Micro LEDs (Volume 106) [1 ed.] 012823041X, 9780128230411

Citation preview

SERIES EDITORS CHENNUPATI JAGADISH Distinguished Professor Department of Electronic Materials Engineering Research School of Physics and Engineering Australian National University Canberra, ACT2601, Australia

ZETIAN MI Professor Department of Electrical Engineering and Computer Science University of Michigan 1310 Beal Avenue Ann Arbor, MI 48109 United States of America

Academic Press is an imprint of Elsevier 50 Hampshire Street, 5th Floor, Cambridge, MA 02139, United States 525 B Street, Suite 1650, San Diego, CA 92101, United States The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, United Kingdom 125 London Wall, London, EC2Y 5AS, United Kingdom First edition 2021 Copyright © 2021 Elsevier Inc. All rights reserved No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. Details on how to seek permission, further information about the Publisher’s permissions policies and our arrangements with organizations such as the Copyright Clearance Center and the Copyright Licensing Agency, can be found at our website: www.elsevier.com/permissions. This book and the individual contributions contained in it are protected under copyright by the Publisher (other than as may be noted herein). Notices Knowledge and best practice in this field are constantly changing. As new research and experience broaden our understanding, changes in research methods, professional practices, or medical treatment may become necessary. Practitioners and researchers must always rely on their own experience and knowledge in evaluating and using any information, methods, compounds, or experiments described herein. In using such information or methods they should be mindful of their own safety and the safety of others, including parties for whom they have a professional responsibility. To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors, assume any liability for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions, or ideas contained in the material herein. ISBN: 978-0-12-823041-1 ISSN: 0080-8784

For information on all Academic Press publications visit our website at https://www.elsevier.com/books-and-journals

Publisher: Zoe Kruze Acquisitions Editor: Jason Mitchell Developmental Editor: Jhon Michael Peñano Production Project Manager: Abdulla Sait Cover Designer: Alan Studholme Typeset by SPi Global, India

Contents Contributors Preface

1. Development of nitride microLEDs and displays

ix xiii

1

Hongxing Jiang and Jingyu Lin 1. Introduction 2. MicroLED structure and processing 3. MicroLED array for high voltage AC/DC-LEDs 4. Realization of the first full-scale active driving microLED microdisplay 5. MicroLED and microLED display characteristics 6. Full color microdisplay development 7. Applications in large flat panel displays and medicine 8. Concluding remarks Acknowledgments References

2. Micro-LEDs for biomedical applications

1 6 13 19 25 36 47 50 50 51

57

Jonathan J.D. McKendry, Erdan Gu, Niall McAlinden, Nicolas Laurand, Keith Mathieson, and Martin D. Dawson 1. Introduction 2. Micro-LED characteristics 3. GaN LED/CMOS chip-scale microfluorimetry with SPADs 4. Micro-LED based optoelectronic tweezers 5. Light-emitting dressings and printed LEDs 6. Optogenetic neural probes and neural interfaces 7. Conclusion Acknowledgments References

3. High external quantum efficiency III-nitride micro-light-emitting diodes

58 60 64 71 78 83 89 89 89

95

Matthew S. Wong, Shuji Nakamura, and Steven P. DenBaars 1. Introduction 2. Size-dependent efficiency 3. Techniques toward high efficiency

95 96 99 v

vi

Contents

4. Long-wavelength devices 5. Conclusion References

4. GaN-on-silicon MicroLEDs for neural interfaces

111 114 114

123

Kanghwan Kim, Fan Wu, Kensall D. Wise, and Euisik Yoon 1. Introduction 2. Michigan probes and optogenetics 3. Optoelectrodes: Optical stimulation combined with electrical recording within an integrated platform 4. GaN-on-silicon microLED optoelectrodes 5. Minimizing stimulation artifact 6. Discussion 7. Conclusion References

5. Quantum-dot-based full-color micro-LED displays

124 126 131 140 152 164 165 166

173

Tingzhu Wu, Yu-Ming Huang, James Singh Konthoujam, Zhong Chen, and Hao-Chung Kuo 1. Introduction 2. Background of full-color micro-LED display 3. Conclusion References

6. Damage-free neutral beam etching for GaN micro-LEDs processing

173 175 198 198

203

Xuelun Wang and Seiji Samukawa 1. Introduction 2. Neutral beam generation source 3. Application of NBE for fabrication of sub-10-nm nanostructures 4. Sub-10-μm GaN micro-LEDs fabricated by NBE 5. Conclusion References

7. From nanoLEDs to the realization of RGB-emitting microLEDs

203 205 209 213 219 220

223

Zhaoxia Bi, Zhen Chen, Fariba Danesh, and Lars Samuelson 1. An industrial perspective on the market pull for the development of microLED displays

223

Contents

2. General description of the challenges, status, and needs for progress 3. An overview of our different material science approaches toward long-wavelength nitride emitters 4. Industrial approaches toward mass transfer of microLEDs 5. Outlook Acknowledgments References

8. Mass transfer for Micro-LED display: Transfer printing techniques

vii

225 226 245 247 248 248

253

Changhong Linghu, Shun Zhang, Chengjun Wang, Hongyu Luo, and Jizhou Song 1. Introduction 2. Mass transfer techniques for Micro-LED displays 3. Transfer printing techniques for Micro-LED assembly 4. Latest development of transfer printing techniques 5. Conclusion References

9. Micro-LED based optical wireless communications systems

253 255 259 261 276 277

281

P. Tian, Jonathan J.D. McKendry, J. Herrnsdorf, S. Zhu, Erdan Gu, Nicolas Laurand, and Martin D. Dawson 1. Introduction 2. Device-related characteristics 3. Micro-LED based high-speed VLC systems 4. Novel optical wireless communication systems based on micro-LED 5. Conclusion Acknowledgments References Further reading

281 283 295 305 316 317 317 321

10. Angular color shift and power consumption of RGB micro-LED displays

323

Fangwang Gou, En-Lin Hsiang, and Shin-Tson Wu 1. Angular color shift of RGB micro-LEDs 2. Power consumption References

323 333 342

viii

Contents

11. Monolithic integration of AlGaInP red and InGaN blue/green LEDs

345

Dong-Seon Lee and Sang Hyeon Kim 1. Introduction 2. Multi-color integration by wafer bonding 3. Toward high-resolution microLED display 4. Conclusion References Index

345 354 369 383 385 389

Contributors Zhaoxia Bi Lund University, Solid State Physics, NanoLund, Lund, Sweden Zhen Chen gl o-USA, Sunnyvale, CA, United States Zhong Chen Department of Electronic Science, Fujian Engineering Research Center for Solid-State Lighting, Xiamen University, Xiamen, China Fariba Danesh gl o-USA, Sunnyvale, CA, United States; gl o AB, Lund, Sweden Martin D. Dawson Institute of Photonics, Department of Physics, University of Strathclyde, Glasgow, United Kingdom Steven P. DenBaars Materials Department; Department of Electrical and Computer Engineering, University of California, Santa Barbara, CA, United States Fangwang Gou College of Optics and Photonics, University of Central Florida, Orlando, FL, United States Erdan Gu Institute of Photonics, Department of Physics, University of Strathclyde, Glasgow, United Kingdom J. Herrnsdorf Institute of Photonics, Department of Physics, University of Strathclyde, Glasgow, United Kingdom En-Lin Hsiang College of Optics and Photonics, University of Central Florida, Orlando, FL, United States Yu-Ming Huang Department of Photonics & Graduate Institute of Electro-Optical Engineering, College of Electrical and Computer Engineering, National Chiao Tung University, Hsinchu, Taiwan Hongxing Jiang Department of Electrical and Computer Engineering, Texas Tech University, Lubbock, TX, United States Kanghwan Kim Department of Electrical Engineering and Computer Science, University of Michigan, Ann Arbor, MI, United States; Center for BioMicrosystems, Brain Science Institute, Korea Institute of Science and Technology, Seoul, South Korea ix

x

Contributors

Sang Hyeon Kim School of Electrical Engineering, Korea Advanced Institute of Science and Technology (KAIST), Yuseong-gu, Daejeon, Republic of Korea James Singh Konthoujam Department of Photonics & Graduate Institute of Electro-Optical Engineering, College of Electrical and Computer Engineering, National Chiao Tung University, Hsinchu, Taiwan Hao-Chung Kuo Department of Photonics & Graduate Institute of Electro-Optical Engineering, College of Electrical and Computer Engineering, National Chiao Tung University, Hsinchu, Taiwan Nicolas Laurand Institute of Photonics, Department of Physics, University of Strathclyde, Glasgow, United Kingdom Dong-Seon Lee School of Electrical Engineering and Computer Science, Gwangju Institute of Science and Technology, Gwangju, Korea Jingyu Lin Department of Electrical and Computer Engineering, Texas Tech University, Lubbock, TX, United States Changhong Linghu Department of Engineering Mechanics, Soft Matter Research Center, and Key Laboratory of Soft Machines and Smart Devices of Zhejiang Province, Zhejiang University, Hangzhou, China Hongyu Luo Department of Engineering Mechanics, Soft Matter Research Center, and Key Laboratory of Soft Machines and Smart Devices of Zhejiang Province, Zhejiang University, Hangzhou, China Keith Mathieson Institute of Photonics, Department of Physics, University of Strathclyde, Glasgow, United Kingdom Niall McAlinden Institute of Photonics, Department of Physics, University of Strathclyde, Glasgow, United Kingdom Jonathan J.D. McKendry Institute of Photonics, Department of Physics, University of Strathclyde, Glasgow, United Kingdom Shuji Nakamura Materials Department; Department of Electrical and Computer Engineering, University of California, Santa Barbara, CA, United States Lars Samuelson Lund University, Solid State Physics, NanoLund; gl o AB, Lund, Sweden

Contributors

xi

Seiji Samukawa Institute of Fluid Science, Tohoku University, Sendai, Japan Jizhou Song Department of Engineering Mechanics, Soft Matter Research Center, and Key Laboratory of Soft Machines and Smart Devices of Zhejiang Province, Zhejiang University, Hangzhou, China P. Tian School of Information Science and Technology, Academy of Engineering and Technology, Fudan University, Shanghai, China Chengjun Wang Department of Engineering Mechanics, Soft Matter Research Center, and Key Laboratory of Soft Machines and Smart Devices of Zhejiang Province, Zhejiang University, Hangzhou, China Xuelun Wang GaN Advanced Device Open Innovation Laboratory, National Institute of Advanced Industrial Science and Technology, Nagoya, Japan Kensall D. Wise Department of Electrical Engineering and Computer Science, University of Michigan, Ann Arbor, MI, United States Matthew S. Wong Materials Department, University of California, Santa Barbara, CA, United States Fan Wu Diagnostic Biochips, Glen Burnie, MD, United States Shin-Tson Wu College of Optics and Photonics, University of Central Florida, Orlando, FL, United States Tingzhu Wu Department of Electronic Science, Fujian Engineering Research Center for Solid-State Lighting, Xiamen University, Xiamen, China Euisik Yoon Department of Electrical Engineering and Computer Science; Department of Biomedical Engineering, University of Michigan, Ann Arbor, MI, United States; Center for Nanomedicine, Institute for Basic Science (IBS) and Graduate Program of Nano Biomedical Engineering (Nano BME), Yonsei University, Seoul, South Korea Shun Zhang Department of Engineering Mechanics, Soft Matter Research Center, and Key Laboratory of Soft Machines and Smart Devices of Zhejiang Province, Zhejiang University, Hangzhou, China S. Zhu School of Information Science and Technology, Academy of Engineering and Technology, Fudan University, Shanghai, China

Preface MicroLED (μLED), initially developed in 2000 contemporaneous with the emergence of blue/white light-emitting diode (LED)-based solid-state lighting, is now recognized as the ultimate display technology and is one of the fastest-growing technologies in the world. Tech giants are currently utilizing it on a wide range of products, including wearable displays for high speed three-dimensional/augmented reality/virtual reality (3D/AR/VR) display applications, high brightness/contrast large flat panel displays and TVs, and as light sources for neural interfaces and optogenetics and visible light communications (Li-Fi). This volume comprises contributions from a group of scientists and engineers who made pioneering and critical contributions to the developments of μLED technology. In Chapter 1, the results of the early development of the first μLED/μLED array as well as the first active Si CMOS driving full-scale high-resolution μLED microdisplay and the status and remaining challenges of microLED technology are discussed. Chapter 2 discusses the advancement of microLEDs for biomedical applications. Methods for realizing high-external quantum efficiency III-nitride microLEDs are presented in Chapter 3. The development of GaN-on-Silicon microLEDs for neural interface is presented in Chapter 4. Various processes and device architectures for achieving full-color microLED displays based on quantum-dots, nanoLEDs, and monolithic integration of AlGaInP red and InGaN blue/ green LEDs are discussed, respectively, in Chapters 5, 7, and 11. The implementation of neutral beam etching processing for realizing damage-free microLEDs is presented in Chapter 6. The development of transfer printing techniques for microLED assembly is summarized in Chapter 8. Chapter 9 describes the development of optical wireless communications systems, Li-Fi, based on microLEDs. Critical issues of angular color shift and power consumption of full-color microLED displays are addressed in Chapter 10. It has been extremely thrilling as well as satisfying for us to witness that the shift in the format of LED to the microsize has generated such intensive efforts in the development of emerging μLED products. Researchers are racing to overcome key technical barriers to bring μLED products to the market. According to MarketWatch, “The global μLED market is valued at $170 million in 2018 and is expected to reach $17 billion by the end of 2025, growing at a CAGR of 78.3% during 2019–2025,” while R&D xiii

xiv

Preface

activities have experienced exponential growth beginning from 2006, with a total of nearly 3000 publication items in 2020 according to Google Scholar. We see a very bright future for μLED. The most significant driving force is the potentially huge market demand for this technology. It is anticipated that collective efforts from the worldwide R&D communities of III-nitrides, LEDs, lighting, displays, optical communications, and optogenetics will not only bring microLED products to mass consumer electronics markets, but also enable them to serve society on the broadest scale by their incorporation into sectors such as medical/health, energy, transportation, communications, and entertainment. HONGXING JIANG JINGYU LIN

CHAPTER ONE

Development of nitride microLEDs and displays Hongxing Jiang∗ and Jingyu Lin Department of Electrical and Computer Engineering, Texas Tech University, Lubbock, TX, United States *Corresponding author: e-mail address: [email protected]

Contents 1. Introduction 2. MicroLED structure and processing 2.1 MicroLED structure and pixel processing 2.2 Fabrication of the first passive-matrix microdisplay 3. MicroLED array for high voltage AC/DC-LEDs 3.1 Monolithic single-chip high-voltage AC/DC-LEDs 3.2 Heterogeneous integrated high-voltage DC/AC LEDs 4. Realization of the first full-scale active driving microLED microdisplay 4.1 Hybrid active driving microLED microdisplays 4.2 Unique features of CMOS IC driver for microLED microdisplays 5. MicroLED and microLED display characteristics 5.1 Brightness characterization 5.2 Temperature dependence 5.3 Operating speed and view angle characterization 5.4 Power conversion efficiency characteristics 6. Full color microdisplay development 6.1 Standard side-by-side RGB sub-pixel approach 6.2 Vertically stacked RGB microLED microdisplay concept 6.3 Demonstration of full color microdisplays based on vertically stacked RGB microLEDs 7. Applications in large flat panel displays and medicine 8. Concluding remarks Acknowledgments References

1 6 6 9 13 13 15 19 19 22 25 25 28 28 31 36 36 38 44 47 50 50 51

1. Introduction MicroLED is currently recognized as the ultimate display technology and is one of the fastest-growing technologies in the world. Global tech giants are poised to utilize microLED on a wide-ranging products from Semiconductors and Semimetals, Volume 106 ISSN 0080-8784 https://doi.org/10.1016/bs.semsem.2021.01.006

Copyright

#

2021 Elsevier Inc. All rights reserved.

1

2

Hongxing Jiang and Jingyu Lin

wearable displays, high speed three-dimensional/augmented reality/virtual reality (3D/AR/VR) displays, pen-projectors, ultrahigh-definition and high brightness/contrast large flat panel displays and TVs, to light source for neural interface and optogenetics as well as for visible light communications (Li-Fi) (Lin and Jiang, 2020). A few examples of these recent developments are illustrated in Fig. 1. The huge opportunity in consumer electronics is one of the major driving forces behind the recent advances of innovative technologies and products based on microLED. According

Fig. 1 A few examples of microLED applications: (A) smart watches; (B) smart glasses; (C) 3D/AR/VR displays; (D) smart phones; (E) dashboard and pico-projectors; (F) light sources for neutral interface—neural cells expressing ChR2 are covered by a 64
64 matrix of μLED array with individual control of their intensity and timing (inset, scale bar, 30 μm). (G) Extremely high-performance large TVs; and (H) miniature personal computers. Panel (F): Reproduced from Grossman, N., Poher, V., Grubb, M.S., Kennedy, G.T., Nikolic, K., McGovern, B., Berlinguer, P.R., Gong, Z., Drakakis, E.M., Neil, M.A., Dawson, M.D., Burrone, J., Degenaar, P., 2010. Multi-site optical excitation using ChR2 and micro-LED array. J. Neural Eng. 7, 016004; Copyright (2010) IOP Publishing.

Development of nitride microLEDs and displays

3

to Business Wire, “Micro-LED market worldwide is projected to grow by US$18.7 Billion, driven by a compounded growth of 78.8%, during 20192025” (https://www.businesswire.com/news/home/20200228005197/ en/Global-Micro-LED-Market-Analysis-Trends-Forecasts-2019-2025). Currently, “almost all the big names in the tech industry see MicroLED as the next big thing” (https://www.ledinside.com/news/2018/2/micro_led_ vs_oled_competition_between_the_two_display_technologies; News Article, 2018) and researchers are racing to overcome key technical barriers to bring microLED products to the market. The concept of microLED (μLED) was first conceived (Fan et al., 2008; Jiang et al., 2001, 2002; Jiang and Lin, 2001, 2003, 2013; Jin et al., 2000a,b, 2001) during the period of rapid development of solid-state lighting based on blue/white LEDs after the invention of III-nitride blue LEDs in the early 1990 (Akasaki, 2015; Akasaki and Amano, 2006; Amano, 2015; Nakamura, 2015; Nakamura et al., 2000). The initial development of microLED was inspired by the well-known evidence from traditional III-V semiconductors that optoelectronic devices including emitters and detectors with micro-cavities possess unique advantages such as low power consumption, high quantum efficiency, enhanced speed, reduced lasing threshold, ability of miniaturization, 2D array formation and integration, and reduced cost (Chang and Campillo, 1996; Iga, 2018; Yamamoto and Slusher, 1993). Various III-nitride microstructures, including microdisks, rings, pyramids, prisms, waveguides, and optically pumped vertical cavity surface emitting lasers (VCSEL) have been successfully fabricated and studied by several research groups prior to 2000 (Bidnyk et al., 1998; Chang et al., 1999; Dai et al., 2001a,b; Jiang et al., 1999; Li et al., 2000; Mair et al., 1997, 1998; Martin et al., 2001; Someya et al., 1999; Zeng et al., 1999a,b). Enhanced quantum efficiencies, optical resonant modes and optically pumped lasing actions were observed in GaN microdisks, rings, pyramids as well as in GaN VCSEL structures (Bidnyk et al., 1998; Chang et al., 1999; Dai et al., 2001a,b; Jiang et al., 1999; Li et al., 2000; Mair et al., 1997, 1998; Martin et al., 2001; Someya et al., 1999; Zeng et al., 1999a,b). The question naturally arises: What unique features will a micro-sized LED and array possess? Although the device architecture of microLED itself is much simpler than that of a VCSEL (Iga, 2018), back then, the nitride research community faced two major technical challenges for realizing blue/green microLEDs. The first issue was the relatively poor p-type conductivity and hence relatively high material resistivity due to the

4

Hongxing Jiang and Jingyu Lin

well-known large Mg acceptor energy level of about 160 meV in GaN (Akasaki, 2015; Akasaki and Amano, 2006; Amano, 2015; Nakamura, 2015; Nakamura et al., 2000), which becomes more problematic as the LED size reduces to micron scale. The second issue was the ratio of etched region to active area increases with a decrease in the microLED size, which tends to enhance the non-radiative recombination of injected electrons and holes due to plasma etching induced damage (Cao et al., 1999) to the sidewalls of μLEDs. As such, the transition from basic research to the realization of the first current injection microLED shown in Fig. 2 was made possible only after these two issues were overcome to some extent ( Jiang et al., 2001, 2002; Jin et al., 2000a,b, 2001). By virtue of its name, microLEDs are microns in size, whereas conventional broad-area LEDs have a dimension of 0.3 mm
0.3 mm or larger. Therefore, displays based on microLEDs offer much higher special resolution than displays based on conventional LEDs. On the other hand, the intrinsic properties of inorganic semiconductor LEDs—self-emissive, high brightness with near perfect black levels, extremely high contrast ratio, high efficiency and hence low energy consumption, wide view angles, and high thermal and mechanical robustness, plus high-resolution and fast operating speed offered by microLEDs, make microLEDs ideal for high 5-20 µm

(b)

(a)

p-contact

p-type GaN

n-type GaN III-nitride buffer Sapphire substrate

EHT = 15.00 KV

optical active media InGaN/GaN MQWs

(c)

Ni/Au p-contact 10 µm

n-type GaN

μ-LED

10 µm

Detector = SE1

Fig. 2 The first current injection GaN microLED: (A) schematic layer structure diagram of an InGaN/GaN MQW wafer for microLED (μLED) array fabrication; (B) schematic illustration of an InGaN/GaN MQW microLED (μLED) array; and (C) SEM image of a fabricated InGaN/GaN MQW μLED array with a μLED pixel diameter of 12 μm and p-type Ni/Au contact diameter of 10 μm. Reproduced from Jin, S.X., Li, J., Li, J.Z., Lin, J.Y., Jiang, H.X., 2000. GaN microdisk light emitting diodes. Appl. Phys. Lett. 76, 631; Copyright (2000) AIP Publishing.

Development of nitride microLEDs and displays

5

dynamic range (HDR) and three-dimensional/augmented reality/virtual reality (3D/AR/VR) display applications. Another important application of microdisplays is for head-up displays (HUD). The drivers of modern vehicles are confronted with an increasing workload in terms of information processing. To meet the demand for a situation-adapted, non-distracting, fast perceptible information display, HUDs can provide some advantages compared to instrument clusters. In a driving situation using head-up display, the driver would see a virtual image reflected from the windshield into his/her eyes, which provide both less distraction from the road scene and less fatigue over driving periods and hence an improved safety ( Jiang and Lin, 2003). On the other hand, a head-up or head wearing microdisplay would not only provide a pilot with vital information, linking them to the aircraft’s systems as well as to their rapidly changing environment, but would also provide hands-free capability, all of which can greatly enhance his/her ability in making split-second decisions and actions that can determine the success or failure of a mission ( Jiang and Lin, 2003). A recent exciting development is the applications of microLEDs for large flat panel displays and TVs with extremely high performances. The first large screen microLED TV was successfully developed by Sony in 2012, named Crystal LED display (https://en.wikipedia.org/wiki/Crystal_LED). Sony’s large-scale microLED TV demonstrated the validity of microLED technology and instigated a furious race in the display industry to bring microLED large flat panel display products to the global consumer electronics markets. To improve the manufacturing yield, the next large microLED TV was a 146-in. μLED TV incorporated a method of combining an array of smaller modules (or panels) to form a large flat panel display, which was showcased with the brand name of “The Wall” by Samsung in 2017 (https://www.samsung.com/us/business/products/displays/direct-view-led/ the-wall/). The most recent microLED large displays including the one showcased by Sony in the 2019 National Association of Broadcasters Show has a format of 16K and a dimension as large as 21m
5.5 m; another 14500 modular MicroLED TV showcased at Consumer Electronics Show (CES) 2020 by LG. These large μLED TVs however still retain all the intrinsic outstanding features of inorganic semiconductor LEDs. Another area in which microLEDs can make a significant impact is for neural interface/optogenetics (Coffey, 2018; Degenaar et al., 2009; Grossman et al., 2010; Gutruf and Rogers, 2018; Habermann et al., 2019; Kim et al., 2013; Klein et al., 2018; McAlinden et al., 2019; Mickle et al., 2019; Poher et al., 2008; Reddy et al., 2019; Scharf et al., 2016; Wu et al., 2015). MicroLEDs also appear

6

Hongxing Jiang and Jingyu Lin

to be particularly suitable for LiFi applications (Carreira et al., 2019; Dawson and Neil, 2008; Hass, 2018; McKendry et al., 2009; Rae et al., 2018). This chapter provides a brief overview on the developments of III-nitride microLEDs. Section 2 describes the early developments and basic III-nitride microLED material structures, fabrication processes and characteristics. Section 3 discusses the development of single-chip high voltage AC- and DC LEDs based on microLED array technology for general illumination and solid-state lighting applications. Section 4 reviews earlier development results of passive microLED microdisplays. In Section 5, the realization of the first active driving InGaN full-sale high-resolution (640
480 pixels) monochrome microdisplay in video graphics array (VGA) format capable of delivering video graphics images is reviewed. In Section 6, recent progresses toward the attainment of full color microdisplays for the realization of small form factor of “wearable” AR/VR smart glass displays are examined. Section 7 provides a brief overview on the prospective uses of microLEDs in areas of large flat panel displays and medicine. In Section 8, we provide a brief general summary and outlook of the microLED technology.

2. MicroLED structure and processing 2.1 MicroLED structure and pixel processing The device layer structure for microLED fabrication is generally the same as those of conventional LEDs. For instance, the wafer structure for the first microLED demonstration is schematically shown in Fig. 2A. Same as the conventional III-nitride blue/green LEDs, the microLED structure was grown on sapphire substrate by metal-organic chemical vapor deposition (MOCVD). As illustrated in Fig. 2A, the LED structure is based on InGaN/GaN multiple quantum wells (MQWs) confined between n-type and p-type GaN carrier injection layers. The emission wavelength can be tuned by adjusting the compositions of In in the MQW active region. For microLEDs operating in the UV wavelengths, the MQW active region contains AlInGaN alloys and n- and p-AlGaN are the carrier injection layers. More specifically, the MQW LED wafer used to fabricate the first microLED array comprised 3.5 μm of Si-doped GaN, 0.1 μm of Si-doped ˚ of AlGaN/GaN, superlattice consisting of alternating layers of 50 A˚/50 A ˚ ˚ ˚ a 50 A of Si-doped GaN, 30 A/30 A undoped InGaN/GaN MQW active layer, 0.14 μm of Mg-doped superlattice consisting of alternating layers of

Development of nitride microLEDs and displays

7

˚ of AlGaN/GaN, and 0.4 μm Mg-doped GaN epilayer ( Jin et al., 50 A˚/50 A 2000a). The LED structure was treated by a rapid thermal annealing at 950 °C for 5 s in nitrogen. This process produced a hole concentration in the p-layers of 5
1017 cm3 (with a hole mobility of 12 cm2/V s) and an electron concentration in the n-layers of 1.6
1018 cm3 (with an electron mobility of 310 cm2/V s) ( Jin et al., 2000a,b). By incorporating the AlGaN/GaN superlattice structure into the LED device layers, the hole concentration in the p-layers was enhanced from 2
1017 to 5
1017 cm3. A schematic diagram and scanning electron microscope (SEM) image of the first microLED array fabricated in the authors’ laboratory are shown in Fig. 2B and C, respectively ( Jin et al., 2000a). As inductive-coupled plasma (ICP) is an established technique for III-nitride device pattern transfer (Cao et al., 1999), ICP dry etch was used to form the microLED array. As shown in Fig. 3A and B, the microLED has a mesa structure with the n- and p-contacts fabricated on the same side because of the insulating sapphire substrate. The process starts from plasma etch down to n-GaN to form the circular mesa (p-GaN and MQW), then another ICP etch is taken down to sapphire for p-pad. As the microLED has a size less than 20 μm, the p-pad must be built on the sapphire substrate due to space limitation. Ti/Al bilayers were deposited with e-beam evaporation on n-GaN to form n-type ohmic contacts with thermal annealing at 600 °C. An insulation layer (SiO2) was then deposited to isolate n-GaN from the p-contact metals. Ni/Au bilayers were deposited on p-GaN to form p-type contacts with thermal annealing at a temperature of 550 °C. After thermal annealing in air, the Ni/Au alloy formed transparent contact on p-GaN layer with a transmission of 70%–75% at visible wavelengths. The last step of surface passivation layer of SiO2 finishes the device fabrication. Fig. 3C shows an atomic force microscope (AFM) image of a fabricated microLED ( Jin et al., 2001). As can be seen from Fig. 3C, the p-type contact was connected to the top p-GaN layer by opening a hole through the insulating dielectric layer. The size of the p-type contact shown in Fig. 3C is about 4 μm in diameter. Fig. 3D shows an optical microscope image taking from the top (p-type contact side) of a representative InGaN/GaN MQW microLED with a diameter d ¼ 12 μm in action. For a direct comparison, Fig. 3E shows an optical microscope image of a conventional indicator LED in action. A conventional indicator LED has a typical chip area of 300 μm
300 μm and is generally encapsulated into the standard 3 mm or 5 mm lamp. A natural subsequent step was to implement various schemes to address μLEDs within a μLED array to create practical devices. Quite distinct from

(b)

(a)

(c) 1.564μm μ LED

0 5μm

15μm 10μm

Insula ng layer

10μm

5μm

15μm 0

(e)

(d)

μLED in action

Conventional LED in action

Fig. 3 (A) The top schematic view and (B) the cross-sectional schematic view of a GaN microLED. (C) AFM image of a fabricated microLED with a diameter of d ¼ 12 μm with p-contact crossing over the mesa. (D) Optical microscope image of a μLED (d ¼ 12 μm) in action. (E) A conventional broad area LED (300 μm
300 μm) fabricated from the same LED wafer in action for comparison. Reproduced from Jin, S.X., Shakya, J., Lin, J.Y., Jiang, H.X., 2001. Size dependence of III-nitride microdisk light emitting diode characteristics. Appl. Phys. Lett. 78, 3532; Copyright (2000) AIP Publishing.

Development of nitride microLEDs and displays

9

those of conventional III–V LED structures, the highly electrically insulating sapphire substrate used in III-nitride LED structures naturally provides an ideal platform for the isolation between individual microLEDs within an array. Moreover, the transparent sapphire substrate also conveniently serves as a nature surface for output coupling, reducing the steps for device packaging. For LED structure grown on conductive SiC or Si substrates, the isolation approach is to incorporate a layer of AlN epilayer template between the substrate and LED structure (Li et al., 2006). Very quickly, many potential applications and products started to emerge for μLEDs and arrays. One of which is a μLED array with independently addressed pixels or microLED microdisplay (μdisplay), which was first introduced by the authors’ group between 2000 and 2001 ( Jiang et al., 2001, 2002). One other product resulted was by connecting a number of μLEDs in series so that the sum of the voltage drop across the individual μLEDs adds up to the voltage of a high-voltage AC or DC supply to create single-chip high-voltage AC/ DC-LEDs to match the infrastructure for lighting (Fan et al., 2007, 2009, 2010, 2012; Jiang and Lin, 2007; Jiang et al., 2005).

2.2 Fabrication of the first passive-matrix microdisplay Passive driving monolithic μLED microdisplays were first developed for low information content microdisplays, in which the μLEDs themselves and the interconnection between these μLEDs (the signal transmission paths, including all the metal lines for n- and p-type contacts) are all integrated on the same GaN wafer. This monolithic integration has the merits of providing easy and quick demonstration and characterization, since the microdisplay itself is an independent package, and the driving circuit can be designed using an off-the-shelf complementary metal-oxide-semiconductor (CMOS) integrated circuit (IC) chip. Passive driving by X-Y scanning has been widely employed because this approach is convenience and low cost. As shown in Fig. 4, the first prototype passive III-nitride blue microdisplay consisting of 10
10 pixels with a pixel size of 12 μm in diameter was soon conceived ( Jiang et al., 2001, 2002). To obtain a working device, a dielectric layer was deposited above the etch-exposed underneath n-type GaN layer to isolate the p-type contacts from the n-type layer. Also shown in Fig. 4A are conducting wires used to make the connection between the n-type ohmic contacts and the contact pads which are used for current injection into n-type ohmic contacts. There are also conducting wires to make the connections between individual pixels through the top p-type ohmic

10

Hongxing Jiang and Jingyu Lin

n-contact pads

(a)

Insula ng layer

p-contact pads

0.5 mm

n-contact grid micro-size light emi ers/detector

(b)

250 μm Fig. 4 The first demonstrative GaN microLED microdisplay: (A) optical microscope image of a bonding scheme that allows us to address each microLED pixel individually to enable the realization of a passive driving microdisplay. (B) Optical microscope images (top view) of a blue microdisplay in action, displaying letters “KSU.” Reproduced from Jiang, H.X., Jin, S.X., Li, J., Lin, J.Y., 2002. Micro-size LED and detector arrays for mini-displays, hyperbright light emitting diodes, lighting, and UV detector and imaging sensor applications. US Patent 6,410,940 (filed June 15, 2000); Jiang, H.X., Jin, S.X., Li, J., Shakya, J., Lin, J.Y., 2001. III-nitride blue microdisplays. Appl. Phys. Lett. 78, 1303, US patent 6,410,940; Copyright (2001) AIP Publishing.

contacts and the pixel control pads which are used for current injection into p-type ohmic contacts. Each pixel has its own control pad. In this array, the state of the pixels can be individually controlled. The operation of these prototype InGaN/GaN MQW microdisplays has been demonstrated. Fig. 4B shows the optical microscope images of a blue microdisplay of Fig. 4A in action, displaying a sequence of letters of “KSU” ( Jiang et al., 2001, 2002). Soon after, several research groups were engaged very early on in pursuing the further developments of microLED technology (Choi et al., 2003, 2004; Lau et al., 2013; Liu et al., 2009, 2013; Ozden et al., 2001).

Development of nitride microLEDs and displays

11

A matrix-addressing scheme was developed to demonstrate a passive-matrix microdisplay with 128
96 pixels (Choi et al., 2004). The concepts of flip-chip bonded microdisplays as well as μLEDs on Si substrates have also been developed (Lau et al., 2013; Liu et al., 2009, 2013). As illustrated in Fig. 5A-C, in a passive driven design, each μLED pixel relates to anode on the column data line and cathode on the row scan line (Fan et al., 2008). When a specific row is scanned, the brightness of each pixel on this row is decided by its column data line which provides a current corresponding to the pixel brightness (gray scale) to turn this pixel on. Each row of pixels is isolated from each other by the deep trench of about 2 μm in width and 3–5 μm in depth. Since the chemical inert nature of III-nitride materials, plasma etch based on chlorine chemistry is generally used, with a typical etch rate of 0.5 μm/min by ICP etching. It takes several minutes to etch through GaN down to sapphire. Such long-time highintensity plasma etch has adverse effects on the surface morphology and etching profile. Surface damage will enhance the nonradiative recombination rate, especially for the microLED with a large etched surface area as compared to its bulk volume. Special steps for surface treatments, including chemical wet etching after plasma etch to remove the damage and compensate the dangling bonds, and oxide/polymer surface protection, have been adopted in the process. Controlling the etch profile is also critical. As an example, with a 45° profile and an etch depth of 3 μm, a nominal 15 μm microLED will have a top diameter of only 9 μm, and the light emitting area is reduced by around 60%! For high resolution microdisplays with even smaller pixel size, the emitting area reduction ratio becomes even higher. This etching profile clearly is not acceptable. By specifically selecting the proper etching mask and plasma etching condition, we were able to increase the etching profile angle to more than 80°, as demonstrated in Fig. 3C ( Jin et al., 2001). To connect a μLED array with hundreds of pins to passive driver/ controller IC chips under the restriction of minimizing the microdisplay module size, it is beneficial to adopt a flip-chip bonding approach (Lau et al., 2013; Liu et al., 2009, 2013). For conventional LEDs, typically the light is emitted from the top GaN surface, and the ohmic contact metal layer occupies a fraction of the p-GaN mesa. For microLEDs, not only the lighting area is dramatically reduced, but the metal contact area is also small, corresponding to a large resistance. A flip-chip approach allows light emitting from the sapphire side, so the whole area of the p-type GaN surface can be occupied by the metal layer. Fig. 5D shows the optical image of a

p-line (metal)

p-pad

p-GaN

p-line

n-line

Transparent metal layer

p-line Polymide

MQW SiO2 SiO2

n-pad

n-line

sapphire sapphire

(a)

(b)

(c)

(d)

Fig. 5 Schematic diagrams of a passive driven GaN microdisplay: (A) the top view; (b) the cross-sectional view; (C) the cross-sectional view after surface planarization. (d) The optical microscopy image of a flip-chip bonded 128
128 microdisplay. The zoom-in image shows the detail of the microdisplay observed from sapphire side with one pixel turning-on. Each microLED pixel has a diameter of 18 μm. Reproduced from Fan, Z.Y., Jiang, H.X., Lin, J.Y., 2008. III-nitride micro-emitter arrays: development and applications. J. Phys. D: Appl. Phys. 41, 094001, Copyright (2008) IOP Publishing.

Development of nitride microLEDs and displays

13

flip-chip bonded microdisplay with 128
128 pixels and a pixel pitch of 22 μm fabricated in authors’ laboratory. In this device, a chip-on-glass (COG) driver and μLED array devicse both were directly bonded on board through gold bumps, and the connections between driver and μLED array were implemented through the photolithography defined metal wires on the board. An off-the-shelf OLED segment/common driver with controller was used to drive the μLED array (Fan et al., 2008). However, these passive-matrix microdisplays are not capable of delivering video graphics images. In the passive driving mode, one can only independently access one row at a time. Furthermore, the addressing time of each pixel, or the time one pixel is in the “on” state, is inversely proportional to the number of lines (rows) in the display matrix. The result is that under pulse driving, the brightness of the pixel is only 1/R (R defines the number of the rows) of the brightness under the equivalent DC current driving (Fan et al., 2008). In other words, to keep the same brightness, the driving current must be increased by R times. For high-information-content (highresolution) video displays, or for very high luminance-requirement sunlight readable displays, the maximum driving current limit is reached, and the light output saturates with further increase in current; hence the light efficiency and thermal dissipation become serious issues. Furthermore, passive displays address each pixel for only a short time and then moves on to address the other pixels in the display. When the pixel is not being addressed, it does not have current flowing through it (i.e., it does not light up). Therefore, for high-information-content displays, the desired driving approach is active matrix driving to be discussed in Section 4.

3. MicroLED array for high voltage AC/DC-LEDs 3.1 Monolithic single-chip high-voltage AC/DC-LEDs The inventions of single-chip high-voltage AC/DC-LEDs were also evolved from on-chip integration of microLED or mini-LED arrays (Ao et al., 2002; Fan et al., 2007, 2009, 2010, 2012; Jiang and Lin, 2007; Jiang et al., 2005; Kal et al., 2010; Lee et al., 2011; Sakai et al., 2008; Yen et al., 2007), which are capable to fully address the key compatibility issue of LEDs with the power grid infrastructure. As of today, GaN single-chip high-voltage AC/ DC-LEDs have been widely commercialized for general illumination as well as for automobile headlight applications.

14

Hongxing Jiang and Jingyu Lin

The insulating nature of sapphire substrate allows serial connection between μLEDs. Fig. 6 illustrates the basic principle on how to build single-chip high-voltage AC/DC LED devices ( Jiang and Lin, 2007; Jiang et al., 2005). The number of linked mini-LEDs is chosen so that the sum of the voltage drops across the individual micro- (or mini-) LEDs adds up to the high voltage of the AC/DC supply. Since LEDs only emit light when they are forward biased, two arrays are created for AC operation, one of which lights up during the first half cycle of the AC power source and the other of which lights up when the polarity of the source is reversed, as illustrated in Fig. 6A. It is obvious that a single array of Fig. 6B can operate under a high-voltage DC source, as needed in application areas such as for automobile headlights. The advantages of single-chip high-voltage LEDs for automobile headlights include simple design, fabrication, and packaging as well as high yield and low cost. Fig. 6C shows the schematic illustration of the layer structure of a single μLED fabricated on sapphire substrate. In order to achieve high voltage operation, a certain number of μLEDs are connected in series, meaning that

+ V

-

V

(a) (c)

( b) +

(d)

-

p-contact p-GaN MQW

n-contact

n-GaN Sapphire substrate

Sapphire substrate

Fig. 6 Schematic conceptual circuit diagrams on how to build (A) a single-chip high-voltage AC-LED device and (B) a single-chip high-voltage DC-LED device from a microLED array; (C) the cross-sectional view of a low voltage DC LED; (D) the cross-sectional view of a single-chip high-voltage AC/DC LED achieved via monolithic integration of plurality of μLEDs by connecting the p-contact of one μLED to the n-contact of its neighboring μLED. Reproduced from Jiang, H.X., Lin, J.Y., Jin, S.X., 2005, 2007. Light emitting diodes for high AC voltage operating and general lighting, US Patents 6,957,899 and 7,210,819; Jiang, H.X., Lin, J.Y., 2007. Light emitting diodes for high AC voltage operating and general lighting. US Patent 7,213,942.

Development of nitride microLEDs and displays

15

the p-contact of one μLED is connected with the n-contact of its neighboring μLED, as illustrated in Fig. 6D (Fan et al., 2007, 2009, 2010, 2012; Jiang and Lin, 2007; Jiang et al., 2005). The serial integration of μLED array on-chip avoids multiple soldering points and thus reduce chip failure rate. Furthermore, with two reverse LED arrays corresponding to the current flow of the positive and negation half-cycle of AC voltage, the light on/off frequency is doubled from 50 to 60 Hz AC frequency to 100–120 Hz, and the effect of light flicking is minimized. Fig. 7A shows the top view of a monolithic single-chip high-voltage AC-LED that integrates multiple mini-LEDs ( Jiang et al., 2005). In this integration scheme, a conventional LED is replaced by an array of μLEDs for 110/120 VAC for lighting applications. Other design variations such as Wheatstone bridge type of circuits have also been employed to build monolithic AC-LEDs on single chips (Yen et al., 2007). Nevertheless, the fundamental design principle for building high-voltage DC/AC LEDs is based on Ohm’s law and illustrated in Fig. 6. Fig. 7B is a photo of several packaged 5 mm indicator high-voltage AC-LED lamps (each with an overall chip size of 0.3 mm
0.3 mm) plugged into the 120 VAC power outlets without power converters. These AC indicator LED lamps are very suitable for uses in indication, signage, night lamps and holiday decorative tree lighting (Fan et al., 2009). Since each LED lamp runs with a current of no more than 0.5 mA, there is almost no limitation on the number of AC-LED lamps connected in the tree lighting string, and the LED string can be directly plugged into the house-hold AC power supply outlets without transformer or rectifier (Fan et al., 2009; Jiang and Lin, 2007; Jiang et al., 2005).

3.2 Heterogeneous integrated high-voltage DC/AC LEDs Further improvement to the monolithic integration scheme can be made by heterogeneously integrating μLED array with a passive/active submount through flip-chip bonding (Fan et al., 2007). Fig. 7C shows the crosssectional view of a flip-chip bonded high-voltage AC/DC LED device with the interconnection between individual μLEDs taking place on the submount (Fan et al., 2007). The submount materials can be chosen so they are electrically insulating and thermally conductive. The submount can be fabricated to contain flip-chip bumps, insulating and metal layers to serially interconnect the μLED array, current limiting resistors and other control

Fig. 7 (A) Top view of a single-chip high-voltage AC-LED via monolithic integration of plurality of μLEDs. (B) Photo of several high-voltage indicator AC-LEDs with a chip size of about 0.3 mm
0.3 mm directly plugged into the 120 VAC power outlets without power converters. (C) The cross-sectional view of a heterogeneous integrated high-voltage AC/DC LED achieved via flip-chip bonding with the interconnection between each individual μLEDs through a submount. (D) Photo of a high-voltage power AC-LED with a chip size of about 1 mm
1 mm directly plugged into a 120 VAC power outlet without power converter. Panel (A): Reproduced from Jiang, H.X., Lin, J.Y., Jin, S.X., 2005, 2007. Light emitting diodes for high AC voltage operating and general lighting. US Patents 6,957,899 and 7,210,819. Panel (C): Reproduced from Fan, Z.Y., Jiang, H.X., Lin, J.Y., 2009. Micro-LED based high voltage AC/DC indicator lamp. US Patent 7,535,028.

Development of nitride microLEDs and displays

17

and driver circuits (Fan et al., 2007). The final device has two or more outlet connections for the supplied power. Depending on the detailed design, the supplied power may be 12 V, 24 V, and other DC voltages, or it may be AC voltages such as 110/120 V, and 220/240 V. The heterogeneously integrated high voltage DC/AC design shown in Fig. 7C (Fan et al., 2007) provides several improvements comparing with the monolithic integration approach illustrated in Figs. 6D and 7A (Fan et al., 2009; Jiang and Lin, 2007; Jiang et al., 2005). By moving more metal layers from the microLED array die to the submount, LED chip fabrication is easier, and the production yield and device reliability will improve. Second improvement is the enhancement of heat dissipation by reducing the thermal resistance. LED performance and lifetime strongly depend on the p-n junction temperature. Furthermore, with flip-chip bonding, now the light is extracted from the transparent substrate side. In the design shown in Figs. 6D and 7A, light is extracted from the nitride device side. The n-contact, p-contact, current spreading layer, and interconnection layer, will partially block or absorb the light extraction. For the heterogeneous integration design in Fig. 7C, since there are no metal contacts or layers on the substrate side, light absorption is avoided, and the light efficiency is increased. For III-nitride LED wafers grown on the sapphire substrates, single-chip high-voltage AC/DC-LEDs are fully compatible with the traditional lowvoltage DC LED production lines. Both the conventional low-voltage DC-LEDs and high-voltage AC/DC-LEDs have the n- and p-contacts fabricated on the same side of the LED wafer due to the electrically insulating nature of sapphire substrate and require plasma dry etching to etch into n-GaN epilayer to form the LED mesa. The sapphire substrate provides a natural base for the isolation of individual microLEDs. Therefore, the chip cost for both low-voltage DC-LEDs and high-voltage AC/DC-LEDs is very similar. However, the major cost saving for high-voltage AC/DC-LEDs will be in the packaging because of the elimination of transformers and other electronics drivers. On the other hand, to fabricate high-voltage AC/ DC-LEDs from LED structures grown on conductive SiC or Si substrates, the isolation between μLEDs can be achieved by incorporating a layer of insulating material between the substrate and LED layer structure. This may require modification to the existing LED layer structures. The highly insulating AlN can provide an ideal solution for μLEDs isolation for LED structures grown on conductive SiC or Si substrates (Fan et al., 2012; Li et al., 2006).

18

Hongxing Jiang and Jingyu Lin

For lighting applications, typical LED chip size is 1 mm
1 mm or greater (named power LEDs). Each mini-LEDs within the power AC-LED array has a similar chip size as the traditional indicator DC-LEDs and experiences the same operating voltage and current density as the traditional DC-LEDs. However, because AC-LEDs are effectively operating under a pulsed mode (60 Hz) and are in the “on” state only for a half of the cycle, it is expected that AC-LEDs can be designed to handle higher current density than the traditional DC-LEDs. Another advantage of the integration approach for obtaining high-voltage AC/DC LEDs is the flexibility of incorporating a protection mechanism on the chip level by limiting the influence of the line voltage variation on the current (Fan et al., 2010). The protection element can be directly integrated on the AC/DC LED chip or it may be integrated on the submount of the flip-chip bonded AC/DC LED device and has the function of reducing or easing the influence of voltage variation on AC/DC LED device performance and lifetime. It is very interesting to realize that a current-limiting resistor can be monolithically integrated on the same AC/DC LED chip and fabricated from the same LED wafer (Fan et al., 2010). Fig. 7D is a photo of a packaged power AC-LED directly plugged into the 120 VAC power outlet without the need of a power converter. A power AC-LED lamp runs under a 120/220 VAC with a current around 20 mA is capable to achieve a high brightness level for general lighting application. One can further integrate plurality of high voltage AC/DC-LED chips on a submount to form a compact AC/DC-LED lamp for ultrahigh brightness applications (Fan et al., 2012). It is important to note that compared to DC-LEDs, no optical power is lost during the AC cycle for AC-LEDs ( Jiang and Lin, 2013). The effective optical output power of an LED under AC operation is the integrated power over a half-period in a 60 Hz cycle, which can be evaluated according to Rπ P ave ¼ π1 0 P ðtÞdt , where the LED optical power is written in terms the forward voltage (V) and current (I) according to the diode equation,  qðV IRÞ  nkT Poutput ¼ CI 0 e  1 , and the voltage produced by an AC power supply is sinusoidal, V ¼ Vo sin 2πft ¼ Vo sin ωt, with the root-mean-square voltage being Vrms ¼ 0.707 V0 ( Jiang and Lin, 2013). In the single-chip AC-LED design shown in Figs. 6 and 7, Vrms is divided among n microLEDs, so Vrms across each microLED will be the same as a traditional DC-LED. By evaluating the above integral, it can be shown that the

Development of nitride microLEDs and displays

19

effective (or average) optical power output of an LED under AC operation is comparable to the output power under DC operation. This is because the diode behavior (turn-on behavior) is embedded in the response ( Jiang and Lin, 2013).

4. Realization of the first full-scale active driving microLED microdisplay 4.1 Hybrid active driving microLED microdisplays In order to realize the full potentials of III-nitride μLED microdisplays, developing schemes to actively and energy efficiently drive the μLED array is essential, since these devices are expected to operate on battery power and the efficiency of all components in the device package directly affect how long it can operate between charging. For high resolution microdisplays, an active driving scheme in which each pixel is geared with its own driver is much preferred to ensure the frame refreshing. The real breakthrough in μLED microdisplay was achieved in 2011 by the authors’ group by demonstrating full-sale high-resolution monochrome blue and green microdisplays in video graphics array (VGA) format (with 640
480 pixels) capable of delivering video graphics images, which was accomplished through the incorporation of an active matrix driving scheme (Day et al., 2011, 2012, 2015; Lin et al., 2011). The challenge for achieving μLED based μdisplay with active driving has been that III-nitride μLEDs cannot be fabricated directly over Si IC circuitry. To overcome this difficulty, we have employed the hybrid μLED microdisplay concept, like the widely deployed scheme of hybrid focal plane array detectors (Lamarre et al., 2001), which utilizes the technique of flip-chip bonding via indium metal bumps. However, the typical size of indium metal bumps used in hybrid focal plane array detectors is around 20μm. Fig. 8A shows a detailed cross-sectional view of a high-resolution microdisplay with pixels arranged in a matrix format (Day et al., 2015). The InGaN microLED array is integrated on Si CMOS driver chips by using flip-chip bonding or aligned wafer bonding. Flip-chip bonding is based on metal bumps. Fig. 8B shows a schematic illustration of a highresolution microdisplay which consists of an array of InGaN microLEDs as pixels arranged into matrix format on a sapphire substrate and the μLED array is heterogeneously integrated on to a Si CMOS IC driver chip

Fig. 8 Schematics of cross-sectional views of a high-resolution microdisplay with microLED pixels arranged in a matrix format: (A) the InGaN microLED array is integrated on Si CMOS driver IC by using flip-chip bonding or aligned wafer bonding. (B) Illustration of flip-chip bonding between μLED matrix array and CMOS driver IC via indium bumps to form a highly integrated microdisplay in one package. (C) The optical microscopy image of a segment of an InGaN μLED microdisplay chip showing a flip-chip bonded package with μLED pixels of 12 μm
12 μm in size and indium bumps of 6 μm in diameter, viewed from the transparent sapphire side. Panel (A): Reproduced from Day, J., Li, J., Lie, D.Y.C., Fan, Z.Y., Lin, J.Y., Jiang, H.X., 2015. CMOS IC for micro-emitter based microdisplay. US Patent 9,047,818. Panels (B) and (C): Reproduced from Day, J., Li, J., Lie, D.Y.C., Bradford, C., Lin, J.Y., Jiang, H.X., 2011. III-nitride full-scale high-resolution microdisplays. Appl. Phys. Lett. 99, 031116; Copyright (2011) AIP Publishing.

21

Development of nitride microLEDs and displays

using indium bump bonding (Day et al., 2011, 2012, 2015; Lin et al., 2011). The polished back surface of the transparent sapphire substrate is used to display images. As shown in Fig. 8A, the pixels share a common cathode (n-type contact) with independently controllable anode (p-type contact). The hybrid integration of the InGaN microLED array with the Si CMOS driving circuit IC chip means that close to millions of the signal connections between the microLED array and the driving circuit have been accomplished in a single flip-chip bonding package through the indium metal bumps. This active matrix display also means that each pixel is geared with its own pixel driver circuit in CMOS that can store data and driving each individual μLED. The interface requirements of this hybridized package are thereby reduced to a few lines required for the signal and power connections. An expanded image under an optical microscope view of a segment of an individually diced μLED array chip is shown in Fig. 8C, which illustrates in more details of the fabricated devices. MicroLED pixels and indium metal bumps fabricated by thermal evaporation on the μLED pixels viewed from the transparent sapphire side are clearly seen and the results demonstrated 6 μm indium bumps with excellent size uniformity. The fabricated blue and green VGA microdisplays (640
480 pixels) have a pixel size of 12 μm
12 μm and a pitch distance of 15 μm. Fig. 9A shows a fully assembled InGaN green microdisplay in action. Fig. 9B represents the input picture, whereas Fig. 9C shows the displayed image of the input picture by a (a)

(b)

(c)

9.6 mm

Fig. 9 (A) A fully assembled InGaN/GaN MQW microdisplay in VGA format (having 640
480 pixel with a pixel size of 12 μm and a pitch distance of 15 μm) operating at a driving current of 1 μA per pixel, with its size relative to a US quarter; (B) input picture of a leopard; (C) a grayscale projected image of the input picture of a leopard from a green VGA InGaN/GaN MQW microdisplay (having 640
480 pixels with a pixel size of 12 μm and a pitch distance of 15 μm) operating at a driving current of 1 μA per pixel. Reproduced from Day, J., Li, J., Lie, D.Y.C., Bradford, C., Lin, J.Y., Jiang, H.X., 2012. Jiang, Full-scale self-emissive blue and green microdisplays based on GaN micro-led arrays. Proc. SPIE 8268, 82681X; Copyright (2012) SPIE.

22

Hongxing Jiang and Jingyu Lin

green VGA InGaN microdisplay (having 640
480 pixel with a pixel size of 12 μm and a pitch distance of 15 μm). The results demonstrated that the fabricated microLED microdisplays can deliver real time video graphics images (Day et al., 2011, 2012, 2015; Lin et al., 2011). Passive displays usually address each pixel for only a short time and then moves on to address the other pixels in the display. When the pixel is not being addressed, it does not have current flowing through it (i.e., it does not light up). Given a refresh rate of 60 Hz or more, the human eye will average the brightness of each pixel so that a seemingly constant image is seen. Since each pixel is only conducting current for a small percentage of the time, a large current pulse must be used, which has an effect on the requirement for current density handling capabilities in the microLEDs as well as on the metal lines of the microdisplay driver. Active displays also address each pixel for a short amount of time before moving on to other pixels in the display. Contrary to the passive displays, however, when the pixel is not being addressed it continues to conduct the desired current until it is time to address the pixel with new data for the next frame. Since the microLED pixel is always on and there is no need for light averaging, lower driving currents can be used. This relieves the design constrains for both the microLED pixel and display driver current densities.

4.2 Unique features of CMOS IC driver for microLED microdisplays An active display driver has four primary parts (Day et al., 2015). First, the pixel data must be interpreted from the outside world and converted to an analog signal. Subsequently, the data must be stored so that the pixels can be addressed at the proper time. Horizontal and vertical shift registers are needed to step through the rows and columns of pixels. Finally, a circuit is needed that will accept the pixel data when it is being addressed and maintain the data between addressing. For a VGA display to achieve a 60 Hz frame rate the standard pixel clock frequency is 25.175 MHz, which translates to a period of only 39.7 ns. The alternative method employed in the design of CMOS IC driver for microLED array is parallel data processing (Day et al., 2015). This involves the use of eight 8-bit binary digital to analog converters (DACs). As the digital data signals come to the device, they are each fanned out to eight digital sample-and-hold

Development of nitride microLEDs and displays

23

circuits. The input clock is fed through an eight-bit shift register whose outputs control the sampling of the eight-digital sample-and-hold circuits. In this manner, each digital sample-and-hold circuit holds every eighth set of data. The output of each digital sample-and-hold circuit is the input of one 8-bit binary DAC. This means each DAC must only convert every eighth set of data, increasing the time allowed for conversion from 39.7 ns to 317.6 ns, making it implementable in the 0.6 μm CMOS using this architecture. The analog output of each DAC is sent through a buffer to the arrays of analog sample-and-hold circuits. The analog sample-and-hold circuits are aligned such that the output of one DAC buffer goes to every eighth analog sample-and-hold circuit so that the pixel data is stored in the appropriate column based on the order it is received by the device. This is a unique method for digital data conversion and sample-andholding. The CMOS pixel driver circuit design is similar to the approach used in a previous work by Biggelaar et al. for OLED displays (Biggelaar et al., 2001). A vertical shift register and a horizontal shift register are used for addressing the pixel array. The vertical shift register has the same number of bits as the height of the display, i.e., 480 bits for the full VGA device. As the shift register progresses, it enables one entire row of pixels at a time. It advances to the next row once the next row’s data has been stored in the analog sampleand-hold circuits. The horizontal shift register has the same number of bits as the width of the display, i.e., 640 bits for the full VGA device. As the shift register progresses it enables the next analog sample-and-hold circuit so that it can be programmed with the proper data. For this design there are two full rows of analog sample-and-hold circuits (Day et al., 2015). One is for sampling, based on the signal from the horizontal shift register as described. The other is used to program the pixels of the current row selected by the vertical shift register. These two rows toggle back and forth being used for sampling and addressing pixels in sync with the vertical shift register. The result of this is that each pixel is addressed for time that it takes to input an entire row of pixel data. An eight-bit shift register along with the digital sample-and-hold circuits are used to interpret the digital input clock and data so that it can be input to the DACs (Day et al., 2011, 2012, 2015; Lin et al., 2011). The analog signal is then passed to the analog sample-and-hold circuitry to be stored until needed for programming the pixel driver. The horizontal shift registers

24

Hongxing Jiang and Jingyu Lin

control which analog sample-and-hold circuit is sampling at any given time so that they are programmed in a sequential order. The vertical shift register steps through the display one row at a time, enabling that row of pixels to be addressed using the data stored in the analog sample-and-hold circuits. The display consists of an array of pixel driver circuits that can be addressed with data current for a short time, before sustaining that current in the microLED until the next addressing cycle (Day et al., 2015). CMOS active matrix 640
480 and 160
120 microdisplay controller ICs with μLED current of 0.5–10 μA were fabricated in a mature 0.6 μm CMOS process by X-Fab, Lubbock TX, USA (Day et al., 2011, 2012, 2015; Lin et al., 2011). This process has been used primarily because it can sustain the relatively higher operating voltages required for III-nitride microLEDs than those for OLEDs and with excellent cost effectiveness. Some of the unique design features of the CMOS ICs for driving microLED array include (Day et al., 2015): (1) Since microLEDs require higher operating voltages than those of OLEDs, the design of the active matrix microdisplay driver differs dramatically from those OLEDs. (2) The microLED array has a unique anode while all pixels in the array share a common cathode that is to be grounded. The CMOS IC includes the necessary circuitry to address the microLED array as an active matrix display, which includes individual pixel drivers as well as periphery circuitry required to program pixel drivers. (3) The design of the active matrix microdisplay driver can be done entirely in PMOS, or a combination of PMOS and NMOS, to account for process limitations of size, spacing and speed. The driver is capable to supply an adjustable driving current of greater than 0.01 μA per pixel for a microLED array. (4) For a typical VGA active matrix microdisplay, each sample-and-hold circuit must be programmed in a mere 40 ns. This stringent timing requirement also extends to the interconnecting DAC (IDAC) and the IDAC buffer. In the CMOS process, high supply voltage is needed to provide enough voltage headroom to drive the microLEDs. The required size of a segmented DAC and the high bandwidth requirement can both be very difficult to achieve. (5) One of the novelties of this CMOS design is to employ parallel data processing in the CMOS design to fulfill the timing requirements of 25.175 MHz of a VGA display at 60 Hz. This parallel processing also reduces the bandwidth needed for each buffer by a factor of eight. This design greatly reduces the power consumption of the microdisplay, and complexity required to achieve high bandwidth. (6) Another novelty is

Development of nitride microLEDs and displays

25

to employ two approaches to control the microLED currents. The first is to characterize the MOSFET and apply the appropriate voltage to the gate to draw the desired current. The second approach is to force the desired current through the MOSFET in a diode configuration and store the gate voltage. (7) Moreover, the conventional use of polysilicon (POLY) capacitors in CMOS design is eliminated. POLY capacitors require additional structure and spacing whereas PMOS capacitors do not. Furthermore, a POLY capacitor requires a two POLY process. In abstaining from the use of the POLY device, one less mask is needed, hence processing costs are also reduced. The PMOS can share common drain, source, and gate connections with other devices in the design, which leads to significant area savings over the POLY capacitor. The results shown in Figs. 8 and 9 demonstrated the very first successful bonding between CMOS IC driver and microLED array via flip-chip bonding using indium bumps as small as 5 μm to create practical full-scale self-emissive microLED microdisplays which are capable of delivering real time video graphics images (Day et al., 2011, 2012, 2015; Lin et al., 2011). Due to the outstanding characteristics of III-nitride microLEDs, the successful realization of this first active driving and functional microLED microdisplay via heterogeneous integration of microLEDs with Si driver IC has attracted immense attention from researchers and developers worldwide and was as a major catalyst for the emergence of the new microLED display industry. With a global effort in the microLED field today, many different applications have started to emerge.

5. MicroLED and microLED display characteristics 5.1 Brightness characterization For the temperature dependence and radiance characterization, passive μLED array chips with 16
16 pixels (with 12 μm pixel size and 15 μm pitch distance) were fabricated. Fig. 10 shows the measured I-V and L-I characteristics of InGaN μLEDs for blue and green pixels (Day et al., 2011, 2012). The optical powers were measured by a calibrated optical power meter placed on the sapphire side of the μLED array when a single pixel was lit. The nearly linear optical power output with increasing forward current is very useful for a range of applications and ensures a high brightness microdisplay with a high dynamic range gray scale. To obtain a sense of the

26

Hongxing Jiang and Jingyu Lin

(a)

20 Power (μW)

0.8 Current(mA)

25

12 x12 μm2 λ = 462 nm

1.0

0.6 0.4

12 x12 μm2 λ = 462 nm

15 10

0.2 5

0.0 0

1

2

3

4

0 0.0 0.2 0.4 0.6 0.8 1.0

5

Current (mA)

Voltage (V)

(b) 0.20 12 x12 μm2 λ = 517 nm

4

Power (μW)

Current (mA)

0.15

5

0.10

0.05

12 x12 μm2 λ = 517 nm

3

2

1 0.00 2.0

2.5

3.0 Voltage (v)

3.5

0 0.00

0.05

0.10

0.15

0.20

Current (mA)

Fig. 10 I-V and L-I characteristics of (A) blue (462 nm) and (B) green (517 nm) μLED pixels. The optical power was measured by a calibrated optical power meter placed on the sapphire side of a μLED array when a single pixel was lit. Reproduced from Day, J., Li, J., Lie, D.Y.C., Bradford, C., Lin, J.Y., Jiang, H.X., 2011. III-nitride full-scale high-resolution microdisplays. Appl. Phys. Lett. 99, 031116; Copyright (2011) AIP Publishing.

microdisplay brightness, the luminance of the green μLED pixels has been characterized. The luminance of μLED was measured by using a bare green μLED array chip (16
16 pixels simultaneously on) in conjunction with a 10-in. integrating optical sphere, and the single pixel radiance shown in Fig. 11A was obtained by dividing the measured value by the number of

27

Development of nitride microLEDs and displays

120

(a)

λ = 517 nm

300 200

8 7

I=0.1 mA Optical Power (a.u.)

light intensity (mcd)

(b)

100 λ= 462 nm

400

6

80

5 4

60

3

40

2

100 At low currents 1 mcd/1 μA

0 0.0

Optical power Operating voltage

voltage(V)

500 12 μm pixel

0.2

0.4

0.6

Current (mA)

0.8

1.0

20 0 -100

1 0 -50

0

50

100

o

Temperature( C)

Fig. 11 (A) Luminance of a single green (517 nm) μLED pixel as a function of driving current and (B) the temperature dependence of the relative emission optical power of an InGaN μLED array. Reproduced from Day, J., Li, J., Lie, D.Y.C., Bradford, C., Lin, J.Y., Jiang, H.X., 2011. III-nitride full-scale high-resolution microdisplays. Appl. Phys. Lett. 99, 031116; Copyright (2011) AIP Publishing.

pixels (16
16 ¼ 256). As shown in Fig. 11A, a 12 μm pixel outputs roughly 1 mcd/μA and the luminance increases almost linearly with driving current (I) for I < 100 μA (Day et al., 2011, 2012). For a microdisplay with a 15 μm pitch (distance between pixels), when every pixel within the array is lit up and operates at 1 μA, the brightness of the microdisplay can be calculated to be 4
106 cd/m2. This luminance level is several orders of magnitude higher than those of liquid crystal displays (LCD) and organic LED (OLED) displays. Based on the pixel characteristics shown in Fig. 10, at I ¼ 1 μA, a green μLED pixel has a voltage of around 2.6 V. This means that the power dissipation within the μLED array is only about 0.2 W for a full VGA (640
480 pixels) microdisplay since normally only a fraction (25%) of pixels are lit up for graphical video image displays. Furthermore, this 1 μA driving current translates to a current density of about 0.7 A/cm2 for a 12 μm pixel, which is about 1/30 of the typical value of 22 A/cm2 [¼20 mA/(0.03
0.03 cm2)] in a conventional indicator LED operating at 20 mA, which have an average chip size of 300 μm
300 μm and an expected operating lifetime exceeding 100,000 h under normal operating conditions of I ¼ 20 mA. These estimates also imply that the operating lifetime of III-nitride μLED array should be comparable to those of indicator LEDs (Day et al., 2011, 2012).

28

Hongxing Jiang and Jingyu Lin

5.2 Temperature dependence Another outstanding feature of InGaN microLED based self-emissive microdisplays is their ability to operate under harsh conditions and at high or low temperatures. The operating temperature (T) dependence of the optical output power of these μLEDs has been measured and the results are depicted in Fig. 11B (Day et al., 2011, 2012; Lin et al., 2011). The temperature dependence of the relative optical power shown in Fig. 11B) was measured by placing the optical power meter outside the optical window of a portable cryostat while the μLED array chip (16
16 pixels simultaneously on) was held inside the cryostat under a driving current of 0.1 mA. The intensity of the μLED emission decreased by about 10% when T was raised from 20 to 100 °C and remained almost constant when T was cooled down from 20 to 100 °C, while the operating voltage at 0.1 mA varied from 4.1 V at 100 °C to 2.9 V at +100 °C. This continuous reduction in the operating voltage with increasing T is most likely due to thermal activation of free holes ( p) described by the process of p  exp.(EA/kT), where the Mg acceptor activation energy (EA) in GaN is about 160 meV. The T dependence of the μLED emission intensity in Fig. 11B represents the lowest thermal quenching ever reported for any type of microdisplay. Furthermore, the measured emission wavelength change with temperature is only about 0.06 nm/°C. The outstanding thermal stability of microLEDs is a direct attribute of III-nitride semiconductors.

5.3 Operating speed and view angle characterization The operating speed of microLED is one of the most critical parameters, especially for application areas in high-speed 3D/AR/VR displays in which fast motion picture response times are needed. Time-resolved electroluminescence (EL) technique was employed to measure the operating speed of microLEDs ( Jin et al., 2001). In the time-resolved EL setup, the microLED was driven by a picosecond pulse generator. The pulse generator has a frequency range 0–1 MHz, pulse height 0–40 V, and pulse width 0–2 ns with a rise time 100 ps. Pulse shape was monitored by a digital oscilloscope. A synchronized trigger from the pulse generator was given to the constant fraction discriminator (CFD) for precise timing. The microLED was placed with its mesa side facing toward the entrance slit of the monochromator and the EL signal was collected by a micro-channel plate photomultiplier tube (PMT). The amplified signal from the PMT was sent to a time-to-amplitude

29

Development of nitride microLEDs and displays

Time (ns) -0.5

0.0

(a)

0

1.0

1.5

2.0

T = 300 K Big LED μ -LED(d=12μm)

-1

System Response

1.0

-2

τ=214 ps

-3 -4 -5

τ=170 ps

-6

τ=30 ps

-7 0.21

(b)

0.20

τoff (ns)

0.5

Relative Intensity (a.u)

Ln(IEL) (A.U.)

-1.0 1

T = 300 K

0.19 0.18

0.8 0.6 0.4 0.2 0.0

0.17

(c)

-80 -60 -40 -20

0

20

40

60

80

Angle (degree)

0.16 0.15 0.14 7

8

9

10

11

12

13

14

15

16

LED Diameter (μm) Fig. 12 (A) Transient responses of a μLED of 12 μm in diameter and a conventional broad-area LED and (B) the size dependence of the turn-off time, toff, of μLEDs. (C) Viewing angle of microLEDs. Panels (A) and (B): Reproduced from Jin, S.X., Shakya, J., Lin, J.Y., Jiang, H.X., 2001. Size dependence of III-nitride microdisk light emitting diode characteristics. Appl. Phys. Lett. 78, 3532; Copyright (2001) AIP Publishing. Panel (C): Reproduced from Fan, Z.Y., Jiang, H.X., Lin, J.Y., 2008. Lin, III-nitride micro-emitter arrays: development and applications. J. Phys. D: Appl. Phys. 41, 094001; Copyright (2008) IOP Publishing.

converter (TAC) through a pico-timing discriminator for time-resolved detection. The signal output of TAC was fed to a multi-channel analyzer, which was connected to a computer to analyze the time dispersion curves. The time resolved EL setup has an overall system response of about 30 ps ( Jin et al., 2001). Fig. 12 plots (a) the EL transient responses of a representative 12 μm microLED and a conventional broad-area LED (300 μm
300 μm) and (b) the size dependence of the “turn-off” time, τoff, of microLEDs. The turn-on response is on the order of the system response (30 ps) and is too fast to be measured. Hence, the operating speed of a microLED is limited by its turn-off time. The turn-off transient is in a form of single exponential decay and its decay lifetime, τoff, can thus be determined. It was found that τoff decreases with a decrease of microLED size and reduced from 0.21 ns for

30

Hongxing Jiang and Jingyu Lin

d ¼ 15 μm to 0.15 ns for d ¼ 8 μm ( Jin et al., 2001). With this fast speed and other advantages such as long operation lifetime, microLED arrays may be used to replace lasers as inexpensive short distance optical links such as between computer boards with high operating frequencies. Much progress has been achieved recently in using nitride microLEDs in the application area of Li-Fi. The increased operating speed partly is a result of an enhanced radiative recombination rate in μLED because of a reduction of piezoelectric field strength due to partial strain relief in the microdisks (Dai et al., 2001b). However, the increased response speed of microLED with deceasing size is also related with an increased non-radiative recombination rate through the etched edge area of microLEDs ( Jin et al., 2001; Shakya et al., 2004). For microLEDs, the ratio of etched edge surface area to active light emitting area increases with a decrease of the microLED size. The etched edge surface area generally has a higher surface recombination velocity which leads to a larger non-radiative recombination rate. Photonic crystal (PC) LEDs were fabricated to evaluate the surface recombination effects of etched edge surfaces (Shakya et al., 2004). PC LEDs with triangular lattice patterns of circular holes with diameters d ¼ 100–200 nm and periodicities a ¼ 300–600 nm were fabricated using electron-beam (e-beam) lithography and ICP dry etching. The holes were etched through into the active layers (Shakya et al., 2004). It was found that the total decay rate, which includes both the radiative and non-radiative decay rates, increases linearly with the total edge areas of etched holes of PCs (Shakya et al., 2004). The results thus clearly indicated the effects of plasma dry etching on the non-radiative and hence on the total recombination rate of conventional and microLEDs. A value of surface recombination velocity of vs ¼ 1.48
105 cm/s was obtained for InGaN/GaN MQWs structure with holes etching through the active layers using ICP dry etching (Shakya et al., 2004). However, the measured surface recombination velocity for the “as grown” undoped GaN epilayers was vs ¼ 5
104 cm/s (Aleksiejunas et al., 2003). The results thus implied that ICP dry etching increased the surface recombination velocity by as much as a factor of 3. Because of the increased surface recombination velocity, the non-radiative decay rate and hence the response speed of microLEDs increases with decreasing in the microLED size. Nevertheless, the fastoperating speed ensures that the III-nitride μLED based microdisplays have a much faster response than LCD and OLED based microdisplays. Another critical application of III-nitride μLEDs is for bio-molecule intrinsic

Development of nitride microLEDs and displays

31

fluorescence property measurements. The fast-operating time of III-nitride μLEDs also ensures that these devices are perfect light sources for fluorescence lifetime measurements of medical and biologic samples. The angular distribution pattern of light emission from μLEDs has been measured through the sapphire substrate. The result presented in Fig. 14C showed that the viewing angle is more than 120° (Fan et al., 2008; Jiang et al., 2001). Hence, microLED displays have a much broader viewing angle compared with that of LCD and has a similar viewing angle compared with OLED displays. The large viewing angle is an intrinsic property of selfemissive displays.

5.4 Power conversion efficiency characteristics The output power of LEDs, Poutput, depends on the internal quantum efficiency (ηint) and extraction efficiency (Cex), Poutput ∝ ηint Cex. Because piezoelectric field strength is reduced due to partial strain relief and current spreading is also improved in microLEDs (Dai et al., 2001b; Tian et al., 2012), ηint is expected to increase in μLEDs. However, we believe that the enhancement in the extraction efficiency, Cex, is probably more important here. The extraction efficiency Cex of a conventional broad-area LED can be estimated by considering the total internal reflection occurring at the LED-air interface. For GaN-air interface, nGaNsin θC ¼ 1, where nGaN ¼ 2.4, the critical angle θc is around 24°, providing an extraction efficiency around 8.3% from one side of the LED ( Jiang and Lin, 2003; Jin et al., 2000b). In contrast, the emitted light is much easier to get out in microLEDs than in the conventional broad-area LEDs ( Jiang and Lin, 2003; Jin et al., 2000b). It is thus expected that Cex is increased in microLEDs. However, owing to the higher ratio of etched sidewall perimeter to active mesa area of μLEDs, the effects of surface recombination velocity become particularly important in microLEDs and have been studies in detail (Wierer and Tansu, 2019). Fig. 13 shows the simulated power conversion efficiency versus current density for green LEDs for (a) a fixed surface recombination velocity of vs ¼ 3
104 cm/s and variable as/V ¼ 4
101, 4
102, 4
103, 4
104 cm1; and (b) a fixed as/V ¼ 8
103 cm1 and variable surface recombination velocity of vs ¼ 1, 10, 100, and 1000 cm/s, where as is the exposed surface area, V the active region volume, and vs the surface recombination velocity. The current density increases as the size of the microLEDs decreases for the same amount of injected current.

32

Hongxing Jiang and Jingyu Lin

Fig. 13 LED power conversion efficiency versus current density for the green LED with: (A) constant vs ¼ 6
103 cm/s and variable as/V ¼ 4
101, 4
102, 4
103, 4
104 cm1; and (B) constant as/V ¼ 8
103 cm1 and variable vs ¼ 1, 10, 100, and 1000 cm/s. Reproduced from Wierer Jr., J.J., Tansu, N., 2019. III-nitride micro-LEDs for efficient emissive displays. Laser Photon. Rev. 13, 1900141; Copyright 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim.

A decrease in the microLED size increases the ratio of as/V from 4
101 to 4
104 cm1 as well as the surface recombination. The results in Fig. 13A clearly revealed that for a fixed microLED size, the power conversion efficiency decreases as the surface recombination velocity increases. Fig. 13B shows the power conversion efficiency or external quantum efficiency (EQE) versus current density for green LEDs with a constant size of 5 μm (as/V ¼ 8
103 cm1) and surface recombination velocity vs varying from 1 to 1
103 cm/s. The low current efficiency recovered only at vs ¼ 1 cm/s, implying that the surface recombination needs to be almost nonexistent in microLEDs in this configuration (Wierer and Tansu, 2019). While for a fixed surface recombination velocity, the power conversion efficiency decreases

Development of nitride microLEDs and displays

33

with an increase in the ratio of the exposed surface area to the volume (as/V) or equivalently with a decrease in the microLED size. Special steps for surface treatments, including thermal annealing, chemical wet etching after plasma etch, and oxide/polymer surface passivation to remove the damage and compensate the dangling bonds, can be adopted in the process. Controlling the etch profile is also critical (Fan et al., 2008; Jin et al., 2001). In one study, it was demonstrated that the sidewall etch defects can be partially recovered by increasing the thermal annealing time, consequently improving the efficiency at low current densities (Tian et al., 2012). More recently, it was shown that the combination of chemical treatment and atomic layer deposition (ALD) passivation effectively addresses the effects of non-radiative recombination (Wong et al., 2019, 2020), leading to the demonstration of size independent peak EQE of μLEDs. It was believed that the use of chemical treatment is to eliminate leakage paths at the sidewall due to

Fig. 14 (A) Schematic of the InGaN dot-in-nanowire tunnel junction LED heterostructure. (B) Scanning electron microscopy image of an InGaN photonic nanocrystal array. The scale bar represents 500 nm. (C) The PL spectrum for an InGaN photonic nanocrystal array measured at room temperature. (D) Schematic of a fabricated microscale LED (not drawn to scale). (E) Variations of relative EQE vs injection current density. Reproduced from Liu, X., Wu, Y., Malhotra, Y., Sun, Y., Mi, Z., 2020. Micrometer scale InGaN green light emitting diodes with ultra-stable operation. Appl. Phys. Lett. 117, 011104; Copyright (2020) AIP Publishing.

34

Hongxing Jiang and Jingyu Lin

plasma damage, whereas ALD passivation removes surface recombination to erase non-radiative recombination sites (Wong et al., 2019, 2020). Advances in damage-free neutral beam etching (NBE) have also been made lately (Zhang et al., 2020; Zhu et al., 2019). It was shown that a successful elimination of the size-dependent efficiency decrease in GaN microLEDs is achieved using NBE (Zhu et al., 2019) and the process of NBE nano-disk etching followed by GaN regrowth represents a promising step forward in the development of microLEDs with a buried active region in a top-down structure (Zhang et al., 2020). Alternatively, it was shown that surface recombination can be largely suppressed by employing a unique core-shell nanowire structure surrounding the QW active region (Nguyen et al., 2013). As illustrated in Fig. 14, green microLEDs based on InGaN nanowires, which exhibit highly stable and efficient emission with a small efficiency droop under high injection current, have been attained (Liu et al., 2020; Nguyen et al., 2013, 2015). A parallel approach currently also under development is the utilization of a selective overgrowth method (Bai et al., 2020), in which the selective overgrowth only takes place within prepatterned SiO2 micro-hole arrays, as illustrated in Fig. 15. This selective overgrowth approach has led to the realization of ultrasmall, ultra-efficient, and ultracompact green μLEDs with a dimension of 3.6 μm and an inter-pitch distance of 2 μm (Bai et al., 2020). Using such a technique, the dry-etching processes for the formation of μLED mesas can be eliminated. The authors further pointed out that the SiO2 micro-hole masks also naturally serve as a surface passivation without requiring any additional processes or etching-back processes, which further simplifies subsequent device fabrication. Consequently, as demonstrated in Fig. 15, ultrahigh brightness of above 107 cd/m2 and peak EQE of 6% in the green spectral region at 515 nm has been achieved (Bai et al., 2020). Unique features of microLED microdisplays are summarized in Table 1 (Day et al., 2011, 2012, 2015; Jiang and Lin, 2013; Lin et al., 2011). Compared with other technologies, microLED microdisplays can provide superior performance. Unlike LCDs that normally require an external light source (Vettese, 2010), microLED microdisplays are self-emissive, which results in both space and power saving, improves image quality, and allows viewing from any angle without color shift and degradation in contrast. LCDs also require the use of heaters when operating in low temperature environments. The luminance level of microLED microdisplays is several orders of magnitude higher than those of LCD and OLEDs. OLEDs must be driven at current densities many orders of magnitude lower than

Fig. 15 Schematic of μLED formation by direct epitaxial approach: (A) SiO2 mask deposition; (B) SiO2 mask patterning; (C) μLED array overgrowth; (D) schematic drawing of a green μLED array; (E) a green μLED array at an injection current density of 9 A/cm2; (F) brightness of a green μLED array chip as a function of injection current density; (G) external quantum efficiency (EQE) of a μLED array chip fabricated by direct epitaxial approach as a function of injection current density. Note that the measurements are performed on a bare chip without any package or coating. Reproduced from Bai, J., Cai, Y., Feng, P., Fletcher, P., Zhao, X., Zhu, C., Wang, T., 2020. A direct epitaxial approach to achieving ultrasmall and ultrabright InGaN micro light-emitting diodes (μLEDs). ACS Photonics 7, 411; Copyright (2020) American Chemical Society.

36

Hongxing Jiang and Jingyu Lin

Table 1 Comparison among various technologies for microdisplays. Technology Liquid crystal Organic LED MicroLED

Mechanism

Backlighting/ LED

Self-emissive

Self-emissive

Luminous efficacy

Medium

Low

High

Brightness (nit or cd/m2)

3000 (full color) 104 (green)

1500 (full color) 103 (yellow)

105 (full color) 107 (blue/green)

Contrast ratio

200:1 (intrinsic)

Very high Greater than 1,000,000:1

Very high Greater than 1,000,000:1

Response time

ms

μs

sub-ns

Frame rate

kHz

MHz

GHz

Resolution (ppi)

500

5000

5000

Operating temperature (°C)

0–60 (requires heater)

50 to 70

100 to 120

Shock resistance

Low

Medium

High

Lifetime

Medium

Medium

Long

Cost

Low

Low

Low

semiconductor LEDs to obtain devices with a reasonable lifetime. For OLEDs to operate at high brightness needed for HDR, they are susceptible to screen burn-in. On the other hand, digital light processing (DLP) and laser beam steering (LBS) devices require the use of rapidly scanning microelectromechanical (MEMS) mirrors and separate light sources such as LEDs or laser diodes (LDs), which adds complexity, volume, and cost to the system. Furthermore, the service lifetimes of MEMS and LDs are shorter than LEDs.

6. Full color microdisplay development 6.1 Standard side-by-side RGB sub-pixel approach Following the demonstration of the first VGA monochrome μLED microdisplay (Day et al., 2011, 2012, 2015; Lin et al., 2011), substantial progresses have been made in the development of monochromatic μLED microdisplays with higher pixel density, smaller pixels, and larger display sizes

Development of nitride microLEDs and displays

37

(https://www.allaboutcircuits.com/news/tiny-displays-mojo-vision-microleddisplay-highest-pixel-density/; https://www.microled-info.com/jbddemonstrates-2-million-nits-and-10000-ppi-micro-led-microdisplays; Chen et al., 2019; Templier, 2016). One of the most vital next steps for μLEDs microdisplays is the realization of full color displays. Since the growth technology for attaining InGaN/GaN MQW red color wafers with sufficiently high quantum efficiencies and emission intensities is still under development (Iida et al., 2020a,b), the community is unable to use monolithic integrated InGaN/GaN MQW red-green-blue (RGB) wafers at this point. One plausible approach is to use three primary color RGB microLED sub-pixels based on AIGalnP (Red) and GaN (Green and Blue) placed laterally on the same plane to form one full color pixel via techniques such as transfer printing and flip chip (Bower et al., 2017; Lau, 2019; Li et al., 2019a). Alternatively, RGB microLED lateral sub-pixels can also be based on blue or UV microLED array with color converter materials such as phosphors or quantum dots (QDs) (Han et al., 2015; Lee et al., 2018; Lin et al., 2017; Luo et al., 2018). InGaN nanowires and nanocolumn (NC) LEDs have shown promising results for full color microLED applications (Bui et al., 2019; Kishino et al., 2020; Monemar et al., 2016; Ra et al., 2016; Samuelson et al., 2016; Yamano and Kishino, 2018). Most recently, a 16
16 array of InGaN/GaN-based NC microLEDs with different emission colors has been attained monolithically via a one-step selective area growth, pointing to the potential of integrated NC microLEDs for realizing full-color microLEDs (Kishino et al., 2020). At the same time, approaches including funnel-tube array and projection lithography patterned QDs on microLED wafers have shown to provide improved conversion efficiencies and reduced crosstalk (Gou et al., 2019; Li et al., 2019b). These approaches have been thoroughly reviewed recently (Ding et al., 2019; Zhou et al., 2020). The approach of placing three RGB sub-pixels laterally on the same plane has the advantages of simple architecture and identical CMOS control circuit design as the monochrome microdisplays. However, the drawback of placing three RGB sub-pixels laterally on the same plane is a reduction in the pixel density as each full color pixel containing three sub-pixels of RGB colors and occupying an area of three times of that of monochrome microdisplays. Consequently, it reduces the resolution or pixel density (D), or equivalently pixel per inch (ppi), by a factor of 31/2 ¼ 1.7. This reduction in resolution is not preferred for applications such as AR and VR microdisplays, which require high resolutions. This scenario is more clearly illustrated in Fig. 16, which plots the dimension of length (l) or width (w) of a

38

Hongxing Jiang and Jingyu Lin

Fig. 16 (A) Top view of representative pixels of a high pixel density display such as a smartwatch. (B) A plot of subpixel length or width versus pixel density (pixels per inch, ppi) showing maximal sizes at a given ppi. Reproduced from Wierer Jr., J.J., Tansu, N., 2019. III-nitride micro-LEDs for efficient emissive displays. Laser Photon. Rev. 13, 1900141; Copyright 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim.

subpixel as a function of the pixel density D (ppi) for different sub-pixel spacing (s) for such a design (Wierer and Tansu, 2019). It can be seen from Fig. 16 that l or w decreases with increasing the pixel density, D. Even for the case of the smallest possible spacing between sub-pixels of s (¼2 μm) on the plot, the maximum achievable D is less than 2000 ppi for l (or w) as small as 3 μm. The pixel density for applications increases when the viewing distance is decreased where our eyes have a resolution acuity of 60 pixels/degree (Arden, 1988; Cuervo et al., 2018). The most aggressive pixel density requirement might be for a VR headset, wherein a viewing distance of 60 mm requires a minimum pixel density of 1455 ppi. Therefore, for high pixel density applications, getting all three primary color pixels to emit through a single pixel aperture is critical for the realization of full-color microdisplays with a small form factor at a system level and to enable the maximum utilization of semiconductor integration techniques in the design of the microLED microdisplays.

6.2 Vertically stacked RGB microLED microdisplay concept As illustrated in Fig. 17, we have disclosed a full color microdisplay architecture based on vertically stacked microLEDs (Fan et al., 2011). In the

(a)

wafer bonding

(c) (b)

n-AlGaInP p-AlGaInP

AlGaInP

p+-GaAs SiO2 n-GaN p-GaN n-GaN

InGaN

GaN buffer Sapphire substrate

Fig. 17 Schematics of a full color microdisplay via vertically stacked RGB color pixels: (A) InGaN and AlGaInP hybrid layer structure with InGaN for blue and green emission, and AlGaInP for red emission; (B) the perspective view of the pixel structure with electrodes distribution; (c) the schematic circuit diagram showing the three-color pixels have common anode with independent cathodes. Reproduced from Fan, Z.Y., Li, J., Lin, J.Y., Jiang, H.X., 2011. Micro-emitter array based full-color microdisplay. US Patent 8,058,663.

40

Hongxing Jiang and Jingyu Lin

design, each pixel includes three vertically stacked red, green, and blue microLEDs and the size of RGB sub-pixels is the same as the size of microdisplay pixel, which is minimized to the greatest extent possible. In this architectural design, microdisplays are based on III-nitride InGaN/GaN MQWs, with the optical active MQW regions containing InGaN with different In-contents to emit three primary RGB colors once red emitting InGaN LEDs are available. Alternatively, the microdisplay can also be based on heterogeneous integration of InGaN and AlGaInP MQW LEDs. In such structures, the highly efficient AlGaInP based semiconductors are used for red emission, and InGaN based semiconductors for green and blue emission. As illustrated in Fig. 17A, blue emitting InGaN/GaN MQW LED is grown directly on sapphire substrate. The green emitting MQW LED structure is then grown on top of blue emitting MQW LED structure. Another red emitting AlGaInP MQW wafer is stacked on top of blue/green wafer using wafer bonding. Sapphire substrate is utilized naturally as the light emission window of the microdisplays. In this fashion, photons with lower energy pass through sub-pixels of higher energy bandgap semiconductors before exiting from the emitting surface; i.e., while blue light will exit the display directly, the green light from each sub-pixel needs to pass through blue sub-pixel first; the red light needs to pass through both blue and green sub-pixels first before emitting to the front surface of the microdisplay. Fig. 17B shows the pixels with contacts distribution, and Fig. 17C is the circuit diagram of three stacked microLEDs (Fan et al., 2011). In Fig. 17, the three emitters in one pixel share a common anode with independently controllable cathode. It could be difficult to design the driving circuit for a microdisplay with such a pixel structure. To facilitate the driving circuit design, a preferred scheme of the pixel structure is shown in Fig. 18A. The three stacked emitters in one pixel share a common cathode as ground with independently controllable anode. To achieve such a pixel structure, the hybrid integrated wafer structure is schematically shown in Fig. 18B and C, in which tunneling junctions for p-GaN can be incorporated. The incorporation of tunneling junction can also minimize the plasma etching induced damage on p-type layers. The device structure is formed by wafer bonding of InGaN and AlGaInP wafers. A high performance microdisplay for full color video information requires a highly integrated CMOS active matrix driving circuit and the related processor. These microdisplays can be integrated on Si IC driver chips by using flip-chip bonding or aligned wafer bonding. Flip-chip bonding is based on indium metal bumps, as demonstrated by the results shown

Fig. 18 Another embodiment of a full color microdisplay via vertically stacked RGB color pixels: (A) The circuit diagram shows the three stacked color pixels share a common cathode and possess independent anodes. (B) The AlGaInP and InGaN hybrid layer structure with tunneling junctions. (C) The schematic circuit diagram showing the RGB pixel with VC as the common ground, and VR, VG, VB for red, green, and blue pixel control, respectively. (D) Illustration of utilizing a Si CMOS-compatible oxide-to-oxide aligned wafer bonding process to integrate the microdisplay with Si IC driver/processor as the backplane to realize high density microdisplays. Reproduced from Fan, Z.Y., Li, J., Lin, J.Y., Jiang, H.X., 2011. Micro-emitter array based full-color microdisplay. US Patent 8,058,663.

42

Hongxing Jiang and Jingyu Lin

Figs. 8 and 9 (Day et al., 2011, 2012, 2015; Lin et al., 2011). For high density microdisplays with pitch distance less than 15 μm, the preferred approach is based on Si CMOS-compatible oxide-to-oxide aligned wafer bonding process to integrate the microdisplay with Si IC driver/processor as the backplane, as illustrated in Fig. 18D (Fan et al., 2011). For small format microdisplays, passive driving scheme may be implemented. Nevertheless, it can be seen that the major challenge of this full color microLED microdisplay approach is to design, fabricate and interconnect electronic circuits to control vertically stacked RGB sub-pixels which are sharing the same optical aperture in the microLED array (Fan et al., 2011). As the size of Si transistors further shrinks and the transistor density on IC further increases, the semiconductor industry has begun to adopt 3D IC architectures, in which the through-Si-via (TSV) technology has been developed to make metal interconnections between different layers. In Si CMOS technology, a TSV or through-chip-via is a vertical interconnect access (via) for electrical connection that passes through a Si wafer or die. TSVs are high-performance interconnect techniques used as an alternative to wire bonding and flip chips to create 3D packages and 3D ICs. Compared to alternatives such as package-on-package, the interconnect and device density are substantially increased, whereas the length of the connections is significantly decreased. If adopted for the development of vertically stacked RGB full color microLED microdisplays, a TSV equivalent technology would be extremely useful for interconnecting electronic circuits to control RGB sub-pixels. Though tremendous progresses have been achieved for III-nitride semiconductor technologies, R & D activities for the development of 3D III-nitride ICs have been very limited. This is because 3D III-nitride ICs presently are probably far ahead of many immediate applications. However, the via technology has been recognized for various important applications of III-nitride semiconductors. For example, developing a stable and reliable backside via etch process is being recognized as one of the critical enabling but challenging techniques for achieving high performance and high yield GaN-on-SiC high electron mobility transistors (HEMTs) monolithic microwave IC (MMIC) devices (Chen et al., 2014; Fan et al., 2015). The realization of GaN via using ICP dry etching and metallization has been previously reported, as the SEM images shown in Fig. 19 clearly revealed that etched through holes and metallization through theses holes have been successfully realized (Chen et al., 2014). Thus, with further development, it is feasible that GaN via technology can be adopted for the

Fig. 19 Through wafer vias in GaN on SiC: (A) SEM images of the clean progression GaN etching to ultrasonic clean; (B) SEM image of the metallized via backside. Reproduced from Chen, C.H., Chang, Y.W., Weng, M.H., Chang, R., Huang, S. H., Wang, F., Wei, Y.F., Hsieh, S., Cho, I.T., Wohlmuth, W., 2014. Method for forming through wafer vias in GaN on SiC devices and circuits. In: CS MANTECH Conference, May 19th–22nd, Denver, Colorado, USA, p. 275.

44

Hongxing Jiang and Jingyu Lin

integration of electronic control circuits for vertically stacked RGB wafers for full color microLED microdisplays. However, to integrate InGaN blue/ green with AlGaInP red wafers, not only we need to develop throughwafer-via for both GaN and AlGaInP wafers, further steps involving substrates removal and wafer bonding are also necessary. Moreover, integration of InGaN blue/green wafer with AlGaInP red wafer also requires forming a via through substrate in a compound semiconductor having microLED arrays on the front side of the substrate (Cramer and Dawson, 2011; Yang et al., 2014). These processes are complicated, nevertheless, there are no fundamental physics issues to adopt them.

6.3 Demonstration of full color microdisplays based on vertically stacked RGB microLEDs Most recently, a full color microLED microdisplay based on monolithic and heterogeneous integration of microLED array with Si CMOS image processor has been demonstrated by Ostendo Technologies, Inc., wherein each pixel is a vertical stack of sub-pixels of red, green and blue microLEDs with RGB light emission sharing the same optical aperture (Yadavalli et al., 2020). As illustrated in Fig. 20A, in these microLED microdisplays, three separate epi-wafers of red, green, and blue emission colors were first processed to create desired size microLED arrays with a 5–10 μm pixel pitch, each of the three processed epi-wafers were sequentially bonded to a single receiving handle wafer followed by substrate removal and backside process to create a handle wafer with RGB microLED pixel arrays stacked vertically on top of each other. Three RGB LED color wafers were electrically isolated using SiO2 layers. Substrate removal was used to stack three primary color wafers vertically in the order of blue, green, and red from glass side (emission side) with one single optical aperture for each pixel of all three colors (Yadavalli et al., 2020). Moreover, as illustrated in Fig. 20A, the vertical sidewalls and the p- and n-contacts to microLED pixels preferentially incorporated reflective metal layers using contact via technology. Each of the three processed epi-wafer (blue, green, and red) was sequentially bonded to a single receiving handle wafer of monolithically fabricated micro-scale pixel-level optical elements array with alignment accuracy better than 1 μm between each bond pair wafers. After bonding of each wafer to the receiving handle wafer, substrate removal was performed by utilizing either laser-lift-off (LLO) or using lapping and wet etching techniques (depending on the LED wafer growth substrate). Metal contacts connected to each sub-pixel of three colors were fabricated using dry etching and via

Fig. 20 See figure legend on next page.

46

Hongxing Jiang and Jingyu Lin

technology (Yadavalli et al., 2020). The three-primary color microLED arrays are fusion bonded with CMOS image processor and are further processed to enable through-silicon vias on the backside of the CMOS image processor. Electrical interconnects of very high density (1–3 million/cm2) at the bond interface carry electrical signals from CMOS image processor to vertically stacked RGB micro-LED layers (Yadavalli et al., 2020). The digital control logic comprises an array of logic cells positioned directly below each pixel and associated control logic located at the periphery of the logic cell array (Yadavalli et al., 2020). The CMOS backplane handles image source data in a serial bit stream format wherein multiple bit words define color component and brightness of each pixel (Yadavalli et al., 2020). This vertical integration of three primary sub-pixels with CMOS control circuits, in our opinion, may represent the design which is very close to what is needed in the final commercial products of full color microLED microdisplays. These vertically integrated full color microLED arrays enable very high performance microdisplays and provide multicolor self-emissive microdisplays with the expected outstanding features of microLEDs, including high resolution, high brightness, low power consumption, compactness, high contrast and high speed. Fig. 20B shows examples of close-up full color images displayed by such a microdisplay. The results shown in Fig. 20B clearly demonstrated outstanding features of high color gamut, pixel density, spatial resolution, and contrast due to vertical integration of three primary color pixels. The demonstrated performance specifications by such a microdisplay so far include: The pitch size can be as small as 5 μm Fig. 20 (A) Schematic illustration a cross-section of QPI® device pixel layer stack by Ostendo Technologies, Inc., wherein Red, Green and Blue micro-LEDs are vertically stacked on top of CMOS image processor driving transistors. Wafer-wafer fusion bonding techniques are utilized to form the QPI device. (B) Examples of camera pictures of full color images displayed by a QPI® chip by Ostendo Technologies, Inc., which is bonded to a flex and driven by a test platform board (prior to integration into smart glasses). (C) Schematic of QPI® Distributed Optical Combining, with devices optically coupled to the edge of AR glass combiner lens (single or multiple devices depending on targeted diagonal field of view for application). Camera picture of prototype small form factor “wearable” wireless AR smart glasses utilizing QPI® devices optically coupled to the edge of AR glass combiner lens. Reproduced from Yadavalli, K., Chuang, C.L., El-Ghoroury, H.S., 2020. Monolithic and heterogeneous integration of RGB micro-LED arrays with pixel-level optics array and CMOS image processor to enable small form-factor display applications. Proc. SPIE 11310, 113100Z-1; Copyright (2020) SPIE.

Development of nitride microLEDs and displays

47

with 1280
720 pixels for high-definition resolution format, white luminance of 3
105 cd/m2, contrast ratio of on/off of 106:1, and a power consummation of less than 300 mW (Yadavalli et al., 2020). The possibility of using these demonstrated full color microdisplays for the realization of small form factor of “wearable” AR smart glasses is illustrated in Fig. 20C and D. As shown in Fig. 20C, the microdisplay is optically coupled to the edge of AR glass combiner lens and images can be projected to be viewed by eyes. These full color microdisplays combining light source and light modulation at the pixel level potentially enable holistic optical and system design for various applications, including AR smart glasses, direct view light field displays, head-up displays and pico-projectors. Prototype small form factor “wearable” wireless AR smart glasses developed have also been demonstrated to have sub-1 W power consumption and has a weight 1 mVpp on the same shank) stimulation artifact in signals recorded from all the channels on prototype μLED optoelectrodes (Wu et al., 2015). Even greater artifact accompanies stimulation with rectangular pulses, as shown in Fig. 13. Since the stimulation artifact prevented the detection of important neural activities such as changes in the local field potentials by overriding them, the use of sinusoidal pulses prevented accurate calculation of the dose and the duration of optical stimuli and also limited the temporal resolution of the stimulation. When the rectangular voltage pulses were supplied to the LEDs, they would generate high-magnitude stimulation artifacts that saturated the neural recording amplifier. As a result, LEDs could not be turned on and off as rapidly as desired to enable stimulation with a sufficiently high (>100 Hz) frequency. Therefore, it is necessary that the stimulation artifact be eliminated or at least reduced to a reasonable level so that faster stimulation signals (ideally with rectangular pulses) could be utilized without significantly degrading the quality of the recorded signals. 5.1.2 Potential sources of stimulus artifact A variety of forms of optically-induced artifact and/or noise in the in vivo signals from electrode arrays have been reported in the literature (Ayling et al., 2009; Budai et al., 2018; Cardin et al., 2010; Han et al., 2009; Jun et al., 2017; Kampasi et al., 2016, 2018; Khurram and Seymour, 2013;

Fig. 13 An example stimulation artifact generated on the prototype microLED optoelectrode. Adapted with permission from Kim et al. (2016); Copyright 2016, IEEE.

154

Kanghwan Kim et al.

Kozai and Vazquez, 2015; Laxpati et al., 2014; Liu et al., 2018; Park et al., 2014; Mikulovic et al., 2016; Scholvin et al., 2015; Wang et al., 2020; Wise et al., 1970). Among them is an artifact due to photoelectrochemical effect (PEC), which is the emission of electrons from the surface of the metal at the metal-electrolyte interface. Because this takes place when a metal electrode is exposed to incident photons with sufficient energy, PEC has been observed in a variety of devices regardless of the substrate material (Ayling et al., 2009; Budai et al., 2018; Cardin et al., 2010; Han et al., 2009; Khurram and Seymour, 2013; Kozai and Vazquez, 2015; Laxpati et al., 2014; Liu et al., 2018; Park et al., 2014). Another form of stimulation artifact results from electromagnetic interference (EMI), a phenomenon that takes place via capacitive and inductive coupling due to the exchange of electromagnetic energy between two adjacent conductive materials that carry electricity. Because EMI takes place when the source is in close proximity to the receiver (target), it has been observed on devices that have light sources directly integrated with recording devices (Kampasi et al., 2016, 2018; Wang et al., 2020). Finally, artifact can be induced by photovoltaic (PV) effect, during which the electrostatic potential of the semiconductor substrate changes due to the absorption of photons. Because most high-density in-vivo extracellular electrode arrays have been made using silicon substrates, artifact has been observed during many experiments in which visible light (λ < 1100 nm) has been used ( Jun et al., 2017; Mikulovic et al., 2016; Wise et al., 1970). A variety of engineering measures have been used to prevent artifact, including replacing metal electrodes with transparent electrodes (Liu et al., 2018; Park et al., 2014), inclusion of EMI shielding structures (Kampasi et al., 2018; Wang et al., 2020), and degenerate doping of the silicon substrate (Scholvin et al., 2015; Wise et al., 1970). Unfortunately, because microLED optoelectrodes are made of heterogeneous materials that are very densely integrated, the sources of artifact in microLED optoelectrodes have not been clearly identified in the past nor have solutions for the elimination of these artifacts. 5.1.3 Electromagnetic interference EMI is inevitable in a system where a source of a large current (or high voltage) and an electrically conductive body connected to a high-impedance load are located in close proximity to each other. On microLED optoelectrodes, metal traces that carry signals of different power levels are densely integrated in order to minimize the width and the cross-sectional area of the implanted

GaN-on-Si μLED optoelectrodes

155

portion of the device. More specifically, the signal recording circuits, especially the metal traces that connect the recording sites to the backend of the probe, are integrated adjacent to the LED drive circuits. This high-density integration increases the mutual capacitance between the traces and, in turn, makes the electrode circuits susceptible to EMI resulting from the LED drive signals. A close observation of the stimulation artifact waveform revealed two sources of EMI artifact on microLED optoelectrodes: the n-GaN layer and the metal traces for the LED drive signal (Kim et al., 2016, 2020). Capacitive EMI can be alleviated by reducing the mutual capacitance between the source and the receiver and placing a shielding layer between the two. The mutual capacitance between two metallic bodies can be reduced most effectively by increasing the distance between those bodies and adding a shielding layer between them. For the microLED optoelectrode, in order to keep the width and the cross-sectional area of the shanks as small as possible, multi-metal-layer structures with embedded shielding layers (Kim et al., 2016, 2020) are utilized. Two- and three-metal-layer structures, in which the metal traces for the LED anode and cathode connections (LED interconnects) and the metal traces for the recording electrodes are defined in separate metal layers, have been utilized. Schematic diagrams of the cross-sections of the multi-metal-layer microLED optoelectrode structures are shown in Fig. 14. On the three-metal-layer configuration, there is a metal layer connected to the ground embedded on the layer between the bottom layer and the top layer (the LED interconnects and the recording electrode interconnects, respectively). Two- and three-metal-layer structures reduce the amplitude of the stimulation artifact by approximately an order of magnitude (from several millivolts peak-to-peak to several hundreds of microvolts peak-to-peak) (Kim et al., 2016, 2020).

5.2 Photovoltaic effect from silicon substrate Even after the reduction of EMI with a multi-metal-layer structure, the magnitude of the stimulation artifact on microLED optoelectrodes was considerably higher than typical neuronal spikes (100 μVpp). An observation of the shape of the stimulation artifact revealed that the residual artifact was mostly from the photovoltaic effect from the silicon substrate. The use of highly-boron-doped silicon substrates for the GaN/InGaN epitaxial growth reduced the magnitude of the stimulation artifact, while not affecting the plug efficiency of the microLEDs on the optoelectrode.

Fig. 14 Cross-sectional diagrams of the single (left) and the multi-metal-layer (middle and right) microLED optoelectrode structures. The sources of the electromagnetic interference are highlighted.

GaN-on-Si μLED optoelectrodes

157

5.2.1 Photovoltaic effect It is well known that the interaction between a photon and a semiconductor results in the generation of electrons and holes in the semiconductor, which in turn results in the generation of photocurrents in the material. Since Wise et al. discovered that room light generates noise in the electrical recording channels of silicon-based devices having intrinsic or lightlydoped silicon substrates (Wise et al., 1970), Michigan Probes have been fabricated with heavily-doped silicon substrates (Scholvin et al., 2015), unless the substrate doping density has to be kept low to accommodate on-chip integrated electronics ( Jun et al., 2017). Even so, there have been reports on photovoltaic-effect (PV)-induced stimulation artifact when silicon probes are used in the vicinity of high-intensity illumination for optogenetics (Mikulovic et al., 2016). The waveform of the stimulation artifact observed on multi-metal-layer microLED optoelectrodes has the same polarity as that of the artifact generated due to illumination on the optoelectrode using an external light source (Kim et al., 2020). When 470-nm LED light illuminated the tips of electrode arrays (i.e., probes) fabricated with a variety of substrates while the tips were submerged in saline, an artifact with the same shape was observed on silicon probes with intrinsic and lightly-boron-doped silicon substrates but not on those with heavily-doped silicon substrates. On the other hand, the same illumination did not generate any artifact on electrode arrays fabricated using soda-lime glass and GaN-on-sapphire GaN/InGaN MQW LED wafers. Fig. 15 shows the variations of the magnitudes and the shapes of the artifact measured on electrode arrays fabricated using various substrates during the spot illumination experiment. These results suggest that the stimulation artifact on multi-metal-layer microLED optoelectrodes is due to the PV effect in the silicon substrate and not from other sources. 5.2.2 Heavily-doped silicon substrates It has been conjectured that the reduced carrier lifetime in heavily-doped silicon helps mitigate light-induced noise on Michigan Probes (Scholvin et al., 2015; Wise et al., 1970). The heavy doping of semiconductors greatly reduces carrier lifetime (Fossum and Lee, 1982; Ross, 1980) and therefore the diffusion lengths of free carriers. This reduced carrier diffusion would reduce the current generated inside the substrate and thus the induced voltage (Scholvin et al., 2015). Thus, the magnitude of the stimulation artifact depends on the free carrier distribution and therefore the doping of the substrate.

Fig. 15 The magnitude and the shape of signals recorded from electrode arrays during a focus blue-light illumination on the electrodes. An obvious lack of (transient) stimulation artifact on devices fabricated using non-silicon substrates can be observed. Adapted with permission from Kim et al. (2020); Copyright 2020, Springer Nature.

GaN-on-Si μLED optoelectrodes

159

An FEM simulation supported this hypothesis: the magnitude of the built-up voltage inside a silicon substrate shrinks with a higher boron substrate doping density (Kim et al., 2020). First, optical illumination induces electron-hole pair generation inside the silicon substrate, and the optically-generated carriers then redistribute inside the substrate depending on their types. The difference between electron and hole distribution patterns gives rise to an electric field inside the substrate and, in turn, changes the electrostatic potential of the substrate-electrolyte interface. An experiment using external illumination on multi-metal-layer microLED optoelectrodes fabricated with wafers with different silicon doping densities confirmed that the magnitude of the light-induced voltage signal is greatly reduced with the use of heavily-boron-doped silicon substrates (Kim et al., 2020). External illumination experiments were repeated with multi-metal-layer microLED optoelectrodes fabricated with GaN-on-Si LED wafers having three different doping densities, and the results confirmed that microLED optoelectrodes fabricated on heavily-boron-doped silicon substrates show the smallest stimulation artifact. 5.2.3 Effect of the substrate on MOCVD-grown InGaN MQW LED performance One of the important requirements for a microLED optoelectrode is that it generates sufficient light while not excessively heating the tissue. There could be concerns with regard to a potentially decreased efficiency of the devices due to the need for GaN epitaxial growth on highly-doped substrates. It was conjectured that a larger mismatch in the lattice constants between the substrate and the epitaxial layer would degrade the epitaxial layer quality and the optical and electrical characteristics of the fabricated devices. Because doping does affect crystal lattice structure and alters the lattice constant (Kucytowski and Wokulska, 2005), the effects of increased substrate doping density on the emission efficiency of the microLEDs needed to be carefully investigated. The results of an experiment conducted with fabricated microLEDs, shown in Fig. 16, suggested that the effect of increased substrate doping is not more significant than that of non-uniformity of the MOCVD grown active layer on a 4-in. wafer (Kim et al., 2020). The electrical and optical characteristics of microLEDs fabricated using LED wafers with intrinsic silicon wafers were in fact inferior to those of microLEDs fabricated using LED wafers with more highly-doped silicon substrates. Much greater variations were observed in the microLED characteristics due to LED locations on

Fig. 16 The electrical and optical characteristics of microLEDs fabricated on GaN-on-Si LED wafers with different silicon substrate doping densities. A greater intra-wafer, location-dependent variation of the characteristics suggests a minimal contribution of the substrate doping density to the LED performance. Adapted with permission from Kim et al. (2020); Copyright 2020, Springer Nature.

GaN-on-Si μLED optoelectrodes

161

a single wafer than there were between those on different wafers at the same location. These results suggested that other factors in the epitaxial growth affect the final plug efficiency of the fabricated microLEDs more significantly than does the substrate doping level.

5.3 MiniSTAR microLED optoelectrodes Stimulation artifact, most of which was attributed to EMI generated from LED drive signals and PV resulting from emitted photons interacting with the Si substrate, was successfully mitigated with multi-metal-layer structures and heavy boron doping of the silicon substrate. In order to emphasize the small stimulation artifact, these microLED optoelectrodes are referred to as minimal-stimulation-artifact (miniSTAR) microLED optoelectrodes. Optical stimulation using the microLEDs on the fabricated miniSTAR microLED optoelectrodes generated stimulation artifacts with greatly reduced amplitudes, ranging from smaller than a few tens of microvolts to a couple of hundred microvolts peak-to-peak (Kim et al., 2020). Fabricated miniSTAR microLED optoelectrodes utilized for in vivo experiments demonstrated the lack of noticeable stimulation artifact. The additional artifact reduction strategy of transient pulse shaping (Kim et al., 2020; Mendrela et al., 2018b) was utilized so that the magnitude of the artifact was reduced to below 50 μV, so that the artifact does not affect neuronal signal detection. An artifact-free opto-electrophysiology experiment was demonstrated using a mouse with ChR2-expressing neurons. 5.3.1 Fabrication of minimal-stimulation-artifact microLED optoelectrodes MiniSTAR microLED optoelectrodes were fabricated in the multi-metallayer configuration, and GaN-on-Si InGaN MQW LED wafers with heavily-boron-doped Si substrates were utilized. The fabrication process was identical to that for the fabrication of the prototype microLED optoelectrodes until the microLED structures were defined. Upon completion of the definition of LED interconnects, the second and the third metal layers, which respectively serve as the shielding layer and the recording electrode interconnects, were defined with ALD Al2O3/PECVD SiO2 bilayer stacks between each of the metal layers. Standard Ti/Ir electrode patterning and two-step DRIE device release steps were utilized to define the electrodes and release the fabricated probes, respectively. Released miniSTAR microLED optoelectrodes were assembled on printed circuit boards (PCBs) that provide connections to the neuronal

162

Kanghwan Kim et al.

signal recording IC and LED drivers. Four-layer PCBs were used on which the traces for the recorded neuronal signals and the LED drive signals were separated by two ground-connected internal layers. Fig. 17 shows a miniSTAR microLED optoelectrode mounted on a printed circuit board. The miniSTAR microLED optoelectrode consists of four shanks, on whose tips the LEDs and the electrodes are evenly distributed. Each of the 5-mm long, 66-μm wide, 30-μm thick shanks contains 8 recording electrodes and 3 LEDs. The exposed surface areas of the microLEDs and the recording electrodes are the same as those on the prototype μLED optoelectrode: 10 μm  15 μm for the LEDs and 11 μm  13 μm for the electrode sites (both W  L). 5.3.2 Transient pulse shaping for artifact-free optoelectrophysiology Bench-top experiments using fabricated miniSTAR optoelectrodes showed that the mean magnitude of the stimulation artifact was greatly reduced with the combined multi-metal-layer structure and high substrate doping (Kim et al., 2020). However, some channels, especially those showing signals

Fig. 17 A photograph of a fabricated miniSTAR microLED optoelectrode mounted on a four-layer printed circuit board. The inset shows a microphotograph of a tip of the optoelectrode. Adapted with permission from Kim et al. (2020); Copyright 2020, Springer Nature.

GaN-on-Si μLED optoelectrodes

163

recorded from electrodes closer to the microLEDs at the center of the shank, still recorded large-amplitude artifacts. An investigation of the distribution of the shape and magnitude of the artifact confirmed that EMI generated from the LEDs contributed to the artifact (Kim et al., 2020), and techniques that can reduce the magnitude of the capacitively-coupled artifact current (Kim et al., 2020; Mendrela et al., 2018b) were adopted to further reduce the artifact magnitude so that it became lower than the detectable threshold voltage (50 μV). The final EMI reduction was carried out by reducing the slew rate of the voltage pulses supplied to the LEDs. First, considering that the typical time response of opsins is in the order of several milliseconds (Boyden et al., 2005), the rise (and the fall) time of pulses was increased to as long as 1 ms. In addition, using the fact that an LED does not allow current flow until the turn-on voltage is reached, the off-time voltages of the LED anodes were set to approximately 2.5 V, just below the turn-on voltages of the LEDs. The combination of these two techniques to reduce the slew rate of the pulses resulted in an additional reduction of the stimulation artifact magnitude, measured using the bench-top setup, to less than 50 μV peak-to-peak on all electrodes (Kim et al., 2020). 5.3.3 In vivo demonstration of artifact-free optoelectrophysiology The successful elimination of the stimulation artifact was validated in vivo using a virus-injected animal. A mouse was injected with CaMKIIa promotor driven ChR2 virus (AAV5-CaMKIIa-hChR2(H134R)-EYFP) in its hippocampus so that the pyramidal neurons express ChR2 on their membranes. A miniSTAR microLED optoelectrode was utilized to elicit and record light-induced activity, and no discernable artifact was recorded on any of the channels. This demonstration of artifact-free recording was carried out with pulses having 1-ms rise/fall times. While all the three LEDs on a shank were turned on (w/ 460-nW radiant flux from each) and off in a sequence, no artifact was detected while the status of each LED was toggled, and the absence of the artifact was further validated with the detection of distortion-free spike waveforms whose timestamps corresponded to those of the LED switching timestamps (Kim et al., 2020). Fig. 18 shows example traces of a stimulation-artifact-free recording session. No adverse effects due to the artifact reduction schemes were observed during the in vivo session.

164

Kanghwan Kim et al.

Fig. 18 Demonstration of the absence of the stimulation artifact on the miniSTAR microLED optoelectrode. Insets show the zoomed-in view of the detected spike (top) and that overriding the traces of the 10 spontaneous spikes from the same unit (bottom). No deformation of the stimulation artifact waveform is observed. Adapted with permission from Kim et al. (2020); Copyright 2020, Springer Nature.

6. Discussion MicroLED optoelectrodes have demonstrated their ability to enable studies of functional connectivity within neural circuits for the first time. These devices are now being disseminated to a number of neuroscience laboratories through an NSF-funded program. It is expected that these devices, fabricated in the miniSTAR configuration, will continue to enable numerous groundbreaking discoveries. There still are a few improvements that can be made in the microLED optoelectrodes, while preserving their identity as devices on which microLEDs and signal recording electrodes are monolithically integrated at very high density. These include multi-color optical stimulation capability, increasing the number of the sites for both optical stimulation and electrical recording, and fabricating the devices on flexible platforms. It will be a challenging task to add the capability to generate photons with different energies (i.e., emit lights with multiple wavelengths) using monolithically integrated microLEDs. Multi-color optical stimulation, as Kampasi et al. demonstrated, could enable multi-modal neuronal activity (e.g., optical excitation with blue light combined with optical inhibition with red light, or optical excitation of two different populations of neurons with photons having different wavelengths) and would facilitate analyses of highly heterogeneous neural circuits (Kampasi et al., 2018). Thankfully, a

GaN-on-Si μLED optoelectrodes

165

number of techniques for the monolithic integration of multi-color LEDs have been recently developed (Chung et al., 2017; Liu et al., 2020; Robin et al., 2019). Once the efficiencies of these monolithically integrated LEDs become comparable to those of the planar GaN/InGaN microLEDs on the microLED optoelectrodes, they can be readily incorporated into the microLED optoelectrode fabrication process to realize multi-color microLED optoelectrodes. While it sounds easy to integrate more LEDs onto the microLED optoelectrodes utilizing batch photolithographic patterning processes, it is challenging to integrate more than a few LEDs on the shank of an optoelectrode. In order to minimize the damage inflicted on the brain tissue, the dimensions (especially the width) of the implanted portion of a probe must be minimized (e.g., to less than 100 μm). Because of limitations in the patterning processes, it is challenging to integrate more than a few LEDs along with the metal lines that provide the electrical connections to the LEDs onto a narrow shank. In the near future, however, techniques that may enable the definition of higher-density metal patterns can probably be adopted to enable the fabrication of large-scale microLED optoelectrodes that maximize the utility of these devices. Finally, the fabrication of flexible microLED optoelectrodes faces a few challenges as well. As already demonstrated by a few research groups, it is possible to release the GaN LED layer from its substrate (sapphire or silicon), batch transfer them onto a new substrate, and then build a device on the new substrate (Ayub et al., 2016, 2017; Reddy et al., 2019). However, considering the inferiority of polymers as water- and ion-barriers, it is unlikely that microLEDs transferred onto a flexible platform and then fabricated thereon will perform as robustly as rigid devices in the harsh in vivo environment. Continued efforts to develop water- and ion-barrier-formation techniques for flexible devices must precede the development of flexible microLED optoelectrodes.

7. Conclusion This chapter has introduced microLED optoelectrodes and discussed the factors that have enabled their successful implementation, especially the maturation of GaN-on-Si epitaxial growth techniques. MicroLED optoelectrodes are fabricated with GaN-on-Si GaN/InGaN MQW wafers having a 460-nm emission wavelength, which is suitable for the excitation

166

Kanghwan Kim et al.

of the most widely used opsin, ChR2. Fabricated microLED optoelectrodes, the first-of-a-kind optoelectrodes with monolithically integrated cell-sized LEDs and signal-recording electrodes, have enabled the precision analysis of the neuronal activity in the brain with a spatial resolution of 60 μm. With techniques such as modification of the doping density of the silicon substrate, the stimulation artifact generated by optical stimulation has been eliminated and the spatial resolution that opto-electrophysiology microLED optoelectrodes provide has greatly improved. Thanks to the low cost of quality GaN-on-Si LED wafers, microLED optoelectrodes are now being fabricated in batch and disseminated worldwide.

References Abaya, T.V.F., Blair, S., Tathireddy, P., Rieth, L., Solzbacher, F., 2012. A 3D glass optrode array for optical neural stimulation. Biomed. Opt. Express 3, 3087–3104. Anikeeva, P., Andalman, A.S., Witten, I., Warden, M., Goshen, I., Grosenick, L., Gunaydin, L.A., Frank, L.M., Deisseroth, K., 2012. Optetrode: a multichannel readout for optogenetic control in freely moving mice. Nat. Neurosci. 15, 163–170. Aravanis, A.M., Wang, L.-P., Zhang, F., Meltzer, L.A., Mogri, M.Z., Schneider, M.B., Deisseroth, K., 2007. An optical neural interface: in vivo control of rodent motor cortex with integrated fiberoptic and optogenetic technology. J. Neural Eng. 4, S143–S156. Ayling, O.G.S., Harrison, T.C., Boyd, J.D., Goroshkov, A., Murphy, T.H., 2009. Automated light-based mapping of motor cortex by photoactivation of channelrhodopsin-2 transgenic mice. Nat. Methods 6, 219–224. Ayub, S., Gossler, C., Schwaerzle, M., Klein, E., Paul, O., Schwarz, U.T., Ruther, P., 2016. High-density probe with integrated thin-film micro light emitting diodes (μLEDs) for optogenetic applications. In: 2016 IEEE 29th International Conference on Micro Electro Mechanical Systems (MEMS), 24–28 Jan. 2016, pp. 379–382. Ayub, S., Gentet, L.J., Fia´th, R., Schwaerzle, M., Borel, M., David, F., Bartho´, P., Ulbert, I., Paul, O., Ruther, P., 2017. Hybrid intracerebral probe with integrated bare LED chips for optogenetic studies. Biomed. Microdevices 19, 49. Barker, A.T., Jalinous, R., Freeston, I.L., 1985. Non-invasive magnetic stimulation of human motor cortex. Lancet 325, 1106–1107. Bito, L., Davson, H., Levin, E., Murray, M., Snider, N., 1966. The concentrations of free amino acids and other electrolytes in cerebrospinal fluid, in vivo dialysate of brain, and blood plasma of the dog. J. Neurochem. 13, 1057–1067. Boyden, E.S., Zhang, F., Bamberg, E., Nagel, G., Deisseroth, K., 2005. Millisecondtimescale, genetically targeted optical control of neural activity. Nat. Neurosci. 8, 1263. Budai, D., Vizva´ri, A.D., Bali, Z.K., Ma´rki, B., Nagy, L.V., Ko´nya, Z., Madara´sz, D., Henn-Mike, N., Varga, C., Herna´di, I., 2018. A novel carbon tipped single micro-optrode for combined optogenetics and electrophysiology. PLoS One 13, e0193836. Buzsa´ki, G., 2004. Large-scale recording of neuronal ensembles. Nat. Neurosci. 7, 446. Buzsa´ki, G., Stark, E., Ber
enyi, A., Khodagholy, D., Kipke, D.R., Yoon, E., Wise, K.D., 2015. Tools for probing local circuits: high-density silicon probes combined with Optogenetics. Neuron 86, 92–105. Canales, A., Jia, X., Froriep, U.P., Koppes, R.A., Tringides, C.M., Selvidge, J., Lu, C., Hou, C., Wei, L., Fink, Y., Anikeeva, P., 2015. Multifunctional fibers for simultaneous optical, electrical and chemical interrogation of neural circuits in vivo. Nat. Biotechnol. 33, 277–284.

GaN-on-Si μLED optoelectrodes

167

Cao, H., Gu, L., Mohanty, S.K., Chiao, J., 2013. An integrated μLED Optrode for Optogenetic stimulation and electrical recording. IEEE Trans. Biomed. Eng. 60, 225–229. Cardin, J.A., Carl
en, M., Meletis, K., Knoblich, U., Zhang, F., Deisseroth, K., Tsai, L.-H., Moore, C.I., 2010. Targeted optogenetic stimulation and recording of neurons in vivo using cell-type-specific expression of Channelrhodopsin-2. Nat. Protoc. 5, 247–254. Chen, X., Ye, H., Kuruvilla, R., Ramanan, N., Scangos, K.W., Zhang, C., Johnson, N.M., England, P.M., Shokat, K.M., Ginty, D.D., 2005. A chemical-genetic approach to studying neurotrophin signaling. Neuron 46, 13–21. Chung, K., Sui, J., Demory, B., Ku, P.-C., 2017. Color mixing from monolithically integrated InGaN-based light-emitting diodes by local strain engineering. Appl. Phys. Lett. 111, 041101. Chung, J.E., Joo, H.R., Fan, J.L., Liu, D.F., Barnett, A.H., Chen, S., Geaghan-Breiner, C., Karlsson, M.P., Karlsson, M., Lee, K.Y., Liang, H., Magland, J.F., Pebbles, J.A., Tooker, A.C., Greengard, L.F., Tolosa, V.M., Frank, L.M., 2019. High-density, long-lasting, and multi-region electrophysiological recordings using polymer electrode arrays. Neuron 101, 21–31. e5. Dadgar, A., Bl€asing, J., Diez, A., Alam, A., Heuken, M., Krost, A., 2000. Metalorganic chemical vapor phase epitaxy of crack-free GaN on Si (111) exceeding 1 μm in thickness. Jpn. J. Appl. Phys. 39, L1183–L1185. Deisseroth, K., 2011. Optogenetics. Nat. Methods 8, 26–29. Fan, B., Kwon, K.Y., Weber, A.J., Li, W., 2014. An implantable, miniaturized SU-8 optical probe for optogenetics-based deep brain stimulation. In: 2014 36th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, 26–30 Aug. 2014, pp. 450–453. Fossum, J.G., Lee, D.S., 1982. A physical model for the dependence of carrier lifetime on doping density in nondegenerate silicon. Solid-State Electron. 25, 741–747. € Fritsch, G., 1870. Uber die elektrische Erregbarkeit des Grosshirns. Arch. Anat. Physiol. Wiss. Med. 37, 300–332. Galvani, L., 1791. De Viribus Electricitatis in Motu Musculari Commentarius. Ex Typographia Instituti Scientiarium, Bononiae. Gradinaru, V., Thompson, K.R., Zhang, F., Mogri, M., Kay, K., Schneider, M.B., Deisseroth, K., 2007. Targeting and readout strategies for fast optical neural control in vitro and in vivo. J. Neurosci. 27, 14231. Guha, S., Bojarczuk, N.A., 1998. Ultraviolet and violet GaN light emitting diodes on silicon. Appl. Phys. Lett. 72, 415–417. Han, X., Qian, X., Bernstein, J.G., Zhou, H.-H., Franzesi, G.T., Stern, P., Bronson, R.T., Graybiel, A.M., Desimone, R., Boyden, E.S., 2009. Millisecond-timescale optical control of neural dynamics in the nonhuman primate brain. Neuron 62, 191–198. Harvey, E.N., 1929. The effect of high frequency sound waves on heart muscle and other irritable tissues. Am. J. Physiol. Legacy Content 91, 284–290. Henze, D.A., Borhegyi, Z., Csicsvari, J., Mamiya, A., Harris, K.D., Buzsa´ki, G., 2000. Intracellular features predicted by extracellular recordings in the hippocampus in vivo. J. Neurophysiol. 84, 390–400. Hong, G., Lieber, C.M., 2019. Novel electrode technologies for neural recordings. Nat. Rev. Neurosci. 20, 330–345. Im, M., Cho, I.-J., Wu, F., Wise, K.D., Yoon, E., 2011. Neural probes integrated with optical mixer/splitter waveguides and multiple stimulation sites. In: 2011 IEEE 24th International Conference on Micro Electro Mechanical Systems, 23–27 Jan. 2011, pp. 1051–1054. Jones, K.E., Campbell, P.K., Normann, R.A., 1992. A glass/silicon composite intracortical electrode array. Ann. Biomed. Eng. 20, 423–437.

168

Kanghwan Kim et al.

Jun, J.J., Steinmetz, N.A., Siegle, J.H., Denman, D.J., Bauza, M., Barbarits, B., Lee, A.K., Anastassiou, C.A., Andrei, A., Aydın, C ¸ ., Barbic, M., Blanche, T.J., Bonin, V., Couto, J., Dutta, B., Gratiy, S.L., Gutnisky, D.A., H€ausser, M., Karsh, B., Ledochowitsch, P., Lopez, C.M., Mitelut, C., Musa, S., Okun, M., Pachitariu, M., Putzeys, J., Rich, P.D., Rossant, C., Sun, W.-L., Svoboda, K., Carandini, M., Harris, K.D., Koch, C., O’Keefe, J., Harris, T.D., 2017. Fully integrated silicon probes for high-density recording of neural activity. Nature 551, 232–236. Kampasi, K., Stark, E., Seymour, J., Na, K., Winful, H.G., Buzsa´ki, G., Wise, K.D., Yoon, E., 2016. Fiberless multicolor neural optoelectrode for in vivo circuit analysis. Sci. Rep. 6, 30961. Kampasi, K., English, D.F., Seymour, J., Stark, E., McKenzie, S., V€ or€ oslakos, M., Buzsa´ki, G., Wise, K.D., Yoon, E., 2018. Dual color optogenetic control of neural populations using low-noise, multishank optoelectrodes. Microsyst. Nanoeng. 4, 10. Khodagholy, D., Gelinas, J.N., Thesen, T., Doyle, W., Devinsky, O., Malliaras, G.G., Buzsa´ki, G., 2015. NeuroGrid: recording action potentials from the surface of the brain. Nat. Neurosci. 18, 310–315. Khurram, A., Seymour, J.P., 2013. Investigation of the photoelectrochemical effect in optoelectrodes and potential uses for implantable electrode characterization*. In: 2013 35th Annual International Conference of the Ieee Engineering in Medicine and Biology Society (EMBC), 3–7 July 2013, pp. 3032–3035. Kim, J.-Y., Tak, Y., Kim, J., Hong, H.-G., Chae, S., Lee, J. W., Choi, H., Park, Y., Chung, U.-I., Kim, J.-R., Shim, J.-I., 2012. Highly efficient InGaN/GaN blue LED on 8-inch Si (111) substrate. In: Proc. SPIE 8262, Gallium Nitride Materials and Devices VII, 8262, 21–26 Jan 2012, 82621D-1-9. Kim, T.-I., McCall, J.G., Jung, Y.H., Huang, X., Siuda, E.R., Li, Y., Song, J., Song, Y.M., Pao, H.A., Kim, R.-H., Lu, C., Lee, S.D., Song, I.-S., Shin, G., Al-Hasani, R., Kim, S., Tan, M.P., Huang, Y., Omenetto, F.G., Rogers, J.A., Bruchas, M.R., 2013. Injectable, cellular-scale optoelectronics with applications for wireless Optogenetics. Science 340, 211. Kim, K., English, D., Mckenzie, S., Wu, F., Stark, E., Seymour, J., Ku, P., Wise, K., Buzsa´ki, G., Yoon, E., 2016. GaN-on-Si μLED optoelectrodes for high-spatiotemporal-accuracy optogenetics in freely behaving animals. In: 2016 IEEE International Electron Devices Meeting (IEDM), 3–7 Dec. 2016. 26.5.1–26.5.4. Kim, K., V€ or€ oslakos, M., Seymour, J.P., Wise, K.D., Buzsa´ki, G., Yoon, E., 2020. Artifact-free and high-temporal-resolution in vivo opto-electrophysiology with microLED optoelectrodes. Nat. Commun. 11, 2063. Kissinger, P.T., Hart, J.B., Adams, R.N., 1973. Voltammetry in brain tissue—a new neurophysiological measurement. Brain Res. 55, 209–213. Kobayashi, R., Kanno, S., Sakai, S., Lee, S., Koyanagi, M., Yao, H., Tanaka, T., 2011. Development of Si neural probe with optical waveguide for highly accurate optical stimulation of neuron. In: 2011 5th International IEEE/EMBS Conference on Neural Engineering, 27 April-1 May 2011, pp. 294–297. Kozai, T.D.Y., Vazquez, A.L., 2015. Photoelectric artefact from optogenetics and imaging on microelectrodes and bioelectronics: new challenges and opportunities. J. Mater. Chem. B 3, 4965–4978. Krnjevi
c, K., 1974. Chemical nature of synaptic transmission in vertebrates. Physiol. Rev. 54, 418–540. Kucytowski, J., Wokulska, K., 2005. Lattice parameter measurements of boron doped Si single crystals. Cryst. Res. Technol. 40, 424–428. Kwon, K.Y., Lee, H.-M., Ghovanloo, M., Weber, A., Li, W., 2015. Design, fabrication, and packaging of an integrated, wirelessly-powered optrode array for optogenetics application. Front. Syst. Neurosci. 9, 69.

GaN-on-Si μLED optoelectrodes

169

Laxpati, N.G., Mahmoudi, B., Gutekunst, C.-A., Newman, J.P., Zeller-Townson, R., Gross, R.E., 2014. Real-time in vivo optogenetic neuromodulation and multielectrode electrophysiologic recording with NeuroRighter. Front. Neuroeng. 7, 40. LeChasseur, Y., Dufour, S., Lavertu, G., Bories, C., Desch^enes, M., Vall
ee, R., De Koninck, Y., 2011. A microprobe for parallel optical and electrical recordings from single neurons in vivo. Nat. Methods 8, 319–325. Lee, J., Ozden, I., Song, Y.-K., Nurmikko, A.V., 2015a. Transparent intracortical microprobe array for simultaneous spatiotemporal optical stimulation and multichannel electrical recording. Nat. Methods 12, 1157–1162. Lee, H.J., Son, Y., Kim, J., Lee, C.J., Yoon, E.-S., Cho, I.-J., 2015b. A multichannel neural probe with embedded microfluidic channels for simultaneous in vivo neural recording and drug delivery. Lab Chip 15, 1590–1597. Li, Y., Baek, K., Gulari, M., Wise, K.D., 2007. A drug-delivery probe with an in-line flowmeter based on trench refill and chemical mechanical polishing techniques. In: SENSORS, 2007 IEEE, 28–31 Oct 2007, pp. 1144–1147. Libbrecht, S., Hoffman, L., Welkenhuysen, M., van den Haute, C., Baekelandt, V., Braeken, D., Haesler, S., 2018. Proximal and distal modulation of neural activity by spatially confined optogenetic activation with an integrated high-density optoelectrode. J. Neurophysiol. 120, 149–161. Liu, X., Lu, Y., Iseri, E., Shi, Y., Kuzum, D., 2018. A Compact Closed-Loop Optogenetics System Based on Artifact-Free Transparent Graphene Electrodes. Front. Neurosci. 12, 132. Liu, X., Wu, Y., Malhotra, Y., Sun, Y., Ra, Y.-H., Wang, R., Stevenson, M., Coe-Sullivan, S., Mi, Z., 2020. Submicron full-color LED pixels for microdisplays and micro-LED main displays. J. Soc. Inf. Disp. 28, 410–417. Lopez, C.M., Putzeys, J., Raducanu, B.C., Ballini, M., Wang, S., Andrei, A., Rochus, V., Vandebriel, R., Severi, S., Van Hoof, C., Musa, S., Van Helleputte, N., Yazicioglu, R.F., Mitra, S., 2017. A neural probe with up to 966 electrodes and up to 384 configurable channels in 0.13 μm SOI CMOS. IEEE Trans. Biomed. Circuits Syst. 11, 510–522. Martinez, J, Pedreira, C, Ison, M.J., Quiroga, R.Q., 2009. Realistic simulation of extracellular recordings. J. Neurosci. Methods 184 (2), 286–293. McAlinden, N., Massoubre, D., Richardson, E., Gu, E., Sakata, S., Dawson, M.D., Mathieson, K., 2013. Thermal and optical characterization of micro-LED probes for in vivo optogenetic neural stimulation. Opt. Lett. 38, 992–994. Mendrela, A.E., Park, S., V€ or€ oslakos, M., Flynn, M.P., Yoon, E., 2018a. A battery-powered opto-electrophysiology neural interface with artifact-preventing optical pulse shaping. In: 2018 IEEE Symposium on VLSI Circuits, 18–22 June 2018, pp. 125–126. Mendrela, A.E., Kim, K., English, D., McKenzie, S., Seymour, J.P., Buzsa´ki, G., Yoon, E., 2018b. A high-resolution Opto-electrophysiology system with a miniature integrated Headstage. IEEE Trans. Biomed. Circuits Syst. 12, 1065–1075. Mikulovic, S., Pupe, S., Peixoto, H.M., Do Nascimento, G.C., Kullander, K., Tort, A.B.L., Lea˜o, R.N., 2016. On the photovoltaic effect in local field potential recordings. Neurophotonics 3 (1), 015002. Miyawaki, A., Llopis, J., Heim, R., McCaffery, J.M., Adams, J.A., Ikura, M., Tsien, R.Y., 1997. Fluorescent indicators for Ca2+ based on green fluorescent proteins and calmodulin. Nature 388, 882–887. Mohanty, A., Li, Q., Tadayon, M.A., Roberts, S.P., Bhatt, G.R., Shim, E., Ji, X., Cardenas, J., Miller, S.A., Kepecs, A., Lipson, M., 2020. Reconfigurable nanophotonic silicon probes for sub-millisecond deep-brain optical stimulation. Nat. Biomed. Eng. 4, 223–231.

170

Kanghwan Kim et al.

Najafi, K., Wise, K.D., Mochizuki, T., 1985. A high-yield IC-compatible multichannel recording array. IEEE Trans. Electron Dev. 32, 1206–1211. Nakayama, T., Hammel, H.T., Hardy, J.D., Eisenman, J.S., 1963. Thermal stimulation of electrical activity of single units of the preoptic region. Am. J. Physiol. 204 (6), 1122–1126. Nobili, L., 1828. Comparaison entre les deux galvanometres les plus sensibles, la grenouille et le moltiplicateur a deux aiguilles, suivie de quelques resultats noveaux. Ann. Chim. Phys. 38, 225–245. Nordhausen, C.T., Rousche, P.J., Normann, R.A., 1994. Optimizing recording capabilities of the Utah Intracortical electrode Array. Brain Res. 637, 27–36. Park, H., Shin, H.-J., Cho, I.-J., Yoon, E.-S., Suh, J.-K.F., Im, M., Yoon, E., Kim, Y., Kim, J., 2011. The first neural probe integrated with light source (blue laser diode) for optical stimulation and electrical recording. In: 2011 Annual International Conference of the IEEE Engineering in Medicine and Biology Society, 30 Aug.-3 Sept. 2011, pp. 2961–2964. Ogawa, S., Lee, T.M., Kay, A.R., Tank, D.W., 1990. Brain magnetic resonance imaging with contrast dependent on blood oxygenation. Proc. Natl. Acad. Sci. 87, 9868–9872. Pama, E.A.C., Colzato, L.S., Hommel, B., 2013. Optogenetics as a neuromodulation tool in cognitive neuroscience. Front. Psychol. 4, 610. Park, D.-W., Schendel, A.A., Mikael, S., Brodnick, S.K., Richner, T.J., Ness, J.P., Hayat, M.R., Atry, F., Frye, S.T., Pashaie, R., Thongpang, S., Ma, Z., Williams, J.C., 2014. Graphene-based carbon-layered electrode array technology for neural imaging and optogenetic applications. Nat. Commun. 5, 5258. Pisanello, F., Sileo, L., Oldenburg, I.A., Pisanello, M., Martiradonna, L., Assad, J.A., Sabatini, B.L., De Vittorio, M., 2014. Multipoint-emitting optical Fibers for spatially addressable in vivo Optogenetics. Neuron 82 (6), 1245–1254. Reddy, J.W., Kimukin, I., Stewart, L.T., Ahmed, Z., Barth, A.L., Towe, E., Chamanzar, M., 2019. High density, double-sided, flexible optoelectronic neural probes with embedded μLEDs. Front. Neurosci. 13, 745. Robin, Y., Hemeret, F., D’Inca, G., Pristovsek, M., Trassoudaine, A., Amano, H., 2019. Monolithic integration of tricolor micro-LEDs and color mixing investigation by analog and digital dimming. Jpn. J. Appl. Phys. 58, SCCC06. Rodger, D.C., Fong, A.J., Li, W., Ameri, H., Ahuja, A.K., Gutierrez, C., Lavrov, I., Zhong, H., Menon, P.R., Meng, E., Burdick, J.W., Roy, R.R., Edgerton, V.R., Weiland, J.D., Humayun, M.S., Tai, Y.-C., 2008. Flexible parylene-based multielectrode array technology for high-density neural stimulation and recording. Sens. Actuators B 132, 449–460. Ross, B., 1980. Survey of Literature on Minority Carrier Lifetimes in Silicon and Related Topics. Lifetime Factors in Silicon. ASTM International, West Conshohocken, PA, pp. 14–28. Rossi, M.A., Go, V., Murphy, T., Fu, Q., Morizio, J., Yin, H.H., 2015. A wirelessly controlled implantable LED system for deep brain optogenetic stimulation. Front. Integr. Neurosci. 9, 8. Royer, S., Zemelman, B.V., Barbic, M., Losonczy, A., Buzsa´ki, G., Magee, J.C., 2010. Multi-array silicon probes with integrated optical fibers: light-assisted perturbation and recording of local neural circuits in the behaving animal. Eur. J. Neurosci. 31, 2279–2291. Rubehn, B., Wolff, S.B.E., Tovote, P., Schuettler, M., L€ uthi, A., Stieglitz, T., 2011. Polymer-based shaft microelectrodes with optical and fluidic capabilities as a tool for optogenetics. In: 2011 Annual International Conference of the IEEE Engineering in Medicine and Biology Society, 30 Aug.-3 Sept. 2011, pp. 2969–2972.

GaN-on-Si μLED optoelectrodes

171

Scharf, R., Tsunematsu, T., McAlinden, N., Dawson, M.D., Sakata, S., Mathieson, K., 2016. Depth-specific optogenetic control in vivo with a scalable, high-density μLED neural probe. Sci. Rep. 6, 28381. Scholvin, J., Kinney, J.P., Bernstein, J.G., Moore-Kochlacs, C., Kopell, N., Fonstad, C.G., Boyden, E.S., 2015. Close-packed silicon microelectrodes for scalable spatially oversampled neural recording. IEEE Trans. Biomed. Eng. 63, 120–130. Schwaerzle, M., Paul, O., Ruther, P., 2017. Compact silicon-based optrode with integrated laser diode chips, SU-8 waveguides and platinum electrodes for optogenetic applications. J. Micromech. Microeng. 27, 065004. Segev, E., Reimer, J., Moreaux, L.C., Fowler, T.M., Chi, D., Sacher, W.D., Lo, M., Deisseroth, K., Tolias, A.S., Faraon, A., Roukes, M.L., 2016. Patterned photostimulation via visible-wavelength photonic probes for deep brain optogenetics. Neurophotonics 4, 1–15. Sharma, H.S., Hoopes, P.J., 2003. Hyperthermia induced pathophysiology of the central nervous system. Int. J. Hyperthermia 19, 325–354. Shim, E., Chen, Y., Masmanidis, S., Li, M., 2016. Multisite silicon neural probes with integrated silicon nitride waveguides and gratings for optogenetic applications. Sci. Rep. 6, 22693. Siegel, M.S., Isacoff, E.Y., 1997. A genetically encoded optical probe of membrane voltage. Neuron 19, 735–741. Son, Y., Lee, H.J., Kim, J., Shin, H., Choi, N., Lee, C.J., Yoon, E.-S., Yoon, E., Wise, K.D., Kim, T.G., Cho, I.-J., 2015. In vivo optical modulation of neural signals using monolithically integrated two-dimensional neural probe arrays. Sci. Rep. 5, 15466. Sparta, D.R., Stamatakis, A.M., Phillips, J.L., Hovelsø, N., van Zessen, R., Stuber, G.D., 2012. Construction of implantable optical fibers for long-term optogenetic manipulation of neural circuits. Nat. Protoc. 7, 12–23. Stark, E., Koos, T., Buzsa´ki, G., 2012. Diode probes for spatiotemporal optical control of multiple neurons in freely moving animals. J. Neurophysiol. 108, 349–363. Stevenson, I.H., Kording, K.P., 2011. How advances in neural recording affect data analysis. Nat. Neurosci. 14, 139–142. Tran, C.A., Osinski, A., Karlicek, R.F., Berishev, I., 1999. Growth of InGaN/GaN multiple-quantum-well blue light-emitting diodes on silicon by metalorganic vapor phase epitaxy. Appl. Phys. Lett. 75, 1494–1496. Wang, J., Wagner, F., Borton, D.A., Zhang, J., Ozden, I., Burwell, R.D., Nurmikko, A.V., van Wagenen, R., Diester, I., Deisseroth, K., 2011. Integrated device for combined optical neuromodulation and electrical recording for chronic in vivo applications. J. Neural Eng. 9, 016001. Wang, L., Ge, C., Wang, M., Ji, B., Guo, Z., Wang, X., Yang, B., Li, C., Liu, J., 2020. An artefact-resist optrode with internal shielding structure for low-noise neural modulation. J. Neural Eng. 17, 046024. Wise, K.D., Angell, J.B., 1975. A low-capacitance multielectrode probe for use in extracellular neurophysiology. IEEE Trans. Biomed. Eng. BME-22, 212–219. Wise, K.D., Angell, J.B., Starr, A., 1970. An integrated-circuit approach to extracellular microelectrodes. IEEE Trans. Biomed. Eng. BME-17, 238–247. Wise, K.D., Sodagar, A.M., Yao, Y., Gulari, M.N., Perlin, G.E., Najafi, K., 2008. Microelectrodes, Microelectronics, and Implantable Neural Microsystems. Proc. IEEE 96 (7), 1184–1202. Wu, F., Stark, E., Im, M., Cho, I.-J., Yoon, E.-S., Buzsa´ki, G., Wise, K.D., Yoon, E., 2013. An implantable neural probe with monolithically integrated dielectric waveguide and recording electrodes for optogenetics applications. J. Neural Eng. 10, 056012. Wu, F., Stark, E., Ku, P.-C., Wise, K.D., Buzsa´ki, G., Yoon, E., 2015. Monolithically integrated μLEDs on silicon neural probes for high-resolution Optogenetic studies in behaving animals. Neuron 88, 1136–1148.

172

Kanghwan Kim et al.

Zhang, J., Laiwalla, F., Kim, J.A., Urabe, H., van Wagenen, R., Song, Y.-K., Connors, B.W., Zhang, F., Deisseroth, K., Nurmikko, A.V., 2009. Integrated device for optical stimulation and spatiotemporal electrical recording of neural activity in light-sensitized brain tissue. J. Neural Eng. 6, 055007. Zhu, D., McAleese, C., McLaughlin, K.K., H€aberlen, M., Salcianu, C. O., Thrush, E. J., Kappers, M. J., Phillips, W. A., Lane, P., Wallis, D. J., Martin, T., Astles, M., Thomas, S., Pakes, A., Heuken, M., Humphreys, C. J., 2009. GaN-based LEDs grown on 6-inch diameter Si (111) substrates by MOVPE. In: Proc. SPIE 7231, Light-Emitting Diodes: Materials, Devices, and Applications for Solid State Lighting XIII, 7231, 24–29 Jan 2009, 723118-1-11. Zhu, D., McAleese, C., H€aberlen, M., Salcianu, C., Thrush, T., Kappers, M., Phillips, A., Lane, P., Kane, M., Wallis, D., Martin, T., Astles, M., Hylton, N., Dawson, P., Humphreys, C., 2011. Efficiency measurement of GaN-based quantum well and light-emitting diode structures grown on silicon substrates. J. Appl. Phys. 109, 014502. Zorzos, A.N., Boyden, E.S., Fonstad, C.G., 2010. Multiwaveguide implantable probe for light delivery to sets of distributed brain targets. Opt. Lett. 35, 4133–4135. Zorzos, A.N., Scholvin, J., Boyden, E.S., Fonstad, C.G., 2012. Three-dimensional multiwaveguide probe array for light delivery to distributed brain circuits. Opt. Lett. 37, 4841–4843.

CHAPTER FIVE

Quantum-dot-based full-color micro-LED displays Tingzhu Wua, Yu-Ming Huangb, James Singh Konthoujamb, Zhong Chena, and Hao-Chung Kuob,* a

Department of Electronic Science, Fujian Engineering Research Center for Solid-State Lighting, Xiamen University, Xiamen, China b Department of Photonics & Graduate Institute of Electro-Optical Engineering, College of Electrical and Computer Engineering, National Chiao Tung University, Hsinchu, Taiwan *Corresponding author: e-mail address: [email protected]

Contents 1. Introduction 2. Background of full-color micro-LED display 2.1 QDs photoresist patterning technique 3. Conclusion References

173 175 184 198 198

1. Introduction Quantum dots (QDs) are semiconductor nanoparticles. They show remarkable optical and electrical characteristics that differ from those of bulk semiconductor materials and are thus referred to as artificial atoms. Due to their exceptional features, QDs have found applications in a wide variety of advanced technologies, including solar cells (Han et al., 2014; Hsu et al., 2018; Pattantyus-Abraham et al., 2010; Schaller and Klimov, 2004; Tsai et al., 2013), photodetectors (Konstantatos and Sargent, 2009), photodiodes (Pal et al., 2012), field effect transistors (Koh et al., 2011), biological systems (Algar and Krull, 2008; Dahan et al., 2003; Marchuk et al., 2012; Selvin, 2014), and light-emitting diodes (LEDs) (Chen et al., 2015; Lin et al., 2016; Liu et al., 2017). QDs have been used as color down-converters for LEDs to realize high-quality displays and effective illumination sources. Quantum dots pumped both electrically and optically are used with LEDs (Huang et al., 2017; Roh et al., 2020; Shirasaki et al., 2013). This involves not only radiative transfer from

Semiconductors and Semimetals, Volume 106 ISSN 0080-8784 https://doi.org/10.1016/bs.semsem.2021.01.005

Copyright

#

2021 Elsevier Inc. All rights reserved.

173

174

Tingzhu Wu et al.

LEDs to QDs, but also Forster resonance energy transfer (FRET), which is responsible for transferring non-radiative energy from LEDs to QDs. FRET, also known as non-radiative resonant energy transfer (NRET), is strong enough to be observed when the LED’s emissive quantum wells are in close contact with the phosphor QD layer. Energy is transmitted by radiative energy transfer and NRET from active LEDs to QDs. The absorption spectrum of acceptors, i.e., QDs, must be consistent with the donor emission spectrum, i.e. the active LED, in the direct radiative energy transfer and in comparison, donors and acceptors in NRET need to be in near contact (Zhang et al., 2012). Both these mechanisms of energy transfer can generate electron-hole pairs in QDs and thus produce radiation. An effective QD-LED device must be configured in such a way that each of these mechanisms can contribute to the energy transfer process with minimal losses. Ultraviolet (UV)/blue wavelength absorbing QDs which emit blue, green, and red colors are used with a UV or blue LED for QD-LED implementation (Hsu et al., 2015; Yang et al., 2014). Researchers have proposed many innovative ideas to create better performing QD-LEDs (Liu et al., 2020). The existence of various choice of materials for QDs leads to their variability across numerous reports. The electroluminescence of QD-based displays show better performance than either QD photoluminescence (PL)-based organic LED (OLED) or liquid crystal technology and offers the best solution with wide color range and pure black color. The QD EL display (QD-LED display, also called QLED or EL-QLED display) pumps electrons and holes into QDs, where they recombine to generate photons directly for the main red, green, or blue display (Smeeton et al., 2019). Nevertheless, we are not categorizing QDs in this chapter on the basis of their material or design, as this digresses from the primary objective of our discussion, i.e., the combined use of optically excited QDs and LEDs to achieve QD-LEDs. In the following sections, lighting sources and display devices based on QD-LEDs that have been proposed in the last few years are addressed briefly. The first approach to integrate QDs in consumer displays was used in recent QD-LED televisions with edge-lit lenses (Konstantatos and Sargent, 2009). This approach used a dispersion of QDs over a strip of LEDs in a polymer embedded in a glass tube on the edges of a screen. This method has many drawbacks including decreased production temperature. The method proved challenging due to the thermal instability of early QDs and required the use of a very thick hermetic tube to ensure continued operational reliability. The QD films in displays are the latest technology of choice for QD-LCD TVs. First, cadmium selenide (CdSe) or indium phosphide (IP)-based quantum dots were imprinted on a blue LED backlight and embedded into prototypical

Quantum-dot-based full-color micro-LED displays

175

LCD matrices, which offered the benefits of QD’s color efficiency. CdSebased quantum dots have been generally used in the past few years due to its higher performance in terms of luminous efficiency and wide color-gamut (Sadeghi et al., 2019; Zhang et al., 2019). However, Cd has negative effects on the environment. QDs made of InP and perovskite are widely recognized as candidates for overcoming this issue. Cadmium-free films are the newest industry standard method for generating QDs for screens (Lee et al., 2018). However, the general consensus is that the development of next-generation QD color filters (QDCFs) will be a bottleneck on the research and development route toward QD-LEDs.

2. Background of full-color micro-LED display Micro-LEDs (μLED) have emerged as a promising display technology in the last few years. Many researchers around the globe support this technology as the best alternative for future displays (Huang et al., 2020; Lin et al., 2016; Wu et al., 2018). μLEDs have the ability to replace conventional display technologies due to the combined effects of their self-emissive feature and inorganic properties of the material. The anticipated excellent performance of μLED displays in terms of luminance, brightness, efficiency, power consumption, contrast ratio, durability, and response time make it a preferred research topic (Ahn et al., 2018; Chen et al., 2019). μLED displays are ideal for a wide range of applications, such as wearable watches, cell phones, auto head-up displays, AR/VR, micro projectors, and high-end televisions ( Jiang and Lin, 2013). For using μLEDs in a display, an array of μLEDs emitting all three primary colors, i.e. red, green, and blue (RGB) has to deploy. Mass-transfer techniques can be used to assemble full-color μLED displays into RGB LED matrices (Corbett et al., 2017). This approach to obtaining full colored μLEDs using μLED chips with primary colors concurrently comes with several critical disadvantages. Firstly, owing to the reduced emissions efficiency, a green gap is formed and there seems to be no suitable material to develop effective green emitting LEDs (Schiavon, 2015). Secondly, LEDs with different colors are commonly grown on different substrates and under various operating conditions. Therefore, their implementation requires a complex circuitry, which is costly. These issues, along with several other major challenges such as low transfer yield, slow transfer time, high manufacturing cost, and difficulty in inspection and repair, inhibit the use of the aforementioned approach to realize full-color μLEDs (Ding et al., 2019; Huang et al., 2019; Lin and Jiang, 2020; Zhou et al., 2020).

176

Tingzhu Wu et al.

An alternative method to realizing full-color LEDs is to incorporate single-color LEDs with some color-converting material, which is typically phosphor (Ahn et al., 2019; Oh et al., 2011; Unithrattil et al., 2014). For LEDs, phosphor-assisted color down-conversion is a mechanism in which after energy down-conversion, a phosphor material absorbs short wavelengths and re-emits relatively longer wavelengths. A blue or violet LED is commonly used in conjunction with a small amount of yellow-emitting phosphor. The colors emitted are tunable by configuring the proportions of direct emissions from the LED and the phosphor content, as well as by adjusting the composition of the phosphor. Industrial white LED sources are typically obtained by mixing single-color nitride LEDs with Y3Al5O12: Ce3+ (YAG: Ce) yellow phosphor emitters. Due to their low energy consumption, this type of white LEDs (WLEDs) used for illumination purposes are dominant in the market. White LEDs based on YAG: Ce yellow phosphor still suffer from significant drawbacks such as poor color quality and low spectral efficiency, which limits their application range (Chen et al., 2013; Kim et al., 2004; Narendran et al., 2004). In the past few years, optically pumped QDs have been introduced as an advanced and innovative form of phosphor. QDs based LEDs (QD-LEDs) find their applications in a wide variety of innovations such as general lighting, backlighting displays, and self-emissive screens (Lai et al., 2018; Li et al., 2019; Yang et al., 2019). QDs have several potential advantages in terms of color quality, performance, and versatile implementation in optoelectronic systems. The benefits of QDs as a color-converting material have been discussed in detail in the section below. By integrating QDs as color-converter components, the output performance of LEDs can be dramatically improved and a color-converted hybrid LED emitting at a longer wavelength can be produced by using the optical and electrical properties of QDs. The process of conversion of colors with the use of QDs is analogous to the conversion of colors using phosphors. However, the phosphor bandwidth emission is very high compared to QDs. The emissions of QDs have FWHM of around 40 nm which is substantially smaller than phosphor emissions. For the proper performance of a hybrid color-converted device, optical absorption is an essential property, and hence the colorconversion materials should have decent absorption to demonstrate high efficiency during the color conversion cycle. QDs can absorb almost all photons with wavelengths slightly shorter than their emission wavelength. Therefore, these exceptional attributes of QDs make them very promising candidates as an efficient color converter.

Quantum-dot-based full-color micro-LED displays

177

Colloidal QDs (CQDs) are used as color down-converters for LEDs to realize efficient illumination sources and high-quality displays. Both electrically and optically pumped QDs are used with light-emitting devices. If CQDs are selected for RGB pixels, care must be taken while choosing the deposition process. There are different methods for depositing QDs for display technology and light emitting applications, including spin coating, mist coating, pulse spray, stamping, and inkjet printing. As mentioned earlier, QD-LED-based display technology is the ultimate choice for future generation displays because of their compelling potential advantages in terms of efficiency, power utilization, contrast ratio, lifetime, and response time. For a higher resolution, LEDs with a size smaller than 100 μm (known as μLEDs) must be used. Hence, displays realized utilizing QDs and μLEDs are often referred to as QD-μLED displays. In 2015, Han et al. (Han et al., 2015) combined ultra-violet (UV) μLED and colloidal QDs to achieve a full-color display. GaN-based UV μLEDs with a small pitch of 40 μm were combined with CQDs of all three primary colors, i.e., red, green, and blue. RGB QDs were deposited on an array of UV LEDs with the help of an aerosol jet (AJ) printer to ensure fine printing that is highly precise and mask-less and to enable non-contact deposition of liquids containing functional materials. The QD solution is aerosolized and entrained in a gas stream in the aerosol jet printing method. There are two methods to generate aerosols: pneumatic and ultrasonic. The flux of aerosols is guided to a print head where a coaxial sheath gas flow aerodynamically focuses it. The proportion of carrier gas flow rate and sheath gas flow rate is the most critical parameter in determining the correct linewidth for spraying. The system parameters need to be optimized during the QD spray. The RGB QDs were sprayed onto the μLED array surface using aerosol jet system in red, green, and blue series, as shown in Fig. 1A. A computer program specifically controlled the spraying process, and the quantity of the QDs could be tracked instantly. A distributed Bragg reflector (DBR) was used to cap the top of the μLED array to reduce the leaked UV light for display applications. A pronounced UV peak (at 395 nm) is an indicator of less efficient pumping whereas the red, green, and blue signals are not strong. With the incorporation of DBR, the strong UV band reflection effectively suppressed the peak of 395 nm and raised the visible intensities by 194% (blue), 173% (green), and 183% (red). Fig. 1B shows the PL spectra of the QDs samples with and without DBR indicating that with DBR, the intensity is higher by increasing the degree of photoexcitation.

178

Tingzhu Wu et al.

Fig. 1 (A) The photo image of the printed QDs on a μLED array under fluorescence microscopy. (B) The relative PL intensity of the sprayed QDs samples with and without DBR. Figures reproduced with permission from Optical Society of America.

Further, in order to solve the crosstalk effect, in 2017, Lin et al. proposed a photoresist mold with a blocking wall to prevent the leakage of UV light and to confine the quantum dots (Lin et al., 2017). The photoresist mold with blocking wall was formed using a simple lithographic technique by using the AZ 5214-E photoresist. The window size is the same as that of the μLEDs and the unetched area forms the blocking wall which is the trench shape of μLEDs. The PR mold is similar to the QD-μLED array of 35  35 μm2 area, comprising of small emissive sections with a pitch size of 40 μm, as shown in Fig. 2A and B. The PR mold covering could potentially reduce the crosstalk of the QDs, and the sidewall is silver coated to reflect the leakage of light from the sidewall and prevent new crosstalk from occurring. The PR mold can limit the QD area of the intended channel and reflect the light leakage from the sidewall with the silver coating as shown in Fig. 2C. This study also resolved the coffee-ring effect on the μLED due to the PR layer deposition. The major coffee-ring effect means that when the QD droplets spread on a solid surface, the ring-like QD particles deposit around the edge. Evaporation of the liquid at the perimeter is faster than at the middle because the perimeter is thinner than the middle. The height profile must therefore preserve the shape of the spherical cap determined by surface tension and must avoid shrinking. During evaporation, the QD solution would float outwards and the QD particles were taken to the surface by the solution. On evaporation, the QD particles were deposited along the perimeter and would produce a coffee-ring. The mechanism is shown in Fig. 3A. Fig. 3C shows the optical microscope (OM) image of the GaN substrate with QD drops, which was observed to produce the

Quantum-dot-based full-color micro-LED displays

179

Fig. 2 (A) Optical microscopy image of 35 μm  35 μm PR square windows. (B) Top-view image of the μLED layout with the pitch of 40.2 μm defined as the intended channel. (C) The QD droplets jetted in the PR mold to confine and resolve the crosstalk effect. Figures reproduced with permission from Optical Society of America.

Fig. 3 (A) Mechanism of the coffee ring effect. (B) Mechanism of the proposed solution to reduce the coffee ring effect by using the PR mold. (C) The coffee ring effect can be observed in the OM image when the QD drops were jetted on the GaN substrate. (D) The PR square can confine the QD drops in the hole and resolve the coffee ring effect. Figures reproduced with permission from Optical Society of America.

coffee ring effect. The mechanism of the PR mold proposed to resolve the coffee-ring effect is shown in Fig. 3B. The mold has been designed to ensure similar liquid thickness along the perimeter and the middle by confining the solution. A comparable thickness of the solvent would induce comparable accelerated evaporation and would prevent the solution from

180

Tingzhu Wu et al.

carrying the QD particles to the perimeter or middle position. Fig. 3D illustrates the possibility of confining QD drops in the hole, thus resolving the coffee-ring effect. To limit the leakage of UV light and to prevent biological damage, the micro display was designed with a DBR structure. The DBR structure is prepared using alternative deposition of 17.5 multilayer pairs of HfO2/ SiO2 on a quartz glass. The DBR’s stopband width and stop center is 80 and 395 nm, respectively. The DBR’s maximum reflectance at 395 nm is 92.5%. The DBR structure was composed of multiple pairs of different dielectric refractive index layers to achieve higher reflectivity in the selected wavelength. The prepared DBR was then used to cover the LED array deposited with QDs, and the calculated electroluminescence (EL) spectrum of the structure is shown in Fig. 4A. Fig. 4A shows a 23%, 32%, and 5% increase in light production from QDs, for the red, green, and blue spectrums, 16 W/O DBR With DBR

Intensity (a. u.)

14

(a)

12

(b)

GB RGB RGB R

10 8 6 4 2 0

450

500 550 600 Wavelength (nm)

650

700

(c) DBR

Fig. 4 (A) The measured EL spectrum of RGB QDs deposited by AJ on a UV μLED array, where the black and red lines represent the devices with and without the DBR, respectively. (B) A row of the RGB pixels driven individually. (C) The full-color μLED covered with DBR layer in full operation. Figures reproduced with permission from Optical Society of America.

Quantum-dot-based full-color micro-LED displays

181

respectively. In Fig. 4B, two rows of RGB pixels were driven, allowing each pixel to be independently modified. The difference of brightness of each pixel is caused by the leakage current from the μLED array. Fig. 4C shows the full-color μLED in full operation and covered with the DBR. In another study on full-color display, Chen et al. demonstrated fullcolor monolithic hybrid quantum dot nanoring μLEDs with improved efficiency using atomic layer deposition and nonradiative resonant energy transfer (Huang Chen et al., 2019). Nanoring (NR) μLED can be produced by strain-induced engineering which can change the color from green to blue by regulating the wall width of the μLED through etching. In addition, several researchers have produced high-quality, full-color μLED displays by incorporating RGB QDs with UV-μLED and gained significant recognition. In Chen’s study, they used electron beam (e-beam) lithography and QD printing to manufacture hybrid quantum dot nanoring μLEDs. The device consisted of three subpixels i.e. a green LED, a blue NR-μLED, and a red QD-NR-μLED. In this study, NRET is proposed as a method of effective color conversion. Fig. 5 illustrates the production process of the hybrid QD-NR-μLEDs and the fabricated device’s cross-sectional view. Metal organic chemical vapor deposition (MOCVD) is used to grow epitaxial layers of InGaN/GaN LEDs having an emission wavelength of 525 nm on c-plane pattern sapphire substrates. The epitaxial layers composed of an undoped active layer InGaN/GaN MQW, a GaN buffer layer, an n-GaN layer, and a p-GaN layer with the MQWs consisting of eight pairs of In0.28Ga0.72N wells and GaN barrier. The bare epitaxial wafer as shown in Fig. 5A is cleaned before the beginning of the production process. Subsequently, e-beam lithography is used to create RGB region with negative photoresist with each green pixel formed in a rectangular mesa, whereas the other two regions are composed primarily of NR series. The photoresist is removed through lift-off process after the deposition of nickel using e-beam evaporation to form a hard mask pattern. Next, inductively coupled plasma reactive ion etching (ICP-RIE) is used to etch the GaN-based material to establish the active area, separate the pn layer, and isolate each subpixel. The passivation layer of Al2O3 is deposited using atomic layer deposition (ALD) technology and, as shown in Fig. 5B, CdSe/ZnS red QDs are dispersed on a blue NR-μLED region by means of super-inkjet (SIJ) printing technique. Spin-on-glass (SOG) etching process is later used to shield the QD layer and separate the pn electrodes by the deposition of a transparent conductive oxide (TCO). The Ni/Au metal deposition for the pn electrodes using the lift-off process is as shown in Fig. 5C. Eventually, the red color

182

Tingzhu Wu et al.

(a)

(b)

(e)

(c)

(d) DBR filter

(f)

TCO film p-GaN MQWs n-GaN Buffer

p electrode

MQWs

Red QD n electrode

Sapphire

Fig. 5 (A) Epitaxial wafer; (B) three subpixels of a green μLED, a blue NR-μLED, and a red QD-NR-μLED; (C) deposition of TCO film and pn electrodes; (D) covering DBR filter; (E) full-color display panel composed of the proposed hybrid QD-NR-μLEDs; (F) cross-sectional view of a single RGB pixel. Figures reproduced with permission from Optical Society of America.

region is covered with a distributed Bragg reflector (DBR) to filter out and reuse blue light, as illustrated in Fig. 5D. Fig. 5E shows the final hybrid QD-NR- μLED based full-color display panel, and the device crosssectional view is shown in Fig. 5F. The temperature-dependent time-resolved photoluminescence (TDTRPL) decay curve of NR-μLED with and without ALD passivation conducted at very low power (20 μW) for a very low excitation power density is shown in Fig. 6A. The total recombination time for each curve is 10.40, 7.80, 3.03 and 2.28 ns using the exponential decay fitting, referring to the calculation at 15 K and 300 K for the NR-μLED without passivation and with passivation, respectively. The carrier lifetime at 15 K is longer than that at 300 K since the radiative recombination is dominant at 15 K, but there is still the effect of non-radiative recombination as it declines. The non-radiative recombination is attributed to the surface defects, and the short carrier lifetime with passivation at 15 K is due to the reduction of surface defects leading to the superior recombination mechanism. Similar findings are also obtained at 300 K, but the recombination mechanism becomes dominated by the non-radiative recombination centers. These results indicate that the passivation layer deposited using ALD technology minimized surface defects by increasing the ratio between surface and volume. Fig. 6B represents the

183

Quantum-dot-based full-color micro-LED displays

Log Intensity (Norm.)

(a)

1

0.01

0.001

40 60 Decay time (ns)

(c) NR-µLED NR-µLED with red QDs Exponential Fitting

0.1

0.01

1E-4 0

a2 / (a1+a2) = 19.7 % tMQW = 3.03 ns thybrid = 1.02 ns hET = 66.4 %

5

10 15 20 Decay Time (ns)

25

30

Log Intensity (Norm.)

Log Intensity (Norm.)

20

0

1

0.001

τMQW at 300K = 3.03 ns τ'MQW at 300K = 2.28 ns

τMQW at 15K = 10.40 ns τ'MQW at 15K = 7.80 ns

1E-4

(b)

w/o passivation at 15K w/ 1 nm passivation at 15K w/o passivation at 300K w/ 1 nm passivation at 300K

0.1

80

100

1 NR-µLED NR-µLED with red QDs Exponential Fitting

0.1

0.01 a¢2 / (a¢1+a¢2) = 17.3 % t′ MQW = 2.28 ns t′hybrid = 1.06 ns h′ET = 53.6 %

0.001

1E-4

0

5

10 15 20 Decay Time (ns)

25

30

Fig. 6 TD-TRPL curves of NR-μLEDs (A) with and without ALD passivation at 15 K and 300 K, respectively, (B) without ALD passivation, and with and without red QDs, and (C) with 1 nm ALD passivation, and with and without red QDs. Figures reproduced with permission from Optical Society of America.

decay curve of NR-μLED without ALD passivation, and with and without red QDs. The decay lifetime is extracted and found to be 3.03 ns for NR-μLED and 1.02 ns for QD-NR-μLED. Furthermore, the NRET rate is 0.65 ns 1, which is higher than the MQW nonradiative decay rate of 0.33 ns 1, which suggests that the efficient NRET process greatly reduced the losses of nonradiative recombination. The QD-NR-μLED NRET efficiency is 66.4% in Fig. 6B and 53.6% in Fig. 6C for QD-NR-μLED with a 1 nm ALD passivation. This decrease in efficiency is caused by an increase in the gap between MQWs and QDs. However, this passivation layer increased the intensity of blue light by 143.7%. In this case, the NRET mechanism depends primarily on the coupling distance between donors and acceptors, namely the NR-μLED QDs and MQWs. Fig. 7A displays the electroluminescence (EL) spectra for the individual RGB colors in hybrid QD-NR-μLED with peaks at 630, 525, and 467 nm, respectively. It is clear that the device possesses narrow EL spectra, which is responsible for a wide color gamut as shown in Fig. 7B, which is approximately 104% and 78.2% according to the national television system committee (NTSC) and Rec. 2020, respectively. This is enough to sustain the full-color performance in display technology.

184

Tingzhu Wu et al.

Fig. 7 (A) EL spectra of RGB hybrid QD-NR-μLEDs. (B) Color gamut of RGB hybrid QD-NR-μLEDs, NTSC and Rec. 2020. Figures reproduced with permission from Optical Society of America.

Changing the thickness of the μLED is necessary to enable the color conversion mechanism from blue to green, or blue to red for full-color displays. Hu et al. demonstrated uniform quantum dots printed in inkjet as color conversion layers for full-color organic LED displays in 2020 (Hu et al., 2020). Fig. 8A illustrates the measurements from a polymer-based QD ink with uniform green and red QDs at micrometer thickness. This indicates that the polymer-based QD ink effectively absorbs blue light better than its solvent-based counterpart. The light conversion efficiency (LCE) from blue to green and red can be tuned by changing the thickness of the QD layer. The LCE of the green QD layer reached 90% at 10.2 μm thickness and 33% for red QDs at 10.5 μm thickness. The color gamut of the QD-OLED could reach 95% according to the BT 2020 standard, as shown in Fig. 8B.

2.1 QDs photoresist patterning technique CQDs are suitable for use as a color conversion layer for μLEDs, and a high contrast ratio can be achieved with QD-based μLED displays. Photolithography techniques can be used for producing high-resolution μLED displays and also to overcome the challenges faced in mass transfer process. QDs can be combined with photoresist (PR) to form QDPR. This approach provides a method to pattern QDPR arrays that can monitor size and thickness while maintaining the benefits of photolithography. This technique overcomes the bottlenecks of the previously reported methods and provides a

Fig. 8 (A) PL spectra and LCE of green and red QD layers at different thicknesses. (B) Color gamut of the QD-OLED display (blue triangle area), in comparison with the color gamut of BT. 2020 (black triangle area) (Hu et al., 2020; Huang Chen et al., 2019). Figure reproduced with permission from Royal Society of Chemistry.

186

Tingzhu Wu et al.

cost-effective, realistic solution in developing high-resolution, large-area devices, particularly full-color μLEDs for display applications. A full-color μLED array was fabricated using semipolar (20–21) blue LEDs with a 50 μm diameter chip size and a green or red QDPR color-conversion layer. This work proved that the semipolar μLED system can achieve full-color display. When compared to c-plane LEDs, semipolar LEDs show much better wavelength shift characteristics and improved efficiency droop. In addition, the use of the photoresist matrix helps in achieving high contrast and higher color stability in μLED. Huang Chen et al. demonstrated a full-color μLED display with high color stability using semipolar (20–21) InGaN LEDs and quantum-dot photoresist. MOCVD can be used to grow (20–21) oriented GaN layer on patterned sapphire substrate (PSS) (Chen et al., 2020). The conventional way of growing semipolar GaN by off-axis slicing of bulk-form GaN substrate grown by hybrid vapor-phase epitaxy is expensive and not desirable for mass production. To address this problem, an advanced orientation-controlled epitaxy (OCE) method, which is a simple epitaxy method involving MOCVD, was used to directly grow GaN material on the sapphire wafer, to produce semipolar GaN. The epitaxial structure of the semipolar device consisted of a bulk GaN buffer layer (5 μm), an n-GaN layer (1.5 μm), an undoped InGaN/GaN MQW active layer, a p-GaN layer (150 nm), and a p-InGaN layer (3 nm). The MQWs were designed using five pairs of 3 nm thick InGaN wells and a 5 nm thick GaN barrier with a wavelength of about 450 nm on emission. In this analysis, the c-plane wafer was also developed with a similar MQW structure. Fig. 9 represents the process flow of μLED array with the lithographic process of black photoresist (PR) matrices on the top of the array. The development of the μLED array started with the deposition of transparent conductive oxide (TCO), accompanied by annealing to form the ohmic p-type contact. Then, 1 μm depth mesa etch was formed by using HCl solution and ICP-RIE, followed by the formation of n-type electrode. Then, 200 nm thick SiO2 is deposited as a passivation layer by plasma-enhanced chemical vapor deposition technique and the μLED array process is completed via hole process and ICP-RIE. Later, the lithography process is used to deposit black PR matrices and QDPR on the semipolar μLED array and a black PR was used to flatten the array to prevent lateral leakage of blue light. As shown in Fig. 9B, the Ni/Au (p-electrode metal) lines were then deposited on the flattened surface to connect each chip. Next, the lithography process sequentially produced the gray photoresist, red QDPR, green

Quantum-dot-based full-color micro-LED displays

187

Fig. 9 Process flow for the fabrication of a full-color RGB pixel array. (A) μLED array process. (B) Black PR matrices and p-electrode metal lines. (C) Red, green, and blue (transparent) pixel lithography process. (D) Color pixel bonding. Figures reproduced with permission from Optical Society of America.

QDPR, and transparent PR to form a color pixel on a highly transparent glass substrate as shown in Fig. 9C and this color pixel on glass was stuck together with μLED in Fig. 9D. Fig. 10A displays the electroluminescence (EL) spectrum at different current densities showing a wavelength peak at 453 nm and full width at half maximum (FWHM) of 24.8 nm under 200 A/cm2 injection current density. This narrow FWHM suggests that the epitaxy offers a high-quality structure and is therefore responsible for the delivery of purer emitted light, matching colors, and color gamut distribution. Fig. 10B gives the experimental data as well as the simulation fit for the external quantum efficiency for both c-plane and semipolar devices. It can be seen from the figure that the simulation fit and the experimental data are in good agreement, and the c-plane device has a larger external quantum efficiency (EQE) than that of the semipolar device. Despite having a lower EQE, the output power of the semipolar device is sufficient for the application of color conversion. In addition, the semipolar device shows an efficiency drop of just 14.7% at a high injection current density of 200 A/cm2, while the c-plane device shows a 55% drop in efficiency under the same conditions. It is suggested that the output power and the

188

Tingzhu Wu et al.

Fig. 10 (A) Electroluminescence spectrum of semipolar μLED with increasing applied current density. (B) Experimental data and simulation curves for normalized external quantum efficiency of semipolar and c-plane μLEDs. (C) Fluorescence microscopy image of RGB pixel. (D) Electroluminescence spectra of red and green pixels (Chen et al., 2020). Figures reproduced with permission from Optical Society of America.

EQE can be further improved by optimizing the active area of the device. The severe efficiency droop in c-plane devices can be explained by two important phenomena i.e. carrier leakage and Auger recombination, which will cause quantum-confined Stark effect (QCSE) leading to low carrier recombination rate. The polarization charges that accumulate at the hetero interfaces of the active region can also induce tilting of the energy band and can separate the overlap of the wave-function distribution, resulting in nonradiative recombination and carrier leakage. Fig. 10C represents the fluorescence microscopy (FLOM) image of RGB pixel matrices on glass showing a high contrast ratio between the gray PR matrices and the color pixels. The gray PR can exhibit higher reflectivity as compared to the black PR, thereby reducing the crosstalk effect between the pixels and increasing the output intensity by inside reflection. Fig. 10D represents the EL spectra of green and red pixels showing peak

Quantum-dot-based full-color micro-LED displays

189

wavelengths of 536 nm and 630 nm, respectively. The corresponding FWHM for the RGB emission spectra are 30.5 nm, 24.5 nm, and 24.8 nm. Hence, these narrow FWHMs specify good performance in color rendering. It is clear that the blue light leakage can be reduced by using QDPR, and it can be further reduced by increasing the thickness of QDPR or changing its composition. Since blue light leakage is minimized, only blue light from QD pixel remains, and each pixel will independently show brilliant colors. Hence, red and green QDPR pixels can filter blue light and significantly enhance the color purity. The color performance of the RGB μLEDs under the injection current density from 1 to 200 A/cm2 is demonstrated in Fig. 11 using CIE 1931 and 1976. The color coordinates for the c-plane device vary from (0.1572, 0.1067) to (0.1483, 0.0379). For the semi-polar device, the variation is from (0.1433, 0.0388) to (0.1490, 0.0317) in CIE 1931 chromaticity diagram. It is found that the color shift (μ’υ´) for semipolar blue μLEDs is 0.0209 and is smaller than that of the c-plane device, which is 0.1374 in CIE 1976. The color gamut of the RGB made from semipolar μLED is almost unchanged when the current density of the injection increases, although there is a variation of about 10% under the same conditions for c-plane device. Also, the red and green pixels display no color shift due to the emission by optical pumping and stability of QDPR. The RGB pixel fabricated using semipolar μLED and QDPR shows a wide-color gamut by achieving 114.4% of National Television Standards Committee (NTSC) space and 85.4% Rec. 2020 in the CIE 1931. Hence, the RGB pixels produced using semipolar μLED and QDPR show excellent color stability and wide color gamut characteristics required for next generation display applications. The historical development of QD printing for use in full-color display at Kuo’s Lab is illustrated in Fig. 12. The overview of the developmental history of QD-based display technology with their description can be found in Table 1. The disadvantages associated with each method are listed. In the beginning, the spray coating machine was applied for large-scale chips. However, due to the non-uniform QD region obtained by this method, it was challenging to achieve repeatability even under the same printing conditions. Moreover, the pulsed spray coating machine cannot control the linewidth precisely. Subsequently, AJ printing was employed on the full-color LED, which combined PR mold and the RGB QD pattern. The thickness of the QD pattern layer is successfully controlled to be below 35 μm.

(a)

(b)

0.9

c-plane

c-plane

0.8

c-plane

0.7

200 A/cm2

c-plane

0.6

1 A/cm2

1 A/cm2

0.100

1 A/cm2

0.275

0.5

0.225

0.6 0.075

0.4 0.5

y

y 470

0.3

0.4

0.050

470

0.3

200 A/cm2

460

0.025

0.2 0.125

0.2 0.1

CIE 1931 0.125

0.150

1 A/cm2

0.1

450 440 0.000 0.100

v'

v' 0.175

0.175

0.0 0.0 0.200

CIE 1931 0.1

0.2

0.3

x

0.4

0.5

0.6

0.7

CIE 1976 0.075 0.150 0.100

x

2 460 200 A/cm

0.200

0.250

200 A/cm2 0.2

u'

0.3

0.4

0.5

0.6

u'

(c)

(d)

0.9

Semipolar

CIE 1976 0.0 0.0 0.1

0.275

Semipolar

Semipolar

0.8

Semipolar

0.6

1 A/cm2

0.100 0.7

200 A/cm2

0.5

0.225

0.6 0.075

0.4 0.5

y

470

v' 0.175

y

1 A/cm2

470

0.3 460

0.025

200 A/cm2

0.150

x

0.175

1 A/cm

0.3

0.2

200 A/cm2

1 A/cm2

0.1

0.1 CIE 1931

CIE 1931 0.125

2

0.125

0.2 450 440

0.000 0.100

v'

0.4

0.050

0.0 0.0 0.200

0.1

0.2

0.3

0.4

x

0.5

0.6

0.7

CIE 1976 0.075 0.100 0.150

460 0.200

u'

0.250

CIE 1976 0.0 0.0 0.1

200 A/cm2 0.2

0.3

0.4

0.5

0.6

u'

Fig. 11 Color gamut of RGB pixel assembly in the CIE 1931 from (A) c-plane and (B) semipolar μLED and QDPR, under various current densities. Color gamut of RGB pixel assembly in the CIE 1976 from (C) c-plane and (D) semipolar μLED and QDPR, under various current densities. Figures reproduced with permission from Optical Society of America.

Pulsed spray coating machine: 30 × 30 cm2 (large scale)

Aerosol jet printing system: 35 µm (precise spray pattern)

SIJ printing system

Valve of sprayed quantity

Solution filling + circulation system

Nozzle QD ink transducer

Pulsed air pressure

Super inkjet printing system: 10 µm (spray precise spray pattern)

Water Substrate

QD ink

Blue NR-µLED

Reb QD

Fig. 12 Developmental history of QD printing system at Kuo’s Lab (Chen et al., 2020; Han et al., 2015). Figure reproduced with permission from Optical Society of America.

Table 1 Development in QD-based display technology. Methods

Description

Pulsed-spray coating machine

The operation is precisely controlled by a computer and its quality of spray was quite stable each time. The disadvantage is that the spray diameters are large and hence could not be used to spray precise patterns

Aerosol jet printing system

The QD solution was aerosolized by ultrasonic vibration. The aerosol jet process began with a mist generator that atomized liquid materials into small droplets. The disadvantage is the large spray area, resulting in the printing of QDs outside the windows area

Super-inkjet printing system

The pressure generated by the oscillating electric field is used by this printing machine to print the QDs. With better control of QD-inks, it provides fine-linewidth pattern. In addition, printing sufficiently dense QDs caused the higher color conversion of the color red and green. However, the method is time consuming

Quantum dots photoresist methods

Black matrix

substrate

Spin Black matrix Spin Red QDPR & Exposure Development

UV-light mask

Spin Green QDPR & Exposure Development

Table reproduced with permission from MDPI.

Quantum dot photoresist method by photolithography is a fast and convenient method. This method can control the thickness of QDPR to prevent the leakage of blue light. However, the disadvantage is the high usage of QD

194

Tingzhu Wu et al.

In recent years, the super ink-jet printing (SIJ) technology appeared along with the widespread application of μLED. The SIJ can maintain the QD pattern layer below 10 μm, and it can further precisely jet the QD solution onto the surface. Moreover, to avoid leakage of the blue incident light, the QDs need to be repeatedly printed to the block, which makes the technique time-consuming. The QDPR method effectively uses QD-based photoresist to achieve QD patterning at fine pitch. The advantages include quick fabrication, thickness-control, and scalability, which could probably become the future trend. Most importantly, the method can solve issues such as the high leakage of the blue incident light and crosstalk effect. Due to its high contact ratio, the incorporation of the black matrix will prevent the crosstalk effect. In addition, the QDPR is a mixture of QDs, photoresist, and TiO2 (Chen et al., 2020), and TiO2 can serve as a mirror to reflect the incident blue light. This results in the outer blue light reducing significantly. Due to its precise alignment, the SIJ printing technique can manufacture a fine-pitch display. However, as it can only be developed at the small-scale, the technique is limited to academic research. In contrast, the benefits of the QDPR approach like scalability, speed, and ability to control thickness make it ideal for commercial use. As the array size of a μLED decreases, its light-emitting efficiency improves while experiencing massive non-radiative recombination from sidewall defects. However, the far-field radiation pattern will deviate from the ideal Lambertian distribution depending on the sidewall emission, resulting in the color shift of μLED displays. To solve this issue, Gou et al. studied the angular color shift of RGB μLED displays from mismatched angular distribution and suggested a model of simulation to support the experimental findings (Gou et al., 2019a). The μLED is an innovative display technology and has magnificent characteristics such as long lifespan, strong readability in sunlight, high dynamic range, low energy consumption, and wide color gamut. The color-conversion method and mass transfer process are two common solutions to achieving the full-color display. However, manufacturing challenges remain in terms of achieving high yield, large size displays such as tablets, monitors, TVs, video walls, etc. (Chen et al., 2018; Gou et al., 2019b). The most widely used commercial LED epitaxy wafer is based on structures of multi quantum wells (MQW), i.e., GaInP/AlGaInP MQWs for red LEDs and InGaN/GaN MQWs for blue and green LEDs for (Bower et al., 2017), resulting in inconsistent angular distribution between

Quantum-dot-based full-color micro-LED displays

195

Fig. 13 Simulated total, top and sidewall emission of (A) red, (B) green, and (C) blue μLEDs at different viewing angles. (D) Measured and simulated data of far-field radiation patterns of RGB μLEDs. (E) Simulated color triangle of the RGB μLED display system and the CIE coordinates of 10 reference colors from 0 to 80 degree viewing angle (Gou et al., 2019a). Figures reproduced with permission from Optical Society of America.

RGB μLEDs due to different epitaxy materials and structures between them. Consequently, there is a color shift of mixed colors in RGB μLEDs displays. A device layout has been suggested to solve this issue that will be addressed further, later. Fig. 13A–C reflects the simulated total, top, and sidewall emissions from RGB μLEDs and it can be seen that the green and blue chip sidewall emissions are far greater than that of red chips due to better absorption in red MQWs than in green and blue MQWs. Fig. 13D displays the far-field radiation patterns of RGB μLEDs where the dots represent the measured data while the solid lines represent the simulation effects. A strong agreement exists between the simulation and the experiment indicating a color gamut of 97% of DCI-P3 standard and 78% of Rec. 2020 standard. From the figure, it is obvious that the light emission from the red chip decreases due to Lambert’s cosine law while the light emission for green and blue μLEDs increases from the normal angle to 40 degree and decreases afterwards. This discrepancy in angular

196

Tingzhu Wu et al.

distributions is caused by the difference in material dynamics of RGB chips and will also cause an angular shift in mixed colors. Fig. 13E displays the RGB μLED display color triangle and the 10-reference color CIE coordinates at viewing angles from 0 to 80 degree with 10 degree intervals. Due to the low cavity effect within μLEDs, no color change is observed at fixed driving current for primary colors. Although the color shift gets worse as the viewing angle increases (as predicted for mixed colors and white dots), it has been observed that the color shift for all colors at 800 is 0.061 and the maximum value for the magenta channel is 0.169, which exceeds the amount that just has been observed (Δμ’v´ < 0.02). This problem will get worse as the size of the μLEDs decreases and the sidewall emissions from green and blue chips increases. To mitigate the color change, green and blue μLED sidewall emissions should be eliminated to obtain the corresponding RGB radiation patterns i.e. Lambertian distribution. Hence, Gou et al. suggested a system mode shown in Fig. 14A, where the system consists of a μLED array and top black matrix outside the emission region. In addition, the space between the μLEDs is filled with resin with a refractive index of about 1.5. All the sidewall emissions from green and blue chips can be completely absorbed by the black matrix resulting in emissions following the Lambertian distribution from the top of RGB μLEDs. Additionally, the author introduces a taper angle, α, shown in Fig. 14A to boost the light intensity from the top emission. The light intensity from the green and blue chip is observed to increase as α increases from 90 to 140 degree and then found to saturate. However, the taper angle has little influence on the red chip. The color shift is extreme when α is greater than 130 degree and the wider taper angle contributes to a narrower angular distribution between green and blue. However, there is no effect on red and thus, a severe color shift occurs. Fig. 14B shows the simulated color shifts of 10 reference colors from 0 to 80 degree viewing angle for RGB μLED display with top black matrix and 120 degree taper angle. It is found that the average color shift at 80 degree is 0.05, and the maximum value for the magenta channel is 0.014, which is below 0.02 and is acceptable for commercial applications. Also, from the simulated radiation patterns for red, green, and blue μLEDs with top black matrix, it is clear that the sidewall emission is completely eliminated due to the presence of the black matrix and the color shift is significantly reduced with matched angular distribution for all the RGB colors as shown in Fig. 14C.

(b)

(a) Top glass BM

BM

α

0.5

BM

α

α

0.4 CIE v′

Resin

BM

Substrate

1.00

30 °

0.50 0.00

0.3 0.2

(c) 90°

0.00

0.75 0.25

(b) 90°

90°

0.00

0.25

(a)

1.00

°

0.25

0.50

30 °

60

°

0.50

0.75

0°

° 60

0.75

30 ° 60

Intensity

1.00

0°

Intensity

0°

Intensity

(c)

0.6

Rec. 2020 DCI-P3 RGB micro-LED

0.1 0 0

0.1

0.2

0.3

0.4

0.5

0.6

CIE u′ Fig. 14 (A) RGB μLED display with top black matrix. (B) Simulated color shifts of 10 reference colors from 0 to 80 degree viewing angle for RGB μLED display with top black matrix and 120 degree taper angle. (C) Simulated radiation patterns for red, green, and blue μLEDs with top black matrix (Gou et al., 2019a). Figures reproduced with permission from Optical Society of America.

198

Tingzhu Wu et al.

3. Conclusion In this article, we thoroughly reviewed the recent advances in the deployment of μLEDs and QD display technologies including μLED development and processing, large-scale μLED conversion and full-color QD based μLED. The development of QD printing technology has been described, including the methods of pulsed spray coating, aerosol jet printing, and super inkjet printing. The use of colloidal QDs as color down-converters for LEDs to produce high-quality, effective illumination sources and displays is described. There are also outlooks on optimizing the optical density of color converters, patterning and deposition, and solving μLED display issues. Finally, we discuss the specialized applications of μLED displays with fabrication of nanoring structures and full-color monolithic μLED structures by strain relaxation and printing quantum dots. Moreover, the combination of quantum dots photoresist and semipolar μLED to achieve wide color gamut by utilizing the wavelength stability of μLED and the narrow bandwidth emission of quantum dots is presented. Advancements in the development of QD-based μLEDs are expected to make this display technology ubiquitous in the near future.

References Ahn, H., Hong, S., Kwon, O., 2018. An active matrix micro-pixelated LED display driver for high luminance uniformity using resistance mismatch compensation method. IEEE Trans. Circuits Syst. Express Briefs 65 (6), 724–728. Ahn, Y.N., Kim, K.D., Anoop, G., Kim, G.S., Yoo, J.S., 2019. Design of highly efficient phosphor-converted white light-emitting diodes with color rendering indices (R1 R15)  95 for artificial lighting. Sci. Rep. 9 (1), 16848. 2019/11/14. Algar, W.R., Krull, U.J., 2008. Quantum dots as donors in fluorescence resonance energy transfer for the bioanalysis of nucleic acids, proteins, and other biological molecules. Anal. Bioanal. Chem. 391 (5), 1609–1618. 2008/07/01. Bower, C.A., et al., 2017. Emissive displays with transfer-printed assemblies of 8 μm  15 μm inorganic light-emitting diodes. Photonics Res. 5 (2), A23–A29. 2017/04/01. Chen, K., et al., 2013. Improving the angular color uniformity of hybrid phosphor structures in white light-emitting diodes. IEEE Electron Device Lett. 34 (10), 1280–1282. Chen, K., et al., 2015. Wide-range correlated color temperature light generation from resonant cavity hybrid quantum dot light-emitting diodes. IEEE J. Sel. Top. Quantum Electron. 21 (4), 23–29. Chen, G., Wei, B., Lee, C., Lee, H., 2018. Monolithic red/green/blue micro-LEDs with HBR and DBR structures. IEEE Photon. Technol. Lett. 30 (3), 262–265. Chen, C., Chen, H., Liao, J., Yu, C., Wu, M., 2019. Fabrication and characterization of active–matrix 960  540 blue GaN-based micro-LED display. IEEE J. Quantum Electron. 55 (2), 1–6, 3300106. https://doi.org/10.1109/JQE.2019.2900540. Chen, S.-W.H., et al., 2020. Full-color micro-LED display with high color stability using semipolar (20 21) InGaN LEDs and quantum-dot photoresist. Photonics Res. 8 (5), 630–636. 2020/05/01.

Quantum-dot-based full-color micro-LED displays

199

Corbett, B., Loi, R., Zhou, W., Liu, D., Ma, Z., 2017. Transfer print techniques for heterogeneous integration of photonic components. Prog. Quantum Electron. 52, 1–17. 2017/03/01/. Dahan, M., L
evi, S., Luccardini, C., Rostaing, P., Riveau, B., Triller, A., 2003. Diffusion dynamics of glycine receptors revealed by single-quantum dot tracking. Science 302 (5644), 442. € ur, U., € Morkoc¸, H., 2019. Micro-LEDs, a manDing, K., Avrutin, V., Izyumskaya, N., Ozg€ ufacturability perspective. Appl. Sci. 9 (6), 1206. Gou, F., et al., 2019a. Angular color shift of micro-LED displays. Opt. Express 27 (12), A746–A757. 2019/06/10. Gou, F., Hsiang, E.-L., Tan, G., Lan, Y.-F., Tsai, C.-Y., Wu, S.-T., 2019b. Tripling the optical efficiency of color-converted micro-LED displays with funnel-tube array. Crystals 9 (1), 39. Han, H.-V., et al., 2014. A highly efficient hybrid GaAs solar cell based on colloidalquantum-dot-sensitization (in eng). Sci. Rep. 4, 5734. Han, H.-V., et al., 2015. Resonant-enhanced full-color emission of quantum-dot-based micro LED display technology. Opt. Express 23 (25), 32504–32515. 2015/12/14. Hsu, S.-C., et al., 2015. Highly stable and efficient hybrid quantum dot light emitting diodes. IEEE Photonics J. 7, 1–10. 10/01. Hsu, S., Huang, Y., Kao, Y., Kuo, H., Horng, R., Lin, C., 2018. The analysis of dualjunction tandem solar cells enhanced by surface dispensed quantum dots. IEEE Photonics J. 10 (5), 1–11. Hu, Z., et al., 2020. Inkjet printed uniform quantum dots as color conversion layers for full-color OLED displays. Nanoscale 12 (3), 2103–2110. https://doi.org/10.1039/ C9NR09086J. Huang Chen, S.-W., et al., 2019. Full-color monolithic hybrid quantum dot nanoring micro light-emitting diodes with improved efficiency using atomic layer deposition and nonradiative resonant energy transfer. Photonics Res. 7 (4), 416–422. 2019/04/01. Huang, H., et al., 2017. Electrically-pumped wavelength-tunable GaAs quantum dots interfaced with rubidium atoms. ACS Photonics 4 (4), 868–872. 2017/04/19. Huang, Y., Tan, G., Gou, F., Li, M.-C., Lee, S.-L., Wu, S.-T., 2019. Prospects and challenges of mini-LED and micro-LED displays. J. Soc. Inf. Disp. 27 (7), 387–401. 2019/07/01. Huang, Y., Hsiang, E.-L., Deng, M.-Y., Wu, S.-T., 2020. Mini-LED, micro-LED and OLED displays: present status and future perspectives. Light Sci. Appl. 9 (1), 105. 2020/06/18. Jiang, H.X., Lin, J.Y., 2013. Nitride micro-LEDs and beyond—a decade progress review. Opt. Express 21 (S3), A475–A484. 2013/05/06. Kim, J.S., et al., 2004. White-light generation through ultraviolet-emitting diode and white-emitting phosphor. Appl. Phys. Lett. 85 (17), 3696–3698. 2004/10/25. Koh, W.-k., Saudari, S.R., Fafarman, A.T., Kagan, C.R., Murray, C.B., 2011. Thiocyanatecapped PbS nanocubes: ambipolar transport enables quantum dot based circuits on a flexible substrate. Nano Lett. 11 (11), 4764–4767. 2011/11/09. Konstantatos, G., Sargent, E.H., 2009. Solution-processed quantum dot photodetectors. Proc. IEEE 97 (10), 1666–1683. Lai, C.-F., Tien, Y.-C., Tong, H.-C., Zhong, C.-Z., Lee, Y.-C., 2018. High-performance quantum dot light-emitting diodes using chip-scale package structures with high reliability and wide color gamut for backlight displays. RSC Adv. 8 (63), 35966–35972. https:// doi.org/10.1039/C8RA07928E. Lee, C.K.W.E., Yurek, J., Ma, R. (Eds.), 2018. A New Frontier for Quantum Dots in Displays. Li, J., Tang, Y., Li, Z., Liang, G., Ding, X., Yu, B., 2019. High thermal performance and reliability of quantum-dot-based light-emitting diodes with watt-level injection power. IEEE Trans. Device Mater. Reliab. 19 (1), 120–125.

200

Tingzhu Wu et al.

Lin, J.Y., Jiang, H.X., 2020. Development of microLED. Appl. Phys. Lett. 116 (10), 100502. 2020/03/09. Lin, H., et al., 2016. Excellent color quality of white-light-emitting diodes by embedding quantum dots in polymers material. IEEE J. Sel. Top. Quantum Electron. 22 (1), 35–41. Lin, H.-Y., et al., 2017. Optical cross-talk reduction in a quantum-dot-based full-color micro-light-emitting-diode display by a lithographic-fabricated photoresist mold. Photonics Res. 5 (5), 411–416. 2017/10/01. Liu, C., et al., 2017. Enhanced color-conversion efficiency of hybrid nanostructured-cavities InGaN/GaN light-emitting diodes consisting of nontoxic InP quantum dots. IEEE J. Sel. Top. Quantum Electron. 23 (5), 1–7. Liu, Z., et al., 2020. Micro-light-emitting diodes with quantum dots in display technology. Light Sci. Appl. 9 (1), 83. 2020/05/11. Marchuk, K., Guo, Y., Sun, W., Vela, J., Fang, N., 2012. High-precision tracking with non-blinking quantum dots resolves nanoscale vertical displacement. J. Am. Chem. Soc. 134 (14), 6108–6111. 2012/04/11. Narendran, N., Gu, Y., Freyssinier, J.P., Yu, H., Deng, L., 2004. Solid-state lighting: failure analysis of white LEDs. J. Cryst. Growth 268 (3), 449–456. 2004/08/01/. Oh, J.R., Park, H.K., Oh, J.H., Kouh, T., Rag Do, Y., 2011. Highly efficient full-color display based on blue LED backlight and electrochromic light-valve coupled with front-emitting phosphors. Opt. Express 19 (17), 16022–16031. 2011/08/15. Pal, B.N., Robel, I., Mohite, A., Laocharoensuk, R., Werder, D.J., Klimov, V.I., 2012. High-sensitivity p–n junction photodiodes based on PbS nanocrystal quantum dots. Adv. Funct. Mater. 22 (8), 1741–1748. 2012/04/24. Pattantyus-Abraham, A.G., et al., 2010. Depleted-heterojunction colloidal quantum dot solar cells. ACS Nano 4 (6), 3374–3380. 2010/06/22. Roh, J., Park, Y.-S., Lim, J., Klimov, V.I., 2020. Optically pumped colloidal-quantum-dot lasing in LED-like devices with an integrated optical cavity. Nat. Commun. 11 (1), 271. 2020/01/14. Sadeghi, S., Khabbaz Abkenar, S., Ow-Yang, C.W., Nizamoglu, S., 2019. Efficient white LEDs using liquid-state magic-sized CdSe quantum dots. Sci. Rep. 9 (1), 10061. 2019/07/11. Schaller, R.D., Klimov, V.I., 2004. High efficiency carrier multiplication in PbSe nanocrystals: implications for solar energy conversion. Phys. Rev. Lett. 92 (18), 186601. 05/05/. Schiavon, D., 2015. Analysis of the Green Gap Problem in III-Nitride LEDs. Dissertation, Open Access Repositorium der Universit€at Ulm. Selvin, P., 2014. New Small Quantum Dots for Neuroscience (Presentation Video) (SPIE BiOS). SPIE. Shirasaki, Y., Supran, G.J., Bawendi, M.G., Bulovi
c, V., 2013. Emergence of colloidal quantum-dot light-emitting technologies. Nat. Photonics 7 (1), 13–23. 2013/01/01. Smeeton, T.M., et al., 2019. 54–1: Invited paper: development of electroluminescent QD-LED displays. In: SID Symposium Digest of Technical Papers. vol. 50, pp. 742–745. no. 1. 2019/06/01. Tsai, Y.-L., et al., 2013. Improving efficiency of InGaN/GaN multiple quantum well solar cells using CdS quantum dots and distributed Bragg reflectors. Sol. Energy Mater. Sol. Cells 117, 531–536. 2013/10/01/. Unithrattil, S., Lee, K.H., Chung, W.J., Im, W.B., 2014. Full-color-emitting CaYAl3O7: Pr3 +,Ce3+ phosphor for near-UV LED-based white light. JOL 152, 176–181. 2014/08/01/. Wu, T., et al., 2018. Mini-LED and micro-LED: promising candidates for the next generation display technology. Appl. Sci. 8 (9), 1557.

Quantum-dot-based full-color micro-LED displays

201

Yang, W., et al., 2014. High density GaN/AlN quantum dots for deep UV LED with high quantum efficiency and temperature stability. Sci. Rep. 4 (1), 5166. 2014/06/05. Yang, Z., et al., 2019. Recent advances in quantum dot-based light-emitting devices: challenges and possible solutions. Mater. Today 24, 69–93. 2019/04/01/. Zhang, F., et al., 2012. Nonradiative energy transfer between colloidal quantum dot-phosphors and nanopillar nitride LEDs. Opt. Express 20 (S2), A333–A339. 2012/03/12. Zhang, H., Su, Q., Chen, S., 2019. Recent progress in the device architecture of white quantum-dot light-emitting diodes. J. Inf. Disp. 20 (4), 169–180. 2019/10/02. Zhou, X., et al., 2020. Growth, transfer printing and colour conversion techniques towards full-colour micro-LED display. Prog. Quantum Electron. 71, 100263. 04/01.

CHAPTER SIX

Damage-free neutral beam etching for GaN micro-LEDs processing Xuelun Wanga,* and Seiji Samukawab a

GaN Advanced Device Open Innovation Laboratory, National Institute of Advanced Industrial Science and Technology, Nagoya, Japan b Institute of Fluid Science, Tohoku University, Sendai, Japan *Corresponding author: e-mail address: [email protected]

Contents 1. Introduction 2. Neutral beam generation source 3. Application of NBE for fabrication of sub-10-nm nanostructures 4. Sub-10-μm GaN micro-LEDs fabricated by NBE 5. Conclusion References

203 205 209 213 219 220

1. Introduction Generally, GaN micro-LEDs have been fabricated by etching a planar InGaN/GaN LED wafer by means of inductively coupled plasma (ICP) ( Jiang et al., 2001; Olivier et al., 2017; Tian et al., 2012). However, in this case, high-density crystalline defects acting as nonradiative recombination centers are inevitably generated on the sidewall surface of the micro-LED mesa owing to ion bombardment and high-energy high-density ultraviolet (UV) photon irradiation from the plasma. With a minor carrier diffusion length on the order of 0.5–1 μm for GaN LEDs grown on c-plane sapphire substrates (Hafiz et al., 2015; Karpov and Makarove, 2002), the injected carriers should be kept away from the sidewall surface at a distance of a few microns to avoid the influence of the ICP-induced defects on the emission efficiency of the micro-LED. This implies that the efficiency of a GaN micro-LED is inevitably reduced by ICP-induced nonradiative defects when the chip size is reduced to below approximately 10 μm. The influence of nonradiative sidewall defects is especially significant at low current Semiconductors and Semimetals, Volume 106 ISSN 0080-8784 https://doi.org/10.1016/bs.semsem.2020.12.001

Copyright

#

2021 Elsevier Inc. All rights reserved.

203

204

Xuelun Wang and Seiji Samukawa

densities because Shockley-Read-Hall nonradiative recombination tends to dominate the recombination process at low current densities (Schubert, 2007), while a high emission efficiency should be maintained over a wide current density range for display applications where the light intensity needs to be varied over a wide dynamic range. The systematic study by Olivier et al. (2017) on the chip size dependence of the external quantum efficiency (EQE) of a GaN LED demonstrated that the EQE of a 10-μm micro-LED at a current density of 1 A/cm2 was 10 times lower than that of a large-area (>100 μm) device. The miniaturization of the micro-LED sizes to below 10 μm is crucial for micro-LED displays for a cost reduction and realization of a high display resolution. To reduce the production cost of micro-LED displays to a level comparable to those of conventional liquid crystal displays (LCDs) and organic LEDs (OLEDs), it has been estimated that the microLED chip size needs to be reduced to approximately 9 and 3 μm for largesize television and smartphone applications, respectively (Paranjpe et al., 2018). Moreover, virtual reality/augmented reality head-mounted display is expected to be one of the major applications of micro-LED displays. In this case, a display resolution higher than 2000 pixels per inch (ppi) is required owing to the very small distance between the display and human eye (Vieri et al., 2018), which implies that the micro-LED chip size should be reduced at least to 5 μm. Therefore, development of high-efficiency GaN micro-LEDs with sizes in the sub-10-μm region is a prerequisite for the realization of high-performance micro-LED displays. KOH wet chemical etching is an effective method for the removal of plasma-induced damage layers. It has been successfully used to improve the emission efficiencies of GaN micro-LEDs (Smith et al., 2020; Wong et al., 2019). However, a crucial issue of the KOH wet etching technique is that an inclined semipolar-like sidewall surface is required because the KOH solution cannot etch the vertical nonpolar planes (m- or a-planes) ( Jung et al., 2013). The inclusion of an inclined sidewall surface will increase the actual size of the micro-LED by a few microns (assuming an epilayer thickness of 3–5 μm), which hinders the realization of a real 3-μm micro-LED. The vertical sidewall is also an important factor determining the efficiency of the LED wafer usage in the chip singulation process. Assuming a singulation spacing of 3 μm for 3-μm micro-LEDs, 75% of the LED wafer will be wasted after the singulation process. Recently, we used a novel damagefree dry etching technique, neutral beam etching (NBE), for the fabrication of GaN micro-LEDs. In this chapter, after a review of the basic concept and fundamental characteristics of the NBE technique, we present our recent results on sub-10-μm GaN micro-LEDs fabricated by NBE.

Damage-free neutral beam etching

205

2. Neutral beam generation source The NBE technique initially developed by Samukawa et al. (2001, 2002) suppresses the incidence of charged particles and UV photon radiation onto the substrate. This enables exposure of the substrate to only the energycontrolled neutral beam (the neutral beam’s kinetic energy can be precisely controlled by the ion-acceleration energy obtained by the applied electric field before neutralization). This enables ultralow-damage nanoscale etching of semiconductor materials, which can suppress the formation of defects at the atomic layer level and control surface chemical reactions with a high precision. A conceptual illustration of the neutral beam source that was evolved from a pulse-modulated plasma with an on/off switching time of 50 μs is shown in Fig. 1 (Samukawa, 2006). This source uses an ICP source and has carbon electrodes for ion acceleration at the top and bottom of the quartz plasma chamber. Gas is introduced from the upper electrode in the form of a shower. Ions accelerated from the plasma pass through apertures (diameter: 1 mm, length: 10 mm) formed in the lower graphite carbon electrode, where the ions are neutralized by colliding with the aperture sidewalls. In a plasma modulated by 50-μs pulses, electrons lose energy during the “off” periods and undergo dissociative attachment with a halogen gas having a

Fig. 1 Schematic of a neutral beam generation source developed using new concepts. Practical neutralization rate and energy could be obtained using negative ions generated efficiently by a pulse-modulated plasma.

206

Xuelun Wang and Seiji Samukawa

large electron affinity (Cl, Br, or F). Consequently, even in a high-density low-pressure plasma, an afterglow plasma consisting of both positive and negative ions is formed during “off ” periods. In this case, when the 600-kHz radio frequency (RF) electric field with direct current bias applied to the plasma is varied and positive and negative ion beams in the Cl plasma pass through the carbon apertures, almost 100% of the negative ions are efficiently neutralized (Samukawa et al., 2001, 2002), whereas approximately 70–80% of the positive ions are neutralized (Mizutani and Nishimatsu, 1988; Shimokawa, 1992). Kubota et al. (2012) used the time-dependent Kohn-Sham equations for a detailed analysis of the neutralization mechanism, which demonstrated that the negative ions transition into electrons with a high probability. This is attributed to the resonant transitions between orbitals with energies close to those of graphite, which neutralized negative ions with a high probability. On the other hand, positive ions are neutralized with a lower probability by low-probability electron transitions owing to multi-stage Auger transitions between orbitals with disparate energy levels. Therefore, a neutral beam formed by neutralizing mainly negative ions with a pulse-modulated plasma has a higher density and lower energy than those obtained with positive ions. The influence of the UV photon irradiation from the plasma to the surface was investigated using a UV lamp during a high-density Cl beam bombardment (Samukawa et al., 2007) to directly understand the effects of the UV photon irradiation, which enhances surface reactions and defect generation during the Cl2 plasma etching. The UV lamp was set at 90° from the neutral beam source, as shown in Fig. 2. A short-arc Xe flash lamp (pulse discharge) was used for irradiation of UV photons on the Si sample surface. Fig. 2 also shows the spectrum (from 200 nm to the visible region) irradiated from the lamp to the substrate surface. A higher photon intensity was observed in the UV region of 220–400 nm than that in the visible region. The power density of the irradiated photons was monitored with a calorimeter and fixed at 38 mW/cm2. The photon irradiation frequency was fixed at 8 Hz (the “on”-time pulse width was 25 ms (full width at half maximum)). To clarify the effects of UV and visible photons on the surface reactions, UV photons were cut below 380 nm with a UV photon filter. As shown in Fig. 3, the effect of the photon irradiation on the etching depth was investigated as a function of the RF bias power and irradiated photon wavelength with and without application of the UV photon filter in the region of 220–380 nm. By varying the RF bias power in the range of 0–80 W, the Cl beam energy could be controlled in the range of 10 to 100 eV. In this

Damage-free neutral beam etching

207

Fig. 2 (A) Illustration of an experimental setup combining the developed neutral beam source and UV lamp. (B) Photon spectrum (from 200 to 800 nm) irradiated from a short-arc Xe flash lamp to the Si substrate surface. The photon irradiation power density and pulse frequency are in the ranges of from 0 to 50 mW/cm2 (3
1016 photons/cm2/s) and 0–8 Hz, respectively (the “on” time was fixed at 25 ms).

experiment, the Si etching depth was measured by atomic force microscopy. The irradiation of UV photons (from 220 to 380 nm) largely increased the etching depth under any RF bias condition, whereas the irradiation of visible photons did not increase the etching depth under the same Cl beam conditions. This demonstrated for the first time that UV light in the range of 220–380 nm largely enhances the Cl reactions with Si. The increase in the etching rate corresponds to an increase in the defect density on the Si surface. These results suggest that the UV photon irradiation of the surface

208

Xuelun Wang and Seiji Samukawa

Fig. 3 Effect of the photon irradiation (power: 38 mW/cm2, frequency: 8 Hz) on the etching depth as a function of the RF bias power during a Cl neutral beam etching process. The irradiated photon wavelength was also changed by applying a UV photon filter. In this experiment, UV photons below 380 nm could be eliminated with the filter.

largely enhances surface reactions during Si etching by the combination with a Cl neutral beam. This is attributed to the generation of crystal defects by UV irradiation on the Si surface. This implies that the UV irradiation has an important role in the generation of surface defects and reactions, even in Cl plasma etching. The UV photon irradiation was also a major source of defect generation in a Cl2-based ICP etching of GaN (Minami et al., 2011). To more directly confirm the generation of crystal defects by UV photon irradiation, the depth profile of the absolute value of the defect density in a 150-nm-thick SiO2 film after irradiation with He, Ar, and O2 plasmas was investigated using electron spin resonance (ESR) measurements, as shown in Fig. 4. High-intensity UV photons with wavelengths of 50, 100, and 120 nm were emitted from the plasma to the surface in the He, Ar, and O2 plasmas, respectively. In all cases, high-density defects were generated in the SiO2 film within depths of a few tens of nanometers from the surface. UV photons from the plasma to the surface generated high-density defects and enhanced the surface reactions within depths of a few tens of nanometers from the surface. These high-density defects remained within depths of a few tens of nanometers on the etched surface, which degraded the performance of optical and electronic devices involving nano/microstructures.

Damage-free neutral beam etching

209

Fig. 4 The depth profile of the absolute value of the defect density in a 150-nm-thick SiO2 film after irradiation of He, Ar, and O2 plasmas was investigated by an ESR measurement. High-intensity UV photons with wavelengths of 50, 100, and 120 nm were emitted from the plasma to the surface in the He, Ar, and O2 plasmas, respectively.

For future nanoscale devices, our NBE technique can be a mainstream approach for atomic-layer, defect-free, and top-down processes in place of plasma processing. In summary, damage-free etching of semiconductor materials could be realized by the proposed NBE technique because both ion bombardment and UV photon irradiation are eliminated. In the conventional ICP process, although the ion bombardment effect can be reduced by reducing the RF bias power (Tripathy et al., 2001), the UV photon irradiation cannot be avoided, which implies that a damage layer always exists on the etched surface with a depth of a few tens of nanometers.

3. Application of NBE for fabrication of sub-10-nm nanostructures In this section, we present some results on the fabrication of sub-10-nm semiconductor nanostructures using the NBE technique to demonstrate its excellent performance as a damage-free etching process for semiconductor materials. Instead of lithography, we used a bio-template, as proposed by Yamashita et al. (Kubota et al., 2005; Yamashita, 2001), as an etching mask with dots with sizes of a few nanometers. As shown in Fig. 5A, the biological supermolecule (protein) ferritin has a diameter of 12 nm and internal cavity of 7 nm. A negative charge exists inside the cavity. When ferritin is added into a solution containing dissolved Fe ions, the Fe positive ions are introduced into the cavity of ferritin molecules to form iron oxide cores with diameters of 7 nm. Ferritin molecules containing these iron cores are selectively placed

210

Xuelun Wang and Seiji Samukawa

Fig. 5 Schematic of the fabrication of sub-10-nm nanostructures by the NBE process using the biological supermolecule ferritin (protein). (A) Self-organized two-dimensional array of the biological supermolecule ferritin. (B) Iron oxide cores left on the sample surface after the protein removal by UV ozone etching. (C) Defect-free ultrafine nanostructures with a size smaller than 10 nm fabricated by NBE with the iron oxide core as the etching mask.

Fig. 6 SEM images of defect-free nanodisk structures (diameter: 7 nm) formed with uniform density and regular arrangement using a self-organized array of ferritin-based cores as an etching mask for Si, Ge, GaAs, and graphene.

in a two-dimensional arrangement on a silicon oxide film. The protein is then removed by UV/ozone treatment or heat processing, while retaining the 7-nm iron cores on the substrate for use as an etching mask (Fig. 5B) (Kubota et al., 2005). Finally, the Cl2-based neutral beam can etch any type of surface material using the etching mask of 7-nm iron cores (Fig. 5C). Fig. 6 shows scanning electron microscopy (SEM) images of nanodisk

Damage-free neutral beam etching

211

Fig. 7 Variation in bandgap energy with the disk thickness in the nanodisk structures of Si, Ge, and AlGaAs.

structures of silicon, germanium, gallium arsenide, and graphene with diameters of approximately 10 nm produced by this process (Samukawa, 2015). Sub-10-nm quantum nanodisks are formed in an array configuration with uniform spacing. Fig. 7 shows the precise control of the bandgap energy in these nanodisk structures with different materials upon variation of the thickness of these disks while maintaining the diameter at 10 nm (Samukawa, 2015). The bandgap can be controlled over a wide range with a high precision by varying the nanodisk size and material. Notably, other quantum dot fabrication techniques cannot provide such flexible and precise bandgap control. A similar process has been applied for fabrication of InGaN/GaN quantum nanodisk (Higo et al., 2017). A 3-period In0.3Ga0.7N (2nm)/GaN (8 nm) multiple quantum wells (MQWs) sample and an In0.3Ga0.7N (2 nm)/GaN single quantum well (SQW) sample were grown on c-plane (0001) sapphire substrates by metal organic vapor phase epitaxy (MOVPE) for structural characterization and optical measurements, respectively. Fig. 8A and B show top-view SEM and side-view transmission electron microscopy (TEM) images of the nanodisk fabricated from the MQWs sample, respectively. The nanodisk diameter measured using the TEM image was as small as approximately 5 nm. This is the minimum feature size of an InGaN/GaN nanodisk fabricated by a top-down dry process. The temperature dependence of the photoluminescence (PL) intensity was investigated to determine the internal quantum efficiency (IQE) of the InGaN/GaN nanodisk, as shown in Fig. 9. The IQEs of the as-grown SQW and the nanodisk were calculated to be 0.3% and 11.6% at 200 K, respectively, assuming that the nonradiative

212

Xuelun Wang and Seiji Samukawa

Fig. 8 (A) Top-view SEM and (B) cross-sectional TEM images of In0.3Ga0.7N/GaN nanodisk structures fabricated by NBE. Reprinted from Higo, A., Kiba, T., Chen, S., Chen, Y., Tanikawa, T., Thomas, C., Lee, C. Y., Lai, Y.-C., Ozaki, T., Takayama, J., Yamashita, I., Murayama, A., and Samukawa, S., 2017. Optical study of sub-10 nm In0.3Ga0.7N quantum nanodisks in GaN nanopillars. ACS Photonics 4, 1851–1857 with the permission of ACS Publications.

relaxation centers are completely frozen (IQE of 100% at 6 K). The significant increase in the IQE of the nanodisk is attributed to the improvement in the spatial overlap between the electron and hole wave functions owing to the relaxation of the quantum confined Stark effect (QCSE). The efficient coupling of electron and hole wave functions increases the oscillator strength of the e1  hh1 transition in the nanodisk. The localized positions of the

Damage-free neutral beam etching

213

Fig. 9 IQEs at various temperatures for an as-grown In0.3Ga0.7N/GaN SQW and quantum nanodisk fabricated by NBE.

electrons and holes in the as-grown SQW are spatially different owing to the band slope caused by the strong QCSE effect induced by intrinsic piezoelectric fields. Therefore, a high IQE of the nanodisk was achieved, 39 times that of the as-grown SQW. These results clearly suggest that damage-free etching of InGaN/GaN nanodisks was achieved by the NBE process.

4. Sub-10-μm GaN micro-LEDs fabricated by NBE A series of square-shaped InGaN/GaN micro-LEDs with mesa sizes in the range of 40
40–6
6 μm2 was fabricated by the NBE process using a blue-emitting (440 nm) GaN LED wafer grown on a c-plane sapphire substrate by MOVPE (Zhu et al., 2019). A similar series of samples was fabricated using the conventional ICP process as a reference. The LED layer structure contained a 5-period InGaN (2 nm)/GaN (12 nm) MQW active layer and 150-nm-thick Mg-doped p-GaN layer. The AlGaN electronblocking layer was not included for simplicity. For the NBE etching, Cl2 was used as the etching gas. ICP and RF powers of 400 and 5 W were used, respectively. The etching pressure was set at 0.1 Pa. The ICP etching

214

Xuelun Wang and Seiji Samukawa

was performed using a commercial ICP machine (Samco, RIE-400iPS) under a pressure of 1 Pa. A mixture of BCl3 (20 sccm) and Cl2 (50 sccm) was used as the etching gas. ICP and RF powers of 150 and 30 W were used, respectively. A SiO2 layer deposited by plasma-enhanced chemical vapor deposition (PECVD) and patterned by photolithography and BHF wet etching was used as the mask for micro-LED mesa etching. Fig. 10A shows a schematic drawing of the fabricated device. After formation of the micro-LED mesa, a 200-nm-thick SiO2 layer was deposited on the sample surface by PECVD at 350°C as a passivation and electrical isolation layer. Square Ni (2 nm)/Au (5 nm) semitransparent p-type

Fig. 10 (A) Schematic of the InGaN/GaN micro-LED fabricated in this study. (B) SEM image of a 6-μm micro-LED. Panel A Reprinted from Zhu, J., Takahashi, T., Ohori, D., Endo, K., Samukawa, S., Shimizu, M., and Wang, X. L., 2019. Near-complete elimination of size-dependent efficiency decrease in GaN micro-light-emitting diodes. Phys. Status Solidi A 216, 1900380-1-6 with the permission of John Wiley and Sons.

215

Damage-free neutral beam etching

electrodes were formed on the mesa top surface in a window opened into the SiO2 layer by vacuum evaporation and liftoff. The size of the p-electrode was designed to be 1 μm smaller than the size of the micro-LED mesa, which provided a spacing of 0.5 μm between the Ni/Au electrode and perimeter of the micro-LED mesa. The sample was then annealed in N2 at 500°C for 2 min to form an Ohmic contact. Subsequently, Cr/Au p- and n-type bonding pads were formed on the surface of the SiO2 electrical isolation layer and n-GaN layer, respectively. Finally, a Au stripe was fabricated to connect the Ni/Au p-electrode and Cr/Au p-type bonding pad. The wafer was lapped to approximately 150 μm and cut into small chips by mechanical blade dicing. Fig. 10B shows an SEM image of a 6-μm micro-LED fabricated by the above process. The fabricated LED chips were bonded to TO-18 packages with Ag paste without resin encapsulation. Light was extracted from the p-GaN side through the Ni/Au semitransparent electrode. The light output power was measured using a calibrated Si photodiode placed at a distance of 2 mm from the LED chip, which can collect approximately 80% of the total emission assuming a Lambertian emission pattern. The device efficiency is analyzed in terms of the EQE defined by EQE ¼

P out e , hνI

where Pout is the light output power measured by the Si photodiode, h is the Planck constant, ν is the frequency, I is the injection current, and e is the electron charge. Fig. 11A and B show low-magnification cross-sectional TEM images of a typical 6-μm micro-LED mesa etched by the NBE and ICP processes, respectively. The sidewall of the NBE mesa showed an intersection angle of approximately 98° with respect to the mesa top (0001) surface, approximately 16° smaller than that of the ICP-etched sample (114 °C). Further, the sidewall of the NBE mesa can be controlled to be almost vertical with respect to the (0001) plane by optimizing the edge geometry of the SiO2 mask (Zhang et al., 2020). As stated in Introduction, this is a very useful characteristic to improve the usage efficiency of epitaxial wafers in the chip singulation process because the spacing required for chip singulation is essentially limited by the resolution of photolithography in this case. The mesa heights of the NBE and ICP samples were approximately 370 and 650 nm, respectively. Fig. 11C and D show high-resolution lattice images of the InGaN/GaN MQWs near the sidewall surfaces of the NBE and ICP-etched samples, respectively. Clear lattice structures were observed on the outmost sidewall surfaces of both NBE and ICP-etched samples,

216

Xuelun Wang and Seiji Samukawa

Fig. 11 Cross-sectional TEM images of 6-μm micro-LEDs fabricated by (A) NBE and (B) ICP process. (C and D) High-resolution lattice images of the InGaN/GaN MQWs active regions near the sidewall surfaces of the NBE and ICP-etched samples, respectively. Reprinted from Zhu, J., Takahashi, T., Ohori, D., Endo, K., Samukawa, S., Shimizu, M., and Wang, X. L., 2019. Near-complete elimination of size-dependent efficiency decrease in GaN micro-light-emitting diodes. Phys. Status Solidi A 216, 1900380-1-6 with the permission of John Wiley and Sons.

which indicates that the etching qualities of both processes were reasonably high. However, points defects, such as N vacancies, which are known as main defects of ICP-etched GaN layers, cannot be easily distinguished by TEM. The dark line near the sidewall surface in both NBE and the ICP-etched samples was an artifact of the TEM observation, probably related to thickness fluctuations of the TEM sample near the surface. Fig. 12A and B show current-density-dependent EQE characteristics of micro-LEDs with different sizes fabricated by the ICP and NBE processes, respectively. All devices were aged at a current density of 50 A/cm2 for a few hours to stabilize their light output powers. As expected, the EQE of the ICP-etched sample exhibited a significant decrease with the reduction in the chip size from 40 to 6 μm. At a current density of 5 A/cm2, the EQE decreased from 3.44% for the 40-μm device to approximately 0.66% for the 6-μm device. In particular, the EQE of the 6-μm device monotonously increased with the current density without reaching a maximum up to the

Fig. 12 EQE as a function of the current density of the micro-LEDs with different sizes fabricated by the (A) ICP and (B) NBE processes. Reprinted from Zhu, J., Takahashi, T., Ohori, D., Endo, K., Samukawa, S., Shimizu, M., and Wang, X. L., 2019. Near-complete elimination of size-dependent efficiency decrease in GaN micro-light-emitting diodes. Phys. Status Solidi A 216, 1900380-1-6. with the permission of John Wiley and Sons.

218

Xuelun Wang and Seiji Samukawa

maximum current density of 80 A/cm2 used in this study, which implies that nonradiative recombination is dominant even at a current density as high as 80 A/cm2. These characteristics are similar to those typically observed in InGaN/GaN micro-LEDs fabricated using the ICP process (Olivier et al., 2017; Tian et al., 2012). In contrast, all NBE devices exhibited EQE-current density characteristics similar to those of large-area InGaN/GaN LEDs. The EQEs of all four devices initially rapidly increased with the increase in current density, reaching a maximum at a current density of approximately 5 A/cm2, and then decreased with the further increase in the current density owing to the well-known efficiency droop effect in GaN LEDs grown on c-plane sapphire substrates. Furthermore, the maximum EQE observed around the current density of 5 A/cm2 for all four devices varied only by less than 10%. The four devices exhibited very similar EQEs at an even lower current density of 1 A/cm2 (approximately 2.58%, 2.63%, 2.74%, and 2.74% for the 40-, 20-, 10-, and 6-μm micro-LEDs, respectively). These results indicate that the size-dependent efficiency decrease typically observed in ICP-etched GaN micro-LEDs, particularly in the sub-10-μm region, has been successfully eliminated, at least down to a chip size of 6 μm, using the NBE process. Fig. 13 summarizes the current density at the peak EQE as a function of the chip size for the micro-LEDs fabricated in this study and those reported

Fig. 13 Current density at the peak EQE as a function of the chip size for GaN micro-LEDs reported in the literature and those fabricated in this study. The open squares, solid circles, and open triangles represent data reported from Olivier et al. (2017), Tian et al. (2012), and Hwang et al. (2017), respectively. The solid triangles and solid stars represent data obtained in this study.

Damage-free neutral beam etching

219

by several other groups. A lower current density at the peak EQE indicates a higher IQE for LEDs with similar Auger recombination rates and thus can be used as a figure of merit characterizing the IQE of a GaN LED. All devices shown in Fig. 13 were grown on c-plane sapphire substrates with an emission wavelength of approximately 450 nm. Therefore, it is reasonable to assume that these devices have similar Auger recombination rates because the Auger recombination rate of an InGaN/GaN LED grown on a c-plane sapphire substrate depends mainly on the emission wavelength (Piprek et al., 2015; Zhang et al., 2009). As can be seen from Fig. 13, the current density at the peak EQE of the micro-LED fabricated by the ICP process rapidly increased when the chip size was reduced to below 10 μm. In contrast, the current density at the peak EQE of the micro-LEDs fabricated by the NBE process was essentially independent on the chip size down to at least the size of 6 μm. This further indicates that the NBE process is a powerful technique for fabrication of high-efficiency sub-10-μm GaN micro-LEDs required for high-resolution and high-brightness micro-LED displays.

5. Conclusion In this chapter, we reviewed the fabrication of sub-10-μm GaN micro-LEDs using a novel damage-free dry etching process, NBE, to overcome the problem of efficiency decrease in the sub-10-μm region in GaN micro-LEDs fabricated by the conventional ICP process. In the NBE process, charged particles are neutralized and UV photons emitted from the plasma are blocked by a carbon aperture placed between the plasma and etching chamber. Only an energy-controlled neutral beam is supplied to the sample surface for etching. In this manner, the two major sources of defect generation in the conventional ICP process, ion bombardment and UV photon irradiation, can be avoided, and thus damage-free dry etching can be realized. A series of InGaN/GaN blue micro-LEDs with sizes in the range of 40
40–6
6 μm2 was fabricated using the NBE process to define the device mesa. All fabricated micro-LEDs exhibited EQE-current density characteristics similar to those of large-area GaN LEDs even in the current density region lower than 1 A/cm2, with a maximum in the EQE curves at a current density as low as approximately 5 A/cm2. Furthermore, all fabricated micro-LEDs exhibited similar maximum EQEs with variations smaller than 10%. These results show that the size-dependent efficiency decrease issue for GaN micro-LEDs fabricated by the conventional ICP

220

Xuelun Wang and Seiji Samukawa

process is successfully eliminated by the NBE process. We believe that the NBE method is a very promising process for fabrication of high-efficiency sub-10-μm InGaN/GaN micro-LEDs required for high-resolution and high-brightness micro-LED displays.

References

€ ur, U., € Metzner, S., Hafiz, S., Zhang, F., Monavarian, M., Avrutin, V., Morkoc¸, H., Ozg€ Bertram, F., Christen, J., Gil, B., 2015. Determination of carrier diffusion length in GaN. J. Appl. Phys 117. 013106-1-4. Higo, A., Kiba, T., Chen, S., Chen, Y., Tanikawa, T., Thomas, C., Lee, C.Y., Lai, Y.-C., Ozaki, T., Takayama, J., Yamashita, I., Murayama, A., Samukawa, S., 2017. Optical study of sub-10 nm In0.3Ga0.7N quantum nanodisks in GaN nanopillars. ACS Photonics 4, 1851–1857. Hwang, D., Mughal, A., Pynn, C.P., Nakamura, S., DenBaars, S.P., 2017. Sustained high external quantum efficiency in ultrasmall blue III-nitride micro-LEDs. Appl. Phys. Express 10. 032101-1-4. Jiang, H.X., Jin, S.X., Li, J., Shakya, J., Lin, J.Y., 2001. III-nitride blue microdisplays. Appl. Phys. Lett. 78, 1303–1305. Jung, S., Song, K.R., Lee, S.N., Kim, H., 2013. Wet chemical etching of semipolar GaN planes to obtain brighter and cost-competitive light emitters. Adv. Mater. 25, 4470–4476. Karpov, S.Y., Makarove, Y.N., 2002. Dislocation effect on light emission efficiency in gallium nitride. Appl. Phys. Lett. 81, 4721–4723. Kubota, T., Baba, T., Kawashima, H., Uraoka, Y., Fuyuki, T., Yamashita, I., Samukawa, S., 2005. Study of neutral-beam etching conditions for the fabrication of 7-nm-diameter nanocolumn structures using ferritin iron-core masks. J. Vac. Sci. Technol. B 23, 534–539. Kubota, T., Watanabe, N., Ohtsuka, S., Iwasaki, T., Ono, K., Iriye, Y., Samukawa, S., 2012. Numerical study on electron transfer mechanism by collision of ions at graphite surface in highly efficient neutral beam generation. J. Phys. D: Appl. Phys. 45. 095202-1-5. Minami, M., Tomiya, S., Ishikawa, K., Matsumoto, R., Chen, S., Fukasawa, M., Uesawa, F., Sekine, M., Hori, M., Tatsumi, T., 2011. Analysis of GaN damage induced by Cl2/SiCl4/Ar plasma. Jpn. J. Appl. Phys. 50. 08JE03-1-4. Mizutani, T., Nishimatsu, S., 1988. Sputtering yield and radiation damage by neutral beam bombardment. J. Vac. Sci. Technol. A 6, 1417–1420. Olivier, F., Tirano, S., Dupre, L., Aventurier, B., Largeron, C., Templier, F., 2017. Influence of size-reduction on the performances of GaN-based micro-LEDs for display application. JOL 191, 112–116. Paranjpe, A., Montgomery, J., Lee, S.M., Morath, C., 2018. Micro-LED displays: key manufacturing challenges and solutions. SID Symp. Dig. Tech. Pap. 49, 597–600. Piprek, J., R€ omer, F., Witzigmann, B., 2015. On the uncertainty of the auger recombination coefficient extracted from InGaN/GaN light-emitting diode efficiency droop measurements. Appl. Phys. Lett. 106, 101101-1-4. Samukawa, S., 2006. Ultimate top-down etching processes for future nanoscale devices: advanced neutral-beam etching. Jpn. J. Appl. Phys. 45, 2395–2407. Samukawa, S., 2015. A neutral beam process for controlling surface defect generation and chemical reactions at the atomic layer. ECS J. Solid State Sci. Technol. 4, N5089– N5094.

Damage-free neutral beam etching

221

Samukawa, S., Jinnai, B., Oda, F., Morimoto, Y., 2007. Surface reaction enhancement by UV irradiation during Si etching process with chlorine atom beam. Jpn. J. Appl. Phys. 46, L64–L66. Samukawa, S., Sakamoto, K., Ichiki, K., 2001. High efficiency neutral-beam generation by combination of inductively coupled plasma and parallel plate dc bias. Jpn. J. Appl. Phys. 40, L779–L782. Samukawa, S., Sakamoto, K., Ichiki, K., 2002. Generating high-efficiency neutral beams by using negative ions in an inductively coupled plasma source. J. Vac. Sci. Technol. A 20, 1566. Schubert, E.F., 2007. Light-Emitting Diode, second ed. Cambridge University Press, Cambridge, UK. Shimokawa, F., 1992. High-power fast-atom beam source and its application to dry etching. J. Vac. Sci. Technol. A 10, 1352–1357. Smith, J.M., Ley, R., Wong, M.S., Baek, Y.H., Kang, J.H., Kim, C.H., Gordon, M.J., Nakamura, S., Speck, J.S., DenBaars, S.P., 2020. Comparison of size-dependent characteristics of blue and green InGaN microLEDs down to 1μm in diameter. Appl. Phys. Lett. 116, 071102-1-5. Tian, P., McKendry, J.J.D., Gong, Z., Guilhabert, B., Watson, I.M., Gu, E., Chen, Z., Zhang, G., Dawson, M.D., 2012. Size-dependent efficiency and efficiency droop of blue InGaN micro-light emitting diodes. Appl. Phys. Lett. 101, 231110-1-4. Tripathy, S., Ramam, A., Chua, S.J., Pan, J.S., Huan, A., 2001. Characterization of inductively coupled plasma etched surface of GaN using Cl2/BCl3 chemistry. J. Vac. Sci. Technol. A 19, 2522–2532. Vieri, C., Lee, G., Balram, N., Jung, S.H., Yang, J.Y., Yoon, S.Y., Kang, I.B., 2018. An 18 megapixel 4.300 1443 ppi 120 Hz OLED display for wide field of view high acuity head mounted display. J. Soc. Inf. Disp. 26, 314–324. Wong, M.S., Lee, C., Myers, D.J., Hwang, D., Kearns, J.A., Li, T., Speck, J.S., Nakamura, S., DenBaars, S.P., 2019. Size-independent peak efficiency of III-nitride micro-light-emitting-diodes using chemical treatment and sidewall passivation. Appl. Phys. Express 12, 097004-1-4. Yamashita, I., 2001. Fabrication of a two-dimensional array of nano-particles using ferritin molecule. Thin Solid Films 393, 12–18. Zhang, M., Bhattacharya, P., Sing, J., 2009. Direct measurement of auger recombination in In0.1Ga0.9N/GaN quantum well and its impact on the efficiency of In0.1Ga0.9N/GaN multiple quantum well light emitting diodes. Appl. Phys. Lett. 95, 201108-1-3. Zhang, K.X., Takahashi, T., Ohori, D., Cong, G.W., Endo, K., Kumagai, N., Samukawa, S., Shimizu, M., Wang, X.L., 2020. High-quality nanodisk of InGaN/GaN MQWs fabricated by neutral-beam-etching and GaN regrowth: toward directional micro-LED in top-down structure. Semicond. Sci. Technol. 35, 075001-1-6. Zhu, J., Takahashi, T., Ohori, D., Endo, K., Samukawa, S., Shimizu, M., Wang, X.L., 2019. Near-complete elimination of size-dependent efficiency decrease in GaN micro-lightemitting diodes. Phys. Status Solidi A 216, 1900380-1-6.

CHAPTER SEVEN

From nanoLEDs to the realization of RGB-emitting microLEDs Zhaoxia Bia, Zhen Chenb, Fariba Daneshb,c,†, and Lars Samuelsona,c,∗ a

Lund University, Solid State Physics, NanoLund, Lund, Sweden gl
o-USA, Sunnyvale, CA, United States gl
o AB, Lund, Sweden ∗ Corresponding author. e-mail address: [email protected] b c

Contents 1. An industrial perspective on the market pull for the development of microLED displays 2. General description of the challenges, status, and needs for progress 3. An overview of our different material science approaches toward long-wavelength nitride emitters 3.1 Nanowires as enablers of radial nanowire LEDs 3.2 Planar InGaN red microLEDs 3.3 Development of GaN and InGaN platelets as templates for RGB microLEDs 4. Industrial approaches toward mass transfer of microLEDs 5. Outlook Acknowledgments References

223 225 226 226 232 237 245 247 248 248

1. An industrial perspective on the market pull for the development of microLED displays The market pull for microLED display development as the nextgeneration displays is strong. The advantages over conventional display technologies are promising and acknowledged by all. LCD’s main disadvantages are slow response, low brightness and contrast, and poor color saturation. OLED displays have the advantages of self-emission, wide viewing angle, high contrast, power saving, fast response, but are limited in maximum brightness and have significant reliability problems. MicroLEDs are also self-emissive, therefore have all the advantages of OLED over LCD. †

Present address: “Fariba Danesh Consulting”, Los Altos, CA, United States.

Semiconductors and Semimetals, Volume 106 ISSN 0080-8784 https://doi.org/10.1016/bs.semsem.2021.01.001

Copyright

#

2021 Elsevier Inc. All rights reserved.

223

224

Zhaoxia Bi et al.

Additionally, microLEDs provide higher brightness (100
), higher contrast ratio, higher resolution, higher speed, longer lifetime, better environment stability and reliability, and wide viewing angles. Each characteristic of microLEDs can benefit different kinds of display applications. MicroLED is two orders of magnitude brighter than LCD and OLED. High-end glass-based AR/MR and head-up display (HUD) devices require a transparent display and a high brightness of 100,000 nits to view in an outdoor environment. This brightness can only be provided by microLEDs. The fact that the active area covered by ultra-small microLEDs leads to very high transparency of the displays, even allowing multiple layers of such displays to be combined. Ambient contrast ratio (ACR) is also an essential specification for outdoor applications, because it is sometimes difficult to read the smartphone with an OLED display panel under high ambient light. The ACR of microLED is much higher than that of OLED, because of the much higher brightness of each subpixel. The resolution of the displays specified in pixels per inch (PPI) for microLED displays can be 10 times higher than that of OLED and LCD, which means a much smaller chip size. PPI is critical in AR, VR, and MR applications because the images are projected and seen through optical lenses. In large displays, the smaller subpixel is key to reducing costs. Tiny microLEDs can be bonded to flexible or stretchable substrates to make special displays for foldable smartphones and wearable electronics. When such a small microLED array adheres to a transparent substrate at a low fill factor, a transparent flat panel display can be realized. The small subpixels also leave enough room between the chips to integrate sensors and drivers to achieve functions such as fingerprint identification, touch screen, and gesture control. Due to the high electron mobility, the response speed of microLED is in the order of nanoseconds, while OLED and LCD are in the order of microseconds and milliseconds, respectively. This feature is significant for visible light communication (VLC) application. High dynamic range (HDR) displays and AR/MR displays also prefer a fast response because these devices need more pixels per image and more frames per second for convenience and safety. MicroLEDs have good reliability, environmental stability, and ability to operate in a broader temperature range of 100 to +120°C, which are vital for outdoor applications. In particular, while typical OLEDs can hardly survive at 80 mA/cm2 for over 100 h, microLEDs have been shown to survive at 3.5 kA/cm2 for over 300 h (Tian et al., 2016), which is the key for VLC

From nanoLEDs to the realization of RGB-emitting microLEDs

225

application where the applied current of the microLEDs is several orders higher than that of general lighting. As a result of the above advantages, microLED has a tremendous opportunity to replace LCD and OLED, and occupy all display applications, including AR, VR, MR, wearables, smartphones, automobiles, laptop screens, TVs.

2. General description of the challenges, status, and needs for progress Despite its many advantages and potential applications, there is no commonly accepted technical approach to realize a microLED display. We think the top two challenges are how to realize the highly efficient long-wavelength LEDs and how to efficiently assemble microLEDs. The solutions for these two challenges will be discussed later. Three different approaches have been used to achieve red emission from microLEDs. Converting 450 nm blue or 410 nm UV light into RGB in displays has been achieved by using Quantum dots on microLEDs. However, the color conversion method has the issues of poor reliability due to quantum dot deterioration, a short lifetime, low conversion efficiency, uneven emission of pixels, color/light crosstalk, and lower PPI. Many teams use AlInGaP LEDs for red emission due to its high external quantum efficiency (EQE) in broad-area LED. However, the efficiency of AlInGaP microLEDs drops faster than III-nitride with microLED sizes below 50 μm. Red InGaN microLEDs are a real alternative to AlInGaP for subpixel sizes of 5 μm and below. Gl
o has demonstrated planar InGaN red microLEDs with an EQE of 12%, tested by a third party in 2019, with a completely encapsulated configuration including a lens for an efficient light outcoupling. The same size red InGaN microLEDs on-chip level have been improved showing 2
EQE in 2020. Efficiency improvements for red InGaN microLEDs on engineered substrates are on-going. Mass transfer and assembly are generally regarded as the most substantial technical challenges to overcome to enable microLED display manufacturing. The proposed techniques to solve this issue are two categories: indirect pick-and-place vs direct wafer-scale transfer. Indirect transfer techniques typically include the assembly of microLEDs on a carrier substrate, a print head, or a cartridge. The direct transfer techniques utilize typical wafer-towafer or die-to-wafer type equipment and process. The choice of each technique and its deployment are primarily driven and impacted by the target PPI and size of the display that the company intends to build.

226

Zhaoxia Bi et al.

Still remaining dream scenario is one where different means of patterning, epi-growth and processing could enable different colors to be formed on one and the same wafer, without the challenge of the post-growth assembly into three-color microLEDs. Such efforts are on-going, although still with unproven success for realistic devices. Other challenges include the reduced EQE with a shrunken chip size. The low EQE in microLED is caused by severe sidewall defects from mesa etch, such as dangling bonds, defects, and structural damages. Solutions include new LED structures, chip architecture designs, and process improvement to reduce the effect of sidewall defects. USCB and several other groups demonstrated size-independent EQE for microLED by wet etching and improved passivation, softer etching conditions, etc. (Ley et al., 2020; Zhu et al., 2019). Defects are inevitably introduced during microLED manufacturing, while the current commercial flat panel displays are almost defect-free. The yield of 100% can only be obtained by filtering the bad product in each step. MicroLEDs need thousands of times more testing than broad-area LEDs. High-speed contactless inspection at multiple steps need to be developed. Technology to manage defects in the panel level also helps to solve this problem. One solution is to repair the dysfunctional subpixel, and another is redundancy by doubling the microLEDs. Many other minor challenges in each process also need to be addressed. For instance, the epitaxial structure needs to be designed to reduce the wavelength drift with current density. In circuit design, it is necessary to have a compensation circuit to improve the wavelength and luminance uniformity of the microLED display. In optical design, the crosstalk needs to be minimized, and the emission angle needs to be tuned based on the applications. It is best to develop the structure and process of each step according to the final product with the overall picture and actual situation in mind.

3. An overview of our different material science approaches toward long-wavelength nitride emitters 3.1 Nanowires as enablers of radial nanowire LEDs C-oriented GaN nanowires (NWs) offer interesting features for LEDs, such as the absence of dislocations and offering six equivalent m-plane side facets. InGaN quantum wells (QWs) grown on such non-polar side facets are free of quantum-confined Stark effect (QCSE). This allows the growth of thicker QWs than on the c-plane, which is helpful to suppress the Auger recombination and reduce the efficiency droop of nitride LEDs

From nanoLEDs to the realization of RGB-emitting microLEDs

227

(Kim et al., 2007; Kioupakis et al., 2011). With a p-GaN shell grown around the InGaN QWs, each NW works as an independent LED. By connecting an array of such NWs, microLEDs of different sizes can be fabricated. Such microLEDs do not suffer from leakage caused by dry etch damage on the edges of the mesa because the mesa is actually formed by growth, instead of by etching. Selective area growth has been widely used to grow GaN NWs with metal-organic chemical vapor deposition (MOCVD, Barrigon et al., 2019). For this method, a growth mask is needed, on which GaN tends not to nucleate. The mask is usually SiOx or SiNx deposited by plasma-enhanced or lowpressure chemical vapor deposition. Arrays of circular openings need to be patterned on the mask with lithography, so that the underlying GaN film grown on sapphire, Si or SiC substrate, gets exposed to the growth ambient and seeds the GaN NW growth from the mask openings. E-beam lithography can be used to fabricate openings smaller than 100 nm, but it is not a practical technique for wafer-size patterning due to the low writing speed. For the openings between 100 and 200 nm, nanoimprint lithography can be used to pattern wafers. As an alternative, displacement Talbot lithography, using a three-dimensional interference pattern formed by a grating illuminated by a coherent light, has been recently deployed in our laboratory for a fast exposure on a wafer up to 600 in diameter (Gomez et al., 2020). The growth fashion of GaN NWs is energetically favorable. However, the selective area growth of GaN tends to kinetically form a hexagonal pyramid at the common growth conditions for GaN films ( Jindal and Shahedipour
Sandvik, 2009). This arises from a low growth rate of 1011 planes, in contrast to the c-plane. In order to form GaN NWs with vertical side facets, the growth has to be conducted with either an ultra-low V/III ratio or a pulsed growth mode (Barrigon et al., 2019; Choi et al., 2012; Hersee et al., 2006; Monemar et al., 2016). Both types of growth obviously deviate from the common GaN growth regimes. The optical characterization of GaN NWs usually shows a strong yellow luminescence band (Lin et al., 2014), and a typical cathodoluminescence spectrum at room temperature is shown in Fig. 1A. The strong yellow luminescence indicates a high level of vacancies and impurities of carbon and oxygen. Such impurities are generally buried in the thick n-GaN layer for the planar GaN growth. The thin n-GaN layer grown for NWs may represent a limitation to optimizing NW devices’ performance. Furthermore, the emissions from single GaN NWs are not uniform along the NWs, as shown in Fig. 1B. Here, a GaN NW was transferred onto an Au-coated Si substrate which does not emit any light under E-beam excitation. The hyper spectrum shows a much stronger GaN band-edge emission

228

Zhaoxia Bi et al.

Fig. 1 Cathodoluminescence characterization of a single GaN NW conducted at room temperature with an average spectrum (A) and a hyper spectrum (B). (C) Temporal evolution of GaN NW growth, shown by cross-sectional TEM images of GaN NWs grown for different times. (A) and (B): Courtesy of Anders Gustafsson. (C): Reprinted with permission from Barrigon, E., Heurlin, M., Bi, Z., Monemar, B., Samuelson, L., 2019. Synthesis and applications of III-V nanowires. Chem. Rev. 119, 9170–9220. Copyright 2019 American Chemical Society.

at the most bottom section of the NW for a length of about 200 nm. The NW grown after this bottom section shows reduced luminescence intensity to the NW top. This variation seems related to the growth development of the GaN NWs shown in Fig. 1C (Barrigon et al., 2019). The initial GaN grown out of
the openings possesses smooth 1012 planes at the top pyramid for a length of about 200 nm. During the subsequent growth, other high-index planes develop from the connection part between the m-planes and the top pyramid, leading to multiple-faceted surfaces of the top pyramid (see the growth for 60 s). The growth for the first 30 s corresponds to the bottom section showing the strong band-edge emission in Fig. 1B. The multiple-faceted top surface after the first 30 s growth may introduce impurities, such as carbon or oxygen,

From nanoLEDs to the realization of RGB-emitting microLEDs

229

into the NWs, leading to the low optical quality with reduced band-edge emission. This may also explain the increased yellow luminescence from the position of 700 nm to the NW top. Besides the strong band-edge emission, the most bottom section of the NW also shows a strong yellow luminescence. Meanwhile, the band-edge emission and the yellow luminescence band become weak simultaneously at the position from about 400 to 700 nm. This indicates that a low yellow luminescence intensity does not necessarily indicate good quality, and other mechanisms may be needed to explain the optical data here. InGaN QWs can be grown around the GaN NWs, forming active shells for LEDs. However, the growth of InGaN QWs on the side facets is usually not uniform, that is the thickness and the indium content tend to increase from the NW bottom to the top (Tchernycheva et al., 2014; Yeh et al., 2012; Zhang et al., 2015). In order to characterize the luminescence variation along the NWs, p-contacts were made to the p-GaN shell locally at different positions on a single NW LED with E-beam lithography, as shown in Fig. 2. The NW LED structure can be found in a previous publication (Monemar et al., 2016) and Fig. 2A shows an SEM image of arrays of GaN NW LEDs. The NW was transferred to a SiOx/Si substrate, and the n-contact was made to the n-GaN core at the NW bottom. Separate p-contacts were made from the NW bottom to the top pyramid. The functional p-contacts were marked from 1 to 5, covering the NW middle to the top (Fig. 2B). The electroluminescence (EL) for the contacts 1–3 shows similar emissions at 474nm (Fig. 2C). However, the EL for the contact 4 at the connection part between the m-planes and the top pyramid shows a red-shift to 490 nm due to the local thicker QWs with a higher indium content. For the contact 5 at the top pyramid, two peaks were observed at 432 and 490 nm. Thin QWs are
formed on the pyramid facets of 1011 due to the low growth rate, leading to a blue-shift of the emission to 432 nm. Meanwhile, carriers injected from contact 5 may diffuse to the QWs around the contact 4, leading to the emission at 490 nm. In order to increase the monochromaticity of the emissions, passivation and etching of the QWs at the NW top were reported (Zhang et al., 2015). Meanwhile, short GaN NWs are suggested to be used to obtain m-plane QWs with improved uniformity on the thickness and the indium content. gl
o has published microLED performance of NW microLEDs processed side by side to standard c-plane only planar microLEDs. High-quality best-performing green InGaN material was grown on 200 sapphire substrates in Aixtron MOCVD reactors with recipes optimized over time for each type

230

Zhaoxia Bi et al.

Fig. 2 EL characterization of a single NW LED. (A) A tilted-view SEM image of arrays of GaN NW LEDs. (B) A single NW LED with an n-contact at the NW bottom to the n-GaN core and several separate p-contacts to the p-GaN shell. (C) EL spectra measured for the p-contacts 1–5 at 10 μA. Courtesy of Kristian Storm.

of growth. The planar epitaxy samples were first processed as mini-LEDs of 200
200 μm2 in a benchmarking exercise. They were measured, showing a luminous efficacy of 110 lm/W at 10 A/cm2 operating current. Then wafers with above planar epitaxy were processed side by side with optimized NW epitaxy wafers with a mask producing microLED sizes ranging from 5
5 μm2 to 20
20 μm2 on each wafer. The devices were then characterized on each wafer side by side. A sample of 10 devices per size was measured over a range of current densities. As shown in Fig. 3, the NW microLEDs’ efficiency peaks at very low current densities of 0.1–0.2 A/cm2, while the planar microLEDs show efficiency peaks at tens of A/cm2. As seen in Fig. 3, the efficiency of the planar microLEDs suffers at low operating current densities because the etch damage of the mesa introduces leakage paths that consume the carriers before radiative recombination takes place. It should also be noted that the planar c-plane LED wafers are typically used for LED sizes beyond 50
50 μm2. Optimizing the standard planar c-plane growth to move the peak efficiency to the low current densities may improve this performance. In addition, reducing the size of microLED

From nanoLEDs to the realization of RGB-emitting microLEDs

231

Fig. 3 (A) A side by side comparison of electrical efficiency, comparing NW green microLEDs to standard planar c-plane green microLEDs at sizes ranging from 5
5 μm2–20
20 μm2. (B) A magnified picture of an all InGaN full-color display using
AB. 20
20 μm2 subpixels on an LTPS backplane. Courtesy of glo

devices from 20
20 μm2 to 5
5 μm2 did not reduce the efficiency of the NW microLEDs. A reduction of the microLED size in the very same regime is well proven to impact the planar material performance. This is well understood to be an impact of the ratio of edge vs bulk material. In the NW devices, this ratio is not greatly affected by the size of the microLEDs since

232

Zhaoxia Bi et al.

the NW each has a p-shell growth that prevents leakage and impact of the size reduction. gl
o has made significant progress in improving peak EQE for planar structures by virtue of cleaning and passivating the etched surfaces of GaN. Such efforts have shown significant improvement with peak EQEs of planar structures exceeding those shown in Fig. 3 for NW LEDs. This peak efficiency at low current densities may be a major advantage or even a necessity in display applications that use low-temperature polycrystalline silicon (LTPS) thin-film transistors (TFT) made on glass. Planar epitaxy structures have been tuned to enable EQE for planar structures showing peaks at low current densities as well. The LTPS technology has difficulty delivering high currents to each subpixel. Producing LTPS material to deliver higher currents requires larger or thicker metal layers in the TFT which then cause a significant level of efficiency penalty in the native efficiency of the back plane. Such a system penalty may make it difficult for microLED displays to compete with OLEDs.

3.2 Planar InGaN red microLEDs 3.2.1 Motivation and challenges in developing InGaN red microEDs Highly efficient InGaN blue LEDs with an EQE of 84.3% (Narukawa et al., 2010) and AlInGaP red LEDs with an EQE of 55% (Krames et al., 1999) are demonstrated. Therefore, many companies use AlInGaP LEDs as a red emission source. However, the efficiency of AlInGaP microLEDs drops faster than III-nitride with microLED sizes below 50 μm due to higher surface recombination velocity and larger carrier diffusion coefficients in AlInGaP. In high-resolution color display applications, the small pixel sizes of a few micrometers are desired. The reported ambipolar diffusion coefficients of InGaN MQWs and InGaP MQWs are 0.1–1 cm2/s (Aleksiej
unas et al., 2014; Malinauskas et al., 2012) and 3 cm2/s (Prasad et al., 1994; Thiagarajan et al., 1991), respectively. Moreover, InGaN LED has a much smaller surface recombination velocity of 3
102 to 104 cm/s than that of AlGaInP red LEDs (106 cm/s) (Kitagawa et al., 2011; Onuma et al., 2012), which makes the sidewall effect in AlGaInP red microLEDs much more severe than that in InGaN red microLEDs. These behaviors make it possible for InGaN red microLED to have a higher EQE than AlInGaP one in subpixel sizes 99.9999%. A 4 K HDTV has approximately 25 million chips. The methods used to assemble microLEDs on a panel are diverse. However, it is helpful to think about the techniques as divided into two categories: (1) indirect and (2) direct wafer-scale transfer. Indirect transfer techniques use an intermediary vehicle such as a carrier substrate, a print head, or a cartridge to transfer the microLEDs. For displays >100 , the cost of the microLEDs is of a key concern. Hence, chip size of 100 . Applications such as smartwatch, smartphone and larger require standard TFT backplanes. The large surface area of the display makes the use of CMOS as a backplane impossible due to the cost. Such displays have resolutions in hundreds of PPI, and may be built with indirect techniques. Examples include MEMS-based carrier print heads used by Luxvue, polymer print heads used by X-celeprint, or cartridges used by Rohinni and PlayNitride. Direct transfer by wafer-to-wafer bonding: The direct transfer techniques use standard wafer-to-wafer or die-to-wafer equipment/processes with some customization. For near-eye displays (AR/VR), at sizes of 2000, the direct techniques are typical. CMOS backplanes are needed because TFT cannot achieve a resolution of  > 800 PPI, and the cost of CMOS backplanes is supportable. In this case, the density of the microLEDs on the wafer may be the same as that of the display. Hence, the actual microLED wafer is brought into contact with the display backplane and the LEDs are bonded onto the CMOS backplane. With the notable exception of gl
o, all others using such direct transfer to date, have produced monochrome displays only (Ou et al., 2018). The expectation is that monochrome displays can then be assembled as RGB with further integration of color converting quantum dots, or with projection optics.

246

Zhaoxia Bi et al.

An example is Plessey. The microLEDs are positioned on their original substrates, brought in contact to the backplane. It has been prepared with bonding pads and eutectics of various heights for each color RGB (Bernard et al., 2019). Indirect transfer with electrostatic transfer head: LuxVue, acquired by Apple in 2014, demonstrated the use of electrostatic forces to transfer microLED chips. In Fig. 10A and B (Bibl et al., 2012), a mass transfer tool with an array of microLED transfer heads is shown. A voltage is selectively applied to the transfer head generating a grasping force between the transfer head and the microLED, that picks up the microLED from the carrier substrate. When the chip is transferred to the desired position, a negative voltage is applied, and the transfer head releases the chip. Indirect transfer with elastomer stamp printing: X-celeprint uses this technology to transfer chips with a flexible polydimethylsiloxane (PDMS) stamp. An array of posts on a transfer stamp is pressed against a spatially matching array of components released from a source wafer to adhere the components to the stamp posts, as shown in Fig. 11 (Cok et al., 2017). The transfer stamp is transported to a destination wafer and the components pressed against a receiving surface on the backplane. The stamp can transfer 82,863 chips with a size of 75 μm
90 μm in a single print. It is claimed that chips down to a size of 3 μm can be printed (Cok et al., 2017). Others: Other examples include fluidic transfer used by e-Lux. The liquid suspension of the microLEDs flows on top of the TFT backplane substrate, and the microLEDs are captured in wells with shapes matching a stud on the microLED bottom. A 12-in. microLED panel of 42–168 PPI with 8,294,400 microLEDs sized at 20 μm can be assembled in 15 min (Wu et al., 2018). This rate is 9216 microLEDs per second.

Fig. 10 (A) An Illustration of a transfer head array utilizing electrostatic force. (B) A detailed structure of a transfer head.

From nanoLEDs to the realization of RGB-emitting microLEDs

247

Fig. 11 Transfer printing with a PDMS stamp. Reprinted with permission from Cok, R.S., Meitl, M., Rotzoll, R., Melnik, G., Fecioru, A., Trindade, A.J., Raymond, B., Bonafede, S., Gomez, D., Moore, T., Prevatte, C., Radauscher, E., Goodwin, S., Hines, P., Bower, C.A., 2017. Inorganic light-emitting diode displays using micro-transfer printing, J. Soc. Inf. Disp. 25, 589–609. Copyright 2017 John Wiley and Sons.

5. Outlook Although the tiny chip size and the challenging display specifications for all applications present substantial problems to be solved in the realization of commercial microLED displays, the many advantages of the technology compared with current LCD and OLED displays are highly compelling. We believe microLED will become the next-generation technology for most applications such as AR, VR, MR, wearable devices, smartphones, automobiles, laptop screens, HDTVs. In addition, new types of displays such as flexible displays and transparent displays integrated into surfaces never before used as displays will emerge. The first microLED display products will likely be high PPI applications such as AR, MR, VR, wearable devices, and projectors. This first generation will prove and commercialize many of the advantages of the microLEDs including higher brightness, higher ambient contrast, higher compactness,

248

Zhaoxia Bi et al.

higher PPI, enhanced robustness, better environmental stability. The next wave microLED display products will likely be automotive HUDs with similar requirements but low PPI. There are currently over 150 companies working on the development and commercialization of microLED technology. Since many problems need to be resolved before mass production, it will take more time to mature the technology to support high-volume consumer markets such as smartphones, TVs, etc. In the long run, microLED will be used as in new products for novel applications, such as VLC, lithography, biomedical and health equipment to aid people with visual or hearing impairments, and so on.

Acknowledgments The authors want to acknowledge invaluable contributions to the development of this research and development from colleagues both at Lund University and at gl
o-USA and gl
o AB. This work was performed with financial support from the Swedish Research Council, the Swedish Foundation for Strategic Research, the Knut and Alice Wallenberg Foundation, and the Swedish Energy Agency. This project has also received funding from the European Union under the project “NWs4LIGHT” (grant no. 280773). This publication reflects only the author’s views, and the funding agency is not responsible for any use that may be made of the information it contains.

References Aleksiej
unas, R., Gelzˇinyt_e, K., Nargelas, S., Jarasˇi
unas, K., Vengris, M., Armour, E.A., Byrnes, D.P., Arif, R.A., Lee, S.M., Papasouliotis, G.D., 2014. Diffusion-driven and excitation-dependent recombination rate in blue InGaN/GaN quantum well structures. Appl. Phys. Lett. 104, 022114. Barrigon, E., Heurlin, M., Bi, Z., Monemar, B., Samuelson, L., 2019. Synthesis and applications of III-V nanowires. Chem. Rev. 119, 9170–9220. Bernard, J., Haas, H., Caplet, S., Bedoin, A., Tournaire, M., Peray, P., 2019. MicroLED displays based on transfer with microtubes interconnections. SID Symposium Digest of Technical Papers 50, 244–247. Bi, Z., Gustafsson, A., Lenrick, F., Lindgren, D., Hultin, O., Wallenberg, L.R., Ohlsson, B.J., Monemar, B., Samuelson, L., 2018. High In-content InGaN nanopyramids: tuning crystal homogeneity by optimized nucleation of GaN seeds. J. Appl. Phys. 123, 025102. Bi, Z., Lenrick, F., Colvin, J., Gustafsson, A., Hultin, O., Nowzari, A., Lu, T., Wallenberg, R., Timm, R., Mikkelsen, A., Ohlsson, B.J., Storm, K., Monemar, B., Samuelson, L., 2019. InGaN platelets: synthesis and applications toward green and red light-emitting diodes. Nano Lett. 19, 2832–2839. Bi, Z., Lu, T., Colvin, J., Sj€ ogren, E., Vainorius, N., Gustafsson, A., Johansson, J., Timm, R., Lenrick, F., Wallenberg, R., Monemar, B., Samuelson, L., 2020. Realization of ultrahigh quality InGaN platelets to be used as relaxed templates for red micro-LEDs. ACS Appl. Mater. Interfaces 12, 17845–17851. Bibl, A, Higginson, J.A, Law, H. F. S., Hu, H. H, 2012. Micro device transfer head heater assembly and method of transferring macro device, US 8 349 116 B1 (filed 2012).

From nanoLEDs to the realization of RGB-emitting microLEDs

249

Choi, K., Arita, M., Arakawa, Y., 2012. Selective-area growth of thin GaN nanowires by MOCVD. J. Cryst. Growth 357, 58–61. Cok, R.S., Meitl, M., Rotzoll, R., Melnik, G., Fecioru, A., Trindade, A.J., Raymond, B., Bonafede, S., Gomez, D., Moore, T., Prevatte, C., Radauscher, E., Goodwin, S., Hines, P., Bower, C.A., 2017. Inorganic light-emitting diode displays using micro-transfer printing. J. Soc. Inf. Disp. 25, 589–609. Coulon, P.M., Alloing, B., Br€andli, V., Vennegues, P., Lerous, M., Zuniga-Perez, J., 2016. Dislocation filtering and polarity in the selective area growth of GaN nanowires by continuous-flow metal organic vapor phase epitaxy. Appl. Phys. Express 9, 015502. Doi, T., Honda, Y., Yamaguchi, M., Amano, H., 2013. Strain-compensated effect on the growth of InGaN/AlGaN multi-quantum well by metalorganic vapor phase epitaxy. Jpn. J. Appl. Phys. 52, 08JB14. Even, A., Laval, G., Ledoux, O., Ferret, P., Sotta, D., Guiot, E., Levy, F., Robin, I.C., Dussaigne, A., 2017. Enhanced In incorporation in full InGaN heterostructure grown on relaxed InGaN pseudo-substrate. Appl. Phys. Lett. 110, 262103. Gomez, V.J., Graczyk, M., Jam, R.J., Lehmann, S., Maximov, I., 2020. Wafer-scale nanofabrication of sub-100 nm arrays by deep UV displacement Talbot lithography. Nanotechnology 31, 295301. Gorczyca, I., Suski, T., Strak, P., Staszczak, G., Christensen, N.E., 2017. Band gap engineering of In(Ga)N/GaN short period superlattices. Sci. Rep. 7, 16055. Hashimoto, R., Hwang, J., Saito, S., Nunoue, S., 2013. High-efficiency green-yellow light-emitting diodes grown on sapphire (0001) substrates. Phys. Status Solidi C 10 (11), 1529–1532. Hersee, S.D., Sun, X., Wang, X., 2006. The controlled growth of GaN nanowires. Nano Lett. 6, 1808–1811. Hwang, J.-I., Hashimoto, R., Saito, S., Nunoue, S., 2014. Development of InGaN-based red LED grown on (0001) polar surface. Appl. Phys. Express 7, 071003. Iida, D., Niwa, K., Kamiyama, S., Ohkawa, K., 2016. Demonstration of InGaN-based orange LEDs with hybrid multiple-quantum-wells structure. Appl. Phys. Express 9, 111003. Iida, D., Zhuang, Z., Kirilenko, P., Velazquez-Rizo, M., Ohkawa, K., 2020. Demonstration of low forward voltage InGaN-based red LEDs. Appl. Phys. Express 13, 031001. Jindal, V., Shahedipour-Sandvik, F., 2009. Theoretical prediction of GaN nanostructure equilibrium and nonequilibrium shapes. J. Appl. Phys. 106, 083115. Khalilian, M., Bi, Z., Johansson, J., Lenrick, F., Hultin, O., Colvin, J., Timm, R., Wallenberg, R., Ohlsson, J., Pistol, M.E., Gustafsson, A., Samuelson, L., 2020. Dislocation-free and atomically flat GaN hexagonal microprisms for device applications. Small 16, 1907364. Kim, K.C., Schmidt, M.C., Sato, H., Wu, F., Fellows, N., Jia, Z., Saito, M., Nakamura, S., DenBaars, S.P., Speck, J.S., 2007. Study of nonpolar m-plane InGaN/GaN multiquantum well light emitting diodes grown by homoepitaxial meta-organic chemical vapor deposition. Appl. Phys. Lett. 91, 181120. Kioupakis, E., Rinke, P., Delaney, K.T., Van de Walle, C.G., 2011. Indirect Auger recombination as a cause of efficiency droop in nitride light-emitting diodes. Appl. Phys. Lett. 98, 161107. Kitagawa, H., Fujita, M., Suto, T., Asano, T., Noda, S., 2011. Green GaInN photoniccrystal light-emitting diodes with small surface recombination effect. Appl. Phys. Lett. 98, 181104. Krames, M.R., Ochiai, H.M., Hofler, G.E., Carter, C.C., 1999. High-power truncatedinverted-pyramid light-emitting diodes exhibiting 50% external quantum efficiency. Appl. Phys. Lett. 75, 2365–2367.

250

Zhaoxia Bi et al.

Ley, R.T., Smith, J.M., Wong, M.S., Margalith, T., Nakamura, S., DenBaars, S.P., Gordon, M.J., 2020. Revealing the importance of light extraction efficiency in InGaN/GaN microLEDs via chemical treatment and dielectric passivation. Appl. Phys. Lett. 116, 251104. Lin, Y.T., Yeh, T.W., Nakajima, Y., Dapkus, P.D., 2014. Catalyst-free GaN nanorods synthesized by selective area growth. Adv. Funct. Mater. 24, 3162–3171. Malinauskas, T., Kadys, A., Grinys, T., Nargelas, S., Aleksiejnas, R., Miasojedovas, S., Mickeviius, J., Tomasˇinas, R., Jarasˇinas, K., Vengris, M., Okur, S., Avrutin, V., Li, X., Zhang, F., Ozg€ ur, U., Morkoc, H., 2012. Impact of carrier localization, recombination, and diffusivity on excited state dynamics in InGaN/GaN quantum wells. Proc. SPIE 8262, 82621S. Mitchell, B., Dierolf, V., Gregorkiewicz, T., Fujiwara, Y., 2018. Perspective: toward efficient GaN-based red light emitting diodes using europium doping. J. Appl. Phys. 123, 160901. Monemar, B., Ohlsson, B.J., Gardner, N.F., Samuelson, L., 2016. Nanowire-based visible light emitters, present status and outlook. In: Dayeh, S.A., Morral, F.I., Jagadish, C. (Eds.), Semiconductors and Semimetals. vol. 94. Academic Press, Burlington, pp. 227–271. Mukai, T., Yamada, M., Nakamura, S., 1999. Characteristics of InGaN-based UV/Blue/ Green/Amber/Red light-emitting diodes. Jpn. J. Appl. Phys. 38, 3976–3981. Narukawa, Y., Ichikawa, M., Sanga, D., Sano, M., Mukai, T., 2010. White light emitting diodes with super-high luminous efficacy. J. Phys. D 43, 354002. Ohkawa, K., Watanabe, T., Sakamoto, M., Hirako, A., Deura, M., 2012. 740-nm emission from InGaN-based LEDs on c-plane sapphire substrates by MOVPE. J. Cryst. Growth 343, 13–16. Onuma, T., Sakai, N., Igaki, T., Yamaguchi, T., Yamaguchi, A.A., Honda, T., 2012. Comparative study of surface recombination in hexagonal GaN and ZnO surfaces. J. Appl. Phys. 112, 063509. Ou, F., Chong, W.C., Xu, Q., Chen, Y., Li, Q., Zhang, L., 2018. Monochromatic active matrix micro-LED micro-displays with >5,000 dpi pixel density fabricated using monolithic hybrid integration process. SID Symposium Digest of Technical Papers 49, 1677–1680. Ozaki, T., Funato, M., Kawakami, Y., 2019. Red-emitting InxGa1-xN/InyGa1-yN quantum wells grown on lattice-matched InyGa1-yN/ScAlMgO4 (0001) templates. Appl. Phys. Express 12, 011007. Prasad, M., Martinez, O.E., Menoni, C.S., Rocca, J.J., Chillam, J.L.A., Hafich, M.J., Robinson, G.Y., 1994. Transient grating measurements of ambipolar diffusion and carrier recombination in InGaP/lnAIP multiple quantum wells and InGaP bulk. J. Electron. Mater. 23, 359–362. Sharma, T.K., Towe, E., 2010. Application-oriented nitride substrates: the key to long-wavelength nitride lasers beyond 500 nm. J. Appl. Phys. 107, 024516. Shioda, T., Yoshida, H., Tachibana, K., Sugiyama, N., Nunoue, S., 2012. Enhanced light output power of green LEDs employing AlGaN interlayer in InGaN/GaN MQW structure on sapphire (0001) substrate. Phys. Status Solidi A 209, 473–476. Tchernycheva, M., Lavenus, P., Zhang, H., Babichev, A.V., Jacopin, G., Shahmohammadi, M., Julien, F.H., Ciechonski, R., Vescovi, G., Kryliouk, O., 2014. InGaN/GaN core-shell sin
oe nanowire light emitting diodes with graphenebased p-contact. Nano Lett. 14, 2456–2465. Thiagarajan, P., Schmerge, J.F., Menoni, C.S., Marconi, M., Martinez, O.E., Rocca, J.J., Hafich, M.J., Lee, H.Y., Robinson, G.Y., 1991. Picosecond absorption dynamics of photoexcited InGaP epitaxial films. Appl. Phys. Lett. 59, 90–92.

From nanoLEDs to the realization of RGB-emitting microLEDs

251

Tian, P., Althumali, A., Gu, E., Watson, I.M., Dawson, M.D., Liu, R., 2016. Aging characteristics of blue InGaN micro-light emitting diodes at an extremely high current density of 3.5 kA cm2. Semicond. Sci. Technol. 31, 045005. Wu, X., Liu, J., Quan, Z., Xiong, C., Zheng, C., Zhang, J., Mao, Q., Jiang, F., 2014. Electroluminescence from the sidewall quantum wells in the V-shaped pits of InGaN light emitting diodes. Appl. Phys. Lett. 104, 221101. Wu, T.Z., Sher, C.W., Lin, Y., Lee, C.F., Liang, S.J., Lu, Y.J., Chen, S.W.H., Guo, W.J., Kuo, H.C., Chen, Z., 2018. Mini-LED and microLED: promising candidates for the next generation display technology. J. Appl. Sci. 8, 1557. Yeh, T.W., Lin, Y.T., Stewart, L.S., Dapkus, P.D., Sarkissian, R., O’Brien, J.D., Ahn, B., Nutt, S.R., 2012. InGaN/GaN multiple quantum wells grown on nonpolar facets of vertical GaN nanorod arrays. Nano Lett. 12, 3257–3262. Yoshikawa, A., Hashimoto, N., Kikukawa, N., Che, S.B., Ishitani, Y., 2005. Growth of InN quantum dots on N-polarity GaN by molecular-beam epitaxy. Appl. Phys. Lett. 86, 153115. Yoshikawa, A., Che, S.B., Yamaguchi, W., Saito, H., Wang, X.Q., Ishitani, Y., Hwang, E.S., 2007. Proposal and achievement of novel structure multiple quantum wells consisting of 1 ML and fractional monolayer InN wells inserted in GaN matrix. Appl. Phys. Lett. 90, 073101. Zhang, H., Jacopin, G., Neplokh, V., Largeau, L., Julien, F.H., Kryliouk, O., Tchernycheva, M., 2015. Color control of nanowire InGaN/GaN light emitting diodes by post-growth treatment. Nanotechnology 26, 465203. Zhang, S., Zhang, J., Gao, J., Wang, X., Zhang, C., Zhang, M., Wu, X., Xu, L., Ding, J., Quang, Z., Jiang, F., 2020. Efficient emission of InGaN-based light-emitting diodes: toward orange and red. Photonics Res. 8, 1671–1675. Zhu, J., Takahashi, T., Ohori, D., Endo, K., Samukawa, S., Shimizu, M., Wang, X.L., 2019. Near-complete elimination of size-dependent efficiency decrease in GaN micro-light-emitting diodes. Phys. Status Solidi A 216, 1900380.

CHAPTER EIGHT

Mass transfer for Micro-LED display: Transfer printing techniques Changhong Linghu, Shun Zhang, Chengjun Wang, Hongyu Luo, and Jizhou Song∗ Department of Engineering Mechanics, Soft Matter Research Center, and Key Laboratory of Soft Machines and Smart Devices of Zhejiang Province, Zhejiang University, Hangzhou, China ∗ Corresponding author: e-mail address: [email protected]

Contents 1. 2. 3. 4.

Introduction Mass transfer techniques for Micro-LED displays Transfer printing techniques for Micro-LED assembly Latest development of transfer printing techniques 4.1 Magnet-controlled transfer printing 4.2 Laser-driven programmable non-contact transfer printing 4.3 SMP-based universal transfer printing 4.4 Laser-driven active micro-structured transfer printing 5. Conclusion References

253 255 259 261 261 263 267 272 276 277

1. Introduction Regarded as the most promising candidate for the next-generation display technology (Wu et al., 2018), Micro-LEDs have excited both the research community and the industry for their great advantages over LCDs and OLEDs in terms of brightness, efficiency, reliability, response time and life span (Huang et al., 2020; Wu et al., 2018; Zhou et al., 2020). However, the commercialization process of Micro-LED displays has been discouraged by the great challenges and prohibitive cost during the integration of Micro-LED chips from the fabrication wafers onto the display panels. For the integration of the full-color and direct-view displays (Lee et al., 2016), massive RGB Micro-LED dies and driving IC chips grown on the Semiconductors and Semimetals, Volume 106 ISSN 0080-8784 https://doi.org/10.1016/bs.semsem.2020.12.002

Copyright

#

2021 Elsevier Inc. All rights reserved.

253

254

Changhong Linghu et al.

Fig. 1 Mass transfer techniques for Micro-LED display.

epiwafers must be transferred accurately onto the backplane with the desired distribution quickly, as shown in Fig. 1. The display products, such as large-area displays, flexible displays, TVs, laptops, smart phones/watches usually have a relatively large area. To relocate the small dies and chips and distribute them onto such a large area quickly is very challenging. The mass transfer process has long been one of the major bottlenecks of the Micro-LED display (Ding et al., 2019). Firstly, the transfer head should be able to pick up the Micro-LED dies and IC chips firmly and release them onto the backplane reliably. This contradiction entails careful choosing of the picking-up and release principles, deliberate design of the transfer head structure and the transfer scheme. What’s more, the small lateral dimensions (56 M UPH). One disadvantage of this method is that an excess of chips is required to be dispersed in the liquid, far more than the chip amount required for the LED display (Ahn et al., 2008). Moreover, failures at some binding sites are also constantly encountered. The design and fabrication of the chips and binding wells further add undesired complexities. Besides, the repair and encapsulation will meet with some difficulties due to the liquid environment (Zhou et al., 2020). Electrostatic transfer (Ashdown and Speier, 2015; Bibl et al., 2018; Chaji and Fathi, 2019; Golda and Bibl, 2017; Hu et al., 2013) picks up the chips using the electrostatic attraction force between the transfer head and the Micro-LED dies. Usually, the center spaces of the convex pillars are designed as integral multiples of the distance between the neighboring Micro-LED dies on the donor substrate. This facilitates accurate registration between the transfer head and the LED dies, and a passive selectivity (i.e., once the transfer head is fabricated, the selection pattern is set). Electrostatic transfer is able to transfer a large amount of Micro-LED dies in a batch cycle (typical transfer speed is about 12 M UPH), and it can manipulate the chips as small as 1 μm. To ensure a sufficient electrostatic force for the reliable gripping, the LED dies must be relatively flat due to the thin nature of the convex pillars on the transfer head. And careful preparation of the dielectric layer is required to protect the LED chips from the harm of the electrostatic interaction. Magnetic transfer technique (Ashdown et al., 2011; Henley, 2018; Wu et al., 2017, 2020) takes the advantage of the controllable magnetic attraction force between the magnetic transfer head and the LED dies. The key points of magnetic transfer are the MEMS electromagnet module and the magnetic coating layer. The programmable electromagnet, consisting of the coils that can be individually addressed, can pick up the Micro-LED chips selectively. This active selectivity makes this transfer method more versatile and powerful, especially used for repairing. However, the complex design of the transfer head inevitably adds to the fabrication cost and confines the pack

258

Changhong Linghu et al.

density of the magnets. To provide the magnetic force for picking up, a magnetic coating layer should be applied on the Micro-LED dies. This coating layer can be removed subsequently and does not affect the performance of the chips. Besides, the coating also improves the compatibility of the magnetic transfer technique with the non-planar LED chips. The typical transfer speed for magnetic transfer is about 0.9 M UPH and the companies focusing on this method are Hcsemitek, Cooledge and ITRI. Laser transfer technique (Delaporte and Alloncle, 2016; Henley, 2018; Lee et al., 2019) applies a laser beam to locally ablate the buffer layer connecting the LED chips and the transfer head. The buffer layer is deliberately designed with the high absorption of the laser energy and the easy ablation at a relatively low thermal power. The selectivity can be very high, depending on the spatial resolution of the laser beam. However, the direct ablation of the buffer layer might cause thermal damage to the chips, and the residuals will also impair the chips’ performance. Uniqarta (Marinov, 2018; Marinov et al., 2012) further improved the laser transfer method by introducing a dynamic release layer. During the release process, only the upper portion away from the chips are ablated. The ablation-generated gas creates a blister in the dynamic release layer without cracking. The expanding blister drives the release of the LED chips. This thermal-mechanical process can prevent the thermal damage and residual contamination efficiently. By scanning of multiple laser beams, the transfer speed can be as high as 500 M UPH (Marinov, 2018). Transfer printing (Linghu et al., 2018; Yoon et al., 2015) represents a set of transfer methods where an elastomer stamp is used to pick up the LED chips from the donor substrate and print them onto the receiver substrate. The intrinsic van der Waals adhesion between the stamp and the LED chips and the controllable adhesion reduction on-demand facilitates reliable picking up and printing (or release) of the LED chips, respectively. This method is very versatile in its good compatibility with different LED chip materials at different scales (the minimal size can be down to nanoscale) (Choi et al., 2018; Kim et al., 2016). The fabrication of the stamp is very easy and simple by molding and curing of the liquid polymer precursors. Since the transfer process relies on tunable adhesion between the stamp and the chips, the stamp can be used repeatedly to further reduce the cost. The stamp can be easily scaled into large area to achieve high throughput and fast transfer speed (Gomez et al., 2017). The transparent nature of the stamp enables the simultaneous monitoring of the registration process. This ensures a high alignment accuracy during the picking-up and printing process

Mass transfer and transfer printing

259

(Bower et al., 2015; Gomez et al., 2016). All these advantages of transfer printing have earned it immense focus and fast development in recent years, as stated in detail in the next session.

3. Transfer printing techniques for Micro-LED assembly In a typical transfer printing cycle, an elastomer stamp is used to approach the LED chips on the donor substrate as shown in Table 1. To pick up the LED chips successfully, the adhesion between the LED chips and the stamp should be stronger than the adhesion between the LED chips and the donor substrate. After picking-up, the LED chips attached to the stamp are moved to the receiver substrate. Then the adhesion between the LED chips and the stamp are reduced to a value smaller than that between the LED chips and the receiver substrate. The underlying mechanism is the competing fracture between the stamp/chip/substrate interfaces (Feng et al., 2007; Kimlee et al., 2014) as shown in Fig. 2A. Usually, the adhesion at the chip/ substrate interface is a constant. Consequently, the key to transfer printing is the modulation of the adhesion at the stamp/chip interface (Fig. 2B) by outer stimuli like peeling speed (Meitl et al., 2006) and so on. A key factor for the performance of a transfer printing technique is the adhesion switchability, which is defined as the ratio of the maximum adhesion strength to the minimum adhesion strength. The greater the switchability, the better the adhesion modulation effect. Another factor is the residual adhesion or the minimum adhesion a transfer printing technique can reach. The residual adhesion determines the reliability of the printing process. Transfer printing process is very flexible in terms of selectivity. The picking-up process can be conducted in a high-throughput way using a flat stamp to retrieve all the chips on the donor substrate at one time (Cho et al., 2016; Kim et al., 2011; Meitl et al., 2006), or in a passive selective way using stamps with predefined surface patterns (Ahmed et al., 2015; Kim et al., 2011) (Fig. 2C-1). While printing, the chips can be released all at one time without selectivity (Cho et al., 2016; Meitl et al., 2006) or actively patterned by localized or patterned stimuli (Linghu et al., 2019, 2020; Luo et al., 2020; Wang et al., 2020) (Fig. 2C-2). In most scenarios, adhesion between the chips and the receiver substrate is required to overcome the attraction force between the stamp and the chips. Consequently, the chips should contact the receiver substrate for printing. The contact printing (Fig. 2C-3) mode is limited to flat receiver substrate. Some state-of-the-art transfer printing techniques

Fig. 2 Principles of transfer printing. (A) The two interfaces in the stamp/ink/substrate structure. (B) Adhesion strength modulated by external stimulus, showing the high (ON) and low (OFF) adhesion state and the switchability. (C) Illustrations of different transfer printing modes.

Mass transfer and transfer printing

261

(Gao et al., 2017; Luo et al., 2020; Saeidpourazar et al., 2012) can reduce the adhesion between the chip and stamp under zero to facilitate non-contact mode printing (Fig. 2C-4). That is, the chips are suspend above the receive substrate before printing. The non-contact mode can break through the limitation of the receiver geometry and material properties.

4. Latest development of transfer printing techniques Typical transfer printing techniques include kinetically controlled transfer printing technique (Meitl et al., 2006) (where the adhesion is modulated by the peeling speed of the stamp), laser-driven non-contact transfer printing technique (Gao et al., 2017; Saeidpourazar et al., 2012) (where a laser pulse is used to cause the differential thermal expansion to drive the release), gecko-inspired transfer printing technique (representing a set of techniques with fibrillar surface assisted by specific manipulation such as retraction angle or lateral movement) (Carlson et al., 2011; Yoo et al., 2014), and aphid-inspired transfer printing technique (representing a set of techniques based on the change of contact area) (Carlson et al., 2012; Kim et al., 2010). The details of these transfer printing techniques have already been reviewed in the literature (Linghu et al., 2018). In this section, we will introduce the latest development of transfer printing techniques.

4.1 Magnet-controlled transfer printing To attain the high-reliability and high-efficiency requirement, Linghu et al. (2019a, 2019b) developed a magnet-controlled transfer printing technique. They use a rapidly tunable and highly reversible adhesive surface inspired by aphids’ adhesion pad as the stamp for transfer printing. The magnetcontrolled stamp features open reservoirs in an elastomer body filled with magnetic particles and sealed with an adhesive membrane. The magnet-controlled transfer printing process is illustrated in Fig. 3A. Out of the magnetic field, the adhesive membrane of the stamp remains flat, which provides a large contacting area between the stamp and chip, and the strong adhesion for picking up. While printing, a magnetic field is applied to cause the deformation of the adhesive membrane and drives the delamination of the chips from the stamp. Pull-off tests (Fig. 3B) showed that the adhesion strength is highly sensitive to the magnetic pressure (i.e., the magnetic force over the area of adhesive membrane) and the retraction speed of the stamp. The adhesion is continuously tunable and can be switched into a specific value by adjusting the magnetic pressure and the

Fig. 3 Magnet-controlled transfer printing. (A) Schematic illustration of the typical process of the magnet-controlled transfer printing. (B) Measured adhesion strength between the stamp and the flat glass substrate under different magnetic pressures as a function of retraction speed. (C) Pattern printing by the scanning of a localized magnetic field. Reproduced with permission from Linghu, C., Wang, C., Cen, N., Wu, J., Lai, Z., Song, J., 2019a. Rapidly tunable and highly reversible bio-inspired dry adhesion for transfer printing in air and a vacuum. Soft Matter 15, 30–37.

Mass transfer and transfer printing

263

retraction speed. Under certain conditions, the adhesion fore between the chips and the stamp is below zero. This is useful because it enables non-contact printing. The switchability within experimental measurement is as high as 104X. The adhesion can be switched rapidly thanks to the high response speed of the magnetic materials to the magnetic field. In their work, Linghu et al. (2019) demonstrated the super capabilities of the magnet-controlled transfer printing by patterned printing of the silicon chips onto the PDMS using a core-shaped electromagnet, as shown in Fig. 3C.

4.2 Laser-driven programmable non-contact transfer printing Non-contact transfer printing is superior to contact mode in that it can eliminate the influence of the receiver on the transfer yield and allow non-contact printing of inks onto arbitrary receivers. As mentioned in the former section, magnet-controlled transfer printing can realize non-contact mode printing under certain scenarios. However, the minimal chip size and the resolution of the selectivity of the magnet-controlled transfer printing is restricted by the localized magnetic field, which is hard to be focused. Besides, the residual adhesion of a flat adhesive membrane of the magnet-controlled stamp is still relatively strong. This constrains the applicability of the non-contact printing of magnet-controlled transfer printing to larger and thicker chips. Another attempt of non-contact transfer printing is driven by a laser beam. It controls the interfacial adhesion via interfacial thermal mismatch due to the laser heating of inks to facilitate the delamination of inks from the stamp (Gao et al., 2017; Saeidpourazar et al., 2012). However, reaching the sufficient thermal mismatch for the ink release requiring a high temperature of near 300 °C (Saeidpourazar et al., 2012), which may cause permanent interfacial damage. To address these challenges, Luo et al. (2020) developed a laser-driven programmable non-contact transfer printing technique via a simple yet robust elastomeric active micro-structured stamp. The stamp has circular cavities filled with air and encapsulated by micro-structured PDMS membranes duplicated from low-cost and easily available sandpapers, which can provide tunable adhesion by heating the air in cavities. There is a metal layer on the cavity’s inner surface, which serves as an absorbing layer of the laser power and protects the ink/stamp interface from thermal damage. Compared to the Magnet-Controlled Transfer Printing Technique, this method can further improve the resolution and applicability, which makes reliable programmable non-contact transfer printing possible.

264

Changhong Linghu et al.

Fig. 4A schematically illustrates the typical process of this transfer printing technique. First of all, inks are picked up from the donor substrate through the adhesion of the stamp, then the stamp carrying arrays of inks is aligned to the receiver substrate, leaving a tiny gap (50–150 μm) between them. At last, the micro-patterned membrane deforms dynamically by heating the designative cavities through the programmable laser beam. The energy release rate of the ink/membrane interface grows with the increase of the membrane deflection, which makes non-contact printing possible. When the energy release rate reaches the critical value, the delamination occurs and inks will be printed onto the receiver substrates. Fig. 4B shows the measured pull-off force under various temperature increases. Below the temperature increase of 40 °C, the pull-off force decreases almost linearly and dramatically. While the temperature increase is above 40 °C, the pull-off force decreases slowly due to the negligible difference of contact area at a high temperature increase. The adhesion switchability, which is defined by strong adhesion divided by weak adhesion, can reach 1006  at a 100 °C temperature increase. Fig. 4C shows the microscopy images of the micro-structured stamp surface. The cavity has the dimension of 160 μm in diameter and 350 μm in depth. The spacing between the centers of neighboring cavities is 240 μm. The stamp membrane (25 μm thick) is duplicated from a cheap sandpaper (2000 mesh). These spiny microstructures promote the growth of interface cracks when the micro-structured membrane is deformed, so that inks can be printed onto any challenging substrates. For example, Fig. 4D demonstrates the printing of a Si platelet onto the PDMS substrate with pyramid surface microstructures. Micro-scale Si platelets (350 μm by 350 μm by 3 μm) and LED chips (400 μm by 200 μm by 90 μm) are taken as inks to demonstrate the programmable non-contact printing capability of the active elastomeric microstructured stamp. Fig. 5A shows the selective printing of a 3  3 array Si platelets. At first, the Si platelets array is picked up from the SOI substrate to the stamp membrane, then the platelets surrounded by the dot line are heated by the laser beam. The delamination occurs in 100 ms and the cross patterned Si platelets are printed on PDMS, while the Si platelets without laser heating still remain on the stamp, and there is no damage to the stamp in the laser heating process. Fig. 5B demonstrates the patterned printing of Si platelets by laser scanning. The Si platelets are printed on PDMS to form a “ZJU” pattern and on AirPods to form a letter “F.” Fig. 5C shows the printing of a single LED chip onto the challenging receiver like PDMS pillar arrays and a steel sphere with the diameter of 1 mm. Fig. 5D demonstrates

Fig. 4 Laser-driven programmable non-contact transfer printing. (A) Schematic illustration of the laser-driven programmable non-contact transfer printing process via an active elastomeric micro-structured stamp. (B) The pull-off force as the function of the temperature increase. (C) Microscopy images of the micro-structured stamp surface. (D) SEM image of a Si platelet printed onto PDMS with pyramid microstructures. Reproduced with permission from Luo, H., Wang, C., Linghu, C., Yu, K., Wang, C., Song, J., 2020. Laser-driven programmable non-contact transfer printing of objects onto arbitrary receivers via an active elastomeric microstructured stamp. Natl. Sci. Rev. 7, 296–304.

Fig. 5 Demonstrations of the laser-driven programmable non-contact transfer printing. (A) Selective printing of five Si platelets (350  350  3 μm) from a 3  3 array by laser heating. (i) Si platelets on the micro-structured stamp after the pick up with those surrounded by the dot line to be heated by the laser beam. (ii) The printed cross pattern on PDMS. (iii) The remaining Si platelets on the stamp. (B) Patterned printing by laser scanning. (i) The printed “ZJU” pattern on PDMS. (ii) Si plates printed onto AirPods to form a letter “F.” (C) Printing a single LED chip (400  200  90 μm) onto challenging receivers like (i) PDMS pillar arrays and (ii) a steel sphere with the diameter of 1 mm. (D) Non-contact printing of the LED chips (400  200  90 μm) onto the concave surface on a mobile phone shell to form a letter “P.” (E) Electrical properties of LED before and after transfer printing onto a notebook. Reproduced with permission from Luo, H., Wang, C., Linghu, C., Yu, K., Wang, C., Song, J., 2020. Laser-driven programmable non-contact transfer printing of objects onto arbitrary receivers via an active elastomeric microstructured stamp. Natl. Sci. Rev. 7, 296–304.

Mass transfer and transfer printing

267

the non-contact printing of the LED chips onto the concave surface on a mobile phone shell to form a letter “P,” which is impossible to realize via contact transfer printing techniques. The electrical properties of LED chips are measured before and after transfer printing onto a notebook. Fig. 5E shows that this laser-driven programmable non-contact transfer printing technique does no damage to electronical inks. In summary, this work proposed a new laser-driven transfer printing technique, whose adhesion switchability can reach three orders of magnitude with a temperature increase below 100 °C. It has provided a new protocol of programmable non-contact transfer printing with greatly improved reliability and efficiency.

4.3 SMP-based universal transfer printing Normally, dry adhesion based on van der Waals interactions is used for the picking-up of the chips. However, the utilization of dry adhesion depending on the area of contact is only suitable for flat objects. The weak grip forces resulting from small contact areas between the stamp and the non-planar geometries of the LED chips will lead to the failure of the picking-up process. Besides, the residual adhesion forces between the stamp and the LED chips at smaller scale would be stronger compared to their gravities, which will hinder the printing of the chips. This would be an intrinsic limitation with the transfer printing techniques using dry adhesion. To solve this problem, Linghu et al. (2020) proposed a universal gripping strategy as illustrated in Fig. 6. A simple shape memory polymer (SMP) block as shown in Fig. 6A is used as the stamp. As a kind of emerging smart material, SMP can memorize their temporary shapes and fully recover to the original shapes upon external simulations such as heat, light, electric current, magnetic fields, radiation waves and moisture (Zhang et al., 2019; Zheng et al., 2015). As Fig. 6B shows, there is a sharply drop off of elastic modulus (3 GPa to 2 MPa) of SMP after outer stimulus is applied, which enables the intimate contact with complex 3D shaped target objects. Keeping the load and removing the outer stimulus (Fig. 6C), SMP will recover to stiff state and retain the shape to lock the 3D objects, which offers a very large grip force (Fig. 6D). One big advantage of the large grip force is that the SMP stamp can easily pick up the objects which have strong adhesion with donor substrate such as Micro-LED arrays on blue tapes. This enables the compatibility of this transfer printing technique with established fabrication technologies

Fig. 6 Schematic illustration of the gripping and release process of the SMP-based universal transfer printing. (A–D) Picking-up process. (E–H) Printing process. Reproduced with permission from Linghu, C., Zhang, S., Wang, C., Yu, K., Li, C., Zeng, Y., Zhu, H., Jin, X., You, Z., Song, J., 2020. Universal smp gripper with massive and selective capabilities for multiscaled, arbitrarily shaped objects. Sci. Adv. 6, eaay5120.

Mass transfer and transfer printing

269

of the LED chips. Meanwhile, the SMP stamp can also pick up 3D shaped objects like pyramids, cylinders and spheres. This characteristic may bring new possibilities for the assembly and the development of displays using unconventional LED devices in the forms of wires (Dai et al., 2015; Wang et al., 2014), disks (Chung et al., 2016), pyramids (Sun et al., 2013) and so on. As Fig. 6D–G shows the SMP deformed by embedded objects can recover its original shape again by applying outer stimulus, which enables the release of target objects (Fig. 6H). To characterize the SMP stamp, a SMP block (5 mm in thickness, 100 mm in diameter) and a steel sphere (5 mm in diameter) are used to obtain the grip force at the locked state and the residual force at the released state under different grip speeds (Fig. 7A). For the measurement of grip force, the steel sphere is embedded into the SMP block at a depth of 3 mm under 120 °C. Then the steel sphere is pulled out at a fixed speed after the SMP block is cooled down to 30 °C. To measure the residual force, the steel sphere is placed onto the SMP block at 120 °C, and after the SMP is cooled down to 30 °C the steel sphere is pulled out at a fixed speed. The results show that the grip force is much larger than the residual force. Fig. 7B shows transfer printing process of a 10 10 array of 300 μm steel spheres. The steel spheres are prepared on the Kapton tape and then transfer printed onto the PDMS substrate using a SMP stamp. To demonstrate the manipulation of arbitrarily shaped micro-scaled objects, the iron particles of 75 μm are moved sequentially with the arrangement transformed from the original bigger rectangle into a smaller rhombus as shown in Fig. 7C. Simultaneously, a single and a cluster of SiO2 spheres (10 μm in diameter) are transfer printed, respectively, as shown in Fig. 7D. As mentioned above, the SMP stamp can easily pick up the Micro-LEDs (including irregular shaped Micro-LEDs) from blue tape due to the strong grip force, which is much more difficult for stamps based on dry adhesion. However, there is a large contact area between the released flat Micro-LEDs and SMP stamp after shape recovery. The resulting large residual force will hinder the printing process. Since the SMP stamp does not rely on the adhesion but the embedding to grip the objects, surface treatment or increasing the roughness of the SMP stamp surface are efficient ways to reduce the residual force. As shown in Fig. 8A, two kinds of surface morphologies are fabricated by molding the silicon wafer and 2000-mesh sand paper templates, respectively. The surface microstructures duplicated from the 2000-mesh sand paper greatly reduce the adhesion strengths as shown in Fig. 8A (III), which can ensure a 100% printing yield. It’s worth noting that

Fig. 7 See figure legend on opposite page.

271

Mass transfer and transfer printing

A

(III)

-15 100

200

300

400

500

0

-15 0

100

200

300

400

F grip

0.4

60 oC

140 120 100

25 oC 0.3 0.2 0.1

0.8 0.4 0

0

500

Lateral displacement (um)

Lateral displacement (um)

B



200

(III)

(II) Amplitude (um)

(I)

150 200 um

Released

100 200 um

Embedded

50

0

200 um



0

Amplitude (um)

Amplitude (um)

0

15

F adhesion

160

Grip Force (N)

100 um

100 um 15

0.5

180

(I I )

Adhesion Strength (kPa)

(I)

0

200

400

600

800

1000 1200

200 um

Lateral displacement (um)

C (I)

(III)

(II)

Current (A)

0.002

ON Blue Tape After Transfer

0.001

0.000

1 mm

5 mm

500 um

1.0

1.5

2.0

Voltage (V)

2.5

Fig. 8 Programmable transfer printing of Micro-LED chips. (A) Surface SEM images, roughness and adhesion strengths of the smooth and the micro-structured SMP universal gripper. (B) Selective printing of the Micro-LED chips by a local laser heating. (C) Programmable printing of the Micro-LED chips with a “ZJU” pattern onto a PDMS substrate with no apparent variation in performance of the LED chips before and after transfer printing. Reproduced with permission from Linghu, C., Zhang, S., Wang, C., Yu, K., Li, C., Zeng, Y., Zhu, H., Jin, X., You, Z., Song, J., 2020. Universal smp gripper with massive and selective capabilities for multiscaled, arbitrarily shaped objects. Sci. Adv. 6, eaay5120.

Fig. 7 Characterization and demonstrations of the SMP-based universal transfer printing. (A) Grip and residual forces of a SMP block exerted on a steel sphere (diameter: 5 mm). (B) Transfer printing process of a 10  10 array of 300 μm steel spheres. (C) Manipulation of irregular shaped micro-scaled iron particles (scale bar: 100 μm). (D) Manipulation of single and a cluster of SiO2 spheres (diameter: 10 μm). Reproduced with permission from Linghu, C., Zhang, S., Wang, C., Yu, K., Li, C., Zeng, Y., Zhu, H., Jin, X., You, Z., Song, J., 2020. Universal smp gripper with massive and selective capabilities for multiscaled, arbitrarily shaped objects. Sci. Adv. 6, eaay5120.

272

Changhong Linghu et al.

the grip forces of these two kinds of SMP surface remain in the same order of magnitude. Duplicating the microstructures of the 2000-mesh sand paper, a SMP stamp with large grip force and weak residual adhesion is fabricated for reliable transfer printing of Micro-LED arrays. By using laser heating, programmed patterns of Micro-LEDs are selectively released and printed onto the target substrate (Fig. 8B and C). And there is no change in performance of the Micro-LED chips before and after transfer printing, as shown by the current–voltage curve shown in Fig. 8C-III. In summary, this work proposed a SMP-based universal transfer printing strategy for arbitrarily shaped objects. The strong grip force of SMP stamp is an obvious advantage for manipulating non-planar structures or target objects with strong adhesion at the objects/donor substrate interface.

4.4 Laser-driven active micro-structured transfer printing Handling of ultrathin chips with the thickness down to 20 μm (Liu et al., 2014) is a great challenge due to the fragility of rigid chips and compliance of elastic chips. By utilizing the shape-conformal stamp with actively actuated surface microstructures, Wang et al. (2020) proposed a cost-effective and high-reliability transfer printing technique that has great advantages in selective and large-area assembly of ultrathin Micro-LED chips. Fig. 9A schematically illustrates the working mechanism of the stamp to tune its adhesion strength by increasing the surface roughness with expandable microspheres. Initially, the embedded microspheres are small enough to ensure a negligible effect on adhesion strength of the strong adhesive layer. Upon external stimuli, the microspheres will expand to form surface microstructures and yield weak adhesion. The corresponding scanning electron microscopy (SEM) images and measured surface roughness of the thermal release tape (TRT) stamp before and after heating at 90 °C, respectively, are shown in Fig. 9B and C. It’s observed that the actively actuated surface microstructures can significantly increase the surface roughness of the TRT stamp (Fig. 9D) and achieve the large adhesion switchability (Fig. 9E), which is crucial to transfer the ultrathin inks. Fig. 9F shows the SEM images of a microscale inorganic LED chip (285 μm by 285 μm by 4.6 μm) adhered on the TRT stamp before and after heating at 90 °C. Compared to the stamps with predefined surface microstructures (Kim et al., 2010; Meitl et al., 2017; Song et al., 2019), the actively actuated surface microstructures of the TRT

273

Mass transfer and transfer printing

A

Large adhesion switchability induced by expandable microspheres

Adhesives

Expandable microspheres

Exposed to external stimuli Backing layer

Backing layer

Strong adhesion

B

25Ȼ

Weak adhesion

Adhesion layer

E

10

G

0

50μm

F 25Ȼ

30

Energy release rate (J/m2)

Roughness (μm)

12

1 2 3 4 5 6 7 Scanning distance (mm)

μ-LED

8

90Ȼ

20

8 6

10

4 2 0

40 0

50μm

D

90Ȼ

80

Backing layer

Backing layer

25Ȼ

120

Height (μm)

90Ȼ

C

c

Adhesion layer

30 40 50 60 70 80 90 100

Heating Temperature (Ȼ)

0

Programmable transfer printing prototype Microscope

Optical Fiber

100 μm

30 40 50 60 70 80 90 100

Heating Temperature (Ȼ)

Laser system

Printed inks on receiver sub.

100 μm

H Printed PDs pattern

Dichroic Mirror

Inks on TRT

Receiver substrate

PDMS

TRT stamp on automated translational stage

n+ doping

2 mm Inks:

unheated;

heated

100 μm

Fig. 9 Laser-driven active micro-structured transfer printing of ultrathin Micro-LED chips. (A) Schematic illustration of the concept design to construct a shapeconformal micro-structured stamp with large adhesion switchability. (B) SEM images and (C) corresponding profile of TRT stamp before and after heating on a hotplate. (D) The measured surface roughness of the TRT stamp and (E) energy release rate of the TRT stamp with the glass slide after being uniformly heated on a hotplate at various temperatures. (F) SEM images of the ultrathin, inorganic μ-LED (285 μm by 285 μm by 4.6 μm) on the TRT stamp before and after heating on a hotplate, respectively. (G) Schematic prototype of the laser-assisted programmable transfer printing system via automated translational stage. (H) Programmable printed Si nanomembrane-based photodetectors with a robot-like pattern on PDMS substrate. Reproduced with permission from Wang, C., Linghu, C., Nie, S., Li, C., Lei, Q., Tao, X., Zeng, Y., Du, Y., Zhang, S., Yu, K., Jin, H., Chen, W., Song, J., 2020. Programmable and scalable transfer printing with high reliability and efficiency for flexible inorganic electronics. Sci. Adv. 6, eabb2393.

274

Changhong Linghu et al.

stamp can ensure large contact area and strong adhesion with the ultrathinLED chip during picking-up and then can dynamically drive the TRT stamp into the weak adhesion state for reliable release upon heating. By combining the merits of laser heating system, the TRT stamp can be easily used for programmable transfer printing. Fig. 9G schematically illustrates the basic apparatus to construct the laser-assisted programmable transfer printing prototype, which consists of a programmable laser system for rapid and highly localized heat delivery, the automated translational stage for the precise position and alignment and an optical microscopy for in situ monitoring. Fig. 9H shows the selectively printed Si photodetectors with a robot-like pattern on the PDMS receiver substrate out of an array of 39  39. The schematic insets show the programmable heating pattern of the inks (upper), and the detail of the Si photodetector including two back-to-back n-p dopant regions (lower), respectively. With the aid of the laser-assisted programmable transfer printing prototype, the selective transfer printing of ultrathin inorganic Micro-LEDs can be easily achieved with high efficiency. Fig. 10A shows the optical image of the wafer-scale ultrathin Micro-LED chips on the TRT stamp harvested from a 4-in. sapphire wafer by the standard laser lift-off process. With the selective transfer printing pattern in Fig. 10B, a 10  10 array of MicroLEDs in a sparse form was selectively printed onto a temporary PDMS receiver substrate via programmable transfer printing prototype and then transfer printed onto the polyimide (PI)-coated glass substrate for subsequent device fabrication. Fig. 10C and D shows the optical images of the transfer-printed Micro-LED array on PDMS substrate and PI coated glass substrate, respectively. The optoelectronic performance of the MicroLED chip during the transfer printing process was measured, as shown in Fig. 10E. The consistency of the current–voltage curves indicates the laser-assisted programmable transfer printing prototype here has negligible effects on the performance of the transferred Micro-LED chips. By assembling the transfer-printed Micro-LEDs into individually addressable array, the ultrathin Micro-LED based flexible display can be created. The ultrathin flexible display consists of the PI supporting layer, PI insulating layers, PI encapsulating layer and the interconnects (Fig. 10F) and covers a total in-plane size of 2.19 cm by 2.79 cm and thickness of around 15 μm (Fig. 10G). The flexible device can be wrapped around a glass tube (diameter, 20 mm) with displayed patterns, i.e., “Z,” “J,” and “U” (Fig. 10H) or adhered onto the skin with the medical tape for applications in wearable displays and phototherapy of psoriasis (Fig. 10I) (Zhang and Wu, 2018).

Mass transfer and transfer printing

275

Fig. 10 Transfer-printed ultrathin-LED array for flexible display and healthcare. (A) Optical image of transferred-LEDs on the TRT stamp from the fabricated 4-in. sapphire wafer by the standard laser lift-off process. (B) Schematic illustration of selectively transfer printing pattern from dense form on the TRT stamp into sparse array for usage. The red squares indicate the μ-LEDs to be transferred from the TRT stamp. (C) Optical image of selectively transfer-printed μ-LEDs in a 10  10 array on the PDMS temporary receiver substrate. (D) Optical image of μ-LEDs transfer printed onto the polyimide substrate from the PDMS temporary receiver substrate using the TRT stamp. (E) I–V curve of the μ-LED on the fabricated sapphire wafer and receiving polyimide substrate, respectively. (F and G) Schematic and optical images of the fabricated inorganic μ-LED array-based flexible display. (H) Optical image of μ-LED array-based flexible display wrapped around a glass tube with displayed letters of ZJU. (I) Flexible display attached on the medical tape and mounted onto skin for phototherapy. Reproduced with permission from Wang, C., Linghu, C., Nie, S., Li, C., Lei, Q., Tao, X., Zeng, Y., Du, Y., Zhang, S., Yu, K., Jin, H., Chen, W., Song, J., 2020. Programmable and scalable transfer printing with high reliability and efficiency for flexible inorganic electronics. Sci. Adv. 6, eabb2393.

276

Changhong Linghu et al.

In summary, this work provided a promising solution to the programmable transfer printing technique with high reliability and efficiency, which is well suitable for the large-area assembly of ultrathin Micro-LEDs. The flat adhesive provides strong adhesion for reliable retrieval and the actively actuated surface microstructures ensure weak adhesion for successful release of inks and surpass previous transfer printing techniques with the predefined surface microstructures.

5. Conclusion In this chapter, we reviewed the mass transfer techniques for Micro-LED displays. The large-area, high-throughput assembly of MicroLED dies and driver chips (i.e., mass transfer) has long been a challenge of the Micro-LED display. This bottleneck has delayed the commercialization process of Micro-LED displays. With the immerse efforts from the research community and the industry, many mass transfer techniques have shown great promise. Among them are conventional pick-and-place method, fluid transfer, electrostatic and magnetic transfer, laser transfer and transfer printing. Pick-and-place currently is a simple solution commercially used due to its compatibility with the established fabrication and sorting process of LED chips and its perfect industrial chain. But it is expected to phase out in the near future for its high cost. Liquid transfer also has a long way to go before its yield can be improved to an acceptable level. Electrostatic transfer, magnetic transfer and laser transfer are marching rapidly toward commercialization. On the contrary, transfer printing techniques own unique advantages in all aspects and are developing quickly. A review of the latest development of the transfer printing techniques is also presented here. The magnet-controlled transfer printing provided a good example with rapid tunability and high switchability. The laser-driven programmable noncontact transfer printing improved the resolution of selectivity using a laser heating. Beside this, the micro-structured surfaces further reduced the residual adhesion, which enables reliable non-contact printing. The SMP-based universal transfer printing and the laser-driven active micro-structured transfer printing addressed the challenges for the assembly of chips with non-conventional geometries and large-area integration of ultrathin-LED chips, respectively.

Mass transfer and transfer printing

277

References Ahmed, N., Dagdeviren, C., Rogers, J.A., Ferreira, P.M., 2015. Active polymeric composite membranes for localized actuation and sensing in microtransfer printing. J. Microelectromech. Syst. 24, 1016–1028. Ahn, J.H., Meitl, M.A., Baca, A.J., Khang, D.Y., Kim, H.S., Rogers, J.A., 2008. Transfer printing techniques and inorganic single-crystalline materials for flexible and stretchable electronics. In: Mitzi, D.B. (Ed.), Solution Processing of Inorganic Materials. John Wiley & Sons, Inc, pp. 407–447. Ashdown, I., Speier, I., 2015. Method and eletrostatic transfer stamp for transferring semiconductor dice using electrostatic transfer printing techniques. United States patent application 9209059. Ashdown, I., Speier, I., Sheen, C.W., Schick, P.M., 2011. Method and magnetic transfer stamp for transferring semiconductor dice using magnetic transfer printing techniques. United States patent application 20110151588. Bibl, A., Higginson, J.A., Law, H.-F.S., Hu, H.-H., 2018. Micro device transfer head heater assembly and method of transferring a micro device. United States patent application 10121864. Bower, C., Meitl, M., Bonafede, S., Gomez, D., 2015. Heterogeneous integration of microscale compound semiconductor devices by micro-transfer-printing. In: IEEE 65th Electronic Components and Technology Conference, San Diego, CA, pp. 963–967. Carlson, A.P., Kimlee, H.J., Wu, J., Elvikis, P., Cheng, H., Kovalsky, A., Elgan, S.L., Yu, Q., Ferreira, P.M., Huang, Y., 2011. Shear-enhanced adhesiveless transfer printing for use in deterministic materials assembly. Appl. Phys. Lett. 98, 264104. Carlson, A.P., Wang, S., Elvikis, P., Ferreira, P.M., Huang, Y., Rogers, J.A., 2012. Active, programmable elastomeric surfaces with tunable adhesion for deterministic assembly by transfer printing. Adv. Funct. Mater. 22, 4476–4484. Chaji, G., Fathi, E., 2019. Integration of micro-devices into system substrate. United States patent application 10468472. Cho, S., Kim, N., Song, K., Lee, J., 2016. Adhesiveless transfer printing of ultrathin microscale semiconductor materials by controlling the bending radius of an elastomeric stamp. Langmuir 32, 7951–7957. Choi, M.K., Yang, J., Hyeon, T., Kim, D.-H., 2018. Flexible quantum dot light-emitting diodes for next-generation displays. npj Flexible Electron. 2, 10. Chow, E.M., Lu, J.P., Bert, J.A., Crawford, L.S., Raychaudhuri, S., Plochowietz, A., Rupp, B.B., Burton, G., Lujan, R.A., Wang, Q., 2017. Micro-object assembly with an optically addressed array. In: 19th International Conference on Solid-State Sensors, Actuators and Microsystems, Kaohsiung, TAIWAN, pp. 682–685. Chung, K., Yoo, H.D., Hyun, J.K., Oh, H., Tchoe, Y., Lee, K., Baek, H., Kim, M., Yi, G., 2016. Flexible gan light-emitting diodes using gan microdisks epitaxial laterally overgrown on graphene dots. Adv. Mater. 28, 7688–7694. Dai, X., Messanvi, A., Zhang, H., Durand, C., Eymery, J., Bougerol, C., Julien, F.H., Tchernycheva, M., 2015. Flexible light-emitting diodes based on vertical nitride nanowires. Nano Lett. 15, 6958–6964. Delaporte, P., Alloncle, A.P., 2016. Laser-induced forward transfer: a high resolution additive manufacturing technology. Opt. Laser Technol. 78, 33–41. Ding, K., Avrutin, V., Izyumskaya, N., Ozgur, U., Morkoc, H., 2019. Micro-LEDs, a manufacturability perspective. Appl. Sci. 9, 1206. Feng, X., Meitl, M., Bowen, A.M., Huang, Y., Nuzzo, R.G., Rogers, J.A., 2007. Competing fracture in kinetically controlled transfer printing. Langmuir 23, 12555–12560.

278

Changhong Linghu et al.

Gao, Y., Li, Y., Li, R., Song, J., 2017. An accurate thermomechanical model for laser-driven microtransfer printing. J. Appl. Mech. 84, 064501. Golda, D., Bibl, A., 2017. Micro device transfer head array. United States patent application 9548233. Gomez, D., Ghosal, K., Meitl, M., Bonafede, S., Prevatte, C., Moore, T., Raymond, B., Kneeburg, D., Fecioru, A., Trindade, A.J., 2016. Process capability and elastomer stamp lifetime in micro transfer printing. In: 66th IEEE Electronic Components and Technology Conference, Las Vegas, NV, pp. 680–687. Gomez, D., Ghosal, K., Moore, T., Meitl, M., Bonafede, S., Prevatte, C., Radauscher, E., Trindade, A.J., Bower, C.A., 2017. Scalability and yield in elastomer stamp microtransfer-printing. In: IEEE 67th Electronic Components and Technology Conference, Lake Buena Vista, FL, pp. 1779–1785. Henley, F.J., 2018. Materials, process and production equipment considerations to achieve high-yield microled mass-production. Dig. Tech. Pap. Soc. Inf. Disp. Int. Symp. 49, 86–89. Hu, H. H., Bibl, A., Higginson, J.A., Law, H.-F.S., 2013. Method of fabricating and transferring a micro device and an array of micro devices utilizing an intermediate electrically conductive bonding layer. United States patent application 8518204. Huang, Y., Hsiang, E.-L., Deng, M.-Y., Wu, S.-T., 2020. Mini-LED, micro-LED and OLED displays: present status and future perspectives. Light Sci. Appl. 9, 105. Jiang, H.X., Jin, S.X., Li, J., Shakya, J., Lin, J.Y., 2001. III-nitride blue microdisplays. Appl. Phys. Lett. 78, 1303–1305. Kahler, J., Heuck, N., Stranz, A., Waag, A., Peiner, E., 2012. Pick-and-place silver sintering die attach of small-area chips. IEEE Trans. Compon. Packag. Manuf. Technol. 2, 199–207. Kim, S., Wu, J., Carlson, A.P., Jin, S.H., Kovalsky, A., Glass, P., Liu, Z., Ahmed, N., Elgan, S.L., Chen, W., 2010. Microstructured elastomeric surfaces with reversible adhesion and examples of their use in deterministic assembly by transfer printing. Proc. Natl. Acad. Sci. U. S. A. 107, 17095–17100. Kim, T., Cho, K.S., Lee, E.K., Lee, S., Chae, J., Kim, J.W., Kim, D.H., Kwon, J., Amaratunga, G.A.J., Lee, S.Y., 2011. Full-colour quantum dot displays fabricated by transfer printing. Nat. Photonics 5, 176–182. Kim, B.H., Nam, S., Oh, N., Cho, S.Y., Yu, K.J., Lee, C.H., Zhang, J., Deshpande, K., Trefonas, P., Kim, J.H., 2016. Multilayer transfer printing for pixelated, multicolor quantum dot light-emitting diodes. ACS Nano 10, 4920. Kimlee, H.J., Carlson, A.P., Grierson, D.S., Rogers, J.A., Turner, K.T., 2014. Interface mechanics of adhesiveless microtransfer printing processes. J. Appl. Phys. 115, 143513. Lee, V., Twu, N., Kymissis, I., 2016. Micro-LED technologies and applications. Inf. Disp. 32, 16–23. Lee, D., Cherekdjian, S., Kang, S., Mishra, K., Ong, P., Xu, X., 2019. Ultra-fine high efficiency microleds with testability and transferability using layer-transfer technology. Dig. Tech. Pap. Soc. Inf. Disp. Int. Symp. 50, 236–239. Linghu, C., Zhang, S., Wang, C., Song, J., 2018. Transfer printing techniques for flexible and stretchable inorganic electronics. npj Flexible Electron. 2, 26. Linghu, C., Wang, C., Cen, N., Wu, J., Lai, Z., Song, J., 2019a. Rapidly tunable and highly reversible bio-inspired dry adhesion for transfer printing in air and a vacuum. Soft Matter 15, 30–37. Linghu, C., Zhu, H., Zhu, J., Li, C., Song, J., 2019b. Mechanics of magnet-controlled transfer printing. Extreme Mech. Lett. 27, 76–82. Linghu, C., Zhang, S., Wang, C., Yu, K., Li, C., Zeng, Y., Zhu, H., Jin, X., You, Z., Song, J., 2020. Universal smp gripper with massive and selective capabilities for multiscaled, arbitrarily shaped objects. Sci. Adv. 6, eaay5120.

Mass transfer and transfer printing

279

Liu, Z., Huang, Y., Chen, J., Yin, Z., 2014. Tunable peeling technique and mechanism of thin chip from compliant adhesive tapes. IEEE Trans. Compon. Packag. Manuf. Technol. 4, 560–568. Liu, H., Liu, Z., Xu, Z., Yin, Z., Huang, Y., Chen, J., 2015. Competing fracture of thin-chip transferring from/onto prestrained compliant substrate. J. Appl. Mech. 82, 101012. Liu, Y., Wei, F., Lu, T., Liu, Z., 2018. GAN HEMT-LED homogeneous integration for micro-led mass transferring. Dig. Tech. Pap. Soc. Inf. Disp. Int. Symp. 49, 660–664. Luo, H., Wang, C., Linghu, C., Yu, K., Wang, C., Song, J., 2020. Laser-driven programmable non-contact transfer printing of objects onto arbitrary receivers via an active elastomeric microstructured stamp. Natl. Sci. Rev. 7, 296–304. Marinov, V.R., 2018. Laser-enabled extremely-high rate technology for μled assembly. Dig. Tech. Pap. Soc. Inf. Disp. Int. Symp. 49, 692–695. Marinov, V.R., Swenson, O.F., Miller, R., Sarwar, F., Atanasov, Y., Semler, M.R., Datta, S., 2012. Laser-enabled advanced packaging of ultrathin bare dice in flexible substrates. IEEE Trans. Compon. Packag. Manuf. Technol. 2, 569–577. Meitl, M.A., Zhu, Z.-T., Kumar, V., Lee, K.J., Feng, X., Huang, Y.Y., Adesida, I., Nuzzo, R.G., Rogers, J.A., 2006. Transfer printing by kinetic control of adhesion to an elastomeric stamp. Nat. Mater. 5, 33–38. Meitl, M.A., Radauscher, E., Rotzoll, R., Raymond, B., Bonafede, S., Gomez, D., Moore, T., Prevatte, C., Fecioru, A., Trindade, A.J., Bower, C.A., 2017. Emissive displays with transfer-printed microscale inorganic LEDs. Dig. Tech. Pap. Soc. Inf. Disp. Int. Symp. 48, 257–263. Paranjpe, A., Montgomery, J., Lee, S.M., Morath, C., 2018. Micro-LED displays: key manufacturing challenges and solutions. Dig. Tech. Pap. Soc. Inf. Disp. Int. Symp. 49, 597–600. Peng, B., Huang, Y., Yin, Z., Xiong, Y., 2012. Competing fracture modeling of thin chip pick-up process. IEEE Trans. Compon. Packag. Manuf. Technol. 2, 1217–1225. Plochowietz, A., Wang, Y., Shreve, M., Crawford, L.S., Raychaudhuri, S., Butylkov, S., Rupp, B.B., Wang, Q., Wang, Y., Kalb, J., 2019. Programmable micro-object assembly with transfer. In: 20th International Conference on Solid-State Sensors, Actuators and Microsystems and Eurosensors, Berlin, Germany, pp. 390–393. Rupp, B.B., Plochowietz, A., Crawford, L.S., Shreve, M., Raychaudhuri, S., Butylkov, S., Wang, Y., Mei, P., Wang, Q., Kalb, J., 2019. Chiplet micro-assembly printer. In: 2019 IEEE 69th Electronic Components and Technology Conference, Las Vegas, NV, pp. 1312–1315. Saeidpourazar, R., Li, R., Li, Y., Sangid, M.D., Lu, C., Huang, Y., Rogers, J.A., Ferreira, P.M., 2012. Laser-driven micro transfer placement of prefabricated microstructures. J. Microelectromech. Syst. 21, 1049–1058. Sasaki, K., Schuele, P.J., Ulmer, K., Lee, J.-J., 2019. System and method for the fluidic assembly of emissive displays. United States patent application 10418527. Schuele, P.J., Sasaki, K., Ulmer, K., Lee, J.-J., 2017. Display with surface mount emissive elements. United States patent application 20170133550. Song, E., Chiang, C., Li, R., Jin, X., Zhao, J., Hill, M., Xia, Y., Li, L., Huang, Y., Won, S.M., 2019. Flexible electronic/optoelectronic microsystems with scalable designs for chronic biointegration. Proc. Natl. Acad. Sci. U. S. A. 116, 15398–15406. Stauth, S., Parviz, B.A., 2006. Self-assembled single-crystal silicon circuits on plastic. Proc. Natl. Acad. Sci. U. S. A. 103, 13922–13927. Sun, B., Zhao, L., Wei, T., Yi, X., Liu, Z., Wang, G., Li, J., 2013. Shape designing for light extraction enhancement bulk-gan light-emitting diodes. J. Appl. Phys. 113, 243104. Takano, S., Takahashi, K., 2001. A high density full color LED display panel on a silicon microreflector. IEEJ Trans. Sens. Micromach. 121, 464–468.

280

Changhong Linghu et al.

Tong, C., Liu, X., Li, W., Liu, Z., 2019. Investigation of full-color solutions for micro-led display. Dig. Tech. Pap. Soc. Inf. Disp. Int. Symp. 50, 771–774. Virey, E.H., Baron, N., 2018. Status and prospects of microled displays. Dig. Tech. Pap. Soc. Inf. Disp. Int. Symp. 49, 593–596. Wang, R., Nguyen, H.P.T., Connie, A.T., Lee, J.G., Shih, I., Mi, Z., 2014. Color-tunable, phosphor-free ingan nanowire light-emitting diode arrays monolithically integrated on silicon. Opt. Express 22, A1768–A1775. Wang, C., Linghu, C., Nie, S., Li, C., Lei, Q., Tao, X., Zeng, Y., Du, Y., Zhang, S., Yu, K., Jin, H., Chen, W., Song, J., 2020. Programmable and scalable transfer printing with high reliability and efficiency for flexible inorganic electronics. Sci. Adv. 6, eabb2393. Wu, M.-H., Fang, Y.-H., Chao, C.-H., 2017. Electric-programmable magnetic module and picking-up and placement process for electronic devices. United States patent application 9607907. Wu, T., Sher, C., Lin, Y., Lee, C., Liang, S., Lu, Y., Chen, S.H., Guo, W., Kuo, H., Chen, Z., 2018. Mini-LED and micro-led: promising candidates for the next generation display technology. Appl. Sci. 8, 1557. Wu, M.-H., Fang, Y.-H., Chao, C.-H., 2020. Magnetic transfer module and method for transferring electronic element. United States patent application 10566125. Yeh, H.J., Smith, J.S., 1994. Fluidic self-assembly for the integration of gaas light-emitting diodes on si substrates. IEEE Photonics Technol. Lett. 6, 706–708. Yoo, B., Cho, S., Seo, S., Lee, J., 2014. Elastomeric angled microflaps with reversible adhesion for transfer-printing semiconductor membranes onto dry surfaces. ACS Appl. Mater. Interfaces 6, 19247–19253. Yoon, J., Lee, S.M., Kang, D., Meitl, M.A., Bower, C.A., Rogers, J.A., 2015. Heterogeneously integrated optoelectronic devices enabled by micro-transfer printing. Adv. Opt. Mater. 3, 1313–1335. Zhang, P., Wu, M.X., 2018. A clinical review of phototherapy for psoriasis. Lasers Med. Sci. 33, 173–180. Zhang, S., Wang, C., Nie, S., Xiao, J., Song, J., 2019. Wrinkling of silicon nanoribbons on shape memory polymers. J. Phys. D 52, 265101. Zheng, N., Fang, G., Cao, Z., Zhao, Q., Xie, T., 2015. High strain epoxy shape memory polymer. Polym. Chem. 6, 3046–3053. Zhou, X., Tian, P., Sher, C.-W., Wu, J., Liu, H., Liu, R., Kuo, H.-C., 2020. Growth, transfer printing and colour conversion techniques towards full-colour micro-led display. Prog. Quantum. Electron. 71, 100263.

CHAPTER NINE

Micro-LED based optical wireless communications systems P. Tiana,*, Jonathan J.D. McKendryb, J. Herrnsdorfb, S. Zhua, Erdan Gub, Nicolas Laurandb, and Martin D. Dawsonb a

School of Information Science and Technology, Academy of Engineering and Technology, Fudan University, Shanghai, China b Institute of Photonics, Department of Physics, University of Strathclyde, Glasgow, United Kingdom *Corresponding author: e-mail address: [email protected]

Contents 1. Introduction 2. Device-related characteristics 2.1 Micro-LED for visible light communication 2.2 Fast color converters for VLC 3. Micro-LED based high-speed VLC systems 3.1 Advanced modulation schemes 3.2 Integrated micro-LED/CMOS for structured VLC and MIMO VLC 3.3 Summary 4. Novel optical wireless communication systems based on micro-LED 4.1 Underwater wireless optical communication based on micro-LEDs 4.2 Micro-LED array based duplex VLC 4.3 Deep-ultraviolet micro-LED communication 5. Conclusion Acknowledgments References Further reading

281 283 283 287 295 295 299 305 305 305 311 315 316 317 317 321

1. Introduction New forms of optical communication using the visible and ultraviolet regions of the electromagnetic spectrum are now being investigated intensively. These include optical wireless communications (OWC) in various environments, such as in atmosphere, underwater and in space, and guided wave optical communications (GWOC) in such as polymer optical fibers (POF), polymer waveguide backplanes (PWBs) and short-wavelength photonic integrated circuits (SW-PICs). Each of these applications’ scenarios Semiconductors and Semimetals, Volume 106 ISSN 0080-8784 https://doi.org/10.1016/bs.semsem.2021.01.003

Copyright

#

2021 Elsevier Inc. All rights reserved.

281

282

P. Tian et al.

offers new opportunities, capabilities and challenges. Typically, there are restrictive requirements of low size, weight, power and cost (SWaP-C) driven by integration, packaging and systems’ deployment considerations. There must also be effective means to interface the devices to electronic control, appropriate solid-state photodetector/camera technology must be utilized, and the optical and scattering properties of the propagation medium must be carefully taken into account in developing the appropriate optical link scenarios and technologies. Compound semiconductor based electroluminescent devices, including light-emitting diodes (LEDs), super-luminescent diodes (SLDs) and laser diodes (LDs) are essential enabling technologies for these applications, in particular those based on gallium nitride (GaN) and its family of alloys. Of these, GaN LED technology is arguably the most mature, and it now offers almost continuous coverage from the deep ultraviolet through to the amber-red region (with efficiency peaking in the blue-violet), with wavelength specificity and spectrally-narrow output determined by the epitaxy and structure design. An important development, beginning over 10 years ago and now a focus of global research, is the realization of the advantages of a new format of gallium nitride LEDs, micro-scale LEDs or “micro-LEDs” for optical communications applications (Rajbhandari et al., 2017b). Micro-LEDs, devices of diameter/side dimension from a few micrometers to a few 10’s of micrometers, are being developed in the first place for novel forms of electronic visual display technology. A strong driver for micro-LEDs in display applications is the high brightness and contrast enabled by these devices. The high-density array-format capability of micro-LEDs and their ready interfacing to silicon complementary metaloxide-semiconductor (CMOS) electronics, important characteristics for displays, also offer sophisticated spatio-temporal pattern projection performance for spatially modulated/multiplexed communications (Griffiths et al., 2020). This can combine communication functions with lighting, sensing, imaging, tracking and location capability, fostering new technological convergences. These capabilities are augmented by the very rapid (100’s MHz–1 GHz) modulation response capabilities of individual microLEDs and the ability, when required, for their output to be imaged and/or coupled efficiently to low-loss polymer or short-wavelength-transparentinorganic waveguides. This chapter reviews the pertinent performance characteristics and selected deployment scenarios for micro-LED devices as sources for optical

Micro-LED based optical wireless communications systems

283

communications systems. It first covers the modulation bandwidth capabilities of the devices, indicating the key role that high current density can play in facilitating this performance. Various forms of fast-response color converters are then described, because certain applications scenarios can advantageously combine a short-wavelength micro-LED or micro-LED array with fluorescent/luminescent down-conversion. After this, advanced modulation and data encoding schemes are presented, which have permitted data rates of several to
10 Gb/s over multi-meter distances. The review then describes integration with CMOS, before moving on to applications. Due to space constraints a non-exhaustive list of applications areas is covered, which includes digital-to-light conversion, underwater wireless optical communication, duplex visible light communications, and deep-ultraviolet communications.

2. Device-related characteristics 2.1 Micro-LED for visible light communication Modulation bandwidth (BW) is a measure of how quickly a device responds to changes in the input signal over time. The modulation bandwidth of a micro-LED can be specified as the frequency value at which the detected optical power (photocurrent) is half of that at direct current (DC), and this is also known as the electrical-to-optical bandwidth or optical 3 dB bandwidth. Alternatively, the modulation bandwidth can be specified as the frequency at which the received electrical power (proportional to the photocurrent squared) is half of that at DC, this is referred to as the electrical-to-electrical, or electrical 3 dB bandwidth. It should be noted that the 3 dB optical bandwidth is equivalent to an 6 dB electrical bandwidth, due to the square law relationship between photocurrent and electrical power. Measurement of the modulation bandwidth is usually achieved by utilizing a network analyzer as illustrated schematically in Fig. 1. Micro-LEDs are usually tested, unpackaged, on chip with a high-speed probe, and the DC component and small signal AC component generated by the network analyzer are combined via a bias-tee to drive the micro-LED. Optics can be used to collimate and focus the light beam onto a high bandwidth photodetector. The received electrical signal is then fed back to the network analyzer. The AC signal is swept in frequency and the frequency response is then obtained on the network analyzer by comparing the output signal power with the input signal power at different frequencies. From the

284

P. Tian et al.

Fig. 1 Setup to measure the modulation bandwidth of micro-LED.

frequency response measured in this way we may ascertain the electrical bandwidth of the link, which is equal to the device bandwidth if all other link components have a higher bandwidth. The use of GaN-based micro-LEDs for high speed visible light communication (VLC) was initially demonstrated by McKendry et al. (2010) using a 450-nm blue micro-LED with diameter of 72 μm. The micro-LED had a 3 dB optical modulation bandwidth of up to 245 MHz, and an errorfree data transmission rate of 1 Gb/s was achieved, with a received light power of 1 mW. The size and wavelength-dependent characteristics of micro-LEDs for VLC were further studied by McKendry et al. (2011). These micro-LEDs were fabricated from commercially available epitaxial wafers grown by MOCVD. The micro-LEDs had peak emission wavelengths at 370, 405, and 450 nm, respectively, and pixel diameters from 14 to 84 μm in 10 μm increments. As shown in Fig. 2A and B, the microLED array shared a common n-contact, and each micro-LED could be individually addressed by p-contact. The modulation bandwidth was measured as a function of the micro-LED peak emission wavelength and diameter as shown in Fig. 2C. The maximum optical bandwidth of the 450 nm microLED was
235 MHz for a 74 μm pixel, but the bandwidth of a 44 μm pixel can achieve
435 MHz. The 24 and 34 μm pixels were defective which limited the maximum achievable current density, causing a deviation in the trend of the bandwidth vs pixel size. Fig. 2D shows the current density dependent modulation bandwidth of 44, 64 and 84μm diameter micro-LEDs. These devices show almost the same modulation bandwidth under the same injection current density, suggesting that current density is key to understand and increase the modulation bandwidth. The micro-LED array was then

A

B

p-GaN

p-contact

n-contact

p-contact

MQW

Si02

n-GaN

100 mm Sapphire substrate

450

Maximum optical -3dB point (MHz)

84 mm diameter 64 mm diameter 44 mm diameter

D 370 nm 405 nm 450 nm 520 nm

400 350 300 250 200 150 100

450

Modulation bandwidth (MHz)

C

400 350 300 250 200 150 100 50

50 10

20

30

40

50

60

Pixel diameter (mm)

70

80

0 0.00

0.01

0.02

0.03

0.04

0.05

Current density (mA/mm2)

Fig. 2 (A) Microscopic image of a 450 nm 8  8 micro-LED array showing two pixels in operation. (B) Schematic structure of the micro-LEDs. (C) Maximum optical modulation bandwidth of micro-LED with different diameters and wavelengths. (D) Modulation bandwidth as a function of the current density of 450 nm micro-LEDs. From McKendry, J. J., Massoubre, D., Zhang, S., Rae, B. R., Green, R. P., Gu, E., Henderson, R.K., Kelly, A.E., Dawson, M. D., 2011. Visible-light communications using a CMOS-controlled micro-light-emitting-diode array. J. Lightwave Technol. 30, 61–67.

286

P. Tian et al.

bonded to an 8  8 CMOS driver, such that each micro-LED was connected to an individual CMOS driver, and the light emission was extracted from the sapphire side. The 3 dB bandwidth of the micro-LED was partly affected by the CMOS but more than 100 MHz bandwidth was still obtained when driven by CMOS, and data transmission was demonstrated from a single CMOS-controlled micro-LED pixel up to 512 Mb/s. Advanced modulation techniques utilizing CMOS-controlled arrays are described in more detail in Section 3.2. Many research groups have attempted to enhance the modulation bandwidth of micro-LEDs by introducing advanced technology or designing new epitaxial structures. The modulation bandwidth characteristics of high-speed micro-LEDs are summarized in Fig. 3. Mainstream methods to improve LED modulation bandwidth include quantum well structure design, electrode design, epitaxial material improvement and so on. Currently, nonpolar micro-LEDs can achieve the highest bandwidths, with blue micro-LEDs having bandwidths of up to 1.5 GHz (Rashidi et al., 2018), and semipolar green micro-LEDs can achieve a bandwidth

Fig. 3 Bandwidth vs current density for high-speed micro-LEDs.

Micro-LED based optical wireless communications systems

287

of 756 MHz (Huang Chen et al., 2020). It is important to note, however, that data capacity in an optical link depends on both bandwidth and received power and this needs to be considered in comparing semi-polar and nonpolar devices with their polar counterparts. The first report on data transmission capability of AlGaInP LEDs in micro-LED geometries (i.e., without the microcavities required by resonant cavity LEDs) was reported by Carreira et al. (2020), and modulation bandwidths up to 170 MHz were achieved. In addition, the modulation bandwidth of deep-UV micro-LEDs, to be discussed further in Section 4.3, is up to 438 MHz (He et al., 2019), which is 3 times higher than the reported bandwidth of other deep-UV LEDs (Kojima et al., 2018). The high-bandwidth characteristics of micro-LEDs will be of great help in increasing the data rates of communication systems. The modulation bandwidth is a key factor impacting on the channel capacity and transmission data rate of VLC systems. It is dependent on many characteristics of the device itself, including the RC time constant, which is affected by the resistance-capacitance effect (Yang et al., 2014), and the carrier lifetime. Micro-LEDs have a small RC time constant due to their small size. High current densities, over kA/cm2, can be achieved with micro-LEDs which can be attributed to low junction temperature and uniform current spreading in micro-LEDs (Gong et al., 2010). High current densities are the key factor in achieving a high modulation bandwidth due to the resulting decrease in differential carrier lifetime and a correspondingly rapid response to high-frequency modulation signals (Green et al., 2013). At low current densities, semipolar and nonpolar GaN micro-LEDs enable higher modulation bandwidths due to their larger electron-hole wavefunction overlaps and hence shorter carrier lifetimes (Monavarian et al., 2018). In addition, the modulation bandwidth of micro-LED is also affected by the large signal modulation depth due to the carrier sweep-out effect of the electric field within the p-i-n region (Tian et al., 2018).

2.2 Fast color converters for VLC We have said that in some scenarios, short wavelength micro-LEDs can be advantageously combined with luminescent color converters. In this section, color conversion of micro-LEDs in the context of VLC is discussed and an overview of fast color-converting materials is given. 2.2.1 Time dynamics of color conversion GaN-based micro-LEDs are enabling new forms of displays and solid-state lighting instrumentation and, as discussed, are suited to the digital light

288

P. Tian et al.

functionalities needed for VLC. However, they are subject to the so-called yellow/green gap, i.e., a spectral region near a wavelength of 560 nm where both InGaN and AlGaInP-based LEDs have poor efficiency, and LEDs fabricated on a single wafer typically emit at a single wavelength. For these reasons, GaN LEDs are often combined with materials called color converters to emit at different wavelengths in the visible (Leitao et al., 2017a). In such implementations, a UV or blue LED chip photo-pumps the colorconverting material, which absorbs the LED photons and re-emits at longer wavelengths. In turn, they can efficiently extend the emission of the device across the visible spectrum, create all the primary colors needed for display within a single device, or enable white LEDs by color mixing. Indeed, the majority of commercial white LEDs sold today consist of a blue GaN LED combined with a yellow-emitting phosphor. Fig. 4A is a simple schematic showing the concept of color conversion for VLC. An LED is modulated

Fig. 4 (A) Principle of time-dependent color conversion; data encoded through modulation of a source is transferred to another wavelength via a color converter (C. C.). (B) Photograph of a blue-emitting micro-LED with a color converter made of colloidal quantum dots encapsulated in flexible glass in direct contact with the micro-LED sapphire substrate, close up of 3 different colors of colloidal quantum dot converters. (C) Photographs of red, yellow/orange and green colloidal quantum dots (CQD)/PMMA color converters under UV illumination. Panel (B): From Foucher, C., Islim, M.I., Guilhabert, B.J.E., Videv, S., Rajbhandari, S., Gomez Diaz, A., et al., 2018. Flexible glass hybridized colloidal quantum dots for Gb/s visible light communications. IEEE Photonics J. 10, 1–12. Panel (C): From Leitao, M.F., Santos, J.M.M., Guilhabert, B., Watson, S., Kelly, A.E., Islim, M.S., et al., 2017a. Gb/s visible light communications with colloidal quantum dot color converters. IEEE J. Sel. Top. Quantum Electron. 23, 1–10.

Micro-LED based optical wireless communications systems

289

so that the LED light carries a signal (represented by 1 and 0 in the schematic). The light is absorbed and the signal is transferred to a longer wavelength by a color converter (C. C.). Fig. 4B demonstrates in practice a possible single color conversion implementation of a micro-LED intended for VLC (Foucher et al., 2018). Fig. 4C illustrates the images of red, yellow/orange and green colloidal quantum dots (CQD)/PMMA color converters under UV illumination (Leitao et al., 2017a). In this particular case, the color converter is made of colloidal quantum dots as discussed further in Section 2.2.2. Color conversion is based on the process of absorption followed by spontaneous emission, and the time dynamics are therefore set by the rate of radiative transition between the lowest excited state and the ground state of the color-converting material. This rate is proportional to the product of the matrix element for the transition and the density of states (Fox, 2010; Rabouw and Donega, 2016). The former is intrinsic to the type of material and to the nature of the lowest transition (dipole allowed or forbidden) in such a material, whereas the second is dependent on the environment of the emitter (Rabouw and Donega, 2016). For an otherwise equivalent transition strength, the rate is also inversely proportional to the emission wavelength. The overall luminescence (excited state) lifetime of a color converter is intrinsically linked to its capability for modulation of light: the longer the excited state lifetime, the lower the BW. The latter is defined as the modulated frequency at which the color-converted light has half the intensity of the color-converted light at DC. The relation between BW and luminescence lifetime (τ) is given by the following formula (Laurand et al., 2012; Zhang et al., 2018). pffiffiffi 3 BW ¼ 2πτ While the full frequency response of a color converter is important, the BW is often used as a parameter to compare the dynamic response of different color-converting materials: the higher the BW the better for VLC. A color converter for VLC should therefore combine a short luminescence lifetime and a high efficiency (Leitao et al., 2017a; Zhang et al., 2018)—meaning a short radiative decay. In many LED devices and systems for lighting, color-converting materials are typically phosphors, i.e., the transition is dipole-forbidden. The modulation bandwidth for a phosphor is a couple of MHz at best, whereas it is several hundreds of MHz for micro-LEDs (Rajbhandari

290

P. Tian et al.

et al., 2017b), as described in Section 2.1. This means that in a colorconverted micro-LED device using phosphors, the color-converted light cannot follow the modulation speed of the micro-LED leading to distortion and reduced signal to noise ratio when implemented as the source of a VLC link. For this reason, VLC links often simply filter out the slow phosphor component of the signal at the receiver. Hence, other faster color converters are being explored, and in the next section we give an overview of these alternatives to phosphors. 2.2.2 Inorganic semiconductor colloidal quantum dots Colloidal quantum dots (CQDs) are luminescent, nanosized (1 Gb/s data transmission by color mixing and Wavelength Division Multiplexing (WDM). Other reports of micro-LEDs with CQD color converters for VLC have been reported (Leitao et al., 2017b, 2019; Mei et al., 2018). In particular, metal halide perovskite CQDs have the fastest known luminescence dynamics among CQDs (Smith and Nie, 2010) and should therefore a-priori lead to faster color converters for VLC as mentioned previously. A white-light source with a bandwidth of 85 MHz was demonstrated by combining a micro-LED and yellow-emitting CsPbBr1.8I1.2 perovskite CQDs (blended in resin) achieving a 300 Mb/s VLC link (using non-return to zero on-off keying modulation scheme, see Section 3.1.2) (Mei et al., 2018). Demonstration of color-converted micro-LED based VLC source to date have relied on integration on a macroscopic scale, often just a color converter in close contact with the micro-LEDs as shown in Fig. 4B. Micro-scale integration using different (in some cases a combination of ) techniques such as ink-jet or spray printing (Lin et al., 2017), contact/transfer printing (Kim et al., 2011), or photolithography (Guilhabert et al., 2008; Park et al., 2016) is however possible; this is an area of current interest for high-density arrays aiming at displays. It is also possible to do the opposite, i.e., to print the micro-LEDs themselves onto the color converter using micro transferprinting as was done in reference (Rae et al., 2017). 2.2.3 Organic semiconductors The properties of organic semiconductors as VLC color converters were recently reviewed by Manousiadis et al. (2020) and therefore we only give a very concise overview here. In the form of conjugated molecules and polymers, they benefit from many of the same processing capabilities as CQDs for forming thin-films and composites while their dynamics are faster (although it can be argued that, depending on the wavelength, perovskite

294

P. Tian et al.

CQDs can be comparable). The typical bandwidth of organic semiconductors ranges from 39 to 470 MHz. The latter record value was obtained for a thin-film of an orange emitting poly (phenylenevinylene) (PPV) derivative having a PLQY of 47% (Vithanage et al., 2017). Micro-LED based white light VLC up to 1.68 Gb/s was demonstrated with a commercial PPV (called super yellow, 200 MHz bandwidth) and OFDM as the modulation scheme (Chun et al., 2014). While the latter was neither in the solid-state nor integrated onto the micro-LED, it clearly showed the potential of organic semiconductor for >1 Gb/s VLC. Aggregation-induced emission luminogens have also recently been studied as color converters for VLC; these are organic-based molecules that are designed to be emissive only in the solid-state. They have shown >100 MHz bandwidth for emission from the blue to the red part of the visible spectrum (Zhang et al., 2018). 2.2.4 Epitaxial quantum well nanoplatelets The third color-converting technology for micro-LED-based VLC is based on epitaxial quantum wells (QWs) (Santos et al., 2015, 2016; Schiavon et al., 2013). These quantum wells are grown separately from the LED structure and are designed for photo-pumping, mitigating (but not eliminating) the issues of wavelength engineering for the green/yellow gap region. The QW structure is then heterogeneously integrated with the micro-LED. Hybrid LED color-converted emission in the green (Santos et al., 2015; Schiavon et al., 2013) and red (Santos et al., 2016) have been reported in this way. The main foreseeable advantage of such an approach is that of stability conferred by an all epitaxial device (albeit obtained by heterogeneous integration). The challenges include a more complicated fabrication and integration, and the difficulty of efficient light extraction and emission in the green-yellow gap. In the report (Santos et al., 2016), a 400 nm-thick AlInGaP multiquantum well membrane released from its GaAs growth substrate was bonded by capillarity (Van der Waals effect) onto the sapphire substrate of a micro-LED. To increase the light extraction from the membrane, a hemispherical lens was added on top. The BW of this color-converting membrane was found to be excitation-dependent and varied between 150 and >300 MHz. The overall conversion efficiency was only around 1%, however, attributed to a low extraction efficiency from the membrane itself. Data rates up to above 800 Mb/s in a short-range (10 cm) free space link, with the remaining unabsorbed blue micro-LED light filtered out, were demonstrated using 4-PAM modulation.

Micro-LED based optical wireless communications systems

295

It is possible to create such a color converter with microsize areas by standard microelectronics processing, and therefore they can be transfer printed directly onto individual micro-LED pixels. Consequently, the approach is compatible with high-density micro-LED arrays. 2.2.5 Summary This section has given an overview of color-converting material technology for micro-LED based VLC. Both CQDs and organic semiconductors benefit from versatile fabrication for integration with micro-LEDs. Organic semiconductors have faster dynamics in general, while CQDs have narrower emission wavelengths and do not require special design considerations for matching their absorption with the emission of micro-LEDs. CQDs research for integration within high-density micro-LED arrays is also progressing fast and may come to be the dominating color-converting technology for micro-LEDs in the near future. A problem with CQDs is that the fastest materials rely on toxic elements. InP CQDs may be a good replacement, being only slightly slower in their temporal response. There is also intense research into Pb-free metal halide perovskites and it will be interesting to see if a solution retaining the fast response of current perovskite CQDs can be found. Finally, another interesting prospect for VLC is that of epitaxial QW color converters if the challenges of wavelength coverage, efficiency and fabrication complexity can be overcome.

3. Micro-LED based high-speed VLC systems 3.1 Advanced modulation schemes 3.1.1 LED data modulation overview As discussed in Section 2.1, micro-LEDs have very high modulation bandwidths compared to standard broad-area LEDs. This in principle allows them to support very high data rate VLC. However, in order to transmit data, it must first be encoded in such a way that it can be robustly transmitted over an optical channel. Fig. 6 shows a simplified block diagram of a VLC system. The input digital bit stream to be transmitted is first encoded using the selected modulation scheme, and then converted by a waveform generator or digital-to-analog converter to a voltage signal that will modulate the output intensity of the LED. This signal is combined with a DC bias using a bias-tee, which is necessary to overcome the forward voltage threshold of the LED and shift the voltage modulation by the waveform generator into a regime where the light output to voltage characteristic of the LED is

296

P. Tian et al.

Fig. 6 A simplified block diagram of a VLC system. The transmitter and receiver sub-systems are shown in blue and orange, respectively.

approximately linear. The output intensity of the LED is then modulated by the applied voltage signal, following the LED’s particular output power vs voltage (L–V) relationship. This optical signal is transmitted over the optical channel, potentially through optics at the transmitter and/or receiver end, and received by a photodetector. At the receiver side, a photoreceiver, such as a PIN or avalanche photodiode, converts the optical signal back into the electrical domain. Decoding electronics then process this signal and attempt to faithfully recover the original input bit stream. Note that, unlike radio frequency (RF) communication, the optical signal is carried only by the instantaneous power of the optical carrier as the optical emission from LEDs is inherently incoherent. There are many modulation schemes that can be chosen to encode data in VLC systems, each with their own trade-offs in terms of complexity, spectral efficiency and power efficiency. For a more detailed discussion, the reader is directed to references such as Haas et al. (2016) or Chi (2018). Here we will briefly summarize two of the modulation schemes most commonly used in VLC demonstrations using micro-LEDs, On-Off Keying (OOK) and Orthogonal Frequency Division Multiplexing (OFDM). 3.1.2 On-off keying (OOK) OOK is one of the oldest and the simplest digital modulation schemes, both in concept and practical hardware implementation. Information is encoded in the intensity of output pulses, which is modulated between digital logic

Micro-LED based optical wireless communications systems

297

Fig. 7 OOK binary optical transmission.

High (“On”) and logic Low (“Off”) levels. In Non-Return-to-Zero (NRZ) OOK, the pulse intensity is held constant during each bit period, as opposed to returning to the “zero” state as is the case in Return-to-Zero (RZ). Fig. 7 provides an illustrated example of an optical OOK-NRZ signal, indicating the logic 0, logic 1 and DC bias levels. At the receiver a threshold can be applied to the received signal to convert the analog photocurrent back into a digital bit stream. With high bandwidths available from micro-LEDs, OOK data rates in excess of 1 Gb/s can be supported. For example, Dinh et al. (2016) reported 2.4 Gb/s using a semipolar micro-LED with a modulation bandwidth of 1 GHz. While OOK offers a good compromise between complexity and performance, it suffers from deleterious effects at higher data rates such as inter-symbol interference (ISI) (Elgala et al., 2011). Other modulation schemes are more robust against these effects and offer higher spectral efficiency. Spectral efficiency indicates how much data can be sent over the available system bandwidth, defined in units of bits/s/Hz, thus modulation schemes with higher spectral efficiency are capable of transmitting higher data rates. OFDM is one such commonly-used scheme, and will be discussed in the subsequent section. 3.1.3 Orthogonal frequency division multiplexing (OFDM) Unlike OOK, which transmits a single bit stream in series, OFDM transmits multiple parallel data streams simultaneously, thus making more

298

P. Tian et al.

efficient use of the available bandwidth by exploiting the available signal to noise ratio. It is also robust against multipath interference. In OFDM, the available bandwidth is subdivided into many sinusoidal carrier signals of different frequency, known as “subcarriers.” OFDM, being a special case of Frequency Division Multiplexing (FDM), uses subcarriers with frequencies chosen so that they are mathematically orthogonal. Each parallel data stream is then encoded onto a subcarrier by modulating its phase and/or amplitude, the modulated subcarriers are then combined together and transmitted simultaneously by the LED. Fig. 8 shows an example block diagram of an optical OFDM system. First, the input data is demultiplexed from a serial bit stream to multiple parallel bit streams. These parallel data streams are then modulated using a modulation scheme such as Quadrature Amplitude Modulation (QAM), and an Inverse Fast Fourier Transform (IFFT) is used to multiplex the QAM symbols onto frequency subcarriers in a time-sequence OFDM symbol. Cyclic prefixes (CPs) are inserted between OFDM symbols to mitigate against ISI. Each OFDM symbol with CP is then written onto a serial stream by a parallel-to-serial converter. This serial stream is then sent to a Digitalto-Analog Converter (DAC) or waveform generator to generate an analog voltage that subsequently modulates the LED output intensity. This process is reversed at the receiver. An Analog-to-Digital Converter (ADC) converts the analog output of the photoreceiver to a digital signal. Each OFDM symbol is stored into a memory by a serial-to-parallel converter, the CPs are removed, a Fast Fourier Transform converts the signal back into the frequency domain, the QAM symbols of the parallel carriers are decoded and the data multiplexed back into a serial bitstream (Armstrong, 2009;

Fig. 8 Block diagrams of the modulation and demodulation processes of OFDM that take place at the transmitter and receiver ends, respectively.

Micro-LED based optical wireless communications systems

299

Rajbhandari et al., 2017b). OFDM also allows adaptive bit and powerloading algorithms to be used to optimize the channel capacity (Haas et al., 2016), by loading more data onto subcarriers with a higher signalto-noise ratio. OFDM has been extensively used in broadband wired and wireless communications as it is robust against ISI and scales well with increasing data rates (Armstrong, 2009). In recent years, OFDM has increasingly been adapted and applied to optical wireless communications. Most of the reported micro-LED VLC demonstrations do not use NRZ-OOK, but some form of OFDM instead, with the highest reported data rates generally being achieved using OFDM. At the time of writing, the highest error-free data rates demonstrated using GaN micro-LEDs are 7.91 Gb/s by Islim et al. (2017) using a single violet-emitting micro-LED and 11.74 Gb/s by Xie et al. (2020) using an array of series-connected micro-LEDs. Table 1 shows a chronological, but not exhaustive, list of published reports of VLC using micro-LEDs starting with the report of 1 Gb/s using OOK-NRZ by McKendry et al. (2010). From this table, the trend toward ever higher data rates and OFDM is apparent, as both the performance of micro-LEDs and the optimization of optical modulation schemes continue to mature.

3.2 Integrated micro-LED/CMOS for structured VLC and MIMO VLC 3.2.1 Background Early generations of micro-LED displays were fabricated in passive-matrix formats, which require rapid line-scanning in order to display arbitrary patterns. Scaling this approach up to large arrays has limitations, as visible flicker can become an issue for arrays larger than 0.01 Megapixels (Griffiths et al., 2020). Therefore active-matrix architectures, where each micro-LED pixel has a dedicated electronic driver, are highly desirable. It is possible to monolithically integrate LEDs and driver electronics in a single GaN-based epitaxial structure, as demonstrated by Lee et al. (2014). However, this requires complex epitaxial growth and processing, and the area of the chip footprint that must be set aside for the drivers necessarily limits the display fill-factor, as is indicated by the schematic cross-section of such a device in Fig. 9. Instead, the most commonly-favored approach is to separately fabricate a micro-LED array and a driver array and vertically-integrate the two using flip-chip bump bonding. This approach allows for high fill-factor

Table 1 Chronological list of selected publications on VLC using micro-LEDs. Modulation Maximum data rate Transmission Publication and year scheme (Gb/s) distance (m)

Comments

McKendry et al. (2010)

OOK-NRZ 1

Back-to-back

McKendry et al. (2011)

OOK-NRZ 0.512

Back-to-back CMOS-controlled micro-LED array

Wun et al. (2012)

OOK-NRZ 1.07

50

Transmission over polymer optical fiber (POF)

Maaskant et al. (2013)

OOK-NRZ 0.5

Not given

Transmission over POF

Zhang et al. (2013)

OOK-NRZ 1.5

Back-to-back Multi-channel transmission using CMOS-controlled micro-LED array

Tsonev et al. (2014)

OFDM

3

0.05

10

10

5

0.5

a

Li et al. (2015)

PAM

Ferreira et al. (2016)

OFDM

Dinh et al. (2016)

OOK-NRZ 2.4

Back-to-back Semi-polar micro-LED

Liu et al. (2017)

OOK-NRZ 1.3/1

3/10

Islim et al. (2017)

OFDM

7.91

Not given

7.48

0.5

Optical MIMO using CMOS-controlled micro-LED array

2.5

0.05

Transmission over 2D waveguide arrays

Rajbhandari et al. (2017a) PAMa a

Transmission over POF

Bamiedakis et al. (2019)

PAM

He et al. (2019)

OFDM

1

0.3

UVC-emission (262 nm)

Xie et al. (2020)

OFDM

11.74/10/2

0.3/5/20

Series-connected micro-LEDs

a

Pulse amplitude modulation.

301

Micro-LED based optical wireless communications systems

Ni/ITO

Source Gate

Drain

SiO2 P-GaN MQWs

Ni Ti/Ai

n-GaN u-GaN VDD Sapphire

Fig. 9 Cross-section schematic of a monolithically-integrated GaN LED and metaloxide-semiconductor field-effect transistor (MOSFET). From Lee, Y., Yang, Z., Chen, P., Hsieh, Y., Yao, Y., Liao, M., Lee, M., Wang, M., Hwang, J., 2014. Monolithic integration of GaN-based light-emitting diodes and metal-oxide-semiconductor field-effect transistors. Opt. Express 22, A1589-A1595.

Fig. 10 Schematic illustration of a micro-LED array bump-bonded to a CMOS driver array. From Griffiths, A. D., Herrnsdorf, J., McKendry, J. J. D., Strain, M. J., Dawson, M. D., 2020. Gallium nitride micro-light-emitting diode structured light sources for multi-modal optical wireless communications systems. Phil. Trans. R. Soc. A 378, 20190185.

micro-LED arrays to be integrated with complex digital drive circuitry, as the driver circuitry is implemented underneath the emissive part of the array. Fig. 10 shows an illustration of this approach. The electronic drivers are fabricated using a standard complementary metal-oxide-semiconductor (CMOS) foundry process and bump bonds, typically either Au or In, are deposited onto contact pads on this array. These pads are aligned with a matching set of pads on the micro-LED array and then force, heat and ultrasonic energy are used to partially melt the bump bonds and establish physical and electrical contact between the two arrays. Light emitted by the pixels is then extracted through the micro-LED sapphire substrate or, if this has been removed, the n-GaN layer. CMOS-controlled micro-LED displays are now being commercialized for applications such as Virtual Reality and Augmented Reality (VR/AR). Fig. 11 shows an example of a 0.2600 diagonal green-emitting display developed by Compound Photonics and Plessey. It features a 1080p (1920 1080)

302

P. Tian et al.

Fig. 11 0.2600 diagonal monochrome native Green 1080p display module produced by Compound Photonics and Plessey. From https://plesseysemiconductors.com/compoundphotonics-plessey-first-microLED-display-module/.

array of micro-LEDs on a 3.015 μm pitch, integrated with CMOS electronics capable of displaying real-time video. 3.2.2 Applications in visible light communications Driver electronics tailored for display applications necessarily have functionalities including pixel brightness control and the ability to refresh the displayed pattern at video rates. These capabilities can also be repurposed for VLC applications, or functionality specific tailored for VLC can be implemented in CMOS. Optical Camera Communications (OCC) is a form of VLC that uses the cameras widely deployed in devices such as smartphones, laptops, and vehicles as receivers (Saha et al., 2015). In OCC the emission from an intensity-modulated light source, such as an LED, is captured using a camera. The intensity of the transmitter vs time can be extracted from the captured video frames, from this the transmitted optical signal can be recovered and wireless data transmission achieved. If an arrayed transmitter is used, such as a micro-LED array, this intensity modulation can be done while simultaneously displaying patterns and images visible to the human eye, which is referred to as “Data-through-displays.” In OCC the possible data rates are limited by the frame capture rate of these cameras, which is typically up to 60 fps although consumer devices are now available with slow-motion capture modes up to 960 fps.

Micro-LED based optical wireless communications systems

303

Encoding data on the displayed 2D patterns themselves allows these data rates to be increased by exploiting the spatial, as well as temporal, domain. Griffiths et al. (2019) demonstrated data transmission rates up to 122.88 kb/s using a 16  16 CMOS-controlled micro-LED projector system and a 960 fps smartphone camera. By using a higher-resolution transmitter, these data rates could in principle be scaled further up to Mb/s. Another example of CMOS-controlled micro-LEDs being applied to VLC is the concept of “Digital to Light” conversion, an example of which was reported by Griffiths et al. (2017). The concept had previously been demonstrated by Fath et al. (2013) using off-the-shelf LEDs and provides a method of generating discrete output power levels by switching on or off binary-weighted groups of LEDs, hence the alternative terminology “Discrete Power Level Stepping.” This allows the output intensity levels required for VLC to be generated without the distortion that can arise by modulating the intensity of a single LED due to the LED’s inherently non-linear relationship between bias voltage and output power. It also allows the output power to be modulated without the requirement for a separate DAC, in effect a quasi-analog optical output is generated directly from a digital input. Fig. 12A illustrates the concept as implemented on a violet-emitting 16  16 micro-LED array. Four groups of pixels in different columns in the array are used, each having a total of 2x pixels, where x was an integer varied from 0 to 3. Assuming that the total emitted power is proportional to the number of active pixels, this means that by switching on a single column at a time that four discrete output power levels could be generated. This is shown in Fig. 12B where an optical Pulse-Amplitude Modulation (PAM) output signal with four discrete levels (4-PAM) was generated for a resulting bit rate of 200 Mb/s. The aforementioned demonstrations give examples of how CMOScontrolled micro-LED arrays that are primarily designed for micro-display applications can also be used for VLC. However, there are examples of such devices that have been custom-designed specifically for VLC. One example was reported by Rajbhandari et al. (2017a). Nine groups of four microLED pixels, each 39 μm in diameter and driven by a separate CMOS driver, were used in a Multiple-Input, Multiple-Output (MIMO) VLC demonstrator system. The emission from each group of micro-LEDs was imaged onto a single receiver in a 3  3 avalanche photodiode (APD) array, allowing up to nine parallel data streams to be transmitted simultaneously. MIMO thus allowed the data capacity to increase linearly with the number

Fig. 12 (A) Schematic illustration and corresponding optical micrograph of a binary-weighted group of micro-LEDs and (B) the resulting 4-PAM signal from this device recorded at 100 M Samples/s. From Griffiths, A. D., Islim, M. S., Herrnsdorf, J., McKendry, J. J., Henderson, R., Haas, H., Gu, E., Dawson, M. D., 2017. CMOS-integrated GaN LED array for discrete power level stepping in visible light communications. Opt. Express 25, A338–A345.

Micro-LED based optical wireless communications systems

305

of independent MIMO channels. Using this configuration, Rajbhandari et al. (2017a) reported a data transmission rate up to 7.48 Gb/s over a distance of 0.5 m.

3.3 Summary This section gave an overview of strategies that have been developed for fully exploiting the high modulation bandwidths and emission patterning capabilities of micro-LED pixels and arrays in order to maximize optical wireless data transmission rates. Advanced modulation techniques adapted from RF wireless communications, such as OFDM, have enabled optical wireless data rates approaching 10 Gb/s to be demonstrated using individual micro-LED pixels, with 10 Gb/s recently being exceeded using seriesconnected groups of micro-LEDs. Integrating micro-LED arrays with control electronics, which provides sophisticated control of the spatio-temporal patterns emitted by the array, provides a further route to extending data rates by utilizing the spatial domain. Finally, convergence with consumer electronics such as high frame rate smartphone cameras can enable novel forms of optical wireless, such as “Data Through Displays.”

4. Novel optical wireless communication systems based on micro-LED 4.1 Underwater wireless optical communication based on micro-LEDs Underwater wireless communication is crucial for oceanography investigation, offshore oil exploration and sea floor monitoring. The available techniques mainly include underwater acoustic communication, underwater radio-frequency (RF) communication, and underwater wireless optical communication (UWOC) (Zhu et al., 2020). Traditional acoustic communication supports long-distance communication of tens of kilometers, but suffers from low bandwidth of
kHz, low data rate of
kb/s, high latency due to low underwater acoustic wave speed, multipath propagation, and Doppler spread. Underwater RF communication has the advantages of low latency, smooth transition through the air/water interface, and tolerance to water turbulence. High frequency RF only offers short-range underwater communication of meters, and RF waves at extra low frequency 30–300 Hz can propagate through conductive salty water but huge antenna and high power are required. Compared with underwater acoustic and RF communication, UWOC has the advantages of high data rate, low

306

P. Tian et al.

latency, and transmission distances up to hundreds of meters, depending on the local water properties. Although the light may be strongly attenuated by water absorption and scattering from suspended particles, the attenuation in the blue-green spectrum region is relatively small, making high-efficiency and high-bandwidth GaN-based blue and green emitters suitable for UWOC. The external quantum efficiency of GaN-based LEDs and lasers has been significantly enhanced in the past two decades, enabling solid-state lighting (SSL) and visible light communication, as well as UWOC. LDs have higher bandwidth and low light beam divergence which would help the long-distance high-speed UWOC, but have disadvantages of high cost, potential eye safety issues and more strict alignment requirements. LEDs are relatively cheap, and high-power LEDs are available. The Lambertian radiation of LED beam restricts the UWOC distance and the corresponding optics needs to be optimized. In addition, the LED bandwidth is still low which limits the highest achievable data rate of UWOC. As discussed earlier in Section 2.1, the 3 dB electrical-to-optical modulation BW of commercial blue LEDs is usually less than tens of MHz, so micro-LEDs have been proposed for UWOC application. UWOC using high-bandwidth micro-LED was first proposed by Tian et al. (2017). In Fig. 13, a micro-LED on sapphire substrate with size of 80 μm  80 μm was used for UWOC. The light was emitted from the sapphire side with a peak emission wavelength of
440 nm. The main difference of the experimental setup between UWOC and VLC is that the communication channel has been changed from free space to water, otherwise the setup is broadly similar to that described in Section 3.1.1. The light emission from the micro-LED was collimated and we can see the transmitted light beam through the 0.6 m water tank filled with tap water due to the scattering effect. Reflective mirrors were used to reflect the light beam within the tank, therefore a longer transmission distance than the tank length was obtained. By employing an OOK modulation scheme for the 160 MHz micro-LED, 800 and 200 Mb/s data rates were demonstrated at distances of 0.6 and 5.4 m, respectively. A single high-bandwidth blue micro-LED has a typical light output power of only several mW, which needs to be improved for long-distance UWOC under high-attenuation underwater environment. Arvanitakis et al. (2020) developed 450-nm-emitting, series-connected micro-LED arrays consisting of six micro-LED pixels with diameter of 60 or 80 μm as shown in Fig. 14. Compared with parallelly connected micro-LEDs, a series connection can reduce the resistance-capacitance (RC) effect mitigating the typical RC limitation on modulation bandwidth for broad-area

Micro-LED based optical wireless communications systems

307

Fig. 13 (A) Schematic setup of UWOC using a micro-LED. Images of (B) the packaged micro-LED, (C) the communication link and (D) the PIN receiver. From Tian, P., Liu, X., Yi, S., Huang, Y., Zhang, S., Zhou, X., et al., 2017. High-speed underwater optical wireless communication using a blue GaN-based micro-LED. Opt. Express 25, 1193–1201.

LEDs, while increasing the available optical transmission power by using multiple micro-LEDs. The device has a light output power of more than 10 mW but still maintains high optical modulation bandwidth of 338.5 MHz at 30 mA for an array of 60 μm micro-LEDs and 263 MHz at 50 mA for an array of 80 μm micro-LEDs. Using OFDM, a UWOC data rate of 4.92 Gb/s was achieved at 1.5 m tap-water distance based on the 60 μm micro-LED array, and using the 80 μm micro-LED array 3.22 and 3.4 Gb/s were demonstrated over 3 and 4.5 m tap water, respectively. Underwater channels cause significant light attenuation which has been extensively researched (Duntley, 1963; Zeng et al., 2016). The Beer Lambert’ law can be used to describe the light attenuation using the formula I ðzÞ ¼ I 0 ecðλÞz , where I is the light output power at the receiver, I0 is the light output power at the transmitter, z is the transmission distance, and c(λ) is the underwater attenuation coefficient. Based on the attenuation coefficients caused by absorption and scattering, four different water types can be classified, i.e., pure sea water, clear ocean water, coastal ocean water, and turbid harbor

A

B

60 lm (E-E -3dB BW) 60 lm (E-O -3dB BW) 80 lm (E-E -3dB BW) 80 lm (E-O -3dB BW)

300

Metal tracks to P-type contact

P-type GaN

n-type GaN

Insulation layer

P-type contact

Quantum Wells

Buffer layer

C

FEC threshold

3x10–3

250 10–3 200

BER

-3dB Bandwidth (MHz)

350

Sapphire

150

(4.92 Gb/s, 1.5x10–3)

3x10–4

100 10–4

50 0 0

10

20

30

Current (mA)

40

50

4.0

4.5

5.0

Data Rate (Gb/s)

Fig. 14 (A) Optical image and schematic structure of series-connected micro-LEDs. (B) 3 dB modulation bandwidth characteristics of two series-connected micro-LEDs. (C) UWOC data rate through 1.5 m tap water. From Arvanitakis, G. N., Bian, R., McKendry, J. J. D., Cheng, C., Xie, E., He, X. Y., Yang, G., Islim, M. S., Purwita, A. A., Gu, E. D., Haas, H., Dawson, M. D., 2020. Gb/s underwater wireless optical communications using series-connected GaN micro-LED arrays. IEEE Photonics J. 12, 1–10.

309

Micro-LED based optical wireless communications systems

Optical An example of the attenuation coefficient (m-1)

Euphotic zone (200 m)

0m

0

LED

Refractive index at 500 nm in Pacific Ocean

LD

0.05 0.1

0m

1.336 1.34

100 m 200 m

200 m

Dysphotic zone

1000 m

(few kilometers)

Aphotic zone (deeper than few kilometers) Clear ocean water

phytoplankton and phytobenthos

sunlight

salinity

Fig. 15 Schematic diagram of the ocean environment. From Zhu, S., Chen, X., Liu, X., Zhang, G., Tian, P., 2020. Recent progress in and perspectives of underwater wireless optical communication. Prog. Quantum Electron. 73, 100274.

water. In addition, the value of the attenuation coefficient c(λ) will also vary with the depth of water. Fig. 15 shows the schematic ocean environment for UWOC. In the vertical direction, we can see the euphotic zone, dysphotic zone and aphotic zone, with different physical characteristics of salinity, phytoplankton and phytobenthos concentration, background sunlight, refractive index, etc. Attenuation of micro-LED UWOC has been studied under different simulated water environment by adding Maalox® antacid, chlorophyll, and sea salt (Tian et al., 2019). In a micro-LED with emission wavelength of 445 nm based UWOC, Maalox® suspension, mainly consisting of Al(OH)3 and Mg(OH)2, was added in water to simulate the light scattering effect in ocean water with suspended particles (Tian et al., 2019). In Fig. 16A, the light attenuation coefficients were calculated under different Maalox® concentrations. According to the attenuation coefficient value, different water types may be simulated by adding Maalox®. The attenuation coefficient could be fitted well by a scattering model, suggesting that the scattering effect is dominant for diluted Maalox® suspension. Furthermore, in Fig. 16B, Swiss chlorophyll was employed to simulate the absorption effect and an absorption model could be used to fit the attenuation excellently. The Swiss chlorophyll is obtained from natural chlorophyll and is soluble in water, being adopted to simulate the attenuation effect of sea water with organic particles, phytoplanktons in particular which can absorb light.

310

P. Tian et al.

Fig. 16 (A) Experimentally obtained water attenuation coefficients and theoretically fitting results at different Maalox® and (B) chlorophyll concentrations of a micro-LED based UWOC system. Inset: the light beam through 100.24 mg/m3 Maalox® and 1207.73 mg/m3 chlorophyll. From Tian, P., Chen, H., Wang, P., Liu, X., Chen, X., Zhou, G., Zhang, S., Lu, J., Qiu, P., Qian, Z., Zhou, X., Fang, Z., Zheng, L., Liu, R., Cui, X., 2019. Absorption and scattering effects of Maalox, chlorophyll, and sea salt on a microLED-based underwater wireless optical communication. Chin. Opt. Lett. 10, 100010.

In summary, UWOC has developed rapidly in the last decade. MicroLED based UWOC systems have been investigated from device fabrication to UWOC channel modeling and simulation. We expect that micro-LED has great potential to work as a low-power high-speed light source for the future underwater “Internet of Things.”

Micro-LED based optical wireless communications systems

311

4.2 Micro-LED array based duplex VLC Previous sections have discussed how micro-LEDs can be used as highbandwidth transmitters for VLC. In this section, we will discuss how micro-LED could also serve as high-bandwidth photodetectors (PDs), and thus enabling both transmitter and detector functionalities to be integrated within a single micro-LED array. Such integrated micro-LED arrays provide a possible approach to solve the well-known problem of emitter and detector integration for duplex VLC. Fig. 17A shows an application scenario for a full-color micro-LED display. Several micro-LEDs can transmit parallel information to the users, and the micro-LEDs in the same chip, working as high-bandwidth detectors, can also detect external optical signals. Thus, for typical micro-LED applications in virtual reality (VR) or augmented reality (AR) displays, micro-LED based VLC has huge potential to supplement and even replace present Wi-Fi or wired networks (Liu et al., 2019). A typical micro-LED structure for photodetector application is shown in Fig. 17B. The micro-LED based photodetector array was fabricated on a sapphire substrate with emission peak wavelength of
450 nm. The device consists of an n-GaN layer, an InGaN/GaN multiple quantum well (MQW), and a p-GaN layer, as well as p- and n-contact separated by a SiO2 layer. All optoelectronic properties are characterized at room temperature using LDs as the illumination source. Fig. 18 shows the I–V characteristics of the PDs as a function of optical pump power density, device size, laser wavelength and reverse bias. With higher optical power density, the

Fig. 17 (A) Illustration of micro-LEDs for applications in both micro-LED display and duplex VLC. (B) Schematic diagram of the micro-LED structure for photodetector array. From Liu, X., Lin, R., Chen, H., Zhang, S., Qian, Z., Zhou, G., Chen, X., Zhou, X., Zheng, L., Liu, R., Tian, P., 2019. High-bandwidth InGaN self-powered detector arrays toward MIMO visible light communication based on micro-LED arrays. ACS Photonics 6, 3186–3195.

312

P. Tian et al.

Fig. 18 I–V characteristics of the PDs with diameters of (A) 40, (B) 60, and (C) 100 μm, under different optical power densities from the 405 nm laser. I–V characteristics as a function of (D) device size, (E) laser wavelength and (F) reverse bias. From Liu, X., Lin, R., Chen, H., Zhang, S., Qian, Z., Zhou, G., et al., 2019. High-bandwidth InGaN self-powered detector arrays toward MIMO visible light communication based on micro-LED arrays. ACS Photonics 6, 3186–3195.

photocurrent increases for PDs with all sizes due to more photo-generated carriers. Photosensitivity of these PDs, defined as the ratio of the photocurrent to the dark current, reaches 109 under zero-bias with an optical power density of 1 W/cm2, but the temperature effect is not taken into account. The high on/off ratio helps to improve the detection of weak optical signals and the signal-to-noise ratio. The responsivity is calculated by the formula R¼

Iph q ¼ η   λ, hc Lin

Micro-LED based optical wireless communications systems

313

where Iph is the photocurrent of the PD, Lin is the external optical power exciting the active region of the PD, q is the elementary charge, h is the Plank constant, c is the light velocity, λ is the laser wavelength, and η is the quantum efficiency of the PD. At zero bias under 405 nm power density of 11 W/cm2, R is 0.24, 0.29, and 0.21 A/W and η is 74%, 88%, and 62% for the 40, 60, and 100 μm PD, respectively. Both R and η increase after increasing the reverse bias as more carriers could be collected to produce photocurrent. The responsivity value is higher than that of Si-PIN detector at 405 nm. Specific detectivity (D*) of the PD is a figure of merit to evaluate the PD’s ability to detect weak signals, which can be expressed as rffiffiffiffiffiffiffiffiffiffiffiffi AμLED D ¼R , 2qid ∗

where AμLED represents the area of the PD, and id is the dark current dominated by the shot noise. Under 405 nm optical power density of 11 W/cm2 and zero bias, the D* is 7.5  1012, 1.5  1013, and 1.3  1013 cm Hz1/2/W1 (or Jones), respectively. The analysis above shows that the micro-LED based PD has excellent responsivity and specific detectivity. Beside responsivity, the micro-LED based PD demonstrates wavelength selectivity as shown in Fig. 18E. The PD responsivities at wavelengths below 450 nm are much higher than those at other wavelengths. The PDs have very low responsivity in the yellow spectra region and the photocurrent mainly comes from the light emission with wavelength below 450 nm. Such PDs are quite suitable for VLC based white-light LEDs mixed by blue LEDs and slow yellow phosphor. The micro-LED based PD also has fast response with 3 dB optical bandwidths of 56.8, 56.2, and 53.5 MHz for 40, 60, and 100 μm devices, respectively, making the PDs suitable for high speed VLC. For the 2  2 MIMO VLC experiment in Fig. 19, two 405 nm lasers (LDs) were used as parallel transmitters, and two corresponding micro-LEDs were employed as receivers. The small divergence angle of the LDs could reduce the optical cross-talk between the two VLC channels. Using OOK modulation, the maximum data rates of 40, 60, and 100 μm devices are 180, 175, and 185 Mb/s at 5 V bias, respectively. Then, the 2  2 MIMO VLC demonstrates double data rate, and we expect that higher data rate is possible using a larger array, e.g., 10  10 micro-LED array (Tian et al., 2014). Higher modulation bandwidth and data rate of InGaN-based micro-detector were reported. Ho et al. (2018) demonstrated an InGaN

314

P. Tian et al.

Fig. 19 (A) Illustration and (B) picture of MIMO VLC using a micro-LED array as a photodetector array. From Liu, X., Lin, R., Chen, H., Zhang, S., Qian, Z., Zhou, G., Chen, X., Zhou, X., Zheng, L., Liu, R., Tian, P., 2019. High-bandwidth InGaN self-powered detector arrays toward MIMO visible light communication based on micro-LED arrays. ACS Photonics 6, 3186–3195.

MQW based 80-μm PD on a sapphire substrate could achieve 3 dB optical bandwidth of 71.5 MHz and 3.2 Gb/s VLC data rate using OFDM modulation scheme. Alkhazragi et al. (2020) further reported a 7.4 Gb/s data rate based on a semipolar micro-photodetector. The PD structure in their work suggests that micro-LEDs could also be potentially fabricated using the same wafer, so we expect that the performance of integrated micro-LED and PD could be further improved. In summary, it can be seen that micro-LEDs could also be employed as PDs. The micro-LED based PD demonstrates excellent characteristics of

315

Micro-LED based optical wireless communications systems

responsivity, wavelength-selectivity, bandwidth and data rate. Such research provides an approach for high-speed, duplex, and MIMO VLC.

4.3 Deep-ultraviolet micro-LED communication Deep-ultraviolet (UV) communication has attracted significant attention due to its advantages of low background noise and non-line-of-sight (NLOS) operation. As shown in Fig. 20, all solar UV radiation in the UV-C band between 200 and 280 nm is absorbed by the ozone layer and upper atmosphere before reaching the ground, so the UV-C background noise is very low for UVC communications. Strong scattering of UV-C light in the air makes NLOS communication link possible, which is an advantage over VLC. It is also argued that deep-UV communication in outer space can be inherently secure as inter-satellite UV-C communication channels are protected from eavesdropping by ground-based receivers due to the atmospheric absorption (He et al., 2019). Similar to the LEDs for VLC, broad-area UV-C LEDs have low modulation bandwidth. The highest reported modulation bandwidth of conventional broad-area UV-C LED is 153 MHz, suggesting the great potential to increase the bandwidth by fabricating UV-C micro-LEDs. Fig. 21 shows a typical deep-UV communication experiment using a 262-nm micro-LED 320-380nm 280-320nm 200-280nm

UVA

UVB

UVC

Re ce iv er l

in

e

of

si g

ht

Ozone sphere

NLOS

Fig. 20 Illustration of UV-C communication.

Tra ns

mit

ter

line

of s igh

C light path Scattered UV

t

316

P. Tian et al.

Fig. 21 (A) UV-C micro-LED communication setup. (B) Electrical and optical characteristics, (C) modulation bandwidth, and (D) data rates of the UV-C micro-LED. From He, X., Xie, E., Islim, M.S., Purwita, A.A., McKendry, J.J., Gu, E., et al., 2019. 1 Gbps free-space deep-ultraviolet communications based on III-nitride micro-LEDs emitting at 262 nm. Photonics Res. 7, B41–B47.

(He et al., 2019). The UV-C micro-LED has an emission area of 566 μm2 and an optical power of 196 μW. The 3 dB electrical modulation bandwidth achieved 438 MHz, much higher than broad-area UV-C LEDs. Using OOK and OFDM modulation schemes, 800 Mb/s and 1.1 Gb/s data rates have been achieved, respectively. There is still huge potential to improve the UV-C communication system. To effectively use the advantages of UV-C micro-LEDs for longdistance NLOS communication, we still need to increase the light output power of the UV-C micro-LED by developing high-quality UV-C material and series-connected micro-LEDs. The data rate could be further increased by developing highly-sensitive high-bandwidth UV-C detectors. Those efforts are expected to lead to the practical application of UV-C communication in the future.

5. Conclusion The intensive worldwide effort underway to develop new forms of high-brightness, high-pixel-density, low power consumption and fast

Micro-LED based optical wireless communications systems

317

response electronic visual display technology based on micro-LEDs is also likely to facilitate dramatic new developments in micro-LED-based optical communications. Micro-LEDs are capable of multi-Gb/s data communications, can be imaged or collimated for free space applications or efficiently coupled into fibers or waveguides as required, are readily compatible with sophisticated control electronics and can be fabricated for operation from the deep-ultraviolet through to the long-wavelength visible. This combination of advantageous characteristics is opening up a wide range of applications scenarios for a new era in visible and ultraviolet communications.

Acknowledgments The University of Strathclyde authors gratefully acknowledge extensive support from EPSRC for micro-LED visible light communications research, including grants EP/ F05999X/1, EP/K00042X/1, EP/M013264X/1, EP/T000974X/1, and EP/S001751/1. Fudan University authors acknowledge the support from National Natural Science Foundation of China (61974031 and 61705041).

References Alkhazragi, O., Kang, C.H., Kong, M., Liu, G., Lee, C., Li, K.H., Zhang, H., Wagstaff, J.M., Alhawaj, F., Ng, T.K., Speck, J.S., 2020. 7.4-Gbit/s visible-light communication utilizing wavelength-selective semipolar micro-photodetector. IEEE Photon. Technol. Lett. 32, 767–770. Armstrong, J., 2009. OFDM for optical communications. J. Lightwave Technol. 27, 189–204. Arvanitakis, G.N., Bian, R., McKendry, J.J.D., Cheng, C., Xie, E., He, X.Y., Yang, G., Islim, M.S., Purwita, A.A., Gu, E.D., Haas, H., Dawson, M.D., 2020. Gb/s underwater wireless optical communications using series-connected GaN micro-LED arrays. IEEE Photonics J. 12, 1–10. Asbeck, P., 1977. Self-absorption effects on the radiative lifetime in GaAs-GaAlAs double heterostructures. J. Appl. Phys. 48, 820–822. Bamiedakis, N., McKendry, J.J., Xie, E., Gu, E., Dawson, M.D., Penty, R.V., White, I.H., 2019. Ultra-low cost high-density two-dimensional visible-light optical interconnects. J. Lightwave Technol. 37, 3305–3314. Carreira, J.F.C., Xie, E., Bian, R., Herrnsdorf, J., Haas, H., Gu, E., Strain, M.J., Dawson, M.D., 2020. Gigabit per second visible light communication based on AlGaInP red micro-LED micro-transfer printed onto diamond and glass. Opt. Express 28, 12149–12156. Chi, N., 2018. LED-Based Visible Light Communications, first ed. Tsinghua University Press, Beijing. Chu, H., Rajbhandari, S., Faulkner, G., Tsonev, D., Xie, E., McKendry, J.J.D., Gu, E., Dawson, M.D., 2016. LED based wavelength division multiplexed 10 Gb/s visible light communications. J. Lightwave Technol. 34, 3047–3052. Chun, H., Manousiadis, P., Rajbhandari, S., Vithanage, D.A., Faulkner, G., Tsonev, D., McKendry, J.J.D., Videx, S., Xie, E., Gu, E., Dawson, M.D., Haas, H., Turnbull, G.A., Samuel, I.D.W., O’Brien, D.C., 2014. Visible light communication using a blue GaN μLED and fluorescent polymer color converter. IEEE Photon. Technol. Lett. 26, 2035–2038.

318

P. Tian et al.

Dinh, D.V., Quan, Z., Roycroft, B., Parbrook, P.J., Corbett, B., 2016. GHz bandwidth semipolar (112¯ 2) InGaN/GaN light-emitting diodes. Opt. Lett. 41, 5752–5755. Duntley, S.Q., 1963. Light in the sea. J. Opt. Soc. Am. A 53, 214–233. Elgala, H., Mesleh, R., Haas, H., 2011. Indoor optical wireless communication: potential and state-of-the-art. IEEE Commun. Mag. 49, 56–62. Fath, T., Heller, C., Haas, H., 2013. Optical wireless transmitter employing discrete power level stepping. J. Lightwave Technol. 31, 1734–1743. Ferreira, R.X., Xie, E., McKendry, J.J., Rajbhandari, S., Chun, H., Faulkner, G., Watson, S., Kelly, A.E., Gu, E., Penty, R.V., White, I.H., O’Brien, D.C., Dawson, M.D., 2016. High bandwidth GaN-based micro-LEDs for multi-Gb/s visible light communications. IEEE Photon. Technol. Lett. 28, 2023–2026. Foucher, C., Islim, M.I., Guilhabert, B.J.E., Videv, S., Rajbhandari, S., Gomez Diaz, A., Chun, H., Vithanage, D.A., Turnbull, G.A., Samuel, I.D.W., Faulkner, G., O’Brien, D.C., Haas, H., Laurand, N., Dawson, M.D., 2018. Flexible glass hybridized colloidal quantum dots for Gb/s visible light communications. IEEE Photonics J. 10, 1–12. Fox, M., 2010. Optical Properties of Solids, second ed. Oxford University Press, Oxford. Gong, Z., Jin, S., Chen, Y., McKendry, J.J.D., Massoubre, D., Watson, I.M., Gu, E., Dawson, M.D., 2010. Size-dependent light output, spectral shift, and self-heating of 400 nm InGaN light-emitting diodes. J. Appl. Phys. 107, 013103. Green, R.P., McKendry, J.J.D., Massoubre, D., Gu, E., Dawson, M.D., Kelly, A.E., 2013. Modulation bandwidth studies of recombination processes in blue and green InGaN quantum well micro-light-emitting diodes. Appl. Phys. Lett. 102, 091103. Griffiths, A.D., Islim, M.S., Herrnsdorf, J., McKendry, J.J., Henderson, R., Haas, H., Gu, E., Dawson, M.D., 2017. CMOS-integrated GaN LED array for discrete power level stepping in visible light communications. Opt. Express 25, A338–A345. Griffiths, A.D., Herrnsdorf, J., Strain, M.J., Dawson, M.D., 2019. Scalable visible light communications with a micro-LED array projector and high-speed smartphone camera. Opt. Express 27, 15585–15594. Griffiths, A.D., Herrnsdorf, J., McKendry, J.J.D., Strain, M.J., Dawson, M.D., 2020. Gallium nitride micro-light-emitting diode structured light sources for multi-modal optical wireless communications systems. Phil. Trans. R. Soc. A 378, 20190185. Guilhabert, B., Elfstrom, D., Kuehne, A.J.C., Massoubre, D., Zhang, H.X., Jin, S.R., Mackintosh, A.R., Gu, E., Pethrick, R.A., Dawson, M.D., 2008. Integration by self-aligned writing of nanocrystal/epoxy composites on InGaN micropixelated light-emitting diodes. Opt. Express 16, 18933–18941. Haas, H., Yin, L., Wang, Y., Chen, C., 2016. What is LiFi? J. Lightwave Technol. 34, 1533–1544. He, X., Xie, E., Islim, M.S., Purwita, A.A., McKendry, J.J., Gu, E., Haas, H., Dawson, M.D., 2019. 1 Gbps free-space deep-ultraviolet communications based on III-nitride micro-LEDs emitting at 262 nm. Photonics Res. 7, B41–B47. Ho, K.T., Chen, R., Liu, G., Shen, C., Holguin-Lerma, J., Al-Saggaf, A.A., Ng, T.K., Alouini, M.-S., He, J.-H., Ooi, B.S., 2018. 3.2 gigabit-per-second visible light communication link with InGaN/GaN MQW micro-photodetector. Opt. Exp. Dermatol. 26, 3037–3045. Huang Chen, S.W., Huang, Y.M., Chang, Y.H., Lin, Y., Liou, F.J., Hsu, Y.C., Song, J., Choi, J., Chow, C.W., Lin, C.C., Horng, R.H., Chen, Z., Han, J., Wu, T., Kuo, H.C., 2020. High-bandwidth green semipolar (20  21) InGaN/GaN micro light-emitting diodes for visible light communication. ACS Photonics 7, 2228–2235. Islim, M.S., Ferreira, R.X., He, X., Xie, E., Videv, S., Viola, S., Watson, S., Bamiedakis, N., Penty, R.V., White, I.H., Kelly, A.E., Gu, E., Haas, H., Dawson, M.D., 2017. Towards 10 Gb/s orthogonal frequency division multiplexing-based visible light communication using a GaN violet micro-LED. Photonics Res. 5, A35–A43.

Micro-LED based optical wireless communications systems

319

Kim, T.H., Cho, K.S., Lee, E.K., Lee, S.J., Chae, J., Kim, J.W., Kim, D.H., Kwon, J.Y., Amaratunga, G., Lee, S.Y., Choi, B.L., Kuk, Y., Kim, J.M., Kim, K., 2011. Full-colour quantum dot displays fabricated by transfer printing. Nat. Photonics 5, 176–182. Kim, J.Y., Voznyy, O., Zhitomirsky, D., Sargent, E.H., 2013. 25th anniversary article: colloidal quantum dot materials and devices: a quarter-century of advances. Adv. Mater. 25, 4986–5010. Kojima, K., Yoshida, Y., Shiraiwa, M., Awaji, Y., Yamamoto, N., Chichibu, S., 2018. 1.6-Gbps LED-based ultraviolet communication at 280 nm in direct sunlight. In: 2018 European Conference on Optical Communication (ECOC). IEEE, pp. 1–3. Laurand, N., Guilhabert, B., McKendry, J., Kelly, A.E., Rae, B., Massoubre, D., Gong, Z., Gu, E., Henderson, R., Dawson, M.D., 2012. Colloidal quantum dot nanocomposites for visible wavelength conversion of modulated optical signals. Opt. Mater. Express 2, 250–260. Lee, Y., Yang, Z., Chen, P., Hsieh, Y., Yao, Y., Liao, M., Lee, M., Wang, M., Hwang, J., 2014. Monolithic integration of GaN-based light-emitting diodes and metal-oxidesemiconductor field-effect transistors. Opt. Express 22, A1589–A1595. Leitao, M.F., Santos, J.M.M., Guilhabert, B., Watson, S., Kelly, A.E., Islim, M.S., Haas, H., Dawson, M.D., Laurand, N., 2017a. Gb/s visible light communications with colloidal quantum dot color converters. IEEE J. Sel. Top. Quantum Electron. 23, 1–10. Leitao, M.F., Islim, M.S., Yin, L., Viola, S., Watson, S., Kelly, A.E., Li, X., Yu, D., Zeng, H., Videv, S., Haas, H., Gu, E., Laurand, N., Dawson, M.D., 2017b. MicroLED-pumped perovskite quantum dot color-converter for visible light communications. In: 2017 IEEE Photonics Conference (IPC), pp. 69–70. Leitao, M.F., Islim, M.S., Yin, L., Viola, S., Watson, S., Kelly, A.E., Dong, Y., Li, X., Zeng, H., Videv, S., Haas, H., Gu, E., Watson, I.M., Laurand, N., Dawson, M.D., 2019. Pump-power-dependence of a CsPbBr3-in-Cs4PbBr6 quantum dot color converter. Opt. Mater. Express 9, 3504–3518. Li, X., Bamiedakis, N., Wei, J., McKendry, J.J., Xie, E., Ferreira, R., Gu, E., Dawson, M.D., Penty, R.V., White, I.H., 2015. μLED-based single-wavelength bi-directional POF link with 10 Gb/s aggregate data rate. J. Lightwave Technol. 33, 3571–3576. Lin, H.Y., Sher, C.W., Hsieh, D.H., Chen, X.Y., Chen, H.M.P., Chen, T.-M., Lau, K.M., Chen, C.H., Lin, C.C., Kuo, H.C., 2017. Optical cross-talk reduction in a quantumdot-based full-color micro-light-emitting-diode display by a lithographic-fabricated photoresist mold. Photonics Res. 5, 411–416. Liu, X., Tian, P., Wei, Z., Yi, S., Huang, Y., Zhou, X., Qiu, Z., Hu, L., Fang, Z., Cong, C., Zheng, L., Liu, R., 2017. Gbps long-distance real-time visible light communications using a high-bandwidth GaN-based micro-LED. IEEE Photonics J. 9, 1–9. Liu, X., Lin, R., Chen, H., Zhang, S., Qian, Z., Zhou, G., Chen, X., Zhou, X., Zheng, L., Liu, R., Tian, P., 2019. High-bandwidth InGaN self-powered detector arrays toward MIMO visible light communication based on micro-LED arrays. ACS Photonics 6, 3186–3195. Maaskant, P.P., Shams, H., Akhter, M., Henry, W., Kappers, M.J., Zhu, D., Humphreys, C.J., Corbett, B., 2013. High-speed substrate-emitting micro-lightemitting diodes for applications requiring high radiance. Appl. Phys. Express 6, 022102. Manousiadis, P.P., Yoshida, K., Turnbull, G.A., Samuel, I.D.W., 2020. Organic semiconductors for visible light communications. Phil. Trans. R. Soc. A. 378, 20190186. McKendry, J.J., Green, R.P., Kelly, A.E., Gong, Z., Guilhabert, B., Massoubre, D., Gu, E., Dawson, M.D., 2010. High-speed visible light communications using individual pixels in a micro light-emitting diode array. IEEE Photon. Technol. Lett. 22, 1346–1348. McKendry, J.J., Massoubre, D., Zhang, S., Rae, B.R., Green, R.P., Gu, E., Henderson, R.K., Kelly, A.E., Dawson, M.D., 2011. Visible-light communications using a CMOS-controlled micro-light-emitting-diode array. J. Lightwave Technol. 30, 61–67.

320

P. Tian et al.

Mei, S., Liu, X., Zhang, W., Liu, R., Zheng, L., Guo, R., Tian, P., 2018. High-bandwidth White-light system combining a micro-LED with perovskite quantum dots for visible light communication. ACS Appl. Mater. Interfaces 10, 5641–5648. Monavarian, M., Rashidi, A., Aragon, A.A., Oh, S.H., Rishinaramangalam, A.K., DenBaars, S.P., Feezell, D., 2018. Impact of crystal orientation on the modulation bandwidth of InGaN/GaN light-emitting diodes. Appl. Phys. Lett. 112, 041104. Park, J.-S., Kyhm, J., Kim, H.H., Jeong, S., Kang, J., Lee, S., Lee, K.T., Park, K., Barange, N., Han, J., Song, J.D., Choi, W.K., Han, I.K., 2016. Alternative patterning process for realization of large-area, full-color, active quantum dot display. Nano Lett. 16, 6946–6953. Piquette, A.P., Hannah, M.E., Mishra, K.C., 2012. An investigation of selfabsorption and corresponding spectral shift in phosphors. ECS Trans. 41, 1–9. Rabouw, F.T., Donega, C.D.M., 2016. Excited-state dynamics in colloidal semiconductor nanocrystals. Top. Curr. Chem. 374, 58–87. Rae, K., Foucher, C., Guilhabert, B., Islim, M.S., Yin, L., Zhu, D., Oliver, R.A., Wallis, D.J., Haas, H., Laurand, N., Dawson, M.D., 2017. InGaN μLEDs integrated onto colloidal quantum dot functionalized ultra-thin glass. Opt. Express 25, 19179–19184. Rajbhandari, S., Jalajakumari, A.V., Chun, H., Faulkner, G., Cameron, K., Henderson, R., Tsonev, D., Haas, H., Xie, E., McKendry, J.J.D., Herrnsdorf, J., Ferreira, R., Gu, E., Herrnsdorf, J., Dawson, M.D., O’Brien, D.C., 2017a. A multigigabit per second integrated multiple-input multiple-output VLC demonstrator. J. Lightwave Technol. 35, 4358–4365. Rajbhandari, S., McKendry, J.J., Herrnsdorf, J., Chun, H., Faulkner, G., Haas, H., Watson, I.M., O’Brien, D., Dawson, M.D., 2017b. A review of gallium nitride LEDs for multi-gigabit-per-second visible light data communications. Semicond. Sci. Technol. 32, 023001. Rashidi, A., Monavarian, M., Aragon, A., Rishinaramangalam, A., Feezell, D., 2018. Nonpolar m-plane InGaN/GaN micro-scale light-emitting diode with 1.5 GHz modulation bandwidth. IEEE Electron Device Lett. 39, 520–523. Saha, N., Ifthekhar, M.S., Le, N.T., Jang, Y.M., 2015. Survey on optical camera communications: challenges and opportunities. IET Optoelectron. 9, 172–183. Santos, J.M., Jones, B.E., Schlosser, P.J., Watson, S., Herrnsdorf, J., Guilhabert, B., McKendry, J.J.D., De Jesus, J., Garcia, T.A., Tamargo, M.C., Kelly, A.E., Hastie, J. E., Laurand, N., Dawson, M.D., 2015. Hybrid GaN LED with capillary-bonded II–VI MQW color-converting membrane for visible light communications. Semicond. Sci. Technol. 30, 035012. Santos, J.M.M., Rajbhandari, S., Tsonev, D., Chun, H., Guilhabert, B., Krysa, A.B., Kelly, A.E., Haas, H., O’Brien, D.C., Laurand, N., Dawson, M.D., 2016. Visible light communication using InGaN optical sources with AlInGaP nanomembrane downconverters. Opt. Express 24, 10020–10029. Schiavon, D., Binder, M., Loeffler, A., Peter, M., 2013. Optically pumped GaInN/GaN multiple quantum wells for the realization of efficient green light-emitting devices. Appl. Phys. Lett. 102, 113509. Smith, A.M., Nie, S., 2010. Semiconductor nanocrystals: structure, properties, and band gap engineering. Acc. Chem. Res. 43, 190–200. Tian, P., McKendry, J.J., Gong, Z., Zhang, S., Watson, S., Zhu, D., Watson, I.M., Gu, E., Kelly, A.E., Humphreys, C.J., Dawson, M.D., 2014. Characteristics and applications of micro-pixelated GaN-based light emitting diodes on Si substrates. J. Appl. Phys. 115, 033112. Tian, P., Liu, X., Yi, S., Huang, Y., Zhang, S., Zhou, X., Hu, L., Zheng, L., Liu, R., 2017. High-speed underwater optical wireless communication using a blue GaN-based microLED. Opt. Express 25, 1193–1201.

Micro-LED based optical wireless communications systems

321

Tian, P., Wu, Z., Liu, X., Fang, Z., Zhang, S., Zhou, X., Liu, K., Liu, M., Chen, S., Lee, C., Cong, C., Hu, L., Qiu, Z., Zheng, L., Liu, R., 2018. Large-signal modulation characteristics of a GaN-based micro-LED for Gbps visible-light communication. Appl. Phys. Express 11, 044101. Tian, P., Chen, H., Wang, P., Liu, X., Chen, X., Zhou, G., Zhang, S., Lu, J., Qiu, P., Qian, Z., Zhou, X., Fang, Z., Zheng, L., Liu, R., Cui, X., 2019. Absorption and scattering effects of Maalox, chlorophyll, and sea salt on a micro-LED-based underwater wireless optical communication. Chin. Opt. Lett. 10, 100010. Tsonev, D., Chun, H., Rajbhandari, S., McKendry, J.J., Videv, S., Gu, E., Haji, M., Watson, S., Kelly, A.E., Faulkner, G., Dawson, M.D., Haas, H., O’Brien, D., 2014. A 3-Gb/s single-LED OFDM-based wireless VLC link using a gallium nitride μLED. IEEE Photon. Technol. Lett. 26, 637–640. Vithanage, D.A., Kanibolotsky, A.L., Rajbhandari, S., Manousiadis, P.P., Sajjad, M.T., Chun, H., Faulkner, G.E., O’Brien, D.C., Skabara, P.J., Samuel, I.D.W., Turnbull, G.A., 2017. Polymer colour converter with very high modulation bandwidth for visible light communications. J. Mater. Chem. C 5, 8916–8920. Wun, J.M., Lin, C.W., Chen, W., Sheu, J.K., Lin, C.L., Li, Y.L., Bowers, J.E., Shi, J.W., Vinogradov, J., Kruglov, R., Ziemann, O., 2012. GaN-based miniaturized cyan light-emitting diodes on a patterned sapphire substrate with improved fiber coupling for very high-speed plastic optical fiber communication. IEEE Photonics J. 4, 1520–1529. Xie, E., Bian, R., He, X., Islim, M.S., Chen, C., McKendry, J.J.D., Gu, E., Haas, H., Dawson, M.D., 2020. Over 10 Gbps VLC for long-distance applications using a GaN-based series-biased micro-LED array. IEEE Photon. Technol. Lett. 32, 499–502. Yang, W., Zhang, S., McKendry, J.J., Herrnsdorf, J., Tian, P., Gong, Z., Ji, Q., Watson, I.M., Gu, E., Dawson, M.D., Feng, L., Wang, C., Hu, X., 2014. Sizedependent capacitance study on InGaN-based micro-light-emitting diodes. J. Appl. Phys. 116, 044512. Zeng, Z., Fu, S., Zhang, H., Dong, Y., Cheng, J., 2016. A survey of underwater optical wireless communications. IEEE Commun. Surv. Tutorials 19, 204–238. Zhang, S., Watson, S., McKendry, J.J.D., Massoubre, D., Cogman, A., Gu, E., Henderson, R.K., Kelly, A.E., Dawson, M.D., 2013. 1.5 Gbit/s multi-channel visible light communications using CMOS-controlled GaN-based LEDs. J. Lightwave Technol. 31, 1211–1216. Zhang, Y., Jiang, M., Han, T., Xiao, X., Chen, W., Wang, L., Wong, K.S., Wang, R., Wang, K., 2018. Aggregation-induced emission luminogens as color-converters for visible light communication. ACS Appl. Mater. Interfaces 10, 34418–34426. Zhu, S., Chen, X., Liu, X., Zhang, G., Tian, P., 2020. Recent progress in and perspectives of underwater wireless optical communication. Prog. Quantum Electron. 73, 100274.

Further reading Lan, H.Y., Tseng, I.C., Kao, H.Y., Lin, Y.H., Lin, G.R., Wu, C.H., 2018. 752-MHz modulation bandwidth of high-speed blue micro light-emitting diodes. IEEE J. Quantum Electron. 54, 1–6. Lan, H.Y., Tseng, I.C., Lin, Y.H., Lin, G.R., Huang, D.W., Wu, C.H., 2020. High-speed integrated micro-LED array for visible light communication. Opt. Lett. 45, 2203–2206. Liao, C.L., Ho, C.L., Chang, Y.F., Wu, C.H., Wu, M.C., 2014. High-speed light-emitting diodes emitting at 500 nm with 463-MHz modulation bandwidth. IEEE Electron Device Lett. 35, 563–565. Lin, C.H., Tu, C.G., Yao, Y.F., Chen, S.H., Su, C.Y., Chen, H.T., Kiang, Y.W., Yang, C.C., 2016. High modulation bandwidth of a light-emitting diode with surface Plasmon coupling. IEEE Trans. Electron Devices 63, 3989–3995.

CHAPTER TEN

Angular color shift and power consumption of RGB micro-LED displays Fangwang Gou, En-Lin Hsiang, and Shin-Tson Wu* College of Optics and Photonics, University of Central Florida, Orlando, FL, United States *Corresponding author: e-mail address: [email protected]

Contents 1. Angular color shift of RGB micro-LEDs 1.1 Angular color shift 1.2 Emission patterns 1.3 Sidewall emissions 1.4 Color shift 1.5 Angular distribution tolerance 2. Power consumption 2.1 Light extraction efficiency 2.2 Ambient contrast ratio 2.3 Chip-size dependent power efficiency 2.4 Uniform LED chip size in RGB subpixels 2.5 Different LED chip sizes in RGB subpixels References

323 323 324 325 328 332 333 333 335 336 339 341 342

1. Angular color shift of RGB micro-LEDs Sidewall emission of a micro-LED causes mismatched angular distributions between AlGaInP-based red micro-LED and InGaN-based blue/ green counterparts due to material difference, which causes visually noticeable angular color shift of RGB micro-LED displays. This issue is analyzed in this chapter.

1.1 Angular color shift Angular color shift is a critical display metric to achieve supreme image quality at a large viewing angle. For a full-color display consists of individual red, green and blue (RGB) emitters, color shifts originate from two factors. Semiconductors and Semimetals, Volume 106 ISSN 0080-8784 https://doi.org/10.1016/bs.semsem.2020.12.003

Copyright

#

2021 Elsevier Inc. All rights reserved.

323

324

Fangwang Gou et al.

The first one is angular color shifts of RGB primary colors. For instance, the microcavity resonance causes emission spectrum in an organic LED display to shift toward a shorter wavelength as the viewing angle increases, which is known as blue shift in an optical cavity (Kim et al., 2017). However, the primary colors usually account only for a small portion of the displayed images. The majority are those colors created by mixing RGB colors with different ratios. As to the mixed colors, the mismatched angular distributions of RGB emitters can also result in an angular color shift, which is a non-negligible or even more critical issue (Tan et al., 2017). For instance, if the angular distributions of blue and green subpixels decline faster than that of red, the white point of the display would look reddish at a large viewing angle. For a micro-LED, as the chip size down to micron scale, although the internal quantum efficiency may droop due to increased non-radiative recombination from sidewall defects, its light extraction efficiency is improved because the light emission from sidewall gradually increases (Choi et al., 2003; Wang et al., 2009). However, the far-field radiation pattern would deviate from ideal Lambertian distribution, depending on the sidewall emission intensity, which is determined by the refractive index of the employed semiconductor material and device structure. For commercial LEDs, the most commonly used epitaxy wafer for red LED is based on GaInP/AlGaInP multiple quantum wells (MQWs), while blue and green LEDs are based on InGaN/GaN MQWs (Bower et al., 2017; Harbers et al., 2007). Therefore, the angular distribution mismatch among RGB micro-LEDs would occur because the red chip uses different epitaxy materials and has different structures from the green and blue ones. As a result, the angular color shift of mixed colors, such as skin tone, by mixing RGB colors at different ratios may become distinguishable by the human eye (Gou et al., 2019a).

1.2 Emission patterns Fig. 1 shows the emission spectra of RGB micro-LEDs with chip size of 35
60 μm at different viewing angles. For AlGaInP-based red microLED, it has metal contact layer, n-cladding AlGaInP, n-type AlInP diffusion barrier, GaInP/AlGaInP MQWs, p-type AlInP diffusion barrier, p-cladding AlGaInP, and p-GaP window layer (Horng et al., 2018). For InGaN-based green and blue micro-LEDs, their structures consist of metal contact layer, p-type GaN, AlGaN electron block layer, InGaN/GaN MQWs and n-type GaN. For each device, the central wavelength does

Angular color shift and power consumption

325

Fig. 1 Emission spectra of (A) red, (BB) green and (C) blue micro-LEDs at 0°, 30° and 60° viewing angles.

Fig. 2 Far-field emission patterns of RGB micro-LEDs.

not shift as viewing angle increases, indicating the cavity effect inside micro-LEDs is negligible. The layout and refractive indices of epitaxial materials for the blue and green micro-LEDs are similar, but they are quite different from those of red chip (Liu et al., 2010; Moser et al., 1994). Fig. 2 shows the emission patterns of RGB micro-LEDs. As viewing angle increases, the light emission from red chip declines following Lambert’s cosine law. In contrast, for green and blue micro-LEDs, the light intensity gets stronger from normal angle to 40° and then decreases, which are more like Batwing distributions. The mismatched angular distributions originate from different materials of RGB micro-LED chips, which will cause angular color shift of mixed colors.

1.3 Sidewall emissions For a traditional LED with chip size at millimeter scale, the light emission from top surface dominates, which is limited by total internal reflection.

326

Fangwang Gou et al.

The critical angle (θc) between the semiconductor and air interface determines the amount of light escaping from LED and it can be calculated according to Snell’s Law:
 n θc ¼ sin 1 air : (1) n where nair and n represents the refractive index of the air and top semiconductor layer, respectively. If we neglect the Fresnel loss at semiconductor/air interface, the light extraction efficiency η of the LED can be estimated using following equation (Lee, 2001): η¼

Ω , 4π

(2)

where Ω is the solid angle of escape cone, which is expressed as: 2π ð

Ω¼

ðθc dϕ sin θdθ:

0

(3)

0

However, as the chip size shrinks, the sidewall emission from micro-LED should also be taken into consideration. In other words, after photons are generated from the MQWs, they can be extracted from all of six surfaces of the chip. Fig. 3 depicts the total, top and sidewall emissions from RGB micro-LEDs. The top emissions of RGB chips are all Lambertian distributions, but the sidewall emission from green and blue chips are much stronger than that from red one. This difference originates from the stronger absorption of red MQW than that of green and blue MQWs. To analyze the light path inside the micro-LED, we take a point-like source located at MQW layer with coordinate (x, y) as an example (Fig. 4A). The chip size is a
b and the distance of MQW from top surface is h1. On the top surface,

Fig. 3 Total, top, and sidewall emissions of (A) red, (B) green, and (C) blue micro-LEDs at different viewing angles.

Angular color shift and power consumption

327

Fig. 4 (A) Top view of micro-LED chip with a point-like source located at (x, y). (B and C) Side views of light emission from the point source with emission angle θi: (A) θi < θc: top emission; (C) θi > 90°  θc: sidewall emissions.

the light can be extracted for the rays with the incident angle θi smaller than θc, as depicted by the black line in Fig. 4B. A completed escape cone can be obtained because of the short distance h1 between MQW and the semiconductor/air interface. The top emission ratio to the total emission intensity can be calculated using Eq. (2). In the opposite direction, the light goes downward and gets reflected by the bottom electrode pad back to top surface [i.e., yellow line in Fig. 4B], which should be also included as top emission. But for the downward light, both reflectance Rs of bottom metal pad and absorption of MQW need to be taken into consideration during calculation. For the emission toward sidewall, the escape cone may not be completed, depending on the position of point-like sources located at the MQW layer. For example, the point source at a short distance from sidewall

328

Fangwang Gou et al.

can get a completed escape cone. However, as the distance increases, the escape cone will become uncompleted because of the small thickness of micro-LED chip (20%) occurs at red color, which means the power saving is image content dependent. For instance, if the image content is rich in red, the power saving will be more obvious.

References Ahn, H.A., Hong, S.K., Kwon, O.K., 2018. An active matrix micro-pixelated LED display driver for high luminance uniformity using resistance mismatch compensation method. IEEE Trans. Circuits Syst. Express Briefs 65, 724–728. Biwa, G., Doi, M., Yasuda, A., Kadota, H., 2019. Technologies for the crystal LED display system. SID Symp. Dig. Tech. Pap. 50, 121–124. Bower, C.A., Meitl, M.A., Raymond, B., Radauscher, E., Cok, R., Bonafede, S., Gomez, D., Moore, T., Prevatte, C., Fisher, B., Rotzoll, R., Melnik, G.A., Fecioru, A., Trindade, A.J., 2017. Emissive displays with transfer-printed assemblies of 8 μm
15 μm inorganic light-emitting diodes. Photonics Res. 5, A23–A29. Bulashevich, K.A., Karpov, S.Y., 2016. Impact of surface recombination on efficiency of III-nitride light-emitting diodes. Phys. Status Solidi RRL 10, 480–484. Chen, H., Tan, G., Wu, S.T., 2017. Ambient contrast ratio of LCDs and OLED displays. Opt. Express 25, 33643–33656. Choi, H.W., Jeon, C.W., Dawson, M.D., Edwards, P.R., Martin, R.W., Tripathy, S., 2003. Mechanism of enhanced light output efficiency in InGaN-based microlight emitting diodes. J. Appl. Phys. 93, 5978–5982. Choi, H.W., Liu, C., Gu, E., McConnell, G., Girkin, J.M., Watson, I.M., Dawson, M.D., 2004. GaN micro-light-emitting diode arrays with monolithically integrated sapphire microlenses. Appl. Phys. Lett. 84, 2253–2255.

Angular color shift and power consumption

343

Chu, C.-H., Wu, F., Sun, S., 2018. High PPI micro-LED display based on PWM technology. SID Symp. Dig. Tech. Pap. 49, 337–338. Ding, X., Tang, Y., Li, Z., Li, J., Xie, Y., Lin, L., 2017. Multichip LED modules with V-groove surfaces for light extraction efficiency enhancements considering roughness scattering. IEEE Trans. Electron Devices 64, 182–188. Dobrowolski, J.A., Sullivan, B.T., Bajcar, R.C., 1992. Optical interference, contrastenhanced electroluminescent device. Appl. Optics 31, 5988. Fairchild, M.D., 2013. Color Appearance Models, third ed. Wiley. Gong, Z., Jin, S., Chen, Y., McKendry, J., Massoubre, D., Watson, I.M., Dawson, M.D., 2010. Size-dependent light output, spectral shift, and self-heating of 400 nm InGaN light-emitting diodes. J. Appl. Phys. 107, 01313. Gou, F., Hsiang, E.L., Tan, G., Chou, P.T., Li, Y.Y., Lan, Y.F., Wu, S.T., 2019a. Angular color shift of micro-LED displays. Opt. Express 27, A746–A757. Gou, F., Hsiang, E.L., Tan, G., Lan, Y.F., Tsai, C.Y., Wu, S.T., 2019b. Tripling the optical efficiency of color-converted micro-LED displays with funnel-tube array. Crystals 9, 39. Harbers, G., Bierhuizen, S.J., Krames, M.R., 2007. Performance of high power light emitting diodes in display illumination applications. J. Disp. Technol. 3, 98–109. Horng, R.H., Chien, H.Y., Chen, K.Y., Tseng, W.Y., Tsai, Y.T., Tarn Tair, F.G., 2018. Development and fabrication of AlGaInP-based flip-chip micro-LEDs. IEEE J. Electron Devi. Soc. 6, 475–479. Hsiang, E.L., He, Z., Huang, Y., Gou, F., Lan, Y.F., Wu, S.T., 2020. Improving the power efficiency of micro-LED displays with optimized LED chip sizes. Crystals 10, 494. Huang, Y., Tan, G., Gou, F., Li, M.C., Lee, S.L., Wu, S.T., 2019. Prospects and challenges of mini-LED and micro-LED displays. J. Soc. Inf. Disp. 27, 387–401. Huang, Y., Hsiang, E.L., Deng, M.Y., Wu, S.T., 2020. Mini-LED, Micro-LED and OLED displays: present status and future perspectives. Light: Sci. Appl. 9, 105. Hunt, R., Pointer, M., 2011. Measuring Colour, fourth ed. John Wiley & Sons. Kao, C.C., Kuo, H.C., Huang, H.W., Chu, J., Peng, Y.C., Hsieh, Y.L., Luo, C.Y., Wang, S.C., Yu, C.C., Lin, C.F., 2005. Light-output enhancement in a nitride-based light-emitting diode with 22° undercut sidewalls. IEEE Photon. Technol. Lett. 17, 19–21. Kim, H., Baik, K.H., Cho, J., Kim, K.K., Lee, S.N., Sone, C., Park, Y., Seong, T.Y., 2007. Enhanced light output of GaN-based light-emitting diodes by using omnidirectional sidewall reflectors. IEEE Photon. Technol. Lett. 19, 1562–1564. Kim, E., Chung, J., Cho, H., Cho, N.S., Yoo, S., 2017. A systematic approach to reducing angular color shift in cavity-based organic light-emitting diodes. Org. Electron. 48, 348–356. Konoplev, S.S., Bulashevich, K.A., Karpov, S.Y., 2018. From large-size to micro-LEDs: scaling trends revealed by modeling. Phys. Status Solidi A 215, 1700508. Lee, S.J., 2001. Analysis of light-emitting diodes by Monte Carlo photon simulation. Appl. Optics 40, 1427–1437. Lee, J.H., Zhu, X., Lin, Y.H., Lin, T.C., Hsu, H.C., Lin, H.Y., Wu, S.T., 2005. High ambient-contrast-ratio display using tandem reflective liquid crystal display and organic light-emitting device. Opt. Express 13, 9431–9438. Lee, J.S., Lee, J., Kim, S., Jeon, H., 2008. GaN light-emitting diode with deep-angled mesa sidewalls for enhanced light emission in the surface-normal direction. IEEE Trans. Electron Devices 55, 523–526. Lin, C., Fang, Y., Kao, M., Huang, P., Chang, F., Yang, L., Wu, C., 2018. Ultra-fine pitch thin-film micro LED display for indoor applications. SID Int. Symp. Digest Tech. Pap. 49, 782–785. Liu, Z., Wang, K., Luo, X., Liu, S., 2010. Precise optical modeling of blue light-emitting diodes by Monte Carlo ray-tracing. Opt. Express 18, 9398–9412.

344

Fangwang Gou et al.

McCamy, C., Marcus, H., Davidson, J.G., 1976. A color-rendition chart. J. Appl. Photogr. Eng. 2, 95–99. Moser, M., Winterhoff, R., Geng, C., Queisser, I., Scholz, F., D€ ornen, A., 1994. Refractive index of (Alx Ga1x)0.5 In0.5P grown by metalorganic vapor phase epitaxy. Appl. Phys. Lett. 64, 235–237. Oh, J.T., Lee, S.Y., Moon, Y.T., Moon, J.H., Park, S., Hong, K.Y., Song, K.Y., Oh, C., Shim, J.I., Jeong, H.H., Song, J.O., Amano, H., Seong, T.Y., 2018. Light output performance of red AlGaInP-based light emitting diodes with different chip geometries and structures. Opt. Express 26, 11194–11200. Ohmae, A., Kataoka, Y., Nishinaka, I., Ohasha, T., Sakoda, G., Biwa, G., 2019. Light Emitting Apparatus, Light Emitting Unit, display Apparatus, Electronic Device and Light Emitting Element. US Patent 10, 168,004 B2. Olivier, F., Daami, A., Licitra, C., Templier, F., 2017. Shockley-Read-Hall and Auger non-radiative recombination in GaN based LEDs: a size effect study. Appl. Phys. Lett. 111, 022104. Pyo, B., Joo, C., Kim, H., Kwon, B., Lee, J., Lee, J., Suh, M.C., 2016. A nanoporous polymer film as a diffuser as well as a light extraction component for top emitting organic light emitting diodes with a strong microcavity structure. Nanoscale 8, 8575–8582. Singh, R., Narayanan Unni, K.N., Solanki, A., Deepak, 2012. Improving the contrast ratio of OLED displays: an analysis of various techniques. Opt. Mater. 34, 716–723. Tan, G., Lee, J.H., Lin, S.G., Zhu, R., Choi, S.H., Wu, S.T., 2017. Analysis and optimization on the angular color shift of RGB OLED displays. Opt. Express 25, 33629–33642. Tian, P., McKendry, J.J., Gong, Z., Guilhabert, B., Watson, I.M., Gu, E., Chen, Z., Zhang, G., Dawson, M.D., 2012. Size-dependent efficiency and efficiency droop of blue InGaN micro-light emitting diodes. Appl. Phys. Lett. 101, 231110. Tomoda, K., Hirao, N., Biwa, G., 2014. Light Emitting Unit and Display Device. US Patent 8, 686, 447 B2. Wang, X.H., Fu, W.Y., Lai, P.T., Choi, H.W., 2009. Evaluation of InGaN/GaN light-emitting diodes of circular geometry. Opt. Express 17, 22311–22319. Wong, M.S., Nakamura, S., DenBaars, S.P., 2020. Review-progress in high performance III-nitride micro-light-emitting diodes. ECS J. Solid State Sci. Technol. 9, 015012. Zhao, Y., Tanaka, S., Pan, C.-C., 2013. Green semipolar (202¯ 1¯ ) InGaN light-emitting diodes with small wavelength shift and narrow spectral linewidth. Appl. Phys. Express 6, 062102. Zhu, Y., Xu, C., Da, X., Liang, T., Zhang, J., Shen, G., 2007. Enhanced output of flip-chip light-emitting diodes with a sidewall reflector. Solid State Electron. 51, 674–677.

CHAPTER ELEVEN

Monolithic integration of AlGaInP red and InGaN blue/green LEDs Dong-Seon Leea,* and Sang Hyeon Kimb,* a

School of Electrical Engineering and Computer Science, Gwangju Institute of Science and Technology, Gwangju, Korea School of Electrical Engineering, Korea Advanced Institute of Science and Technology (KAIST), Yuseong-gu, Daejeon, Republic of Korea *Corresponding authors: e-mail address: [email protected]; [email protected] b

Contents 1. Introduction 2. Multi-color integration by wafer bonding 3. Toward high-resolution microLED display 4. Conclusion References

345 354 369 383 385

1. Introduction The display industry has developed many technologies to provide a better visual display, which is a ubiquitous device in our lives. Shapes, form factors, and functions have changed and the performance has improved significantly in meeting the demands of the end users. Recently, the microLED display has been actively considered as a next-generation self-emitting display. Many industrial and research players such as Samsung, X-celeprint, LETI, Sony, PlayNitdie, etc. have developed and demonstrated microLED displays (Huang et al., 2020; Li and Liu, 2020). In the manufacture of displays using microLEDs, specific characteristics, such as brightness, light quantity, efficiency, and life span, are considered important factors. However, the processes associated with reducing the size of LEDs also reduce the efficiency, especially by the sidewall effect (Lin and Jiang, 2020). In addition to these process problems, LEDs with sizes 1/10 of those of existing LEDs should be lightweight enough to be transported in large quantities, quickly and accurately (Lin and Jiang, 2020; Wong et al., 2020). This process is referred to as transfer technology and includes not Semiconductors and Semimetals, Volume 106 ISSN 0080-8784 https://doi.org/10.1016/bs.semsem.2021.01.004

Copyright

#

2021 Elsevier Inc. All rights reserved.

345

346

Dong-Seon Lee and Sang Hyeon Kim

only the technique of separating and moving LEDs from their growth substrates to the desired substrates but also connecting the transferred LEDs to electronic devices such as electrodes or TFTs, of the desired structure. Therefore, when the miniaturized LED can maintain high efficiency by adopting new process technologies, other methods such as separating the manufactured LEDs from their growth substrates as well as picking up, transferring, and connecting the separated LEDs onto the desired substrates or electronic devices are also required. These transfer techniques can be classified into three major methods, as shown in Fig. 1, namely, the pick-and-place method, the monolithic method, and other methods (Virey, 2017a). The technologies can also be distinguished by the resolution of the displays to be manufactured. The pick-and-place method is used to classify technology according to the source of power with which the individual LEDs and one pick-up head are matched to lift the LEDs. The head picks up the individual LEDs and places them onto a desired surface and location, which gives this method advantages: easy manufacturability for full-color displays, applicability to existing TFT backplane structures, and manageable defects and errors. However, the transfer speed and accuracy must be improved for mass production, and disadvantages such as the necessity to develop specific process technologies, equipment, and facilities for lifting individual LEDs must be addressed. Moreover, the currently used technologies have limitations; for example, it is difficult to change the initially set resolution once the process line is established. The current pick-and-place technology is suitable for displays requiring low PPI, and technologies that achieve high throughput must be expressly developed. The transfer method for the monolithic array can be classified into the hybridization and monolithic integration steps. The hybridization method involves lifting a microLED array comprised of a large number of LEDs and connecting the array to the TFTs at previously formed locations. The monolithic integration method is used to fabricate LEDs and then grow other devices such as TFTs onto the previously formed LED layers. Both methods can reliably transfer the initial LED array and pixel pitch of the growth substrate, so they have advantages for implementing the desired pixels on displays depending on the LED spacings and sizes produced on the growth substrate; hence, these methods are good for application to technologies that require high overall PPIs. However, LEDs that are grown on substrates generally operate at a single wavelength, so they cannot be directly used to realize full-color displays. In applications requiring multicolor light sources, further processes are needed to arrange the phosphor or

Fig. 1 Transfer methods for microLEDs (Choi et al., 2017; Lee, 2019; Virey, 2017a).

348

Dong-Seon Lee and Sang Hyeon Kim

QDs in micro size. Thus, the development of new technologies in related fields is essential. In addition, since large numbers of LEDs are transferred at once, an appropriate temperature is required during the transfer process for melting the solder. This can cause problems such as position mismatch during the transfer between the LEDs and electronic devices due to the differences in the thermal expansion coefficients of the different substrates. Other new technologies, which do not clearly fit into the two methods described earlier, have also been developed for mass production and transfer as noted under “other method” in Fig. 1. These technologies have been developed to transfer large numbers of microLEDs rapidly, and they are being studied by various research institutes and organizations. For example, the fluidic method, which is being developed by eLux, is a technology that separates microLEDs individually from their growth substrates and places them into a fluid, which is then poured on the target substrate for placement of the microLEDs in the grooves on the target substrate (Lee, 2019). Meanwhile, the Korea Institute of Machinery and Materials (KIMM) has introduced a roll-to-roll-based mass transcription technology for high-throughput transfer (Choi et al., 2017). Depending on the purpose and product group to which these LEDs are applied, the requirements for the size, intensity, and efficiency of the LEDs are different; thus, we need to consider the species of the light source and the corresponding transfer method from the light source development stage. Further differentiated transfer technology is required based on the process and the electrical structures in the backplane. Therefore, all other subsequent processes need to be changed accordingly, and the level of PPI that is applied to the product must be taken into account before designing the light source. Research institutes and organizations that develop displays using monolithic-based transfer technology provide some of the greatest advantages, namely, mass production and high resolution. As shown in Fig. 1, the monolithic transfer technology can be classified into hybridization and monolithic integration, and both technologies involve considerable efforts to improve accurate connections between the driver ICs and LEDs. Since the monolithic method transfers the light source of a single wavelength, as noted above, the technology for fabricating full-color displays by applying various color converters must be supported. In this process, when light passes through the filter or color converters, light loss may occur, and several studies are underway to solve this problem as depicted in Fig. 2 (Han et al., 2015; Templier et al., 2016; Tull et al., 2015). The proposed solutions

Fig. 2 Monolithic transfer technologies (Han et al., 2015; Templier et al., 2016; Tull et al., 2015).

350

Dong-Seon Lee and Sang Hyeon Kim

include techniques for not only making micro-sized RGB arrays using conventional single-color LEDs with QDs or using phosphor but also stacking the RGB LEDs to increase the amount of light intensity, which can be weakened due to a decreased size of the LEDs (Kim et al., 2019). Osram’s blue-LED/red-phosphor/green LED stacking technology is expected to solve some of the electrical connection problems caused by differences in the red and green/blue growth substrates. This method has the advantage of expressing all colors through structural adjustments, but there are problems associated with the difference in lifetimes of the LEDs due to differences in the heat resistance characteristics between the phosphor and LEDs, which make it difficult to control the individual RGB LEDs (Virey, 2017b). Fig. 3B illustrates the technology developed by Aledia, which produces microLEDs by applying nano-rod or micro-rod LEDs and has been studied for a long time now (Gilet and Robin, 2018). The LEDs of this structure, which have been studied traditionally, can be manufactured at a size level of less than several microns and have 3D shapes, so they can propagate light in the desired directions according to the arrangement of the reflective electrodes. Although various problems, such as difficulty in forming the upper electrodes, reduced light intensity, and insufficient full-color generation from a single light source, still remain, the matured knowledge acquired during the long development should be a strong advantage of this approach. Furthermore, this technology has additional advantages, such as capability for growth on Si substrates, which enable cheaper fabrication and easier electrical connections to Si-based driver ICs, such as TFTs and CMOS. These rod-based LEDs can be transferred in large quantities using roll-to-roll technology, and it is expected that LEDs of different colors can be transferred to flexible substrates and stacked to realize RGB stacking. Ostendo’s method involves metal wafer bonding to stack each RGB microLED on a Si CMOS driver and then form a photonic structure to extract light through waveguide cladding (El-Ghoroury et al., 2015). This approach is expected to allow a large amount of emitted light because the size of each pixel is similar to that of the whole pixel LED; however, forming a photonic structure (hole) in the stacked structure for the light extraction would remain as a challenging task. Furthermore, not only is the manufacturing process complicated, but a limited area for the light extraction by the waveguide can lower the extracted light power. These efforts to develop stacked microLEDs are proceeding more actively with the confirmation of their potential as a breakthrough for

Fig. 3 (A) Osram’s blue-LED-phosphor-green LED-stacking technology; (B) Aledia’s GaN on a Si 3D microwire LED; (C) Ostendo’s 3D integrated RGB structure (El-Ghoroury et al., 2015; Gilet and Robin, 2018; Virey, 2017b).

352

Dong-Seon Lee and Sang Hyeon Kim

low optical output power due to their small size, starting with research to combine single light sources with driver ICs such as TFT and CMOS. Stacked RGB microLED-based displays are being actively researched as these structures are likely to be a breakthrough for addressing the reduced light output problems due to smaller sized LEDs. These bonding technologies include not only bonding using metal but also capping technology using insulators to enable each LED or driver IC to operate independently, wiring design for driving, and connections using solder bumps. Therefore, the optimization of bonding technology to deposit insulators and bonding metals as well as to connect them using pressure or temperature to interconnect GaN-based materials and GaAs-based materials is key to realizing displays with microLEDs. In particular, these technologies should be considered for manufacturing microLED displays based on stacking technology. As shown in Fig. 4, several research groups have developed technologies for bonding the LEDs with driver ICs by realizing bump structures using capillary structures, fixing the LEDs to the driver ICs using an insulator, and isolating the LEDs by depositing the insulator (Chun et al., 2014; Geum et al., 2019; http://www.monolithic3d.com/ 3d-micro-displays.html). These efforts have led to the development of LED bonding technologies with various materials, as shown in Fig. 4. This excellent technology is important to realizing stacked LEDs, so it has been studied for display manufacturing using LEDs and has attracted great attention in studies involving dual- or full-color LED displays. As shown in Fig. 4, some technologies insert electrode materials through holes after stacking RGB LEDs (Fig. 4A), achieve electrode bonding using conductive alloys such as ITO (Fig. 4B) and bonding materials (Fig. 4C). Typical stacking technologies stack the LEDs by applying pressure to the individual LEDs with different colors with metal and/or oxide bonding materials. These processes can provide a strong bonding interface and good electrical interconnections, but there are several potential problems such as the damage during the process due to the large thermal and/or physical stress. Therefore, to mitigate this, there has been many studies to realize the high-performance stacked LEDs and ultimately to fabricate a full-color display based on a stacked LED. In this chapter, we describe and discuss our recent demonstration of multi-color stacked LEDs by adhesive bonding. Furthermore, a potential route to realize ultra-high-resolution displays will be covered with derived technological issues such as pixel—driver integration and reduced efficiency in scaled LEDs.

Fig. 4 (A) RGB stacked device of the monolithic 3D Inc. and (B) schematic of the fabrication process. (C) Study of the fabrication of an ultra-resolution display through wafer bonding by Prof. Kim’s group (Chun et al., 2014; Geum et al., 2019; http://www.monolithic3d.com/ 3d-micro-displays.html).

354

Dong-Seon Lee and Sang Hyeon Kim

2. Multi-color integration by wafer bonding i) Blue-green integration by wafer bonding One of the various bonding methods for color integration is shown in Fig. 5 (Kang et al., 2017a). The passive-matrix based blue-green stacked LEDs with a bonding method was reported using SU-8 as a bonding material. This technology is not a metal or oxide-based bonding technology, so it has the advantage that there is little possibility of damage due to pressure or temperature in the LED itself. The blue and green LEDs are grown on different substrates and fabricated separately. The size of each blue and green LED was 75
75 μm and the pitch size was 100 μm. Passive-matrix microLED arrays were fabricated separately for blue and green using the processes shown in Fig. 5. The separately fabricated blue and green microLED arrays were bonded using SU-8 at 250 °C and 1 kg/cm2 pressure. The fabricated stacked blue-green microLED array was bonded without defects as shown in the SEM image of Fig. 5C. As shown in Fig. 6, an 8
8 microLED array with blue-green stacked structures can operate blue and green LEDs individually or simultaneously, and through this driving method, it not only emits individual blue and green colors but also expresses cyan color. Moreover, the blue and green LED can operate in the desired color and at the desired location. The individual operating of the stacked LEDs shows the advantage of expressing various colors between blue and green through a change in the injection current and voltage as shown in Fig. 7. The ability to control the colors reveals that the stacked microLED structure can be applied in displays. However, since the sapphire substrate of each LED was bonded without removing them, the transmittance was reduced. Moreover, there was transmittance reduction caused by the metal line itself. To solve this problem, it was necessary to optimize the design of metal lines, increase the transmittance of the bonding layer, or reduce the thickness of the DSP sapphire. One of the methods for solving these problems is to remove the growth substrates, and this method is explained later in this chapter. ii) Green-red integration by adhesive bonding The high efficiency red LED manufacturing technology announced to date uses a GaAs substrate, which has a different conductivity from InGaN/ GaN-based blue-green stacked microLEDs with non-conductive sapphire

Fig. 5 (A) Fabrication steps of a vertically-stacked passive-matrix microLED array: (i) ICP etching of GaN for formation of the row and all pixels of the microLED array, (ii) deposition of ITO on p-GaN and deposition of Ti/Au layers on n-GaN, (iii) SU-8 pattern for isolation between the metal electrodes and deposition of Cr/Au layers, (iv) wafer bonding of the blue and green LED chip. (B) microscopy images before (top) and after (bottom) alignment of the blue LED chip stacked on green LED chip. (C) SEM image of the cross-sectional interface between the blue and green LED chip (Kang et al., 2017a).

Fig. 6 EL spectrum and microscope image for (A) only blue emission, (B) only green emission, (C) blue and green emission at different pixel positions, and (D) blue and green emissions at the same pixel position (Kang et al., 2017a).

Fig. 7 The EL spectra measured by changing the duty ratio by Pulse Width Modulation (PWM) for (A) green subpixel and (B) blue subpixels. (C) CIE 1931 x-y chromaticity diagram of the device by various PWM voltages. (D) Series of microscope images at M (1,2) of the microLED array for various applied PWM voltages (Kang et al., 2017a).

358

Dong-Seon Lee and Sang Hyeon Kim

substrates. The differences in the conductivities of the substrate generally disturb the fabrication of a passive matrix display using a red LED. In order to fabricate RGB LED arrays using a passive matrix, different methods have been typically conducted to date: forming a stacked structure after the red LED is bonded to the non-conductive material, giving conductivity to the GaAs substrate itself, or etching the GaAs substrate to reveal the back side of the substrate. Therefore, the green-red microLED stacking process using heterogeneous substrates was performed as shown in Fig. 8, with the aim of solving the problems caused by the conductive differences between substrates (Kang et al., 2017b). The method integrated InGaN/ GaN-based green LEDs and AlGaInP-based red LEDs using bonding methods. To fabricate the stacked green-red LED arrays, a process using a bonding technology is necessary. The nitride-based green LED epi grown by MOCVD was fabricated and the LED was formed through a conventional LED process. The entire green LED was coated with SU-8 as a bonding material, and then the phosphide-based red LED wafer was covered upside down on the green LED structure and the bonding process was conducted. In order to provide a flat and uniform interface by preventing air-void formation at the interface of the two LEDs, the bonded LEDs were placed in a vacuum state for 40 min during the process as shown in Fig. 8C. After all the air in the interface was removed, the LEDs were rigidly bonded through the curing process at 200 °C under atmospheric pressure for 2 h. Then, the GaAs substrate was removed through the etching process using an ammonia/hydrogen peroxide solution. After that, red pixels were formed as shown in Fig. 8E after the photolithography process and electrode formation process. Finally, a green-red stacked LED was fabricated by removing the bonding materials on the pad region. This fabrication method addressed the problem of microLEDs requiring at least two or more transfer steps in the conventional technology. This method also has the advantage of lowering the possibility of yield degradation because the entire thin film is transferred only once and all the pixels are formed only by a lithography process, which is different from the method in which each fabricated pixel is transferred individually to the final substrate. Fig. 9 shows the optical and electrical characteristics of the stacked green-red LEDs. In this case, the bonded red LED is compared with the red LEDs in the wafer state. For the green LED, the characteristics of the green LED before (Fig. 8A) and after bonding are compared. For the red LED, as shown in Fig. 9A, the optical characteristics improved

Fig. 8 Schematic fabrication process of the dual-color LEDs: (A) fabrication of InGaN green LEDs, (B) adhesive bonding of a red LED wafer on the fabricated green LEDs, (C) curing of the adhesive material inserted between the red LED wafer and fabricated green LEDs, (D) GaAs substrate removal by chemical wet etching, (E) mesa pattern of the red LED thin film, and (F) formation of the n- and p- electrodes of the red LEDs (Kang et al., 2017b).

360

Dong-Seon Lee and Sang Hyeon Kim

Fig. 9 Electrical and optical properties of the transferred red and green LED subpixels: (A) current-voltage (I-V) and electrical input power-light output power (EIP-LOP) characteristics and (B) electroluminescence (EL) spectra and microscope images of the transferred red LEDs and those on a wafer, and (C) I-V and EIP-LOP characteristics and (D) EL spectra and microscope images of the green LEDs before and after bonding. All LEDs were operated at an EIP of 20 mW (Kang et al., 2017b).

significantly after the GaAs substrate was removed. This was because the light extraction increased as the GaAs substrate, which absorbed a significant amount of generated light, was removed. However, the electrical characteristics tended to decrease compared to the reference samples through etching and bonding processes. It is not a problem that the LED characteristics themselves deteriorated after the bonding, but a problem that occurs due to the change of contact position coming from the inverted structure. The electrical characteristic degradation due to the inverted structure can be improved through the epi design for the inverted structure. On the other hand, as shown in Fig. 9C for a green LED, there is almost no change in the electrical and optical characteristics after the fabrication process. The green-red stacked LED is driven through four electrodes as shown in Fig. 8F. Fig. 10 reveals that, it can be operated in desired positions of the array individually and have a great advantage when applied in display

Monolithic integration of AlGaInP red and InGaN blue/green LEDs

361

Fig. 10 Performance of additive color mixing on the dual color LEDs realized by adhesive bonding (A) EL spectra of the dual color LEDs for the seven color modes, (B) CIE color coordinates, (C) photographs (left) and microscope images (right) showing colored light emissions (Kang et al., 2017b).

applications because it can express all colors between green and red depending on the applied current and voltage. iii) Blue-green integration by selective area growth Although the above-mentioned stacked technology for blue and green has the advantage that the light intensity is increased because the area of the LED light emitting layer is wider than the structure of the RGB array arranged laterally, it has problems of optical loss due to the SU-8 and metal line. Therefore, the electrical characteristics need to be carefully considered. Fabricating stacked blue-green LEDs using selective area growth (SAG) technology, as shown in Fig. 11, can improve optical loss problems (Kong, 2017). Selective area growth technology is a process technology that forms a mask using a dielectric and regrows a material only in an area not covered with the dielectric mask. This technique is widely used in epitaxial growth methods in a way that is less damaging and contaminating than dry etching. Since the blue and green LED structures use the same material, two-color blue and green LEDs are monolithically integrated on the same sapphire substrate using SAG. The stacked LED structure has a simple fabrication process with good efficiency. A blue-green stacked LED is fabricated by growing the blue LED structure on top of the p-GaN of a general green LED structure. This method has no structure to prevent light from escaping, so there is no factor that causes light reduction, and it has a simple structure and uses the existing LED growth method and process method. In addition, since the structure is formed through regrowth, the yield problem of the previously announced technologies is minimized, but it has a problem in that the performance of the pregrown LED device can be reduced due to the regrowth.

362

Dong-Seon Lee and Sang Hyeon Kim

Fig. 11 (A) Schematic illustration of a multi-junction InGaN/GaN LED (B) SEM image of the final device (C) color changes shown on a CIE 1931 x-y chromaticity diagram from blue (460 nm) to green (520 nm) with first and last points from the independently operated blue and green emission (D) representative EL spectra in a blue, cyan, and green mode at 20 mA current injections (Kong, 2017).

The stacked green-blue LED structure using SAG method is shown in Fig. 11A. First, SiO2 was deposited on top of a grown green LED wafer, and a SAG mask was formed using a photolithography method; then wet etching using a buffered oxide etchant was performed to selectively remove only the part that was to be used to make a blue LED. Therefore, the plasma damage could be minimized by excluding the ICP dry etching process. The blue LED structure was regrown using a SiO2 mask already made on the top of the green LED structure. In this process, the regrowth temperature and thickness of each layer function as important factors. If the growth temperature is too high or the epi layer is too thick, it adversely affects the lower green LED structure, so it is important to optimize the growth temperature and the thickness of the layer. The LED structure has an electrode as shown in Fig. 11A, which can be connected to an external circuit that is operated individually.

Monolithic integration of AlGaInP red and InGaN blue/green LEDs

363

The fabricated blue-green stacked LED can independently emit the green light at the region that does not have regrowth and blue light at the regrowth region. Blue light (460 nm) is emitted from the outer ring part and green light (520 nm) is emitted from the inner circle part. The LED operation of this structure can be independent or simultaneous, as that of the green-red LED in Fig. 11, and an overall color between blue and green can be realized, as shown, depending on the injection current and voltage. Various cyan colors are possible when blue and green are simultaneously emitted. Therefore, it is expected that the stacked structure can be applied to display applications through the independent or simultaneous control of each color. iv) Hybrid RGB integration by SAG and adhesive bonding The RGB LED array was fabricated through the process shown in Fig. 12 by combining the growth technique for the green-blue stacked LED structure and the bonding technique described in Fig. 8. After the SU-8 was spincoated as shown in Fig. 12B on the green-blue stacked LED structure, which was formed by the Fig. 12A process [i], the red LED was bonded and the GaAs substrate of the red LED was removed [ii]. The structure of the red LED was formed through a patterning process (Fig. 13). The electrode and the reflecting layer structure were then formed [iii]. Finally, except for the area for bonding of the red LED, the overall removal of SU-8 [iv] resulted in the structure of the RGB LED array. The LED array can be operated individually by each red, green, and blue LED through five electrodes as shown in Fig. 14A, and various colors can be realized through the control of the injection current and voltage. Fig. 14B also shows an equivalent circuit diagram for operating each RGB LED individually or simultaneously. In this circuit, the part of interest is the circuit part of the green and blue LED. In the case of green and blue LEDs, the circuit part needs to be independently formed because an anode is commonly used and current and voltage are applied. For this purpose, the common anode was connected to the GND (ground) to separate the circuits of the two LEDs. The red LEDs were independently formed with electrodes and formed to have circuits separated from the green and blue LEDs. Finally, this independent circuit structure was made to enable each LED to be driven independently or simultaneously as shown in the bottom picture of Fig. 14B, which enabled the independent control of the amount of current and voltage applied to each LED, thereby enabling the desired color to be realized with the desired brightness. The color expression ability of the Hybrid RGB LED array is verified by the CIE color coordinates shown in Fig. 15A and compared with the

Fig. 12 Schematic of the fabrication process for hybrid RGB LEDs: (A) fabrication process of the blue/green dual-color LEDs using selective area growth (SAG), and (B) the process for the formation of the red pixels using adhesive bonding (Kang et al., 2018).

Monolithic integration of AlGaInP red and InGaN blue/green LEDs

365

Fig. 13 Schematic of the hybrid RGB LEDs. Top and cross-sectional views of the final device (left). Microscopic images of the hybrid RGB LEDs in (top to bottom) blue, green, red, and white color modes (right) (Kang et al., 2018).

Fig. 14 (A) Circuit configuration schematic of the hybrid RGB LEDs and (B) equivalent circuit diagrams of the circuit configuration (Kang et al., 2018).

Fig. 15 Color performances of the hybrid RGB LEDs: (A) chromaticity coordinates, (B) photographs for 10 color modes, and (C) EL spectra for two- and three-color mixing (Kang et al., 2018).

Monolithic integration of AlGaInP red and InGaN blue/green LEDs

367

BT2020 color space. BT2020 is a color evaluation category of the UHD TV display. In general, the display company is developing products with 75% of BT2020. For the Hybrid RGB LED array, the color space can express a fairly wide color space at about an 80% level compared to the BT2020 space. In order to express a more abundant color, it is expected that the overall color can be realized by approaching BT2020. The color will be improved when process and epi structure optimizations are carried out to realize the wavelength band of green and blue LEDs with a short wavelength. As described previously, the desired color can be expressed by controlling the injection current and voltage ratio of the three light sources, so that various colors such as white pink can be realized, where such colors cannot be expressed with a monochromatic wavelength. However, there is a problem in that the internal indium intermixing or inter diffusion in the quantum wells of the pregrown LEDs during the regrowth process at high temperature causes deterioration of the characteristics. Therefore, the fabrication technology optimization and epitaxial structure optimization are further required to avoid this problem. v) Full adhesive bonding technique In the regrowth of blue LED on a green LED structure during MOCVD, the regrowth temperature and thickness control are crucial factors in minimizing the influence of the green LED. Even if the blue LED is formed through this optimization, the electrical characteristics of the regrown LED are not good enough because it has a relatively long current path compared to LEDs made by conventional fabrication. Therefore, if we can fabricate the RGB LED pixel array using only a bonding technology, it must have advantages over that using the previous regrowth method. Unlike the full color LED fabrication method using the adhesive bonding technology and SAG method together, the RGB LED array fabrication using only the bonding method does not go through the high temperature growth process again after making the green LED, so it can improve the problem of the degradation of the device through the inter-diffusion of indium. All LEDs used in this technology are from commercialized Epis; in particular blue LED Epi uses GaN on Si epi. As shown in the process flow chart of Fig. 16, first, the bonding material is coated (b) on the conventional green LED array (a), and the Si is removed after bonding with the blue LED wafer (c). Then, the blue LED is formed (d) and red LED wafer is bonded using a bonding material (e), and then the GaAs substrate is removed (f ). Finally, red pixels are formed (g) and bonding materials on green and blue LEDs are removed (h) and red-green-blue stacked LEDs are fabricated.

Fig. 16 Schematic of the fabrication steps for the full-color inorganic LEDs by the adhesive bonding technique (Mun et al., 2021).

Monolithic integration of AlGaInP red and InGaN blue/green LEDs

369

Fig. 17 Schematic of the RGB LEDs by full adhesive bonding. (A) Cross-sectional views of the final device. (B) Optical microscopic image of the RGB LEDs in the white color mode (Mun et al., 2021).

In Fig. 17, you can see the cross-sectional structure of the fabricated RGB LED array using SU-8 (a) and the optical microscopic image of the actual operating LED (b). Each LED is bonded using SU-8, and the insulation between the bonded LEDs is also formed using SU-8. The LEDs have two electrodes for each color, and the red, green, and blue can be controlled independently through a total of six electrodes. The electrical characteristics of the LEDs shown in Fig. 18A, confirm that the series resistance of the red LED is higher than that of the blue and green LEDs. This is due to the low doping concentration of the surface where the metal contact is formed, which is the same reason that was described for the R-G bonding part. The power ratio of three LEDs was controlled to express the desired color, and all three color emissions were very efficient because they used commercial Epi. The color space was 83% compared to the BT2020 color space, and we expect that it can express a fairly wide color space. If the process and epi structure optimizations are performed to realize the wavelength of green LEDs with a shorter wavelength, we expect that a color expression similar to BT2020 will be possible.

3. Toward high-resolution microLED display As described above, a microLED display is composed of several million LEDs with a micrometer size as a pixel and full-color display needs R (red), G (green), B (blue) sub-pixels, which make fabrication difficult. InGaN-based green and blue LEDs can be grown on the same substrates, but typical AlGaInP-based red LED should be grown on GaAs, which is

Fig. 18 Color performances of the RGB LEDs after full adhesive bonding: (A) I-V characteristics of the LEDs by color, (B) chromaticity coordinates and photographs for seven different modes, and (C) EL spectrum of the RGB LEDs in white mode (Mun et al., 2021).

Monolithic integration of AlGaInP red and InGaN blue/green LEDs

371

a different substrate. These things have promoted several kinds of integration technologies reported so far, such as “pick and place technology,” “roll transfer technology,” etc. However, there are still many critical issues in those methods such as limited resolution, transfer yield, throughput, redundancy, and repairing of the pixels after transfer. Therefore, the fabrication of high-resolution microLED pixels is a very challenging task because of the absence of a fabrication technology that enables high resolution. Conventional “pick and place” and “roll transfer” needs alignment between each pixel transfer with the mechanical way, which creates a limitation of the alignment accuracy up to several microns. This alignment issue is not only for fabricating R/G/B pixels, but it is crucial to integrating the pixels with driver integrated circuits (ICs). Furthermore, high-resolution directly indicates that small pixels are needed for the display, which requires another important technology to form a stable sidewall surface. Otherwise, high-efficiency, which is one of the important advantages of an inorganic microLED, cannot be realized due to a large amount of non-radiative recombination at the sidewall surface. Therefore, fabricating a high-resolution display still requires the development of many important technologies. In this chapter, we introduce the vertical pixel integration and impact of the surface preparation on the quantum efficiency of the microLEDs. The conventional fabrication method for a microLED display typically uses pixel transfer techniques, which include various approaches: electrostatic MEMS, elastomer stamping, magnetic stamping, roll transfer, etc. These approaches create mechanical issues, which limit the resolution. On the other hand, we can actually get a hint from the 3D integration technology in the semiconductor industry. There are two different methods of 3D integration: Si via (TSV)-based 3D integration and monolithic 3D (M3D) integration (Geum et al., 2016; Kang et al., 2018; Kim et al., 2018; Panth et al., 2014; Zhang et al., 2018). In the TSV-based approach, two wafers are processed separately and the two wafers are mechanically aligned to make an electrical interconnection between the wafers. Therefore, the alignment accuracy is not that high, so one TSV can make an electrical connection at the block level, which includes approximately 10,000 transistors. This is very similar to the process in conventional microLED display fabrication (pixel by pixel transfer). Another approach of M3D is now a very hot research topic in the semiconductor industry. Different from the TSV-based 3D integration, the bottom-level transistors are processed, and top-level semiconductor active

372

Dong-Seon Lee and Sang Hyeon Kim

layers are formed in M3D integration. Then, top-level transistors are processed with very precise lithographic alignment. In principle, this process sequence provides a device to device level interconnection. Thus, here, this concept can be directly adopted in the microLED display. Therefore, we recently proposed a new approach to fabricate highresolution microLED displays using M3D integration technology. Fig. 19 shows the schematic process flow of the vertically stacked high-resolution microLED display on the driver IC. Here, the process starts from the whole layer transfer on the driver IC. By transferring the red, green, and blue multi-quantum well (MQW), vertically stacked sub-pixels can be realized. Then, we sequentially process the LED for each pixel, and metallization and interconnection between pixels and driver ICs finalize the fabrication process. This is why M3D integration can also be referred to as sequential 3D integration. With this process, very precise physical alignment between the sub-pixels can be realized by lithographic alignment during the fabrication, which has proven the “nanometer scale” alignment in CMOS technologies. Furthermore, this precise alignment is similarly applied to the monolithic integration between pixels and the driver IC, which makes it possible to fabricate high-resolution displays. On the other hand, one of the real challenges in vertically stacked pixels is optical crosstalk between the sub-pixels, as shown in Fig. 20. The light emitted from the top LED (typically blue), which has a larger bandgap than does the bottom LED, can excite the carrier in the bottom MQW or can be absorbed in the highly doped contact layer in bottom LEDs. This optical

Fig. 19 Schematic process flow of the vertically stacked high-resolution microLED display by M3D/sequential 3D process.

Monolithic integration of AlGaInP red and InGaN blue/green LEDs

373

Fig. 20 Technological issue of vertically stacked microLED pixel, which is the optical cross-talk between different sub-pixels.

crosstalk has already been seen in our previous demonstrations (Kang et al., 2017c). In the vertically stacked red/green LEDs, EL spectra of red at around 635 nm were observed when the injection current in green LEDs was sufficiently large. Therefore, it is very important to minimize the optical crosstalk and optical loss between the sub-pixels, especially for the red/green interface because the typical contact layer in a red LED is GaAs, and it shows a non-negligible optical absorption. Therefore, to prevent this optical crosstalk, an optically functional layer that reflects short-wavelength (green and blue) and transmit long-wavelength (red) at the same time is needed, as shown in the right figures in Fig. 20. This filtering layer can be designed at the bonding interface. First, we simply considered the distributed Bragg reflector (DBR) to realize the filtering function. Using pairs of SiO2, and SiNx, we designed the DBR following the equation t ¼ λ/(4nmaterial), where t, λ, and nmaterial are the thickness, target wavelength, and reflective index of the material. With 78 nm-thick SiO2 and 51 nm-thick SiNx, we simulated the spectral reflectance using a transfer matrix method. Fig. 21A shows the simulated optical reflectance of DBR with a different number of pairs. With an increase in the number of pairs in the DBR, the peak reflectance values increased and the peak of the reflectance became narrower. It should be noted that the DBR has a high reflection for green and blue and a high transmission for red. Based on these simulations, we fabricated the DBR on a sapphire substrate by plasma-enhanced chemical vapor deposition. Fig. 21B shows the transmission characteristics of the samples as a function of the wavelength; the transmission characteristics had a trend similar to the simulation. The larger number of pairs of DBRs exhibited steeper transmittance curves at around 400–500 nm, indicating high reflection for blue light. On the other hand, the samples had a quite high transmittance at around 630 nm,

Fig. 21 (A) Simulated reflectance spectra of SiO2/SiNx DBRs with various numbers of pairs (3, 5, 7, 9 pairs). (B) Measured transmittance of the fabricated DBR on a sapphire substrate (Geum et al., 2019).

Monolithic integration of AlGaInP red and InGaN blue/green LEDs

375

indicating high transmittance for the red light. We believe that a similarly designed filtering layer can be introduced at the bonding interface, which will prevent optical crosstalk between the sub-pixels. To see the feasibility of the high-resolution microLED formation by M3D integration, we started to form vertically stacked pixels by the adhesive wafer bonding. We first formed a red LED on top of the green LED in our previous work (Kang et al., 2017c). As shown in Fig. 10, this stacked LED showed very broad coverage of the CIE chart from red to green. More recently, to verify the potential process technology for dense pixel formation, we formed vertically stacked red and blue MQW by the insulator direct wafer bonding, which is a more industrial approach than adhesive bonding. First, we bonded the red MQW on blue MQW/sapphire substrates by wafer bonding. Here, we inserted the DBR layer on a red MQW/GaAs substrate before the wafer bonding. This functioned as a filtering layer to prevent optical crosstalk between red and blue sub-pixels. Then, we formed a very dense pixel array using electron beam lithography and SiCl4-based etching with a Ni/SiO2 hard mask. Fig. 22 shows the cross-sectional TEM image of the fabricated red on a blue LED structure, showing that the red MQW was well bonded on the blue MQW without any voids at the bonding interface. Furthermore, there were DBR layers of multiple pairs of SiO2, and SiNx at the bonding interface. Fig. 23 shows the tilted SEM image of the fabricated dense pixels. Pixels with pitch sizes from 1 μm to 300 nm were well fabricated, and red light emission by optical pumping was confirmed after the pixelization. Here, it should be noted that the smallest pattern was larger than 60,000 PPI. This experimental demonstration directly indicates the potential fabrication of the ultra-high-resolution display.

Fig. 22 Cross-sectional TEM image of the fabricated vertically stacked red on blue LED substrate (Geum et al., 2019).

0.6 µm

0.6 µm

1 µm

1 µm

Optical pumping

Pixel width 1 µm

Pixel width 0.7 µm

Pixel width 0.5 µm

Pixel width 0.3 µm

MQW

Fig. 23 Birds-eye view of the fabricated high-density vertically stacked microLEDs with different pitch sizes (Geum et al., 2019).

Monolithic integration of AlGaInP red and InGaN blue/green LEDs

377

As shown in Fig. 22, DBR can be introduced at the bonding interface of the red on blue stacked LED structure. To evaluate the optical filtering function, we carried out a PL investigation on this sample. Fig. 24 shows the PL intensity of the red LED with optical pumping from the blue substrate side with a different number of DBR pairs. Here, we used a 405 nm laser as a pump laser. A clear PL response at around 630 nm, corresponding to red emission was observed in the sample without DBR because the pump laser can reach red MQW passing through the bonding interface and exciting the carrier. On the other hand, with the increasing number of pairs of DBR, the PL intensity decreased due to the reflection of pump laser on the DBR at the bonding interface. These results strongly suggest that the optical crosstalk between the sub-pixels can be minimized by the optical design of the bonding interface. As described above, M3D integration provided a very good physical alignment between the sub-pixels, but this was not only for the dense pixel fabrication. M3D integration can be also applied to the driver-pixel integration shown in Fig. 19. We recently demonstrated very similar device integration with the driver-pixel integration, which is an III-V imager integrated directly over Si CMOS readout integrated circuitry (ROIC). We consider III-V imager and ROIC as pixel and driver IC, respectively. Fig. 25A shows the cross-sectional TEM image and photographic image of the InGaAs imager on SOI MOSFETs. The photographic image shown in the inset confirmed the InGaAs imager on the bottom SOI MOSFETs. M3D vertical integration of the devices can be clearly seen in the TEM as well. Here, the typical process temperature of the III-V imager is quite low less than 400 °C, so there was no damage in the bottom circuitry during the integration process. This was confirmed in the transfer curves of SOI MOSFETs before and after the imager integration process in Fig. 25B. This strongly suggests that top III-V/bottom Si CMOS is one of the most promising structures to maximize the benefits of M3D integration. To mimic the readout operation, we carried out the electrical interconnection between the top imager and bottom SOI MOSFETs with a typical one transistor—one photodiode (1T-1PD) connection. Then, we observed the readout operation of the integrated PD by measuring the output voltage (Vout) in a 1T-1PD configuration with a changing incident light intensity and gate voltage applied to SOI MOSFETs. Fig. 25C shows the Vout characteristics of the 1T-1PD. Vout gradually decreased with an increase in the light intensity and Vout increased when the transistor was turned on by increasing the gate voltage, which indicated that light signal can be read by this simple circuit configuration realized by M3D integration.

Pump power 1260 μW 720 μW 524 μW 230 μW 51 μW

(c) 7 pairs

Pump power 1260 μW 720 μW 524 μW 230 μW 51 μW

PL intensity (a.u.)

(b) 3 pairs

PL intensity (a.u.)

PL intensity (a.u.)

(a) w/o DBR

Pump power 1260 μW 720 μW 524 μW 230 μW 51 μW

580 600 620 640 660 680

580 600 620 640 660 680

580 600 620 640 660 680

Wavelength (nm)

Wavelength (nm)

Wavelength (nm)

Fig. 24 Photoluminescence spectra with 405 nm pump laser from the fabricated vertically stacked microLEDs without DBR and with 3 pairs, 7 pairs DBRs (Geum et al., 2019).

379

Monolithic integration of AlGaInP red and InGaN blue/green LEDs

A

B

C

1550 nm illumination

0.10

VOUT (V)

0.08 0.06 0.04

VD=0.1V

0.02 VGS=0.4V to 1V

with 25mV step 0.1

1

10

Light intensity (µW) Fig. 25 (A) Cross-sectional TEM of the InGaAs PD on SOI MOSFETs. (B) Transfer curves of the SOI MOSFETs before and after PD integration. (C) Readout operation of InGaAs PD by SOI MOSFET (Geum et al., 2020).

We have not applied this configuration in microLED pixel—driver IC integration, but this technology can be directly transferred to the pixeldriver integration, which will be done soon. On the other hand, another critical challenge for high-resolution microLED displays would be lower quantum efficiency (QE) than the expected value. Fig. 26 shows the typical QE curve of an inorganic LED as a function of the current density. Peak QE appears at quite a high current density region larger than 100 A/cm2. This peak position becomes even higher in a scaled LED due to the increased portion of the non-radiative surface recombination (Chen et al., 2015; Daami et al., 2018; Lee et al., 2011),

380

Dong-Seon Lee and Sang Hyeon Kim

Fig. 26 Typical IQE spectra of the inorganic LED and the preferred current density region by the application.

whereas the desired current density for the high-resolution display is less than10 A/cm2. These results clearly indicate that QE in the targeted current density is very low in a scaled LED, nearly zero in certain cases. Therefore, decreasing the peak position (current density) of QE and/or minimizing the increase of the peak position with pixel scaling is a very important technological issue for high-resolution displays. We suggested two approaches to solve these issues. First, we pulled down the QE peak position by the re-design of the epi-structure. The dominating carrier recombination mechanism at each current density can be different. Before reaching the peak position, Shockley-Read-Hall (SRH) recombination is dominant, which is related to the material quality. At the peak position, radiative recombination is the dominant mechanism. At high a current density range exceeding the peak position, auger recombination is the dominant carrier recombination mechanism. Therefore, to pull down the peak position, the ratio between each carrier recombination mechanism needs to be considered. To do this, we suggested modifying the carrier density distribution by changing the quantum well (QW) structure from an electron dominant QW to a hole dominant QW (Li et al., 2019). Fig. 27 shows the simulated IQE curve of conventional and proposed LED structures. Regardless of the number of QWs, the current density at the peak EQE is quite high in conventional QWs. On the other hand, a high IQE at a low current was achieved in the proposed QW structure by changing the QW structure to have a hole dominant carrier distribution. Another approach to increase the QE at the low current density is to minimize the non-radiative recombination, which is the reason for the

Monolithic integration of AlGaInP red and InGaN blue/green LEDs

Internal quantum efficiency (a.u.)

1.0

381

ΔEc/ΔEv ~0.45 (proposed SQW) ΔEc/ΔEv ~1 (Conventional MQW)

0.8

0.6

ΔEc/ΔEv ~1 (Conventional SQW)

Current density x2500 reduction

0.4

0.2

0.0 10-10 10-8 10-6 10-4 10-2 100 102 104

Current density (A/cm2) Fig. 27 Simulated IQE of conventional LED and low-current specialized LED.

increased current density at the peak QE and the drop of the QE itself in the scale LED. Critical points in the scaled LED would be a significant contribution of the non-radiative surface recombination following the scaling. The increased portion of the surface with respect to the LED area decreases the QE and the current density at the peak QE region. First, to see the impact of the non-radiative recombination center at the sidewall of the scaled LED, we carried out time-resolved photo-luminescence (TRPL) measurement for scaled red LED samples with various mesa sizes. Fig. 28A shows the TRPL decay curve measured from the scaled LEDs with different pitch sizes from 1 to 0.3 μm. Decay time gradually decreased with the dimension scaling of the LED, indicating the clear sidewall contribution on the carrier recombination. To analyze these characteristics, we extracted the effective decay times (τeff) using a linear sum of the weighted triple exponentials. τeff decreased with a scaling of LED width, as shown in Fig. 10B. With a 1/τeff ¼ 1/τbulk + 4
SRV/d relationship, where τbulk, d, and SRV are the carrier lifetime in bulk material, the width of the LED, and the surface recombination velocity, respectively, we extracted an SRV value of 4100 cm/s in our red LED. This value itself is not exceptionally high and a quite moderate value among non-passivated III-V nanowires, whereas the surface with this value significantly impacts the optical properties. Since the SRV can be defined as σvthDit, where σ, vth, and Dit are the capture cross-section, thermal velocity, and interface trap density, respectively,

A

B

C

Fig. 28 (A) Experimental TRPL decay curves measured at room temperature for different μ-LED sizes and fitting results of the decay curve for a 1 μm sized LED, as shown in the inset. (B) Inverse effective decay time as a function of inverse LED widths for calculating the SRV value and fitting parameters for calculating the effective decay times as inserted in the inset table. (C) Comparison of the TRPL decay curves of the bulk-layer red LED layer and unpassivated and passivated 1.0 μm-width LEDs with sulfur passivation and the enhancement ratio shown in the inset (Geum et al., 2019).

Monolithic integration of AlGaInP red and InGaN blue/green LEDs

383

Dit was approximately 7.4
1012 eV1 cm2 in the sample, which is a quite large number. Therefore, to reduce this trap density, we applied sulfur treatment and Al2O3 passivation by atomic layer deposition, which is quite common in MOSFET society (del Alamo et al., 2013; Suzuki et al., 2012; Zhang et al., 2018). Here, sulfur treatment was carried out by dipping the sample in (NH4)xS solutions. The sulfur treatment and Al2O3 passivation produced a remarkable improvement in the TRPL decay curve, as shown in the right figure. The TRPL decay curve was almost recovered by this passivation. As described above, it will be very important to investigate and develop passivation technology to eliminate non-radiative centers at the sidewall of the LED to guarantee a high QE in scaled LEDs.

4. Conclusion In recent years, microLED-based displays have been steadily developing, and one of the technologies, the “RGB LED array monolithic integration” technology, offers considerable prospects. A pick-and-place-based display realizes a display by separating the LED wafers of each color, R, G, and B into a chip and arranging the pixels horizontally through a transfer process. While the transfer method is not yet suitable for use in high resolution applications due to general limitations and mechanical accuracy limitations as well as yield degradation problems, monolithic integration can significantly increase resolution, making it a suitable technology for AR/VR applications. In addition, in a stacked RGB LED array, the area of the active region itself is wider than that in a display manufactured by dividing one pixel area into sub-pixels for each color, so it has the advantage of stronger light emission and more vivid color realization with the same injection current and voltage. In addition, compared to a structure that implements RGB colors using phosphors on top of a single-wavelength LED, individual LEDs can be independently controlled to show rich and unique colors, which is also a striking advantage. Nevertheless, the progress of research on technologies related to monolithic integration is rather slow. In this chapter, to solve the problems caused by this transfer method, we introduced a device manufacturing method using only a semiconductor patterning process after stacking red, green, and blue wafers. We presented a bonding technology for wafer stacking, and a method that realized full color by combining a selective growth method and the bonding technology and a method that realized full color using only the bonding technology. It was integrated on a single wafer without any structural defects, and by controlling

384

Dong-Seon Lee and Sang Hyeon Kim

the optical power ratio of each pixel, the desired color was created with a color space of up to 80% of the BT2020 area. At present, implementation of a stacked LED using such a regrowth and bonding material still requires optimization of structure, epilayer characteristics, and growth processes. It is also necessary to solve the problem of decreasing light emission in some color since the color changes according to the injection amount of current and voltage. In the realization of next-generation low-power, high-efficiency displays, the technology of arranging microLEDs in a stacked structure is a technology that must be further studied. In particular, in the development of displays that require high clarity and rich colors even when the pixel size is small, the technical advantages of such a stacked LED manufacturing technology need to receive much attention. Furthermore, it is also quite important to consider pixel driver IC integration for high-resolution displays. Therefore, we proposed integration technologies using an M3D-like process that we introduced for sub-pixel formation. This process will potentially provide very accurate alignment between pixels and the driver IC and also between sub-pixels as well. To address this issue, we introduced a post-pixelization process after the layer transfer of MQW layers. We demonstrated ultra-high-resolution stacked LEDs exceeding 60,000 PPI estimated from their pattern size. These results strongly suggest that the M3D-like process will be the only technology that supports high-resolution, which cannot be realized by conventional technologies such as “pick and place.” Furthermore, as a reference circuit configuration, we demonstrated 1T-1PD M3D integration showing clear readout operation without any degradation of the bottom Si devices after the integration process. This technology will be applied to the pixel—driver IC integration. In addition to our proof-of-concept demonstration of the stacked LED, light extraction without optical crosstalk between sub-pixels is very important. Therefore, we showed the feasibility of the filtering functionality of the bonding interface using a multi-stacked dielectric forming DBR, which clearly achieved reduced optical crosstalk between blue and red pixels. Lastly, one of the most important issues for high-resolution displays is the realization of high QE at a low current density. Here, in this chapter, we considered two approaches to improve QE. One was to re-design the MQW to have the QE peak at a low current density by changing the dominant carrier in the QW from electron to hole. With this re-designed QW, we achieved remarkable improvement in QE at a low current density. Another approach was to minimize the non-radiative recombination at

Monolithic integration of AlGaInP red and InGaN blue/green LEDs

385

the sidewall surface by passivating the recombination center with sulfur and Al2O3. With a careful investigation by TRPL, we found that this passivation was very effective in recovering the TRPL decay time, which is quite comparable to the bulk reference layer without the sidewall surface. Throughout our experimental demonstration, it should be noted that process development for M3D-like pixel integration and further systematic investigation of the surface passivation must be taken into account for manufacturable high-resolution microLED displays in the future.

References Chen, S.L., Chen, W.M., Ishikawa, F., Buyanova, I.A., 2015. Suppression of non-radiative surface recombination by N incorporation in GaAs/GaNAs core/shell nanowires. Sci. Rep. 5, 1–9. Choi, M., Jang, B., Lee, W., Lee, S., Kim, T.W., Lee, H.J., Kim, J.H., Ahn, J.H., 2017. Stretchable active matrix inorganic light-emitting diode display enabled by overlay-aligned roll-transfer printing. Adv. Funct. Mater. 27, 1606005. Chun, J., Lee, K.J., Leem, Y.-C., Kang, W.-M., Jeong, T., Baek, J.H., Lee, H.J., Kim, B.-J., Park, S.-J., 2014. Vertically stacked color tunable light-emitting diodes fabricated using wafer bonding and transfer printing. ACS Appl. Mater. Interfaces 6, 19482–19487. Daami, A., Olivier, F., Dupre, L., Henry, F., Templier, F., 2018. Electro-optical sizedependence investigation in GaN micro-LED devices. SID Symp. Dig. Tech. 49, 790–793. del Alamo, J.A., Antoniadis, D., Guo, A., Kim, D.-H., Kim, T.-W., Lin, J., Lu, W., Vardi, A., Zhao, X., 2013. InGaAs MOSFETs for CMOS: recent advances in process technology. In: 2013 IEEE International Electron Devices Meeting. pp. 2–1. El-Ghoroury, H.S., Chuang, C.-L., Alpaslan, Z.Y., 2015. Quantum photonic imager (QPI): a novel display technology that enables more than 3D applications. In: SID Symposium Digest of Technical Papers. vol. 26, pp. 371–374, 1. Geum, D.M., Park, M.S., Lim, J.Y., Yang, H.D., Song, J.D., Kim, C.Z., Yoon, E., Kim, S.H., Choi, W.J., 2016. Ultra-high-throughput production of III-V/Si wafer for electronic and photonic applications. Sci. Rep. 6, 1–10. Geum, D.-M., Kim, S.K., Kang, C.-M., Moon, S.-H., Kyhm, J., Han, J., Lee, D.-S., Kim, S., 2019. Strategy toward the fabrication of ultrahigh-resolution micro-LED displays by bonding-interface-engineered vertical stacking and surface passivation. Nanoscale 11, 23139. Geum, D.M., Kim, S.K., Lee, S., Lim, D., Kim, H.J., Choi, C.H., Kim, S.H., 2020. Monolithic 3D integration of InGaAs photodetectors on Si MOSFETs using sequential fabrication process. IEEE Electron Device Lett. 41 (3), 433–436. Gilet, P., Robin, I.-C., 2018. Nanostructures on silicon to solve the active display paradigms. In: SID Symposium Digest of Technical Papers 52-1, pp. 684–687. Han, H.-V., Lin, H.-Y., Lin, C.-C., Chong, W.-C., Li, J.-R., Chen, K.-J., Yu, P., Chen, T.-M., Chen, H.-M., Lau, K.-M., Kuo, H.-C., 2015. Resonant-enhanced full-color emission of quantum-dot-based micro LED display technology. Opt. Express 23, 25. Huang, Y., Hsiang, E.-L., Deng, M.-Y., Wu, S.-T., 2020. Mini-LED, Micro-LED and OLED displays: present status and future perspectives. Light: Sci. & Appl. 9, 105. Kang, C.-M., Kong, D.-J., Shim, J.-P., Kim, S., Choi, S.-B., Lee, J.-Y., Min, J.-H., Seo, D.-J., Choi, S.-Y., Lee, D.-S., 2017a. Fabrication of a vertically-stacked passivematrix micro-LED array structure for a dual color display. Opt. Express 25 (3), 2489–2495.

386

Dong-Seon Lee and Sang Hyeon Kim

Kang, C.-M., Kang, S.-J., Mun, S.-H., Choi, S.-Y., Min, J.-H., Kim, S., Shim, J.-P., Lee, D.-S., 2017b. Monolithic integration of AlGaInPbased red and InGaN-based green LEDs via adhesive bonding for multicolor emission. Sci. Rep. 7, 10333. Kang, C.M., Kang, S.J., Mun, S.H., Choi, S.Y., Min, J.H., Kim, S., Shim, J.P., Lee, D.S., 2017c. Monolithic integration of AlGaInP-based red and InGaN-based green LEDs via adhesive bonding for multicolor emission. Sci. Rep. 7, 1–9. Kang, C.-M., Lee, J.-Y., Kong, D.-J., Shim, J.-P., Kim, S.-H., Mun, S.-H., Choi, S.-Y., Park, M.-D., Kim, J., Lee, D.-S., 2018. Hybrid Full-Color inorganic light-emitting diodes integrated on a single wafer using selective area growth and adhesive bonding. ACS. Photonics. 5 (11), 4413–4422. Kim, S.H., Kim, S.K., Shim, J.P., Geum, D.M., Ju, G., Kim, H.S., Lim, H.J., Lim, H.R., Han, J.H., Lee, S., Kim, H.S., Bidenko, P., Kang, C.M., Lee, D.S., Song, J.D., Choi, W.J., Kim, H.J., 2018. Heterogeneous integration toward a monolithic 3-D chip enabled by III–V and Ge materials. IEEE J. Electron Devices Soc. 6, 579–587. Kim, H.M., Ryu, M., Cha, J.H.J., Kim, H.S., Jeong, T., Jang, J., 2019. Ten micrometer pixel, quantum dots color conversion layer for high resolution and full color active matrix micro-LED display. J. Soc. Inf. Disp. 27, 347–353. Kong, D.-J., 2017. Color tunable monolithic InGaN/GaN LED having a multi-junction structure. Opt. Express 24 (6), 2489–2495. Lee, J.J., 2019. Technology and latest results on fluidic assembly of micro LEDs. In: Micro LED Display Conference, p. 2019. Lee, Y.J., Lee, C.J., Chen, C.H., 2011. Effect of surface texture and backside patterned reflector on the AlGaInP light-emitting diode: high extraction of waveguided light. IEEE J. Quantum Electron. 47 (5), 636–641. Li, Y.-L., Liu, Y.-T., 2020. MicroLED display: the next-generation display technology. Proc. SPIE Adv. Display Technol. X 11304, 113040H. Li, N., Han, K., Spratt, W., Bedell, S., Ott, J., Hopstaken, M., Libsch, F., Li, Q., Sadana, D., 2019. Ultra-low-power sub-photon-voltage high-efficiency light-emitting diodes. Nat. Photonics 13, 588–592. Lin, J.Y., Jiang, H.X., 2020. Development of microLED. Appl. Phys. Lett. 116, 100502. Mun, S.-H., Kang, C.-M., Min, J.-H., Choi, S.-Y., Jeong, W.-L., Kim, G.-G., Lee, J.-S., Kim, K.-P., Ko, H.-C., Lee, D.-S., 2021. Highly efficient full-color inorganic LEDs on a single wafer by using multiple adhesive bonding. In progress. Panth, S., Samal, S., Yu, Y.S., Lim, S.K., 2014. Design challenges and solutions for ultrahigh-density monolithic 3D ICs. In: 2014 SOI-3D-Subthreshold Microelectronics. Technology. Unified Conference. S3S 2014, pp. 1–2. Suzuki, R., Taoka, N., Yokoyama, M., Lee, S., Kim, S.H., Hoshii, T., Yasuda, T., Jevasuwan, W., Maeda, T., Ichikawa, O., Fukuhara, N., Hata, M., Takenak, M., Takagi, S., 2012. 1-nm-capacitance-equivalent-thickness HfO2/Al2O3/InGaAs metaloxide-semiconductor structure with low interface trap density and low gate leakage current density. Appl. Phys. Lett. 100 (13), 132906. Templier, F., Dupre, L., Tirano, S., Marra, M., Verney, V., Olivier, F., Aventurier, B., Sarrasin, D., Marion, F., Catelain, T., Berger, F., Mathieu, L., Dupont, B., Gamarra, P., 2016. GaN-based emissive microdisplays: a very promising technology for compact, ultra-high brightness display systems. In: SID Symposium Digest of Technical Papers 75–1, pp. 1013–1016. Tull, B.R., Basaran, Z., Gidony, D., Limanov, A.B., Im, J.S., Kymissis, I., Lee, V.W., 2015. High brightness, emissive microdisplay by integration of III-V LEDs with thin film silicon transistors. In: SID Symposium Digest of Technical Papers. vol. 46, pp. 375–377, (1). Virey, E., 2017a. “MicroLED displays: hype and reality, hopes and challenges.,” Yole report., p. 21.

Monolithic integration of AlGaInP red and InGaN blue/green LEDs

387

Virey, E., 2017b. “MicroLED displays: hype and reality, hopes and challenges,” Yole report., p. 200. Wong, M.S., Nakamura, S., DenBaars, S.P., 2020. Review—progress in high performance III-nitride micro-light-emitting diodes. ECS J. Solid State Sci. Technol. 9, 015012. Zhang, L., Ou, F., Chong, W.C., Chen, Y., Li, Q., 2018. Wafer-scale monolithic hybrid integration of Si-based IC and III–V epi-layers—a mass manufacturable approach for active matrix micro-LED micro-displays. J. Soc. Inf. Disp. 26, 137–145.

Index Note: Page numbers followed by “f ” indicate figures and “t” indicate tables.

A Acoustic traps, 71 Adhesive bonding green-red integration, 354 hybrid RGB integration, 363 Aerosol jet (AJ) printer, 177 ALD. See Atomic layer deposition (ALD) passivation AlGaInP LEDs, 113 Aligned wafer bonding, 19–21, 40–42 AlInGaP LEDs, 225, 232 Ambient contrast ratio (ACR), 224, 335–336 Analog-to-digital converter (ADC), 298–299 Angular color shift, RGB micro-LEDs ambient contrast ratio, 331 angular distribution tolerance, 332–333 emission patterns, 324–325 far-field radiation pattern, 324 GaInP/AlGaInP multiple quantum wells (MQWs), 324 internal quantum efficiency, 324 Lambertian distribution, 329, 332–333 Macbeth ColorChecker, 330–331 primary colors, 323–324 RGB radiation patterns, 329 sidewall emissions, 325–328 top black matrix, 329, 330f Anti-reflection (AR) coating, 336–337 Artifact-free optoelectrophysiology in vivo demonstration, 163 transient pulse shaping, 162–163 Atomic force microscopy (AFM), 7, 206–208 Atomic layer deposition (ALD) passivation, 33–34, 100–109 Auger recombination, 187–189 Avalanche photodiode (APD) array, 303–305

B Bandgap energy, 209–211, 211f Batwing distribution, 332–333 Beer Lambert’ law, 307–309

Bio-template, 209–211 Blue nanoLEDs, 238–240, 239f Bump-bonding process, 68

C Cadmium selenide (CdSe)-based quantum dots, 174–175 Capping technology, 350–352 Carrier leakage, 187–189 Cell-type-specific optogenetic neuronal stimulation, 131, 132f Channelrhodospin-2 (ChR2), 83–84, 142, 157 Chemical mechanical polishing (CMP), 242–243, 242f Chemogenetics, 130–131 Chinese Hamster Ovary (CHO) cells, 76–78 Clausius-Mossotti factor, 73 Coffee-ring effect, 178–180, 179f Colloidal quantum dots (CQDs), 177, 184–186, 287–293 Color-conversion method, 184, 194–195 Commercial off-the-shelf (COTS) miniature LEDs, 134 Complementary metaloxide-semiconductor (CMOS) electronics, 9, 65–67, 282, 284–286, 299–305 Constant fraction discriminator (CFD), 28–29 Consumer Electronics Show (CES), 5–6 Contact/transfer printing, 293 CQDs. See Colloidal quantum dots (CQDs) Cyclic prefixes (CPs), 298–299

D Deep-brain optical stimulation devices, 133–134, 135f Designer Receptors Exclusively Activated by Designer Drugs (DREADD), 130–131 Dielectrophoresis (DEP), 71–73, 73f 389

390 Digital-to-analog converters (DACs), 22–23, 298–299 Direct wafer-scale transfer, 245–246 Distributed Bragg reflector (DBR), 177, 180–181, 373, 377

E E-beam lithography, 240 Elastomeric stamp, 79–81 Electrical stimulation, 130 Electroluminescence (EL) technique, 28–29 NW LED, 229, 230f of RGB hybrid QD-NR-μLEDs, 183, 184f of RGB quantum dots (RGB QDs), 180–181, 180f of semipolar μLED, 187–189, 188f Electron spin resonance (ESR), 208–209 Electrophysiology, 124, 126 Electrostatic transfer, 257 eLux technology, 345–348 Epitaxial quantum wells (QWs), 294–295 External quantum efficiency (EQE), 31–33, 95–96, 103–104, 108–109, 203–204, 214–219, 217–218f, 225, 335, 337–338 Extraction efficiency, 31

F Fast color-converters epitaxial quantum well nanoplatelets, 294–295 inorganic semiconductor colloidal quantum dots, 290–293 organic semiconductors, 293–294 time dynamics, 287–290 Ferritine, 209–213, 210f First passive-matrix microdisplay, fabrication of, 9–13 Flip-chip bonding approach, 11–13, 19–21, 40–42 Fluid transfer, 256–257 Fluorescence-based spectroscopy (fluorimetry), 64 Forster resonance energy transfer (FRET), 173–174 Frequency division multiplexing (FDM), 297–298

Index

G Gallium nitride (GaN), 282 NWs, 226–227 cathodoluminescence characterization, 227–229, 228f electroluminescence, 229, 230f optoelectronic technology, 58–59 platelets, 238–240 Gallium nitride/indium gallium nitride (GaN/InGaN) multi-quantum-well (MQW) LEDs, 125–126, 143, 144f, 145–147, 157, 159–161 GaN:Eu LED, 236 GaN-on-silicon microLED optoelectrodes conceptual drawing, 125–126, 125f emission efficiency, 143–145 fabrication process, 145–147, 146f monolithic integration, 141 near-cellular-resolution opto-electrophysiology in vivo opto-electrophysiology, 148–149 optical stimulation, demonstration of, 150–151, 151f radiant flux, 147–148 photovoltaic effects, 157 electrical and optical characteristics, 159–161, 160f heavily-doped silicon substrates, 157–159 stimulation artifacts electromagnetic interference, 154–155 potential sources, 153–154 prototype microLED optoelectrode, 152–153, 153f “Geiger mode,”, 67 Green emitters, 111–113 Green platelet LEDs, 243–244, 244f Guided wave optical communications (GWOC), 281–282

H Head-up displays (HUD), 4–5 High electron mobility transistors (HEMTs), 42–44 High-resolution microLED display, 345–348, 369–383

391

Index

Hybrid-integrated prototype LED optoelectrodes, 139, 140f Hybrid MQW red LED, 234

I Indirect transfer techniques, 245 Indium bump bonding, 19–21 Indium tin oxide (ITO), 72, 103–104 Inductively coupled plasma (ICP), 7, 203–204, 213–218 InGaN multiple quantum wells (MQWs), 234 platelets, 241–243, 242f pyramid growth, 240–241 quantum wells, 229 red microLEDs, 225 bandgap engineering, 234–236, 235f growth condition optimization, 236–237, 237f motivation and challenges, 232–233 substrates, templates, or structures, 234 InGaNOS, 234 Ink-jet or spray printing, 293 Integrated circuit (IC) chip, 9 Interconnecting DAC (IDAC), 24–25 Internal quantum efficiency (IQE), 211–213, 218–219, 233 Inter-symbol interference (ISI), 297 Inverse Fast Fourier Transform (IFFT), 298–299

J Just noticeable color difference (JNCD), 332

K KOH. See Potassium hydroxide (KOH) Korea Institute of Machinery and Materials (KIMM), 345–348

L Lambertian distribution, 196 Lambertian emission pattern, 214–215 Lambertian radiation, 305–306 Laser diodes (LDs), 34–36, 58–59, 282 Laser-driven active micro-structured transfer printing, 272–276, 273f Laser-driven programmable non-contact transfer printing, 263–267, 265–266f

Laser-lift-off (LLO), 44–46 Laser transfer technique, 258 “Li-Fi,”, 62–63 Light conversion efficiency (LCE), 184, 185f Light-emitting diodes (LEDs), 58–59, 173–174, 282 Liquid crystal displays (LCDs), 27, 71–72 Low-temperature polycrystalline silicon (LTPS), 232

M Maalox suspension, 309 Magnet-controlled transfer printing, 261–263, 262f Magnetic transfer technique, 257–258 Magnetic tweezers, 71 Mass-transfer techniques, 175, 194–195, 245–246, 246–247f MBE. See Molecular beam epitaxy (MBE) M3D integration technology, 371–372 Mercury discharge lamps, 65 Metal-organic chemical vapor deposition (MOCVD), 6–7, 109–110, 181–182, 186–187 Metal organic chemical vapor deposition (MOCVD)-grown InGaN MQW LEDs, 143, 159–161 Metal organic vapor phase epitaxy (MOVPE), 211–213 Metal-oxide-semiconductor field-effect transistor (MOSFET), 299, 301f Michigan probes, 126, 128–129, 129f, 137–138f Microelectrode arrays (MEAs), 84 Microelectromechanical (MEMS) mirrors, 34–36 Micro-LED based optical wireless communications (OWC) systems novel optical wireless communication systems based on deep-ultraviolet (UV) communication, 315–316 underwater wireless optical communication (UWOC), 311–315 visible light communication (VLC), 283–287

392 Micro-LED technology, biomedical applications characteristics electrical, spectral and optical power, 61–62 micro-LED device format, 60–61 temporal characteristics, 62–63 gallium nitride LED technology, 58–59 GaN LED/CMOS chip-scale microfluorimetry, SPADs, 67–68 micro-LED CMOS driver design, 65–66 short optical pulse generation, 67 single-chip fluorescence microsensor, 68–71 time-resolved microfluorimetry, 64–65 light-emitting dressings and printed LEDs heterogeneous integration, 78–81 smart optical bandages, homogenous irradiance, 81–83 optical source technology, 58–59 optoelectronic tweezers (OET) demonstration and application, 76–78 device integration and operation, 74–76 micro-LED light source, 74 micro-manipulation and, 71–72 structure and operating principle, 72–73, 72f optogenetic neural probes and neural interfaces absorption spectra, 83–84 channelrhodospin-2 (ChR2), 83–84 in vitro experiments, 84–85 in vivo micro-LED-based devices, 85–88 opsin toolkit, 83–84 spatiotemporal modulation, 84 photonic technologies, 58 Micro-light-emitting diodes (Micro-LEDs), 95–96, 223–224 challenges, 225–226 displays, 175–197 advantages, 253 mass transfer techniques, 254f, 255–259 transfer printing techniques, 259–261 InGaN red, 232–237 long-wavelength devices, 111

Index

green emitters, 111–113 red emitters, 113 mass transfer, 245–246, 246–247f mesa etching, 213–215 MOCVD-grown tunnel junctions, 109–110 nanowires (NWs), 226–232 electrical efficiency, 229–232, 231f RGB, 237–244 size-dependent efficiency, 96–98 techniques toward high efficiency, 99 chemical treatment, 105–109 sidewall passivation, 99–104 tunnel junctions, 109–110 Micro-transfer-printing, 79–81 Minimal-stimulation-artifact (MiniSTAR) microLED optoelectrodes fabrication, 161–162 in vivo demonstration, 163 transient pulse shaping, 162–163 μLEDs. See Micro-light-emitting diodes (Micro-LEDs) MOCVD. See Metal-organic chemical vapor deposition (MOCVD) Modulation bandwidth (BW), 283 Molecular beam epitaxy (MBE), 109 Monolithically-integrated microLEDs, 141, 145–147 adhesive bonding, green-red integration, 354 differentiated transfer technology, 348 eLux technology, 345–348 full adhesive bonding technique, 367 high-resolution microLED display, 345–348, 369–383 hybridization method, 345–348 hybrid RGB integration, SAG and adhesive bonding, 363 Osram’s blue-LED/red-phosphor/green LED stacking technology, 350 Ostendo’s method, 350 pick-and-place method, 345–348 selective area growth, blue-green integration, 361 stacking technologies, 352 TFTs, 345–348 wafer bonding, blue-green integration, 354

393

Index

Multiple-input, multiple-output (MIMO), 299–305 Multiple quantum wells (MQWs), 182–183, 186–187, 211–213, 324, 326–328

N

Nanoring (NR) μLED, 181–183 Nanowire LED electrical efficiency, 229–232, 231f electroluminescence, 229, 230f Nanowires (NWs), 226–232 Neuronal activity intracellular recording, 127 modulation methods, 130–131 monitoring methods, 124 multi-channel extracellular recording of, 126–128, 127f Neutral beam etching (NBE) technique, GaN micro-LEDs external quantum efficiency (EQE), 203–204 generation source, 205–209 inductively coupled plasma (ICP), 203–204 KOH wet chemical etching, 203–204 liquid crystal displays (LCDs), 203–204 nonradiative sidewall defects, 203–204 organic LEDs (OLEDs), 203–204 Shockley-Read-Hall nonradiative recombination, 203–204 sub-10-nm nanostructures, application of, 209–219 Nitride microLEDs and displays brightness characterization, 25–27 CMOS IC driver, microLED microdisplays, 22–25 consumer electronics, 1–3, 5–6 first passive-matrix microdisplay, fabrication of, 9–13 full color microdisplay development demonstration, vertically stacked RGB micro LEDs, 44–47 standard side-by-side RGB sub-pixel approach, 36–38 vertically stacked RGB micro LED microdisplay concept, 38–44 heterogeneous integrated high-voltage DC/AC LEDs, 15–19

high dynamic range (HDR), 4–5 hybrid active driving micro LED microdisplays, 19–22 large flat panel displays and medicine, applications in, 47–50 monolithic single-chip high-voltage AC/DC-LEDs, 13–15 operating speed and view angle characterization, 28–31 structure and pixel processing, 6–9 temperature dependence, 28 III-nitride microstructures, 3–4 Non-contact transfer printing, 263–267 Non-polar GaN substrates, 112–113 Non-radiative Auger recombination coefficients, 63 Non-radiative recombination, 380–381 Non-radiative resonant energy transfer (NRET), 173–174 Non-return-to-zero (NRZ), 296–297 NWs. See Nanowires (NWs)

O On-off keying (OOK), 296–297 Optical Camera Communications (OCC), 302 Optoelectrodes bidirectional interfaces, 157–159 optical stimulation, 131–140 Optoelectronic tweezers (OET) demonstration and application, 76–78 device integration and operation, 74–76 micro-LED light source, 74 micro-manipulation and, 71–72 structure and operating principle, 72–73, 72f Optogenetics, 129–133 Optrodes, 133–134 Organic LED (OLED) displays, 27 Orientation-controlled epitaxy (OCE) method, 186–187 Orthogonal frequency division multiplexing (OFDM), 292–293, 297–299 Osram’s blue-LED/red-phosphor/green LED stacking technology, 350 Ostendo’s method, 350

394

Index

P

Q

PECVD. See Plasma-enhanced chemical vapor deposition (PECVD) Phosphor, 176 Photobiomodulation, 81 Photodynamic therapy (PDT), 81 Photolithography techniques, 11–13, 184–186, 293 Photoluminescence (PL) intensity, 211–213 Photomultiplier tubes (PMTs), 65 Photonic crystal (PC) LEDs, 30–31 Photon irradiation, effect of, 206–208, 208f Photoresist mold, 178 Phototherapy, 81–82 Pick-and-place transfer method, 255–256, 345–348 Pixels per inch (PPI), 224, 247–248 Planar silicon microprobes, 126–129 Planck constant, 214–215 Plasma damage removal, 105–106 Plasma-enhanced chemical vapor deposition (PECVD), 99–104, 100f, 213–214 Platelets, 238, 239f Polydimethylsiloxane (PDMS), 79–83, 79f, 246, 247f Polymer optical fibers (POF), 281–282 Polymer waveguide backplanes (PWBs), 281–282 Potassium hydroxide (KOH), 105–108 Power consumption, RGB micro-LEDs ambient contrast ratio, 335–336 chip-size dependent power efficiency, 336–339 LED chip sizes, in RGB subpixels, 341–342 light extraction efficiency, 333–335 power efficiency function, 339 uniform LED chip size, in RGB subpixels, 339–340 Power conversion efficiency characteristics, 31–36 Printing system, 189, 191f Programmable transfer printing, 269–272, 271f Pull-off tests, 261–263, 262f Pulse-amplitude modulation (PAM), 303 Pulse width modulation (PWM), 335, 337

Quadrature amplitude modulation (QAM), 298–299 Quantum-confined Stark effect (QCSE), 187–189, 211–213, 226–227 Quantum-dot-based full-color micro-LED displays, 174–197 coffee-ring effect, 178–180, 179f colloidal QDs (CQDs), 177 color conversion, 184 development in, 189, 192–193t electroluminescence (EL) spectrum, 180–181, 180f magnificent characteristics, 194 Quantum dots (QDs), 173–174 printing system, 189, 191f Quantum dots photoresist (QDPR) patterning technique, 184–197 Quantum wells (QWs), 294–295

R Radiative Auger recombination coefficients, 63 Radio frequency (RF) wireless networks, 62–63, 206–208, 213–214 Readout integrated circuitry (ROIC), 377 Red emitters, 113 Red, green and blue (RGB) micro-LED displays, 237–244 angular color shift ambient contrast ratio, 331 angular distribution tolerance, 332–333 emission patterns, 324–325 far-field radiation pattern, 324 GaInP/AlGaInP multiple quantum wells (MQWs), 324 internal quantum efficiency, 324 Lambertian distribution, 329, 332–333 Macbeth ColorChecker, 330–331 primary colors, 323–324 RGB radiation patterns, 329 sidewall emissions, 325–328 top black matrix, 329, 330f power consumption ambient contrast ratio, 335–336 chip-size dependent power efficiency, 336–339

395

Index

LED chip sizes, in RGB subpixels, 341–342 light extraction efficiency, 333–335 power efficiency function, 339 uniform LED chip size, in RGB subpixels, 339–340 Red platelet LEDs, 243–244, 244f Resistance-capacitance (RC) effect, 287, 306–307 Retinal ganglion cells (RGCs), 84–85 Return-to-zero (RZ), 296–297

S ScAlMgO4, 234, 235f Scanning electron microscopy (SEM), 7, 209–211, 272–274, 273f Selective area growth (SAG) technology, 361–362 blue-green integration, 361 hybrid RGB integration, 363 Semi-polar GaN substrates, 112–113 Semipolar μLED, 187–189, 188f Shape memory polymer (SMP)-based universal transfer printing, 267–272, 268f, 270–271f Shockley-Read-Hall (SRH) nonradiative recombination, 96–97, 99, 380 Short-wavelength photonic integrated circuits (SW-PICs), 281–282 Sidewall damage, 96–97 Sidewall passivation, 99–104 Single-photon avalanche diodes (SPADs), 67–68 Single quantum well (SQW), 211–213 Snell’s Law, 325–326 Solid-state lighting (SSL), 305–306 Spectral efficiency, 297 Spin-on-glass (SOG) etching process, 181–182 Spray coating machine, 189 Stamp printing, 246, 247f Subcarriers, 297–298 Super ink-jet printing (SIJ) technology, 194 Superluminescent diodes (SLDs), 58–59, 282 Surface recombination, 97, 98f

T Tetramethylammonium hydroxide (TMAH), 105–106 Thermal release tape (TRT) stamp, 272–274, 275f III-nitride material system, 96–97 Through-Si-via (TSV) technology, 42, 371 Time-correlated single photon counting (TCSPC), 64–65 Time-dependent Kohn-Sham equations, 206 Time-resolved photo-luminescence (TRPL) measurement, 381–383 Time-to-amplitude converter (TAC), 28–29 Transfer printing techniques, 258–261 latest development, 261–276 principles, 259, 260f

U Ultraviolet (UV) photon irradiation, 203–204, 206–209, 219–220 Underwater wireless optical communication (UWOC), 311–315 U.S. BRAIN initiative, 124

V Vertical cavity surface emitting lasers (VCSEL), 3–4 Video graphics array (VGA) format, 19, 22–23 Virtual reality and augmented reality (VR/AR), 301–302 Visible light communication (VLC), 283–287 fast color-converters for epitaxial quantum well nanoplatelets, 294–295 inorganic semiconductor colloidal quantum dots, 290–293 organic semiconductors, 293–294 time dynamics, 287–290 integrated micro-LED/CMOS, structured VLC/MIMO VLC, 299–305 micro-LED for, 283–287 data modulation, 295–296 on-off keying (OOK), 296–297 orthogonal frequency division multiplexing (OFDM), 297–299

396

W Wafer bonding blue-green integration, 354 multi-color integration, 354–369 Wall-plug efficiency (WPE), 233

Index

Waveguide optoelectrodes, 136–138, 137–138f Wavelength division multiplexing (WDM), 292–293 White LEDs (WLEDs), 176