(a)
80 40 60 ZnSe Tickness [nm] (b)
100
120
Figure 7.24 (a) Schematic of a top-emission OLED with dielectric layer (ZnSe) and (b) experimental and simulated intensity in the normal direction.61 (Reprinted with permission from Riel, H., Karg, S., Beierlein, T. and Rieß, W., Tuning the emission characteristics of top-emitting organic light-emitting devices by means of a dielectric capping layer: an experimental and theoretical study. J. Appl Phys., 94, 5290. Copyright (2003), American Institute of Physics)
shown in Figure 7.24(b). The dielectric layer can enhance the outcoupling light intensity in the forward direction. Also, angle-independent spectra can be achieved by tuning the dielectric thickness. By engineering the transparent, semitransparent and reflective characteristics of the electrodes, some interesting displays can be fabricated. Based on the transparent OLED, the display can show information from the both sides.62 Moreover, a double-sided OLED display with different content is also possible by engineering the optical characteristics of the electrode.63 A potential application for such a display is in clam-shape mobile phones, which need two displays. Here, one double-sided display can replace the two displays, and possibly the backlight unit, which can greatly reduce the thickness of the mobile phone.
7.5.2 Polymer OLEDs When distinguished by molecular weight, there are two main kinds of OLEDs. For a molecular weight less than 1000 g mol−1 , the device is called a small-molecule OLED, which can usually be thermally evaporated under high vacuum. For a molecular weight over 10 000 g mol−1 , the device is called a polymer OLED (or polymeric light-emitting device, PLED). In 1990, a polymer OLED was fabricated by Burroughes et al.5 using the spin-coating technique. A thin film of poly(p-phenylene vinylene) (PPV) with yellowish-green emission was used with a device configuration of Al/Al2 O3 /PPV/Al. The emitting spectra and efficiency of the conjugate polymer can be tuned by suitable molecular design. Figure 7.25 shows some common repeating units of polymer EL materials, such as polyphenylenes (PPP), polyfluorene (PF) and PPV. In 1991, Braun and Heeger reported a red R2
R1
n
n PPP
PF
n PPV
Figure 7.25 Molecular structures of some polymer EL materials.
Organic light-emitting devices
205
polymer LED containing poly[2-methoxy-5-(2-ethylhexyloxy)-1,4-phenylenevinylene] (MEH-PPV).64 Ohmori et al. reported a blue-emitting polymer LED, containing poly(9,9-dialkylfluorene)s, with an EL peak emission wavelength of 470 nm.65 The long-chain characteristics of polymer molecules provide lots of flexibility for chemists to design molecules to improve the carrier transport characteristics and recombination coefficient, and to adjust the emission wavelength spanning the whole visible range. The driving voltage of polymer OLEDs is typically lower than that of small-molecule ones due to the higher conductivity. Also, the uniformity of polymer films can be greatly improved by suitable molecule design, which decreases the recrystallization phenomenon which is sometimes observed in amorphous thin films in small-molecule OLEDs. Since the polymer will be decomposed upon thermal treatment before it obtains enough kinetic energy for evaporation, thin-film formation of polymer OLEDs is by the solution process. This consists of dissolving the polymer into solution and then coating the solution on the substrate. The solvent is vaporized under vacuum or by thermal treatment, and the polymer thin film is formed. Residual solvent in the organic thin film is sometimes an impurity quencher which results in nonradiative recombination. Also, it diffuses during device operation or storage, which shortens the device lifetime. To achieve low driving voltage, high recombination efficiency and high emission efficiency, a multilayer structure is typically desired for optimizing the optical and electrical characteristics in an OLED. However, it is not easy (although possible) to fabricate a polymer OLED with more than two layers. Since typical organic nonpolar solvents will dissolve most EL polymers, it is common to apply PEDOT:PSS, which can be dissolved in a polar solution (water soluble), as the HIL and HTL prior the EL polymer thin-film formation. Several methods have been proposed to fabricate a multilayer polymer OLED. Self-assembly methods66 or electropolymerization provide options to avoid the solubility problem.67 Another approach is to have a series of functional polymers that could be spin-coated on the substrate. This requires polymers that are soluble in polar and nonpolar orthogonal solvent systems.68
7.5.3 Device fabrication Since OLEDs and PLEDs are sensitive to moisture and oxygen, the thin-film process should be carried out in vacuum or inert gas environments. After the organic thin-film formation, the sample should be encapsulated or passivated without exposure to the ambient environment. Hence, for an OLED display, the circuitry fabricated by conventional semiconductor processes (e.g. lithography, etching and deposition) should be produced before the thin-film process. Typically, a conventional OLED is fabricated on a glass substrate. The transparent conductive material ITO can be used as the anode. Light emission from the organic layers propagates through the transparent ITO and glass substrate. However, the resistivity of ITO is typically 10−4 cm−1 which is one to two order of magnitude higher than that of the reflective metals, such as aluminum and silver, and this high resistance results in voltage drops of the ITO transmission lines, which in turn induces: (1) nonuniformity of panel luminance and (2) delay and distortion of the control signals. Hence, some assisting anode, such as aluminum or chromium, could be used outside the emission region to improve the current conduction. Since the conductivities of organic materials are quite low, the current pathways are only in the vertical direction (perpendicular to the glass substrate) and will not propagate horizontally (parallel to the glass substrate), which means the light emission occurs at the regions where anode and cathode overlap observed from the vertical direction. An insulation layer, such as polyimide, is needed to define the pattern of the emission region on the ITO and beneath the organic layers. Such a layer is critical since it provides a flat surface and buries the circuitry underneath which is important for OLED thin film deposition due to its very thin nature (100–200 nm). Thin-film transistors (TFTs) may be needed for AM driving techniques. In an OLED display, at least two TFTs in one pixel are needed due to the diode junction characteristics. Mobility of the transistors should be as high as possible to provide sufficient current, i.e. electron–hole pairs, for photon generation. Mobility decrease and threshold shift are two typical phenomena observed in TFT degradation due to the current stress.
206
Introduction to Flat Panel Displays
Once the backbone of the OLED display is fabricated, the organic thin-film and encapsulation processes should be accomplished without exposure to the ambient environment. For small-molecule OLEDs, the organic materials can be sublimed through thermal evaporation under high vacuum. By controlling the temperature, the deposition rate, of as low as 0.1 nm s−1 , can be controlled precisely. For polymer materials, due to their large molecular weight, solution processes are needed. Two common fabrication methods are spin-coating and ink-jet printing. After film formation, the devices are transferred for encapsulation under inert environments for longer lifetime. Depositing passivation layer(s) upon the OLED directly is also possible which is beneficial for reducing the thickness of the display and has potential for flexible substrate applications.
7.5.3.1 Thin-film formation For small-molecule OLEDs, a common method to deposit the organic thin film is by thermal evaporation under high vacuum, since the molecular weight is low enough for sublimation. As illustrated in Figure 7.26, the organic layers and the cathode materials are deposited sequentially on the patterned ITO glass substrate. Since there is no wet process allowed after the thin-film deposition, the patterns of the organic layers and the metal are defined by the shadow mask, which is positioned near to the glass substrate. This is a metal plate with holes, where the organic vapor from the thermal sources can pass through and deposit on the substrate. Typically, the substrate is kept at room temperature, which means the evaporated molecules cannot obtain enough kinetic energy to move on the glass substrate and distribute randomly in an amorphous manner. An amorphous thin film is important for better surface uniformity. The deposition rate is controlled by the evaporation temperature. Higher temperature results in faster deposition rate, and higher throughput. Device performances, such as efficiency and lifetime, are also dependent on the deposition rate.69 However, when the deposition temperature is too high, organic material may decompose which results in material degradation and device failure. To obtain a uniform thin film, the distance between the substrate and the organic source should be as large as possible, which results in a low usage rate of the organic material. Typically, only 10 % or less is deposited on the glass substrate, while the other 90 % or so of materials are coated on the wall of the chamber. This not only wastes the organic material, but also limits the panel size. A linear source, which is a combination of ‘point’ thermal sources as a ‘linear’ one, can effectively reduce the size of the chamber, increase the material usage rate, improve the throughput and enlarge the substrate size (as large as 1100 mm × 1250 mm; Gen V glass substrate).70 Substrate
base layer
Anode
organic layer cathode
cathode Figure 7.26 Thermal evaporation.
Organic light-emitting devices
Cathode Organic layers Anode Glass
Red Green Blue
207
White
Red Green Blue
Cathode Organic layers Anode Color filter Glass
Blue
Cathode Organic layers Anode CCM layer Glass
Red Green Blue
Figure 7.27 Fabrication methods for full-color OLED: (a) lateral emitter, (b) white OLED with color filter and (c) color change material (CCM).
For a full-color display, red, green and blue subpixels are needed. It is straightforward to use red, green and blue OLEDs for this purpose. However, as discussed above, the conventional lithography process is not available for OLED thin-film formation. Typically, there are three fabrication methods for full-color OLED displays (Figure 7.27): (1) use of a lateral emitter, (2) use of a white OLED with color filters and (3) use of a blue OLED with color change materials (CCMs). As shown in Figure 7.28, lateral emitter involves depositing red, green and blue OLEDs in different zones though fine-pitch shadow masks. Device efficiency and operation lifetime is higher for lateral emitter, compared with the other two fabrication methods, since the OLEDs for three primary colors can be optimized independently. However, typically, the subpixel pitch is about 70 m, which means the shadow mask cannot be too thick (∼100 m). Since the glass substrate is as large as 370 × 470 mm (Gen II glass), it is not easy for handling and processing. When the shadow mask is used continuously for a long time (e.g. a week), the accumulation of the evaporated organic materials near the hole edge of the fine-pitch shadow mask may reduce the hole size, which in turn results in dimension reduction and misalignment of the deposited region. Also, the thermal expansion of the metal shadow mask may also result in misalignment in the high-temperature deposition environment. The distance between the
Figure 7.28 Fine-pitch shadow mask for a full-color OLED.13 (Reprinted from Hung, L.S. and Chen, C.H., Recent progress of molecular organic electroluminescent materials and devices. Mater. Sci. Eng. R 39, 143. Copyright (2002), with permission from Elsevier)
208
Introduction to Flat Panel Displays
shadow mask and the substrate should be kept small for better alignment. However, the surface of the substrate is easily scratched by the shadow mask if they are too close, which results in difficulties in device fabrication. To simplify the process, the conventional color filter technique, as used for LCDs, can be used for fullcolor OLEDs. By fabricating a white OLED, without a complex fine-pitch shadow mask, and blocking the light through color filters at each subpixel can effectively improve the fabrication yield. However, similar to the case of LCDs, two-thirds of the light is absorbed by the color filter, which means the efficiency and lifetime will decrease. Operation lifetime describes the luminance decay with time, which is often different for different colors of OLEDs. For the lateral emitter, different degradation rate may result in a color shift with time. However, the color stability remains good when using a white OLED with color filter. Another possible way to improve the efficiency and simplify the fabrication is by using a CCM, which can absorb blue light, and generate red and green light. As shown in Figure 7.27, EL from the blue OLED itself is used for the blue subpixel. By using appropriate CCMs on the other two subpixels, red and green light can be generated. However, material selection, fabrication process and material stability still remain issues for this technique. By using the laser-assisted pattern technique, a high resolution and large-size substrate can be realized at the same time with the lateral emitter arrangement.71 Laser-induced thermal imaging (LITI) is one of the most important fabrication technologies; its operation principle is shown in Figure 7.29. An organic thin film is first deposited uniformly on a donor film. Then, upon laser illumination on certain regions, light-to-heat conversion (LTHC) absorbs the light and converts it to heat, which induces the thermal expansion of the LTHC with high spatial resolution, and in turn releases the organic thin film onto the substrate. The resolution of this technology depends on the beam size of the laser beam. Limitation of the substrate size comes from the moving distances of the laser head. By increasing the numbers of laser heads, the throughput can be increased. Displays with resolution of 302 ppi and glass substrate of 730 mm × 920 mm (Gen IV) have been demonstrated with LITI technology. Also, Sony Corporation has demonstrated a 27.3-inch OLED display with full-HD resolution (1920 × 1080) using laser-induced pattern-wise sublimation (LIPS) technology.72 However, since the film formation process is different from that of thermal evaporation, the operation lifetime of OLEDs formed using laser-assisted patterning is sometimes shorter. The fabrication methods for polymer OLEDs must be ‘solution’processes since the materials cannot be sublimed due to their high molecular weight. When these materials are heated, they decompose before they obtain enough kinetic energy for sublimation. Chloroform, dichloroethane, toluene and xylene are commonly used solvents. After dissolving the polymer materials into the solvent, the solution is spin-coated or
Figure 7.29 Schematic of the LITI process.71 (Source: Lee, S.T. et al. (2007) LITI (laser induced thermal imaging) technology for high-resolution and large-sized AMOLED. SID Tech. Dig., 1588. Reproduced by permission of SID)
Organic light-emitting devices
209
ink-jet printed onto the substrates. By drying off (e.g. under vacuum or heating) the solution, polymer thin films can be obtained. Spin-coating is a common method for semiconductor fabrication. By controlling the rotational rate and viscosity of the solution, thin films with precise thickness (resolution of several nanometers) can be obtained. For full-color fabrication, different solutions can be cast in different positions. Ink-jet printing, which is a conventional process in printing technology, can be used.73 Seiko Epson Corporation has demonstrated a 40-inch polymer OLED display using multi-head ink-jet printing technology.
7.5.3.2 Encapsulation and passivation Many organic materials are sensitive to moisture and oxygen. Also, the oxidation of metal electrodes may degrade the OLED properties.74 Therefore, the isolation of the OLED from oxygen and water is necessary for lifetime extension. Typically, OLEDs are hermetically sealed with another cover using an adhesive such as UV-cured epoxy, as shown in Figure 7.30. The cover can be another glass, or a metal lid. Possible pathways for environmental moisture and oxygen are from the adhesive and the boundary of the adhesive and the glass substrate or the cover. To reduce water and oxygen permeability into the OLED, the gap between the substrate and the cover should be as small as possible to reduce the ‘channel width’. At the same time, the width of the epoxy should be as wide as possible to increase the ‘channel length’. Also, the interface of the epoxy with glass substrate and cover may form a channel if the bonding strength is not good enough. Even though the epoxy can entirely prevent attack from the environments, there are still possible residual solvents from the fabrication prior to the thin-film deposition, which will vaporize gradually and degrade the OLED. Hence, a desiccant such as CaO is incorporated to react with any water which is residual or diffuses through epoxy seal, as shown in Figure 7.30.75 It is also possible to replace the cover with a passivation layer directly upon the OLED of several micrometers thickness. In this way the thickness of the display can be reduced to that of about one glass substrate (∼0.55 mm). Also, the process flow of OLED fabrication becomes simpler. Since the organic materials cannot resist high temperature, a process temperature lower than 100 ◦ C is needed for the passivation layer. This passivation layer should not have cracks or pinholes which provide the channels for environmental attack. The thermal stress should be as small as possible so that the organic and metal thin films underneath are not damaged. Recent research has provided several concepts to form good passivation layers with low water vapor permeation rate (WVPR), such as the deposition of polymer by PECVD,76 single adhesive layer with an inorganic cap,77 multiple layers consisting of repeatable organic/inorganic structure78 and the incorporation of desiccant and barrier layer.75 The main advantages of using a polymer as passivation layer are mechanical and chemical stability and strong adhesion to the underlying layer.76 The merits of inorganic materials are transparency and abrasion resistance. However, a low WVPR is only achieved using a thick layer (hundreds of micrometers), which results in a large mechanical stress that can crack the film beneath the passivation layer. To reduce the stress without increasing the WVPR, an adhesive layer is prepared before the deposition of the inorganic passivation layer.77 The formation of pinholes is inevitable in a thick inorganic passivation layer. So, to fabricate
Cover Substrate
Desiccant Adhesive Cathode Organic layers ITO
Device Substrate
Figure 7.30 Encapsulated OLED.
210
Introduction to Flat Panel Displays
a pinhole-free film and block the passage of water vapor through grain boundaries, inorganic/organic multiple layers are adopted.74, 79, 80 Such a passivation layer not only can be deposited on the OLED, but also on a flexible substrate, e.g. poly(ethylene terephthalate) (PET), to reduce the WVPR from 10−1 –101 g m−2 per day down to levels to meet the requirements of OLEDs (<5 × 10−6 g m−2 per day for a storage lifetime longer than 10 000 h).76 Simultaneously, the surface roughness of the flexible substrate, which yields dark spots after the fabrication process, is modified by the application of multilayer structure. The height of spikes on the PET surface is reduced from 150 to <10 Å. By embedding getter particles into the polymer substrate or polymer lid, it is possible not only to block water vapor from outside, but also to absorb it before it contacts the device.
7.5.3.3 Device structures for AM driving As discussed in Chapter 3, silicon-based TFTs on AM backplanes are a sensitive to visible light which should be shielded by a black matrix. This means some pixel size would be sacrificed without emitting light for the TFT layout. So the aperture ratio of an AM-OLED is typically low, as shown in Figure 7.31(a), which results in low display luminance. This would be even worse if more than two transistors were used. Top emission is one of the most promising technologies for achieving high aperture ratio AMOLEDs, as shown in Figure 7.31(b).81 Typically, a top-emission OLED has a high reflective anode and a semitransparent or transparent cathode. Hence, light comes out from the semitransparent cathode side. The TFTs can be hidden underneath the reflective anode and hence the aperture ratio is increased. As shown in Equations (3.17) and (3.18), increasing the W /L value can increase the current density. Since the TFT and OLED are stacked together without interference, the whole pixel area can be used for the TFT. Hence, it is possible to use an amorphous silicon (a-Si) TFT to drive an OLED even though the mobility value is low. However, the fabrication process and the optical design of such a device are quite different from those of a conventional bottom-emission OLED. Degradation of a-Si TFTs under current stress is another issue which needs to be considered. Configurations of TFTs and OLEDs for AM-OLED displays are still being developed. LIGHT Cathode Emitting Material ITO
TFT Glass
Seal
Transparent Cathode Metal anode
Emitting Material
TFT LIGHT
Glass
Figure 7.31 Cross-section of an AM-OLED: (a) bottom-emission and (b) top-emission OLED configuration.81 (Reprinted from Lee, C.J., Pode, R.B., Moon, D.G. and Han, J.I., Realization of an efficient top emission organic light-emitting device with novel electrodes. Thin Solid Films, 467, 201. Copyright (2004), with permission from Elsevier)
Organic light-emitting devices
211
7.5.4 Electrical and optical characteristics In this section, the J–V characteristics and the optical performances of a two-layer OLED are demonstrated and analyzed, experimentally and numerically. The HTL and ETL (EML) materials are NPB and Alq3 , respectively. Their device band diagram are shown in Figure 7.32. Five two-layer OLEDs are fabricated with varying HTL and ETL thicknesses. The HTL thickness varies from 200 to 1000 Å, at a step of 200 Å, while the ETL thickness varies from 1000 to 200 Å in order to keep the total organic thickness at 1200 Å. Table 7.2 shows the device configuration. Figure 7.33 shows the J–V characteristics of the experimental and simulated results. We note that the current density increases sharply with increasing driving voltage, which is a typical SCLC and TCLC behavior. The driving voltage of device 1 is highest, and decreases with thicker HTL materials. The electron and hole mobilities of Alq3 and NPB are 10−6 –10−3 and 10−4 –10−3 cm2 /V−1 s−1 , respectively, at a field strength of 0.1–1 MV cm−1 . This means that hole transport is much faster than electron transport. This also implies the electric field mainly drops across at the Alq3 layer because of its low electron mobility. Figure 7.34 shows the distributions of charge density and recombination rate of devices 1, 3 and 5. We can see that many holes and electrons are blocked at the HTL–ETL interface due to the energy barriers of LUMO and HOMO value differences between NPB and Alq3 , respectively. Note that the barrier for holes (0.3 eV) is smaller than that for electrons (0.6 eV) which means the holes can penetrate into the ETL, to recombine with the electrons, as shown in Figures 7.34(a)–(c). Such a carrier pile-up near the HTL–ETL interface confines the carriers, which increases the recombination efficiency. Figures 7.34(d)–(f) show the recombination rate (R), which follows Langevin theory. We can also see that the recombination mainly takes place in the ETL and neat HTL–EML interface due to the carrier pile-up.
2.4 eV 3.0 eV
Al 3.2 eV
NPB ITO 4.9eV
Alq3
5.4 eV 5.7eV Figure 7.32 Energy diagram of the two-layer OLED.
Table 7.2
Layer structures of the two-layer OLEDs.
Unit 1 2 3 4 5
ITO glass ITO glass ITO glass ITO glass ITO glass
HTL (NPB) (Å)
ETL (Alq3 ) (Å)
LiF (Å)
Al (Å)
200 400 600 800 1000
1000 800 600 400 200
12 12 12 12 12
1500 1500 1500 1500 1500
212
Introduction to Flat Panel Displays
0.10 1 2 3 4 5
0.08 0.06
Current density (A/cm2)
Current density (A/cm2)
0.10
0.04 0.02 0.00
2
4
6 8 Bias (V)
10
12
1 2 3 4 5
0.08 0.06 0.04 0.02 0.00
2
4
6 8 Bias (V) (b)
(a)
electrons holes
10
1016 14
1012 0
20
1020
10
14
1012 0
20
40 60 80 Position (nm)
40 60 80 Position (nm) (d)
1014 1012 1010 0
20
1022 1020
100 120
40 60 80 100 120 Position (nm) (c)
HJ-5 (NPB 100nm/Alq3 20nm)
1024 1022 1020 1018
1018 20
1016
100 120
HJ-3 (NPB 60nm/Alq3 60nm)
1024
1018 0
electrons holes
1018
(b)
rnp (cm–3s–1)
1022
1016
100 120
HJ-1 (NPB 20nm/Alq3 100nm)
1024 rnp (cm–3s–1)
40 60 80 Position (nm) (a)
10
18
No. 5 (NPB 100nm/Alq3 20nm)
1020
electrons holes
rnp (cm–3s–1)
18
No. 3 (NPB 60nm/Alq3 60nm) Carrier density (cm–3)
Carrier density (cm–3)
No. 1 (NPB 20nm/Alq3 100nm)
10
12
(a) Experimental and (b) simulated J–V characteristics of devices 1 to 5.
Carrier density (cm–3)
Figure 7.33
10
0
20
40 60 80 100 120 Position (nm) (e)
0
20
40 60 80 100 120 Position (nm) (f)
Figure 7.34 Simulated results of (a–c) carrier density and (d–f) recombination rate distributions of devices 1, 3 and 5.
Figure 7.35(a) shows the luminance versus current density for devices 1 to 5. We can see that the luminance increases then decreases as the HTL thickness increases. As shown in Figure 7.34, maximum recombination occurs at the HTL–ETL interface. To have constructive wide-angle interference, the recombination position must be located at λ/4n from the reflective cathode, derived from Equation (7.39). Considering the peak wavelength of PL emission at 530 nm and that the refractive index of the organic material is about 1.6, the λ/4n value is about 82 nm, which means device 2 should be close to the optimization condition of the constructive wide-angle interference. Experimental spectra of the devices are shown in Figure 7.35(b). The EL emission spectral peaks are at 558, 544, 536, 534 and 526 nm for devices
Organic light-emitting devices
213
1 2 3 4 5
3000 2000 1000 0 0.00
0.03 0.06 0.09 0.12 Current Density (A/cm2) (a)
Figure 7.35
1 2 3 4 5
1.0
Intensity (a.u.)
Luminance (cd/m2)
4000
0.15
0.8
0.6 480
520 560 Wavelength (nm) (b)
600
Measured (a) luminance versus current density and (b) EL spectra of devices 1 to 5.
1 to 5. Such a blue shift can be understood from the wide-angle interference. Since the interference is wavelength dependent, the enhancement ratio blue-shifts on increasing the recombination zone towards the cathode, i.e. increasing the HTL thicknesses.
7.5.5 Degradation mechanisms Lifetime is one of the major obstacles to wide commercial applications for OLEDs. Although the operation principles of LEDs and OLEDs are similar, the lifetime of OLEDs is much shorter (typically less than a tenth) than that of LEDs. The main reasons are that (1) OLEDs have a thinner layer structure, (2) organic materials are sensitive to moisture and oxygen and (3) organic materials are not as robust as semiconductors due to larger vibration energy. Degradation in OLEDs can be categorized as three independent modes: (1) dark-spot formation, (2) catastrophic failure and (3) intrinsic degradation. Dark-spot formation is the increase in nonemissive regions with time. This occurs even during OLED storage,82 and the growth rate may be accelerated by operation. The formation mechanisms of the dark spots are electrochemical reactions of the organic (or electrode) materials with environments. To slow down dark-spot formation, as described in Section 7.5.3.2, encapsulation or passivation is needed to prevent oxygen and moisture from attacking the OLED. There are many possible root causes which lead to the dark-spot phenomenon. Dark spot means nonemission at a certain area rather than luminance decrease in the whole emissive zone, which means structural defects in OLEDs produce the dark spot. The root causes of the structure defects include the following: (1) the layer between cathode and organic layers produces bubble structure which impedes electron injection and loses contact between the cathode and organic layer, which leads to a nonemissive zone;83 (2) pinholes on the cathode provide a pathway for water and oxygen diffusion which oxidize the cathode metal and react with the organic materials;84 and (3) intense decomposition of the ITO and organic film caused by local temperature rises releases volatile species which lead to local delamination and to oxidation of the cathode.85 Figure 7.36 shows OLED images during operation taken under an optical microscope, just after device fabrication and after storage at 85 ◦ C/100 % relative humidity for three hours. It is clear that the dark spots exist when the device is fabricated, and they grow with time. Typically, control of substrate cleanliness and smoothness and device fabrication process reduces the density of structural defects, and the potential of dark spots is also reduced. Another way is using hermetic sealing of the OLED in an inert gas environment, which can control dark-spot degradation.86 Heating the substrate between the organic evaporation87 can also reduce the dark-spot area. But this can change some organic layer characteristic at the same time.
214
Introduction to Flat Panel Displays
(a)
(b)
Figure 7.36 Images of EL from an OLED showing dark spots (a) when the device is just fabricated and (b) after storage at 85 ◦ C/100 % relative humidity for 3 h.
Cathode ETL HTL ITO Glass Figure 7.37
ITO spike in an OLED which results in catastrophic failure.
Catastrophic failure means a luminance decrease and huge current increase because of electrical shorts caused by defects in organic layers.88 This also originates from morphological defects, such as ITO spikes. Since the organic layer thickness is only 100 nm, substantial roughness of ITO results in the topological change of the subsequent organic thin films. As shown in Figure 7.37, if there is an ITO spike, the organic thin films will be thinner there. When the device is operated under high electric field or after a long time, the progressive electric short phenomenon appears because of collapse of the defect position. To improve this, it is straightforward to maintain the morphology of the OLED. After device fabrication, it is also possible to burn out localized conduction filaments by applying a reverse bias.89 Even though the morphology and encapsulation of an OLED are completely perfect, uniform luminance decay and driving voltage increase with time under constant current driving can still be observed, which is called intrinsic degradation. There are many reports on the mechanisms of OLED degradation. (1) Ions from the electrodes diffuse to the light-emitting zone: indium ions from the ITO anode and metal ions from the cathode (Mg:Ag or LiF/Al) will diffuse through into the EML.90, 91 They form fluorescence quenchers and decrease the luminance under a fixed driving current. Meanwhile, the mobile ions also create a built-in voltage that increases the driving voltage under the same electrical current. (2) Impurities that exist in the organic materials migrate in devices: Zou et al. reported that AC-driven devices may have longer lifetimes than those of DC-driven devices because the impurities may be drawn back in the reverse bias of the AC mode.92 (3) Decomposition of light-emitting materials: Cao et al. reported that the loss of conjugation of PPV was the main decay mechanism and reported that the intrinsic lifetime of PPV-based PLEDs was limited by the amount of charge flowing through the devices rather than AC or DC drive modes.93 (4) Loss of contact of electrode–organic layers: this type of decay is not intrinsic and the device may be recovered by recoating the cathode.94
Organic light-emitting devices
215
Also, one of the major degradation mechanisms which may account for the long-term stability is the ‘unstable cation’ model. As mentioned in Section 7.5.2, the heterojunction in an OLED helps to confine the carriers which improves the current efficiency. However, at the same time, the carriers pile up at the HTL–EML interface. Since the injected holes from the HTL to the EML form cations in the EML material which are chemically unstable, this will accelerate the formation of nonradiative trapping centers and result in luminance decay and voltage increase of a device.95 Popovic and Aziz designed an experiment to observe this degradation mechanism.96 Figure 7.38 shows the PL intensity versus time of 5 nm thick Alq3 , sandwiched between two thick NPB and tetraphenyltriazine (TPT) layers. NPB and TPT are the HTL and ETL materials, providing hole and electron transport through the Alq3 thin layer under a bias. We can see that, after hole transport through the Alq3 layer, a gradual decrease of the PL intensity is observed, revealing a continuous decrease in the PL intensity of the Alq3 on prolonged current flow. The decrease in the PL intensity of the Alq3 points to its degradation as a result of hole transport, thus revealing that cationic Alq3 species are unstable and that their degradation products are fluorescence quenchers. On the other hand, while significant PL decrease is observed in the case of transporting holes into Alq3 , the PL stays remarkable constant in the case of transporting electrons. These results provide evidence that injection of holes into the Alq3 layers of OLEDs is one of the main factors in device degradation. When aging an OLED, not only does the luminance decay but also the voltage increases with constant driving current. Lee et al. have verified that a major degradation mechanism in an OLED is cation formation by conducting a series of experiments with the same organic materials and different device structures.97 If the metal ion diffusion was the major degradation mechanism, a thicker HTL or ETL might result in longer lifetime. However, the operation lifetime of the thicker device was no better. On the other hand, the lifetime correlates with the power efficiency, as shown in Figure 7.39. As mentioned before, higher power efficiency means better energy transfer from electrical to optical power. Hence, the organic degradation comes from the chemical reaction which shows a higher reaction rate with higher temperature, i.e. lower power efficiency. With the same power efficiency, we can see the lifetime is longer with a thicker HTL, i.e. electron-rich device. A thicker HTL means the hole density is less near the recombination interface and there is less cation formation.
1.2 Electrons
Normalized PL
1 0.8
Holes
0.6 0.4 N
N
N
N
N
N
0.2
TPT
0
0
10
30 20 Time (hours)
40
50
Figure 7.38 PL as a function of time for hole and electron only devices.96 (Source: Popovic, Z.D. and Aziz, H., Reliability and degradation of small molecule-based organic light-emitting devices (OLEDs). IEEE J. Sel. Top. Quantum c 2002 IEEE) Electron., 8, 362.
216
Introduction to Flat Panel Displays
1400
Lifetime (hrs)
1200
1000
800
600 4
6 8 Power Efficiency (Im/W)
10
Figure 7.39 Dependence of lifetime on power efficiency. Dashed line and square symbols correspond to electronrich devices. Solid line and triangular symbols correspond to hole-rich devices.97 (Reprinted from Lee, J.H., Huang, J.J., Liao, C.C. et al., Operation lifetimes of organic light-emitting devices with different layer structures. Chem. Phys. Lett., 402, 335. Copyright (2005), with permission from Elsevier)
A linear correlation was also found between the trap formation rate and the quencher formation rate at the HTL–ETL interface, as shown in Figure 7.40. Since voltage increase and luminance decay directly correlate to trap and nonradiative center formation, a linear relationship is found which means the thermally assisted chemical reaction induces organic material degradation near the recombination interface, which in turn quenches the photons and traps the carriers. 80 100 Lum.Decay y(%)
Lum. Decay Rate* (dEn*dEML) (%/hrs*Å)
60
60 6 40 20
Voltage (Y)
8
80
4
0 0
40
80 120 Time (hrs)
140
40
20 0.0
0.4 0.8 1.2 Voltage Raising Rate (mV/hrs)
1.6
Figure 7.40 Voltage increase rate versus the luminance decay rate times the total thickness of EML and ETL. The inset shows typical curves of luminance decay and voltage as a function of time.97 (Reprinted from Lee, J.H., Huang, J.J., Liao, C.C. et al., Operation lifetimes of organic light-emitting devices with different layer structures. Chem. Phys. Lett., 402, 335. Copyright (2005), with permission from Elsevier)
Organic light-emitting devices
217
Since the lifetime of an OLED is as long as 10 000 h under normal operation, some accelerating measurement and extrapolating methods are needed to estimate the lifetime value systematically in a shorter period.98, 99 However, acceleration might introduce other aging mechanisms, making the estimation of the lifetime a complex issue. Under high luminance aging, a typically used relation is L0n t1/2 = constant,
(7.43)
half line (h)
100000
10000
1000
200
400
600
800
L0 (cd/m2) Figure 7.41 Half-life versus initial luminance on a double-logarithmic scale.100 (Reprinted with permission from Féry, C., Racine, B., Vaufrey, D. et al., Physical mechanism responsible for the stretched exponential decay behavior of aging organic light-emitting diodes. Appl. Phys. Lett., 87, 213502. Copyright (2005), American Institute of Physics)
200 85°C
70°C 50°C
150
Luminance (cd/m2)
25°C
100
50
0
1
10 100 1,000 Operating Time (hrs)
10,000
Figure 7.42 Luminance versus time for different environment temperatures on a semilogarithmic scale.101 (Reprinted with permission from Parker, I.D., Cao, Y. and Yang, C.Y., Lifetime and degradation effects in polymer light-emitting diodes. J. Appl. Phys., 85, 2441. Copyright (1999), American Institute of Physics)
218
Introduction to Flat Panel Displays
where n is the acceleration coefficient and t 1/2 is the half-life. As shown in Figure 7.41, the best fit has been obtained with n = 1.7.100 From Figure 7.41 we can see that the half-life is 40 000 h under 200 cd m−2 . If the luminance is increased to 10 000 cd m−2 for acceleration tests, we obtain a half-life of 51 h. Another problem is that the luminance may not decrease monotonically, which means more than one mechanism may influence the OLED degradation during aging tests.101 This increases the complexity in analyzing the results. Figure 7.42 shows luminance versus time for different environment temperatures. We can see that the luminance remains constant, increases and then drops rapidly. There are three distinct mechanisms to account for the degradation behaviors. Also, with different temperatures, the lifetime performance changes.
7.6 Improvement of internal quantum efficiency To improve the IQE, full utilization of triplet excitons is necessary which can increase the OLED luminance by a factor of four, as described in Section 7.3.2. By introducing a heavy metal into the organic materials, the spin–orbital coupling efficiency can be enhanced and strong phosphorescence due to the recombination from the triplet excited state is observed, which makes 100 % IQE possible. A tandem structure, which is series connection of several OLEDs, was proposed to increase the luminance and extend the operation lifetime. Since the OLEDs are stacked together, when one electron–hole pair is input through the external circuit, it appears as if more than one photon id emitted, which means the IQE of this system seems larger than 100%. Since the phosphorescent materials cannot cover the whole visible spectrum (especially the pure blue region), a combination of fluorescent and phosphorescent emitters is needed to harvest the excitons or to sensitize the triplet recombination, for achieving high IQE at the desired wavelength. These two methods are commonly used for broad-band OLEDs, such as white devices.
7.6.1 Phosphorescent OLEDs Typically, phosphorescent emission is a slower and less efficient process than fluorescent emission since the decay of triplet states in conventional organic materials is generally not allowed due to the conservation of spin symmetry.102 This restricts the upper limit of the IQE of an OLED to 25 %, and 75 % of the energy from the triplet excited energy is wasted. However, by introducing a heavy metal atom in a molecule, the radiative decay of triplet dopant is possible. There are two mechanisms accounting for the EL emission from a phosphorescent dopant. The first is through energy transfer and the second is carrier trapping. As shown in Figure 7.43, there are two means of energy transfer in OLEDs. First, Förster energy transfer is a long-range process where donor (D) transfers to acceptor (A) by dipole–dipole coupling (Section 7.3.4.1). However, Förster transfer requires that the transition from excited state to ground state of both D and A must be allowed according to conservation of spin symmetry. Thus, this only transfers energy to the singlet state of the acceptor. In contrast, Dexter energy transfer, a short-range process, transfers energy by exchanging electrons of donor and acceptor. Unlike Förster energy transfer, Dexter energy transfer requires only the total spin to be conserved. It allows both singlet–singlet and triplet–triplet energy transfer. Also, direct trapping of carriers is important for the emission of phosphorescent OLEDs. Since the energy gap of a triplet dopant is usually smaller than that of the host, the triplet dopant may trap carriers and then lead to the direct formation of excitons at dopant sites. As described above, with the strong spin–orbital coupling of heavy-metal complexes, phosphorescent materials can break up the spin-forbidden rule of non-radiative relaxation of the triplet state. Therefore, high-efficiency phosphorescent OLEDs can be realized. One of the most successful green phosphorescent material is fac-tris(2-phenylpyridine)iridium (Ir(ppy)3 ) with 4,4 -N,N -dicarbazolebiphenyl (CBP) as the host, which demonstrates peak current efficiency and power efficiency of 26 cd A−1 and 19 lm W−1 under a luminance of 100 cd m−2 .103 Figure 7.44 shows the device structure and the organic structures.
Organic light-emitting devices
(a)
219
host singlet
guest
host
(b)
dye singlet
energy transfer
host singlet
light
dye singlet
heat
(c)
host singlet
dye singlet
dye triplet ISC
(d)
host singlet
light
dye singlet
light Figure 7.43 Energy transfer mechanisms in a guest–host system of an OLED: (a, b) energy transfer of the fluorescent dopant; (c, d) energy transfer of the phosphorescent dopant.102 (Reprinted with permission from Baldo, M.A., O’Brien, D.D., Thompson, M.E. and Forrest, S.R., Excitonic singlet-triplet ratio in a semiconducting organic film. Phys. Rev. B, 60, 14422. Copyright (1999) by the American Physical Society)
Since CBP is a bipolar transport material, i.e. capable of transporting holes and electrons, one more layer, the HBL, is inserted to confine the excitons in the EML. 2,9-Dimethyl-4,7-diphenyl-1,10-phenanthroline (BCP) is widely used as a HBL due to its wide bandgap and large HOMO value, which can hinder the diffusion of excitons formed in the EML and prolong their residence time, thereby increasing the probability of energy transfer from the host to the phosphorescent dopant. However, the low glass transition temperature (T g ) of BCP (83 ◦ C) may lead to device instability. Thus, an alternative, aluminum(III) bis(2-methyl-8-quinolinato)-4-phenylphenolate (BAlq), with a higher T g value can be used which shows higher stability.104 By using a p–i–n structure to improve the injection of carriers, a green phosphorescent OLED with the highest power efficiency of 62 lm W−1 and current efficiency of 61 cd A−1 at 1000 cd m−2 has been demonstrated.105, 106 By using platinum(II) 2,3,7,8,12,13,17,18-octaethyl-12H,23H-porhine (PtOEP)107 as a red dye, an OLED can be realized with an external quantum yield of 5.6 % using a CBP host.108 However, due to the relatively long phosphorescent lifetime (∼50 s), use of PtOEP tends to lead to triplet–triplet annihilation at high current density. Several other red phosphorescent materials with shorter lifetimes have been proposed, such as bis(2-(20-benzo[4,5-a]thienyl)pyridinato-N,C30)iridium(acetylactonate) (Btp2Ir(acac)) and Ir(3-piq)2 (acac).109 112 Due to the very short exciton lifetime of Ir(3-piq)2 (acac), 1.04 s, it achieves the highest current efficiency of 23.94 cd A−1 under a voltage of 8.29 V.112 Considering the development of blue phosphorescent materials, there are difficulties in using energy transfer to excite deeper blue phosphorescent materials. With the increasing energy gap of phosphorescent dopant, a larger energy gap of the host is needed. Also, even with a suitable host, the larger energy gap of
220
Introduction to Flat Panel Displays
LUMO Levels
2.6 ev
3.2 ev
2.7 ev
3.7 ev exciton formulatiom region
ITO 400 A
α-NPD
4.7 ev
400 A Alq3
MgAg
60 A BCP
200 A In(PPY)3 In CBP
5.7 ev
6.0 ev
HOMO Levels 6.3 ev 6.7 ev lr N
N
N 3
(a)
N
N
(b)
CH3
CH3
(c)
Figure 7.44 Device structure and chemical structures.103 (Reprinted with permission from Baldo, M.A., Lamansky, S., Burrow, P.E. et al., Very high-efficiency green organic light-emitting devices based on electrophosphorescence. Appl. Phys. Lett., 75, 4. Copyright (1999), American Institute of Physics)
phosphorescent dopant may result in difficulty for carrier injection. An efficient blue phosphorescent material is FIrpic which exhibits a current efficiency and power efficiency of 12 cd A−1 and 5 lm W−1 under 100 cd m−2 .113
7.6.2 Tandem structure The operation principle of a tandem device is to inject one electron–hole pair into the EML to produce multiple electron–hole pairs through a charge generation layer (CGL) connected between each OLED unit, as shown in Figure 7.45.114, 115 The CGL can be composed of a metal thin film114 or a tunneling junction.116, 117 In a tandem device, current efficiency, in terms of cd A−1 , would scale linearly with the number of single OLED units, even more than unity. Due to the high current efficiency, a tandem device can achieve the same luminance at lower driving current and generate less heat, which in turn increases the operation lifetime.118 By connecting different color units, high-efficiency white OLEDs can be obtained. With suitable design of the layer structures, devices with high external quantum efficiency and high color purity can also be achieved. One of the key factors in a tandem device is the design and the fabrication of the CGL which can be conductive or insulating thin films. First we discuss the conductive case. With a proper layout design, conductive CGLs can also be used as the contact pads of vertical stacking OLEDs that emit three primary colors, as shown in Figure 7.46. In this case, each OLED unit can be driven independently and can be used for fabricating a full-color display with high aperture ratio, named a stacked OLED (SOLED).119, 120 Typically, high work function materials, like ITO, gold and nickel, are used as the anode of an OLED device for effectively injecting holes into the p-type organic materials. In contrast, low work function materials, like alkali metals and alkaline earth metals, are good candidates for cathode materials.13 However, in a tandem OLED, a common metal CGL is necessary which can inject holes and electrons from the anode and cathode to the p- and n-type organic materials, respectively. Techniques for
Organic light-emitting devices
221
– +
Emissive unit CGL
CGL – +
Emissive unit CGL
– +
– +
(a)
Emissive unit
(b)
Figure 7.45 Structures of (a) a conventional OLED and (b) a tandem device.114 (Source: Matsumoto, T., Nakada, T., Endo, J. et al. (2003) Multiphoton organic EL device having charge generation layer. SID Digest, 979. Reproduced by permission of SID)
Material
VR
VB
VC
ITO CuPc Alq3 Alq3:PtOEP α−NPD Cupc ITO Cupc Alq3 Alq′2OPh α−NPD Cupc ITO Cupc Alq3 α−NPD ITO Glass
Thickness Functions 570 Å 55 Å 150 Å 350 Å 400 Å 55 Å 570 Å 55 Å 150 Å 350 Å 400 Å 55 Å 570 Å 55 Å 500 Å 400 Å 1500 Å – 1mm
}
Transparent contact Buffer ETL Red emiting ETL HTL Hole injection layer
}
Transparent contact Buffer ETL Red emiting ETL HTL Hole injection layer
}
Transparent contact Green emitting ETL HTL Transparent anode Substrate
Figure 7.46 Schematic cross-section with conductive CGLs.119 (Reprinted with permission from Gu, G. et al., Transparent stacked organic light emitting devices: I. Design principles and transparent compound electrodes. J. Appl. Phys., 86, 4067. Copyright (1999), American Institute of Physics)
222
Introduction to Flat Panel Displays
– – –
+ + + +
– – – –
+ + + Figure 7.47 Energy band diagram of a tandem semiconductor LED with magnified reverse biased tunneling junction band energy diagram.118 (Reprinted with permission from Guo, X. et al., Thermal property of tunnel-regenerated multiactive-region light-emitting diodes. Appl. Phys. Lett., 82, 4417. Copyright (2003), American Institute of Physics)
fabricating transparent or top-emitting OLEDs can be used here as the conductive CGL in a tandem device. The concept of the insulating CGL, using a tunneling junction, is illustrated as follows. A reverse biased tunneling n–p junction is sandwiched between two OLED units. After an electron–hole pair is recombined in the previous OLED unit, the electrons in the p-side HOMO will tunnel into the n-side LUMO which can generate radiative recombination again in the next unit. Such a technique is widely used in semiconductor LEDs and laser diodes (LDs) to increase IQE, improve thermal stability and decrease threshold current of a LD, as shown in Figure 7.47.121, 122 In semiconductor devices, such a tunneling junction is fabricated with n- and p-type materials of high carrier concentration by controlling the doping ratio during semiconductor epitaxy. However, intrinsic organic materials usually have large bandgaps, i.e. 2 to 3 eV. Therefore the intrinsic concentration of thermally generated free carriers is generally negligible, i.e. less than 1010 cm−3 .23 Similar to semiconductor materials, impurities are doped into the organic materials that can either transfer an electron to the LUMO states (n-type doping) or remove an electron from the HOMO states (p-type doping) to generate a free electron or hole (Section 7.5.1.3).
7.6.3 White OLEDs For lighting applications, fluorescent lamps have been used for decades. However, the mercury in fluorescent lamps leads to serious environmental pollution. Therefore, a nonpolluting and high-efficiency lighting technology, such as LEDs or OLEDs, is required.123 Also, these two technologies can serve as planar backlights of LCDs. Although semiconductor LEDs have the advantages of high color purity and long lifetime, the problem of differential aging among LEDs in an array causes serious nonuniformity problems, especially for a large panel. White light emission from an OLED can be realized by mixture of two complementary colors or three primary colors. The former method has the advantages of simpler structure. However, it results in the color shifting with various driving voltages and low efficiency compared to monochromatic OLEDs.124, 125 The color-shifting phenomenon results from recombination zone shifting, contributing to field-dependent carrier mobilities. Recently, color stability has been achieved by evaporating a mixed-solution target containing a blue host (BANE) and red dopant (DCM2) as the EML.126 The proposed structure is shown
Organic light-emitting devices
223
LiF/Al TPBI
DCM2 doped BANE
NPB Glass Substrate Figure 7.48 Structure using a mixture of three colors in the same EML.126 (Reprinted from Jou, J.-H. et al., Efficient, color-stable fluorescent white organic light-emitting diodes with an effective exciton-confining device architecture. Org. Electron., 7, 8. Copyright (2006), with permission from Elsevier)
in Figure 7.48. The best power efficiency is 6.5 lm W−1 (9.6 cd A−1 ) at 12 cd m−2 with CIE coordinates of (0.346, 0.343). All color variations are less than (0.007, 0.006) between 100 and 10 000 cd m−2 . White OLEDs with tandem structures, which are several OLED components in series, have been proved to be able to enhance the efficiency,127 as illustrated in Figure 7.49. The efficiency, 22 cd A−1 , of the tandem structure of two stacked components is more than double the efficiency, 8 cd A−1 , of the single device. This amplification comes from not only the nature of tandem structure (one electron–hole pair can generate more than one photon), but also the microcavity effect. However, it suffers from less color purity and spectrum shifting at various view angles. Another concept, different from that mentioned above, of utilizing both singlet and triplet excitons provides dramatically high efficiency at high luminescence and eases the color-shifting problem at various injection current densities simultaneously.128 The principle of management of singlet and triplet
Cathode WOLED Cathode
Connecting Layer
WOLED
WOLED
Glass Substrate
Glass Substrate
8 cd/A, CIE(0.29, 0.42)
22 cd/A, CIE(0.29, 0.42)
Figure 7.49 Architectures of white OLEDs in single (left) and tandem (right) devices, and their corresponding efficiencies and CIE coordinates.127 (WOLED = white OLED.) (Reprinted with permission from Chang, C.-C., Chen, J.-F., Hwang, S.-W. and Chen, C.H., Highly efficient white organic electroluminescent devices based on tandem architecture. Appl. Phys. Lett., 87, 253501. Copyright (2005), American Institute of Physics)
224
Introduction to Flat Panel Displays
S
Forster Transfer
Energy
S S
T HOST excition formation zone
Diffusion Transfer
T BLUE fluorescent dopant
T T
RED GREEN phosphorescent dopant
Figure 7.50 Energy transfer for exciton harvesting white OLED, which utilizes both singlet and triplet excitons.128 (Reprinted by permission from Macmillan Publishers Ltd: Nature 440, 908, Y. Sun et al., Management of singlet and triplet excitons for efficient white organic light-emitting devices. Copyright (2006))
excitions is illustrated in Figure 7.50. First, excitons are generated in the host molecules. Second, singlet excitons are sequentially transferred to blue fluorescent dopants and triplet excitons to red and green phosphorescent dopants. In order to implement this management, fluorescent dopants must be placed in the recombination zone and phosphorescent dopants must be appropriately doped slightly away from the recombination zone. This method not only eases the sharp roll-off behavior of efficiency due to triplet–triplet annihilation, but also reduces the color-shifting problem. Consequently, even at higher luminance, ∼500 cd m−2 , the external quantum efficiency and power efficiency remain high at 18.4 % and 23.8 lm W−1 . To reduce manufacturing costs, the solution process is more suitable than vacuum deposition. A highly efficient white OLED with a combination of a solution-processed phosphorescent blue OLED with an appropriate down-conversion phosphor system has been demonstrated.129 Blue light generated in the blend layer excites the OSRAM phosphors coated on the other side of a glass substrate. This system has a power efficiency of 25 lm W−1 , current efficiency of 39 cd A−1 and CIE coordinate of (0.26, 0.40). This highly efficient solution-based white OLED provides a cheaper manufacture for potential mass production.
7.7 Improvement of extraction efficiency Typically, the external quantum efficiency (EQE), ηex , of an OLED is related to the IQE, ηin , and the light extraction efficiency, ηext , as130 ηex = ηin ηext , (7.44) where ηin is the product of the fluorescent quantum efficiency of light-emitting material ηF , the fraction of recombinations that result in singlet excitons χ and the fraction of carriers recombined in the EML ηre :131 ηin = ηF χ ηre .
(7.45)
For the commonly used ETL material Alq3 , ηF is only about 30 %. It can be improved to nearly 100 % by introducing high-efficiency dopant materials into such a matrix. From spin statistics, EL from an OLED exhibits a value of χ of 1/4 for typically fluorescent materials. Values of ηre of nearly 100 % can be achieved under charge-balance condition by adjusting the thicknesses of the organic layers. From the above discussions, we can see that the upper limit of the ηin value is only about 25 %.132
Organic light-emitting devices
225
However, it is found that the maximum ηext is only about 26.4 % even when the device structure is optimized. Most of the emitted light is confined within the OLED device due to the total internal reflection (TIR) at the interface between glass substrate and air, and waveguide effects of high refractive index organic layers.133 Below are discussed the methods to improve the extraction efficiency. 1. Eliminate the ITO/organic waveguide phenomenon. Trapping of light in a medium is called the waveguide phenomenon, for instance the high-index guided (ITO/organic) mode. The easiest way to alleviate this effect is to reduce the thickness of those layers (e.g. ITO and organic layers); however, this is not practical. Gu et al. used a shaped substrate (ITO) to redirect the edge light to the normal direction, which increases EQE by a factor of 1.9 ± 0.2 over similar OLEDs fabricated with flat glass substrates.134 Yahiro and Tsutsui formed an extremely low refractive index silica aerogel layer (n = 1.03) between ITO and glass substrate, demonstrating an 80 % improvement.135 Peng et al. inserted anodic aluminum oxide (AAO) nanoporous films between ITO transparent electrode and glass substrate and reported successfully extracting light from glass substrate with an apparent improvement (50 %).136 Lee et al. introduced a two-dimensional photonic crystal (PC) pattern into the glass substrate of an OLED.137 An increase in the extraction efficiency of over 50 % was successfully achieved because the periodic modulation converts the guided waves into external leaky waves. Matterson et al. demonstrated that Bragg scattering induced by a corrugated substrate can be used to increase the efficiency and control the spectrum and polarization of light from a polymer OLED.138 2. Avoid TIR at the interface of air and substrate. Increasing the surface roughness of the substrate could help the substrate mode to be coupled out. Yamasaki et al. fabricated an OLED with a structure consisting of scattering medium and monolayer of silica microspheres to couple out the waveguide mode of light in the thin-film emissive devices.139 Shiang and Duggal have given a more detailed discussion about the design of the scattering medium.140 They mixed the different concentrations of high refractive index ZrO2 (d = 0.6 mm) into polydimethylsilicon resin (PDMS, n = 1.42) to get a maximum enhancement of 40 %. Microlens-array films (MAFs) can be used to reduce the incident angle and avoid TIR to increase the external mode. Madigan et al. used microlenses, spherically shaped patterns, on the back of the device substrate. The emission intensity of OLEDs at normal viewing angle and the total external emission efficiency ηext have been increased by factors of 9.6 and 3.0, respectively.141 However, Möller and Forrest discovered that attached microstructures may also result in image blur,142 and Lin et al. used a micropyramid for brightness enhancing applications.143
Homework problems 7.1 Is it possible to fabricate a single-layer OLED? Why? If it is possible, what are the criteria for the organic material and the electrodes? 7.2 Assume there is a system with one host and two guest materials. One guest gives singlet blue emission, and the other triplet yellow emission. Sketch the possible energy transfer and the restrictions from the host to the guests to achieve 100 % IQE. 7.3 Derive the approximate SCLC equation as follows: SCLC: J = με(V 2 /L 3 ); where J is the current density, μ the carrier mobility, ε the static dielectric constant, V the driving voltage and L the device thickness. Current conduction: J = ρv; where ρ is the injected free charge concentration and v the drift velocity. Capacitance: Q = CV ; where Q is the total injected free charge and C the capacitance. (You may need Q = ρAL, v = μ(V /L), A = device cross-section.) (continued)
226
Introduction to Flat Panel Displays
7.4 Derive the approximate TCLC equation as follows: TCLC: J = εμNc(ε/ε N 0 kT t )l (V (l+1) /L (2l+1) ); where J is the current density, μ the carrier mobility, ε the static dielectric constant, V the driving voltage, L the device thickness and l = T t /T. (Assuming the free and trap carrier concentrations are defined as follows. Free carrier concentration distribution: n = N c exp[(F − E c )/kT ]; trap carrier concentration distribution: nt = kT t N 0 exp[(F − E c )/kT t ]; N c : effective density of states in the conduction band; F: quasi Fermi level; E c : bottom edge of the conduction band energy; k: Boltzmann constant; T : temperature in kelvin; T t : characteristic temperature.) Current conduction: J = ρv; where ρ is the injected free charge concentration and v the drift velocity. Capacitance: Q = CV ; where Q is the total injected free charge and C the capacitance. (You may need Qt = ρt AL, v = μ(V /L), n = ρ/ε, nt = ρt /ε, A = device cross-sectional area.) 7.5 For a planar structure, does a top- or bottom-emission OLED exhibit higher EQE? Why?
References 1. Sze, S.M. (2001) Semiconductor Devices: Physics and Technology, 2nd edn, John Wiley & Sons, Ltd. 2. Baldo, M.A. and Forrest, S.R. (2001) Interface-limited injection in amorphous organic semiconductors. Phys. Rev. B, 64, 085201. 3. Pope, M., Kallmann, H. and Magnante, P. (1963) Electroluminescence in organic crystals. J. Chem. Phys., 38, 2042. 4. Tang, C.W. and Vanslyke, S.A. (1987) Organic electroluminescent diodes. Appl. Phys. Lett., 51, 913. 5. Burroughes, J.H., Bradley, D.D.C., Brown, A.R. et al. (1990) Light-emitting diodes based on conjugated polymers. Nature, 347, 539. 6. Kagan, J. (1993) Organic Photochemistry: Principles and Applications, Academic Press, New York. 7. Gierschner, J., Mack, H.G., Luer, L. and Oelkrug, D. (2002) Fluorescence and absorption spectra of oligophenylenevinylenes: vibronic coupling, band shapes, and solvatochromism. J. Chem. Phys., 116, 8596. 8. Michl, J. and Bonacic-Koutecky, V. (1990) Electronic Aspects of Organic Photochemistry, John Wiley & Sons, Inc., New York. 9. Tang, C.W., Vanslyke, S.A. and Chen, C.H. (1989) Electroluminescence of doped organic thin films. J. Appl. Phys., 65, 3610. 10. Klessinger, M. and Michl, J. (1995) Excited States and Photochemistry of Organic Molecules, John Wiley & Sons, Inc., New York. 11. Yip, W.T. and Levy, D.H. (1996) Excimer/exciplex formation in van der Waals dimers of aromatic molecules. J. Phys. Chem., 100, 11539. 12. D’Andrade, B.W., Brooks, J., Adamovich, V. et al. (2002) White light emission using triplet excimers in electrophosporescent organic light-emitting devices. Adv. Mater., 14, 1032. 13. Hung, L.S. and Chen, C.H. (2002) Recent progress of molecular organic electroluminescent materials and devices. Mater. Sci. Eng. R, 39, 143. 14. Petersson, G.P., Svensson, C.M. and Maserjian, J. (1975) Resonance effects observed at the onset of Fowler– Nordheim tunneling in thin MOS structures. Solid-State Electron., 18, 449. 15. Pope, M. and Swenberg, C.E. (1999) Electronic Processes in Organic Crystals and Polymers, 2nd edn, Oxford University Press. 16. Gill, W.D. (1972) Drift mobilities in amorphous charge-transfer complexes of trinitrofluorenone and poly-nvinylcarbazole. J. Appl. Phys., 43, 5033. 17. Choi, W.K., Delima, J.J. and Owen, A.E. (1986) Model for the variations in the field-dependent behavior of the Poole–Frenkel effect. Phys. Stat. Sol. (b), 137, 345. 18. Davids, P.S., Campbell, I.H. and Smith, D.L. (1997) Device model for single carrier organic diodes. J. Appl. Phys., 82, 6319. 19. Rusu, G.I. (1993) On the current–voltage characteristics of some thin-film sandwich structures of the metal/organic semiconductor/metal type. Appl. Surf. Sci., 65–66, 381.
Organic light-emitting devices
227
20. Staudigel, J., Stößel, M., Steuber, F. and Simmerer, J. (1999) A quantitative numerical model of multilayer vapor-deposited organic light emitting diodes. J. Appl. Phys., 86, 3895. 21. Parker, I.D. (1994) Carrier tunneling and device characteristics in polymer light-emitting diodes. J. Appl. Phys., 75, 1656. 22. Kao, K.C. and Hwang, W. (1981) Electrical Transport in Solid: With Particular Reference to Organic Semiconductors, Pergamon Press. 23. Brutting, W., Berleb, S. and Muckl, A.G. (2001) Device physics of organic light emitting diodes based on molecular materials. Org. Electron.. 2, 1. 24. Meyer, H., Haarrer, D., Naarmann, H. and Hohold, H.H. (1995) Trap distribution for charge carriers in poly(paraphenylene vinylene) (PPV) and its substituted derivative DPOP-PPV. Phys. Rev. B, 52, 2587. 25. Markov, D.E., Hummelen, J.C., Blom, P.W.M. and Sieval, A.B. (2005) Dynamics of exciton diffusion in poly(p-phenylene vinylene)/fullerene heterostructures. Phys. Rev. B, 72, 045216. 26. Kalinowski, J. (2004) Organic Light-Emitting Diodes: Principles, Characteristics and Processes, Marcel Dekker, New York. 27. Forrest, S.R. (2004) Exciton formation statistics under electrical injection in organic semiconductor thin films. J. Luminescence, 110, 378. 28. Cao, Y., Parker, I.D., Yu, G. et al. (1999) Improved quantum efficiency for electroluminescence in semiconducting polymers. Nature, 397, 414. 29. Adachi, C., Kwong, R. and Forrest, S.R. (2001) Efficient electrophosphorescence using a doped ambipolar conductive molecular organic thin film. Org. Electron., 2, 37. 30. Pfeiffer, M., Forrest, S.R., Zhou, X. and Leo, K. (2003) A low drive voltage, transparent, metal-free n–i–p electrophosphorescent light emitting diode. Org. Electron., 4, 21. 31. Kawamura, Y., Goushi, K., Brooks, J. et al. (2005) 100% phosphorescence quantum efficiency of Ir(III) complexes in organic semiconductor films. Appl. Phys. Lett., 86, 071104. 32. Greenham, N.C., Friend, R.H. and Bradley, D.D.C. (1994) Angular dependence of the emission from a conjugated polymer light-emitting diode: implications for efficiency calculations. Adv. Mater., 6, 491. 33. Saleh, B.E.A. and Teich, M.C. (1991) Fundamentals of Photonics, John Wiley & Sons, Inc., New York. 34. Takada, N.,Tsutsui, T. and Saito, S. (1993) Control of emission characteristics in organic thin-film electroluminescent diodes using an optical-microcavity structure. Appl. Phys. Lett., 63, 2032. 35. So, S.K., Choi, W.K., Leung, L.M. and Neyts, K. (2003) Interference effects in bilayer organic light-emitting diodes. Appl. Phys. Lett., 82, 466. 36. Riel, H., Karg, S., Beierlein, T. and Rieß, W. (2003) Tuning the emission characteristics of top-emitting organic light-emitting devices by means of a dielectric capping layer: an experimental and theoretical study. J. Appl. Phys., 94, 5290. 37. Braun, D. and Heeger, A.J. (1991) Visible light emission from semiconducting polymer diodes. Appl. Phys. Lett., 58, 1982. 38. Hsiao, C.H., Lee, J.H. and Tseng, C.A. (2006) Probing recombination-rate distribution in organic light-emitting devices with mixed-emitter structure. Chem. Phys. Lett., 427, 305. 39. Kalinowski, J., Palilis, L.C., Kim, W.H. and Kafafi, Z.H. (2003) Determination of the width of the carrier recombination zone in organic light-emitting diodes. J. Appl. Phys., 94, 7764. 40. Shi, J. and Tang, C.W. (1997) Doped organic electroluminescent devices with improved stability. Appl. Phys. Lett., 70, 1665. 41. Fox, J.L. and Chen, C.H. (1988) Benzopyrano[6,7,8-i,j]quinolizine-11-one lasing dyes and intermediates for their preparation. US Patent 4736032. 42. Chen, C.H., Klubek, K.P. and Shi, J. (1999) Red organic electroluminescent materials. US Patent 5908581. 43. Hosokawa, C., Higashi, H., Nakamura, H. and Kusumoto, T. (1995) Highly efficient blue electroluminescence from a distyrylarylene emitting layer with a new dopant. Appl. Phys. Lett., 67, 3853. 44. Shirota, Y., Kuwabara, Y., Inada, H. et al. (1994) Multilayered organic electroluminescent device using a novel starburst molecule, 4,4 ,4 -tris(3-methylphenylphenylamino)triphenyamine, as a hole transport material. Appl. Phys. Lett., 65, 807. 45. Wu, C.C., Wu, C.I., Sturm, J.C. and Kahn, A. (1997) Surface modification of indium tin oxide by plasma treatment: an effective method to improve the efficiency, brightness, and reliability of organic light emitting devices. Appl. Phys. Lett., 70, 1348. 46. Tadayyon, S.M., Grandin, H.M., Griffiths, K. et al. (2004) CuPc buffer layer role in OLED performance: a study of the interfacial band energies. Org. Electron., 5, 157.
228
Introduction to Flat Panel Displays
47. Nuesch, F., Rothberg, L.J., Forsythe, E.W. et al. (1999) A photoelectron spectroscopy study on the indium tin oxide treatment by acids and bases. Appl. Phys. Lett., 74, 880. 48. Hung, L.S., Tang, C.W. and Mason, M.G. (1997) Enhanced electron injection in organic electroluminescence devices using an Al/LiF electrode. Appl. Phys. Lett., 70, 152. 49. Shaheen, S.E., Jabbour, G.E., Morrell, M.M. et al. (1998) Bright blue organic light-emitting diode with improved color purity using a LiF/Al cathode. J. Appl. Phys., 84, 2324. 50. Hung, L.S., Tang, C.W. and Mason, M.G. (1997) Enhanced electron injection in organic electroluminescence devices using an Al/LiF electrode. Appl. Phys. Lett., 70, 152. 51. 1 X. Zhou, M. Pfeiffer, J. Blochwitz, A. Werner, A. Nollau, T. Fritz, and K. Leo. Very-low-operating-voltage organic light-emitting diodes using a p-doped amorphous hole injection layer. Appl. Phys. Lett. 78, 410 (2001). 52. Endo, J., Matsumoto, T. and Kido, J. (2002) Organic electroluminescent devices with a vacuum-deposited Lewis-acid-doped hole-injecting layer. Jpn. J. Appl. Phys., 41, L358. 53. Pfeiffer, M., Leo, K., Zhou, X. et al. (2003) Highly efficient organic light emitting diodes by doped transport layers. Org. Electron., 4, 89. 54. Lee, J.H., Wu, M.H., Chao, C.C. et al. (2005) High efficiency and long lifetime OLED based on a metal-doped electron transport layer. Chem. Phys. Lett., 416, 234. 55. He, G., Schneider, O., Qin, D. et al. (2004) Very high-efficiency and low voltage phosphorescent organic light-emitting diodes based on a p–i–n junction. J. Appl. Phys., 95, 5773. 56. D’Andrade, B.W., Forrest, S.R. and Chwang,A.B. (2003) Operational stability of electrophosphorescent devices containing p and n doped transport layers. Appl. Phys. Lett., 83, 3858. 57. Xie, Z., Hung, L.S. and Zhu, F. (2003) A flexible top-emitting organic light-emitting diode on steel foil. Chem. Phys. Lett., 381, 691. 58. Riel, H., Karg, S., Beierlein, T. et al. (2003) Phosphorescent top-emitting organic light-emitting devices with improved light outcoupling. Appl. Phys. Lett., 82, 466. 59. Chen, C.W., Hsieh, P.Y., Chiang, H.H. et al. (2003) Top-emitting organic light-emitting devices using surfacemodified Ag anode. Appl. Phys. Lett., 83, 5127. 60. Han, S., Feng, X. and Lu, Z.H. (2003) Transparent-cathode for top-emission organic light-emitting diodes. Appl. Phys. Lett., 82, 2715. 61. Riel, H., Karg, S., Beierlein, T. and Rieß, W. (2003) Tuning the emission characteristics of top-emitting organic light-emitting devices by means of a dielectric capping layer: an experimental and theoretical study. J. Appl Phys., 94, 5290. 62. Liu, K.C., Teng, C.W., Lu, Y.H. et al. (2007) Improving the performance of transparent PLEDs with LiF/Ag/ITO cathode. Electrochem. Solid State Lett., 10, J120. 63. Ko, C.W., Hu, S.H., Li, S.H. et al. (2005) Development of 1.5-inch full color double sided active matrix OLED with novel arrays design. SID Tech. Dig., 961. 64. Braun, D. and Heeger, A.J. (1991) Visible light emission from semiconducting polymer diodes. Appl. Phys. Lett., 58, 1982. 65. Ohmori, Y., Uchida, M., Muro, K. and Yoshino, K. (1991) Blue electroluminescent diodes utilizing poly(alkylfluorene). Jpn. J. Appl. Phys., 30, L1941. 66. Ho, P.K.H, Kim, J.S., Burroughes, J.H. et al. (2000) Molecular-scale interface engineering for polymer lightemitting diodes. Nature, 404, 481. 67. Chou, M.Y., Leung, M.K., Su, Y.O. et al. (2004) Electropolymerization of starburst triarylamines and their application to electrochromism and electroluminescence. Chem. Mater., 16, 654. 68. Ma, W., Iyer, P.K., Gong, X. et al. (2005) Water/methanol-soluble conjugated copolymer as an electron-transport layer in polymer light-emitting diodes. Adv. Mater., 17, 274. 69. Lee, C.B., Uddin, A., Hu, X. and Andersson, T.G. (2004) Study of Alq3 thermal evaporation rate effects on the OLED. Mater. Sci. Eng. B, 112, 14. 70. Long, M., Grace, J.M., Freenman, D.R. et al. (2006) New capabilities in vacuum thermal evaporation sources for small molecule OLED manufacturing. SID Tech. Dig., 1474. 71. Lee, S.T., Suh, M.C., Kang, M.C. et al. (2007) LITI (laser induced thermal imaging) technology for highresolution and large-sized AMOLED. SID Tech. Dig., 1588. 72. Hirano, T., Matsuo, K., Kohinata, K. et al. (2007) Novel laser transfer technology for manufacturing large-sized OLED displays. SID Tech. Dig., 1592. 73. Iino, S. and Miyashita, S. (2006) Printable OLEDs promise for future TV market. SID Tech. Dig., 1463.
Organic light-emitting devices
229
74. Kwon, S.H., Paik, S.Y., Kwon, O.J. and Yoo, J.S. (2001) Triple-layer passivation for longevity of polymer light-emitting diodes. Appl Phys Lett., 79, 4450. 75. Lewis, J.S. and Weaver, M.S. (2004) Thin-film permeation-barrier technology for flexible organic light-emitting devices. IEEE J. Sel. Top. Quantum Electron., 10, 45. 76. Kho, S., Cho, D. and Jung, D. (2002) Passivation of organic light-emitting diodes by the plasma polymerized para-xylene thin film. Jpn. J. Appl. Phys., 41, L1336. 77. Kim, G.H., Oh, J., Yang, Y.S. et al. (2004) Lamination process encapsulation for longevity of plastic-based organic light-emitting devices. Thin Solid Films, 467, 1. 78. Lee, J.H., Kim, G.H., Kim, S.H. et al. (2004) Longevity enhancement of organic thin-film transistors by using a facile laminating passivation method. Synth. Met., 143, 21. 79. Weaver, M.S., Michalski, L.A., Rajan, K. et al. (2002) Organic light-emitting devices with extended operating lifetimes on plastic substrates. Appl Phys Lett., 81, 2929. 80. Burrows, P.E., Graff, G.L., Gross, M.E. et al. (2001) Ultra barrier flexible substrates for flat panel displays. Displays, 22, 65 (2001). 81. Lee, C.J., Pode, R.B., Moon, D.G. and Han, J.I. (2004) Realization of an efficient top emission organic lightemitting device with novel electrodes. Thin Solid Films. 467, 201. 82. Fung, M.K., Gao, Z.Q., Lee, C.S. and Lee, S.T. (2001) Inhibition of dark spots growth in organic electroluminescent devices. Chem. Phys. Lett., 333, 432. 83. Aziz, H., Popovic, Z., Tripp, C. et al. (1998) Degradation processes at the cathode/organic interface in organic light emitting devices with Mg:Ag cathodes. Appl. Phys. Lett., 72, 2642. 84. Lim, S.F., Ke, L., Wang, W. and Chua, S.J. (2001) Correlation between dark spot growth and pinhole size in organic light-emitting diodes. Appl. Phys. Lett., 78, 2116. 85. Gardonio, S., Gregoratti, L., Melpignano, P. et al. (2007) Degradation of organic light-emitting diodes under different environment at high drive conditions. Org. Electron., 8, 37. 86. Burrows, P.E., Bulovic, V., Forrest, R. et al. (1994) Reliability and degradation of organic light emitting devices. Appl. Phys. Lett., 65, 2922. 87. Chan, M.Y., Lai, S.L., Wong, F.L. et al. (2003) Efficiency enhancement and retarded dark-spots growth of organic light-emitting devices by high-temperature processing. Chem. Phys. Lett., 371, 700. 88. Kim, Y., Choi, D., Lim, H. and Ha, C.S. (2003) Accelerated pre-oxidation method for healing progressive electrical short in organic light-emitting devices. Appl. Phys. Lett., 82, 2200. 89. Luo, Y., Aziz, H., Popovic, Z.D. and Xu, G. (2006) Correlation between electroluminescence efficiency and stability in organic light-emitting devices under pulsed driving conditions. J. Appl. Phys., 99, 054508. 90. Kitamura, M., Imada, T. and Arakawa, Y. (2003) Organic light-emitting diodes driven by pentacene-based thin-film transistors.Appl. Phys. Lett., 83, 3410. 91. Hosokawa, C., Eida, M., Matsuura, M. et al. (1997) Organic multi-color electroluminescence display with fine pixels. Synth. Met., 91, 3. 92. Zou, D.C., Yahiro, M. and Tsutsui, T. (1998) Spontaneous and reverse-bias induced recovery behavior in organic electroluminescent diodes. Appl. Phys. Lett., 72, 2484. 93. Cao, Y., Yu, G., Parker, I.D. and Heeger, A.J. (2000) Ultrathin layer alkaline earth metals as stable electroninjecting electrodes for polymer light emitting diodes. J. Appl. Phys., 88, 3618. 94. Ni, S.Y., Wang, X.R., Wu, Y.Z. et al. (2004) Decay mechanisms of a blue organic light emitting diode. Appl. Phys. Lett., 85, 878. 95. Aziz, H., Popovic, Z.D., Hu, N.X. et al. (1999) Degradation mechanism of small molecule-based organic light-emitting devices. Science, 283, 1900. 96. Popovic, Z.D. and Aziz, H. (2002) Reliability and degradation of small molecule-based organic light-emitting devices (OLEDs). IEEE J. Sel. Top. Quantum Electron., 8, 362. 97. Lee, J.H., Huang, J.J., Liao, C.C. et al. (2005) Operation lifetimes of organic light-emitting devices with different layer structures. Chem. Phys. Lett., 402, 335. 98. Tsai, C.H., Liao, C.H., Lee, M.T. and Chen, C.H. (2005) Highly stable organic light-emitting devices with a uniformly mixed hole transport layer. Appl. Phys. Lett., 87, 243505. 99. Mori, T., Mitsuoka, T., Ishii, M. et al. (2002) Improving the thermal stability of organic light-emitting diodes by using a modified phthalocyanine layer. Appl. Phys. Lett., 80, 3895. 100. Féry, C., Racine, B., Vaufrey, D. et al. (2005) Physical mechanism responsible for the stretched exponential decay behavior of aging organic light-emitting diodes. Appl. Phys. Lett., 87, 213502.
230
Introduction to Flat Panel Displays
101. Parker, I.D., Cao, Y. and Yang, C.Y. (1999) Lifetime and degradation effects in polymer light-emitting diodes. J. Appl. Phys., 85, 2441. 102. Baldo, M.A., O’Brien, D.D., Thompson, M.E. and Forrest, S.R. (1999) Excitonic singlet–triplet ratio in a semiconducting organic film. Phys. Rev. B, 60, 14422. 103. Baldo, M.A., Lamansky, S., Burrow, P.E. et al. (1999) Very high-efficiency green organic light-emitting devices based on electrophosphorescence. Appl. Phys. Lett., 75, 4. 104. Kwong, R.C., Nugent, M.R., Michalski, L. et al. (2002) High operational stability of electrophosphorescent devices. Appl. Phys. Lett., 81, 162. 105. He, G., Schneider, O., Qin, D. et al. (2004) Very high-efficiency and low voltage phosphorescent organic light-emitting diodes based on a p–i–n junction. J. Appl. Phys., 95, 5773. 106. He, G., Pfeiffer, M., Leo, K. et al. (2004) High-efficiency and low-voltage p–i–n electrophosphorescent organic light-emitting diodes with double-emission layers. Appl. Phys. Lett., 85, 3911. 107. Baldo, M.A., O’Brien, D.F., You, Y. et al. (1998) Highly efficient phosphorescent emission from organic electroluminescent devices. Nature, 395, 151. 108. O’Brien, D.F., Baldo, M.A., Thompson, M.E. and Forrest, S.R. (1999) Improved energy transfer in electrophosphorescent devices. Appl. Phys. Lett., 74, 442. 109. Adachi, C., Baldo, M.A., Thompson, M.E. et al. (2001) High-efficiency red electrophosphorescence devices. Appl. Phys. Lett., 78, 1622. 110. Jiang, X., Jen, A. K.-Y., Carlson, B. and Dalton, L.R. (2002) Red electrophosphorescence from osmium complexes. Appl. Phys. Lett., 80, 713. 111. Song, Y.H., Yeh, S.J., Chen, C.T. et al. (2004) Bright and efficient, non-doped, phosphorescent organic redlight-emitting diodes. Adv. Func. Mater., 14, 1221. 112. Li, C.L., Su, Y.J., Tao, Y.T. et al. (2005) Yellow and red electrophosphors based on linkage isomers of phenylisoquinolinyliridium complexes: distinct difference in photophysical and electroluminescence properties. Adv. Func. Mater., 15, 387. 113. Adachi, C., Kwong, R.C., Djurovish, P. et al. (2001) Endothermic energy transfer: a mechanism for generating very efficient high-energy phosphorescent emission in organic materials. Appl. Phys. Lett., 79, 2082. 114. Matsumoto, T., Nakada, T., Endo, J. et al. (2003) Multiphoton organic EL device having charge generation layer. SID Dig., 979. 115. Liao, L.S., Klubek, K.P. and Tang, C.W. (2004) High-efficiency tandem organic light-emitting diodes. Appl. Phys. Lett., 84, 167. 116. Yang, R.Q. and Qiu, Y. (2003) Bipolar cascade lasers with quantum well tunnel junctions. J. Appl. Phys., 94, 7370. 117. Guo, X., Shen, G.D., Wang, G.H. et al. (2001) Tunnel-regenerated multiple-active-region light-emitting diodes with high efficiency. Appl. Phys. Lett., 79, 2985. 118. Guo, X., Shen, G.D., Ji, Y. et al. (2003) Thermal property of tunnel-regenerated multiactive-region light-emitting diodes. Appl. Phys. Lett., 82, 4417. 119. Gu, G., Parthasarathy, G., Burrows, P.E. et al. (1999) Transparent stacked organic light emitting devices: I. Design principles and transparent compound electrodes. J. Appl. Phys., 86, 4067. 120. Gu, G., Parthasarathy, G., Tian, P. et al. (1999) Transparent stacked organic light emitting devices: II. Device performance and applications to displays. J. Appl. Phys., 86, 4076. 121. Kim, J.K., Hall, E., Sjölund, O. and Coldren, L.A. (1999) Epitaxially stacked multiple-active-region 1.55 m lasers for increased differential efficiency. Appl. Phys. Lett., 74, 3251. 122. Korshak,A.N., Gribnikov, Z.S. and Mitin, V.V. (1998) Tunnel-junction-connected distributed-feedback verticalcavity surface-emitting laser. Appl. Phys. Lett., 73, 1475. 123. D’Andrade, B.W. and Forrest, S.R. (2004) White organic light-emitting devices for solid-state lighting. Adv. Mater., 16, 1585. 124. Cheng, G., Zhao, Y., Zhang, Y. et al. (2004) White organic light-emitting devices using 2,5,2 ,5 -tetrakis(4 biphenylenevinyl)-biphenyl as blue light-emitting layer. Appl. Phys. Lett., 84, 4457. 125. Liu, T.-H., Wu, Y.-S., Lee, M.-T. et al. (2004) Highly efficient yellow and white organic electroluminescent devices doped with 2,8-di(t-butyl)-5,11-di[4-(t-butyl)phenyl]-6,12-diphenylnaphthacene. Appl. Phys. Lett., 85, 4304. 126. Jou, J.-H., Chiu, Y.-S., Wang, R.-Y. et al. (2006) Efficient, color-stable fluorescent white organic light-emitting diodes with an effective exciton-confining device architecture. Org. Electron., 7, 8.
Organic light-emitting devices
231
127. Chang, C.-C., Chen, J.-F., Hwang, S.-W. and Chen, C.H. (2005) Highly efficient white organic electroluminescent devices based on tandem architecture. Appl. Phys. Lett., 87, 253501. 128. Sun, Y., Giebink, N.C., Kanno, H. et al. (2006) Management of singlet and triplet excitons for efficient white organic light-emitting devices. Nature, 440, 908. 129. Krummacher, B.C., Choong, V.-E., Mathai, M.K. et al. (2006) Highly efficient white organic light-emitting diode. Appl. Phys. Lett., 88, 113506. 130. Do, Y.R., Kim, Y.-C., Song, Y.-W. et al. (2004) Enhanced light extraction efficiency from organic light emitting diodes by insertion of a two-dimensional photonic crystal structure. J. Appl. Phys., 96, 7629. 131. Gu, G. and Forrest, S.R. (1998) Design of flat-panel displays based on organic light-emitting devices. IEEE J. Sel. Top. Quantum Electron., 4, 83. 132. Moon, D.G., Pode, R.B., Lee, C.J. and Han, J.I. (2004) Transient electrophosphorescence in red top-emitting organic light-emitting devices. Appl. Phys. Lett., 85, 4771. 133. Nakamura, T., Tsutsumi, N., Juni, N. and Fujii, H. (2004) Improvement of coupling-out efficiency in organic electroluminescent devices by addition of a diffusive layer. J. Appl. Phys., 96, 6016. 134. Gu, G., Garbuzov, D.Z., Burrows, P.E. et al. (1997) High-external-quantum-efficiency organic light-emitting devices. Opt. Lett., 22, 396. 135. Yahiro, M. and Tsutsui, T. (2001) Influence of device configuration on external quantum efficiency in organic light-emitting devices. Proc. MRS, 660, 5. 136. Peng, H.J., Ho, Y.L., Yu, X.J. and Kwok, H.S. (2004) Enhanced coupling of light from organic light emitting diodes using nanoporous films. J. Appl. Phys., 96, 1649. 137. Lee, Y.-J., Kim, S.-H., Huh, J. et al. (2003) A high-extraction-efficiency nanopatterned organic light-emitting diode. Appl. Phys. Lett., 82, 3779. 138. Matterson, B.J., Lupton, L.M., Safonov, A.F. et al. (2001) Increased efficiency and controlled light output from a microstructured light-emitting diode. Adv. Mater., 13, 123. 139. Yamasaki, T., Sumioka, K. and Tsutsui, T. (2000) Organic light-emitting device with an ordered monolayer of silica microspheres as a scattering medium. Appl. Phys. Lett., 76, 1243. 140. Shiang, J.J. and Duggal, A.R. (2004) Application of radiative transport theory to light extraction from organic light emitting diodes. J. Appl. Phys., 95, 2880. 141. Madigan, C.F., Lu, M.H. and Sturm, J.C. (2000) Improvement of output coupling efficiency of organic lightemitting diodes by backside substrate modification. Appl. Phys. Lett., 76, 1650. 142. Möller, S. and Forrest, S.R. (2002) Improved light out-coupling in organic light emitting diodes employing ordered microlens arrays. J. Appl. Phys., 91, 3324. 143. Lin, L., Shia, T.K. and Chiu, C.J. (2000) Silicon-processed plastic micropyramids for brightness enhancement applications. J. Micromech. Microeng., 10, 395.
8 Field emission displays 8.1 Introduction A field emission display (FED) has a simple structure and a high luminance efficiency. This type of display does not need a backlight, color filter, polarizer or the other optical films that are required in a liquid crystal display (LCD). Therefore, the structure of a FED is simpler than that of a LCD. Additionally, FEDs have a shorter response time, a wider viewing angle and a larger temperature range than LCDs.1 They can display static pictures and motion pictures, in cold ambient and hot ambient, for personal use and public use.2 The structure of a FED is similar to that of a cathode ray tube (CRT). Both FEDs and CRTs use phosphors to produce brightness and depend on a vacuum to maintain the lifetime of electron emission. The operating mechanism of FEDs involves field emission electrons to excite a phosphor and generate luminance. Field emission uses a high electric field rather than a thermionic approach to extract electrons in a vacuum.3
8.2 Physics of field emission Electrons may be emitted from a solid surface into a vacuum by many mechanisms, including thermionic emission, photoemission and field emission.4, 5 Thermionically emitted electrons are thermally excited over the potential energy barrier while photoemitted electrons are excited over the potential energy barrier by incoming photons. In field emission, electrons tunnel though the surface potential energy barrier, which has been thinned and shaped by a strong electric field. A sharp emitter structure usually offers a strong electric field even when the applied voltage is low. Since the ions generated by the residual gas can damage the emitter, less residual gas and a lower operating voltage are needed. To achieve a low operating voltage, an emitter material with a low work function and a sharp structure is required. A vacuum is also required to eliminate residual gas and thus reduce emitter damage which, in turn, increases emitter lifetime.
8.2.1 Work function and field enhancement For a metal surface without an applied electric field, the surface potential energy E for a typical Fermi level E F and work function Ψ is as presented in Figure 8.1,6 where Z is the distance from the metal surface and the surface barrier is the sum of E F and Ψ . This diagram is for electrons at a metal surface in the absence of an electric field.
Introduction to Flat Panel Displays c 2008 John Wiley & Sons, Ltd
J.-H. Lee, D.N. Liu and S.-T. Wu
234
Introduction to Flat Panel Displays
Surface barrier
Energy E (ev)
Work function ψ
Surface barrier due to image charge effect
Fermi level EF Metal
0 Figure 8.1
Distance Z (Å)
Surface potential energy diagram in the absence of applied electric field.
Energy E (ev) Surface barrier lowered by applied electric field with contribution from image charge effect
Work function ψ
Em Fermi level EF
Metal
0
Zm
Distance Z (Å)
Figure 8.2 Surface potential energy diagram with applied electric field.
The straight line of the surface barrier is smoothed by the image charge effect. However, the maximum height of the barrier is as high as the sum of the Fermi level E F and the work function Ψ . An applied electric field bends the surface barrier line, as shown in Figure 8.2, where E m is the maximum surface barrier. Z m is the distance from the metal surface to the position the maximum of the surface barrier. This shaping of the potential energy is the result of the image potential energy and a strong electric field. As the applied electric field increases, the surface barrier is bent more and becomes thinner. When the applied electric field is as high as 107 V cm−1 , the surface barrier is sufficiently thin to enable electrons to tunnel through the barrier.
Field emission displays
235
Since the image potential energy E i is equal to −e2 /4πε4Z, where ε is the permittivity in a vacuum, the surface potential energy E(z) that is seen by an electron on the vacuum side of the metal–vacuum interface is asymptotically given by E(z) = EF + Ψ −
e2 , 4πε4Z
(8.1)
where E(z), E F and Ψ are in eV. When an electric field F is applied to the surface, the surface potential energy that is seen by an electron becomes E(z) = EF + Ψ −
e2 − eFz 4πε4Z
(8.2)
and the maximum surface barrier E m is given by Em = EF + Ψ −
e3 F 4πε
(8.3)
or √ Em = EF + Ψ − 3.79 × 10−5 F eV, where F is in V m−1 . The surface barrier is maximum at Z m where e Zm = 4πε4F
(8.4)
(8.5)
or √ Zm = 1.9 × 10−5 Fm.
(8.6)
Several equations have been used to describe field emission. One equation, derived by Fowler and Nordheim in 1928,7 ignores the lowering of the barrier by the image effect.8, 9 This form of the Fowler– Nordheim equation is J(F) =
√ √ 6.2 × 10−6 EF 2 −6.8 × 107 Ψ 3 . √ F exp F (Ψ + EF ) Ψ
(8.7)
where current density J is in A cm−2 , electric field F is in V cm−1 and Fermi level E F and work function Ψ are in eV. Another equation considers the lowering of the barrier by the image effect.10, 11 This modified Fowler–Nordheim equation is typically presented in the form √ 1.54 × 10−6 2 −6.8 × 107 Ψ 3 v(y) J(F) = F exp . Ψ t 2 (y) F
(8.8)
where emission current density J is in A cm−2 , electric field F is in V cm−1 , work function Ψ is in eV and the Schottky lowering of the work function barrier is given by √ y = 3.79 × 10−4 F/Ψ .
(8.9)
The functions v(y) and t(y) have been computed, and are presented in Figure 8.3.12 The approximation t 2 (y) = 1.1 and v(y) = 0.95− y2 can be applied to Equation (8.8). Equations (8.7) and (8.8) clearly reveal that the electric field dominates the current density. Restated, a strong electric field is required to yield a high current density.
236
Introduction to Flat Panel Displays
V (y); t2 (y) 1.5 t2(y) ~ 1.1 t2(y)
1.0 v(y) v(y) ~ 0.95-y2 0.5
0
0.0
0.5
1.0 y
Figure 8.3 Comparison of approximate forms with exact solutions of the Fowler–Nordheim field emission functions v(y) and t 2 (y).
In addition to field emission, thermionic emission can become important as the temperature rises. In thermionic emission, the metal is heated to excite electrons and to move them over a potential energy barrier. The equation of thermionic emission is J = 120T 2 e−Ψ/kT ,
(8.10)
where emission density J is in A cm−2 , absolute temperature T is in K, work function Ψ is in eV and the Boltzmann constant k is 8.625 × 10−5 eV K−1 .13 For thermionic emission, temperature dominates the current density; in field emission, the electric field dominates the current density. Figure 8.4 shows the range of thermionic current emission and field emission for an emitter with a work function of 4.5 eV,14, 15 indicating that, in a strong enough field, most of the emission is field emission. However, at sufficiently high temperature and low electric field, the emission is mainly thermionic emission.
8.2.2 Vacuum mechanism A vacuum is required to reduce the amount of residual gas. Meanwhile, the probability of undesired ionization is reduced as the distance from the cathode to the anode decreases to less than the electron mean free path (λc ). The equation for the electron mean free path is λc =
T cm, 273pPc (v)
(8.11)
where Pc (v) is the average number of collisions made by an electron as it travels through 1 cm in a gas at a pressure of 1 torr at 0 ◦ C, v is the potential applied to generate the electron velocity, T is the absolute temperature in kelvin and p is the pressure in torr.16 Since Pc (v) is typically about 70, p must be less than 3.14 torr for T = 300◦ K and d = 50 m. However, the lifetime of the device increases with
Field emission displays
237
Temp. (1000°C) 4 Thermionic emission 3
2
1
Field emission
0
6
10 Field (×107 v/cm)
Figure 8.4 Ranges of thermionic emission and field emission for an emitter with a work function of 4.5 eV.
a higher vacuum since ion bombardment decreases. The typical vacuum in a field emission device is in the region of 10−7 torr.
8.3 FED structure and display mechanism A FED panel comprises a field emission array (FEA) plate and a phosphor plate. The FEA plate is a structure that generates field emission.17 Additionally, gated field emission is typically required to modulate the emission of electrons.18 Adding a third electrode (gate) between the anode and the cathode causes gated field emission. The emitter is on top of the cathode. The gate is typically much closer to the emitter than to the anode to control (or modulate) the electron emission. This emission current is extracted from the emitter and is a function of the voltage between the gate and the cathode. The voltage between the anode and the gate governs the magnitude of the emission current flowing to the anode and to the gate. A structure in which the gate is placed above the cathode is called a vertical structure. Figure 8.5 shows the vertical structure of a gated conic emitter in a FED.19 In this figure, r is the radius of the emitter, which is typically hundreds of angstroms; d is the diameter of the gate opening, which is typically a few or a few tens of micrometers; h is the height of the emitter, which is typically a few or a few tens of micrometers; x is the tip elevation above the top edge of the gate, which is typically less than 1 m; S ag is the distance between the anode and the gate, which is typically from a few tens of micrometers to a few millimeters; V ge is the voltage between the gate and the emitter; and V ag is the voltage between the anode and the gate. Notably, the electric field F = f (r, d, h, s, V ge ) and the emission current density is J = f (F). In Figure 8.5, electrons are emitted from the emitter and vertically excite RGB phosphors. The phosphors used in the field emission device are of the electron-excited type. However, the gate and the cathode can also be arranged in a horizontal configuration, as presented in Figure 8.6. The gate is placed at the same elevation as the cathode. Since the emitted electron must fly over the gate to reach the anode, the gate receives more current than the anode. Modeling experiments show that the anode of the vertical structure collects more current than the anode of the horizontal structure under the same operating conditions. This is a shortcoming for a horizontal gated emitter, although this type of gated emitter has a simple structure and process steps.
238
Introduction to Flat Panel Displays
Light Phosphor plate
Substrate
Phosphor
Anode BM
Vag
e–
Sag
Vge h
FEA plate
e–
Spacer
e– Emitter
d x
Gate Dielectric Cathode Substrate
Figure 8.5 Vertical structure of gated conic emitter in a FED.
Light
Spacer
Substrate
Phosphor
Phosphor plate
Anode BM
e–
e–
e– Gate
Emitter Cathode
FEA plate
Substrate Figure 8.6 Horizontal gated emitter in a FED.
8.4 Emitter The emitter has an important role in FEDs. Field emitter structures can be conic, wedge-shaped or tubular.20 The emitting region of a conic emitter is its tip, while those of the other structures are edges. The many emitters include the Spindt emitter, the carbon nanotube (CNT) emitter and the surface conduction emitter (SCE).21 The Spindt emitter is a sharp cone while the CNT emitter has nanometric diameter carbon tubes. The SCE uses a material called PdO with a nanometric gap structure to generate surface electrons. Table 8.1 compares these emitters. These forms of field emission need a high vacuum of Table 8.1
Comparison of Spindt emitter, CNT emitter and SCE.
Item
Spindt
CNT
SCE
Feature Operating voltage (V) Major process Key issue
Sharp conic structure Few tens Conic emitter formation Large area of evaporation
Nano-diameter of carbon tube Few tens to hundreds CNT formation CNT activation or CNT growth rate
Nano-gap of slit Few tens Nano-gap formation High gate drawn current
Field emission displays
239
typically 10−7 torr. The operating voltages are typically from a few tens to a few hundreds of volts. Since a higher operating voltage of driver integrated circuits (ICs) is more expensive, higher operating voltage indicates higher driver costs. Therefore, the driver costs for Spindt emitters and SCEs are low due to their relatively low operating voltages. The major processes of formation of the Spindt emitter, the CNT emitter and the SCE are conic formation, CNT formation and nano-gap formation, respectively. The key difficulties in the formation of the Spindt emitter, the CNT emitter and the SCE are the large area of evaporation, the challenge of CNT activation and the high gate drawn current, respectively. The high gate drawn current of the SCE is due to its horizontal gated structure. This horizontal gated structure is helpful to make a nano-gap slit between the cathode and the gate. In spite of the convenience to make a nano-gap slit for the horizontal gated structure, the emitted electron of SCE for this structure must fly over the gate and results in high gate drawn current. The effective current applied to the anode is therefore reduced and results in lower current efficiency of the SCE. Compared with the SCE, the Spindt and CNT emitters using vertical gated structures have lower gate drawn currents and higher anode current. In other words, Spindt and CNT emitters have lower power consumption for the same operating voltage. Moreover, the emitter elevation also affects the gate drawn current for the vertical gated structure. When the emitter elevation is low, the gate drawn current will be high. For the Spindt emitter, one can control the emitter elevation and the gate drawn current effectively. However, the Spindt emitter suffers from the uniformity difficulty for large-area evaporation. As regards the CNT emitter, the control of emitter elevation is a challenge and the emitter elevation is typically low. Therefore, the CNT emitter usually has higher gate drawn current than the Spindt emitter. In addition, the CNT emitter also suffers from the effective activation difficulty or the low growth rate of the tube. Notably, a smaller value of r of an emitter corresponds to not only a higher emission current density J, but also a smaller emitting area. If the emitter is too sharp, then the emission current I may be decreased because the product of J and A is smaller. The emission currents vary from different tips.22 Restated, one emitter may require a lower emitting voltage while another emitter may require a higher emitting voltage. The variation of the emitting voltage among the emitters results in the problem of uniformity and raises the challenge of modulating the gray levels of the display.
8.4.1 Spindt emitter Many materials, including semiconductors, can be used in Spindt field emitters.23, 24 Ideally, the field emitter should be a material with a high melting point to withstand a large current, a low work function to provide large emission and low vapor pressure to maintain the necessary vacuum in a sealed device.25 An emitter should also be sharp to generate a sufficiently strong electric field for electron emission at low voltage.26 28 Low-voltage operation reduces the probability of dielectric breakdown. Table 8.2 presents the common emitters silicon,28 33 tungsten,34, 35 molybdenum,36 38 LaB6 39, 40 and tantalum41, 42 used in field emission devices, along with some of their properties. Among these emitters, tungsten has the highest melting point and the lowest vapor pressure while silicon has the lowest emitter radius. Since silicon can be formed using standard semiconductor fabrication Table 8.2
Most common Spindt field emitters.
Item
Si
W
Mo
LaB6
Ta
Melting point (◦ C) Work function (eV) Vapor pressure (torr)
1410 4.50 10−6 at 1200 ◦ C <10
3410 4.50 10−11 at 1800 ◦ C <200
2617 4.50 7 × 10−7 at 1800 ◦ C 400
>1500 2.66 —
2996 4.25 5 × 10−10 at 1800 ◦ C <200
Reported emitter radius (Å)
—
240
Introduction to Flat Panel Displays
Gate Cathode
Dielectric Cathode, dielectric and gate patterning
Axial rotation Directional Sacrificial evaporation layer Directional evaporation with axial rotation to form sacrificial layer
Vertical evaporation Axis rotation
Vertical evaporation with axis
Sacrificial layer removal
Figure 8.7 Typical process flow for a conic emitter.
technologies into sharp tips, it has been studied and used widely as a field emitter despite its lower melting point and higher vapor pressure than other materials such as tungsten, molybdenum and tantalum.43 Field emitters must be sharp since the electric field varies with sharpness and electron emission depends strongly on the electric field.44 Sharp emitters also enable electron emission devices to operate at low voltage.45 Figure 8.7 shows a typical process flow to form a conic emitter.46 At the beginning of the process, the cathode, the dielectric and the gate are patterned. Then, directional evaporation with axial rotation is performed to form a sacrificial layer. After the sacrificial layer is formed, vertical evaporation with axial rotation is performed to form a conic emitter. The final step in this process is to remove the sacrificial layer to form a gated field emitter.
8.4.2 CNT emitter Spindt-type (sharp-cone-type) emitters have been widely studied in the last few decades. However, the evaporation used in the Spindt emitter process for forming displays with large areas is difficult. Restated, this is the main difficulty that maintains the uniformity of the sharp-conic emitter when a larger display is desired. Accordingly, the use of a CNT as an emitter is an alternative approach of forming FEDs.47 49 CNT is a carbon material with a nanosized structure. Since it is tubular with a diameter of a few tens or hundreds of nanometers, it is naturally sharp and can act as a field emitter.
Field emission displays
241
Cathode patterning Cathode patterning Gate
Dielectric
Photoresist Dielectric, gate and photoresist deposition
UV exposure and photoresist developing
Gate and dielectric etching
Photoresist removing CNT CNT coating and activating
Figure 8.8 Typical indirect approach for forming CNT emitters.
Indirect and direct approaches can be adopted to form CNT emitters. In an indirect approach, the CNT is grown by arcing and then broken up into a powder. A solvent is added to this powder of CNT to form a paste.50 A CNT emitter is then formed by screen printing the CNT paste onto a substrate and then activating it.51 Figure 8.8 shows the typical process, in which the maximum process temperature is close to about 450 ◦ C.52 At the beginning of this process, the cathode, the dielectric and the gate are patterned. The following process is to coat the CNT paste onto the substrate and then activate the CNT, while the CNT paste is prepared in a separated process. Since the CNT is formed by high-temperature arcing, it has high purity. However, the uniformity of the CNT emitter cannot be effectively controlled since pulverizing and screen printing randomly redistribute the CNT.53 Accordingly, the direct approach to the formation of a CNT emitter is adopted to solve the uniformity problem. Two methods, chemical vapor deposition (CVD) and electrophoretic deposition (EPD), are used to form the CNT emitter.54 56 Figure 8.9 shows the typical process of CVD for forming CNT emitters. The cathode, dielectric and gate are patterned at the beginning of this process. The following process is the deposition of a catalyst. A typical catalyst is Fe/Ni/Co and the process temperature is about 600 ◦ C. The process temperature can be reduced by using the plasma CVD process. The reaction for forming the CNT is C2 H2 → 2C + H2 .
(8.12)
Although the CVD process yields a more uniform CNT, the growth rate is a little slow. Accordingly, EPD can be adopted as an alternative approach for growing CNT. Figure 8.10 shows a deposition mechanism
242
Introduction to Flat Panel Displays
Cathode patterning Cathode patterning Dielectric
Gate
Photoresist Dielectric, gate and photoresist deposition
UV exposure and development of photoresist
Gate and dielectric etching; Removal of photoresist Catalyst Catalyst deposition CNT CNT growth Figure 8.9 Typical CVD process for forming CNT emitters.
DC power
Substrate Electrode Solution Charged CNT Electrode
Figure 8.10 Typical deposition mechanism in the EPD process.
of the EPD process. In this deposition mechanism, the substrate is placed in an electrophoresis solution. The CNTs in the solution are charged and then move to the electrode. Following the motion, the CNTs are deposited onto the electrode. CNTs can be grown directly onto the substrate using either CVD or EPD. Although the uniformity obtained using the CVD or EPD approach exceeds that obtained using the arcing method, the purity obtained is lower.
Field emission displays
243
8.4.3 Surface conduction emitter In a SCE device, a palladium oxide (PdO) film is used as an emitter and a nano-gap is used to extract the electrons from the surface of he PdO film.57 These electrons are called surface conduction electrons. The structure of the SCE display differs a little from the display structure using the Spindt-type emitter and the CNT emitter. Figure 8.11 shows a typical structure of a SCE display, which is commonly called a surface electron display (SED). In this figure, the electron is horizontally extracted from the emitter to the gate. An anode voltage is applied to collect the emitted electron. The gap between the gate and the cathode is about 10 nm. The cathode and the gate are made of platinum. The emitter is PdO. The platinum film is patterned by photolithography while the PdO film of the emitter is deposited by ink-jet printing. Figure 8.12 shows a typical process flow for a SCE. At Light
Spacer
Substrate
Phosphor
Phosphor plate
Anode BM
e–
e–
e– Gate
Emitter Cathode
FEA plate
Substrate
Figure 8.11 Typical structure of a SCE display (surface electron display).
Gate Cathode
Gate and cathode patterning
Substrate
Emitter
Emitter formation
Slit
Slit formation
Figure 8.12 Typical process flow for a SCE.
244
Introduction to Flat Panel Displays
the beginning of this process, the gate and cathode are patterned. The following step is to coat a layer of emitter material. After the coating of the emitter material is completed, a slit is formed with nanometric dimensions. This slit formation is critical since a slit with a nanostructure is required.
8.5 Panel process The panel process involves a FEA plate and a phosphor plate and comprises assembly and aging processes. Figure 8.13 shows a typical process flow. One of the main processes for the FEA plate is the emitter formation process while the major process for the phosphor plate is the phosphor formation process. Assembly and aging processes are associated with panel alignment, vacuum-tight sealing and characteristic stabilization. As stated in a previous section, the three main emitters are the Spindt emitter, the CNT emitter and the SED emitter. Figure 8.14 shows the display process of the Spindt emitter. The major process for the FEA plate was described in the previous section. For the phosphor plate, black matrix (BM) formation is commonly used to increase the contrast ratio of the display. Following BM formation, phosphor formation is performed. Notably, red (R), green (G) and blue (B) phosphors are deposited and patterned in phosphor formation. After the FEA plate and the phosphor plate have been formed, assembly and aging are conducted so that the process of the FED panel is completed. Figure 8.15 shows the display process of a CNT-type emitter. The preceding section described the main process for a FEA plate. The CNT paste is prepared in a separate process, as presented in Figure 8.15. At the beginning of the CNT paste preparation, CNTs are formed and then purified. Following CNT purification, a solvent and a binder are typically added to form the CNT paste. As to the phosphor plate, BM formation is commonly employed to increase the contrast ratio of the display. After BM formation has been completed, R, G and B phosphors are deposited and patterned. After the FEA plate and phosphor plate are formed, the assembly and aging processes are conducted to complete the FED panel.58 Figure 8.16 shows the display process for a SCE. The previous section described the main process for the FEA plate. Additionally, PdO paste is prepared in separate processes, as presented in Figure 8.16. The preparation of the PdO paste begins with formation of PdO powder. Then, a solvent and binder are typically added to form the PdO paste. As to the phosphor plate process, BM formation is commonly adopted to increase the contrast ratio of the display. After BM formation, R, G and B phosphors are deposited and patterned during phosphor formation. Following the formation of the FEA plate and the phosphor plate, the assembly and aging process complete the FED panel.59 Screen printing,60 photolithography and sandblasting are the three major approaches for forming layers in the FED process. Among these major approaches, screen printing is the most commonly used approach for the FED process. This approach can be adopted for the cathode, gate electrode, dielectric, BM and phosphors layers. Figure 8.17 shows the side view of a typical screen-printing process.
FEA plate process
Phosphor plate process
Assembly and aging processes
Figure 8.13 Typical process flow for a FED.
Field emission displays
245
FEA plate
Phosphor plate
Cathode formation
BM formation
Dielectric formation
Phosphor formation
Gate formation
Sacrificial layer evaporation
Emitter formation and sacrificial layer removal
Assembly and aging processes
Figure 8.14 Typical process flow of FED with Spindt emitter.
FEA plate
Phosphor
Cathode formation
BM formation
Dielectric formation
CNT
Gate formation
CNT paste
Emitter formation
Phosphor formation
Assembly and aging processes
Figure 8.15 Typical process flow for FED with CNT emitter.
246
Introduction to Flat Panel Displays
FEA plate
Phosphor plate
PdO powder
Cathode and gate
BM formation
PdO paste
Emitter film formation
Phosphor formation
Emitter slit formation
Assembly and aging processes
Figure 8.16 Typical process flow for FED with SCE.
Squeezee moving
Screen frame
Screen Squeezee Paste Space
Printed pattern
Substrate
Figure 8.17 Side view of screen printing.
The screen mask, the paste and the printing machine are three major elements of screen printing.61 A good paste should easily pass through a screen mask when a shear force is applied to the paste. However, the pattern of the paste should become solid and stand still when no shear force is applied, such that it does not diffuse and high resolution is maintained. After the paste is deposited, it must be dried and fired. Drying removes the solvent: the process temperature is typically under 150 ◦ C. Firing removes the binder and melts the particle at a typical process temperature of more than 400 ◦ C. Figure 8.18 shows the typical process temperature profile with process time. The critical aspect of the drying process is the uniformity of drying from the outer surface to the inner core as well as from the edge to the center of the paste layer. In the firing process, the binder must be fully removed and the accumulated stress must be reduced as much as possible.
Field emission displays
247
Temp.(°C) 600 500
Step Cooling
Firing Debinding
400 300 200
Continuous Cooling
100 0 0.0
0.5
1.0
1.5
2.0 Time (hr.)
Figure 8.18 Typical process temperature profile.
Photolithography is another important approach used in the FED process which is also commonly used in display and semiconductor processes. This process uses a photoresist to form a pattern and subsequent etching of the desired material. Stripping the photoresist leaves the desired pattern.
8.6 Field emission array plate techniques An optimal field emission device must have the following features: (1) a sharp emitter; (2) an emitter with a low work function to enhance electron emission;61 (3) an emitter with a high melting point to withstand the high temperature caused by resistive heating;62 (4) a small gate opening for a gated emitter structure to increase the electric field;63 and (5) a thick dielectric film between the electrodes to maintain the necessary operating voltage without breakdown or significant leakage. Field emission devices include many devices that use an electron source, such as a scanning tunneling microscope.64 The substrate in a FED not only provides a base for the film or paste deposition but also isolates the vacuum. Therefore, substrates with a smooth surface and a robust mechanical strength are required. The additional issue associated with FED substrates is thermal expansion due to the high temperature of the FED process. Sodalime glass is typically used in a FED substrate with a thickness of typically 2.8 mm. Since thermal expansion may produce an error in the image position,65 sodalime glass with a high strain point is used to eliminate the thermal expansion of the substrate and control the error in the image position. As the FED panel is a high-vacuum device, the substrate receives an air pressure of about 1 atm from outside of the panel. Accordingly, deployment of spacers at particular positions of the FEA plate or phosphor plate is demanded. Deformation can occur when the spacer deployment is insufficient. Meanwhile, the glass should also be sufficiently strong to sustain the pressure difference. Additionally, the substrate in the FEA plate typically must have an exhausting hole. During the evacuation, normal air is pumped through this exhausting hole. The floating process is commonly adopted for the formation of sodalime glass, as shown in Figure 8.19. At the beginning of this process, raw material is placed in the melting furnace. Molten glass is formed and then floats on the surface of the molten metal. This molten glass therefore has a smooth surface. Glass is formed upon annealing. The floating process is a traditional process for forming soadlime glass and is cost-effective. An electrode in a FEA plate typically comprises a cathode and a gate electrode. The cathode provides an electrical base for the emitter and the cathode is typically silver paste with a thickness of 10 m. The process for forming the cathode typically is screen printing, as shown in Figure 8.20. The silver paste
248
Introduction to Flat Panel Displays
Input raw materials
Melting furnace Floating bath
Molten glass
Molten metal
Annealing
Glass
Figure 8.19 Typical floating process for forming glass.
Screen mask
Screen printing
Drying
Firing
Figure 8.20 Typical cathode formation using screen printing.
is screen-printed through a patterned screen mask. The pattern is formed on the substrate and then dried and fired. A firm and solid pattern of electrode is therefore formed. The alternative approach for forming a cathode is to deposit photosensitive silver paste on the substrate and then to form a pattern by photolithography. The line width resolution achieved using the photolithographic approach is 20 m, which is better than the 50 m of the screen-printing approach. Screen printing is a commonly used approach for forming a cathode since the material cost is lower and the process steps are simpler.66 The dielectric is the layer between the gate electrode and the cathode. The layer must be a good insulator to reduce leakage current. The dielectric layer is typically formed by screen printing with a thickness of 20 m or using a CVD process with a thickness of 2 m. The thickness of this film is dominated by the need to minimize leakage current. The gate electrode is used to control or to modulate the emission from the cathode, which is typically silver paste with a thickness of 10 m. The cathode is typically formed by screen printing. The alternative approach for forming the gate electrode is to deposit a photosensitive silver paste on the dielectric layer and then to form a pattern photolithographically.
8.7 Phosphor plate techniques The phosphor plate comprises substrate, anode and phosphor layers. Its major function is to generate brightness. The substrate in the phosphor plate is similar to the substrate used in the FEA plate. The anode in the phosphor plate is an electrical base for phosphor. The anode voltage can reach hundreds or tens of thousands of volts. The phosphors used in FEDs are electron-excited.67 The luminescence mechanism is cathodoluminescence.68 The distance between the gate and the anode is kept short to keep electron emission as a
Field emission displays
Table 8.3
249
Common phosphors used in low-voltage applications.
Item
Blue-green Blue
ZnO:Zn ZnS:Zn ZnS:Ag,Al ZnS:Cu,Al Zn0.65 Cd0.35 S:Ag,Cl (ZnCd)S:Ag Y2 O2 S:Eu
Green Red
Table 8.4
Relative efficiency (%)
Luminance efficiency (lm W−1 )
Excitation energy (V)
8 — — 250 — 150–300 —
10 −1
5.0–10.0 0.5–0.8 0.3–0.6 1.0–1.5 4.5 0.5–1.0 0.5–1.0
200–400 200–400 200–400 200–400 200–400 200–400 200–400
3 — 4 —
Phosphors commonly used in high-voltage applications.
Item
Blue Green Red
Decay time (s)
ZnS:Ag ZnS:Cu,Al Y2 O2 S:Eu
Decay time (s)
Relative efficiency (%)
Luminance efficiency (lm W−1 )
Excitation energy (kV)
30–50 30–50 200
21 17–23 13
— 40–65 —
10–30 10–30 10–30
proximity focusing and not to spread wide. The applied voltage must be low since the distance is short. Table 8.3 presents the phosphors commonly used in low-voltage applications.69, 70 Most of the materials have low luminance efficiency in the range of the applied voltage. ZnO phosphor is a favorable exception with relatively high luminance efficiency.71 However, the green saturation of ZnO is not high, with the green color being mixed with light blue color. Accordingly, ZnO is commonly adopted in monochrome FEDs but it is not commonly used in color FEDs. Additionally, the luminance efficiency varies with the applied voltage. Restated, the luminance efficiency typically increases with increasing applied voltage. Although a short distance between gate and anode can ensure a sufficiently narrow distribution of emission spot sizes, only low-voltage phosphors, which typically have low luminance efficiency, can be used.72 To increase luminance efficiency, high-voltage phosphor is used, as shown in Table 8.4.73 These phosphors have relative high efficiency around the applied voltage. However, the distance between the gate and the anode must be sufficiently large that the device will not break down. Notably, the distribution of phosphor particle sizes is also important.74 Figure 8.21 shows the normal distribution of the sizes of phosphor particles. Since a uniform distribution of phosphor particle sizes typically provides a higher pixel resolution, large and small phosphor particles should be removed before or during the phosphor process. Outgas from the surface layers of the FED device may contaminate and degrade the phosphor. As the phosphor is degraded, the brightness is reduced. The phosphor layer is typically formed by screen printing with a thickness of 30 m for each color. Figure 8.22 shows the typical screen-printing process of the phosphor layer. In this process, phosphor paste is deposited followed by drying and firing. Drying removes the solvent material and firing removes the binder material.
8.8 Assembly and aging techniques The assembly process binds an FEA plate and a phosphor plate into a display panel with vacuum-tight sealing and evacuation.75 After assembly, aging is required to expose defects and stabilize the quality of the display. Figure 8.23 presents a typical assembly and aging process. At the beginning of this process, a spacer and sealing layer are formed. Then, panel alignment, sealing, evacuation and aging are performed.
250
Introduction to Flat Panel Displays
Amount
Small size
Normal size
Large size
Particle size Figure 8.21 Normal distribution of sizes of phosphor particles.
Red phosphor Screen printing red phosphor and drying Green phosphor Screen printing green phosphor and drying Blue phosphor Screen printing blue phosphor and drying; Firing Figure 8.22 Typical screen-printing process for forming a phosphor layer.
Sealing layer
Spacer
Spacer and sealing layer formation
Panel alignment Light
Sealing, evacuation and aging
e–
e–
Figure 8.23 Typical assembly and aging process.
e–
Field emission displays
251
8.8.1 Spacer The spacer maintains a uniform space between the FEA plate and the phosphor plate.76 Therefore, the height of the spacer determines the distance between the gate and anode. In a high-vacuum device, such as a FED, the spacer supports the substrates and prevents them from deforming or cracking. Accordingly, the spacer not only should be a high-vacuum material but also must be tough with high compressive strength. Its width is preferably less than tens of micrometers, since the area of the spacer cannot display. For an insufficiently thin width of spacer, the resolution of the display is limited. The spacer is typically from 100 to 1000 m in height.77 The spacer height is determined by the applied voltage. In general, a high voltage is required to yield high luminance efficiency. At a low voltage of a few hundred volts, a short spacer, about 100 m long, is sufficient to withstand the applied voltage. For a high anode voltage of more than a few thousand volts, long spacers, of over 1000 m, are required to prevent breakdown. Notably, electron emission spreads widely when the spacer is long. Hence, the emission spot in the phosphor plate becomes large.78 In all cases, the spacer must have a high aspect ratio structure. The material of the spacer is typically dielectric. Accordingly, the spacer is easily charged when electrons travel towards the anode. The charging mechanism in the spacer can undesirably discharge during display. This phenomenon causes undesirable arcing inside the display cell. Slight arcing disturbs the display quality. Seriously arcing damages the device. To prevent these phenomena, a spacer material must be a slightly conductive dielectric rather than a normal dielectric. Screen printing, placement and dispensing approaches can be adopted to form a spacer. Since the screen-printing approach must be repeated many times, the spacer width is not easily controlled and the process is time-consuming. Therefore, the placement approach, shown in Figure 8.24, is an alternative and effective approach for forming the spacer. In this process, sealing pastes are deposited at the desired locations of the FEA plate. After this step has been completed, spacers are placed on top of these sealing pastes and then the sealing paste is cured. The dispensing approach provides a higher aspect ratio, which is the ratio of the rib height to the width.79
8.8.2 Sealing layer formation and panel alignment Typical sealing layer materials are glass frits and glass powder, which is a vacuum material and provides a vacuum-tight sealing.80, 81 The material must have a low melting point, and thus be compatible with glass substrate. The use of epoxy material, commonly used in LCDs, as a sealing layer is not appropriate
FEA plate formation
Adhesive Adhesive layer formation
Spacer Spacer placement and curing Figure 8.24 Placement approach for spacer formation.
252
Introduction to Flat Panel Displays
in the FED process since outgas is usually given out of the epoxy material after the sealing process. This outgas can contaminate the display cell and degrade the performance of the display. The first step of the sealing process is to dispense a sealing layer onto the surrounding area of a FEA plate or phosphor plate. The next step is to align the FEA plate and the phosphor plate. The accuracy of this alignment is critical and can affect the display quality. The main challenge of this process is the alignment shift during the high-temperature sealing process. The alignment of the FEA plate and phosphor plate is typically maintained by clippers. When the FEA plate and the phosphor plate are aligned and clipped, an additional sealing paste must also be deposited in the area that surrounds the small opening of the FEA plate. Then, an exhausting tube is placed on the deposited area of the small opening of the FEA plate. An alternative approach, called tubeless sealing, does not use an exhausting tube. The advantages of this tubeless sealing approach are the faster evacuation and thinner panel.82 The evacuation is faster because the panel vacuum is obtained from a ready vacuum of the whole chamber and this chamber vacuum can pump down before the panel process. The panel can be thinner when a tip-off tail of the exhaust tube is absent. However, the gas given off by the sealing paste using tubeless sealing can be sealed inside of the panel and can then contaminate the panel. Tubeless sealing demands extra work to clean the contamination inside of the panel.83
8.8.3 Sealing The sealing process melts the material of the sealing layer to bind the FEA plate to the phosphor plate permanently.84 The gas given off during the sealing process contaminates the emitter. Because of the thermal characteristics of the glass frits and glass powder, the sealing is typically performed at high temperature. The typical sealing process temperature is 450 ◦ C. Clipper selection is important since the clipper may lose binding strength during sealing. For an improper selection of clipper, the panel alignment may be shifted during sealing. Moreover, the glass substrate may crack during sealing at high temperature.
8.8.4 Evacuation and sealing off The evacuation begins in the viscous flow region, in which the mean free path is short and the vacuum pumping speed is rather high. When a high-vacuum environment is achieved, the evacuation is in the molecular flow regime, in which the mean free path is large and the vacuum pumping speed is rather slow. Since the display cell depends on a high-vacuum environment, a vacuum process to evacuate all of the gas from each display cell is critical. In addition to the physical vacuum pump, chemical getting is adopted in FED processing to assist the vacuum pump and adsorb these impurities. The getter is the main material in a chemical pump.85 This getter is a material that can absorb gas to yield a vacuum.86 Radiofrequency heating is commonly adopted to activate and heat the getter but not the glass substrate. Improper heating of the getter can crack the substrate. The additional function of the getter is to absorb the impurities which sometimes are poison gases. Since the poison gases may be emitted into the display cell in the sealed panel, this poison gas must be removed from the sealed panel. The getter can act as a poison gas absorber and maintain a clean gas environment in the display cell. Getters are of two types: evaporation getters (EGs) and nonevaporation getters (NEGs). The EG type has a larger effective area for reaction with gas but the evaporated material may evaporate onto the undesired area and contaminate the device. The tipping-off of the exhausting tube must be performed carefully so that gas given off during the tip-off process can evacuate out and not be left inside of the panel. Therefore, a preliminary tip-off process is typically needed to evacuate the gas given off so that only very little given-off gas is left inside the panel. In order to perform the tip-off process, electric heating or gas heating on the exhausting
Field emission displays
253
Vacuum (torr)
10–3 10–4 Panel vacuum
10–5 10–6
Chamber vacuum
10–7 0
Figure 8.25
50
100
150
200 Time (min.)
Pump chamber vacuum and panel vacuum for a 3-inch panel with 200 m high spacers.
tube is commonly used. Electric heating uses electricity to heat a ceramic and the heated ceramic is used to tip off the exhausting tube, while gas heating uses gas directly to tip off the exhausting tube. Most importantly, electrical heating needs more apparatus than gas heating, although electrical heating is relatively controllable. Notably, the pressure of the vacuum in the pump chamber differs by one to two orders of magnitude from that in the panel. Figure 8.25 shows plots of the chamber vacuum and the panel vacuum for a 3-inch panel with 200 m high spacers.
8.8.5 Aging The purpose of aging is to expose defects. Contamination of the emitter surface and defects in the dielectric and in the electrode are revealed during the aging process. Aging can also stabilize the emission since it can polish or smooth the emitter surface and remove surface contamination from the emitter.
8.9 System techniques FED circuits perform power supply, signal processing and scan/data driving functions. Figure 8.26 shows a typical block diagram. In this diagram, the FED panel is driven by a driver circuit while the signal processing circuit provides a video signal to the panel. Since a few tens of volts must be applied to operate a FED panel, a high-voltage driver IC is required. Additionally, amplitude and pulse width modulation are commonly adopted in FEDs because field emission devices have linear current–voltage (I–V ) and high response characteristics.87 Among scan addressing, direct addressing, matrix addressing and other addressing approaches, matrix addressing can address more data in a short address time. Additionally, each pixel is electrically connected to each row and each column in matrix addressing. For VGA (640 × 480) data format, the number of scan lines is 480 and the number of data lines is 640. Figure 8.27 shows typical matrix addressing. As shown in Figure 8.27, data 1 to 640 are sent to the panel when row 1 is scanned. After row 1 is scanned, row 2 is scanned. Data 1 to 640 are sent to panel when row 2 is scanned. Such scanning proceeds sequentially from row 1 to row 480 to display.
254
Introduction to Flat Panel Displays
Power supply
Signal process
Driver
FED panel Figure 8.26 Typical FED system.
Scan (row 1) Scan (row 2) Scan (row 3)
Scan (row 480) Data (Column 1)
Data (Column 640)
Figure 8.27 Typical matrix addressing.
Homework problems 8.1 Draw a trajectory of electrons that are emitted from emitter to anode for distances of 100 m and 5000 m. Does the spot size of the electrons at a distance of 100 m differ from that at 5000 m? 8.2 Classify the emission when the electric field is 5 × 107 V cm−1 and the temperature is 500 K. Does field emission dominate? 8.3 Describe the process constraints of the Spindt emitter, CNT emitter and SCE.
References 1. Utsumi, T., (1991) Keynote address vacuum microelectronics: what’s new and exciting. IEEE Trans. Electron Dev., ED-38, 2276. 2. Itoh, S., Tanaka, M., Tonegwa, T. et al. (2007) Development of field-emission display. SID Dig., 1297. 3. Palevsky, A., Gammie, G. and Koufopoulos, P. (1994) Field-emission displays: a 10,000-fL high efficiency field-emission display. SID Dig., 55.
Field emission displays
255
4. Gomer, R. (1961) Theory of field emission, in Field Emission and Field Ionization, Harvard University Press, Cambridge, MA, p. 1. 5. Dyke, W.P. and Dolan, W.W. (1956) Field emission, in Advances in Electronics and Electron Physics, Vol. 8 (ed. L. Marton), Academic Press, New York, p. 89. 6. Modinos, A. (1984) Electron emission from free-electron metals, in Field, Thermionic, and Secondary Electron Emission Spectroscopy, Plenum, New York, p. 1. 7. Fowler, R.H. and Nordheim, L. (1928) Electron emission in intense electric fields. Proc. R. Soc. Lond., A119, 73. 8. Nordheim, L.W. (1928) The effect of the image force on the emission and reflection of electrons by metals. Proc. R. Soc. Lond., A121, 626. 9. Stern, T.E., Gossling, B.S. and Fowler, R.H. (1929) Future studies in the emission of electrons from cold metals. Proc. R. Soc. Lond., 124, 699. 10. Spindt, C.A., Brodie, I., Humphey, L. and Westerberg, E.R. (1976) Physical properties of thin-film field emission cathodes with molybdenum cones. J. Appl. Phys., 47 (12), 5248. 11. Good Jr, R.H. and Muller, E.W. (1956) Field emission, in Handbook Der Physik, Vol. XXI, Electron-Emission Gas Discharge I, Springer-Verlag, Berlin, p. 176. 12. Burgess, R.E. and Kroemer, H. (1953) Corrected values of Fowler–Norheim field emission functions v(y) and s(y). Phys. Rev., 90 (4), 515. 13. Gomer, R. (1961) Theory of field emission, in Field Emission and Field Ionization, Harvard University Press, Cambridge, MA, p. 1. 14. Modinos, A., (1984) Electron emission from free-electron metals, in Field, Thermionic, and Secondary Electron Emission Spectroscopy, Plenum, New York, p. 1. 15. Murphy, E.L. and Good Jr, R.H. (1956) Thermionic emission, field emission, and the transition region. Phys. Rev., 102 (6), 1464. 16. Brodie, I. (1989) Physic considerations in vacuum microelectronic devices. IEEE Trans. Electron Dev., ED36, 2641. 17. Tsai, T., Su, T., Liu, D. and Huang, J. (1993) Optimization of micro tip design. Symposium on Semiconductor Modeling and Simulation, p. 109. 18. Liu, D., Ravi, T.S., Bagley, B.G. et al. (1992) Fabrication of self-aligned gated field emitters. J. Micromech. Microeng., 2, 21. 19. Tanaka, M., Obara, Y., Naito, Y. et al. (2004) Development of a 11.3-inch VGA full-color FED. SID Dig., 832. 20. Liu, D., Ravi, T.S., Gmitter, T. et al. (1991) Fabrication of wedge-shaped silicon field emitters with nm-scale radii. Appl. Phys. Lett., 58 (10), 1042. 21. Okuda, M., Matsutani, S., Asai, A. et al. (1998) Electron trajectory analysis of surface conduction electron emitter displays (SEDs). SID Dig., 185. 22. Palmer, W.D., Mancusi, J.E., Ball, C.A. et al. (1994) Measured DC performance of large arrays of silicon field emitters. IEEE Trans. Electron Dev., 41 (10), 1866. 23. Liu, D. and Marcus, R.B. (1994) Characterization of silicon field emission microtriodes. J. Vac. Sci. Technol. B, 12 (2), 672. 24. Stratton, R. (1962) Theory of field emission from semiconductors. Phys. Rev., 125 (1), 67. 25. Brodie, I. (1975) Bombardment of field-emission cathodes by positive ions formed in the inter electrode region. Int. J. Electron., 38, 541. 26. Ravi, T.S., Marcus, R.B. and Liu, D. (1991) Oxidation sharpening of silicon tips. J. Vac. Sci. Technol. B, 9 (6), 2733. 27. Marcus, R.B., Ravi, T.S., Gmitter, T. et al. (1991) Atomically sharp silicon and metal field emitters. IEEE Trans. Electron Dev., 38 (10), 2289. 28. Marcus, R.B., Ravi, T.S., Gmitter, T. et al. (1990) Formation of silicon tips with < 1 nm radius. Appl. Phys. Lett., 56 (3), 236. 29. Orvis, W.J., Ciarlo, D.R., McConaghy, C.F. et al. (1989) A progress report on the Livermore miniature vacuum tube project. Proc. IEDM, 89, 529. 30. Chesnokov, V.V., Gailes, V.M., Domakhina, A.M. et al. (1991) Field-emission microcathodes in microelectronics devices. IEEE Trans. Electron Dev., ED-38, 2389. 31. McGruer, N.E., Warner, K., Singhal, P. et al. (1991) Oxidation sharpened gated field emitter array process. IEEE Trans. Electron Dev., ED-38, 2389. 32. Duzshev, N.A., Eremchenko, D.V., Kozlov,A.I. et al. (1990) Wedge-type array cold cathode stability. Proceedings of the 3rd International Vacuum Microelectronics Conference, Williamsburg, VA, p. 5.
256
Introduction to Flat Panel Displays
33. Betsui, K. (1991) Fabrication and characteristics of Si field emitter arrays. Proceedings of the 4th International Vacuum Microelectronics Conference, Nagahama, Japan, p. 26. 34. Kanemaru, S. and Itoh, J. (1991) Fabrication and characterization of lateral field-emission triodes. IEEE Trans. Electron Dev., ED-38, 2334. 35. Nureki, C. and Araragi, M. (1991) Planar field emission devices with three-dimensional gates structures. Proceedings of the 4th International Vacuum Microelectronics Conference, Nagahama, Japan, p. 78. 36. Spindt, C.A., Brodie, I., Humphey, L, and Westerberg, E.R. (1976) Physical properties of thin-film field emission cathodes with molybdenum cones. J. Appl. Phys., 47 (12), 5248. 37. Kondo, Y., Matsuura, J. and Kimura, H. (1991) Characteristics of field emitter arrays for light source. Proceedings of the 4th International Vacuum Microelectronics Conference, Nagahama, Japan, p. 56. 38. Spindt, C.A., Holland, C.E., Rosengreen, A. and Brodie, I. (1991) Field-emitter arrays for vacuum microelectronics. IEEE Trans. Electron Dev., ED-38, 2355. 39. Givargizov, E.I., Kudintseva, C.A. and Chubun, N.N. (1988) Field-emission arrays from refractory materials. Proceedings of the 1st International Vacuum Microelectronics Conference, Williamsburg, VA, p. 8. 40. Djubua, B.C. and Chubun, N.N. (1991) Emission properties of Spindt-type cold cathodes with different emission cone material. IEEE Trans. Electron Dev., ED-38, 2314. 41. Fukuda, S. and Betsui, K. Fabrication and characteristics of tantalum field emitter arrays. Proceedings of the 4th International Vacuum Microelectronics Conference, Nagahama, Japan, p. 144. 42. Komatsu, H. (1991) Fabrication and characteristics of vacuum microelectronic devices with a lateral field electron emission cathode. Proceedings of the 4th International Vacuum Microelectronics Conference, Nagahama, Japan, p. 48. 43. Liu, D. (1993) Fabrication and characterization of gated silicon field emitters. ASID ’93, p. 165. 44. Brodie, I. (1989) Physic considerations in vacuum microelectronic devices. IEEE Trans. Electron Dev., ED-36, 2641. 45. Liu, D., Ravi, T.S. and Marcus, R.B. (1992) Gated micron-size electron field emitters. 50th DRC Conference, p. IIIA-5. 46. Itoh, S., Watanabe, T., Ohtsu, K. et al. (1993) Investigation of cathodoluminescence display device with field emission cathodes. Jpn. J. Appl. Phys., 32, 3955. 47. Kim, J.M., Lee, N.S., Chung, D.S. et al. (2001) New emitter techniques for field emission displays. SID Dig., 304. 48. Dijon, J., Boronat, J.F., Fournier, A. et al. (2004) Towards a low cost high quality carbon nanotubes field emission display. SID Dig., 820. 49. Park, K.C., Ryu, J.H., Yu, Y.Y. et al. (2007) Controlled electron emission current with carbon nanotube emitters array grown with RAP process. SID Dig., 1417. 50. Cho, Y.R., Choi, W.S., Shin, H.Y. et al. (2004) Effects of binder materials in CNT paste on field emission properties for CNT-cathode applications. SID Dig., 368. 51. Lee, J.A., Lee, Y.D., Moon, S.I. et al. (2004) Effect of surface treatment on screen-printed carbon-nanotube films by argon plasma. SID Dig., 356. 52. Lee, C.C., Lin, B.N., Hsiao, M.C. et al. (2005) Development of CNT-FED by printing method. SID Dig., 1716. 53. Dijon, J., Fournier, A., De Monsabert, T.G. et al. (2006) A status on the emission uniformity of CNT FED technology. SID Dig., 1744. 54. Pribat, D., Cojocaru, C.S., Xavier, S. et al. (2007) Field emission from nanostructures: carbon nanotubes and metal nanowires. SID Dig., 1409. 55. Hsiao, C.Y., Chan, T.F., Lee, S.H. et al. (2005) Electrophoresis deposition method to fabricate CNT-FED cathode in water base solution. SID Dig., 411. 56. Lim, S.H., Moon, J.H., Yoon, H.S. et al. (2004) Field emission from carbon nanotubes grown by a triode-type DC plasma enhanced CVD. SID Dig., 924. 57. Yamamoto, K., Nomura, I., Yamazaki, K. and Uzawa, S. (2005) Fabrication and characterization of surface conduction electron emitters. SID Dig., 1933. 58. Fink, R.L., Thuesen, L.H., Ginsberg, V. et al. (2006) Twenty-five inch diagonal carbon nanotube field emission display. SID Dig., 1748. 59. Oguchi, T., Yamaguchi, E., Sasaki, K. et al. (2005) A 36-inch surface-conduction electron-emitter display (SED). SID Digest, 1929. 60. Reimer, D.E. (1988) Analytical engineering model of the screen printing process: I. Solid State Technol., August, 107.
Field emission displays
257
61. Sakamoto, S. and Ogawa, Y. (1995) Screen printing for fabrication of PDPs. International Display Workshops ’95, p. 41. 62. Dolan, W.W., Dyke, W.P. and Trolan, J.K. (1953) The field emission initiated vacuum arc: II. The resistively heated emitter. Phys. Rev., 91 (5), 1054. 63. Spindt, C.A., Holland, C.E., Rosengreen, A. and Brodie, I. (1991) The effects of size and shape on field-emitter array performance. Proceedings of the 4th International Vacuum Microelectronics Conference, Nagahama, Japan, p. 44. 64. McCord, M.A., Chang, T.H.P., Kern, D.P. and Speidell, J.L. (1989) An STM controlled field emission microlens electron source, in Vacuum Microelectronics 89 (ed. R.E. Turner), Institute of Physics Publishing, Bristol, p. 165. 65. Ford, P.W. and Veyhl, E.W. (1993) Image position errors due to plate bending. SID Dig., 983. 66. Hayakawa, H., Sakuda, K., Kuwada, R. et al. (1997) Photoimageable thick film black conductor system for FPD. International Display Workshops ’97, p. 547. 67. Vecht, A. (1994) Advances in phosphor materials for display application. IDRC ’94, p. 86. 68. Yabe, M., Iwashima, M. and Omoto, K. (2007) FED spacer fabrication with nozzle dispense. SID Dig., 1201. 69. Chakhovskoi, A.G., Kesling, W.D., Trujillo, J.T. and Hunt, C.E. (1994) Phosphor selection constraints in application of gated field-emission microcathodes to flat panel displays. J. Vac. Sci. Technol. B, 12 (2), 785. 70. Nagata, T., Nakayama, T. and Murakami, H. (2007) Low voltage cathodoluminescent properties of Y2 O3 -based nano-particle phosphors. SID Dig., 1328. 71. Tyan, J.H., Lu, J.Y., Liu, D.N. et al. (1995) Phosphor characterization for FED applications. First International Conference on the Science and Technology of Display Phosphors, Nov. 14–16, 1995, San Diego, CA, p. 145. 72. Tanaka, M., Nohara, Y., Tamura, K. et al. (1998) Luminance improvement of low-voltage full-color FEDs. SID Dig., 193. 73. Dean, K.A., Li, H., Coll, B.F. et al. (2006) High brightness, high voltage color field emission display technology. SID Dig., 1845. 74. Ozawa, L. (1994) Phosphor screen for cathode ray picture tubes for entertainment purposes, in Application of Cathodoluminescence to Display Devices, Kodansha, Tokyo, p. 7. 75. Reisman, A. (1978) Single-cycle gas panel assembly. IBM J. Res. Develop., 22 (6), 596. 76. Nonomura, K. and Honjo, K. (2005) Influence of spacer on image quality. SID Dig., 426. 77. Yotani, J., Uemura, S., Nagasako, T. et al. (2007) Practical CNT-FED structure for high-luminance characterdisplays. SID Dig., 1301. 78. Dean, K.A., Howard, E., Johnson, M.R. et al. (2007) A caron nanotube field emission display with simple electron beam trajectory control. SID Dig., 1305. 79. Yabe, M., Iwashima, M. and Omoto, K. (2007) FED spacer fabrication with nozzle dispense. SID Dig., 1201. 80. Ishizeki, K., Kuroki, Y., Ueki, M. et al. (2006) A novel hermetic-sealing material for FED. SID Dig., 1756. 81. Roth, A. (1966) Permanent seals, in Vacuum Sealing Techniques, Pergamon Press, Oxford, p. 23. 82. Shimosato, Y., Seki, T., Yoshimura, Y. and Nagoshi, T. (1998) Cart pusher in-line exhausting system for PDPs. International Display Workshops ’98, p. 559. 83. Shimosato, Y., Seki, T. and Yoshimura, Y. (1997) Vacuum process for plasma display panel without exhaust pipe. International Display Workshops ’97, p. 539. 84. Alpha, J.W. (1976) Glass sealing technology for displays. Opt. Laser Technol., December, 259. 85. Kohl, W.H. (1980) Getter materials, in Handbook of Materials and Techniques for Vacuum Devices, Reinhold Publishing, New York, p. 545. 86. Caloi, R.M., Carretti, C. and Amiotti, M. (1997) Gettering technology in plasma display panels. International Display Workshops ’97, p. 543. 87. Huang, C.N. and Liang, C.C. (2007) A study on PWM driving schemes for a 20 inch VGA carbon nanotube field emission display. SID Dig., 1309.
Index A
C
absorption, 11, 23, 41, 42, 54, 66, 67, 79, 80, 81, 139, 142, 147, 165, 166, 168, 179, 180, 181, 183 acceptor, 37, 38, 138, 147, 148, 149, 150, 184, 185, 218 achromatic, 21, 75 active matrix (AM), 31, 47, 196 alkali, 202, 203, 220 alkaline, 202, 220 aluminum chelate, 178, 197 ambipolar, 196 anion, 189, 202 anthracene, 177 antibonding, 178 antinode, 195, 196 aperture ratio, 57, 58, 59, 87, 92, 95, 122 aqueous humor, 12 arc discharge, 110–111 aromatic diamine, 178, 197 arylamine, 203 atomic orbital, 177, 178, 179
candela (cd), 25, 26, 27, 199, 200 charge generation layer (CGL), 220, 221 chromaticity diagram, 16, 17, 18, 19, 20, 21, 23 Commission Internationale de l’Eclairage (CIE), 11, 12, 16, 17, 18, 19, 20, 21, 23 ciliary muscle, 12, 13 cold cathode fluorescent lamp (CCFL), 23, 24, 59, 139 color change material (CCM), 207 color difference, 11, 19, 21, 24, 25 color gamut, 21, 22, 59, 91, 101, 139, 145, 158, 160, 168, 170, 172 color matching experiments, 16, 17 color mixing, 16, 20 color space, 11, 16, 20, 21, 27 color temperature, 22, 23 colorimetry, 12, 15 concentration quenching, 187, 199 conduction band, 33, 34, 36, 37, 40, 137, 141, 144, 148, 150, 156, 177 cone cell, 11, 13, 14, 15, 28 continuity equation, 157, 189, 192 cornea, 12 correlated color temperature, 23 critical angle, 139, 161, 165, 193, 194
B Beer-Lambert law, 187 blackbody radiator, 22, 23 blackbody locus, 23 blind spot, 12, 13, 14 bonding, 32, 33, 37, 38, 39, 40, 178, 179, 209 Born-Oppenheimer approximation, 179
Introduction to Flat Panel Displays c 2008 John Wiley & Sons, Ltd
D Dexter energy transfer, 184, 185, 218 donor, 37, 38, 147, 148, 149, 150, 184, 185, 218 drift-diffusion current equation, 189
J.-H. Lee, D.N. Liu and S.-T. Wu
260
Index
E
G
effective mass, 35, 36, 40, 148, 149, 191 eigenenergy, 179 electrode quench, 187, 196 electroluminescence (EL), 137, 141, 144, 146, 177, 178, 189, 195, 198, 199, 200, 203, 204, 205, 208, 212, 213, 214, 218, 221, 224 electronic state, 179, 180, 181, 182, 183, 184, 185, 188 electron-injection layer (EIL), 178, 189, 194, 200 electron-transporting layer (ETL), 178, 189, 194, 197, 198, 199, 200, 202, 203, 211, 212, 214, 215, 216, 221, 224 emitting layer (EML), 178, 189, 194, 197, 198, 199, 200, 203, 211, 214, 215, 216, 219, 220, 222, 223, 224 encapsulation, 145, 161, 196, 206, 213, 214 energy state, 32, 33, 34, 37, 44, 158, 177, 178, 179 epitaxy, 139, 143, 149, 154, 155, 156, 161, 163 excimer, 180, 184, 185, 186, 187, 199 exciplex, 180, 184, 185, 186, 187 external electrode fluorescent lamp (EEFL), 23, 24 external quantum efficiency (EQE), 142, 194, 196, 203, 220, 224, 225 extraction efficiency, 139, 155, 158, 161, 165, 166, 167, 168, 194, 224, 225 extrinsic degradation, 196 eyeballs, 13 eye lens, 12, 13
Gaussian Lens formula, 12 gray scale, 75, 76, 78, 86, 100, 131–131, 239 glow discharge, 110–111
F Fabry-Perot cavity, 194 Fermi level, 34, 38, 44, 147, 149, 150, 202, 233–235 field emission display (FED), 23, 24, 233, 237–238, 240, 244–245, 247, 249, 252, 254 flat fluorescent lamp (FFL), 23, 24, fluorescence, 179, 181, 183, 184, 186, 188, 193, 200, 214, 215 Förster energy transfer, 184, 185, 218 Fowler-Nordheim(FN) tunneling, 189, 235–236 Franck-Condon principle, 179, 180, 183, 185 Frenkel exciton, 193
H Hamiltonian, 183 heavy atom effect, 183 heterojunction, 148, 153 hole-blocking layer (HBL), 178, 219 hole-injection layer (HIL), 178, 189, 199, 200, 203, 205 hole-transporting layer (HTL), 178, 189, 194, 197, 198, 199, 200, 202, 205, 211, 212, 213, 214, 215, 216, 221 highest occupied molecular orbital (HOMO), 177, 178, 179, 180, 181, 196, 197, 200, 201, 202, 203, 211, 219, 220, 222 hot cathode fluorescent lamp (HCFL), 23, 24
I ideality factor, 190 illuminance, 25, 26 impurity quenching, 187 indium tin oxide (ITO), 58, 73, 74, 81, 95, 96, 97, 98, 114, 118–119, 139, 165, 178, 189, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 205, 206, 209, 210, 211, 213, 214, 220, 221, 225 ink-jet printing, 196, 206, 209, 243 intermolecular process, 184, 187 internal conversion, 183, 188 internal quantum efficiency (IQE), 182, 195, 218, 222, 224 intersystem crossing (ISC), 183, 193, 219 intrinsic degradation, 196, 213, 214
J Jablonski diagram, 180, 183, 184, 188
L Langevin theory, 138, 155, 156, 190, 193, 211 laser-assisted pattern technique, 208 laser-induced pattern-wise sublimation (LIPS), 208 laser induced thermal imaging (LITI), 208
Index
light-emitting diode (LED), 23, 24, 27, 31, 32, 57, 59, 101, 137 light-to-heat conversion (LTHC), 208 lithography, 115–116, 119, 121, 205, 207, 243–244, 247–248 lowest unoccupied molecular orbital (LUMO), 177, 179, 180, 181, 196, 197, 200, 201, 203, 211, 220, 222 lumen (lm), 17, 25, 26, 27 luminous flux, 25, 26, 27 luminance, 25, 26 luminous intensity, 25, 26, 27 lux, 25, 26
M MacAdam ellipses, 19, 20 metamerism, 12, 28 microcavity, 195, 203, 223 mobility, 31, 39, 40, 41, 42, 43, 45, 46, 53, 54, 55, 148, 149, 152, 177, 187, 189, 192, 193, 195, 197, 198, 202, 210, 211 molecular orbital, 177, 178, 179, 180 monochromatic light locus, 17 Mott-Gurney equation, 192
O optic nerve, 12, 13 organic light-emitting device (OLED), 31, 47, 49, 50, 52, 53, 177
P passivation, 42, 54, 96, 97, 149, 196, 206, 209, 210, 213 passive matrix (PM), 47, 48, 49, 50, 53 Pauli’s exclusion principle, 32, 34, 178, 182 photometry, 12, 18 photopic, 13, 14, 25 photoreceptor, 13, 16 phosphorescence, 177, 179, 182, 183, 184, 193, 218 Planck constant, 22, 35, 191 Poisson’s equation, 147, 150, 189, 192 polaron, 193 polyethylene Terephthalate (PET), 210 polymetric light-emitting device (PLED), 204, 205, 214, 225 Poole-Frenkel (PF) model, 189 primary color, 11, 15, 16, 21, 27, 28
261
Q quantum yield, 187, 188, 219 quenching, 187
R radiometry, 16, 18 recombination, 137, 138, 139, 140, 141, 142, 148, 153, 154, 155, 156, 157, 158, 159, 165, 177, 183, 184, 189, 190, 191, 192, 193, 195, 196, 197, 198, 200, 205, 211, 212, 213, 215, 216, 218, 222, 224 reflection, 11, 12, 60, 91, 92, 93, 99 refractive index, 57, 67, 68, 69, 75, 139, 143, 155, 161, 168, 193, 194, 195, 212, 225 retina, 12, 13 Richardson-Schottky (RS) thermionic emission, 177, 190 rod cell, 13, 14
S Schottky barrier, 138, 203 scotopic, 13, 14 semiconductor, 31, 32 shadow mask, 206, 207, 208 singlet exciton, 177, 179, 180, 182, 183, 188, 193 space-charge limited conduction (SCLC), 177, 189, 192, 211 spin-coating, 196, 204, 206, 209 spin-orbital coupling, 183, 218 Stokes shift, 146, 147, 181, 183, 189, 218, 219, 223, 224
T thermal evaporation, 177, 203, 206 thin-film transistor (TFT), 31, 57, 58, 59, 60, 63, 68, 75, 78, 83, 91, 92, 93, 95, 96, 101 trap-charge limited conduction (TCLC), 177, 189, 192, 211 trichromatic space, 11, 15 triplet exciton, 177, 179, 180, 182, 183, 188, 193, 218, 219, 223, 224 triplet-triplet annihilation, 219, 224 tristimulus values, 15, 17, 20 two-transistor and one-capacitor (2T1C), 52, 53
262
U uniform color system, 19, 25
V valence band, 33, 34, 36, 37, 38, 39, 40, 44, 137, 141, 144, 148, 150, 153, 156, 158, 164, 177 vibrational state, 180, 181, 183
Index
viewing angle, 13, 16, 57, 59, 60, 64, 71, 73, 75, 76, 77, 78, 79, 80, 83, 86, 88, 89, 90, 91, 92, 93, 95, 96, 101, 225 visual axis, 13 vitreous humor, 12
W water vapor permeation rate (WVPR), 209, 210 workfunction, 233–237, 239, 247